From capital to habitus: class differentiation of family educational patterns in urban China

Based on Bourdieu’s concepts of “capital” and “habitus” in his class theory, this paper examines the differentiation of capital investment and parenting habitus on children’s education between the Chinese urban middle class and lower class. By analyzing 2009 survey data of students in grades 4 and 8 in urban areas, the authors found that middle-class parents had significant advantages in capital investment, but showed no significant differences in parenting attitudes when compared to lower-class families. This finding indicates that the current Chinese middle class largely relies on capital possession, but displays few differences in class habitus compared with the lower class. The so-called class crystallization is maintained primarily through economic capital, but not through distinctions of inner dispositions.

The Chinese middle class

The Chinese academia started the discussion of the middle class in the mid-1980s. Today, this research topic receives extensive attention (Li 2009). While the question of whether China has a middle class or not is still open (Zhou and Chen 2010), scholars are more concerned about the size and attitude of the middle-class population, as well as its impact on current Chinese society.

The size of the middle class

Cheng Li (2010) believes that China’s middle class consists of three main groups: the economic group, the political group, and the educated group. The size of the middle class depends on the defined criteria. While still under debate, scholars agree that there are four commonly defined criteria: having a high and stable income, engaging in professional or managerial occupations, having a high level of education, and being able to maintain a comfortable standard of living and consumption (Chunling Li 2010). Due to different criteria, scholars estimate the size of the middle class differently. For example, Zhou (2005) uses monthly income above 5000 RMB in 2005, having a white-collar occupation, and having access to formal college education as the standards; Zhou estimates that in the five major cities of Beijing, Shanghai, Guangzhou, Nanjing, and Wuhan, the proportion of the middle class is only 11.9 %. Similarly, using income, occupation, and education as indicators, Peilin Li and Yi Zhang use CGSS2006 data and estimate that in 2006 China’s urban middle class probably accounted for about one-quarter of the urban population. They also find that self-identified “middle stratum” is more stable and more easily form of common social attitudes and behavioral preferences (Li and Zhang 2008). However, the scholarly definition, individual subjective identified definition, and public perception of the middle class are far from each other. The public perception of the middle class only describes the upper middle class rather than the entire middle class (Chunling Li 2010). Hence, during China’s rapid transformation, the clear boundaries of the Chinese middle class and class identity are still in formation, and research on political attitudes and consumption patterns of the middle class corroborate this assertion.

Political attitudes and consumption patterns of the middle class

Many scholars believe that the middle class can play a role in social stability (Qiang Li 2001; Zhao 2005), and act as a stabilizer of social structure and buffer zone in social conflicts (Li and Li 2007). However, skeptics point out that the middle class is not necessarily a social stabilizer (Zhang 2008) and may act as a subversion or alienation (Hu and Hu 2008). Zhou (2005) believes that Chinese middle-class individuals are “avant-garde in consuming, backward in politics.” They are staunch allies of the government, unlike the middle class in some countries where they are the key to democratic transition of power. The reason for this is that in addition to the conservative and radical nature of the Chinese middle class, it also strongly depends on the state (Li 2008).

Groups within the Chinese middle class depend on different economic bases, which determine whether these individuals’ political attitudes are radical or conservative. The new middle class can be divided into state-owned economic groups and private sector groups (respectively referred to as “endogenous” and “exogenous” middle class). The former has strong intergenerational continuity in close proximity with the government, and is relatively conservative in political and consumption awareness; the formation of the latter is characterized more by the market economy and is thus more radical in political and consumption awareness (Li and Li 2007; Chunling Li 2010). Zhang (2008) also finds that the “new middle class” is more critical than the “old middle class” and is more skeptical of the government and institutions. However, because the new middle class is one of the biggest benefiters of the reform and opening, it is still a strong supporting force of social stability (Chunling Li 2011).

Although economic basis deeply influences political attitudes, few scholars use only income and economic ability to define the middle class (Cheng Li 2010; Chunling Li 2010). The formation and reproduction of class have both an economic as well as a cultural dimension; both are indispensable (Crompton 2008). In terms of media exposure, compared with other classes, the Chinese middle class spends more time on reading books rather than exposure to television, newspapers, or the Internet. Middle-class individuals read more books and journals with more knowledgeable information (Zhou 2005; Guo 2009); thus, class differentiation on reading is emerging. According to the consumption stratification, Qiang Li argues that since the late 1990s, China’s cultural consumption has changed from a class-convergent “cultural hybrid mode” to a class-differentiated cultural pattern (2011, 296-97).

