BUYING FUTURE HAPPINESS Why do people invest not enough in education?

ERASMUS UNIVERSITY ROTTERDAM Erasmus School of Economics Marketing

Master Thesis Supervisor: Dhr Donkers

Name: Roxanne Kuipers Examnumber: 300970 Email: [email protected] Thesis: Master Date: 25-10-2009 Phone number: 06-23740848 Preface

I have written this paper to complete my Master in Marketing at the Erasmus University Rotterdam. Writing this thesis was very interesting and I have learnt a lot. I would like to thank the people who have supported me in writing this paper. First, I would like to thank my supervisor dhr. Donkers for his time and advice. I would also like to thank my partner and family for their inexhaustible support. Finally, I would like to thank SINL BV for their help, and off course thanks to all the respondents who have participated in the survey.

Roxanne Kuipers Rotterdam, November 2009

2 Executive Summary

3 Contents

Preface...... 2 Executive Summary...... 3 1. Introduction to the research...... 3 1.1 Motivation...... 3 1.2 The structure...... 3 1.3 Research questions...... 3 2. What is the difference in investments in education between foreigners and domestic people?...... 3 2.1 Introduction...... 3 2.2 Two general reason for not having a start qualification...... 3 2.3 Drop outs...... 3 2.4 Level of education...... 3 2.5 Conclusion...... 3 3. What are the causes of the fact that people invest not enough in education?...... 3 3.1 The Theory of Planned Behavior in general...... 3 3.2 The Theory of Planned Behavior in practice...... 3 3.3 The influence of the perception of the state of the labor market...... 3 3.4 The influence of the future time perspective...... 3 3.5 The influence of cultural and social background...... 3 3.6 The influence of motivation and earlier performance...... 3 3.7 Control for Ability...... 3 4. Method...... 3 4.1 Sample...... 3 4.2 Procedure...... 3 4.3 Measures and Coding...... 3 5. Results...... 3 5.1 The Data...... 3 5.2 Factor analysis...... 3 5.3 Regression analyse...... 3 5.4 Test of hypotheses Theory of Planned Behavior...... 3 6. Discussion...... 3 6.1 Hypotheses 2 and 3...... 3

4 6.2 Hypotheses 4 and 5...... 3 6.3 Hypothesis 6...... 3 6.4 Hypotheses 7 and 8...... 3 6.5 Hypotheses 9...... 3 6.6 Hypothesis 10...... 3 6.7 Hypothesis 11...... 3 6.8 General conclusion...... 3 6.9 Limitations and suggestions for further research...... 3 7. What can a company, like SINL BV, do to stimulate people to invest more and better in education?...... 3 Appendixes...... 3 Appendix 1: Temporary bibliography Thesis...... 3 Appendix 2: Framework Theory of Planned Behavior...... 3 Appendix 3: Survey...... 3 Appendix 5: Missing Values...... 3 Appendix 5: Demographic Characteristics...... 3 Appendix 6: KMO and Bartlett’s Test...... 3 Appendix 7: Rotated Component Matrix...... 3 Appendix 8: Reliability Analysis...... 3

5 1. Introduction to the research

1.1 Motivation

I am working for SINL BV, a school that gives shortened Intermediate Vocational Trainings, especially for adults who are in re-integration projects. SINL BV gives their schooling with the guarantee that the students will get a job after they finished their education. In my job at SINL BV, I work daily with people who have not invested enough in education in their younger years and I would like to study why. I am also interested in the question what companies that sell education, like SINL BV, can do about the phenomenon that people do not invest enough in education.

1.2 The structure

The research will consist of a theoretical and a practical part. In the first, theoretical, part of the thesis a closer look is taken at the difference in investments in education between foreigners and domestic people. Also the general causes are explored, which are described in earlier papers, of the fact that people invest not enough in education. In the theoretical part the framework of the Theory of Planned Behavior will be used (appendix 2).

In the second, practical, part of the thesis a survey is hold. Dutch people and foreigners were asked to fill in a questionnaire with questions about among others: their cultural and social background, their education, jobs of their families and their motivation to follow education in their younger years and now.

After collecting the data, a regression analyse and a factor analyses is applied to learn which variables have a significant influence on investing in education.

6 1.3 Research questions

There are a lot of papers written about investments from the government in education. But there is not so much written about the investments of individuals in education. There are some general articles who write about decision making with respect to the future time perspective, but specific literature about the question: ‘why do people invest not enough in education?’ is missing. Therefore, this thesis will be relevant for companies, like SINL BV, who sell education and who would like to know what they can do to stimulate the investments of individuals in education. To fill the gap in literature this thesis is focused on the following main question:

Why do people invest not enough in education?

The main question will be answered with the help of the following research questions:

1. What is the difference in investments in education between foreigners and domestic people? 2. What are the causes of the fact that people invest not enough in education? 3. What can a company, like SINL BV, do to stimulate people to invest more and better in education.

7 2. What is the difference in investments in education between foreigners and domestic people?

2.1 Introduction

Before answering the main question “Why do people invest not enough in education?” it is important to verify which people invest not enough in education. In this chapter the difference between the investments in education of foreigners and domestic people are explored. When it is known which kind of people invests (in general) not enough in education, it is possible to identify more easily the reasons why these people invest not enough in education.

Jaco Dagevos (2006) concludes in his research that, since a couple of years, the position of foreigners in the Dutch labor market is weak. Dagevos (2006) noticed a few causes of this phenomenon, two of the most important aspects are the level of training and the control of the Dutch language. Especially adolescents and foreigners face difficulties if they apply for a job. They have not enough experience or their level of education is too low (Dagevos, 2006). The low level of education could be an explanation for the high unemployment level.

Korvorst & van der Mooren (2007) agree in this with Dagevos (2006). They also found in their research that foreigners have many times more no job than domestic people. To test if foreigners invest not enough in education with regard to domestic people, the following hypothesis is formed:

H1: Foreigners invest more frequently not enough in education than domestic people

In this paper the term foreigners refers to people who are born in a foreign country or people who have one or two parents who are born in a foreign country. Foreigners are classified in western foreigners and non-western foreigners, with non-western foreigners coming from Turkey, Africa, Asia or Latin-America (Traag & van der Velden, 2007). In this thesis the term foreigners are used for both groups.

8 In the Netherlands the term start qualification is used to determine the minimal level of education that people should have. The minimal level of education is required to have a good chance at a sustainable job (Bierings & de Vries, 2007). In the Netherlands the term start qualification means that the minimum certificate of education people need is HAVO (school of higher general secondary education), VWO (pre-university education) or MBO at level two (intermediate vocational education).

Before starting to answer hypothesis 1, the first thing to clarify is what is meant by investing not enough in education. In this thesis we speak of not investing enough in education when someone is able (based on intelligence, qualification and experience) to reach a certain level of education but the person did not reach this level of education. The respondents within SINL BV are tested on intelligence, qualifications and experience and are, based on these tests, able to reach the MBO level of education.

2.2 Two general reason for not having a start qualification

Almost 66% of the people who do not have a start qualification have never followed education that gives them the opportunity to follow HAVO, VWO or MBO at level 2. This means that there are two reasons for people why they do not have a start qualification 1) they did not start an education that meets the requested level and 2) they did not successfully end an education at the requested level (Beckers & Traag, 2005). In 2006, more than nine million people between 15-64 year lived in the Netherlands without following education. Bierings & de Vries (2007) saw that 31% of these people had no start qualification. The people who are elder than 25 years had most of the times no start qualification because of reason one. And the adolescents had most of the time no start qualification because of reason two. In 2006 there were almost 600,000 adolescents who did not get to school. Almost 40% of them had no start qualification (Bierings & de Vries, 2007).

2.3 Drop outs

In the school year 2005/2006, 3.9% of the students left their education without a start qualification. It was also a fact that non-western foreigners more frequently left their education too early than domestic people. The percentage dropout by foreigners was 7% and by the domestic students it was 3% (Jaarboek onderwijs 2009). The students who are from Antillean and Aruban origin, showed the highest percentage dropouts. The Yearbook education 2009 also noticed that the percentage dropouts were higher at the first generation non-western foreigners than at the next generation.

9 Beckers & Traag (2005) also noticed that domestic people have several times more a start qualification than foreigners. In the year 2004, 62% of the domestic adolescents who did not study had a start qualification, as against 46% of the non-western foreigners that had a start qualification.

Domestic people and western-foreigners are in general more successful in their education than non-western foreigners. The level of education of people form Turkey and Morocco is behind the level of education of other people (Jaarboek onderwijs 2009)

2.4 Level of education

Most dropouts are students of the MBO. MBO education contains four level of education and it is proved that: the higher the level of education, the lower the percentage of dropouts (Jaarboek onderwijs 2009). Of the students who followed MBO level 1, 39% of the students left the education after one year without a start qualification. At MBO level 2, the percentage was 15%, and at level 3, 7%. The lowest percentage dropouts were seen at MBO level 4, namely 5% (Jaarboek onderwijs 2009).

2.5 Conclusion

As is clear from the evidence above, some people are more likely to invest not enough in education than other people. The following points are noticed:

 Non-western foreigners invest more frequently not enough in education than western foreigners and domestic people;  First generation foreigners invest more frequently not enough in education than foreigners from the next generation;  Students, who follow the lowest level of the MBO, are more likely to leave education without a start qualification and so invest more frequently not enough in education than students at the higher levels of MBO.

10 Based on the evidence in this chapter it can be concluded that hypothesis 1: Foreigners invest more frequently not enough in education than domestic people, is true. If we take a more detailed look it can be said that the group of people that invest the least in education is the following: Non-western first generation foreigners who study at the lowest level of the MBO. In the following chapter a close look is taken at the reasons why people invest not enough in education.

11 3. What are the causes of the fact that people invest not enough in education?

Chapter two put forward that some people invest not enough in education. In chapter three the general causes of this phenomenon will be explored. In this exploration the Theory of Planned Behavior will be used

3.1 The Theory of Planned Behavior in general

Icek Ajzen is the founder of the Theory of Planned Bevavior, this theory is an extension of the theory of reasoned action and is created to predict and explain human behavior (Ajzen, 1991). The research question “What are the general causes of the fact that people invest not enough in education?” will be built on the Theory of Planned Behavior.

In the Theory of Planned Behavior, human behavior is guided by three kinds of beliefs: Behavioral Beliefs, Normative Beliefs and Control Beliefs (Appendix 2). Behavioral Beliefs are beliefs about the likely outcomes of human behavior, which also include the evaluations of these outcomes. Normative Beliefs are beliefs about the normative expectations of others. Normative Beliefs also include the motivation of human to react at these expectations of others. Finally, Control Beliefs are beliefs about the presence of factors that influence human behavior, which also include the power of these factors (Ajzen 2002).

Behavioral Beliefs, Normative Beliefs and Control Beliefs lead respectively to, a favorable/unfavorable attitude toward a certain behavior, social pressure or subjective norms and perceived behavior control (Ajzen, 2002). These three outcomes of the beliefs lead to the intention to perform a given behavior. Finally, Actual Behavior Control (non-motivational factors like the availability of opportunities and resources) influences the perceived behavioral control and the intention to perform a given behavior. Actual Behavior Control, like opportunities, also influences the decision of people to carry out their intentions (they will do so if the opportunity arises) (Ajzen, 2002).

Appendix 2 shows that the three kinds of beliefs are also influencing each other. Behavior Beliefs influences Control Beliefs and vice versa. Normative Beliefs influences Control Beliefs and vice versa. And finally, Behavior Beliefs influences Normative Beliefs and vice versa.

12 3.2 The Theory of Planned Behavior in practice

In this thesis the Theory of Planned Behavior will be the foundation for answering research question two: What are the general causes of the fact that people invest not enough in education? In appendix 2 a schematic overview of the Theory of Planned Behavior is given, which will be explained below.

Behavior The research will concentrate on the following human behavior: investing not enough in education.

Control Beliefs Chapter 3.6 looks at the influence of motivation and earlier performance on the investments in education. We will see that the Behavioral Beliefs, the Normative Beliefs and the Actual Behavior Control are influencing the Control Beliefs. And the Control Beliefs are influencing the Behavioral Beliefs and the Normative Beliefs. So the Control Beliefs directly and indirectly influences the intention and the behavior.

Behavioral Beliefs In chapter 3.4 the influence of the future time perspective on investing in education will be discussed. The perception of the state of the labor market will be discussed in chapter 3.3. We will see that the Control Beliefs and the Normative Beliefs are influencing the Behavioral Beliefs. And the Behavioral Beliefs are influencing the Control Beliefs and the Normative Beliefs. So the Behavior Beliefs directly and indirectly influences the intention and the behavior.

Normative Beliefs Chapter 3.5 looks at the influence of the cultural and social background of people on the investment in education. We will see that the Control Beliefs and the Behavioral Beliefs are influencing the Normative Beliefs. And the Normative beliefs are influencing the Control Beliefs and the Behavioral Beliefs. So the Normative Beliefs directly and indirectly influences the intention and the behavior.

13 Actual Behavioral Control Chapter 3.7 controls for ability. A closer look is taken at the skills and capabilities people have. Further it has to be verified whether or not people have had the money to invest in education. These elements of Actual Behavioral Control will directly influence the Control Beliefs and the Behavior. Indirectly will it also influence the Behavioral en Normative Beliefs.

14 3.3 The influence of the perception of the state of the labor market

Since many years, the position of foreigners on the labor market is inferior, if compared to the position of the domestic people (Dagevos, 2006). The foreign adolescents have the worst positions; foreign adolescents are twice as much unemployed as domestic adolescents. Especially in a broad labor market, the foreign people have a difficult position on the labor market (Dagevos, 2006). This is among others due to their education level, their lack of control of the Dutch language and the other factors that will be explained in this thesis.

The Human capital theory The human capital theory states that there is a positive relationship between the amount of education people have followed and their production/salary (Becker 1975). The decision to follow (further) education or not is a cost/benefits trade off. The investments in human capital involve an initial cost and the individual hopes to gain a return on this investment in the future (Blundell et all. 1999). The initial costs consist of among others: tuition course fees and the forgone wages while the individual was studying instead of working. Usually more education means a higher productivity and therefore a higher wage. In summary: the individual gives up some income during the period and he or she will follow education in return for increased income in the future (Blundell et all. 1999).

This outcome suggests that if the labor market is perceived as broad (i.e., less work and lower wages) people are more likely to invest in education. In this case people give up some proportion of their income now because their chances on the labor market are not so good at the moment. Therefore it will be more profitable to study longer. In this thesis the definition of the CWI will be used to specify whether the labor market is broad or narrow. If there are on every ten open vacancies less than 14 people without a job, the labor market is narrow. When there are more than 25 people without a job, the labor market is broad.

The influence of the government on the (perception of the) state of the labor market The government is also able to influence the state of the labor market. The government is among others responsible for the decisions around social security and minimum wages. Because of the existence of social security, people can become less motivated to get a job, especially individuals with a low wage (near by the minimum wage). The existence of the minimum wage can also have a negative impact on the unemployment rate. This is especially the case if individuals have a low productivity; employers have to pay them more

15 wage than the employee delivers to them, so the employer will decide to hire a more productive individual (dissertatie RUG). The influence of investing enough in education on the perception of the state of the labor market Research results suggest that adolescents who do not go to school are more successful on the labor market if they have at minimum a start qualification. For example in 2004: 84 % of the adolescents with a start qualification who did not get to school had a paid job for more than 12 hours a week, as against to 67% of the adolescents without a start qualification (Beckers & Traag, 2005). Beckers & Traag (2005) also found that the unemployment rate among adolescents without a start qualification was higher (17%) than the unemployment rate among adolescents with a start qualification (10%). Bierings & de Vries (2007) agree with this, they found that in 2006 46% of the people without a start qualification had no paid job for more than 12 hours a week as against to 24% of the people that had a start qualification. In the research of Beckers & Traag also comes true that people with a start qualification have more chance on a fixed treaty (82%) than people without a start qualification (75%).

The importance of the binding with the labor market The intensity in which people feel bounded to the labor market is influencing the perception they have of the labor market. This is called the distance to the labor market. Bierings & de Vries noticed three groups of people with a distance to the labor market: 1. People who are immediately available for a job and they are actively looking for it (they have a small distance to the labor market). 2. People that would like to work but they are not immediately available and/or they are not actively looking for it (this people have a larger distance to the labor market than group 1). 3. People that do not want to work or can not work, they are inactive (this group has the largest distance to the labor market).

People without a start qualification and elderly people have a relatively large distance to the labor market (Bierings & de Vries, 2005). But as we see above, people that can not work or do not want to work have the largest distance.

