Child behavior problems and parental psychological distress in Chinese families of children with autism: The putative moderating role of parental social support and cultural values

The wellbeing of parents of children with autism residing in mainland China remains understudied. We aimed to examine whether and how parental perceived social support, individualism, and collectivism acted together to moderate the relationships between child behavior problems and parental psychological distress in Chinese parents of children with autism. With convenience and snowball sampling, data on 268 primary caregiver parents of children with autism were collected from an online cross‐sectional survey. Linear regression analysis indicated that child behavior problems were significantly associated with increased psychological distress in Chinese parents of children with autism. There was no evidence to support the stress‐buffering model of social support in moderation analysis of the association between child behavior problems and parental psychological distress. Nonetheless, increased social support was associated with lower levels of parental psychological distress. Moderated moderation analyses did not support a role for individualism or collectivism as a moderator of the putative buffering role of social support. However, there was evidence that parental individualism was associated with increased parental psychological distress. Our findings highlight that child behavior problems are a robust correlate of parental psychological distress, and parental social support may act as a compensatory factor promoting less psychological distress rather than having a protective role. The role of social support and cultural values in the wellbeing of parents of children with autism in China requires additional exploration, including longitudinal research designs.


Lay Summary
Child behavior problems typically decrease the wellbeing of parents of children with autism, whilst perceived support increases wellbeing.However, these relationships have been rarely explored in Chinese parents.We asked parents of children with autism living in mainland China about their psychological distress, perceived support and cultural values, and their children's behavior problems.We found that higher levels of child behavior problems and individualistic cultural values were associated with increased parental psychological distress, and a higher level of perceived support and was associated with less psychological distress.These findings confirm the importance of addressing child behavior problems and increasing parental perception of social support in improving wellbeing of parents of children with autism.More work is needed to understand why parental

