Open Access

Discordance between subjective and objective social status in contemporary China

The Journal of Chinese Sociology20152:14

DOI: 10.1186/s40711-015-0017-7

Received: 20 September 2015

Accepted: 9 November 2015

Published: 26 November 2015

Abstract

This study investigates the divergence between subjective and objective social status in contemporary China, using the unique data assembled from 10 waves of nationally representative surveys from 2003 to 2012. We construct the measure of status discordance by subtracting objective status generated by latent class analysis from self-rated social position. We find that more than half of urban residents in China underestimated their genuine social positions, while more than half of rural residents overestimated it. Regression analyses demonstrate that (1) family income, education, and occupational prestige are negatively correlated with status discordance; (2) the perceived upward social mobility is positively correlated with status discordance; and (3) the effects of objective indicators and perceived mobility differ between urban and rural residents.

Keywords

Status discordance Perceived social mobility Subjective social status

Background

From the 1990s, sociologists in China have been interested in how people see themselves as being located in the social hierarchy. A large body of Chinese literature has been devoted to examining such interchangeably used concepts as subjective social status, subjective social status, and class identity (Bian and Lu 2002; Dong 2007; Gao 2013; Feng 2011; Li et al. 2005; Li and Zhang 2008; Liu 2001, 2002; Lu 2010; Zhao 2005). These studies revealed that as in Western societies, subjective social status among the Chinese is influenced by but does not completely coincide with objective social position. Unlike Western societies where there is a strong tendency for people to express middle-class identity (Centers 1949; Hodge and Treiman 1968; Jackman and Jackman 1983; Wright and Singelmann, 1982; Evans et al. 1992; Kelley and Evans 1995; Oddsson 2010; Hout 2008), however, these studies also show that Chinese people on average have lower social status. Although there is a rich literature regarding perceived status, only a few studies extensively explore the relationship between subjective and objective social positions (Qiu and Han 2014; Sosnaud et al. 2013) using cross-sectional data. Furthermore, during the transition period in China, characterized by expanding inequality and social turmoil, it is of both theoretical and policy significance to investigate the discordance between subjective and objective social status, since status inflation or status deflation can have important governance implications.

This study, therefore, aims to complement the literature by systemically assessing the divergence between subjective and objective social status among the Chinese. Based on the nationally representative data assembled from 10 waves of two leading projects (Chinese General Social Survey (CGSS) and Chinese Social Survey (CSS)) between 2003 and 2012, we construct a unidimensional measure of objective social position using latent class analysis and exploit the discrepancy between subjective and objective status to construct a measure for status discordance. We describe in detail the structure, characteristics, and evolution of status discordance over the decade and further explore how the discordance is determined in a sociologically meaningful way, using multiple approaches including the ordinary least squares (OLS) model, hierarchical linear model (HLM), and a generalized ordered logit (gologit) model. In a step forward from studies based on Western populations or on much fewer and less representative Chinese samples, we endeavor to present a broad picture of status discordance in China over the recent decade and to discover whether findings on self-rated social position in other countries, especially the USA and Europe, apply to China, a formerly communist country undergoing profound and fast social changes.

Status discordance: concept and literature

The divergence between subjective and objective social positions, or, as Hodge and Treiman (1968) called it, “status inconsistency,” is well-known but not well-understood in the literature. Hout (2008) attributes such status ambiguity to imperfect correlations among income, education, and occupation. However, very few empirical studies focused on this topic explore why and how the discordance is formed. One exception is Sosnaud et al. (2013), analyzing the relationship between subjective social status and voting among Americans. In this study, the authors found that around half of the Americans have concordant status identity,” while around one third have a “deflated perception” (self-rated status being lower than objective status) and about 17 % have an “inflated perception” (self-rated status being higher than objective status). Using the discrepancy between objective and subjective status as the dependent variable to fit the multi-logistic models onto data, they found that race and education are closely related to status discordance.

Literature in Mandarin has addressed this topic from the 2000s. Wang and Li (2002) argued that one’s subjective social status may differ depending on objective social conditions. Li (2005) found that the correlation between objective indicators (e.g., income, education, occupation, and consumption) and subjective social status is not strong. Li (2004) revealed that those on the middle rung of the socioeconomic ladder tend to underestimate their genuine social positions. Lei (2009) revealed that white-collar migrants in Shanghai have mixed perceived social status. Similar findings regarding the existence of status discordance came from a study of rural residents by Lu and Zhang (2006), analyzing data from villages in Zhejiang Province. In a more recent study, Qiu and Han (2014) analyzed urban samples from CGSS2010 and found that more Chinese overestimate their true social status in urban area. And people with higher occupational prestige tend to deflate status, while those who are on the lower rungs of the prestige ladder tend to have inflated social status. As for the mechanism of status discordance, these studies provided three aspects of explanations. First, self-rated status hinges on perceived social justice or subjective wellbeing (Qiu and Han 2014). Second, both social elites and those with lower status on the bottom rung of the social ladder have stronger class identity than the middle classes (Li 2004). Third, social mobility generates social fragmentation and therefore leads to greater divergence between subjective and objective social positions (Li 2005).

That being said, however, previous studies regarding status discordance have two major limits. First, the measure of status discordance in previous studies may be problematic since they simply used the EGP schema (after Erikson, Goldthorpe, Portacarero 1982) aggregated from occupational categories, Liuxin schema based on social power (Qiu and Han 2014), or even occupational prestige itself (Lei 2009) to represent objective status and therefore ignored the other pivotal dimensions such as income and education. For example, although Sosnaud et al. (2013, p. 90) defended the use of EGP schema by stressing that “most important for our purposes, the EGP has been the predominant schema in the class-voting literature,” a unidimensional measure of objective social status commanding income, education, and occupation would have been more convincing. Another example is the heated debate about using occupation or other factors to construct the class scheme in China (Liu 2005, 2007; Li et al. 2012; Lin and Wu 2010; Li 2005, pp. 53–127; Fan 2014). Second, most previous China studies on subjective social status, as a matter of fact, have been focused on either specific groups or certain cities with cross-sectional research design, meaning that the findings are neither generalizable to the entire country nor sufficient to track the temporal trends. Thirdly, such studies lack a systematic illustration of the mechanism by which status discordance is formed. That is, the determinants of status discordance have not been systematically investigated. Our study, therefore, aims to fill in this gap.