The above empirical research on political attitudes and consumption patterns of the middle class reflects the two basic contrasting ideas. One view is that the Chinese middle class, although increasing in scale, is still not a “class” with unified social attitudes and behavioral orientation (Li and Zhang 2008). Another view is that China’s class structure is moving from economic stratification to cultural stratification, and class structure is increasingly solidifying and even formed a cleavage (Sun 2002; Li 2004). How can we understand this seemingly contradictory phenomenon? Neo-Marxists believe that the formation of a new class identity requires common class experiences and class consciousness. Class experiences emphasize that identification is a reflection of actual social and economic situations stemming from a long-term interaction with the objective world; class consciousness emphasizes that class identity is a social construct concerning people’s active role in the historical process (Thompson 1966). In other words, the transition between Marx’s (1959) “class in itself” to “class for itself” requires individual experience as an intermediary bridge, otherwise the objective social and economic conditions do not automatically translate into class identity. Therefore, understanding the formation of a class needs to go beyond the opposition between objectivism and subjectivism, of which Bourdieu’s class theory is undoubtedly a classic paradigm.

After the reform and opening, Chinese society provided a natural laboratory for discussing the dynamic changes of social stratification. The reform and opening up is a watershed that separated “political stratification” from “economic stratification” in Chinese society (Li 1997). After experiencing radical and chaotic revolutions, upper-class culture and lifestyles have almost disappeared (Qiang Li 2011). This provides an excellent opportunity for us to observe the emergence and development of cultural stratification after the reform. Bourdieu’s (1984) emphasis on economic capital, cultural capital, and stratification of habitus provides a good framework for this analysis.

Bourdieu’s class theory: capital and habitus

Bourdieu’s theories of class are scattered among his books and articles (Weininger 2005). In Bourdieu’s view, if one principle exists that applies to various aspects of social life, it is “distinction.” In a divided society, no individual, group, or class can escape the logic of distinction (Bourdieu 1984). Distinctions between individuals or groups mainly occur in material relations and symbolic relationships, namely objective status position and internal ideas and tastes (its different combinations form “habitus”). The former limits attitudes and actions of individuals, and the latter is the individual’s inner experiences and active construction of the living world (Wacquant 2013). Therefore, Bourdieu declares that class is defined by its “being” and “being-perceived” at the same time (1990, 135).

Bourdieu defines class as a group of individuals that shares a common nature and the same external living conditions. He proposes using economic, social, cultural, and symbolic capital to analyze the individual’s external living conditions in the social space, with habitus analyzing internalized personal characteristics. In Bourdieu’s opinion, modern society is a result of different cross-penetrations of “fields” that form the “social space.” Individuals and classes use a variety of forms of capital in the social space to struggle with each other and to defend or take an advantageous social position. The same social position will result in the same or similar living conditions, thus shaping a similar class habitus (Liu 2003). Capital, field, and habitus are three core concepts of Bourdieu’s theory of class. This paper focuses on capital and habitus.

Bourdieu defines capital as “the set of actually usable resources and powers” (1984, 114), including economic capital, cultural capital, social capital, and symbolic capital. In the social space, the three main dimensions of class formation are the amount of capital, the proportion of capital formation, and the evolution of the historical track of status position. During class formation, economic and cultural capitals are considered the most important forms of capital. Bourdieu (1984) mainly uses the share of these two types of capital to locate individuals in the social space. Since labor time^{Footnote 1} required to obtain different forms of capital differs, reproducibility is thus different, which means various capitals’ capability of being passed on differ. Thus, different segments of capital have different strengths as class barriers. Compared with economic capital, the transfer of cultural capital is more intimate, riskier, and more difficult to pass on, but once cultural capital is obtained, its role as a class barrier is very strong.

However, the division of classes is generated not only from the economic and social conditions in the external social conditions (capital), it also depends on a unique lifestyle associated with a particular social position, the habitus formation (Bourdieu 1984). Habitus is a system of social disposition internalized in daily behaviors. It is the accumulation of individual cognitive and motivational systems and an internalization of the common objectives of social rules and organizational values. It is reflected in the individual in a less-conscious and sustainable manner, which embodies thinking, perceptions, and actions with cultural characteristics (Liu 2003). Bourdieu’s habitus goes beyond the inherent flaws in objectivism and subjectivism: habitus links social structure and practices, and is shaped by the social structure, as well as regulating practices (Brubaker 2004).

