Special Article Class Awareness in Japan and the U. S. Expansion and Stability*

Toru Kikkawa Sho Fujihara Osaka University Osaka University

Abstract In this paper, we examine the determinants of subjective . This topic used to be discussed extensively in the U.S., although few consider the present state of social status. We focus on the sequence of temporal change in Japan in comparison with that in the U.S. Thus, we apply a multigroup MIMIC model to the longitudinal dataset from both societies. We feature subjective social status as the dependent variable. This is measured as a latent concept by three indicators: class identification, self-ranked , and satisfaction with standard of living. As independent variables, we examine the general aspects of objective social status: education, occupation, and household income. Significance tests of the model fit indicate that objective social status gradually gains influence to subjective social status in Japan, while the determinant structure is almost equally maintained in the U.S. Expanding class awareness observed in Japan calls for researchers to pay attention to future consequences not only in the U.S., but also other countries.

Keywords: subjective social status, education, Japan, multigroup MIMIC model

1 Introduction

Representing one’s subjective social status, class identification holds the central position of class consciousness.1) Close association of and class consciousness has been expected under the influence of the Marxist thesis that “the mode of production of material life conditions the general process of social, political, and life” (Marx [1859] 1977:21). Investigation of the relationship between objective and subjective social statuses was one of the major issues of stratification studies in the 20th century. In such studies, occupational position is regarded as the core of objective social status. Subsequently, the scope is enhanced by adding education and income as components of objective social status according to Weberian understanding. On the other hand, one’s class identification, which reflects a sense of belonging to a particular social class, such as “upper

* We thank the 2005 SSM Research Committee for allowing us to use the SSM surveys. We also thank the SSP Project for allowing us to use the SSP-I2010 survey. This study is supported by the ISM Collaboration Program 24-J-4204 in 2012 and the JSPS Grant-in Aid for Scientific Research (S) 23223002. Corresponding author: Toru Kikkawa, Graduate School of Human Sciences, Osaka University, 1-2 Yamadaoka, Suita, Osaka 565-0871, Japan. E-mail: [email protected]

⌮ㄽ࡜᪉ἲ(Sociological Theory and Methods) 2012, Vol.27, No.2: 205224 205 ⌮ㄽ࡜᪉ἲ class,” “,” “,” or “lower class,” is assumed to the subjective counterpart. Using data from the U.S. collected in 1947, Centers (1949) examines the relationship as a pioneer. He stresses the relevance of occupational status, and his analysis shows the significance of education as a determinant of class identification. Analyzing the 1964 survey data from the U.S., Hodge and Treiman (1968) reveal that education, occupation, and income have independent contributions to one’s class identification, although the total amount of explanation (R-square = .196) is far insufficient for the theoretically expected state of accordance. Jackman and Jackman (1973, 1983) also observe the influence of occupation, education, and household income on class identification of American men and women. They claim that the parameter estimates in the 1970s are similar in magnitude to those obtained with data from the 1960s. These findings imply that a multiple-determinant structure, which consists of the effects of education and income as well as occupation, is time invariant in the U.S. Indeed, various studies report the supplemental effects of properties, social networks, ethnicity, gender, urbanity, and age on class identification, even though education, occupation, and income maintain the relevance in most cases (Cannon 1984; Gross 1953; Hodge and Treiman 1968; Hout 2008; Macy 1989; Yamaguchi and Wang 2002). Nowadays, with an exception accomplished by Hout (2008), less effort is devoted to examining the relationship between socio-economic status and subjective social status. This is not because of an insufficiency in survey data collection, but because researchers have lost interest in examining the current objective-subjective association. As the data shows results similar to those of previous studies, researchers deem this multiple-determinant structure, which predicts about one-fifth of variance in class identification, to be an invariant assumption.2) Because of substantially unchanging predictability, the topic that was once extensively discussed tends to be pushed aside as a thing of the past. Nevertheless, this paper dares to shed light on the changing trend of the objective-subjective association in data from contemporary Japan.

2 Analytical Framework

2.1 Measuring Subjective Social Status with Multiple Indicators Despite the theoretical presupposition, previous studies derive the fact that occupational status is neither a unique nor dominant determinant of class identification in today’s society. Hout (2008) empirically shows ambivalence stemming from status inconsistency and ambiguity about the borderline that separate the working class from the middle class diminish the influence of occupational status in the U.S. This irrelevancy of occupation is frequently associated with the proclamation of the death of the class (Bauman 1982; Beck 1992). Therefore, it is no more meaningful to stick to peoples’ identification of a particular class category, which consists of a set of occupations. Rather, we should investigate subjective social status in a more general manner. In this aspect, Kluegel et al. (1977) introduce a structural equation modeling with multiple indicators to measure subjective social status (subjective class identification in their words), instead

206 Class Awareness in Japan and the U. S. of using a single indicator of classic class identification. They suggest that the use of the multiple indicator method corrects for attenuation in objective-subjective association.3) Based on these arguments, we apply the multiple indicator method to measure subjective social status as a latent concept. We pick up two complementary measurement variables in addition to classic class identification; self-ranked social position and satisfaction with standard of living. We measure subjective social status as a latent concept with these three items. Self-ranked social position with a 10-point scale has developed as an alternative version of class identification. Looking at the General Social Survey (GSS), this version was introduced in several survey-years along with classic class identification. It is also used in the International Social Survey Programme (ISSP) and comparative surveys in East Asia; the and Survey (SSM) and East Asian Social Surveys (EASS). As the numerical scale ensures that there is no semantic uncertainty, we assume that the scale is a less-biased version for class identification. However, it is also true that responses with numerical values do not reflect the sense of belonging to concrete social groups. Therefore, classic class identification and self-ranked social position ought to be utilized complementarily. We also bring satisfaction with standard of living into focus. Life satisfaction is generally treated as a central item of well-being. However, as far as examining subjective social status, we may understand that it represents one’s evaluation of their standard of living. Class identification and satisfaction with standard of living practically correlate with each other to a certain degree (Centers 1949; Kikkawa 2000; Yamaoka 2008).

2.2 Temporal Changes in the Effects: A Different Story in Japan The consequences of the studies of subjective social status in Japan are different from those in the U.S. in several ways. In the 1950s when the SSM started, the linkage of social status and class identification was treated as a matter of great concern. The linkage reflected not only the scholastic arguments of Marxist class theory against a pluralistic stratification theory but also a broad social concern about the proportion and composition of each class category. Despite this, researchers found that the objective-subjective association was unexpectedly weak. The story was similar in the U.S. However, the observed relationship was even smaller in Japan. One of the reasons is that Japanese people do not have the habit of referring to words such as “middle class” or “working class” in perceiving social position; rather, they tend to place themselves in a stratum. Therefore, the Japanese version of class identification has slightly different response categories; that is, it has a five-point ordinal scale of “the upper,” “the upper middle,” “the lower middle,” “the upper low,” and “the lowest.” Thus, the version, which does not contain the category of “working class,” has been inquired into in Japanese longitudinal stratification surveys, instead of the classic class identification commonly inquired into in Europe and the U.S. Another trait in Japan is the strong concern to temporal changes in the proportion of the response categories, specifically the increase of “the middle” strata. Along with the high growth of the