Mechanisms of discordance: theories and hypotheses

To understand status discordance, we must first of all focus on subjective social status. We know that self-rated social position in essence can be seen as a substitute for or component of class consciousness. Therefore, both individual-level and contextual-level factors contribute to its formation (Hodge and Treiman 1968; Wright 1997; Morris and Murphy 1966; Lopreato and Hazelrigg 1972; Jackman and Jackman 1973). Evans and Kelley (2004) classified previous theories of subjective social status into two types. One is the reality argument, or Marx and Durkheim’s argument, that one’s subjective social status is a direct reflection of his/her objective position in the social hierarchy. The distribution of subjective social status in a society, therefore, changes over the course of economic development and differs under different political systems (e.g., communism vs. capitalism). The other, the reference group argument, stresses that subjective social status is influenced by subjective sampling and therefore reflects one’s immediate social milieu (Woelfel and Haller 1971), concordant with the social comparison theory (Festinger 1954; Hoffman et al. 1954). Furthermore, Chinese sociologists have proposed similar arguments such as “relative deprivation” (Liu 2001). That is, no matter whether they are higher class or lower class, in a society featuring drastic social changes, anyone can experience status discordance.

Consequently, a mixture of the above two approaches can constitute the foundational theory of status discordance. Here, we follow Kelley and Evans (1995) and label it the “Reality and Reference” (R&R) blend argument which incorporates both “the social-psychological forces towards centrality and the social economic forces towards dispersion” (Evans and Kelley 2004, p. 4). This argument in general echoes the Weberian idea of class identity in prioritizing a subjective interpretation of one’s objective social position. It stresses that one’s status discordance is not only affected by his/her objective social indicators such as income, education, and occupation but also hinges on one’s immediate social networks. The “R&R-blend argument” has been supported by many empirical studies. For instance, previous findings suggested that most people understand class labels (Jackman and Jackman 1983) and routinely place themselves on a certain rung of the social ladder (Centers 1949; Moorhouse 1976; Wright 1985; Evans et al. 1992). Although objective social position—in particular education, income, and occupation—plays a remarkably stable role in forming subjective social status (Hout 2008), it does not perfectly predict self-placement in the social hierarchy (Hodge and Treiman 1968; Kelley and Evans 1995; Oddsson 2010). As a matter of fact, numerous studies across countries have revealed that people of all economic backgrounds tend to see themselves as being positioned in the middle class (Evans et al. 1992; Evans and Kelley 2004; Adair 2001; Hout 2008) and subjective social status is more of an indicator of social-psychological identity (Wright 1997; Adair 2001). In addition, aside from individual attributes, context effects such as the role of the wealth of a country, income inequality, employment, or political ideology at the aggregated level are also found to be significant in shaping one’s self-rated social position (Evans and Kelley 2004; Andersen and Curtis 2012; Curtis 2015).

The R&R-blend argument applies to subjective social status and status discordance in any given society. However, they can also be affected by individual trajectory in terms of social mobility (Wright and Shin 1988). Specifically, China is undergoing unprecedentedly drastic social transition.1 Therefore, one’s objective position on the social ladder could have been greatly changed or complicated during the process of fast stratification and urbanization (Lu 2010). For example, a newly rich person may possess merely a high school diploma, but when it comes to class consciousness, he/she may declare higher class identity than a graduate working as a primary school teacher. These are in fact the Chinese versions of Yoshino’s merchant seaman quoted in Hodge and Treiman (1968) and are perhaps more complicated simply because of the vast scale and accelerated pace of the social transition. In addition, one of the results of the dramatic social changes could be that people find it more difficult to use a reference group to locate themselves in the social hierarchy because of the mismatches between other people’s education, income, and occupations. If the reference group as the benchmark fails to be suitable for “horizontal comparison,” one may turn to, or at least rely more on, “historical comparison” using his/her perceived class mobility. If it is the case, subjectively perceived mobility would explain more variations of subjective social status than objective positions would. Consequently, we propose a “mobility” argument as the other mechanism of status discordance paralleling the R&R-blend argument. Based on the R&R-blend argument and the mobility argument, we hypothesize the following:
  • H1 Status discordance can be commonly observed in transitioning China

    The higher the objective status is, the less likely one is to identify his/herself as being lower class, and the less room there is to overestimate the genuine social position. If we measure the status discordance by subtracting objective status from subjective status, we can further propose the following three hypotheses regarding the effects of three major objective social indicators of class, based on the R&R-blend argument:

  • H2.1 Educational attainment is negatively correlated with status discordance

  • H2.2 Occupational prestige is negatively correlated with status discordance

  • H2.3 Income level is negatively correlated with status discordance

    Considering the huge socioeconomic gap between rural and urban China (Lu 2009; Wang 2010), we expect that the effects of the three objective indicators differ between rural and urban residents (Gao 2013). In particular, with the expanding social radius and rural–urban migration, the reference groups of rural residents, especially of those who have links with urban areas through family or work, or have lived in cities for a short time, may show deflated status (Gao 2013; Zhang 2011). Meanwhile, some urban residents may have higher self-rated class identity since their reference groups consist of more rural–urban migrants. Finally, rural–urban migrants may be earning more and are on average better educated than rural dwellers, but they are very likely to place themselves in the bottom of the society simply because their referenced groups have changed from village peers to urban neighbors. Therefore, we propose the following:

  • H3 The effects of educational attainment, income, and occupational prestige are different between rural and urban residents

    Based on the mobility argument, we also expect that one’s perceived mobility, or subjective social mobility or relative class change (Liu 2002), is strongly correlated with status discordance. We follow Bourdieu (1984) to group self-reported mobility into three classes: perceived upward mobility, downward mobility, and no mobility.

  • H4 Compared to those who have experienced no horizontal social mobility, people who have perceived upward mobility tend to have higher status discordance.

    Since the opportunity of social mobility is higher and better among urban residents than rural residents, we further propose the following:

  • H5 The roles of perceived social mobility differ between rural and urban people.

Data, measurements, and methods

Combine data

The richness of the data from two leading nationwide survey projects, namely CGSS and CSS, has provided us the opportunity to explore subjective social status among the Chinese in detail. The CGSS was initiated by the Hong Kong University of Science and Technology and the Renmin University of China in 2003 and had been jointly conducted together with several leading universities in China annually or biannually. The CSS was initiated in 2006 by the Chinese Academy of Social Sciences. Each wave of CGSS covered 5000–12,000 households drawn from 125 counties in around 30 provinces, while each wave of CSS covered 7000–8000 households from 128 counties in around 30 provinces. Although the two projects are conducted by different institutions, they are both continuous cross-sectional, of very similar multi-stage stratified national probability sampling, and share a lot of key questions with identical wording, answer categories, and sequencing.