Based on the evidence above it can be expected that a person’s perception of the state of the labor market is influencing their investments in education. This will be tested with the following hypothesis:

16 H2: The perception of the state of the labor market influences the investments of people in education

The perception of the state of the labor market and the theory of planned behavior In chapter 3.2 is noticed that the perception of the state of the labor market is a Behavioral Belief (a belief about the likely outcome of human behavior). This Behavioral Belief influences the Control Beliefs and Normative Beliefs. However, the Control Beliefs and the Normative Beliefs are also influencing the Behavioral Beliefs. Below will explained shortly how some of these beliefs can influence each other and this will be controlled in the research.  The Behavioral Belief Perception of the state of the labor market has a direct influence on the Control Belief Motivation & Earlier Performance. As noticed before, if the labor market is perceived as broad (less work and lower wages) people are more likely to invest in education. In this thesis will be explored, with hypothesis 3, if the perception of the state of the labor market has a direct influence on motivation and earlier performance.

H3: The Behavioral Belief Perception of the state of the labor market has a direct influence on the Control Belief Motivation & Earlier Performance

 The Behavioral Belief Perception of the state of the labor market has a direct influence on the Normative Belief Cultural & Social Background. As will be explained in chapter 3.5 a person who has to make the decision to invest in education can be stimulated by family, friends and the environment. Hypothesis 2 test if a person’s perception of the state of the labor market is influencing their investments in education. If hypothesis 2 is true, a person will be more stimulated by family, friends and the environment if they have a positive perception of the labor market.

In chapter 3.7 I will control for ability. I will take a closer look at the Resources, Skills and Capabilities people have. These elements of Actual Behavioral Control will indirectly influence the Behavioral Beliefs via the Control Beliefs and the Normative Beliefs, and directly influence the Behavior.

Conclusion For answering hypothesis 2: The perception of the state of the labor market influences the investments of people in education, in the survey will be included:  Do people perceive the labor market as broad?

17  Do people perceive the social security system as less motivating to get a job?  Do people feel a relatively small or large distance to the labor market?

18 3.4 The influence of the future time perspective

As already mentioned, the human capital theory says that the decision to follow (further) education or not is a cost/benefits trade off. The problem with education is that people have to invest in the near-future but the benefits will result in the distant-future. Peetsma (2000) found in her research that there is a positive relationship between school investments and future time perspective. This indicates that the more people invest in education, the better their future time perspective. Soman (1998) concludes in his research that temporal delay of the incentives of an event has a negative influence at the efforts/benefits of the event. Temporal delay between the choice about an event and the redemption causes a systematic underweighting of future efforts. This result suggests that people may find it difficult to evaluate the cost/benefits trade off of education.

The concept of future time perspective In general, future time perspective can be described as a representation or conceptualisation of a particular life domain in terms of time (Peetsma 2000). With a particular life domain you can think of such as professional career or social relations. The future time perspective is characterized by extension and valence (Lens, 1986 in Peetsma, 2000). Below the definitions for these terms are given (Peetsma 2000):  Perspective: the representation of events or objects in the near or distant future  Extension: the degree of remoteness of the representation in time  Valence (or Relvance): the value of a certain life domain in the future The concept of future time perspective will be interpreted as a variant of the concept of “Attitude” so to define the concept of future time perspective the definition of Peetsma (2000) will be used: “an attitude towards an object (life domain) over a certain period of time in the future”.

Desirable versus Feasible options Trope & Liberman (2003) saw in their research that temporal distance changes people’s responses to events in the future. This is because of the fact that the temporal distance changes the way people mentally represent those future events. The greater the temporal distance, the more future events are represented in terms of abstract features. Therefore it can be more difficult to decide about an event in the future than about a near-future event. If people have to make a decision about something that is going to happen in the more distant future, the person will think in more abstract terms and the accent is on the questions how desirable the option is.

19 If people have to make a decision about something that is going to happen in the near future, the person will think in more concrete terms and the accent is on the question how feasible the option is (Zhao et all, 2007). In short, people will prefer more desirable options for the distant future and more feasible options for the near future. This is also the case when people have to choose between investing more in education or not. The problem with this decision is that people have to invest in the near future if they want to see an outcome in the more distant future (most of the time the result will come after four years of studying). People are more likely to invest in education if they think that the outcome of this investment is desirable in the distant future.

The length of the future time perspective If people think about the future they are probably thinking about goals, plans and projects and how they are going to achieve them. They are directing present actions to achieve goals in the future. Simons et all, (in McInerney, 2004) refer to this future thinking as an important cognitive element. But the future is not for everyone of the most importance. So it is also important to look of what people are doing to achieve the desired future goals and how important it is for people to achieve these future goals. Simons et all, (in McInerney, 2004) found a relationship between the length of the future time perspective and the perceived utility of what people are doing to achieve future goals. People with a longer future time perspective see their present behavior as more important to achieve future goals than people with a shorter future time perspective. Hypothesis 4 is formed to test if the future time perspective is influencing the investments people make in education:

H4: The future time perspective people have is influencing the investments of those people in education

Awareness As we have seen in the paragraph above, to make a good decision about the future, it is necessary to think about the future and to be aware of the importance of it. McInerney et all (1998) examined in their research why some students are more or less motivated to study than other students. They found that students that were less motivated found schooling irrelevant and they did not see a clear link between the value of education and achieving future goals (McInerney, 2004).

20 Solutions Zhao et all (2007) demonstrate in their research that outcome simulation (encouraging people to think about the desirable outcome of fulfilling the goal) helps people to change their preferences in the near future so that it is consistent with their natural distant-future preferences. As noticed in the paragraph “Awareness” it is important that students make plans and set future goals. Students also need to know what is needed to achieve these future goals. (McInerney, 2004). Also Miller and Brickman (in McInerney, 2004) conclude in their research that is it important to hold valued future goals because this will motivate student to study and invest enough in education. Process simulation (encouraging people to imagine the step-by-step process of reaching a certain goal) helps people to change their preference in the distant future so that is consistent with their natural near-future perspective.

The influence of the future time perspective and the theory of planned behavior As noticed in chapter 3.2 the influence of the future time perspective is a Behavioral Belief (a belief about the likely outcome of human behavior). This Behavioral Belief also influences the Control Beliefs and Normative Beliefs. The Control Beliefs and the Normative Beliefs are also influencing the Behavioral Beliefs. Below will explained shortly how some of these beliefs can influence each other and this will be controlled in the research.  The Behavioral Belief future time perspective of a person has a direct influence on the Control Belief Motivation & Earlier Performance. As noticed above, people will choose more likely to invest in education if they think that the outcome of this investment is desirable in the distant future. So if people have a positive time and future perspective they will be more motivated to invest in education. This will be tested with hypothesis 5

H5: The Behavioral Belief future time perspective of a person has a direct influence on the Control Belief Motivation & Earlier Performance

 The Behavioral Belief future time perspective of a person has a direct influence on the Normative Belief Cultural & Social Background. Hypothesis 4 test if the future time perspective is influencing the investments people make in education. If hypothesis 4 is true and people think that investing enough in education is desirable in the distant future they will be more likely stimulated by friends, family and the environment to invest enough in education.

21  The Normative Belief Cultural & Social Background directly influences the Behavioral Belief future time perspective. In chapter 3.5, hypothesis7 test if a person’s cultural and social background is influencing their investments in education. If hypothesis7 is true we will test wit hypothesis 6 if the Normative Belief Cultural & Social Background directly influences the Behavioral Belief future time perspective

H6: The Normative Belief Cultural & Social Background directly influences the Behavioral Belief future time perspective

In chapter 3.7 there will be controlled for ability. A closer look is taken at the Resources, Skills and Capabilities people have. These elements of Actual Behavioral Control will indirectly influence the Behavioral Beliefs via the Control Beliefs and the Normative Beliefs and directly influence the Behavior.

Conclusion For answering hypothesis 4: The future time perspective students have is influencing the investments of those people in education, in the survey will be included:  Do people see investing in education as a near or more distant future event?  Do people find investing in education desirable?  Do people have a short or long future time perspective with respect to investing in education? And are they aware of the importance of investing enough in education.

22 3.5 The influence of cultural and social background

The theory of reasoned action assumes that decision makers can make a free choice. But Dutch research show that this is, as we look at the education decision, not always true. In their decision making process where people have to decide if they want to invest further in education, people are probably influenced by certain factors. Below the influence of the social and cultural background will be explored.

The influence of social capital Before looking at the social factors that are influencing the investing decision of people to invest enough in education or not, first the term social capital has to be defined. To define social capital the definition of Putnam (2000 in: de Koning et all, 2008) is used. He says: “social capital is the surplus value of social networks, which exist because from these networks comes the tendency to do things for each other without getting a direct compensation”. An example of social capital is parents who are helping their children with their homework. But social contacts also exist when people are looking for a job or when people have a membership of an association (de Koning et all, 2008). There are several studies that found a significant positive effect between performance and social capital (La Porta e.a, 1997, Knack & Keefer, 1997, in: de Koning et all, 2008). A person’s social skills can be of influence at his or her social capital. These social skills can also be important when looking at the performances in education.

The social factors can be divided in two groups. The first group is about the family itself: the stability, the combination etcetera. The second group is about the environment and the social network of the family.

First group of social factors As noticed before, the first group of social factors is about the family itself. Several studies show that the expectations of the parents have an influence on the decisions someone is taking about his education. Kortlik & Harrisson (1989, in Disertatie RUG) found in their research that more than 50% of the students find their parent’s opinion important while making decisions about their education. Parents also have influence on the performance of their child at school, but they are especially responsible for motivating their child in the decision process to invest in education (Need, ..). The social network of the parents is important with respect to motivating children to follow education. Therefore looking at the second group of social factors is necessary.

23 Second group of social factors The second group is about the environment and the social network of the family. The influence of friends and teachers is of less importance than the influence of the parents on decisions about education. Weerdenburg (1987, in: Disertatie RUG) shows in his research that 54% of the students found the influence of their parents of most importance, and 13% of the students found the influence of teachers of most importance.

Several studies provide evidence of a relationship between the social environment and decisions about education. For example the research of Tesser (1986, in disertatie RUG) showed that students from a higher social environment are following longer education than students from a lower social environment. And Reender & van der Velden (184, in disertatie RUG) found that if the parents of a student have no work, these students are leaving school more frequently without a start qualification.

Some explanations about the influence of the social environment is found in the work of Boudon (1974). Boudon says that there are primary and secondary effects. The primary effect is the influence of the social environment on the performances in education. The secondary effect is about the influence of the social environment on decisions about education. When looking at this secondary effect, Boudon states that when people have to make a crucial decision about something they choose the alternative that has the least costs. The costs of investing enough in education are higher for people from lower social environments than for people from higher social environments. Boudon gives a definition of these costs: the social distance between the students on the one side and the parents and environment on the other side. For example, if the parents of a student have invested not enough in education, the student will choose more frequently to invest also not enough in education because when he does, the social distance between him en his parents will become larger.

Thomas and Wetherell (1974, in: disertatie RUG) also found a negative relationship between the social environment and the degree in which the student is interested in earning money. They saw that the students from a lower social environment where more interested in a short future perspective, so they choose more frequently not to invest more in education but they want to earn money now. And the student from higher social environment where more interested in a longer future perspective so they decided more frequently to invest more in education so that they will probably earn more money in the future.

24 The influence of the cultural/ethnical background Another difference between the lower and the higher social classes are the cultural resources. Under these cultural resources we understand the values, the norms and the lifestyle of people (Van de Velde et all, 1996 in: van de Wiele 2002). As noticed before, the lower social classes may find investing in education less important because of their financial situation. Also in the values and norms of the lower social classes is noticed they find investing in education relatively less important that the higher social classes (Kochyut, 1993).

The ethnical origin from people can be also be an explanation for the question why foreigners leave education to early. Foreigners face all kinds of problem with respect to integration (de Vries & Wolbers, …). These foreigners have to deal with cultural differences, less information to the labor market and most of the time they speak the Dutch language unsatisfactory (de Vries & Wolbers,…). Because of the fact that foreigners have more frequently insufficient knowledge about the labor market and the system of education, foreigners have unrealistic high expectations about their child’s education. Especially Turkish and Moroccans parents want their children to achieve the highest possible education (Ledoux, 1993; in: Disertatie RUG). This is an explanation why there are so many dropouts by these students. The expectation is that these problems will become less important if foreigners are longer in the country.

Based on the evidence above it can be expected that a person’s cultural and social background is influencing their investments in education. This will be tested with the following hypothesis:

H7: The social and cultural background people have is influencing the investments of those people in education

The influence of the culture and social background and the theory of planned behavior As noticed in chapter 3.2 the influence of the cultural and social background is a Normative Belief (a belief about the normative expectations of others). This Normative Belief also influences the Control Beliefs and Normative Beliefs. The Control Beliefs and the Behavioral Beliefs are also influencing the Normative Beliefs.

25 Below will explained shortly how some of these beliefs can influence each other and this will be controlled in the research.  The Normative Belief Cultural and Social background has a direct influence on the Control Belief Motivation & Earlier Performance. Hypothesis 7 tests if a person’s cultural and social background is influencing their investments in education. If hypothesis 7 is true we will test with hypothesis 8 if a person who is stimulated by friends, family and the environment to invest in education, will be more motivated to invest enough in education.

H8: The Normative Belief Cultural and Social background has a direct influence on the Control Belief Motivation & Earlier Performance

 The Behavioral Belief future time perspective of a person has a direct influence on the Normative Belief Cultural & Social Background. Hypothesis 4 test if the future time perspective is influencing the investments people make in education. If hypothesis 4 is true and people think that investing enough in education is desirable in the distant future they will be more likely stimulated by friends, family and the environment to invest enough in education.  The Behavioral Belief Perception of the state of the labor market has a direct influence on the Normative Belief Cultural & Social Background. As will be explained chapter 3.5 a person who has to make the decision to invest in education can be stimulated by family, friends and the environment. Hypothesis 2 test if a person’s perception of the state of the labor market is influencing their investments in education. If hypothesis 2 is true, a person will be more stimulated by family, friends and the environment if they have a positive perception of the labor market.

In chapter 3.7 there will be controlled for ability. A closer look is taken at the Resources, Skills and Capabilities people have. These elements of Actual Behavioral Control will indirectly influence the Normative Beliefs via Control Beliefs and the Behavioral Beliefs and directly influence the Behavior.

26 Conclusion For answering hypothesis 7: The cultural and social background of people is influencing the investments of those people in education, in the survey will be included:  Do people have enough knowledge about the labor market to make a good decision about investing in education or not?  Do people have enough knowledge about the education system to make a good decision about investing in education or not?  Do people feel that there parents are expecting to much from them?  Do people feel supported by their family to follow enough education?  Do people feel supported by their social network to follow enough education?  Are the parents of the student both working?  Do the student lives in a lower or higher social environment?

27 3.6 The influence of motivation and earlier performance

In several researches a direct relationship is found between performance of students and their participation in education (Webbink et all, 1993 in: Disertatie RUG). Sometimes this seems logical, because every type of education has its own requirements and if students do not meet these requirements they have to leave the education. But at other times the participation in education is not influenced by the real performance of the student but by the perceptions the students has about his or her own performance.

The action-effect-affect chain As noticed above, the visions students have about themselves, their performance and their education are influencing their own perceptions about their education (Boekaerts & Simons, 1993). For example: someone is doing a test to get permission to follow a type of schooling, unfortunately he failed to pass the test. If this person fails this test several times, the person will think that he is not able to do the test (action), and therefore he will not be allowed to follow that type of education (effect). And if he knows that his parents will be disappointed the person will feel uncomfortable (affect). Boekaerts & Simons (1993) are calling this the negative action-effect-affect chain and if this chain is going to continue for this person (he fails all tests), the person will go to anticipate. So before he will start the test, he will think: I will not be able to do this test. Because of this type of anticipating, the person is more likely to fail the test (Boekearts & Simon, 1993).

The attribution theory Also Wiener (1986) has been doing research about perceptions. Wiener named these perceptions of success: prospective attributes. The attribution theory of Wiener shows that the effect of a persons behavior, will be ascribed by this person and his environment to certain causes (Boekearts & Simon, 1993).

The attribution theory of Wiener consists of three dimensions:

1. Intern / Extern This dimension refers to the place of control, does the student ascribe the cause by him or herself (intern) or does the student ascribe the cause outside of him- or herself (extern)? 2. Stabile / Non-Stabile This dimension refers to the stability of the factors, does the student ascribe the cause variable (Non-Stabile) or permanent (Stabile)?

28 3. Drivable / Non-Drivable This dimension refers to the drivability of the factor, could the student have done better (Drivable) or could he not (Non-Drivable)?

