https://orcid.org/0000-0003-1136-6852
-
PDF
- Split View
-
Views
-
Cite
Cite
Yifei Hou, J Jill Suitor, Childhood Parental Disfavoritism and Chinese Adults’ Psychological Well-Being in Middle and Later Life: The Moderating Effect of Gender, The Journals of Gerontology: Series B, Volume 78, Issue 12, December 2023, Pages 2102–2110, https://doi.org/10.1093/geronb/gbad138
Close - Share Icon Share
Abstract
Parental differential treatment of children, particularly disfavoritism, has been found to detrimentally affect adult children’s psychological well-being in the United States. However, no study has investigated the long-reaching influence of parental disfavoritism in China, where there is an absence of equal treatment norms. Drawing from theories of social comparison, life course, and gender dynamics in China, we tested how perceptions of childhood parental disfavoritism affect midlife and older Chinese adults’ depressive symptoms, and how the effects differ by own and parent’s gender.
Random-intercept models were used based on a sample of 17,682 midlife and older Chinese adults, drawn from 5 waves of China Health and Retirement Longitudinal Study.
Recollections of childhood parental disfavoritism were associated with higher depressive symptoms among Chinese adults. Perceptions of paternal disfavoritism predicted both men’s and women’s depressive symptoms, whereas perceptions of maternal disfavoritism predicted women’s depressive symptoms only. Paternal disfavoritism was more detrimental than maternal disfavoritism, but only for men. Maternal disfavoritism was more detrimental for women than men.
These findings shed light on the universality of the long-reaching detrimental effect of perceptions of parental disfavoritism across cultures as well as the unique gendered patterns in China shaped by patriarchy. Findings suggest that the implementation of Three-Child Policy in China should be accompanied with parental education programs involving fathers on equal treatment of children.
Perceiving parents treating siblings better than self is a bitter emotional experience. Scholars have coined this phenomenon “parental disfavoritism,” which, together with parental favoritism, reflects the perceived hierarchical position of siblings in parental differential treatment (PDT; Suitor et al., 2016). Studies conducted in the United States and Europe have well-documented that PDT, particularly disfavoritism, is associated with children’s lower psychological well-being both in their childhood and adulthood (Boll et al., 2005; Jensen et al., 2013; Suitor et al., 2017). Recently, two studies adopting the life course perspective have found that recalled PDT continued to affect children’s well-being in adulthood (Davey et al., 2009; Peng et al., 2018).
Unlike societies where social norms encourage parents to treat children equally (Parsons, 1942), PDT has not been stigmatized as much and is more culturally tolerated in China (Barrett Singer & Weinstein, 2000). To date, however, little is known about whether PDT, particularly disfavoritism, induces similar long-term detrimental effects in China. Furthermore, even less is known about how gender as a social structural factor shapes the long-term effect of parental disfavoritism, despite a large body of literature revealing the essential role of gender in shaping family relations in China (Xie, 2013).
Understanding the long-term impact of PDT, particularly disfavoritism in China, however, is particularly timely as multichild families become the ideal type of family advocated by the policy. Research in the United States (Kong et al., 2022) has suggested that poor parental treatment in childhood may lead to challenges to optimal care of parents in later life, in part through its effect on adult children’s psychological well-being (Peng et al., 2018). Thus, studying the impact of parental disfavoritism on Chinese adults’ well-being is an essential first step in improving the mental health of the next generation who serve an important role in supporting older parents in their middle and later adulthood.
To advance knowledge in these aspects, we asked the following questions: (1) Is parental disfavoritism recalled from childhood associated with increased depressive symptoms among midlife and older Chinese adults? (2) How does the effect depend on own and parent’s gender? To address these questions, we used data from China Health and Retirement Longitudinal Study (CHARLS) 2011–2018.
Background
Childhood Parental Disfavoritism and Adults’ Psychological Well-Being
Across the past four decades, a large body of research in the United States and Europe has documented the detrimental effect of PDT, particularly disfavoritism, on children’s psychological well-being (Boll et al., 2005; Harris & Howard, 1985; Jensen et al., 2013; Suitor et al., 2017). According to theories of social comparison and equity, individuals compare themselves to others to gather information. Specifically, perceiving oneself as inferior or under-benefited relative to others results in unhappiness (Festinger, 1954; Suls & Wheeler, 2012).
Recently, two studies in the United States have extended this line to study the effect of recalled childhood PDT on well-being in adulthood. By calculating the deviation of each child’s recalled quality of maternal and paternal from the family mean, Davey and colleagues (2009) constructed indirect measures of PDT and found that adult children who recalled maternal affection lower than the family average reported lower positive affect and higher negative affect, whereas no effect of within-family differences in relationships with fathers was found (Davey et al., 2009). Although this study revealed the long-term impact of within-family differences in parent–child relations, some scholars argue that the indirect measures of parenting differences could hardly capture children’s direct perceptions of favoritism/disfavoritism, especially when children acknowledge that these parenting differences respond to their different needs (Jeannin & Van Leeuwen, 2015). Using more direct measures, Peng and colleagues found that recalling that mothers tended to “play favorites” among children in childhood predicted adult children’s depressive symptoms (Peng et al., 2018). However, both studies have focused on the United States, and neither has measured the target children’s direct perception of parents’ disfavoritism towards themselves in childhood. Therefore, these studies could not shed light on the long-term effect of childhood parental disfavoritism on adults’ well-being in China.
Up to now, only three studies have examined the effect of PDT in China. Two studies in Hong Kong have found that PDT adversely affected adolescents’ psychological well-being (De Man et al., 2003; Ng et al., 2020). Using a sample from Henan province, Luo and colleagues found that parental favoritism predicted the favored young children’s psychosocial problems (Luo et al., 2020). These studies, however, all captured the effect in children’s early life, and therefore, could not shed light on the persistence of the effect of PDT on well-being in adulthood.
There are reasons to anticipate that childhood parental disfavoritism will have similar detrimental effects on Chinese adults’ well-being, and the effect could last long enough into children’s adulthood. Studies have documented that social comparison and psychological responses to injustice are similar or possibly even more severe in China, due to the more collective orientation in China than in the United States or Europe (Kim et al., 2015). Furthermore, both the principle of life span development of the life course theory (Elder, 1974; Thomas & Znaniecki, 1918) and a growing body of studies have confirmed the long-reaching influences of childhood adversity on psychological well-being in adulthood in China (Xiang & Wang, 2021; Yang et al., 2020). Thus, childhood parental disfavoritism should be seen as a within-family form of childhood adversity that affects children’s mental health via social comparison with siblings (Festinger, 1954; Suls & Wheeler, 2012). It distinguishes from other forms of adversity, such as childhood abuse and household dysfunction (e.g., Felitti et al., 1998), which only capture differences in childhood experiences between isolated individuals or between-families. Such within-family comparison processes could have long-term effects on children’s mental health through perceptions of unfairness, resulting in anger (Tabibnia & Lieberman, 2007). In addition, parental disfavoritism also increases sibling tension (Peng et al., 2018), which could have long-term implications for children’s psychological well-being. Therefore, we propose that childhood parental disfavoritism will be associated with children’s lower psychological well-being in China, and the effect is likely to last well into adulthood.
