Exploring how parental mental health and rearing styles relate to children’s mental health: a cross-sectional study among migrant and local primary school students in China

Abstract

Objectives

To exploratorily examine whether parental rearing styles mediate the association between parental mental health and children’s anxiety tendencies and to explore whether these associations differ between local urban and rural-to-urban migrant children in China.

Design

Cross-sectional survey.

Setting

Two primary schools in Hangzhou, eastern China: one primarily serving rural-to-urban migrant children and one serving local urban children.

Participants

A total of 929 children in grades 4–6 and 1273 parents participated. Inclusion criteria were parental consent and child assent; exclusion criteria were inability to complete questionnaires.

Primary and secondary outcome measures

Children’s mental health was assessed using the Mental Health Test; parental anxiety and depression were measured with the Generalised Anxiety Disorder-7 and Patient Health Questionnaire-9. Parental rearing styles were assessed with the Egna Minnen Beträffande Uppfostran and the Hereford Parents’ Attitudes Survey. Mediation and structural equation modelling were applied to test associations.

Results

In exploratory analyses, maternal anxiety was significantly associated with children’s self-blaming and sensitivity tendencies. Parental rearing styles, particularly overprotection and emotional warmth, mediated the relationship between maternal anxiety and children’s mental health outcomes. Mediation effects were stronger among urban children compared with migrant children.

Conclusions

Our results indicate that maternal anxiety is associated with children’s mental health, with parental rearing styles acting as mediators. Although the cross-sectional and exploratory nature limits causal inference, the findings suggest that interventions addressing both parental mental health and rearing practices may be helpful in promoting children’s mental well-being, particularly in migrant populations. Further longitudinal research is needed to clarify these pathways and to understand how contextual factors shape parent-child dynamics.

STRENGTHS AND LIMITATIONS OF THIS STUDY.

• This study simultaneously collected data from both children and their parents using standardised and validated instruments, allowing a more comprehensive assessment of family-level influences on children’s mental health.

• A cluster sampling approach was adopted to recruit participants from both migrant and local schools within the same district, helping to control for geographic and socioeconomic confounding.

• Analyses were exploratory and based on cross-sectional data, which limits causal inference and means findings should be interpreted as hypothesis-generating rather than confirmatory.

• The response rate from fathers was relatively low, which reduced the statistical power of analyses involving paternal data.

• Local and migrant children were recruited from two different schools, which may introduce school-level confounding despite efforts to minimise differences by selecting schools from the same district.

Introduction

Rural-to-urban migration is a pervasive phenomenon in China, propelled by the rapid pace of modernisation and urbanisation. This migration involves a wide array of populations, with migrant children facing distinct mental health challenges compared with their urban peers. Studies have consistently shown that migrant children tend to exhibit higher levels of anxiety and other mental health problems than local children. For example, meta-analyses have found that migrant children report significantly higher anxiety across most subscales of the Mental Health Test (MHT) compared with local urban children.¹ These mental health difficulties in childhood are particularly concerning, as they are strongly linked to an elevated risk of persistent mental health issues in adulthood, including depression, anxiety, substance abuse and suicidal ideation.² Additionally, early childhood mental health problems are associated with long-term physical health issues, such as cardiometabolic dysregulation.³

Research has shown that children migrating with their parents tend to report better health outcomes than left-behind children, with lower levels of depression.⁴ However, despite this, there is still a significant gap in the mental health of migrant children compared with local urban children, with no comprehensive strategies in place to address these disparities effectively. As of 2021, the National Bureau of Statistics of China reported approximately 9.84 million migrant children in primary schools and this number is expected to rise, especially with ongoing governmental efforts to improve social support for migrant workers and their families. Specialised schools have been established in urban areas to accommodate migrant children, presenting an opportunity for targeted interventions aimed at improving their mental health outcomes. However, the specific risk factors driving the increased prevalence of mental health issues in migrant children remain insufficiently defined.