Students who ascribe the good education performances to internal factors and bad education performances to external factors, will feel more positivity if they do something with success. If they fail they will minimize their negative feelings and blame others for their bad performance (Boekearts & Simons, 1993). If a student ascribes education performances in this way, they have a more positive self image and feel more positive against their education. Students who have more positive feelings against their education will be more motivated to invest enough in education. If students ascribe their bad education performance consequently to external factors, they will probably get the feeling that they are out of control. This may lead to passivity and helplessness. Students who have these emotions will feel more legitimate to invest not enough in education because they do not blame there self for performing bad (Boekearts & Simons, 1993). Also Peetsma (1990, in Disertatie RUG) noticed that students who ascribe their performance to internal/drivable factors will have a more positive education perspective than students who ascribe there performance to external/non-drivable factors.

The Big Five As noticed before, the personal characteristics of a person can influence his performance in education and therefore his decision to invest enough in education or not. To define the personal characteristics we will use the Big Five personality traits (Eijck & de Graaf, 2001):

1. Extraversion: this trait refers to the degree in which someone is directed to the outside world. 2. Agreeableness: this trait is associated with the skills someone need to create interpersonal relationships. 3. Conscientiousness: this trait is associated with persistence and the typical characteristics by this trait are carefulness and systemic working. 4. Emotional Stability: this trait is associated with have the space to deal with challenges without panicking and the typical characteristic by this trait is self regulation. 5. Openness: this trait is associated with the intellectual curiosity and the typical characteristics by this trait are creativity and inventiveness.

Eijck & de Graaf (2001) showed in their research that there is no influence of the social environment on the personal characteristics of people. There showed by men, the five personal characteristics are having a significant influence on the performance in education.

29 Extraversion and Agreeableness are having a negative effect and Conscientiousness, Emotional Stability and Openness are having a positive effect. For women, only the characteristic Extraversion has a negative effect and Emotional stability a positive effect. Eijck & de Graaf (2001) ascribe the difference between men and women to the fact that women participate relatively short in education and had their personal characteristics not the change to become optimal.

In this thesis will be explored if motivation and earlier performance is influencing the decision to invest enough in education or not. Therefore hypothesis 9 is formed

H9: The motivation and earlier performance of people with respect to their study, are influencing the investments of those people in education

The influence of motivation and earlier performance the theory of planned behavior As we have seen in chapter 3.2 the influence of motivation and earlier performance is a Control belief (a belief about the presence of factors that influences human behavior). This Control Belief also influences the Behavioral Beliefs and Normative Beliefs. We also see that the Behavioral Beliefs and the Normative Beliefs are influencing the Control Beliefs. I will explain shortly how some of the beliefs can influence each other and I will control for this in my research.  As noticed in chapter 3.5 the Normative Belief Cultural and Social background has a direct influence on the Control Belief Motivation & Earlier Performance. Hypothesis 7 tests if a person’s cultural and social background is influencing their investments in education. If hypothesis 7 is true we will test with hypothesis 8 if a person who is stimulated by friends, family and the environment to invest in education, will be more motivated to invest enough in education.

H8: The Normative Belief Cultural and Social background has a direct influence on the Control Belief Motivation & Earlier Performance

 As noticed in chapter 3.3 the Behavioral Belief Perception of the state of the labor market has a direct influence on the Control Belief Motivation & Earlier Performance. As noticed before, if the labor market is perceived as broad (less work and lower wages) people are more likely to invest in education. In this thesis will be explored, with hypothesis 3, if the perception of the state of the labor market has a direct influence on motivation and earlier performance.

30 H3: The Behavioral Belief Perception of the state of the labor market has a direct influence on the Control Belief Motivation & Earlier Performance

 As noticed in chapter 3.4 The Behavioral Belief future time perspective of a person has a direct influence on the Control Belief Motivation & Earlier Performance. As noticed above, people will choose more likely to invest in education if they think that the outcome of this investment is desirable in the distant future. So if people have a positive time and future perspective they will be more motivated to invest in education. This will be tested with hypothesis 5

H5: The Behavioral Belief future time perspective of a person has a direct influence on the Control Belief Motivation & Earlier Performance

In chapter 3.7 there will be controlled for ability. A closer look is taken at the Resources, Skills and Capabilities people have. These elements of Actual Behavioral Control will directly influence the Control Beliefs and the Behavior.

Conclusion For answering hypothesis 9: The motivation and earlier performance of people with respect to their study are influencing the investments of those people in education, in the survey will be included:  To which factors (as we look to the three dimension of the Attribution theory) ascribe people their good or bad performance in education?  In which degree have people the personal characteristics as describe in the Big Five personality traits?

31 3.7 Control for Ability

The previous chapters showed that there are several internal and external factors that are influencing the decision to invest enough in education. Before all these factors plays a part it is important to determine if someone is able and capable to invest in enough in education.

As noticed in chapter 2.2 it is important to realize that the term, investing not enough in education, is different for everyone. People who are able to reach the University but did only reach the MBO level of education did not invest enough in education. But people who are not able to reach MBO level two but did reach MBO level one, has invested enough in education. So before starting to conclude if people have invested enough in education, the level of education that people are able to reach has to be known. In this thesis not investing enough in education means that someone is able (based on intelligence, qualification and experience) to reach a certain level of education but the person did not reach this level of education.

The influence of the degree of maturity It is possible that someone is not able to invest enough in education because this person is not mature enough. Because of relatively immaturity, people will face a discrepancy between what they can en what is expected from them (Westerberg et all, 2009). Students in our education system have to be able to make choices on their own and they need the discipline to work independent. But not every student has the skills to be able to do this. Westerberg et all (2009) noticed a relationship between the development of the brains and the development of these skills. In the adolescence the brain is developing and during this adolescence the student will be more capable to steer, plan, monitor, control and evaluate their own learning behavior. From the age of 18 or 19 years, students have more skills that are necessary to be successfully at school.

Blair & Diamond (2008) also showed in their article how immaturity influences the decision to invest not enough in education. For example: if a person does not have the skills to listen carefully to the teacher and is not able to not finish its assignments, this person will have less pleasure in going to school. The person will probably have to deal with a low self esteem and the teachers will expect less of this person. All together this will probably lead to a higher chance that this person will not invest enough in education.

32 In their research Westerberg et all (2009) noticed that persons who finished school to early and did not invest in enough in education gave the following internal reasons for this:  Low motivation  Not doing homework frequently  Wrong choices They also gave the following external reasons:  Education was to difficult  Chaos in their education Westerberg et all (2009) showed in their research that these persons who did not invest enough in education had a relatively low psychosocial and cognitive development. This means that they have among others: less empathy, less self regulation and a lower tolerance of frustration. They are also less able to carefully plan, organize and accomplish a job.

With hypothesis 10 will be tested if the degree of maturity is influencing the decision to invest not enough in education:

H10: The degree of maturity of someone is influencing the decision to invest enough in education.

The influence of the financial situation As explored before, it is necessary to invest in education to have a desirable result of the education in the future. The consequence of this result is that people need the financial inputs to invest in education.

Students that are coming from a family who do not have these financial inputs are less able to invest enough in education than students who have enough financial inputs (van de Wiele, 2002). It is a fact that these financial inputs are becoming less important for investing in education because education is cheaper through the support of the government. But this does not mean that there are not any differences between the higher and lower social classes. Nicaise (1996) argued therefore that the fact that education is becoming cheaper had more advantages for the higher social classes than for the lower social classes. Also Kochuyt (1993 in; van de Wiele 2002) argued that the people from lower social classes have more difficulty with the costs of education than the people from higher social classes. This is due to the fact that there are several costs next to the education cost that have to be payed to the school. These several costs include for example transport costs and the cost for excursions.

33 As seen in the Human Capital Theory, investing in education is a cost/benefit trade-off. If people choose to invest in education for several years, they have no income in these years. Especially in the lower social classes this can be a problem because these people need to care for there family. Another reason why people from a lower class more frequently decide to earn money now instead of investing in education and probably earn more money in the future, is because of the fact that people from lower classes have relatively less information about the labor market and less insurance that they will get enough gain from their investment (van de Wiele, 2002). Also students that have negative intentions toward investing in education (for example if they find it boring) will probably choose to invest not enough in education because of the loss of income (Disertatie RUG).

Hypothesis 11 is formed to test if the financial situation of people is influencing the decision to invest in education.

H11: The financial situation is influencing the decision to invest enough in education.

The influence of the degree of maturity/the financial situation and the theory of planned behavior As we have seen in chapter 3.2 the influence of the degree of maturity and the influence of the financial situation are Actual Behavioral Controls (non-motivational factors like availability of opportunities and resources). As noticed in the previous chapters this Actual Behavior Control directly influences the Control Beliefs and the Behavior. Indirectly, Actual Behavior Control influences the Behavioral Beliefs and the Normative Beliefs.

Conclusion For answering hypothesis 10: The degree of maturity of someone is influencing the decision to invest enough in education, in the survey will be included:  Do people show behavior (for example less empathy, less self regulation, a lower tolerance of frustration, and less able to carefully plan, organize and accomplish a job) that can be ascribed to a degree of immaturity.

For answering hypothesis 11: The financial situation is influencing the decision to invest enough in education, in the survey will be included:  Do people have the financial inputs to invest enough in education?

34 4. Method

4.1 Sample In the survey, 200 respondents were asked to participate. Two different groups were targeted. The first group consist of 100 respondents who are following there education at SINL, most of these respondents are foreigners. This group has invested not enough in education in their younger years. This is known because they were tested before they started with their education at SINL. From the results of these test we know that these respondents were able (based on intelligence, qualification and experience) to reach a higher level of education but the person did not reach this level of education yet. Within this first group, respondents with different demographic characteristics participated. The second group consist of 100 respondents who do not follow their education at SINL. These respondents were travelling by train in October 2009 and they have different demographic characteristics. These two different groups were used to explore the differences between the two groups and to control the hypotheses. Two different groups were also needed to get a general impression of the causes why people invest not enough in education.

4.2 Procedure The respondents were asked to participate in a survey about their behavior with respect to investing in education (Appendix 3). The respondents were not told that the intention of the survey was to get an answer on the question why people invest not enough in education. This is done to prevent that respondents get a negative attitude against the survey; they could have thought that they were asked to participate because they invested probably not enough in education. In the introduction of the survey is told that the questionnaire is anonymous, to take care that the respondents feel free to answer the truth. In the introduction of the survey an example question is used to explain the questions used in the survey.

The questionnaire consists of 53 questions. These questions were constructed based on the theoretical research. Question one till five are used to evaluate the outcomes. Question one and two contains two of the dependent variables that have to be explored. Question three till five are used to know the highest level of education the respondent had reached.

35 Question six till 13 were used to measure perceived behavioral control. Questions about perceived behavior control are used to capture respondent’s confidence that they are capable of performing the behavior (Ajzen, 2006). In this thesis the behavior is: investing in education. In this section there are for example question about the difficulty and possibility to perform the behavior.

For the second hypothesis, question 14 till 20 are constructed. These questions relate to the respondent’s perception of the state of the labor market. Questions 21 till 25 were constructed for the third hypothesis, these questions relate to the future time perspective of the respondents. To explore the fourth hypothesis, question 26 till 39 were asked. These questions are going about the cultural and social background of the respondents. Question 40 till 42 are about the respondent’s motivation and earlier performance. Question 40 and 41 contains two of the dependent variables that have to be explored. For the fifth hypothesis, question 43 till 47 are used. These questions relate to a respondent’s degree of maturity. Finally, question 48 till 53 are asked to get demographic information about the respondents. This information is asked at the end of the survey to be sure that the respondents will not be influenced by these questions before they answered the survey.

4.3 Measures and Coding To measure the outcome of most questions, a seven-point likert scale was used. Respondents were asked to encircle the number that correspond whit their opinion. The scale is going from 1 (very bad/few etc) till 7 (very good, much etc) and so it is insert in SPSS. There are a few questions that are coded differently; these are explained in Appendix 3A.

36 5. Results

5.1 The Data

Missing values Before the analysis of the data could be done, the data has to be explored. There were four respondents who forget to fill in a whole page of the questionnaire. Because of this fact, these respondents were left out of the analysis. After the elimination of these four respondents, the data contains 39 missing value, these missing value have been replaced with the mean score of the variable that correspond with each missing value. The number of missing values (39) is relatively small if it is compared to the sample (200). The 39 missing values are also scattered across the questions. Therefore each question is answered by minimal 98.5% percent of the respondents (appendix 5) Because of these facts, replacing the missing values with the mean will not influence the results that much (Field, 2005).

Demographic characteristics After exploring the missing values, the demographics of the respondents are explored. The group of 200 respondents contains of respondents with different levels of education (Appendix 5.1). But between the two groups there are differences. In the first group, respondents from SINL, 21.6% from the respondents reached HAVO and 38.1% reached the MBO 1 or 2 level. However in the second group, respondents not from SINL, 7.4% reached HBO. This means that in the first group the highest level of education is relatively lower than in the second group.

Most of the respondents, 77%, lived their entire live in the Netherland (Appendix 5.2). But again there is a difference between the two groups. In the first group, 53.6% of the respondents lived their entire live in the Netherlands and 46.4% did not. In the second group, 93.9% of the respondents lived their entire live in the Netherlands. Because of this it is logical that at the question ‘How many years do you live in the Netherlands?’ 77% of the respondents answers that they live their entire live at the Netherlands (Appendix 5.3). And in the second group, 93,9% have answered this. However, in the first group, 52.6% lived 16 till 20 years in the Netherlands. This means that in the second group are relatively more people who did not live their entire live in the Netherlands than in the first group. Overall 97.4% of the respondents have the Dutch Nationality (Appendix 5.4)

37 Of all the respondents 62.6% have parents who are both not working (Appendix 5.5). There are no serious differences between the first and the second group. Most of the respondents has one or two members in there family (inclusive the respondent) who are working (Appendix 5.6). In the first group, 40% of the respondents has one family member who is working and 46.3% has zero or two family members who are working. In the second group, 65.7% of the respondents has two working member in their family

On the question ‘In which country are you born?’ 72.3% of the respondents answered that they were born in the Netherlands and 21% was born in a non-western country (Appendix 5.7). Also 6.7% of the respondents were born in a western country. Again there is a difference between the two groups. In the second group, 93.3% of the respondents are born in the Netherlands. In the first group this percentage is 50%. In the first group 42.7% of the respondents were born in a non-western country. Around the same results are given when respondents have to answer the question ‘In which country are your parents born?’.

The respondents have different age (Appendix 5.8). In the first group 73.2% of the respondents are between 20 and 40 years old. In the second group 36.4% of the respondents are older than 50 years. This means that the respondents in the second group are relatively older than the respondents in the first group.

Most of the respondents (62.6%) describe their family income, when they lived at their parents, as average (Appendix 5.9). There are no serious differences on this result between the two groups. Almost 26% of the respondents have at this moment a family income between €900 and 1500 euro (Appendix 5.10). The respondents are spread over all income categories. Again there are differences between the first and the second group. In the first group, 51.5% of the respondents have an income between €900 and 1500 euro and 25.8% of the respondents have an income beneath €900 euro. In the second group only 6.2% of the respondents have an income beneath €900 euro. Respectively 28.9% and 34% of the respondents in the second group have an income between €1500 and € 3000 or above €3000. It can be concluded that the respondents in the second group have a relatively higher income than the respondents in the first group.

38 5.2 Factor analysis

After the data has been explored, a factor analysis is done. In the survey, different aspects are included to give an answer on the hypotheses and the main question. A factor analysis would help to know whether these different variables are driven by the same underlying variable (Field, 2005). A factor analysis identifies a group or cluster of variables, this is useful to understand the structure of the variables in the survey and it will reduce the data set to a smaller size witch is more manageable (Field, 2005).

Assumptions Two assumptions are checked before starting the factor analysis. First, the reliability of the sample, this is dependent on the size of the sample. As said before this research has a sample size of 200 respondents. To test the reliability, the Kaiser-Meyer-Olkin Measure of sampling adequacy (KMO) is used (Appendix 6). This statistic varies between 0 and 1; a value of 1 indicates that the factor analyses would be reliable. In this research KMO is 0.713, this is a good indication that the factor analyse would be reliable (Field, 2005).

Secondly, Bartlett’s test of sphericity is used to test if the original correlation matrix is an indentity matrix (Appendix 6). For a factory analysis it is needed that there are some relationships between the variables, otherwise no clusters would be found. In this research Bartlett’s test of sphericity is significant (0,000), this means that the correlation matrix is not an identity matrix and therefore relationships between the variables could be found (Field, 2005).

Data Screening To do factor analysis, variables are needed that correlate well but not perfectly (Field, 2005). Therefore the correlation matrix has to be screened. Any variable that does not correlate with any other variable should be eliminated. In this research there are no variables that do not correlate with any other variable, so no variables are eliminated at this stage. Also variables that correlate to perfectly should be eliminated. The variables with a correlation coefficient hihger than 0.9 should be eliminated. Therefore two variables are eliminated in this research, question 4 and 27. This was expected because question 4 was included in the questionnaire to check if the respondents understand question 3. Because question 3 is included in the factor analysis, question 4 is not needed anymore. The same is truth for question 5, so this variable is also eliminated.