Therefore, we hypothesized that:
H1. Perceptions of childhood parental disfavoritism are associated with midlife and older Chinese adults’ higher depressive symptoms.
The Moderation Effect of Own Gender
Classic theories of gender role development (Chodorow, 1978; Gilligan, 1982) have argued that women are socialized beginning in childhood to be especially sensitive to others’ emotions. Consistent with these arguments, empirical studies in the United States have found that women, relative to men, are both more involved in their social relations and are affected more intensely by those relationships (Birditt et al., 2009). Given that research has shown similar gender norms in China emphasizing women’s compliance and sensitivity (Hu, 2015), it is likely that Chinese women are more sensitive to childhood parental disfavoritism than men. Thus, we suggest that childhood parental disfavoritism may have a greater effect on Chinese women’s than men’s depressive symptoms.
However, affect control theory, together with patrilineal and patrilocal norms in China, can be used to propose an alternative hypothesis. Affect control theory posits that receiving less reward than expected results in negative emotions (Heise, 1987). China has been well-known for its preference for sons over daughters (Das Gupta et al., 2003). Sons carry the family name, whereas daughters are regarded as joining their husbands’ families after marriage. Thus, the son preference norm could lead men to believe that they should be equally valued compared to their brothers and should be treated more favorably relative to their sisters. In addition, sons’ continued interaction with their parents may make the pain of their childhood parental disfavoritism constantly hurt, whereas daughters “marry out” to their husbands’ families and the influence of their childhood parental disfavoritism may decrease over time. Thus, it is possible that childhood parental disfavoritism may have a greater effect on Chinese men’s than women’s depressive symptoms. Therefore, we proposed two alternative hypotheses:
H2a. Perceptions of childhood parental disfavoritism are associated with more depressive symptoms among women than men.
H2b. Perceptions of childhood parental disfavoritism are associated with more depressive symptoms among men than women.
The Moderation Effect of Parent’s Gender
The literature on the socialization of women and power dynamics in Chinese families provides the bases for competing hypotheses regarding the moderating effect of parents’ gender.
On the one hand, classic theories of gender have argued that women are socialized to be “kin-keepers,” which leads them to invest more in their social relationships than men (Chodorow, 1978). This is reflected in mothers’ and fathers’ differential involvement with children. Studies in China as well as in the United States have shown that mothers have closer relationships and interact more intensively with children than fathers (Suitor et al., 2011; Xu et al., 2017). In addition to a worldwide norm of “motherhood mandate” associating childrearing as women’s femininity, Chinese women’s role as caretakers is also facilitated by the traditional patriarchal ideology and the unique way women embody it, viewing childrearing as women’s gender resources (Zuo & Bian, 2001). Because Chinese mothers provide more important interpersonal resources for children than fathers, it is possible that maternal disfavoritism is more detrimental than paternal disfavoritism.
On the other hand, theories of power and social influence suggest a different picture (Emerson, 1962). Chinese families are traditionally known as patriarchal. Despite socialist reformation enabling Chinese women to participate in education and labor force, the power dynamics between couples are still very traditional. Fathers bring more economic resources and are more authoritative in the family than mothers (Xie, 2013). Thus, it is possible that paternal disfavoritism carries more weight in affecting children’s well-being due to fathers’ control over socioeconomic resources that would be valuable to adult children. In addition, studies have shown that Chinese fathers are less caring and more likely to adopt authoritarian parenting than mothers (Shek, 2000; Xu et al., 2017). Thus, it is possible that paternal disfavoritism could take harsher forms and be implemented with fewer objections compared to maternal disfavoritism, the effect of which might be softened by the closer mother–child relations.
Thus, we propose alternative hypotheses:
H3a. Perceptions of childhood maternal disfavoritism are associated with midlife and older Chinese adults’ higher depressive symptoms than paternal disfavoritism.
H3b. Perceptions of childhood paternal disfavoritism are associated with midlife and older Chinese adults’ higher depressive symptoms than maternal disfavoritism.
Interaction Between Own and Parent’s Gender: Gender Similarity
In addition to the arguments proposed above, it is also likely that the effect of childhood parental disfavoritism on Chinese adults’ well-being will be stronger when there is a match between own and parent’s gender.
Theories of social learning and identity formation have argued that children identify with same-gender parents to develop gender-appropriate behaviors (Bandura & Walters, 1977). Consequently, parents who share the children’s gender play a particularly essential role in children’s identity development and psychological well-being. This gender-based identification can make disfavoritism from parents who share the children’s gender particularly harmful, because children are more sensitive to treatment from parents who share their gender and that disfavoritism from them poses greater difficulty for the children’s identity formation. Consistent with this theory, research in the United States has found that the detrimental effect of parental disfavoritism is more pronounced among same-gender parent–child dyads (Harris & Howard, 1985).
Furthermore, literature has shown that children’s need to identify with same-gender parents to develop gender-appropriate behaviors is similar across cultures (Bandura & Walters, 1977). Thus, we think gender similarity in shaping the effect of parental disfavoritism is likely to work similarly in China. In fact, the broader literature on children’s and adolescents’ development in China has supported a stronger influence of fathers on sons’ and mothers’ on daughters’ socioemotional development (Li et al., 2014). Given that traditional patrilocal norms emphasize sons’ continued involvement with natal families across the life course, we think the important influence of paternal disfavoritism is likely to continue into sons’ adulthood. In comparison, although daughters are mainly expected to involve with their husbands’ families after marriage according to traditional patriarchal norms, a growing body of literature has revealed the increasingly important connections between married daughters and their natal mothers. For example, older mothers often serve as an important source of support for daughters in caring for grandchildren in China (Chen et al., 2011). Thus, we think that the difference in the effect of parental disfavoritism based on gender similarity found in children’s early years is likely to continue into children’s adulthood. Thus, we hypothesize that:
H4. Childhood parental disfavoritism is associated with midlife and older Chinese adults’ higher depressive symptoms when own and parents’ gender are the same.
Method
Data
We used data from CHARLS. The CHARLS is a biennial longitudinal survey of Chinese residents aged 45 and older, aiming to understand socioeconomic determinants and consequences of aging in China. This cohort has not yet been affected by the birth-control programs in China aiming to control population growth, such as the One-Child Policy implemented in the 1980s. Thus, the multiple-children family setting provides the ideal basis for us to investigate the effect of parental disfavoritism.
The sample of CHARLS was selected using multistage probability proportion to size sampling. These procedures yielded a nationally representative sample of Chinese individuals aged 45 years or older. More detailed information about CHARLS sampling strategy is available elsewhere (Zhao et al., 2014). The first wave was collected in 2011, followed up every 2 years, with the latest wave conducted in 2018. A life-history survey was conducted in 2014. Perceptions of childhood parental disfavoritism were asked in 2014 only, whereas information on depressive symptoms was collected in 2011, 2013, 2015, and 2018. We used multiple waves of data 2011–2018 to fully utilize the longitudinal structure of the survey.
Analytic Samples
We limited our sample to respondents from multichild families who had been raised by two biological parents to eliminate the confounding influences of family structure. Eligible respondents must have participated in at least one wave of the main surveys (2011, 2013, 2015, 2018) and the 2014 Life History Survey.