Ten waves of CGSS and CSS asked respondents to explicitly express their social status by choosing a position in the scaled social hierarchy. As shown in Table 1, three waves are from CSS (2006, 2008, and 2011) and seven waves from CGSS (2003, 2005, 2006, 2008, 2010, 2011, and 2012), giving a total sample of 82,823 adult Chinese.2 Among them, 80,958 respondents had explicitly answered the question of subjective status. That is to say, only less than 3 % of the respondents denied status by failing or refusing to offer an answer to the close-ended questions of choosing the group which they believe they belong to in the social hierarchy. This strongly shows that the Chinese are as much class-aware and willing to answer status questions as their peers in the USA (Hout 2008), although one might have expected that the legacy of egalitarian norms rooted in the communist ideology could have caused Chinese to less recognize the labels of social strata.
Table 1

Data from 10 waves of CGSS and CSS, 2003–2012

 

2003

2005

2006

2008

2010

2011

2012

Total

CGSS

        

 All cases

5894

10,372

10,151

6000

11,785

5620

11,765

61,587

 

 Cases with SSS

4933

10,372

9641

6000

11,730

5607

11,712

 

59,995

CSS

         

 All cases

7061

7139

7036

21,236

 

 Cases with SSS

6994

7045

6924

 

20,963

Total

       

82,823

80,958

SSS subjective social status

Because the data were combined from two projects spanning 10 years, the measurement consistency of subjective social status is worth close inspecting. In fact, CGSS started using the 10-level scale with visual aid of a 10-rung ladder on the questionnaire from 2008,3 while CSS and other waves of CGSS adopted traditional 5-level scales (upper, upper middle, middle, lower middle, and lower class). For compatibility, we transformed the 10-rung ladder into the 5-level scale.4 Besides, CGSS 2003 asked the respondents to choose the position of “your family,” while all the other waves used the wording of position of “yours.” Although intuitively the two are compatible, it is problematic to simply merge them for a longitudinal analysis if the data shows considerable divergence. Fortunately, because CGSS 2005 and 2006 asked both the questions about family and individual class identity, we had the opportunity to conduct the test. We found that more than 92 % declared exactly the same class, and 99.7 % when using a three-level scale. The dominating concordance assures us that the wording problem does not pose threats to our analysis.

To measure objective social status, as jointly determined by education, income, and occupation, we deliberately dropped those who had never worked, which resulted in a sample of 76,946.5 Due to missing data, our working sample for analysis shrunk to 68,054. These individuals were sampled from 31 provinces, accounting for all provinces in mainland China. However, of the 68,054 respondents around 41 % lived in rural areas, while around 59 % in urban areas, meaning that urban residents were oversampled.6 In addition, given the sampling design of CGSS and CSS, respondents were selected from households with different numbers of adults. Therefore, for each wave, we use sampling weights to compute representative figures for the annual general population in China. Finally, considering that the annual sample sizes are different from the real annual population in China from 2003 to 2012, we further use sampling weights to correct for this.7 In fact, our working sample is hitherto the largest data sets ever investigated in social class and mobility studies in China.

Dependent variable

To construct the measure of status discordance, we first use latent class analysis (LCA) to generate a unidimensional measure of objective status in so much we can quantify the divergence between objective and subjective social status. The LCA hypothesizes the existence of unobserved (latent) categorical variables to explain the associations among a constellation of observed (manifest) categorical variables. As regards this paper, the latent categorical variable refers to objective social status, while the manifest categorical variables are three major indicators in the surveys, namely income, education, and occupational prestige. In light of LCA model, when informants’ responses on manifest variables are mutually independent on the condition of being a member of one of the latent classes, the typology could be identified according to the number of latent classes at a given significant level. Having generated the categorical variable of objective social status, we quantify the divergence between subjective social status and objective social status. Specifically, since we have a five-category subjective status variable, as long as we show that a five-category objective status variable can be identified according to LCA statistical criteria, we can simply subtract the objective status from the subjective status. Note that if the objective status has more than five categories according to LCA statistical criteria, we simply choose the five-category objective status for compatibility because the subjective status is a five-category dummy.

In this paper, an exploratory LCA model8 is employed, owing to that it is not only ideal for absence of an explicit theory about the nature of social strata but also can provide information about the adequacy of proposed assumptions. The particular approach is clear cut: to begin by fitting a one-class model to the data, followed by a two-class model, a three-class model, a four-class model, etc., and continuing until a model is identified with adequate fitness (Goodman 1974a, b). When a model fits data satisfactorily, the relevant statistics such as latent class probabilities and latent conditional probabilities could tell us more about the objective social status structure. Further, all the respondents could then be assigned to the appropriate latent classes to create a new categorical variable immediately denoting his/her objective position in the social hierarchy. The detailed discussion on the LCA model, the model goodness-of-fit, and the model usefulness can be referred to McCutcheon (1987).

We present the fit statistics from the LCA models assuming from one to seven classes in Table 2. Three variables, education (five categories), family income (five categories), and EGP schema (six categories), are used for generating the latent variable of objective social positions.9 As is shown, the seven models’ likelihood ration chi-square statistic L 2 (and its p value), the associated Bayesian information criteria (BIC and adjusted BIC) as parsimony indices, and degrees of freedom are all listed. Specifically, model 1 reported in the table is equivalent to testing a model with one single latent class (complete independence), while model 2 tests two classes, model 3 tests triple classes, model 4 tests four classes, etc. The p value shows that model 7 is sufficient to end the exploratory analysis with seven latent classes, given a 0.001 alpha level.10 Since the subjective social class is measured by a five-category dummy, we chose the five-category variable generated from model 5 to compute status discordance and report the estimated conditional probabilities of the five latent classes in the Appendix. In the robustness check, we also collapse the seven-category status variable from model 7, fitting data satisfactorily into a five-category one. Subtracting the objective status variable from the subjective status variable, we obtain a measure of status discordance ranging from −4 to 4. In addition, we also draw on Sosnaud et al. (2013) and classify status discordance into three types: concordant status, inflated status, and deflated status. Importantly, deflated status may pose threats to governance since people underestimate their genuine social positions on the social ladder. We will later describe status discordance over the decade in detail.
Table 2

Latent class analysis model selection: fit statistics

Models

L 2

p(L 2)

BIC (L 2)

Adj-BIC (L 2)

df

1

One-class

511,41.580

0.000

626,367.370

626,329.234

107

2

Two-class

8056.536

0.000

583,426.991

583,347.540

94

3

Three-class

867.578

0.000

576,382.697

576,261.932

81

4

Four-class

163.355

0.000

575,823.139

575,661.060

68

5

Five-class

103.286

0.000

575,907.735

575,704.341

55

6

Six-class

72.321

0.003

576,021.435

575,776.727

42

7

Seven-class

38.993

0.102

576,132.772

575,846.750

29

Independent variables

We present major independent variables in Table 3. At the individual level, the three key independent variables of interest are family income, schooling years, and international socioeconomic index (ISEI),11 together making up one’s objective social position. Demographic control variables include gender, age, marital status (1 = married; 2 = divorced/widowed; 3 = single), employment status (1 = employed; 2 = unemployed; 3 = retired), and residential status (1 = rural; 2 = urban). In addition, since the CGSS and CSS both asked respondents to report how they felt about their social position compared to before,12 we also control for a triple-class dummy of perceived mobility (1 = upward mobility; 2 = downward mobility; 3 = no mobility). Notice that in the regressions, we use the natural logarithm value of family income.
Table 3