From Bourdieu’s point of view, habitus is acquired as a group thought, behavior, and mode of leisure activities. Obtaining habitus is not necessarily done consciously, but individuals occupying the same economic, cultural, social, and symbolic capital tend to have the same or similar temperament system (see Crompton 2008). Although members of the same class do not have identical experiences, there is no doubt that members of the same class are more likely than members of other classes to encounter similar situations (Bourdieu 1990). Thus, Crompton (2008) argues that a combination of various forms of capital constitutes habitus. Specific class habitus is the product of a series of objective norms, so it produces behaviors that exclude all “extravagances” (Bourdieu 1990, 56). On this basis, Bourdieu points out that due to different economic conditions, there are two basic classes of taste. One pursues luxury and freedom, and the other one pursues necessities. The former is a habitus of the ruling class that has a good economic standing, and the latter is reflected in the working-class habitus (Liu 2003). Lifestyle (status) shows the differences between the classes, and the concept of habitus connects lifestyle and class position. Habitus is reflected in a series of practical activities in such areas as consumption, which gradually forms symbolic boundaries between individuals occupying different positions in the class structure and further legitimates the class structure.

Habitus, a product of history, is based on past experiences (i.e., evolving of class position), in particular by experiences at an early age, including gender division of labor, household objects, modes of consumption, and family experiences such as parent-child relationships (Bourdieu 1990). Therefore, habitus is not directly determined simply by the various types of capital. For example, attending a concert does not indicate that all the listeners possess musical culture; only those who are familiar with the internal logic of these musical works are the real occupants (Bourdieu 1984). Capital is like concert tickets, while habitus is the appreciation for music.^{Footnote 2} Even possessing a ticket does not necessarily mean that the holder can immediately truly enjoy music. In other words, it is possible that there is a dislocation between habitus and capital.^{Footnote 3}

Thus, habitus, together with economic capital and cultural capital, is an element of social groups’ and classes’ construction of the borders between each other. Capital is more involved in the building of social boundaries, while habitus is closely related to the construction of the symbolic boundary (Fan 2012). However, the permeability of the two types of boundaries differs. Habitus needs historical accumulation; it cannot be like Zhao wei tian she lang, mu deng tian zi tang (A peasant in the morning becomes a guest of the emperor by the dawn), which cannot be achieved in a day. Compared with capital, acquiring habitus needs a longer time and has poorer permeability, so it is a more-rigid class barrier that cannot be easily penetrated.

Hence, Bourdieu’s theory of capital and habitus provides a useful perspective for inspecting the formation and changes in the class structure of Chinese society. By analyzing the differences in the dimensions of capital and habitus of the current classes, we can get a glimpse of the current stage of the formation of Chinese class structure. Sun (2008) states that in the late 1990s China began the process of finalizing the class structure, a process marked by four characteristics: the class boundary begins to form, including both tangible and intangible boundaries of living and culture; internal identity begins to form; interclass mobility begins to decline; and the reproduction of classes increases. His “cleavage society” thesis argues that the gaps in culture and many other aspects of social life increasingly emerge. The main cleavage is the wealth gap, which is the root of all kinds of differences and oppositions in Chinese society (Sun 2002). The question is whether the wealth gap has led to a cleavage in culture and lifestyle. In modern society, education has become an important stepping stone for obtaining social status, and is the core element of cultural capital. We therefore explore answers to this question by investigating the similarities and differences in the possession of capital and habitus in the education reproduction process between the middle class and lower class.

Intergenerational transmission of cultural capital

In his research on class reproduction, Bourdieu proposes three aspects to determine individuals’ class status: socioeconomic status, class habitus, and cultural and social capitals (Bourdieu and Passeron 1990). The first and third factors are undoubtedly forms of capital in Bourdieu’s theory and are external conditions; the second factor is habitus, part of individual dispositions.

External living conditions (capital) and internal disposition system (habitus) are the two core dimensions of Bourdieu’s definition of class. However, many consider that habitus cannot be directly observed, and can only be understood explanatorily (Weininger 2005). Bourdieu (1984) mainly observes lifestyles, especially consumer behaviors, to show the habitus behind them but he is also aware that the majority of cultural consumption activities are related to economic costs; for example, going to the theater requires expensive tickets. The first part of this article also pointed out the decisive role of economic capital in determining the political attitudes and consumption patterns of the Chinese middle class. Therefore, a key to using Bourdieu’s class theory in quantitative research is to find variables that measure dispositions and habitus that are not directly affected by capital, especially economic capital. We believe that in the study of educational reproduction, upbringing and parenting orientation is a more appropriate indicator of habitus.