207 ⌮ㄽ࡜᪉ἲ economy before the mid-1970s, there was observed a gradual shift in status identification from “the lower” to “the middle.” About 75% of Japanese people identified themselves as belonging to “the middle” strata, which consists of “the upper middle” and “the lower middle.” The media reported the upgrading as the accomplishment of the Mass Middle Society, as they preferred to grasp the tide not from the objective status structure itself, but from the distribution of the people’s evaluation. This change was regarded as a symbolic phenomenon in the era of “Japan as Number One” (Vogel 1979). However, a general upgrading has not been observed since the mid-1970s. In consequence, this approximately normal distribution was maintained for forty years. Japanese researchers referred to the stagnancy as the saturation of the middle.5) Concerning the objective-subjective association, Naoi (1979) concluded that social status did not convincingly predict one’s status identification in the 1970s. Studies in the 1980s and 1990s also did not achieve a clear-cut explanation for the upgrading and saturation of the distribution (Hara 1990; Mamada 1990). This provisionally implied that there was actually no functional relationship between the proportion of “the middle” response and economic trends (Seiyama 1990). After the 1990s, subjective social status studies were deadlocked because of the insufficiency of explanation. It is worth noting that the reason for the decline in the study in Japan was diminishing predictability, though it was unchanging predictability in the U.S. Thereafter, Kikkawa (2000) re-examined the uncertainty. He applied OLS regression to the objective-subjective association in the data from 1975, 1985, and 1995 to identify a changing trend in the determinants. This could be understood as a delayed catch-up in methodology to American studies. He found several trends; first, in 1975, it was true that status identification was almost independent of objective social status. This was remarkable in comparison with succeeding survey years. He described the situation as people having lost sight of rational judgment for class identification because of high economic growth. Second, he observed that household income became an important determinant in 1985, at the beginning of “the bubble economy.” Third, by 1995, he found that a multiple-determinant structure, in which occupation, income, and education indicate significant effects, formed. Fourth, he pointed out that the R-squares increased in those twenty years. In summary, status identification had gradually changed its formation process and had become more predictable. Recent studies show that the determinant structure has been maintained and R-squares further increased in the 2000s and thereafter (Kanbayashi 2010; Kikkawa 2006, 2008; Sirahase 2010; Sudo 2009). Overall, Japan is almost the only society in which subjective social status remains a core issue of stratification studies. It is also notable that Japan and the U.S. are the only societies wherein we find comparative longitudinal surveys which cover the state of objective and subjective social status in these 30 years, namely the SSM and the GSS. Nevertheless, with respect to the temporal change, comparative studies are insufficient6) despite researcher interest.

2.3 Research Question It seems true that Japan has caught up with the U.S. with respect to the magnitudes and patterns of

208 Class Awareness in Japan and the U. S. determinants. However, the consequences are not simple. In the U.S., the determinant structure retains similar patterns for many decades. Therefore, researchers do not expect delicate transformations caused by current social or economic conditions. As a result, few examines the influences of the economic boom in the early 2000s, economic recession after the Lehman shock, or recent anti-social disparity movements on people’s status identification. In contrast, recent Japanese studies imply that the determinant structure can transform drastically along with changes in social environments, even in the post-industrial era. Today, it is said that Japan has shifted from a “generally middle class” society to a “generally unequal” society. It is also said that Japanese people’s appreciation of wealth and social position, in other words class consciousness, has been greatly transformed by rising and inequality. If these presumptions are true, what is happening in the determinant structure of subjective social status in the U.S. as well as Japan? In this paper, we examine the latest reality in Japan and compare it to U.S. reality. We confirm the trends that former Japanese studies have described in a more sophisticated manner. Our main concern is how largely subjective social status is influenced by objective social status in contemporary Japan, in which social disparity is becoming an important issue.

3 Data, Variables, and Statistical Models

3.1 Data The data sets in this study are derived from national surveys conducted in the 1980s, 1990s, and the year 2010. To be used as a reliable measure of objective social status, these surveys needed to contain information about respondents’ educational attainment, occupational status, and annual household income. They also needed to contain class identification, self-ranked social position, and satisfaction with standard of living. Japanese data comes from the SSM Survey of 1985 and 1995, and the 2010 Stratification and Social Psychology Interview Survey (SSP-I).7) The SSP-I 2010 is designed to have strict comparability with the SSM 1985 and the SSM 1995.8) It reflects the latest reality of social status in Japan. Because the SSM 1985 lacks the self-ranked social position of female respondents, the analysis for Japanese women is conducted using the SSM 1995 and the SSP-I 2010. Data for the U.S. was gathered from the GSS conducted from 1972 to 2010 (the 1972–2010 GSS cumulative data). It covers a long time period and contains information about respondent’s subjective and objective social statuses. We use data obtained in 1987, 2000, and 2010 because the GSS asked the same three questions concerning subjective social status only in the three periods. The GSS sampled Spanish and English speakers since 2006. However, the following analysis is restricted to data for English-speaking respondents to retain comparability with data gathered before 2006. We use WTSSALL and OVERSAMP weights to consider the sub-sampling of non-respondents, differential household size, and oversamples of African Americans (Smith et al. 2011).

209 ⌮ㄽ࡜᪉ἲ

3.2 Variables This paper examines the effects of objective social status on subjective social status. To avoid attenuation caused by measurement errors and extract unidimensional scales of subjective social status, we take a latent variable approach, that is, creating a latent variable of subjective social status (Kluegel et al. 1977). To measure the latent concept of subjective social status, we use three variables that indicate multidimensional aspects of subjective social status; class identification (four categories for the U.S. and five categories for Japan), self-ranked social position (10 categories for both countries), and satisfaction with standard of living (three categories for the U.S. and five categories for Japan). The questions used in the surveys and descriptive statistics for these three indicators are provided in Table 1. Because questions for class identification and satisfaction with standard of living slightly differ between the U.S. and Japan, we do not assume subjective social status to be identical between the two countries.9) Nevertheless, we presume that the latent variables are comparable between the U.S. and Japan because one of the three variables for subjective social status, self-ranked social position, is commonly implemented. And actually, this is the best remedy for the inevitable differences in survey designs. Age and squared age at the time of the survey are used as control variables. Educational attainment is coded into four categories; less than high school (not graduated from high school), high school, junior college, and university.10) Occupation of respondents is coded according to the EGP class schema (Erikson and Goldthorpe 1992), which consists of professionals and managers (I + II); routine non-manual workers (III); non-farm petty (IVab); technicians, supervisors, and skilled manual workers (V + VI); semi- and un-skilled manual workers, farmers and manual workers in agriculture (IVc + VIIab); and unemployed. Household income is transformed using the natural logarithm because it is positively skewed. Sex and survey year are used as grouping variables for the multigroup analysis. Descriptive statistics for independent variables are shown in Table 2. We restrict the data to respondents aged 25–60 and omit cases with one or more missing variables. For Japan, this yields 1,706 men in 1985, 740 men and 827 women in 1995, and 662 men and 803 women in 2010. For the U.S., this yields 411 men and 525 women in 1987, 311 men and 398 women in 2000, and 490 men and 612 women in 2010.