Descriptive statistics for variables used in models of subjective social status (China, 2003–2012)

Variables

Proportion (%)

Mean (sd)

Perceived mobility of class

Upward mobility

35.48

 
 

Downward mobility

32.38

 
 

No mobility

32.14

 

Work status

Unemployed

14.48

 
 

Employed

74.76

 
 

Retired

10.76

 

Residential status

Rural residents

51.14

 
 

Urban residents/migrants

48.86

 

Marital status

Married

88.43

 
 

Single

5.26

 
 

Divorced/widowed

6.31

 

Male = 1

 

0.502 (0.50)

CCP member = 1

 

0.103 (0.31)

Age

 

45.42 (14.06)

Years of schooling

 

8.41 (4.20)

ISEI

 

34.66 (15.35)

Family annual income (Yuan)

 

40,662 (120,168)

N = 68,054

  

Percentages for categorical variables and means for continuous variables are reported; mean score of SSC is also shown; numbers in parentheses are standard deviations

Because the data are combined from 10 waves of surveys spanning 10 years and from two projects, we include year dummies and project dummies (1 = CGSS, 0 = CSS) to control for changing macroeconomic conditions or unobserved common aggregate shocks. To test for a trend in the relation between subjective and objective conditions such as education, income, and occupational prestige, we incorporate linear year and interactions of linear year and relevant indicators of objective positions. Furthermore, when doing single-level analysis, we control for dummies of province.

Methods and models

We first of all provide an overall picture of status discordance in China from 2003 to 2012 and compare it with that in the USA. Then we perform multivariate regression to explore how status discordance is determined by objective status indicators and other individual attributes in a sociologically meaningful way. We start by fitting the OLS model to treat status discordance as a continuous variable. Considering that individuals are nested in provinces and cities, we proceed to fit HLMs. Finally, we use the triple-category dummy of status discordance as the dependent variable and fit the gologit model since the test of parallel lines had not been passed, paying close attention to the mechanism of “deflated status.” To address the rural–urban difference, we include interaction terms between residential status (rural vs. urban) and a set of individual attributes, rather than naïvely fitting models to subsamples of rural or urban respondents, respectively.

Status discordance in China: a description

We present the annual distributions of both objective and subjective social status in Fig. 1. Regarding objective social position, on average over the 10 years, around 16.05 % of Chinese respondents were upper middle or upper class, 18.78 % were located on the middle rung, and around 65.17 % belonged to the lower middle or lower class. This is quite different to the structure of the subjective social status, for which the relevant percentages are 0.08, 6.08, 40.63, 31.19, and 21.29 %. In fact, the mean score of objective social status among the Chinese is 3.06, which is much higher than the mean score of subjective status, 2.34. Figure 1 generally shows that most people were located in the middle lower and lower classes, and the shape of the objective status is onion-like. In comparison, most people claimed to belong to the middle lower or lower classes, and the shape of subjective status is more plump-waisted, resembling half of a bowling bottle. In addition, compared to objective status, the central tendency is more salient with subjective status.
Fig. 1

Distributions of subjective and objective social status overtime

The mean score of status discordance as a continuous variable is −0.093, with a standard deviation of 1.42. That is to say, over 2003–2012, on average, Chinese reported lower subjective status than objective status. The structure of status discordance in each year is quite similar. As the bold black line in Fig. 3 shows, on average, around 24 % of people overestimated their class by 1 point (status discordance = 1), while around 16 % underestimated by 1 point (status discordance = −1) (Fig. 2).
Fig. 2

Annual distribution of status discordance (percent)

If we follow Sosnaud et al. (2013) in using a triple-class typology (inflated, deflated, concordance) to describe the status discordance, we will get 39.74 % of respondents with inflated status, 31.11 % with deflated status, and 29.14 % with concordant status. And in the USA, Sosnaud et al. (2013) revealed that the responding percentages are 52, 24, and 24 %. In other words, more than half of Americans accurately located themselves on the social ladder, while in China, this ratio was only 29.14 %. Therefore, H1 has been confirmed: status discordance is very common among the Chinese, and it is even more common than in the USA. In addition, Fig. 3 demonstrates that deflated status had increased from 2003 to 2001, while inflated status had the opposite trend.
Fig. 3

Tendencies of triple-category status discordance (percentage). S < O = deflated, S = O = concordant, and S > O = inflated

Among urban residents, the percentages of “concordant status,” “inflated status,” and “deflated status” are 28.74, 20.32, and 50.94 % while those of rural residents are 29.54, 58.71, and 11.75 %. Consequently, it looks like urban residents tend to underestimate their social class while rural residents tend to overestimate it. We use a dotted line to denote status discordance in urban areas and a solid line to present that in urban areas in Fig. 4 and find that the tendency for urban residents to claim deflated status or for rural residents to claim inflated status is stable over time.
Fig. 4

Rural–urban difference of status discordance

Results and discussion

Linear regression

We present in Table 4 the results from OLS and HLMs (random coefficient models) predicting continual status discordance,13 after establishing the presence of inflated and deflated status. Models OLS 1 and HLM 1 are baseline models, while models OLS 2 and HLM 2 incorporated relevant interaction terms. According to Table 4, both OLS models and random coefficient models show similar patterns of how status discordance is shaped. Holding other factors fixed, people with higher educational attainment tend to have less status discordance, measured by subtracting objective status from subjective status. Likewise, those with higher family income or better jobs are more likely to report less status discordance, other things being equal. Therefore, H2.1, H2.2, and H2.3 are supported.
Table 4

Ordinary least square and hierarchical linear models of status discordance China 2003–2012

 

OLS 1

OLS 2

HLM 1

HLM 2

     

Random coefficient

Random coefficient

 Schooling years (SY)

−0.110***

(0.0052)

−0.0675***

(0.0050)

−0.116***

(0.0064)

−0.0723***

(0.0054)

 Household income (HI)

−0.0351***

(0.0082)

−0.00999

(0.0117)

−0.0354***

(0.0085)

−0.0077

(0.0125)

 Occupation (ISEI)

−0.0191***

(0.0011)

−0.0286***

(0.0012)

−0.0190***

(0.0013)

−0.0280***

(0.0013)

 Perceived mobility

        

  Upward mobility (UM)

0.284***

(0.0230)

0.317***

(0.0247)

0.279***

(0.0232)

0.316***

(0.0244)

  Downward mobility (DM)

−0.193***

(0.0220)

−0.226***

(0.0262)

−0.197***

(0.0217)

−0.227***

(0.0251)

 Age

0.0051***

(0.0007)

0.0055***

(0.0007)

0.0053***

(0.0007)

0.0056***

(0.0007)

 Male

−0.0717***

(0.0112)

−0.0966***

(0.0107)

