Unequal parenting in China: A study of socio-cultural and political effects

This study examines the parental socio-cultural and political effects on parenting practices in China. Based on the China Education Panel Survey, we construct a new typology of parenting styles – intensive, permissive, authoritarian and neglectful – and focus on intensive parenting as a particular mode in which the more privileged families in China use superior cultural and political resources to reinforce their advantages. We show that parents in higher class positions, with higher education and with membership in the leading Chinese Communist Party (CCP) tend to adopt intensive parenting as a means of securing all-round development and obtaining favourable academic achievement for their children. Parenting styles thus reflect a more complicated feature of social stratification in China than in Western societies.


Introduction
Recent empirical evidence contains a remarkably rich but divergent body of findings on the impacts of family resources on parenting styles in reform-era China.Two major theses stand out.The first one, as exemplified by Hong and Zhao (2014), stresses similarity, while the second, as represented by Huang (2018) and Yang and Zhao (2020), focuses on disparity, in parenting practices.China used to be an agricultural society but the reforms that started in 1978 have created unprecedented 'room at the top', with many opportunities for upward mobility.Yet, as the Chinese occupational structure is still in constant flux, the burgeoning middle classes have not become full-fledged yet, nor have they developed distinctive parenting strategies, values and styles.It is thus argued that all parents in China, regardless of class, place great importance on children's education and that there are no clear class differences in parenting styles.This argument sounds convincing.However, the other argument is also persuasive.The reforms have led to growing socio-economic inequalities in Chinese society, which naturally lead to deepening divisions in parenting behaviour, with higher-class parents adopting permissive parenting practices.
Both arguments have garnered evidence for support.For instance, even very poor peasants in remote areas hold high expectations for their children (Li & Xie, 2020), wishing them to seize opportunities for educational success and realise their dreams of climbing up the class ladder.On the other hand, it is undeniable that the rapid increase in socio-economic inequality in post-reform China has led to widening gaps in parental resources and parenting styles.Both arguments root the different parenting styles in socio-economic and cultural disparities (Li et al., 2017;Li & Wen, 2021;Li & Zhu, 2022), at the expense of the far more important factor of political influence on parenting practices, namely the parental affiliation with the leading Chinese Communist Party (CCP) over and above the impacts of parental class position and educational qualifications.
The present research seeks to make a contribution to the existing literature on Chinese parenting in four ways.First, using nationally representative data from the China Education Panel Survey (CEPS), we conceptualise and measure the typologies of parenting styles in contemporary China.We do this taking into account China's changing social context.Over the past few decades, China has witnessed a rapid rise of the middle class, a large expansion in higher education, a sharp drop in fertility and changing norms towards parenting.Compared with previous generations, a large portion of the new generation of Chinese parents (born in the 1970s and 1980s) have only one child, and most children have benefited from the economic development and the expansion of higher education that started in the late 1990s.The rapid socio-economic changes have also intensified inequality and competition.Many studies have documented the growing rigidification of the class structure and deepening division between people in different social positions in post-reform China (Li, 2021;Wu, 2019).The economic uncertainty and rising inequality in recent years would engender a heightened sense of anxiety among Chinese parents.Driven by the 'loss aversion' mentality (Kahneman, 2011), parents in China are now increasingly encouraged to raise children in a competitive way in which children learn to adapt to ever more intensive market competition via enhanced aspiration, motivation and self-direction reinforced by all available resources.Although existing studies recognise the shift towards intensive parenting in English-speaking countries (e.g.Doepke & Zilibotti, 2019;Yerkes et al., 2021), there is little research that focuses on intensive parenting in China based on large-scale nationally representative data (Zhu, 2022).What also remains unknown is the overall picture of parenting styles in the current context in China and the distinct parenting categories among Chinese adolescents.
Second, we aim to examine the effect of parental social, cultural and political resources on parenting styles in China.This question is of particular interest because although more recent studies on Chinese parenting styles have re-emphasised the inequalities of parenting in China, they are limited in conceptualising and measuring parenting styles and they fail to acknowledge the dynamic nature of Chinese social structure since the reform and opening up, as well as the unique complexities of status acquisition in the context of China (Huang, 2018;Tian & Jing, 2021).A notable shift in scholarly research worldwide is towards the emphasis on 'intensive parenting', a term coined by Hays (1996) to describe the child-rearing activity in which parents sacrifice their own needs and spend a tremendous amount of time, energy and money on their children.Parenting, as Budds et al. (2017) and Irwin (2018) note, is conditioned by socio-economic resources, with differences manifesting themselves mainly between working-and middle-class families in Western societies.In Lareau's classic study of the impact of family class on child-rearing, she finds that middle-class American parents are actively engaged in a purposeful, goal-oriented concerted cultivation style of parenting whereas working-class American parents adopt a more laissez-faire, negligent accomplishment of natural growth style of parenting (Lareau, 2003).This raises the question of whether class position is as strongly linked to parenting style in China as it is in Western societies, whether political factors would play a role as important as class in China's situation, and whether Lareau's qualitative findings can be corroborated by using large-scale nationally representative survey data in China.By examining whether parental social resources, i.e. their class position, educational qualifications and membership in the Chinese Communist Party (CCP), have significant effects on parenting strategies and styles that might shape children's outcomes, this study will provide a more comprehensive grasp than hitherto available of the factors that shape parenting practices in China.
Third, it is hoped that our study of differences in parenting styles may provide a useful contrast with the case in Western countries, such as the United Kingdom and the United States, on which most studies on the inequalities of parenting styles are centred.In the UK the impact of parents' class or education on parenting styles is found to be insignificant (Chan & Koo, 2011), and in the US, the most influential study is based on qualitative research with no objectively measured class position (Lareau, 2003).In China, we expect parental class, education and political status all play important roles in parenting styles.Further, cultural and political resources, as captured by education and political affiliation, are assuming a dominant role in cadre selection and career progression (Bian et al., 2001;Li, 2003).More specifically, political resources make it easier to access cultural and economic privileges, and this political power is resistant to change (Kraaykamp & Nieuwbeerta, 2000).Under this system, marketisation is promoted by the state and previously politically advantaged groups reproduce the established social order through a series of capital holding and favour exchange.Parental party affiliation is likely to have a strong impact on parenting styles (e.g.CCP members are more likely to adopt parenting styles which are more conducive to their children's development).Given this, we would expect that political resources play similarly important roles in intergenerational transmission in China.
A final contribution is an assessment of whether differences in parenting styles are related to gaps in children's outcomes.While prior research recognises the role of parenting styles in child development, especially in school achievement and psychological development (Chan & Koo, 2011;Lamborn et al., 1991), few studies have provided a rigorous test on the impact of parenting styles on child outcomes.We use fixed effects models to address confounding and other biases that result from school characteristics.To produce a more robust estimate of the results, we also employ an Inverse Probability Weighted Regression Adjustment (IPWRA) approach to obtain causal estimates of the average treatment effect (ATE) of selection into a particular parenting style in multiple treatment settings.The problem of endogeneity becomes more complicated when treatment has multiple categories.This approach is a generalised propensity score procedure that uses a multinomial logit model with the inverse of the predicted probabilities as a sampling weight, which is particularly effective in balancing observables and estimating multiple treatments.Once again, if parenting styles can shape child development and set children on different paths and outcomes, then the finding might contribute to existing knowledge of the intergenerational reproduction of social inequality.

