Quality-Oriented Education Policy in Urban China and Its Association with Social Reproduction of Educational Inequality

Quality-oriented Education Policy in Urban China and its Association with Social Reproduction of Educational Inequality

Jie Wang

A thesis in fulfilment of the requirements for the degree of

Doctor of Philosophy

Social Policy Research Centre

Faculty of Arts and Social Sciences

January 2018

1 ORIGINALITY STATEMENT ‘I hereby declare that this submission is my own work and to the best of my knowledge it contains no materials previously published or written by another person, or substantial proportions of material which have been accepted for the award of any other degree or diploma at UNSW or any other educational institution, except where due acknowledgement is made in the thesis. Any contribution made to the research by others, with whom I have worked at UNSW or elsewhere, is explicitly acknowledged in the thesis. I also declare that the intellectual content of this thesis is the product of my own work, except to the extent that assistance from others in the project's design and conception or in style, presentation and linguistic expression is acknowledged.’

Signed ……………………………………………

Date ………………………………………………

COPYRIGHT STATEMENT ‘I hereby grant the University of New South Wales or its agents the right to archive and to make available my thesis or dissertation in whole or part in the University libraries in all forms of media, now or here after known, subject to the provisions of the Copyright Act 1968. I retain all proprietary rights, such as patent rights. I also retain the right to use in future works (such as articles or books) all or part of this thesis or dissertation. I also authorise University Microfilms to use the 350 word abstract of my thesis in Dissertation Abstract International (this is applicable to doctoral theses only). I have either used no substantial portions of copyright material in my thesis or I have obtained permission to use copyright material; where permission has not been granted I have applied/will apply for a partial restriction of the digital copy of my thesis or dissertation.'

Signed ……………………………………………

Date ………………………………………………

AUTHENTICITY STATEMENT ‘I certify that the Library deposit digital copy is a direct equivalent of the final officially approved version of my thesis. No emendation of content has occurred and if there are any minor variations in formatting, they are the result of the conversion to digital format.’

Signed ……………………………………………

Date ………………………………………………

III PLEASE TYPE THE UNIVERSITY OF NEW SOUTH WALES Thesis/Dissertation Sheet

Surname or Family name: Wang

First name: Jie Other name/s:

Abbreviation for degree as given in the University calendar: PhD

School: Social Policy Research Centre Faculty: Arts and Social Sciences

Title: Quality-oriented Education Policy in Urban China and its Association with Social Reproduction of Educational Inequality

Abstract 350 words maximum: (PLEASE TYPE) This thesis explores the impact of Quality-Oriented Education (QOE) policy in urban China on social inequality. In order to do so the thesis examines the association between the family background of students and their transitions from junior high school (JHS) to regular senior high school (RSHS) (Transition One) and from RSHS to university (Transition Two). The original hypothesis of the study was that the social reproduction of educational inequality would intensify during and after the implementation of QOE. This hypothesis was partially confirmed, but QOE was shown to have inconsistent effects on the two transitions. Alongside the implementation of QOE was the unexpected rise in popularity of commercial extracurricular tutoring (CET). The study found that sufficient hours of CET reduced differences in academic achievement between students with different levels of family cultural capital as indicated by the education of father. QOE shortens the time of students spend in school for primary school and JHS more than for RSHS. Transition One results in rejection of large numbers of underprivileged students and therefore CET has less effect on students in RSHS, because those who are likely to be less competitive have already been selected out. QOE is associated with the increase in educational inequality in Transition One but with the decrease in inequality in Transition Two. The mechanism for this is that students from families with lower cultural capital are more likely to attend CET for sufficient hours in RSHS than in primary or JHS due to the difference in at-school time and the Transition One rejection. Thus, as CET became more popular during and after the implementation of QOE, the effects of family cultural capital decreased but only in respect of Transition Two. The amount of CET required to make a difference in primary and JHS is too expensive for families whose resources are diluted by multiple children and thus the effects of sibling number on Transition One increase in Periods Two and Three compared with Period One. Parents as the indirect educational service receivers are interviewed since they should play a key role in the QOE reform and might be expected to be concerned about the justice of this significant educational reform. The continuation of the examination-oriented system keeps parents evaluating the current educational system as just. Parents also deny the injustice of CET by blaming the victim with the Neoliberalism discourse and arguing there is no absolute equality.

Declaration relating to disposition of project thesis/dissertation

I hereby grant to the University of New South Wales or its agents the right to archive and to make available my thesis or dissertation in whole or in part in the University libraries in all forms of media, now or here after known, subject to the provisions of the Copyright Act 1968. I retain all property rights, such as patent rights. I also retain the right to use in future works (such as articles or books) all or part of this thesis or dissertation.

I also authorise University Microfilms to use the 350 word abstract of my thesis in Dissertation Abstracts International (this is applicable to doctoral theses only).

………………………………………… ……………………………………..… ……….……………………...…….… Signature Witness Signature Date

The University recognises that there may be exceptional circumstances requiring restrictions on copying or conditions on use. Requests for restriction for a period of up to 2 years must be made in writing. Requests for a longer period of restriction may be considered in exceptional circumstances and require the approval of the Dean of Graduate Research.

FOR OFFICE USE ONLY Date of completion of requirements for Award: 31/01/2018

IV Table of Contents

Acknowledgements ...... iv Abbreviations ...... v List of Figures ...... vi List of Tables ...... vii Chapter One. Introduction ...... 1 I. Introduction ...... 1 Rationale for introducing QOE ...... 1 Proposed changes under the QOE reform ...... 2 Inequality of students’ education influenced by QOE and parents’ evaluation ...... 3 II. Research Question, Approach and Outline of Chapters ...... 7 III. Introduction to Chinese Education System ...... 10 Public school, educational stages and tuition fees ...... 10 Primary and secondary school size ...... 12 Admission approach ...... 13 Educational expansion ...... 14 IV. Introduction to Quality-oriented Education ...... 15 Definition of Quality-oriented Education ...... 15 Difference between EOE and QOE ...... 16 Timeline of implementing QOE policy ...... 17 Chapter Two. Literature Review ...... 19 I. Social Reproduction of Educational Inequality ...... 19 II. Trends in Social Reproduction of Educational Inequality in Contemporary China ...... 21 Indicators of family background and social reproduction of educational inequality ...... 22 The trend in social reproduction of educational inequality after the Cultural Revolution 24 Explaining changes in social reproduction of educational inequality by periods ...... 28 Implications for research questions and methodology of the thesis ...... 31 III. Quality-oriented Education and Educational Inequality ...... 34 “Su zhi” (quality), “su zhi jiao yu” (QOE) and neoliberalism ...... 34 The intention of QOE policy: whether it was designed to improve equality? ...... 36 QOE policy implementation and its effects on educational equality ...... 37 IV. QOE and Justice Evaluation of Educational Resource Distribution ...... 42 Chapter Three. Analytical Framework and Methodology ...... 48

I. Analytical Framework ...... 48 II. Methodology ...... 51 Chapter Four. The Change of Social Reproduction of Educational Inequality Before, During and After QOE...... 55 I. Introduction ...... 55 II. Data, Variables and Model ...... 61 Dependent variables ...... 62 Independent variables ...... 62 Method and modelling procedures ...... 67 III. Results ...... 68 Change in a student’s opportunity of transiting from JHS entering RSHS ...... 68 Change in a student’s opportunity of transiting from RSHS to University ...... 74 IV. Summary and Discussion ...... 79 Chapter Five. A Different Version of the Story to Explain the Changes ...... 85 I. Introduction ...... 85 II. Data and Methods ...... 87 III. Findings and Analysis ...... 91 What changes has QOE caused in reality? ...... 91 Out-of-school (extracurricular) time ...... 91 Unexpected consequence (CET) and its association with family background factors ...... 99 Difference in QOE between compulsory education and RSHS ...... 117 IV. Summary and Discussion ...... 119 Chapter Six. Can CET Improve Students’ Academic Performance? ...... 132 I. Introduction and Hypothesis ...... 132 II. Data, Variables, and Model ...... 140 Chinese Family Panel Studies and China Education Panel Survey ...... 140 Academic achievement ...... 142 CET ...... 143 Covariates ...... 144 Method and modelling procedures ...... 146 III. Results ...... 152 Treatment-effects analysis ...... 152 CET and family background factors ...... 156 Interaction effects analysis ...... 158

ii IV. Summary and Discussion ...... 163 Chapter Seven. QOE, CET and Educational Inequality: Parents’ Justice Evaluation ...... 167 I. Introduction, Methods and Data ...... 167 II. Results ...... 170 Parents’ general evaluation of QOE ...... 170 Parents’ justice evaluation of CET: the justice principle of need ...... 178 Why didn’t parents think CET was an unnecessary inequality? ...... 188 III. Summary and Discussion ...... 196 Chapter Eight. General Discussion...... 200 I. Summary ...... 200 Implementation gap and the popularity of CET, especially the academic CET ...... 202 Family background effects and their different associations with CET ...... 205 JHS vs. RSHS and effects of their difference on family background-CET association ...... 208 Impact of other social and policy changes ...... 209 “No absolute equality”: Parents’ justice evaluation of QOE and its consequence CET .... 210 II. Relationship to Previous Research ...... 214 Trend in social reproduction of educational inequality and effects of QOE...... 214 QOE, other social changes and their relationship with the popularity of CET ...... 219 Family socioeconomic factors, CET and academic achievement ...... 222 The justice evaluations of QOE policy and its unexpected consequence CET ...... 224 Chapter Nine. Conclusion ...... 230 I. Theoretical and Policy Implications of Findings ...... 230 II. Limitations and Future Research ...... 234 References: ...... 236 Appendix ...... 264

iii Acknowledgements

I would firstly like to express my sincere gratitude to my supervisors Professor Deborah Brennan and Professor Ilan Katz for your support, motivation, patience and generosity throughout the process of my thesis researching and writing.

I would also like to thank my thesis panel member Associate Professor Xiaoyuan Shang, my current and former postgraduate coordinators Associate Professor Bruce Bradbury, Dr Fiona Hilferty and Dr Trish Hill, and my Master thesis supervisor Professor Xiaotian Feng for your insightful comments and kind assistance.

My particular thanks go to my friends Zimin Tan, Dr Phillipe Adam, Associate Professor Bingqin Li and my Rally-up writing group members Professor Karen Fisher, Dr Yuvisthi Naidoo, Dr Katrina Moore and Dr Jan Idle whose support and encouragement have been of such great value along this challenging and exciting way. To my HDR fellows Cliff Chen, Jialiang (Angela) Cui, Peter Davidson, Qian Fang, Quanchai Kerddaen, Deborah Lutz, Sallie-Anne Moad, Naveed Nor, Dr Alec Sewell, David Scott, Cathy Thomson, Cris Townley, and Horas Wong, many thanks for your friendship.

I would like to address my acknowledgement to those parent, teacher and student interviewees and gatekeepers in my field work. Your valuable contribution makes this thesis. Thank you.

To my girlfriend Chunqing Dai, I want to say hearing your voice is always the happiest moment after a long work day. I owe you a great debt of gratitude.

I am most grateful to my parents Jialing Liu and Songbai Wang, who are always being there for me. I owe it all to you.

iv Abbreviations

CET Commercial extracurricular tutoring

CEPS China Education Panel Survey

CFPS Chinese Family Panel Studies

EOE Examination-oriented education

JHS Junior high school

MMI Maximally maintained inequality

PSM Propensity score matching

QOE Quality-oriented education

RSHS Regular senior high school

SHSAE Senior high school admission examination

v List of Figures

Figure 1.1 Rate of student transitions from JHS to RSHS and from RSHS to university between 1991 and 2011 ...... 15 Figure 2.1 Research on education and stratification in developing countries...... 21 Figure 3.1 The impact of QOE policy on social reproduction of educational inequality and parents’ justice evaluation of the impact ...... 48 Figure 3.2 Classifying mixed methods research in terms of priority and sequence ...... 53 Figure 3.3 Priority and sequence of the thesis’s mixed methods ...... 54 Figure 6.1 Testing Hypothesis 6.3 ...... 150 Figure 6.2 Overlap plots of propensity score ...... 155 Figure 6.3 Predictive student achievement by father’s education by tutorial group .. 162

vi List of Tables

Table 1.1 Educational Stages in China ...... 11 Table 4.1 Year Ranges of Cohorts across Periods ...... 57 Table 4.2 Descriptive Statistics for All Variables in Analyses ...... 66 Table 4.3 Logistic Models Predicting Probability of Student Transiting from JHS to RSHS ...... 70 Table 4.4 AME / APP of Family Background Factors on Transition from JHS to RSHS .... 72 Table 4.5 Logistic Models Predicting Probability of Student Transiting from RSHS to University ...... 76 Table 4.6 AME / APP of Family Background Factors on Transition from RSHS to University ...... 79 Table 4.7 Hypothesized Changes in Social Reproduction of Educational Inequality ...... 81 Table 4.8 AME and APP of Family Background Factors on Educational Transitions Without Educational Expansion Measures ...... 83 Table 5.1 Interviewee Number of Each Group and Related Information ...... 90 Table 6.1 AME and APP of Family Background Factors on Educational Transitions .... 136 Table 6.2 Trends in Effects of Family Background on Educational Transitions ...... 137 Table 6.3 Descriptive Statistics for Treatment-effects Analysis ...... 148 Table 6.4 Descriptive Statistics for Interaction-effects Analysis ...... 151 Table 6.5 Fixed Effects Model predicting a student’s examination score ...... 153 Table 6.6 Comparing Coefficients and Standard Errors of Different Models ...... 156 Table 6.7 Logistic Regression of Tutoring on Family Background and Others ...... 157 Table 6.8 Interaction Effects Analysis between CET and Family Background ...... 160 Table 6.9 Interaction Effects Analysis with Four Tutoring Levels ...... 161 Table 7.1 Typology of Education Services under Parents’ Evaluation ...... 183

vii Chapter One. Introduction

I. Introduction

In 1993, the Chinese central government formally launched a reform policy named “Quality-oriented Education” (su zhi jiao yu, 素质教育1, hereinafter QOE) (CCCPC and State Council 1994). QOE was to replace the old “Examination-oriented Education” (ying shi jiao yu, 应试教育, hereinafter EOE) system and to educate students for all-around development. The reform is comprehensive and includes changes in educational ideology, the curriculum, teaching approaches and evaluation methods (Liu and Dunne 2009). However, even though the old EOE system was not designed for educational equality and students from privileged families had advantages in educational transitions, this reform still raises concern from both scholars (Lou 2011; Wang 2011a) and ordinary parents (Li 2017a) that it will lead to a significant increase in the existent advantages of privileged students, or in other words, an increase in “social reproduction of educational inequality”2 in China, which term is used by Wu (2008) and defined as the persistent linkage between family origin and children’s educational attainment. This thesis aims to explore the impact of QOE policy on social reproduction of educational inequality and parents’ evaluation of the impact in terms of social justice, which has not been thoroughly explored by previous research.

Rationale for introducing QOE The changes to Chinese education system are comprehensive since the abandonment of Maoism and the inauguration of Reform and Opening-up in the late 1970s – from early to higher education, from formal to shadow education and from academic to vocational education (Vickers and Zeng 2017). Stimulated by the needs of the emerging market economy in the Reform and Opening-up era, especially the preparation of a

1 QOE is also translated as “education for all-around development” in some other articles (e.g., Liu and Zhang 2009). 2 There are other terms used to describe the same phenomena, such as social reproduction of educational inequality and educational inequality by family origin (background). 1 modernised workforce, since 1985 the Chinese central government has launched a series of reforms aimed at the education system (Liu and Dunne 2009). Besides the practical needs of the market economy, these reforms have also reflected the new influence of the Western educational theories and practices that replaced the previous influence from the Soviet Union in the early stage of the Reform and Opening-up era (Tan and Chua 2015; Yan 2002: 5-17). The QOE policy, introduced in 1993, is the hallmark of these reforms (Cravens, Liu and Grogan 2012). The old EOE was accused of overemphasising the selection function of education, in other words, acquiring academic knowledge and examination skills by rote and repeated training, which imposes heavy academic burden upon students, damages their wellbeing and stifles their creativity (Qiao 2014; Jin and Tang 2004). In contrast, the aim of QOE is to establish a new vision for education to alleviate the academic burden, to develop all students rather than a few “intellectuals” (rather than overemphasising the selection function of education), to develop their creativity, problem-solving skills and lifelong learning attitudes, and to enhance students’ all-around “qualities” (quality, “su zhi”, 素质) including morality, academic potential, physical health (sport) and aesthetic appreciation (art) (CCCPC and State Council 1999; Little 2000; Jin and Tang 2004).

Proposed changes under the QOE reform The QOE reform was designed to introduce new educational ideology, curriculum, teaching and learning methods and admission methods (Liu and Dunne 2009). In this study, I consider the specific measures of curriculum, teaching approaches and admission methods associated with social reproduction of educational inequality and that provide the clues for the research questions and assumptions. First, the curriculum for different educational levels has been changed to various extents. One of the most notable changes in the QOE curriculum is that sport and art are intended to be given more importance. In EOE, sport and art courses were often reduced or even excluded to give way to academic subjects that are examined in the admission process to higher level school. Second, QOE is aimed at introducing new teaching and learning approaches, which are student-oriented, pay attention to students’ own experiences, and emphasise students’ active and initiative participation in the teaching process (Guo 2010). Instead of the teacher lecturing all the time, as was the case in EOE, the new approaches

2 encourage students to participate more actively in the course, for example by making presentations and require students to explore, discuss and cooperate in class. The changes in the content of curriculum and teaching and learning methods are mainly the result of the “curriculum reform”, with the last two rounds of curriculum reform (1993– 2000 and since 2001) highly associated with the QOE policy (Xie, Ma and Zhang 2013). Fourth, even though standardised examinations organised by educational authorities are still the major approach for admission to different levels of education, some elite or “key” (“zhong dian”, 重点) schools and universities are allowed to admit a few “talented” students through their own admission systems. This usually involves an interview, examination of knowledge not taught by formal schools and/or recruiting students with an achievement in Science Olympiads or other competitions such as humanities and social sciences, art or sport. Besides introducing these new reform measures, the QOE policy has also tried to alleviate the heavy academic burden on students, which was one of the most crucial reasons for reforming the old EOE system (Jin and Tang 2004: 45-51). Specific measures have included shortening the time spent in schools on weekdays, prohibiting extra courses at weekends and reducing homework with the intention of encouraging students to participate more in non-academic extracurricular activities. At the end of this chapter, I provide a more detailed introduction to the QOE policy, which includes the official definition of QOE, the difference between EOE and QOE, and the timeline of implementing QOE Policy.

Inequality of students’ education influenced by QOE and parents’ evaluation Even though the policy documents and official media advocate the numerous merits of QOE, an unintended consequence of the four QOE measures is that they are likely to increase the probability of social reproduction of educational inequality. I discuss the new QOE teaching and learning methods at first. Enlightened by Bernstein’s (1975) linguistic reproduction and Bourdieu and Passeron’s (1977) cultural reproduction, I assume that compared to the lecturing and cramming method that dominated EOE teaching and learning, the new QOE teaching and learning methods that encourage initiative, exploration and creativity could adopt the linguistic and cultural forms and practices of the middle and upper social classes, and thus benefit students from those social classes. However, my study is not focused on this aspect because the changes of

3 teachings and methods might be subtle and complicated meaning that research on their effects on social reproduction of educational inequality would require long-term participant observation and, more importantly, according to interviews with school teachers, the new teaching and learning methods have not really been employed in practice.

The more notable phenomenon attracting greater concern is the rise of commercial extracurricular tutoring (also known as shadow education, cram school, private supplementary tutoring and so on; hereinafter CET). According to Bray and Kwok’s (2003: 612) definition, CET or private supplementary tutoring is only concerned with “academic subjects taught in mainstream schools”, “provided by tutors (and the companies which employ these tutors) for financial gain”, and “provided outside school hours, mostly in the evenings, at week-ends, and during vacations”. In China, there are usually three main categories of providers of CET: “commercial enterprises, individuals such as university students who operate on an informal basis, and school teachers who provide tutoring in addition to their regular duties” (Zhang and Bray 2016: 1). According to a nationwide research report about shadow education in China (Yang 2016), CET as a commercial industry has been developing only for approximately two to three decades and the number of attendees is still increasing. The report found that in 2014 36.7% of the total students in China attended CET and in metropolises such as Beijing, Shanghai, Guangzhou and Shenzhen, the rate could reach up to 70%. In Chapter Five (pp. 99-101), I will describe CET further and define the CET term used in my thesis.

With the implementation of QOE, particularly its measure of alleviating students’ academic burden, the time students spend in school was supposed to be reduced. However, CET, which in China is not only for academic subjects but also for art, sport, or other non-academic subjects, has become popular (Liu 2016), and takes up time that was previously allocated to the school curriculum. In particular, the popularity of art and sport CET can be easily related to the curriculum reform, which gives more attention to students’ art and sporting qualities. Compared to the almost free compulsory education

4 (primary school and junior high school [JHS]) and low fee regular senior high school1 (RSHS), CET is for-profit, and thus rather expensive. Therefore, many underprivileged students cannot afford CET fees, with the equipment and one-to-one or studio tutors2 of many art and sport CETs even not affordable for lower-middle class- families. However, CET in academic subjects directly prepares the attendees for the competitive admission examinations into RSHS and university. Even CET in non-academic subjects such as art, sport and Mathematical Olympiad are usually aimed at the new special admission process that recruits a small proportion of students with “strong points” or “all-around development”, especially in the aspects of art and sport in the name of QOE reform, for example the “autonomous university enrolment” (“zi zhu zhao sheng”, 自主招生) as an addition to the regular national university admission examination (“gao kao”, 高考) (Xun and Wang 2011). Therefore, both the academic and non-academic CET under the QOE policy are likely to benefit the more privileged students in the competitive educational admission examinations. Several studies have recognised this phenomenon (Zhang and Bray 2015; Zhang and Bray 2016; Lou 2013; Wang 2011c). Based on the above observation, I propose the first research question of this study: How has the social reproduction of educational inequality changed alongside the implementation of QOE policy? My hypothesis is that it has been increasing, which will be demonstrated. However, according to the QOE policy document (CCCPC and State Council 1994), there was no measure designed to promote CET aimed at the competitive educational admission examinations. Therefore, the second research question of this study is: Is the popularity of CET associated with the implementation of QOE policy and, if so, how is CET as a mediator associated with QOE and then associated with students’ admission opportunities?

If the above associations exist, the following research question is: how are the parents’ evaluation of QOE policy in terms of justice impacted by the popularity of CET under the

1 In the senior high school stage, there are two tracks after the senior high admission examination usually organised by the city-level governments: the academic track namely RSHS and the vocational track including a variety of vocational senior high schools, and the RSHS admission is much preferable but more competitive. 2 Compared with lecturing in a big classroom, the one-to-one, studio or small-group tutoring is more often used in art and sport CET.

5 QOE policy? It is important for the success of QOE policy to figure out parents’ evaluation since they are the key stakeholder group, and, on behalf of their children, they are the receivers of educational services of QOE. In addition, parents in China are usually sensitive to a government policy affecting their children’s educational admission opportunities and an “unjust” policy could result in parents’ direct actions, for example a demonstration or street protest1. However, when several education authorities tried to inhibit the popularity of CET2, parents were strongly opposed to this policy (Ruan 2011; Hu 2014). There could be a number of reasons for parental support of CET. My research interest is whether parents’ opposition indicates they evaluate CET under the QOE policy as just, whether the evaluation of justice is simply because there is no significant increase in educational inequality under the QOE policy, or because parents intentionally dismiss the educational inequality involved in the mechanism of CET under the QOE policy. If the answer is the latter, what arguments will parents employ to justify the fact that underprivileged students could be worse-off in education admissions as they cannot afford expensive CET?

As I mentioned in the first paragraph, this thesis aims to research the impact of QOE policy on social reproduction of educational inequality and parents’ evaluation of the impact in terms of social justice. However, I only focus on the educational inequality in students’ transition opportunities from JHS to RSHS and then from RSHS to university (see the third section of this chapter for an introduction to the Chinese educational system) in an urban area for the following reasons. First, the QOE reform, initially aimed at primary education and regular secondary education, was extended to vocational secondary education and higher education sectors at the end of the 1990s (Yu 2008). However, since vocational education and higher education were already vocation- oriented, the focus of the reform was on the compulsory education (primary school and JHS) and RSHS. In terms of the reasons for my study focusing on the urban area, at first there was a significant structural gap in education between the urban and rural areas in

1 An incident in 2015 involved a massive protest against the Ministry of Education who allocated a part of the quota of university enrolment places from Jiangsu and Hubei Province to several middle and western undeveloped provinces (see Ai (2016) for a news report). 2 For example, the Department of Education in Chengdu City prohibited the public education institutions from lending their facilities to CET institutions (Ruan 2011).

6 China due to the household registration system (“hu kou”, 户口) (Yang 2006a; Zhang 2005) and some researchers have observed much greater difficulty in implementing QOE in rural areas than in urban areas due to the shortage of resources (Wang 2006). There is evidence that the examination-oriented education is still dominant in rural areas (Wang 2006), with this thesis aiming to research the effects of QOE. Second, If CET plays an important role in the association between QOE and social reproduction of educational inequality, the inequality in access to CET between urban and rural students should be treated differently from the inequality between different socio-economic groups because the former inequality is highly associated with the factors in the supply of CET such as more universities with students desiring to work as part-time tutors and more CET centres due to the population density with public transport in the urban area (as cited in Zhang and Bray 2015; as cited in Zhan et al. 2013). Third, the phenomena of peasant labour migration1 (Zhu 2002; Solinger 1999) and the “left-behind children”2 (Tan 2011) in rural areas could be additional factors explaining the change of educational reproduction in rural areas, which makes the discussion too complicated to be covered by just one PhD thesis.

II. Research Question, Approach and Outline of Chapters

To achieve the above research aim, in Chapter Two I first conduct a literature review. Based on the literature review, Chapter Three proposes the analytical framework of the thesis—The impact of QOE policy on social reproduction of educational inequality and

1 In general, peasant labourers are still registered as “peasant” under the “hu kou” or household registration system. They are not an urban citizen and often cannot enjoy the welfare in the city including urban public schools. If they want their children to obtain an education in the urban region, they usually have to pay an expensive extra fee to a public school or send their child to a private school where the quality if teaching is usually very poor. To date, the problem is still “being processed” (He 2017). Furthermore, even if they can take their children with them, child care is often a problem since they usually work late or live in a dormitory. In addition, many of them change their work location frequently while changing their child’s enrolment is not that easy. Because of the above and other unmentioned reasons, many parents have to leave their children at their rural home, where they are taken care of by their grandparents. In China, these children are called “left-behind children” (“liu shou er tong”, 留守儿童) and their education achievement is deeply affected by separating from their parents (for a review, see Tan (2011)). A more narrative report about these children in English has been published recently by BBC News (Sudworth 2016). 2 See the above footnote.

7 parents’ justice evaluation of the impact—and provides a summary of the methodology employed in the thesis to explain why my research uses mixed methods and the type of mixed methods applied. Greater detail of methods utilised are reported in the method section of each of the results chapters (Chapters Four to Seven) for the convenience of reading. According to the theoretical framework, the research questions of the four results chapters are as follows: (1) What is the changing pattern of social reproduction of educational inequality along with the implementation of QOE? (Chapter Four) (2) What is the mechanism of QOE influencing the social reproduction of educational inequality? Is the popularity of CET the mechanism? If so, how is CET as a mediator associated with QOE and then associated with students’ admission opportunities? (Chapters Five and Six) (3) What is the parents’ perceived changes of social reproduction of educational inequality? Do they evaluate those changes as being fair or not? What are their explanations for the evaluation? (Chapter Seven).

The framework also guides the analyses in the following chapters and the research approach (i.e., mixed methods mainly including second-hand survey data analysis and interview) are explained in greater detail in the four results chapters. In this chapter, I briefly describe the analysis methods of each chapter. The aim of the first three of the four results chapters is to display the “reality” of changes of social reproduction of educational inequality under the QOE reform, which is the basis for parents perceiving and then making their justice evaluation. In Chapter Four, I employ the binary logistic regression model, more specifically Wu’s (2013) revised version of Mare’s (1981) logistic response model of school continuation (transition) to separately estimate the interaction effects of the QOE policy period dummy variable and family background indicators on a student’s probability of entering RSHS and then their probability of transiting from RSHS to university. This aims to examine whether the effects of family background indicators on a student’s transition opportunity differ between before, during and after the implementation of QOE policy.

8 In Chapter Five, I further investigate the mechanism of how the QOE policy changes the association between students’ family background and their opportunities of entering RSHS and then transiting from RSHS to university to illustrate the probable causes for the change patterns revealed in Chapter Four by drawing on interviews with parents, students and school teachers and analysing relevant policy documents. I first report what specific changes the QOE policy probably caused in reality instead of treating the policy text as the practice and then illustrate how the implementation of QOE policy probably makes a major contribution to the popularity of CET. Third, I describe how CET as an unexpected consequence or by-product of QOE policy mediates the association between family background indicators and a student’s educational achievement. I draw the key hypotheses from the above description and test them in Chapter Six with second-hand survey data.

In Chapter Six, I first demonstrate that CET can improve students’ academic achievement, using the fixed effect model and propensity score matching methods. Then, I estimate the effects of family background indicators on a student’s probability of attending CET with the binary logistic model. Finally, I investigate whether attending CET can probably benefit underprivileged students more than their privileged peers by estimating the difference in effects of family background indicators on a student’s academic achievement between CET attendees and non-attendees, and predicting whether the academic achievement gap between attendees and non-attendees will narrow as the family background indicators increase.

In Chapter Seven, I employ the data from qualitative interviews with parents to illustrate parents’ general views of the reform replacing EOE with QOE, their view of QOE’s unexpected consequence, i.e., the popularity of CET, in terms of its necessity and how to fund it, and their justice evaluation of CET and their arguments for the evaluation.

In Chapter Eight I begin by summarising the main empirical findings displayed in the previous four chapters and discussing whether the parents’ arguments for the justice evaluation of the QOE policy, especially its unexpected consequence are supported by the reality of the social reproduction of educational inequality under the QOE or how

9 the arguments are related to the reality. In addition, I relate my findings to previous research to show how my research fits with other studies in the field to make a contribution to knowledge.

I conclude in Chapter Nine by discussing the theoretical implications of this chapter and its implications for current policies and future reforms, the limitations of the study and the future research required to solve those limitations.

III. Introduction to Chinese Education System

To facilitate readers who may not be very familiar with the context of the thesis, I briefly describe the education system in China, which is focused on primary school and junior and regular (academic) senior high school stages as other parts of my research are.

Public school, educational stages and tuition fees In general, the education system in China is strictly regulated by educational authorities at different government levels (Zhao and Zhang 2016) and most educational institutions are public. Private schools in primary and secondary education were officially legalised in 1997 by the State Council (the central government of China) (State Council 1997); however, until 2009 more than 93%1 of students in those two educational stages were still attending public schools.

The educational stages in China usually include early childhood education, primary school, junior and senior high school and tertiary education (see Table 1.1). Primary school and JHS are compulsory. Senior secondary education includes regular (academic) and vocational senior high schools while the tertiary education level includes three-year

1 The proportion is calculated by the author using the data source from the National Bureau of Statistics of China (NBSC 2016a, 2016b). In terms of private schools, the teaching quality of some private schools set up for children of “peasant-worker” in the city is rather poor (see Footnote 1, p. 7) and urban students usually do not enter these schools. Meanwhile, many good private schools are foreign language schools, and thus their students will not attend RSHS and university admission in China and will study aboard (Mangin 2015). Therefore, compared to public schools, the effects of private schools on the social reproduction of educational inequality based on the RSHS and university admission in China could be rather small. 10 vocational colleges and four-year bachelor’s degree universities 1 , and then postgraduate education.

Table 1.1 Educational Stages in China Typical Age Educational Stages Level Compulsory 2–5 Early childhood education # Varies No 6–11 Primary school Grade 1–6 Yes 12–14 Junior high school Grade 7–9 Yes 15–17 Regular/vocational senior high school # Grade 10–12 No 18–22 3-year vocational college/4-year bachelor’s Varies No degree university ≥23 Postgraduate* Varies No Note: * including professional or academic master’s degree and doctoral degree. # The tuition is free in some underdeveloped minority regions and a few coastal developed cities.

The rates for public and private school and university tuition and/or sundry fees are usually set by city-level and/or provincial governments 2 and are not adjusted for inflation but increased every few years. Some underprivileged students are entitled to fees abatement or waiving of fees and other subsidies. Even though compulsory education in urban areas did not become free until 2008, the fees for textbooks and sundries (no tuition fees) in a public primary school or JHS were rather low compared to family income. For example, the annual fees (CN¥ 666 or approximately AU$ 125 for two semesters3) was less than 4% of the annual disposable income per capita of urban residents of Nanjing Municipality (CN¥ 17,538 or approximately AU$ 3,3094) in 2006. RSHS tuition has never been free but is also low. For example, the annual tuition fees of a top public RSHS (CN¥ 1,700 RMB or approximately AU$ 320 for two semesters5) was less than 4% of the annual disposable income per capita of urban residents of Nanjing Municipality (CN¥ 46,104 or approximately AU$ 8,7006) in 2015. RSHS tuition fees have been frozen since 2004 while the annual disposable income per capita keep increasing.

1 The three-year vocational college and four-year bachelor’s degree university are both called “university” in the thesis, unless stated otherwise. 2 The administrative divisions of the People's Republic of China consist of the provincial level, the prefectural (city) level, the county (district) level, the township level and the village level (informal) from high to low. 3 Source: Nanjing Daily (Shi and Qian 2006) 4 Source: the website of Bureau of Statistics of Nanjing Municipality (BSNM 2016). 5 Source: the website of Nanjing Municipality (Price Control Administration, Finance Bureau and Bureau of Education 2017a). 6 Source: the website of Bureau of Statistics of Nanjing Municipality (BSNM 2016) 11 The tuition fees for VSHSs of different types and qualities in Nanjing Municipality varied from CN¥ 800 to 3,1001 (frozen rate since 2000) in 2017. In terms of vocational and academic undergraduate programs, students started to be charged tuition fees in 1989. The university tuition fees were largely increased between 1997 and 20052 and the increase rate largely exceeded the increase rate of family income during the same periods, which was seriously criticised for “industrialization of education” and damaging the educational fairness (Yang 2006a: 106-126). Under this serious criticism, university tuition fees were frozen for several years. In 2001, the annual tuition fees for a normal bachelor’s degree program of a university in Nanjing Municipality were approximately CN¥ 4,600 (approximately AU$ 920) while the annual disposable income per capita of urban residents of Nanjing Municipality was only CN¥ 8,848 (approximately AU$ 1,770)3. The tuition fees were frozen from 2011 to 2013 while the annual disposable income per capita of urban residents of Nanjing Municipality increased to CN¥ 39,881 (approximately AU$ 7976) 4 in Nanjing Municipality in 2013. Finally, postgraduate education is still free for most students of academic master’s degrees and PhD programs after balancing subsidy and scholarship against tuition fees while the professional degree programs are not.

Primary and secondary school size Primary and secondary schools in China are usually rather large in size. For example, the government of Nanjing Municipality requires that primary schools must have at least three classes per grade and JHS or RSHS at least six classes per grade and each class usually contains 45 students (GNM 2005). Therefore, the students of a school are usually from different communities with diverse family backgrounds.

1 Source: the website of Nanjing Municipality (Price Control Administration, Finance Bureau and Bureau of Education 2017b). 2 Different provinces had different schedules for the university tuition fees increase, but the large increases usually happened between 1997 and 2005 (Yang 2006a: 106-126; Chen and Zong 2014). 3 Source: the website of Bureau of Statistics of Nanjing Municipality (BSNM 2016) 4 Source: the website of Bureau of Statistics of Nanjing Municipality (BSNM 2016) 12 Admission approach According to Compulsory Education Law (Article 12, 2015 Amendment, 义务教育法 [2015 年修订] 第 12 条), school-age children can enrol in a public school near their residence without taking any examination (SCNPC2015a). However, some parents like to choose a public school that is outside their school district (if these schools still have extra enrolment places after satisfying local students’ enrolment) or a private school for its higher teaching quality. This is despite the fact that the Compulsory Education Law (Article 12, 2015 Amendment) and education authorities are trying to relatively equalise the teaching quality of compulsory education schools to decrease the proportion of “school-choosing” students. For school-choosing, parents need to pay an extra fee and their children need to take a special entrance assessment even though any formal examination is prohibited by education authorities (Yang 2017). School-choosing is not encouraged by governments and in some cities, for example, Hangzhou City, the public schools in the compulsory education stage have been prohibited from accepting students outside their school district since 2012 (Jin 2012).

If they want to enrol in a senior high school, students need to take the senior high school admission examination (“zhong kao”, 中考, hereinafter SHSAE), which is usually organised by a city-level education authority. However, as China is a unitary state SHSAE is still regulated by the directives of the Ministry of Education in principle (e.g. one directive issued in 2016 Guiding Opinions of the Ministry of Education on Further Advancing Reforms to the Student Examination and Admission System to Senior High School (教育部关于进一步推进高中阶段学校考试招生制度改革的指导意见 ) (Ministry of Education 2016). Students go through an application system where they may choose the city-owned or district (county)-owned senior secondary schools where they wish to study in an order of preference. Whether they receive an offer of admission from a RSHS or a VSHS is usually dependent on their SHSAE scores. There is usually a clear hierarchy of RSHSs in a city and the ranking depends on the admission possibility of these RSHSs’ students to a university or a good university. Similar to a university, the ranking of a RSHS is associated with its admission scores. In addition, the admission scores of a RSHS are usually higher than a VSHS because the students of the former have

13 a much higher probability of entering higher education. Actually, the majority of vocational senior high schools are not aimed at preparing their students for higher education. In 20081, the proportion of JHS graduates admitted to vocational senior high schools compared to the whole JHS graduates admitted to vocational or regular senior secondary schools was approximately 40% (Ministry of Education 2009).

The process of admission to university is similar to the one to senior high school. However, the UAE is usually organised by the provincial-level educational authorities under the regulation of the Ministry of Education. Students can apply to higher education institutions both inside and outside their provinces. Each provincial-level administrative division is assigned a quota of students to be admitted to different kinds of universities by the Ministry of Education before the higher education entrance examination, and thus students compete at the provincial level rather than at the national level. However, even though the Ministry of Education claimed it was further improving the fairness of the admission quotas between different regions (Ministry of Education 2017), the formula for calculating the quota remains a secret and the fairness of the quota is in doubt, which resulted in 2016 in a mass protest of parents in three provinces against allocating some of their province’s admission quota to other underdeveloped regions (Ai 2016).

Educational expansion Even though access to RSHS and university is still competitive, China has largely expanded access to senior high schools including RSHS since the passing of the Compulsory Education Law ( 义务教育法 ) in 1986 (Lin and Zhang 2006). Higher education has also undergone expansion and from 1999 to 2005 and the expansion was dramatic due to a designed “university expansion policy” (Lin and Zhang 2006; Li 2010). The following figure displays the rate trends of student transitions from JHS to RSHS and from RSHS to university between 1991 and 20112, which are the transitions and the

1 The year when the youngest cohort of the sample in the analysis of Chapter Four attended SHSAE. 2 The corresponding birth year of those students can be found in Table 4.1 of Chapter Four. It should be noted that the well-known “one-child policy” started in 1979 and ended in 2013, and thus my quantitative and qualitative research may consider the effects of siblingship size but not the effects resulting from the changes to the one-child policy. 14 period that my research is focused on (Figure 1.1). The rate of student transitions from JHS to RSHS generally increased between 1994 and 2001, with the increase accelerated between 1999 and 2002 (more than 10%). Those entrants to RSHS between 1999 and 2002 usually went through the transition from RSHS to university between 2002 and 2005, and thus to a large extent offset the effects of university expansion policy. However, the university expansion policy between 1999 and 2001 had already largely increased RSHS graduate transition opportunity to university from less than 50% to more than 70%. The previous studies about the effects of university expansion policy on social reproduction of educational inequality will be reviewed in the next chapter.

100% 90% 80% 70% 60% 50% 40% 30% 20% 10% 0%

JHS to RSHS RSHS to University

Figure 1.1 Rate of student transitions from JHS to RSHS and from RSHS to university between 1991 and 2011. Data from the National Bureau of Statistics (NBSC 2016d, 2016c).

IV. Introduction to Quality-oriented Education

Definition of Quality-oriented Education The definition of the Quality-oriented Education was clearly articulated in the government document titled Some suggestions on current actively moving forward the implementation of the Quality-oriented Education in primary and secondary schools (关于当前积极推进中小学实施素质教育的若干意见), which was released by the State

15 Education Commission (another name of the Ministry of Educational in 1985-1998) in 1997(State Education Commission 1997) and was widely cited. The document said, The aim of QOE is to improve the nationals’ quality. QOE is the education which is based on the regulations of the Law of Education with a view of the long-term development of the educatees and the society, with the basic purposes of serving all the students and improving their all-around qualities and with the characters of being focused on cultivating educatees’ attitude and ability and promoting them to develop in the aspects of morality, intellectuality and physique lively, actively and initially. QOE is trying to make students learn to be, learn to know, learn to work, learn to live, learn to exercise and learn to aesthetically appreciate and to build the foundation for them of being a socialist citizen with ideality, morality, knowledge and discipline.

Difference between EOE and QOE Jin and Tang (2004: 65) summarized the differences between EOE and QOE in the seven following aspects: (1) The service recipient: EOE was focused on a few “clever” students but ignored the majority while QOE is designed to serve all the students. (2) The teaching aim: EOE aimed to impart knowledge but ignored the education of morality, physique, aesthetics, mental health and vocational skills; in contrast, QOE is designed for students’ all-around development. (3) The cultivation of ability: EOE was only focused on the training of skills but ignored the cultivation of ability while QOE emphasizes the cultivation of all kinds of abilities. (4) The teaching method: EOE required students to learn by rote and repetition with heavily academic burden while QOE inspires students to learn by exploring, makes student learn lively, actively and initially and alleviates students’ academic burden. (5) The evaluation: EOE used examinations to select “good” students and viewed the examination scores as the main or even sole selection criterion while QOE uses the evaluation for students’ development and uses diverse evaluation approaches and diverse agents to carry out evaluating.

16 (6) The teaching content: EOE had a difficult content and stuck to the divide of different disciplines, overlooked the interdisciplinary and applied content, divorced its content from life to some degree, and ignored practice. In contrast, QOE lowers the depth and difficulty of its content, weakens the divide of different disciplines, emphasizes the interdisciplinary knowledge, relates its content to students’ life, and puts theory into practice. (7) The period of learning: EOE restricted learning into the school age while QOE emphasizes the lifetime development and lifetime teaching and learning.

Timeline of implementing QOE policy The reason why 1993 is viewed as the year of launching QOE is that in 1993 it was stated that “the primary and secondary school shall change from the ‘examination-oriented education’ into integrally improving the nationals’ quality” in a joint national platform on education of the central committee of the ruling party Communist Party of China and the State Council (CCCPC and State Council 1994). To implement that platform, in 1993 and 1994 the Ministry of Education issued the directive documents for alleviating academic burden on primary school and high school students (State Education Commission 1994, 1993). This directive was directly aimed at the EOE and required local educational authorities and schools to allow students to have more time for non- academic courses and extracurricular activities to develop their all-round “qualities” that had been ignored in the EOE system. In the autumn of the same year when the platform was issued, the fourth-round curriculum reform started to implement the new curriculum plan across the county, which gave prominence to developing students’ all- round qualities (Xie, Ma and Zhang 2013).

Six years later, the Central Committee of the Communist Party of China and the State Council (1999) issued another joint national platform on education titled Decision on deepening the educational reform and moving fully forward the quality-oriented education ( 中共中央国务院关于深化教育改革全面推进素质教育的决定 , hereinafter Decision). Following the Decision was another directive document for alleviating academic burden on primary school and high school students issued by the

17 Ministry of Education in 2000 (Ministry of Education 2000) and the fifth-round or “new curriculum reform” that started in 2001 emphasised the development of students’ creative spirit and ability and was more comprehensive and deeper than the previous rounds of curriculum reform (Xie, Ma and Zhang 2013).

Besides the academic burden alleviation and curriculum reform (teaching and learning content and methods), the admission approach reform is also the key item in QOE policy. However, compared to the other two reform items, reform of the old examination- oriented admission system progressed very slowly (Shi 2016; Yan 2011; Zhu 2009; Sargent et al. 2011). The most significant change was abolishing the JHS admission examination; however, this was conducted in the 1980s (Sun, Kang and Zhu 2009), which was before the launch of the QOE policy. The latest progress is marked by one overarching policy document titled Suggestions on implementing the further reform of the institutions of examination and admission (关于深化考试招生制度改革的实施意见 ), which was passed by the Political Bureau of the Central Committee of the Communist Party of China on 29 August 2014 and was released to the public by the State Council on 4 September 2014 (State Council 2014b). However, it was only implemented in some pilot areas in 2017 for the first time, and thus was not experienced by the cohort of my research. Therefore, my research uses 1993 and 2000, which are the years that launched a new national platform relevant to QOE, as the cut-off years for dividing the three different periods, i.e. before, during and after the implementation of QOE and compares the changes in social reproduction of educational inequality across these three periods. I further explain why my research choose these two years as the cut-off years in the introduction section of Chapter Four. My research also chooses 1976 as the first birth year of the cohort in the first QOE policy period. 1976 is the year when the Cultural Revolution ended and the literature review (Chapter Two) will explain why my research only focuses on the period after 1976.

18 Chapter Two. Literature Review

This Chapter firstly describes the general conceptual framework of the research area of social reproduction of educational inequality and where my research locates in this framework and then reviews the research on trends in social reproduction of educational equality in the contemporary urban China, the association between the Quality-oriented Education (QOE) policy and educational inequality, and the social justice theories and justice evaluation which will be employed for the discussion on parents’ evaluation of QOE policy in terms of educational “su zhi” (justice or fairness, 公平).

I. Social Reproduction of Educational Inequality

Studies of social reproduction of educational inequality or how characteristics of the family of origin are associated with educational outcomes constitute a broad field and have a long tradition in sociology and relevant disciplines of educational and policy research, and thus they have been comprehensively summarised by previous Annual Reviews articles (Collins 2009; Raudenbush and Eschmann 2015; Breen and Jonsson 2005; Werfhorst and Mijs 2010). Even though I do not intend to repeat their work in order to focus on studies that illustrate the theoretical and empirical contributions of research in China, in this section, I do briefly describe the key research topics in this area that have served as a foundation or stimulus for my research and how my research fits into this area.

Buchmann and Hannum (2001) provided a figure displaying the conceptual framework for this field of study (see Figure 2.1) and even though the studies they reviewed were in developing countries, this framework covers the key research topics on education and stratification. According to Buchmann and Hannum’s (2001: 78) figure, educational inequality was “shaped by a wide range of factors on multiple levels” and was “a consequence of dynamic interrelationships between family decisions about education

19 (commonly referred to as demand)”, which involved a range of family background determinants such as socioeconomic status (SES), structure and resources, and “the provision of educational opportunities (commonly referred to as supply)”, which involved school and community factors. The school factors should include the teaching and learning language and expressive culture, which usually benefited students from the middle-class and this research area has a long tradition in the literature on the education of students in developed countries (see Collins (2009) for a review). However, according to Buchmann and Hannum’s (2001) review, the effects of teaching and learning language and expressive culture was not the focus of research on the social reproduction of educational inequality in developing countries. Since Buchmann and Hannum (2001) reviewed the research on education and stratification in developing countries, a prominent part of their framework was how the macro-structural elements of non- developed or non-western countries, “including state policies and global forces, shape educational stratification through their effects on the demand for education or the structure and supply of schooling” (p. 78). Buchmann and Hannum’s (2001) research further reviewed studies on the effects of education on economic outcomes and social mobility in developing countries, where studies followed “the tradition of status attainment research in the United States” (p. 79).

Figure 2.1 has been removed due to copyright restrictions. Figure 2.1 Research on education and stratification in developing countries. Reprinted from “Education and stratification in developing countries: A review of theories and research,” by C. Buchmann and E. Hannum, 2001, Annual Review of Sociology, 27(1), p.79. Copyright 2001 by Annual Reviews.

The later four reviews on the inequality of educational outcome have different focuses but can fit into the framework. Breen and Jonsson’s (2005) review employed almost the same framework as Buchmann and Hannum (2001) did; however, it was more focused on developed countries. The other three reviews revolved around the school factors (including the classroom). As an anthropologist, which is different to other authors, Collins (2009) reviewed the research on how the conduct and organisation of classrooms and curricula in schools played a causal role in perpetuating not only cultural and

20 economic inequalities, but also linguistic inequality in different social classes, which reflected the traditional attentions of sub-disciplines of Anthropology and the interplay between family and school factors. Werfhorst and Mijs (2010) reviewed the research on the impact of national-level educational institutions on inequality in student achievement and concluded that compared to countries with a more standardised educational system, countries with a more strongly differentiated educational system (between school tracking) tend to have higher levels of inequality of educational opportunity by social class (and race/ethnicity), which reflected the interplay between macro-structural forces and school factors. Raudenbush and Eschmann (2015) focused on reviewing studies on the role of the school itself in inequality of educational outcomes by social class and concluded that free, universal schooling promotes social equality through equalising access to school and making disadvantaged children benefit more from access. In addition, the authors found that this equalising effect would be more pronounced for younger children than for older children.

My research also fits into Buchmann and Hannum’s (2001) framework. According to Werfhorst and Mijs’s (2010) typology, research on inequality of educational outcome can be classified into two types, namely inequality in terms of dispersion of educational outcomes (student test scores or people’s years of education) and educational inequality of opportunity in terms of the influence of social class (and race/ethnicity) on students’ test scores or (direct) admission opportunity. My research focuses on the latter. Based on Buchmann and Hannum’s (2001) framework, my research mainly involves the QOE policy as one of the macro-structural forces, the family and school factors and the explained variable, namely the inequality of educational opportunity by social class. However, different to Buchmann and Hannum’s (2001) framework, my research is only focused on the inequality of educational outcome and will not further use it as an explanatory factor to predict social mobility, which association thus will not be reviewed in this chapter.

II. Trends in Social Reproduction of Educational Inequality in Contemporary China

21 To be able to review trends in social reproduction of educational inequality in contemporary China, I systematically searched both Chinese and English journal databases, namely Chinese Social Science Citation Index and Social Sciences Citation Index and focused on quantitative research with a nationwide sample and a cohort in a long period due to my research aim. I identified 18 relevant articles published in the past 15 years. These articles explored various dimensions of educational inequality including gender (e.g., Yang, Huang and Liu 2014; Wu and Zhang 2010; Wu 2012), ethnicity (e.g., Liu 2008; Li 2010), urban-rural (e.g., Wu and Zhang 2010; Wu 2011; Tam and Jiang 2015), and cities with different sizes in eastern and non-eastern China (e.g., Liu 2008). However, since my research is focused on the change in educational inequality by family background in contemporary China, I focused the most attention on the 12 relevant articles that wholly or partially researched such a change (Liu 2008; Li 2003; Li 2006; Liu 2006; Wu 2010; Li 2010; Golley and Kong 2013; Wu 2014; Li 2014a; Chen et al. 2015; Wu 2008; Lu and Treiman 2008; Wu 2013). In this section, I first report the findings of reviewing previous studies regarding what indicators of family background researchers were concerned with and why. Then, I report what the trends in social reproduction of educational inequality in contemporary China previous studies suggest. The third theme is how to explain the above changes. Finally, I discuss the method issued in the previous studies to explain why the trends displayed by previous studies were not consistent.

Indicators of family background and social reproduction of educational inequality In Buchmann and Hannum’s (2001) conceptual framework, the family background factors included SES (usually including education, occupation and income), family structure and size, and family resources. The 12 recent studies on social reproduction of educational inequality of contemporary China did not exceed that framework in terms of the family background factors. All these studies used one (usually education) or more family SES domains to indicate a student’s family background to predict a student’s educational achievement. The retrospective data collection used in the majority of studies made reporting past family income almost impossible since family income varied frequently and was incomparable between different birth cohorts’ due to inflation. Therefore, the majority of studies did not use family income as one of the indicators of family background. One exception was a study that treated the father's pre-retirement

22 income as a control variable (Chen et al. 2015). Parental education (usually father’s education) was the most common indicator of family background and, when explaining the mechanism behind how parental education influences children’s education, those studies usually referred to the classical cultural capital theory (Bourdieu and Passeron 1977), especially the embodied capital (e.g., person's habitus) and objectified capital (e.g., family book collection) to explain the differentiation of children’s educational achievement (e.g., Wu 2013; Wu 2008). Parental occupation (again, usually father’s occupation) was the other common indicator of family background used by previous research and was measured in Ganzeboom and Treiman’s (1996) International Socioeconomic Index (ISEI) (e.g., Wu 2013) or classified into a few categories (e.g., Li 2003), which depended on the original measurement of the survey data that the studies utilised. To explain the mechanism of how parental occupation influenced children’s education, Wu (2013) adopted the “relative risk aversion” mechanism of rational action theory in educational decision-making (Breen and Goldthorpe 1997). Wu (2013) claimed that the father’s occupation indicated the class status of the family and families in the higher class had stronger aspirations for their children continuing in education to avoid the downward social mobility while families in the lower class, especially in the lowest class had nothing to lose but the upward opportunity if they left education.

Besides the SES of the family, the structure and size and resources of the family have also been used to indicate family background. The common approach is to treat the siblingship size as an indicator of the joint effects of family size and family resources based on the resource dilution hypothesis that on average, the more children the lower the quality of each child because each children’s share of family resources decreased as the sibling number increased and the quality could be indicated by educational achievement (Blake 1981). However, in some other studies on education and stratification in developing counties, siblings might actually contribute to household resources and were positively associated with children’s schooling (see Buchmann and Hannum (2001) for a review).

Besides the above factors, other family background indicators used by previous studies include whether a parent had membership to the Communist Party of China (Wu 2008),

23 child’s ethnicity (Wu 2013), family’s “hu kou” status (household registration, rural or urban “hu kou”) (Li 2003) and the family structure (nuclear, single-parent or extended family) (Liu 2008). However, my research is focused on the effects of QOE on students’ educational inequality by family background in urban areas rather than comparing the effects of different family background indicators on students’ educational achievements, and thus only uses parent education and occupation and child’s siblingship size to indicate family background.

The trend in social reproduction of educational inequality after the Cultural Revolution The cohorts in the above 12 articles have different ranges of birth year and the oldest cohort were born in 1930 (Chen et al. 2015) while the youngest cohort was born in 1996 (Wu 2013). Between 1930 and 1996 there were at least four periods that had distinctive micro-structural contexts namely the period of the Republic of China (before 1949), Mao’s Communist China before the Cultural Revolution (1949–1966), the Cultural Revolution (1966–1976), and the Reform and Opening-up (after 1976, or more precisely since 1978). Four articles treated the period after the Cultural Revolution as a whole era (Chen et al. 2015; Lu and Treiman 2008; Wu 2008; Li 2003) while the other studies had a more detailed description of and research on the variation in social reproduction of educational inequality after the Cultural Revolution. This distinction between the two categories above is that the former usually contain macro-political processes such as the Cultural Revolution and the Reform and Opening-up in their explanation or discussion while the latter are usually focused on specific policies or mechanisms such as tuition increase and educational expansion. My research is focused on the period of Reform and Opening-up, and thus the following section mainly reviews the trend in social reproduction of educational inequality after the Cultural Revolution and gives more attention to the trend in the urban sample if there is a sole or separate description on it.

Recent studies on changes in educational inequality by family background across periods in contemporary China have discovered that compared to the Cultural Revolution period, the educational inequality between different levels of family background grew during the Reform and Opening-up period (Li 2003; Chen et al. 2015). In terms of specific

24 indicators of family background, Wu (2008) examined the association between cultural capital and children’s educational transition opportunities separately into junior high school (JHS), into senior high school and into university in urban China across three periods, namely the period from the communist regime establishment in 1949 to 1965 before the beginning of the Cultural Revolution, the period of the Cultural Revolution and the period after the Cultural Revolution (1977–1996). Wu (2008) found that the effects of cultural capital on entering senior high school and university were not signiﬁcant during the Cultural Revolution but did become significant in the Reform and Opening-up era after the Cultural Revolution. This was consistent with Lu and Treiman’s (2008) research on the effects of sibling size (indicating the “resource-dilution" hypothesis) on urban children’s number of years of schooling. Lu and Treiman (2008) found a U-shaped pattern of (negative) effects of siblingship size across the above three periods, which meant that after the Cultural Revolution families with fewer children regained advantages owing to the resurging competition in investing resources in education. These findings are consistent with the conclusion from Buchmann and Hannum’s (2001) review of several earlier studies (e.g., Deng and Treiman 1997; Zhou, Moen and Tuma 1998).

The above studies did not contain more detailed description or analysis of trends in social reproduction of educational inequality after the Cultural Revolution while other studies covered a longer period of cohort birth years and divided the post-Cultural Revolution era into a few sub-periods. It should be noted that the cohorts in these studies on social reproduction of educational inequality after the Cultural Revolution are not exactly the same (Wu 2012; Liu 2008; Wu 2014; Li 2010; Li 2006; Golley and Kong 2013; Li 2014a; Tam and Jiang 2015). In addition, the findings were not consistent in terms of whether the trend in educational inequality by family background after the Cultural Revolution was increasing upward or with fluctuations. In terms of effects of parental education, Wu (2012) found that the educational inequality in the transition from junior to senior high school1 caused by the differentiation of father’s education

1 The primary school and JHS constitute compulsory education, and thus the transition between them are different from the latter two transitions. Therefore, in this section I will not review the research on the transition from primary school to JHS.

25 consistently increased in 1989–1998 (educational transition year, cohort born around 1974–1983) and then 1999–2008 (cohort born around 1984–1993) compared to 1978– 1988 (cohort born around 1963–1973), which was also supported by Liu’s (2008) research (cohort born around 1960–1982). However, in terms of the transition from senior high school to university, the effects of the father’s education on a child’s transition opportunity did not change significantly based on Wu’s (2012) research (cohort born around 1960–1990) and Li’s (2010) research (cohort born around 1975– 1985). Using a different method of the sequential logit model, Wu (2014) directly employed the continuous birth cohort year variable rather than a set of cohort periods to interact with the father’s school year number and found that the degree of educational inequality by father’s education continuously increased after the Cultural Revolution; however, its acceleration tends to be moderate. In addition, Wu’s (2014) research provided a more detailed description of the trend of intergenerational education transmission for the birth cohorts from 1949 to 1985 separately in the transitions from primary to JHS, from junior to senior high school and from senior high school to university. Applying Buis’ (2010) method of decomposition of an estimate of inequality of educational outcome into a weighted sum of probabilities passing three transitions, the research found that the increasing inequality of the transition from junior to senior high school over time was a more important contributor to the general increase in educational inequality by father’s education than the other two transitions. This supported Wu’s (2012) findings that the increase of educational inequality was significant in the transition from junior to senior high school but not the transition to university.