INTRODUCTION
Concurrent evidence exists in both Western and Chinese contexts showing that parents of children with autism have elevated parenting stress and mental health difficulties (Barroso et al., 2018;Lu et al., 2015;Schnabel et al., 2020;Su et al., 2018).Higher levels of parental stress and mental health problems may also affect outcomes for children with autism via parenting beliefs and practices (Chen et al., 2021;Osborne & Reed, 2010) and potentially reduced intervention effectiveness (Karst & Van Hecke, 2012).Thus, understanding what contributes to parental wellbeing could inform support and service design for both parents and children with autism.
Research to date has been focused on examining and identifying correlates of wellbeing in parents of children with autism including child-related, parent-related, and context-related characteristics (see Enea & Rusu, 2020 for a review).Child behavior problems are typically associated with parental wellbeing over and above other child characteristics such as adaptive behavior skills and autism symptom severity (Yorke et al., 2018).Additionally, longitudinal research has established the relationship between child behavior problems and parental wellbeing (Benson, 2023;Totsika et al., 2013).However, whether this association holds in Chinese parents raising a child with autism and what factors may affect or moderate this association remains less well understood.
Social support has been conceptualized and evidenced as one of the most important coping resources in improving wellbeing of parents of children with autism ( Čoli c et al., 2022;McCubbin & Patterson, 1983).Parental perceived support has been demonstrated as a predictor of parental wellbeing (Robinson & Weiss, 2020).Following a stress-buffering model of social support (Cohen & Wills, 1985), parental perceived support may also mitigate the effect of stressors on parental wellbeing.However, there are mixed findings regarding whether parental perceived support serves as a moderator between child behavior problems and parental wellbeing (Halstead et al., 2017;Higgins et al., 2023).An association between parental perceived support and wellbeing has also been found in studies of Chinese parents of children with autism (Ban et al., 2021;Wang, 2021).However, it is unclear whether there is a buffering role of parental perceived support between child behavior problems and parental wellbeing in Chinese parents of children with autism.
Culture is increasingly considered an important ecological systems factor in understanding the parental and familial experience of raising a child with autism (Al Khateeb et al., 2019;Glidden et al., 2021).Previous studies have used culture as a means of comparing how parents from Eastern and Western countries adapt to having a child with autism (Chao et al., 2018;Lin et al., 2011).Nevertheless, both national and international studies rarely consider and measure cultural beliefs.Research on general populations has supported that wellbeing is related to social interdependence, particularly in collectivistic cultures (Markus & Kitayama, 1991).In addition, the receipt, acceptance, and perception of support are characterized by culture (Beehr & Glazer, 2001).Therefore, to better understand the wellbeing of Chinese parents of children with autism, cultural beliefs need to be considered explicitly.
Research usually characterizes culture with values endorsed by individuals (Hofstede, 2001).Individualism and collectivism are commonly used to conceptualize cultural values.Specifically, individualism refers to the extent to which individuals consider themselves independent from their culture group, and collectivism is defined as the extent to which individuals consider themselves to be related to others in their culture group (Hofstede, 1984).Individualism and collectivism are often globally mapped in Eastern and Western cultures, to characterize them as individualistic and collectivistic cultures; and used as a lens to interpret country differences in families' experiences.For example, a higher level of parental distress in Japanese parents compared with Italian parents may be attributed to parental cultural tendencies to adhere to common norms and values in collectivistic societies when managing the challenges of having a child with autism (Giannotti et al., 2021).However, individualism and collectivism may coexist in one culture and can both be endorsed to some extent by an individual at the same time.An intraculture (in contrast to interculture) approach is thus more useful to capture variations in individuals' endorsement of cultural values within one culture (e.g., He et al., 2021).
The aim of this study was to examine how social support, individualism, and collectivism act together to moderate the relationship between child behavior problems and parental psychological wellbeing in Chinese parents of children with autism residing in mainland China.First, whether child behavior problems were associated with parental psychological distress was examined.Based on substantial existing evidence, we hypothesized that child behavior problems would significantly and positively predict parental psychological distress (Hypothesis 1).
Second, whether parental perceived support moderated the relationship between child behavior problems and parental wellbeing was investigated.As the stressbuffering model of social support proposes (Cohen & Wills, 1985), we hypothesized that the association between child behavior problems and parental psychological distress would be weaker when parents perceive higher levels of social support (Hypothesis 2).Third, given that only limited evidence was available we did not generate directional hypotheses, but we explored whether the buffering role of parental perceived support was associated with parental individualism or collectivism.

Research design
The study used an online cross-sectional survey.Parents were invited to complete an online questionnaire by primary parental caregivers which assessed child behavior problems, parental perceived support, individualism and collectivism, and psychological distress.Additionally, participants were also asked to answer a set of demographic questions.Inclusion criteria for the study included: (1) a primary parental caregiver of a child with autism aged between 4 and 15 years; (2) the child with autism had a formal diagnosis from a professional (e.g., doctors, pediatricians, psychologists, or psychiatrist)-according to parent report; and (3) the parent and child with autism resided in mainland China and lived in the same family home.No exclusion criteria were specified.Convenience and snowball sampling strategies were used for data collection.The study was preregistered on the Open Science Framework (https:// osf.io/zjgd9/).

Procedure
The study was approved by the University of Warwick research ethics committee (application reference number 199/2-122).Data collection was initiated and conducted following the ethical approval of the study.An invitation letter that included a study advert was first sent via email or WeChat to special education schools, private rehabilitation services, parent support groups, and university settings where most Chinese children of autism and their parents receive services and support.The study advert involved a brief introduction to the study and a link to the study webpage, where a summary of the study, the participant information leaflet and an anonymous online (Qualtrics) survey link were embedded.The survey included the participant information leaflet, the consent statement, and the main body of the survey.The study advert was distributed to potential participants by the organizations described above.Participants could not access the survey without first providing consent.