39 Also the elimination of question 27 was expected, this question was included in the questionnaire to check if the respondents understand question 28. Because question 28 is included in the factor analyse, question 27 is not needed anymore.

Before starting the factor analysis, the variables question 42 and 48 are also eliminated. Question 42 is excluded because all respondents scored very high on this question. This question was about the personal characteristics of the respondents. For example, every respondent found him or herself very friendly and emotional stable. It is expected that respondents find it difficult to look objectively at their personal characteristics and therefore the answers on question 42 are not reliable. Question 48 is excluded because almost all respondents had the Dutch Nationality so it would not be a predicting variable. After the eliminated of the question described above, multicollinearity should not be a problem (Field, 2005).

Extraction and Rotation Kaiser’s criterion is used to retain factors; factors with eigenvalues greater than 1 are retained. Because of this, 12 factors are extracted and together they explain 72.2% of the total variance. The component matrix showed that most of the variables load highly onto the first factor, it is requested that all loading less than 0.4 will be suppressed in the output. The table reproduced correlations showed that the percentage of nonredundant residuals with absolute values greater than 0.05 is 19%. This percentage should be less than 50% and the smaller it is, the better (Field, 2005), therefore in this research there are no grounds for concern.

To optimize the factor structure a rotation is done. Varimax is choosen as the method for rotation. It is expected that the factors are independent and varimax tries to load a smaller number of variables highly onto each factor. Therefore the clusters of factors would be better interpretable (Field, 2005). The rotated component matrix (Appendix 7) shows that 12 factors are extracted. Appendix 7A show a scheme with the variables/questions in each factor.

40 Reliability Analysis For each factor the reliability of the scale have to be checked. A reliability analyse is used to measure the consistency of the questionnaire. To do a reliability analyse all questions that were reverse phrased are reverse scored (Field, 2005). The factors with a Cronbach’s α above 0.7 indicates that the measure is consistent

 Factor 1 has a Cronbach’s α of 0.802 and therefore the measure of this factor is consistent. This factor is going about the knowledge of the respondents about the Dutch education system and the Dutch labor market. The factor also measures if the respondent was mature enough to invest in education. In summary this factor measure if it is possible for the respondent to invest in education.

 Factor 2 has a Cronbach’s α of 0.814 and therefore the measure of this factor is consistent. This factor is going about the future time perspective of the respondents and the influence of the financial situation to invest in education.

 Factor 3 has a Cronbach’s α of 0.834 and therefore the measure of this factor is consistent. This factor is going about the demographic characteristics of the respondent.

 Factor 4 has a Cronbach’s α of 0.774 and therefore the measure of this factor is consistent. This factor is going about the influence of the cultural and social background. This factor is also going about making the right investments with enough information.

 Factor 5 has a Cronbach’s α of 0.774 and therefore the measure of this factor is consistent. This factor is going about the maturity of the respondent.

 Factor 6 has a Cronbach’s α of 0.659 and because this is close to 0.7 the measure of this factor is consistent. This factor is going about the influence of the social environment and the difficulty to follow education.

 Factor 7 has a Cronbach’s α of 0.731 and therefore the measure of this factor is consistent. This factor is going about the importance of investing in education and the financial consequences of investing in education

41  Factor 8 has a Cronbach’s α of 0.673 and because this is close to 0.7 the measure of this factor is consistent. This factor is going about the possibility to invest in education, based on the financial situation.

 Factor 9 has a Cronbach’s α of 0.812 and therefore the measure of this factor is consistent. This factor is going about the respondent’s perspective of the state of the labor market.

 Factor 10 has a Cronbach’s α of 0.057 and therefore the measure of this factor is not consistent. In this factor there is no connection between the questions and therefore this factor is eliminated from the analysis. Decided is not to use these question in the analysis because question 17 and 18 are covered in the other factors and question 34 seems to have no influence on the investments in education (66.6% of the respondent do not feel discriminated in a job interview).

 Factor 11 has a Cronbach’s α of 0.565, this is not so close to 0.7 but the measure of this factor seems to be consistent because there is a connection between the variables. This factor is going about the age of the respondents and the way they feel that there parents are expecting to much from them at the area of education.

 Factor 12 has a Cronbach’s α of 0.496 and therefore the measure of this factor is not consistent. In this factor there is no connection between the questions and therefore this factor is eliminated from the analysis. Decided is not to use question 12 in the analysis this question is covered in the other factors. Question 3 might be an important question so this question will be included in further analysis.

42 5.3 Regression analyse

This research contains four dependent variables, question 1, 2, 40 en 41. To explore the influence of each factor and question 3 on these dependent variables, multiple regression and binary logistic regression are used.

Multiple regressions seek to predict an outcome from several predictors. To explore the dependent variables question 1 and 2, multiple regression is used because the dependent variable is continuous and the independent variables are categorical or interval.

Multiple regression with dependent variable question 1 To explore the influence of the factors and question 3 on the value that people give themselves about how they have invested in education, three multiple regressions are accomplished.

1. In the first multiple regression all respondents (group 1 and group 2) are used. Forced entry is the method of regression that is chosen. In this method all predictors are forced into the model simultaneously (Field, 2005). This method is chosen because the chosen predictors are included for good theoretical reasons. The selected method for dealing with missing data points is to exclude cases listwise. This method is chosen to prevent that the data will be influenced.

The correlation matrix is checked to control for multicollinearity. Because there is no substantial correlation (R > .9) between predictors there is no multicollinearity. With the Durbin-Watson statistic, the assumption of independent errors is controlled. If the values are close to two this means that the assumption has almost certainly been met (Field, 2005). In this research the Durbin-Watson statistic is 2.091 which mean that the assumption of independent errors is tenable (Figure 1). Figure 1 also shows that the predictors explain for 58.7% the outcome variable.

Figure 1

Model Summaryb

Change Statistics Adjusted Std. Error of R Square Durbin- Model R R Square R Square the Estimate Change F Change df1 df2 Sig. F Change Watson 1 ,766a ,587 ,559 ,71938 ,587 21,288 11 165 ,000 2,091 a. Predictors: (Constant), Q3aangepast, Factor9, Factor5, Factor7, Factor3, Factor6, Factor11, Factor8, Factor1, Factor4, Factor2 b. Dependent Variable: Q1

43 The model above is a significant fit of the data overall because (as showed in the ANOVA table) the F-Ratio is very significant.

Figure 2 shows which predictors are making a significant contribution to the model. Factor 1, 2 and 4 have a significant positive effect on the outcome variable. Factor 5 and 6 have a significant negative effect on the outcome variable And factor 3, 7, 8, 9, 11 and question 3 have no significant effect on the outcome variable. Because there are no VIF greater than 10 and no tolerance below 0.2 figure 2 also shows that collinearity is not a problem for this model.

Figure 2

Coefficientsa

Unstandardized Standardized Coefficients Coefficients 95% Confidence Interval for B Correlations Collinearity Statistics Model B Std. Error Beta t Sig. Lower Bound Upper Bound Zero-order Partial Part Tolerance VIF 1 (Constant) 1,988 ,508 3,911 ,000 ,984 2,991 Factor1 ,518 ,067 ,534 7,787 ,000 ,387 ,650 ,638 ,518 ,390 ,533 1,877 Factor2 ,147 ,057 ,201 2,605 ,010 ,036 ,259 ,154 ,199 ,130 ,419 2,386 Factor3 -,103 ,075 -,081 -1,381 ,169 -,251 ,044 -,109 -,107 -,069 ,726 1,377 Factor4 ,516 ,076 ,475 6,767 ,000 ,365 ,666 ,564 ,466 ,339 ,508 1,969 Factor5 -,132 ,047 -,180 -2,826 ,005 -,225 -,040 ,044 -,215 -,141 ,620 1,612 Factor6 -,388 ,072 -,363 -5,394 ,000 -,530 -,246 ,140 -,387 -,270 ,555 1,803 Factor7 -,071 ,059 -,073 -1,221 ,224 -,187 ,044 ,272 -,095 -,061 ,707 1,414 Factor8 -,026 ,068 -,025 -,386 ,700 -,160 ,108 ,293 -,030 -,019 ,608 1,646 Factor9 ,038 ,036 ,063 1,070 ,286 -,032 ,108 ,017 ,083 ,054 ,725 1,380 Factor11 ,035 ,050 ,043 ,702 ,484 -,063 ,133 ,057 ,055 ,035 ,676 1,479 Q3aangepast ,034 ,036 ,061 ,954 ,342 -,037 ,105 ,333 ,074 ,048 ,610 1,640 a. Dependent Variable: Q1

The casewise diagnosic of this model shows that there are 9 cases that are outside the limits. We have a sample of 200 respondents, so it is reasonable to expect about 10 cases (5%) to have standardized residuals outside of these limits. This means that there is no reason for concern in this model.

To explore the outliers in this model, the standardized residuals are explored. Because less than 5% of the cases have absolute values above 2 and no more than 1% have absolute values above 2.5 there is no reason for concern (Field, 2005). To check if cases might be influencing the mode, Cooks distance is explored. Every value above 1 indicates that a case might be influencing, in this research very few cases have values above 1 and given that the other statistics are fine, this is probably no cause for concern (Field, 2005). In this research is also looked at the DFBeta statistics, an absolute value above 1 indicates that a case would have a large influence on the regression parameters (Field, 2005). In this research there are no absolute values of DFBeta above 1 so there will not be a problem.

44 To check the assumption of linearity and homoscedasticity, the standardized residuals are plotted against the standardized predicted values. Because the graph looks like a random array of dots, the assumption of linearity and homoscedasticity are met. To check the normality of residuals a histogram and normal probability plot are made. These figures show that the model has normal distributed residuals.

2. In the second multiple regression only the respondens from group 1, people who are following their education at SINL, are used. The same choices as in the first multiple regression are made. So, forced entry is the method of regression that is chosen. And the selected method for dealing with missing data points is to exclude cases listwise.

The correlation matrix is checked to conclude that there is no multicollinearity. In this research the Durbin-Watson statistic is 2.167 which mean that the assumption of independent errors is tenable (Figure 3). Figure 3 also shows that the predictors explain for 39.6% the outcome variable.

Figure 3

Model Summaryb

Change Statistics Adjusted Std. Error of R Square Durbin- Model R R Square R Square the Estimate Change F Change df1 df2 Sig. F Change Watson 1 ,629a ,396 ,298 ,95645 ,396 4,053 11 68 ,000 2,167 a. Predictors: (Constant), Q3aangepast, Factor2, Factor3, Factor6, Factor8, Factor5, Factor7, Factor11, Factor9, Factor4, Factor1 b. Dependent Variable: Q1

The model above is a significant fit of the data overall because the F-Ratio is very significant.

Figure 4 shows which predictors are making a significant contribution to the model. Factors 1 and 4 have a significant positive effect on the outcome variable. Factor 6 have a significant negative effect on the outcome variable And factor 2, 3, 5, 7, 8, 9, 11 and question 3 have no significant effect on the outcome variable. Because there are no VIF greater than 10 and no tolerance below 0.2 figure 2 also shows that collinearity is not a problem for this model. Figure 4

Coefficientsa

Unstandardized Standardized Coefficients Coefficients 95% Confidence Interval for B Correlations Collinearity Statistics Model B Std. Error Beta t Sig. Lower Bound Upper Bound Zero-order Partial Part Tolerance VIF 1 (Constant) 1,888 1,372 1,376 ,173 -,849 4,625 Factor1 ,314 ,129 ,282 2,434 ,018 ,056 ,571 ,403 ,283 ,229 ,659 1,516 Factor2 -,024 ,150 -,019 -,161 ,873 -,323 ,275 -,214 -,019 -,015 ,650 1,539 Factor3 -,054 ,120 -,043 -,453 ,652 -,294 ,185 -,060 -,055 -,043 ,976 1,024 Factor4 ,603 ,141 ,496 4,279 ,000 ,322 ,884 ,472 ,461 ,403 ,662 1,511 Factor5 -,036 ,083 -,046 -,432 ,667 -,202 ,130 ,011 -,052 -,041 ,785 1,274 Factor6 -,383 ,126 -,345 -3,036 ,003 -,635 -,131 -,045 -,346 -,286 ,689 1,452 Factor7 -,011 ,146 -,008 -,078 ,938 -,303 ,280 ,214 -,009 -,007 ,753 1,328 Factor8 ,187 ,154 ,125 1,218 ,228 -,119 ,493 ,231 ,146 ,115 ,849 1,177 Factor9 ,019 ,077 ,026 ,244 ,808 -,135 ,172 -,012 ,030 ,023 ,789 1,268 Factor11 ,052 ,117 ,048 ,446 ,657 -,181 ,285 -,003 ,054 ,042 ,780 1,282 Q3aangepast -,018 ,075 -,023 -,233 ,816 -,168 ,133 -,016 -,028 -,022 ,940 1,064 a. Dependent Variable: Q1 45 This model is controlled for outliers, influencing cases, homoscedasticity and linearity and there is no reason for concern. The same results as in the first multiple regression are found.

3. In the third multiple regression only the respondents from group 2, people who have not followed their education at SINL, are used. The same choices as in the first and second multiple regression are made.

The correlation matrix is checked to conclude that there is no multicollinearity. In this research the Durbin-Watson statistic is 2.159 which mean that the assumption of independent errors is tenable (Figure 5). Figure 5 also shows that the predictors explain for 92.9% the outcome variable.

Figure 5

Model Summaryb

Change Statistics Adjusted Std. Error of R Square Durbin- Model R R Square R Square the Estimate Change F Change df1 df2 Sig. F Change Watson 1 ,964a ,929 ,920 ,29414 ,929 100,966 11 85 ,000 2,159 a. Predictors: (Constant), Q3aangepast, Factor11, Factor7, Factor2, Factor9, Factor3, Factor4, Factor6, Factor1, Factor8, Factor5 b. Dependent Variable: Q1

The model above is a significant fit of the data overall because the F-Ratio is very significant.

Figure 6 shows which predictors are making a significant contribution to the model. Factors 1 2, 4, 9, 11 and question 3 have a significant positive effect on the outcome variable. Factors 5, 6, 7 and 8 have a significant negative effect on the outcome variable And factor 3 has no significant effect on the outcome variable. Because there are no VIF greater than 10 and only 1 tolerance below 0.2 figure 6 also shows that collinearity is not a problem for this model.

Figure 6

Coefficientsa

Unstandardized Standardized Coefficients Coefficients 95% Confidence Interval for B Correlations Collinearity Statistics Model B Std. Error Beta t Sig. Lower Bound Upper Bound Zero-order Partial Part Tolerance VIF 1 (Constant) ,388 ,781 ,496 ,621 -1,165 1,941 Factor1 ,732 ,047 ,835 15,715 ,000 ,640 ,825 ,838 ,863 ,454 ,296 3,379 Factor2 ,420 ,054 ,386 7,742 ,000 ,312 ,528 ,575 ,643 ,224 ,336 2,979 Factor3 ,203 ,158 ,072 1,285 ,202 -,111 ,516 -,420 ,138 ,037 ,263 3,803 Factor4 ,195 ,058 ,194 3,381 ,001 ,080 ,310 ,648 ,344 ,098 ,254 3,934 Factor5 -,203 ,046 -,282 -4,393 ,000 -,295 -,111 ,065 -,430 -,127 ,203 4,936 Factor6 -,414 ,059 -,363 -7,077 ,000 -,531 -,298 ,326 -,609 -,205 ,318 3,148 Factor7 -,228 ,035 -,281 -6,508 ,000 -,298 -,159 ,322 -,577 -,188 ,449 2,227 Factor8 -,155 ,047 -,169 -3,281 ,002 -,250 -,061 ,353 -,335 -,095 ,313 3,191 Factor9 ,115 ,029 ,192 4,015 ,000 ,058 ,172 ,024 ,399 ,116 ,365 2,742 Factor11 ,161 ,033 ,223 4,864 ,000 ,095 ,226 ,090 ,467 ,141 ,398 2,515 Q3aangepast ,149 ,047 ,286 3,148 ,002 ,055 ,243 ,609 ,323 ,091 ,101 9,884 a. Dependent Variable: Q1

46 This model is controlled for outliers, influencing cases, homoscedasticity and linearity and there is no reason for concern. The same results as in the first and second multiple regression are found.

In summarize, this model appears to be accurate for the sample and generalizable for the population, the assumptions seem to be met. Between the three multiple regression there are a couple of differences. First of all, in the third model with only the respondents who have not followed their education at SINL, the predictors explain the outcome variable very good (for 92.9%). In the, second, model with only the respondents who are followed their education at SINL, this was only 39.6%. The second difference is the significant influence of the variables. In figure 7 below is summarized which variable is influencing which group.