210 Class Awareness in Japan and the U. S.

Table 1. Descriptive Statistics for the Indicators of Subjective Social Status

Japan US Men Women Men Women 1985 1995 2010 1995 2010 1987 2000 2010 1987 2000 2010 Class Identification (1) Class Identification (4) Lower 5.7 5.0 5.0 2.8 3.6 Lower class 4.9 3.5 7.1 6.9 5.8 9.5 Upper-lower 18.0 16.2 19.2 13.3 14.3 Working class 48.2 50.2 49.6 49.9 46.5 51.6 Lower-middle 49.4 52.6 48.8 52.8 53.7 Middle class 45.5 42.8 41.4 39.4 44.2 36.6 Upper-middle 25.1 25.4 26.1 30.0 27.8 1.5 3.5 1.8 3.8 3.5 2.3 Upper 1.8 .8 .9 1.1 .6 Self-Ranked Social Position (2) Self-Ranked Social Position (5) 1…Top .2 .7 .2 .4 .5 1…Top 2.0 3.5 4.7 2.5 2.0 6.2 2 .4 1.1 .8 1.0 .5 2 2.9 3.2 3.5 3.1 2.3 3.1 3 3.9 4.6 4.7 6.8 3.1 3 15.8 14.2 15.5 8.8 7.8 10.6 4 10.7 13.1 18.4 12.7 12.0 4 18.0 13.5 14.5 14.3 17.6 15.4 5 25.1 26.0 23.7 32.0 31.5 5 28.7 32.2 34.1 30.3 33.4 39.2 6 23.5 24.9 20.9 22.7 23.9 6 13.4 16.1 12.5 16.8 15.1 10.8 7 19.9 18.2 17.7 15.8 17.2 7 8.0 11.3 9.0 12.0 11.3 9.2 8 10.3 7.0 10.0 6.5 8.6 8 6.1 3.9 4.1 6.9 6.3 3.3 9 4.1 2.3 2.4 1.2 1.9 9 2.4 1.3 1.0 2.5 2.3 1.1 10…Bottom 1.9 2.2 1.4 .9 .9 10…Bottom 2.7 1.0 1.2 3.1 2.0 1.1 Satisfaction with Standard of Living (3) Satisfaction with Standard of Living (6) Dissatisfied 3.1 4.9 3.6 2.7 3.0 Not satisfied at all 27.7 19.9 34.5 28.2 32.9 38.1 Moderately dissatisfied 10.4 14.6 7.7 10.3 7.1 More or less satisfied 48.2 47.0 47.4 49.3 44.2 44.3 Neither 25.1 23.2 19.0 19.4 15.4 Pretty well satisfied 24.1 33.1 18.2 22.5 22.9 17.7 Moderately satisfied 43.3 42.6 37.5 45.3 45.0 Satisfied 18.2 14.7 32.2 22.4 29.5 N 1,706 740 662 827 803 411 311 490 525 398 612 Note: Questions asked in the surveys are as follow; (1) Assuming we can divide present-day Japanese society into these five levels, which do you consider yourself to fall into? (2) Assuming we divide society into 10 levels according to this scale, which group do your consider yourself to fall into? (3) Are you satisfied or dissatisfied with life in general? (4) If you were asked to use one of four names for your social class, which would you say you belong in: the lower class, the working class, the middle class, or the upper class? (5) In our society there are groups which tend to be towards the top and those that are towards the bottom. Here we have a scale that runs from top to bottom. Where would you put yourself on this scale? (6) We are interested in how people are getting along financially these days. So far as you and your family are concerned, would you say that you are pretty well satisfied with your present financial situation, more or less satisfied, or not satisfied at all?

Table 2. Descriptive Statistics for Independent Variables Japan US Men Women Men Women 1985 1995 2010 1995 2010 1987 2000 2010 1987 2000 2010 Age Age (Mean) 42.4 44.2 44.9 43.6 44.0 39.6 41.4 43.4 39.6 41.1 42.2 (S.D.) 9.6 9.5 9.8 9.6 9.6 9.3 9.5 10.3 10.0 9.6 10.5 Education Less than high 29.2 18.1 7.4 18.1 4.0 16.8 10.3 11.0 17.0 9.8 8.2 High school 45.7 50.7 53.6 56.5 61.4 49.6 52.7 50.8 54.7 55.8 48.9 Junior college 3.0 2.0 4.2 16.8 18.7 4.4 9.7 7.4 6.7 8.5 10.1 University 22.2 29.2 34.7 8.6 15.9 29.2 27.3 30.8 21.7 25.9 32.8 Occupation I+II 28.3 33.2 33.7 9.6 15.6 41.6 41.5 38.0 40.6 42.0 44.4 III 12.7 10.3 11.2 26.6 25.4 6.1 5.8 9.6 28.6 24.4 25.7 IVab 16.2 15.5 11.5 5.4 7.9 6.1 4.2 9.2 4.4 6.0 4.4 V+VI 17.0 20.1 17.5 6.1 6.1 21.9 23.5 19.4 6.1 3.8 4.6 IVc+VIIab 22.7 17.4 19.3 19.0 14.9 22.9 23.5 20.0 15.6 19.6 16.3 Unemployment 3.2 3.4 6.8 33.4 30.1 1.5 1.6 3.9 4.8 4.3 4.6 Household income (1) Non-logged (Mean) 565.0 839.3 691.1 791.6 613.9 328.9 524.3 635.5 278.4 445.1 562.1 (S.D.) 352.6 640.6 520.0 476.5 372.8 173.0 307.6 433.8 177.5 317.2 426.9 Logged (Mean) 6.181 6.568 6.290 6.487 6.180 5.595 6.046 6.108 5.330 5.744 5.935 (S.D.) .632 .561 .880 .681 .870 .729 .744 1.013 .900 .984 1.053 N 1,706 740 662 827 803 411 311 490 525 398 612 Note: (1) Unit: ten thousand Japanese yen for Japan and one hundred dollars for the U.S.

211 ⌮ㄽ࡜᪉ἲ

3.3 Statistical Models First, we conduct the multigroup confirmatory factor analysis to measure latent subjective social status and examine whether the assumption of weak measurement variance that constrains factor loadings to be invariant across grouping variables is valid. Then, we use the multiple indicators and multiple causes (MIMIC) model to investigate the effect of observed objective social status on latent subjective social status. The equations of MIMIC model can be written as follows:

ܡൌૃߟ൅ߝ ߟൌߛᇱܠ൅ߞ

Here, y are indicators of the latent subjective social status ሺߟሻ, and ܠ are the set of variables for objective social status that predict ߟ (see also Figure 1). To observe whether the effects of objective social status on subjective social status change over time, we apply the multigroup analysis to the MIMIC model. The multigroup MIMIC model enables us to test changes in the effects of objective social status on subjective social status (ߛᇱ) between periods, while retaining the assumption of measurement invariance between sexes and periods (ૃ). We might be able to test the differences of the effects of objective social status on subjective social status between men and women as well as between periods. However, the distributions of education and occupation are different between sexes, specifically in Japan (as shown in Table 2). Therefore, we do not set equality constraints on the structural parameters of men and women, and concentrate on testing changes in the effects of objective social status between periods. We conduct the analyses for the U.S. and Japan separately, using the five-group MIMIC model for Japan (three groups for men and two groups for women) and the six-group MIMIC model for the U.S. (three groups for men and women, respectively) as shown in Figure 2. All analyses are conducted using Mplus ver. 6.2 (Muthén and Muthén 2010).

Objective Social Status x Latent Variable η Indicators y

ζ Age Class Identification ε1

Education Subjective Self-Ranked Social ε2 Social Status Position Occupation

Household Income Satisfaction with ε3 Standard of Living

Note: The indicators for subjective social status are not the same between the U.S. and Japan. Figure 1. The MIMIC Model

212 Class Awareness in Japan and the U. S.

1985→1995→2010 1987→2000→2010

ζ ζ Age Age

Education Education Subjective Subjective Men Social Status Social Status Occupation Occupation

Household Income Household Income

ζ ζ Age Age

Education Education Subjective Subjective Women Social Status Social Status Occupation Occupation

Household Income Household Income

1995→2010 1987→2000→2010 Japan US Figure 2. The Multigroup MIMIC Model (Grouping Variables: Sexes and Periods)