Although CHARLS sampled respondents aged 45+ as the “major respondents,” their spouses enrolled in the sample automatically, resulting in a small proportion (3.5%) of the original sample younger than 45. Because this small proportion of respondents were not representative of the population at that age, we excluded respondents who were younger than 45.
A total of 11,998, 13,470, 11,364, 16,228, and 14,734 eligible respondents had participated in the 2011, 2013, 2014 Life History, 2015, and 2018 surveys, accordingly. The original data set contained 14.9% missing data. To reduce bias induced by non-response and attrition, we handled missing data with multiple imputations using Stata 17 chained equations. The final analytical sample consists of 17,682 individuals in 11,149 households providing information for 55,682 person-year observations.
Measures
Dependent variable
Depressive symptoms were measured by the Center for Epidemiologic Studies-Depression (CES-D) 10-item scale constructed by Harmonized CHARLS team based on data sets in 2011, 2013, 2015, and 2018. Respondents were asked to rate their frequencies of 10 depressive feelings over the last week: (a) felt depressed; (b) everything was an effort; (c) sleep was restless; (d) was happy; (e) felt lonely; (f) bothered by little things; (g) could not get going; (h) had trouble keeping mind on what is doing; (i) feel hopeful about the future; and (j) feel fearful. Responses to each item were coded 0 = rarely or none of the time, 1 = some or little of the time, 2 = occasionally or a moderate amount of time, and 3 = most or all of the time. Items “was happy” and “fell hopeful about the future” were reverse coded, and responses were summed to create a scale of 0–30 (Mean = 8.26, SD = 6.26, skewness = 0.90, kurtosis = 3.35). The Cronbach’s alpha reliability of the items ranged from 0.76 to 0.81 across waves.
Explanatory variables
The 2014 survey asked respondents “Did your (male/female) guardian treat your siblings better than you when you were growing up, very, a little, somewhat, or not at all?” Because we have limited our sample to respondents raised by two biological parents, these questions collected information on paternal/maternal disfavoritism, respectively. Based on these questions, we constructed two sets of explanatory variables (a) any childhood parental disfavoritism; and (b) disfavoritism by type.
Any childhood parental disfavoritism
About 18.97% of respondents reported that either their father or mother treated siblings better than themselves (i.e., answered “very,” “a little,” or “somewhat” in either question). Because the proportion is relatively small, we combined these categories together and constructed a binary indicator, which was coded “1 = childhood disfavoritism” if the respondent reported either father or mother treated siblings better than themselves (very-somewhat). Otherwise, we coded it as “0 = no disfavoritism.”
Disfavoritism by type
Because paternal and maternal disfavoritism were highly correlated (tetrachoric correlation rho = 0.92), suggesting that there was a high degree of congruence between father and mother in disfavoring the child. To compare the relative effect of paternal and maternal disfavoritism, we constructed a categorical variable of the type of disfavoritism as our explanatory variable, which was coded: 1 = no disfavoritism, 2 = paternal disfavoritism only, 3 = maternal disfavoritism only, 4 = both paternal and maternal disfavoritism.
Moderator
The moderators in this study are own and parent’s gender. The explanatory variables have already measured paternal and maternal disfavoritism separately. In addition, child’s gender was measured as 0 = men, 1 = women.
Control variables
To obtain unbiased estimates, we controlled for the following confounding factors that may affect both parental disfavoritism and psychological well-being.
Time-invariant variables
Because resource scarcity and dilution (Jenkins et al., 2003) predict both PDT and adults’ psychological well-being, we controlled for childhood starvation (1 = yes, 0 = no) and the number of siblings calculated by the number of siblings who survived to age 6 and older. In addition, children’s health status, birth order and gender composition of the sibship have also been found to predict parental differentiation, which may also affect mental health in their adulthood (Harris & Howard, 1985). Thus, we controlled for childhood health condition (0 = same or better; 1 = poorer) constructed from questions asking respondents about childhood health status compared to other children of the same age. We controlled for whether the respondent was firstborn (1 = firstborn, 0 = others) and proportion of sons, both of which were constructed based on respondents’ reported numbers of older and younger brothers and sisters in the family. Because social support may mitigate the effect of childhood adversity (McLafferty et al., 2018), we controlled for childhood friendship (0 = no good friend, 1 = have good friends). Because mental health may transmit generationally, which has also been found to predict parental differentiation (Crouter et al., 1999), we controlled for parents’ mental illness (1 = no abnormality, 1 = abnormality). This variable was constructed based on no respondents’ reported information about whether the female/male guardian had abnormality of mind when they were young. We also controlled for the respondent’s age and educational attainment (1 = secondary schools [grades 6–12] and above; 0 = less than secondary school), which predicted mental health.
Time-varying variables
We controlled for the respondent’s age as a time-varying variable. Because social statuses may predict mental health (Pienta et al., 2000), we controlled for marital status (0 = not married, 1 = married), retirement status (0 = not retired, 1 = retired), and residential area (0 = urban, 1 = rural). Besides, we controlled for respondents’ self-reported health status (ranged 1–5, higher score = poorer health) and health behaviors measured by whether the respondents did light activities (0 = no, 1 = yes) for at least 10 min every week. Finally, we included the wave of the interview as dummy variables to control for period effects.
Table 1 presents descriptive statistics for the sample.