Figure 7 All respondents Group 1 Group 2 Factor 1 + + + Factor 2 + 0 + Factor 3 0 0 0 Factor 4 + + + Factor 5 - 0 - Factor 6 - - - Factor 7 0 0 - Factor 8 0 0 - Factor 9 0 0 + Factor 22 0 0 + Question 3 0 0 + + = significant positive influence - = significant negative influence 0 = no significant influence

47 Multiple regression with dependent variable question 2 To explore the influence of the factors and question 3 on the degree in which people would have liked to invest more in education, again three multiple regressions are accomplished.

1. In the first multiple regression all respondents (group 1 and group 2) are used. Forced entry is the method of regression that is chosen. The selected method for dealing with missing data points is to exclude cases listwise.

The correlation matrix is checked to conclude that there is no multicollinearity. In this research the Durbin-Watson statistic is 1.886 which means that the assumption of independent errors is tenable (Figure 8). Figure 8 also shows that the predictors explain for 51.7% the outcome variable.

Figure 8

Model Summaryb

Change Statistics Adjusted Std. Error of R Square Durbin- Model R R Square R Square the Estimate Change F Change df1 df2 Sig. F Change Watson 1 ,719a ,517 ,485 1,35745 ,517 15,869 11 163 ,000 1,886 a. Predictors: (Constant), Q3aangepast, Factor9, Factor5, Factor7, Factor3, Factor6, Factor11, Factor8, Factor1, Factor4, Factor2 b. Dependent Variable: Q2

The model above is a significant fit of the data overall because the F-Ratio is very significant.

Figure 9 shows which predictors are making a significant contribution to the model. Factor 2, 3, 4, 11 and question 3 have a significant positive effect on the outcome variable. Factor 1, 7 and 9 have a significant negative effect on the outcome variable And factor 5, 6 and 8 have no significant effect on the outcome variable. Because there are no VIF greater than 10 and no tolerance below 0.2 figure 2 also shows that collinearity is not a problem for this model.

Figure 9

Coefficientsa

Unstandardized Standardized Coefficients Coefficients 95% Confidence Interval for B Correlations Collinearity Statistics Model B Std. Error Beta t Sig. Lower Bound Upper Bound Zero-order Partial Part Tolerance VIF 1 (Constant) -,334 ,983 -,340 ,734 -2,275 1,606 Factor1 -,248 ,127 -,145 -1,958 ,052 -,498 ,002 -,045 -,152 -,107 ,538 1,857 Factor2 ,280 ,110 ,218 2,549 ,012 ,063 ,497 ,481 ,196 ,139 ,405 2,467 Factor3 ,508 ,145 ,227 3,512 ,001 ,222 ,793 ,376 ,265 ,191 ,710 1,408 Factor4 ,453 ,145 ,239 3,123 ,002 ,167 ,740 ,295 ,238 ,170 ,507 1,972 Factor5 -,060 ,089 -,046 -,671 ,503 -,235 ,116 ,226 -,052 -,036 ,625 1,599 Factor6 ,152 ,133 ,083 1,141 ,256 -,111 ,416 ,357 ,089 ,062 ,559 1,790 Factor7 -,302 ,110 -,176 -2,734 ,007 -,520 -,084 -,212 -,209 -,149 ,713 1,403 Factor8 -,205 ,130 -,112 -1,583 ,115 -,461 ,051 ,211 -,123 -,086 ,593 1,685 Factor9 -,225 ,067 -,214 -3,341 ,001 -,357 -,092 ,062 -,253 -,182 ,720 1,389 Factor11 ,532 ,095 ,372 5,573 ,000 ,344 ,721 ,431 ,400 ,303 ,664 1,505 Q3aangepast ,204 ,068 ,211 2,990 ,003 ,069 ,338 ,333 ,228 ,163 ,596 1,679 a. Dependent Variable: Q2

48 This model is controlled for outliers, influencing cases, homoscedasticity and linearity and there is no reason for concern. The same results as in the first multiple regression are found.

2. In the second multiple regression only the respondents from group 1, people who are following their education at SINL, are used. The same choices as in the first multiple regression are made. So, forced entry is the method of regression that is chosen. And the selected method for dealing with missing data points is to exclude cases listwise.

The correlation matrix is checked to conclude that there is no multicollinearity. In this research the Durbin-Watson statistic is 1.577 which means that the assumption of independent errors is relatively tenable (Figure 10). Figure 10 also shows that the predictors explain for 28.9% the outcome variable.

Figure 10

Model Summaryb

Change Statistics Adjusted Std. Error of R Square Durbin- Model R R Square R Square the Estimate Change F Change df1 df2 Sig. F Change Watson 1 ,538a ,289 ,171 1,41993 ,289 2,440 11 66 ,013 1,577 a. Predictors: (Constant), Q3aangepast, Factor4, Factor3, Factor5, Factor9, Factor8, Factor7, Factor11, Factor1, Factor6, Factor2 b. Dependent Variable: Q2

The model above is a significant fit of the data overall because the F-Ratio is very significant.

Figure 11 shows which predictors are making a significant contribution to the model. Factors 3 and 11 have a significant positive effect on the outcome variable. Factor 1 and 9 have a significant negative effect on the outcome variable And factor 2, 4, 5, 6, 7, 8, and question 3 have no significant effect on the outcome variable. Because there are no VIF greater than 10 and no tolerance below 0.2 figure 2 also shows that collinearity is not a problem for this model.

Figure 11

Coefficientsa

Unstandardized Standardized Coefficients Coefficients 95% Confidence Interval for B Correlations Collinearity Statistics Model B Std. Error Beta t Sig. Lower Bound Upper Bound Zero-order Partial Part Tolerance VIF 1 (Constant) 5,047 2,099 2,404 ,019 ,856 9,238 Factor1 -,386 ,195 -,248 -1,978 ,052 -,775 ,004 -,209 -,237 -,205 ,686 1,458 Factor2 ,018 ,226 ,010 ,081 ,936 -,433 ,469 ,031 ,010 ,008 ,663 1,507 Factor3 ,405 ,181 ,236 2,242 ,028 ,044 ,766 ,254 ,266 ,233 ,973 1,028 Factor4 ,077 ,212 ,046 ,366 ,716 -,345 ,500 -,097 ,045 ,038 ,678 1,475 Factor5 ,068 ,125 ,063 ,542 ,589 -,182 ,318 ,064 ,067 ,056 ,787 1,270 Factor6 ,036 ,185 ,024 ,193 ,848 -,333 ,404 ,025 ,024 ,020 ,684 1,462 Factor7 -,343 ,217 -,186 -1,582 ,118 -,776 ,090 -,244 -,191 -,164 ,776 1,288 Factor8 -,074 ,228 -,036 -,324 ,747 -,528 ,381 -,118 -,040 -,034 ,849 1,178 Factor9 -,352 ,114 -,363 -3,083 ,003 -,580 -,124 -,248 -,355 -,320 ,777 1,287 Factor11 ,329 ,180 ,217 1,822 ,073 -,032 ,689 ,170 ,219 ,189 ,759 1,317 Q3aangepast -,019 ,111 -,018 -,170 ,865 -,241 ,203 -,041 -,021 -,018 ,943 1,060 a. Dependent Variable: Q2

49 This model is controlled for outliers, influencing cases, homoscedasticity and linearity and there is no reason for concern. The same results as in the first multiple regression are found.

3. In the third multiple regression only the respondents from group 2, people who have not followed there education at SINL, are used. The same choices as in the first and second multiple regression are made.

The correlation matrix is checked to conclude that there is no multicollinearity. In this research the Durbin-Watson statistic is 2.427 which mean that the assumption of independent errors is relatively tenable (Figure 12). Figure 12 also shows that the predictors explain for 79.1% the outcome variable.

Figure 12

Model Summaryb

Change Statistics Adjusted Std. Error of R Square Durbin- Model R R Square R Square the Estimate Change F Change df1 df2 Sig. F Change Watson 1 ,889a ,791 ,764 ,86425 ,791 29,275 11 85 ,000 2,427 a. Predictors: (Constant), Q3aangepast, Factor11, Factor7, Factor2, Factor9, Factor3, Factor4, Factor6, Factor1, Factor8, Factor5 b. Dependent Variable: Q2

The model above is a significant fit of the data overall because the F-Ratio is very significant.

Figure 13 shows which predictors are making a significant contribution to the model. Factors 3, 4, 5, 9, 11 and question 3 have a significant positive effect on the outcome variable. Factors 1, 2, 6 and 8 have a significant negative effect on the outcome variable and factor 7 has no significant effect on the outcome variable. Because there are no VIF greater than 10 and only 1 tolerance below 0.2 figure 6 also shows that collinearity is not a problem for this model.

Figure 13

Coefficientsa

Unstandardized Standardized Coefficients Coefficients 95% Confidence Interval for B Correlations Collinearity Statistics Model B Std. Error Beta t Sig. Lower Bound Upper Bound Zero-order Partial Part Tolerance VIF 1 (Constant) -18,077 2,295 -7,876 ,000 -22,641 -13,513 Factor1 -,421 ,137 -,280 -3,072 ,003 -,693 -,148 ,100 -,316 -,152 ,296 3,379 Factor2 -,465 ,160 -,249 -2,914 ,005 -,782 -,148 ,345 -,301 -,144 ,336 2,979 Factor3 3,848 ,464 ,802 8,301 ,000 2,926 4,770 -,032 ,669 ,411 ,263 3,803 Factor4 ,713 ,169 ,414 4,210 ,000 ,376 1,050 ,480 ,415 ,209 ,254 3,934 Factor5 ,632 ,136 ,512 4,652 ,000 ,362 ,902 ,248 ,451 ,231 ,203 4,936 Factor6 -,606 ,172 -,310 -3,523 ,001 -,948 -,264 ,436 -,357 -,175 ,318 3,148 Factor7 -,161 ,103 -,115 -1,558 ,123 -,366 ,044 -,224 -,167 -,077 ,449 2,227 Factor8 -,771 ,139 -,490 -5,534 ,000 -1,048 -,494 ,189 -,515 -,274 ,313 3,191 Factor9 ,395 ,084 ,385 4,689 ,000 ,227 ,562 -,055 ,453 ,232 ,365 2,742 Factor11 1,029 ,097 ,834 10,604 ,000 ,836 1,222 ,452 ,755 ,526 ,398 2,515 Q3aangepast 1,354 ,139 1,520 9,752 ,000 1,078 1,630 ,296 ,727 ,483 ,101 9,884 a. Dependent Variable: Q2

50 This model is controlled for outliers, influencing cases, homoscedasticity and linearity and there is no reason for concern. The same results as in the first and second multiple regression are found.

In summarize, this model appears to be accurate for the sample and generalizable for the population, the assumptions seem to be met. Between the three multiple regression there are a couple of differences. First of all, in the third model with only the respondents who have not followed their education at SINL, the predictors explain the outcome variable relatively good (for 79.1%). In the, second, model with only the respondents who are followed their education at SINL, this was only 28.9%. The second difference is the significant influence of the variables. In figure 14 below is summarized which variable is influencing which group.

Figure 14 All Respondents Group 1 Group 2 Factor 1 - - - Factor 2 + 0 - Factor 3 + + + Factor 4 + 0 + Factor 5 0 0 + Factor 6 0 0 - Factor 7 - 0 0 Factor 8 0 0 - Factor 9 - - + Factor 22 + + + Question 3 + 0 + + = significant positive influence - = significant negative influence 0 = no significant influence

51 Binary Logistic regression with dependent variable question 40 Binary logistic regression is used here to predict outcome variables that are categorical. To explore the influence of the factors and question 3 on the question if people would have preferred it to invest earlier in education, three binary logistic regressions are accomplished.

1. In the first binary logistic regression all respondents (group 1 and group 2) are used. Forced entry is the method of regression that is chosen. The selected method for dealing with missing data points is to exclude cases listwise.

The classification table shows that the model correctly classifies 54.6% of the people. And in this model the value of the constant is 1,184. The Residual chi-square is 93.331 and this value is significant. This means that the coefficients for the variables not in the model are significantly different from zero. So, the addition of one or more of these variables to the model will significantly affect its predictive power (Field, 2005).

From the Omnibus tests of model coefficients it can be concluded that the value is significant and this means that overall the model is predicting Q40 better than it was with only the constant included. To explore if the observed data are significantly different from the predicted values of the model, the Hosmer & Lemeshow’s goodness-of-fit test is used. A non-significant value for this test indicates that the model does not differ significantly from the observed data (Field, 2005). In this research the value is significant (0.041) which means that it is possible that the model is predicting the real-world data not that well.

As noticed above, if only the constant was included, the model correctly classified 54.8% of the people. But if the predictors are included this has risen to 82.7%. Figure 15 shows which predictors are making a significant contribution to the model. If the coefficient from the Wald statistic is significantly different from zero, we can assume that the predictor have a significant influence to the prediction of the outcome (Field, 2005). Factor 2, 8 and 9 have a significant positive effect on the outcome variable. Factor 5 and 6 have a significant negative effect on the outcome variable And factor 1, 3, 4, 7, 11 and question 3 have no significant effect on the outcome variable.

Figuur 15 (output 16, variables in the equation)

52 This model is controlled for multicollinearity, outliers, influencing cases, homoscedasticity and linearity and there is no reason for concern.

2. In the second binary logistic regression only the respondents of group 1, people who are following their education at SINL, are used. Forced entry is the method of regression that is chosen. The selected method for dealing with missing data points is to exclude cases listwise.

The classification table shows that the model correctly classifies 77.3% of the people. And in this model the value of the constant is -1,126. The Residual chi-square is 7.141 and this value is not significant. This means that none of the variables excluded from the model could make a significant contribution to the predictive power of the model (Field 2005). With a stepwise regression, the residual chi-square is also not significant. Therefore the analysis is terminated at this stage.

3. In the third binary logistic regression only the respondents of group 2, people who have not followed their education at SINL, are used. Forced entry is the method of regression that is chosen. The selected method for dealing with missing data points is to exclude cases listwise.

The classification table shows that the model correctly classifies 85.9% of the people. And in this model the value of the constant is 1,804. The Residual chi-square is 81.306 and this value is significant. So, the addition of one or more of these variables to the model will significantly affect its predictive power (Field, 2005).

From the Omnibus tests of model coefficients it can be concluded that the value is significant and this means that overall the model is predicting Q40 better than it was with only the constant included. To explore if the observed data are significantly different from the predicted values of the model, the Hosmer & Lemeshow’s goodness-of-fit test is used. In this research the value is not significant (1) which means that the model is predicting the real- world data fairly well.

53 As noticed above, if only the constant was included, the model correctly classified 85.9% of the people. But if the predictors are included this has risen to 100%. The table variables in the equation show that no predictors are making a significant contribution to the model. Therefore it is decided to use a stepwise regression. The model than correctly classifies 87.9% of the people.

Figure 16 shows that only factor 9 and 1 are included significantly in the model. The factors 9 and 1 have a significant positive influence on the outcome variable.

Figuur 16 (output 21, variables in the equation)

This model is controlled for multicollinearity, outliers, influencing cases, homoscedasticity and linearity and there is no reason for concern.

In summarize, this model appears to be accurate for the sample and generalizable for the population, the assumptions seem to be met. Between the three multiple regression there are a couple of differences. First of all, in the third model with only the respondents who are not followed their education at SINL, the predictors correctly classifies the people very good (87.9%). In the, first model with all the respondents this was 82.7%. The second difference is the significant influence of the variables. In figure 17 below is summarized which variable is influencing which group.

Figure 17 All Respondents Group 1 Group 2 Factor 1 0 0 0 Factor 2 + 0 0 Factor 3 0 0 0 Factor 4 0 0 0 Factor 5 - 0 0 Factor 6 - 0 0 Factor 7 0 0 0 Factor 8 + 0 0 Factor 9 + 0 + Factor 11 0 0 0 Question 3 0 0 0 + = significant positive influence - = significant negative influence 0 = no significant influence

54 At question 40 was also asked what the reason was that people did not invest earlier in education. The answers differ a lot but there are two reasons that were named frequently. The first reason that people named frequently was about their children (38.2% of the respondents named this reason). People answered that they did not invest earlier in education because had to take care of their children. The second reason that people named frequently was about their financial situation (30.3% of the respondents named this reason). People answered that they did not have the financial possibilities to study in the past.

Binary Logistic regression with dependent variable question 41 To explore the influence of the factors and question 3 on the question if people would have preferred it to perform better in education, again three binary logistic regressions are accomplished.

1. In the first binary logistic regression all respondents (group 1 and group 2) are used. Forced entry is the method of regression that is chosen. The selected method for dealing with missing data points is to exclude cases listwise.

The classification table shows that the model correctly classifies 65.3% of the people. And in this model the value of the constant is 0.633. The Residual chi-square is 40.532 and this value is significant. So, the addition of one or more of these variables to the model will significantly affect its predictive power (Field, 2005).