Participants
A total of 493 parents from 22 provinces of mainland China (mainly Henan, Chongqing, Sichuan, Jiangsu, Guangdong, Shanghai, Zhejiang, Heilongjiang) followed the survey link, with 374 consenting to participate in the study.After two types of incomplete responses were first identified and excluded 268 parents were included for analysis in the study.The first type of incomplete response was full missingness where participants did not respond to any part of the survey except for the consent statements (N = 63).The second type was very early drop-outs where only a few child demographic questions were answered and there was not any information provided on the main study variables of interest (N = 43).
In terms of child demographics (Table 1), 76.9% of children with autism were male and the mean child age was 8.9 years (SD = 3.4); 39.6% of children were reported to have a co-occurring disability, most of whom were reported to have an intellectual disability/developmental delay (86.8%).
As shown in Table 2, 75.0% of participants were mothers.The mean parent age was 37.7 years (SD = 5.8).More than half of parents had a university degree or equivalent (57.1%) and were employed either full-time or part-time (50.0%).In terms of family characteristics, 37.7% of families reported at least one of three deprivation indicators (objective poverty, subjective poverty, and economic hardship, see measure section below).

Parental psychological distress
The Kessler-10 (K10, Kessler et al., 2002) was used to measure parental psychological distress.It consists of 10 items on how participants have been feeling in the last 4 weeks, rated on a 5-point Likert rating scale, with a total score of 0-40.It has been used with good reliability and validity in a Chinese population (Bu et al., 2017).In the current study, the reliability coefficient for the K10 was 0.96 (McDonald's omega).

Child behavior problems
The Behavior Problems Inventory-Short Form (BPI-S, Rojahn et al., 2012) was used to measure behavior problems of children with autism.It consists of a total of 30 items completed by parents, assessing child selfinjurious behavior, stereotyped behaviors, and aggressive/destructive behavior rated on a scale from 0 (never) to 4 (hourly) (total score range 0-120).The Chinese translation of the BPI-S has good reliability and validity in a Chinese population (An et al., 2015;Wei et al., 2021).The reliability coefficient for the BPI-S total frequency score in the current study was 0.91 (McDonald's omega).

Parental perceived social support
The multidimensional scale of perceived social support (MSPSS, Zimet et al., 1988) (Li et al., 2018).The reliability coefficients of the individualism and collectivism subscales in the current study were both 0.80 (McDonald's omega).

Family deprivation
Family deprivation was measured with a composite index, which was created by incorporating three dichotomized indicators described below, namely: subjective poverty, objective poverty, and economic hardship.The three dichotomized indicators were summed into one deprivation composite index score ranging from zero to three with higher values indicating higher levels of deprivation.The composite index score was further dichotomized to identify families experiencing deprivation (having one or more deprivation indicators) versus not experiencing deprivation (having no deprivation indicators).
Subjective poverty was measured with the question "how well would you say you are managing financially these days."A five-point response scale, with 1 = living comfortably and 5 = finding it very difficult, was dichotomized by collapsing "finding it quite difficult" and "finding it very difficult" to indicate subjective poverty, and "living comfortably," "doing alright," and "just about getting by" as indicating no poverty.Objective poverty was measured by using a dichotomous question "is the total annual income of your household in the past year lower or higher than" followed by a given household income value.The household income value was calculated based on information about the province and residence area the participant had selected from an earlier drop-down menu and the number of people reported as being in the household.The provincial yearly minimum living security standard poverty line per capita produced annually by each provincial government separately for urban and rural areas in mainland China (Qi & Wu, 2019) was multiplied by the number of people in the household to obtain a bespoke comparison income value for the participant.The household was classified as experiencing objective poverty if the household income in the past year was below the given household income value.
Economic hardship was measured with the question "Suppose you had to raise 5000 Yuan for an emergency, how hard would it be for you to get that money?"A fourpoint response scale, with 1 = "I could easily raise the money" and 4 = "I don't think I could raise the money," were dichotomized by collapsing "I would have to do something drastic to raise the money (e.g., selling an important possession)" and "I don't think I could raise the money" to indicate economic hardship, and "I could easily raise the money" and "I could raise the money, but it would involve some sacrifices" to indicate no economic hardship.This question was a similar format to that used in national surveys in other countries (e.g., Sanson et al., 2002).