Descriptive Statistics (N = 11,149 Households; 17,682 Respondents; 55,682 Person-Year Observations)
| . | Total . | Men . | Women . | ||||
|---|---|---|---|---|---|---|---|
| . | Mean or % . | SD . | Range . | Mean or % . | SD . | Mean or % . | SD . |
| Time-varying variables | |||||||
| CESD in 2011 | 8.26 | 6.26 | 0–30 | 7.19 | 5.74 | 9.13 | 6.51 |
| CESD in 2013 | 7.76 | 5.77 | 0–30 | 6.81 | 5.19 | 8.63 | 6.12 |
| CESD in 2015 | 7.81 | 6.34 | 0–30 | 6.67 | 5.75 | 8.85 | 6.65 |
| CESD in 2018 | 8.5 | 6.47 | 0–30 | 7.33 | 5.87 | 9.54 | 6.79 |
| Age in 2011 | 57.04 | 8.93 | 45–97 | 57.85 | 8.62 | 56.41 | 9.12 |
| Married (ref: not married) | 88.81 | 92.02 | 86.00 | ||||
| Retired (ref: not retired) | 29.95 | 23.52 | 35.59 | ||||
| Rural residence (ref: urban) | 64.08 | 64.83 | 63.44 | ||||
| Health status | 2.95 | 0.96 | 1–5 | 2.84 | 0.97 | 3.02 | 0.96 |
| Light activities (ref: no light activities) | 82.45 | 82.78 | 82.16 | ||||
| Wave of interview (ref: 2011) | |||||||
| 2013 | 24.89 | 24.92 | 24.86 | ||||
| 2015 | 26.21 | 26.72 | 25.76 | ||||
| 2018 | 27.38 | 27.95 | 26.87 | ||||
| Time-invariant variables | |||||||
| Any disfavoritism (ref: no disfavoritism) | 18.97 | 18.72 | 19.18 | ||||
| Type of disfavoritism (ref: no disfavoritism) | |||||||
| Paternal disfavoritism only | 2.56 | 2.42 | 2.68 | ||||
| Maternal disfavoritism only | 5.06 | 4.55 | 5.51 | ||||
| Both paternal and maternal disfavoritism | 11.35 | 11.76 | 10.99 | ||||
| Women (ref: Men) | 53.24 | 0 | 100 | ||||
| Childhood starvation (ref: no starvation) | 68.90 | 72.36 | 65.97 | ||||
| Childhood worse health (re: better or the same) | 13.06 | 12.31 | 13.83 | ||||
| Number of siblings | 4.13 | 1.82 | 0–15 | 3.91 | 1.85 | 4.02 | 1.80 |
| First born (ref: later born) | 30.73 | 29.74 | 31.60 | ||||
| Proportion of sons | 40.54 | 39.47 | 41.48 | ||||
| Secondary school or above (ref: below secondary school) | 34.50 | 45.22 | 25.08 | ||||
| Childhood friendship (ref: no good friend) | 47.34 | 47.40 | 47.29 | ||||
| Parents’ mental illness (ref: no abnormality) | 3.77 | 3.73 | 3.87 | ||||
| . | Total . | Men . | Women . | ||||
|---|---|---|---|---|---|---|---|
| . | Mean or % . | SD . | Range . | Mean or % . | SD . | Mean or % . | SD . |
| Time-varying variables | |||||||
| CESD in 2011 | 8.26 | 6.26 | 0–30 | 7.19 | 5.74 | 9.13 | 6.51 |
| CESD in 2013 | 7.76 | 5.77 | 0–30 | 6.81 | 5.19 | 8.63 | 6.12 |
| CESD in 2015 | 7.81 | 6.34 | 0–30 | 6.67 | 5.75 | 8.85 | 6.65 |
| CESD in 2018 | 8.5 | 6.47 | 0–30 | 7.33 | 5.87 | 9.54 | 6.79 |
| Age in 2011 | 57.04 | 8.93 | 45–97 | 57.85 | 8.62 | 56.41 | 9.12 |
| Married (ref: not married) | 88.81 | 92.02 | 86.00 | ||||
| Retired (ref: not retired) | 29.95 | 23.52 | 35.59 | ||||
| Rural residence (ref: urban) | 64.08 | 64.83 | 63.44 | ||||
| Health status | 2.95 | 0.96 | 1–5 | 2.84 | 0.97 | 3.02 | 0.96 |
| Light activities (ref: no light activities) | 82.45 | 82.78 | 82.16 | ||||
| Wave of interview (ref: 2011) | |||||||
| 2013 | 24.89 | 24.92 | 24.86 | ||||
| 2015 | 26.21 | 26.72 | 25.76 | ||||
| 2018 | 27.38 | 27.95 | 26.87 | ||||
| Time-invariant variables | |||||||
| Any disfavoritism (ref: no disfavoritism) | 18.97 | 18.72 | 19.18 | ||||
| Type of disfavoritism (ref: no disfavoritism) | |||||||
| Paternal disfavoritism only | 2.56 | 2.42 | 2.68 | ||||
| Maternal disfavoritism only | 5.06 | 4.55 | 5.51 | ||||
| Both paternal and maternal disfavoritism | 11.35 | 11.76 | 10.99 | ||||
| Women (ref: Men) | 53.24 | 0 | 100 | ||||
| Childhood starvation (ref: no starvation) | 68.90 | 72.36 | 65.97 | ||||
| Childhood worse health (re: better or the same) | 13.06 | 12.31 | 13.83 | ||||
| Number of siblings | 4.13 | 1.82 | 0–15 | 3.91 | 1.85 | 4.02 | 1.80 |
| First born (ref: later born) | 30.73 | 29.74 | 31.60 | ||||
| Proportion of sons | 40.54 | 39.47 | 41.48 | ||||
| Secondary school or above (ref: below secondary school) | 34.50 | 45.22 | 25.08 | ||||
| Childhood friendship (ref: no good friend) | 47.34 | 47.40 | 47.29 | ||||
| Parents’ mental illness (ref: no abnormality) | 3.77 | 3.73 | 3.87 | ||||
Descriptive Statistics (N = 11,149 Households; 17,682 Respondents; 55,682 Person-Year Observations)
| . | Total . | Men . | Women . | ||||
|---|---|---|---|---|---|---|---|
| . | Mean or % . | SD . | Range . | Mean or % . | SD . | Mean or % . | SD . |
| Time-varying variables | |||||||
| CESD in 2011 | 8.26 | 6.26 | 0–30 | 7.19 | 5.74 | 9.13 | 6.51 |
| CESD in 2013 | 7.76 | 5.77 | 0–30 | 6.81 | 5.19 | 8.63 | 6.12 |
| CESD in 2015 | 7.81 | 6.34 | 0–30 | 6.67 | 5.75 | 8.85 | 6.65 |
| CESD in 2018 | 8.5 | 6.47 | 0–30 | 7.33 | 5.87 | 9.54 | 6.79 |
| Age in 2011 | 57.04 | 8.93 | 45–97 | 57.85 | 8.62 | 56.41 | 9.12 |
| Married (ref: not married) | 88.81 | 92.02 | 86.00 | ||||
| Retired (ref: not retired) | 29.95 | 23.52 | 35.59 | ||||
| Rural residence (ref: urban) | 64.08 | 64.83 | 63.44 | ||||
| Health status | 2.95 | 0.96 | 1–5 | 2.84 | 0.97 | 3.02 | 0.96 |
| Light activities (ref: no light activities) | 82.45 | 82.78 | 82.16 | ||||
| Wave of interview (ref: 2011) | |||||||
| 2013 | 24.89 | 24.92 | 24.86 | ||||
| 2015 | 26.21 | 26.72 | 25.76 | ||||
| 2018 | 27.38 | 27.95 | 26.87 | ||||
| Time-invariant variables | |||||||
| Any disfavoritism (ref: no disfavoritism) | 18.97 | 18.72 | 19.18 | ||||
| Type of disfavoritism (ref: no disfavoritism) | |||||||
| Paternal disfavoritism only | 2.56 | 2.42 | 2.68 | ||||
| Maternal disfavoritism only | 5.06 | 4.55 | 5.51 | ||||
| Both paternal and maternal disfavoritism | 11.35 | 11.76 | 10.99 | ||||
| Women (ref: Men) | 53.24 | 0 | 100 | ||||