From the Omnibus tests of model coefficients it can be concluded that the value is significant and this means that overall the model is predicting Q41 better than it was with only the constant included. In this research the value of the Hosmer & Lemeshow’s goodness-of-fit test is not significant (0.206) which means the model is predicting the real-world data fairly well.

As noticed above, if only the constant was included, the model correctly classified 65.3% of the people. But if the predictors are included this has risen to 75%. Figure 18 shows which predictors are making a significant contribution to the model. The factors 2 and 5 have a significant positive effect on the outcome variable. The factors 7 and 8 have a significant negative effect on the outcome variable. And factor 1, 3, 4, 6, 9, 11 and question 3 have no significant effect on the outcome variable. With a stepwise regression the results do not become better.

55 Figuur 18 (output 20, variables in the equation)

This model is controlled for multicollinearity, outliers, influencing cases, homoscedasticity and linearity and there is no reason for concern.

2. In the second binary logistic regression only the respondents of group 1, people who are following their education at SINL, are used. Forced entry is the method of regression that is chosen. The selected method for dealing with missing data points is to exclude cases listwise.

The classification table shows that the model correctly classifies 57.5% of the people. And in this model the value of the constant is 0.312. The Residual chi-square is 9.5 and this value is not significant. This means that none of the variables excluded from the model could make a significant contribution to the predictive power of the model (Field 2005). With a stepwise regression, the residual chi-square is also not significant. Therefore the analysis is terminated at this stage.

3. In the third binary logistic regression only the respondents of group 2, people who have not followed their education at SINL, are used. Forced entry is the method of regression that is chosen. The selected method for dealing with missing data points is to exclude cases listwise.

The classification table shows that the model correctly classifies 72.7% of the people. And in this model the value of the constant is 0.981. The Residual chi-square is 75.005 and this value is significant. So, the addition of one or more of these variables to the model will significantly affect its predictive power (Field, 2005).

From the Omnibus tests of model coefficients it can be concluded that the value is significant and this means that overall the model is predicting Q41 better than it was with only the constant included. In this research the value of theHosmer & Lemeshow’s goodness-of-fit test is not significant (1) which means that the model is predicting the real-world data fairly well.

56 As noticed above, if only the constant was included, the model correctly classified 72.7% of the people. But if the predictors are included this has risen to 100%. The table variables in the equation show that no predictors are making a significant contribution to the model. Therefore it is decided to use a stepwise regression. The model than correctly classifies 85.9%.

Figure 19 shows that only factor 5 is included significantly in the model. Factor 5 has a significant positive influence on the outcome variable.

Figuur 19 (output 24, variables in the equation)

This model is controlled for multicollinearity, outliers, influencing cases, homoscedasticity and linearity and there is no reason for concern.

In summarize, this model appears to be accurate for the sample and generalizable for the population, the assumptions seem to be met. Between the three multiple regression there are a couple of differences. First of all, in the third model with only the respondents who have not followed their education at SINL, the predictors correctly classifies the people very good (85.9%). In the first model, with all the respondents, this was 75%. The second difference is the significant influence of the variables. In figure 20 below is summarized which variable is influencing which group.

Figure 20 All Respondents Group 1 Group 2 Factor 1 0 0 + Factor 2 + 0 0 Factor 3 0 0 0 Factor 4 0 0 0 Factor 5 + 0 + Factor 6 0 0 0 Factor 7 - 0 0 Factor 8 - 0 0 Factor 9 0 0 0 Factor 11 0 0 0 Question 3 0 0 0 + = significant positive influence - = significant negative influence 0 = no significant influence

57 At question 41 was also asked what the reason was that people did not perform better in education. The answers differ a lot but there a two reasons that were named frequently. The first reason that people named frequently was about the education system (30.9% of the respondents named this reason). People answered that they did not perform better in education because they did not understand the education system or they found that the education system has limitations. The second reason that people named frequently was about their own character (45.6 % of the respondents named this reason). People answered that they did not have the discipline to study or they had other priorities.

5.4 Test of hypotheses Theory of Planned Behavior

In this section the hypothesis, who are made to see the influence of the different beliefs in the Theory of Planned Behavior, is tested. To test this, linear regression analyse is used.

Test of hypothesis 6 To explore if the Normative Belief Cultural & Social Background directly influences the Behavioral Belief future time perspective, hypothesis 6 was formed:

H6: The Normative Belief Cultural & Social Background directly influences the Behavioral Belief Time & Future perspective

To test hypothesis 6, three linear regression analyses are accomplished. To test this hypothesis the influence of factor 4 and 6 on factor 2 will be explored.

1. In the first linear regression all the respondent are used. The same choices as in the multiple regression analyses are made. So the selected method for dealing with missing data points is to exclude cases listwise.

Figure 21 shows that factor 4 and 6 explain for 12.9% the outcome variable factor 2.

Figure 21

Model Summaryb

Change Statistics Adjusted Std. Error of R Square Durbin- Model R R Square R Square the Estimate Change F Change df1 df2 Sig. F Change Watson 1 ,359a ,129 ,120 1,38524 ,129 14,289 2 193 ,000 1,101 a. Predictors: (Constant), Factor4, Factor6 b. Dependent Variable: Factor2

The model above is a significant fit of the data overall because the F-Ratio is very significant.

58 Figure 22 show that factor 6 has a significant positive influence on factor 2. Factor 4 does not have a significant influence on factor 2.

Figure 22

Coefficientsa

Unstandardized Standardized Coefficients Coefficients 95% Confidence Interval for B Correlations Collinearity Statistics Model B Std. Error Beta t Sig. Lower Bound Upper Bound Zero-order Partial Part Tolerance VIF 1 (Constant) 1,671 ,577 2,893 ,004 ,532 2,809 Factor6 ,432 ,111 ,301 3,883 ,000 ,213 ,652 ,349 ,269 ,261 ,749 1,335 Factor4 ,145 ,117 ,096 1,234 ,219 -,086 ,376 ,247 ,088 ,083 ,749 1,335 a. Dependent Variable: Factor2

This model is controlled for outliers, influencing cases, homoscedasticity and linearity and there is no reason for concern. The same results as in the first multiple regression are found.

2. If the influence of factor 4 and 6 on factor 2 is tested with only the respondents who are following their education at SINL other results are found. Factor 4 and 6 than explain for 7.8% the outcome variable factor 2. The F-ratio is significant so the model is a significant fit of the data overall. Factor 6 does not have a significant influence on factor 2. But factor 4 has a significant negative influence on factor 2.

3. If the influence of factor 4 and 6 on factor 2 is tested with only the respondents who have not followed their education at SINL other results are found. Factor 4 and 6 than explain for 30.6% the outcome variable factor 2. The F-ratio is significant so the model is a significant fit of the data overall. Factor 6 does not have a significant influence on factor 2. But factor 4 has a significant positive influence on factor 2.

59 6. Discussion

In chapter two, hypothesis 1 was explored and the result of this exploration is that foreigners invest indeed more frequently not enough in education than domestic people. This means that hypothesis 1 is true. In Chapter 5.3 and 5.4 the other hypotheses are explored and in this chapter the results will be discussed.

6.1 Hypotheses 2 and 3 In chapter 5.3 hypothesis 2 is tested, and for testing this hypothesis factor 9 is used. In the results is showed that factor 9 has no significant influence on question 1 as it was tested with all the respondents and within group 1. However, within group 2, factor 9 has a significant positive influence on question 1. This means that, for the respondents who have not followed their education at SINL, their perception of the state of the labor market has a significant positive influence on how they perceive their investments in education. So, people with a positive perception of the labor market (they think that there will be enough jobs) find frequently more times that they have invested enough in education. This outcome supports hypothesis 2. Within group 1, the perception of the state of the labor market has no influence on how they perceive their investments in education. There could be two reasons for this difference. As showed in chapter 4, in group 1 are relatively more people with a lower education than in group 1. It is possible that the respondents from group 1 interpreted the questions wrong. Another possibility is that, because of the fact that group 1 have relatively more foreign respondents, the respondents might find it difficult to create a perception of the state of the labor market.

From the results is also known that factor 9 has a significant negative influence on question 2 as it was tested with all the respondents and within group 1. However within group 2, factor 9 has a significant positive influence on question 2. This means that for the respondents who have not followed their education at SINL, their perception of the state of the labor market has a significant positive influence on the degree they wish to invest more in education. So, people with a positive perception of the labor market (they think that there will be enough jobs) find relatively more often that they did not want to invest more in education. This outcome supports hypothesis 2. Within group 1, the perception of the state of the labor market has a significant negative influence on the degree they wish to invest more in education. This is a strange outcome because this means that people, who perceive the state of the labor market as positive, find relatively more that they want to invest more in education.

60 Although this seems strange in general terms, for group 1 there could be an explanation. As noticed above it could be possible that the respondents interpreted the questions wrong. Another explanation could be the fact that all the respondents from group 1 has not invested enough in education and therefore they will probably all answer that they have wanted to invested more in education.

Factor 9 has a significant positive influence on question 40 as it was tested with all the respondents and within group 2. However, within group 1, factor 9 has no influence on question 40. This means that for the respondents who have not followed their education at SINL, their perception of the state of the labor market has a significant positive influence on the question if they wanted to invest earlier in education. So, people with a positive perception of the labor market (they think that there will be enough jobs) find relatively more often that they did not want to invest earlier in education. This outcome supports hypothesis 2. Within group 1 there is no influence of the perception of the state of the labor market on the question if they wanted to invest earlier in education. As noticed above this could be because the people in group 1 might find it difficult to to create a perception of the state of the labor market.

Finally, question 9 has no influence on question 41. This means that the perception of the labor market has no influence on the question if people had wanted to perform better in education. From the evidence above it can be concluded that H2, the perception of the state of the labor market influences the investments of people in education, is partly true. The hypothesis only holds for the respondents within group 2.

From the evidence above it can be also concluded that hypothesis 3, the Behavioral Belief Perception of the state of the labor market has a direct influence on the Control Belief Motivation & Earlier Performance, is partly true. The perception of the labor market has no influence on the question if people had wanted to perform better in education. But if the results are take from all the respondents, or from only group 2, the perception of the state of the labor market has a significant positive influence on the question if they wanted to invest earlier in education.

61 6.2 Hypotheses 4 and 5 In chapter 5.3 hypothesis 4 is tested, and for testing this hypothesis factor 2 is used. The results show that factor 2 has a significant positive influence on question 1 if it was tested with all the respondents and within group 2. Within group1, factor 2 has no influence on question 1. This means that for the all the respondents together en for only the respondents who have not followed their education at SINL, their future time perspective has a significant positive influence on how they perceive their investments in education. So, people with a more distant future time perspective find frequently more often that they have invested enough in education. This outcome supports hypothesis 4. Within group 1, the future time perspective has no influence on how people perceive their investments in education. There could be two reasons for this difference. As showed in chapter 4, in group 1 are relatively more people with a lower education than in group 1. It is possible that the respondents from group 1 interpreted the questions wrong. Another possibility is that, because of the fact that group 1 have relatively more foreign respondents, the respondents might find it difficult to have a realistic future time perspective.

From the results is also known that factor 2 has a significant positive influence on question 2 when it is tested with all the respondents. Within group 1 factor 2 has no influence on question 2. Within group 2 factor 2 has a significant negative influence on question 2. When the test was done with all the respondents, factor 2 had a significant positive influence on question 2. This means that people with a more distant future time perspective find relatively more often that they did not want to invest more in education. This outcome supports hypothesis 4. Within group 1, the future time perspective has no influence on the degree people wish to invest more in education. As noticed above it could be possible that the respondents interpreted the questions wrong or that the respondents might find it difficult to have a realistic future time perspective. For the respondents who have not followed their education at SINL, their future time perspective has a significant negative influence on the degree they wish to invest more in education. So, people with a more distant future time perspective find relatively more often that they want to invest more in education. This is the opposite conclusion as was noticed with all the respondents. An explanation for this could be that the question was wrongly interpreted or that a high percentage from all the respondents has indeed invested not enough in education, independent from their future time perspective.

62 Factor 2 has only significant positive influence on question 40 as it was tested with all the respondents. However, within group 1 and 2, factor 2 has no influence on question 40. This means that for all the respondents, their future time perspective has a significant positive influence on the question if they wanted to invest earlier in education. So, people with a more distant future time perspective find relatively more often that they did not want to invest earlier in education. This outcome supports hypothesis 4.

Factor 2 has only a significant positive influence on question 41 as it was tested with all the respondents. This means that the future time perspective of people have a significant positive influence on the question if people had wanted to perform better in education. So, people with a more distant future time perspective find relatively more often that they did not had wanted to perform better in their education. This outcome supports hypothesis 4. From the evidence above it can be concluded that H4: the future time perspective people have, are influencing the investments of those people in education, is partly true.

From the evidence above it can be also concluded that hypothesis 5, The Behavioral Belief Time & Future perspective of a person has a direct influence on the Control Belief Motivation & Earlier Performance, is partly true. People with a more distant future time perspective find relatively more often that they did not had wanted to perform better in their education. This outcome supports hypothesis 5. And people with a more distant future time perspective find relatively more often that they did not want to invest earlier in education. This outcome also supports hypothesis 5.

6.3 Hypothesis 6 In chapter 5.4 hypothesis 6 is tested, and for testing this hypothesis factor 4 and 6 are used. The results show that factor 4 has no influence on factor 2 if it was tested with all the respondents. But within group 1, factor 4 has a significant negative influence on factor 2 and within group 2 factor 4 has a significant positive influence on factor 2. This means that from the people who have not followed their education at SINL, their cultural and social background has a significant positive influence on their future time perspective. So people who are supported by their cultural and social background in their investments in education have relatively more often a more distant future time perspective. This outcome supports hypothesis 6. Within group 1, the opposite result was found. There could be two reasons for this difference. It is possible that the respondents from group 1 interpreted the questions wrong.

63 Another possibility is that, because of the fact that group 1 have relatively more foreign respondents, the respondents might find it difficult to have a realistic future time perspective.

The results from the test of hypothesis 6 also show that factor 6 has a significant positive influence on factor 2. But within group 1 and 2 there is no significant influence. This means that the cultural and social background of people has a significant positive influence on their future time perspective. So people who are supported by their cultural and social background in their investments in education have relatively more often a more distant future time perspective. This outcome supports hypothesis 6.

From the evidence above it can be concluded that H6: the Normative Belief Cultural & Social Background directly influences the Behavioral Belief Time & Future perspective, is partly true.

6.4 Hypotheses 7 and 8 In chapter 5.3 hypothesis 7 is tested, and for testing this hypothesis factor 4 and 11 are used.

The results show that factor 4 has a significant positive influence on question 1. This means that for the all the respondents together their social and cultural background has a significant positive influence on how they perceive their investments in education. So, people who are supported by their social and cultural background in their investments in education find frequently more often that they have invested enough in education. This outcome supports hypothesis 7.

From the results is also known that factor 4 has a significant positive influence on question 2 when it is tested with all the respondents and within group 2. Within group 1 factor 2 has no influence on question 2. This means that people who are supported by their social and cultural background in their investments in education find relatively more often that they did not want to invest more in education. This outcome supports hypothesis 7. Within group 1, the social and cultural background has no influence on the degree people wish to invest more in education. As noticed above it could be possible that the respondents interpreted the questions wrong.

Factor 4 has no significant influence on question 40 and 41. This means that the social and cultural background of people have no influence on the question if people wanted to invest earlier in education or on the question if people had wanted to perform better in education.

64 The results show that factor 11 has a significant positive influence on question 1, but only within group 2. Within group 1, factor 11 has no significant influence on question 1.This means that, for the people within group 2, social and cultural background has a significant positive influence on how they perceive their investments in education. So, people who are supported by their social and cultural background in their investments in education find frequently more often that they have invested enough in education. This outcome supports hypothesis 7.

From the results is also known that factor 11 has a significant positive influence on question 2. This means that people who are supported by their social and cultural background in their investments in education find relatively more often that they did not want to invest more in education. This outcome supports hypothesis 7.

Factor 11 has no significant influence on question 40 and 41. This means that the social and cultural background of people have no influence on the question if people wanted to invest earlier in education or on the question if people had wanted to perform better in education.

From the evidence above it can be concluded that H7: the social and cultural background people have is influencing the investments of those people in education, is partly true. It only holds for the respondents within group 2.

From the evidence above it can be also concluded that H8, the Normative Belief Cultural and Social background has a direct influence on the Control Belief Motivation & Earlier Performance, is not true. Because, factor 4 and 11 has no significant influence on question 40 and 41. This means that the social and cultural background of people have no influence on the question if people wanted to invest earlier in education or on the question if people had wanted to perform better in education.