Cultural capital is one type of cultural resources that a person or family possesses. In general, it is associated with the dominant concept or symbol in the society (Jæger 2011), thus individuals or groups with cultural capital can use it to enhance their advantages with better access to social resources and social status. Bourdieu (1986) points out that cultural capital has three forms: the embodied state, the objectified state, and the institutionalized state. While accumulated cultural capital covers the entire socialization process, it is also the most hidden of all forms of capital in the reproduction channels. Applying this concept, scholars have conducted many empirical studies, but the operational definition of cultural capital remains controversial. Most quantitative researchers use the participation of children or their parents in highbrow cultural activities, such as visiting museums, galleries, or concerts, as a common indicator. This measurement is also known as “embodied cultural capital” (Byun et al. 2012). However, other scholars argue that this operational definition is too narrow, and includes reading habits and literary environment, family educational resources, extracurricular activities, and other indicators (Jæger 2011). Studies have shown that these cultural capital indicators have significant effects (see related review Jæger 2011) on children’s academic performance and educational attainment. Among them, the number of books at home is an important indicator of cultural capital. It has a significant positive effect on children’s educational performance and attainment in many countries (Evans et al. 2010; Zhao and Hong 2012). Children in families with higher socioeconomic status are more involved in highbrow cultural activities (Sullivan 2001; Xu and Hampden-Thompson 2012). Some empirical studies in Western countries find that the more advanced the cultural activities a child participates in, the better the child’s academic performance (Jæger 2011); however, the generalization of this relationship to other societies is still questionable (Byun et al. 2012).

Cultural capital and economic capital are closely related, and cultural participation is more influenced by economic capital, but participation does not indicate cultural tastes. The latter is more subject to habitus during the process of socialization (Yaish and Katz-Gerro 2012). We use the number of books in the home, children’s participation in cultural activities, and children’s attendance at shadow education as three variables to measure input of domestic capital, and to distinguish them from the parenting orientations mentioned below as an indicator of habitus. Cultural capital investment may be significantly different between classes. We therefore propose:

Hypothesis 1: Chinese parents show obvious class differences in their investment in children’s education capital. Compared to lower-class parents, middle-class parents have significant advantages in the number of books in the home, children’s participation in cultural activities, and children’s attendance in shadow education.

Stratification of parenting orientation

As an indicator of habitus, parenting orientation is less constrained by economic capital.^{Footnote 4} At the same time, parenting style is closely related to children’s self-esteem, subjective well-being, adventurous spirit, and educational attainment. In psychological research, Baumrind (1971) first proposed common classifications of parenting, namely “authoritative,” “authoritarian,” and “permissive.” Maccoby and Martin (1983) divide Baumrind’s three types of parenting into two dimensions, responsiveness and demandingness; with interaction, this produced four parenting styles: “authoritative,” “authoritarian,” “indulgent,” and “neglectful,” classify parenting into three categories: “parents have the final say,” “children call the shots,” and “mutual decisions.” Pong et al. (2005) add a fourth category, “laissez-faire,” in which the parents and children rarely make decisions. This classification basically corresponds to Maccoby and Martin’s four categories. A large number of empirical studies show that when comparing the authoritarian and negative types of parenting, children from authoritative families are more mature, independent, and have more social responsibility and achievements (Chen 2002). International comparative studies note that the argument that authoritative parenting is more favorable has a certain universality and transcends racial and cultural boundaries (Pong et al. 2010).

Western psychology focuses on individual psychological characteristics, and thus some psychologists claim that the current research on parenting styles focuses too much on individual psychological variables and ignores other important contextual factors of class, and that learning a child’s class origin often better predicts the child’s future outcomes than knowing the child’s early psychological characteristics (Frank 2013). Using Bourdieu’s capital-habitus framework, Lareau (2003) provides a detailed description and analysis of family parenting styles of the middle class and the working class. She finds that the middle class adopts a “concerted cultivation” pattern, arranging activities for children and not hesitating to intervene in children’s activities. In comparison, the working class adopts the “accomplishment natural growth” pattern; they do not arrange activities for children and delegate a good deal of child-rearing responsibilities to schools. Although both parenting styles have strengths and weaknesses within the family, in public life middle-class children showed obvious advantages compared to their counterparts in working-class families because “differences in the cultural logic of child rearing are attached to unequal currency in the broader society” (Lareau 2003, 244).