| Childhood starvation (ref: no starvation) | 68.90 | 72.36 | 65.97 | ||||
| Childhood worse health (re: better or the same) | 13.06 | 12.31 | 13.83 | ||||
| Number of siblings | 4.13 | 1.82 | 0–15 | 3.91 | 1.85 | 4.02 | 1.80 |
| First born (ref: later born) | 30.73 | 29.74 | 31.60 | ||||
| Proportion of sons | 40.54 | 39.47 | 41.48 | ||||
| Secondary school or above (ref: below secondary school) | 34.50 | 45.22 | 25.08 | ||||
| Childhood friendship (ref: no good friend) | 47.34 | 47.40 | 47.29 | ||||
| Parents’ mental illness (ref: no abnormality) | 3.77 | 3.73 | 3.87 | ||||
| . | Total . | Men . | Women . | ||||
|---|---|---|---|---|---|---|---|
| . | Mean or % . | SD . | Range . | Mean or % . | SD . | Mean or % . | SD . |
| Time-varying variables | |||||||
| CESD in 2011 | 8.26 | 6.26 | 0–30 | 7.19 | 5.74 | 9.13 | 6.51 |
| CESD in 2013 | 7.76 | 5.77 | 0–30 | 6.81 | 5.19 | 8.63 | 6.12 |
| CESD in 2015 | 7.81 | 6.34 | 0–30 | 6.67 | 5.75 | 8.85 | 6.65 |
| CESD in 2018 | 8.5 | 6.47 | 0–30 | 7.33 | 5.87 | 9.54 | 6.79 |
| Age in 2011 | 57.04 | 8.93 | 45–97 | 57.85 | 8.62 | 56.41 | 9.12 |
| Married (ref: not married) | 88.81 | 92.02 | 86.00 | ||||
| Retired (ref: not retired) | 29.95 | 23.52 | 35.59 | ||||
| Rural residence (ref: urban) | 64.08 | 64.83 | 63.44 | ||||
| Health status | 2.95 | 0.96 | 1–5 | 2.84 | 0.97 | 3.02 | 0.96 |
| Light activities (ref: no light activities) | 82.45 | 82.78 | 82.16 | ||||
| Wave of interview (ref: 2011) | |||||||
| 2013 | 24.89 | 24.92 | 24.86 | ||||
| 2015 | 26.21 | 26.72 | 25.76 | ||||
| 2018 | 27.38 | 27.95 | 26.87 | ||||
| Time-invariant variables | |||||||
| Any disfavoritism (ref: no disfavoritism) | 18.97 | 18.72 | 19.18 | ||||
| Type of disfavoritism (ref: no disfavoritism) | |||||||
| Paternal disfavoritism only | 2.56 | 2.42 | 2.68 | ||||
| Maternal disfavoritism only | 5.06 | 4.55 | 5.51 | ||||
| Both paternal and maternal disfavoritism | 11.35 | 11.76 | 10.99 | ||||
| Women (ref: Men) | 53.24 | 0 | 100 | ||||
| Childhood starvation (ref: no starvation) | 68.90 | 72.36 | 65.97 | ||||
| Childhood worse health (re: better or the same) | 13.06 | 12.31 | 13.83 | ||||
| Number of siblings | 4.13 | 1.82 | 0–15 | 3.91 | 1.85 | 4.02 | 1.80 |
| First born (ref: later born) | 30.73 | 29.74 | 31.60 | ||||
| Proportion of sons | 40.54 | 39.47 | 41.48 | ||||
| Secondary school or above (ref: below secondary school) | 34.50 | 45.22 | 25.08 | ||||
| Childhood friendship (ref: no good friend) | 47.34 | 47.40 | 47.29 | ||||
| Parents’ mental illness (ref: no abnormality) | 3.77 | 3.73 | 3.87 | ||||
Analytical Strategy
Because our explanatory variables were time-invariant, we used random-effect models to adjust for the longitudinal structure of the data. Because spouses were enrolled in the sample, analyses based on the full sample clustered at both subject and household levels. Because only one person per couple was selected once conditional on gender, the analyses based on separate gender samples only had person-year observations nested within persons.
We reported robust standard errors to address potential heteroskedasticity. Because only cross-sectional weights were provided in CHARLS, we did not use weights in our main analyses. However, we provided a sensitivity analysis to test the extent to which sampling weights affected our results using one-wave data and applying cross-sectional weights (Supplementary Material 1).
We began by using the variable any childhood parental disfavoritism to predict Chinese adults’ depressive symptoms. To test whether the effect of parental disfavoritism varied by own gender, we interacted any childhood parental disfavoritism with the respondents’ gender. To study how the effect differs by parent’s gender, we used disfavoritism by type to predict Chinese adults’ depressive symptoms and compared the effects of paternal disfavoritism only with maternal disfavoritism only.
To study how own and parents’ gender interact to shape the effect of childhood parental disfavoritism, we stratified the sample to men and women and used disfavoritism by type to predict depressive symptoms. We compared the effects within the same model (across parent’s gender) and across models by child’s gender based on Z-test (Paternoster et al., 1998). We took this gender-stratified approach because results were easier to show and interpret. However, we conducted a sensitivity analysis that took an alternative approach by interacting the types of disfavoritism and respondents’ gender, and testing the differences in the average marginal effects by own and parents’ gender (Supplementary Material 2).
In addition, we conducted sensitivity analyses to test the potential influences of other adverse childhood experiences (Supplementary Material 3). Finally, research has shown that stress triggers parental differentiation and impairs mental health (Crouter et al., 1999). Therefore, parental differentiation could be more common during historical periods that featured resource scarcity and unfavorable social-cultural conditions. Thus, we tested how cohort may shape the effect of parental disfavoritism (Supplementary Material 4).
Results
Regression Analyses
We began our analyses by using any childhood disfavoritism to predict midlife and older Chinese adults’ depressive symptoms. As shown by the first row of coefficients in Model 1 in Table 2, having experienced any disfavoritism by either father or mother was associated with a 0.65 (p < .01) unit increase in midlife and older Chinese adults’ depressive symptoms.