Conceptualising parenting styles
Parenting styles and their prototypes have been extensively studied by child development psychologists since the 1960s.The first detailed study on parenting style was Baumrind's (1971) seminal work, which proposes three parenting configurations: 'authoritative', 'authoritarian' and 'permissive'.Baumrind's research was based on observations made in laboratory and home settings, and on structured interviews with parents.Maccoby and Martin (1983) advance Baumrind's research by viewing parenting styles as a combination of two dimensions of responsiveness and demandingness.However, a major criticism of Baumrind's and subsequent researchers' work was their overly rigid two-factor model (Greenspan, 2006).For example, Becker (1964) introduces a third dimension labelled as 'anxious involvement versus calm detachment' and emphasises the notion of overprotective parents, who are high on warmth and control and who show more emotional involvement than other types of parents.Pulkkinen (1982) uses parenting involvement items to distinguish child-centred parenting from parent-centred parenting.Doepke and Zilibotti (2019) note that in sharp contrast with the permissive parenting that prevailed in the 1970s in the United States, intensive parenting characterised by a child-centred, heavily involved, time-intensive and demanding approach to child-rearing has become widespread across a number of industrialised countries over the last three decades, which features the active parental involvement in children's daily life, mainly in education-oriented activities.In sum, child development psychologists, educationalists and sociologists have developed the rudiments of intensive parenting.Even though their research mainly focuses on Western countries, their findings on intensive, child-centred parenting seem well suited to China's situation in the 21st century.

Class and parenting styles
Sociologists from the perspective of social stratification generally focus on how parental social resources, with particular regard to parental social class, contribute to the formation of parenting styles.Kohn (1959) was among the first to indicate that people's class position shapes the values of parenting.Classes vary in value orientations, in aspirations for life goals and in socio-economic resources to reinforce values and aspirations.Parenting values and orientations are rooted in different employment statuses and occupational conditions, which are captured by class (Goldthorpe, 1987).More specifically, middle-class parents place a higher value on self-direction whereas working-class parents adhere more to conformity and obedience (Kohn, 1959).In a ground-breaking ethnographic study, Lareau (2003) argues that middle-class and working-class parents adopt different strategies and habitus of family life, and she dubs the disparities as 'concerted cultivation' and 'accomplishment of natural growth'.According to Lareau, middle-class parents actively foster their children's talents, skills and social competence by virtue of organised leisure activities, language use and intervention in institutional settings.By contrast, working-class parents taking the natural growth approach are less involved in their children's activities.
Lareau's evidence is based on a small sample of 12 families, yet her findings on classaligned parenting styles have gained wide recognition and have been verified by quantitative studies using nationally representative data in Western countries.For example, Baker and Barg (2019), drawing on data from the Millennium Cohort Study in the UK, show that higher-class parents place more emphasis on 'thinking for self' than on obedience, as compared with their peers situated in routine manual positions.Similarly, in a paper based on the Youth Panel of the British Household Panel Survey, Chan and Koo (2011) show that authoritative parenting is more prevalent in professional-managerial households.With regard to engagement with teachers, Li et al. (2003) find that middleclass parents are around five times as likely to interact with teachers than their workingclass peers, and the same is found in France (Barg, 2019).