One potential risk factor that may help explain the mental health challenges faced by migrant children is the mental health of their parents. Parental mental health has been shown to significantly affect children’s well-being, both through genetic inheritance and environmental influences. For example, children of parents with anxiety or depressive disorders are at increased risk of developing these disorders themselves^{5 6} and maternal depression during pregnancy has been associated with a variety of child mental health problems, including the internalising and externalising behaviours.⁷ Migrant workers, who are parents of migrant children, are at higher risk for mental health issues, with studies showing that 28% of migrant workers experience depressive symptoms,⁸ a rate higher than that of non-migrant workers.^{9 10}

Parental mental health also affects parenting behaviours. Anxious parents often engage less effectively with their children, displaying withdrawal¹¹ or overprotectiveness,¹² which can exacerbate children’s mental health issues. Moreover, parental rearing styles, such as emotional warmth, overprotection and rejection, have been found to significantly influence children’s mental health outcomes. Parenting has been shown to account for a substantial portion of the variance in child anxiety and depression, with 4% of the variance in anxiety and 8% in depression.¹³ A recent meta-analysis further reported that parental overprotection was positively associated with both internalising and externalising problems in children.¹⁴ Research has also highlighted that lack of emotional warmth, being over-involved, rejecting and punitive in parental rearing skills are linked to internet addiction disorder and other comorbid mental issues in adolescents.¹⁵ Studies in migrant families have also reported that paternal emotional warmth can reduce social anxiety in migrant adolescents, while maternal overprotection exacerbates it.¹⁶

Building on this body of research, our study aims to explore the hypothesis that parental rearing skills may serve as a mediator in the relationship between parental mental health and the mental well-being of migrant children. Specifically, we examine how the mental health of parents influences their rearing styles and attitudes, and how these parenting practices, in turn, affect the mental health of their children. To test this hypothesis, we conducted a cross-sectional survey involving children and their parents from two primary schools in Hangzhou, China—one serving local urban children and the other serving migrant children. Our primary objective was to examine whether parental rearing styles mediate the relationship between parental mental health and children’s mental health using structural equation modelling (SEM). The secondary objectives were: (1) to exploratorily explore whether the mediation pathways differ between local urban and rural-to-urban migrant children in China and (2) to compare the mental health status of children and parents, as well as parental rearing styles and attitudes reported by both parents and children, between the two groups, in order to contextualise the observed mediation effects.

Methods

Research procedures and participants

We conducted a cross-sectional survey between September and November 2022, employing a cluster sampling strategy. The study was carried out in two primary schools in Hangzhou, the capital of Zhejiang province, one of the most developed provinces on the east coast of China. To select the participating schools, we first identified all seven primary schools serving rural-to-urban migrant children in Hangzhou and randomly chose one. We then randomly selected a primary school for local children within the same district as the selected school for migrant children. Both selected schools agreed to participate in the study, with support from both the head teacher of each class and school-based mental health professionals.

We initially explained the study to parents and obtained consent for their children’s participation. Additionally, we invited at least one parent to participate by distributing online questionnaires. Subsequently, all students in grades 4, 5 and 6 whose parents provided consent were surveyed. Prior to completion, each item of the questionnaire was thoroughly explained by a trained mental health professional. Participants needed to be able to complete the survey independently.

Public involvement

The public were not involved in the design, conduct, reporting or dissemination plans of this research.

Measurement

Demographic characteristics

Demographic and socioeconomic data were collected from children and parents, including age, sex, grade, only-child status, parental education levels, family income and frequency of time spent with children. Parental education was coded as an ordinal variable from 1 (primary school or less) to 7 (Ph.D.). Annual family income was categorised as an ordinal variable from 1 (< 5000 CNY) to 6 (>100 000 CNY). The frequency of spending time with children was also coded as an ordinal, with values ranging from 1 (almost every day) to 8 (less than once a year), where higher values indicated less frequent interaction.

Children’s mental health

We used the MHT scale, which was adapted from Kurt Suzuki’s ‘Diagnostic Test of Anxiety Tendency’ into Chinese by Zhou,¹⁷ to assess the anxiety tendencies in children. The Cronbach’s α coefficient for the total scale was 0.92, and those for the subscales ranged from 0.63 to 0.80 in samples of students from grade 4 in primary school through senior high school.¹⁷ Cronbach’s α was also calculated for each MHT subscale using the ‘psych’ package in R in our dataset. The standardised alpha values ranged from 0.80 for the Impulsive Tendency subscale to 0.85 for the Study Tendency subscale, indicating good internal consistency across all subscales. This scale consists of 100 items, each scored 1 for ‘yes’ and 0 for ‘no’.^{1 18} The scale includes eight subscales: learning anxiety, interpersonal anxiety, loneliness, self-blaming tendency, sensitive tendency, somatic anxiety, phobia anxiety and impulsive tendency. A total score of 8 or higher on any subscale indicates a psychological issue. The scale also includes a validity subscale, where a score of 8 or more suggests unreliable response. We excluded subjects with unreliable responses from the analysis.