Mixed Linear Model Predicting Midlife and Older Chinese Adults’ Depressive Symptoms in Total Sample and Stratified by Gender
| . | Total . | Men . | Women . | |
|---|---|---|---|---|
| Variables . | Model 1 . | Model 2 . | Model 3 . | Model 4 . |
| Any disfavoritism | 0.65** (0.09) | |||
| Type of disfavoritism | ||||
| Paternal only | 1.13**,a (0.22) | 1.05**,a (0.33) | 1.23** (0.31) | |
| Maternal only | 0.59**,a (0.16) | 0.37a,b (0.21) | 0.90**,b (0.23) | |
| Both paternal and maternal | 0.54** (0.11) | 0.63** (0.15) | 0.58** (0.16) | |
| Women | 1.59** (0.06) | 1.58** (0.06) | ||
| Childhood starvation | 0.96** (0.07) | 0.95** (0.07) | 0.77** (0.10) | 1.18** (0.11) |
| Childhood poorer health | 1.07** (0.11) | 1.09** (0.11) | 1.15** (0.16) | 1.33** (0.16) |
| Number of siblings | 0.02 (0.02) | 0.02 (0.02) | 0.03 (0.03) | 0.04 (0.03) |
| Firstborn | −0.07 (0.07) | −0.08 (0.07) | −0.03 (0.10) | −0.09 (0.11) |
| Proportion of sons | 0.20 (0.16) | 0.19 (0.16) | −0.08 (0.22) | 0.41 (0.24) |
| Age | −0.00 (0.00) | −0.00 (0.00) | −0.00 (0.01) | −0.01 (0.01) |
| Secondary school and above | −0.71** (0.07) | −0.71** (0.08) | −0.76** (0.09) | −0.84** (0.12) |
| Married | −1.49** (0.12) | −1.49** (0.12) | −1.45** (0.18) | −1.55** (0.15) |
| Retired | 0.16* (0.06) | 0.16* (0.06) | 0.36** (0.10) | −0.03 (0.08) |
| Rural residence | 1.22** (0.08) | 1.21** (0.08) | 0.97** (0.10) | 1.26** (0.11) |
| Health status | 1.64** (0.03) | 1.64** (0.03) | 1.54** (0.04) | 1.83** (0.04) |
| Light activities | −0.31** (0.10) | −0.30** (0.09) | −0.25 (0.13) | −0.45** (0.14) |
| Childhood friendship | 0.39** (0.07) | 0.39** (0.07) | 0.33** (0.10) | 0.58** (0.10) |
| Parent’s mental illness | 1.42** (0.20) | 1.45** (0.20) | 1.93** (0.30) | 1.29** (0.28) |
| Wave of survey | ||||
| 2013 | −0.33** (0.06) | −0.33** (0.06) | −0.23** (0.08) | −0.40** (0.08) |
| 2015 | −0.17** (0.07) | −0.18** (0.07) | −0.29** (0.08) | −0.07 (0.09) |
| 2018 | 0.36** (0.07) | 0.36** (0.07) | 0.20* (0.09) | 0.51** (0.09) |
| Constant | 2.39** (0.32) | 2.37** (0.32) | 2.90** (0.42) | 3.36** (0.44) |
| Random effects | ||||
| Level-3 variance | 6.63 | 6.60 | ||
| Level-2 variance | 5.33 | 5.32 | 9.66 | 13.34 |
| Level-1 variance | 19.15 | 19.16 | 16.69 | 21.28 |
| N (household) | ||||
| N (person) | 55,682 | 55,682 | 25,963 | 29,719 |
| N (person-year observation) | 17,682 | 17,682 | 8,350 | 9,332 |
| . | Total . | Men . | Women . | |
|---|---|---|---|---|
| Variables . | Model 1 . | Model 2 . | Model 3 . | Model 4 . |
| Any disfavoritism | 0.65** (0.09) | |||
| Type of disfavoritism | ||||
| Paternal only | 1.13**,a (0.22) | 1.05**,a (0.33) | 1.23** (0.31) | |
| Maternal only | 0.59**,a (0.16) | 0.37a,b (0.21) | 0.90**,b (0.23) | |
| Both paternal and maternal | 0.54** (0.11) | 0.63** (0.15) | 0.58** (0.16) | |
| Women | 1.59** (0.06) | 1.58** (0.06) | ||
| Childhood starvation | 0.96** (0.07) | 0.95** (0.07) | 0.77** (0.10) | 1.18** (0.11) |
| Childhood poorer health | 1.07** (0.11) | 1.09** (0.11) | 1.15** (0.16) | 1.33** (0.16) |
| Number of siblings | 0.02 (0.02) | 0.02 (0.02) | 0.03 (0.03) | 0.04 (0.03) |
| Firstborn | −0.07 (0.07) | −0.08 (0.07) | −0.03 (0.10) | −0.09 (0.11) |
| Proportion of sons | 0.20 (0.16) | 0.19 (0.16) | −0.08 (0.22) | 0.41 (0.24) |
| Age | −0.00 (0.00) | −0.00 (0.00) | −0.00 (0.01) | −0.01 (0.01) |
| Secondary school and above | −0.71** (0.07) | −0.71** (0.08) | −0.76** (0.09) | −0.84** (0.12) |
| Married | −1.49** (0.12) | −1.49** (0.12) | −1.45** (0.18) | −1.55** (0.15) |
| Retired | 0.16* (0.06) | 0.16* (0.06) | 0.36** (0.10) | −0.03 (0.08) |
| Rural residence | 1.22** (0.08) | 1.21** (0.08) | 0.97** (0.10) | 1.26** (0.11) |
| Health status | 1.64** (0.03) | 1.64** (0.03) | 1.54** (0.04) | 1.83** (0.04) |
| Light activities | −0.31** (0.10) | −0.30** (0.09) | −0.25 (0.13) | −0.45** (0.14) |
| Childhood friendship | 0.39** (0.07) | 0.39** (0.07) | 0.33** (0.10) | 0.58** (0.10) |
| Parent’s mental illness | 1.42** (0.20) | 1.45** (0.20) | 1.93** (0.30) | 1.29** (0.28) |
| Wave of survey | ||||
| 2013 | −0.33** (0.06) | −0.33** (0.06) | −0.23** (0.08) | −0.40** (0.08) |
| 2015 | −0.17** (0.07) | −0.18** (0.07) | −0.29** (0.08) | −0.07 (0.09) |
| 2018 | 0.36** (0.07) | 0.36** (0.07) | 0.20* (0.09) | 0.51** (0.09) |
| Constant | 2.39** (0.32) | 2.37** (0.32) | 2.90** (0.42) | 3.36** (0.44) |
| Random effects | ||||
| Level-3 variance | 6.63 | 6.60 | ||
| Level-2 variance | 5.33 | 5.32 | 9.66 | 13.34 |
| Level-1 variance | 19.15 | 19.16 | 16.69 | 21.28 |
| N (household) | ||||
| N (person) | 55,682 | 55,682 | 25,963 | 29,719 |
| N (person-year observation) | 17,682 | 17,682 | 8,350 | 9,332 |
Notes: Robust standard errors in parentheses;
**p < .01,
*p < .05.
aSignificant differences within model at p < .05.
bSignificant differences across models at p < .05.