Beyond class: Education and political affiliation
Over and above class, there are other parental resources such as cultural and political capitals which shape parents' knowledge, beliefs, values and goals about parenting and which in turn impact parenting styles.
Previous research has shown a close link between parents' education and parenting styles.Drawing on a large sample of high school students in California, Dornbusch et al. (1987) show that highly-educated parents tend to adopt less authoritarian and permissive styles and more authoritative parenting.In addition to parenting styles, researchers find that education is positively related to parental time spent with children.Using data from the American time use survey, Guryan et al. (2008) find that highly educated parents spend more quality time with their children and are more effective in enriching their children through face-to-face interaction.Highly-educated parents view time spent with children as an essential investment in human capital.Maternal education is particularly related to the quality time mothers spend with their children, and more highly educated mothers tend to spend more quality time reading to their children and less time watching television with them (Timmer et al., 1985).
Research has also shown that political resources impact parenting in Western countries.Supporters or members of a political party tend to adopt specific viewpoints, attitudes and behaviours.Furthermore, parents' political party affiliation deeply influences children's family life.For example, in the United States, liberals are less likely than conservatives to report that their parents adopted harsh and punitive parenting practices (Adorno et al., 1950).Drawing on the Pew Research Center's Gender and Generations Survey, Elder and Greene (2016) find that Democratic fathers embrace less authoritarian and more egalitarian attitudes towards child-rearing and are more engaged in the day-today care of their children than are their Republican peers, who are more likely to hold authoritarian and traditional views about parenting, and to stress obedience over independence and self-reliance.
Political resources have been demonstrated to play a more salient role in former socialist countries, such as Bulgaria and Slovakia, than in Western countries due to the predominant leadership position held by the Communist Party members.For example, one study by Kraaykamp and Nieuwbeerta (2000) examines the effect of parental party affiliation on children's high-culture participation in the former socialist countries of Eastern Europe.The authors argue that in the communist era, political power plays an important role in the redistribution process.Nearly all reallocations of income, goods and power take place under the Party leadership, which plays an indispensable role in gaining access to high culture both directly and indirectly through parental cultural resources.The commonality of state socialist countries is the role played by the Communist Party, which assumes paramount power in society.Using large-scale surveys held in five former Soviet-bloc countries (Bulgaria, the Czech Republic, Hungary, Poland and Slovakia), the authors find that even four years after the fall of communism, parental affiliation with the Communist Party can still increase children's high-culture participation.

Inequalities of parenting in China since 1978
As a basis for a theoretical synthesis, the distinctive roles of education and Communist Party membership played in the Chinese social stratification process have been emphasised by several scholars (Walder et al., 2000;Zhou et al., 1996).Recent research suggests that the transition from the state-socialist economy to the marketoriented economy promotes a shift from traditional to modern parenting values in the sense that the market-oriented social environment rewards creativity, initiative-taking and competitiveness, altering the perception of obedience and humbleness, causing Chinese parents to be more inclined towards a democratic parenting style favouring children's individuality, originality and competition (Wang, 2014).This leads us to consider how parental background resources affect parenting styles in the current context of China.
Deeply influenced by Confucianism and the imperial examination system, Chinese families have had a long tradition of attaching great importance to children's education as a means of achieving upward social mobility.In the course of economic transformation, education becomes one of the most important determinants of elite occupations (Walder et al., 2000).The market-oriented reforms provide new opportunities and alternative avenues for getting ahead in the market sector, contributing to rising returns to education in the reform era (Bian & Logan, 1996), and this is also true in the state sector where individuals with college education were more likely to be recruited by government agencies (Zhou et al., 1996).However, beginning in 1998, the large expansion of higher education has brought increasing proportions of the population into universities, devaluing the lower and intermediate levels of educational certificates, and mounting the pressure for yet higher levels of educational attainment, as Collins revealed in his Credential Society (1979).Consequently, parents constantly adjust their parenting styles to aid achievement in response to the potential credential inflation resulting from educational expansion and the ever-growing competition in the school system.The role of high-stake examinations and fierce competition in the school system may help explain why Chinese parents are increasingly adopting achievement-oriented intensive parenting, as reflected in the evergreater investment in time, effort and money in children's education and development.For example, in order to improve their children's chances of gaining admission to an elite university, Chinese parents not only spend a lot of time helping children with their homework but also spend a lot of money enrolling children in expensive private tuition courses.Until recently, classrooms in Chinese schools were filled with students in the evenings and during weekends, with commercial organisations recruiting existing teachers to do cram teaching and making huge profits, which put a great deal of extra pressure on students, brought a heavy financial burden to parents, and greatly exacerbated the social inequality between the rich and the poor. 1  While most existing research focuses on parental class, income and education in explaining parenting differences, we give particular note to the role of parental political resources in shaping parenting styles, and we take parents' membership in the Chinese Communist Party (CCP) as an indicator of political resources.The Party officials often occupy higher positions, which are directly linked with power, and with the control and command of superior resources in China's socio-economic structure.The CCP assumes the predominant leadership position in the country, with over 95 million members.Most (if not all) of the political apparatuses and socio-economic institutions are headed by CCP members.With the recruitment into the Party being highly selected and strictly screened, the CCP has enlisted most of the elite members in society, who are in turn rewarded with more socio-economic advantages relative to non-members.The CCP also attaches great importance to cultivating Party members and taking strict steps to enforce Party discipline: for example, urging members to be strict with themselves and tackling hedonism and extravagance.In this sense, Party affiliation is not just a matter of capital, but also a matter of normative alignment, both of which could influence parenting norms and practices.It is the interweaving of the possession of superior resources and the holding of specific beliefs and dispositions that can be explicitly or implicitly passed on from one generation to the next within the politically loyal family, that seems to be most important in influencing Party members' parenting styles, as we are going to see.