Parental rearing skills assessment

Two established scales were employed to assess parental rearing skills—one completed by the parents and the other by the children. The Chinese version of Egna Minnen Betraffande Uppfostran (Swedish for ‘My memories for upbringing’) for children (EMBU-C),¹⁹ translated by Yue et al,²⁰ was used to evaluate parental rearing styles. The scale includes six domains: emotional warmth, overprotection, over-interference, favouritism, rejection and punishment.²⁰ Overprotection and over-interference were combined into one domain for mothers, based on principal component analysis among healthy Chinese populations.²⁰ The EMBU-C contains 66 items rated on a 4-point Likert scale (1=never; 4=always), completed by the children. The Chinese version of the Hereford Parents’ Attitudes Survey (PAS)²¹ was used to assess parental attitudes in five domains: self-confidence in parental roles, attribution of the child’s behaviour, mutual understanding, and acceptance of the child’s behaviour and feelings. The Chinese version of PAS demonstrated a Cronbach’s α of 0.85, with inter-scale correlations ranging from 0.39 to 0.61.²² It has 40 items rated on a 5-point Likert scale (1=completely disagree; 5=completely agree), completed by parents.

Parents’ mental health

To assess parental anxiety, we used the Generalised Anxiety Disorder-7 (GAD-7)²³ scale, which contains 7 items. The first three items address the core criteria of Generalised Anxiety Disorder as per Diagnostic and Statistical Manual of Mental Disorders-IV (DSM-IV). Responses are scored from 0 (not at all) to 3 (nearly every day), and a cut-off of 10 was used to indicate anxiety in our analysis.²⁴ For assessing depression, the depression module of the Patient Health Questionnaire (PHQ-9)²⁵ was used. PHQ-9 consists of the nine items based on DSM-IV criteria for depressive disorder, scored from 0 (not at all) to 3 (nearly every day). A cut-off of 15 was used to indicate moderately severe depression.²⁶

Analysis plan

We employed mediation analysis within an SEM framework to examine our hypothesis that parental rearing skills may mediate the relationship between parental mental health and children’s mental well-being. In the model, the independent variable (X) was parental mental health, the dependent variable (Y) was children’s mental health, and the mediator (M) was parental rearing style. We first explored the associations between X→M and X→Y, followed by the associations between M→Y. Finally, we examined the mediation effects. Because X, M and Y each comprised multiple variables derived from different scales and subscales, we applied false discovery rate (FDR) correction (q<0.05) to adjust for multiple comparisons in the X→M, X→Y and M→Y analyses separately.

Statistical analysis

There were no missing data in the collected responses as all survey questions were set as mandatory. To ensure data quality, we used the validity subscale of the MHT to identify and exclude students who might have provided unreliable responses. Consequently, a complete-case analysis was conducted on the full sample of parents and students whose validity subscale scores were below 8.

Demographic characteristics between migrant and local urban children were compared using two-sample t-tests for continuous variables and the χ² test for categorical variables. To compare the mental health status of children and parents between the two groups, logistic regression analyses were performed using each binary mental health outcome as the dependent variable. As an exploratory step, unadjusted models were first tested. For analyses of children’s mental health, we further included demographic and socioeconomic variables that show significant differences between groups as covariates to explore their independent effects. To compare parental rearing styles and attitudes reported by both parents and children between the two groups, linear regression analysis was performed with EMBU-C scores and PAS scores as outcomes.

Next, we analysed the association between X→Y and M→Y using logistic regression. In these models, the MHT binary variables were treated as the outcomes, with parent groups (anxious/depressed mothers/fathers), EMBU-C scores and PAS scores as independent variables. To test whether the effect is different between local and migrant children, we included an interaction term between children’s group status (migrant vs local) and these independent variables. Additionally, linear regression was used to examine the association between X→M. Binary variables were used for X and Y (mental health outcomes) to facilitate interpretation and ensure clinical relevance. The underlying statistical assumptions for the linear and logistic regression models were rigorously assessed using the car package in R. For the linear regression models, residual plots were visually inspected to confirm that the assumptions of linearity, independence and homoscedasticity were satisfied, and quantile-quantile (Q-Q) plots were used to assess the normality of residuals. For the logistic regression models, the Box-Tidwell test was used to evaluate linearity in the logit. In addition, the variance inflation factor was calculated for all models to assess potential multicollinearity.