−0.0774***

(0.0107)

−0.0975***

(0.0102)

 CCP membership

−0.0608***

(0.0197)

−0.0339*

(0.0185)

−0.0566***

(0.0187)

−0.0343*

(0.0180)

 Marital status

        

  Divorced/widowed

−0.118***

(0.0244)

−0.123***

(0.0236)

−0.121***

(0.0234)

−0.125***

(0.0234)

  Single

−0.159***

(0.0244)

−0.124***

(0.0230)

−0.148***

(0.0230)

−0.123***

(0.0219)

 Work status

        

  Employed

0.0489***

(0.0151)

0.0861***

(0.0154)

0.0607***

(0.0146)

0.0912***

(0.0154)

  Retired

0.028

(0.0237)

−0.0207

(0.0218)

0.0079

(0.0242)

−0.025

(0.0225)

 Urban residents

−0.479***

(0.0308)

−0.456***

(0.0317)

−0.460***

(0.0296)

−0.438***

(0.0310)

 Linear year

0.0021

(0.0047)

0.0047

(0.0045)

0.0028

(0.0045)

0.0046

(0.0044)

 CGSS

0.167***

(0.0254)

0.180***

(0.0249)

0.167***

(0.0244)

0.179***

(0.0242)

 SY# urban residents

  

−0.0907***

(0.0062)

  

−0.0852***

(0.0057)

 HI# urban residents

  

−0.0491***

(0.0142)

  

−0.0527***

(0.0149)

 ISEI# urban residents

  

0.0167***

(0.0016)

  

0.0154***

(0.0013)

 UM# urban residents

  

−0.0725**

(0.0337)

  

−0.0785**

(0.0331)

 DM# urban residents

  

0.0704***

(0.0245)

  

0.0659***

(0.0229)

 Province dummies

Yes

Yes

 Constant

−0.657***

(0.0451)

−0.616***

(0.0482)

−0.152***

(0.0480)

−0.188***

(0.0481)

sd (SY)

    

0.0317***

(0.0046)

0.0200***

(0.0027)

sd (HI)

    

0.0396***

0.0062

0.0387***

(0.0072)

sd (ISEI)

    

0.0064***

(0.0012)

0.0045***

(0.0008)

sd (random intercept)

    

0.1685***

(0.0296)

0.1674***

(0.0298)

sd (residuals)

    

1.0295***

(0.0092)

1.0189***

(0.0100)

Observations

68,054

 

68,054

 

68,054

 

68,054

 

(1) Estimations are based on weighted data; robust standard errors adjusted for clustering on provinces are shown in parentheses in OLS models. (2) Reference categories: female, non-CCP membership, married, employed, no mobility; project CSS; year 2003; Beijing

***p < 0.01

Models OLS 2 and HLM 2 show that the effect of objective indictors differs between rural and urban areas. Specifically, according to model OLS 2, controlling for other variables, status discordance of a rural resident with a high school diploma is less than that of a rural resident with a junior school diploma by 0.41 (−0.0675*6 = −0.405). However, the effect among urban residents is around 1 (−0.0675–0.0907)*6 = 0.96), almost double than among rural residents. Similar to education, the effect of family income on status discordance is significant too. In particular, the impact of income is not statistically significant among rural residents, while it is negative among urban residents, probably because of the low variation of family income in rural areas. Besides, the negative role of occupational prestige in status discordance is more salient for rural residents. Based on these findings, H3 is empirically supported.

According to Table 4, not surprisingly, perceived upward mobility would encourage people to report higher status discordance, as compared with perceived no mobility, controlling for other predictors. Taking model OLS 1 for example, those who experienced upward mobility on average have 0.28 higher status discordance than those without perceived mobility, around one third of the class scheme. If we compare those with perceived upward mobility with those with downward mobility, the difference of status discordance would be 0.47, comparable to the difference between a rural and urban resident. This finding generally supports H4. Furthermore, we find that the correlation between perceived mobility and status discordance is different between rural and urban areas, according to models OLS 2 and HLM 2. In particular, the negative role of downward mobility is relatively small among urban residents, while upward mobility has a larger role among rural residents, providing empirical evidence for H5.

Since respondents on the top (bottom) rung of the social ladder cannot claim higher (lower) subjective status, we fit models to subsamples of those belonging to the middle class, upper middle class, and lower middle class to see whether the results are changed by deleting samples to rule out the ceiling effect, which can be a confounder. In Table 5, we present the results from subsample models, controlling for the same covariates as those in Table 4 (although we only report coefficients of the three major objective indicators, perceived mobility, and their interaction terms). As Table 5 shows, both OLS and HLM models have almost identical patterns to what we present in Table 2, providing us further evidence supporting H1 through H2.
Table 5

Models of status discordance ruling out the ceiling effect China 2003–2012

 

OLS 3

HLM 3

Schooling years (SY)

−0.0294***

(0.0070)

−0.0324***

(0.0070)

Household income (HI)

−0.0921***

(0.0266)

−0.0857***

(0.0253)

Occupation (ISEI)

−0.0130***

(0.0010)

−0.0127***

(0.0011)

Perceived mobility

    

Upward mobility (UM)

0.281***

(0.0320)

0.278***

(0.0319)

Downward mobility (DM)

−0.260***

(0.0350)

−0.265***

(0.0343)

SY# urban residents

−0.0253***

(0.0074)

−0.0232***

(0.0077)

HI# urban residents

0.0941***

(0.0198)

0.0851***

(0.0205)

ISEI# urban residents

0.0126***

(0.0012)

0.0116***

(0.0011)

UM# urban residents

−0.0214

(0.0384)

−0.0231

(0.0380)

DM# urban residents

0.125***

(0.0330)

0.127***

(0.0323)

Other controls

Yes

Yes

Observations

39,346

 

39,346

 

(1) Estimations are based on weighted data; robust standard errors adjusted for clustering on provinces are shown in parentheses in OLS models. (2) Reference categories: female, non-CCP membership, married, employed, no mobility; project CSS; year 2003; Beijing

***p < 0.01

Gologit regression

We use the triple-category dummy to measure status discordance and fit the logistic model. The paralleling line test shows that the gologit model should be fitted. We present the relevant results in Table 5, reporting the model predicting the probability of underestimating status (deflated status) in models 1 and 2 and the probability of overestimating status (inflated status) in models 3 and 4. In general, we get quite similar results to what we found in Tables 4 and 5, supporting all five of our hypotheses. For example, education, family income, and occupational prestige all predict the probability of deflated status. The only exception is that the interaction terms of residential status and income are insignificant. Likewise, perceived upward mobility predicts less probability of inflated status, while downward mobility does the opposite. And the role of mobility is different between rural and urban areas (Table 6).
Table 6

Gologit models of status discordance China (2003–2012)

 

Gologit1

Gologit2

Gologit3

Gologit4

 