Dataset
The data for this study are taken from the second wave of the China Education Panel Survey (CEPS, 2014(CEPS, -2015)).The survey adopts a multistage design with probability proportional to size (PPS), starting with two cohorts -the seventh and ninth graders in the 2013-2014 academic year.The baseline survey randomly selects a school-based, nationally representative sample of 19,487 students (10,279 seventh graders and 9208 ninth graders) in 438 classes of 112 schools in 28 county-level units in mainland China.In the follow-up survey conducted in the 2014-2015 academic year, it tracked 9449 out of the 10,279 grade 7 students in the baseline survey, who were now in grade 8, with a followup rate of 91.9%.The analysis of the present study is based on a sample of 9449 students aged between 12 and 18 in the 2014-2015 academic year. 2  Because of item non-response, we use multiple imputation with chained equations to handle the problem of missing data and produce 20 imputed datasets in Stata (Royston et al., 2009).We impute all missing data except for the dependent variables (von Hippel, 2007), resulting in the varying sample size, with 9449 for the analysis of the effect of family background resources on parenting styles and fewer than 9449 cases for the analysis of the effect of parenting styles on children's outcomes. 3Descriptive information, including the percentage missing of all variables in the unimputed sample, is displayed in Appendix A (see Online Supplementary Material).

Measures
Parenting styles.In order to identify parenting styles in contemporary China, three dimensions of parenting -demandingness, responsiveness and involvement -were included as inspired by theoretical expectations by leading scholars as discussed above. 4 Demandingness was measured by three items, including two questions to the children: (1) 'Do your parents care and are they strict with your behaviour at school?' (the response categories are: 'they don't care', 'they do care about it but are not strict', 'they are very strict about it'); (2) 'Do your parents care and are they strict with your Internet access time?' (the response categories are: 'they don't care', 'they do care about it but are not strict', 'they are very strict about it'); and one question to the parents: (3) 'When having different opinions, do you usually persuade or force your child to agree with you?' (the response categories are: 'yes', 'no').Likewise, responsiveness was measured by three items regarding verbal discussion or communication, dining together and intimacy, including one question to the parents: (1) 'How often do you discuss his/her worries and troubles with this child?' (the response categories are: 'never', 'sometimes', 'often'); and two questions to the children: (2) 'How often do you have dinner with your parents?' (the response categories are: 'never', 'sometimes', 'often'); (3) 'How close are you to your parents?' (the response categories are: 'not close', 'not too close nor too far', 'very close').Finally, involvement was measured by time spent on organised leisure activities, time spent on schoolwork and parental intervention in school; namely, two questions to the children: (1) 'How much time on average did you spend on extra-curricular activities on weekends?' (the response categories are: 'never or less than 2 hours', 'about 2-4 hours', 'more than 4 hours'); 5 (2) 'How often did your parents check up on your homework last week?' (the response categories are: 'never', 'one or two days', 'greater than or equal to three days'); and one question to the parents: (3) 'How many times have you contacted your child's teacher at school this semester?'(the response categories are: 'never', 'once', 'two times or more'). 6  Parental resources.Our main explanatory variables pertain to parental social, cultural and political resources, which are captured by measures on parents' class position, education and Communist Party affiliation, respectively.Regarding parental class, we adopt a fiveway schema based on standard practice in class analysis in China (Li, 2021;Zhao & Li, 2019): (1) higher and lower levels of professional and managerial salariat, (2) routine non-manual, (3) self-employed with or without employees, (4) lower-grade technician, skilled and unskilled manual workers in industry and commerce, and (5) agricultural workers.We coded the parental class by using father's or mother's class position, whichever is higher.
In line with Buis (2013), parental cultural resources are indicated by parents' education, using the higher of either parent's educational attainment.And this variable was coded into three categories: low (lower middle school or below); higher secondary (high school or lower technical, called zhongzhuan in Chinese); and tertiary (first degree or above).
Parental political resources are based on parents' CCP memberships, which are crossclassified into four categories: both CCP, father CCP, mother CCP and neither CCP. 7 These key variables are, of course, intricately related.For instance, 42.7% of mothers and 45.3% of fathers who are CCP members are in professional-managerial salariat positions, as compared with 5.0% and 10.3% of non-CCP mothers and fathers; and 61.4% of the families where parents are CCP members are in salariat positions, as compared with 12.0% of the non-CCP families.Over the last few decades, the CCP has attached great importance to recruiting highly-educated members -university graduates -who are then promoted into leadership positions.