Mediation models and SEM models were fitted using the lavaan R package. We calculated 95% CI and p values based on 1000 bootstrap resamples. In the mediation model, independent variables were those about parental mental health factors significantly associated with children’s mental health and the outcome was children’s mental health. Mediators were the parental rearing styles/attitudes that were significantly associated with both the independent variables and the outcome. Continuous scores, rather than binary variables, were used, and all variables have been standardised in the models. Significant mediators were then included in the same model for SEM analysis, with non-significant mediators removed to improve model fit. Model fit was assessed using the root-mean-square error of approximation (RMSEA), the comparative fit index (CFI) and the Tucker-Lewis index (TLI), where RMSEA <0.1 and CFI/TLI >0.9 indicated good model fit.

FDR correction was applied to adjust for multiple comparisons. A significance level of p<0.05 (two-sided) or FDR q<0.05 was used for all analyses. All statistical analyses were performed using R software V.4.1.0 (https://www.r-project.org/).

Results

Demographic characteristics of students and their parents

All eligible students in the selected schools were invited to participate, resulting in 1001 children aged 9–13, including 518 (51.7%) local children and 483 (48.3%) migrant children (table 1). Among the participants, migrant children had a significantly higher proportion of boys (χ²=6.27, p=0.01) and fewer only-children (χ²=74.19, p<0.001) compared with local children. No significant differences were observed in terms of mean age, grade distribution and the response rates of fathers and mothers across the two groups.

Table 1. Demographic and socioeconomic characteristics of children and their parents.

	Local children, n=518 (51.7%)		Migrant children, n=483 (48.3%)		t	P value
	Mean	SD	Mean	SD
Age	10.75	0.90	10.82	0.92	1.19	0.28
	n	%	n	%	χ2
Sex (boy)	267	51.50	288	59.60	6.27	0.01
Only child Status	247	47.70	133	27.50	74.19	<0.001
Grade					1.11	0.89
Fourth grade	169	32.60	158	32.70
Fifth grade	176	34.00	172	35.60
Sixth grade	173	33.40	153	31.70
Income level reported by children					90.90	<0.001
Low	7	1.40	17	3.50
Lower-middle	21	4.10	90	18.60
Middle	237	45.80	241	49.90
Upper-middle	224	43.20	103	21.30
High	29	5.60	32	6.60
Number of father participants	261	50.40	221	45.80	1.96	0.16
Number of mother participants	411	79.30	381	78.90	0.01	0.92
	Mean	SD	Mean	SD	t	P value
Father’s age	41.18	5.67	39.35	5.78	12.27	<0.001
Mother’s age	39.13	4.46	37.03	5.07	38.37	<0.001
Father’s educational levels	5.07	0.99	2.87	1.08	540.06	<0.001
Mother’s educational levels	4.86	0.86	2.67	1.03	1060.30	<0.001
Income level reported by father	3.05	0.71	2.04	0.90	191.22	<0.001
Income level reported by mother	3.05	0.61	2.03	0.90	353.34	<0.001
Frequency of accompanying children (father)	1.59	0.89	1.58	1.00	0.02	0.89
Frequency of accompanying children (mother)	1.27	0.82	1.44	0.90	8.29	0.004

Among 1273 parents who participated in the study, a larger proportion were mothers (n=792, 79.1%) compared with fathers (n=481, 48.2%). Mothers also reported spending more time with their children than fathers did. Additionally, mothers of local children spent more time with their children than mothers of migrant children (t=8.29, p=0.004). Parents of migrant children tended to be younger, had lower education levels and reported lower family incomes compared with parents of local children (p<0.001).

Differences in the mental health of children and parents, parental rearing styles and attitudes between the two schools

Following quality control, MHT scores from 929 children were included in the analysis. The prevalence of anxiety was 6.5% (n=23) among local children and 15.3% (n=58) among migrant children (table 2). The most common issue reported was learning anxiety, with a prevalence of 31.2% (n=127) for local children and 54.1% (n=205) for migrant children. Without adjusting for covariates, migrant children exhibited significantly higher levels of anxiety across all subscales compared with local children (figure 1A). After adjusting for the demographic and socioeconomic characteristics of children and their fathers, the difference remained statistically significant. However, when adjusting for the characteristics of children and their mothers, only the difference in learning anxiety remained significant (table 2).

Table 2. Difference in children’s anxiety tendency between the two schools after adjusting for demographic and socioeconomic variables.