The trend in effects of sibling numbers on children’s educational opportunity after the Cultural Revolution, which indicates the family resource-dilution hypothesis, was similar to the trend in effects of fathers’ education reviewed above. Wu (2012) and Liu (2008) both found increasing negative effects of larger siblingship size on children’s transition opportunity from junior to senior high school while its effects on children’s transition

26 opportunity from senior high school to university did not increase significantly according to Wu’s (2012) research1.

However, the previous studies’ findings on the trend in effects of parental occupation on child’s educational opportunity after the Cultural Revolution were more complicated. Wu (2012) found the effects of fathers’ ISEI on both children’s transition opportunities from junior to senior high school and from senior high school to university did not change significantly. While Li’s (2010) research supported Wu’s (2012) findings in terms of children’s transition opportunities from senior high school to university, Liu (2008) found the inequality associated with the father’s occupation in the child’s transition opportunity from JHS to university was increasing.

It should be noted that the studies reviewed in the above section used a nationwide sample and the previous studies have demonstrated the increasing educational inequality between the rural and urban students after the Cultural Revolution (Tam and Jiang 2015; Wu 2011; Li 2014a; Pang 2016; Wu 2013). The research on trends in equality inequality by family background using urban samples did not have exactly the same findings as the research using the nationwide sample. Golley and Kong (2013) examined the association between children’s and parents’ number of years of schooling without controlling other covariates for 10 5-year birth cohorts of children from 1941 to 1990 and found there was a clear trend of rising persistently in social reproduction of educational inequality for the urban birth cohorts from 1976 to 1990, which supported the research using the nationwide sample. However, Li (2006) found that compared to the early reform period (1977–1991)2, during the late reform period (1992–2003) the inequality of an urban child’s transition opportunity based on their father’s education was decreasing in both transitions from junior to senior high school (cohort born around 1977 to 1988) and from senior high school to university (cohort born around 1974 to 1985). These findings are opposite to the findings from the studies that used nationwide samples. Furthermore, in Li’s (2006) findings, the trend in inequality of a child’s

1 Liu (2008) did not research the transition opportunity from senior high school to university. 2 Those are educational transition years rather than cohort birth years. In terms of the transition from junior to senior high school in the early reform period, the cohorts were born around 1962 to 1976 and for the transition from senior high school to university they were born around 1959 to 1973. 27 transition opportunity from junior to senior high school based on their father’s occupation decreased while the inequality increased in terms of a child’s transition opportunity from senior high school to university due to the sharply increasing advantage of the management group over the blue-collar group.

Explaining changes in social reproduction of educational inequality by periods The previous studies had different results in terms of the trend in social reproduction of educational inequality after the Cultural Revolution; however, they proposed a set of self-consistent theories to explain their own results and since they used different samples with different cohort birth years, it is difficult to directly compare two studies to find out which explanation is more reasonable. However, a variety of explanations proposed by these studies helped me obtain a more comprehensive understanding of the trend in social reproduction of educational inequality during this period of a variety of social and policy changes with these explanations classed into three categories: the drives increasing, decreasing, and suppressing the change of educational inequality by family background factors after the Cultural Revolution.

The result of the increase in educational inequality by family background was expected after the Cultural Revolution. The most important drive was the micro-structural force of the Reform and Opening-up and the nation sharply changed from “policies designed to promote equality” to “policies designed to promote economic development” (Lu and Treiman 2008; Li 2003). Therefore, contrary to the Cultural Revolution period when the Chinese government intentionally expanded the opportunities for children from lower- class families and higher-class families were under attack and less able to transmit educational advantage, after the Cultural Revolution a “merit-based” educational system (mainly the examination system rather than recommendation based on family origin1) was introduced to select the qualified labour force for economic development and the effects of family origins on social reproduction of educational inequality once again became more salient (Li 2003; Wu 2008). Besides abolishing the “radical

1 The children of workers, peasants and soldiers could have the opportunity of being recommended to enter university during the Cultural Revolution when the examination system was abolished. 28 egalitarian agenda” in the education policies of the Reform and Opening-up (Buchmann and Hannum 2001), the reform in the educational field was transforming the “centrally- planned system” into a “market-oriented system” and the consequence was that education was “industrialized” or “commodified” rather than keeping education as a kind of social welfare and the educational field as part of the Chinese society was moving toward a “market-oriented system of social stratification” (Li 2003; Wu 2008; Yang 2006a: 97-143). In terms of the specific mechanism as to how a market-oriented education system can drive the social reproduction of educational inequality, Wu (2013) adopted rational action theory in educational decision-making (Breen and Goldthorpe 1997) to explain such a mechanism: compared to the “social-welfare” educational system, the educational cost increased in the market-oriented educational system while the expected return decreased due to the educational expansion (especially the university expansion) of causing graduate redundancy and low salary and the prosperity of manufacturing causing an increase in blue-collar job number and salary. Thus, underprivileged families found it easier to make a “rational” decision of making their children drop out of education. Considering that income inequality has been sharply increasing in mainland China after the Cultural Revolution (Wu and Perloff 2005; Xie 2010), the differentiation in the above educational choice between privileged and underprivileged families could become more and more obvious.

However, not all studies found that the social reproduction of educational inequality increased. Some studies found there was a decrease, for example, in social reproduction of transition opportunity inequality from junior to senior high school in the urban area between the early and the later Reform and Opening-up periods (Li 2006) while other studies found there was no significant change, for example, in social reproduction of transition opportunity inequality from senior high school to university between the cohort born in 1975–1979 and the cohort born in 1980–1985 (Li 2010). It is interesting that both the increasing and unchanging trends were explained by the same set of theories: educational expansion and maximally maintained inequality (MMI) theory1

1 The social reproduction of educational inequality does not only occur in the educational transition between different educational stages, but also occurs in the admission opportunity into a higher-quality institution in the same stage (e.g., elite vs ordinary university). To the latter kind of social reproduction of educational inequality, effectively maintained inequality theory as a modification of MMI theory was 29 (Raftery and Hout 1993). In the previous chapter, I have described the educational expansion in China experienced by students who usually were born or entered primary school after the Cultural Revolution. However, educational expansion and distribution of educational opportunities are two separate processes (Mare 1980; Wu 2010). Previous studies found that the educational expansion in China did not necessarily or instantly result in a weaker association between social origin and education (Li 2010; Wu 2013; Li 2006). These findings were similar to foreign authors’ studies on the association between educational expansion and social reproduction of educational inequality in other countries (Raftery and Hout 1993; Shavit and Blossfeld 1993), and thus employed the MMI theory that their colleagues utilised. According to MMI theory, “if the demand for a given level of education is saturated for the upper classes, that is, if some origin- specific transition rates approach or reach 100 percent, then the association between social origin and education is weakened” (Raftery and Hout 1993). Therefore, Li (2006) employed MMI theory to explain the association between family background and children’s transition opportunity to senior high school was weakened in the late reform period because the educational enrolment rate in senior high school had possibly increased to the threshold of meeting the demand from the upper-class families. Li (2006) and Li (2010) further employed the MMI theory to explain that the social reproduction of inequality in higher education admission opportunity was unchanged because the demand of upper-class families for higher education had not achieved the threshold of relative saturation. Even though it seems impossible to set up a generalised statistical model predicting the threshold of saturation owing to the different admission systems between different countries, Li’s (2017b) research on the inequality in higher education admission opportunity between students from urban and rural areas supported the existence of the threshold in the effects of Chinese educational expansion on social reproduction of educational inequality between the rural and urban students and designed a statistical model to calculate the threshold based on the normal distribution and different means of the university admission examination scores between rural and urban students. According to Li (2017b) threshold dependent

proposed by (Lucas 2001). However, in this thesis I am only focused on the former kind of social reproduction of educational inequality, and thus will not discuss the effectively maintained inequality theory further. 30 inequality (TDI) theory, which was proposed to modify the application of Raftery and Hout (1993) MMI theory to China, the educational expansion was supposed to make the inequality of educational opportunity by family background (the urban or rural “hu kou” in this research) first increase and then decrease in China if lowering the admission score resulting from the expansion started with a high score and the scores of both lower and upper-class students were normally distributed and the latter had a larger mean. If Li’s (2017b) threshold dependent inequality theory is correct, the increase in social reproduction of educational inequality mentioned above can have another reasonable explanation as well.

After Li’s (2017b) modification, MMI theory appears to be able to explain the increasing, decreasing and unchanged social reproduction of educational equality and why the trends of social reproduction of educational equality are different between different educational stages. However, in terms of the same educational stage, this theory cannot explain why the trend of educational inequality in university enrolment opportunity based on family cultural capital (father’s education) was opposite to the one based on family resources (father’s occupation) that was found by Li (2006). In addition, the threshold in Li’s (2017b) threshold dependent inequality theory is based on the normal distribution and different means of examination scores between students with different family backgrounds. if some new policies could change the distribution and mean gap of those examination scores, new theories explaining the change of social reproduction of educational equality could be proposed based on those new policies without abandoning Li’s (2017b) threshold dependent inequality theory.

Implications for research questions and methodology of the thesis From the above review on studies of the trend in social reproduction of educational inequality, researchers came to similar conclusions that educational inequality was increasing during the Reform and Opening-up period compared to the Cultural Revolution period. However, studies of the trend across different cohorts and in terms of different educational transitions during the Reform and Opening-up period had different or even opposite findings. The reasons for these discrepancies include different criteria for dividing cohort groups (e.g., Li (2006) vs. Wu (2012)), single (e.g., Li 2006) or

31 multiple family socioeconomic factors (e.g., Li 2010), different combinations of multiple family socioeconomic factors (e.g., Li (2010) vs. (Wu 2012)), the general sample (e.g., Wu 2012) or urban sample (e.g., Li 2006) and so on. Some studies also used different indicators for educational outcome, for example different to other studies that only researched the general higher education admission opportunity, Yue (2015) divided universities into elite and ordinary universities and Liu (2006) further divided them into three categories. Even though previous studies did not provide a consistent conclusion about the trend in social preproduction of educational inequality in the current Reform and Opening-up period, their theories or discussions are still helpful for me to form my research questions and will assist with my own discussion. In addition, I also utilised and combined some methods from their research, which will be described in the methodology section of the results chapters.

I briefly discuss the methodological implications for my research. Even though a few earlier studies used the model of multivariable linear regression of a child’s years of education to estimate and compare the effects of family background on educational inequality between different periods (e.g., Li 2003), the later studies in China were influenced by Mare’s (1981) research and utilised the model of logistic regression of successful transition probability to avoid the structural bias resulting from the increase of average years of education of the whole population (e.g., Li 2006). In addition, since the estimated coefficients are not comparable between different logistic models using different cohorts for the coefficients and the error variances are not separately identified (Breen, Karlson and Holm 2010), the recent studies simply added an interaction between period variables and family background indicators into one logistic model (e.g., Wu 2013; Li 2014a). In terms of the sample, the previous studies have made a consistent conclusion that the educational inequality between the rural and urban residents increased after the Cultural Revolution (Tam and Jiang 2015; Wu 2011; Li 2014a; Pang 2016; Wu 2013). Even though the identity of rural or urban “hu kou” (户口, household registration) was viewed as one of the indicators for family background (political status in Mao’s era and occupational status in the Reform and Opening-up era) by researchers, it is also an indicator for population geographical factors in distributing

32 educational resources, for example, the policy of cancelling small teaching centres and merging small schools in rural areas due to low population density (Li 2014a). The population geographical factors associated with the rural “hu kou” also includes the phenomena of peasant labour migration (Zhu 2002; Solinger 1999) and the “left-behind children” (Tan 2011), which I mentioned in Chapter One (pp. 6-7). Due to the phenomena of peasant labour migration, Golley and Kong’s (2013) research divided the sample into three categories namely urban, rural and migrant children, which makes the research more complicated. Therefore, I only use the urban example to focus my research on the relatively “pure” family background indicators associated with educational inequality and the findings of my studies cannot be generalised to rural or migrant populations.

In terms of implications of the previous studies reviewed in this section for forming my research questions, the research questions of those studies were focused on the effects of “micro-structural forces” (the term used in Buchmann and Hannum’s (2001) conceptual framework) including the institutional change of the Reform and Opening- up and/or some specific policy changes such as educational expansion on the association between “family factors” and child’s “educational outcome” (i.e., social reproduction of educational inequality). However, except the reintroduction of a “merit-based” educational admission system, the “micro-structural forces” researched in these studies had an effect on social reproduction of educational inequality only through family factors (the mechanism of rational action theory in family’s educational decision-making) or directly on educational outcomes (educational expansion)1. However, in Buchmann and Hannum’s (2001) conceptual framework, school factors also have important effects on social reproduction of educational inequality and when a new “micro-structural force” such as the QOE policy in my research is introduced, not only the admission system but also the learning and teaching process in school (such as content and methods) could be changed. The change of school factor could not only directly affect students’ educational outcome but also influence families’ reaction to those changes, for example, how the families arrange their children’s out-of-school activities due to those changes in school,

1 Actually, Buchmann and Hannum’s (2001) conceptual framework did not clearly illustrate the direct effect of micro-structural forces on educational outcome. 33 which could affect children’s educational outcome. Since the new family arrangement could be associated with family background, the new “micro-structural force” could finally have an effect on the social reproduction of educational inequality.

III. Quality-oriented Education and Educational Inequality

“Su zhi” (quality), “su zhi jiao yu” (QOE) and neoliberalism Some researchers have already suggested the neoliberal policies were appropriated by Chinese government to reform and restructure its education system (Mok and Lo 2007; Mok 2006). In terms of neoliberalism as applied in the government’s educational policy, Davies and Bansel (2007: 248) suggested that the neoliberal state is “giving power to global corporations” and installing “apparatuses and knowledge through which people are reconfigured as productive economic entrepreneurs of their own lives”, which is different from the previous administrative state who is “responsible for human wellbeing, as well as for the economy”(p.248), and thus “an appropriate(d) neoliberal subject” or educatee is supposed to “float free of the social and takes up responsibility for its own survival in a competitive world, where only the fittest survive” (p.258). My research also uses the above definition of neoliberalism.

It is not supervising that the discussion on the relationship between QOE (“su zhi jiao yu”, 素质教育) and educational inequality was involved in a general topic about the association between neoliberalism and human “su zhi” (素质, quality) discourse (e.g., Anagnost 2004; Yan 2003; Kipnis 2007). Through the lens of socio-cultural anthropology, “su zhi” was employed in the processes of governing contemporary China in many ways and its discourse informed policy-making decisions (Kipnis 2007) including Chinese educational reforms (Kipnis 2006). Through the circulation of “su zhi” discourse in reform-era China, a form of common sense was achieved that “su zhi” was not something that naturally inhered in the body but was a supplement that had to be built into the body as human capital measuring the body’s worth (Anagnost 2004). Therefore, building “su zhi” into the child was a neoliberal subject production separating the middle-class family from the rural ones (Anagnost 2004). Kipnis (2007) summarised the use of neoliberalism in the analysis of “su zhi” discourse as using the “blame the

34 victim (by denying structural factors due to belief in a just world 1)”-type neoliberal discourse, drawing equivalence between “su zhi” and money and equating non-market forms of measuring “su zhi” with the market (see Yan’s (2003) discussion of psychological tests applied in labour recruitment). Kipnis (2007) agreed that occasionally some concepts of “su zhi” were appropriately labelled neoliberal in some points of the above analyses but he argued against framing “su zhi” discourse as a form of neoliberalism, which could limit the operation of “su zhi” discourse. This is because, besides its ability to fit neoliberal discursive contexts, “su zhi” discourse, more importantly, reinforced the hierarchical and authoritarian structure in China. Because of the discipline characteristic of socio-cultural anthropology, I find the above discussion is focused on abstract concepts, and thus does not attempt to discuss the specific mechanism or policy measures of QOE that may increase (or decrease) the inequality of educational opportunity between different socioeconomic group. More importantly, these studies pay attention to the prominence of the whole post-Mao or economic reform era after the cultural revolution (1966–1976) while the QOE policy only started in the early 1990s. While it is incisive to employ the abstract neoliberal “su zhi” discourse in the economic reform era to explain parents’ investment in their children’s education, other studies were inclined to attribute parents’ behaviours to the influence of traditional culture deriving from the imperial China period (Cai 2012; Luo 2009). To avoid getting entangled with such a controversy and as a policy research thesis, my study is focused on the less abstract level of discussion.

“Marketization” and “privatization” are two concepts that often appeared in the neoliberalism discussion (e.g., Kipnis 2007; Mok, Wong and Zhang 2009) and were employed to explain the intensiﬁed educational inequalities between different regions with different levels of economic development by the previous studies. Relevant educational policies after the 1990s included the promotion of school choice, growing individual contributions to tuition (especially for higher education), the rise of private

1 Lerner and Miller (1978: 1030) illustrated the belief in a just world (BJW) theory that “individuals have a need to believe that they live in a world where people generally get what they deserve”. When confronting a fact conflicting with this BJW view, individuals will “try to ignore, reinterpret, distort, or forget it—for instance, by finding imaginary merits to the recipients of fortuitous rewards, or assigning blame to innocent victims” (Bénabou and Tirole 2006: 700).

35 for-profit providers with curricular and extracurricular education, and “transformed” schools1 (Mok, Wong and Zhang 2009: 97-144; Crabb 2010; Yang 2006a). However, even though some studies did mention QOE in their discussion on “marketization” and “privatization” (e.g., Crabb 2010; Anagnost 2004), they did not conclude whether it was the QOE policy that resulted in the “marketization” and “privatization” of Chinese education, which then increased educational inequality. Therefore, the following two questions have been explored in the following sub-section of literature review: (1) Was QOE policy designed to improve educational equality or intensify educational inequality? (2) Regardless of the policy intention, what is the relationship between QOE policy and educational inequality in practice?

The intention of QOE policy: whether it was designed to improve equality? QOE policy was originally designed to reduce the excesses of Examination-oriented Education (EOE). In addition, the purpose of QOE was to improve the workforce’s “su zhi” (quality), namely practical skills and innovative ability that were seen as critical to raising China’s global competitiveness and realising its modernisation by the ruling Communist Party of China and the central government (Dello-Iacovo 2009; Yu 2008). Even though improving educational equality was not a major intention of QOE initially, as an argument against EOE focusing teaching resources on only a few students with high examination scores while ignoring the majority of students (Yu 2008), the fact that primary and high schools should serve all students (mian xiang quan ti xue sheng, 面向全体学生) was written into the first overarching QOE policy document The Outline of Education Reform and Development of China (中国教育改革和发展纲要) (Yu 2008; CCCPC and State Council 1994).

The crucial component of QOE namely “all-around development” is a potential factor in intensifying educational inequality, especially when the “all-around development” is conducted in the family environment after school. When a stronger version of academic burden alleviation policy occurred at the beginning of 2000 (Ministry of Education 2000) and in principle opened possibilities for students to have more extracurricular time to

1 Those schools were “transformed” from a public school to a public-owned but privately managed school.

36 participate in more “creative” activities, China Education Daily, which is a governmental newspaper directly under the Ministry of Education, immediately printed a series of articles and editorials giving suggestions to parents and students on how to spend the upcoming long winter school break (as cited in Woronov 2008). Through analysing those articles and editorials, Woronov (2008) found the implicit assumption was that “parents—even those working full-time—can and should find the time to organize and supervise children’s vacation days to the minute” (p. 414) and more importantly parents should organise high-quality vacation or after-school activities that could be unaffordable for some working-class families. However, in the two overarching QOE policy documents issued in 1993 and 1999 (CCCPC and State Council 1999, 1994), the formal schools were required to take the major responsibility of facilitating students’ “all-around development” including developing the resource-consuming art and sport qualities while parents should assist schools to improve their children’s “moral quality”. Even though in the 1990s the educational marketisation was a government policy practice (Yang 2006a: 99-106), after the mid-2000s the Communist Party of China and Chinese education authorities paid more attention to the educational equality and the equality of QOE in the compulsory education stage (Tao and Yuan 2010; Yu 2008). In brief, there is no research concluding whether or not in the 1990s the QOE policy intention was to shift more educational responsibility from the government to the family.

QOE policy implementation and its effects on educational equality Despite two decades of implementing QOE reform, the EOE model still predominates as the basic education model in China (Zhu 2009; Wu 2016; Liu and Dunne 2009; Shouse and Ma 2015; Kipnis 2011; Liao and Yuan 2017; Chen 2014; Gu and Zhang 2010). Education remained focused on the academic subjects (so-called “intelligence quality”) that led to senior high school and university admission examinations while neglecting the development of moral, physical and aesthetical qualities (Gu and Zhang 2010; Wang 2014). Even though the current curriculum reform was based on the long-period “laying the major foundation” work and experience from the pilot areas and then gradually implementing these nationally (Law 2014) and there were some “shining” and “successful” examples (Dello-Iacovo 2009; Ning and Lv 2012; Lockette 2012), most studies indicated that schools encountered significant problems implementing the new

37 teaching and learning methodology at the core of the curriculum reform and there was a disconnection between policies and practices in reality (Dello-Iacovo 2009; Marton 2006). Many school teachers still adopted the teacher-centred teaching approach (e.g., whole-class lecturing and intensive drillings) rather than student-centred teaching approaches such as small group work, discovery methods and project-based inquiries even if they had been trained in the new methods (Third Research Group 2006; Wang 2011a; Wang 2014). They just employed the new teaching and learning methods as “meticulously staged performances” when inspected by education officials (Wu 2016). The third key reform of QOE policy, namely the admission approach, was always on the agenda of the educational policy makers and has resulted in several adjustments (Yan 2011). However, every time a new adjustment was proposed, it led to a nationwide controversy resulting in the examination-oriented admission system (widely perceived as “just and fair” due to its procedure justice) barely changing (Yan 2011; Shi 2016; Sargent et al. 2011; Zhu 2009). The “bottleneck” of the QOE reform was the current admission system, especially the university admission system (Yan 2011) because the university admission system is the “baton” to direct the QOE reform and determine whether the QOE can really replace the old EOE (Wen and Liu 2014; Zheng 2002). Along with hindering the development of all-around qualities and student-centred teaching methodology, the “obstinate” examination-oriented admission system has also seriously weakened the academic burden alleviation policy (Qiao 2014; Law 2014). It was not only because the academic burden alleviation policy was not fully complied with by some schools or local educational authorities, but the more important factor was that the government policy could only regulate the schools (including private schools in China) but not the parents (Ruan 2011). As a result, the academic burden that school teachers place on students has been slightly alleviated, but parents put more pressure on their children through the commercial extracurricular tutoring (hereinafter CET) (Yan 2011; Ruan 2011; Hu 2014).

Previous research has indicated that QOE could increase educational inequality and empirical studies have focused on the urban-rural inequality such as the emphasis on the observations that “su zhi” was urban-orientated and the rural teachers lacked training in the new teaching methods (Lou 2011; Wang 2011a). The educational

38 inequality also increased between urban students from different socioeconomic backgrounds, especially when the “all-around development” was developed with the help of parents during after-school time because underprivileged urban families could not afford the fees for developing some “qualities” of their children, for example, the “strong point of art” (Wang 2011c: 104-105). However, as I reviewed above, if the admission system was still examination-oriented and focused on academic subjects, these quality-oriented inequalities caused by the QOE policy should not significantly increase the inequality of admission opportunities.

Besides CET in non-academic subjects, Wang (2011c: 101-104) found that underprivileged urban students could not afford CET in academic subjects through field work, which increased the inequality in academic performance between students from different socioeconomic backgrounds. This was supported by several quantitative studies that found that family background (usually including family income, parent’s education, socio-economic status and “hu kou”) was associated with the probability of academic attending extracurricular tutoring (Chen and Bai 2015) and a similar association was found in studies without differentiating between academic and non- academic extracurricular tutoring (Liu and Bray 2017; Li and Qiu 2016; Liu and Xie 2014). Besides family background factors, the family size or more specifically the number of children is also related to the probability of attending CET due to its association with the effects of family income and educational expectations for children (Liu 2012).

However, Wang’s (2011c) research and other above studies did not discuss whether the extracurricular tutoring in academic subjects had an association with the QOE policy or not, which I have identified as a research gap. Some studies noted the possible relationship between the QOE policy and the demand for CET as a small part of their discussion or finding (Xu 2009; Zhang and Bray 2015; Lou 2013). For example, Lou (2013) briefly mentioned the academic burden alleviation policy and the fact that the Mathematical Olympiad performance is one of the admission criteria to a “key” (elite) school or class that has resulted in the popularity of extracurricular tutoring. However, those observation was neither systematic nor based on empirical research and, more importantly, even though academic burden alleviation policy is one of the QOE sub-

39 policies, there is no thorough analysis on the relationship between the whole QOE policy and extracurricular tutoring.

Even though the effects of QOE have not been thoroughly researched, it should be noted that previous research proposed many other explanations for parents sending their children to attend extracurricular tutoring. Zhang and Bray (2015) conducted a systematic review of those explanations or factors and categorized them into four levels: the micro (student and family), the institutional (school and CET company), the community (urban or rural) and the system (cultural and social contexts) level, and suggested those explanations or factors could be related. Through analysing those the explanations for attending CET, I find there is another dimension that can be added into the above categorization: attending CET as a personal or family educational decision versus as a social phenomenon or fashion. As a personal or family educational decision, whether to attend CET is associated with family background factors. Thus, the decision to attend CET interacts with the research question whether CET can work as the mechanism of social reproduction of educational inequality (Research Question 2, see Chapter One, p. 8) and has been reviewed above. Those factors can also be categorized into the micro level in Zhang and Bray’s (2015) categorization. Other micro-level factors, which have not been reviewed by Zhang and Bray (2015), include the desire for upward social mobility (Hu 2014; Li 2015), intensive parenting style of emphasising a careful plan of structured activities for children (Davies 2004) and parents using consuming to avoid the responsibility of “true” parenting (Hu 2014). Those factors could be considered in further research about the reasons for families with different socio-economic backgrounds have different probabilities of sending their children to attend CET.

In addition, my research is concerned with attending CET as a social phenomenon or “fashion”, and the explanations at this level can also be classified into two categories: the long-term existence versus the short-term rise of CET. In terms of the long-term existence of CET, the explanations include the cultural and social contexts such as traditional Confucian cultural emphasis on education (Hu 2014; Bray 2006), the high- stakes examinations (Liu 2012; Kim and Lee 2010), and the large-scale and highly stratified market for credentials (the proportion of students crossing provincial borders

40 to attend university) (Davies 2004). In addition, at the policy level the equalisation policies such as employing standardised textbooks and the school district system without entrance examinations1 in the compulsory education stage resulted in parents’ strong need of differentiated educational services to help their children obtain or retain the advantage in the competitive admission into schools of post-compulsory education (Liu and Xie 2014; Bray and Kwok 2003; Park, Byun and Kim 2011). However, these factors had already existed for a long period before the popularity of extracurricular tutoring. When the CET has existed as a popular educational institution, there are other factors maintaining its popularity. At the institutional level, the popularity of CET is encouraged by schools using the reports of examinations organized by large tutoring enterprises to recruit high performers or directly entrusting tutoring enterprises to train and organize their students to attend academic competitions, and it is also encouraged by school teacher who can financially benefit from the CET mechanism through working as a part-time tutor in their non-working time or through kickbacks for referral of tutees (Zhang and Bray 2017a; Zhang and Bray 2017b; Zhang 2014). In addition, the aggressive advertising from CET companies maintained the popularity of CET as well (Zhan et al. 2013). At the personal level, Huo and Xu (2013) and Yu and Ding (2011b) employed the Game Theory or the prisoners’ dilemma to explain this popularity, i.e., parents worried that their children would get left behind if their children did not attend extracurricular tutoring but other families’ children did. A similar argument was the peer pressure on both parents and children due to other children attending extracurricular tutoring (Hu 2014; Bray 2013). However, those factors still cannot explain the initial rise rather than existence of CET as a popular social fashion. Therefore, my research is focused on seeking new social or policy changes, for example the new QOE policy, and determining whether there is an association between these changes and the popularity of extracurricular tutoring and further explore whether the association changes social reproduction of educational inequality. It should be noted again that my thesis does not argue that QOE policy is the exclusive factor explaining the rise of CET. Other macro- level factors include “neoliberalism” policies to decentralize control of schooling,

1 In the system, a school-age student usually enters the school nearest to his/her home without undertaking an admission examination.

41 replace the government role under the planned economy with educational credentials in allocating job opportunities, rapidly expand formal education, and empower private education raise performance, and empower private education (Zhang and Bray 2017b) and the dramatic economic growth (Zhang 2014: 436). Even though those factors were only briefly mentioned without elaboration and could have existed approximately a decade before the popularity of CET, they are still worth considering when the further research try to conduct a comprehensive review of factors in the rise of CET, which however is not the aim of my thesis.

As the possible mediator, CET should be not only associated with family background factors but also students’ academic achievement. Some studies neither differentiated the effects of art and sport tutoring and academic tutoring nor discussed the relationship between QOE policy and extracurricular tutoring, but they used quantitative methods to reveal the effects of extracurricular tutoring on students’ educational achievement. Pang (2016) compared the urban-rural gaps in admission opportunity before and after the emergence of the extracurricular tutoring market and claimed that urban students from non- “key” (elite) JHSs benefited more from extracurricular tutoring, which played a much more significant role in enlarging the urban-rural gap. Through structural equation modelling, Li and Qiu (2016) demonstrated that CET was a mediator between family background and students’ academic performance while its path coefficient to academic performance was only significant in the general national sample model but not in the urban or rural sample model. Therefore, the effects of CET on urban students’ academic performance needs further demonstrating after determining the role of CET in the association between QOE policy and social reproduction of educational inequality.

IV. QOE and Justice Evaluation of Educational Resource Distribution

The significant discrepancy between the QOE policy documents and its implementation has been investigated by educational policy researchers as reviewed in the above section. Besides the external factors, for example the residual influence of traditional imperial examination (“ke ju”, 科举) culture (Cai 2012), the QOE policy-making process

42 itself was also under criticism and blamed for that discrepancy. One critique was that politicians and civil servants predominated the educational policy-making without enough consultation with key stakeholders, including teachers, parents and students (Zhu 2009; Ruan 2010). The policy evaluation was insufficient, unprofessional and perfunctory (Yang 2006b; Liu 2002; Yang 2006c) and there is no previous research reporting any result of QOE policy evaluation involving the key stakeholder namely parents. QOE policy evaluation involving parents was crucial because parents’ private interest of making their children survive the examination-oriented admission was not consistent with the policy makers’ public interest of improving the population’s quality and this interest conflict was one of the most important reasons for the plight of QOE reform (Ruan 2010, 2008, 2011).

Even though CET is closely related to QOE policy, there is no previous research on parents’ evaluation of extracurricular tutoring, i.e., whether students should attend extracurricular tutoring or whether it should be the family’s responsibility for the tutorial fees. Ruan (2011) research reveals such a study on parents’ opinion was necessary through analysing the directive released by the Department of Education in Chengdu City prohibiting government-supervised educational institutions and their employees from organising, participating in and assisting with CET and prohibiting any school from recognising any tutorial institution’s grade report for admission. This directive incurred a massive opposition from parents and Ruan (2011) claimed there was an interest conflict between the educational authorities and parents in terms of extracurricular tutoring, and thus it was necessary to determine and analyse parents’ concern.

One intention of the tutorial prohibiting policy was to promote educational equality (Ruan 2011). Does parents’ opposition to such a policy mean they do not support educational equality? As I reviewed above, previous research reveals the “bottleneck” of QOE reform is the examination-oriented admission system but every attempt to reform the current admission system has incurred a massive concern about admission “gong ping” (公平, justice/fairness, and in other situations it may refer to equality or equity) because the current examination system, especially the “gao kao” (university admission examination) is viewed as one of the only fair institutions that Chinese trust

43 (Fong 2006; Sargent et al. 2011) and the “inflexible” public preference for admission “su zhi” overrode the critiques of the current admission system (Yan 2011).

Admission “gong ping” focuses on the justice of distributing educational resources, especially the enrolment positions of regular senior high schools (RSHSs) and universities in my research. However, there are few empirical studies relevant to Chinese people’s perception of educational “gong ping”. A similar research area is Chinese people’s perception of distributive justice on income and since income inequality has been sharply increasing in mainland China over the last decades 1 , a number of relevant studies have emerged including research on the perceived income justice (Li 2014b; Liu and Hu 2016; Meng 2012; Li and Wu 2012; Wang 2011b; Ma and Liu 2010; Whyte 2009; Sun 2009; Wu 2009). These studies are relevant to my research to show what extent Chinese people perceive the distributive inequality as injustice, what principle they employ to evaluate the distributive justice and what affects their evaluation. However, since these studies are not directly referring to the distributive justice of educational resources, I only review them briefly in the following section.

Even though in recent surveys (after 2000) respondents perceived the existence of income inequality, they were inclined to evaluate the current income distribution as just (Meng 2012; Whyte 2009; Wu 2009; Li 2014b). However, the urban and rural citizens had different judgements. Compared to the rural citizens, urban citizens were more likely to perceive income distribution as unfair (Whyte 2009), which was supported by Wang (2011b) research on income justice evaluation of the residents in Shanghai, China. It should be noted that the above studies were not analysing respondent’s justice evaluation of their own income (micro justice) but rather the income inequality of difficult occupational groups or of the whole community or society (macro justice). Even though distributive justice could be evaluated at the micro-social level regarding a person’s immediate circumstances and at the macro level of groups or society or the state as a whole (Hegtvedt and Markovsky 1995; Alwin 2000), both justice evaluations rarely converged (Wegener 1991).

1 See Li’s (2014b) review on the Gini Coefficient of China. 44

According to Alwin’s (2000) introduction to social justice, distributive justice would be “presumably” achieved if the generally accepted evaluative principles were implemented through a clear and just set of procedures. However, rather than consensus, the distributive justice involved “a range of competing principles- rights or entitlements, equality of outcomes, equality of opportunity, equity or proportionality of rewards, and the satisfaction of basic needs” (Alwin 2000: 2695). Meng (2012) compared the current Chinese people’s preference of principles of distributive justice and found equality of opportunity was more recognised than equality of outcomes. Similar conclusions also appeared in both Wu’s (2009) and Li’s (2014b) research, which disaffirmed the conventional sense that absolute egalitarianism of income was favoured by Chinese people, at least after the Cultural Revolution. However, for the urban citizens working in state-owned enterprises and senior citizens who had more experience of Mao’s era and the planned economy, they were more likely to perceive the current income inequality as unjust. Li and Wu (2012) claimed that this revealed the effects of egalitarianism. In terms of the egalitarianism in the Chinese education field, several studies have discussed the “compensatory education” in western countries and its policy implications for Chinese education (Yang 2009a). Compensatory education involves “redemptive egalitarianism” or “redemptive philosophy” (Coleman 1966; Coleman et al. 1966; Husen 1975; Coleman 1973; Bell 1973; Frankel 1973) and studies have discussed whether society should respond to the moral problem of redeeming the individual who “has been born with less favourable genes and in less favourable circumstances” (Husen 1975: 16-19). However, studies on the “compensatory education” in China did not research whether the redemptive egalitarianism was one of principles held by Chinese parents when they evaluated the distributive justice of educational resources.

Li and Wu (2012) summarised theories regarding how to explain the difference in distributive justice principles between different groups (or cohorts). The first set of theories focused on social stratification and included the self-interest theory (Sears and Funk 1991) and an opposite theory concerned with education’s enlightenment effects (Robinson and Bell 1978). The second set of theories focused on notions, including the

45 senses of value held by a specific group such as egalitarianism, individualism, fatalism and ascriptivism (Wegener and Liebig 1995; Verwiebe and Wegener 2000) and the ideologies at the social level such as socialism and capitalism (Svallfors 1993; Jæ ger 2006). The third set of theories focused on the effects of the socialist states transition to the market economy including two competing theories namely adaptation theory and reaction hypothesis (Listhaug and Aalberg 1999). Several studies about Chinese people’s perceived distributive justice applied the data collected in China in recent years to test some of the above theories (e.g., Li and Wu 2012; Sun 2009); however, since Chinese parents’ distributive justice of educational resources have not been empirically explored, my research will employ the inductive approach rather than the above deductive approach even though these theories are useful for inclusion in my discussion.

Wang (2011b) summarised several social psychological theories including the social comparison theory, the sense of relative deprivation theory, the attribution theory and the fair world theory (victim-blaming theory) and mixed them with some of the above theories regarding how to explain distributive justice principles between different groups to form his own classification system of literature review. However, even though these psychological theories can contribute to formulating individual subjective judgment regarding the justice of a given reward outcome (see Alwin’s (2000) review on the development of the formulation of justice evaluation), they should not be classified with the theories explaining “the effects of social structure and of the observer’s position in the stratification structure” (Jasso and Wegener 1997: 393) on intergroup variation of distributive justice evaluation. Therefore, my research will pay more attention to the theories reviewed in Li and Wu’s (2012) research.

In this chapter, I have reviewed the previous research relevant to my research aim, which is the impact of QOE policy on social reproduction of educational inequality in contemporary urban China and parents’ evaluation of the impact in terms of social justice. Based on the literature review, the next chapter (Chapter Three) will propose the analytical framework of the thesis and provide a summary of the methodology employed to explain why my research uses mixed methods and the type of mixed methods applied. The following four results chapters (Chapters Four to Seven) apply

46 different quantitative or qualitative methods and the details of methods are reported in the method section of each chapter for the convenience of reading.

47 Chapter Three. Analytical Framework and Methodology

I. Analytical Framework

My thesis aims to explore the impact of Quality-oriented Education (QOE) policy on social reproduction of educational inequality in contemporary urban China and parents’ evaluation of the impact in terms of social justice. Based on Buchmann and Hannum’s (2001) conceptual framework of research on education and stratification in developing countries and relevant previous studies reviewed in the preceding chapter, I propose the analytical framework named “the impact of QOE policy on social reproduction of educational inequality and parents’ justice evaluation of the impact” for the thesis (Figure 3.1).

Figure 3.1 The impact of QOE policy on social reproduction of educational inequality and parents’ justice evaluation of the impact

48 As I discussed in the first section of Chapter Two, my research only focuses on the inequality of educational outcome and its explanatory factors and does not further explore the association between educational outcome and social mobility as Buchmann and Hannum's (2001) framework did. Specifically, the educational outcome is indicated by students’ transition opportunities from junior high school (JHS) to regular senior high school (RSHS) and from RSHS to university. In terms of the explanatory factors in the inequality of educational outcomes, my research mainly involves the macro-structural forces, the family and school factors. The community factors may be indirectly mentioned when they are associated with the family or school factors but they are not the focus of my thesis, which is different from Buchmann and Hannum’s (2001) framework regarding the community as part of the educational supply of the school. The macro-structural force that the thesis focuses on is the QOE policy and it is to be revealed by comparing the extent of the inequality of educational outcome by family background between three QOE policy periods, namely before, during and after the implementation of QOE policy. However, it should be emphasised that besides the QOE policy, it is possible that other macro-structural forces that changed during the same period as implementation of the QOE policy might have effects on the change of association between family background factors and a student’s educational achievement. My thesis is not to make an argument that QOE policy is the exclusive factor in explaining such a change but to illustrate the possible association between QOE policy and social reproduction of educational inequality, which has not been previously studied in any great detail. My analysis framework assumes the implementation of QOE policy as a macro-structural element shapes educational stratification via its effects on families’ demand for education or the structure and supply of schooling. A family’s decision about or demand for education is associated with the family’s cultural capital (indicated by the parent’s education, usually the father’s), family’s resources (diluted by siblings) and family’s social status (indicated by parent’s occupation). Compared to the family’s demand for education as an element in the private sphere, the educational supply from schools as an element in the public sphere is supposed to be more directly and deeply influenced by the QOE as a government policy. The aspects of schooling that the QOE policy reformed mainly includes at-school time, the curriculum, pedagogy and the assessment approach.

Besides the above factors that have been mentioned in Buchmann and Hannum’s (2001) framework, my analytical framework exhibits how family and school factors interact to produce educational stratification, which was viewed as a limited focus of research effort in Buchmann and Hannum’s (2001) review article. The commercial extracurricular tutoring (hereinafter CET) is the focus of my research in the aspect of school-family interaction, yet this was not mentioned in Buchmann and Hannum’s (2001) framework. Just as its alternate names of supplementary education imply, CET is supplementary to the formal education provided by schools and should be part of the supply of schooling. However, compared to formal schools (even private schools in China), which are strictly regulated by educational authorities and thus are rather similar, CET is to a large extent under the influence of demand from families, and thus the tutorial content, pedagogy and schedule varies between different suppliers. Parents (sometimes after discussing with their child) decide what kind of tutoring their child should attend by comparing various tutorial suppliers or they do not send their child to attend any CET and/or tutor their child by themselves. It is reasonable for my research to assume that CET being supplementary to the formal education is associated with a student’s educational outcome. Even though QOE policy did not directly encourage the development of the CET tutoring industry, previous research has implied their possible association, for example the association between the QOE policy aim of “all-around development” of moral, intellectual, physical and aesthetic qualities and the popularity of art and sport tutoring as I reviewed in the previous chapter. My research aims to explore their association in greater detail and is based on empirical evidence.

The above analysis on the impact of QOE policy on social reproduction of educational inequality in the urban area is based on relatively objective data and my research aims to further explore parent’s subjective evaluation of the above impact of QOE policy in terms of social justice. Since this quantitative part of my research is exploratory, the analysis framework does not confine the exploration to a few specific preconceived conceptual relationships but rather provides a guideline. The social reproduction of educational inequality in the urban area under QOE policy can be viewed as the distributive outcome of educational resources (admission opportunity into RSHS and

50 university) and when researching parent’s evaluation of the distributive outcome in terms of justice or fairness, based on Alwin’s (2000) introduction to social justice theories, my research focuses on two key elements that determine the evaluation outcome: what evaluative principle (such as rights or entitlements, equality of outcomes, equality of opportunity, equity or proportionality of rewards and the satisfaction of basic needs) parents are employing and whether or not they think the distributive procedure is clear and just.

II. Methodology

My research uses a “mixed methods” approach, which means integrating quantitative and qualitative research within a single project (Bryman 2012). The purpose of conducting mixed methods research is to achieve richer results with breadth and depth of understanding and corroboration (Johnson, Onwuegbuzie and Turner 2007). In terms of the debate as to whether quantitative and qualitative research can be combined, I follow the technical version taken by the majority of mixed methods researchers that recognises that the connections between quantitative and qualitative research are “with distinctive epistemological and ontological assumptions” but they “are not viewed as fixed and ineluctable”, and the technical version “gives greater prominence to the strengths of the data-collection and data-analysis techniques with which quantitative and qualitative research are each associated” (Bryman 2012: 631).

More importantly, using mixed methods necessary to meet the aim of the research project, which is to explore the impact of QOE policy on social reproduction of educational inequality in the urban area and parents’ evaluation of the impact in terms of social justice. To achieve this aim, I first use quantitative research methods to determine whether the implementation of QOE policy is associated with changes in the inequality of educational opportunity by family background (see Chapter Four). At this stage, the deductive approach is employed to design my quantitative hypothesis after reviewing previous research and I hypothesise that the implementation of QOE policy is associated with the increase in the inequality of educational opportunity by family background and test the relevant hypotheses with the data from a nationwide survey.

51 Then, I commence the field work utilising qualitative methods to explain the results in Chapter Four, which follows the inductive principle. I interview school teachers, students and their parents and collect qualitative data on how the implementation of QOE has changed students learning life in and out of school by comparing parents’ own childhood with their children’s and how students from different family background have a different learning life under the QOE policy, which could mean that they have benefited differently from the changes in terms of their educational outcome (see Chapter Five). The qualitative findings in Chapter Five are derived from one city as a field work site, and thus are not able to be generalised into the nationwide urban areas of China. However, some key findings can be hypothesised and there are suitable quantitative data from another nationwide survey to test those hypotheses (see Chapter Six). Based on the above three results chapters of displaying and explaining the “reality” of changes in social reproduction in urban schools under the QOE policy, my research finally explores parents’ evaluation of those changes in distributing educational resources by family background under the QOE policy in terms of social justice or fairness. Since the research at this stage involves understanding parents’ subjective feelings and perceptions of what the just or fair principle, procedure and outcome of distributing educational resource should be, the qualitative approach is the appropriate method for the research reported in Chapter Seven and I collect this qualitative data by interviewing the same group of parents as I did for the research reported in Chapter Five.

In terms of the types of mixed methods research, Bryman (2012) proposed a nine-type classification of mixed methods (see Figure 3.2) in terms of two criteria: the priority decision (which method is dominant?) and the sequence decision (which method precedes which?). For each type, the upper case indicates which one is the main data- collection approach, the arrow refers to the sequence, and the “+” simply means that the two data-collection approaches were almost concluded concurrently.

Figure 3.2 has been removed due to copyright restrictions.

Figure 3.2 Classifying mixed methods research in terms of priority and sequence. Capitals and lower case indicate priority; arrows indicate sequence; + indicates concurrent.

However, my research is more complicated. The quantitative and qualitative approaches have more or less equal weight in my research. The two quantitative chapters (Chapters Four and Six) separately use two different second-hand datasets, namely the Chinese Family Panel Studies (CFPS [family] 1, Wave 2010 and 2012) and the China Education Panel Survey (CEPS [education], Wave 2013), which will be introduced in greater detail in Chapters Four and Six. My field work collecting qualitative data for the analyses in Chapters Five and Seven were concluded in 2015. However, if the two quantitative datasets were collected by myself and the qualitative data for the two qualitative finding chapters were collected separately, according to the internal logic of the four chapters the sequence of data-collection for the four finding chapters should have been as follows: (1) the first approach was quantitative in Chapter Four to describe the changes of social production in urban schools under QOE policy; (2) the second approach was qualitative in Chapter Five to further explore the specific clues to explain the above changes; (3) the third approach was quantitative in Chapter Six to test the key hypotheses based on the results from Chapter Five; and (4) while collecting data for all the above three chapters, simultaneously a qualitative approach was employed in Chapter Six to explore parents’ evaluation of the phenomenon and mechanism of the changes of social production in urban schools under QOE policy explored in the above three chapters in terms of social justice (see Figure 3.3).

Figure 3.3 Priority and sequence of the thesis’s mixed methods

When reporting the data collection sequence of the thesis’s mixed methods in the above paragraph, I have described the logic of combining the quantitative and qualitative

1 Since the abbreviation for these two datasets, i.e., CFPS and CEPS, are very similar, I attach a note in brackets to differentiate them.

53 approaches for the following four results chapters. Bryman (2006) identified approximately 16 ways of combining quantitative and qualitative research via content analysis of articles using mixed methods. According to Bryman’s (2006) typology, my thesis uses three different ways of combining the four chapters: (1) the “QUAN (Chapter Four) → QUAL (Chapter Five)” uses the way of “explanation” or “unexpected results”, which means the qualitative research (Chapter Five) is used to explain the results or unexpected results from the quantitative research (Chapter Six); (2) the “QUAL (Chapter Five) → QUAN (Chapter Six)” uses the way of “confirm and discover”, which entails the qualitative research generates hypotheses (Chapter Five) and then the quantitative research tests them (Chapter Six); and (3) the objective chapters (Chapters Four to Six) “+” subjective “QUAL (Chapter Seven) uses the way of “diversity of views”, which combines the researcher’s perspective through the quantitative research (and the semi- structured qualitative research in my thesis) to uncover relationships between variables and the participants’ perspective through the qualitative research to reveal meanings. For the convenience of reading, the details of different quantitative or qualitative methods used in the following four results chapters (Chapters Four to Seven) will be reported in the method section of each chapter.

54 Chapter Four. The Change of Social Reproduction of Educational Inequality Before, During and After QOE

I. Introduction

In the first chapter, I briefly described the background, content and implementation process of the Quality-oriented Education (QOE) policy and its possible connection with the fact that children from a lower socioeconomic family could be worse off in education. Even though there are several quantitative research studies describing the change in opportunity for educational transition in China over the past decades (Liu 2008; Li 2003; Li 2006; Liu 2006; Wu 2010; Li 2010; Golley and Kong 2013; Wu 2014; Li 2014a; Chen et al. 2015; Wu 2008; Lu and Treiman 2008; Wu 2013), none of them connects the change to the introduction of QOE policy. Accordingly, I have identified a research niche exploring the connection between the implementation of QOE policy and changes in social production. Therefore, the research question of this chapter is what is the changing pattern of association between students’ opportunities for educational transition and their family’s socioeconomic factors before, during and after the implementation of the QOE policy? In other words, this chapter aims to determine whether there is a possibility that QOE could moderate the association between family socioeconomic factors and children’s educational attainment. This chapter is impossible to demonstrate a causal relationship between QOE and social reproduction of educational inequality since other policies or other social factors could also play a role, however, a mutual verification between the quantitative results of this chapter and the qualitative evidences of next chapter will improve the reliability of each chapter’s conclusion.

In the first chapter, I briefly described the educational system in the post-Mao era and stated the importance of transitions from junior high school (JHS) to regular senior high school (RSHS) for students’ eventual educational attainment. The transition to RSHS is so crucial that even if a student enters a vocational senior high school instead of a RSHS,

55 she or he has a much lower opportunity of entering a university1 because a vocational senior high school usually do not offer the curriculum for the ordinary university admission examination2. In addition, since vocational education and higher education were already vocation-oriented, the focus of the QOE reform was on the primary and regular secondary education systems3, as they were more significantly affected by the Examination-oriented Education (EOE). Therefore, for the aim of my research, in this chapter my research focuses on the transition from JHS to RSHS and then from RSHS to university.

In Chapter Two, I reviewed theories explaining how family socioeconomic factors affect children’s educational achievement. These include Bourdieu’s cultural capital theory (Bourdieu and Passeron 1977), the resource dilution theory about number of siblings of children (Blake 1981) and the rational action theory in educational decision-making (Breen and Goldthorpe 1997). Wu (2013) merges the cultural capital theory and the resource dilution theory into the rational action theory by separately regarding parent’s cultural capital and the number of siblings as the operational indicators of two mechanisms in the rational action theory, namely “differences in ability and expectations of success” and “differences in resources”. The assumptions of the moderation effects of QOE are based on the above theories.

The two key policy goals of QOE as I described in Chapter One are to alleviate students’ academic burden and to encourage their all-around development. It is not unusual for policies to fail to achieve their intended goals or to produce unintended outcomes. This could explain the possible inconsistences between the hypotheses and the findings and will be elaborated in this chapter and Chapter Five. In this chapter, I present several

1 In the period from 1994 to 2011 university admission year, 64.8% RSHS students entered university while only 19.3% VSHS students entered university (Data source: the Chinese Family Panel Studies (CFPS [family]). 2 VSHS students can take UAE by teaching themselves or enter a special program of three-year VSHS plus two or three-year junior college with or without taking a special examination. 3 There are very few vocational JHSs in the post-Mao era. They are usually in rural areas and their students usually enter the labour force after graduating. In 2008, the proportion of students enrolled in vocational JHSs out of the total JHS students was 0.19% (N = 108,168), which is calculated with the data from the National Bureau of Statistics of China (2014).

56 hypotheses based on the policy documents released by the educational authorities. The QOE policy was formally launched in 1993. In 1993, 1994 and 2000, the Ministry of Education issued the directive documents for alleviating academic burden in primary school and high school students (State Education Commission 1994, 1993; Ministry of Education 2000). This directive required that local educational authorities and schools allow students to have more time for non-academic courses and extracurricular activities to develop their all-round “qualities”, especially the qualities of sport and art that had been ignored in the EOE education system. This was one of the purposes of QOE policy stated in a joint national platform on education (CCCPC and State Council 1994). Therefore, the research uses 1993 and 2000 as the cut-off years for dividing the three different periods, i.e. before, during and after the implementation of QOE, with 1993 and 2000 being the years that students enter JHS. Therefore, students in the second and third periods have spent at least three years in school under the influence of QOE policy. Table 4.1 reports the year ranges of the three cohorts in terms of their usual year of birth, the year when they enrolled in JHS, in RSHS and in university (including three-year vocational junior college program and four-year bachelor’s degree program). Although for different analyses I use different year ranges, the respondents in the same policy period are among the same cohort.

Table 4.1 Year Ranges of Cohorts across Periods Before QOE During QOE After QOE Cohort Period 1st cohort 2nd cohort 3rd cohort Usual Year of Birth 1976–1980 1981–1987 1988–1993 When Enrolling in JHS 1988–1992 1993–1999 2000–2005 When Enrolling in RSHS 1991–1995 1996–2002 2003–2008 When Enrolling in University 1994–1998 1999–2005 2006–2011

If the academic burden alleviation policy has been implemented effectively since 1993, students would have more extracurricular or after-school time after this date. Even though students could spend some of the time on peer activities, I assume the amount of time available to be at home and interacting with parents increased generally. In addition, some parents would like to use the time to send their child to an extracurricular tutoring school. According to Bourdieu’s cultural capital theory (Bourdieu and Passeron 1977; Bourdieu 1973, 1986), cultural capital can be transmitted from parents to their

57 children at home and the transmission approach includes learning and practicing linguistic and cultural competence in parent-child activities, being affected by parent’s aspirations for education or knowledge or access to cultural goods or atmosphere in the household. Since they are organised by parents, extracurricular tutoring or activities such as involvement in sports clubs can be treated as a type of cultural goods, which are shaped by different cultures of different classes. Previous research has demonstrated the association between family cultural capital as indicated by parental education and children’s probability of attending extracurricular tutoring and attending extracurricular tutoring is associated with student’s better school performance (Liu and Xie 2014). Since the curriculum and admission examination should be adjusted to giving more attention to student’s all-around development, in other words, the sport and art qualities under the QOE policy, students with more family cultural capital and attending more art and sport tutoring should perform better in school and get more advantages in the RSHS and university admission in the QOE era than in the previous EOE era. Even though in EOE era, the differentiation in school and admission performance between students with different family cultural capital should exist as well, students had less extracurricular time to make that difference even larger since “schools may serve as equalizers if the variation in school environments is smaller than the variation in non-school environments” (Downey, von Hippel and Broh 2004: 613). In addition, as I mentioned in Chapter One, compared to the lecturing and cramming method dominating the EOE teaching and learning, the new QOE teaching and learning methods encouraging initiative, exploration and creativity could adopt the linguistic and cultural forms and practices of the middle and upper social classes and thus benefit the students from these social classes, which assumption is based on the theories of Bernstein’s (1975) linguistic reproduction and Bourdieu and Passeron’s (1977) cultural reproduction. Therefore, I assume the more time students spend at home, on parent-arranged activities or with their parents, the more likely it is that the transmission of cultural capital could occur. The cultural transmission from parents to children could finally result in differences in ability and expectations of success between students from different socioeconomic backgrounds and then result in social reproduction in the field of education. Therefore, I propose the following hypothesis:

58 Hypothesis 4.1: With the implementation of QOE, the positive effects of a family’s cultural capital on a student’s opportunity of entering RSHS will gradually increase between 1991 and 2008.

Hypothesis 4.2 is the same as Hypothesis 4.1 except that it describes the change of student’s opportunity of entering higher education. Hypothesis 4.2: With the implementation of QOE, the positive effects of a family’s cultural capital on a student’s opportunity of transiting from RSHS to university will gradually increase between 1991 and 2008.

When parents want to invest resources in their children to keep their relatively high SES or to climb the social ladder, a higher sibling number could result in a smaller share of these resources distributed to each child according to the resource dilution theory (Blake 1981). It should be emphasised that these resources are not the public services such as compulsory education that almost every child can access but additional private investments unequally enjoyed by students with more family financial resources. When QOE started to be implemented, the gap of private investment in children’s education had the opportunity to enlarge significantly. The most obvious phenomenon was that extracurricular tutoring became more and more popular. Since extracurricular tutoring is provided by private suppliers for profit, the fees could become a financial burden for some families, especially when they have more than one child attending these institutions. Therefore, I argue that: Hypothesis 4.3 With the implementation of QOE, the opportunity gap of entering RSHS will gradually expand between 1991 and 2008 between the only-child students and students with one or more siblings.

Again Hypothesis 4.4 is the same as Hypothesis 4.3 except that it describes the change in a student’s opportunity of entering higher education. Hypothesis 4.4 With the implementation of QOE, the opportunity gap of transiting from RSHS to university will gradually expand between 1991 and 2008 between the only-child students and students with one or more siblings.

59 Breen and Goldthorpe’s (1997) rational action theory includes the third mechanism of relative risk aversion, which is different from Bourdieu’s cultural capital theory and Blake’s resource dilution theory. The mechanism states that the higher the social status or social prestige a group occupies, the stronger is their motivation for pursuing education because people with the highest social status need the most education to maintain their current status while people with lowest social status only lose the opportunity of upward mobility but do not need to worry about downward mobility if they choose not to purse education further. According to this theory, with the implementation of QOE I assume parents have more control over their children’s education in terms of arranging their increased extracurricular time, and thus the difference of worrying about downward mobility from different socioeconomic groups will have more influence in the era of QOE. To test this assumption, I use Wu’s (2013) strategy of operationalising the relative risk aversion mechanism with the father’s occupational status as an indicator of social status or social prestige and propose the following hypotheses: Hypothesis 4.5: With the implementation of QOE, the positive effects of a father’s occupational status on a student’s opportunity of entering RSHS will gradually increase between 1991 and 2008.

Hypothesis 4.6 is the same as Hypothesis 4.6 except that it describes the change in a student’s opportunity of entering higher education. Hypothesis 4.6: With the implementation of QOE, the positive effects of a father’s occupational status on a student’s opportunity of transiting from RSHS to university will gradually increase between 1991 and 2008.

In the first chapter, I described the significant gap in education between urban and rural people in China and suggested there were the two reasons why the thesis focused on the urban areas: one was there were much greater difficulty in implementing QOE in rural areas, and the other reason was the two phenomena, i.e., “peasant labour migration” and the “left-behind children” would make the discussion too complicated if the research included the rural sample (see the detailed discussion in Chapter One, p. 6- 7). Besides these two reasons, there is a technical reason relevant to the data-collection.