Outcome variables.The outcome variables in the analysis were academic achievement, depression and risky behaviour.Academic achievement comes from the average test scores in three subjects: Chinese, mathematics and English.Depression is based on children's responses to 10 items concerning the frequency of having the following negative feelings in the last week.The sub-questions include 'feeling blue', 'too depressed to focus on anything', 'unhappy', 'not enjoying life', 'having no passion to do anything', 'sad or sorrowful', 'nervous', 'excessive worry', 'feeling something bad will happen' and 'too energetic to concentrate in class'.The response categories include 4 = always, 3 = often, 2 = sometimes, 1 = seldom and 0 = never.Risky behaviours include 10 items that assess the frequency of such behaviour as 'cursing or saying swearwords', 'quarrelling with others', 'having a fight with others', 'bullying the weak', 'having a violent temper', 'unable to concentrate', 'skipping classes, being absent or truanting', 'cheating in exams', 'smoking or drinking alcohol' and 'going to net bars or video arcade'.Each item was measured on a five-point scale (4 = always, 3 = often, 2 = sometimes, 1 = seldom and 0 = never).Total scores for these two variables were calculated by adding up the ratings for their respective items, with higher values denoting higher depression and risky behaviour.
Controls.We controlled for several confounding variables that are likely to have an impact on parenting styles and child outcomes: gender, hukou, age, ethnicity, migration status, number of siblings, family structure, frequent parental quarrels, presence of grandparents in the household and provision of grandchild care.Gender was measured by children's sex (male = 1, female = 0).Hukou was measured by the household registration type (rural = 1, urban = 0).Ethnicity was measured by ethnic identity (minority ethnic groups = 1, Han = 0).Migration was measured by children's migration status (migrant = 1, local = 0).Number of siblings was measured by the number of brothers and sisters.Family structure included four groups:(1) living with mothers and fathers; (2) living with mother; (3) living with father; (4) both parents absent.Frequent parental quarrels indicate whether parents quarrel a lot (yes = 1, no = 0).The presence of grandparents indicates whether grandparents live in the same household with the child (yes = 1, no = 0).Provision of grandchild care indicates whether grandparents take care of the child's daily life (yes = 1, no = 0).We also controlled for cognitive ability, which has been shown to affect child outcomes, especially academic achievement (Yang & Zhao, 2020).Cognitive ability was measured by the standardised cognitive test scores that were computed by the CEPS based on the three-parameter logistic (3PL) item response theory model.

Analytic plan
The first task in our study is to identify distinct patterns of parenting practices.We do so by using latent class analysis (LCA), which groups individuals based on similar patterns of individual characteristics by an array of observed variables.LCA posits that the phenomenon of interest might be considered as a categorical latent variable instead of a continuous one. 8Next, we explore the association between measures of parental background resources and parenting styles using multinomial logistic regression models accounting for clustering by schools.Finally, we introduce a set of school-fixed effects to estimate the effects of parenting styles on a wide range of child outcomes, including standardised test scores, depression and risky behaviour.The fixed effects contain any observed and unobserved school characteristics that affect child outcomes, allowing us to control for differences in school quality and eliminate bias that originates in the school context. 9

Identifying typologies of parenting styles
We use latent class analysis (LCA) to identify the typology of parenting styles among Chinese families.To determine the optimal number of groups, we started by one latent class and compared the model fit indices and statistics iteratively.Table 1 shows the degrees of freedom, test statistic G 2 , p-value of the Lo-Mendell-Rubin likelihood ratio (LMR LR) test and p-value of the adjusted Lo-Mendell-Rubin likelihood ratio (ALMR LR) test and information criterion indices.The G 2 statistic expresses the correspondence between observed and predicted response patterns.The significance test is in favour of the four-class model because the p-value of both the LMR LR and the ALMR LR tests that compare the four-class model with the five-class model becomes statistically insignificant (p >= 0.05) in the five-class model, suggesting no more significant improvement in model fit by including an additional class into the model and thus the four-class model is preferred.Also, the statistic of the Bayesian information criterion (BIC) suggests that the four-class approach is superior as it has the lowest value.Thus, considering the model fit statistics as well as the meaning arising from the LCA membership classification, we decided to adopt the four-class model.
When individuals were classified into a four-class membership, typologies of parenting styles were defined according to the pattern of conditional item-response probabilities.Table 2 reports the estimated relative size and the conditional probability of the latent classes.In terms of the relative size, the four latent classes account for 16, 39, 33 and 12% of the sample respectively.Turning to the conditional item-response probabilities, parents in latent class 1 appear to be very strict with children, set strict time limits for Internet use, and may even force children to obey; however, they do not communicate much, nor do they seem to have intimate relationships, with their children, suggesting an authoritarian parenting style.Parents in category 2 often discuss children's worries or troubles, dine and have close relationships with children, but exercise rather limited parental discipline and control.This category is labelled as permissive parenting.Category 3 is associated with 'high demandingness', 'high responsiveness' and 'high involvement'.These parents are achievement-oriented and are more likely to send children to various extra-curricular activities, check their schoolwork and interact with educators.This parenting style is called intensive parenting. 10The scores for latent class 4 are low on all items -a parenting style which is rather neglectful.