	Local children, n=518 (51.7%)	Migrant children, n=483 (48.3%)	No adjustment, OR (95% CI)	Adjusted A, AOR (95% CI)	Adjusted B, AOR (95% CI)	Adjusted C, AOR (95% CI)	Adjusted D, AOR (95% CI)
Learning anxiety	127 (31.2%)	205 (54.1%)	2.6 (1.94 to 3.48)***	2.27 (1.67 to 3.09)***	2.36 (1.40 to 3.98)**	1.97 (1.02 to 3.78)*	4.54 (1.43 to 14.37)*
Interpersonal anxiety	10 (2.4%)	27 (7.1%)	3.17 (1.51 to 6.64)**	2.94 (1.36 to 6.36)**	1.62 (0.46 to 5.72)	11.4 (2.59 to 50.25)**	6.53 (0.49 to 87.69)
Loneliness	3 (0.7%)	14 (3.7%)	5.37 (1.53 to 18.83)**	5.31 (1.46 to 19.26)*	12.47 (0.88 to 175.71)	20.26 (2.56 to 160.32)**	Uninformative (0 to ∞)
Self-blaming tendency	37 (8.8%)	81 (21.4%)	2.82 (1.86 to 4.28)***	2.5 (1.61 to 3.87)***	1.95 (0.94 to 4.01)	5.86 (2.36 to 14.58)***	6.54 (1.43 to 30.01)*
Sensitive tendency	25 (6.1%)	56 (14.8%)	2.68 (1.64 to 4.4)***	2.39 (1.42 to 4.01)**	1.38 (0.58 to 3.29)	3.11 (1.01 to 9.54)*	0.94 (0.10 to 9.10)
Somatic anxiety	22 (5.3%)	49 (12.9%)	2.65 (1.57 to 4.48)***	2.36 (1.37 to 4.08)**	1.09 (0.43 to 2.76)	4.92 (1.61 to 15.09)**	1.65 (0.25 to 10.69)
Phobia anxiety	11 (2.6%)	34 (9.0%)	3.66 (1.83 to 7.34)***	3.53 (1.71 to 7.29)***	2.6 (0.80 to 8.42)	4.51 (1.19 to 17.13)*	6.63 (0.56 to 79.10)
Impulsive tendency	4 (1.0%)	18 (4.7%)	5.14 (1.72 to 15.31)**	5.27 (1.69 to 16.42)**	4.39 (0.79 to 24.43)	1.49 (0.14 to 16.20)	0.24 (0.00 to 41.92)
MHT total	23 (6.5%)	58 (15.3%)	2.61 (1.57 to 4.33)***	2.06 (1.21 to 3.50)**	1.34 (0.55 to 3.30)	3.74 (1.31 to 10.73)*	1.36 (0.21 to 8.80)

*P value <0.05; **p value <0.01; ***p value <0.001.

Adjusted A: adjusted for children’s sex, only-child status and family income reported by them.

Adjusted B: in addition to the adjustments in A, mothers’ age, educational level, family income reported by them and time spent with children by mothers were also adjusted.

Adjusted C: in addition to the adjustments in A, fathers’ age, educational level, family income reported by them and time spent with children by fathers were also adjusted.

Adjusted D: all variables in A, B and C were adjusted.

MHT, Mental Health Test.

Figure 1. The association among the mental health of students and parents, and parental rearing skills, and their differences between migrant and local urban children. (A) Differences in the mental health of parents and children between local and migrant groups (logistical regression). (B) Differences in the parenting skills between parents of local and migrant children (linear regression). (C) The effect of parents’ mental health on their rearing styles (linear regression). (D) The effect of parents’ mental health on their rearing attitudes (linear regression). (E) The effect of parents’ mental health on children’s mental health (logistic regression). No covariates were included in these exploratory analyses. EMBU, Egna Minnen Betraffande Uppfostran; MHT, Mental Health Test; PAS, Hereford Parents’ Attitudes Survey.

The prevalence of anxiety assessed by GAD-7 was significantly higher in both fathers (18.1% vs 9.2%) and mothers (19.4% vs 11.2%) of migrant children compared with those of local children (figure 1A). Depression, as assessed by PHQ-9, was also significantly more common in mothers of migrant children (25.7% vs 18.5%). Although the prevalence of depression was higher among fathers of migrant children (24.0% vs 19.9%), the difference was not statistically significant.

Parental rearing styles, assessed using the EMBU-C, showed significant differences between the two groups. Migrant children reported lower levels of emotional warmth and higher levels of overprotection/over-interference, favouritism, rejection and punishment by both fathers and mothers. Regarding parental attitudes, except for the attribution factor, mothers of migrant children scored lower on all three other PAS factors compared with mothers of local children (figure 1B). Fathers showed no significant differences in the attribution or understanding factors.