4 Results

4.1 Measurement Models for Subjective Social Status We construct the latent variable of subjective social status from three indicators; class identification, self-ranked social status, and satisfaction with standard of living. To compare the structures of the determinants of subjective social status across two or more groups, an equivalence of measurement is required. Thus, the assumption of weak measurement invariance that imposes equality constraints on factor loadings across sexes and periods is tested. For Japan, the results of the multigroup confirmatory factor analysis show that indices for the model fit11) are acceptable (ɖଶȀ†Ǥ ˆǤ ൌ ͵Ǥͺ͵͹ǡ  ൌ Ǥͻͻ͵ǡ  ൌ ǤͲͷͷǡ  ൌ ǤͲ͵͹), although the ɖଶ test does not indicate very good fit (ɖଶ ൌ ͵ͲǤ͸ͻ͸ǡ †Ǥ ˆǤ ൌ ͺǡ ݌ ൏ ǤͲͲͳ). Table 3 shows that the standardized factor loadings for both class identification and self-ranked social position are relatively high and similar, and the standardized factor loading of satisfaction with standard of living is the lowest, indicating that, in Japan, both class identification and self-ranked social position are central indicators of subjective social status. The assumption of weak measurement invariance for the U.S. is also not supported by the ɖଶ test of model fit (ɖଶ ൌ ͵͵ǤͶʹ͸ǡ †Ǥ ˆǤ ൌ ͺǡ ݌ ൏ ǤͲͲͳ). However, other indices of this model are acceptable (ɖଶȀ†Ǥ ˆǤ ൌ ͵Ǥʹͻ͸ǡ  ൌ Ǥͻ͹Ͷǡ  ൌ ǤͲ͹ͳǡ  ൌ ǤͲͶ͵). Table 4 indicates that the standardized factor loading for class identification is the highest, followed by that of self-ranked social position, and the standardized loading for satisfaction with standard of living is the lowest, meaning that the core indicator of subjective social status in the U.S. is class identification. The results of the multigroup confirmatory factor analyses suggest that, in both the U.S. and Japan, the principal indicators of subjective social status are class identification and/or self-ranked social position. Therefore, the latent variables of subjective social status covers the characteristics of classic class identification, with which previous research has been mainly concerned. Moreover, the latent variables of subjective social status are comparably measured in the U.S. and Japan, although very strict comparability is not assured. In the following analyses, we apply the assumption of weak measurement invariance.12)

213 ⌮ㄽ࡜᪉ἲ

Table 3. Standardized Factor Loadings on Subjective Social Status for Japan

1985 1995 2010 Men Est. S.E. p Est. S.E. p Est. S.E. p Class Identification .727 .018 .000 .878 .019 .000 .816 .021 .000 Self-Ranked Social Position .760 .018 .000 .860 .019 .000 .827 .022 .000 Satisfaction with Standard of Living .397 .016 .000 .417 .019 .000 .424 .021 .000 1985 1995 2010 Women Est. S.E. p Est. S.E. p Est. S.E. p Class identification .840 .019 .000 .785 .023 .000 Self-Ranked Social Position .839 .020 .000 .786 .021 .000 Satisfaction with Standard of Living .394 .018 .000 .396 .020 .000

Table 4. Standardized Factor Loadings on Subjective Social Status for the U.S.

1987 2000 2010 Men Est. S.E. p Est. S.E. p Est. S.E. p Class Identification .730 .036 .000 .644 .045 .000 .635 .038 .000 Self-Ranked Social Position .591 .036 .000 .566 .043 .000 .550 .034 .000 Satisfaction with Standard of Living .522 .032 .000 .459 .035 .000 .485 .032 .000 1987 2000 2010 Women Est. S.E. p Est. S.E. p Est. S.E. p Class Identification .673 .034 .000 .687 .036 .000 .629 .037 .000 Self-Ranked Social Position .589 .034 .000 .643 .037 .000 .549 .029 .000 Satisfaction with Standard of Living .536 .031 .000 .522 .033 .000 .508 .033 .000

4.2 Japan Table 5 shows the results of the multigroup MIMIC model for Japan with all free structural parameters between grouping variables, sexes, and periods. This multigroup MIMIC model fits the data well (ɖଶ ൌ ʹʹ͹Ǥ͹͹Ͷǡ †Ǥ ˆǤ ൌ ͳͳͺǡ ɖଶȀ†Ǥ ˆǤ ൌ ʹǤ͵ͷͶǡ  ൌ Ǥͻ͸ʹǡ  ൌ ǤͲ͵ͺǡ  ൌ ǤͲʹͲ). To investigate changes in parameters between periods, we constrain two or three parameters so that they are the same and test whether these equality constraints significantly worsen the statistical fit of the model relative to the model with all free structural parameters (the base model).13) The results for Japan are shown in Table 6. The ߂ɖଶ and ߂†Ǥ ˆǤ respectively indicates increases in the ɖଶ and †Ǥ ˆǤ after implementing the equality constraints relative to the base model (ɖଶ ൌ ʹ͹͹Ǥ͹͹Ͷ, †Ǥ ˆǤ ൌ ͳͳͺ). When p-value obtained from the chi-square test of ߂ɖଶ with ߂†Ǥ ˆǤ is less than .05, the assumption of the same effects between periods is judged to be invalid. In this situation, there are certain changes in the effects of variables between the periods. For example, the ߂ɖଶ for the effect of not graduating from high school (less than high) for men in the 1985 = 1995 = 2010 column is 10.402 with 2 d.f., and the resultant p-value is .006. This indicates that the assumption of the same effects of not graduating from high school between 1985, 1995, and 2010 is rejected, indicating that the effects significantly changed over time. The subsequent columns for less than high school show that p-values are less than .05 in the 1985 = 1995 column and in the 1995 = 2010 column, but more than .05 in the 1985 = 2010 column. Thus, there exist significant changes between 1985 and 1995 and between 1995 and 2010. Table 5 indicates that the negative effect of not graduating from high school found in 1985 disappears in 1995, and then reappears in 2010. Table 6 indicates that the total effects of all variables changed between 1985, 1995, and 2010

214 Class Awareness in Japan and the U. S.

(education: p = .007; occupation: p = .003; and household income: p < .001). However, the second column of Table 6 (1985 = 1995) indicates that the effects of all variables are stable between 1985 and 1995 (p > .05), except for the effect of not graduating from high school (p = .025). Therefore, we can conclude that the effect of objective social status on subjective social status remains stable between 1985 and 1995. On the other hand, the assumptions that the total effects of education, occupation, and household income are the same between 1985 and 2010, and between 1995 and 2010 are rejected, respectively. This is because there are significant changes in the effect of objective social status in 2010. Returning to Table 5, the effects of education and occupation increase between 1985 and 2010 and between 1995 and 2010. Compared to those who graduated from high school, those who did not graduate from high school evaluated their subjective social status as lower, and those who had attended university tend to view themselves on higher levels of social status in 2010. Professionals and managers (I + II) evaluate their subjective social status higher in 2010 than in 1985 and 1995. However, the effect of household income shows a different trend, weakening between 1995 and 2010.14) The resultant R-squares are .187, .173, and .352 in 1985, 1995, and 2010, respectively, which means that a set of variables on objective social status explains more variance of subjective social status in 2010 than before, despite the weakening effect of household income. For Japanese women, we can see a similar pattern as men; the effect of objective social status differs between 1995 and 2010. Although change in the total effect of education is not significant between 1995 and 2010 (p = .100), change in the effect of junior college is significant (p = .042) and change in the effect of university is marginally significant (p = .062). As Table 5 shows, the effect of education increases between the two periods. The total effect of occupation does not indicate any significant change (p = .191), but the effects of being a less-skilled manual worker (IVc + VIIab) and unemployed change between 1995 and 2010 (p = .009 and p = .030, respectively). Compared to female professionals and managers (I + II), less-skilled female manual workers (IVc + VIIab) do not evaluate their subjective social status lower in 1995, but do so in 2010, while unemployed females, most of which are homemakers, tend to view themselves on higher levels of social status in 1995, but not in 2010. The effect of occupation for Japanese women is low in 1995, but appears in 2010, suggesting that, for Japanese women, their own occupation has become more important in considering their own social status. The effect of household income for Japanese women shows a similar trend for Japanese men; the effect of household income weakens between 1995 and 2010.15) Although the effect of household income decreases in 2010, the effects of education and occupation increase; therefore, the R-squares show a tendency to increase from .187 in 1995 to .226 in 2010. These results suggest that, for Japanese women, the multiple-determinant structure of subjective social status appears between 1995 and 2010. As a whole, the effects of education and occupation on subjective social status have increased over time both for men and women, while the effect of household income has decreased even though it still has significance. Previous studies investigated the determinants of class identification (Kikkawa 2000, 2006) and self-ranked social position (Sirahase 2010) in a rather descriptive way and found

215 ⌮ㄽ࡜᪉ἲ that the multiple-determinants structure has been formed and the effect of objective social status has increased in Japan. The result of this paper confirms previous research in a more rigorous way by using the multigroup MIMIC model.