Socio-economic and political differences in parenting styles
We first have a look at the association between parental class, education and CCP membership on the one hand, and parenting style on the other, as shown in Table 3.It is evident that parental class, education and political status are strongly associated with parenting styles.For example, 42% of salariat parents adopt the 'intensive' style and only 7% adopt the 'neglectful' style.By contrast, parents in manual working positions are only half as likely to pursue an 'intensive' style (at 22%) but are nearly three times as likely to be 'neglectful' (at 20%).Parents in agricultural positions are similarly disadvantaged as manual workers in cities.
It is noticeable that parental education and Party affiliation have similar impacts as does parental class, with gaps in terms of percentage-point differences between the most advantaged and the most disadvantaged groups within each domain being of similar magnitudes.For instance, parents with degree-level qualifications are twice as likely as those with less than high school qualifications to have intensive parenting styles (at 49 and 25% respectively), and parents with both having CCP membership are also around 20 percentage points more likely to have an intensive style than those where neither has CCP membership (at 45 and 27% respectively).The foregoing analysis shows the importance of taking into account parental political affiliation as well as parental class and education.The analysis is, however, at the gross level, and we wish to see whether each of the three domains would have important effects net of other structural and demographic factors.We conduct more refined analysis below, with results shown in Table 4.
We conduct a multinomial logistic regression accounting for clustering by schools to assess the relative impacts of key factors of chief sociological importance on different types of parenting. 11Table 4 shows some important features.Firstly, family social, cultural and political differences are not only shown in parenting styles in general, but in intensive parenting in particular.Compared to neither parent being a CCP member, parents both affiliated with the Party are more likely to adopt intensive (rather than authoritarian, neglectful and permissive) parenting.And respondents whose fathers are members of the Party but whose mothers are not Party members are more likely to pursue intensive and authoritarian parenting than permissive and neglectful parenting.Moreover, parents' education seems particularly meaningful.Compared with parents who have less than a high school degree, well-educated parents (having a high school degree or above) are more likely to have intensive rather than authoritarian, permissive or neglectful parenting practices.Our results show the clear importance of parental education and political resources in predicting intensive parenting.It is also striking that Communist Party affiliation impacts parenting styles in a gendered way.While paternal Party membership alone is significant in predicting intensive and authoritarian parenting, maternal Party membership alone does not have any significant impact.These results are in line with the   fact that male CCP members are often in leadership roles in the country's socio-political hierarchy.It is also interesting to note that, among the four categories of Party membership, families with both parents being Party members are more likely to adopt intensive parenting than the other three categories.It is possible that in these families, the political capital of two CCP parents facilitates access to various kinds of resources and creates an affinity for child-rearing practices.Furthermore, from Table 4 it is clear that parents' occupational class is of somewhat lesser importance in predicting intensive parenting than are parental education and CCP membership.Other things being equal, many of the parameters for parents' occupational class are statistically non-significant.This finding seems at odds with the prevailing results in existing research on parenting strategies in Western countries, particularly in the United States, where it is often found that intensive parenting is more prevalent in middle-class families.Since most of these studies are qualitative in nature (e.g.Hays, 1996;Lareau, 2003) and since US researchers do not often use fine-grained occupation-based class measures in quantitative studies (Dornbusch et al., 1987;Ishizuka, 2019), whether occupational class in Western countries plays a more important role in shaping intensive parenting than in China remains to be tested, but it is fairly safe to say that Lareau's qualitative findings cannot be well translated to China because the effects of parents' occupational class are not the sole determinants of intensive parenting.Another interesting result is that the parents' occupational class also appears relevant in predicting permissive parenting.More specifically, those with parents in either salariat or routine non-manual occupations were more likely to adopt permissive rather than authoritarian and neglectful parenting practices when compared to parents in the manual working class.Therefore, we have found some evidence that class positions still, to varying degrees, impact parenting styles in contemporary China.
Finally, we obtain important findings for the individual characteristics.For example, non-intact families are more likely to be authoritarian, permissive or neglectful rather than intensive.This result does not differ substantially from the result in the United Kingdom (Chan & Koo, 2011), which implies that the presence of both parents is a key factor determining a child-centred parenting approach.Additionally, a child's gender influences parenting styles in the sense that parents are more likely to be authoritarian and intensive and less likely to be permissive and neglectful when their child is male instead of female.This could be attributed to the Chinese son preference, which indicates that boys are still given more importance than girls even after 40 years of reform and opening up, and Chinese parents are more likely to be more demanding, responsible and cultivating when they have boys.It is also shown that the sibship size impacts parenting styles in the sense that fewer siblings can significantly predict intensive and permissive parenting.In terms of the household registration system (hukou), it can be seen that rural parents are more likely to adopt authoritarian parenting compared with urbanites.Considering the rural-urban divide in parenting styles revealed in previous qualitative research (Lu & Chang, 2013), our results further indicate that rural parents in China are still more likely to preserve parenting strategies that are consistent with traditional values and beliefs, such as valuing strictness and control.