Lareau’s (2003) binary distinction of parenting is based on three main aspects. First, different classes of parents invest differently in extracurricular activities, and children of the middle class have more opportunities to participate in formal, adult-led activities such as concerts. Second, different classes of people differ in the interaction patterns between parents and teachers. Third, interaction patterns of parents and children differ. Lower-class parents use an imperative tone more frequently, and middle-class parents discuss issues with their children more (Cheadle 2008). The first and second aspects of differences belong to the capital dimensions of education (cultural capital and social capital), and the third aspect belongs to parenting habitus. Using the concerted cultivation model, middle-class parents are more likely to uphold an authoritative parenting style, while lower-class parents are more inclined to use an authoritarian or permissive parenting style under the “accomplishment natural growth” model. Lareau (2003) believes that in many cases, parenting styles are not deliberate but are naturally revealed in everyday life, which she refers to as Bourdieu’s habitus.

Research on Chinese parenting styles is mainly conducted by psychologists. The influence of family socioeconomic status on children’s educational attainment via parenting has received little attention. Theoretical and empirical research studying class differentiation on parenting style is even rarer. Based on the above review, we argue that if China does indeed have a clear-cut class crystallization, then the parents of different classes should show significant differences in parenting orientation. Combined with the above thesis of a cleavage society or class crystallization, we propose:

Hypothesis 2: As an operationalized variable of habitus, Chinese parents show obvious class differences in parenting orientation toward children. Middle-class parents are more likely to utilize an authoritative parenting style, and lower-class parents are more likely to utilize an authoritarian or permissive parenting style.

Data

The dataset of this study is drawn from the “National Scientific Literacy Survey of Youth” conducted in 2009. The project was commissioned by the Ministry of Science and Technology, jointly implemented by the National Institute of Educational Sciences and the Chinese Academy of Science and Technology Development. The overall survey populations are fourth and eighth graders enrolled in schools in the country’s 286 cities (all municipalities, provincial capitals, and prefecture-level cities and municipal districts, excluding the jurisdiction of counties and county-level cities). The results of this study can thus be generalized only to the Chinese urban population, including some rural migrant workers living in cities. This survey applied the multistage random sampling method. Using the number of students in the school districts within the city as the basis and applying probability proportional to size (PPS) sampling, this survey sampled 128 municipal districts in all of the country’s municipal districts. It then randomly selected a primary and a secondary school in each region. Then in each primary school, a fourth-grade class, each junior high school, and an eighth-grade class were randomly selected. Finally, all students in the selected classes and their parents were interviewed. Each student questionnaire corresponded to one parental questionnaire, which was brought home for their guardians to complete. The final sample included 61 fourth-grade classes and 57 eighth-grade classes in 20 provinces. Qualified student and parental questionnaires totaled 6079 and 6028, respectively.

We removed nonparent respondents in parental questionnaires, and cases missing child’s gender. The final analytic sample size was 5411. Because about 19 % of the cases had missing values, we used multiple imputations to deal with this issue. We included all variables in the multiple imputation models. After imputation, we ended up with ten independent data sets, and ran statistical analysis for each of these datasets. The analysis results are summarized in the “Conclusions” section. Handling of missing data and statistical analysis were conducted by using Stata 12.0. All descriptive results and statistical model output were statistics after the missing value imputation. The results from the multiple imputation methods are similar to the results obtained by the case-wise deletion.

Variables

Dependent variables

The dependent variable is divided into two groups, capital and habitus. Capital includes participation in cultural activities, number of books in the home, and attendance at shadow education. Participation in cultural activities is a count variable. We asked whether the child went to the following places during the past year: science and technology museums, natural history museums, planetariums, botanical gardens, zoos, libraries, children’s palaces, or galleries. For each venue, a visit was coded as 1, otherwise 0. The eight scores were then added up to represent the number of cultural activities in which the child participated. The value of this variable is an integer between 0 and 8. The number of books in the home as a continuous variable ranged from the midpoint value for each category option (0 = 0 books, 5 = 1–10 books, 15 = 11–20 books, 35 = 21–50 books, 75 = 51–100 books, 150 = 101–200 books, 350 = 200, or more books). Attendance at shadow education is a dichotomous variable; participation in shadow education was coded as 1, otherwise 0. Shadow education involves two forms: private tutoring and attending tutoring classes, including composition, Mathematical Olympiad cg and other educational classes, and technological and arts classes.

Parenting orientation represents habitus; it was measured by the following three questions:

On the education of children, do you agree with the following statements? (1 = completely agree, 2 = agree, 3 = disagree, 4 = completely disagree)

1. A.

   Children contradicting the teacher is absolutely intolerable.

2. B.

   After children are out of class and have completed their homework, what they do with their time is their own choice. Parents do not need to control it.

3. C.

   All child-related things, regardless of their importance, must first be discussed with the children.

Habitus measurements are ordinal variables (1–4 refers to completely agree to completely disagree). Agreeing with statement A indicates an authoritarian tendency; agreeing with B indicates a laissez faire tendency; and agreeing with C indicates an authoritative tendency.