Parent and family demographic variables
Parents were asked about their level of complete education (primary school, junior high school, senior high school, secondary vocational and technical school, college, university graduate degree, and university postgraduate degree).Responses were dichotomized into a single variable, by collapsing "college," "university graduate degree" and "postgraduate degree" to indicate university education, and "primary school," "junior high school," "senior high school," "secondary vocational and technical school" to indicate not having a university education.Household composition was measured using a single item which had three categories (single-parent household, twoparent household, and extended family household).Family size was measured based on a continuum of the total number of people living in the household, ranging from 2 to 20 people.Parent employment status was measured using a single item on current employment status with three responses (employed full-time, employed part-time, and not employed).A dichotomised score was created with "employed full-time" and "employed part-time" collapsed to indicate being employed and with "not employed" indicating not being employed.

Analysis
All statistical analyses were pre-registered and performed using R-Version 4.3.0.Descriptive statistics, including means, standard deviations, and bivariate Pearson correlations were computed for all continuous variables.As shown in Figure 1, for the first research aim, a simple linear regression model was fitted to examine whether child behavior problems (independent variable, X) were associated with parental psychological distress (dependent variable, Y).For the second aim, PROCESS moderation model 1 (Hayes, 2017) was used to investigate whether the relationship between child behavior problems and parental psychological distress was conditional on parental perceived support (Moderator, W).For the third aim, the PROCESS moderated moderation model 3 (Hayes, 2017)  F I G U R E 1 The moderated moderation model of parental perceived support as a primary moderator and parental individualism/ collectivism as a secondary moderator in the relationship between child behavior problems and parental psychological distress.
examine whether the moderation of parental perceived support was conditional on parental individualism and collectivism (Moderator, Z).Bootstrap methods with 5000 resamples were used to verify moderation results and to produce 95% confidence interval (Hayes, 2017).Model diagnostics were performed by plotting residuals to verify whether all computed linear regression models were appropriate, including QQ plot for normality, and residuals versus predicted values plot for homoscedasticity and linearity (Chen & Chen, 2021).Sensitivity analysis was used to test the robustness of findings after the inclusion of two blocks of control variables (block 1: family deprivation and parental education level; block 2: household composition, family size, and parental employment status).Control variables were selected according to their theoretical importance to the outcome variable (Raina et al., 2004;Tint & Weiss, 2016).Given the sample size (Harrell, 2015), two multiple linear regressions were fitted introducing each block of control variables for each analysis.Whether models with the addition of control variables improved the predictive power of main analysis models was evaluated by performing model comparisons using the likelihood ratio test and information criterion such as Akaike information criterion and Bayesian information criterion (BIC) values.If the p value of the likelihood ratio test is larger than 0.05 or there is no difference between information criterion values, it was concluded that the model with additional control variables did not provide a better fit.
Missing data were first described with frequencies and percentages calculated for variables included in the main analysis models (see Table S1).Next, data were iteratively checked if missing at completely random (MCAR) by using the Little's test (Little, 1988), MCAR, or missing not at random by calculating relationships between missingness and observed data.Results of the Little's MACR test indicated that data were missing completely at random (χ 2 (220) = 251, p = 0.076), inferring that model results with complete case analysis were not biased by data missingness.

Descriptive analysis
The means, standard deviations, and bivariate correlations for main study variables are presented in Tables 3  and 4. All independent variables showed significant correlations with the dependent variable (i.e., parental psychological distress) in expected directions.Child behavior problems (r = 0.23, p < 0.001) and parental individualism (r = 0.29, p < 0.001) were both significantly positively correlated with parental psychological distress.Parental perceived support (r = À0.39,p < 0.001) and parental collectivism (r = À0.15,p = 0.05) were both significantly negatively correlated with parental psychological distress.