60 As I mentioned above, there is large-scale migration from rural to urban areas, which is due to not only peasant labour migration but also to land acquisition, house-buying, entering university and so on. Some of the immigrants could change their household registration from the rural “hu kou”1 (more accurately “agricultural hu kou”) to the urban “hu kou” (“non-agricultural hu kou”); however, many people cannot or choose not to change their “hu kou”. For example, the people whose lands were acquired by the government could choose to keep their current “hu kou” status for economic interest but they could enjoy the urban welfare at the same time, which is different from peasant labour migrants. Therefore, if a survey like the one used by my analysis uses a person’s “hu kou” when she or he was 12 years old as a criterion to access whether their child could enter an urban school (at least the JHS if they just change their “hu kou” at 12 years old) and access tutoring services that are more available in urban areas, it is usually correct for a person with the urban “hu kou” because it is rare to change “hu kou” from the urban one to the rural one. However, this criterion might be incorrect for a person with rural “hu kou” like these land-acquisition peasants, even though it was used by many surveys. Therefore, the urban sample is more accurate in terms of whether participants received education in rural or in urban areas. In summary, because of the above three reasons, the current research is only focused on the urban sample.

II. Data, Variables and Model

I use the data from the Chinese Family Panel Studies (CFPS [family] 2, Waves 2010 and 2012), which is a country-wide survey and contains individual-, family-, and community- level longitudinal data in contemporary China. At the individual level, the size of the 2010 sample was 452,590. The data used in this chapter are from a sub-dataset collected with the adult questionnaire and only include the respondents who have at least graduated from JHS.

1 From 2014 the central government started the “hu kou” reform and promised to merge the “agricultural hu kou” and “non-agricultural hu kou” into singe “residential hu kou” (State Council 2014a), but this policy change did not influence the cohort in my research. 2 The data are from CFPS (family), which are funded by 985 Program of Peking University and carried out by the Institute of Social Science Survey of Peking University. The CFPS (family) website address is http://www.isss.edu.cn/cfps/EN/.

Dependent variables Because the analysis is focused on two educational transitions: from JHS to RSHS and from RSHS to university, the dependent variables are whether the participant entered a RSHS or not (Yes = 1) and whether the participant who had entered a RSHS eventually entered university or not (Yes = 1). The cases in the analysis of transition from JHS to RSHS are old enough to have taken the RSHS admission examination. Similarly, the cases in the analysis of transition from RSHS to university must be old enough to have taken the university admission examination and the university education does not include online or continuing program.

Independent variables Main explanatory variables: The main explanatory variables include policy period, father’s years of education, sibling number and father’s ISEI. The policy period variable consists of two dummy variables During QOE and After QOE with the Before QOE as the reference. In the introduction to this chapter, I described the criteria for dividing the three QOE policy periods. For the analysis of transition from JHS to RSHS, the policy periods are the ranges of years when students graduated from JHS or enrolled in RSHS if they could (see the fifth row in Table 1). For the analysis of transition from RSHS to university, the policy periods are the ranges of years when students graduated from RSHS or enrolled in university if they could (see the last row in Table 1). The hypotheses proposed in the introduction describe the effects of family’s cultural capital, resources and social prestige, which are indicated by father’s years of education, sibling number and father’s ISEI, respectively. A father’s years of education is defined as years of schooling of the interviewee’s father when the interviewee was 14 years old. I treat illiterate or semi-literate as 0 years, primary school as 6 years, JHS as 9 years, senior high school as 12 years, junior college as 15 years, and four-year university or more as 16 years. To reduce the effects of outliers on the model-fitting and to reveal the possible speciality of only-child for the Chinese one-child policy, I recode the sibling number into a category variable measured by two dummy variables with only-child as the reference, one sibling and two siblings or more. I recode more than two siblings as three because all the three policy periods are in the era of the strict one-child policy and cases with

62 more than two siblings are extraordinary and less than 8% of the analysed sample. The father’s ISEI is a continuous score from ISEI for the 1988 International Standard Classification of Occupation (ISCO88) coding of father’s occupation (Ganzeboom and Treiman 1996; Ganzeboom et al. 1992). The father’s occupation is stated as the one when the interviewee was 14 years old.

Control variables: I refer to Tam and Jiang’s (2015) strategy to control the educational expansion by adding two variables to the model: initial enrolment rate of the first year of the policy period and enrolment rate of the year when the interviewee took the admission examination. Through controlling the two variable’s effects simultaneously, the estimated effects of enrolment rates in a given admission district can reflect the effects of deviations from the initial enrolment rate (Tam and Jiang 2015). In the brief introduction to Chinese education in Chapter One, I described admission districts and the quota system in both RSHS and university admission processes. For the RSHS admission, the admission district is the county or urban district but there is no local county-level admission quota data for estimating the enrolment rate of every RSHS admission district and, more importantly, the case is only identified at the province level. Therefore, I use the enrolment rate data at the province level. For the RSHS admission cohort in year T, I estimate the number of students (N) in a respondent’s age cohort using the number of graduates from primary schools (P) for the same province in year T

p minus 3, i.e., N (T-3). The time lag of three years is because of the normal time difference between graduation from primary school and RSHS admission. The province-specific enrolment rate of RSHS is the number of students admitted to RSHS for the RSHS

p admission cohort in year T divided by N (T-3). All the graduation and admission student numbers are from the open online database of the National Bureau of Statistics of China1.

However, in terms of university admissions, the number of students admitted per province provided by the National Bureau of Statistics does not only include local

1 The web site address is http://www.stats.gov.cn/.

63 students but also students from other provinces1 because provincial universities are required to mainly admit local students while the universities regulated by the Ministry of Education must recruit a considerable number of students from other provinces. Furthermore, there is no public document or research article that explains how to divide the proportion of local students to non-local students. To approximately estimate the local RSHS students’ university enrolment rate, Tam and Jiang (2015) used the average of the national and the provincial university enrolment rates, which are the numbers of

p students enrolled in university for the nation or a province in year T divided by N (T-6). The time lag of six years is due to the normal time difference between graduation from primary school and university admission. Tam and Jiang’s (2015) university enrolment rate (I call it “gross university enrolment rate”) includes the transition from vocational senior high school to university2, whereas my analysis is focused on the transition from RSHS to university, and thus at first used the same calculation method except for

p R replacing N (T-6) with N (T-3), which is the number of graduates from RSHS (R) for the same province in year T minus 3. I call this rate “net university enrolment rate”. Nevertheless, after undertaking a logit-regression on the policy period variable, initial university enrolment rate (in 1994) and university enrolment rate, I found neither the gross nor net university enrolment rate was significant and the KHB-method (Breen, Karlson and Holm 2010; Karlson 2015; Karlson, Holm and Breen 2012) test did not find a significant coefficient difference between their reduced model (without educational expansion variables) and full model (with educational expansion variables). This indicated that the university enrolment rate measured by this method did not succeed in controlling the effects of educational expansion. Thus, I kept applying “net” and “gross” university enrolment rates with different proportions of the national to provincial university enrolment rates and with the same logit-regression and KHB testing method as above. After several iterations, I discovered that the gross university enrolment rate with a proportion of 7/10 national and 3/10 provincial university enrolment rate was the most

1 The number of international university students is very small. In 2011, the proportion of international students in all university students enrolled in a three or four-year university program was 0.51% (N=118,837) (Ministry of Education 2012; NBSC 2014). 2 Students in VSHS can take a specific admission examination to enter a joint junior-college program or learn by themselves to take the ordinary university admission examination; however, the number of students with this kind of transition is very small and this kind of transition is rarely relevant to QOE.

64 appropriate for controlling the effects of educational expansion. One reasonable explanation for this composition of university enrolment rate is that the Ministry of Education can control the university enrolment quota given to every province but cannot or does not want to control the RSHS enrolment quota. However, since the era of imperial examination (“ke ju”, 科举) (Xie 2009), the central authority has to consider educational equality between different provinces in terms of the admission as the previous “ke ju” or the current “gao kao” (高考, the university admission examination) is the most important social mobility pathway. Since different provinces have different RSHS enrolment rates, the Ministry of Education could only moderate the university enrolment quota allocated to each province to achieve a balanced gross university enrolment. Even though the central government has a much higher authority over the provincial governments in China, the interest of provinces with more universities should be considered as the provincial governments provide the majority of the budget for the provincial universities and some for universities directly under ministries and commissions of the Central Government. Therefore, the current variable “gross” university enrolment rate with a proportion of 7/10 national and 3/10 provincial university enrolment rate is significant.

Since the enrolment rate calculated above is just an expedient measure rather than an accurate value provided by the educational authorities, in the summary section of this chapter I will discuss whether the measure for the enrolment rate in the analysis still could not control the main effects of educational expansion, and whether the results can only be explained by the educational expansion and exclude other alternative explanations such as the QOE policy. In addition, the enrolment rate variables in both transition analyses are not individual level but are admission district level variables. Therefore, I use the cluster-robust method to ensure that the standard errors reported in the analysis results are valid for clustering within an admission district. The critical reason why I do not present results based on multilevel mixed-effects models is because these models with interaction effects are hardly concave, which is similar to Tam and Jiang’s (2015) encounter and I use a similar method to theirs.

65 Another control variable is gender, with female as the reference because of the traditional consideration of gender inequality in Chinese education. In the beginning, I controlled ethnicity as well but during the analysis I found this variable was not significant especially when controlling for the family background factors. Therefore, to focus on the effects of family background factors, I exclude ethnicity as Tam and Jiang (2015) did as well.

Table 4.2 reports descriptive statistics of variables in the analyses of the transition from JHS to RSHS and the transition from RSHS to university. Because there is a considerable number of missing values in the father’s ISEI, I used the multiple-imputation method to manage missing data and repeated the same models as the ones presented in the results section of this chapter. I found that the results with missing imputation presented a similar trend of social reproduction in schools across the three QOE policy periods as the one presented in the results section of this chapter. The results with missing imputations can be found in Appendixes 1 and 2 of this thesis. In addition, I applied sampling weights provided by CFPS (family) to correct for oversampling and compute figures representative of China’s general population. The cluster-robust method can also make the reported standard errors valid for weighted estimations.

Table 4.2 Descriptive Statistics for All Variables in Analyses From JHS to RSHS From RSHS to University Variable N Mean S. D. N Mean S. D. DV: Success of Transition 909 0.585 0.493 528 0.648 0.478 Policy Period 977 531 Before 344 0.352 0.478 131 0.247 0.431 During 370 0.379 0.485 204 0.384 0.487 After 263 0.269 0.444 191 0.369 0.483 Initial enrollment rate 977 14.29 4.856 531 5.577 2.861 Enrollment rate 977 25.86 12.49 531 21.11 11.53 Male (Ref=Female) 977 0.467 0.499 531 0.463 0.499 Father’s years of education 929 9.765 3.464 506 10.29 3.432 Sibling number 976 530 0 sibling 470 0.482 0.500 283 0.534 0.499 1 sibling 332 0.340 0.474 176 0.332 0.471 2 siblings or more 174 0.178 0.383 71 0.134 0.341 Father’s ISEI 668 42.18 17.14 359 43.88 17.85

66 Method and modelling procedures Since both the dependent variables are binary, in the current study I employed the logistic regression model, more specifically Wu’s (2013) revised version of Mare’s (1981) logistic response model of school continuation to separately estimate the effect of the interaction between the policy period variable and family background factors. This aimed to examine whether or not the effects of family background factors on a student’s transition opportunity differ between the different QOE policy periods. Wu (2013) made two revisions of Mare’s (1981) model: (1) instead of comparing different cohorts of birth, Wu directly compared different cohorts of educational transition; and (2) instead of modelling with cohorts separately and simply comparing coefficients of different models without significance testing, Wu (2013) included all cohorts into one sample and used the transition period as the independent variable. Wu (2013) then separately tested the significance of the differences in family background factors’ coefficients between different periods by estimating the interaction effects between the period variable and family background factors. I employed Wu’s (2013) model and utilised Stata 14 for all analyses (StataCorp 2015b). An example is the model estimating the interaction effects between the period variable and the father’s years of education. The model that I fit is: Pr(퐸푛푡푒푟푖푛푔 = 1)

= 퐹(훽0 + 훽1퐼푛푖퐸푟푙푖푗 + 훽2퐸푟푙푖푗 + 훽3푀푖푗 + 훽42푃푟푑2푖푗 + 훽43푃푟푑3푖푗

+ 훽5퐹푒푑푢푖푗 + 훽61푆푖푏1푖푗 + 훽62푆푖푏2푖푗 + 훽7퐹푖푠푒푖푖푗 + 훽82푃푟푑2푖푗 ∗ 퐹푒푑푢푖푗

+ 훽83푃푟푑3푖푗 ∗ 퐹푒푑푢푖푗) where, 퐹(푧) = 푒푧⁄(1 + 푒푧) is the cumulative logistic distribution.

In this equation, 퐼푛푖퐸푟푙푖푗 and 퐸푟푙푖푗 are the initial value of enrolment rate measure and the enrolment rate measure of student i in province j, respectively. 푀푖푗 indicates a student’s gender. The policy period variable is indicated by two dummy measures with the first policy period as the reference, 푃푟푑2푖푗 indicates a dummy measure for the second policy period and 푃푟푑3푖푗 indicates the third one. Two family background factors namely the father’s years of education and the father’s ISEI are indicated by

퐹푒푑푢푖푗 and 퐹푖푠푒푖푖푗 , respectively, while the third family background factor namely sibling number is indicated by two dummy measures 푆푖푏1푖푗 푎푛푑 푆푖푏2푖푗 that stand for

67 student i with one sibling or with two siblings or more, respectively, and the reference is with no sibling. 푃푟푑2푖푗 ∗ 퐹푒푑푢푖푗 and 푃푟푑3푖푗 ∗ 퐹푒푑푢푖푗 indicate the interaction items and here I take the interaction between the policy period measures and the measure for the father’s years of education. In addition, I employ the Huber-White adjustment (Huber 1966; White 1980) to correct the error term and cluster the error at the province level.

III. Results

In this section, I present the results from the two sets of logistic models that predicted the probability of a student transiting from JHS to RSHS and the probability of a student transiting from RSHS to university. Each set included two simple models testing the control effects of educational rate variables, a full model containing all the variables, and the models with the interaction effects between the policy period and family background factors.

Change in a student’s opportunity of transiting from JHS entering RSHS Table 4.3 reports the results of the logistic models predicting the probability of a student entering RSHS. Models 1 (M1) and 2 (M2) show the results as to whether adding the two variables Initial Value of Enrolment Rate and Enrolment Rate can control the effects of educational expansion or not. After adding these two educational expansion variables, the coefficients of the two dummy variables of Policy Periods became smaller and insignificant while the enrolment rate was significant. I applied the KHB-method to further test the above change with the original data1 (Breen, Karlson and Holm 2010; Karlson 2015; Karlson, Holm and Breen 2012) and the results show that for both dummy variables the coefficient differences between the reduced model (without educational expansion variables) and the full model (with educational expansion variables) are significant (z = 2.57 for during QOE and z = 2.55 for after QOE, with before QOE the

1 This is because the current version of KHB cannot run with multiple imputed data. However, the case number with missing values in the policy period variable and educational expansion variables are only 68 (N = 977 in Models 1 and 2 of Table 4.3; N = 909 for the KHB results). After running Models 1 and 2 with the original data, there was no significant difference from the results in Table 4.3. 68 reference). The results indicate that the two enrolment rate variables can control the effects of educational expansion, which enabled me to use the policy period variable to test the effects of its contemporary policy QOE.

Model 3 (M3) presents the logistic regression results of both explanatory variables and control variables on the probability of a student entering RSHS. Since my focus is on the change of coefficients of family’s socioeconomic factors between different policy periods, I do not describe M3 in detail. Models 4 (M4) to 6 (M6) present the interaction effects between policy periods and family background factors and are the main part of the results report. The bold italic values in the table were estimated by running the same regression but changing the reference of dummy variables of the policy period variable. These values show the changes in coefficients of family background factors between the “after” and “during” period (the “during” period is the reference).

69 Table 4.3 Logistic Models Predicting Probability of Student Transiting from JHS to RSHS M1 M2 M3 M4 M5 M6 During vs. Before& 0.670***(3.97) 0.336(1.50) 0.170(0.46) -1.089(-1.52) 0.843(1.52) -1.152(-1.51) After vs. Before 1.460***(4.63) 0.675(1.52) 0.763(1.33) -0.127(-0.13) 1.334*(1.98) 0.544(0.80) Initial enroll. rate -0.022(-1.12) -0.042(-1.30) -0.046(-1.40) -0.042(-1.26) -0.046(-1.41) Enrollment rate 0.031**(2.60) 0.039*(2.46) 0.041*(2.57) 0.037*(2.29) 0.042**(2.62) Male -0.427*(-2.24) -0.407*(-2.12) -0.454*(-2.46) -0.444*(-2.55) Father’s yrs. of edu. 0.081*(2.46) 0.003(0.05) 0.079*(2.43) 0.075*(2.13) Sibling No. 1 sibling vs. None 0.058(0.19) 0.062(0.20) 1.029*(2.22) 0.025(0.08) 2+ siblings vs. -0.599+(-1.77) -0.626+(-1.78) -0.292(-0.64) -0.646+(-1.91) None Father’s ISEI 0.016**(2.59) 0.015*(2.51) 0.018**(2.80) 0.002(0.22) Interaction Father’s Edu. Sibling No. Father’s ISEI During vs. Before 0.130+(1.94) -1.356**(-3.26) # 0.030+(1.73) -0.819(-0.99) ## After vs. Before 0.089(1.16) -1.360**(-3.15) # 0.002(0.15) 0.291(0.61) ## After vs. During^ -0.041(-0.49) -0.004(-0.01) # -0.029*(-2.56) 1.110(1.21) ## _cons -0.211(-1.29) -0.347(-1.31) -1.447**(-3.12) -0.703(-1.14) -1.975***(-3.46) -0.767(-1.36) n 909 909 613 613 613 613 Pseudo R2 0.052 0.058 0.136 0.141 0.149 0.145 Note: t statistics in parentheses. + p < 0.10, * p < 0.05, ** p < 0.01, *** p < 0.001. The values in the constant and pseudo R2 are from the regression when treating the first policy period as the reference. & In the table, the latter of each two-measure comparison is the reference. For example, “before” is the reference in “During vs. Before”. # The results indicate the interaction between a policy period dummy measure and the dummy measure for one sibling and ## for two siblings or more. ^ The bold italic values are estimated by running the same regression but changing the reference of dummy variables of the policy period variable.

70 Column 1 of Table 4.4 presents the average marginal effects (AME) of a father’s years of education and IESI on a student’s probability of entering RSHS and the average predicted probabilities (APP) of entering RSHS of student groups with different sibling numbers across three QOE policy periods. To clearly display the estimation results of significance level (at least p < 0.05) for the above trends of social reproduction in school and to facilitate the discussion following this results section, I manually drew the figures in Column 2 of Table 4.4. For example, in Row 1, Column 1 of Table 4.4 the trend in the effects of a father's education on a student’s transition opportunity from RSHS to university forms a reverse V-shape; however, both the changes from before QOE to during QOE and from during QOE to after QOE are not significant. Thus, I simply display the trend with a horizontal line in Row 1, Column 2 of Table 4.4. Even though the increase in effects of father’s ISEI from the “before” to “during” period is only significant at the 90% significant level, I still use a reverse V-shape to indicate the trend because if I treated the two effects of father’s ISEI in the “before” and “during” periods as being equal, there would be a similar significant difference between the “before” and “after” period as the one between in the “during” and “after” periods. However, the estimation results in Table 4.3 suggest there is no significant difference in the effects of a father’s ISEI on a student’s probability of entering RSHS between the “during” and “after” periods. I describe those results and trends as follows.

71 Table 4.4 AME / APP of Family Background Factors on Transition from JHS to RSHS 1. AME / APP 2. Trend with significance

0.18

0.15 .03 0.12

.02 0.09 0.06

.01 0.03

0 0

AME Education of Father's Before QOE During QOE After QOE

1. Before QOE During QOE After QOE

0.85

.8 0.75 × ^ .7 0.65

.6 0.55 ×

.5 0.45

.4 0.35 Before QOE During QOE After QOE

Before QOE During QOE After QOE 2.APP by Sibling Number 0 Sibling 1 Sibling 2 Siblings no sibling 1 sibling 2+ siblings

0.04

.008

0.03 .006 0.02 + .004 0.01

.002 0

3.AME of Father's ISEI 0 Before QOE During QOE After QOE

Before QOE During QOE After QOE

Note: AME = average marginal effects. APP= average predicted probability. The ^ in the second figure of Column 2 indicates the gap between one sibling and the other two sibling indicators in the after QOE period is logically but not statistically significant. The X in the second figure of Column 2 indicates the increase in the coefficients of one sibling and two siblings or more between two periods is logically but not statistically significant since the increase is significant between the first and third periods. The + in the third figure of Column 2 indicates the increase is logically significant but only at the 90% significance level. The values indicated by the plots in each figure of Column 2 are not completely consistent with the predictions as they have been adjusted to draw the trend figures.

72 Model 4 (M4) in Table 4.3 suggests that the coefficients of father’s years of education are only significant in the second (“during”) period and, as mentioned above, Model 4 suggests the changes in coefficients of father’s years of education between three policy periods are not statistically significant at the 95% or higher confidence level. The trend formed by the average marginal effects of father’s ISEI during the three QOE policy periods can also be viewed in the two figures of the first row in Table 4.4. Therefore, the results about the effects of family’s cultural capital are not consistent with Hypothesis 4.1.

Model 5 (M5) further tests the hypothesis of the effects of sibling number on a student’s opportunity of entering RSHS. The first figure in the second row of Table 4.4 is formed by the APP of entering RSHS with three different sibling number levels across three QOE policy periods and after pair-wise comparisons of those margins or APP, I manually drew the second figure in the second row to clearly elaborate on this trend. The gap between the RSHS admission probability of only-child students and students with one sibling is significant in the before QOE period; however, unexpectedly, students with one sibling have approximate 2.8 (e1.029≈2.8) times the RSHS admission odds of only-child students, which needs discussing further. The gap significantly narrows and reverses in the second period, which is that the RSHS admission odds of students with one sibling is approximately 70% (e1.029-1.356≈0.7) of the ones of only-child students. In the third period, the gap almost does not change compared with the one in the second period. Even though the result is not consistent with the expression of Hypothesis 4.3, according to the first figure in the second row of Table 4.4 the reason for the gap narrowing and reversing is that the admission probability of only-child students increases while the probability of students with one sibling remains almost unchanged across the three QOE policy periods, which is consistent with the original intention of Hypothesis 4.3 in that families with more financial resources per child benefit from the QOE policy. Since there is no significant difference in the RSHS admission probability between only-child students and students with two siblings or more for each QOE policy period, the gap trend between students with one sibling and student with two siblings or more is similar to the one above, which also requires further discussion (see the summary of Chapter Five).

Model 6 (M6) presents the changes of the effects of father’s ISEI between different policy periods. The results indicate that the coefficient of the father’s ISEI is only significant during the implementation of QOE but is not significant before or after. During the implementation of QOE, as father’s ISEI increases one unit, the odds of student entering RSHS increases by approximately 3.4% (e0.033-1≈0.034) while the effect size of father’s ISEI before or after the implementation of QOE is almost zero. Considering the range of father’s ISEI is between 19 to 90, the inequality in the odds of entering RSHS between students with the highest and the lowest father’s IESI increases by approximately 240% in the second QOE period. The coefficient of father’s ISEI only increases at the 90% significance level from the first period to the second period but does not change significantly between the first and third period; however, it decreases significantly from the second period to the third period. After comparing the coefficients of father’s ISEI during the three policy periods and testing the significance of the differences between these three policy periods, it is possible that there is an inverted V trend (see figures in the third row of Table 4.4, row title “AME of father’s ISEI”) in the change of the effects of a father’s ISEI on a student’s opportunity of transiting from JHS to RSHS, which is not consistent with Hypothesis 4.5. I will propose some possible explanations for this discrepancy especially regarding the unexpected dropping from the second period to the third period.

Change in a student’s opportunity of transiting from RSHS to University Table 4.5 reports the results of logistic models predicting the probability of a student transiting from RSHS to higher education (three-year junior college or four-year bachelor-degree program). Similar to the strategy shown in Table 4.3, Models 1 and 2 in Table 4.5 report the testing results for whether adding the two variables Initial Enrolment Rate and Enrolment Rate when measuring the provincial transition opportunity from RSHS to university1 can control the effects of expansion to higher education or not. After adding these two educational expansion variables, the coefficients of two dummy variables of Policy Periods became smaller and insignificant.

1 The detailed estimation method for these two transition rate variables has been described in the variable section of this chapter. 74 Even though it is not significant at the 95% significance level in Model 2, the transition rate becomes significant when adding another control variable and the explanatory variables in Model 3. Again, I applied the KHB-method to further test the above changes. After adding the two transition rate variables, the coefficient differences between the reduced and full models (N = 528) were significant for during QOE (z = 2.00, before QOE is the reference) but not significant at the 95% confidence level for after QOE (z = 2.89). However, after adding gender and several explanatory variables (only excluding father’s ISEI because of its missing value issue) into the full model (N = 502), the coefficient differences were significant for both dummy variables of the policy period variable (z = 2.36 for during QOE and z = 1.99 for after QOE). Therefore, the two transition rate variables could control the effects of educational expansion in the transition from RSHS to university.

75 Table 4.5 Logistic Models Predicting Probability of Student Transiting from RSHS to University M 1 M 2 M3 M 4 M 5 M 6 During vs. Before& 1.561***(9.32) 0.734(1.51) 0.858(1.36) 1.659(1.37) 2.219**(2.82) 1.231(1.10) After vs. Before 1.214***(4.03) -0.237(-0.30) 0.126(0.12) 3.037*(1.98) 0.562(0.48) 0.182(0.13) Initial enroll. rate -0.014(-0.27) 0.033(0.30) 0.039(0.34) 0.011(0.09) 0.032(0.29) Enrollment rate 0.061+(1.93) 0.044(1.04) 0.038(0.91) 0.049(1.13) 0.044(1.05) Male -0.273(-0.83) -0.372(-1.07) -0.386(-1.19) -0.273(-0.83) Father’s yrs. of edu. 0.041(1.20) 0.173*(2.10) 0.055+(1.71) 0.044(1.30) Sibling No. 1 sibling vs. None -0.522(-1.61) -0.625+(-1.91) 0.537(1.16) -0.526(-1.60) 2+ siblings vs. None -1.171*(-2.30) -1.336**(-2.94) 0.091(0.09) -1.172*(-2.27) Father’s ISEI 0.020**(2.88) 0.021**(2.81) 0.019**(2.80) 0.023(1.22) Interaction Father’s edu. Sibling No. Father’s ISEI During vs. Before -0.086(-0.71) -2.353***(-4.02) # -0.009(-0.42) -3.070**(-2.92) ## After vs. Before -0.275**(-2.64) -0.408(-0.49) # -0.001(-0.06) -0.978(-0.93) ## After vs. During^ -0.189**(-2.67) 1.945**(2.82) # 0.007(0.48) 2.092**(2.72) ## _cons -0.404*(-2.25) -0.734**(-3.16) -1.498**(-2.81) -2.651**(-2.89) -2.233***(-4.08) -1.670*(-1.99) n 528 528 356 356 356 356 Pseudo R2 0.062 0.081 0.142 0.157 0.173 0.133 Note: t statistics in parentheses. + p < 0.10, * p < 0.05, ** p < 0.01, *** p < 0.001. The values in the constant and pseudo R2 are from the regression when treating the first policy period as the reference. & In the table, the latter of each two-measure comparison is the reference. For example, “before” is the reference in “During vs. Before”. # The results indicate the interaction between a policy period dummy measure and the dummy measure for one sibling and ## for two siblings or more. ^ The bold italic values are estimated by running the same regression but changing the reference of dummy variables of the policy period variable.

Models 4 to 6 in Table 4.5 presents the coefficients of family background factors in different policy periods and the change of their coefficients between different policy periods through estimating the interaction effects. Model 4 shows that the coefficients of father’s years of education keep decreasing from the first period to the third period (also see the figure in Row 1, Column 2 of Table 4.6) and in the third period it become negative, which makes the difference between the coefficient in the first period and third period significant. More specifically, before the implementation of QOE, as father’s education increases one year, the odds of student transiting from RSHS to university increase by approximately 19% (e0.173-1≈0.19) while after the implementation of QOE, as father’s education increases one year, the odds of student transiting from RSHS to university do not increase but decrease by approximate 10% (1-e0.173-0.275≈0.10). Even though the negative effects of father’s education in the third QOE period is not statistically significant after further testing, these results are still opposite to Hypothesis 4.2, which will be further explored in Chapter Five.

In contrast to the narrowing trend in the gaps of RSHS admission probability between students with different sibling number levels (see the figures in Table 4.4), the gaps in the transition opportunity from RSHS to university is only significant in the second QOE policy period (see figures in the second row of Table 4.6, row Title: “APP of Sibling Number”), which are that the transition odds of students with one sibling are approximately 16% (e0.537-2.353≈0.16) of the ones of only-child students, and that the transition odds of students with two or more siblings is approximately 8% (e0.537-3.070≈ 0.08) of the ones of only-child students. This gap is caused by the significant increase in only-child students’ probability of transiting from RSHS to university in the second QOE policy period and the transition probability of students with one sibling and with two siblings or more remains unchanged. However, in the third period the gaps in RSHS admission probability between students with different sibling number levels disappear again similar to the first period. These results indicate that the dilution of more siblings on families’ resources were more significant in the second period than in either the first or third period. Thus, the possibility of a student with more siblings entering university after graduating from RSHS in the second period is the lowest among the three periods.

This is not consistent with the linear assumption in Hypothesis 4.4 and will be further explored in Chapter Five.

However, it is unexpected that there is no significant change of coefficients in father’s ISEI between the three policy periods (see Model 6 of Table 4.5 and figures in the third row of Table 4.6, row title: “AME of Father's ISEI”). Therefore, hypothesis 4.5 has not been confirmed. More importantly, the result is different from the inverted V change tendency about the effects of a father’s ISEI on the probability of a student entering RSHS, as the figures in the third row of Table 4.4 showed, which will be further explored in Chapter Five.

Table 4.6 AME / APP of Family Background Factors on Transition from RSHS to University 1. AME / APP 2. Trend with significance

0.2 .03

0.1 .02

.01 0

0 Before QOE During QOE After QOE

-0.1 -.01

-0.2 -.02

1.AME of Father's Education Before QOE During QOE After QOE Average Marginal Effects of Father's Yrs of Edu on Probility of Entering University 1

0.8 .9

0.6 .8

.7 0.4 × × .6

0.2 .5

.4 0 Before QOE During QOE After QOE

2.APP by Sibling Number Before QOE During QOE After QOE 0 Sibling 1 Sibling 2 Siblings no sibling 1 sibling 2+ siblings

0.03

.0045 .004

0.02 .0035

.003 0.01 .0025

.002 0 3.AME of Father's ISEI Before QOE During QOE After QOE Before QOE During QOE After QOE Average Marginal Effects of Father's ISEI on Probility of Entering University across 3 QOE policy peirods Note: AME= average marginal effects. APP= average predicted probability. The × in the 2rd figure of Column 2 indicates the increase or decrease of coefficients of 2 siblings is logically but not statistically significant because the gap between the coefficients of 1 sibling and 2 or more siblings in the during QOE period is significant. The values indicated by the plots in each figure of Column 2 are not completely consistent with the predictions as they have been adjusted to draw the trend figures.

IV. Summary and Discussion

In this chapter, I have estimated the change of effects of family background factors on students’ opportunity of entering RSHS and their opportunity of transiting from RSHS to university between three different QOE policy periods (1987–2006). These factors

include father’s years of education, sibling numbers and father’s ISEI. To test the significance of these above changes, namely coefficient differences of family background factors between three policy periods in logistic models, I estimated the interaction effects between family background factors and the policy period variable. Following is a brief summary of the results.

I predicted that the positive effects of a family’s cultural capital (indicated by father’s education, Hypotheses 4.1 and 4.2), resources (reversely indicated by sibling number, Hypotheses 4.3 and 4.4) and family social status (indicated by father’s ISEI, Hypotheses 4.5 and 4.6) on a student’s educational transitions from JHS to RSHS and then from RSHS to university would keep increasing alongside the implementation of QOE policy (see Table 4.7 for the convenience of comparison between the hypothesis figures, i.e., Table 4.7 and result figures, i.e., Tables 4.4 and 4.6). However, the results are much more complicated than the above straightforward hypothesis. In the transition from JHS to RSHS, only the narrowing RSHS admission probability gap between students with one sibling (originally having higher probability) and only-child students is consistent with Hypothesis 4.3 that the family with more financial resources per child will benefit more from the QOE policy. However, except for the above gap trend, the gap trend relevant to the effects of two siblings or more is not consistent with the hypothesis, i.e., the gap between students with one sibling (originally having higher probability) and students with two siblings or more does not expand as the hypothesis expected but narrows and there is no gap between students with two siblings or more and only-child students across the three QOE policy periods. In addition, different from Hypothesis 4.1, the effects of the father’s years of education do not vary significantly as the implementation of QOE proceeds and different from Hypothesis 4.5, the effects of father’s ISEI increases between the first and second period and then decreases between the second and third period. The results indicate, in terms of RSHS admission probability that students with a larger family culture capital do not benefit more from the QOE period and for students with a higher family social status, their advantage only increases in the second period but decreases from the second to third period.

Table 4.7 Hypothesized Changes in Social Reproduction of Educational Inequality 1.Transition from JHS to RSHS 2. Transition from RSHS to University

0.18 0.18 0.15 0.15 0.12 0.12 0.09 0.09 0.06 0.06 0.03 0.03 0 0 Before QOE During QOE After QOE Before QOE During QOE After QOE

Effects ofEffects Father's Education

1 1

0.8 0.8

0.6 0.6

0.4 0.4

No. 0.2 0.2

0 0 Before QOE During QOE After QOE Before QOE During QOE After QOE

2.Transition Probability by Sibling Probability by 2.Transition 0 Sibling 1 Sibling 2 Siblings 0 Sibling 1 Sibling 2 Siblings 0.005 0.005

0.004 0.004 0.003 0.003 0.002 0.002 0.001 0.001 0 3.Effects of Father's ISEI 0 Before QOE During QOE After QOE Before QOE During QOE After QOE

Furthermore, all the effect trends of family background factors on the transition from JHS to RSHS are different to the hypotheses and from the ones on the transition from RSHS to university. On the transition from RSHS to university, different from Hypothesis 4.2, the effects of family’s cultural capital decrease as the implementation of QOE progresses while the advantage of students with more family resources become more salient only in the second period and not in the first or third period, which is different from Hypothesis 4.4. The effects of family’s social status do not change significantly across the three different periods, which is different from Hypothesis 4.6.

Since the control variables of enrolment rates are only expedient measures for educational expansion rather than accurate values provided by the educational authorities, I performed a further check on their effects. After repeating the models in Tables 4.3 and 4.5 without the two measures for educational expansion, the results obtained show a similar pattern of the changes in the inequality of educational opportunity between social groups with different levels of three socioeconomic indicators (see Table 4.8). Therefore, it is possible that the measures for enrolment rates in the analysis could not control the main effects of educational expansion; Or irrespective of the extent of educational expansion, the trend of social reproduction in schools during this period is the same as that indicated in Table 4.8. The critical question is can the above results only be explained by the education expansion and exclude the QOE policy as an alternative explanation? According to Li’s (2017b) threshold dependent inequality theory, which is proposed to modify the application of Raftery and Hout’s (1993) maximally maintained inequality theory to China, the educational expansion is supposed to make the inequality of educational opportunity between different social groups first increase and then decrease. If the educational expansion was the exclusive factor, the trends of social reproduction in schools should have formed a “^” shape (if the expansion in this period started with a low enrolment rate) or should have been downward (if the expansion in this period started with a high enrolment rate). However, according to Table 4.8, the trends in the effects of three family background factors on a student’s RSHS admission opportunity do not form a consistent “^” shape and the three trends in a student’s transitional opportunity from RSHS to university are different from each other, not form a consistent downward shape. The possible explanation is that the distribution pattern of students’ examination scores by father’s education is different from the pattern by sibling number or father’s ISEI. However, there is no previous research making a similar conclusion and there is no suitable dataset to test this explanation.

Table 4.8 AME and APP of Family Background Factors on Educational Transitions Without Educational Expansion Measures From JHS to RSHS From RSHS to University

.03

.04

.02

.01

0 -.02

AME of Father's Education Before QOE During QOE After QOE Before QOE During QOE After QOE

APP by Sibling Number Before QOE During QOE After QOE Before QOE During QOE After QOE no sibling 1 sibling 2+ siblings no sibling 1 sibling 2+ siblings

.008

.006 .006

.005 .004

.004 .002

.003 0

AME of Father's ISEI

.002

Before QOE During QOE After QOE Before QOE During QOE After QOE

Note: AME = average marginal effects. APP = average predicted probability.

Even if the explanation based on educational expansion is correct, the QOE policy still can be considered as an alternative explanation, which may not exclude the effects of educational expansion but have a combined effect on the social reproduction in schools with educational expansion and other concurrent polices and social changes. That is because the educational expansion is to lower the gate line of admission scores in

general and whether the inequality increase or decreases depends on a fixed threshold, which is based on the normal distribution and a fixed gap of examination score means between students with different levels of family background factors. The explanation based on educational expansion does not consider the possible change in the gap of examination score means, which is to be explained by the QOE policy.

However, different from what the hypotheses in this chapter expected, the implementation of QOE policy was not always increasing the gap of examination score means between privileged and underprivileged students, and a similar problem to the explanation based on educational expansion is these three trends in effects of family background factors on a student’s transitional opportunity are different from each other. To explain the above discrepancy between my research hypotheses and the results obtained, I consider the possibility of the gap between the QOE policy design and its implementation, the critical differences among the three family background factors and between the transition from JHS to RSHS and the one from RSHS to university in terms of their association with the QOE policy, and alternative contemporaneous policies or social changes that could have affected the policy outcomes. Some possible explanations for the failure to confirm some of those research hypotheses based on my existing knowledge before I entered the field work were discussed. In the next chapter, I use findings from the field work with school teachers, parents and students to elaborate on the discrepancy between research hypotheses and analysis results. Some findings support my previous possible explanations while other findings are different for those possible explanations or propose the possible joint effects of other policy or social factors. To avoid confusion, I present these well-integrated explanations of the quantitative results of this chapter at the end of the next chapter, which presents the qualitative findings of my field work.

Chapter Five. A Different Version of the Story to Explain the Changes

I. Introduction

According to the government documents relevant to the Quality-oriented Education (QOE) , in the beginning of Chapter Four I assumed that there should be more out-of- school time due to the academic-burden alleviation policy, more time for sport and art courses or extracurricular activities due to the curriculum and examination reform, and more teaching and learning activities to develop creativity and problem-solving skills due to the teaching methods reform during and after the implementation of QOE. These policies and reforms are the key measures of QOE and were hypothesised to benefit the privileged students more than the underprivileged ones. Therefore, in Chapter Four I assumed that, compared to the period before the implementation of QOE, family background factors would have greater effects on students’ admission opportunities to regular senior high school (RSHS) and university in the periods during and after the implementation of QOE. However, not all of these assumptions have been confirmed using statistical hypothesis-testing in Chapter Four and some results contradict the original assumptions. In addition, the effect directions of family background factors in the RSHS admission analysis are not consistent with the ones in the university admission analysis. Even in the same admission analysis, the effect directions of family background factors are not consistent with each other. The above discrepancies need to be explained with the help of qualitative findings in this chapter.

If the original assumption is theoretically correct, i.e. that privileged students could benefit more than underprivileged students from the QOE policy in RSHS and university admissions, the above findings could be explained by two factors: (1) the QOE policy has not been fully implemented in reality and (2) there are other concurrent policies and social changes interacting with the effects of the QOE policy. Some researchers have observed the conflict between the central government’s policies on QOE reform and local practices, noting that these could result from the almost unchanged examination- oriented admission system and the fact that preparing students for the admission

examinations remains the top priority for parents, students, teachers, schools and even local authorities (Liu and Dunne 2009). In addition, the past 40 years are the era of change from Mao’s era to the Reform and Opening-up era and from the planned economy to the market economy. The QOE reform has only been conducted in this era, and thus this political macro context and many new policies relevant to education occurred during the same period. For example, private education as part of the Reform and Opening-up after Mao’s era was officially legalised in 1995, approximately two years after the QOE reform commenced. The Reform and Opening-up and the market economy have also stimulated fast economic growth, an increase in family income, and a huge increase in disparities in Chinese society with the Gini Coefficient rising significantly1. These above and other factors should be considered when explaining the discrepancies between the hypotheses and the results in Chapter Four.

As I mentioned in the summary section of Chapter Four, some possible explanations for the failure to confirm some of the research hypotheses were established before I entered into the field work. These explanations are based on my previous knowledge and experiences. In this chapter, I will use evidence from my field work to support and elaborate on these explanations, which will be presented in the summary and discussion section of this chapter to associate the field work findings with the quantitative results in Chapter Four. First, I set out the research questions of this chapter: (8) What changes has the QOE policy brought in reality? (9) What is the difference among three different family background factors in their association with educational outcome? (10) What is the difference between the QOE policy designed for or implemented in the educational stage before RSHS admission and the one in RSHS before university admission? (11) Besides QOE, what other concurrent policies or social changes could be related to these changes? The second question will restrict the scope of the first and third questions since my research is focused on the relationship between family background factors and students’

1 See Li’s (2014b) review on the change in the Gini coefficient of China. 86

educational achievements. Since other concurrent policies or social changes take effects along with the implementation of QOE, they will be reported in the three sub-sections corresponding to the first three questions rather than be reported in a separate sub- section in the finding and analysis section. My field work includes interviews with parents, school teachers and students. The interview topics were not restricted to the above four questions; however, the results I present in this chapter will be focused on these four questions.

II. Data and Methods

The interview is the main method of data acquisition for the qualitative part of my thesis. However, when interviewees’ descriptions involve a government policy or a social phenomenon, I also refer to government documents and statistics, newspaper reports and other public information to support or contrast their descriptions. I conducted my field work in Nanjing Municipality1, the capital of Jiangsu Province in the east coast of China, for the reason of access convenience. All the schools involved in my field work are located in the central city area and are public schools. The major reason why I focused on public schools is that the majority of students in the compulsory education (primary school and junior high school (JHS)) and RSHS attend public school (more than 93% in 20092). More than three quarters of parent interviewees were recruited through two JHSs in the first-round recruitment. According to the professor who helped me contact the “gatekeepers” of these two schools, one school’s ranking of an average score in the senior high school admission examination (hereinafter SHSAE). SHSAE is usually upper-middle in Nanjing Municipality and the other one is middle. Similar to many other urban public JHSs, these two schools have a large student population (more than 1,000), and thus their school districts cover many residential blocks with different

1 One primary and one JHS teachers in another city were briefly consulted via online communication tools about a national educational QOE program policy namely “Sport and Art 2+1”, which I heard about after the field work. This implementation of the program could be implied when school teachers in Nanjing talked about their schools’ non-academic activities, but the name of the program was not directly mentioned by them. 2 The proportion is calculated by me from the data source of the National Bureau of Statistics of China (2016a, 2016b).

kinds of socioeconomic characteristics. The three schools from which the remaining five parents were recruited in the second round are also in the central city area with a middle or upper-middle academic ranking and a large student enrolment. To compare the difference in the implementation of QOE between three different stages of basic education, I also recruited teachers from one primary school, one JHS and one RSHS. The JHS where the teacher interviewees were recruited from is one of the two schools in the first-round recruitment of parent interviewees. This school and the RSHS belong to a school consortium. Since there is no formal admission examination for entering JHS, I relied on the vice-president of this primary school informing me that its teaching reputation was middle to upper-middle in that district of Nanjing Municipality. The RSHSs ranking of an average score in the university admission examination is usually an upper-middle in Nanjing Municipality.

I recruited parent interviewees whose children were attending JHS because they could provide information relevant to their children’s learning life in the previous primary school and the current JHS. Since almost half of JHS students are rejected from RSHS through the senior high school examination process1, my recruitment strategy enabled me to include the voices of students who might be not able to enter RSHS and their parents. Third, during my interviews with teachers in RSHS, I found the information provided was sufficient to explain the difference between the effects of family background factors on students’ admission opportunities into RSHS and university. Therefore, I did not conduct the third-round recruitment to involve parents of RSHS students. Finally, limitations of budget and time were critical factors in designing the recruitment strategy.

During the field work, I conducted two rounds of recruiting parent interviewees. The first round was with the help of one of my previous professors in Nanjing University who introduced me to the two JHS headmasters. With the headmasters’ consent and assistance, I requested five class advisers to distribute an interview invitation letter to their students’ parents with the help of these students’ delivery. These students were in

1 Before 2014, the rejection rate was even higher (see Chapter One, p.15) 88

the second year of JHS (Grade 8) and each class had approximately 50 students. The letter enquired about parents’ potential willingness to participate in an interview and it contained a short survey on parent’s occupation, education, a student’s extracurricular tutoring and other background information. As a result of this process, 39 parents expressed an interest in participating. Based on the occupation and education information obtained from the survey, I recruited 17 parents with different levels of occupational prestige and education. However, there appeared to be only three parents who were interviewed who might belong to the lower-middle socioeconomic group1. Therefore, during my field work in the primary school, I requested the dean of students, to invite a few parents from the lower-middle socioeconomic group to be interviewed. In this second-round recruitment, five more parents agreed to participate and their children were Grade 7 students from three different JHSs, which I have described above and three of them might belong to the lower-middle socioeconomic group. Before doing the field work, my original design was to divide parent interviewees into two (higher and lower) or three (higher, middle, and lower) groups different socioeconomic ; however, during interviewing I did not find clear stratification of these socioeconomic groups in parents’ behaviours with raising their children and views on the education system even though there could have been some obvious differences between parents with the highest occupational prestige and those with the lowest occupational prestige. if I could recruit them. Therefore, I will not present parent interviewee’s background information by classifying them into three groups.

Parent interviewees were born between 1968 and 1979 and included five fathers and seventeen mothers. After parents were interviewed, seven student participants (three boys and four girls) with different levels of academic performance (i.e., the total examination score of last semester) and different parental socioeconomic backgrounds were recruited on a voluntary basis and with their parents’ consent. In addition, three primary school teachers, three JHS teachers and two RSHS teachers were interviewed, among which one junior and one RSHS teachers were also parents who were

1 The criteria of the lower-middle socioeconomic group in my field work of 2015 were that the annul family income was below CN¥ 100,000 (approximately AU$20,000, if they agreed to tell me), both the student’s father’s and mother’s education were below the bachelor’s degree and the father’s occupation was not white-collar. 89

interviewed. Table 5.1 reports the interviewee number of each group and related information.

Table 5.1 Interviewee Number of Each Group and Related Information Role Number Notes Parent 22 JHS students’ parents, two parents working as a teacher student 7 Junior high school 3 Primary school Teacher 3# Junior high school 2# Regular senior high school Note. # one teacher interview in this group was also interviewed as a JHS student’s parent and included in the number of the parent interviewee group.

The majority of interviews were one-on-one except one interview with a couple and one interview with a father and his son. The interviews were conducted by me in Chinese from April to June 2015. The majority of interviews with parents and teachers lasted approximately one hour and the interviews with students lasted approximately thirty minutes. The interviews were semi-structured (see Appendix 3 for a list of interview topics) and I ensured that parents felt free to speak on topics beyond those (see the research in the introduction of this chapter) I had identified in advance. For the parent interviewees, I mainly asked parents to compare their own at-school and out-of-school life when they were in primary and JHS with their children’s, which included learning and entertainment activities and how parents were involved in children’s education. For student interviewees, I asked them to describe their at-school and out-of-school life to confirm what the parents described. Since I have not found any contradiction between students and parents’ descriptions, I mainly report parents’ descriptions. The interviews with teachers were more professional. I directly asked them to compare the difference between the QOE and the Examination-oriented Education (EOE), especially in teaching and learning methods, curriculum and evaluation methods, and how the QOE policies have been implemented in reality.

Since the interviews were based on a set of specific questions, the analysis started with a clear coding framework; however, I was very open to including new categoriesto fully explore the difference between the policy in the documents and the practice in reality.

The coding was conducted by me using NVivo software (QSR 2016). The findings presented in this chapter directly follow the sequence of four research questions in the introduction of this chapter.

III. Findings and Analysis

What changes has QOE caused in reality? As I discussed in Chapter One and above, I am focused on the changes that QOE has caused in reality and that were supposed to be relevant to the educational inequality between different socioeconomic groups. Therefore, in this section I will report findings about changes (or non-changes) in out-of-school time, teaching and learning methods, non-academic courses and activities at school and admission examinations, which are separately related to four key elements of QOE policy, namely the policy of alleviating students’ academic burden, teaching and learning method reform, curriculum reform and evaluation method reform.

Out-of-school (extracurricular) time Based on information gained from parents and school teachers, out-of-school time has increased compared to their own childhood, namely the period before QOE. Out-of- school time includes the evening on weekdays and the weekends. With respect to the evening on weekdays, parental interviewees expressed a range of different views. For example, Yang’s mother and Yu’s mother claimed the at-school time has been shortened. Yu’s mother: It has been shortened. I've just been talking to my friend this morning. Everyone has a shorter (at-school time). Except we (her child) have a lot of extracurricular tutoring outside of school now, there is no any other change. When we were children, it was the school teachers who were in charge of our study. Put it (teaching and learning) into school. We needed to continue lessons long after 6 pm. Now classes are over very early. Students are not allowed to attend extracurricular tutorials organised by school.

Yang’s mother even remembered she had to attend an evening lesson till 9 pm when she was in JHS. However, some parents had not been required to attend evening lessons, and thus claimed “there was no significant differences” (Jian’s father). The at-school time in parents’ childhood seemed dependent on which city the parent interviewee grew up in or which school they attended. More precisely, it depended on whether the then local educational authorities or school masters attached enough importance to the admission rate ranking of their district or school. However, in the QOE era “the admission rate ranking is still important” (Teacher Mrs Qian). Therefore, the question is whether the academic burden alleviation policy has been strictly implemented by the local educational authorities or obeyed by the school headmasters. The trend is the inspection of the implementation of alleviation policy from the superior educational authorise is “stricter and stricter” (Teacher Mrs Wang). According to my interviewees, the five JHSs in my research field work had no evening lesson in recent years.

Parents and teachers unanimously agreed that the at-school time at weekends has significantly decreased. They advised that their own children or students did not go to school at weekends nowadays, which is different from when they were in primary and JHS. Before the field work (or in Chapter Four) I assumed the academic burden- alleviation policy could be the main cause of this difference; however, the real cause is a little more complicated than that. As Teacher Mrs Wang said, even though a few JHSs still asked their students to attend a half-day class on Saturday, the majority of her district’s schools complied with the “stricter and stricter” prohibition on school- organised lessons at weekends from the local educational authorities, which indicates the QOE academic burden alleviation policy did work more or less. However, the more important factor is the labour policy of shortening weekly working hours from 48 to 44 hours and then to 40 hours per week, which were separately released by the central government in 1994 and 1995 (State Council 1994, 1995). These two years were almost the same time as the central educational authority released the academic burden alleviation policy. This new factor of labour policy change had not been considered before the field work. As part of the labour force, school teachers were subject to this policy change and without teachers working at weekends students “benefited” as well. When parent interviewees, for example Xiang’s mother, were school aged, they

attended school every Saturday (before 1994) or every other Saturday (between 1994 and 1995). Before QOE, Saturday school days were regarded as a normal part of the school week and almost every student had to attend them; by contrast, in the QOE era there are only illegal extracurricular tutorials at a few schools on weekends. Therefore, on average the out-of-school time has increased in the QOE era. It should be noted that even though the labour policy change made a big contribution to the above increase, without the strict implementation of the academic burden alleviation policy many more school headmasters would arrange illegal extracurricular tutorials at weekends as Teacher Mrs Wang described.

In Chapter Four I stated that the reason why I wanted to find out whether students’ out- of-school time had increased after the introduction of QOE was because the extent of parent socioeconomic factors’ influence is supposed to be associated with how much out-of-school time is in their control. If they control their children’s out-of-school time, parents can arrange extra activities such as the commercial extracurricular tutoring (hereinafter CET) for their children. However, if the homework load assigned by school teachers increased along with the increase of out-of-school time, the parent-controlled time might not increase. I reviewed a series of regulation policies on student’s homework load limitation in primary and JHS issued by the Ministry of Education (or the State Education Commission, another name of the Ministry of Educational in 1985-1998) (Ministry of Education 1983, 2000; State Education Commission 1994, 1993, 1988) and discovered there was no obvious difference in homework load limitation before and after 1993, the year when QOE was launched. The limitation is 1.5 hours in JHS and 1 hour or less in senior grades of primary school while in early grades of primary school there should be no written homework. Therefore, under the stricter and stricter inspection of the implementation of alleviation policy in the QOE era, students’ extracurricular time for homework at least should not increase compared to the EOE era. During the interview, Shan’s father and Dan’s mother still complained that their children had too much homework to complete, but student Yu and student Lin told me they could finish their homework assigned by school teachers very quickly (in about half an hour for student Yu). Even though they needed to work on the assignments from the CETs, that is a different issue. Therefore, since the time for homework did not increase, the

“net” out-of-school time increased during the QOE era with the help of policies including no school at weekends.

Teaching and learning methods in textbook and in practice In terms of teaching and learning method reforms, QOE policy requires teachers to inspire students and help students learn by exploration, in an active engaged manner, which is opposite to the EOE teaching methods that required students to learn by rote and repetition (Jin and Tang 2004: 65). However, according to teachers I interviewed, these new QOE teaching and learning methods were only employed occasionally, particularly in “public lectures” that are viewed and “learned” by other schools’ teachers or inspected by the superior educational officials. Teacher Mrs Zhao explained why the QOE teaching and learning methods were not really feasible. Teacher Mrs Zhao: With respect to the teaching and learning by exploration, if the overall ability level of children in the class is relatively high, in that case the implementation (of new methods) may be very smooth.... But if the overall ability of students in the class is relatively weak and if we need to slowly develop them from the outset to achieve this level of competence and then use this type of inquiry-based iterative teaching and learning in class, it will be in fact very difficult. We are not likely to complete teaching tasks. ... The class hours are very limited. We have the senior high school admission examination. The skills at answering examination questions certainly need being taught to them. Because those skills need being practiced repeatedly ... Of course, these (new) methods have many advantages, but the biggest problem is their efficiency is not high.

As stated by the interviewed school teachers, the admission examinations in the QOE era were still in favour of the old EOE teaching and learning methods of requiring students to learn by repeatedly training rather than the new QOE methods of encouraging students to learn by exploration. Since the examination score was almost the only admission requirement, teachers and students had to keep using the old but

efficient methods to obtain a high admission examination score regardless of the QOE teaching and learning method reforms. Teacher Mrs Zhao concluded that “in reality there is no big difference in teaching methods and student’s learning methods”, but she also added “if there is a difference, I think the burden on teachers and students has been heavier and heavier.” As a school teacher, Nan’s mother described a technical reason why the burden had been heavier: Nan’s mother: The junior high school began to change the textbook around 2000, 2001. The ideology of (new textbooks) is that the teaching materials can help students continue to discover and explore. (They) use this kind of student self-learning perspective to write teaching materials. So, the knowledge structures in the (new) textbooks are neither as clear as before, nor given directly. ... During the lecturing and the reviewing the teacher certainly will give them (students) a knowledge structure, but (writing) it (down in the textbook by students themselves) depends on that students listen attentively and the teacher teaches it well.

The year 2001 Nan’s mother mentioned is the beginning year of the teaching and learning method reform in the current fifth round curriculum reform. To survive the almost unchanged admission examination, teachers and students have to spend some at-school learning time dealing with the desynchronisation between the new textbooks reflecting the new ideas of QOE teaching and learning methodology and the real teaching and learning activities. However, the at-school learning time is limited, therefore some students turn to extra assistance outside the school.

A few more non-academic Courses and Activities at school The fourth (1992–2000) and fifth (since 2001) rounds of curriculum reform are closely relevant to QOE policy and gave prominence to student all-round development of morality, intelligence, physique and aesthetics (Xie, Ma and Zhang 2013) rather than just focusing on the development of intelligence (academic courses) as the old EOE system did. Parents talked more about the changes in education of physique and aesthetics than about changes to moral education. In the previous chapters, I proposed that the new

attention to a student’s qualities of physique and aesthetics could increase the inequality of admission opportunities between different socioeconomic groups in urban areas and the two conditions of the foregoing change were (1) sport and art subjects were included in the admission evaluation and (2) in order to achieve a high performance in sport and art subjects, additional family investment, for example, CET, was required. Based on results from my field work, parents suggested their children’s physical and art education had been paid greater attention by the school compared to their own school experience or at least had not been paid less attention. However, Yu’s mother claimed that the lesson time for physical education had not been misappropriated for academic subjects any longer because physical education has been included in the SHSAE while the art courses have not. As a school teacher, Nan’s mother confirmed the different treatments to these two different courses, which was related to a technological cause: Nan’s mother: In our school, four classes1 take the PE lesson together. Male and female students take different sport subjects. So, it is possible that the four classes are taught by eight teachers at the same time. You can choose your own favourite sports subject. ...You (an academic subject teacher) cannot ask all the eight teachers to give their teaching time to you… that is, eight categories, eight types of students.

The arrangement of multiple physical electives is not an innovation by Nan’s mother’s school. It is consistent with the national “sport, art 2+1 program”2 to develop “students’ different potentials and strong points”, which is one of the key principles of QOE policy. The schools that my interviewees worked at or interviewees’ child attended usually organised different kinds of sports and art activities such as sports meetings and art festivals. However, the more at-school time for sports and art did not match parents’

1 There are usually several classes in one grade in Chinese schools. 2 This is a nationwide program in accordance with the order from the MOE named “Sport, Art 2+1 Program” as a sub-policy to implement the quality-oriented education (GOME 2011). This program requires primary and JHSs to organise different kinds of courses and activities to make every student master at least two sports and one art technique. The order document contains an appendix listing approximately 60 sports and 20 art techniques. 96

investment in their children’s extracurricular activities. Many parents I interviewed sent their children to attend extracurricular sport and/or art tutorials in the early grades of primary school, however, when their children approached or entered JHS, they only or mainly attended tutorials in academic subjects. I can understand that students drop the art tutorials because art subjects are not included in the SHSAE, but why do they drop sport tutorials as well when this is included in SHSAE?