Predicting deflated status

Predicting inflated status

Schooling years (SY)

0.229*** (0.010)

0.180*** (0.012)

−0.130*** (0.009)

−0.0927*** (0.010)

Household income (HI)

0.207*** (0.023)

0.218*** (0.028)

−0.0215 (0.020)

−0.0286 (0.026)

Occupation (ISEI)

0.0196*** (0.002)

0.0342*** (0.002)

−0.0427*** (0.002)

−0.0535*** (0.002)

Perceived mobility

    

 Upward mobility (UM)

−0.594*** (0.045)

−0.674*** (0.048)

0.594*** (0.045)

0.674*** (0.048)

 Downward mobility (DM)

0.291*** (0.043)

0.480*** (0.057)

−0.461*** (0.051)

−0.480*** (0.057)

Age

−0.0102*** (0.001)

−0.0099*** (0.001)

0.0102*** (0.001)

0.0099*** (0.001)

Male

0.212*** (0.022)

0.223*** (0.023)

−0.212*** (0.022)

−0.223***(0.023)

CCP membership

0.0168 (0.040)

−0.0195 (0.038)

0.0609 (0.042)

0.0195 (0.038)

Marital status

    

 Divorced/widowed

0.157** (0.065)

0.164** (0.065)

−0.286*** (0.054)

−0.290*** (0.053)

 Single

0.158*** (0.041)

0.144*** (0.041)

−0.158*** (0.041)

−0.144*** (0.041)

Work status

    

 Employed

−0.174*** (0.031)

−0.220*** (0.042)

0.174*** (0.031)

0.145*** (0.030)

 Retired

0.059 (0.057)

0.081 (0.063)

0.162*** (0.046)

0.113*** (0.044)

Urban residents

0.883*** (0.066)

0.768*** (0.058)

−0.742*** (0.054)

−0.768*** (0.058)

SY# urban residents

 

0.0882*** (0.009)

 

−0.0882*** (0.009)

HI# urban residents

 

−0.00111 (0.025)

 

0.00111 (0.025)

ISEI# urban residents

 

−0.0215*** (0.002)

 

0.0215*** (0.002)

UM# urban residents

 

0.155** (0.065)

 

−0.155** (0.065)

DM# urban residents

 

−0.212*** (0.052)

 

0.212*** (0.052)

Linear year

−0.0316*** (0.008)

−0.0348*** (0.009)

0.014 (0.010)

0.0157* (0.010)

CGSS

−0.442*** (0.049)

−0.454*** (0.049)

0.364*** (0.047)

0.382*** (0.045)

Province dummies

0.181** (0.080)

0.259*** (0.069)

−2.148*** (0.062)

−2.053*** (0.074)

Constant

Yes

Yes

Yes

Yes

Observations

68,054

68,054

68,054

68,054

(1) Average marginal effects based on weighted data are reported; robust standard errors adjusted for clustering on provinces are shown in parentheses. (2) Reference categories: female, non-CCP membership, married, employed, no mobility; project CSS; year 2003; Beijing

***p < 0.01; **p < 0.05

Before closing this section, we proceed to conduct a sensitivity analysis to check if our analysis is robust across various model specifications and measures of important variables. To save space, we do not report the regression results here, but they can be accessed upon request. Specifically, we also fitted the following: (1) multi-level models treating waves as the third level where provinces are nested in, (2) a multi-level ordered logistic model treating social status as an ordered multinomial dummy, (3) models excluding wave 2003 since it contains only urban residents, (4) models on the data from 2010 to 2012 using a 10-level scale for subjective social status, and (5) models on subsamples such as wave-specific samples, province-specific samples, CGSS samples, CSS samples, non-religious samples, religious samples, urban samples, and rural samples. All in all, although not reported here, the similar results of these tests show that our analysis above is robust and reliable.

Conclusions

Using a unique dataset assembled from 10 waves of CGSS and CSS, we investigate status discordance in China or the divergence between subjective social status and objective social status. Based on the large sample of a decade, we used latent class analysis to generate a unidimensional measure of objective status, which uses one’s education, income, and occupational prestige. Then we compare it with subjective status to construct a variable of status discordance. Our method improves upon previous approaches using occupation to measure objective status and status discordance, making the results here more convincing. Regarding regression models, we used OLS, HLM, and Gologit models to perform regression, under different working assumptions. The multi-model strategy ensures that our findings are robust.

Our empirical description shows that the proportion of deflated social status in China is higher than that among Americans, suggesting that people may have more difficulties in recognizing their genuine social position during fast social transition, like what China is undergoing. This situation is further complicated by the fact that more than half of urban residents had deflated status, while more than half of rural residents had inflated status. All these findings potentially point to the pattern of the divergence between subjective and objective status in a transitional society with gigantic rural–urban differences in terms of socioeconomic conditions, in comparison with the USA or Europe.

We pay close attention to the role of both objective indicators and subjective factors in status discordance. Our results obtained from a multiple-model approach in general echo Sosnaud et al.’s (2013) finding that divergence between subjective and objective social status is shaped in a sociologically meaningful way. Importantly, we reveal that in China over the last decade (1) higher levels of education, family income, and occupational status may lead to deflated class and the effects of them are different between rural and urban China and (2) status discordance also hinges on perceived social mobility in that upward mobility predicts inflated status, and the implications of mobility also differ between rural and urban areas.

In general, these findings provide strong evidences for our five hypotheses derived from both the R&R-blend argument and the mobility argument. In fact, we argue that a blended theoretical framework incorporating both the R&R-blend and mobility arguments may shed more light on our understanding of the mechanisms of status discordance. Since the R&R-blend argument is position-focused and the mobility argument is process-focused, we propose that a blending “position-process” framework could be used for future studies. Finally, our analyses also have two policy implications.14 First, deflated status is obviously harmful for governance. Therefore, for a concordant status or inflated status, the policy makers need to not only improve objective social conditions such as income and education but also make more people feel they are or can be experiencing upward social mobility. Second, considering the rural–urban difference in terms of the determinants of status discordance, the government should endeavor to narrow the socioeconomic gap between rural and urban areas.

As mentioned from the beginning, China provided class/status researchers a rare opportunity to explore the distribution, determinants, and discordance regarding subjective social status under societal transformation. The large-scale and drastic social changes among highly heterogeneous and institutionally segregated population have given rise to different patterns of class/status identification in comparison with that in the USA, Europe, or other countries. As such, the subjective social status and status discordance are a key dimension of importance to understanding the mechanism of how objective positions and perceived mobility affects social stability, especially in a society like China which is undergoing dramatic social and economic transformation. The findings in our study, some concordant with previous research while some otherwise, therefore are helpful to call on more academic attention to be devoted to this line of research.