Independent variables

Occupational stratification was mainly based on Chunling Li’s (2010) classification of the middle class with minor changes, including the upper middle class, the lower middle class, and the lower class. The upper middle class includes: people in charge of agencies and enterprises, middle managers, senior professionals, general institutional leaders/civil servants, business employees, private entrepreneurs (with eight or more employees). The lower middle class includes: grassroots managers, other professional and technical personnel/general technicians, enterprise and institutional staff, skilled workers, soldiers, policemen, firefighters, and self-employed/small landowners (with eight employees or less). The lower class includes: commercial and service industry workers, unskilled workers, agricultural workers, forestry workers, animal husbandry workers, fishery workers, freelancers, and the unemployed.

Subjective class identification

The survey asked parents, “Where do you think your family’s socioeconomic status ranks locally? 1. The upper class; 2 the upper middle class; 3. the middle class; 4. the lower middle class; 5. the lower class; 6. unable to classify; 7. I do not know.” We collapsed the upper class, the upper middle class, and the middle class into one category, “the upper and middle class,” followed by “the lower middle class” and “the lower class.” Finally, we listed “unable to classify” and “I do not know” as a separate category.^{Footnote 5}

Education

We divided education into two categories according to whether the interviewed parent had received higher education (including college, undergraduate, and above), (0 = no, 1 = yes).

Household monthly income

This was incorporated into models after taking the natural logarithm.

We also added a number of control variables: children’s grade year (0 = fourth grade, 1 = eighth grade), gender of the interviewed parent (0 = mother, 1 = father), children’s ethnicity (0 = minorities, 1 = Han), and parents’ registered household status (0 = rural, 1 = urban). Descriptive statistics for relevant variables are shown in Table 1.

Table 1 shows a significant inconsistency between occupational class and subjective class identification. If classified according to occupations only, more than half of the parents belong to the lower class (nonskilled industrial workers, farmers, service workers, freelancers, and the unemployed). The upper middle class accounted for 19 %, and the lower middle class accounted for 28 %. However, in the subjective class, 26 % of the interviewed parents identified themselves as the lower middle class, only 12.3 % chose the lower class, while 41.7 % chose the upper middle class (of which only 4.7 % believed they were in the upper middle class and 37 % in the middle class). We thus incorporated both the occupational class and the subjective class in the regression model to examine their different roles.^{Footnote 6}

Table 2 presents the descriptive statistics of parenting orientation. The results show that the majority agree with statements A and C, and the majority do not agree with statement B. For the “intolerable to contradict the teacher” statement, authoritarian parents are the majority; for “discuss children’s activities with them,” the majority are authoritative parents; and for “children can choose their own activities,” there are less-“laissez-faire” parents.^{Footnote 7}

We used regression models to examine the class effects on educational capital investment and parenting orientation, and then compared the effects across the models. According to the dependent variable’s characteristics, we used negative binomial regression, general linear regression (OLS), logistic regression, and ordinal logit regression models. Participation in cultural activities is a count variable for which we generally used the Poisson regression. However, when there was an overdispersion problem (e.g., the variance is greater than the mean), negative binomial regression is more appropriate than the Poisson regression (Long and Freese 2001). The number of books in the home can be regarded as a continuous variable, so we used general linear regression. Attendance at shadow education is a binary variable, and we used logistic regression. Parenting orientations are all ordinal dependent variables, and were thus analyzed using ordinal logit regression.^{Footnote 8} Due to the sampling process that adopted cluster sampling, we therefore controlled for the cluster effect in all models.^{Footnote 9}

Class differences in capital and habitus

Models 1–3 in Table 3 present the effects of class on the variables of cultural capital, and models 4–6 in Table 4 present the effects of class on parenting habitus. It is important to note that incorporating variables from both parents are more appropriate in the capital models (Table 3), but for the ease of comparison, we incorporated the same variables in the capital models and the habitus models. In fact, incorporating both parents’ variables did not significantly alter the results of the models.^{Footnote 10}

The class variables show a strong impact in the capital models (models 1–3). Household income and education are important indicators for defining the middle class. Therefore, we only discuss the influences of the family income, parental education, and parental class position. Per capita household income significantly increases children’s participation in the cultural activities, the number of books in the home, and the possibility of children attending shadow education. Parents with higher education show significant positive effects in all three models. Parents’ occupational classes also have significant effects. Compared with parents in the occupational lower class, those in the upper and lower middle class provide more opportunities for their children to participate in the cultural activities, more books in the home, and attendance at shadow education. In terms of subjective class, compared to parents identified as the lower class, those in the subjective upper and lower middle class also involve their children more in the cultural activities. Parents in the subjective upper middle class provide more books in the home, but the number of books does not differ between the subjective lower middle class and the lower class. There was no significant difference among the different subjective classes in attending shadow education (model 3). Therefore, hypothesis 1 was basically confirmed: compared to the lower class, middle-class families have significant advantages in capital investment in their children’s education.