As I described in Chapter One, the SHSAE is organised by the city or county-level education authorities but is still regulated by the directives of the Ministry of Education in principle. Therefore, I took Nanjing Municipality’s SHSAE as an example. In 2015, the full score obtainable for the physical education examination was 40, which is only approximately 6% of the full SHSAE score and, more importantly, the average score of the physical education examination was 39.4 and more than 80% of students achieved the full score (Qian 2016). Compared to other highly competitive academic subject examinations in the SHSAE such as Mathematics, Chinese, English, Physics and Chemistry, an official of the Department of Education of Nanjing Municipality said bluntly that “the physical examination in the SHSA is an examination for inspiration and policy rather than for selection” (Qian 2016). In addition, the university admission examination even excludes the examination of physique entirely, not to mention the examination of art1. Therefore, I have not found convincing evidence that the increased attention and investment to physical and art education is directly associated with the change in the inequality of admission opportunities between different socioeconomic groups in the urban areas.

Almost unchanged examination-oriented admission and its general consequence The above presentation and analysis of the field work results focused on the changes that the QOE policy created. It shows that, at least in some instances, the changes were more limited than anticipated by the policy. Further, the examination-oriented senior high school and university admission approach has remained almost unchanged. In

1 The policy of giving bonus points to students with art, sport and/or other “strong points” in the UAS has existed since 1984 (Li and Yang 2011), which is before the implementation of QOE, and thus not considered a factor in the change of social reproduction of educational inequality in this research. 97

Chapter One, I mentioned the new “autonomous university enrolment” (“zi zhu zhao sheng”, 自主招生) as a supplement to the regular national university admission examination. The autonomous university enrolment was introduced in the name of QOE in 2003 and usually utilise not only the traditional examination approach, but also personal statement, interviewing and other supporting documents as the admission evaluation approaches. However, only a few key universities are allowed to utilise the autonomous university enrolment system to recruit no more than 5% of each university’s annual newly enrolled undergraduate students and more importantly, those 5% students still need to take the normal university admission examination and at least get the minimum gate scores for the key university category1 (see Xun and Wang (2011) for a detailed description of AUE). Passing the autonomous university enrolment evaluation only gives those students some bonus points in the university admission examination, which only helps “good” students enter a better university or a better undergraduate programme rather than helps “bad” but “special” students enter university. The new introduction of autonomous university enrolment did not affect student’s probability of entering university, which is the outcome variable in the analysis of Chapter Four and thus are not considered further by my research. Therefore, the finding is still that the examination-oriented senior high school and university admission approach has remained almost unchanged.

The evaluation system, especially the university admission approach is the “baton” to direct QOE reform and determine whether the QOE can really replace the old EOE (Wen and Liu 2014; Zheng 2002). The QOE reform has been in progress for approximately 20 years; however, as a critical part of the QOE reform the admission evaluation reform has progressed very slowly. Until 2014, the goal written in the new policy document Implementation Opinions of the State Council on Deepening Reforms to the Student Examination and Admission System (国务院关于深化考试招生制度改革的实施意见) (State Council 2014b) was still “changing the tendency of overemphasizing the admission examinations in education”.

1 In the admission after UAE, universities are classified into several categories and each category has an admission requirement for minimum gate scores. To get admitted into a university, a student need to meet both the university and the category’s requirements for minimum gate scores. 98

Unexpected consequence (CET) and its association with family background factors In a more specific level, the unchanged EOE admission examination combined with the changes caused by the QOE reform resulted in an unexpected by-product, namely the popularity of CET. In the following section, I will describe what CET is and analyse why it has become popular. Then, I will describe and analyse its association with educational inequality.

What is commercial extracurricular tutoring? CET in my research broadly corresponds to Bray and Kwok’s definition of “private supplementary tutoring” (Bray and Kwok 2003). Bray and Kwok’s private supplementary tutoring is only concerned with academic subjects taught in mainstream schools, however, according to my field wok, CET in Nanjing is usually directly concerned with subjects included in a competitive admission examination. Therefore, those subjects do not include the subjects examined by “an examination for policy” such as the examination of physique I described above. Some CET institutions offer other subjects that are not taught in mainstream schools, such as Mathematical Olympiad, instrumental performance and the Go game. However, these subjects could be indirectly associated with the special admission process into primary school and JHS (compulsory education) 1 because when some parents try to send their children to a good private school or a better public school outside their own school district (school- choosing, “ze xiao”, 择校), their children’s performance in these non-academic subjects could be a bonus during the admission process. In that case, non-academic subject tutoring is also included in my discussion.

However, it should be noted that non-academic CET is usually more expensive than academic CET because non-academic CET usually employs expensive teaching forms such as the one-on-one, studio, and small-group tutoring forms while academic CET is in the form of a lecture in a 15-30 student classroom which means one tutor’s salary will

1 There are also students attending Mathematical or other Science Olympiads in RSHS, but unlike in primary school, only a few gifted RSHS students attend and are usually tutored by school teachers in order to represent the school in competition. 99

be paid by approximately two dozen students and those students do not need to buy or rent any expensive equipment like a piano for an instrumental performance tutorial. In addition, even though many students attend both academic and non-academic CETs in the early stage of education, they usually spend more or most time in academic CET especially when the SHSAE approaches according to parents in my filed work. Except for a few students (actually only one student in my research), most students usually attend this kind of academic CETs enrolling 15-35 students in each classroom, which capacity is only a little smaller than the classroom in a normal public school, rather than one-on- one academic CET.

CET has gradually become a large industry in the last thirty years. In 1985, the Decision of the Central Committee of the Communist Party of China on Reforming the Educational System (中共中央关于教育体制改革的决定) started to faintly acquiesce private education as part of the Reform and Opening-up after Mao’s era and in 1995 it was officially legalised into Education Law of the People's Republic of China (03-18-1995) (Yan 2007). According to the Sohu Research Institute of Education’s (2014) report, in the early developmental stage of the extracurricular tutorial industry there were few professional companies and this kind of service was usually provided by public school teachers, retired teachers and university students. Even the pioneers in this industry such as the founders of the Xueda (学大) company and the Tomorrow Advancing Life (好未来) company started from a family tutor to the CEO of an extracurricular tutorial company. For example, the Tomorrow Advancing Life became a formal company only in 2003. Therefore, CET was gradually becoming popular and it has kept increasing in popularity in the recent years (Wang 2013). According to this report, many companies have multiple branches in different cities. For example, the Xueda company has the most city branches (408 teaching centres in 77 cities). CET became a large industry with a CN¥ 650 billion (approximately AU$ 130 billion) output value in 2013 and formed a competitive market with a lot of companies participating in (Sohu Research Institute of Education 2014).

Besides the extracurricular tutoring provided by commercial companies, some schools provide low-cost tutorials, especially for senior students in JHS or students in RSHS who

100

are facing the admission examination and these tutorials are outside the normal or “legal” school hours and require almost every student to attend. I do not include these tutorials in the CET discussion of this study. Due to a similar reason stated in Bray and Kwok’s (2003) paper that CET in pre-primary and post-secondary levels is less vigorous and contains r different mechanics and issues, my research mainly discusses CET in primary and secondary schools.

Why CET could become popular: the connection with QOE policy The popularity of CET is a consensus among my parent and teacher interviewees. According to Jia’s mother who is a JHS teacher, approximately 80% of the students she taught were attending CET. This statement is consistent with Yang’s (2016) nationwide research report described in Chapter One (p. 4). Many previous studies have proposed different explanations for the popularity of CET in different countries (see Chapter Two for a review, pp.40-42). My research is focused on the factors relevant to the QOE policy. In the above section, I presented the changes as well as the unchanged admission system in the QOE reform. From the results of my field work, it was obvious that parents hope that CET can help their children survive in the changes caused by QOE reform and achieve a better score in the unchanged senior high school and university admission examinations. It should be noted that the popularity of CET itself as an unexpected by- product of QOE is a reciprocal cause for its popularity, which could result in the bandwagon effect or change the at-school teaching and learning mechanism to force other students to attend. This bandwagon effect not only resulted from a simple peer pressure but also resulted from a worry that if they did not follow the fashion as other parents did, they would be blamed by their children if their children could not enter a good university or have a satisfactory job in the future. Dan’s mother: I can’t not follow (those parents to invest in children’s CET). I am afraid that in the future my child would say “why didn’t you send me to attend (CET)? We had the (financial) condition. You didn’t send me to attend and I am not studying well, not as well as others (who attended CET)”. We are afraid to face that situation. We don’t want our child to regret in the future and I don’t want to regret.

101

Besides CET itself, other factors for the popularity of CET, which are relevant to the QOE policy but not directly to the admission examinations, will be more detailedly presented and analysed in the following section.

Parents Are Busy First, I present the factors that are not directly related to the purpose of obtaining an advantage in admission examinations. When out-of-school time increases, it is not necessarily CET that is the only activity that helps student occupy their additional extracurricular time. Other activities could be parent-child, peer and individual activities. However, even though their children were approximately 13 years old, some parent interviewees did not want to leave them alone at home. One common reason was “few children have the learning initiative. And then in the extracurricular tutorials, she will be able to restrain herself more (not spend too much time on entertainments) ... If I left her alone at home on Saturday, she would start watching TV, would not do homework. Then I would rather send her to the extracurricular tutorials” (Si’s mother). Although the reason was still related to obtaining academic achievement, Si’s mother did not expect her daughter to spend all her extracurricular time studying. Actually, on Sunday Si’s mother and father usually accompanied Si to some recreational activity. Even though the reason Si’s mother proposed is still relevant to learning, the more important question is why Si’s mother or father could not personally supervise their children doing homework on Saturday. The reason is that both of them are required to work on Saturdays. A similar reason for many parents sending their child to evening tutorials is that they have to work overtime. As I mentioned before, in 1995 Chinese labour policy changed the legal working time from 48 hours and 6 days a week to 40 hours and 5 days a week. Si’s father and mother’s occupations were just normal middle-level or senior white-collar workers in normal business companies. Was there a special reason for their long-term working overtime? According to the “report on China Labour-force Dynamic Survey (2013)” (CSS 2013), over 38% of the employees in China worked overtime in the previous month of surveying time. According to its 2015 report (Cai 2015), the overtime working percentage in corporations with foreign capital was the highest, 52.66% and the average weekly work time length was 41.5 hours in the previous month. The percentage was 39.39% in state-owned enterprises and 37.52% in government sectors. Therefore, with respect to Si’s father and mother there was no

102

special reason for their frequently working overtime, it is just a normal social phenomenon encountered by many working parents. In addition, the World Bank World Development Indicators Database presents that Chinese female labour force participation rate was approximately 75.2% in 2010, which is even higher than OECD counties’ average rate of 65.4% (as cited in NBSC 2013). Therefore, for many two-income families, the CET institution becomes a suitable care centre for their children in the increased out-of-school time.

Lack of Peer Activities If parent-child activities are not enough to absorb the additional out-of-school time, is peer activity a good alternative? Peers could not only be playmates, but also be study-mates. Even though in the one-child policy period most students could not “have an elder sibling to tutor them” (Lin’s father) as their parents could have, the study-mates from the same class or neighbours for example should have been able to replace the role of siblings to help each other in study. However, according to my interviews with parents and their children, students had very few playmates or study- mates in their extracurricular time, other than online playmates. Chang’s father described that his son “only when finishing the examinations, they (his son and his son’s classmates) would occasionally go out together, maybe to see a movie, or when a classmate held a birthday party, they would go together”. The reasons for this phenomenon are complicated and interesting. However, to remain focused on the key arguments of the thesis, I will only briefly present the factors for this phenomenon that arose from analysing my interviewees’ responses. First of all, the lack of playmates and the popularity of CET form a reciprocal causation relationship. Yu’s mother: All the other children went to that place (commercial extracurricular tutorials). Most parents sent their children there. I said how low- grade students could understand the Go game, chess, painting… these kinds of things. The major function of the extracurricular tutorials is (to provide a place where) parent can communicate, children can play together… There are not too many parents holding the idea that how well my child must learn these.

103

At school, students have classmates as playmates; however, during extracurricular time students may find that their playmates need to attend different tutorials and it could be very hard to find a timeslot when all or even both of them are available (Ran’s mother).

Besides the above factor, interviewees also mentioned other possible factors for the low amount or lack of peer activities compared to the parents’ own childhood, which include: (1) one-child policy1 results in less peers in general (Can’s mother), no sibling (Lin’s father), weakening of peer communication ability (Teacher Mr Wu) and parents’ overprotection (Yan’s mother); (2) the collapse of the “dan wei” (work unit, 单位) system turns the acquaintance community into the stranger society and parents do not feel relaxed enough to allow their children to play with strange children (Can’ mother and Yu’s mother); (3) even if parents allowed it, the school-choosing phenomenon could result in kids from the same neighbourhood enter different schools and then they are unable to form a close friendship (Can’s mother); and (4) in the digital era, there is more attractive entertainment such as online video games that allow children to play alone but are not regarded as a healthy activity by parents (Yu’s mother and Teacher Mr Wu). For all of the above reasons or some of them, many parents preferred sending their children to attend CET and/or had to do so.

Learning sport/art needs a professional tutor and helps make attending CET accustomed Professional Sport or Art Tutorial In terms of the contribution of increasing attention to developing students’ strong points and potential in sport and art as a reason for the popularity of CET, it is easy to understand that the normal school courses are only able to offer basic knowledge and training in sports and art. For example, Jun’s mother thought her son had a strong potential in painting and planned to send him to a professional studio to learn because the school’s fine arts teacher was not a professional painter. Even if the school teacher was good at painting, she or he could not spend too much time on one student. I have not found direct evidence or research conclusion that

1 Even though my quantitative and qualitative research did not take account of the effects resulting from the changes in the one-child policy since the corresponding birth year of all these students in my research is in the effective period of one-child policy (1979 -2013), some parent interviewees were born before the one-child policy period and thus mentioned this factor. 104

the undersupply of good quality art and sports teaching in normal public schools is one of the reasons why the QOE policy encourages the development of private education as a supplement to public education, however, the relevant policy documents articulate encouraging both the development of students’ qualities of physique and aesthetics and the development of private education. Therefore, when the QOE policy attached more importance to students’ art and sport qualities, especially promising to offer some admission bonuses to students with an outstanding potential in art or sports, unsurprisingly the art and sports tutorial became popular. Even if their students could not achieve an admission bonus, “learning sport or art is better than watching TV or playing online video games at home” (Yu’s mother). Therefore, the art and sports tutorials became popular.

Even though most students could not achieve an admission bonus and the RSHS and university admission is still academic examination-oriented, in the early schooling stage the RSHS admission is still far away and students have lots of extracurricular time in the QOE era and thus parents usually think “learning sport or art is better than watching TV or playing online video games at home” (Yu’s mother). Therefore, the art and sport tutorials become popular. When parents discover that their children are not outstanding enough in sport or art to achieve an admission bonus and more importantly, the transition admission is arriving, it is easy for parents just to switch their children’s sport or art tutorials to academic tutorials since “attending tutorial has been part of children’s everyday life” (Si’s mother).

Better academic performance For many parents, especially the underprivileged parents, the most significant reason for sending their children to CET is to improve their children’s academic performance or maintain their edge. It is understandable that students’ extracurricular learning is more efficient with the help of professional tutors and in the highly competitive atmosphere of the large tutoring class than under their parents’ supervision or by themselves. However, what is the contribution of the QOE policy to pushing students to attend CET if they want to improve their academic performance and survive admission examinations? First of all, I return to the change in learning and teaching methods at school. As I presented and analysed above, many teachers and

105

students have to spend more energy and time in adopting the new QOE textbook and other teaching and learning arrangements to the old but more efficient teaching and learning methods to achieve a good score in the admission examination because the admission examination is in favour of the old methods. However, this adaption makes the tight teaching and learning time at school even less and some students may find that the adaption makes their learning not as smooth a process. Therefore, CET steps in with its own straightforward and well-structured handouts containing summaries of knowledge points and examination-answering skills. Teacher Mrs Zhao directly related the popularity of CET to the fancy but unpractical QOE teaching and learning methods: Teacher Mrs Zhao: According to senior teachers, in their time the teaching methods were the most straightforward, simplest one. Almost all the teaching and learning activities were school-based. There was no extracurricular tutorial institution. Those children would not want to attend extracurricular tutoring because the learning at school was enough for them to deal with admission examinations. And of course, people of that time did not have that awareness (of sending their children to extracurricular tutoring), not have that idea. But when the school education gave children more freedom to explore (which is opposite to the previous teacher-dominated lecturing) and their academic performance could not improve, they naturally had to find another way.

To encourage teachers and students to employ the new teaching and learning methods and ultimately to improve students’ ability to solve new problems, QOE policy has tried to change some old question styles in the admission examination, which were criticised for encouraging learning by rote and repeatedly training and stifling creativity. Teacher Mrs Zhao described what the new style of examination questions looked like: Teacher Mrs Zhao: For example, there are some knowledge points which will be taught in RSHS but are related to some knowledge points taught in JHS. In order to examine students’ ability, these RSHS knowledge points will appear in JHS examination questions as

106

additional information, and thus JHS students have to answer some RSHS questions.

Teacher Mrs Zhao acknowledged these questions were designed to select “good” students into key RSHSs; however, few parents want to admit that their children are “not good” and just ask their children to ignore these questions in examinations, which means giving up the opportunity of entering a good RSHS and then entering a university. Teacher Mrs Zhao further described how these students could address this problem. Teacher Mrs Zhao: Luckily, for about fifty or sixty percent of the students, no matter whether they really understand (the knowledge) or not, they may make the correct answer mechanically through repeatedly listening to teachers’ explanations, repeatedly exercising, repeatedly strengthening (examination question-solving skills). Or they encountered and answer similar questions more than several times, so they could remember how to answer and just apply it mechanically. In that case, sometimes they might get part of exam scores.

However, as I described above the teaching and learning time at school has decreased leading to less at-school time in total and more sport and art activities and the less efficient new teaching methods have made the teaching and learning time even lower. According to Teacher Mrs Zhao’s description, school teachers usually have no time to repeat the teaching content several times or arrange repetitive exercises in examination question-solving skills in class. Therefore, “now a lot of parents choose sending their children to extracurricular tutorials. Those tutors will explain the content taught by school teachers again. They may help children deepen the impression (of the knowledge) or propose a different (learning) method to see which method children can accept more easily.” (Teacher Mrs Zhao)

Confronting the lower amount of teaching and exercising time and the new style of examination questions, Teachers Mr Wu and Mrs Zhao acknowledged CET as useful or necessary for many students and some parent interviewees also mentioned, “school

107

teachers encouraged their children to attend the CET” (Min’s mother, Xi’s mother and Yi’s father). However, with respect to this kind of encouragement, some parents had very negative speculation over school teachers’ motivation. I am not sure whether some parents misunderstood the reason why school teachers could not meet the needs of students in terms of the admission examination or whether some school teachers’ violating the regulations to participate in CET made parents suspicious. Xi’s mother: Lots of knowledge is not taught by school teachers in class or they only teach something simple. Then you cannot solve difficult (examination) questions. In fact, you even cannot answer some ordinary questions. Then they teach each other’s students at their own extracurricular tutorials. I think those teachers are terrible.

Even if school teachers did not benefit from the CET industry as part-time tutors, some parents still claimed “it (that lots of students attending extracurricular tutoring) is good for (school) teachers. If the academic performance of their students is too bad, school teachers will receive lots of pressure from different aspects.” (Yi’s father) However, Jia’s mother as a school teacher said “this is not for my own achievement, nor for our school’s glory. I just want that in the admission exam your opportunity will not be affected by the subject I teach.” Whatever motivation it involves, school teachers’ encouragement is one of the factors that parent interviewee mentioned or implied as being behind the popularity of CET.

It is interesting that the popularity of CET reciprocally changed the at-school teaching. CET not only helps student review what school teachers have taught but also previews what students will learn at school. Therefore, for many students, the normal learning procedure is “students learn the textbook content for the first time at extracurricular tutorials in the (summer or winter) vocation, then (in the normal semester) school teachers teach it again, and at the Saturday tutorials students learn it again and deeper.” (Jia’s mother as a JHS teacher) However, “because students attending extracurricular tutoring account for 80% (of the total), the teacher cannot get the correct feedback from students on which knowledge points should be explained carefully in the classroom teaching. In other words, what information the teacher receives is that students can

108

understand all the knowledge very smoothly.” (Jia’s mother) Therefore, if school teachers felt most students could understand what they taught including some different knowledge points, the teaching would be undertaken quickly. Those students who did not attend CET would be ignored and become worse off if their learning ability was not strong enough to understand the difficult knowledge points by themselves easily without detailed elaboration. This is one of the reasons why Teacher Mr Wu acknowledged that if their children’s learning ability was not strong enough, parents had better send their children to CET, with this suggestion being given out of good intention.

Special admission in the compulsory education The above factors are mainly used to explain the popularity of CETs in preparing students for ordinary school examinations and ordinary senior high school and university admission examinations. In addition, there are other CETs designed for non-government organised admission processes, especially for the school choosing process for entering primary school and JHS. A good primary school or a good JHS can help students achieve a better examination result in the RSHS and university admissions. As I described above, the senior high school and university admissions are still examination oriented. I take the transition from primary school and JHS as an example since the school choosing phenomenon is more popular for this transition than for the transition from kindergarten to primary school. It should be noted that most students still enter JHS via the normal admission approach (see Chapter One, “Introduction to Chinese Education System”) but owing to a similar reason for sending their children to attend art or sport CET, in my filed work parents knew entering “Nan Wai” (南外, Nanjing Foreign Language School, a top junior high school) was almost impossible for their children but “learning something more was not a bad thing” (Yu’s mother).

During the EOE era, teacher Mrs Wu described “you had a junior high school to enter without taking the admission examination and another one to apply for through taking the competitive admission. If you had the ability, you could enter a key junior high school”, which is also reflected in a local governmental review (Wang 1999). Since the educational authorities started the no-examination JHS admission policy as a measure to alleviate students’ academic burden in 1994 (State Education Commission 1994),

109

these top JHSs, which parents want their children to enter but are not in their own school district, are usually allowed to recruit a small proportion of students into the so- called “QOE plot classes” regardless of these student’s original school districts. However, they are prohibited from organising any normal admission examination relevant to knowledge in the primary school textbooks, which is a specific measure to stifle the resurgence of EOE and to promote QOE. Some foreign language schools are only allowed to organise foreign language admission tests, but these tests largely exceed the learning requirement of a primary school’s English course. However, for other non-foreign language schools, how can they decide whether to admit a student or not? Besides the “connection” and “secretive donation” on occasion (Shan’s mother and Ran’s mother), many schools use the “comprehensive evaluation”: Teacher Mrs Zhao: I have seen some primary school students’ CV used in the school-choosing process… He (a student) took a thick stack of documents, thicker than my supporting documents for job application, really without exaggeration. Then I simply scanned them. In addition to some normal transcripts, there was a thick stack of various certificates of merit such as Mathematical Olympiad competition, English competition… and even some interest-oriented classes, such as calligraphy class, dancing class, and English speech competition, all kinds of competitions… from the first grade to the sixth grade they constantly participated in these competitions.

As I discussed above, since art and sport achievements are not the crucial part of the senior high school examination, why did these JHSs still care about these two qualities? Yu’s mother exposed the logic behind the above admission method as “you have not spent all the time on the academic subjects to achieve high scores in them. You can learn this (academic subject) very well and meantime learn that (non-academic subject) very well, which demonstrates you have a very strong learning ability”. Even though the opportunity to enter a QOE pilot class is very limited, parents still do not give up on the chance and having a sport or art skill could be “very helpful in their children’s further social life” (Si’s mother). Therefore, these specialised CET became relatively popular.

110

However, according to my parent and teacher interviewees, the most important subjects for admission are the Mathematical Olympiad and New Academic Benchmark English Level Test (Teachers Mrs Zhou and Mrs Zhao and Ran’s mother). In view of the popularity of CET in the Mathematical Olympiad and foreign languages, the Ministry of Education issued an administrative order prohibiting public schools from using any competition certificate or certificate of level test in any subjects as admission supporting documents (Ministry of Education 2014). However, some top JHSs are publicly owned but commercially operated (see Zhou and Mao (2007) for an introduction of this type of schools), and thus they have been able to escape from the regulation of this order. In addition, even though some local educational authorities completely stopped any academic subject competition or level test organised by official organisations, it is very hard to prohibit unofficial competitions and tests. A rather interesting phenomenon is that these competitions and tests are organised by some large CET companies (Jiao 2015), which was also mentioned by teacher Mrs Zhou and Jian’s mother. Ran’s mother claimed that particular schools recognised the competition certificates issued by specific CET companies owing to a certain exchange of interests. Since their results became the admission criteria of many top JHSs, these competitions and tests look like quasi-JHS admission examination and a lot of students have to attend these CET because you cannot learn the knowledge and skills, which these competitions and tests involve, from normal school classes. The official JHS admission examination was cancelled for promoting QOE; however, the quasi-JHS admission examination organised by CET companies stepped in and many students still have to endure the EOE learning childhood.

Different associations between different family background factors and CET In Chapter Four, I analysed the association between family background and students’ admission opportunity into RSHS and university. The three elements of a family’s socioeconomic background I focused on were cultural capital, family recourses, and social status (prestige). In the above sections, I presented and analysed the association between attending CET and students’ academic performance in normal learning life and admission. If I assume CET is a mediator between a family’s socioeconomic background

111

and a student’s admission opportunity, I should present and analyse the association between a family’s background and attending CET.

Cultural Capital In terms of cultural capital, when I requested one of my field work gatekeepers 1 , who was also a junior school teacher with some administrative responsibility, to help me recruit parents to interview, she reminded me that it might be very difficult to recruit parents from low socioeconomic backgrounds because she thought they were of low “su zhi” (素质, quality, see the literature review) and did not have a high educational level and thus might be not interested in my topic. As she expected, in the total 25 parents I recruited, only 6 seemed to be from the lower-middle socioeconomic group according to their occupations and annual income and none of them belonged to the lowest socioeconomic group. Therefore, I assume my recruitment result more or less indicates parents with low cultural capital (educational levels) have lower extents of willingness to send their children to CET and I will test this assumption in the next chapter. In this chapter, I will present another aspect of the association between parents’ cultural capital and their children’s CET: choosing what kind of CETs.

According to a recent report published by the Chinese Society of Education (2016), more than 80% of parent respondents strongly agree or agree that CET is an essential part of compulsory education. Therefore, the popularity of CET could result in that whether to attend CET has less effects on students’ academic achievement while what kind of tutorial parents choose for their children could be a more important factor. Besides financial capital, how to choose between different tutorials could be related to parent’s cultural capital and I present a few typical examples briefly. In Chapter Four, I assumed that the increase of out-of-school time could result in more parent-child interaction and thus the effects of parental cultural capital would increase. However, after sending their children to extracurricular tutoring, some parents did not think they had completely fulfilled their responsibility. Teacher Mrs Zhou only sent her daughter to an ordinary mathematical tutorial and a Mathematical Olympia tutorial. When her daughter was in the summer holidays before entering her senior year of primary school, Mrs Zhou heard

1 I did not formally interview this gatekeeper during my field work. 112

that if a student achieved the “excellence” grade in Level 5 of the New Academic Benchmark English Test, some top JHS could admit this student in advance. Even though working as a public primary school teacher in China meant that Mrs Zhou’s annual income might not be as much as a commercial white-collar worker having the same educational level, she was able to obtain some useful admission information not known to most other parents simply because she was in the educational field. After obtaining this information, Mrs Zhou immediately “found a one-on-one English tutor to cram her (Mrs Zhou’s daughter) for the Level 5 of New Academic Benchmark English Test. Luckily, after three months we succeed. We got the excellent grade. So, she got the stepping stone of English subject (to a top junior high school)”.

With respect to ordinary (not one-on-one or other special-formed) CET, finding a good tutor is important because even for the same tutorial subject in the same tutorial company there are usually several tutors and their quality of teaching can be rather uneven (CSE 2016). In addition, information for finding a good tutor is often scarce or unreliable. Shan was attending a tutorial organised by a reputed public-school teacher. This tutorial was not open to “outsiders” as public-school teachers are not allowed to participate in CET. Shan could attend because her mother was a research officer in an educational technology company and reached the teacher through a friend in the educational industry. Again, Shan’s family annual household income was in the second lowest 20% group but since Shan’s mother worked in the education industry, she had enough cultural capital (adding some social capital) to find an experienced tutor.

For many other parents, certain big tutorial companies allowed parents to be present at the tutorial and parents could decide to keep their children attending or switch to another tutor or company. Another important reason a company gave for the above policy is according to a third-party survey they commissioned more than 80 parent respondents wanted to be present to help their children review after their tutorial1. However, choosing a tutor or helping their children after the tutorial requires parents to

1 See the article in the company’s website (http://ssh.speiyou.com/z2014/2014/gxkt/pt.html). 113

have a high educational level. Without enough cultural capital, the positive effects of CET on children’s educational achievement could be rather weak.

Financial Resources Compared to cultural capital and social prestige, family’s resources, especially financial resources were mentioned more often by interviewees. All the parents I interviewed had sent their children to CET and obviously they could afford the associated fees. Even though some parent interviewees acknowledged that the tutorial fees were expensive for many Chinese families, they thought that most families could afford them because “parents could live frugally so children could have a better future. That’s what every parent believes” (Dan’s mother). Dan’s mother’s observation was supported by the report from the Chinese Society of Education that 31.6% of parental respondents claimed no matter how much it cost, they would still like to send their children to extracurricular tutoring and 26.6% of parents were willing to spend half of the family income on their child’s extracurricular tutoring (CSE 2016). In terms of the affordability of tutorial fees, the tutorial fees in my field work site (Nanjing Municipality) provide an example. According to the website of a popular CET company some interviewees’ children attended, the 2015 price for one normal tutorial of 100 minutes and 15–35 students was CN¥ 120 (approximately AU$ 24) and there were 7 lessons in a winter term, 15 in a spring term, 10 in a summer and 15 in an autumn term.1 If a student only took English and Mathematics tutorials, which are the most popular and important subjects, the total tutorial fees in 2015 was approximately CN¥ 11,400 (approximately AU$ 2,280). In 2015, the yearly average personal income of urban residents in Nanjing Municipality was CN¥ 46,104 (approximately AU$ 9,200) and CN¥ 22,542 (approximately AU$ 4,500) for the lowest 20% (BSNM 2016). Therefore, if the family had two parents and one child, the yearly tutorial fees for two subjects made up approximately 8%of the household income (of three people, not two adults), CN¥ 138,312 (approximately AU$ 27, 600) for an average household and about 17% of the household income, CN¥ 67,626 (approximately AU$ 13,525) for the lowest 20%. According to Yang’s mother, her annual household income was approximately CN¥ 100, 000 (approximately AU$ 20,

1 The information is retrieved from the company’s website http://snj.speiyou.com/e/20150415/552ddc5803334.shtml.

114

000), which is within the second lowest 20% group. Yang attended three tutorials including Mathematics, English and Chinese and Yang’s parents spent approximately CN¥ 70,000–80,000 (approximately AU$ 14,000–16,000) on their children. The expenditure included subsistence costs, tutorial fees, and other educational and entertainment fees, which is “except for the necessary cost of living, all the other money is spent on my child” (Yang’s mother). Since Yang attended a public JHS, the tuition, textbooks and exercise books were free and other fees for social activities (CN¥ 100 at most per semester in 2015), school uniforms, and at-school meals or boarding were paid if parents wanted to buy1. Therefore, the tutorial fees make up the majority of Yang’s parents educational expenditure.

The above description and analysis is with regard to the current period, namely the third QOE policy period (after 2000). How affordable was CET before and during the implementation of QOE policy? Unfortunately, I have not found any document on the price of CET during different periods. However, if the popularity indicates some extent of affordability, we can find some clues from parent interviewees’ recollection on the development process of CET. Lin’s father: When did it (CET) start being popular? I am not very sure. But I feel after the late 90s, after 2000, its development enters the peak age. It should be basically consistent with the curve of national economic. There are two aspects. (First,) because of the expansion of education, there were more university graduates. Many of them with teaching ability were accepted by extracurricular tutorial institutions. Second, along with the economic development the household income has increased. The family income could afford such a cost.

The timeline that Lin’s father mentioned is consistent with the above brief description on the development history of CET in this chapter (see the subsection titled “What is commercial extracurricular tutoring”). CET became a large industry and formed a competitive market with a lot of companies participating in, which could further lower

1 The information is retrieved from the official website of Qixia District Government http://xxgk.njqxq.gov.cn/qxq/qxqjyj/201503/t20150303_135319.html. 115

the unit price for one-hour CET. In the above description, even as a family in the second lowest 20% household income group, Yang’s mother said she and her husband could still afford Yang’s educational expenditure.

However, there are two points that should be noted in terms of the payment ability of lower-middle income families for CET. The first one is that Yang has no sibling. If Yang’s mother had another child, their household income would not be enough to afford raising two children at the current expenditure level on Yang. The one-child policy1 appears to be one of the key reasons mentioned by interviewees when I discussed the affordability of tutorial fees. Yu’s mother: You say it (CET) is expensive… in fact, for Chinese families now we only have one child, most of us only have one child. The majority of family expenditure is spent on child. Therefore, it is very possible that the families with more children have become worse off as a result of the popularity of CET. The second point is that CET Yang attended was in the form of a lecture in a 15–30 student classroom and its content was academic, more precisely, Mathematics and English, which means one tutor’s salary will be paid by about two dozen students and those students do not need to buy or rent any expensive equipment such as a piano for an instrumental performance tutorial. The average tutorial cost one student needs to pay could be relatively low. On the contrary, if Yang wanted to take a one-on-one family English tutorial like Ran, or a painting studio tutorial like Jun, it is very possible that his parents could not afford the fees. However, luckily, since lower socioeconomic families in urban China usually have only one child and the EOE admission system only requires students’ good performance in academic subjects, many of them could afford the current fees for academic CET.

Social Status (Prestige) As I mentioned in the above section, the current CET began with the form of private one-on-one family tutoring in the early 1990s rather than the 15–30 student tutorial similar to the classroom in a formal school. Hiring a private tutor into the household can trace its history back to ancient China. However, only the nobility or

1 As I mentioned before, the one-child policy started in 1979 and ended in 2013 and thus both the quantitative and qualitative research in this thesis are not affected by the changes to the one-child policy. 116

the big country landowners were able to hire a private tutor. Even for many rich businessmen it could be very hard to find an average private tutor because in ancient China the normal social status of a businessman without any official title was lower than an ordinary peasant, not to mention the private tutor usually owned a scholarly honour of official rank (“gong ming”, 功名) and was viewed a potential future imperial official. With the influence of this cultural tradition, it could be understood that when it resurged in the 1990s, private family tutoring had some “symbolic value” of a higher social class (Baudrillard 1998) and especially in an era with a strict “hu kou” (household registration) system it also had some “symbolic value” of the urban “hu kou”. When the family tutoring was in a non-academic subject, the “symbolic value” would be much more obvious: Yan’s mother: Because when I was young I lived in the rural area, I felt really envious when seeing a child playing an instrument. I thought only a rich family child could do that. I thought only an urban family child could do that. I feel now that you (referring to Yan) have an opportunity to learn an expertise; I think this kind of opportunity is certainly very good. At that time, an ordinary family hiring a private tutor was very odd and might incur some gossip from the neighbourhood and “if your child needed tutoring, you could ask your relative or friend’s kid to help” (Lin’s father).

Difference in QOE between compulsory education and RSHS Before entering my field work site, I learned that there are some differences in QOE between the compulsory education (primary and JHS) and RSHS, which is also supported by my field work. Of course, there are differences between primary and JHS or, more precisely, the early primary school and other grades in compulsory education and between the senior year and non-senior years. However, since the previous chapter only analysed the RSHS admission and university admission, here I focused on the differences between compulsory education and RSHS.

The difference in at-school time between compulsory education and RSHS is endorsed by the official educational documents. The central educational authority authorised the provincial educational authorities to regulate the at-school time but required them to

117

ensure that students in primary school should have more than 9 hours of sleep, at least 9 hours in JHS and at least 8 hours in senior high school (State Education Commission 1994). Therefore, the Department of Education in Jiangsu Province prescribed that the at-school hours should not exceed 6 hours in primary school, 7 hours in JHS and 8 hours in senior high school1. In terms of the homework load assigned by school teachers, the limitation is 1.5 hours in JHS and 1 hour or less in senior grades of primary school while in early grades of primary school there should be no written homework. However, the central educational authority authorised the provincial educational departments to make their own regulation on the limitation of RSHS homework load (State Education Commission 1994). The limitation in Jiangsu Province is two hours according to the same document regulating at-school hours above. The central educational authority did not explain why it did not make a uniform policy on the limitation of RSHS homework load. My speculation is because it knows that confronting the highly competitive university admission examination it is impractical to strictly limit the homework load in RSHS, especially in some provinces with a very low university admission rate or with a very small quota of top university enrolments. The evidence from the field work are consistent with the policy documents. According to a RSHS teacher Mrs Zheng, the out- of-school time in RSHS in the QOE era did not increase significantly compared to the EOE era, which is different from primary and JHS. The RSHS where Mrs Zheng works is one branch of a school consortium, which also includes a JHS. It is normal that different schools may have different attitudes toward implementing QOE policies, and thus it may not be accurate to compare one independent JHS with one independent RSHS. However, even in this school consortium its JHS branch does not require students at least in first and second year to attend the evening class while its RSHS does, not to mention the RSHS senior year students who also need to attend school on Saturday regardless of the ostensible prohibition from the educational authorities. These school organised tutorials are very common in RSHS and “schools usually claim these tutorials run on a voluntary basis but almost every student needs to attend” (Teacher Mrs Zheng). In summary, the

1 The information was retrieved from the website of a top RSHS in Jiangyin City, Jiangsu Province (http://www.jsjyyz.com/Item/1688.aspx) on 30 January 2017 because the regulation was issued too long ago (2009) to be found from a government website and the CNKI yearbook database on educational policies has no relevant item. 118

out-of-school time excluding the time for homework assigned by school teachers in compulsory education is significantly longer than in RSHS.

In terms of teaching and learning methods, even though the new curriculum reform in the QOE policy tried to change the old EOE teaching and learning methods, the almost unchanged university admission examination was still in favour of the old EOE methods and thus these proposed new QOE methods were not really applied in reality (Teacher Mrs Zheng), which is similar to what happened in compulsory education, especially in junior high school. The old EOE methods in RSHS may be more stubborn than in junior high school because the new curriculum reform in RSHS started latter than in primary and junior high school. The new curriculum reform in compulsory education started at several national pilot districts in 2001 and in the September semester of 2005 the first- year students in primary and junior high school across the country started to use the new curriculum (Yang and Zhu 2003). However, the new curriculum reform in RSHS started in four pilot provinces in 2004 and in the September semester of 2012 the RSHS freshmen in Guangxi Province as the last cohort used the new curriculum.

Since the at-school time in RSHS is used to prepare students for the university admission examination and is so limited in the view of school teachers that they require students to attend school in the evening and on Saturdays, there is “no more at-school time for more non-academic courses or extracurricular activities”, which is different from primary and JHS (Teacher Mrs Zheng and Nan’s mother as a RSHS teacher).

IV. Summary and Discussion

To explain the discrepancies between the research hypotheses and results of the changing pattern of social reproduction of educational inequality across three QOE policy periods in urban China found in Chapter Four, in this chapter, I further explored the mechanism of how QOE policy changes the association between students’ family background and their transition opportunities from JHS to RSHS and then from RSHS and university through drawing on interviews with parents, students and school teachers

119

and analysis of relevant policy documents. Before explaining the discrepancies found in Chapter Four, I first summarise what specific changes the QOE policy brought in reality and how QOE policy probably makes a major contribution to the popularity of CET, an unexpected consequence or by-product of QOE policy. Then, along with summarising how three family background factors are differently associated with CET, I propose a possible explanation for different trends in the associations between three different family background indicators and a student’s transition opportunity from JHS to RSHS (i.e., social reproduction of educational inequality). In addition, along with summarising the difference in the implementation of QOE policy and the difference in the situation of student attending CET between compulsory education and RSHS, I propose a possible explanation for the difference between the trend of social reproduction of educational inequality in terms of transition opportunity from JHS to RSHS and the one in terms of transition opportunity from RSHS to university. During the field work, I also paid attention to other policy and social changes that were concurrent with QOE. However, these “other” policy and social changes are usually associated with QOE policy or, its unexpected consequence, CET. For example, the change in the labour policy of daily and weekly working times facilitated the academic burden-alleviation measure of QOE policy to increase the student’s out-of-school time and the continuous increase in family income could contribute to the popularity of CET. Therefore, I will discuss the effects of these “other” policy and social changes on social reproduction of educational inequality along with QOE policy and CET. The exception is the sharp increase in university tuition fees as I described in the introduction to Chinese education in Chapter One, which I think has a special effect on the trend in the association between family resources and university admission opportunity and will be discussed separately. During proposing the explanations for those discrepancies, I draw the key hypotheses from the above possible explanations, which will be tested in Chapter Six with second-hand survey data.

According to the field work, the QOE policy in addition to the new labour policy and other social and economic changes, has increased students’ out-of-school time especially in the primary and JHS (Years 1 to 9). As noted previously, QOE aimed to alleviate students’ academic learning burden and to help them have all-around development including more sport and art qualities. In addition, the QOE policy

120

introduced new learning and teaching methods, which are reflected in the new textbooks. However, the examination-oriented admission system remained relatively unchanged and school teachers reported that they found the old methods were more efficient to prepare students for examinations. Under the curriculum reform of QOE, there have been a few more non-academic courses and activities at school even though the total at-school time has decreased. For many families, having their child or children entering a RSHS and then entering a university are still the most important ways of climbing the social ladder or keeping their current socioeconomic status. Due to the almost unchanged, fiercely competitive examination-oriented RSHS and university admission processes, it appears that the majority of the additional out-of-school time has been occupied by CET rather than leisure activities. The majority of CET is focused on academic subjects to compensate for the shortened school day and is to prepare students for RSHS and university admission examinations, which is different from my original assumption in Chapter Four that most CETs are in sport or art. This qualitative finding is also supported by the data from the China Education Panel Survey the China Education Panel Survey (CEPS [education], Wave 2013) that in the urban area approximately 36% of CET attendees only attended academic CETs, 36% attended both academic and non-academic CETs and 29% only attended non-academic CETs. Even CET in art, sports and other non-academic subjects is to prepare students for the special admission in the name of QOE; however, these special admissions only recruit a very small number of students and even before the implementation of QOE, the special recruitment or bonus points added to the total examination scores had existed (since 1984 for university admission) (Jiang 2006). Therefore, before other opposite evidences emerge, I assume the effects of special admission and the non-academic CET on the change of social reproduction of educational inequality are limited and focus on the academic CET.

In addition to the fact that the examination oriented admission has not changed as a result of QOE, the popularity of CET is associated with the following factors relevant to the changes brought by QOE: (1) for the extra out-of-school time other forms of activities such as parent-child activities and peer or sibling activities are not available; (2) the general school teaching is not specialised enough to develop students’ art or sports

121

strengths, which results in the emerging of art and sport CET for students in the early educational stage but indirectly promote the popularity of academic CET by devolving students’ habit of attending CET; (3) since the QOE teaching and learning methods are not fit for the old examination oriented admission, the old EOE teaching and learning methods are used in reality but students need extra help to adopt to the dissonance between the QOE textbooks and the EOE teaching and learning methods; (4) schools have less time to prepare students for admission examinations because under the QOE reform they have to shorten at-school time in general in addition to allocating more time for non-academic courses and activities; and (5) the popularity of extracurricular tutoring itself is a factor because if many students have previewed learning content in the tutorials, school teachers could expedite the at-school teaching progress, which will disadvantage students who do not attend CET, and thus force them to attend.

However, whether to attend CET or not is associated with family background factors, and thus on average attending CET may give more edge to students who can afford the CET fees, while students who cannot afford may be worse off in the competition for admission opportunities, especially considering that these students’ at-school learning time decreases and the at-school teaching proceeds faster because other students have done a lot of preparation during CET. In the beginning of Chapter Four, I assumed that substantial CETs focused on art and sport subjects and only the privileged family could afford those CETs. Therefore, I hypothesised that the effects of family background factors on students’ educational transition opportunities should linearly increase as the QOE reform proceeded. However, the fact that the majority of CETs focus on academic subjects has a significant implication for social reproduction of educational inequality. Academic CETs usually utilize the old EOE teaching form favoured by the unchanged EOE admission system, which is lecturing approximately two dozen students in a big classroom at the same time. Compared to the one-on-one, studio and small-group teaching forms which are often utilised in art and sport CETs, the teaching form utilised by academic CETs make each student’s share of a tutor’s salary smaller. In addition, different from many non-academic CETs, academic CETs do not require large expenditure on learning equipment such as expensive music instruments or sport gears. Along with the implementation of QOE, CET became industrialised and formed a

122

competitive market, which further lower the unit price for one-hour CET. Therefore, even though it could still be unfordable for the lowest socioeconomic families, the total expenditure on attending academic CET could be affordable for many lower-middle families with one child after cutting down living expenditure on non-education aspects owing to a strong educational aspiration, which is different from my original assumption. Therefore, the quantitative analysis results in Chapter Four showed a different pattern (see Table 4.4 and Table 4.6 in Chapter Four) from my original hypothesis that the effects of family background factors on students’ educational transition opportunities should linearly increase as the QOE reform proceeded. The following section will employ the findings from this qualitative result chapter to explain the discrepancies between the hypotheses and results in Chapter Four.

The popularity of CET can explain the change in the association between family background factors and student educational transition opportunity across the three QOE policy periods. The first basic assumption drawn from the field work is that attending CET can improve students’ academic performance, in other words, increase their educational transition opportunity. I first explain the social reproduction of educational inequality in terms of the transition opportunity from JHS to RSHS (see Table 4.4 in Chapter Four). In the compulsory education (primary school and JHS) families with more resources per child (i.e. those with fewer children) were able to afford the CET fees more easily and students from only-child families had greater transition opportunities during the second and third QOE policy periods as CET became more popular at the same time.

However, the above result is mainly based on the comparison between only-child students and sibling with one sibling. The changing pattern in the RSHS admission opportunities of students with more than one sibling is more complicated. This pattern is similar to the pattern of only-child students rather than the pattern of students with one child. One possible explanation is that under the strict one-child policy a family with three children or more is rather extraordinary, especially in the urban area. Some of these family could be poor or at least their children would be worse off because their family resources were largely diluted and the demand for family resources became

123

larger along with the implementation of QOE. However, other families could be rather rich because they could afford the large fine for violating the one-child policy1. Therefore, even though their family resources were diluted by multiple children as well, the fees for several children’s CETs were still affordable. If those rich families were the majority of these families with more than two children, it was not surprising that their children’s RSHS admission opportunities increased along with the implementation of QOE. However, since I have no suitable dataset to test the above explanation, my explanation mainly focusses on the changing patterns of only-child students and students with one sibling.

The popularity of CET was also consistent with the insignificant increase (in the second QOE period) and then decrease (in the third QOE period) of the third family background factor’s effects (father’s ISEI, i.e., family’s social status), which is explained by the qualitative findings that CET started with the “symbolic value” of the higher social status group (Baudrillard 1998) in the second period but in the third period it might lose such a value as CET became popular even among some lower-middle social status families with “ambitious” expectations of social mobility. The trend of effects of family’s cultural capital is similar to family’s social status but its changes are not significant. According to the field work, families with more cultural capital (indicated by father’s education) were more willing to send their children to CET and more able to find better CET institutions or private tutors, and thus their children would obtain additional advantages over children from families with less cultural capital if attending CET could improve students’ academic achievement. Therefore, I expected the effects of family cultural capital on social reproduction of education inequality would increase along with the implementation of QOE. However, the results in Chapter Four do not support my expectation.

What factor could counteract the additional advantage of students from families with more cultural capital? Even though families with more cultural capital could be more

1 The data were collected before the one-child policy was adjusted in 2013. In Nanjing Municipality, before 2013, for a couple with more than two children, the fine was 10 to 16 times the annual per capita disposable income of urban households of last year (SCPCJP 2002). 124

willing to send their children to attend CET, there could still be some families with low cultural capital but high educational expectation for their children who sent their children to CET. If the law of diminishing marginal utility in economics could be applied to the effects of CET on children with different family cultural capital, then CET attendees with less family cultural capital would benefit more from attending CET than the ones with more family cultural capital, which also requires testing in Chapter Six.

It should be emphasised that the above mechanism is only supposed to explain the situation in primary school and JHS and its outcome, i.e., the transition opportunity from JHS to RSHS. This does not explain the situation in RSHS and its outcome, i.e., the transition opportunity from RSHS to university. The changing pattern of the association between family background factors and student transition opportunity from RSHS to university (see Table 4.6 in Chapter Four) is different from the one above. According to the difference in the implementation of QOE between the compulsory educational stage (primary school and JHS) and RSHS that I presented in the finding section of this chapter, the possible reasons for this difference are as follows.

There are several possible explanations for the trend in the effects of family cultural capital (indicated by father’s education). My explanation for the decrease in the effects of family cultural capital on student opportunities of entering college during and after the implementation of QOE policy is that attending CET could, to a large extent, improve children with less family cultural capital than those with more family cultural capital, which I name “equalisation effects”. However, why do those equalisation effects only appear in the transition from RSHS to university rather than the transition from JHS to RSHS mentioned above? As I presented in the finding section, interviewees suggested that besides the fact that at-school academic teaching and learning time had become shorter, the out-of-school tutoring had changed the at-school teaching, which was faster than before CET became popular and thus it was harder for students who did not attend tutoring or did not have strong self-learning abilities to follow. Therefore, if the majority (or a number before reaching a threshold) of those students were from a less-cultural capital family, on average the tutoring could not only add more edge to the students with more family cultural capital but also reduce the learning opportunity of students

125

with less family cultural capital from formal school teaching and make them worse off, which I name as “differentiation effects”.

When will the equalisation effects of CET overcome its differentiation effects and make the effects of family’s cultural capital decrease? I assume it depends on two conditions. One is whether there is a large enough proportion (reaching a threshold) of students with less family cultural capital attending CET (or sufficient CET, i.e., the amount of CET hours could also be important) since the condition of the equalisation effects occurring relies on attending CET. Compared to primary school and JHS, in RSHS I assume the gap of willingness to send their children to CET for parents with different cultural capital has been narrowed because the rejection rate of the RSHS admission is high 1 and the average family cultural capital (with smaller standard deviation) of these survivors is supposed to be much more than the average of the whole cohort when they are in primary and JHS.2 There is a significant association between family cultural capital and students’ opportunity of entering RSHS at least in the second and third QOE policy periods (see Model 3 of Table 4.3 in Chapter Four). The other condition is whether the academic at-school teaching and learning time has been significantly shortened or not, which is also related with whether the school teaching runs too fast for those students without attending CET. As I described above, school teaching does change in primary school and JHS but appears not to be significantly changed in RSHS because the out-of- school time in RSHS is much shorter and many RSHSs organise their own “voluntary” extracurricular tutorials and those who do not attend CET in RSHS will not be worse off like the ones in compulsory education will be. Therefore, in RSHS rather than in compulsory education, the equalisation effects defeat the differentiation effects and ensure that the effects of family’s cultural capital on students’ transition opportunity from RSHS to university decrease. In addition, if attending “enough” CET (not just attending CET) is also the condition for ensuring that the equalisation effects to occur,

1 For example, the admission rate is still only approximately 50% in Jiangsu Province in 2015 even after a several-year increase. 2 See Table 4.2 in Chapter Four. The average years of father’s education of students taking SHSAE including the ones not entering RSHS (M = 9.765, S.D. = 3.464, N = 929) is significantly less than students experiencing the transition from RSHS to university including the ones not entering university (M = 10.29, S.D. = 3.432, N = 506), t (1430) = 2.75, p = 0.006. 126

the fact of shorter out-of-school time in RSHS than in compulsory education means the standard for the “enough” could be lower in RSHS than in compulsory education. This could mean that parents with low family cultural capital are more willing to send their children to attend CET since the long CET time means high fees, and thus could suppress parents’ willingness to invest in their children’s CET.

But why do the effects of father’s ISEI (indicating family’s social status or prestige) on students’ transition opportunity from RSHS to university not significantly decrease during and after the implementation of QOE policy as the effects of father’s education do? The reason why the joint effects of differentiation and equalisation only become significant in the effects of father’s education is because the outcome is educational admission opportunity, and thus it could be more sensitive to the changes in the effects of an educational factor such as father’s education. The mechanism could be that CET directly replaces parental tutoring (including learning attitudes and habits, which are directly related to parent’s education) to affect students’ educational admission opportunity. In addition, the mechanism of “symbolic value” of attending CET does not occur on students’ transition opportunity from RSHS to university as it does on the transition from JHS to RSHS. This could also be explained by the higher mean and smaller deviation in father’s ISEI among students in RSHS due to the low RSHS admission rate rejecting students with lower father’s ISEI. However, the hypothesis on father’s ISEI is not confirmed with the data in Table 4.2 of Chapter Four, which could result from the larger standard deviation in father’s ISEI among students experiencing the transition from RSHS to university. This could be due to the “strong motivation for upward social mobility” from a few families with low social status who have the desire for upward social mobility (Hu 2014; Li 2015). However, the “symbolic value” of attending CET is subjective. If a student is able to enter a RSHS, the opportunity of entering university and upward social mobility is on the horizon (Fong 2006), which could have already added the “face” (“mian zi”, 面子, honour) to the family and lifted family social status, especially for those lower social status families (Zhai 2009). Thus, for those lower social status families sending children to CET was not an “out of the social league” action, even during a period (the second QOE period in RSHS) when CET was not too popular to own the “symbolic value”. Therefore, it is reasonable that the narrower gap in the possibility

127

of attending CET between RSHS students with different father’ ISEI cannot result in a significant change in the effects of a father’ ISEI on a student’s transition opportunity from RSHS to university.

Third, in terms of the effects of sibling number on the transition from RSHS to university, which inversely indicates the effects of family resources, its trend should have not changed significantly. If so, it could have been explained by the above discussion that because of the shorter out-of-school time and more school-organised “voluntary” tutorials in RSHS there is less “room” for CET to take effect in RSHS than in primary or JHS. It cannot be denied that the CET still exists in RSHS and has become more popular at least than before QOE. However, because of less out-of-school time in RSHS compared to primary and JHS, the potential total tutoring courses a student has time to attend are fewer, and thus the total fees are less than primary and JHS. Therefore, more underprivileged families could afford the CET fees if they thought that the tutorials or “evening and Saturday lessons” organised by RSHS were still not enough for their children. In addition, it is very possible that the high rejection rate of RSHS admission lifts the average family resources of RSHS students and, more obviously in my research, it is a more usual phenomenon that all the family resources are enjoyed by one child in RSHS than in primary school and JHS1. Therefore, CET is comparatively more affordable for families with less resources or at least the gap in the CET time amount between students with different family resources is not large enough to result in privileged students benefitting so much more from CET that the trend of effects of family resources per child changed during the CET era (during and after implementation of QOE policy).

However, the above theory can only explain why there is not significant change in the effects of sibling number between the first and second QOE period, but the effects of family resources per child did increase in the second QOE period but then decreased to the original (first QOE period) level during the third QOE period. Therefore, what other policy or social change happened during the second QOE period? As I described in the introduction chapter, there are several one-off upsurges in university tuition in the

1 See Table 4.2 in Chapter Four, X2 (2, N = 1506) = 6.10, p < 0.05. 128

second policy period as part of the industrialisation and expansion policy of higher education, which could explain the odd increase of the effects of sibling number in the second period and then its drop in the third period. That’s because s family income keeps increasing while each university tuition upsurge is just a one-off rather than indexed and has been frozen for several years since 2006 and thus in third QOE period the university tuition is no longer that high that it is beyond many underprivileged families’ paying capacity. Besides the effects of increase in university tuition, there is no clue that the QOE policy and its specific outcome, CET, significantly changed the extent of the effects of family resources per child on transition opportunities from RSHS to university since it was shown that there were no significant changes between the effects of sibling number in the first period and in the third period.

It should be noted that different from the changing pattern of the effects of the variable of “more than one sibling” on students’ RSHS admission opportunities, which is similar to the changing pattern of the effects of the variable “no sibling” (see Row Two in Table 4.4), on students’ transition opportunities from RSHS to university the changing pattern of the effects of “more than one sibling” is similar to “one sibling” (see Row Two in Table 4.6). In the above explanation of the changing pattern of the sibling number’s effects on students’ RSHS admission opportunities, I have mentioned that many families with more than two children could be rather rich because they could afford the large fine for violating the one-child policy. In the transition from RSHS to university, it is possible that these families are so rich that even though their family resources are diluted by multiple children, the upsurges in university tuition are still affordable from them. However, why does this possibility not appear in the changing pattern of the sibling number’s effects on students’ transition opportunities from RSHS to university so that families with only child and families with more than two children have a similar changing pattern due to their similar payment capabilities? According to the figure of average predicted probability of students’ transition from JHS to RSHS (see the figure in Row Two, Column One of Table 4.4), the increase in transition opportunity of students with two siblings or more mainly occurs during the third QOE period, which is different from a continuous increase in transition opportunity of only-child students. The reason could be that the cohort of the third QOE period were born between 1988 to 1993 and the privately-run

129

enterprise was legalized in 1988 (State Council 1988). Since the officials of governments, political parties, state-owned enterprises and public institutions (including schools, universities, research institutes and hospitals) have to comply with the strict one-child policy, the group who can afford the large fine for seriously (more than two children) violating the one-child policy without other serious consequences such as losing their jobs are mainly the owners and senior staffs of privately-run enterprises. Therefore, as the privately-run enterprises became thriving during the third QOE period, their owners and senior staffs could send their children to attend a lot of CETs, which help them catch up with their only-child peers. However, since the effects of upsurges in university tuition happened in the second QOE period, the privately-run enterprises were not thriving enough in that period and thus most of the cohort with more than one sibling could not come from families where the parents were owners or senior staffs of privately-run enterprises and not have more payment capabilities than their peers with one sibling. Therefore, those students with more than one sibling are as worse off as their peers with one sibling in terms of the transition opportunity from RSHS to universities owing to the upsurges in university tuitions. Again, since I have no suitable dataset to test the above explanation, my explanation mainly focusses on the changing patterns of only-child students and students with one sibling.

The above explanations have given lots of attention to the role of CET. However, it should be emphasised again that CET or, even its policy circumstance the QOE policy, should not be the exclusive factor in explaining the change of the association between family background factors and students’ educational transition opportunity. In addition, even in the above tutoring mechanism, what I have proposed are only a few assumptions based on the findings from my field work or deduced from some relevant theories. Several previous researchers also support part of my assumption. For example, Xue and Ding (2009) found that the primary and JHS students’ participation rate in CET was higher than RSHS students based on an urban survey in 2004, which supports my explanation for the difference in the effects of CET between the transition from JHS to RSHS (Table 4.4) and on the transition from RSHS to university (Table 4.2). However, for other assumptions that have not been confirmed by previous researchers, I cannot demonstrate all of them via hypothesis testing because of the limited quantitative data

130

I have. In the next chapter, I will propose three critical hypotheses based on the above assumptions that I am able to test.