Table 5. Results of the Multigroup MIMIC Model for Japan 1985 1995 2010 Men Est. S.E. p Est. S.E. p Est. S.E. p Standardized Factor loadings Class Identification .727 .015 .000 .885 .016 .000 .815 .017 .000 Self-Ranked Social Position .760 .016 .000 .852 .016 .000 .828 .018 .000 Satisfaction with Standard of Living .399 .016 .000 .419 .019 .000 .425 .021 .000 Unstandardized regression coefficients Age Age / 10 –.064 .163 .694 –.165 .246 .503 –.329 .256 .199 (Age / 10) 2 .009 .019 .633 .016 .028 .575 .041 .029 .152 Education Less than high –.122 .044 .005 .077 .077 .320 –.326 .101 .001 High school (fixed at 0) .000 ―― .000 ―― .000 ―― Junior college .212 .100 .034 .135 .188 .475 –.012 .128 .925 University .119 .047 .011 .186 .066 .005 .263 .060 .000 Occupation I+II (fixed at 0) .000 ―― .000 ―― .000 ―― III –.163 .058 .005 –.198 .097 .040 –.322 .089 .000 IVab –.073 .055 .187 –.120 .084 .153 –.260 .090 .004 V+VI –.106 .057 .061 –.014 .082 .866 –.388 .081 .000 IVc+VIIab –.141 .054 .009 –.287 .087 .001 –.568 .080 .000 Unemployment –.340 .108 .002 –.378 .156 .016 –.653 .123 .000 Household Income Logged-household income .304 .030 .000 .376 .054 .000 .159 .033 .000 R-square Subjective Social Status .187 .021 .000 .173 .028 .000 .352 .035 .000 N 1706 740 662 1985 1995 2010 Women Est. S.E. p Est. S.E. p Est. S.E. p Standardized Factor loadings Class Identification .852 .016 .000 .798 .019 .000 Self-Ranked Social Position .826 .016 .000 .774 .018 .000 Satisfaction with Standard of Living .395 .018 .000 .399 .020 .000 Unstandardized regression coefficients Age Age / 10 –.009 .218 .967 .090 .226 .691 (Age / 10) 2 .004 .025 .886 –.004 .026 .881 Education Less than high –.315 .068 .000 –.288 .119 .015 High school (fixed at 0) .000 ―― .000 ―― Junior college –.016 .064 .808 .162 .059 .006 University .071 .085 .404 .271 .065 .000 Occupation I+II (fixed at 0) .000 ―― .000 ―― III .107 .085 .209 –.110 .071 .125 IVab .009 .122 .945 –.273 .099 .006 V+VI .157 .121 .195 –.089 .107 .403 IVc+VIIab .007 .092 .937 –.320 .084 .000 Unemployment .189 .084 .024 –.048 .070 .497 Household Income Logged-household income .325 .035 .000 .209 .027 .000 R-square Subjective Social Status .187 .027 .000 .226 .031 .000 N 827 803 2 Note: χ = 277.774, d.f. = 118, p < 0.001. CFI = .962, RMSEA = .038, SRMR = .020.

216 Class Awareness in Japan and the U. S.

Table 6. Increase in the χ2 Statistics (Δχ2) after Equality Constraints on Parameters of the Base Model for Japan 1985 = 1995 = 2010 1985 = 1995 1985 = 2010 1995 = 2010 Men Δχ2 Δd.f. p Δχ2 Δd.f. p Δχ2 Δd.f. p Δχ2 Δd.f. p Education Less than high 10.402 2 .006 5.027 1 .025 3.396 1 .065 9.937 1 .002 High school (fixed at 0) Junior college 1.906 2 .386 .131 1 .717 1.905 1 .168 .415 1 .519 University 3.598 2 .165 .687 1 .407 3.569 1 .059 .748 1 .387 Total 17.544 6 .007 5.696 3 .127 10.530 3 .015 11.970 3 .007 Occupation I+II (fixed at 0) III 2.248 2 .325 .097 1 .755 2.236 1 .135 .885 1 .347 IVab 3.130 2 .209 .216 1 .642 3.127 1 .077 1.295 1 .255 V+VI 11.984 2 .002 .868 1 .352 8.101 1 .004 10.578 1 .001 IVc+VIIab 19.525 2 .000 2.014 1 .156 19.509 1 .000 5.645 1 .018 Unemployment 3.973 2 .137 .039 1 .843 3.656 1 .056 1.920 1 .166 Total 26.190 10 .003 5.761 5 .330 20.333 5 .001 11.855 5 .037 Household Income Logged household income 16.035 2 .000 1.357 1 .244 10.671 1 .001 11.706 1 .001 1995 = 2010 Women Δχ2 Δd.f. p Education Less than high .038 1 .845 High school (fixed at 0) Junior college 4.130 1 .042 University 3.495 1 .062 Total 6.244 3 .100 Occupation I+II (fixed at 0) III 3.795 1 .051 IVab 3.185 1 .074 V+VI 2.324 1 .127 IVc+VIIab 6.864 1 .009 Unemployment 4.694 1 .030 Total 7.419 5 .191 Household Income Logged household income 6.855 1 .009 2 Note: Base model: χ = 277.774, d.f. = 118.

4.3 The U.S. Table 7 shows the results of the multigroup MIMIC model for the U.S. with all free structural parameters between grouping variables. This multigroup MIMIC model represents a good fit (ɖଶ ൌ ʹͳͶǤ͹ͻ͹ǡ †Ǥ ˆǤ ൌ ͳͶʹǡ ɖଶȀ†Ǥ ˆǤ ൌ ͳǤͷͳ͵ǡ  ൌ Ǥͻͷ͵ǡ  ൌ ǤͲ͵͵ǡ  ൌ ǤͲʹͻ). The first column of Table 8 (1987 = 2000 = 2010) illustrates whether the effect of the variable changes over time in the U.S. We can see no significant change (p > .05) in the effect of objective social status on subjective social status except for that of household income both for men and women.16) As shown in Table 7, education and household income affect how individuals view their own social status both for men and women; individuals who complete university and those who have higher household income see their own social status as higher. The R-squares are high and stable. The objective social status along with age explains around 60% of variances of subjective social status for men and around 50% for women at each period. In general, the objective-subjective association is stronger and more stable in the U.S. than in Japan. Although the effect of occupation is weak as shown by Hout (2008), the analysis of the recent GSS data using the multigroup MIMIC model

217 ⌮ㄽ࡜᪉ἲ suggests that the multiple determinant structure of subjective social status found by Hodge and Treiman (1968) and Jackman and Jackman (1973, 1983) is largely maintained. The multiple-determinant structure of subjective social status of U.S. women is basically as stable as that of U.S. men. We can see, however, indicative changes for U.S. women: as shown in Tables 7 and 8, the effect of education increases and that of household income decreases between 2000 and 2010.17)