Footnotes
1

There has been an impressive accumulation of studies focusing on social stratification and mobility in post-reform China (Walder 1986; Nee and Matthews 1996; Bian and Logan 1996; Griffin and Zhao 1993; Nee 1996; Parish and Michelson 1996; Xie and Hannum 1996; Zhou 2000; Wu 2002; Wu and Treiman 2004; Xie and Wu 2008, to name a few. See Bian 2002 for a cogent review).

 
2

The overall response rate of CGSS and CSS is around 75 %.

 
3

The use of 10-scale measure of SSS can be referred in Evans and Kelley (2005), Adler et al. (2000).

 
4

Specifically, we recoded top 10 and 9 to upper class, 8 and 7 to upper middle, 6 and 5 to middle, 4 and 3 to lower middle, and 2 and 1 to lower class.

 
5

For those who are retired, the ISEI is computed based on their last job.

 
6

The urban population had not outnumbered rural population in China until 2011. In addition, all waves covered both rural and urban China except CGSS 2003 which were sampled from urban residents. We included CGSS 2003 to get more waves for time trend analysis. In the sensitive test, however, we will exclude CGSS 2003 to see if our results are robust.

 
7

Weights are calculated as follows, taking the year of 2010 as an example: First, a household weight (HWT) equal to the ratio of the number of adults in the household to the mean number of adults per household (estimated separately for the urban and rural samples) was computed, an appropriate weight to use when we analyze the urban and rural samples separately. Second, since in 2010 49.68 % of the population that year (1.339 billion) lived in urban areas, a population weight (PWT) was computed separately for the urban and rural samples. For the urban population, PWT = (1.339 billion*0.4968 /urban sample size)*HWT; for the rural population, PWT = (1.339 billion*0.50.32/rural sample size)*HWT. Finally, weights are then normalized to the original sample size of 2010: WEIGHT1 = PWT/mean (PWT). Since the 2010 population is 1.339 billion and we have a sample of 11,730, the weights of cases in 2010 are further corrected by WEIGHT2 = WEIGHT1*1.339/11,730. Then weights are finally normalized to the original overall sample size from 2003 to 2010: WEIGHT = WEIGHT2/mean(WEIGHT2). Similar method can be referred in Wu and Treiman (2004).

 
8

See the difference between exploratory LCA (unrestricted) and confirmatory LCA (restricted) in McCutcheon (1987:27).

 
9

We did not use continuous variables to represent income, education and occupations and perform latent profile analysis because (1) the continual measure of income and education has lower validity and efficiency compared to categorical income variable and (2) these two methods lead to quite similar results.

 
10

Since LCA analyses may have several local maxima, we tried different initial values and got quite similar results all with numbers of latent class more than five.

 
11

For those who are retired, the ISEI is computed based on their last job.

 
12

Some waves asked respondents to compare with 10 years before while other waves to compare with 5 years before. We therefore first of all used subsamples for relevant waves to check if the measure inconsistency causes any divergence in analysis and found no prominent difference in the role of perceived mobility.

 
13

We also fitted random intercept models. The likelihood ratio test shows that random coefficient models are better.

 
14

We stress that the R&R-blend argument is position-focused in the sense that it does not address the potential role of perceived social mobility in subjective social status. Since social mobility is process-focused, we propose the notion of “position-process.”

 

Declarations

Acknowledgements

This research was supported by the Youth Project of the National Social Science Foundation “Civic Participation under the Perspective of Social Mobility” (13CSH020) and sponsored by the Zhejiang Academy of Social Science (Sociology of Development). Early draft of this paper was presented in the Conference of Social Stratification and Mobility in winter 2014, Nanjing. We thank Xin Liu, Xiaogang Wu, Yuxiao Wu, Yaojun Li, Yong Gao, Richard Williams, Ding Li, Qi Xu, and Zifan Li and anonymous reviewers for their helpful comments and discussion. All faults are solely ours.

Open AccessThis article is distributed under the terms of the Creative Commons Attribution 4.0 International License (http://creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, and reproduction in any medium, provided you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made.

Authors’ Affiliations

(1)
School of Social and Behavioral Sciences, Nanjing University
(2)
Institute of Sociology, Zhejiang Academy of Social Sciences