131

Chapter Six. Can CET Improve Students’ Academic Performance?

I. Introduction and Hypothesis

In the previous chapter, I found the major change that resulted from implementation of the Quality-oriented Education (QOE) policy and was relevant to students’ educational transition opportunities was the popularity of commercial extracurricular tutoring (hereinafter CET). In the discussion section of Chapter Five, I utilised the qualitative findings from my field work to propose a possible explanation for how CET could influence the changing patterns of association between family background factors and students’ educational transition opportunity before, during and after the implementation of the QOE policy. In particular, I proposed the effects of CET could be used to explain why the estimated trends of social reproduction in the transitions from junior high school (JHS) to regular senior high school (RSHS) and from RSHS to university across three QOE policy periods were different from the hypotheses in the beginning of Chapter Four and why there was a difference between the trend in transition to RSHS and the trend in transition to university. Here, I briefly summarise the explanations proposed in the discussion section of Chapter Five and specify the hypotheses that I have derived from these arguments to test the effects of CET.

Based on results obtained from the field work, the QOE policy has increased students’ out-of-school time with the help of the new labour policy, especially during primary school and JHS, which aimed to alleviate student academic learning burdens and to help them have more extracurricular time to develop all-around qualities including sport and art qualities. However, due to the fiercely competitive senior high school and university admission examinations, it appears that the majority of the additional out-of-school time has been occupied by CET lessons instead of leisure activities. However, most CET is not focused on developing students’ sport or art qualities but on academic subjects to improve students’ current academic performance and, in the long term, to prepare students for academic examination-oriented RSHS and university admission. The need

132

for CET has increased since formal schools 1 capability for preparing students for admission has been damaged by other QOE reform measures including the unpractical new teaching methods and textbooks, the increased sport and art courses and school- based activities, in addition to the sub-policy of academic burden alleviation. CET is not only able to help students better prepare for academic examinations, but its popularity expedites the at-school teaching process. Moreover, the at-school learning time decreases, and thus makes students who do not attend CET worse off in the competition for admission opportunities. According to those findings from my field work, I propose the first hypothesis in this chapter: Hypothesis 6.1: Compared to students who do not attend CET, students who do attend CET exhibit better academic performance.

CET as a mediator should not only connect with the response variable, namely students’ academic performance, but also with the explanatory variable, namely family background factors. CET is not a formal or compulsory education and it is rather expensive for many underprivileged families. Therefore, some parents with low cultural capital may think they have carried out their duty by sending their children to formal schools and have no strong willingness to send their children to CET. Even though the majority of CET focus on academic subject, which employ lecturing in a big classroom as the teaching method and do not require other large expenditure like music instruments or sport gears, and thus the CET fees paid by each student are still affordable, many families with more than one child may have less financial ability to send all their children to CET. However, when CET became more popular, i.e., is no longer viewed as a luxury of the upper-class (social status, not only relevant to income), which occurred in the third QOE policy period, there may be no significant difference in the possibility of attending CET between families with different levels of social status (prestige).

Based on the above assumptions, Hypothesis 6.2 includes three sub-hypotheses, which separately suggest the association between CET and each of the three family

1 In China, even private schools are under the strict regulation of the educational authorities. 133

background factors discussed above. It should be noted that these hypotheses only refer to the third QOE period, which is due to the limitation of the current dataset. Hypothesis 6.2.1: Students from a family with more cultural capital indicated by the father’s education are more likely to attend CET. Hypothesis 6.2.2: Students with fewer siblings (indicating greater financial capital per child) are more likely to attend CET. Hypothesis 6.2.3: There is no significant difference in the possibility of attending CET between students with different family social prestige indicated by the father’s job.

The above hypotheses are based on the qualitative findings mainly through interviewing JHS parents, students and teachers (see Chapter Five) and the quantitative results of the pattern for the impact of family background factors on students’ transition opportunities from junior to RSHS across three QOE policy periods (see the column of “Transition from JHS to RSHS” in Table 6.1, which is a copy of Column 1 Prediction in Table 4.4 in Chapter Four). However, the transition pattern from RSHS to university is different (see the column of “Transition from RSHS to University” in Table 6.1, which is a copy of Column 1 Prediction in Table 4.6 in Chapter Four and compare those two columns in Table 6.1). In the previous chapter, I proposed some possible explanations for these differences. The basic explanation is that compared to students from JHS, RSHS students on average have more family cultural capital with smaller variances, fewer siblings diluting family resources and higher “subjective” family social status, which is due to the low RSHS admission rate rejecting many students with lower family backgrounds. In addition, the extracurricular time is less in RSHS than in JHS and the other compulsory educational stages, namely primary school, under the at-school time regulation of educational authorities. This means that the time allocated to CET is less in RSHS. Therefore, compared to primary and JHS, in RSHS the gap in CET between students from different family backgrounds should be less. However, with only these assumptions, the pattern relevant to the university admission should have been an insignificant but still similar trend as the one relevant to the RSHS admission. In Chapter Five, I also proposed additional assumptions to explain why there is an insignificant V- shape trend in father’s ISEI effects on the university admission rather than a similar

134

reverse V-shape effect trend on the RSHS admission, why there is a declining trend of father’s education effects on the university admission rather than a similar insignificant reverse V-shape trend in the RSHS admission, and why the gap of university admission opportunity between only-child students and non-only-child students narrows from the second to third QOE policy period while the gap of RSHS admission opportunity remains unchanged.

Even though Chapter Five explains all the above differences in social reproduction trends between compulsory education and RSHS, it should be noted that not all these changes across the three QOE policy periods are significant in Table 6.1 and, thus, I briefly display the significant changes (usually at the 95% significant level) in Table 6.2 (which consists of the second column of both Figure 4.1 and 4.2 in Chapter Four). For example, in Table 6.1 the trend in the effects of a father's ISEI on a student’s transition opportunity from RSHS to university forms a V-shape; however, both the changes from before-QOE to during-QOE and from during-QOE to after-QOE are not significant. Thus, I just display the trend with a horizontal line in the first figure of Table 6.2. Currently, I only have suitable data including variables measuring the situation of RSHS students’ attending CET in China, which was collected during the third QOE policy period. Therefore, I can only test part of the assumptions proposed in Chapter Five to explain the discrepancies between the transition from JHS to RSHS and the one from RSHS to university displayed in Table 6.2.

135

Table 6.1 AME and APP of Family Background Factors on Educational Transitions 1.Transition from JHS to RSHS 2.Transition from RSHS to University

.03

.02

.01

.02

-.01

.01 -.02

0 Before QOE During QOE After QOE

1.AME of Father's Education Average Marginal Effects of Father's Yrs of Edu on Probility of Entering University

Before QOE During QOE After QOE

.4 .4

Before QOE During QOE After QOE Before QOE During QOE After QOE

2.APP by Sibling Number no sibling 1 sibling 2+ siblings no sibling 1 sibling 2+ siblings

.0045

.008

.004

.006

.0035

.003

.004

.0025 .002

AME ISEI of Father's .002

3. 0 Before QOE During QOE After QOE Average Marginal Effects of Father's ISEI on Probility of Entering University across 3 QOE policy peirods Before QOE During QOE After QOE

Note: AME= average marginal effects. APP= average predicted probability.

136

Table 6.2 Trends in Effects of Family Background on Educational Transitions 1. Transition from JHS to RSHS 2. Transition from RSHS to University

0.18 0.2 0.15 0.1 0.12

Education 0.09 0 0.06 Before QOE During QOE After QOE -0.1 0.03 0 -0.2 Before QOE During QOE After QOE 1. AME of Father's

0.85 0.8 0.75 × 0.6 0.65 ^ × × 0.55 × 0.4 0.45 0.2 0.35 0

APP Siblingof Number Before QOE During QOE After QOE Before QOE During QOE After QOE

2. 0 Sibling 1 Sibling 2 Siblings 0 Sibling 1 Sibling 2 Siblings

0.04 0.03

0.03 0.02 0.02 +

0.01 AME of Father's 0.01

3. ISEI 0 0 Before QOE During QOE After QOE Before QOE During QOE After QOE

Note: AME= average marginal effects. APP= average predicted probability. The ^ in the 2nd figure of Column 1 indicates the gap between one sibling and the other two sibling indicators in the after-QOE period is logically but not statistically significant. The X in the 1st figures of Row 2 indicates the increase or decrease in the coefficients of one sibling and two siblings or more between two periods is logically but not statistically significant since the increase is significant between the 1st and 3rd periods. The × in the 2rd figure of Column 2 indicates the increase or decrease of coefficients of two siblings is logically but not statistically significant because the gap between the coefficients of one sibling and two siblings or more in the during QOE period is significant. The + in the 3rd figure of Column 2 indicates the increase is logically significant but only at the 90% significance level. The values indicated by the makers in each figure are not completely consistent with the predictions as they have been adjusted to draw the trend figures.

137

My hypothesis testing is focused on explaining why the trend in the effects of father’s education on university admission is different to the one on RSHS admission because a V-shaped trend in the father’s ISEI effects on the university admission opportunity is not significant, which can be explained by the gap of “subjective” social status between students is rather small in RSHS, which could result in the difference in the probability of attending CET between students with different family social statuses is insignificant even though the variance of “objective” social status indicated by father’s ISEI in RSHS is larger than in JHS. The several one-off increases in university tuition during the second QOE period explains why the gap relevant to the effects of sibling number on university admission opportunity only expands during the second QOE period. However, the significant declining effects of father’s education that indicates family cultural capital need further explaining.

In Chapter Five, I proposed the mechanism of equalisation effects of CET, i.e., attending CET could improve students’ education who come from lower cultural capital families to a greater extent than the ones from higher cultural capital families because the former on average have lower academic performance and CET could have some greater marginal utility. However, why does the mechanism of equalisation effects only become significant in the RSHS? This is because there is an adequate number of students from lower cultural capital family attending CET in RSHS, otherwise if only the students from higher cultural capital family attended CET and the tutoring might improve students’ academic achievement (the mechanism of differentiation effects), the gap of admission opportunity between students from higher and those from lower cultural capital families will not narrow but will broaden. According to the “basic explanation”, a higher mean and smaller variance of father’s education in RSHS could result in the difference in probability of attending CET between students with different family cultural capital in RSHS is smaller than in JHS and therefore the mechanism of equalisation effects becomes significant in RSHS. In addition, compared to a father’s ISEI and sibling number, a father’s education could be more directly influencing the student’s education and attending CET is directly utilising the more-equalised quasi-school education to replace the highly-differentiated family education. This should be more associated with family cultural capital rather than family financial resources or class status in terms of the

138

academic subjects examined by RSHS and university admissions. Therefore, the mechanism of equalisation effects mainly takes place in the pattern of the effects of father’s education and, considering the feasibility of the current datasets I have, I propose the following hypothesis, which is only relevant to family’s cultural capital while the other two family-background factors’ effects will be tested in the below analysis. Hypothesis 6.3: The gap between tutorial attendees and non-attendees is narrowing as their family cultural capital increases.

It should be noted that even though I try to use the above mechanisms to explain the changes both in the student’s RSHS and university admission opportunities, I only have data about students’ CET information in primary school and JHS. Therefore, the results should be cautiously used to explain the changes in opportunity for university admission. However, they could still be very useful because the mechanism of CET is more significant in JHS than in RSHS. More importantly, the transition from JHS to senior high school to some extent is more critical than the university admission because in China if a student enters a vocational senior high school rather than a RSHS it is very difficult for him or her to enter university and in recent years the admission rate from JHS to RSHS (approximately 50% in Jiangsu Province in 2015) is much lower than the one from RSHS to university (approximately 90% in Jiangsu Province in 2015). If CET can help students enter RSHS, it has made a big contribution to student university admissions. Therefore, even though the following analysis is only focused on the effects of tutoring on students’ academic performance in JHS, the results can still help to understand the QOE policy and its association with educational inequality.

Paralleling the previous consideration on the rural sample in Chapter Four, the analysis presented in this chapter only involves the urban sample. In addition to the reasons discussed before, the specific concern for the current study is that students in rural areas rarely attend CET. According to the Chinese Family Panel Studies (CFPS [family], Wave 2010) data, only approximately 5% of primary school and JHS students whose family are living in rural areas attend CET, while the percentage in urban areas is approximately 28%. More importantly, I assume that many students who do not attend CET in rural areas have a critical but different reason from their peers in urban areas. This reason is

139

because of the average lower family income and lower population density in rural areas, which means less market demand and higher operating costs for the tutorial companies and, therefore, the provision of CET in rural areas is more limited than in urban areas. This difference could make the discussion about tutoring much more complicated. Therefore, the sample in this analysis only comes from the urban data.

I employ different methods to test Hypothesis 6.1, 6.2 and 6.3 separately. First, I employ the treatment-effects method (propensity score matching, PSM) to test Hypothesis 6.1, namely the treatment effects of CET on a student’s academic performance since PSM has been identified as a method to reduce selection biases (Dehejia and Wahba 1999; Shadish, Cook and Campbell 2002). Second, I employ the logistic regression model to test Hypothesis 6.2, namely the association between family background factors and CET attendance. Third, I use an interaction-effects analysis between different levels of tutoring time amounts and family background factors to test Hypothesis 6.3, namely whether the extent of improvement in academic achievement are different between tutorial attendees from different family cultural capital.

II. Data, Variables, and Model

Chinese Family Panel Studies and China Education Panel Survey The study uses data from the China Education Panel Survey (CEPS [education])1 in 2013, which is different from the Chinese Family Panel Studies (CFPS [family]) used in Chapter Four. CEPS (education) is a county-wide survey and contains individual-, class-, school-, and community-level longitudinal data in contemporary China. The reasons why I used CEPS (education) as supplementary data are as follows: First, different from CFPS (family), CEPS (education) does not require students to report their academic achievement (examination score or ranking) themselves but directly acquires the mid-

1 CEPS (education) is funded by the Science Research Foundation of Renmin University of China, the Social Survey Foundation of National Survey Research Centre (NSRC) at Renmin University of China and the National Science Foundation (USA), and was carried out by NSRC. The CEPS website address is http://www.chinaeducationpanelsurvey.org/. 140

term examination scores of Mathematics, Chinese and English subjects 1 from the surveyed schools, which are the Secondary Sampling Unit (SSU). The examination scores of Mathematics, Chinese and English subjects are school-wide rather than nationwide. The examination score for each subject has been standardised within each grade and each school. Since my research is only focused on the difference in academic achievement between CET attendees and non-attendees and the interaction between CET and family background factors rather than the effects of school or district characteristics, the grade-school standardised scores are able to be used to make a comparison within the same grade of the same school. Thus, they are an appropriate measure for the propose of this research considering that in China there is no national assessment of academic performance for students in primary school and JHS equivalent to the National Assessment Program – Literacy and Numeracy (NAPLAN) in Australia. On the contrary, the self-reported examination score or ranking in the CFPS (family) is not an accurate and appropriate measure to make a comparison since the tertiary sampling unit in the CFPS (family) is the family and the examination taken by the survey respondents are different. The second reason is relevant to the sample size. CEPS (education) contains a national sample and its participants are only students from Grades 7 and 9 in the school year 2013–14 (students in the first and third year of JHS), which produces a very large sample size (11,521). Even though the final sample size for the following analysis is only approximately 1,500, it is still large enough. I will discuss the detailed selective criteria later. While it contains a national sample as well and a comparable “objective” outcome variable, “Have entered RSHS or not”, the current CFPS (family) only has approximately 200 cases who are from the urban areas and contain both the RSHS admission result and tutoring information, which is rather small for a national sample considering China’s population and for an estimation containing quite a few variables. Third, as its name suggests, CEPS (education) has many more variables relevant to education than CFPS (family), which is important for testing Hypothesis 6.1 since the treatment-effects analysis needs to control for as many confounding variable effects as possible (Garrido et al. 2014).

1 In JHS of China, Mathematics, Chinese and English are required courses for each student and they are also compulsory subjects in college admission examinations no matter what degree program a student will apply for. 141

It should be noted that the cases from CEPS (education) in the analysis of this chapter are not part of the cohorts in the policy period analysis in Chapter Four. The usual year of birth in the analysis of Chapter Four is between 1976 and 1993. However, in this chapter the sample from CEPS (education) was students from Years 7 and 9 at the surveying time of 2013 and who were born in 1998—1999 school year1 and in 2000— 2001 school year. Because there is no other dataset with more appropriate samples, I had to use the current samples and I assume the two samples in this chapter can be used to explain the situation in the third QOE policy period and report the cohort difference here.

As I mentioned in the introduction, the estimations in this chapter only utilised the urban sample. Since the extracurricular time is the prerequisite for CET, the samples also excluded students who were in boarding school. I restricted the sample in this analysis to participants who were in a non-full-boarding2 school or who were not boarding at school. In addition, other restrictions were that the school was not in the rural area, the household registration (“hu kou”) was “non-agricultural” and fathers were not peasants. Even though the number of private primary school and JHSs is very small (as mentioned in the brief introduction to the Chinese education system in Chapter One), the sample from CEPS (education) were in a public school since the research tries to exclude the effects of private school and CEPS (education) had a relevant variable to differentiate them.

Academic achievement The outcome variable Academic Achievement is measured by students’ average score in the mid-semester examinations in English, Mathematics and Chinese. The examination score of each subject has been standardised within each grade and each school and then its mean has been adjusted to 70 and its standard deviation to 10.

1 In China, a school year usually begins in September. 2 Some schools allow part of their students to board at school if the student’s home is far away from the school. 142

CET CEPS (education) only provided the data of amounts of academic CET time each student respondent attended in the week before conducting the questionnaire, but in its questionnaire, there was a set of multiple responses as to what CET the student respondent attended (without specifying the time range). Since the outcome variable is student’s academic achievement, I focus on the effects of academic CET. Hypotheses 6.1 and 6.2 are only relevant to the situation as to whether the student has attended CET, and thus this variable is dichotomous (i.e., Tutoring). When testing Hypothesis 6.3, considering the possible different results from different amounts of tutorial time, the tutorial time was measured with two dummy variables indicating that the student had some tutorial time (the amount is greater than 0 but not more than 5 hours per non- holiday week, i.e., Tutoring-some) or had much tutorial time (greater than 5 hours per non-holiday week and less than 10 hours, i.e., Tutoring-much), with a reference category being those who had no tutorial time. The reason why I chose 5 hours as the dividing point is because usually a tutorial lesson for one subject lasts approximately 2.5 hours according to the field work and the CEPS (education) data shows that approximately 44% of the attending-tutorial students did not have a tutorial time of over 5 hours in the urban non-boarding sample. This is consistent with some parents’ argument that they tried to make sure their children had at least one day off for their normal school assignments and some time for relaxation. However, it is still an arbitrary criterion. The reasons why I did not include the cases with 10 tutorial hours or more per week and categorise them into a third dummy variable are that the number of those cases was small and, more importantly, the possibility of abnormality such as very poor academic performance is higher in this category since greater than 10 tutorial hours plus several hours for homework means no time for recreation. In addition, unfortunately, I could not find an approach or a better variable in the data to deduct the tutorial time of non- academic subjects such as sport or art from the total tutorial time, which could affect the accuracy of the analysis. However, according to the findings from the field work in Chapter Five, the majority of parents paid most attention to their child’s academic performance and it could be inferred that most of the tutoring time was spent on academic subjects.

143

Covariates Family background measures I included three family background factors in Chapter Four: father’s education, sibling number and father’s occupational status. Father’s education and sibling number were measured the same way in this chapter as they were in Chapter Four. In terms of father’s occupational status, CEPS (education) only provided a nominal variable of 10 categories to indicate parent’s occupation. The peasant has been excluded from the urban sample as I mentioned above and the “unemployed” (approximately 2% of the urban sample) and “others” (approximately 6%) were also excluded. For the efficiency of estimation and convenience of interpretation, I merged the remaining seven categories into three: worker, small businessman and professional (including company manager and public servant)1. The variable was transformed into two dummy variables with a reference category of those whose father’s job was worker, which is different from the estimation with the data from CFPS (family) that used father’s ISEI instead of the variable of three- category occupation in Chapter Four. It should be noted that in this chapter the two father’s socioeconomic factors (father’s education and father’s job) reflect their state when the survey was implemented while in Chapter Four they reflect their state when the student was 14 years old.

Student’s gender, intelligence and diligence Student’s gender is a common control variable. I also include two other student quality measures namely intelligence and diligence. The general intelligence can predict academic achievement very well, especially the standardised achievement test scores (Frey and Detterman 2004; Deary et al. 2007; Duckworth, Quinn and Tsukayama 2012). The test was developed on the basis of the ability tests of Taiwan Education Panel Survey 2001 (Yang, Tam and Huang 2003) and CEPS (education) provided a psychometric report as well (Wang and Li 2015). The general intelligence score was estimated from the original test score with the 3-parameter logistic model of item response theory (Baker and Kim 2004). Student diligence is supposed to be correlated to student academic achievement and was a composite consisting of three items asking students to indicate

1 I replaced the three-category variable with the original seven-category variable to run the estimation and found the two results were similar. Therefore, I decided to use the simple one. 144

on a four-point Likert scale to what extent they agree with the statement about their working hard on study (i.e., diligence).

Grade retention and class ranking I included a measure for whether a student had repeated a grade (i.e., retention) because it indicates a student’s previous achievement and could influence parent’s investment desire in their child’s education. In addition, class is the tertiary sampling unit of CEPS (education). The class-level measure is the average academic performance ranking of a student’s class in the grade,1 which is measured with two dummy variables (Middle to Upper-Middle and Highest) with a reference category of Lowest to Lower- Middle. The class ranking was originally evaluated and reported by the head class teacher2 with a five-point rating scale and only approximately 3% of head class teachers reported that their class ranking was one of the lowest. For the reason of achieving the balance of covariates across treatment and comparison groups within blocks of the propensity Score I re-categorised the five-point rating scale into three dummy variables. The reason for including class ranking is due to the hidden or disguised key class system, which is nominally forbidden by the educational authorities. Similar to a key school that I described in Chapter 1 or a selective school in Australia, a key class has more experienced teachers and more students with good examination scores than an ordinary class and different teaching quality and other factors associated with class ranking (e.g., the atmosphere of competition, the sense of honour) are supposed to influence students’ academic achievements.

At-home parents and parental educational expectation I controlled for whether both parents were at home or at least one parent was not at home (i.e., parents) in the estimation with CEPS (education) data because parents’ care could also affect children’s academic performance. CEPS (education) provided this variable because it had a special research interest in the phenomena of peasant labour migration and its consequence of the “left-behind children” as I mentioned in Chapter

1 In China, a grade usually consists of several classes. 2 In China, students have their permanent classroom while it is the teachers that come and go. A class head teacher needs to teach a course, as well as taking on the additional responsibility of supervising and advising students. 145

Four. Even though the reason for both parents not being at home in the urban sample are not due to peasant labour migration, I still included this variable because the propensity score matching or weighting method encourages including as many covariates as possible in the estimation. Parent’s expectations of their child’s final educational attainment (years of education) was correlated to the student’s academic achievement and was included in the model. Parent’s expectation of students’ final educational attainment was measured as father’s education (i.e., parent’s expectation).

Grade-School ID Since the outcome variable, namely the examination score, has been standardised within each grade and each school, I include a large set of dummy variables (i.e., grade- school ids) indicating a specific grade of a specific school where a student was located to run a fixed effects models to control the effects of the grade and school characteristics and the district where the school was located. In addition, since the research aimed to compare the academic achievement between CET attendees and non-attendees, I excluded the grade-school groups in which all students were CET attendees or all of them were non-attendees. The number of grade-school groups included in the analysis was 36 (12 groups were excluded).

Method and modelling procedures To test Hypothesis 6.1, I began my analysis with the fixed effects model with data from CEPS (education): 퐾 푀

Y푖푗푘 = ∑ 휇푘퐺푆푘 + ∑ 훽푚푋푚푖푗푘 + 휀푖푗푘 푘=2 푚=1 where, Y represents an academic achievement outcome for student i in class j in grade- school k. A set of dummy variables (퐺푆푘) for grade-school groups allows the intercepts to vary between different 퐺푆푘 groups. The explanatory and control covariates (푋푚푖푗푘) include the treatment indicator (tutoring), a set of household measures, a student’s gender intelligence and diligence, a set of measures for a student’s grade retention and class ranking, and a set of measures for whether both parents were at home and parent’s educational expectations as described in Table 6.3.

146

Consistent with Gottfried and Le (2016), I employed the Huber-White adjustment (Huber 1966; White 1980) to correct the error term and cluster the error at the class level as students in the sample are nested in the tertiary sampling unit, i.e., class.

Even though the above methods tried to control for as many covariates as possible, the selection bias was still an obstacle to demonstrate the effects of tutoring on academic performance. Therefore, the research used the propensity score matching (PSM) to control selection bias with the same sample and variables as above. Besides matching strategies, there are some weighting strategies to estimate average treatment effects (ATEs). I used the code fragments integrated in Stata 14 (-teffects psmatch-) (StataCorp 2015b). The detailed description of the above method can be found in Stata’s manual (StataCorp 2015c: 247-256,296-304).

I used CEPS (education) rather than CFPS (family) as the dataset to run the treatment- effects analysis because this dataset included more covariates potentially related to the outcome of a student’s academic performance to reduce bias even if some covariates were thought to be only associated with the outcome and not the treatment (Brookhart et al. 2006; Austin 2011). Actually, it is beneficial to include all potentially relevant covariates if the dataset is adequately large as the “noise” produced by potentially irrelevant covariates may not obscure any reduction in bias that is gained by their inclusion (Imbens 2000; Brookhart et al. 2006; Ho et al. 2007; Garrido et al. 2014). Table 6.3 presents differences between the attendees and non-attendees of CET with respect to the outcome, student, household and class, which motivated the treatment-effect analyses described in the following section. As I mentioned before, the variables in Table 6.3 were also used in the two regression models before conducting the treatment-effect analyses.

147

Table 6.3 Descriptive Statistics for Treatment-effects Analysis Tutoring Non-tutoring Mean SD Mean SD Outcome Examination Score*** 72.93 7.63 70.82 9.02 Family background measures Father’s Years of Education*** 13.11 3.03 11.74 3.26 Sibling number 0 sibling*** 0.80 0.40 0.60 0.49 1 sibling*** 0.15 0.35 0.32 0.47 2 siblings or more 0.05 0.22 0.08 0.27 Father’s Job Worker** 0.29 0.46 0.37 0.48 Small Businessman 0.24 0.43 0.26 0.44 Professional*** 0.47 0.50 0.37 0.48 Gender (male)* 0.45 0.50 0.53 0.50 Intelligence*** 0.51 0.83 0.19 0.92 Diligence 3.37 0.70 3.31 0.69 Retention (Yes)*** 0.02 0.14 0.09 0.29 Class ranking Lowest to Lower-Middle 0.15 0.36 0.14 0.35 Middle to Upper-Middle 0.67 0.47 0.79 0.41 Highest 0.18 0.38 0.07 0.25 Parents (both at home) *** 0.12 0.32 0.23 0.42 Parent's expectation (Yrs. of Edu. )*** 16.83 4.44 15.79 4.31 Grade-School ID N=36 Note. n=455 attendees of CET (Tutoring), 769 non-attendees (Non-tutoring). *p<.05, ** p<.01, *** p<.001 between Tutoring and Non-tutoring.

As Tutoring is a binary variable, I used the fixed effects binary logistic regression model to test Hypotheses 6.2 (including Hypothesis 6.2.1 to 6.2.3). The model that I attempted to fit is: 퐾 푀

Pr(푇푢푡표푟푖푛푔 = 1) = 퐹 (∑ 휇푘퐺푆푘 + ∑ 훽푚푋푚푖푗푘) 푘=2 푚=1 where, 퐹(푧) = 푒푧⁄(1 + 푒푧) is the cumulative logistic distribution.

Table 6.3 can be used to present the descriptive statistics about the difference between the attendees and non-attendees of CET with respect to the family’s socioeconomic measures, namely father’s education, sibling number and father’s job.

148

The above estimations using fixed effects linear regression are corrected for oversampling and computed figures representative of China’s general population with the sampling weights provided by CEPS (education). However, for treatment-effects analysis, “as far as we know it's not really clear in the literature how to accommodate sample weights in the context of matching” and “the recommendation to date seems to be to ignore sampling weights”1 (Leuven and Sianesi 2003). In addition, the two Stata 14 code fragments integrated used for the estimation (-teffects psmatch-; -xtlogit-) do not allow the sample weights to be included either. Therefore, I did not include the sample weights into the treatment-effects estimation and correspondingly, the fixed effects binary logistic regression model predicting the probability of being in the treatment group is not corrected for oversampling. Actually, after checking the results from the fixed effects linear regression without the sampling weights, the conclusion does not need changing.

To test Hypothesis 6.3, I employed the interaction-effect analysis between tutoring time levels and father’s education. Since father’s education indicating family cultural capital is a continuous variable and is supposed to be positively associated with a student’s academic performance, the hypothesis will be confirmed if its slope is significantly smaller for students with some or much tutorial time than students without tutorial time. In this situation, the gap of academic achievement between students without tutorial time and with some or much tutorial time is larger among students with less family cultural capital than students with more family cultural capital, which indicates the academic achievement of students from a family with less cultural capital will be improved more than the ones with more family cultural capital if they both attend tutoring (see Figure 6.1). Even though I assumed the above gap only occurred among students with different family cultural capitals, in the next section the interaction-effect analysis will also report the results of whether the gap also occurred among students with different family resources per child (reversely indicated by sibling number) and class status (indicated by father’s job) for the purpose of analysis integrity.

1 See the help document integrated into the Stata user-written command. 149

100 90 80 70 Tutoring-none 60 Tutoring-some 50 Tutoring-much 40 30

20 Examination Scores Examination 10 0 0 Year Father's education 22 Years

Figure 6.1 Testing Hypothesis 6.3

In the interaction-effect analysis, the outcome indicator is a continuous variable, and thus is included in a linear regression model together with family background factors, the variable of different tutoring time levels and other control variables. For a few similar reasons as the method testing for Hypothesis 6.2, especially the concave problem, I only clustered the error at the class level and did not present results based on multilevel mixed-effects models in the following section. The cluster-robust can also make the reported standard errors valid for weighted estimations.

Different to Chapter Four, I did not use the multiple-imputation method because the data from CEPS (education) were rather complete, and thus I only used the original data without missing imputation. The CEPS (education) provided the sampling weights to correct for oversampling and compute figures representative of China’s general population. Table 6.4 presents the descriptive statistics for the estimation.

150

Table 6.4 Descriptive Statistics for Interaction-effects Analysis Variable Obs. Mean SD Min Max Outcome

Examination Score 1,224 71.60 8.59 37.69 91.22 Tutoring time None 769 0.63 0.48 0 1 Some 240 0.20 0.40 0 1 Much 215 0.18 0.38 0 1 Household measures Father’s Years of education 1,224 12.25 3.25 0 19 Sibling number 0 sibling 828 0.68 0.47 0 1 1 sibling 313 0.26 0.44 0 1 2 siblings or more 83 0.07 0.25 0 1 Father’s Job Worker 419 0.34 0.47 0 1 Small Businessman 307 0.25 0.43 0 1 Professional 498 0.41 0.49 0 1 Gender (male) 1,224 0.50 0.50 0 1 Parents (both at home) 1,224 0.19 0.39 0 1 Class ranking Lowest to Lower-Middle 22 0.15 0.35 0 1 Middle to Upper-Middle 517 0.75 0.44 0 1 Highest 631 0.11 0.31 0 1 Intelligence 1,224 0.31 0.90 -2.03 2.71 Diligence 1,224 3.33 0.69 1 4 Retention (Yes) 1,224 0.07 0.25 0 1 Parent's expectation (Yrs. of Edu.) 1,224 16.18 4.39 0 22

151

III. Results

Treatment-effects analysis Fixed effects regression results Table 6.5 presents the fixed effects regression results of the effects of having attended CET on students’ examination scores, which indicates their academic outcome. Model 1 displays only the effects of the treatment of attending CET without controlling for family background factors. Model 2 additionally includes family background factors and students’ gender, which are similar to the key explanatory variables in the analysis in Chapter Three. Model 3 controls for a set of variables including a student’s intelligence and diligence, grade retention, class ranking, at-home parents and parent’s educational expectation. These additional control variables could be the mediators between the outcome and family background measures and may obscure the effects of family background measures. Thus, I used two separate models to determine their effects. In all three models, the effects of each grade-school group on the intercept are not displayed to save space.

All three estimation results indicated that attending CET was associated with having a higher examination score, even when controlling in the models for different personal, household, class, grade and school measures. For example, in Model 3 the examination score of students who attended CET are approximately one point higher than those who did not attend. Compared to the coefficient of father’s education, the additional one point is nearly equivalent to a four-year difference in father’s education. Even considering that the tutoring could be a mediator between father’s education and the outcome, which could result in a smaller coefficient of father’s education in this model, I ran the same model excluding the variable of tutoring and the coefficient of father’s education was 0.27 (clustered S.E. = 0.08, p < 0.01). Thus, comparatively the coefficients of tutoring in the above models are still considerable.

152

Table 6.5 Fixed Effects Model predicting a student’s examination score DV: Examination Score (1) (2) (3) Treatment: Tutoring(Yes) 2.813*** 2.049** 1.036* (0.73) (0.61) (0.51) Household measures Father’s Years of education 0.429*** 0.268** (0.10) (0.08) Sibling No. (None=0) 1 sibling -1.874* -1.542* (0.75) (0.63) 2 siblings or more -3.503* -2.351* (1.40) (1.05) Father’s Job (Worker=0) Small Businessman -0.268 -0.125 (0.58) (0.56) Professional 1.262+ 0.710 (0.66) (0.60) Gender (male) -3.741*** -3.484*** (0.45) (0.43) Intelligence 4.128*** (0.33) Diligence 0.215 (0.34) Retention (Yes) -1.132 (1.41) Class Ranking (Lower than Middle =0) Middle to Upper-Middle 3.947*** (1.05) Highest 6.345*** (1.56) Parents (both at home) -0.415 (0.65) Parent's expectation (Yrs. of edu.) 0.349*** (0.07) Grade-School ID Omitted. constant 70.534*** 67.733*** 58.919*** (0.45) (1.31) (2.32) n 1224 1224 1224 R2 0.051 0.154 0.356 Note. Coefficient estimates and standard errors in parentheses. + p < 0.10, * p < 0.05, ** p < 0.01, *** p < 0.001.

153

Treatment Effects Results (PSM) The above fixed effects regression results cannot rigorously control for the possibility that some parents might be predisposed to send their children to CET if I assume that whether to do so or not were available options. Therefore, I re-examined the results with a treatment-effects estimation design. As the first step involved in the treatment- effects analysis, the process of choosing covariates (including the grade-school dummy variables) to include in the propensity score contained numerous trade-offs between covariates’ effects on bias and efficiency (Garrido et al. 2014) and numerous iterations of testing the balance of those covariates across treatment and comparison groups after the preliminary estimation of the propensity score. Even though there is no rule regarding how much imbalance is acceptable in a propensity score, the proposed maximum standardised differences of the propensity score for specific covariates between the treatment and comparison groups range from 10 to 25% (Garrido et al. 2014; Austin 2009; Stuart, Lee and Leacy 2013). The ratio of variances in the propensity score of a specific covariate between the treated and comparison groups should be near 1 or fall in the range between 0.5 and 2 if the treatment and comparison groups are balanced (Garrido et al. 2014; Rubin 2001). During this process, some covariates that appeared to be barely associated with the outcome were deleted or were categorised. The confounding factors in Table 6.1 are the results of this process. In addition, one or two grade-school groups with only a few cases (3–6 cases) are excluded from the models by deleting their corresponding dummy covariates of grade-school ID to ensure covariate balance (Table 6.6). In addition, the overlap assumption that each individual has a positive probability of receiving each treatment level was checked and found not to be violated via the integrated code fragment—teffects overlap— (StataCorp 2015b) after estimating the propensity score (see Figure 6.2).

154

density

0 .2 .4 .6 .8 1 Propensity score, Tutoring=0

Tutoring=0 Tutoring=1

Figure 6.2 Overlap plots of propensity score (number of neighbour =3)

Table 6.6 provides the two estimation results of the ATE in the population from PSM after separately specifying the number of matches per observation (number of neighbours, nn) as five and four. The PSW with a smaller number of neighbours reduces the estimation bias at the cost of the precision of estimated treatment effect (Grilli and Rampichini 2011). The above number of neighbour specification results were determined after numerous trade-offs between the bias and efficiency and numerous iterations of checking the covariate balance. The PSM 1 with five neighbour specifications exclude one grade-school group (three cases) for the covariate balance and after the PSM, all the covariates have standardised differences of the propensity score <10% except for two covariates (retention, 10.4% and a grade-school dummy variable, 10.8%). All of them have the ratio of variances ranging between 0.5 and 2. In PSM 2 with four neighbour specification and excluding two grade-school groups (six cases), the covariate balance was also achieved (standardised differences <10% and ratio of variance near one for all covariates). In Table 6.6, the estimation results of the effects of treatment are also shown, namely having attended CET, on students’ academic performance (examination scores) from the fixed effects regression analyses for the purpose of comparison. The Fixed Effects Models 2 and 3 use the same variables

155

as Model 3 of Table 6.5 except for excluding one or two grade-school groups. Both PSM 1 and PSM 2 indicate that after students are matched on the propensity to attend CET, having attended extracurricular tutoring is associated with better academic performance (approximately 0.5-point examination score higher). Even though both the effect sizes of the treatments in the two PSM models are approximately half of the ones in fixed effects models, it is still approximately equivalent to the effect size of two years of father’s education. Therefore, Hypothesis 6.1 has been confirmed.

Table 6.6 Comparing Coefficients and Standard Errors of Different Models Model N Coefficient Std. Err. Note Fixed Effects Model 1 1224 1.04* 0.51 See Model 3 of Table 6.5 for details

Fixed Effects Model 2 1221 1.08* 0.51 one grade-school group PSM 1 (nn# =5) 1221 0.50* 0.23 excluded

Fixed Effects Model 3 1218 1.15* 0.51 two grade-school groups PSM 2 (nn# =4) 1218 0.50*** 0.10 excluded # nn: number of matches per observation

CET and family background factors To reduce any confusion, the treatment-effects analysis should include covariates theoretically related to both outcome and treatment. Therefore, the logistic regression analysis of whether having attended CET was dependent on family background factors and other confounding factors was conducted before I obtained the treatment-effects analysis results. Table 6.7 presents the above logistic regression results. Consistent with Table 6.5, Model 1 excludes the possible mediators between family background factors and tutoring. According to Model 1, the only-child status is significantly associated with a larger probability of students’ having attended CET while the effects of father’s years of education and job are not significant. Compared to students without a sibling, the odds of having attended CET of students with one sibling decrease by approximately 45% (1-e-0.595≈0.448); However, there is no difference between students without a sibling and students with two siblings or more. The effects of family background factors on students attending CET are not changed largely after controlling more covariates in Model 2. In brief, Hypothesis 6.2.1 (father’s education) is not confirmed. Hypothesis

156

6.2.3 (father’s job) has been confirmed while only part of Hypothesis 6.2.2 (sibling number) has been confirmed.

Table 6.7 Logistic Regression of Tutoring on Family Background and Others Outcome: Tutoring (Yes) (1) (2) Household measures Father’s Years of Education 0.048+ 0.033 (1.82) (1.22) Sibling Number (Non-sibling=0) 1 sibling -0.595** -0.603** (-3.15) (-3.14) 2 siblings or more 0.038 0.181 (0.12) (0.56) Father’s job (Worker =0) Small Businessman 0.122 0.097 (0.65) (0.50) Professional -0.018 -0.082 (-0.10) (-0.45) Gender (male) -0.386** -0.365** (-2.81) (-2.61) Intelligence 0.249** (2.63) Diligence -0.131 (-1.26) Retention (Yes) -0.606 (-1.50) Class Ranking (Lower than Middle =0) Middle to Upper-Middle -0.227 (-0.90) Highest 0.203 (0.57) Parents (both at home) -0.394* (-1.97) Parent's expectation (Yrs. of edu.) 0.039* (2.38) Grade-School ID Omitted. _cons not reported N 1224 1224 pseudo R2 0.021 0.042 Note. t statistics in parentheses. + p < 0.10, * p < 0.05, ** p < 0.01, *** p < 0.001. For Models 1 and 2, number of grade-school groups = 35.

157

Interaction effects analysis To test Hypothesis 6.3, Table 6.8 presents the interaction-effects analysis results between different tutoring time levels and family background factors on students’ academic achievements. Model 1 in Table 6.8 shows that without controlling for other covariates, having attended extracurricular tutorials is strongly associated with students’ high academic achievement and if a student attended some extracurricular tutorials and the tutorial time amount was not greater than five hours per week, her or his examination score could be approximately three points higher than students without any tutoring. In terms of students with greater than five hours of tutorial time (> 5 and <= 10 hours), the increase is approximately 2.6 points. However, there is no significant difference between the effects of the attending some-tutorial group and the attending much-tutorial group after changing the reference from non-tutoring group to much- tutoring group. However, after controlling for the family background factors and other control measures, the above increases of examination scores become insignificant, which are inconsistent with the results from the PSW analysis. Besides the statistical factor of dividing the attending CET group into two sub-groups (some-tutoring and much-tutoring), another possible reason for the above not significant difference has been proposed in Hypothesis 6.3, i.e., attending CET could have different effects on the academic achievement of students from different socioeconomic groups.

Therefore, Models 3 to 5 separately present the interaction-effects analysis results between three family background factors (father’s education, sibling number and father’s job) and tutoring time levels. Only the effects of the father’s years of education on students’ academic achievement significantly showed some variation between different tutoring time levels (see Model 3 in Table 6.8). As a father’s educational achievement increases in one year, the student’s examination score increases by approximately 0.4 in the non-tutoring group. The coefficient of father’s education in the some-tutoring group is not significantly different from the none-tutoring group; however, in the much-tutoring group the student’s examination score did not increase but decreased approximately 0.08 (0.381-0.460 = -0.079). Even though the decrease is insignificant after further testing or, in other words, in the much-tutoring group, the student’s academic achievement score rarely changed as father’s education increased,

158

it is still significantly different from the positive effects of father’s education on the academic achievement of students who did not attend CET.

The last Model in Table 6.9 includes a third dummy variable of “tutoring” indicating a group of students attending more than 10 hours tutoring per week and the effects of the father’s education on students’ academic achievement in this group does not significantly increase compared to the group of students who did not attend extracurricular tutoring.

Therefore, has the above estimation confirmed Hypothesis 6.3 or not? I used predictive margins and margins plot techniques of Stata post-estimation (StataCorp 2015a: 1359- 1416, 1440-1474) to draw the figure of predictive margins (potential-outcome means) for each level of tutoring time when the father’s education was set from 0 to 19 years (see Figure 6.3). According to Figure 6.3, the gap in the examination scores of students with less family cultural capital between the Tutoring-Much group and Tutoring-None (or Tutoring-Some) is much larger than the one between students with more family cultural capital. More specifically, for a student whose father does not attain formal education, the difference in the examination scores between those who attend 5 to 10- hour tutoring per week and those who attend any tutoring is approximately 7 points in the centesimal grade, however, for a student whose father attains 15-years of formal education, the difference is almost non-existent. Therefore, the result basically supports Hypothesis 6.3 even though the exceptional situation of “Tutoring > 10” should be discussed further.

159

Table 6.8 Interaction Effects Analysis between CET and Family Background DV: Examination Score (1) (2) (3) (4) (5) Tutoring (No=0) Some 3.031*** 0.862 2.790 1.054 1.458 (3.78) (1.34) (1.28) (1.35) (1.37) Much 2.561** 1.232+ 7.059* 1.623* 0.560 (3.00) (1.94) (2.50) (2.40) (0.41) Father’s Years of Education 0.270** 0.381*** 0.267** 0.269** (3.19) (3.58) (3.15) (3.15) Sibling No.(None=0) 1 sibling -1.551* -1.541* -1.272+ -1.535* (-2.47) (-2.47) (-1.84) (-2.42) 2 siblings or -2.363* -2.441* -1.744 -2.373* more (-2.27) (-2.36) (-1.48) (-2.24) Father’s job (Worker=0) Small- -0.128 -0.140 -0.119 -0.636 Business- (-0.23) (-0.25) (-0.21) (-0.94) man 0.708 0.680 0.740 1.059 Professional (1.19) (1.14) (1.24) (1.30) Parents (both at home) -0.417 -0.404 -0.439 -0.380 (-0.64) (-0.62) (-0.69) (-0.59) Gender (male) -3.49*** -3.53*** -3.46*** -3.49*** (-8.07) (-8.01) (-8.04) (-7.90) Class Ranking (Lower than Middle =0) Middle to Upper-Middle 3.941*** 3.887*** 3.965*** 3.930*** (3.74) (3.77) (3.72) (3.71) Highest 6.375*** 6.435*** 6.426*** 6.435*** (4.12) (4.25) (4.13) (4.23) Intelligence 4.133*** 4.145*** 4.110*** 4.134*** (12.39) (12.42) (12.63) (12.44) Diligence 0.216 0.174 0.199 0.191 (0.64) (0.52) (0.59) (0.55) Retention (Yes) -1.124 -0.954 -1.200 -1.078 (-0.80) (-0.68) (-0.86) (-0.75) Parent's expectation (Yrs. of edu.) 0.349*** 0.344*** 0.351*** 0.347*** (4.88) (4.91) (4.91) (4.89) Grade-School ID Omitted. Interaction Effects: F’s yrs. Sibling Father’s Tutoring of edu. No. Job Some -0.154 -0.617^ 0.874# (-0.91) (-0.38) (0.69) -0.945^^ -1.676## (-0.45) (-1.21) much -0.460* -1.070^ 2.063# (-2.15) (-0.86) (1.32) -2.469^^ 0.247## (-0.77) (0.15) cons 68.87*** 53.50*** 52.45*** 53.37*** 53.66*** (59.91) (21.57) (21.22) (21.38) (21.36) N 1224 1224 1224 1224 1224 R2 0.051 0.356 0.360 0.357 0.359 Note. Coefficient estimates and clustered standard errors (in parentheses). + p < 0.10, * p < 0.05, ** p < 0.01, *** p < 0.001. ^ 1 sibling, ^^ 2 or more siblings. # Small businessman, ##. Professional.

160

Table 6.9 Interaction Effects Analysis with Four Tutoring Levels DV: Examination Score (1) (2) Tutoring (No=0) Some 0.955 2.640 (1.51) (1.20) Much 1.212+ 6.848* (1.94) (2.49) > 10 hrs. -0.045 0.542 (-0.06) (0.17) Father’s Years of Education 0.272*** 0.371*** (3.66) (3.70) Sibling No. (None=0) 1 sibling -1.450* -1.447* (-2.56) (-2.55) 2 siblings or -1.552 -1.625 more (-1.55) (-1.63) Father’s job (Worker=0) Small Businessman -0.378 -0.385 (-0.74) (-0.75) Professional 0.915+ 0.892+ (1.84) (1.79) Parents (both at home) -0.556 -0.538 (-0.96) (-0.93) Gender (male) -3.466*** -3.506*** (-9.11) (-9.06) Class Ranking (Lower than Middle =0) Middle to Upper-Middle 3.827*** 3.784*** (4.01) (4.03) Highest 4.987*** 5.034*** (3.67) (3.76) Intelligence 4.133*** 4.136*** (13.23) (13.26) Diligence 0.200 0.164 (0.68) (0.55) Retention (Yes) -1.479 -1.342 (-1.12) (-1.03) Parent's expectation (Yrs. of edu.) 0.338*** 0.335*** (5.53) (5.55) Grade-School ID Omitted. Interaction Effects: Tutoring Some -0.135 (-0.79) much -0.445* (-2.11) >10 hrs. -0.0490 (-0.21) cons 54.06*** 53.12*** (23.98) (22.61) N 1412 1412 R2 0.355 0.358 Note. Coefficient estimates and clustered standard errors (in parentheses). + p < 0.10, * p < 0.05, ** p < 0.01, *** p < 0.001.

161

LinearPrediction (ExamScore)

66 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 Father's years of education Non-tutoring Some-tutoring Much-tutoring > 10 hrs

Figure 6.3 Predictive student achievement by father’s education by tutorial group1

1 The prediction is based on the estimation of Model 2 in Table 6.9. 162

IV. Summary and Discussion

In this chapter, I applied the fixed effects linear regression model, treatment-effects analysis, logistic regression and interaction-effects analysis to test my findings from the field work. The proposed hypotheses of this chapter included that attending CET in urban China could improve students’ academic achievement (Hypothesis 6.1), attending CET was positively associated with family cultural capital (Hypothesis 6.2.1) and with family resources per child (Hypothesis 6.2.2), but not associated with family social prestige (Hypothesis 6.2.3), and attending CET negatively moderated the association between family cultural capital and children’s academic performance (Hypothesis 6.3). First, the analysis in this chapter confirms Hypothesis 6.1 that attending CET has a significant positive impact on students’ academic achievement. Even though the improvement brought by CET in the examination scores seems not very large (in general approximately 0.5% -1.15%), in the admission system that “every point is crucial and worth fighting for” (分分必争) (Deng 2016), it is difficult for parents, teachers and students to neglect the effects of CET and to abandon an opportunity of admission into a better RSHS or university. In addition, students’ attending CET is not significantly associated with family cultural capital indicated by father’s education and thus Hypothesis 6.2.1 is not confirmed. Only-child students are more likely to attend CET than students with one sibling, but there is no difference in the probability of attending CET between only-child students and students with more than two siblings and thus Hypothesis 6.2.2 is only partially confirmed. Attending CET is not significantly associated or with father’s job indicating family social status (prestige), which confirms Hypothesis 6.2.3. Third, the research demonstrates that attending CET improves the academic achievement of students with less family cultural capital more than students with more family cultural capital, but this difference is only significant when they attend tutoring for about 5-10 hours per week, which partially confirms Hypothesis 6.3.

I propose some possible explanations briefly for the discrepancies between the hypotheses and the above results. First, in terms of the insignificant association between family cultural capital (indicated by father’s education) and students probability of attending CET (Hypothesis 6.2.1), it should be noted that the data from CEPS used to

163

test the hypotheses was collected in 2013 and if CET continues becoming more popular after the third QOE policy period (2003–2008, the year of graduating from JHS) due to the reasons revealed in Chapter Five, in 2013 the differentiation in willingness of sending children to attend CET among families with different cultural capital should decrease. Actually, this result could be expected based on the insignificant decrease in the effects of father’s education on students’ transition opportunity from JHS to RSHS between during and after the implementation of QOE policy (see Table 6.1 the first figure in the column of Transition from JHS to RSHS). During the after QOE period, the equalisation effects1 of QOE were becoming more obvious and the differentiation effects of QOE were decreasing as the gap in willingness of investing in students’ CET among families with different cultural capital was decreasing and more and more students with less family cultural capital attended CET. If the trend remained until 2013, the differentiation in willingness of sending children to attend CET among families with different cultural capital could no longer be significant. Similarly, the significant gap in the probability of attending CET between only-child students and students with one sibling in this chapter is consistent with the gap in the transition opportunity from JHS to RSHS between only- child students and students with one siblings during the third QOE periods in Chapter Four. In addition, the result that there is no difference in the probability of attending CET between only-child students and students with more than one sibling is also consistent with the narrowed gap in the transition opportunity from JHS to RSHS between these two groups (see the figure in Row Two, Column One of Table 6.1). This consistence also appears in the effects of father’s job (see the figure in Row Three, Column One of Table 6.1). Therefore, even though Hypothesis 6.2.1 is not confirmed, the explanation for the trend of the effects of father’s education on students’ transition opportunity could still be reasonable.

In addition, for family resources indicated by sibling number, even though I claimed that CET is becoming more affordable, the affordability can usually only be applied to one- child families. Therefore, in the estimation, the significant negative effects of the dummy variable “one sibling” indicates that compared to one-child families, for two-children

1 The explanation of the equalisation and differentiation effects of QOE can be found in the summary and discussion section of Chapter Five. 164

families the total expenditure for tutoring is still too high. However, why are the effects of the dummy variable “two siblings or more” insignificant but positive? In the urban area, the number of families with three children or more is very small (7% in the sample from this chapter). In addition, as I discussed in the summary section of Chapter Five, some of these families with two siblings or more could be poor but other families could be rather rich because they could afford the large fine for violating the one-child policy. Therefore, limited cases and conflicting family resource situation indicated by the variable “two siblings or more” may explain the insignificance of this variable and thus this research focus on the comparison of CET-attending probability between only-child students and students with one sibling.

In Figure 6.3, almost all the potential means of academic achievement of students with different cultural capital in the fourth tutoring time level (attending more than 10 hours of tutorials per week) are smaller than the ones in the Tutoring-Some and Tutoring- Much groups, which could be explained that in the fourth tutoring time level more parents were so desperate for their child’s low examination scores to be improved that they forced their child to attend many tutorials. However, this reason, instead of the effects of tutoring, could also be used to explain the difference in the coefficients of father’s education between the Tutoring-Much level and Tutoring-None (or -Some) level: if students had rather low examination scores, it was possible that parents with less cultural capital were less willing to invest a lot (e.g., fees for more than 10 hours of tutoring) on their child’s tutoring than parents with more cultural capital. This results in the Tutoring-Much group examination scores between students with less family cultural capital and with more family cultural capital being not significantly different (the almost even slope line in Figure 6.3). This explanation is possible; however, it seems contradictory to the slope of the line in the Tutoring > 10 level because in this group the above selection bias of excluding students with less family capital and low examination sores should have been more obvious for the more tutoring time requiring stronger willingness to invest and there should have been a flat or even down slope. However, the current sloped line in the Tutoring > 10 level in Figure 6.3 indicates a positive association between a father’s education and a student’s academic achievement.

165

Therefore, the explanation of tutoring effects suggested in Hypothesis 6.3 seems more reasonable.

In this chapter, I demonstrated that CET could improve students’ academic performances and those students from a family with more resources were more likely to attend this kind of tutoring. On the other hand, it appears that students from a family with less cultural capital could benefit more from tutoring than the ones with more family cultural capital. Is there any tension between the above phenomena? Do the QOE policy and its unexpected consequence CET intensify educational inequality or alleviate it? Did parents feel this tension? If they did, what was their perspective on it? Did they think the current QOE system including CET was just or unjust? In the next chapter, I will describe and discuss parent’s evaluation of the QOE policy, especially its by-product or consequence, i.e. the CET, in terms of social justice.

166

Chapter Seven. QOE, CET and Educational Inequality: Parents’ Justice Evaluation

I. Introduction, Methods and Data

In Chapter Four, I analysed the change in the effects of three family background factors on students’ transition opportunities from junior high school (JHS) to regular senior high school (RSHS) and from RSHS to university before, during and after the implement of Quality-oriented Education (QOE) policy. Only the results of the trend in the effects of family resources (diluted by sibling number) on students’ transition opportunities from JHS to RSHS supported my hypotheses that the QOE policy was associated with the increase in social reproduction of educational inequality. The results of the effects of the other two family-background factors, namely family’s cultural capital (indicated by father’s education) and family’s social status (indicated by father’s ISEI), did not support my hypotheses. I also found that the three factors had different effects on the transition from JHS to RSHS compared to the transition from RSHS to university.

Then, in Chapter Five I used the findings from the interviews with parents, teachers and students and the policy documents to explain the inconsistencies between assumptions and analysis results and the discord of changing trends of the effects of family background factors between RSHS admission and university admission. The explanation suggested that the changes created by the QOE reforms and the unchanged examination-oriented admission method jointly were associated with the popularity of commercial extracurricular tutoring (hereinafter CET), which was then used to explain the above inconsistencies and discord.

In Chapter Six, I used quantitative data again to test three critical hypotheses derived from the explanations proposed in Chapter Five and found (1) attending CET can improve students’ academic performance, (2) whether or not students attend CET is only associated with family resources diluted by students’ sibling number but not with father’s education and occupation at least in 2013 (after implementation of QOE) and

167

(3) attending a certain (substantial) amount of CET is likely to benefit students from lower cultural capital families more than from higher cultural capital families.

The results presented in Chapter Six reveal some disagreements. On the one hand, the educational inequality could increase because under the QOE reform a part of the time in compulsory1, cheap public-school teaching time was replaced by “extracurricular” time used for CET, which parents with more financial resource were more likely to send their children to attend and that could improve attendees’ academic performance. On the other hand, the majority of CET focus on academic subjects with the teaching form of lecturing in a big classroom, which reduces each student’s share of CET cost and thus the CET fees should be not completely unaffordable for lower-middle families with one child. if ambitious underprivileged parents manage to send their children to attend sufficient CET, those children could benefit more from this tutoring than their privileged peers. However, for those most underprivileged students and their families who cannot afford enough CET, is the current so-called QOE with CET a just system for them compared to the previous cheap public-school teaching time? Even though social reproduction of education inequality in terms of the transition from RSHS to university has decreased and it might be easier for them to afford enough CET in RSHS as the standard of “enough” could be much lower in RSHS than in primary school and JHS, the key problem for underprivileged students is many of them are unable to enter the RSHS and therefore enjoy the more equalised transition from RSHS to university because the transition to RSHS is more competitive even after educational expansion (see the comparison of transition rate between the two transitions in Chapter One, p.15). In addition, these underprivileged students might be able to afford some but not enough CET to survive the competition of entering RSHS. Therefore, should the policy makers consider changing the current QOE policy to make it more just (fair)?

1 Even though the RSHS is not part of the compulsory education, there are rarely urban parents who refuse supporting their child to enter RSHS once he/she received a competitive admission offer. In addition, the regular high school tuition is very low compared to the average family income (approximately 2% in Yangzhou City in 2014) and the governments provide underprivileged students with subsidies. 168

When I reviewed the government documents relevant to the QOE policy, I discovered there was rarely any record of official public consultation during the process of designing and implementing the QOE policy or its sub-policies. My research offers an informal evaluation of the QOE policy in terms of educational inequality. Before I discuss the policy implication of my research, it is important to listen to the voice of educational service consumers beyond education officials and experts. Therefore, in this chapter, my research is focused on parents’ evaluation of QOE, especially QOE’s unexpected consequence namely the popularity of CET in terms of social justice. For the topic studied in this chapter, the reason why I did not ask students, the direct consumers of educational services, to discuss their justice evaluation of QOE policy rather than ask their parents, the indirect consumers or the purchasers of educational services, is because their parents have experienced both QOE and the Examination-oriented Education (EOE). The evaluation should be based on the comparison between the two educational systems and use the EOE as a reference to evaluate the changes in the old EOE system caused by the new QOE policy. In addition, parents have more experience of social stratification and inequality and other social institutions and social changes, which is critical for answering my research questions. The research questions I wanted to have answered from talking to parents are as follows: (1) What are parents’ general views of the QOE? Should we have the QOE instead of EOE? (2) What are parents’ views of QOE’s unexpected consequence of the popularity of CET? Is CET replacing part of public-school teaching really necessary? (3) What are parents’ view of CET in terms of educational inequality? Through interviewing parents and asking them the above research questions, I wanted to reveal parents’ concerns and thoughts about the QOE policy. In my view, it is important for policy makers to take the views of parents into account when considering appropriate responses to educational inequality.