Table 7. Results of the Multigroup MIMIC Model for the U.S. 1987 2000 2010 Men Est. S.E. p Est. S.E. p Est. S.E. p Standardized Factor loadings Class Identification .719 .036 .000 .670 .055 .000 .689 .041 .000 Self-Ranked Social Position .545 .037 .000 .528 .048 .000 .520 .033 .000 Satisfaction with Standard of Living .516 .035 .000 .469 .037 .000 .515 .032 .000 Unstandardized regression coefficients Age Age / 10 −.144 .225 .524 .076 .261 .771 −.244 .190 .200 (Age / 10) 2 .021 .027 .420 −.004 .030 .890 .024 .023 .289 Education Less than high −.051 .075 .501 .046 .096 .632 −.142 .080 .076 High school (fixed at 0) .000 ―― .000 ―― .000 ―― Junior college −.066 .122 .586 .097 .100 .333 .133 .087 .125 University .274 .064 .000 .446 .073 .000 .329 .058 .000 Occupation I+II (fixed at 0) .000 ―― .000 ―― .000 ―― III −.157 .106 .136 −.140 .115 .223 .004 .084 .958 IVab −.106 .081 .189 −.026 .132 .844 .054 .087 .534 V+VI −.069 .081 .394 .013 .079 .874 −.100 .065 .123 IVc+VIIab −.167 .076 .029 −.053 .074 .467 −.086 .075 .250 Unemployment .174 .113 .122 .233 .292 .425 −.036 .124 .771 Household Income Logged-household income .329 .043 .000 .299 .054 .000 .204 .028 .000 R-square Subjective Social Status .574 .074 .000 .639 .085 .000 .571 .066 .000 N 411 311 490 1987 2000 2010 Women Est. S.E. p Est. S.E. p Est. S.E. p Standardized Factor loadings Class Identification .700 .041 .000 .679 .033 .000 .673 .035 .000 Self-Ranked Social Position .601 .037 .000 .615 .038 .000 .528 .027 .000 Satisfaction with Standard of Living .549 .033 .000 .531 .037 .000 .527 .036 .000 Unstandardized regression coefficients Age Age / 10 −.152 .235 .517 −.400 .219 .068 −.053 .183 .771 (Age / 10) 2 .024 .027 .383 .052 .026 .042 .006 .021 .795 Education Less than high −.065 .077 .398 .101 .090 .262 −.132 .091 .145 High school (fixed at 0) .000 ―― .000 ―― .000 ―― Junior college .049 .080 .545 .070 .083 .394 .081 .068 .230 University .161 .069 .020 .160 .068 .018 .320 .051 .000 Occupation I+II (fixed at 0) .000 ―― .000 ―― .000 ―― III −.110 .068 .106 −.084 .065 .191 −.156 .054 .004 IVab −.155 .107 .147 .087 .109 .424 .147 .096 .125 V+VI −.235 .107 .027 −.157 .095 .098 −.007 .120 .956 IVc+VIIab −.178 .091 .050 −.115 .081 .157 −.057 .070 .417 Unemployment −.085 .128 .507 .341 .170 .045 .060 .115 .605 Household Income Logged-household income .306 .041 .000 .309 .034 .000 .188 .029 .000 R-square Subjective Social Status .455 .067 .000 .518 .070 .000 .497 .064 .000 N 525 398 612 2 Note: χ = 214.797, d.f. = 142, p = 0.0001. CFI = .953, RMSEA = .033, SRMR = .029.

218 Class Awareness in Japan and the U. S.

Table 8. Increase in the χ2 Statistics (Δχ2) after Equality Constraints on Parameters of the Base Model for the U.S. 1987 = 2000 = 2010 1987 = 2000 1987 = 2010 2000 = 2010 Men Δχ2 Δd.f. p Δχ2 Δd.f. p Δχ2 Δd.f. p Δχ2 Δd.f. p Education Less than high 2.449 2 .294 .602 1 .438 .784 1 .376 2.623 1 .105 High school (fixed at 0) Junior college 1.852 2 .396 1.092 1 .296 1.889 1 .169 .089 1 .765 University 2.999 2 .223 1.852 1 .396 .360 1 .549 1.495 1 .221 Total 7.007 6 .320 3.592 3 .309 3.079 3 .380 4.255 3 .235 Occupation I+II (fixed at 0) III 1.871 2 .392 .026 1 .871 1.415 1 .234 1.060 1 .303 IVab 1.716 2 .424 .273 1 .602 1.877 1 .171 .284 1 .594 V+VI 1.069 2 .586 .462 1 .497 .098 1 .754 1.200 1 .273 IVc+VIIab 1.064 2 .587 1.042 1 .307 .485 1 .486 .075 1 .784 Unemployment 1.541 2 .463 .060 1 .806 1.803 1 .179 .753 1 .386 Total 8.336 10 .596 1.155 5 .949 5.957 5 .310 4.429 5 .489 Household Income Logged household income 6.423 2 .040 .120 1 .729 5.708 1 .017 2.454 1 .117 1987 = 2000 = 2010 1987 = 2000 1987 = 2010 2000 = 2010 Women Δχ2 Δd.f. p Δχ2 Δd.f. p Δχ2 Δd.f. p Δχ2 Δd.f. p Education Less than high 3.790 2 .150 2.528 1 .112 .321 1 .571 4.743 1 .029 High school (fixed at 0) Junior college .035 2 .983 .018 1 .894 .049 1 .824 −.062 1 䇷 University 4.856 2 .088 −.129 1 ― 3.244 1 .072 3.541 1 .060 Total 10.350 6 .111 2.314 3 .510 4.248 3 .236 8.899 3 .031 Occupation I+II (fixed at 0) III .707 2 .702 .032 1 .858 .171 1 .680 .733 1 .392 IVab 4.812 2 .090 2.702 1 .100 5.054 1 .025 .113 1 .737 V+VI 2.746 2 .253 .178 1 .673 2.140 1 .144 .957 1 .328 IVc+VIIab 1.070 2 .586 .194 1 .660 1.058 1 .304 .200 1 .655 Unemployment 4.889 2 .087 5.973 1 .015 .739 1 .390 1.940 1 .164 Total 14.206 10 .164 7.725 5 .172 8.387 5 .136 5.686 5 .338 Household Income Logged household income 9.313 2 .010 −.111 1 ― 4.970 1 .026 8.243 1 .004 2 2 Note: Base model: χ = 214.797, d.f. = 142. Δχ is the Satorra-Bentler scaled chi-square difference test TRd.

5 Conclusion and Discussion

In this paper, we apply a multigroup MIMIC model to compare the temporal change of subjective social status in Japan and the U.S. The results show different consequences in the two countries. Previous Japanese studies reported that the objective-subjective association got stronger in these three decades in Japan. Our sophisticated elaboration confirms this trend: At the beginning, when Japan enjoyed economic prosperity, income was the principal determinant of subjective social status and the predictive power of the causal model was moderate. Then, education gradually became influential, and, recently, the effect of occupation increased. These additive sequences of determinants result in an improvement of association between objective and subjective social status. The results indicate expanding class awareness in contemporary Japan. It is notable that the expansion occurred after Japan reached the post-industrial phase, in which its socio-economic system did not strikingly change the composition. On the other hand, it was believed that a multiple-determinant structure might be time invariant in the U.S. Through our study, we reconfirm

219 ⌮ㄽ࡜᪉ἲ the consistency of the basic pattern of association. The Japanese consequence suggests that the determinant structure may change in a short period; the state of subjective social status may sensitively reflect macro social environments. It must be a matter of course that people not only become sensitive to their objective status but also have a broader outlook under the situation of destabilization and fluidity, indeed, the latest determinant structure in Japan seems to converge with the long-lasting state in the U.S.18) I is also reasonable that the symbolic value of one’s educational attainment becomes meaningful, because people perceive more possibilities to refer it in such situations. If this is true, a similar expansion of class awareness might occur in the U.S., reflecting recent economic recession, or anti-disparity movements such as the Occupy Movement in 2011. In fact, we see a recent symptom that the effect of education increases, while that of household income decreases for women in the U.S. between 2000 and 2010.19) Meanwhile, studies exploring the association between health and subjective social status have started to accumulate (Adler et al. 2000; Demakakos et al. 2008; Oprerario et al. 2004; Singh-Manoux et al. 2005; Wolff et al. 2010). This reflects the fact that psychological processes are more critical than the direct influence of socio-economic status when we explain health matters. Thus, the state of subjective social status needs to be constantly observed by sociologists.