However, in the parenting habitus models (models 4–6), the variables of classes show no significant consistent impact on parenting orientation. In addition, the family income has no significant effect on the three parenting attitudes. Only in models 4 and 5 are well-educated parents who are more inclined to disagree with the statement “children contradicting the teacher is absolutely intolerable,” and are more likely to let the children choose their own activities after completing homework. In model 6, compared with occupational lower-class parents, upper middle-class parents are significantly more likely to disagree with the statement “All child-related things, regardless of their importance, must first be discussed with the children.” That is, upper middle-class parents show a weaker authoritative tendency than those in lower class parents. Other occupational classes and subjective class show no significant effects. Thus, the impact of class-related variables on parenting orientation is very weak, and sometimes shows significant impact on the opposite direction as the hypothesis. Therefore, parenting orientation as a habitus does not differ between the middle class (either the upper or the lower middle class) and the lower class. Hypothesis 2 is thus not confirmed.

Bourdieu’s theory of class proposes that capital and habitus are two key elements in educational reproduction. Capital includes participation in cultural activities and cultural material resources, and habitus focuses on subjective attitudes and dispositions. Analyzing national survey data on urban students, this study showed that in urban China there are no distinct class differences in habitus, but significant class differences in the capital dimension. Middle-class families hold undisputable advantages in capital investment in their children’s education. However, in terms of habitus, in urban China, the middle class does not exhibit the authoritative parenting that their Western counterparts do, compared with the authoritarian or permissive parenting style upheld by the lower class.

Qiang Li (2004 ; 2011) argues that China has formed a cultural stratification system, and the class structure tends to be rigid. Sun (2002) further proposes the cleavage society thesis. However, this study finds that a significant gap between classes still occurs mainly in the external capital dimension highly associated with economic resources. In the internal dimensions as habitus, the middle class was not significantly different from the other classes. We believe that the current phenomenon of cultural stratification is mainly the direct outcome of economic stratification in consumption, rather than in taste and habitus. The current Chinese “middle class” is primarily affluent in income, but shows a little class distinctiveness in aspects like attitudes and habitus. Undeniably, many middle-class members have learned a unique individual habitus, but a stable class habitus is still out of reach.

Compared to Western societies where social structure is relatively more stable, China’s social structure has undergone tremendous changes in the 30 years after the reform and opening up. This provides a good opportunity to better observe and analyze the processes of the formation of social class structure. This study indicates that Chinese society is still somewhat distant from class crystallization, and is on the way to being finalized. We believe that the current class barriers still rely mainly on external conditions like money, and internal barriers such as class attitudes and identity are still under construction. The line from Xiaogang Feng’s film “Not the best, but the most expensive” provides an excellent footnote of the current conditions of the Chinese middle class. Of course, this does not mean that the capital dimension and habitus dimension are separate from each other. Economic capital undoubtedly holds a leading role in current Chinese society. Weber (1946) points out that property is not an essential element, but in the long run, property is clearly an extremely stable element in obtaining status. Bourdieu (1986) notes that economic capital is the foundation of all other forms of capital, but only in the analytical sense. The conversion from economic capital to other forms of capital needs time, and it takes even more time for capital to shape habitus. Bourdieu also notes that individual mobility in the social space is not random. On the one hand, the individual is subject to the social forces shaping the social space; on the other hand, individuals can use their own inertia to resist these forces (Bourdieu 1984). In the current Chinese society, the economic factor is undoubtedly a decisive force, but the formation of a stable and distinctive middle class with clear boundaries will still take some time. Individual inertia needs to be steeped in the structure for a longer time to form a consistent class habitus. Therefore, in addition to investigating consumption and political attitudes, we should expand the study of the middle class to a broader area in order to deeply understand the life chances that they encounter in the transforming society, as well as their perceptions of opportunity and structure.