Mixed Linear Model Predicting Midlife and Older Chinese Adults’ Depressive Symptoms in Total Sample and Stratified by Gender
| . | Total . | Men . | Women . | |
|---|---|---|---|---|
| Variables . | Model 1 . | Model 2 . | Model 3 . | Model 4 . |
| Any disfavoritism | 0.65** (0.09) | |||
| Type of disfavoritism | ||||
| Paternal only | 1.13**,a (0.22) | 1.05**,a (0.33) | 1.23** (0.31) | |
| Maternal only | 0.59**,a (0.16) | 0.37a,b (0.21) | 0.90**,b (0.23) | |
| Both paternal and maternal | 0.54** (0.11) | 0.63** (0.15) | 0.58** (0.16) | |
| Women | 1.59** (0.06) | 1.58** (0.06) | ||
| Childhood starvation | 0.96** (0.07) | 0.95** (0.07) | 0.77** (0.10) | 1.18** (0.11) |
| Childhood poorer health | 1.07** (0.11) | 1.09** (0.11) | 1.15** (0.16) | 1.33** (0.16) |
| Number of siblings | 0.02 (0.02) | 0.02 (0.02) | 0.03 (0.03) | 0.04 (0.03) |
| Firstborn | −0.07 (0.07) | −0.08 (0.07) | −0.03 (0.10) | −0.09 (0.11) |
| Proportion of sons | 0.20 (0.16) | 0.19 (0.16) | −0.08 (0.22) | 0.41 (0.24) |
| Age | −0.00 (0.00) | −0.00 (0.00) | −0.00 (0.01) | −0.01 (0.01) |
| Secondary school and above | −0.71** (0.07) | −0.71** (0.08) | −0.76** (0.09) | −0.84** (0.12) |
| Married | −1.49** (0.12) | −1.49** (0.12) | −1.45** (0.18) | −1.55** (0.15) |
| Retired | 0.16* (0.06) | 0.16* (0.06) | 0.36** (0.10) | −0.03 (0.08) |
| Rural residence | 1.22** (0.08) | 1.21** (0.08) | 0.97** (0.10) | 1.26** (0.11) |
| Health status | 1.64** (0.03) | 1.64** (0.03) | 1.54** (0.04) | 1.83** (0.04) |
| Light activities | −0.31** (0.10) | −0.30** (0.09) | −0.25 (0.13) | −0.45** (0.14) |
| Childhood friendship | 0.39** (0.07) | 0.39** (0.07) | 0.33** (0.10) | 0.58** (0.10) |
| Parent’s mental illness | 1.42** (0.20) | 1.45** (0.20) | 1.93** (0.30) | 1.29** (0.28) |
| Wave of survey | ||||
| 2013 | −0.33** (0.06) | −0.33** (0.06) | −0.23** (0.08) | −0.40** (0.08) |
| 2015 | −0.17** (0.07) | −0.18** (0.07) | −0.29** (0.08) | −0.07 (0.09) |
| 2018 | 0.36** (0.07) | 0.36** (0.07) | 0.20* (0.09) | 0.51** (0.09) |
| Constant | 2.39** (0.32) | 2.37** (0.32) | 2.90** (0.42) | 3.36** (0.44) |
| Random effects | ||||
| Level-3 variance | 6.63 | 6.60 | ||
| Level-2 variance | 5.33 | 5.32 | 9.66 | 13.34 |
| Level-1 variance | 19.15 | 19.16 | 16.69 | 21.28 |
| N (household) | ||||
| N (person) | 55,682 | 55,682 | 25,963 | 29,719 |
| N (person-year observation) | 17,682 | 17,682 | 8,350 | 9,332 |
| . | Total . | Men . | Women . | |
|---|---|---|---|---|
| Variables . | Model 1 . | Model 2 . | Model 3 . | Model 4 . |
| Any disfavoritism | 0.65** (0.09) | |||
| Type of disfavoritism | ||||
| Paternal only | 1.13**,a (0.22) | 1.05**,a (0.33) | 1.23** (0.31) | |
| Maternal only | 0.59**,a (0.16) | 0.37a,b (0.21) | 0.90**,b (0.23) | |
| Both paternal and maternal | 0.54** (0.11) | 0.63** (0.15) | 0.58** (0.16) | |
| Women | 1.59** (0.06) | 1.58** (0.06) | ||
| Childhood starvation | 0.96** (0.07) | 0.95** (0.07) | 0.77** (0.10) | 1.18** (0.11) |
| Childhood poorer health | 1.07** (0.11) | 1.09** (0.11) | 1.15** (0.16) | 1.33** (0.16) |
| Number of siblings | 0.02 (0.02) | 0.02 (0.02) | 0.03 (0.03) | 0.04 (0.03) |
| Firstborn | −0.07 (0.07) | −0.08 (0.07) | −0.03 (0.10) | −0.09 (0.11) |
| Proportion of sons | 0.20 (0.16) | 0.19 (0.16) | −0.08 (0.22) | 0.41 (0.24) |
| Age | −0.00 (0.00) | −0.00 (0.00) | −0.00 (0.01) | −0.01 (0.01) |
| Secondary school and above | −0.71** (0.07) | −0.71** (0.08) | −0.76** (0.09) | −0.84** (0.12) |
| Married | −1.49** (0.12) | −1.49** (0.12) | −1.45** (0.18) | −1.55** (0.15) |
| Retired | 0.16* (0.06) | 0.16* (0.06) | 0.36** (0.10) | −0.03 (0.08) |
| Rural residence | 1.22** (0.08) | 1.21** (0.08) | 0.97** (0.10) | 1.26** (0.11) |
| Health status | 1.64** (0.03) | 1.64** (0.03) | 1.54** (0.04) | 1.83** (0.04) |
| Light activities | −0.31** (0.10) | −0.30** (0.09) | −0.25 (0.13) | −0.45** (0.14) |
| Childhood friendship | 0.39** (0.07) | 0.39** (0.07) | 0.33** (0.10) | 0.58** (0.10) |
| Parent’s mental illness | 1.42** (0.20) | 1.45** (0.20) | 1.93** (0.30) | 1.29** (0.28) |
| Wave of survey | ||||
| 2013 | −0.33** (0.06) | −0.33** (0.06) | −0.23** (0.08) | −0.40** (0.08) |
| 2015 | −0.17** (0.07) | −0.18** (0.07) | −0.29** (0.08) | −0.07 (0.09) |
| 2018 | 0.36** (0.07) | 0.36** (0.07) | 0.20* (0.09) | 0.51** (0.09) |
| Constant | 2.39** (0.32) | 2.37** (0.32) | 2.90** (0.42) | 3.36** (0.44) |
| Random effects | ||||
| Level-3 variance | 6.63 | 6.60 | ||
| Level-2 variance | 5.33 | 5.32 | 9.66 | 13.34 |
| Level-1 variance | 19.15 | 19.16 | 16.69 | 21.28 |
| N (household) | ||||
| N (person) | 55,682 | 55,682 | 25,963 | 29,719 |
| N (person-year observation) | 17,682 | 17,682 | 8,350 | 9,332 |
Notes: Robust standard errors in parentheses;
**p < .01,
*p < .05.
aSignificant differences within model at p < .05.
bSignificant differences across models at p < .05.
Next, we interacted any childhood disfavoritism with the respondents’ gender and reestimated the model. However, no substantive relationship between the interaction term and Chinese adults’ depressive symptoms was found (B = 0.14, n.s., results not shown), suggesting little difference in the overall effect of any childhood disfavoritism by respondents’ own gender.
Then, to study how the effects differ by parent’s gender, we used the type of recalled childhood disfavoritism to predict midlife and older Chinese adults’ depressive symptoms. As shown in Model 2, paternal disfavoritism only (B = 1.13, p < .01), maternal disfavoritism only (B = 0.59, p < .01), and both paternal and maternal disfavoritism (B = 0.54, p < .01) predicted children’s increased depressive symptoms in their adulthood. Comparing the effects of paternal and maternal disfavoritism, we found that having been disfavored by fathers only was associated with higher depressive symptoms than having been disfavored by mothers (Z = 2.06, p < .05).
Finally, we stratified the sample into men and women and used disfavoritism by type to predict their depressive symptoms. Models 3 and 4 present results for men and women accordingly. Model 3 showed that for men, paternal disfavoritism only was associated with a 1.05-unit increase in depressive symptoms (p < .01). The effect of maternal disfavoristism only, however, was nonsignificant (B = 0.37, n.s.). Comparison across the effects of paternal and maternal disfavoritism suggested that perceiving oneself as having been disfavored by fathers only was associated with higher depressive symptoms than having been disfavored by mothers only for men (Z = 1.77, p < .05).