The effect of parental mental health on their rearing styles/attitudes and the mental health of children

Mothers with anxiety had a significantly lower level of emotional warmth and a higher level of overprotection/over-interference, punishment and rejection, as reported by their children, after FDR adjustment (q_FDR<0.05) (figure 1C). These mothers also had lower scores across all subscales of PAS (q_FDR<0.05) (figure 1D). Their children exhibited higher levels of self-blaming tendency (q_FDR=0.009) and sensitive tendency (q_FDR=0.009) (figure 1E). Mothers with depression also reported significantly lower levels of emotional warmth (q_FDR=0.03) and lower scores on all subscales of PAS (q_FDR<0.05). Their children demonstrated a higher level of sensitive tendency, though the difference was not significant after FDR adjustment (p=0.008, q_FDR=0.09). No significant effect of fathers’ mental health on the mental health of their children was found. When testing the interaction between the children’s group status and parental mental health, we found a significant interaction between mothers’ anxiety and children’s somatic anxiety. Specifically, mothers’ anxiety was found to increase somatic anxiety in local children only (OR=3.80 (95% CI 1.37 to 10.59), p=0.01) (online supplemental figure S1).

The effect of parental rearing styles/attitudes on the mental health of students

The association between parental rearing styles/attitudes and the mental health of children was shown in table 3 and online supplemental table S1. Emotional warmth was negatively associated with all subscales of the MHT, while overprotection/over-interference, rejection and punishment were positively associated. Significant interactions were found for multiple factors of the EMBU-C, with urban children being more affected by parental rearing styles than migrant children. However, the directions of these associations were consistent across both groups (online supplemental table S2). Except for the favouritism factor, all other EMBU-C factors were associated with children’s mental health in both groups.

Table 3. Association between maternal rearing styles/attitudes and children’s mental health^*.

	Learning anxiety	Interpersonal anxiety	Loneliness	Self-blaming tendency	Sensitive tendency	Somatic anxiety	Phobia anxiety	Impulsive tendency
	OR (95% CI)	OR (95% CI)	OR (95% CI)	OR (95% CI)	OR (95% CI)	OR (95% CI)	OR (95% CI)	OR (95% CI)
EMBU-C
Emotional warmth	0.97 (0.95 to 0.98)***	0.96 (0.94 to 0.99)**	0.94 (0.91 to 0.98)**	0.98 (0.97 to 1.00)*	0.97 (0.95 to 0.99)**	0.96 (0.95 to 0.98)***	0.97 (0.95 to 0.99)*	0.95 (0.92 to 0.98)**
Overprotection/over-interference	1.08 (1.06 to 1.10)***	1.08 (1.03 to 1.12)**	1.06 (1.00 to 1.12)	1.08 (1.05 to 1.11)***	1.07 (1.04 to 1.10)***	1.07 (1.04 to 1.10)***	1.09 (1.05 to 1.13)***	1.08 (1.03 to 1.14)**
Favouritism	1.04 (0.99 to 1.08)	1.06 (0.97 to 1.16)	1.05 (0.92 to 1.21)	1.06 (1.01 to 1.12)*	1.05 (0.98 to 1.12)	1.07 (1.00 to 1.15)	1.03 (0.94 to 1.12)	1.08 (0.96 to 1.22)
Rejection	1.11 (1.07 to 1.15)***	1.11 (1.05 to 1.17)***	1.13 (1.06 to 1.21)**	1.11 (1.07 to 1.15)***	1.11 (1.06 to 1.15)***	1.14 (1.09 to 1.19)***	1.14 (1.09 to 1.20)***	1.14 (1.07 to 1.21)***
Punishment	1.10 (1.06 to 1.14)***	1.09 (1.04 to 1.14)**	1.13 (1.07 to 1.20)***	1.09 (1.05 to 1.13)***	1.09 (1.04 to 1.13)***	1.10 (1.05 to 1.14)***	1.12 (1.07 to 1.16)***	1.11 (1.05 to 1.18)***
PAS
Self-confidence	0.93 (0.89 to 0.96)***	0.84 (0.76 to 0.92)***	0.83 (0.73 to 0.95)*	0.91 (0.86 to 0.95)***	0.93 (0.88 to 0.99)*	0.95 (0.89 to 1.01)	0.92 (0.86 to 1.00)	0.90 (0.80 to 1.00)
Attribution	0.98 (0.95 to 1.02)	0.89 (0.81 to 0.97)*	0.93 (0.81 to 1.06)	0.96 (0.91 to 1.01)	0.97 (0.91 to 1.03)	0.93 (0.87 to 1.00)	0.99 (0.91 to 1.07)	1.00 (0.89 to 1.11)
Acceptance	0.92 (0.88 to 0.96)***	0.96 (0.87 to 1.06)	0.89 (0.77 to 1.02)	0.93 (0.88 to 0.99)*	0.94 (0.88 to 1.00)	0.98 (0.91 to 1.05)	0.91 (0.83 to 0.99)*	0.93 (0.82 to 1.04)
Understanding	0.94 (0.91 to 0.98)*	0.94 (0.86 to 1.04)	0.84 (0.72 to 0.98)*	0.93 (0.88 to 0.98)*	0.93 (0.87 to 0.99)*	0.96 (0.90 to 1.03)	0.96 (0.89 to 1.05)	0.90 (0.80 to 1.02)