The interview method was utilised for the research presented in this chapter as in the previous qualitative chapter, Chapter Five, and the data is part of the production from the field work described in Chapter Five. Therefore, here I will not repeat the description of methods and data of Chapter Five; however, in this chapter I only present and analyse

169

interviews with 22 parents including 2 interviewees with dual roles as a parent and a school teacher.

II. Results

Parents’ general evaluation of QOE Real QOE is necessary, but it is not practical now QOE reform was derived from critiques of the heavy academic burden on students and the overemphasis on academic performance in the examination-oriented education system in the early 1990s. The parent interviewees supported the reform direction despite criticising its implementation, especially the aspect of implementing the policy to try and alleviate students’ academic burden. Jian’s mother: Too painful, absolutely painful, more painful than in my school-days I feel. From my school-days they had started saying we should reform the (EOE) education, should alleviate the academic burden … now there is heavier and heavier burden on children. Their schoolbags are heavier than ours, aren’t they? They spend more time doing homework in the evening than we did.

Not only were their children suffering from the lingering of the old EOE, parents were suffering as well. Jia’s mother: According to parents I contacted, no matter whether their children’s academic performance was good or poor, they all felt anxious… One of my colleagues, her child has a very high academic ranking in the grade. My colleague is still very anxious, worrying about that the score in one subject may be not 100% enough for her child to enter that top RSHS. Then what did she do? She coaxed and pestered her child unceasingly until her child agreed to attend another extracurricular tutorial. In short, parents are always in the struggle with children. Put the pressure on our children and make ourselves anxious.

170

Besides their anxiety concerning how to improve their children’s academic performance, many parents, mainly mothers, also spent lots of after-work time accompanying their children doing homework in the evening and attending extracurricular tutoring at weekends. Under this extent of pressure “plus working in the daytime, children are very tired and parents are very tired as well” (Jia’s mother). Therefore, some mothers just gave up their own career or chose a low-paying but flexible job. Yang’s mother: Parents’ time is mainly spent on children. Just take me for example. I have given up a lot of jobs which are more suitable for me. Why did I choose this job? That’s because it gave me time to take care of my son. Because we, uh, a family’s common goal is for the child, to bring him up to success. A family must sacrifice one; sacrifice one to bring him up.

It should be noted that all three students, namely Jian, Jia and Yang, had no siblings and were already in their second year of JHS; however, taking care of them and supporting their learning were still time-consuming for their mothers, which in some extreme cases could impact on their parents’ relationship. For example, You’s mother had told her son that “(When necessary) I would give up everything for you, even your father”.

Besides alleviating student’s academic burden, the “real” QOE reform was supposed to encourage students’ all-around development. In my interview, all the parents acknowledged that besides the academic achievement or, more accurately, the scores of academic subjects1 in admission examinations, there were some other important qualities that should also receive attention during their children’s education. Even though different interviewees emphasised different qualities, all these qualities are included in the popular slogan “morality, intelligence, physique, aesthetic, all-round development”, which is part of the propaganda for the QOE policy and mentioned with different phrasings in the policy documents (CCCPC and State Council 2010; State

1As I mentioned in Chapter Five, in recent years, the total score of the senior high school admission examination (hereinafter SHSAE) usually includes a small proportion of physical education (e.g., approximately 6% in 2016 SHS admission of Nanjing but more than 80% of students achieved the full PE score) while the college admission examination only consists of academic subjects.

171

Education Commission 1993), and sometimes these qualities also include housework procedures (Chen 1998; He 1992).

According to the then Vice Premier Li’s (2000) explanation1 of the QOE policy, it includes basic qualities, which every student should have, and individual strong points and potentials, which could be different from other students. It is not surprising that compulsory education should take the responsibility of improving students’ basic qualities. Actually, the majority of the QOE policy documents are focused on school education. As to the specific measures of these policies, the teachers I interviewed claimed their schools increased or at least stopped misappropriating the class time of physical education for academic subjects, which has been confirmed by parents interviewed as well. However, the art courses have not been that lucky. Yu’s mother: Physical education uh... because PE is one of subjects in the senior high school admission examination, the PE teacher can be relatively tough. You (academic subject teachers) can’t misappropriate my PE class time. The other teachers (in music and fine arts) don’t have that (privilege) ... (My daughter) said she couldn’t remember she had attended a music or fine arts lesson.

Yu was a second-year JHS student. The SHSAE would arrive approximately one year later (the end of third year in JHS) and it is so competitive that only approximately half of JHS in Jiangsu Province in 2015 could enter RSHS. However, because of the nine-year compulsory education almost all the primary school students can enter JHS. Therefore, primary schools, parents and students themselves can allow students to enjoy more time for physical education and art education. At this stage, some schools not only make efforts to improve students’ basic qualities but also try to find and develop the “strong points and potentials” for each student. According to Teacher Mrs Zhao, her school

1 This explanation derives from a speech of government report delivered by the then Vice Premier Li Lanqing during the third National Conference on Educational Work, which is the top conference on education policies and was held five times from 1985 to 2016 by the central committee of the Communist Party of China and the State Council. Therefore, this explanation should be the most accurate one for what the policy makers want the QOE policy to be.

172

organised several “interest classes” (courses, groups or clubs) and students could attend one or more depending on their interests.

A journalist to whom I spoke informally told me of a similar story that she had reported (Dai and Li 2016). This measure was not just one school’s individual pioneering project but was a nationwide program in accordance with the order from the Ministry of Education named “Sport, Art 2+1 Program” as a sub-policy to implement QOE (COME 2011). This program requires primary school and JHS to organise different kinds of courses and activities to ensure that every student becomes skilled in at least two sport techniques and one art technique. The order document contains an appendix listing approximately 60 sport techniques and 20 art techniques. It seems impossible to require every school to offer courses or activities for all these techniques; however, if the purpose of this policy is to find and develop “individual strong points and potentials” as the policy states, each student should have enough elective sport and art courses to try and decide which are their strong points or potentials. The difficulty in offering that many elective courses does not require too much research effort to reveal. One of the parents I interviewed acknowledged that “one interest course may have a very large enrolment while another may have a very small enrolment. There will be a resource allocation problem, which is too complicated and systematic for school” (Lin’s father). Regardless of the order from the top education authority of China, in reality “a few schools may organise some interest classes or activities while the others may just dismiss students early and just let students go to their own (private) extracurricular tutorials” (Teacher Mrs Zhao). However, eventually most students just go to academic extracurricular tutorials instead of the sport or art tutorials. In my assessment, the financial issue is likely to be responsible for the difference between the above two kinds of schools. In China, the revenue of a public primary school or JHS mainly come from the financial appropriation of town-level and county-level governments (Sun and Du 2010), and thus these governments’ final revenue scales usually determine whether these schools have enough resources to offer enough elective sport and art courses. One school in Dai and Li’s (2016) report was one of the best public primary schools in a very rich county-level city of Eastern China and it could offer more than ten sports and art interest groups. However, in a nearby “not-that-rich” town, Teacher Mr Chu said his

173

school just asked “the music teacher to direct a chorus and the fine arts teacher to teach the folk watercolour, and in afternoons to arrange a little bit more sports activities. Every year the school arranges two cultural festivals. That’s enough to write a report (to the educational bureau)”. Another “not-that-rich” school had a “clever” strategy, which was to focus its resources on tennis so that a large number of students learned and played tennis (according to Teacher Mrs. Feng’s description) regardless of whether tennis was their strong points or potentials. Therefore, tennis became the school’s “strong point and potential”, which could be used for publicity but was not consistent with the original intention of the “Sport, Art 2+1 Program” policy.

Since not every school can offer enough courses to meet all students’ interests, some parents step in. Nan’s mother: The development of qualities is not completely the school’s responsibility. At weekends, you have two days at home, one day for learning, in the other day your parents definitely can take you to learn other things, such as piano, go to the museum, or visit other places... It’s not like our school must take care of everything of your child. Otherwise what would parents do? … For schools, they can only conduct a general, I mean, the quality education suitable for a relatively large number of students.

What Nan’s mother described seems a perfect collaboration between school and family working together to achieve the goal of QOE. However, even though almost all interviewed parents sent their children to attend some art or sport extracurricular tutorial in the early grades of primary school, when they approached or entered JHS they only or mainly attended tutorials in academic subjects and the reason parents gave for this was that the SHSAE was approaching.

Even though schools have increased sport and art education time and more parents send their child to attend art and sport extracurricular tutoring, the interviewed parents still did not think what their children were receiving was the real QOE. Some parents’ comments were very blunt.

174

Min’s mother: Now what the school mainly cares is the examination score, only score... But student’s interest, all-around development, I feel, in this aspect, they did not make any effort, or do anything. The score decides the school’s reputation or school leaders’ promotion. Yu’s mother: The quality-oriented education is just a proposal made by politicians. (It has) never been really implemented. Shan’s mother: Our education is exactly examination-oriented education. You must have a high examination score… The academic burden has already been very heavy. They said they would alleviate it, but nothing happened. My daughter’s schoolbag, dozens of kilograms! Jian’s mother: The policy is just a policy. The actual implementation is not like that.

Parents blamed school or education authorities for not fully or even partially implementing the QOE policy while Nan’s mother as a RSHS teacher thought parents were responsible for the unsuccessful implementation of QOE policy. Nan’s mother: The atmosphere is parents will vote with their feet. In fact, what parents most care is whether the school’s teaching is good or not, whether your students can have a high admission examination score, how many of your students can enter a top regular senior high school… Parents almost won’t ask other things. Your school’s activities are good; will I choose your school? No. In this atmosphere, there is no way for schools (to conduct the QOE) because schools serve these parents.

Nan’s mother claimed parents only cared about their children’s academic achievement or, more precisely, the senior high school admission results and many parents sent their children to extracurricular tutoring that was focused on academic subjects and occupied lots of children’s recreation time that should have been spent in sport, art or other non- academic activities. However, the parents I interviewed did want their children to have all-round development. In the above description, I implied that the competitive RSHS admission, which is oriented to academic examination, was the key reason for “QOE

175

descending to just a slogan” (Jia’s mother), but did parents really want to change this kind of admission?

Why not change the EOE admission? Because it’s just. In my interview, I described some suggestions as to how the current academic examination-oriented admission system could be changed, such as the application system for admission in America with personal statements, recommendations, certifications of sport, art or other specialities, and social activities besides the standardised examination results. However, most parents expressed some serious concerns about those proposals.

First, they praised the objectivity of the old EOE admission that was based solely on the standardised examination and thought the admission method I described was too subjective. Chang’s father: You think Qi Baishi’s (an influential Chinese painter) painting is better. He thinks Zhang Daqian’s (another painter) painting is better. Well, the third person may think another one’s painting is better. There isn’t a uniform standard. Yu’s mother: If the admission depended on the application documents (without examination), it would be entirely decided by personal judgment, well, by school leaders’ judgement. Well just according to their own subjective consciousness, my certificate is useful, but your certificate is not useful, then China will be more chaotic.

However, in the current Chinese and English examinations students still need to write essays, which can be seen as a subjective method as well. Actually, if I expand the quotations from the interviews with Chang’s father and Yu’s mother, what they are really concerned about was revealed. Chang’s father: Under the current (social) environment, there may be no another way to admit students. Because there are too many factors, if there wasn’t a uniform measure, someone who has connections, who has a background ...

176

Yu’s mother: Because the Chinese society is a connection society … I think this (application-admission) system would be a chaos if it was applied in China. Those really common children would not dream of entering Peking University or Tsinghua University. But I think as long as there are humans there will be all kinds of connections. I didn’t apply for a foreign university, perhaps it is not as pure as I image.

At least these essays in the senior high school and university admission examinations are marked anonymously while the application has much more possibility of being affected by social connections. A lot of parents I interviewed had a similar opinion. Jia’s mother as a JHS teacher remembered some proposals made by policy makers were strongly opposed by parents. Jia’s mother: Yes, there were some policy makers who wanted to change this (examination-oriented admission) system, but opposition to it was very fierce, which mainly came from parents. They feel that this current system is still rather fair. If you changed it, for example, as someone suggested that we could employ the western-style application-admission system, using personal statement or something like that… but the opposition voice is very strong. It is because any good thing, when we implement it, will lose its credibility because of our own selfish interest. I think it’s very normal that parents oppose it.

Current education system is just because of the examination-oriented admission Since the examination-oriented admission system has not been changed, parent interviewees were then only asked to evaluate the current education system in terms of social justice rather than the QOE system in theory. The majority of parents evaluated the current education system, no matter whether its nature is the QOE or EOE system, as “just” (Yang’s mother) or “comparatively the most just” (Jia’s mother) or at least “not unjust” (Min’s mother) based on their general impression or personal experience, for example “we haven’t encounter any unjust situation (in terms of my son’s education)” (Yi’s father). The only exception was You’s mother who thought the current

177

examination-oriented admission was “in favour of girls” while her child was a boy. However, the reason why most parent interviewees evaluated the current educational system as being just was because of the unchanged examination-oriented admission. Yu’s mother: Education is the most just industry in China. … It (the admission) only depends on the examination scores. Isn’t it the most just one? The “black case work” or other things such as depending on the connections, or parents’ ability are relatively few.

Parents’ justice evaluation of CET: the justice principle of need Parents thought they and their children had to endure the academic examination- oriented admission because they needed a “gong ping” (公平, just/fair) admission system, which is the core of the justice of the education system. “gong ping” is the word I used in the interview because promoting “gong ping” is written in the top educational platform (CCCPC and State Council 2010) by the central government and widely propagandized and should be more familiar to parents than other similar Chinese words. I will discuss what parent interviewees mean by “gong ping” later in this chapter: whether it is a kind of justice achieved through equity or equality. But before that, I report parent’s justice evaluation of QOE’s unexpected consequence, i.e., the popularity of CET, which appears to be damaging the justice of the QOE admission system. In Chapter Five, I discussed that the above admission system is one of the major reasons for the popularity of the privately funded extracurricular tutoring and in Chapter Six the quantitative findings support that the probability of attending CET is associated with families’ financial resources per child and meanwhile attending extracurricular tutoring can improve student’s academic performance. Therefore, what do parents think of CET? Is it just, especially for the underprivileged students who cannot afford CET?

“CET is just, at least not unjust”, but is this true? All of the parent interviewees sent their children to these privately funded extracurricular tutorials teaching academic subjects and a few of them also sent their children to attend non-academic CET. When I asked some of them whether the extracurricular tutoring was useful or not, they gave an affirmative reply even though a

178

few emphasised its effects also depended on children’s “learning willingness”. Every parent claimed their family could afford the tutorial fees and among them only three parents (two from the lower-middle socioeconomic group, namely Yang’s mother and Shan’s mother) mentioned the fees were pretty expensive. When I mentioned that many low SES families might not be able to afford the tuition fees, parents did not deny this. However, when I continued to ask parents whether or not they thought it was unjust for students from low SES families, they did not agree with this, which seems contradictory to their perceived “reality” and needs further exploring. During the interviews, parents explained their views, which I will describe and discuss later in this section. Before that, I present a typological model with two dimensions (“standardised vs. personalised” and “public vs. private funding”) to summarise parents’ evaluation of CET under the current so-called “quality-oriented education” system. Even though this model reveals two aspects of key arguments for its justice from parent interviewees: (1) arguing for CET in the current so-called QOE system by using the justice principle of need and (2) arguing against applying the justice principle of equality into CET, in this section, I first report how parent interviewees argued for the necessity of CET. This also reveals where parent interviewees drew their conclusions from, even though it might be not true.

Standardised versus personalised qualities: what to teach In its initial design, the QOE policy tried to take into account both the common basic qualities of students and the diversity of individual strong points and potentials to achieve all-around development of students. In the policy documents, the QOE would be for all students (“mian xiang quan ti xue sheng”, 面向全体学生) and be adapted to each student’s aptitude (“yin cai shi jiao”, 因材施教) (State Education Commission 1994, 1997; CCCPC and State Council 1999). In the above paragraphs, I described the conflict between these two different goals and they formed the first dimension in my typological model, namely “standardised – personalised”. It should be noted that the QOE policy did not always take a balanced position but shifted its focus in different sub- policies, which I will discuss later. To achieve the above goals, the QOE policy documents described the principles containing what to teach and how to teach. The “standardised” end of this dimension indicates that the teaching content and methods should be suitable for all students while the ideal “personalised” end indicates that each student

179

should have the custom-built teaching content and methods in accordance with each student’s aptitude. It is easy to understand that the possible changes in teaching methodology are too professional to be realised by parents if the changes do exist. Therefore, I mainly discuss the teaching content in this chapter. The conflict between the standardised and personalised teaching content was also reflected by parents’ opinion on whether or not the school-based extracurricular tutoring should replace CET besides the normal at-school teaching. Jian’s mother supported this idea, but Lin’s father emphasised that different children might have different needs in terms of extracurricular tutoring. Lin’s father: You need to attend Math extracurricular tutorials; I need to attend Physics tutorials; he may need to attend some sport training; she may need to attend Literature tutorials… If no one was allowed to attend, that would be like returning to the school teaching (evening lessons), right? We should allow a diversified system.

Actually, besides CET, some parents also strongly support private school by emphasising the difference between students and the right of choice. Private schools could be seen as the expansion of extracurricular tutoring from the after-school time into the at-school time. You’s mother: If everyone was not allowed to choose school, you just went to the school you should go and the educational resources were distributed equally, in fact this would deprive us of the right of choice. There should be more private schools or other forms of school because I know the difference between people is large, so I think schools should be diversified, suitable for different people.

Public versus private funding: who pays for education expenditure? Education expenditure is always a focus for most parents. Since the September semester of 2008, primary schools and JHSs have been free to attend. Therefore, in my interview the normal school tuition was not discussed in great detail while the extracurricular tutorial fees were still a concern for some interviewees. Even though interviewed parents said they could afford the private extracurricular tutoring fees, some parents

180

still wanted the formal schools rather than the private CET institutions to take more responsibility for extracurricular tutoring as they experienced in their childhood. All of them attended public schools including government schools, “dan wei”’s (单位, socialist work unit) schools and rural local council’s (village) schools as their childhood was in the period before the introduction of private schools or commercial tutorial institutions. Jian’s mother: When we were students, we had interest classes, right? How could we have any commercial extracurricular tutorials? Never heard of it, right? It (the popularity of CET) is the problem of school education. You could offer some gifted education courses, which will solve the problem, right? Why must (the school) put (these courses) into the market? Why did they increase the burden on students’ parents? These tutorials are very expensive.

However, as discussed in the above section, the school-based extracurricular tutoring could be a heavy financial burden for the governments especially when the governments want to meet different kinds of students’ quality development, which was mentioned by one interviewed couple who emphasised that the current market economy was different from the previous planned economy. Shan’s father: It is impossible that the governments provide extra funding. Even if they were paid the normal salary rate to teach the extracurricular tutorials, teachers would be reluctant. Shan’s mother: That’s true. Obviously, they (school teachers) could have been earned more salary (in tutorial companies) while in your place (public school) it (the salary) is almost like working for free. It is impossible. I think it is now a market economy society. China has stopped talking about communism.

Some parents whose occupations are businessmen or businesswomen opposed such a suggestion that the government should ban CETs and then public schools should take over this tutoring. Lin’s father: This is a little opposite to the recent direction of the national government’s policy. The government wants to deregulate, let it be,

181

as long as the society has this demand. Because this is a reasonable and legitimate thing, it should be guided, should be encouraged.

Besides the extracurricular tutoring, a few parents thought private funding should play a more important role in the entire education system. You’s mother: In terms of the good educational resource I should be allowed to use money to get it. If I want to enter a good school, I will pay a little more tuition. If I were a poor student, I would apply for a scholarship. Isn’t that good? I think it also reflects the harmony of society. If you have the ability, you can make more contribution. If you have no money, you can only have this road (applying for a scholarship). I think it could be like that. Should the rich people go to hell? The (educational) system should have a single model (examination-oriented admission).

Different, even opposing, views on whether the basic education should rely on public funding or private funding occurred frequently during the interviews. The topics did not only involve extracurricular tutoring, but also included private schools, school-selecting1 and subsidies to underprivileged students. The reason why I describe the above conflicting views among parents is to reveal the dimension of public-private funding from parents’ evaluation of the current education policies. The “public funding” end of this dimension indicates that expenditures on student’s at-school and after-school education are funded by governments 2 while the “private funding” component indicates that these expenditures are funded by students’ family members, usually their parents.

1 School-selecting means that parents send their children to enter a school outside their school district. The parent-selected school is not necessarily a private school but charges a heavy extra fee and sometimes it requires students to attend a special admission evaluation. 2 The socialist “dan wei” (work unit) system ended in the 1990s and county-level governments were required to take the major financial responsibility for the village schools in the early 2000s, nowadays public schools in China are jointly funded by different level governments from central governments to town-level governments. 182

Four categories, EOE versus QOE, and unnecessary inequality From the above description and analysis, I identified two dimensions as being the core elements of parents’ evaluation of the current education policies, namely the dimension of standardised-personalised qualities and the dimension of public-private funding. With these two dimensions, I formed an orthogonal typological display (see Table 7.1). This typology table reveals four categories of education activities. Since the current compulsory education in China is almost free and the curricular teaching and learning activities are regulated under the academic benchmark1, which is issued by the Ministry of Education, I only describe and analysis the extracurricular activities, which are also the focus of this thesis. In addition, as I have described in Chapter Five and in the beginning of this section, the current extracurricular activities are influenced by both QOE and EOE. Therefore, in each category, I will describe examples for both QOE and EOE activities even though some of them may not be close enough to the ideal type.

Table 7.1 Typology of Education Services under Parents’ Evaluation Goal: Adapt teaching to each student’s aptitude (因材施教) Standardised qualities Personalised qualities Category 1: “opportunity Category 3: “utopian equity” Goal: equality” e.g. longer tutoring e.g. need-based extracurricular Public Provide time at public school (evening tutoring at public schools funding education classes in the senior years) (catching-up or gifted after-school for tutorial; “Sport, Art 2+1 Program”) students of Category 4: “unnecessary Category 2: “differentiation” different Private inequality” e.g. market-based e.g. market-based need-based 有 2 classes ( funding schoolised extracurricular extracurricular tutoring (one-to- 教无类) tutoring (CET) one private tutoring)

The first category is the extracurricular activities to develop students’ standardised qualities with public funding. A typical measure to achieve this goal is to have free public schools admit all students and then to extend the at-school time to teach students the

1 The current Version 2011 for the academic benchmark of compulsory education was issued in February 2012 by the MOE and implemented in the September semester of 2012. 2 I use the term “schoolised” to describe the phenomena that CET institution operates like a normal school and use the same teaching method as the normal school, which is not personalized but lecturing about two dozen students at the same time. 183

same curricular content utilising the same methodology. An example that closely approximates this measure is the evening class that lasts for approximately one-hour after the normal school time ending of 5 pm and is usually designed for all Year 9 students in a JHS. This was described by Jia’s mother who is a JHS teacher. Usually in the additional at-school time these students need to continue learning the academic content to be prepared for the approaching SHSAE, which could be used to exemplify the EOE. In other grades, student may also be asked or encouraged to attend an after- school class or activity organised by the school. For example, the tennis class described above by Teacher Mrs. Feng is a school-wide sport course, which could be used in the school report to the education authority as a highlight of the school’s achievement in implementing QOE. In Category 1 of the typology table (Figure 7.1) every student has almost an equal opportunity to access the same educational goods, which I label as “opportunity equality”.

The second category is opposite the “opportunity equality” category and aims to develop each student’s personal aptitudes with their own private funding. A typical measure is to take personal need-based extracurricular tutorials paid by their parents. In my interview, some privileged parents told me they employed a tutor for their child to provide a one-on-one tutorial after school or at weekends, with the tutorial content being either academic or non-academic. For example, You’s mother and Ran’s mother separately employed experienced former school teachers to tutor their children in academic subjects while Jun’s mother sent her child to a famous painter to learn painting. However, it is obvious that these one-on-one private tutorials are too expensive to be enjoyed by underprivileged students. If this kind of extracurricular private tutorials took a major role in student’s education, the “differentiation” of educational achievement between students with different family backgrounds would be inevitable. Thus, Category 2 is labelled as “differentiation”.

The third category describes an ideal extracurricular educational arrangement for students, which develops each student’s individual strong points and potential and is free under public funding. This category also reflects two traditional Confucian educational standpoints or ideals, namely “provide education for students of all

184

different classes” (“you jiao wu lei”, 有教无类1) and “adapt teaching to each student’s aptitude” (“yin cai shi jiao”, 因材施教) (Liang and Sun 2009). The two goals were also reflected in the QOE policy documents that the QOE would be for all students (“mian xiang quan ti xue sheng”, 面向全体学生) and be adapted to each student’s aptitude (“yin cai shi jiao”, 因材施教) (State Education Commission 1994, 1997; CCCPC and State Council 1999; Li 2000) as I mentioned above. However, it should be noted that the “for all students” in the policy documents is more concerned with students who have different levels of academic performance (Yu 2008) even though academic performance is usually associated with students’ socioeconomic backgrounds. To achieve these two ideals, free public schools are supposed to design an individual extracurricular activity plan for each student based on his or her personal needs, such as gifted, catching-up and special educational programs in academics, sports and art subjects. However, it appears a utopian proposal as designing and delivering so many need-based programs is too budget-consuming to be affordable for the current Chinese governments. Therefore, I labelled Category 3 as “utopian equity”. In reality, there are some examples that show signs of the “utopian equity”. During the EOE era, students were allocated to key classes and non-key classes based on their previous academic achievements. The key classes had teachers2 with more experience who ran the teaching content faster or in greater depth than non-key classes. Sometimes the key-class students could attend extracurricular tutorials for Mathematical Olympiad or other special academic competitions. However, these classes did not charge students more tuition. Therefore, the key-class system could be viewed as a kind of publicly funded educational activity based on student’s academic quality even though it only had a rough classification of students’ academic qualities and did not design a “personalised” teaching program for each student. However, the key-class system was seriously criticised by some parents

1 The literal meaning of “you jiao wu lei” (有教无类) is to provide education without excluding anyone in a particular category. There are different definitions for the “lei” (类, category), including the rich or poor, the noble or ordinary, the talented or mediocre, the good or bad, and so on. I chose the definitions concerning the social stratification because “you jiao wu lei” (有教无类) is opposite to “xue zai guan fu” (学在官府) and means to bring education from the public noble school to the ordinary people (Yu 1987: 25-26; Meng et al. 2010: 57-71). 2 Chinese schools usually enrol a number of students, and thus in each subject and in each grade, there are usually more than one teacher with different levels of experience.

185

for its “examination-orientation” and “some students could feel frustrated and self- abased if they are allocated to the non-key class” (Xiang’s mother). Therefore, according to Yu’s mother even though her child’s school still established some key classes, this arrangement was secret and is in violation of the current QOE policies. A QOE example for the “utopian equity” could be the “Sport, Art 2+1 Program” I described above. Attending this program is free but it is financially impossible for every school to offer 80 sport and art tutorials in the program list to meet the different aptitudes of each students.

The extracurricular educational activities in the last category use private funding to develop some basic qualities that each student should have, which are opposite to the “utopian equity” category. The current CET institutions or cram schools perfectly exemplify this category. These institutions mainly offer tutorials consistent with normal schools’ academic subjects. Some of them also offer some special courses such as Mathematical Olympiad, painting or instrumental performance; however, these courses are not to develop students’ standardised qualities required by the academic benchmark, and thus will be excluded from this category. As I quoted from parent interviewees’ descriptions in Chapter Five, these tutorial institutions usually operate similar to a normal school with a teacher delivering a lecture to a few dozen students in a classroom. Especially during a summer or winter break, they look similar to running an intensive short school semester. For example, one of the top five tutorial institutions offer courses previewing the content of academic subjects of the following semester in a continuing seven-day period over a winter break and according to my interviewees their children and many of their children’s classmates have attended this kind of courses.

A minority of students may need catching-up tutorials; however, it seems a wasteful duplication of effort for the society that so many students, even some with very good academic performances, need to preview the learning content in a summer or winter commercial tutorial, which will be taught by school teachers after the summer or winter break. This learning content is standardised and should be taught to every student. Some interviewees such as Dan’s mother and Teacher Mrs Zhao revealed one of the critical reasons for this was that the at-school teaching and learning time was not

186

enough for the admission examination, which I have described in Chapter Five. However, not every family can afford their children’s CET fees and as I mentioned in Chapter Five a challenge for these underprivileged students is that the at-school teaching is quicker than it should be because many students have previewed the teaching content and the teaching time is limited. Therefore, underprivileged students will be worse off in the admission examination, which is my conclusion from the previous chapters. What I do want to emphasise here is under the current QOE policy and the unchanged examination-oriented admission system it seems inevitable that many parents send their children to these commercial tutorial institutions to survive in the admission examination.

However, the major reason they choose commercial tutorial institutions is not because these institutions have better teaching staff but the normal pubic school cannot offer such services due to the academic burden-alleviation regulation of the QOE policy. To date, I have not seen a research article claiming the commercial instructions on average have better teaching staff than the normal public schools. Even if they did, the governments could use lots of measures including financial and administrative ones to attract these teachers to public schools. Even one parent (Jian’s mother) questioned why the current QOE policy encouraged the privately funded institutions instead of the public schools to develop students’ standardised quality as I described above. If these normal public schools really want to or are allowed to run extracurricular tutoring, they could even replace many CET companies to offer different level courses including catching-up and advanced courses. This is because the student load in an urban school is usually very large and in one grade a course is usually taught by several teachers for several classes and it is easy for these teachers to offer three or four different levels of extracurricular tutorials as elective courses. After all, a famous nationwide tutorial company only offers four different levels of tutorials in the Grade 8 Mathematics course and each tutorial has 15–30 enrolment positions rather than the form of one-on-one personalised teaching. Of course, even if the public school ran some extracurricular tutorials for all students, some very privileged parents may still prefer employing very experienced tutors to run one-on-one tutorials at their own expense, but at least the underprivileged students would not be excluded from the extracurricular learning

187

activities. Therefore, I label the current privately funded CET as “unnecessary inequality” (the label for Category 4).

Why didn’t parents think CET was an unnecessary inequality? From the above section, I identify that as one of the most important consequences of QOE the current CET system looks like a type of “unnecessary inequality” based on the two important criteria of parents to evaluate the current QOE policy, which are also affected by Chinese Confucian tradition. However, when I asked parents whether CET was a kind of inequality, the majority of parents except one did not think so or at least did not agree that the inequality was unnecessary or strongly criticised it. Before I describe and analyse their explanations, I first discuss the meaning of “gong ping” (公平), the word I used during the interview. There are several English words whose meanings are similar to “gong ping”, such as equality, equity and justice, and therefore I need to clarify what the parents meant in English by talking about “gong ping” even though I have revealed the answer above by using the words “unnecessary inequality”.

The meaning of “gong ping” (公平): equality When interviewees were talking about “gong ping” (公平) or justice1, what exactly was they referring to – equality or equity or both2? Even though some researchers claim that the “gong ping” in education should have the same meaning as equity (Cui 2013; Chu and Yang 2008), most parents in my interview emphasised or implied the “quantitative property” of “gong ping”, which is consistent with the “equality” rather than with the “equity” since the latter contains a “qualitative property” (Secada 1989: 82). Lin’s father directly elaborated on “gong ping” ’s quantitative property when I asked his view on the claim that the extracurricular tutoring could result in inequality owing to the expensive fees.

1 Since justness and fairness are today often used interchangeably, I will not discuss the deeper conceptual difference between them as Andre and Velasquez (1990) did, but just use one of them. 2 The reason why I only discuss the difference between those two terms is because my research can be viewed as a study in the distributive justice of educational resources and the main basis of distributive justice “must be” equality and equity (Espinoza 2007). In other words, both equity and equality aim to promote justice. 188

Lin’s father: If so, then this would be very simple. Could we just audit every family’s wealth and portion out all the money equally? You have ten dollars, I have five, he has three, and the total is eighteen dollars. Then each of us gets six dollars. Can we do this? We can’t.

Only You’s mother emphasised the “qualitative property” of “gong ping” on one occasion when she strongly complained about the “gong ping” of the current examination-oriented admission system because she thought that the female students had some biological edge that enabled them to do better in examinations than male students, even though they were given the same examination papers (her child was a boy). However, on another occasion, You’s mother used the same meaning of “gong ping” as other parents did. Therefore, according to the parents that I interviewed, the “gong ping” in my description and analysis refers to “equality”. Returning to the key question of this section, why didn’t parents think that CET is an unnecessary inequality?

Emphasise the role of children’s initiative and endeavour in academic achievement Even though they acknowledged CET could help improve their children’s academic performance, parents emphasised that children’s initiative and endeavour played a more important role. For example, Min’s mother told a typical inspirational news story about a child from an underprivileged family. Min’s mother: There was been a news report. A child, whose parents were greengrocers1, had never attended any extracurricular tutorials, but his academic performance ranked first in every discipline. In contrast, she continued to talk about her own child that “while my child has attended all different kinds of extracurricular tutorials since she was little and I also tutored her (in English), (her academic performance is) still not good.” Through this comparison she made a conclusion that, “I don’t think it is correct that once you have the economic condition to send your child to attend extracurricular tutoring, your child will be better than others at academic performance. I don’t think so. It still depends on the child's ability in learning and understanding.” There are several similar expressions from other

1 In China, a greengrocer only runs a very small business in a bazaar and usually has a low socio-economic status. 189

parents in my interview and their arguments were based on the “evidences” from the news media, TV shows, and old proverbs that students who attended extracurricular tutoring did not necessarily have better academic performance.

One parent went further to express a strong opinion, which could be seen as victim- blaming. You’s mother: There are successful people whose parents are rich as well while there are also successful people who started from nothing. We can't let everyone enjoy the same goods. For some people whose IQ is not normal, or who has extremist ideas, there is nothing you can do to help them. If he doesn't work hard to fit into the society, it is his own problem.

It’s not certain that whether other privileged parents held back a similar opinion because of political correctness; however, in recent years the topic of so-called “hating the riches” is rather popular and discussion on this topic has even appeared in several academic journal articles (e.g., Zhang 2014; Hao 2014).

Emphasise parents’ effort Besides children’s effort, the interviewees also emphasised the role of parents’ efforts in their children’s academic achievement. In response to questions about whether CET fees were too expensive to be affordable for some underprivileged parents and could result in inequality, several interviewees claimed if these parents had made more effort to allocate their income towards their children’s education, they could have afforded the fees. Yu’s mother: You say it (extracurricular tutoring) is expensive … In fact, for Chinese families nowadays, we only have one child; most families only have one child. The majority of family's expenditure is spent on the child.

Several other parents thought the key issue was not about the income but the “notion” that “parents should pay lots attention to children’s learning process” (Si’s mother and Ping’s mother). These parents suggested that with this notion underprivileged families

190

could find other cheap approaches to help their children such as “reading newspaper” (Si’s mother) and “self-studying with the help of online learning materials” (Yi’s father).

However, the emphasis on parents’ own efforts went further again to criticising those underprivileged parents who did not work hard enough to earn more money, from which seems a sense of victim blaming as well. Jun’s mother: Well, how can I say it? There has been the Reform and Opening-up up and economic development since Deng Xiaoping (late 1970s), right? Some people who dared to take risks became rich. Some people who were conservative, who didn’t dare to take risks, fell behind. There is no way to change. Those people, who only considered playing mah-jong, did not study hard, can only have a low wage. It must be that the early bird has a worm to eat. Ping’s mother: But there are a lot of things you can achieve if you try hard. There may be no much development space for my parents because of that era. But we had more development space. We have made life better through our own efforts and then we can let them (our children) live better. I think this is just and reasonable.

It was interesting to hear parent interviewees mentioning the change of social context as they are the eyewitness of the change from Mao’s era to the Reform and Opening- up era and they also expressed their optimism about the upward social mobility even though recently there is more pessimistic discussion about the phenomenon that “a poor and humble family cannot turn their child into an upper-class person any longer”1.

There is no absolute equality at all and you can go to public school Besides asserting that equality can be achieved by students’ and parents’ own efforts, parent interviewees acknowledged that equality in education did exist but justified it by

1 The wide discussion derived from an online blog article titled “In our era a poor and humble family cannot turn their children into an upper-class person any longer” (我们这个时代, 寒门再难出贵子), which was published on the online forum website “Tianya Community” (www.tianya.cn) in 2013; However, I cannot find the original source, but a reprinted article can be viewed online (http://bbs.tianya.cn/post- free-3443970-1.shtml). This article stirred up lots of follow-up articles in newspapers and on the internet. 191

claiming that “there is no absolute equality at all” (Xiang’s mother). One parent used the unsuccessful communism trail in Mao’s era to demonstrate her argument. Yu’s Mother: In any society, there is a hierarchy (stratification), just like the difference between different people. It doesn't matter whether it's good or bad. You want it (the society) to become a completely good one, to become a communist society. That's impossible. It is to ignore the difference. This is a violation of the (social) law. Nan’s mother also mentioned the communistic society but thought it was a beautiful and very long-term goal. However, different from other parents I interviewed, You’s mother strongly criticized people who tried to pursue the absolute equality. You’s Mother: I think for people who want equality there is something wrong with their personalities because there is no absolute equality.... That is to say, we should all learn to adapt to the society.... I told my children “there is no absolute equality, only the relative one. You play your part; you do what you should do.” Through denying the existence of absolute equality, You’s mother further lowered her standard of equality and justified buying the educational admission opportunity. You’s Mother: I think it is okay because he has paid for it (admission opportunity). He has to spend his time earning money. When you spend your time learning, others spend their time earning money. That is the only difference.

Even though other parents did not set such a low bar standard of equality for education, the belief of inexistence of absolute equality, especially the failure of Mao’s egalitarianism communistic trail1, gave them a good reason for feeling at ease and justified when they encountered my question that some underprivileged children could not afford the extracurricular tutoring fees.

1 One relevant research is Lu (2006) discussion on the relationship between the collectivization movement in Mao’s era and the making of peasant egalitarianism mentality.

192

In addition, the free and “not bad” public schools, which I have described in the introduction chapter, are the basis of parent interviewees’ claim that “I think this free tuition policy is rather just. That is the equality for all the children” (Ping’s mother). Therefore, every child can enter a public school in the compulsory education stage and after compulsory education, the tuition of RSHS is not expensive at all 1. And more importantly, even some of underprivileged students cannot attend extracurricular tutorials, Si’s mother believed that “school teachers’ teaching is enough for children” and at least “in a good school their teachers’ ability is rather strong. If you can absorb all the content taught by school teachers, it will be enough for admission”. As Jian’s mother said, “in the current Chinese education system, the teaching level of most public schools is pretty good” and “the number of private junior and senior high school is pretty small, pretty small”. Therefore, some parents thought the public school is the basis of the “relative” educational equality.

Compared with other social fields, education is relatively equal The relativity of equality in education is not only in contrast to absolute equality, but also with the level of equality in other social fields. Therefore, even though some parents acknowledged that the opportunity of attending CET was not completely equal, but the CET’s context, the educational filed was relatively equal compared to other social fields, which could more or less offset the inequality of CET. Yu’s mother: (after attending CET,) you still need to attend examinations. Compared with other industries in China, education is relatively equal. It is because of the examination. Everyone has to take an examination. Isn’t it the most equal system to use exam scores (to admit student)? The covert operations or other things, such as relying on connections, relying on parents’ ability, are relatively less (in the education industry).

1 For example, in Yangzhou City, the tuition for a semester of a key RSHS in 2014 is CN¥840 (approximately AU$185) which has stagnated for 15 years while the average per capita disposable annual income reached CN¥24,167(approximately AU$5300) in 2014.

193

Yu’s mother only singled out the examination system as the key evidence for her argument of the relative equality of education. It should be noted that the equality in her example of the examination system may contain the meaning of procedural justice and it is not surprising that many parents think the procedural justice is a kind of equality that every student in one admission district answers the same examination paper. However, here I want to emphasise that parent interviewees made a conclusion of relative educational equality by comparison to other social fields.

Other things can more or less offset educational inequality caused by CET The above description and analysis is focused on “relative”, but no matter how relative it is, CET is still a kind of inequality. How did parents respond to this unpleasant idea? Based on the results from the interviews, I summarise the following institutional arrangements or elements in education, which parents I interviewed thought could at least to some extent offset the inequality caused by CET.

The first institutional arrangement is the subsidy and scholarship system; however, parents such as Ping’s mother and You’s mother were talking about the effects of subsidy and scholarship on the whole education’s equality because CET institutions do not have a similar arrangement. However, the existence of subsidy and scholarship may relieve parents’ feelings about educational inequality in general, which also has an indirect positive effect on their feelings about the inequality of extracurricular tutoring. A similar indirect effect may also result from the second institutional arrangement namely examination-oriented admission system, which I have already discussed above and do not provide more information on here.

Besides the educational institutional arrangements, Si’s mother added that, “I feel that some very low-income families, parents are very mentally tough, which is helpful for their child in the future.” This view reflects the popular proverb that “children from poor families grow up faster” and is expressed in many literatures and art works including American dream movies. The popularity of this view might be able to relieve the sense of educational inequality as well.

194

Equality still very important, but stress personal effects rather than social welfare Even though they did not strongly criticise the unnecessary inequality and in a sense accepted the Chinese version of “whatever is, is right” (Yi’s father), they still thought that equality should be the goal in society: Yi’s father: All men are created unequal. (The researcher: Do you think whether our society needs to pursue equality as a goal?) Of course, it is necessary, and each of us is pursuing equality. Parent interviewees especially emphasised the importance of educational equality: Ran’s mother: I am not an expert in this area, but my general feeling is that China now is talking about its goals such as being rich, strong, democratic, etc. and the equality is also one of the core values of our country. So what is the embodiment of equality? It’s that everyone equally enjoy education resources, isn’t it? Now the rural children’s education resources, those children from poor families will have no promising future. You can’t. You don’t have the resources. That is the inequality.

However, as I presented above, many parent interviewees mainly stressed parents’ efforts: Ping’s mother: We have made life better through our own efforts and then we can let them (our children) live better. I think this is just and reasonable.

Even though some parents mentioned the governmental responsibility, they set a very low standard for social welfare. Most parent interviewees thought a free public school plus a little subsidy for the basic living cost was enough for underprivileged students. However, when I mentioned the extracurricular tutoring fees, parent interviewees thought it was unnecessary: Dan’s mother: It is not good to give them the extra money for the extracurricular tutoring. Anyway, I think it is pretty enough not to ask them pay school tuitions. I think the (subsidy) standard is almost like that. Therefore, their standard of equality is consistent with their perspectives that “there isn’t absolute equality”.

195

III. Summary and Discussion

In this chapter I present parents’ evaluation of the QOE policy and its unexpected consequence CET, especially in terms of social justice. By researching parents’ view, I would like to reveal their concerns for the future QOE policy reform to provide a healthy and equal educational service to their children. The findings show that parents criticise the old EOE system, especially its heavy academic burden on young students. However, the ideal QOE presented in the government policy document paper has not been truly implemented because of the limited resources of ordinary public schools and, more importantly, the almost unchanged examination-oriented admission method. However, it is this unchanged examination-oriented admission system that makes parents still evaluate the current so-called “QOE” system as being just. Nevertheless, the QOE policy did bring some changes to the education system and one of the most important is the popularity of CET. Actually, the almost unchanged examination-oriented admission method is probably one of the key reasons behind the popularity of CET. Compared to the free compulsory education and low-cost public RSHS, CET is much more expensive and causes new educational inequality since the aim of attending CET is to improve educational performance. The parents who I interviewed could afford the tutorial fees and they acknowledged that these fees are too high to be affordable for many low- income families, which could make these families’ children worse off in academic performance and admission opportunities. Since the other components of QOE have not significantly changed the social reproduction of educational inequality, the research is more focused on the changed component, namely the popularity of CET, to find out parent’s justice evaluation of CET and its association with parent’s justice evaluation of the current “QOE” in general.

Interestingly, parent interviewees did not think the unaffordable CET confronted by students from low income families is not “gong ping” (公平, just), which is consistent with their general evaluation of the current “QOE”. The main reasons suggested by parents include (1) CET can meet the needs of developing students’ personalised qualities especially sport and art qualities and it can also meet the personalised need of strengthening or recuperating performance in one or some academic subjects and (2)

196

the government is not likely to fund the extracurricular tutoring, especially a wide variety of tutorials for personalised need. Through analysing parent interviewees’ arguments, I identified two dimensions among parents’ evaluation of the current “QOE”, especially its consequence the CET: one is standardised vs. personalised quality and the other one is public vs. private funding. The intersection of the above two dimensions forms a four-category typology and the current CET in a few academic subjects with the form of lecturing one or two dozen students is located in the category of using private funding to develop standardised qualities. This was labelled as “unnecessary inequality” because the free public school can meet this need using the old “key class, ordinary class” system during extracurricular time.

Since schools are prohibited from employing the “key class, ordinary class” system and, more importantly, from running their own tutorials during extracurricular time, it is understandable that parents think that CET is not “unnecessary”. However, why do parent interviewees think it is not a kind of “inequality”? I identify a few reasons frequently mentioned by parent interviewees: (1) interviewees inclined to use one or two examples (or maybe exceptions) to argue that success mainly depends on students’ own initiative and endeavour; (2) they blame parents who cannot afford the tutorial fees for not making enough effort; (3) they think public schools have provided the basic equality and people should not expect absolute equality; (4) they think because of the examination-oriented admission, the current education system is relatively equal, at least in terms of procedural justice, which could more or less offset the inequality in the opportunity of attending CET; and (5) other things such as the subsidy and scholarship system and children becoming strong from hardship essentially offset the educational inequality. Even though parent interviewees still welcome an equal society, especially in the aspect of education, their overemphasising on personal efforts for educational achievement and low standard of equality may encourage the “unnecessary inequality”.

Even though my description and analysis of “unnecessary inequality” is focused on CET, some parent interviewee also mentioned and supported private schools with the argument that different students may adapt to different schools. However, the question is that, besides resources, what other difference could exist between the public and

197

private schools under the city-wide or province-wide examination-oriented admission system in China? A similar question could be proposed if the policy makers want to find out whether CET could be replaced by school-based extracurricular tutoring. If the teaching methods utilised by the private school teachers could prepare students better for the admission examinations, the public-school teachers could employ these methods as well or those private school teachers could be hired by public schools. After all, the public and private schools have similar teaching environments, namely a big classroom containing a few dozen students and even the best private school cannot provide one- on-one teaching service. Nowadays, many private schools in China are so-called “transformation schools” (“zhuan zhi xue xiao”, 转制学校). After the transformation, these schools are still owned by the governments but are managed by their own school headmaster team. It should be noted that even before the transformation, these schools have usually been top public schools with a high transition rate into the next-level school stage (Zhou and Mao 2007). What the transformation changed was the tuition increase rather than these schools’ EOE teaching methodology or their consistent pursuing of a high transition rate. Actually, it is some junior high transformation schools that are blamed for intensifying students’ academic burden because they select students via different kinds of admission evaluation methods, and thus encourage parents to send their children to lots of CET (Yang 2005) as I described in Chapter Five. In summary, no matter how different they are, all students except some attending foreign language schools for studying abroad in the further have to learn the same textbooks and confront the same examinations, which mean standardised qualities. Therefore, the difference between students is not a justified reason for promoting the development of private school in the current EOE admission system.

One limitation of the research in this chapter is that the parent interviewees are from lower-middle to upper-middle socioeconomic groups. It was not possible to recruit parents from the lowest socioeconomic group, in particular those who cannot afford CET fees were not recruited. These parents’ voices may also be hard to be heard during a possible public consultation for further policy reform. It may not be necessary due to the suggestion from one gatekeeper of my field work that these parents are not educated enough and do not to attach enough importance to their children’s education.

198

The reason could be that these parents are busy with their livelihood and do not have the leisure to attend some policy consultation that their children cannot directly benefit from because they will graduate very soon. Therefore, in my future research or a relevant public consultation on how to recruit these parents is still a big challenge.

The policy implication will be discussed in Chapter Nine, but before that I will summarize the findings of the whole thesis and compare my research findings with previous ones in Chapter Eight.

199

Chapter Eight. General Discussion

In this chapter, I discuss the findings of my research to illustrate how these findings can answer the research questions of my thesis and discuss my contribution to scientific knowledge. First, I summarise the major findings from both the quantitative and qualitative research for the two major research questions of my thesis: (1) Does the Quality-oriented Education (QOE) policy influence the social reproduction of educational inequality? If so, what is the mechanism? If not, why did the effects not happen as was expected? (2) How do parents evaluate the mechanism? Is it a just mechanism? Second, I discuss the relationship between my research findings and previous literature to clarify how my results integrate with previous literature in this field and what my contribution to knowledge is. It should be noted that as I mentioned in the first introduction, my research is focused on the urban area of China, and thus the findings summarised and discussed in this chapter are only relevant to education in the urban area, except as otherwise prescribed.

I. Summary

My original hypothesis on the influence of the QOE policy on social reproduction of educational inequality was that the policy should benefit privileged students more than underprivileged students and based on the increased inequality parents would evaluate the QOE policy as unjust. In terms of the mechanism, I assumed that by increasing students’ extracurricular or, more accurately, at-home time, and giving more attention to the development of students’ art and sport qualities, the QOE policy should increase the influence of family cultural capital, financial resources and social status. However, the results of hypothesis testing revealed that the trends in the effects of these three different family background factors on students’ transition opportunities from junior high school (JHS) to regular senior high school (RSHS) and from RSHS to university are complicated (see Table 4.4 and Table 4.6 in Chapter Four). In contrast to the hypothesised effects of QOE, there is not a continuous upward trend of social reproduction of educational inequality over the three implementation periods of QOE

200

policy namely before, during and after QOE periods. Further, the trend of social reproduction of educational inequality in the transition opportunity into RSHS is different from the one into university. The most surprising result was that the effects of families cultural capital on students’ transition opportunities into university significantly decreased with the implementation of QOE (benefiting those families with lower cultural capital), which is not only contrary to the upward trend of family resource- dilution effects (disadvantaging those families with more children) on students’ transition opportunities into RSHS, but also directly opposite to the research assumption (see Chapter Four, Hypothesis 4.2). In the context that income inequality has been sharply increasing in mainland China over the last decades1, this surprising finding is contrary to common sense and needs further explanation. In terms of parents’ evaluation, even though I did not show the above analysis results to parents before I interviewed them, most parent interviewees did view the current education system in the name of QOE as being relatively just (see Chapter Seven), which was opposite to my original assumption.

To explain these discrepancies between the hypotheses and results in terms of social reproduction of educational inequality (see Chapter Four), I utilised the qualitative findings extracted from my interviews with JHS students and their parents, and teachers from primary school, JHS, RSHS and from government documents. Although the current datasets I have access to are too limited to test all the hypotheses deriving from the explanations from my field work, some crucial explanations are supported by quantitative analyses (see Chapter Six) following the field work (see Chapter Five). There are four explanations for those discrepancies based on my field work: (1) differences between the QOE policy goals in the text and its implementation and the unexpected consequence, the popularity of academic commercial extracurricular tutoring (hereinafter CET) (the implementation gap and academic CET), (2) different effects of different family background factors (family background effects), (3) differences between JHS and RSHS (JHS vs. RSHS) and (4) alternative contemporaneous policies or social changes that could have affected the policy outcomes (the impact of other social and

1 See Li’s (2014b) review on the Gini Coefficient of China. 201

policy changes). In terms of the discrepancies in parent’s evaluation of QOE between the assumptions and findings, parents have directly explained their own reasons for their evaluation (see Chapter Seven) and I find most reasons are related to the above four explanations, especially to the implementation gap and academic CET, which will be summarised in the following paragraphs.

Implementation gap and the popularity of CET, especially the academic CET According to my field work reported in Chapter Five, the implementation of four key measures (also the sub-goals) of QOE reform (alleviating student’s academic burden, adding more non-academic courses into the curriculum, adopting new teaching and learning methods and changing the examination-oriented admission approach) did not achieve the QOE policy goals of changing the examination-oriented education, or improve student’s all-around qualities and develop their creativity, problem-solving skills and lifelong learning attitudes. Even though the implementation of QOE policy did not achieve the policy goals, it did make some changes to students’ learning lives both in school and at home, which was reflected by the popularity of CET. Extracurricular tutoring has a long history in China; however, its popularity and becoming a prosperous large industry accompanied the implementation of the QOE policy. According to my research, CET played an important role in the change of social reproduction of educational inequality, especially in explaining the discrepancies mentioned in the above paragraph. However, it should be noted that the popularity of CET is an unexpected consequence of the discrepancy between the QOE policy goals and its implementation.

The unsuccessful implementation of QOE policy makes CET feasible and necessary. In terms of feasibility, as one of the key measures of QOE, the academic burden alleviation policy increases students’ out-of-school time. Almost at the same time as the alleviation policy began being implemented, the general labour policy of shortening working hours took effect in schools and removed the compulsory Saturday school day. These policies created some room in students’ calendars to allow CET to infiltrate because the policies could only alleviate the academic burden on students assigned by schools; however, they have no ability in controlling parents to assign even more academic burden on their

202

children during their extracurricular time. In addition, although the “calendar room” created by the QOE policy does not necessarily need to be filled by CET, other activities are often not available, i.e., (1) school organised tutoring is forbidden by educational authorities, (2) many parents are too busy to accompany or supervise their children (mainly when their children are at an early school age) because of working overtime and high female labour force participation, and (3) it is often difficult to organise a peer activity because peers have different tutoring schedules and other social policies or changes such as one-child policy also make the potential study or play mates become less.

The most important reason is that the QOE policy has not achieved one of its goals of changing the examination-oriented admission approach into RSHS and university, which should be the key to the success of QOE reform, and it is necessary for students to attend CET to become well prepared for the admission examinations. The admission approach reform is always on the agenda of education policy makers and has resulted in several adjustments (Yan 2011). However, every time a new adjustment has been proposed, it leads to a nationwide controversy about the justice of the new adjustment, which has resulted in the examination-oriented admission system (widely perceived as “just and fair” due to its procedural justice) barely changing (Yan 2011; Shi 2016; Sargent et al. 2011; Zhu 2009). Therefore, despite changes in the education system, admission to RSHS and university is still focused on academic subjects rather than the all-around development of moral, intelligence, physical and aesthetical qualities that the QOE policy advocates. In terms of compulsory education (primary school and JHS), even though the school district system1 allows students to enter primary school and JHS without examination, in the name of QOE some “good” schools could organise a special admission to recruit a few extra students and the certificates (including for sports, art, advanced English and Mathematical Olympiad 2 ) from CET institutions or other

1 In the system starting in the late 1980s before the implementation of the QOE policy, a school-age student usually enters the school nearest to his/her home without undertaking an admission examination. 2 These courses are not taught by normal schools or are not advanced enough in the normal school’s curriculum. The special admission usually pays more attention to student’s achievement in advanced English and Mathematical Olympiad, but the sport and/or art achievements can be a bonus. Therefore, the school can make an argument that the special admission is a pilot practice of QOE. 203

organisations could be the critical application materials. Even though only a limited number of students can enter these “good” schools via special admission, the majority of parents do not want to abandon this opportunity. More importantly, when their children get used to CET, it is easy to switch CET from courses for the special admission to the ones for normal RSHS and university admissions when they feel their children cannot pass the special admission and the first normal admission examination, i.e., the senior high school admission examination, is approaching, and thus in the later primary school age and JHS age, most CETs that students attend are relevant to academic subjects. Furthermore, the special admission for student’s non-academic “strengths” may benefit privileged students more; however, as it has existed (with different titles) since 1984 (before the implementation of QOE) (Jiang 2006), its effects on social reproduction of educational inequality should be rather stable while my research is to seek a factor resulting in some changes. This is different from my original assumption that non-academic CET could play an important role in students’ learning activities, which would probably result in social reproduction of educational inequality (see the introduction section of Chapter Four). The fact that the majority of CET focus on academic subjects has a significant but different implication for social reproduction of educational inequality. Academic CETs usually are less expensive than non-academic CETs due to the differences in teaching form and learning equipment. Therefore, even though it could still be unfordable for the lowest socioeconomic families, the total expenditure on attending academic CET could be affordable for many lower-middle families with one child after cutting down living expenditure on non-education aspects due to a strong educational aspiration, which is different from my original assumption (see the introduction section of Chapter Four).

Attending CET could help students achieve better academic performance, which has been demonstrated in Chapter Six (Hypothesis 6.1), and thus students could be well prepared for admission examinations. This could not be only because learning is more efficient with the help of professional tutors rather than under parents’ supervision or by oneself but also because of the changes created by the curriculum content and teaching and learning method reform of the QOE policy. According to my field work reported in Chapter Five, these changes include that: (1) schools have less time to

204

prepare students for admission examinations due to the shorter at-school time in general and more time allocated for non-academic courses and activities under the curriculum reform, and (2) students need extra help to adopt to the dissonance between the new QOE textbooks, which are designed for new QOE teaching and learning methods, and the old teaching and learning methods used in reality, which are more efficient for preparing students for admission examinations than the new methods are. Therefore, when the changes brought by the curriculum content and teaching and learning method reform of the QOE policy damage the school’s ability to prepare students for RSHS and university admission examinations, even many school teachers encourage their students to attend CET, which in turn ensures that CET becomes popular.

Family background effects and their different associations with CET The popularity of CET is associated with the unsuccessful implementation of QOE policy, but how can this association explain the different trends in effects of three family background factors on students’ educational transition opportunities? Based on the results from the field work in Chapter Five, I propose the explanation that because these three family background factors have different implications (further explained below), their associations with CET are different, and through the mediator effects of CET between family background factors and student educational transition opportunity, the three family background factors’ associations with student’s educational transition opportunity become different. I will only summarise the explanations for the differences in these factors’ effects on the transition from JHS to RSHS at first and in the next section, I will explain why and how the association between family background factors and CET changes after students enter RSHS owing to the difference between JHS and RSHS.