By comparison, Model 4 showed that for women, paternal disfavorisitism only and maternal disfavoritism only were associated with a 1.23 (p < .01) and 0.90 (p < .01) unit increase in their depressive symptoms, accordingly. Comparison between the effects of paternal and maternal disfavoritism for women suggested no substantive differences (Z = 0.87, n.s.).
Comparison of the effects of paternal treatment only and maternal disfavoritism only across gender models suggested that having been treated unfairly by mothers only was associated with a higher increase in depressive symptoms for women than men (B1 = 0.37, n.s. for men; B2 = 0.90, p < .01 for women; Z = 1.70, p < .05). In contrast, no difference in the effects of paternal disfavoritism only for men and women was found (B1 = 1.05, p < .01 for men; B2 = 1.23, p < .01 for women; Z = 0.40, n.s.).
Supplementary Analyses
Supplemental analyses suggested that results were substantially the same comparing our main results (see Tables 1 and 2) with the weighted results (Supplementary Material 1), models with interaction terms of types of disfavoritism and gender (Supplementary Material 2). In addition, childhood parental disfavoritism had an independent effect on depressive symptoms in adulthood, with other types of childhood adversities controlled, supporting our main conclusions (Supplementary Material 3). Finally, no evidence of variation in the effect of childhood parental disfavoritism across cohorts was found, and our conclusions remained unchanged with cohort effects considered (Supplementary Material 4). This was plausible given that parental disfavoritism affected children’s mental health across the life course mainly via comparison within the family.
Discussion
Our aim in this article was to examine whether childhood parental disfavoritism has a long-term detrimental effect on midlife and older Chinese’s psychological well-being, and how the effect varies by parent’s and child’s gender. Drawing on theories of social comparison and the life course, we hypothesized that childhood parental disfavoritism was associated with higher depressive symptoms among midlife and older Chinese adults (H1). Drawing on theories of gender role development and affect control, we proposed competing hypotheses on the moderating effect of child’s gender (H2a, H2b). Based upon theories of socialization of women and power dynamics in Chinese families, we proposed competing hypotheses on the moderating effect of parent’s gender (H3a, H3b). Finally, drawing on theories of social learning and identity formation, we hypothesized that childhood parental disfavoritism would be more detrimental when own and parent’s gender were the same (H4).
Findings supported some of our hypotheses. First, recollections of childhood parental disfavoritism were associated with higher depressive symptoms among midlife and older Chinese adults, supporting H1. This finding contributes to the literature on PDT by revealing the detrimental effect of parental disfavoritism across cultures and persistence across the life course (Barrett Singer & Weinstein, 2000; De Man et al., 2003; Luo et al., 2020; Ng et al., 2020; Peng et al., 2018). It also complements the life course perspective and literature on midlife and older adults’ mental health by showing that health inequality in mid- and later life is deeply rooted in childhood experiences of parental disfavoritism within the family. Specifically, before our study, most scholars conceptualize adverse childhood experiences that capture the differences in childhood experiences between-families (e.g., Felitti et al., 1998). Our study, together with those conducted in the United States and Europe (Davey et al., 2009; Peng et al., 2018; Suitor et al., 2017), demonstrated that childhood parental disfavoritism as a form of within-family differentiation could have a long-lasting adverse effect on children’s mental health.
Second, no difference in the effect of any childhood parental disfavoritism across own gender was found. This nonfinding turned out to be explained by the hidden heterogeneity across the effects of fathers’ and mothers’ disfavoritism for men and women. Specifically, paternal disfavoritism was found more detrimental than maternal disfavoritism in the total sample, supporting H3b. This pattern is consistent with theories of power and social influence, suggesting a greater influence of fathers on children’s long-term psychological development shaped by patriarchy (Emerson, 1962; Xie, 2013). Examination of the effects by own and parent’s gender suggested that paternal disfavoritism was found more detrimental than maternal disfavoritism only for men. Maternal disfavoritism was more detrimental for women than men. Therefore, H4 was partially supported. This pattern highlights the influence of gender similarity in shaping the effect of childhood parental disfavoritism on mental health in mid- and later life (Harris & Howard, 1985).
These findings show both consistent and different patterns from those found in the United States and Europe, where research has remained inconclusive about the relative influence of paternal and maternal differential treatment on children’s well-being (Meunier et al., 2011). Our findings suggest a universal detrimental effect of fathers’ disfavoritism regardless of child’s gender in China as shaped by the patriarchal power relations. These findings are likely to be generalizable to societies where fathers hold similar power in the family and that have similar fathering practices as in China.
These findings also have practical implications. With multichild family becoming the ideal type of family advocated by the policy, the Chinese government has set up social institutions and parental education programs to promote the science of parenting. However, most programs recruit parents voluntarily (Sun, 2020). Because most primary caretakers are mothers, fathers tend to opt out of these programs. Our study has suggested that it is critical for the nation to consider accompanying the Three-Child Policy with parental education programs involving both fathers and mothers on equal treatment of children.
Several questions remain unanswered in our research that we hope future scholars could study. First, PDT is a multidimensional concept including favoritism and disfavoritism, and relational and evaluative dimensions. Future studies could use more detailed measures of PDT and examine its patterns and consequences in China. Second, our study used data that only selected one respondent per family of origin to provide information on childhood parental disfavoritism. Future research sampling multiple children from the same family could allow examination of the influences of sibling characteristics. Third, childhood parental disfavoritism was measured retrospectively, the validity of which is under debate. However, longitudinal data with prospective designs that trace individuals across several decades are rare in China. Thus, the judicious use of retrospective childhood conditions is still recommended (Haas & Bishop, 2010). Literature has suggested that gender, socioeconomic status, current health status, time since the event (i.e., age; Auriat, 1993; Kamo, 2000), and the severity of childhood experiences (Hardt & Rutter, 2004) are factors that potentially affect recall accuracy. Although we have controlled for these factors in our analyses, future studies using prospective design are recommended. Fourth, due to the lack of direct measures of sibling tension and sense of unfairness, we were not able to directly test the mechanism using our data. Therefore, we have called for future scholars to directly test the mechanisms via which childhood parental disfavoritism affected mental health in adulthood. Finally, future studies expand the study of PDT in more diverse cultural contexts.
Taken together, our study has contributed to the understanding of the universality of the detrimental effect of parental disfavoritism across cultures, its persistence across the life course, and the unique gendered patterns in China shaped by patriarchy. Future scholars should explore patterns and consequences of more diverse dimensions of PDT and identify effective ways to implement interventions.
Funding
The development of the Harmonized CHARLS was funded by the National Institute on Ageing (R01 AG030153, RC2 AG036619, R03 AG043052).
Conflict of Interest
None.
Data Availability
This analysis uses data or information from the Harmonized CHARLS data set and Codebook, Version C as of April 2018 developed by the Gateway to Global Aging Data. For more information, please refer to https://g2aging.org.
Author Contributions
Y. Hou took lead in designing the study, analyzing data, and writing the article. J. Jill Suitor contributed to conceptualizing the research questions and writing and revising the manuscript.