Notes 1) Jackman and Jackman say; “Note the distinction between class identification on the one hand and class consciousness on the other. The former deals with perceived self-location in the socio-economic structure; while the latter deals with a series of issues arising from this perceived self-location, such as the intensity of the perception, and its effects on political orientation” (Jackman and Jackman 1973: 569-70). We may count certain types of public opinions, emotion such as affection and antagonism, and cultural traits, too. Among these broad aspects, political attitude has particularly played an important role in examining the relevance of social class. However, it is out of the scope of this paper. 2) Exploring other post-industrial societies like England, Taiwan, Korea, Germany, and Norway, the determinants do not widely vary (Arita 2009; Ishida 2010; Knudsen 1988; Marsh 2002). It is noteworthy Andersen and Curtis (2012) introduce multilevel logit analysis to examine the impact of economic conditions in 44 countries including developing ones. 3) As measurement items, they use five variables asking respondents to show which social class they felt their life, in general, occupation, lifestyle, income, and influence are most of like. 4) The SSM Survey has been conducted every ten years since 1955 by Japanese researchers of stratification study. There are six surveys that have been pursued. They feature a series of questions including respondents’ current social status and family attributions, social backgrounds, and attitudes and opinions related to social stratification. The datasets are the most valuable source of stratificational reality in contemporary Japan (Hara 2011; Kikkawa 1998; Sato et al. 2011). 5) From another viewpoint, it can be understood that the distributions of status identification in those forty years are normal, although those in the 1950s and 1960s were irregularly distorted (Mori 2008). 6) With respect to cross-sectional cultural comparison, Arita (2009, 2011) compares the determinant structure among societies in East Asia. Ishida (2010) compares Germany, the U.S., and Japan in a similar manner. According to these researchers, Japan is commonplace in terms of the magnitudes of the

220 Class Awareness in Japan and the U. S.

determinant structure, although their results vary in the size of effects. In terms of the relationship between socio-economic status and psychological functioning, in general, we should note that there are several important studies comparing the U.S. and Japan (Kohn et al. 1990; Lincoln and Kalleberg 1990; Wright 1997). 7) Data were collected by the Institute of Statistical Mathematics in collaboration with Osaka University in 2010 and partially in 2011. The sample size in the design was 3,500. The valid cases are 1,763 (50.4%). 8) Although the sixth SSM survey was successfully conducted in 2005, most psychological variables were set aside from the interview questionnaire. Class identification and self-ranked social position were inquired via self-administration, and thus we cannot assure the comparability. 9) Because certain Japanese respondents in 1985 graduated from the old educational system, we assign the corresponding levels in the new educational system to those respondents (e.g., coding high school in the old system into junior college in the new system). 10) As there are no accurately compatible variables in the datasets from both societies, we use five-point-graded general life satisfaction for Japan and three-point-graded financial satisfaction for the U.S. 11) The ߯ଶ statistic value divided by the degrees of freedom is used to evaluate the model fit with a ɖଶȀ †Ǥ ˆǤof less than two indicating that the model fit is very good. CFI: The comparative fit index is used with a CFI of above .900 being considered desirable (Bentler 1990; Hu and Bentler 1995). RMSEA: The root mean square error of approximation is also used. A RMSEA of less than .050 is considered as being indicative of a close fit of the model in relation to the degrees of freedom (Browne and Cudeck 1993). A SRMR (Standardized Root Mean Square Residual) of less than .050 generally indicates an adequate fit. 12) The choice of measurement models (measurement invariance, partial measurement invariance, or configural invariance) made little difference in the results of the analysis and our conclusion. 13) Another method to compare the regression coefficients of two different groups (݃ͳ and ݃ʹ) is to conduct ଶ ଶ .Ǥ ݁Ǥ௚ଶ, with †Ǥ ˆǤ ൌ ܰ௚ଵ ൅ܰ௚ଶ െͶݏ Ǥ ݁Ǥ௚ଵ൅ݏa t-test of ݐൌሺܾ௚ଵ െܾ௚ଶሻȀට 14) Standardized coefficients are .307, .296, and .203 in 1985, 1995, 2010, respectively. 15) Standardized coefficients are .346 in 1995 and .301 in 2010. 16) Only the effect of household income decreased from 1987 to 2010 significantly (߂߯ଶ ൌͷǤ͹Ͳͺǡ†ǤˆǤൌ ͳǡ ݌ ൌ ǤͲͳ͹). Standardized coefficients of household income are .507 in 1987, .497 in 2000, and .475 in 2010. 17) Standardized coefficients of household income for the U.S. women are .517, .626, and .427 in 1987, 2000, and 2010, respectively. 18) Using the World Value Survey data from 44 countries, Andersen and Curtis (2012) demonstrate the relationship between household income and class identity tends to be strongest in countries with a high level of income inequality. Although, Japan is not a highly unequal society, the recent expansion of economic disparity may cause the recent trend of class awareness in Japan. 19) Our result shows the effect of household income slightly decreases in both countries in the last decade, though it remains significant. As for the symptom, we need to observe the future consequences.

References Adler, Nancy E., Elissa S. Epel, Grace Castellazzo, and Jeannette R. Ickovics. 2000. “Relationship of Subjective and Objective Social Status with Psychological and Physiological Functioning: Preliminary Data in Healthy, White Women.” Health Psychology 19(6):586–592. Andersen, Robert, and Josh Curtis. 2012 “The Polarizing Effect of Economic Inequality on Class Identification: Evidence from 44 Countries.” Research in Social Stratification and Social Mobility 30:129–41.

221 ⌮ㄽ࡜᪉ἲ

Arita, Shin. 2009. “Hikaku wo Tsujite Miru Higashi Asia no Syakai Kaiso Kozo: Shoku-gyo ga Motarasu Hoshu Kakusa to Syakaiteki Fubyodo” (A Comparative Study of Social Stratification in East Asian Societies: The Reward Differences between Occupation and Social Inequality). Japanese Sociological Review 59(4):663–81. Arita, Shin. 2011. “Higashi Asia no Syakai Kaiso Kozo Hikaku.” [A Comparative Analysis of Social Stratification Structures in East Asia] Pp. 273–87 in Gendai no Kaiso Syakai 2: Kaiso to Idou no Kozo (Contemporary Stratified Society volume 2: Structures of Stratification and Mobility), edited by Hiroshi Ishida, Hiroyuki Kondo, and Keiko Nakao eds. Tokyo: University of Tokyo Press. Bauman, Zygmunt. 1982. Memories of Class. London: Routledge. Beck, Ulrich. 1992. Risk Society: Toward a New Modernity. London: Sage. Bentler, Peter M. 1990. “Comparative Fit Indices in Structural Models.” Psychological Bulletin 107:238–46. Browne, Michael W., and Robert Cudeck. 1993. “Alternative Ways of Assessing Model Fit.” Pp. 136–62 in Testing Structural Equation Models, edited by Kenneth. W. Bollen, and J. Scott Long. Newbury Park, California: Sage. Cannon, Lynn Weber 1984. “Trends in Class Identification among Black Americans from 1952 to 1978.” Social Science Quarterly 65(1):112–26. Centers, Richard. 1949. The Psychology of Social Classes: A Study of Class Consciousness. Princeton: Princeton University Press. Demakakos, Panayotes, James Nazroo, Elizabeth Breeze, and Michael Marmot. 2008. “Socioeconomic Status and Health: The Role of Subjective Social Status.” Social Science & Medicine 67(2):330–40. Erikson, Robert and John H. Goldthorpe. 1992. The Constant Flux: A Study of Class Mobility in Industrial Societies. Oxford: Oxford University Press. Gross, Neal 1953. “Social Class Identification in the Urban Community.” American Sociological Review 18:398–404. Hara, Junsuke. 1990. “Joron” (Preface.) Pp. 1–21 in Gendai Nihon no Kaiso Kozo 2: Kaiso Ishiki no Dotai (Social Stratification in Contemporary Japan Vol. 2 Dynamics in Class Attitudes), edited by Junsuke Hara. Tokyo: University of Tokyo Press. Hara, Junsuke. 2011. “An Overview of Social Stratification and Inequality Study in Japan: Towards a ‘Mature’ Society Perspective.” Asian Journal of Social Science 39(1):9–29. Hodge, Robert W. and Donald J. Treiman. 1968. “Class Identification in the United States.” American Journal of Sociology 73(5):535–47. Hout, Michael. 2008. “How Class Works: Objective and Subjective Aspects of Class Since the 1970s.” Pp. 25– 89 in Social Class: How Does it Work? edited by Annette Lareau and Dalton Conley. New York: Russel Sage Foundation. Hu, Li-Tze, and Peter M. Bentler. 1995. “Evaluating Model Fit.” Pp. 76–99 in Structural Equation Modeling: Concepts, Issues, and Applications, edited by Rick H. Hoyle. Thousand Oaks: Sage. Ishida, Hiroshi. 2010. “Does Class Matter in Japan? Demographics of Class Structure and Class Mobility from a Comparative Perspectives.” Pp. 33–56 in Social Class in Contemporary Japan: Structure, Sorting and Strategies, edited by Hiroshi Ishida and David H. Slater. Abingdon: Routledge. Jackman, Mary R. and Robert W. Jackman. 1973. “An Interpretation of the Relation between Objective and Subjective Social Status” American Sociological Review 38(5):569–82. Jackman, Mary R. and Robert W. Jackman. 1983. Class Awareness in the United States. Berkley: University of California Press. Kanbayashi, Hiroshi. 2010. “Chu-Ishiki no Houwa to Senzai suru Henka.” (Saturation and Latent Changes of the “Middle” Status Identification: Note on Status Identification in Postwar Japan (2)) Tohoku Gakuin Daigaku Kyouyou Gakubu Ronshu 157:1–24.