Implications of parenting styles
Our final set of analyses examines whether parenting styles affect children's academic, psychological and behavioural outcomes.Table 5 displays the school fixed effects results with the standardised test scores, depression and risky behaviour as dependent variables.
The results indicate that parenting styles matter a great deal in determining these outcomes: compared with neglectful parenting, the other three parenting styles are significantly related to an increase in standardised test scores and a decrease in depression and risky behaviour.Specifically, intensive parenting can greatly improve academic achievement and alleviate depressive symptoms and risky behaviour, followed by permissive, authoritarian and neglectful parenting.

Sensitivity analyses
We conduct several sensitivity analyses to check the robustness of our empirical findings.Firstly, the measurement of parenting styles is constructed with six child-reported indicators and three parent-reported indicators. 12Given that the 'parent questionnaire' has not always been filled in by parents, and 8.7% are 'other adult guardians' in the household, we rerun latent class analysis excluding those who were not parents.The results are consistent with those of the full sample, revealing the same story (see Appendix C in Online Supplementary Material).Secondly, CCP membership is closely intertwined with social class and education, which may result in multicollinearity in the multinomial logistic regression model. 13The same model is used without including political affiliation given the high levels of correlation.The omission of this variable did not substantively change the point estimates of parental class and education (the results displayed in Appendix D).Finally, we apply the Inverse Probability Weighted Regression Adjustment (IPWRA) approach while accounting for clustering by schools to estimate the average treatment effect (ATE) of parenting styles on child outcomes when all control variables are included.Appendix E displays the IPWRA estimates of the ATE of parenting styles on children's academic achievement, depression and risky behaviour.The results confirm that intensive parenting has the greatest beneficial effect on academic achievement and the strongest alleviative effect on depression and risky behaviour, followed by permissive, authoritarian and neglectful parenting. 14