*FDR q value <0.05; **FDR q value <0.01; ***FDR q value<0.001.

^*The association between paternal rearing styles/attitudes and children’s mental health was presented in online supplemental table S1, as including them here would make the table excessively large.

EMBU-C, Egna Minnen Betraffande Uppfostran for children; FDR, false discovery rate; PAS, Hereford Parents’ Attitudes Survey.

Parental rearing attitudes, as reported by parents themselves, have less influence on children’s mental health than rearing styles reported by children. Mothers’ self-confidence in parental role was negatively associated with all MHT subscales, except for somatic anxiety, phobia anxiety and impulsive tendency (table 3). This association was primarily observed in urban children (online supplemental table S2). The attribution factor was negatively associated with interpersonal anxiety, while the acceptance factor was negatively associated with learning anxiety, self-blaming tendency and phobia anxiety. The understanding factor was negatively associated with learning anxiety, loneliness, self-blaming tendency and sensitive tendency. However, when we separated the subjects into local and migrant children, none of the associations were significant, probably due to a smaller sample size. For fathers’ rearing attitudes, significance was only found for associations between the acceptance factor and both learning anxiety and somatic anxiety.

Mediation and SEM models

Since only mothers’ anxiety significantly affected the self-blaming tendency and sensitive tendency of their children, a mediation model was fitted using mothers’ anxiety as the independent variable. We found that four subscales of parental rearing styles/attitudes significantly mediated the effect of mothers’ anxiety on children’s sensitive tendency, including rejection (EMBU-C), overprotection/over-interference (EMBU-C), emotional warmth (EMBU-C) and self-confidence (PAS) (online supplemental table S3). Additionally, five subscales mediated the effect of mothers’ anxiety on the self-blaming tendency of children, including rejection (EMBU-C), overprotection/over-interference (EMBU-C), emotional warmth (EMBU-C), acceptance (PAS) and self-confidence (PAS).

In the final SEM (figure 2; TLI=0.97; CFI=0.94; RMSEA=0.09), a latent variable of anxiety tendency, combining both sensitive and self-blaming tendencies, was created. Three mediators were significantly associated with both mothers’ anxiety and the anxiety tendency of children: overprotection/over-interference (EMBU-C), emotional warmth (EMBU-C) and self-confidence (PAS).

Figure 2. Structural equation model of the relationship among mothers’ anxiety, rearing skills and children’s anxiety tendency. EMBU-C, Egna Minnen Betraffande Uppfostran for children; PAS, Hereford Parents’ Attitudes Survey.

Discussion

We conducted a cross-sectional study to exploratorily investigate the relationship between the mental health of children and parents, and parental rearing styles and attitudes, comparing two distinct groups: one from a school serving local children and the other from a school serving migrant children. Significant disparities were observed in both the mental health of parents and children, as well as in the parental rearing styles and attitudes. Parental rearing practices were associated with the mental health of both generations, with parental reports being more strongly associated with parents’ mental health, while children’s reports had a greater impact on their own mental health. Interestingly, the association between parental rearing skills and children’s mental health also appeared stronger among urban children than migrant children. Mothers’ anxiety was found to significantly increase the self-blaming and sensitive tendency in children from both groups, with somatic anxiety being specifically heightened in local children. Structural equation modelling revealed that several factors of parental rearing, including overprotection/over-interference (EMBU-C), emotional warmth (EMBU-C) and self-confidence (PAS), significantly mediated the relationship between mothers’ anxiety and children’s anxiety tendency.