In terms of family’s cultural capital indicated by the father’s education, its association with CET reflects parents’ willingness to send their children to attend CET. Chinese parents are influenced by the traditional culture emphasising education (Hu 2014; Bray 2006) and have the desire for upward social mobility (Hu 2014; Li 2015), and thus the willingness to invest in students’ education will be always rather strong and there is no evidence that it has changed significantly across the three QOE policy periods. However, if CET fees are still too high to be affordable, even after the family has cut down living

205

expenditure on non-educational aspects, that willingness to invest in CET could be suppressed. The changes associated with parents’ willingness to invest in CET coincided with the implementation of the QOE policy, where the common form of extracurricular tutoring changed from a private family tutorial usually run by one tutor to a large-scale and school-like institution with a large enrolment. Many institutions have branches in different cities. CET became industrialised and formed a competitive market, which further lower the unit price for one-hour CET, and thus compared to one-on-one family tutoring in the early QOE policy years, it became more affordable in the later QOE policy period (see Chapter Five, “what is commercial extracurricular tutoring”). Then, many lower cultural capital parents’ willingness to invest in CET was no longer suppressed, which disconfirmed the null hypothesis that students from a family with more cultural capital are more likely to attend CET in JHS in 2013 (one year in the third QOE period) (see Chapter Six, Hypothesis 6.2.1, see results in Table 6.7).

How can the above theory explain the trend of effects of family cultural capital on children’s transition opportunities from JHS to RSHS? Attending CET could help students from lower family cultural capital catch up with their peers from higher family cultural capital, and to counteract the increasing educational disadvantage of the lower cultural capital environment at home by increasing out-of-school time as the implementation of alleviating academic burden policy in the QOE reform. However, in primary school and JHS, the out-of-school time might increase too much and lower family cultural capital parents’ willingness to invest may be not suppressed by the fees for one or two-hour tutorials; however, when their children need more than five-hour tutorials to catch up with their peers, that willingness could be suppressed because they may be able to afford the fees for one-hour tutorial but not for more than 5-hour tutorials. Therefore, even though the pattern does insignificantly display that students from higher family cultural capital obtain greater advantage when CET emerges only during the second QOE period, yet that advantage decreases when CET becomes popular enough to be affordable during the third QOE period, the effects of family cultural capital on RSHS admission opportunities do not significantly vary across the three QOE periods. This explanation has been supported by an interaction analysis between three different tutorial time levels and student’s father’s education on student’s academic achievement

206

in Chapter Six (Hypothesis 6.3, see results in Table 6.9 and Figure 6.3). This analysis showed that compared with the student group who do not attend any tutorial, the effects of family cultural capital on academic achievement in the student group attending 5 to 10-hour tutorials per week is significantly smaller; however, there is no significant difference in the student group attending no more than 5-hour tutorials per week.

The estimation of students’ transition probability across three QOE periods in Chapter Four could not find a suitable dataset to include a variable that was able to directly measure family financial resources, no mention to the one calculated at constant price. However, the effects of family financial resources can be reflected by the mechanism of family resource dilution by including the variable of students’ sibling numbers in estimations. Students’ sibling number is also associated with whether to attend CET, where only-child students are more likely to attend CET than students with one sibling (see Chapter Six, Hypothesis 6.2.2). Before CET became popular, students with a sibling had a higher probability of entering RSHS than only-child students did, which could be explained that the former could receive tutoring from their elder sibling or be inspired by sibling competition in academic achievement (see Chapter Five, “Lack of peer activities”). However, when CET emerged and became popular, the advantage of students with a sibling disappeared and there was shown to be an insignificant trend that only-child students were gaining an advantage (see the Figure in Colum 1 Row 2 in Table 4.4 of Chapter Four). This is because one-child families do not need to dilute family resources among several children and even an only-child student from a lower-middle income family could afford CET (see Chapter Five, “Financial resources”), which would be more obvious in the third QOE period because the average income of Chinese family has kept increasing during this period. After basic living needs have been met, the need for education is strong, which could be deduced from Maslow's (1987) hierarchy of needs.

The effects of social status (occupational prestige or class category) indicated by a father’s ISEI on a student’s transition opportunity from JHS to RSHS forms an inverted V pattern between the three QOE periods. One possible explanation for this strange

207

(Maslow 1987) pattern could be that the slight increase between the first and second period is due to the “symbolic value” (Baudrillard 1998) of CET that was believed to be a luxury of higher social classes when CET just emerged during the second QOE period, but as CET became more popular in the third QOE period and many lower-middle family students started to attend CET, the “symbolic value” was lost, and thus the effects of occupational status on RSHS admissions decreased (see Chapter Five, “Social status (prestige)” ). This explanation is supported by one of my hypothesis testings, which suggested that there is no significant difference in the probability of attending CET between students having fathers with different occupations estimated using data collected after the implementation of QOE (see Chapter Six, Hypothesis 6.2.3).

JHS vs. RSHS and effects of their difference on family background-CET association The above summary focused on the variation tendencies of the effects of family background factors on students’ transition opportunities from JHS to RSHS across three QOE periods; however, those tendencies are different from the ones of students’ transition opportunities from RSHS to university. There are two basic differences between JHS and RSHS, which probably explain the above tendency differences (see Chapter 5, “Difference in QOE between compulsory education and RSHS”). The first difference involves a student’s learning life in the JHS stage (maybe including primary school as the earlier stage of compulsory education) and in the RSHS stage. Compared to students in compulsory education, RSHS students legally stay longer at school under the QOE policy even without considering the popular phenomenon of “voluntarily” staying late in the evening or “voluntarily” going to school at weekends. The longer at- school time means less margin of the increased extracurricular time, and thus less “room” for CET to influence admission. Less extracurricular time in RSHS also means the “room” for the gap of attended CET hours between students from various levels of family background factors should be smaller in RSHS than in primary and JHS. Second, compared to students in JHS, RSHS students on average had a higher level of family background factors in RSHS, which is due to the low RSHS admission rate rejecting many students from underprivileged families. Therefore, due to the association between family background factors and attending CET, the higher average level of family background factors in these factors probably mean that compared to primary and JHS,

208

in RSHS more students from the “lower level” (a higher standard of “lower” in RSHS than in JHS) of family background could attend CET and the gap of attended CET hours between students from various levels of family background factors should be smaller. Even though there is no suitable dataset for testing the above hypotheses on the difference in the association between attended CET hours and family background factor between JHS and RSHS, I assumed that the above reasoning could explain why the effects of a father’s ISEI were not significant. In terms of effects of cultural capital on a student’s university admission probability, which could be more sensitive even to a slight increase of CET in RSHS, since more and more students from lower family cultural capital could attend CET in RSHS during the second and third QOE periods, the benefit of attending CET could help these students counteract the disadvantage of lacking support when learning at home during out-of-school times. This is one of the disadvantages these students suffered from before the implementation of QOE. In addition, as I discussed above, the extracurricular time “room” for the existence of CET should be smaller in RSHS. Therefore, compared to students in JHS, students only need to afford a few hours of CET, besides that their families have higher paying capability on average. Therefore, CET attending willingness of families with lower cultural capital in RSHS are less suppressed and their children are more likely to take advantage of CET and attend sufficient CET (as I discussed above, the amount of attended CET hours is meaningful) to catch up with other peers, and thus in RSHS the positive effects of CET improving those students’ admission opportunities became significant and the effects of the family cultural capital gap on students’ transition opportunities from RSHS to university decreases as CET became more and more popular during the second and third QOE periods.

Impact of other social and policy changes According to the above explanation, the gap in the transition opportunity from RSHS to university between only-child students and students with one sibling should not have changed significantly across the three QOE policy periods similar to the effects of fathers’ ISEI; however, the result is that the gap increases between the first and second periods but then decreases between the second and third period. What happened during the second period? As I mentioned in the analytical framework of Chapter Three, my thesis

209

is not making the argument that the QOE policy is the exclusive factor explaining the change in social reproduction of educational inequality during the last three decades, and thus in Chapter Four the estimation was particularly to control the effects of one concurrent policy change, namely educational expansion. However, from the results of the literature review, the educational expansion cannot explain why the trends in effects of different family background factors on educational outcomes are different. I sought other social and policy changes, which possibly have a significant effect on the association between sibling number and student transition opportunity from RSHS to university. The increased university tuition fees could be the critical factor. As described in the introduction chapter, there were several one-off upsurges in university tuition fees between 1997 and 2005 as a part of the industrialisation and expansion policy of higher education and most upsurges happened in the second QOE policy period (1999– 2005, the year of transition from RSHS to university). In the third QOE period, the family income kept increasing but each university tuition upsurge was merely a one-off rather than indexed and had been frozen for several years due to the intervention from the central government in 2006 (Chen and Zong 2014). These social and policy changes relevant to university tuitions could explain why the odd gap in the transition opportunity from RSHS to university between only-child students and students with one sibling increased during the second QOE period and then e dropped during the third period.

“No absolute equality”: Parents’ justice evaluation of QOE and its consequence CET According to the above summary, the inequality in the transition opportunity from RSHS to university by family cultural capital decreases across the three QOE policy periods as attending sufficient CET could help students from lower family cultural capital catch up with their privileged peers in terms of academic performance and the sufficient amount of CET time required in RSHS is not too large to be affordable for students with lower family cultural capital and thus does not suppress their families’ willingness of sending them to attend CET. However, the QOE policy created more extracurricular time in primary school and JHS than in RSHS, and thus the standard of sufficient amount of CET hours required in primary school and JHS could be higher than in RSHS and thus more students with lower family cultural capital cannot attend sufficient CET since their

210

willingness of attending sufficient CET could be suppressed by the expensive fees due to their families’ limited payment capacity. That’s because one-hour of CET may be affordable for many lower-middle income families but when their children need to attend more than five-hours of CET to catch up to their privileged peers in JHS, CET fees could be too expensive to be affordable even if their families are willing to send them to attend, not to mention that the lowest income family cannot even afford one-hour of CET. Therefore, the positive effects of attending CET on the academic achievement of students with lower family cultural capital student are insignificant in a general trend in the transition opportunity inequality from JHS to RSHS by family cultural capital. These underprivileged students cannot benefit from CET and they could actually be worse off after CET started to become popular during and after the implementation of QOE since the other students could now do better in admission examinations with the help of sufficient CET. This mechanism is more clearly displayed in the effects of family resources dilution mechanism on students’ transition opportunities from JHS to RSHS since family resources diluted by multiple children are more directly related to family’s payment capability for children’s CET fees. Compared to only-child students, students with one sibling are worse off after CET started to become popular during and after the implementation of QOE. According to the above summary, this differentiation in attending CET between students with different family backgrounds is more obvious in JHS than RSHS and it is a perceived reality of inequality by JHS parents during the interviews. However, even though part of free compulsory education teaching time was replaced by rather expensive CET owing to the QOE policy and many underprivileged students could be worse off, why did the parent interviewees still evaluate the QOE policy and its unexpected consequence CET as being just in terms of its effects on social reproduction of educational inequality?

First, in terms of CET’s policy context that the QOE policy has shortened the at-school time but retained the old examination-oriented admission system, parents acknowledge that the QOE, especially its measure of alleviating student’s heavy academic burden, is necessary but not practical because the admission system is almost unchanged and is still examination-oriented as any proposed new adjustment could lead to a nationwide controversy about its justice. In addition, the majority of parents have concerns for the

211

proposed admission method reform (such as interviewing and application with personal statements and certificates for extracurricular activities) being subjective and procedurally unjust. The popularity of CET is the unexpected consequence of this policy context and parents’ evaluation of CET in terms of justice could be influenced by their justice evaluation of CET’s policy context, i.e., the QOE policy.

In terms of CET itself, some parents’ justice evaluations could not be based on the “reality” of CET. In Chapter Seven, through analysing parent interviewees’ arguments of justifying CET itself, I identified two dimensions of CET: one is standardised-personalised quality (what to teach) and the other one is public-private funding (who pays the fees). The intersection of the above two dimensions forms a four-category typology. According to my analysis, even though the QOE policy advocated developing each student’s strong points (personalised quality), the implementation of the QOE policy did not successfully change the old admission approach with standardised examinations (standardised quality). The implementation of the QOE policy successfully limited or damaged the public school’s ability of preparing students for the standardised admission examination; however, most of the current CET institutions operate just like a normal public school, teach courses within the official curriculum and prepare students for the official standardised educational admission examination. CET is requiring private funding (parents paying for CET) to develop standardised qualities (academic subjects in the official educational admission examinations), and thus I labelled this as “unnecessary inequality”. The “unnecessary” means that the public school should have been able to meet students’ needs to be well prepared for the standardised admission examination1 if such an ability were not limited or damaged by the QOE policy. The “inequality” means that compared to the free public-school teaching, attending sufficient CET (more than five hours per week in JHS in my analysis) is too expensive for many underprivileged students and they could be worse off in admission examinations. However, my analysis of the justice of CET is different from parents’ justice evaluation in terms of its necessity of function and equality of affordability.

1 Students do have different learning abilities, but public schools could have used the old “key class” system or one subject with different level elective courses in the extracurricular time, which ironically are what most CET institutions are using, to meet different needs of students with different learning abilities. 212

Parents argue that CET is necessary because it can meet the personalised and diversified needs of students in developing specific sport or art qualities. Even though sport and art CETs are usually for early school age children and most students usually attended or switched to attending academic CET later, especially when the admission examination was approaching, parents do seek the justification from the existence of sport and art CETs since the public school is not likely to provide a wide variety of sport and art courses for students with different interests. Another personalised and diversified need that parents argue only CET can meet is strengthening or recuperating students’ performances in a specific academic subject. Without considering that most of the current CETs are not designed to meet such a personalised need, I have argued that the needs of most students could have been met by free or cheap public-school teaching1. However, in reality the current public-school system under the QOE policy does give many parents the impression that the public school could not meet such a “personalised” need. Therefore, parents think it is just and fair that CET for personalised need should be paid for by private funding and use the justice principle of need or “that which is real is reasonable” to justify CET based on the above arguments.

In terms of inequality, parents still think that CET is just or at least within an acceptable injustice with the following arguments: (1) they incline to use one or two examples (or maybe exceptions) to overemphasise the role of students’ own efforts in admission success and blame parents who cannot provide their children with enough financial support for not making enough effort, and (2) the fading importance, or even stigma, of communist egalitarianism makes parents justify some extent of inequality by arguing that absolute equality has never existed and that the free public school and other public support has provided basic equality. These arguments are consistent with parents’ general attitudes to equality in society in that parents agree that equality is important for society and meanwhile emphasise the personal efforts to achieve a better life and are not in favour of high social welfare.

1 See the footnote in the previous page. 213

II. Relationship to Previous Research

In this section, I relate my research findings to the previous literature to illustrate my contribution to knowledge, i.e., supporting, contradicting and supplementing previous research. Each sub-section corresponds to each results chapter (from Chapters Four to Seven) for the convenience of reading.

Trend in social reproduction of educational inequality and effects of QOE Previous studies on the educational inequality during the Reform and Opening-up period contained different strategies for dividing cohort groups, different samples (from the urban or general population) and/or different combinations of family socioeconomic factors, and thus demonstrated different or even opposite findings. Therefore, since my cohort dividing strategy was based on the year of implementation of QOE, which has not been investigated in any previous research, it is not surprising that the findings of my study are different to previous research. Since other studies used samples from the general population and achieved a similar conclusion that there was an increasing educational equality between the rural and urban areas (Wu 2013; Li 2010, 2014a; Liu 2006; Tam and Jiang 2015), I only compared my findings with Li’s (2006) research, which was also focused on the urban sample, to illustrate my contribution to this research topic. It should be noted that Li (2006) compared the admission opportunity inequality involving senior high school and university admission between the early (1971–1991, educational transition year) and late (1992–2003) Reform and Opening-up period while my study was focused on the late Reform and Opening-up period and investigates RSHS admissions between three QOE periods from 1991 to 2008 and university admissions from 1994 to 2011. Even though my study covers a later cohort and excludes the VHS students from both admissions, with the assistance of Li’s (2006) research the whole picture of the urban educational inequality during the Reform and Opening-up period can be viewed more clearly.

Li’s (2006) research suggested that the opportunity inequality in senior high school admission resulting from family cultural capital (father’s education) and class category (father’s occupation) decreased between the early and late Reform and Opening-up

214

period and Li (2006) explained that because of the popularity of senior high school education (including RSHS and all types of vocational senior high schools) the “battlefield” of social reproduction of educational inequality moved to the admission of higher education. However, after controlling for the effects of educational expansion, my research revealed that during the late Reform and Opening-up period (the before and during QOE period in my research) the transition from JHS to senior high school was still a battlefield but the RSHS admission opportunity was more competitive than the whole senior high school education including senior VHS. Since the explanatory variables in my research included sibling number, which was not controlled for in Li’s (2006) research, I found that in the battlefield the key weapon is not the family cultural capital or class status but the focused family resources, namely whether the family could focus family resources on one child or not.

In terms of the transition opportunity from RSHS to university, even though Li’s (2006) research did not exclude the vocational senior high school students as my study did, both studies have similar findings that the effects of family cultural capital on the university admission decreased. Li’s (2006) research also found that the opportunity inequality in university admission resulting from class category (father’s occupation) increased between the early and late Reform and Opening-up periods, which was mainly due to the quick increase in the advantage of students from the management class over students form the blue-collar class (the blue-collar class is the reference, the advantage of the ordinary white-collar class and the professionals only slightly decreased). Li (2006) suggested that the reason for the increased advantage of the management class was that the marketisation of economic reform significantly increased the management class’s financial resources in addition to their abundant power and network resources. However, it is doubtful that the general social survey, where Li’s (2006) data came from, included wealthy entrepreneurs or the politicians who are so powerful that can “affect” (e.g., by bribing or by using connections) the “gao kao” (university admission examination), which is such a crucial national examination that the maximum term of imprisonment for cheating in this examination is seven years (SCNPC2015b). Therefore, it is likely that the ranking of the survey respondents of management class should be not too high in the company or government hierarchy. If those parents’ resources are not

215

abundant enough to affect their children’s university admission opportunity, then what other factor could result in their increasing advantage? The two theories proposed in my study on the associations between sibling number and CET and between CET and university tuition upsurge may be part of the explanation. The family resources dilution effects indicated by sibling number increased during the second QOE period in my research (1999-2005) compared to the previous period. This period partly overlapped the late Reform and Opening-up period (1992-2003) in Li’s (2006) research. My research found that the only-child students had more advantage not only in university admissions owing to the university tuition upsurges, but also in RSHS admission owing to the popularity of CET during the second QOE period. Since Li (2006) did not exclude vocational senior high school students from the analysis on students’ university admission, the effects of family background factors on students’ RSHS admission opportunity (students who cannot be admitted by RSHS usually enter vocational senior high school) would be retained and integrated into the effects of family background factors on university admission. That’s because vocational senior high school students have much less opportunity of entering university compared to RSHS students as I described in Chapter One (“Introduction to Chinese educational system”), which has also been confirmed by Wu’s (2013) research. Therefore, the theory on the RSHS admission in my research can be used to explain the university admission in Li’s (2006) research.

But how can the theory on the effects of sibling number be used to explain the effects of father’s occupation in Li’s (2006) research? Regardless of the marketisation of economic reform, the majority of the management class in China should still consist of officials of governments and political parties, state-owned enterprises and public institutions (including schools, universities, research institutes and hospitals) and they are usually members of the Communist Party of China. If they wanted to get promoted from a clerk or a professional to a management position, these officials must have strictly complied with the one-child policy (1979–2013), which covered half of the late Reform and Opening-up period (1992–2003 as their children’s university admission years and their children born from about 1974–1985) in Li’s (2006) research. If an official breached the one-child policy, he or she would be punished with documented serious warning or removal from the leadership position (CCCPC 1997), and thus even the

216

current two Chinese top leaders President Xi Jinping and Premier Li Keqiang only have one child. My research discovered that the class status itself is not the main contribution to the increasing advantage of the privileged family but the significantly increasing advantage of only-child families (the usual family structure of management class in China) in the university admission between the first (1994–1998) and second QOE period (1999–2005), which can explain why the advantage of the management class increased in Li’s late Reform and Opening-up period (1992–2003). The above discussion reveals that the change in the effects of family background factors on students’ university admission opportunity could be clearer revealed by using the sample of RSHS students and adding the sibling number indicating the family resources dilution mechanism as my study did.

In addition, my research covered a longer time range (the extra third QOE period) than Li’s (2006) research and had a new finding that even though the advantage of only-child students in the transition from RSHS to university admission significantly increased during the second QOE period, it then decreased to the original (first QOE period) level during the third QOE period. I have explained that the sole increase during the second QOE period was associated with a few sharp one-off increases in university tuition fees and since the tuition has been frozen during the third QOE period along with the continuous increase in the average family income, the tuition affordability gap between only-child and non-only-child families has closed during the third QOE period. During the third QOE period, the effects of family social status remained unchanged while the effects of family cultural capital continued a downward trend from the first to second QOE periods in my research, in other words, from the early to late Reform and Opening- up period in Li’s (2006) research. The specific mechanism for this change could be the CET in RSHS, which to a considerable extent equalises the learning resources during the out-of-school time (cultural capital at home) and provide those underprivileged students an opportunity of catching up with or sometimes exceeding 1 their peers. However, even though the educational inequality in the university admission

1 I heard an observation described by a professor at Department of Sociology, Nanjing University, during one of my postgraduate courses that in Nanjing University the cleaning staffs’ children often did better in examinations than professors’ children because the former were envious of their professorial colleagues’ occupation, and thus pushed their children very hard including sending them to attend lots of CETs. 217

opportunity is decreasing, it should still be noted that the equalisation trend could mainly benefit the families from the lower-middle to middle level of cultural capital rather than the ones with the lowest level, which has been observed by (Yu 2016) who stated, “the lowest class give up (their children’s) education, the middle class are overly anxious (about their children’s education) while the upper class just walk out on Chinese ‘gao kao’”. In addition, along with educational equality, the original argument for the QOE should be recalled as well that the examination-oriented education system is damaging students’ physical and mental health and creativity since CET becomes a new heavy academic burden instead of the burden assigned by formal schools (Yan 2011; Ruan 2011, 2008; Hu 2014).

My research used QOE and its unexpected consequence CET to explain the changes in the associations between family background factors and students’ admission opportunities. However, as I discussed in the summary section of Chapter 4, even though I tried to control the effects of educational expansion, which is a critical factor used by many previous studies to explain the changes in social reproduction of educational inequality (e.g., Li 2010; Wu 2013; Li 2006), these effects may not be excluded owing to the limitation of data without accurate transition rates. However, even without excluding the effects of educational expansion, the QOE policy still can be considered as an alternative explanation, which may have a combined effect on the social reproduction of educational inequality with the educational expansion and other concurrent polices and social changes. That is because according to Li’s (2017b) threshold dependent inequality theory, the educational expansion is to lower the admission scores in general and whether the inequality increase or decreases depends on a fixed threshold, which is based on the normal distribution and a fixed gap of examination score means between students with different levels of family background factors, while the implementation of QOE could change the threshold by changing the gap of examination score means, for example, improving the average examination scores of students with lower family cultural capital via its unexpected consequence CET, and thus could change the social reproduction of educational inequality in urban China along with the effects of the educational expansion.

218

QOE, other social changes and their relationship with the popularity of CET Implementation of QOE Even though it is different from previous research as it provides direct evidence and more detailed observations from field work, my study still supports the conclusions of previous studies that most QOE measures are not implemented successfully in reality, which include that: (1) in general students have less academic burden assigned by schools due to the QOE sub-policy of alleviating students’ academic burden but the burden (usually CET) assigned by parents increases (Yan 2011; Ruan 2011; Hu 2014); (2) the new teaching and learning methodology of QOE is rarely employed in class (Third Research Group 2006; Wang 2011a; Wang 2014); (3) the increase in non-academic courses and activities at school is very limited and the teaching and learning is still highly focused on academic subjects (Gu and Zhang 2010; Wang 2014); and (4) the old examination-oriented admission system has barely changed (Shi 2016; Yan 2011; Sargent et al. 2011; Zhu 2009), which is the “bottleneck” of the QOE reform and the key cause of the slow progress of the above three QOE sub-reforms (Wen and Liu 2014; Zheng 2002; Yan 2011).

Even though the academic burden-alleviation measure of the QOE policy has been successful, at least in terms of the academic burden assigned by schools, my research found that the QOE policy is not the only contributor for this change. As stated in the first chapter, there have been other concurrent policies and social changes along with the QOE policy that have influenced student learning life. Compared with other research on academic burden alleviation, my study not only included parents’ observation of changes in their children’s out-school time but also proposed a new explanation that in addition to the academic burden alleviation of QOE policy, a concurrent policy namely the labour policy of shortening working hours from 48 to 44 hours per week in 1994 and then to 40 hours in 1995 has contributed to the decreases of students’ at-school time as well.

The association between the implementation of QOE policy and the popularity of CET My research focused on the change of social reproduction of educational inequality and CET becoming more and more popular since the late 1990s rather than the existence of

219

CET itself is the changing factor used as my explanation. Therefore, unlike previous research my study does not discuss the causes such as traditional culture emphasising education (Hu 2014; Bray 2006) and the desire for upward social mobility (Hu 2014; Li 2015), which should be rather stable. On the contrary, my study thoroughly and empirically investigated the relationship between the unsuccessful implementation of QOE policy and the popularity of CET, which has been summarised in the first section of this chapter. The key argument of my study is that due to QOE reform, the formal school’s capability of preparing students for admission examinations has been reduced, and thus parents have to buy CET services to keep their children competitive. This is different from the viewpoints of previous research that claimed that the need for CET was due to the interaction effects between the equalisation policy by employing standardised textbooks and the school district system1 in the compulsory education stage and the unchanged competitive admission into the higher level education, which resulted in parents’ strong need of differentiated educational services to help their children get or keep the edge in the admission process (Liu and Xie 2014; Bray and Kwok 2003; Park, Byun and Kim 2011). However, even though some parents do buy one-on- one family CET services, the most popular form of CET service still operates like a normal school2 and there is no previous finding that the teaching content or methods used by CET companies are significantly different. Rather than using CET to obtain an edge, the primary aim of attending CET could be to make sure their children “not to lose the game at the starting line” and with almost everyone attending CET their children could equally compete in the admission. This seems an irony that the academic and financial burden is getting heavier while the competition stress is the same as the period when students had a longer at-school time. This irony has also been illustrated by Huo and Xu (2013) and Yu and Ding (2011a) using the Game Theory.

1 Entering a school nearest to the student’s home. 2 Cheng and Chen (2014) also used the term “schoolised” (“xue xiao hua”, 学校化) in their research on CET. However, the term in their research was to describe the phenomena that some public schools formally invited a CET company to coordinate a course (usually English) in their school. However, I use the term “Schoolised” to describe the phenomena that many CET institution operates like a normal school and use the same teaching method as the public school, which is not personalized one-on-one tutoring but lecturing about two dozen students at the same time. 220

Other policy and social changes contributing to the popularity of CET The implementation of QOE policy is not the only factor resulting in the popularity of CET. I also find some causes discussed by previous research could be related to policy or social changes discussed by my research. Therefore, they could be not only the causes for the existence of CET but also the indirect causes for CET becoming popular. First, Davies (2004) suggested that the demand for tutoring might be part of a wider strategy of intensive parenting for many families, “in which parents place a great premium on education, value a cognitively stimulating environment for their children, and closely monitor their children’s activities. This style of parenting emphasizes a careful plan of structured activities for children”. However, this intensive parenting, especially carefully planning and closely monitoring their children’s activities is not only related to family’s financial resources but also time-consuming, which phenomenon has been described in Chapter Five (how parents choose CET institutions for their children) and also been observed by Hu (2014). Therefore, if the intensive parenting is adopted by more and more parents since the middle or late 1990s, which is partly indicated by the popularity of CET, the labour policy of shortening weekly working hours mentioned above could also contribute to the popularity of intensive parenting because it allows some parents to have more after- work time and to spend that time on their children’s education even though many other parents are still very busy due to working overtime. Second, Hu (2014) suggested that in the “consumer society” some parents had the view that children’s development depended on the financial investment from parents, especially the investment in CET and claimed this view reflected those parents using consuming to avoid the responsibility of “true” parenting, which is different form the “intensive parenting”. If this viewpoint is correct, CET becoming popular may be related to the continuous decrease in the Engel Coefficient of urban citizens in China from 51.4% to 36.3% (NBSC 2016e), which means after the basic living needs have been fully satisfied parents spend more money on the high educational ambition for their children. Since the compulsory education is almost free and the tuition of RSHS keeps rather low for the urban citizens regardless of the increase in family income and providing children with the basic food, clothes and public-school education is no longer able to demonstrate their sense of responsibility for children, the investment in CET would be an alternative for many parents. Third, another social change is the popularity of CET

221

itself, which leads to CET keeping or becoming more popular and the previous research found both parents and children were influenced by peer pressure during the popularity of CET (Hu 2014; Bray 2013). However, my study found the Bandwagon effect on some parents’ investment in their children’s CET not only results from peer pressure but also from a worry that if they did not follow the fashion as other parents did and their children ended up with an unsatisfactory admission results, they would be blamed by their children for “irresponsibility” when they grew up without a satisfactory job. Therefore, those parents just do what other parents do and may not care too much whether CET can benefit their children or not.

Family socioeconomic factors, CET and academic achievement It is reasonable to assume that the higher the level of a student’s family socioeconomic factors the larger the probability of attending CET. My study found that students with more family focused resources (reversely indicated by sibling number) are more likely to attend CET, which is parallel with the findings of recent studies on Chinese students (Li and Qiu 2016; Chen and Bai 2015; Liu and Xie 2014). However, different from those studies, my study found that there is no significant association between family cultural capital (indicated by father’s education) and students’ probability of attending CET. This difference may result from that in my research I particularly controlled the effects of which school the student attended (inherently including the effects of city where the school located) and thus compared to the effects of parents’ own education, the peer pressure from other parents or the fashion in the same school may have more significant effects, which has been revealed in the findings from my field work. In addition, my study particularly investigated the effects of family social status (indicated by father’s occupation) and discovered there is no significant difference in the probability of attending CET between students from different family class status. I explain this result by the possibility that CET is no longer a luxury with the symbol value of upper class, which is similar to Chen and Bai’s (2015) conclusion that the CET is viewed as a “necessity” rather than a luxury by many parents although their conclusion was based on the descriptive finding that the majority of primary students (69.6%) in Haidian District, Beijing City attended CET.

222

Even though the effects of father’ education is not significant, my research found another factor related to families’ cultural capital, i.e., parents’ expectation of children’s educational level has a significant association with children’s CET attending probability, which is opposite to Chen and Bai’s (2015) research. Since this factor is only a control variable, my discussion will not focus on its effects, but on its implication for the sample- choosing of my research. One explanation for the difference between my finding and Chen and Bai’s (2015) is that the sample in my analysis in Chapter Six was second-year JHS students while Chen and Bai’s (2015) sample was third- and fourth-year primary school students and in the early education stage many parents may mainly treat CET as an after-school care facility rather than as an approach to achieving their educational expectations. This explanation could also be applied to the difference in gender effects where my study showed that girls are more likely to attend CET while Chen and Bai’s (2015) findings were the opposite. This is because in the early education stage such after-school care, usually in the form of non-academic CET, may not be viewed as a necessity for some families, and thus those families may be only willing to “spoil” a boy rather than a girl. While in the later educational stage, especially when the RSHS and university admissions are approaching, CET may become a necessary approach to changing the child’s and then his or her family’s fate regardless of whether the child, especially an only child under the one-child policy, is a girl or a boy (Fong 2006). Actually, my research discovered that in the past 20 years, girls in urban areas of China have a higher probability of entering RSHS than boys do (see Chapter Four), and thus those parents may find that it is more worthwhile to invest in their daughter’s rather than son’s education including CET. Even though neither parents’ expectation nor students’ gender is the main explanatory variable of my research, this difference reveals again that parents’ parenting behaviour can be influenced by the educational stage of their child and they could be more concerned about how to improve their children’s admission opportunity when their children are closer to approaching the admission examination, which have been supported by the findings in Chapter Five on parents’ explanations of changing their children’s non-academic CETs into academic CETs when the RSHS admission is approaching. This assumption justifies the strategy of my study that uses a sample of JHS students from the China Education Panel Survey rather than a

223

sample including primary school students from the Chinese Family Panel Studies to conduct an analysis concerning educational admission.

Parents’ goal in investing in their children’s CET is to boost their children’s academic performance and improve their chances of success in the admission examination. Through the propensity-score matching method and controlling more personal, class-, school- and county-level effects, my study more rigorously demonstrates the positive effects of attending CET on students’ academic performance, which is a parallel result to the findings of recent research on the effects of CET in China that utilised different analysis methods (Liu and Xie 2014; Li and Qiu 2016). However, different from these two previous studies, my research further investigated whether the extent of improving academic achievement by attending CET is different between students with less or more family cultural capital and found that the gap between students attending 5 to 10 hours of tutoring per week and non-attendees is narrowing as family capital increases, which means the students with less family cultural capital benefit more from attending CET than those with more family cultural capital. Even though in the group of 5 to 10-hour tutorial attendees, students with less family cultural capital in general have only caught up with their privileged peers rather than exceeded them, CET can still be viewed as an approach at least to keeping more students with less family cultural capital in the tunnel of educational upward transition. This finding is reasonable when considering that the two arguments stated that most of the current CET is operating like a normal school, which is found by my research, and that “even if schools are unfair, they may serve as equalizers if the variation in school environments is smaller than the variation in non- school environments” (Downey, von Hippel and Broh 2004: 613-614).

The justice evaluations of QOE policy and its unexpected consequence CET Even though my field work did not find that parents were aware of CET’s function of benefiting a few underprivileged students, parents in my interviews did claim that the current QOE policy and its unexpected consequence of CET are just or at least acceptable injustice. However, these parents’ evaluation of CET as being just is still contradictory to their perceived inequality that many underprivileged students cannot afford CET. I summarised the reasons parents mentioned in the interviews and since there is no

224

previous research directly on parents’ justice evaluation of QOE and its consequence of CET in China, I will discuss these reasons with general social justice theories and some previous research on Chinese justice evaluation of income distribution in the following part.

The crucial reason for the ineffectiveness of QOE reform and the popularity of the “schoolised” CET is the stubborn examination-oriented admission system, especially the “bottleneck” of the QOE reform namely the “gao kao” (university admission examination). Parallel to the findings of previous research that the examination- oriented admission system, especially “gao kao” is one of the only institutions that Chinese trust and regard as fair (Fong 2006; Sargent et al. 2011), my study finds parents evaluate the “gao kao” as just and are concerned for the justice of the proposed admission method reform. According to distributive justice theory, there are three components of the distributive justice process: (1) the principles for the allocation of goods, (2) the procedure that governs the application of those allocative principles and (3) the distributive outcomes, which will be “presumably” achieved if the general accepted principles are implemented through a clear and just set of procedures (Alwin 2000). In terms of the “gao kao”, parents appreciate that it distributes the scarce university enrolment places according to students’ scores achieved in a standardised test examining their acquirement of academic knowledge, and thus the allocation principle is consistent with John Rawls’s principle of equality of fair opportunity that “offices and positions are to be open to all under conditions of equality of fair opportunity—persons with similar abilities and skills are to have equal access to offices and positions” (as cited in Buchanan and Mathieu (1986), p.27). Parents do not express a demand for equality of outcomes, for example, using the lottery method, or a demand for returning to the principle of rights or entitlements, for example, as a worker-peasant- soldier university student of the Cultural Revolution period. In addition, the procedural justice of “gao kao” is upheld not only by the laws, particularly the Criminal Law as the “gao kao” is “a national examination prescribed by law” (SCNPC2015b), but also by parents’ direct actions for example a demonstration or street protest1 when a change

1 The usual reports about parents’ direct actions relevant to “gaokao” are that parents voluntarily regulate the traffic to reduce the noise when their children are in an English listening test. However, as I 225

to the admission part of the “gao kao” system was proposed, which is not ordinary after the Tiananmen Square protests of 1989.

The equality of the fair opportunity principle of the “gao kao” system is perceived as just by parents. However, according to Husen’s (1975: 16-19) theory of equality of educational opportunity, in the individual level the “equality” can be conceived of in three ways: (1) as a starting point, (2) as a treatment and (3) as a final goal. If I borrow the format of Alwin’s (2000) statement above, the equality of the final goal should be “presumably” achieved if all individuals start their formal school career on an equal footing and if during this career they are treated equally regardless of their genetic equipment and social origin. Even if we do not consider the equality of educational opportunity from the view point of genetics and the unequal treatment due to different school environments, can the popularity of CET as a for-profit institution preparing its consumers for admission examinations make students take the “gao kao” on an equal footing?

Since it is not part of compulsory education, CET should be regarded as a type of limited educational resource to be allocated. The principle and procedure of distributing the resources of CET should obviously be consistent with the principles of the free market as CET is a kind of service sold by different for-profit companies. The principles of the free market should contradict Husen’s (1975: 16-19) principle of equality of educational opportunity, especially as a starting point and thus it is interesting to explore why parents in the interviews still claim that CET is a just mechanism as part of the educational admission system. After all, usually CET is not to help students learn a dispensable hobby or successfully graduate but is closely related to the competition for the educational admission opportunity and it almost plays a similar role as a normal school in the educational admission, which means CET should not be a luxury but a necessity enjoyed by every student for educational equality (Chen and Bai 2015). In

mentioned in Chapter One (Footnote 1, p.6) there was a massive protest against the MOE allocating a few of the quota of university enrolment places from Jiangsu and Hubei Province to some middle and western undeveloped provinces in 2015 (see Ai (2016) for a news report). Even though this protest was only relevant to the admission part of “gaokao” (including examination and admission), it demonstrated how important the gaokao was for many student parents that they were willing to take serious actions. 226

addition, parents have acknowledged that not all families can afford their children’s CET fees.

In my research, parents’ arguments against the necessity of CET such as “the admission opportunity mainly depends on children’s effort” and “the public-school teaching is enough” have been seriously weakened by their choice of sending their own children to attend CET. Parents’ very justification of their justice evaluation of CET could be the argument that “a responsible parent should be able to afford their child’s CET fees if they make enough efforts”, which claims the allocation of CET depends on the recipient’s merits and follows the principle of equality of fair opportunity as the allocation of the formal educational resources. First, I briefly discuss this argument from two social psychological points of view: (1) the Belief in a Just World theory (BJW) and (2) Chinese perception of self. According to the BJW theory, “individuals have a need to believe that they live in a world where people generally get what they deserve.” (Lerner and Miller 1978: 1030) When confronting a fact conflicting with this BJW view, for example in my research parents understand underprivileged students who cannot afford the necessary CET are in an unfair position, some parents will “try to ignore, reinterpret, distort, or forget it—for instance, by finding imaginary merits to the recipients of fortuitous rewards, or assigning blame to innocent victims” (Bénabou and Tirole 2006: 700). Therefore, it is possible that the parents interviewed attributed parents being able to send their children to CET to their “working hard” and blamed the others for not “making enough efforts” to maintain their belief in a just world, specifically, a just educational field.

However, another question is that even if some parents did not make enough effort, is it fair that their children pay the consequences for their parents’ “fault”? This question has been involved in a debate around redemptive egalitarianism or redemptive philosophy (Coleman et al. 1966; Coleman 1973, 1966; Husen 1975; Frankel 1973; Bell 1973), which discusses whether society should answer the moral problem of redeeming the individual who “has been born with less favourable genes and in less favourable circumstances” (Husen 1975: 18). My discussion will not penetrate the educational philosophy but is focused on the phenomenon that the interviewed parents’ arguments

227

imply that they do not support redemptive egalitarianism of all students starting their formal school career on an equal footing except public support for the basic compulsory education. Yang (2009b) provided an interesting theory of indigenous Chinese psychology about Chinese self, which could be used to understand the above arguments of the interviewed parents. First, Yang (2009b) differentiated two Chinese concepts namely the “Big Self” (da wo, 大我) and “Small Self” (xiao wo, 小我). The “Small Self” indicated the “self” in the narrow sense, which was within the limit of one self’s body while the “Big Self” indicated the group the “Small Self” belonged to such as the family, community, society and state. Then, Yang suggested Chinese traditional culture, especially Confucianism, encouraged or required an individual to keep self-deskilling (xiu shen, 修身) in order to sacrifice his or her “Small Self” for the “Big Self” and further to combine the “Small Self” and the “Big Self” into one unit and to forget the “Small Self”. In this culture, people may view children and their parents as one self-unit, and thus when one part of “Big Self” (parents) have not made enough effort, it is not unjust for the other part (children) to suffer the consequence. Even though in many different cultures many parents were reported to sacrifice themselves for their children in an emergency, it is still meaningful to consider the role of cultural factors in the popularity of CET if we find that CET appears to be more popular in the countries, regions and communities influenced by Chinese Confucianism. Liu and Xie (2016) discovered that in United States the cultural orientation of Asian families was different from that of white families in ways that mediated the effects of family socioeconomic status on children's academic achievement.

The above argument of blaming underprivileged parents for “not making enough efforts” to provide their children with CET is consistent with some interacting social thoughts in China nowadays. As Kipnis (2007) pointed out, blaming the victim (by denying structural factors) in the “su zhi” discourse was one form of neoliberalism. In particular, with respect to the above disadvantage encountered by students with one or more siblings found by my research, blaming their parents for not making enough effort to send their child to CET is consistent with the previous government propaganda for the one-child policy that a responsible parent and citizen should reduce the number of children but improve the quality of the only child (you yu, 优育). In addition, from the interviews I

228

found the discourse used by parents interviewed also involved a direct description of the fading of communist ideology in the post-Mao era to explain why they do not advocate for absolute egalitarianism, which is then used to justify their reluctance to advocate for a more equal educational system. Even though this topic is sensitive in mainland China, a few articles have directly acknowledged the existence of the crisis of belief of communism (Chen 2008; Cai 2002) and parents’ attitude to a equalised educational system is consistent with the findings of previous research disaffirming the conventional sense that absolute egalitarianism of income is favoured by Chinese people (Li 2014b; Wu 2009). Third, parents interviewed who blamed the victim and opposed a higher level of public subsidy for students’ education (e.g., subsidy for underprivileged students’ CET fees) was also reflecting the stigmatisation of welfare and (not only the communist but also the western European) welfare state since the “communist” experiment of Mao’s era failed and the Reform and Opening-up period started (Zheng 2013).

This chapter summarised the main findings of this thesis and compared those findings with previous research along with discussion on their theoretical implications. The final chapter will summarize the theoretical and policy implications of this thesis and report the limitations of this study for further research.

229

Chapter Nine. Conclusion

In this chapter, I first discuss the theoretical implications and policy implications of my research, especially in terms of the design and evaluation of Quality-oriented Education (QOE) and possibilities for further policy improvement. Then, I discuss the limitations of the research and make recommendations for future research.

I. Theoretical and Policy Implications of Findings

Even though there are a number of studies on the trend of social reproduction of educational inequality between different social classes in contemporary China (since 1949), my study’s contribution to the research area is exploring the association between QOE policy and social reproduction of educational inequality in urban China with empirical methods for the first time. Particularly, my study describes the trend of social reproduction of educational inequality in the urban area before, during and after implementation of QOE using survey data. In contrast to previous research that occasionally mentioned the possible associations between QOE, the commercial extracurricular tutoring (hereinafter CET), and educational admission, my study investigates in greater detail how the QOE policy has improperly designed and unsuccessfully implemented and thus is associated with the popularity of CET and how CET could help prepare attendees for the admission examinations during the QOE era. More importantly, different from previous research usually concerning the negative aspects of CET to intensify the disadvantages of underprivileged students in academic achievement, my research finds the positive function of CET to benefit students from families with low cultural capital more in academic achievement. This is because, unlike their privileged peers, without the repetitively affordable schoolised CET it is very difficult for those underprivileged parents to have sufficient family cultural capital (e.g., hiring a luxurious one-on-one private tutor or tutoring their children by themselves) in order to provide their children with a favourable learning environment at home to “meaningfully” spend the additional extracurricular time due to the QOE measure of

230

shortening at-school time. Finally, even though there were a few studies on Chinese people’s evaluation of income in terms of justice, those studies were based on a few quantitative variables lacking a deep understanding of the connotation of justice and the factors forming such a justice evaluation. More importantly, rarely has there been previous research on Chinese people’s evaluation of educational policy in terms of justice, which is one of my study’s contributions to the knowledge in this area.

Even though urban parents’ positive evaluation of the current educational system in terms of justice may indicate that it may not be necessary for Chinese policy makers to treat the educational equalisation policy in the urban area as a priority since the pressure from parents for a more equalised education system is not strong enough, it should not mean there is no intension in the status quo. The fact that many parents give priority to maintaining the justice of admission rather than alleviating their children’s academic burden does not mean the problem of heavy academic burden on students is not important. Parents are deeply concerned that the heavy academic burden could impair their children’s wellbeing, which conception is possibly constructed by the propaganda for QOE since alleviating the heavy academic burden on students was the original justification for promoting the new QOE and abolishing the old Examination- oriented Education (EOE). If the QOE policy cannot succeed in alleviating the heavy academic burden, quite a few parents may finally lose confidence in the Chinese educational system, which has been implied in a current research and opinion poll on the reasons for parents sending their underage children to study overseas (Hao 2017; Zhou 2013; Fang et al. 2013). Other arguments in the propaganda for QOE such as that education should improve student’s all-around qualities and develop their creativity, problem-solving skills and lifelong learning attitudes could result in parents’ negative attitudes toward the current Chinese educational system as well. The human capital and cash outflow from the Chinese educational industry could be intensified along with the scale of middle class enlarging since compared to the lower class, the middle class are more likely to be influenced by the above propaganda, to be able to afford expensive international student tuition, and to be losing their advantage in the university admission according to the findings in my research. Therefore, the relationship between

231

the unsuccessful QOE reform and the popularity of studying abroad should attract more attention from policy makers.

However, reforming the old EOE system while maintaining parents’ belief in its justice would not be easy. It has been well researched by this and previous studies that the competitive examination-oriented admission system is the bottleneck for the QOE reform. The national standardised university admission examination is the foundation of parents’ general belief in the justice of the educational system in China. Unless it can find another admission approach that parents can equally trust, the government would need to be very cautious about changing the current admission approach. If not changing the approach, how about changing the admission content? The unsuccessful implementation of the QOE policy has indicated that without including the non- academic subjects in the admission examination (not simply examining their knowledge on paper1), shortening at-school time only results in the source of academic burden on students changing from school to parents through sending their children to CET, which is mainly focused on academic subjects rather than developing students’ other qualities like art and sport. However, even though including non-academic subjects in the admission examination could make the QOE reform achieve its two critical goals, namely alleviating students’ academic burden and developing their all-round quality, there are some obvious challenges as follows: (1) how to develop a standardised examination of the various performing parts of those non-academic subjects like art and sport, which parents can evaluate as being just and (2) more difficultly, how to fund the education of those non-academic subjects. As I mentioned when I was discussing the “Sport, Art 2+1 Program”, under the order of the Ministry of Education in Chapter Seven (pp. 173-174), even if each school only offered a few courses selected from the program appendix consisting of 60 sport and 20 art techniques to narrowly meet the QOE policy requirement of developing each student’s “strong points and potentials”, the additional investment in new teaching staff and facilities would be a rather heavy financial burden on Chinese central and local governments according to their current and foreseeable

1 If the admission examination of non-academic subjects only tests the knowledge on paper, it is rarely different from the current examination of academic subjects prepared by rote. In my field work, a rumour of including music knowledge in the paper examination for senior high school admission has been strongly criticised by parents for adding more “academic” burden on their children. 232

budgets. In addition, if the non-academic subjects are included in the admission examination and the policy of shorting at-school teaching time is maintained, it is possible that the social reproduction of educational inequality will be intensified again since the lower-middle class may afford the schoolised tutoring but they may not afford the fees for a luxurious teaching studio of art or sport club and the art instrument or sport gears. If the above challenges could be solved through economic and technical development in China, the adventurous policy implications mentioned here can be worth considering for policy makers. The last issue relevant to the proposal of including non-academic subjects in the admission examination is when implementing the new admission in some pilot areas, those areas must form a special independent admission district to ensure parents whose children take the new admission approach that the enrolment rate in the special independent admission district is at least now lower than in other regular admission districts employing the old admission approach. If the reform reduces students’ admission opportunities compared with the others without being affected by the reform, it will suffer a strong opposition from parents who think their children’s interests are damaged by an unfair reform1.

Yang (2006c) discussed the policy evaluation problems resulting in the improperly designed and unsuccessful implementation of QOE2 and suggested that the problems included a lack of evaluations for many sub-policies or measures of QOE, a lack of integrity and perfunctory formalism of evaluations. Yang’s (2006c) research was published approximately 10 years ago but until recently I still could not find an evaluation report of QOE. If the educational authorities have commissioned some evaluation projects, the transparency of releasing the evaluation reports to the public is a problem that should be solved. Another issue that Yang (2006c) did not emphasise is the methodology used in the educational evaluation. The current research in education

1 This concern derives from a story during my field work that the QOE policy had been implemented better in Nanjing Municipality than in other cities of Jiangsu Province for several years; however, since those cities are included in the same university admission district, parents in Nanjing Municipality found their children were worse off in the university admission compared to other cities, and thus applied pressure on the educational authority of Nanjing Municipality and the educational authority had to change its dynamics of implementing the QOE policy. 2 Even though some sub-policies or measures of QOE (e.g. shortening at-school time) have been completed, in general the implementation of QOE is still in progress. 233

in China still contains a lack of empirical research methodology (Yuan 2017; Jin 2017) and without empirical research methodology, the integrity of policy evaluation will be in doubt. Actually, since the QOE sub-policies and measures (e.g., the new curriculum reform from 2001 to 2010) are usually firstly implemented at pilot areas, it could have been possible to conduct some natural experiments by comparing the pilot areas and the non-pilot areas. In particular, it is better to demonstrate the positive effects of “enigmatic” pedagogy reform on students’ creativity, problem-solving skills and lifelong learning attitudes using an experiment method to persuade parents and teachers with suspicion to support the reform.

II. Limitations and Future Research

Due to the limitation that the nationwide sample size for an 18-year cohort was only 356 cases in the logistic regression model predicting probability of students transiting from regular senior high school (RSHS) to university in Chapter Four, I did not differentiate the admission probabilities between the three-year vocational university, the ordinary four-year bachelor-degree university and the elite university (e.g., the universities listed in the national elite university development projects “Project 211” and “Project 985”) for robustness and prediction accuracy. However, just as the difference between my finding that the opportunity inequality in “regular” senior high school admission increased and Li’s (2006) finding that the opportunity inequality in the whole (including regular and vocational) senior high school admission decreased suggested, the trends of social reproduction in the admission opportunity into various categories of universities university could be different due to these categories’ different degrees of competitiveness, which could be investigated by future research if there is a dataset with a larger sample size.

In addition, one critical finding of my study is that because attending CET could improve the academic achievement of students with lower family cultural capital more than their peers with higher family cultural capital, the educational inequality between students with different family cultural capital is decreasing as CET becomes more popular for the implementation of QOE policy; however, the condition is quite a few families with lower

234

cultural capital can afford CET fees. Even though the schoolised form of the current CET could reduce its cost and even the lower-middle class family with ambitious educational expectation for their children can afford the tutorial fees, my study found that some upper-middle class families began to buy the expensive one-on-one tutoring. The implication of the above result is that future studies should investigate whether one-on- one tutoring is better at improving students’ academic achievement than the schoolised CET is. If so and the one-on-one tutoring forms a distinction between different class fractions, the educational inequality between students with different amounts of cultural capital may increase again as more and more students from upper-middle class family attend one-on-one tutoring and become better prepared for admission examinations than their peers who can only afford the schoolised CET. Another limitation relevant to CET is that the current available dataset only provides the information on junior high school (JHS) students’ attending CET and my research has to draw on the difference in the implementation of QOE policy between JHS and RSHS and the difference in average levels of family background factors between JHS and RSHS students to reason their difference in attending CET. Therefore, future research could collect a new dataset containing the information on both JHS and RSHS students’ attending CET to retest the hypotheses based on the findings of my research.

Finally, due to the budget, time and network limitations of my research the parent interviewees recruited in the field work were from lower-middle to higher-middle socioeconomic groups, and thus future research should investigate the justice evaluation of the QOE policy in the current educational system made by parents from the low and the high socioeconomic groups if these groups are willing to participate.

235

References: Ai, W. (2016) 调配高考名额中国家长请愿抗议 [Re-allocate university admission quota, Chinese parents protesting], 法国国际广播电台 [Radio France Internationale], Accessed 30 May 2017. http://cn.rfi.fr/中国/20160515-调配高考名额-中国家长请愿抗议. Alwin, D. F. (2000) 'Social Justice' In Encyclopaedia of Sociology, in Borgatta, E. F. and R. L. Montgomery (ed.), New York: Macmillan Publishing Co.: 2695-2711. Anagnost, A. (2004) 'The corporeal politics of quality (suzhi)', Public Culture, 16 (2): 189- 208. Andre, C. and M. Velasquez (1990) Justice and fairness (Issues in Ethics - V.3, N. 2 Spring 1990), Markkula Center for Applied Ethics, Accessed 23 June 2016. https://legacy.scu.edu/ethics/publications/iie/v3n2/homepage.html. Austin, P. C. (2009) 'Balance diagnostics for comparing the distribution of baseline covariates between treatment groups in propensity‐score matched samples', Statistics in Medicine, 28 (25): 3083-3107. ——— (2011) 'A tutorial and case study in propensity score analysis: an application to estimating the effect of in-hospital smoking cessation counselling on mortality', Multivariate Behavioral Research, 46 (1): 119-151. Baudrillard, J. (1998) The Consumer Society: Myths and Structures. London: SAGE Publications. Bell, D. (1973) The Coming of Post-industrial Society: A Venture in Social Forecasting. New York: Basic Books. Bénabou, R. and J. Tirole (2006) 'Belief in a just world and redistributive politics', The Quarterly Journal of Economics, 121 (2): 699-746. Bernstein, B. (1975) Class, Codes, and Control. London: Routledge & Kegan Paul. Blake, J. (1981) 'Family size and the quality of children', Demography, 18 (4): 421-442. Bourdieu, P. 1996. Cultural reproduction and social reproduction. In: Brown, R., ed. The Annual Conference of the British Sociological Association, 7-10 April 1973 University of Durham. London: Routledge, 71-112. ——— (1986) 'The forms of capital' In Handbook of theory and research for the sociology of education, in Richardson, J. G. (ed.), New York: Greenwood Press: 241-258.

236

Bourdieu, P. and J.-C. Passeron (1977) Reproduction in Education, Society and Culture. Beverly Hills: Sage. Bray, M. (2006) 'Private supplementary tutoring: comparative perspectives on patterns and implications', Compare: a Journal of Comparative and International Education, 36 (4): 515-530. ——— (2013) 'Shadow education: comparative perspectives on the expansion and implications of private supplementary tutoring', Procedia-Social and Behavioral Sciences, 77: 412-420. Bray, M. and P. Kwok (2003) 'Demand for private supplementary tutoring: conceptual considerations, and socio-economic patterns in Hong Kong', Economics of Education Review, 22 (6): 611-620. Breen, R. and J. H. Goldthorpe (1997) 'Explaining educational differentials towards a formal rational action theory', Rationality and Society, 9 (3): 275-305. Breen, R. and J. O. Jonsson (2005) 'Inequality of opportunity in comparative perspective: recent research on educational attainment and social mobility', Annual Review of Sociology, 31 (1): 223-243. Breen, R., K. B. Karlson and A. Holm (2010) Total, direct, and indirect effects in logit models (working paper). Aarhus: Centre for Strategic Educational Research, DPU, Aarhus University. Brookhart, M. A., S. Schneeweiss, K. J. Rothman, R. J. Glynn, J. Avorn and T. Stürmer (2006) 'Variable selection for propensity score models', American Journal of Epidemiology, 163 (12): 1149-1156. Bryman, A. (2006) 'Integrating quantitative and qualitative research: how is it done?', Qualitative Research, 6 (1): 97-113. ——— (2012) 'Chapter 27 Mixed methods research: combining quantitative and qualitative research ' In Social Research Methods, in Bryman, A. (ed.), Oxford, UK: Oxford University Press. Buchanan, A. and D. Mathieu (1986) 'Philosophy and justice' In Justice: Views from the Social Sciences, in Cohen, R. L. (ed.), New York: Plenum: 11-45. Buchmann, C. and E. Hannum (2001) 'Education and stratification in developing countries: a review of theories and research', Annual Review of Sociology, 27 (1): 77-102.

237

Buis, M. L. (2010) Inequality of educational outcome and inequality of educational opportunity in the Netherlands during the 20th century. Ph.D. thesis, Vrije Universiteit Amsterdam. Bureau of Statistics of Nanjing Municipality (BSNM) (2016) 南京统计年鉴 2015 表 4— 2 城镇居民家庭全年人均可支配收入 [Nanjing Statistical Yearbook 2015 Table 4-2 Per Capita Disposable Annual Income of Urban Households], Bureau of Statistics of Nanjing Municipality, Accessed 26 January 2017. http://221.226.86.104/file/nj2004/2016/renmin/4-2.htm. Cai, H. (2015) 中国劳动力动态调查：2015 年报告 [Report on China Labor-force Dynamic Survey (2015)]. Beijing: 社会科学出版社 [Social Science Academic Press]. Cai, S. L. (2012) '我国教育改革的困境与文化反思 [Plight of Education Reform and Cultural Retrospection in China]', 教育理论与实践 [Theory and Practice of Education], 32 (4): 32-35. Cai, X. (2002) '试论社会转型时期的 “信仰危机” [Analysis of the "crisis of belief" in the social transition era]', 当代世界与社会主义 [Contemporary World and Socialism] (2): 55-59. Center for Social Survey at Sun Yat-sen University (CSS) (2013) 中国劳动力动态调查： 2013 年报告 [Report on China Labor-force Dynamic Survey (2013)]. Beijing: 社会科学文献出版社 [Social Science Academic Press]. Central Committee of the Communist Party of China (CCCPC) (1997) 27 February. 中国共产党纪律处分条例（试行）[The Regulation of the Communist Party of China on Disciplinary Actions], Article 145. Central Committee of the Communist Party of China (CCCPC) and State Council (1994) 中发[1993]3 号 [No. 3 [1993] of the CPC Central Committee]. 13 February. 中国教育改革和发展纲要 [Outline of the Plan for Education Reform and Development of China]. ——— (1999) 13 June. 中共中央国务院关于深化教育改革全面推进素质教育的决

定 [Decision on Deepening the Educational Reform and Moving fully forward the Quality-Oriented Education].