Association between child behavior problems and parental psychological distress
For research aim 1 (results displayed in Table 5), the simple regression model examining the association between child behavior problems and parental psychological distress was overall significant (F(1, 200) = 11.64,p < 0.001), accounting for approximately 6% of the variance in parental psychological distress.Child behavior problems were significantly positively associated with parental psychological distress (b = 0.17, SE = 0.05, p < 0.001).

Moderating effect of parental perceived support
For research aim 2 (results presented in Table 6), the overall moderation model was significant (F(3, 193) =15.69, p < 0.001), explaining 20% of the variance of parental psychological distress.There was, however, not a significant interaction between child behavior problems and parental perceived support on parental psychological distress (b = 0.02, SE = 0.04, p = 0.658).Increased perceived support was associated with lower levels of parental psychological distress (b = À2.46,SE = 0.44, p < 0.001).

Moderated moderation role of parental individualism and collectivism
For research aim 3 (results summarized in Table 7), the overall moderated moderation models were significant for analysis of parental individualism as a second moderator (F(7, 143) = 6.47, p < 0.001) and for analysis of parental collectivism as a second moderator (F(7, 151) = 5.59, p < 0.001), each model explaining 24% and 21% of the variance of parental psychological distress respectively.With the inclusion of parental individualism or collectivism, the interaction term for child behavior problems and parental perceived support remained nonsignificant.In addition, there were no significant threeway interactions between child behavior problems,

Sensitivity analysis
With the sequential inclusion of the two blocks of control variables to the main analysis models, the pattern of results of the main effect model of child behavior problems on parental psychological distress (see Table S2), the moderation model of parental perceived support (see Table S3), and the moderated moderation model of parental individualism (see Table S4) and collectivism (see Table S5) did not change, suggesting these model results are robust after the inclusion of the control variables.Although the inclusion of the first block of control variables (family deprivation and parental educational level) significantly improved the predictive power of the model (based on ΔR 2 ), the originally estimated model for each research aim was still retained for the sake of model succinctness and because we prespecified the additional of control variables as a sensitivity analysis only.Further checking the contribution of the first block of control variables to main analysis models, family deprivation was consistently significantly associated with parental psychological distress across models ( p < 0.001), suggesting that parents living in more deprived circumstances have higher levels of psychological distress.