Conclusion and discussion
For decades, sociologists have been concerned with the social distributions of parenting styles.This work contributes to existing knowledge of parenting styles by identifying the typologies of parenting styles and studying the effects of parental socio-cultural and political resources on parenting in contemporary China.Our main results can be summarised as follows.Firstly, we have been able to identify parenting practices as structured by socio-cultural political resources in China.We find that parental class, education and CCP membership are important and salient predictors of intensive parenting.Our findings show that parents in salariat position, with higher education and with CCP memberships, tend to adopt intensive parenting styles, which fits well with previous findings that socioeconomic and cultural resources can be passed on from one generation to the next via socialisation and interaction (Bourdieu, 1973), but our analysis is the first to bring all three structural factors into simultaneous analysis and has found significant political effects over and above all other factors in the model.
What is of particular note is our finding that parental CCP membership plays a unique and salient role in parenting practices, as shown in Table 4, and that even with all other factors being considered in the models.Given the well-known selectivity of CCP membership and the changing composition of intellectual and technical elites within the CCP (Dickson & Rublee, 2000), one can expect CCP members to command socio-economiccultural resources well beyond those of non-members in similar occupational positions, and it is these politically-underlined resources that enable them to make fuller use of available opportunities in nurturing their children's all-round development through both intensive and extensive involvement in education, such as spending more quality time interacting with children.To paraphrase Kahneman (2011), CCP members in leading professional-managerial positions would resort to whatever resources in their possession to try to prevent intergenerational déclassement and to invest in child's human capital, more so than in Western countries.Our analysis clearly shows the social stratification of parenting practice in China due to China's unique socio-political institutions.
Secondly, our analysis provides an interesting contrast with Lareau's (2003) qualitative findings by revealing that intensive parenting in contemporary China is jointly shaped by parental class, education and political affiliation instead of class playing a predominant role.This is because education and Communist Party membership act as 'gatekeepers' to higher income and occupational status in the post-reform era (Bian & Logan, 1996;Walder et al., 2000).This suggests that in China, class distinctions may mostly stem from cultural and political transmission, at least in predicting intensive parenting.Our findings give us a deeper understanding of how parenting styles reflect the deeply entrenched socio-economic-political differences that characterise contemporary Chinese society in a different way from Western societies.
Finally, this study provides evidence that intensive parenting is associated with better academic performance, lower depressive symptoms and less risky behaviour.This suggests that investing in an intensive child-rearing environment is conducive to achieving desirable child outcomes.These findings are attractive because they suggest that parental efforts put into raising their children and their time, resources and knowledge inputs can be converted into human capital and favourable outcomes that are helpful in ensuring future social and economic status.It is not parental background per se but parenting styles also matter (see Appendix F in Online Supplemental Material).
Overall, our study presents a comprehensive analysis of the underlying factors in parenting styles in China.To the best of our knowledge, this is the first systematic study which identifies intensive parenting as a particularly prominent feature of socio-economic-political inequality in China.For policy implications, the intensive parenting culture may induce excessive competition among parents, reduce young people's willingness to get married, lower the birth rate, and exacerbate social divisions.Although China announced in 2016 the end of the one-child policy that had lasted for 35 years, recent research has shown that the fertility intentions for a second child are very low (Jiang et al., 2015).With intensive parenting, parents will have to compete with one another in making ever greater investments in both emotional and financial resources to ensure that their children have the best attainment possible.Therefore, it is of great importance for the government to take measures that can ease parents' and children's pressures and reduce excessive competition.Another concern for policymakers is that class-aligned differences in parenting styles may contribute to a worsening of social inequality.Policies should help the most disadvantaged groups.We notice that the Chinese government recently banned extra-curricular fee-paying classes, which is a welcome development.Yet even more important would be to take measures that can directly target socio-economic inequality at root.Unequal conditions generate unequal opportunities, which produce unequal outcomes.Parenting practices are a reflection of current inequalities and a harbinger of future inequalities.
1.This practice was recently banned by the Government.2. Further analysis shows that 88% of the students were aged 13 or 14; 2% were aged 12; 8% were aged 15-16.And there were six students aged 17, and two students aged 18 (accounting for 0.01%).Among the eight oldest students, seven were from the rural areas, and their parents were farmers or unskilled manual workers.China is a vast country with uneven socioeconomic development.A small number of outliers should not be highly surprising.3. We ran sensitivity analyses using imputed values for the dependent variables in our models.
Results are consistent with models deleting cases missing on the dependent variables (the table is available upon request).4. Parental demandingness refers to the way in which parents set limits for the child's behaviour and show a willingness to confront the child who disobeys; parental responsiveness refers to the extent to which parents are sensitive to children's special needs and demands by being supportive, communicative and acquiescent (Baumrind, 1991); and notably, parental involvement is introduced to refer to parents' willingness to take an active role in children's development and education by incorporating various organised activities, supporting or influencing their learning process and, when necessary, interfering with or even intruding into their school life. 5. CEPS asks students 'how much time on average did you spend on extra-curricular activities' including, 'on cram school courses' and 'on hobby classes'.We added up the two items and generated our score for the time spent on extra-curricular activities.6. Descriptive statistics for the indicator variables used in the LCA are reported in Appendix B in Online Supplementary Material.7. A reviewer commented that it is important to consider parents' position in the CCP hierarchy, which affords the parents with differential political capital.However, no information is available on the levels of the Chinese bureaucratic hierarchy in the dataset, and we are unable to distinguish high-ranking and low-ranking cadres.We thank the reviewer for the insight and will conduct further research when relevant information becomes available.8.The model can be specified by exp( ) (exp( ) 1 where units are assumed to belong to one of C discrete classes c = 1, . .., C. The prior probability that a unit j is in class c, π jc , is a model parameter.If unit j is in class c, the conditional response probability that item i takes on the values a s , s = 1, . .., S i , is modelled as a multinomial logit.9.The primary empirical specification for school fixed effects can be written as Y P X e ij j i j i j i j = + + + α β1* where Y ij refers to the academic achievement, depression and risky behaviour of student i in school j and α j represents school fixed effects, which captures the combined effects of omitted school characteristics, and β 1 in this model can be interpreted as the overall degree of parenting differences in child outcomes when incorporating the influence of school characteristics.Pij refers to parenting styles.X ij refers to a vector of covariates of student i in school j and e ij is the residual.10.It is noticeable that intensive parenting scores high on all items except for valuing obedience.This is consistent with Lareau's qualitative observations on concerted cultivation.According to Lareau (2003), parents who engage in this style of parenting are less likely to use clear directives.11.The robust variance estimator is used to allow for intra-school correlation.12. Due to data constraints, we had to use child-reported measures with parent-reported measures to construct parenting typologies.Parents' evaluations of their children are more likely to be subject to parental social desirability bias.We await new data based on child-reported measures only in constructing parenting typologies in future.13.A closer inspection shows that the variance inflation factors (VIFs) are below the suggested cutoff point of 2.5, which indicates that multicollinearity may not be a concern in the multinominal logistic model.14.One of the reviewers commented that it is necessary to use the seemingly uncorrelated regression model (SUR) to estimate the effects of parenting style on children's outcomes because there might be unobserved factors that affect all three outcome variables (they might be highly correlated with each other), and residuals in the three separate models are correlated.

Table 1 .
Fit statistics for latent class analysis.

Table 2 .
Estimated relative size and conditional probability of the latent classes.

Table 3 .
Distribution of parenting styles by parental class, education, CCP membership and economic condition (row per cent within each variable).

Table 4 .
Odds ratios for multinomial logistic regression of parenting styles on independent variables.

Table 5 .
Fixed effects estimates of the impact of parenting styles on academic, mental and behavioural outcomes.Controls include parental class, education, parents' CCP membership, gender, hukou, age, ethnicity, migration status, number of siblings, family structure, frequent parental quarrels, presence of grandparents in households, grandparents providing care and cognitive ability.Standard errors in parentheses.