222 Class Awareness in Japan and the U. S.

Kikkawa, Toru. 1998. “Guest Editor’s introduction: Trends in Social Stratification in Contemporary Japan.” International Journal of Sociology 28(1):3–10. Kikkawa, Toru. 2000. “Changes in the Determinant of Class Identification in Japan.” International Journal of Sociology 30(2):34–51. Kikkawa, Toru. 2006. Gakureki to Kakusa Fubyodo (Education and Social Inequality: Contemporary Educational Credentialism in Japan). Tokyo: University of Tokyo Press. Kluegel, James R., Royce Singleton, Jr., and Charles E. Starnes. 1977. “Subjective Class Identification: A Multiple Indicator Approach.” American Sociological Review 42(4):599–611. Knudsen, Knud. 1988. “Class Identification in Norway: Explanatory Factors and Life-cycle Differences.” Acta Sociologica 31(1):69–79. Kohn, Melvin L., Atsushi Naoi, Carrie Schoenbach, Carmi Schooler, and Kazimierz Slomczynski. 1990. “Position in the Class Structure and Psychological Functioning in the United States, Japan, and Poland.” American Journal of Sociology 95:964–1008. Lincoln, James R. and Arne L. Kalleberg. 1990. Culture, Control, and Commitment: A Study of Work Organization and Work Attitudes in the United States and Japan. New York: Cambridge University Press. Macy, Michael W. 1989. “Classes of Positions or Classes of Persons: A Test of the ‘Empty Places’ Hypothesis.” National Journal of Sociology 3(2):159–98. Mamada, Takao. 1990. “Kaiso Kizokuishiki” (Class Identification.) Pp. 23–45 in Gendai Nihon no Kaiso Kozo 2: Kaiso Ishiki no Dotai (Social Stratification in Contemporary Japan Vol. 2. Dynamics in Class Attitudes), edited by Junsuke Hara. Tokyo: University of Tokyo Press. Marsh, Robert M. 2002. “Social Class Identification and Class Interest in Taiwan.” Comparative Sociology 1(1):17–41. Marx, Karl. [1859] 1977. Economic Manuscripts: A Contribution to the Critique of Political Economy. Moscow: Progress Publishers. Mori, Naoto. 2008. “Sou-Churyu no Shiso towa Nandattanoka: Chu Ishiki no Genten wo Saguru” (What was the thought of ‘All Middle Class Identification’) Pp. 233-270 in Shiso Chizu 2. Tokyo: NHK Books. Naoi, Michiko. 1979. “Shakai Kaiso to Kaikyu Ishiki” (Social Status and Class Consciousness.) Pp. 365–88 in Nihon no Kaiso Kozo (Social Stratification in Japan), edited by Ken’ichi Tominaga. Tokyo: University of Tokyo Press. Muthén, Linda K. and Bengt O. Muthén. 2010. Mplus User’s Guide Sixth Edition. Los Angeles: Muthén & Muthén. Operario, Don, Nancy E. Adler, and David R. Williams. 2004. “Subjective Social Status: Reliability and Predictive utility for Global Health.” Psychology & Health 19(2):237–246. Sato, Yoshimichi, Fumiaki Ojima, Hiroshi Ishida, Hiroyuki Kondo, Keiko Nakao, Yuriko Saito, and Kazuto Misumi, eds. 2011. Gendai no Kaiso Syakai Zen 3 Kan (Contemporary Stratified Society 3 volumes). Tokyo: University of Tokyo Press. Seiyama, Kazuo. 1990. “Chu Ishiki no Imi” (Middle Class Identification in Post-War Japan). Sociological Theory and Methods 5(2):51–71. Shirahase, Sawako. 2010. “Japan as a Stratified Society: With a Focus on Class Identification.” Social Science Japan Journal 13(1):31–52. Singh-Manoux Archana, Michael G. Marmot, and Nancy E. Adler. 2005. “Does Subjective Social Status Predict Health and Change in Health Status Better than Objective Status?” Psychosomatic Medicine 67(6):855– 61. Smith, Tom W, Peter Marsden, Michael Hout, and Jibum Kim. 2011. General Social Surveys, 1972-2010 [machine-readable data file]. /Principal Investigator, Tom W. Smith; Co-Principal Investigator, Peter V.

223 ⌮ㄽ࡜᪉ἲ

Marsden; Co-Principal Investigator, Michael Hout; Sponsored by National Science Foundation. --NORC ed.-- Chicago: National Opinion Research Center [producer]; Storrs, CT: The Roper Center for Public Opinion Research, University of Connecticut [distributor], 2011. Sudo, Naoki. 2009. Kaiso Ishiki no Dainamikusu: Naze, Soreha Genjitsu kara Zureru Noka (The Dynamics of Class Identification: Why does not it correspond to the real world?). Tokyo: Keiso Shobo. Vogel, Ezra F. 1979. Japan as Number One: Lessons for America. Cambridge: Harvard University Press. Wolff, Lisa S., S. V. Subramanian, Dolores Acevedo-Garcia, Deanne Weber, and Ichiro Kawachi. 2010. “Compared to Whom? Subjective Social Status, Self-Rated Health, and Referent Group Sensitivity in a Diverse US Sample.” Social Science & Medicine 70(12):2019–28. Wright, Eric Olin. 1997. Class Counts: Comparative Studies in Class Analysis. Cambridge: Cambridge University Press. Yamaguchi, Kazuo and Yantao Wang. 2002. “Class Identification of Married Employed Women and Men in America.” American Journal of Sociology 108(2):440–75. Yamaoka, Kazue. 2008. “Social Capital and Health and Well-Being in East Asia: A Population-Based Study.” Social Science & Medicine 66(4):885–99.

(Received: April 6, 2012 /Accepted: July 31, 2012)

224