238

——— (2010) 中发[2010]12 号 [No. 12 [2010] of the CPC Central Committee]. 8 July. 国家中长期教育改革和发展规划纲要(2010-2020 年) [The Outline of the National Plan for Medium and Long-Term Education Reform and Development (2010-2020)]. Chen, A. J. (2008) '当代中国转型时期 “信仰危机” 的原因探究 [Explore the factors in the "crisis of belief" in the transition era of Contemporary China]',当代世界与社会主义 [Contemporary World and Socialism] (3): 161-163. Chen, B. L. and X. X. Bai (2015) '家庭社会经济地位, 家长同辈群体压力与城镇小学生补习——基于北京市海淀区小学调查 [SES, peer pressure on parents, and private tutoring of primary students in Beijing: evidence from a survey in Haidian district]', 清华大学教育研究 [Tsinghua Journal of Education], 36 (5): 102-109. Chen, C. R. (2014) '从博弈论视角看素质教育实施之困境 [Research on the plight of implementing the quality-oriented education from the perspective of game theory]', 教育研究与实验 [Educational Research and Experiment] (1): 22-26. Chen, J. Y. and X. H. Zong (2014) '高校学费变动的社会经济压力分析--基于南京地区高校的调查 [The Social Economic Pressure of College Tuition Rising: A Survey of Colleges and Universities in Nanjing]', 高校教育管理 [Journal of Higher Education Management], 8 (6): 75-79. Chen, Y. Y., S. Naidu, T. H. Yu and N. Yuchtman (2015) 'Intergenerational mobility and institutional change in 20th century China', Explorations in Economic History, 58: 44-73. Chen, Z. L. (1998) '学习党的十五大精神深化对教育战略地位的认识 [Learn the decision of the Fifteenth Party Congress and deepen the understanding of the strategic status of education]', 教育研究 [Educational Research] (7): 3-9. Cheng, T. J. and X. L. Chen (2014) '杂乱有章: 辅导机构学校化与学校辅导机构化 [Orders in Disorders: Schooling and Tutoring Moving toward Each Other]', 教育学报 [Journal of Educational Studies], 10 (1): 109-120. Chinese Society of Education (CSE) (2016) 中国辅导教育行业及辅导机构教师现状调查报告 [Report on the supplementary education industry and tutors in China]. Beijing: 中国教育学会 [The Chinese Society of Education].

239

Chu, H. Q. and H. Y. Yang (2008) '教育公平的原则及其政策含义 [The Essence of Educational Equity and Its Policy Significance]', 教育研究 [Educational Research] (1): 10-16. Coleman, J. S. (1966) 'Equal schools or equal students?', Public Interest (4): 70-75. ——— (1973) 'Equality of opportunity and equality of results', Harvard Educational Review, 43 (1): 129-137. Coleman, J. S., E. Q. Campbell, C. J. Hobson, J. McPartland, A. M. Mood, F. D. Weinfeld and R. York (1966) Equality of Educational Opportunity. Washington, D.C.: National Center for Educational Statistics (DHEW). Collins, J. (2009) 'Social reproduction in classrooms and schools', Annual Review of Anthropology, 38 (1): 33-48. Crabb, M. W. (2010) 'Governing the middle-class family in urban China: educational reform and questions of choice', Economy and Society, 39 (3): 385-402. Cravens, X. C., Y. R. Liu and M. Grogan (2012) 'Understanding the Chinese superintendency in the context of quality-oriented education', Comparative Education Review, 56 (2): 270-299. Cui, Q. S. (2013) 教育公平视阈下 W 县义务教育经费的历史考察 (1978-2011 年) [Historical Survey of W County Compulsory Education Funds from the Perspective of Educational Equity from the Year of 1978 to 2011]. Ph. D. thesis, East China Normal University. Dai, C. Q. and D. G. Li (2016) 海宁免费“晚托班”：家长齐“点赞” [Haining City's free after-school care is welcomed by parents], 嘉兴日报 [Jiaxing Daily]. 23 September, available at http://www.cnjxol.com/xwzx/jxxw/szxw/content/2014- 09/23/content_3165313.htm accessed 25 May 2017. Davies, B. and P. Bansel (2007) 'Neoliberalism and education', International Journal of Qualitative Studies in Education, 20 (3): 247-259. Davies, S. (2004) 'School choice by default? Understanding the demand for private tutoring in Canada', American Journal of Education, 110 (3): 233-255. Dehejia, R. H. and S. Wahba (1999) 'Causal effects in non-experimental studies: Re- evaluating the evaluation of training programs', Journal of the American Statistical Association, 94 (448): 1053-1062.

240

Dello-Iacovo, B. (2009) 'Curriculum reform and 'Quality Education' in China: an overview', International Journal of Educational Development, 29 (3): 241-249. Deng, H. (2016) 聚焦高考综合评价招生：淡化分分必争与一考定终身 [Focus on admitting students using the comprehensive evaluation in the university admission examination: play down "fighting for every point" and "one examination decides a student's fate"], 中国新闻网 [China News], Accessed 30 December 2017. http://www.chinanews.com/gn/2016/09-19/8007080.shtml. Deng, Z. and D. J. Treiman (1997) 'The impact of the cultural revolution on trends in educational attainment in the People's Republic of China', American Journal of Sociology, 103 (2): 391-428. Downey, D. B., P. T. von Hippel and B. A. Broh (2004) 'Are schools the great equalizer? Cognitive inequality during the summer months and the school year', American Sociological Review, 69 (5): 613-635. Espinoza, O. (2007) 'Solving the equity–equality conceptual dilemma: a new model for analysis of the educational process', Educational Research, 49 (4): 343-363. Fang, Y., L. He, X. Y. Liu and P. Wang (2013) '我国低龄留学生发展状况研究报告 [The report on the development of underage Chinese students studying abroad]', 中国青年研究 [China Youth Study] (11): 5-25. Fong, V. L. (2006) Only Hope: Coming of Age under China's One-Child Policy. Stanford: Stanford University Press. Frankel, C. (1973) 'The new egalitarianism and the old', Commentary, 56 (3): 54. Ganzeboom, H. B. and D. J. Treiman (1996) 'Internationally comparable measures of occupational status for the 1988 International Standard Classification of Occupations', Social Science Research, 25 (3): 201-239. Ganzeboom, H. B. G., P. M. Degraaf, D. J. Treiman and J. Deleeuw (1992) 'A standard international socioeconomic index of occupational-status', Social Science Research, 21 (1): 1-56. Garrido, M. M., A. S. Kelley, J. Paris, K. Roza, D. E. Meier, R. S. Morrison and M. D. Aldridge (2014) 'Methods for constructing and assessing propensity scores', Health Services Research, 49 (5): 1701-1720.

241

General Office of the Ministry of Education (GOME) (2011) 教体艺厅 [2011]4 号. 26 April. 关于在义务教育阶段中小学实施“体育、艺术 2+1 项目”的通知 [Notice of Implementing the “Sport, Art 2+1 Program” in Primary and Secondary Schools of Compulsory Education]. Golley, J. and S. T. Kong (2013) 'Inequality in intergenerational mobility of education in China', China & World Economy, 21 (2): 15-37. Gottfried, M. A. and V.-N. Le (2016) 'Full-versus part-day kindergarten for children with disabilities effects on academic and social-emotional outcomes', American Educational Research Journal, 53 (3): 708-744. Government of Nanjing Municipality (GNM) (2005) 宁政发 [2005]141 号. June. 南京市普通中小学办学条件标准（修订版）[The Standards for Regular Primary and Secondary School Running Conditions of Nanjing Municipality (Revision)]. Grilli, L. and C. Rampichini (2011) Propensity scores for the estimation of average treatment effects in observational studies. Paper presented at the Training Sessions on Causal Inference, Bristol, UK, 28-29 June. Gu, M. Y. and M. S. Zhang (2010) ' 推进素质教育 [Advance the quality-oriented education]', 教育研究 [Educational Research] (7): 9-14, 38. Guo, H. (2010) '新课改与“穿新鞋走老路 [The New Curriculum Reform: Wearing New Shoes to Walk on the Old Path]', 课程·教材·教法 [Curriculum, Teaching Material and Method], 30 (1): 3-11. Hao, M. J. (2017) 调查：超半数家长认为出国读高中能够快乐学习 [More than half of parents think studying in aboard senior high schools can make learning happier], 人民网教育频道 [people.cn-Education], Accessed 25 June 2017. http://edu.people.com.cn/n1/2017/0612/c220605-29333690.html. Hao, Q. H. (2014) '仇富型网络群体性事件的成因及其治理 [Causes of Internet Group Events Arising from Hatred of the Rich and their Management Strategy]', 齐鲁学刊 [Qilu Journal] (2): 77-81. He, D. C. (1992) '在国家教委 1991 年工作会议上的总结讲话 [The summative speech at the 1991 work conference of the State Education Commission]' In 中国教育年鉴 1991 [China Education Yearbook 1991], in Zhang, B. Q. (ed.), Beijing: 人民教育出版社 [People's Education Press]: 14-18.

242

He, H. J. (2017) 共享改革发展教育红利让“外来娃”融入城市 [Share the educational benefit of reform and development and integrate "immigrant children" into the urban], 未来网 [K168.cn], Accessed 17 June 2017. http://news.k618.cn/dj/201705/t20170529_11526220.html. Hegtvedt, K. A. and B. N. Markovsky (1995) 'Justice and injustice' In Sociological Perspectives on Social Psychology, in Cook, K., G. Fine and J. House (ed.), Boston: Allyn and Bacon. Ho, D. E., K. Imai, G. King and E. A. Stuart (2007) 'Matching as nonparametric pre- processing for reducing model dependence in parametric causal inference', Political analysis, 15 (3): 199-236. Hu, J. (2014) '“减负”中的家长阻力分析 [Analysis of the opposition to "academic burden alleviation policy" from parents]', 中国教育学刊 [Journal of the Chinese Society of Education] (1): 19-22. Huber, P. J. 1967. The behavior of maximum likelihood estimates under nonstandard conditions. In: Cam, L. M. L. and J. Neyman, eds. Proceedings of the Fifth Berkeley Symposium on Mathematical Statistics and Probability, Volume 1: Statistics, 1966 Berkeley. University of California Press, 221-233. Huo, B. K. and S. S. Xu (2013) '中国学生学习增负减负的博奕: 囚徒困局的视角 [The game of alleviating students' academic burden: the perspective of prisoners' dilemma]', 全球教育展望 [Global Education], 42 (4): 17-26. Husen, T. (1975) Social Influences on Educational Attainment: Research Perspectives on Educational Equality. Paris: Organisation for Economic Co-operation and Development. Imbens, G. W. (2000) 'The role of the propensity score in estimating dose-response functions', Biometrika, 87 (3): 706-710. Jæ ger, M. M. (2006) 'Welfare regimes and attitudes towards redistribution: the regime hypothesis revisited', European Sociological Review, 22 (2): 157-170. Jasso, G. and B. Wegener (1997) 'Methods for empirical justice analysis: part 1. framework, models, and quantities', Social Justice Research, 10 (4): 393-430. Jiang, L. Z. (2006) '保送生制度: 从政治化到人本化 [Admission to colleges based on recommendation: from political considerations to merit system] ', 郑州大学学

243

报 (哲学社会科学版) [Journal of Zhengzhou University (Philosophy and Social Sciences Edition)], 39 (2): 82-84. Jiao, W. (2015) 武汉民间奥数利益链调查 [Investigation into the profit chain of Mathematical Olympiad extracurricular tutoring industry in Wuhan Municipality], 长江商报 [Changjiang Times]. 22 October, available at http://news.sina.com.cn/c/sd/2015-10-22/doc-ifxizwsf8718194.shtml accessed 25 May 2017. Jin, D. D. (2012) 杭州公办小学初中择校名额一个不留 [No enrolment position for school-choosing students in public primary and junior high schools of Hangzhou], 浙江在线 [Zhejiang Online], Accessed 26 June 2017. http://zjnews.zjol.com.cn/05zjnews/system/2012/04/07/018392714.shtml. Jin, X. Y. (2017) 中国教育研究应转向实证研究范式 [Chinese educational studies should shift to the empirical research paradigm], 光明日报 [Guang Ming Daily]. 2 March, available at http://edu.people.com.cn/n1/2017/0302/c1053- 29117547.html accessed 23 June 2017. Jin, Y. M. and Y. G. Tang (2004) 中国素质教育政策研究 [Research on the Quality- Oriented Education of China]. Jinan: 山东教育出版社 [Shandong Education Press]. Johnson, R. B., A. J. Onwuegbuzie and L. A. Turner (2007) 'Toward a definition of mixed methods research', Journal of mixed methods research, 1 (2): 112-133. Karlson, K. B. (2015) 'Another Look at the Method of Y-Standardization in Logit and Probit Models', The Journal of Mathematical Sociology, 39 (1): 29-38. Karlson, K. B., A. Holm and R. Breen (2012) 'Comparing regression coefficients between same-sample nested models using logit and probit a new method', Sociological Methodology, 42 (1): 286-313. Kim, S. and J.-H. Lee (2010) 'Private tutoring and demand for education in South Korea', Economic Development and Cultural Change, 58 (2): 259-296. Kipnis, A. (2006) 'Suzhi: a keyword approach', The China Quarterly, 186 (186): 295-313. ——— (2007) 'Neoliberalism reified: suzhi discourse and tropes of neoliberalism in the People's Republic of China', Journal of the Royal Anthropological Institute, 13 (2): 383-400.

244

Kipnis, A. B. (2011) 'Subjectification and education for quality in China', Economy and Society, 40 (2): 289-306. Law, W. W. (2014) 'Understanding China's curriculum reform for the 21st century', Journal of Curriculum Studies, 46 (3): 332-360. Lerner, M. J. and D. T. Miller (1978) 'Just world research and the attribution process: Looking back and ahead', Psychological Bulletin, 85 (5): 1030-1051. Leuven, E. and B. Sianesi (2003) PSMATCH2: Stata module to perform full Mahalanobis and propensity score matching, common support graphing, and covariate imbalance testing (4.0.11). [Stata user-written command]. [Accessed 27 August 2016]. Li, B. (2017a) 怎样让素质教育真正硬气起来（人民时评） [How to make the Quality- oriented Education stronger (People's Review)], 人民日报 [People's Daily]. 12 April, available at http://paper.people.com.cn/rmrb/html/2017- 04/12/nw.D110000renmrb_20170412_1-05.htm accessed 25 May 2017. Li, C. L. (2003) 'Social and Political Changes and Inequality in Educational Opportunities: On the Impact of Family Background and Institutional Factors on Educational Attainment (1940-2001)', Social Sciences in China (4): 62-79. Li, C. L. (2010) '高等教育扩张与教育机会不平等——高校扩招的平等化效应考查 [Expansion of higher education and inequality in opportunity of education: a study on effect of "Kuozhao" policy on equalization of educational attainment]', 社会学研究 [Sociological Studies], 25 (3): 82-113. ——— (2014a) '教育不平等的年代变化趋势（1940-2010）——对城乡教育机会不平等的再考察 [The changing trend of educational inequality in China (1940-2010): re-examining the urban-rural gap on educational opportunity]', 社会学研究 [Sociological Studies], 29 (2): 65-89. Li, D. (2014b) 中国民众收入分配认知的共识、分歧与变迁 [Chinese People's Perception of Income Distribution: Consensus, Difference and Transition]. Beijing: 中国人民大学国家发展与战略研究院 [National Academy of Development and Strategy, Renmin University of China]. ——— (2017b) '阈值依赖的教育扩张与教育机会不平等——以Ａ省某年高考数据为例 [Threshold dependent inequality caused by education expansion: the case of

245

national college entrance examination]', 社会学研究 [Sociological Studies], 32 (3): 71-94. Li, J. and X. G. Wu (2012) '收入不平等与公平分配: 对转型时期中国城镇居民公平观的一项实证分析 [Income inequality and distributive justice: an empirical analysis of urban residents’ view of equity in transitional China]', 中国社会科学 [Social Sciences in China] (3): 114-128. Li, L. Q. (2000) '深化教育改革，全面推进素质教育，为实现中华民族的伟大复兴而奋斗 ( 摘要)—— 在第三次全国教育工作会议上的报告 [Deepen the educational reform, fully advance the quality-oriented education, and strive to achieve the great renewal of the Chinese nation (abstract) - delivered at the 3rd National Education Work Conference]' In 中国教育年鉴 2000 [China Education Yearbook 2000], in The Editorial Office (ed.), Beijing: 人民教育出版社 [People's Education Press]: 14-16. Li, N. (2015) '影子教育在我国的发展趋势 [The development trend of shadow education in China]', 中国教育学刊 [Journal of the Chinese Society of Education] (5): 54-57. Li, Y. (2006) '制度变迁与教育不平等的产生机制——中国城市子女的教育获得 (1966—2003) [Institutional change and educational inequality: mechanisms in educational stratification in urban China (1966—2003)]', 中国社会科学 [Social Sciences in China] (4): 97-109. Li, Z. J. and Z. Yang (2011) '我国高考加分政策演变的制度分析——基于历史制度主义的分析范式 [The transition of extra credit policies in the China college entrance examination: the analysis from the perspective of institutionalism paradigm]', 清华大学教育研究 [Tsinghua Journal of Education], 32 (1): 61-67. Li, Z. L. and Z. Q. Qiu (2016) '家庭背景如何影响儿童学业成就?——义务教育阶段家庭社会经济地位影响差异分析 [Family background and children’s academic performance: evidence from the compulsory education in China]', 社会学研究 [Sociological Studies], 31 (4): 121-144.

246

Liang, Q. Y. and G. C. Sun (2009) '孔子因材施教的理论基础及启示 [Confucius’ theoretical basis of teaching students according to their aptitude and its enlightenment]', 教育研究 [Educational Research] (11): 87-91. Liao, W. and R. Yuan (2017) 'Understand an emerging "failure" of an equality-oriented teacher policy in China: A job search perspective', International Journal of Educational Research, 81: 71-82. Lin, J. and Y. Zhang (2006) 'Educational expansion and shortages in secondary schools in China: The bottle neck syndrome', Journal of Contemporary China, 15 (47): 255- 274. Listhaug, O. and T. Aalberg (1999) 'Comparative public opinion on distributive justice', International Journal of Comparative Sociology, 40 (1): 117-140. Little, A. W. (2000) 'Globalisation, qualifications and Livelihoods: towards a research agenda', Assessment in Education: Principles, Policy & Practice, 7 (3): 299-312. Liu, A. and Y. Xie (2016) 'Why do Asian Americans academically outperform Whites? The cultural explanation revisited', Social Science Research, 58: 210-226. Liu, F. X. (2002) '素质教育政策与《美国 2061 计划》 [The Quality-oriented Education Policy and "Project 2061" of United States]', 教育发展研究 [Research in Educational Development] (10): 27-31. Liu, J. (2012) 'Does cram schooling matter? Who goes to cram schools? Evidence from Taiwan', International Journal of Educational Development, 32 (1): 46-52. Liu, J. and M. Bray (2017) 'Determinants of demand for private supplementary tutoring in China: findings from a national survey', Education Economics, 25 (2): 205-218. Liu, J. M. (2006) '高等教育扩展与入学机会差异: 1978～2003 [Expansion of higher education in China and inequality in entrance opportunities: 1978-2003]', 社会 [Society], 26 (3): 158-179. Liu, J. M. (2008) '中国基础教育领域中的机会不平等及其变化 [Changing patterns of inequality of educational opportunity in basic education in China]', 中国社会科学 [Social Sciences in China] (5): 101-116. Liu, X. and A. N. Hu (2016) '中国公众的收入公平感：一种新制度主义社会学的解释 [Perception of income fairness: a sociological new institutionalist explanation]', 社会 [Chinese Journal of Sociology], 36 (4): 1-24.

247

Liu, Y. J. and M. Dunne (2009) 'Educational reform in China: tensions in national policy and local practice', Comparative Education, 45 (4): 461-476. Liu, Y. Y. and Y. Xie (2014) 家庭背景、课外辅导与儿童的学业成绩 [Family background, private tutoring, and children’s academic achievement in recent China]. Beijing: 北京大学中国社会科学调查中心 [Institute of Social Science Survey, Peking University]. Liu, Z. Y. (2016) '“拼娃时代”, 我们的教育该如何安放 [The "child-pitting" era: where our education should go]', 人民周刊 [People’s weekly] (22): 18-19. Lockette, K. F. (2012) 'Creativity and Chinese education reform: a review of literature', International Journal of Global Education, 1 (4): 34-39. Lou, J. (2011) 'Suzhi, relevance, and the New Curriculum: a case study of one rural middle school in Northwest China', Chinese Education & Society, 44 (6): 73-86. Lou, S. Z. (2013) '“影子教育”治理的困境与教育政策的选择' [The dilemma of "shadow education" and the choice of basic education policy]', 教育发展研究 [Research in Educational Development] (18): 76-79. Lu, H. L. (2006) ' 集体化与农民平均主义心态的形成 [The experiences of collectivization and the making of peasant egalitarianism mentality]', 社会学研究 [Sociological Studies], 21 (6): 147-164. Lu, Y. and D. J. Treiman (2008) 'The effect of sibship size on educational attainment in China: period variations', American Sociological Review, 73 (5): 813-834. Lucas, S. R. (2001) 'Effectively maintained inequality: education transitions, track mobility, and social background effects', American Journal of Sociology, 106 (6): 1642-1690. Luo, H. Y. (2009) '素质教育政策执行失效的归因分析 [Research on the factors in the ineffective implementation of the Quality-oriented Education Policy]', 上海教育科研 [Shanghai Research on Education] (10): 13-15. Ma, L. and X. Liu (2010) '中国城市居民的分配公平感研究 [People’s perception of distributive justice in urban China]', 社会学研究 [Sociological Studies], 25 (5): 31-49.

248

Mangin, V. (2015) China's new found love of private education, Accessed 16 August 2017. http://www.bbc.com/capital/story/20151221-the-chinese-turning-their-backs- on-the-state. Mare, R. D. (1980) 'Social background and school continuation decisions', Journal of the American Statistical Association, 75 (370): 295-305. ——— (1981) 'Change and stability in educational stratification', American Sociological Review, 46 (1): 72-87. Marton, A. M. (2006) 'The cultural politics of curricular reform in China: a case study of geographical education in Shanghai', Journal of Contemporary China, 15 (47): 233-254. Maslow, A. H. (1987) Motivation and Personality. 3rd ed. New York: Addison-Wesley Educational Publishers Inc. Meng, T. G. (2012) '转型期中国公众的分配公平感: 结果公平与机会公平 [Chinese people’s perception of distributive justice in transitional China: outcome justice and opportunity justice]', 社会 [Chinese Journal of Sociology], 32 (6): 108-134. Meng, X., C. , X. X. Chen, R. F. Zhang and Z. M. Zhou (2010) 中国古代教育史资料 [The Historical Material of Education in Ancient China]. Shanghai: 华东师范大学出版社 [East China Normal University Press]. Ministry of Education (1983) 31 December.关于全日制普通中学全面贯彻党的教育方针、纠正片面追求升学率倾向的十项规定 [The Ten-point Regulation on Fully Implementing the Party's Educational Policy and redressing the tendency of blindly pursuing the high promotion rate]. ——— (2000) 教基 [2000] 1 号. 3 January. 关于在小学减轻学生过重负担的紧急通知 [Urgent Notice of Alleviating the Overweight Academic Burden on Primary School Students]. ——— (2009) 2008 年教育统计数据 [2008 Educational Statistics], 教育部网站 [The website of Ministry of Education], Accessed 18 September 2016. http://old.moe.gov.cn/publicfiles/business/htmlfiles/moe/s7567/index.html. ——— (2012) 2011 年全国来华留学生数据统计 [2011 National Statistics of International Students in China], 教育部网站 [The website of Ministry of Education], Accessed 12 September 2016.

249

http://www.moe.gov.cn/publicfiles/business/htmlfiles/moe/s5987/201202/13 1117.html. ——— (2014) 教基一 [2014] 1 号. 14 January. 关于进一步做好小学升入初中免试就近入学工作的实施意见 [Implementation Opinions on Further Effectively Conducting the Policy of Junior High School Admitting Students in their Vicinity without Examination]. ——— (2016) 教基二[2016] 4 号. 18 September. 教育部关于进一步推进高中阶段学校考试招生制度改革的指导意见 [Guiding Opinions of the Ministry of Education on Further Advancing the Senior High School Admission Examination System Reform]. ——— (2017) 实现五个确保力促教育公平教育部部署 2017 年普通高等教育招生计划管理工作 [Achieve five Assurances, vigorously promote educational equality: the Ministry of Education arranging the administration work of 2017 university admission quota], 教育部网站 [The Website of Ministry of Education], Accessed 26 July 2017. http://www.moe.edu.cn/jyb_xwfb/gzdt_gzdt/moe_1485/201705/t20170510_3 04226.html. Mok, K.-H. (2006) Education reform and education policy in East Asia: Routledge. Mok, K. H. and Y. W. Lo (2007) 'The impacts of neo-liberalism on China’s higher education', Journal for Critical Education Policy Studies, 5 (1): 316-348. Mok, K. H., Y. C. Wong and X. L. Zhang (2009) 'When marketisation and privatisation clash with socialist ideals: educational inequality in Urban China', International Journal of Educational Development, 29 (5): 505-512. National Bureau of Statistics of China (NBSC) (2013) '就业人员和劳动报酬 5-1 劳动参与率 [Employed persons and wages of staff and workers 5-1 The labor participation rate]' In 国际统计年鉴 2012 [International Statistical Yearbook 2012], in National Bureau of Statistics of China (ed.), Beijing: 中国统计出版社 [China Statistics Press]: 121. ——— (2014) 年度数据：教育（最近 10 年）[Annual Statistics: Education (last ten years)], National Bureau of Statistics of China Accessed 12 September 2016. http://data.stats.gov.cn/easyquery.htm?cn=C01.

250

——— (2016a) 年度数据--教育-- 各级各类民办教育在校学生数 [Annual Statistics - - Education -- Enrolment of Students of Non-government Formal Education by Type and Level], National Bureau of Statistics of China Accessed 23 March 2016. http://data.stats.gov.cn/easyquery.htm?cn=C01. ——— (2016b) 年度数据--教育--各级各类学历教育在校学生数 [Annual Statistics -- Education -- Enrolment of Students of Formal Education by Type and Level]. National Bureau of Statistics of China Accessed 23 March 2017. http://data.stats.gov.cn/easyquery.htm?cn=C01. ——— (2016c) 年度数据--教育--各级各类学历教育毕业生数 [Annual Statistics -- Education -- Graduate Number of Students of Formal Education by Type and Level], National Bureau of Statistics of China Accessed 23 March 2016. http://data.stats.gov.cn/easyquery.htm?cn=C01. ——— (2016d) 年度数据--教育--各级各类学历教育招生人数 [Annual Statistics -- Education -- Entrant Number of Students of Formal Education by Type and Level], National Bureau of Statistics of China Accessed 23 March 2016. http://data.stats.gov.cn/easyquery.htm?cn=C01. ——— (2016e) 城乡居民家庭人均收入及恩格尔系数（1988- 2012）[Per Capita Annual Income and Engle’s Coefficient of Urban and Rural Households (1988- 2012)], National Bureau of Statistics of China Accessed 25 April 2016. http://data.stats.gov.cn/easyquery.htm?cn=C01. Ning, L. H. and N. H. Lv (2012) ' 我国基础教育新课程改革的十年审思与展望 [Reflection and prospect of the ten-year New Curriculum Reform of basic education in China]', 教育研究与实验 [Educational Research and Experiment] (5): 57-62. Pang, S. M. (2016) '市场转型, 教育分流与中国城乡高等教育机会不平等 (1977- 2008) [Market transition, educational differentiation and urban-rural inequality in Chinese higher education (1977-2008)]', 社会 [Chinese Journal of Sociology], 36 (5): 155-174. Park, H., S.-y. Byun and K.-k. Kim (2011) 'Parental involvement and students’ cognitive outcomes in Korea focusing on private tutoring', Sociology of Education, 84 (1): 3-22.

251

Price Control Administration, Finance Bureau and Bureau of Education (2017a) February. 2017 年南京市普通高中收费标准 [2017 Fees for Regular Senior High school of Nanjing Municipality]. http://www.nanjing.gov.cn/zxfw_6528/xinzhutifuwu/jyfw_20963/gzjy/gzsfbz/2 01702/t20170213_4361419.shtml. ——— (2017b) February. 2017 年南京市普通中专、职业中专、职业高中收费标准 [2017 Fees for Vocational Senior High school of Nanjing Municipality]. http://www.nanjing.gov.cn/zxfw_6528/xinzhutifuwu/jyfw_20963/zyjy_21148/z yjysfbz/201702/t20170213_4361415.shtml. Qian, H. Y. (2016) 南京 2017 年中考体育跳绳长跑满分标准提高 [The full-mark criteria for the skipping-rope and the long-distance running in the sport examination of 2017 senior high school admission examination become higher], 南京日报 [Nanjing Daily]. 20 September, available at http://js.people.com.cn/n2/2016/0920/c360307-29026306.html accessed 25 May 2017. Qiao, X. H. (2014) '国内近二十年关于 “减负” 问题的研究述论 [Review on the academic-burden alleviation research in China in the last twenty years]', 山西师大学报: 社会科学版 [Journal of Shanxi Normal University (Social Science Edition)], 41 (S4): 32-34. QSR (2016) NVivo Pro 11 for Windows. [Software]. . Raftery, A. E. and M. Hout (1993) 'Maximally maintained inequality: expansion, reform, and opportunity in Irish education, 1921-75', Sociology of Education, 66 (1): 41- 62. Raudenbush, S. W. and R. D. Eschmann (2015) 'Does schooling increase or reduce social inequality?', Annual Review of Sociology, 41 (1): 443-470. Robinson, R. V. and W. Bell (1978) 'Equality, success, and social justice in England and the United States', American Sociological Review, 43 (2): 125-143. Ruan, C. W. (2008) '由冲突到分享: 素质教育政策新视角 [From conflicting to sharing: the new perspective of Quality-oriented Education policy]', 中国教育学刊 [Journal of the Chinese Society of Education] (3): 7-10.

252

——— (2010) '教育利益协调:素质教育政策的制度创新及其路径 [Educational interests’ coordination: the institutional innovation of quality-oriented education policy and the path]', 教育发展研究 [Research in Educational Development] (20): 52-57. ——— (2011) '"素质教育新规”触动谁的敏感神经 [Whose sensitive nerves have been touched by "new regulation of Quality-oriented Education"]', 教育研究 [Educational Research] (6): 17-22. Rubin, D. B. (2001) 'Using propensity scores to help design observational studies: application to the tobacco litigation', Health Services and Outcomes Research Methodology, 2 (3): 169-188. Sargent, T., M. Y. Chen, Y. J. Wu and C. T. Chen (2011) 'Wearing new shoes to walk the old road: the negotiation of opposing imperatives in high school new curriculum classes in China' In Impact and Transformation of Education Policy in China (International Perspectives on Education and Society, Volume 15) in Huang, T. and A. W. Wiseman (ed.), Bingley: Emerald Group Publishing Limited: 79-98. Sears, D. O. and C. L. Funk (1991) 'The role of self-interest in social and political attitudes', Advances in experimental social psychology, 24: 1-91. Secada, W. G. (1989) 'Educational equity versus equality of education: An alternative conception' In Equity in Education, in Secada, W. G. (ed.), London: Falmer Press: 68-88. Shadish, W. R., T. D. Cook and D. T. Campbell (2002) Experimental and Quasi- Experimental Designs for Generalized Causal Inference. Boston Houghton Mifflin Company. Shavit, Y. and H.-P. Blossfeld (ed.) (1993) Persistent Inequality: Changing Educational Attainment in Thirteen Countries. Social Inequality Series. Boulder: Westview Press. Shi, H. L. (2016) '“综合素质评价”改革朝哪里去? [Where will the "Comprehensive Quality Evaluation" reform go?]', 教育科学研究 [Educational Science Research] (4): 22-26. Shi, J. and H. Y. Qian (2006) 新学期中小学幼儿园收费标准出台 [The Fees for kindergarten, primary and secondary school in the new semester], 南京日报

253

[Nanjing Daily]. 10 February available at http://news.sohu.com/20060210/n241764235.shtml accessed 3 March 2017. Shouse, R. C. and C. W. Ma (2015) 'Leadership and creativity in East Asian schools', Asia Pacific Education Review, 16 (4): 491-499. Sohu Research Institute of Education (2014) 中国 K12 课外辅导行业现状及走向（2014 年教育行业白皮书特刊） [The Current Situation and Trend of K12 Extracurricular Tutoring Industry in China (The Special Issue on 2014 White Paper of Educational Industry)]. Beijing: 搜狐教育研究院 [Sohu Research Institute of Education]. Solinger, D. J. (1999) Contesting Citizenship in Urban China: Peasant Migrants, the State, and the Logic of the Market. Berkeley: University of California Press. Standing Committee of the National People's Congress (SCNPC) (2015a) 主席令第 25 号 [Order No. 25 of the President]. 24 April. 义务教育法（2015 修正） [Compulsory Education Law (2015 Amendment)], Article 12 ——— (2015b) 主席令第 30 号 [Order No. 30 of the President]. 29 August. 刑法修正案(九) [Amendment (IX) to the Criminal Law of the People's Republic of China], Article 25 Standing Committee of the People's Congress of Jiangsu Province (SCPCJP) (2002) 江苏省人大常委会公告第 21 号. 24 October 江苏省人口与计划生育条例 [Regulations of Jiangsu Province on Population and Family Planning], Article 44. StataCorp (2015a) Stata Base Reference Manual (Release 14). College Station, TX: StataCorp LP. ——— (2015b) Stata Statistical Software (Release 14). [Software]. . ——— (2015c) Stata Treatment-Effects Reference Manual: Potential Outcomes/Counterfactual Outcomes (Release 14). College Station, TX: StataCorp LP. State Council (1988) 国务院令第 4 号 [Order No. 4 of the State Council]. 25 June. 中华人民共和国私营企业暂行条例 [Provisional Regulations of the People's Republic of China on Private Enterprises].

254

——— (1994) 国务院令第 146 号 [Order No. 146 of the State Council]. 3 February. 关于职工工作时间的规定 [Provisions of the State Council on Working Hours of Employees (1994)], Article 3. ——— (1995) 国务院令第 174 号 [Order No. 174 of the State Council]. 25 March. 国务院关于职工工作时间的规定 [Provisions of the State Council on Working Hours of Employees (1995)], Article 3. ——— (1997) 国务院令第 226 号 [Order No. 226 of the State Council]. 31 July. 社会力量办学条例 [Provisions for Schools Run by Social Resources]. ——— (2014a) 国发〔2014〕25 号 [No. 25 [2014] of the State Council]. 24 July. 国务院关于进一步推进户籍制度改革的意见 [Opinions of the State Council on Further Promotion of Reform of the Household Registration System]. ——— (2014b) 国发〔2014〕35 号[No. 35 [2014] of the State Council]. 3 September. 国务院关于深化考试招生制度改革的实施意见 [Implementation Opinions of the State Council on Deepening Reforms to the Student Examination and Admission System]. State Education Commission (1988) 教督 [1988] 001 号. 11 May. 关于减轻小学生课业负担过重问题的若干规定 [Several Provisions on Alleviating the Overweight Academic Burden on Primary School Students]. ——— (1993) 教基 [1993] 3 号. 24 March. 关于减轻义务教育阶段学生过重课业负担、全面提高教育质量的指示 [Instructions on Alleviating the Overweight Academic Burden on Students at the Compulsory Education Stage and fully Improving the Educational Quality]. ——— (1994) 10 November. 关于全面贯彻教育方针，减轻中小学生过重课业负担的意见 [Opinions on Fully Implementing the Educational Policy and Alleviating the Overweight Academic Burden on Primary and Secondary School Students]. ——— (1997) 教办 [1997] 29 号. 29 October. 关于当前积极推进中小学实施素质教育的若干意见 [Several Opinions on Currently Vigorously Advancing the Implementation of Quality-oriented Education in Primary and Secondary Education].

255

Stuart, E. A., B. K. Lee and F. P. Leacy (2013) 'Prognostic score–based balance measures can be a useful diagnostic for propensity score methods in comparative effectiveness research', Journal of clinical epidemiology, 66 (8): S84-S90. e81. Sudworth, J. (2016) Counting the cost of China’s left-behind children, BBC News, Accessed 16 September. http://www.bbc.com/news/world-asia-china- 35994481. Sun, M. (2009) '市场转型与民众的分配公平观 [Market transition and people’s perception of the principles of distributive justice in contemporary China]', 社会学研究 [Sociological Studies], 24 (3): 78-88. Sun, M. T., C. P. Kang and X. L. Zhu (2009) '改革开放以来中国就近入学政策的内容分析 [Content analysis of Nearby Enrolment Policy in China since the Reform and Opening up]', 教育理论与实践 [Theory and Practice of Education], 29 (9): 16- 20. Sun, Z. J. and Y. H. Du (2010) '中国义务教育财政制度改革: 进展, 问题与建议 [The financial institution reform of Chinese compulsory education: progress, problem and suggestion]', 华中师范大学学报: 人文社会科学版 [Journal of Huazhong Normal University (Humanities and Social Sciences)], 49 (1): 113-119. Svallfors, S. (1993) 'Dimensions of inequality: a comparison of attitudes in Sweden and Britain', European Sociological Review, 9 (3): 267-287. Tam, T. and J. Jiang (2015) 'Divergent urban-rural trends in college attendance: state policy bias and structural exclusion in China', Sociology of Education, 88 (2): 160- 180. Tan, C. and C. S. K. Chua (2015) 'Education policy borrowing in China: has the West wind overpowered the East wind?', Compare: a Journal of Comparative and International Education, 45 (5): 686-704. Tan, S. (2011) '中国农村留守儿童研究述评 [Review on studies of the left-behind children in the rural area of China]', 中国社会科学 [Social Sciences in China] (1): 140-141. Tao, X. P. and Z. G. Yuan (2010) '加强统筹协调促进教育公平 [Enhance the overall coordination, promote the educational equality]', 教育研究 [Educational Research] (7): 10-11.

256

Third Research Group (2006) '新时期素质教育面临的突出问题与应试教育愈演愈烈的原因及其对策建议 [The problems of quality education in the new period and the causes and suggestions for the ever stronger presence of test-oriented education]' In 对策与建议: 2005-2006 年度教育热点、难点问题分析 [Countermeasures and suggestions: analyses on educational hot issues and difﬁcult problems in year 2005-2006], in Zhu, X. (ed.), Beijing: 教育科学出版社 [Educational Sciences Publishing House]: 85-93. Verwiebe, R. and B. Wegener (2000) 'Social inequality and the perceived income justice gap', Social Justice Research, 13 (2): 123-149. Vickers, E. and X. Zeng (2017) Education and Society in Post-Mao China: Taylor & Francis. Wang, C. (2013) '规范我国教育培训市场的思考 [Thoughts on normalizing the educational tutoring industry in China]', 教学与管理 [Journal of Teaching and Management] (3): 7-9. Wang, D. (2011a) 'The dilemma of time: student-centered teaching in the rural classroom in China', Teaching and Teacher Education, 27 (1): 157-164. Wang, F. (2011b) 'Perceptions of distributive justice of the residents in contemporary China' s large cities: an empirical study based on the survey in Shanghai', Chinese Journal of Sociology, 31 (3): 155-183. Wang, H. (1999) ' 天津市全面加强初中建设，大面积提高教育质量 [Tianjin Municipality comprehensively enhancing the development of junior high school and massively improving the educational quality]' In 中国教育年鉴 1988 [China Education Yearbook 1988], in Board, T. E. (ed.), Beijing: 人民教育出版社 [People's Education Press]: 91-95. Wang, J. Y. (2006) ' 农村中小学实施素质教育的困难与对策 [Difficulties and countermeasures in the implementation of Quality-oriented Education in rural primary and secondary schools]', 教育研究 [Educational Research] (11): 41-46. Wang, L. (2014) 'China's janus-faced approach to su zhi education: a capability perspective', Journal of Human Development and Capabilities, 15 (4): 308-319. Wang, P. (2011c) 转型期城市贫困家庭子女义务教育的比较研究 [Comparative Research on the Compulsory Education of Urban Poor Families during the Social Transition Period]. Ph. D. thesis, Fudan University.

257

Wegener, B. (1991) 'Relative deprivation and social mobility: structural constraints on distributive justice judgments', European Sociological Review, 7 (1): 3-18. Wegener, B. and S. Liebig (1995) 'Hierarchical and social closure conceptions of distributive social justice: a comparison of East and West Germany' In Social justice and political change: Public opinion in capitalist and post-communist states, in Kluegel, J. R., D. S. Mason and B. Wegener (ed.), New York: Walter de Gruyter: 263-284. Wen, D. M. and Y. B. Liu (2014) '高考改革何以 “牵一发而动全身” [Why the university admission examination reform is far-reaching]', 中国高等教育 [China Higher Education] (24): 19-22. Werfhorst, H. G. V. d. and J. J. B. Mijs (2010) 'Achievement inequality and the institutional structure of educational systems: a comparative perspective', Annual Review of Sociology, 36 (1): 407-428. White, H. (1980) 'A heteroscedasticity-consistent covariance matrix estimator and a direct test for heteroscedasticity', Econometrica: Journal of the Econometric Society, 48 (4): 817-838. Whyte, M. K. (2009) '中国民众如何看待当前的社会不平等 [Views of Chinese citizens on current inequalities]', 社会学研究 [Sociological Studies], 24 (1): 96-120. Woronov, T. E. (2008) 'Raising quality, fostering "creativity": ideologies and practices of education reform in Beijing', Anthropology & Education Quarterly, 39 (4): 401- 422. Wu, J. (2016) 'Educational discipline, ritual governing, and Chinese exemplary society: why China's curriculum reform remains a difficult task', Policy Futures in Education, 14 (6): 721-740. Wu, X. (2009) 'Income inequality and distributive justice: a comparative analysis of Mainland China and Hong Kong', China Quarterly, 200 (1): 1033-1052. ——— (2010) 'Economic transition, school expansion and educational inequality in China, 1990–2000', Research in Social Stratification and Mobility, 28 (1): 91-108. Wu, X. and J. M. Perloff (2005) 'China's income distribution, 1985–2001', The Review of Economics and Statistics, 87 (4): 763-775.

258

Wu, X. and Z. Zhang (2010) 'Changes in educational inequality in China, 1990–2005: Evidence from the population census data' In Globalization, Changing Demographics, and Educational Challenges in East Asia, in Hannum, E., H. Park and Y. G. Butler (ed.), Bingley: Emerald Group Publishing Limited: 123-152. Wu, X. G. (2011) 'The household registration system and rural-urban educational inequality in contemporary China', Chinese Sociological Review, 44 (2): 31-51. Wu, X. W. (2014) '中国教育不平等的变动趋势 : 队列视角的考察 [Trends of education inequality in China: a cohort perspective]', 人口研究 [Population Research], 38 (6): 76-91. Wu, Y. X. (2008) 'Cultural capital, the state, and educational inequality in China, 1949- 1996', Sociological Perspectives, 51 (1): 201-227. Wu, Y. X. (2012) '中国城乡居民教育获得的性别差异研究 [Gender gap in educational attainment in urban and rural China]', 社会 [Chinese Journal of Sociology], 32 (4): 112-137. ——— (2013) 'Educational Opportunities of Chinese Rural and Urban Residents in 1978- 2008: Inequality and Evolution', Social Sciences in China (3): 58-75. Xie, Y. (2010) '认识中国的不平等 [Understanding the inequality in China]', 社会 [Chinese Journal of Sociology], 30 (3): 1-20. Xie, Y., Y. P. Ma and Z. P. Zhang (2013) '新中国真的发生了八次课程改革吗? [Is it true that the curriculum reform has been really carried out for eight times in New China?]', 教育研究 [Educational Research] (2): 125-132. Xu, Z. F. (2009) 兰州市家庭选择课外辅导的影响因素之质性研究 [Qualitative research on the impact factors why households choose the after-class tutoring in Lanzhou]. Master's Degree thesis, 西北师范大学 [Northwest Normal University]. Xue, H. P. and X. H. Ding (2009) '中国城镇学生教育补习研究 [A study on additional instruction for students in cities and towns in China]', 教育研究 [Educational Research] (1): 39-46. Xun, Z. F. and Q. H. Wang (2011) ' 自主招生 : 精英角逐的场域 [Independent recruitment: a battle field for elite students]', 清华大学教育研究 [Tsinghua Journal of Education], 32 (2): 56-63.

259

Yan, G. C. (2002) 素质教育概论 [Introduction to the Quality-Oriented Education]. Guangzhou: 广东教育出版社 [Guangdong Education Publishing House]. ——— (2007) '田野札记: 中国民办教育制度演进的逻辑 [Filed work notes: the logic of the development of Chinese private education system]', 民办教育研究 [Private Education Research], 6 (4): 1-6. ——— (2011) '差异社会中的机会均等:素质教育展开的制度困境 [The opportunity equality in a differentiated society: the institutional difficulties of implementing the quality-oriented education]', 华中师范大学学报 ( 人文社会科学版 ) [Journal of Huazhong Normal University (Humanities and Social Sciences)], 50 (3): 139-143. Yan, H. (2003) 'Neoliberal governmentality and neohumanism: organizing suzhi/value flow through labor recruitment networks', Cultural Anthropology, 18 (4): 493- 523. Yang, D. P. (2005) 改革“应试教育”的治标与治本 [The temporary and permanent approaches to reforming the "examination-oriented education"], 人民日报 [People's Daily]. 24 November available at http://news.xinhuanet.com/comments/2005-11/24/content_3828725.htm accessed 23 May 2017. ——— (2006a) 中国教育公平的理想与现实 [The Ideal and Reality of Educational Equality in China]. Beijing: 北京大学出版社 [Peking University Press]. Yang, J., X. Huang and X. Liu (2014) 'An analysis of education inequality in China', International Journal of Educational Development, 37: 2-10. Yang, N. L. (2016) 中小学课外辅导领域的教育创新及探索 [Innovation and exploring in the extracurricular tutoring industry of primary and secondary school]. Paper presented at 辅导机构教师专业规范的探索与实践研讨会 [The Workshop on the tutor's codes of practice in the extracurricular tutoring industry], Beijing, 27 December. Yang, P. Y. (2017) 择校热蔓延民办校"小升初"招考转入地下教育部禁令成空文 [School-choosing fashion spreading, the private junior high school's admission examination camouflaging, the Ministry of Education's ban vain], Wei lai wang

260

未来网 [K618.cn], Accessed 26 July 2017. http://edu.news.k618.cn/yc/201706/t20170613_11690643.html. Yang, Q. G. (2009a) '发展型家庭补偿教育政策的构建 [The policy construction of compensatory education for developing families]', 教育科学 [Education Science], 25 (5): 15-20. Yang, R. Y. (2006b) '关于素质教育政策体系、内容的政策学分析与建议 [Analysis and suggestions on the educational policy system and content of quality education from the perspective of policy]', 教育理论与实践 [Theory and Practice of Education], 26 (9): 15-19. ——— (2006c) '素质教育政策运行过程中的评估问题分析 [The analysis of the evaluation issue in the process of implementing the quality-oriented education policy]', 江西教育科研 [Jiangxi Education Research] (7): 11-13. Yang, X. M. and M. J. Zhu (2003) '基础教育课程改革 [Curriculum reform of basic education]' In 中国教育年鉴 2002 [China Education Yearbook 2002], in Zheng, S. S. (ed.), Beijing: 人民教育出版社 People's Education Press: 140. Yang, Z. F. (2009b) '中国人的人己观: “自己” 的“大我”与“小我” [Chinese conceptions of Self and Others: the "Big Self" and the "Small Self" in the "Self"] ' In 如何理解中国人: 文化与个人论文集 [How to Understand Chinese People: the Collection of Culture and Self Research Articles], in Yang, Z. F. (ed.), Chongqing: 重庆大学出版社 [Chongqing University Press]: 321-355. Yu, H. and X. Ding (2011a) 'How to get out of the prisoners dilemma: Educational resource allocation and private tutoring', Frontiers of Education in China, 6 (2): 279-292. ——— (2011b) 'How to get out of the prisoners’ dilemma: Educational resource allocation and private tutoring', Frontiers of Education in China, 6 (2): 279. Yu, J. F. (2008) ' 促进人的全面发展提升国民综合素质 [Promoting people's development in all-around way upgrading the national comprehensive quality]', 教育研究 [Educational Research] (12): 3-10. Yu, X. L. (2016) 底层放弃教育，中产过度焦虑，上层不玩中国高考 [The lowest class give up education, the middle class are too anxious while the upper class just walk

261

out on Chinese ‘gao kao’”], 网易订阅 [163.com Subscription], Accessed 6 June 2017. http://dy.163.com/wemedia/article/detail/C02QR4J20514A4NM.html. Yu, Y. S. (1987) 士与中国文化 [Scholar-gentlemen and Chinese culture]. Shanghai: 上海人民出版社 [Shanghai People's Publishing House]. Yuan, Z. G. (2017) '中国教育需要实证研究 [Chinese education needs empirical studies]', 中国教育学刊 [Journal of the Chinese Society of Education] (2): 3. Yue, C. J. (2015) 'Expansion and equality in Chinese higher education', International Journal of Educational Development, 40: 50-58. Zhai, X. W. (2009) '是 “关系”, 还是社会资本 [Guanxi or social capital?]', 社会 [Society], 29 (1): 109-121. Zhan, S., M. Bray, D. Wang, C. Lykins and O. Kwo (2013) 'The effectiveness of private tutoring: Students’ perceptions in comparison with mainstream schooling in Hong Kong', Asia Pacific Education Review, 14 (4): 495-509. Zhang, W. (2014) 'The demand for shadow education in China: Mainstream teachers and power relations', Asia Pacific Journal of Education, 34 (4): 436-454. Zhang, W. and M. Bray (2015) 'Shadow education in Chongqing, China: Factors underlying demand and policy implications', KEDI Journal of Educational Policy, 12 (1): 83-106. Zhang, W. and M. Bray (2016) 'Shadow education: The rise and implications of private supplementary tutoring' In Spotlight on China: Changes in Education under China's Market Economy, in Guo, S. and Y. Guo (ed.), Rotterdam: Sense Publishers: 85-99. Zhang, W. and M. Bray (2017a) 'Equalising schooling, unequalising private supplementary Tutoring: Access and tracking through shadow education in China' Oxford Review of Education, (Advance online publication). Zhang, W. and M. Bray (2017b) 'Micro-neoliberalism in China: public-private interactions at the confluence of mainstream and shadow education', Journal of Education Policy, 32 (1): 63-81. Zhang, Y. L. (2005) '从数字看教育不公 [Discuss the educational inequality between the urban and rural areas with the statistical data]', 中国改革 [China Reform] (12): 22-24.

262

Zhao, X. L. and X. Zhang (2016) '办学体制改革: 教育现代化的关键选择 [The school- running system reform: the critical choice of educational modernization]', 天津师范大学学报: 基础教育版 [Journal of Tianjin Normal University (Elementary Education Edition)], 17 (4): 1-6. Zheng, G. C. (2013) '让国民福利与国民经济同步发展 [Make the national welfare and national economy to develop simultaneously]', 中国人民大学学报 [Journal of Renmin University of China] (2): 1. Zheng, R. L. (2002) '试析高考的指挥棒作用 [Analysis of the function of university admission examination as a "conductor baton"]', 厦门大学学报: 哲学社会科学版 [Journal of Xiamen University (Arts & Social Sciences)] (2): 7-10. Zhou, G. H. and Z. H. Mao (2007) '我国中小学转制学校研究述评 [A review of the primary and secondary "transformation" schools in China]', 上海教育科研 [Shanghai Research on Education] (9): 25-28. Zhou, M. S. (2013) '教育国际化背景下我国低龄留学原因及利弊探析 [An analysis on the causes and the pros and cons of Chinese young students studying abroad under the background of educational internationalization]', 比较教育研究 [Comparative Education Review], 35 (10): 13-16. Zhou, X., P. Moen and N. B. Tuma (1998) 'Educational Stratification in Urban China: 1949- 94', Sociology of Education, 71 (3): 199-222. Zhu, L. (2002) 中国民工潮 [The Tide of migrant workers in China]. Fuzhou: 福建人民出版社 [Fujian People's Publishing House]. Zhu, Y. X. (2009) '新中国 60 年教育历程及反思 [The sixty year history of New China's education and reflection]', 中国教育学刊 [Journal of the Chinese Society of Education] (11): 7-12.

263

Appendix Appendix 1 Logistic Models Predicting Probability of Student Transiting Entering RSHS with Missing Value Imputed Data M 1 M 2 M3 M 4 M 5 M 6 During vs. Before& 0.666***(3.98) 0.340(1.53) 0.205(0.83) -0.454(-0.84) 0.738*(2.11) -0.757(-1.13) After vs. Before 1.457***(4.79) 0.674(1.52) 0.504(1.11) -0.430(-0.67) 1.217*(2.45) 0.537(0.78) Initial enroll. rate -0.024(-1.17) -0.028(-1.15) -0.029(-1.20) -0.025(-1.05) -0.031(-1.25) Enrollment rate 0.031*(2.47) 0.033**(2.65) 0.035**(2.76) 0.031*(2.43) 0.036**(2.74) Male -0.135(-1.02) -0.118(-0.88) -0.136(-1.02) -0.151(-1.15) Father’s yrs. of edu. 0.055+(1.96) 0.006(0.14) 0.049+(1.87) 0.051+(1.77) Sibling No. 1 sibling vs. None -0.026(-0.11) -0.010(-0.04) 0.659+(1.71) -0.053(-0.24) 2+ siblings vs. None -0.439+(-1.84) -0.441+(-1.88) 0.118(0.46) -0.491*(-2.07) Father’s ISEI 0.015*(2.30) 0.014*(2.28) 0.017**(2.62) 0.006(0.64) Interaction Father’s edu. Sibling No. Father’s ISEI During vs. Before 0.069(1.18) -0.838*(-2.31) # 0.022(1.45) -0.849(-1.58) ## After vs. Before 0.094+(1.67) -1.196**(-2.85) # -0.003(-0.22) -0.930(-1.51) ## After vs. During^ 0.025(0.38) -0.358(-1.05) # -0.025*(-2.00) -0.081(-0.13) ## _cons -0.210(-1.33) -0.320(-1.21) -1.238**(-3.21) -0.809+(-1.95) -1.718***(-3.92) -0.804+(-1.82) N 977 977 977 977 977 977 Pseudo R2 0.052 0.058 0.085 0.088 0.093 0.092 Note: t statistics in parentheses. + p < 0.10, * p < 0.05, ** p < 0.01, *** p < 0.001. The values in the constant and pseudo R2 are from the regression when treating the first policy period as the reference. & In the table, the latter of each two-measure comparison is the reference. For example, “before” is the reference in” During vs. Before”. # The results indicate the interaction between a policy period dummy measure and the dummy measure for one sibling and ## for 2 or more siblings. ^ The bold italic values are estimated through running the same regression but changing the reference of dummy variables of the policy-period variable.

264

Appendix 2 Logistic Models Predicting Probability of Student Transiting from RSHS to University with Missing Value Imputed Data M 1 M 2 M3 M 4 M 5 M 6 During vs. Before& 1.561(9.32) 0.739(1.52) 0.491(0.99) 0.908(0.90) 1.247+(1.77) 0.585(0.57) After vs. Before 1.203***(4.00) -0.237(-0.30) -0.512(-0.62) 1.287(0.95) -0.552(-0.66) -0.882(-0.68) Initial enroll. rate -0.013(-0.24) -0.045(-0.83) -0.046(-0.84) -0.059(-1.21) -0.044(-0.83) Enrollment rate 0.060+(1.92) 0.062+(1.87) 0.059+(1.82) 0.071*(2.27) 0.063+(1.89) Male -0.394(-1.39) -0.467(-1.57) -0.494+(-1.82) -0.379(-1.32) Father’s yrs. of edu. 0.048(1.25) 0.117*(2.04) 0.062(1.62) 0.050(1.27) Sibling No. 1 sibling vs. None -0.686*(-2.13) -0.749*(-2.51) -0.138(-0.31) -0.696*(-2.12) 2+ siblings vs. None -0.947***(-3.40) -1.010***(-4.13) -0.215(-0.44) -0.968***(-3.35) Father’s ISEI 0.017*(2.15) 0.018*(2.13) 0.016*(2.03) 0.016(0.98) Interaction Father’s edu. Sibling No. Father’s ISEI During vs. Before -0.041(-0.55) -1.491*(-2.38) # -0.003(-0.14) -1.721*(-2.28) ## After vs. Before -0.170*(-2.07) 0.088(0.15) # 0.008(0.40) -0.517(-0.65) ## After vs. During^ -0.129*(-2.03) 1.579**(2.87) # 0.011(0.68) 1.204(1.43) ## _cons -0.404(-2.25) -0.738**(-3.18) -1.075**(-2.62) -1.669*(-2.44) -1.522**(-2.72) -1.034(-1.35) N 531 531 531 531 531 531 Pseudo R2 0.062 0.081 0.131 0.138 0.150 0.134 Note: t statistics in parentheses. + p < 0.10, * p < 0.05, ** p < 0.01, *** p < 0.001. The values in the constant and pseudo R2 are from the regression when treating the first policy period as the reference. & In the table, the latter of each two-measure comparison is the reference. For example, “before” is the reference in” During vs. Before”. # The results indicate the interaction between a policy period dummy measure and the dummy measure for one sibling and ## for 2 or more siblings. ^ The bold italic values are estimated through running the same regression but changing the reference of dummy variables of the policy-period variable.

265

Appendix 3 Interview Topics

For parents  Compared with the examination-oriented education you received before, how is your child’s education different?  What are the advantages and disadvantages of the quality-oriented education?  What is your investment in your child?  How does your child’s education affect your family’s finance and your life as a parent?  Compared with other parents, do you invest the same amount, more or less in your child?  What is your opinion about the reform of university admission examination?  Which social class will benefit more from the quality-oriented education?  Do you think the quality-oriented education is just or not? Why?

For students  Could you please describe one of your ordinary school days?  What are your extracurricular activities?  Have you taken some extracurricular tutoring? If yes, what does it involve? Do you think the extracurricular tutoring is useful? If not, why?

For teachers  What is the difference between the quality-oriented education and the examination-oriented education? (Teaching method, curriculum, the amount of school time, the extracurricular activities, the evaluation method…)  Who benefit more from the quality-oriented education, rich or poor family?

266