As an exploratory research, this article naturally has some limitations. First, the fact that the dimensions of habitus have no class difference may be due to the extremely complex internal components of China’s middle class. As Wang and Davis (2010) point out, in China’s rapid transformation, a big class analysis model does not apply, and a finer microstructure of class analysis may be more useful in understanding the different life opportunities and attitudes of various groups. Similarly, the researchers also believe that in Western societies, the middle class has become increasingly fragmented, making it difficult to form a collective consciousness of class identity (Crompton 2008). Second, the measurement of parenting orientation in this study still needs refinement, and there may be differences between the ideological orientations and actual practices of parenting behaviors. Data on daily parenting practices could better reflect the daily exercised but implicit habitus of parents. These issues require further exploration and data collection in the future.

1. 1.

   Bourdieu (1986) uses a general “labor time” as the unit of conversion between forms of capital.

2. 2.

   This appreciation is also considered an “embodied” form of cultural capital, but we use it only as a metaphor to explain the relationship between capital and habitus. We are not intending to join the debate on what cultural capital is.

3. 3.

   When discussing the relation between class habitus and personal habitus, Bourdieu notes that habitus based on personal experiences in the past can be inconsistent with the current class habitus. This dislocation is particularly highlighted in a rapidly changing society. Bourdieu coined the phrase the “hysteresis effect” to describe this dislocation (1990, 60–62).

4. 4.

   The amount of time spent with children is deeply affected by economic capital. For example, taking care of children may constrain the mother from full-time employment, and leads to the loss of “labor time” (Bourdieu 1986). Therefore, parenting orientation emphasizes the form of communication.

5. 5.

   The analysis found that participants who chose unable to classify and I do not know were between lower middle class and lower class in terms of education and income. However, there were some cases whose occupations belonged to the management, professional, or technical classes. We separated this group out, but it is not discussed in this paper.

6. 6.

   We also operationalized parenting orientation into a dichotomous variable of “agree and disagree” to construct logistic regression models; the results are similar to those obtained from the ordinal regression models.

7. 7.

   We ran regression models separately for fourth-grade and eighth-grade students, and the results were basically the same. We merged the two groups for the sake of simplicity.

8. 8.

   A better approach is to incorporate variables in the occupational hierarchy and in the subjective class identification separately, but the model results obtained revealed no major differences. The two class variables were entered together for the sake of simplicity.

9. 9.

   The ratio of statements A and B seems to indicate that there are differences in different situations. On the one hand, due to our limited measurements, it is difficult to fully measure parents’ parenting orientation. On the other hand, it also shows that many parents may not have a consistent parenting attitude but are instead influenced by different, even conflicting, values.

10. 10.

    An anonymous reviewer mentioned that there are possible class differences in the gender proportion of parents filling out the questionnaire, which in turn further affect their parenting orientation. We thus used father’s occupation as a reference point to examine the effects of gender of the interviewed parents. When the father belonged to the upper middle class, 62 % of the interviewed parents were mothers; when the father belonged to the lower middle class, 58 % of the interviewed parents were mothers; and when the father belonged to the lower class, 54 % of the interviewed parents were mothers. These results indicate that there were gender differences in filling out the parental questionnaires. However, the results from models in Table 3 show that gender did not significantly impact parenting philosophy. Even when running the analysis separately for gender, the regression results obtained were consistent with those in Table 3. In other words, whether it is fathers or mothers, parenting philosophy showed no class differences. We therefore believe that the gender of the parent who completed the survey does not have an impact on the conclusions of this article.

This research received support from The Co-Innovation Center of Civil Morality and Social Custom, Research Center of National Moral Survey, and “The National College Fundamental Science Fund of China” (Number: 2242014S20080). The manuscript of this paper was presented at session “social classification and mobilization” in Annual Meeting of Chinese Sociology Association (Guiyang) in 2013 and “Quantitative Social Science Research Workshop” held by Yingdong Huo Research School of HKUST at Nansha, Guangzhou on December 2013. The authors appreciate advice and suggestions from Xiaogang Wu, Yuxiao Wu, Yu Li, Jun Li, Yucheng Liang, Yaqiang Qi, and other conference participants as well as the anonymous reviewers and editors. The authors are solely responsible for the content of this paper.

Affiliations

1. Department of Sociology, School of Humanities, Southeast University, 2 Sipailou, Nanjing, 210096, China

   • Yanbi Hong

2. Research Center of Science, Technology and Social Development, The Chinese Academy of Science and Technology for Development, 8 Yuyuantan South Road, Haidian District, 100038, Beijing, China

   • Yandong Zhao

Corresponding author

Correspondence to Yanbi Hong.

Competing interests

The authors declare that they have no competing interests.

Authors' contribution

YH and YZ carried out the study, participated in the literature review and statistical analyses, and drafted the manuscript. All the authors read and approved the final manuscript.

Reprints and Permissions