DISCUSSION
The results of this study corroborated hypothesis 1 that child behavior problems are significantly associated with parental psychological distress.Contrary to hypothesis 2, parental perceived support did not moderate the relationship between child behavior problems and parental psychological distress.As for the exploration of whether parental cultural values influence the buffering role of parental perceived support, there was no evidence of the further moderation function for parental individualism or collectivism on parent psychological distress.Consistent with current research (Yorke et al., 2018), child behavior problems were associated with parental psychological distress in this sample of parents of children with autism residing in mainland China.Perhaps regardless of nationality or culture, behavior problems of children with autism are a stable and important correlate of parental wellbeing ( Čoli c et al., 2022).Replicating findings from some previous studies in Western cultural contexts (Higgins et al., 2023;Robinson & Weiss, 2020), we found no evidence of a moderating role for parental perceived support in the relationship between child behavior problems and parental wellbeing as would be predicted by a stress-buffering model of social support across cultures.One plausible explanation for the failure to find evidence of this buffering role may be the way that social support is conceptualized and measured.Parental overall perception of sources of support (from families, friends, and significant others) was examined in the present study.Findings may well have been different if we had clearly distinguished commonly studied different dimensions of social support (e.g., instrumental or emotional support, Dunst, 1984;Shepherd et al., 2020).However, we did confirm (consistent with previous research in China ;Ban et al., 2021) that parental perceived support is associated with parental psychological distress, suggesting that social support may serve a compensatory rather than a protective function (Higgins et al., 2023;Robinson & Weiss, 2020).
We also found no evidence that cultural beliefs enter into a moderated moderation process with social support in understanding parental psychological distress, although individualism was associated with increased psychological distress as a main effect.This finding may support a personalityÀcultural clash hypothesis, whereby individualism values discrepant from typical values (collectivism in Chinese society) could be a risk factor for individuals' mental health outcomes (Caldwell-Harris & Ayçiçegi, 2006).In the Chinese context where interdependence is dominant, parents of children with autism with a primarily independent view of self may experience higher levels of psychological distress.The discrepancy between a parental individualist orientation and societal collectivist values has been identified as a stressor for parental psychological health (Caldwell-Harris & Ayçiçegi, 2006).This discrepancy may also be associated with additional risks to the parental coping process (e.g., less social capital and connectedness with ingroups, Chen et al., 2015) that is particularly undermined in the context of having a child with autism.Contrary to the personalityÀcultural clash hypothesis, collectivism did not show the anticipated negative association with parental psychological distress.The reason for this is unclear.
T A B L E 7 Moderated moderation analyses of parental individualism or collectivism as a moderator of the moderation role of parental perceived support between child behavior problems and parental psychological distress.Family deprivation as a covariate consistently showed a significant association with parental psychological distress across hypothesized models, suggesting that it is an additional contextual stressor for psychological wellbeing of parents with children with autism (Emerson et al., 2010;Perry, 2004).This finding is broadly in line with the perspective of ecological and system theories (e.g., Bronfenbrenner, 1986;Cox & Paley, 1997) which contend that each individual is affected by the family context.Family deprivation may also influence parental psychological wellbeing via mediating (e.g., social participation and co-parenting, Sim et al., 2018) and interactive effects (e.g., parent gender, Seymour et al., 2018) with other variables.

Limitations and future research
The present study has some limitations.First, examining the study variables as overall global (rather than dimensional) constructs may preclude finding the possible existence of moderating effects.Future research assessing specific dimensions underlying the overall global constructs of study variables could provide a more nuanced understanding of relationships among these variables.Explorations based on other alternative conceptualizations of social support are also warranted.Second, a more nuanced perspective on the conceptualization and operationalization of parental culture values may be required.For example, in addition to conceptualizing parental culture values as bipolar, there may be multiple distinct underlying dimensions (e.g., horizontal and vertical individualism and collectivism, Triandis & Gelfand, 1998).Third, the study sample is predominantly highly educated mothers.Limited heterogeneity in sample characteristics may restrict the generalization of research finding to broader social demographics.Future research could mitigate this by introducing strategies to over-sample fathers to participate and those from a wider variety of societal groups.Fourth, the cross-sectional design of the study cannot be used to establish causal inference of relationships among examined variables.Research such as the current study requires replication with mainland Chinese families in longitudinal research designs.Finally, cases of early dropout from the survey could not be examined clearly in data missingness check because of a lack of completion of main study variables.

Practical implications
The replication of a significant association between child behavior problems and social support with parental wellbeing in this sample of Chinese parents of children with autism highlights that interventions to reduce child behavior problems and increase social support could be tested in the Chinese context.Given the association between individualism beliefs and parental psychological distress, interventions targeting improvements in parental wellbeing in China will need to be adapted carefully and tested in the Chinese cultural context.Families with children with autism in China with high levels of behavior problems, low levels of social support and individualist cultural beliefs in potential conflict with the prevailing collectivist culture may be at particular risk for high levels of parental psychological distress and thus in more need of services and supports.In addition, family deprivation was a contextual risk for parental psychological distress.Thus, when developing and delivering services and support for children of autism and their parents, practitioners should take into consideration additional parental and familial demands such family socioeconomic context.
Bivariate correlation matrix for study variables.Simple regression analyses of child behavior problems in predicting parental psychological distress.Moderation analyses of parental perceived support as a moderator of the relationship between child behavior problems and parental psychological distress.
T A B L E 4