References

  1. Adair, Stephen. 2001. Immeasurable differences: a critique of the measures of class and status used in the General Social Survey. Humanity and Society 25(1):57–83.Google Scholar
  2. Andersen, Robert & Josh Curtis. 2012. The polarizing effect of economic inequality on class identification: evidence from 44 countries. Research in Social Stratification and Mobility 30(1):129–41.Google Scholar
  3. Bian, Yanjie, and Hanlong Lu. 2002. Reform and socioeconomic inequality: shanghai Citizens’ perceived status. In Market transition and social stratification: the analyses of China by American sociologists, ed. Yanjie Bian, 509–531. Beijing: SDX Joint Publishing Company (China) [In Chinese].Google Scholar
  4. Bian, Yianjie. 2002. "Chinese Social Stratification and Social Mobility." Annual Review of Sociology 28:91–116.View ArticleGoogle Scholar
  5. Bourdieu, P. 1984. Distinction: a social critique of the judgement of taste. Cambridge: Harvard University Press.Google Scholar
  6. Wang, Chunguang and Li, Wei. 2002. Subjective construction and objective reality of social class in contemporary China, Jiangsu Social Sciences (4):95–100. [in Chinese].Google Scholar
  7. Erikson, R., Goldthorpe, J. H., & Portocarero, L. 1982. Social fluidity in industrial nations: England France and Sweden. British Journal of Sociology, 33(1):1–34.View ArticleGoogle Scholar
  8. Wright, E.O. 1997. Class Counts: Comparative Studies in Class Analysis. Cambridge: Cambridge University Press.Google Scholar
  9. Morris, Richard T., and Raymond J. Murphy. 1966. "A paradigm for the study of class consciousness. "Sociology and social research 50(3):297–313.Google Scholar
  10. Lopreato, Joseph, and Lawrence E. Hazelrigg. 1972. Class, conflict, and mobility: Theories and studies of class structure. San Francisco: Chandler House Press.Google Scholar
  11. Woelfel, Joseph, and Archibald O. Haller. 1971. "Significant others, the self-reflexive act and the attitude formation process." American Sociological Review 36(Febrary):74–87.Google Scholar
  12. Centers, Richard. 1949. The psychology of social classes: a study of class consciousness. Princeton: Princeton University.Google Scholar
  13. Moorhouse, H. F. 1976. "Attitudes to Class and Class Relationships in Britain." Sociology 10(3):469–96.View ArticleGoogle Scholar
  14. Wright, Erik Olin. 1985. Class. London: Verso.Google Scholar
  15. McCutcheon, Allan L. 1987. Latent class analysis. London: Sage.Google Scholar
  16. Curtis, Josh. 2015. Mobility and class identity in 35 societies: the role of economic conditions. The European Sociological Review, forthcoming.Google Scholar
  17. Dong, Yunsheng. 2007. Status inconsistency and class structuration, vol. 1, Journal of Jilin University (Social Sciences Edition)th ed. [In Chinese].Google Scholar
  18. Evans, M.D.R., and Jonathan Kelley. 2004. Subjective social location: data from 21 nations. International Journal of Public Opinion Research 16(1):3–38.View ArticleGoogle Scholar
  19. Evans, M.D.R., Jonathan Kelley, and Tamas Kolosi. 1992. Images of class: public perceptions in Hungary and Australia. American Sociological Review 57(n4):461–82.Google Scholar
  20. Fan, Xiaoguang. 2014. Permeability of boundaries and inequality: discussing consequences of social stratification. Beijing: Social Sciences Academic Press (China) [In Chinese].Google Scholar
  21. Feng, Shizheng. 2011. Class identity and social stability in China’s social transition. Heilongjiang Social Sciences (3):127–33. [In Chinese].Google Scholar
  22. Festinger, L. 1954. A theory of social comparison processes. Human Relations 7(2):117–140.Google Scholar
  23. Gao, Yong. 2013. Why has status identification declined: on the changes in the basis of status identification. Society (4):83–102. [In Chinese].Google Scholar
  24. Goodman, Leo A. 1974a. The analysis of systems of qualitative variables when some of the variables are unobservable. American Journal of Sociology 79:1179–259.View ArticleGoogle Scholar
  25. Goodman, Leo A. 1974b. Exploratory latent structure analysis using both identifiable and unidentifiable models. Biometrika 61:215–31.View ArticleGoogle Scholar
  26. Hodge, Robert, and Donald Treiman. 1968. Class identification in the United States. American Journal of Sociology 73(5):535–47.View ArticleGoogle Scholar
  27. Hoffman, P., L. Festinger, and D. Lawrence. 1954. Tendencies toward group comparability in competitive bargaining. Human Relations 7(2):141–159.View ArticleGoogle Scholar
  28. Hout, Michael. 2008. How class works in popular conception: most Americans identify with the class their income, occupation, and education implies for them. In Social class: how does it work? ed. Annette Lareau and Dalton Conley. New York: Russell Sage Foundation.Google Scholar
  29. 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.Google Scholar
  30. Jackman, Mary R., and Robert W. Jackman. 1983. Class awareness in the United States. Berkeley: University of California Press.Google Scholar
  31. Kelley, Jonathan, and M.D.R. Evans. 1995. Class and class conflict in six Western nations. American Sociological Review 60(2):157–78.View ArticleGoogle Scholar
  32. Lei, Kaichun. 2009. Analysis on the current situation and reasons of White-Collar New Migrants’ SES identification deviation. Youth Studies (4):60–70. [In Chinese].Google Scholar
  33. Li, Chunling. 2004. Identity of social class. Jiangsu Social Sciences 6:108–22. [In Chinese]Google Scholar
  34. Li, Chunling. 2005. Cleavage and fragment: an empirical analysis on the social stratification of the contemporary China. Beijing: Social Sciences Academic Press (China) [In Chinese].Google Scholar
  35. Li, Peilin, and Yi Zhang. 2008. The scope, identity, and social attitudes of the middle class in China. Society (2):1–19. [In Chinese]Google Scholar
  36. Li, Peilin, Yi. Zhang, Yandong Zhao, and Liang Dong. 2005. Social conflicts and class consciousness in China today. Beijing: Social Sciences Academic Press (China) [In Chinese].Google Scholar
  37. Li, Lulu, Guangqiang Qin, and Jianwei Chen. 2012. The construction of authority-based Stratum System: an analysis based on work situation and organizational authority. Sociological Studies (6):46–76. [In Chinese].Google Scholar
  38. Lin, Zonghong, and Xiaogang Wu. 2010. China’s institutional change, class structure transformation and income inequality: 1978–2005. Society (6):1–40. [In Chinese].Google Scholar
  39. Liu, Xin. 2001. Class consciousness of China’s urban residents during the transitional period. Sociological Studies (3):8–17. [In Chinese].Google Scholar
  40. Liu, Xin. 2002. Relative deprivation status and class cognition. Sociological Studies (1):81–90. [In Chinese].Google Scholar
  41. Liu, Xin. 2005. Understanding the mechanism of social stratification in contemporary China. Sociological Studies (5):1–25. [In Chinese].Google Scholar
  42. Liu, Xin. 2007. Class structure and the middle class location in urban China. Sociological Studies (6):1–14. [In Chinese].Google Scholar
  43. Lu Xueyi. 2009. Break dual structures and improve the economic-social integration. Social Science Research (4):1–5. [In Chinese].Google Scholar
  44. Lu, Yilong. 2010. Class consciousness and class identity of rural residents: structure or construction. Jiangsu Social Science (1):114–24. [In Chinese].Google Scholar
  45. Lu, Fuying, and Zhaoshu Zhang. 2006. Objective stratification and subjective identity. Chinese Journal of Population Science(3):38–43. [In Chinese].Google Scholar
  46. McCutcheon, Allan L. 1987. Latent class analysis. Newbury Park, CA: Sage Publications.Google Scholar
  47. Oddsson, Gudmundur Aevar. 2010. Class awareness in Iceland. International Journal of Sociology and Social Policy 30(5/6):292–312.Google Scholar
  48. Qiu, Liping, and Yu. Han. 2014. Urban Residents’ class status identification deviation. In The summer 2014 meeting of China social stratification and mobility conference, 227–42. Wuhan: Sociology Department of Wuhan University [In Chinese].Google Scholar
  49. Sosnaud, Benjamin , David Brady and Steven M. Frenk. 2013. Class in name only: subjective status identity, objective status position, and vote choice in American Presidential Elections. Social Problems 60(1):81–99.Google Scholar
  50. Wang, Chunguang. 2010. The Urban and Rural Structure. In Social structure of contemporary China, ed. Xueyi Lu. Beijing: Social Sciences Academic Press [In Chinese].Google Scholar
  51. Wright, Erik Olin & Kwang-Yeong Shin. 1988. Temporality and class analysis: a comparative study of the effects of class trajectory and class structure on class consciousness in Sweden and the United States. Sociological Theory 6(1):58–84.Google Scholar
  52. Wu, Xiaogang, and Donald J. Treiman. 2004. The household registration system and social stratification in China 1955–1996. Demography 41(2):363–84.Google Scholar
  53. Zhang, Yi. 2011. The tendency of Chinese social class structure changes. Studies on Socialism with Chinse Characteristics (3):65–74. [In Chinese].Google Scholar
  54. Zhao, Yandong. 2005. The causes and outcomes of absence of middle class identity. Zhejiang Social Sciences (2):86–92. [In Chinese].Google Scholar

Copyright

© Chen and Fan. 2015