6.5 Hypotheses 9 In this research there is a partly evidence for H9: the motivation and earlier performance of people with respect to their study are influencing the investments of those people in education. This due to the fact that the question about motivation are eliminated from the analysis because all the respondents scored themselves very high on these question. This might be because people find it difficult to look objectively to their own motivation. But with question 3 it was possible to test for the aspect ‘earlier performance’.

65 The results show that question 3 has a significant positive influence on question 1, but only within group 2. Within group 1 or with all the respondents, question 3 has no significant influence no question 1. This means that the earlier performance of people within group 2 has a significant positive influence on how they perceive their investments in education. So, people with a higher education find frequently more often that they have invested enough in education. This outcome supports hypothesis 9.

From the results is also known that question 3 has a significant positive influence on question 2 when it is tested with all the respondents and within group 2 or with all the respondents. Within group 1, question 3 has no influence on question 2. This means that people with a higher education find relatively more often that they did not want to invest more in education. This outcome supports hypothesis 9. Within group 1, earlier performance has no influence on the degree people wish to invest more in education. An explanation for this result could be that in group 1 are relatively more people with no or little education, so most of the time there was no earlier performance.

Question 3 has no significant influence on question 40 and 41. This means that earlier performance have no influence on the question if people wanted to invest earlier in education or on the question if people had wanted to perform better in education.

From the evidence above it can be concluded that H9: the motivation and earlier performance of people with respect to their study, are influencing the investments of those people in education, is partly true.

6.6 Hypothesis 10 In chapter 5.3 hypothesis 10 is tested, and for testing this hypothesis factor 1 and 5 are used.

The results show that factor 1 has a significant positive influence on question 1. This means that for the all the respondents together their maturity has a significant positive influence on how they perceive their investments in education. So, people who are more mature find frequently more often that they have invested enough in education. This outcome supports hypothesis 10. But the results of factor 5 show that factor 5 has a significant negative influence on question 1, when tested with all the respondents or within group 2. This suggest the opposite of the results of factor 1, namely that people who are more mature find frequently more often that they have invested not enough in education.

66 From the results is also known that factor 1 has a significant negative influence on question 2. This means that people who are more mature find relatively more often that they did want to invest more in education. Also in this case the results of factor 5 give the opposite effect. Factor 5 is only of influence within group 2, here factor 5 has a significant positive influence on question 2. This means that people who are more mature find relatively more often that they did not want to invest more in education. This outcome supports hypothesis 10.

Factor 1 has no significant influence on question 40. This means that the maturity of people has no significant influence on the question if they wanted to invest earlier in education. Factor 5 only showed a significant negative influence on question 40 with all the respondents. This means that if people are more mature the relatively find more often that they did not wanted to invest earlier in education. This outcome supports hypothesis 10.

Factor 1 has only significant positive influence on question 41 as it was tested within group 2. And factor 5 has only a significant positive influence on question 41 as it was tested within group 2 or with all the respondents. This means that the maturity of people have a significant positive influence on the question if people had wanted to perform better in education. So, people who are more mature find relatively more often that they did not had wanted to perform better in their education. This outcome supports hypotheses 10.

From the evidence above it can be concluded that H10: the degree of maturity of someone is influencing the decision to invest enough in education, is partly true. Unfortunately a few contradiction are noticed while analysing the results of hypotheses 10. This could indicate that the results of this hypothesis are unreliable. This could be due to the fact that people may find it difficult to look objectively to their own maturity.

6.7 Hypothesis 11 In chapter 5.3 hypothesis 11 is tested, and for testing this hypothesis factor 8 is used.

The results show that factor 8 has a significant negative influence on question 1 when it was tested within group 2. When the test was done with all the respondents or within group 1, factor 8 has no influence on question 1. This means, for the respondents in group 2, that the financial situation has a significant negative influence on how they perceive their investments in education. So, people who have or had the financial possibilites to invest in education find frequently more often that they have invested not enough in education. This outcome supports hypothesis 11.

67 An explanation for this result could be that people who had no financials possibilities to invest enough in education answered that they have invested enough in education based on their possibilities.

From the results is also known that factor 8 has a significant negative influence on question 2 when it was tested within group 2. This means that people who have or had the financial possibilites to invest in education find relatively more often that they did want to invest more in education. This outcome supports hypothesis 11.

Factor 8 only showed a significant positive influence on question 40 with all the respondents. This means that people who have or had the financial possibilities to invest in education find more often that they did not wanted to invest earlier in education. This result is the opposite of the other results. It could be that the question is interpreted wrongly.

Factor 8 has only significant negative influence on question 41 as it was with all the respondents. This means that the financial possibilities of people have a significant negative influence on the question if people had wanted to perform better in education. So, people who have or had the financial possibilities to invest in education find relatively more often that they did had wanted to perform better in their education. This outcome supports hypothesis 11.

From the evidence above it can be concluded that H11: the financial situation is influencing the decision to invest enough in education, is partly true. Unfortunately a contradiction is noticed while analysing the influence of factor 8 on question 40. This could indicate that the results of this hypothesis are unreliable. This could be due to the fact that respondents interpreted the question wrongly.

6.8 General conclusion From the analysis above it can be concluded that the following factors are influencing the investments that people make in education: the perception of the state of the labor market, the future time perspective, the social and cultural background, the earlier performance, the degree of maturity and the financial situation of people.

68 6.9 Limitations and suggestions for further research From the general conclusion is known wich factors are influencing the investments that people make in education. But, like any study, this research has its limitations.

First of all, two groups are used with different demographics. As noticed before, the first group has a lower level of education, contains more foreigners and their financial possibilities are lower. The second group has a higher level of education, contains more domestic people and their financial possibilities are higher. These two groups are the opposite from each other. These two groups dominate the sample, and it is possible that the results of this research are less applicable for other groups. Therefore it is recommended to do further research into the causes of not investing enough in education for other groups of people.

Secondly, not all the possible causes of the fact that people invest not enough in education are studied in this research. Therefore it is recommended to do further research in to other causes of not investing enough in education.

Third, the results of the research can be influenced negatively by the fact that some questions might interpreted wrongly. This might have happened because one group has a low level of education and a lot of foreign people are included. It is also possible that people have given the answers that are socially wished. A good example is the question in this research about personal characteristics (question 42). This question is eliminated from the research because all respondents scored themselves highly on the positive personal characteristics. Therefore it is recommended to do further research in this aspect and it would be better to observe the personal characteristics of people than to ask the person itself.

69 7. What can a company, like SINL BV, do to stimulate people to invest more and better in education?

As noticed before, the following factors are influencing the investments that people make in education: the perception of the state of the labor market, the future time perspective, the social and cultural background, the earlier performance, the degree of maturity and the financial situation of people. Most of these factors are difficult to be influenced by the school. But there are factors that the school can improve and which stimulate people to invest more and better in education.

At SINL there are a lot of students with problems, this can be behavior problems, debts, family problems etcetera. The “wetenschappelijk raad voor het regerinsbeleid” named this students overassigned students. Most of these overassigned students live in chaotic families, chaotic neigberhoods and chaotic schools. These are students who wants to get their diploma but because of all the problems it becomes all too much and the student will leave the school to early (WRR, 2009). Therefore these students invests not enough in education.

Structure The most important factor that the school can improve is the structure. Overassigned students need clear rules, clear teaching objectives and no poor execuses (WRR, 2009). Therefore the school will need good teachers who can maintain the rules but who can also create a close connection with the students. It is important that students feel appreciated and that they have faith in their teachters. Feelings of connection lead to a better performance at school and a lower dropout level (WRR, 2009). So, teachers need to learn, not only how to teach knowledge and skills, but also how to create a connection with the student. Beijing a teacher have to become more comprehensive.

Connection Feelings of connection between the school and the overassigned student can be improved by creating small teams who are responsible for a few students. The advantage of this structure is that teachers not only teach the students knowledge and skills but also have to take care of the student’s social-emotional needs (WRR, 2009). These teams should have frequently meeting to discuss the situation of the student and to decide wich method to use.

70 Integrated teams The solution is not only better teachers, but a whole better schoolteam. Because the overassigned students face most of the times serieus problems, it is necessary to operate on this early. Different people should work closely together to take care of the overassigned students. People who are important for the overassigned students are among other; the caretaker, the confident and the director.

Mixing If there are a lot of overassigned students in one school or class this will lead to commotion and less structure (WRR, 2009). It is therefore important to mix overassigned students with “normal” students.

Link to the practice Another step that schools can take to stimulate their studens to invest enough in education, is to create more professional education which are aligned to the pratice. If students can make the link between theory and practice, they will know why they are at school and what their perspective are on the labor market. This will stimulate the students to invest enough in education instead of working (WRR, 2009).

Conclusion

It can be concluded from the evidence above that it is of the most importance that the school support the student to invest enough in education. The school should not wash their hands of the overassigned students but they have to take care of them. All the people who have responsibilities to the overassigned students should pull together.

71 Appendixes

Appendix 1: Temporary bibliography Thesis

References from books and articles - Ajzen, I. (2002). “ Constructing a TpB Questionnaire: Conceptual and Methodological Considerations. Download from: www.people.umass.edu/aizen/pdf/tpb.measurement.pdf (27-04-2009). - Baumeister, Roy F. & Todd F. Heatherton (1996), ‘Self-Regulation Failure: An Overview’. In: Psychological Inquiry, 7 (1), 1-15. - Bierings, H. & Vries, R de (2007), ‘Jongeren en ouderen zonder startkwalificatie op de arbeidsmarkt’. In: Sociaaleconomische trends, 3e kwartaal 2007. - Blondal, S., Field, S & Girouard, N (2002). ‘Investment in Human Capital Through upper- secondary and tertiary education’. In: OECD Economic Studies, no. 34. - Blundell, R., Dearden, L., Meghir, C & Sianesi, B (1999). ‘Human Capital Investment: The Returns from Education and Training to the Individual, the Firm and the Economy’. In: Fiscal Studies, 20 (1), 1-23. - Dagevos, J (2006). “Hoge (jeugd) werkloosheid onder etnishce minderheden”. In: Sociaal Cultureel Planbureau, januari 2006. - Hoch, S.J. & Georgee F. Loewenstein (1991). ‘Time-Inconsistent Preferences and Consumer Self-Control’. In: Journal of Consumer Research, 17 (March), 492-507. - Kahneman, D. & Lovallo, D (1993). ‘ Timed Choices and Bold Forecasts: A Cognitive Perspective on Risk Taking’. In: Management Science, 39 (1), 17-31. - Koning, J. de., Gravesteijn-Ligthelm, J. & Tanis, O. (2008). “Wat bepaalt succes van allochtonen op de arbeidsmarkt?”. In: SEOR Working paper, no.2008/1. - Korvorst, M & van der Mooren, F. (2008). “Jongeren die niet meer leren, maar ook niet werken”. In: Sociaaleconomische trends. - Leclerc, F. & Bernd, H (1995). ‘Waiting Time and Decision Making: Is time like Money?’. In: Journal of Consumer Research, 22 (1), 110-119. - Nicaise, I (1997). ‘Armoede en menselijke kapitaal’. In: Onze Alma Mater, 4, 463-480. - Soman, D (1998). ‘The Illusion of Delayed Incentives: Evaluating Future Effort-Money Transaction’. In: Journal of Marketing Research, 35 (November), 427-437. - Traag, T. & Velden, R.K.W. van der (2007), ‘Voortijdig schoolverlaten vmbo, De rol van individuele kenmerken, gezinsfactoren en de school’. In: Sociaaleconomische trends, 2e kwartaal 2007.

72 - Trope, Y. & Liberman, N (2003). ‘Temporal Construal’. In: Psychological Review, 110 (3), 403-421. - Tubergen, F & Werforst, H (2006). ‘Post-Immigration Investments in Education’. Utrecht University & University of Amsterdam, paper prepared for the RC-28 meeting in Nijmegen, 11-14 May 2006. - Zhao, M., Hoeffler, S & Zaubermann, G (2007). ‘Mental Simulation and Preference Consistency over Time: The Role of Process-Versus Outcome-Focused Thoughts’. In: Journal of Marketing Research, augustus, 379-388.

Methodological references - Babbie, Earl., The practice of Social Research. Belmont, 2007 (Thomson Wadsworth) - Brink van den, A. W., ‘Scriptie-Schrijven. Rotterdam, 1995 (Buro Onderwijszaken Economische Faculteit Rotterdam) - Vromen, J.J., Syllbabus Methodology & Ethiek van de Economie. Rotterdam. 2006/2007 (Erasmus Universiteit Rotterdam)

Portfolio of sources - www.cbs.nl - www.oecd.com

73 Appendix 2: Framework Theory of Planned Behavior

74 Appendix 3: Survey

Enquête Investeren in toekomstig geluk?!

Zoals bij jou bekend, is investeren in onderwijs erg belangrijk. Voor het afronden van de Master Marketing aan de Erasmus Universiteit doe ik dan ook onderzoek naar het investeergedrag van mensen in onderwijs. De enquête wordt volledig anoniem afgenomen.

Alvast bedankt voor je medewerking aan dit onderzoek!

Voorbeeldvraag De meeste vragen in deze scriptie maken gebruik van een schaal met 7 cijfers. Ik wil je vragen het cijfer te omcirkelen dat het beste jouw ervaring weer geeft. Je kunt niet meerdere cijfers omcirkelen. Hieronder een voorbeeldvraag, deze hoef je niet te beantwoorden.

Het Nederlandse weer is:

Zeer slecht :___1__:___2__:___3__:___4__:___5__:___6__:___7__: Zeer goed

De 7 cijfers moeten dan als volgen geïnterpreteerd worden:

1 Zeer slecht 2 Behoorlijk slecht 3 Een beetje slecht 4 Neutraal 5. Een beetje goed 6. Behoorlijk goed 7. Zeer goed

75 Start van de enquête Ik wil je vragen om de vragen rustig door te lezen en zo goed mogelijk te beantwoorden. Er zijn geen goede of foute antwoorden, ik ben geïnteresseerd in jouw persoonlijke ervaringen. Sommige vragen lijken misschien hetzelfde maar deze belichten toch andere kwesties. Lees daarom de vragen goed door.

Belangrijke opmerking: Met genoeg investeren in onderwijs wordt bedoeld of je de voor jouw hoogste haalbare opleiding hebt afgerond of nog aan het volgen bent, op basis van jouw intelligentie, vaardigheden en ervaringen.

1. Hoe vind je dat je zelf geïnvesteerd hebt in onderwijs? Zeer weinig :___1__:___2__:___3__:___4__:___5__:___6__:___7__: Zeer veel

2. In welke mate had je graag meer willen investeren in onderwijs? Niet :___1__:___2__:___3__:___4__:___5__:___6__:___7__: In extreme mate

3. Van welk niveau (of vergelijkbaar met welke niveau) is je hoogst behaalde diploma?  Lager dan MAVO  MBO 2  MAVO  MBO 3  HAVO  MBO 4  VWO  HBO  MBO 1  Universitair  Anders namelijk …………………………………………………………………………………………

4. Welke opleidingen heb je gevolgd en afgemaakt? -………………………………………………………………………………………………… -………………………………………………………………………………………………… -………………………………………………………………………………………………… -…………………………………………………………………………………………………

76 5. Welke opleidingen heb je gevolgd en niet afgemaakt? -………………………………………………………………………………………………… -………………………………………………………………………………………………… -………………………………………………………………………………………………… -…………………………………………………………………………………………………

6. In welke mate vind je investeren in onderwijs belangrijk voor je toekomst? Zeer onbelangrijk :___1__:___2__:___3__:___4__:___5__:___6__:___7__: Zeer belangrijk

7. Als ik het had gewild had ik een hogere opleiding kunnen volgen. Zeer oneens :___1__:___2__:___3__:___4__:___5__:___6__:___7__: Zeer eens

8. De meeste mensen die belangrijk voor mij zijn vinden dat ik genoeg heb geïnvesteerd in onderwijs. Zeer oneens :___1__:___2__:___3__:___4__:___5__:___6__:___7__: Zeer eens

9. Het volgen van een goede opleiding was voor mij in het verleden: Onmogelijk :___1__:___2__:___3__:___4__:___5__:___6__:___7__: Mogelijk

10. Het volgen van een goede opleiding is voor mij: Zeer makkelijk :___1__:___2__:___3__:___4__:___5__:___6__:___7__: Zeer moeilijk

11. De verantwoordelijkheid voor het genoeg investeren in onderwijs ligt geheel bij mijzelf. Zeer oneens :___1__:___2__:___3__:___4__:___5__:___6__:___7__: Zeer eens

12. Veel mensen in mijn omgeving hebben niet genoeg geïnvesteerd in onderwijs. Zeer oneens :___1__:___2__:___3__:___4__:___5__:___6__:___7__: Zeer eens

13. Mijn omgeving verwacht van mij dat ik genoeg investeer in onderwijs. Zeer oneens :___1__:___2__:___3__:___4__:___5__:___6__:___7__: Zeer eens

77 14. Toen je ging investeren in onderwijs had je het idee dat er weinig werk was in de arbeidsmarkt. Zeer oneens :___1__:___2__:___3__:___4__:___5__:___6__:___7__: Zeer eens

15. Toen je ging investeren in onderwijs had je het idee dat er veel meer werklozen waren dan vacatures. Zeer oneens :___1__:___2__:___3__:___4__:___5__:___6__:___7__: Zeer eens

16. De hoogte van de eventueel te verkrijgen uitkering heeft invloed op jouw keuze om verder te investeren in onderwijs. Zeer oneens :___1__:___2__:___3__:___4__:___5__:___6__:___7__: Zeer eens

17. Het verschil tussen loon en uitkering is te klein om mensen te stimuleren aan het werk te gaan. Zeer oneens :___1__:___2__:___3__:___4__:___5__:___6__:___7__: Zeer eens

18. Met een diploma op minimaal MBO niveau 2 heb je meer kans op een goede baan dan zonder diploma’s. Zeer oneens :___1__:___2__:___3__:___4__:___5__:___6__:___7__: Zeer eens

19. Ik heb genoeg geïnvesteerd in onderwijs om een baan te vinden. Zeer oneens :___1__:___2__:___3__:___4__:___5__:___6__:___7__: Zeer eens

20. Ik heb het gevoel te weinig kennis en ervaring te hebben om kans te maken op een goede baan. Zeer oneens :___1__:___2__:___3__:___4__:___5__:___6__:___7__: Zeer eens

21. Ik heb in het verleden te weinig geïnvesteerd in onderwijs omdat de toekomst toen nog ver weg leek. Zeer oneens :___1__:___2__:___3__:___4__:___5__:___6__:___7__: Zeer eens

22. Ik investeer liever in een korte opleiding omdat het gewenste resultaat dan niet lang op zich laat wachten. Zeer oneens :___1__:___2__:___3__:___4__:___5__:___6__:___7__: Zeer eens

78 23. Investeren in onderwijs zorgt ervoor dat ik later meer ga verdienen. Zeer oneens :___1__:___2__:___3__:___4__:___5__:___6__:___7__: Zeer eens

24. De kosten van het volgen van een opleiding heb ik later zo terug verdiend. Zeer oneens :___1__:___2__:___3__:___4__:___5__:___6__:___7__: Zeer eens

25. Als ik moest kiezen tussen betaald werk nu of een opleiding van 4 jaar en daarna beter betaald werk dan kies ik ervoor om nu te gaan werken. Zeer oneens :___1__:___2__:___3__:___4__:___5__:___6__:___7__: Zeer eens 26. Ik heb/had genoeg kennis van de arbeidsmarkt om te bepalen welke opleiding ik wil(de) volgen. Zeer oneens :___1__:___2__:___3__:___4__:___5__:___6__:___7__: Zeer eens

27. Woon jij al je gehele leven in Nederland? Ja / Nee

28. Indien je niet je gehele leven in Nederland hebt gewoond, hoeveel jaar woon je dan nu in Nederland? -…………………………………………………………………………………………………

29. Toen ik mijn opleiding koos had ik genoeg informatie om een goede beslissing te nemen. Zeer oneens :___1__:___2__:___3__:___4__:___5__:___6__:___7__: Zeer eens

30. Het Nederlandse onderwijssysteem is mij bekend en duidelijk. Zeer oneens :___1__:___2__:___3__:___4__:___5__:___6__:___7__: Zeer eens

31. Ik ken de mogelijkheden die er in Nederland zijn tot het volgen van een opleiding. Zeer oneens :___1__:___2__:___3__:___4__:___5__:___6__:___7__: Zeer eens

32. Ik heb het gevoel dat mijn ouders soms te veel van mij verwachten op het gebied van onderwijs. Zeer oneens :___1__:___2__:___3__:___4__:___5__:___6__:___7__: Zeer eens

79 33. Mijn ouders hebben altijd achter mij gestaan bij mijn onderwijskeuzes. Zeer oneens :___1__:___2__:___3__:___4__:___5__:___6__:___7__: Zeer eens

34. Bij sollicitatieprocedures voel ik mij wel eens gediscrimineerd. Zeer oneens :___1__:___2__:___3__:___4__:___5__:___6__:___7__: Zeer eens

35. Ik krijg genoeg steun van mijn familie om de juiste investeringen te maken op het gebied van onderwijs. Zeer oneens :___1__:___2__:___3__:___4__:___5__:___6__:___7__: Zeer eens

36. Ik krijg genoeg steun van mijn sociale omgeving (vrienden, kennissen, collega’s etcetera), om de juiste investeringen te maken op het gebied van onderwijs. Zeer oneens :___1__:___2__:___3__:___4__:___5__:___6__:___7__: Zeer eens

37. Ben je wel eens gestopt met een opleiding omdat je niet werd gesteund door jouw leraren? Ja / Nee

38. Werken allebei je ouders? Ja / Nee

39. Uit hoeveel werkende leden bestaat jouw gezin -………………………………………………………………………………………………….

40. Had je eerder willen investeren in onderwijs? Ja / Nee Indien je ja hebt geantwoord ga verder met vraag 40A. Indien je nee hebt geantwoord ga door met vraag 41.

80 40 A. Wat zijn de drie voornaamste redenen dat je niet eerder hebt geïnvesteerd in onderwijs? -………………………………………………………………………………………………… -………………………………………………………………………………………………… -…………………………………………………………………………………………………

41. Had je beter willen presteren in de door jouw gevolgde opleidingen? Ja / Nee Indien je ja hebt geantwoord ga verder met vraag 41A. Indien je nee hebt geantwoord ga door met vraag 42.

41 A. Wat zijn de drie voornaamste redenen dat je niet beter hebt kunnen presteren in deze opleidingen? -………………………………………………………………………………………………… -………………………………………………………………………………………………… -…………………………………………………………………………………………………

42. In welke mate beschik jij over de volgende eigenschappen;

Extraversie: Je bent gericht op de buitenwereld en vind het prettig om dingen samen te doen. Je hebt geen moeite met het praten in gezelschap en je kan goed voor jezelf opkomen. Zeer weinig :___1__:___2__:___3__:___4__:___5__:___6__:___7__: Zeer veel

Vriendelijkheid Zeer weinig :___1__:___2__:___3__:___4__:___5__:___6__:___7__: Zeer veel

Zorgvuldigheid Zeer weinig :___1__:___2__:___3__:___4__:___5__:___6__:___7__: Zeer veel

Emotionele Stabiliteit: Je bent vaak tevreden met jezelf en je bent niet snel van slag. Je voelt je snel ontspannen en je bent niet vaak emotioneel. Zeer weinig :___1__:___2__:___3__:___4__:___5__:___6__:___7__: Zeer veel

81 Openheid voor nieuwe ervaringen/ideeën: Je bent nieuwsgierig en gaat graag op zoek naar nieuwe ervaringen. Zeer weinig :___1__:___2__:___3__:___4__:___5__:___6__:___7__: Zeer veel

43. Toen ik begon met investeren in onderwijs was ik hier volwassen en zelfstandig genoeg voor. Zeer oneens :___1__:___2__:___3__:___4__:___5__:___6__:___7__: Zeer eens

44. Toen ik begon met investeren in onderwijs vond ik het lastig om de zaken rondom mijn opleiding goed te plannen en organiseren. Zeer oneens :___1__:___2__:___3__:___4__:___5__:___6__:___7__: Zeer eens

45. Toen ik begon met investeren in onderwijs vond ik het lastig om mij in te leven in anderen. Zeer oneens :___1__:___2__:___3__:___4__:___5__:___6__:___7__: Zeer eens

46. Toen ik begon met investeren in onderwijs vond ik het lastig om een taak helemaal tot eind goed te volbrengen. Zeer oneens :___1__:___2__:___3__:___4__:___5__:___6__:___7__: Zeer eens

47. Als ik meer financiële middelen had gehad had ik meer geïnvesteerd in onderwijs. Zeer oneens :___1__:___2__:___3__:___4__:___5__:___6__:___7__: Zeer eens

48. Ben je in het bezit van de Nederlandse Nationaliteit? Ja / Nee

49. In welk land ben je geboren? -…………………………………………………………………………………………………

50. In welke land zijn je ouders geboren? -…………………………………………………………………………………………………

82 51. Wat is jouw leeftijd?  Jonger dan 20 jaar  Tussen de 20 en 30 jaar  Tussen de 30 en 40 jaar  Tussen de 40 en 50 jaar  Ouder dan 50 jaar

52. Hoe zou jij het gemiddelde gezinsinkomen beschrijven toen je nog bij je ouders woonde?  Een onder gemiddeld inkomen  Een gemiddeld inkomen  Een boven gemiddeld inkomen

53. Wat is nu jouw netto gezinsinkomen per maand?  Minder dan 900 euro  Tussen de 900 en 1500 euro  Tussen de 1500 en 3000 euro  Boven de 3000 euro  Ik weet het niet

Bedankt voor je medewerking!

Appendix 3A  Question 3, 4, 5: These question are coded based on the level of education. No diploma is 0, beneath MAVO is 1, MAVO is 2, HAVO/VWO/MBO 1 and MBO2 are 3, MBO 3 is 4, MBO 4 is 5, HBO is 6 and university is 7;  Yes or No question: These question are coded so that yes is 1 and no is 2;  Question 28: These questions are coded based on the years people live in the Netherlands. 0-5 years is 1, 5-10 years is 2, 10-15 years is 3, 15-20 years is 4, more than 20 years is 5 and their entire live is 6.  Question 40A en 41A: These questions are coded in different categories. Each reason has a number.

83  Question 49 and 50: These questions are coded in different countries. Each country has a number.  Question 51, 52 and 53: These questions are coded so that the lowest option gets the lowest code and the highest option get the highest code.

84 Appendix 5: Missing Values

Descriptive Statistics

Std. Analysis Mean Deviation(a) N(a) Missing N Q3 5,0510 3,17481 196 0 Q4 4,8061 3,28995 196 0 Q5 2,2398 3,22787 196 0 Q6 6,3622 1,15310 196 0 Q7 4,9592 1,92909 196 0 Q8 5,2821 1,47019 196 1 Q9 4,8718 1,88625 196 1 Q10 4,1436 1,58515 196 1 Q11 5,6289 1,49405 196 2 Q12 4,3351 1,55781 196 2 Q13 5,1231 1,46593 196 1 Q14 3,3795 1,99476 196 1 Q15 3,7231 1,95730 196 1 Q16 3,4513 2,16533 196 1 Q17 4,4103 1,96531 196 1 Q18 5,9133 1,28008 196 0 Q19 5,4592 1,53368 196 0 Q20 2,9745 1,74367 196 0 Q21 3,1333 2,09582 196 1 Q22 4,0102 2,07300 196 0 Q23 5,8769 1,34555 196 1 Q24 5,3061 1,54195 196 0 Q25 3,9082 2,16066 196 0 Q26 5,1735 1,44666 196 0 Q27 1,2296 ,42165 196 0 Q28 ,8061 1,58297 196 0 Q29 5,4643 1,53046 196 0 Q30 5,6051 1,38600 196 1 Q31 5,6769 1,22923 196 1 Q32 3,3918 1,89226 196 2 Q33 5,7128 1,59151 196 1 Q34 2,9487 2,06992 196 1 Q35 5,3010 1,84146 196 0 Q36 5,6000 1,33743 196 1 Q37 1,8163 ,38821 196 0 Q38 1,6224 ,48602 196 0 Q39 2,0876 1,47719 196 2 Q42.1 5,6173 1,16421 196 0 Q42.2 6,1582 ,90610 196 0 Q42.3 5,9439 1,08695 196 0 Q42.4 4,9643 1,15636 196 0 A42.5 6,0923 ,88965 196 1 Q43 5,3420 1,58103 196 3 Q44 3,9485 1,87491 196 2 Q45 3,1237 1,77884 196 2

85 Q46 3,3557 1,73421 196 2 Q47 4,0821 2,24258 196 1 Q48 1,0255 ,15807 196 0 Q49 1,3436 ,59906 196 1 Q50 1,3692 ,57919 196 1 Q51 3,2245 1,24473 196 0 Q52 2,0462 ,61011 196 1 Q53 3,0206 1,38104 196 2 a For each variable, missing values are replaced with the variable mean.

86 Appendix 5 : Demographic Characteristics

Appendix 5.1

What is the highest level of the diplomas you have?

No diploma Beneath MAVO MAVO HAVO VWO MBO 1 MBO 2 MBO 3 MBO 4 HBO University

Appendix 5.2

Have you lived your entire live in the Netherlands?

Yes No

87 Appendix 5.3

How many yours do you live in the Netherlands?

Entire live 0 - 5 years 6 - 10 years 11 - 15 years 16 - 20 years More than 20 years

Appendix 5.4

Do you have the Dutch nationality?

Yes No

88 Appendix 5.5

Are your parents both working?

Yes No

Appendix 5.6

89 Appendix 5.7

Appendix 5.8

What is your age?

Younger than 20 years 20 - 30 years 31 - 40 years 41 - 50 years Older than 50 years

90 Appendix 5.9

Appendix 5.10

91 Appendix 6: KMO and Bartlett’s Test

Kaiser-Meyer-Olkin Measure of Sampling Adequacy. ,713

Bartlett's Test of Approx. Chi-Square 5046,395 Sphericity df 946 Sig. ,000

92 Appendix 7: Rotated Component Matrix

Rotated Component Matrixa

Component 1 2 3 4 5 6 7 8 9 10 11 12 Q30 ,777 Q26 ,751 Q31 ,747 Q43 ,656 Q21 ,805 Q22 ,637 Q20 ,575 Q47 ,551 Q16 ,504 ,472 Q7 ,472 Q49 ,909 Q50 ,883 Q28 ,862 Q39 ,471 Q36 ,726 Q8 ,622 Q13 ,615 Q19 ,553 Q29 ,431 ,488 Q11 ,468 Q45 ,832 Q44 ,771 Q46 ,587 Q37 Q35 ,660 Q10 ,634 Q33 ,632 Q38 ,510 ,409 Q23 ,804 Q24 ,739 Q6 ,422 ,582 Q25 ,418 Q52 ,744 Q53 ,710 Q9 ,404 Q14 ,868 Q15 ,795 Q17 ,732 Q34 ,419 ,407 ,443 Q18 -,416 Q51 ,799 Q32 ,656 Q12 ,787 Q3 ,431 ,493 Undefined error #11401 - Cannot open text file "c:\program files\spsseval\en\windows\spss.err": No s Undefined error #11408 - Cannot open text file "c:\program files\spsseval\en\windows\spss.err": No s a. Rotation converged in 21 iterations.

93 Appendix 7A: Scheme Factors

Factor Questions 1 30, 31, 26, 43 2 21, 22, 47, 7, 20,16 3 49, 50, 28,39 4 13, 36, 19, 11, 8, 29 5 44, 45,46 6 35, 10, 33, 38 7 23, 24, 6,25 8 52, 53,9 9 14,15 10 17, 34,18 11 51, 32 12 12, 3

94 Appendix 8: Reliability Analysis

Factor 1

Reliability Statistics

Cronbach's Alpha Based on Cronbach's Standardized Alpha Items N of Items ,792 ,802 4

Factor 2

Reliability Statistics

Cronbach's Alpha Based on Cronbach's Standardized Alpha Items N of Items ,817 ,814 6

Factor 3

Reliability Statistics

Cronbach's Alpha Based on Cronbach's Standardized Alpha Items N of Items ,781 ,834 4

Factor 4

Reliability Statistics

Cronbach's Alpha Based on Cronbach's Standardized Alpha Items N of Items ,772 ,774 6

95 Factor 5

Reliability Statistics

Cronbach's Alpha Based on Cronbach's Standardized Alpha Items N of Items ,744 ,744 3

Factor 6

Reliability Statistics

Cronbach's Alpha Based on Cronbach's Standardized Alpha Items N of Items ,653 ,659 4

Factor 7

Reliability Statistics

Cronbach's Alpha Based on Cronbach's Standardized Alpha Items N of Items ,719 ,731 3

Factor 8

Reliability Statistics

Cronbach's Alpha Based on Cronbach's Standardized Alpha Items N of Items ,572 ,673 3

Factor 9

Reliability Statistics

Cronbach's Alpha Based on Cronbach's Standardized Alpha Items N of Items ,812 ,812 2

96 Factor 10

Reliability Statistics

Cronbach's Alpha Based on Cronbach's Standardized Alpha Items N of Items ,192 ,057 3

Factor 11

Reliability Statistics

Cronbach's Alpha Based on Cronbach's Standardized Alpha Items N of Items ,531 ,565 2

Factor 12

Reliability Statistics

Cronbach's Alpha Based on Cronbach's Standardized Alpha Items N of Items ,414 ,496 2

97