The higher prevalence of mental health problems among migrant children and their parents has previously been well-documented.^{1 10} Risk factors for the poor mental health of migrant workers include long working hours and poor pay, lack of social support, job dissatisfaction and other stressors.⁸ Most of these risk factors are strongly related to low socioeconomic status (SES). In this study, most of the differences in anxiety tendencies between migrant children and urban children became non-significant after adjusting for demographic and socioeconomic characteristics of parents. Although low parental SES is the main risk factor for the mental health problems of both parents and children, addressing it is challenging. As parental SES is a multidimensional concept that affects children through different pathways, targeting specific mechanisms may offer more effective intervention strategies. A review of the association between parental SES and children’s mental health suggested that children from lower SES families are more likely to experience bullying by peers due to perceived shortcomings.²⁷ However, this pathway is less applicable to our sample, as migrant children attend different schools from local children, and their parents share similar SES levels. Moreover, parents with low SES are often unable to provide an adequate living environment for their children. The influence of living environment on psychological stress and mental health has been well-established, with factors such as inadequate housing, crowding, social disorder, violence and lack of green spaces acting as stressors.²⁸ Recent reviews of the residential environment of migrant workers in China have highlighted issues like marginalised housing conditions, poor neighbourhood environments and residential segregation, all of which have significant impacts on mental health.²⁹ Finally, parents with low SES tend to lack knowledge about parenting skills. Previous research among Chinese parents showed a positive correlation between SES and authoritative parenting (rational and issue-oriented) and a negative correlation with authoritarian parenting (controlling and punitive).³⁰ Our findings also support this, showing that parents of migrant children exhibit more negative parenting styles (eg, over-protection, over-interference, favouritism, rejection and punishment) and fewer positive parenting behaviours (eg, emotional warmth, self-confidence, acceptance and understanding).

Mothers’ anxiety affects children’s anxiety tendency by altering parenting styles and attitudes, specifically by increasing overprotection/over-interference (EMBU-C) and decreasing emotional warmth (EMBU-C) and self-confidence (PAS). Overprotection/over-interference refers to parents being fearful, controlling and overly concerned about their child’s safety, and having high achievement expectations. Cognitive theories for anxiety suggest that anxiety narrows attention and creates an attentional bias towards threats.³¹ Mothers with anxiety may be more attuned to potential threats in their children’s environments, leading them to engage in overprotective behaviours.^{12 32} Emotional warmth is shown through support, loving attention, stimulation and acceptance. A previous American study found that warmth, as measured by the Parental Acceptance and Rejection/Control Questionnaire, was associated with fathers’ depression but not mothers’.³³ This inconsistency may be due to differences in questionnaires used and cultural differences between European parents and Japanese parents.³⁴ While no study has specifically linked maternal anxiety to self-confidence in parenting, one comparative study found that anxious patients had significantly lower positive self-appraisal (ie, a less favourable view of themselves) than euthymic patients.³⁵ Furthermore, anxiety is often comorbid with depression, which can lead to feelings of worthlessness, self-blame and a lack of self-confidence.

In addition to these behavioural explanations, attachment-based emotion regulation theories offer further insight into how maternal anxiety may transmit risk for anxiety to children. Attachment theory proposes that children develop emotion regulation skills through repeated interactions with their primary caregiver; when caregivers are highly anxious, caregiving may become overprotective, inconsistent or less emotionally available, which can foster children’s hypervigilance and self-focused worry. Empirical research in adults supports the central role of attachment-related processes and emotion regulation in anxiety,³⁶ and our study suggests that similar mechanisms may operate within parent–child relationships to transmit risk from anxious caregivers to children. Future longitudinal work directly measuring parental attachment, caregiving behaviour and child emotion regulation will be important to test this pathway more rigorously.

Only mothers’ anxiety, rather than mothers’ depression, or the anxiety and depression of fathers, was significantly related to children’s mental health. We also found that mothers spent more time with their children and a higher percentage of mothers participated in this study, which may partly explain why mothers exert a greater influence on children’s well-being. Our findings align with other studies in China, which have reported stronger effects from mothers on children’s mental health compared with fathers. For example, maternal parenting stress had a more substantial effect on adolescent depressive symptoms than paternal stress.³⁷ In traditional Chinese culture, mothers are generally expected to take on primary caregiving roles, while fathers are seen as the primary breadwinners. Although this dynamic is gradually changing, it has not completely transformed over time. The effect of mothers’ depression on children’s sensitive tendency was significant before correcting for multiple comparisons (p=0.008, q_FDR=0.09), suggesting that the difference between mothers’ anxiety and depression is smaller than that between mothers’ and fathers’ mental health. We infer that mothers with depression may be less engaged in parenting than those with anxiety and therefore have a lesser impact on their children. While we have not found studies directly comparing the parenting skills of mothers with anxiety and mothers with depression, we base this inference on the core symptoms of depression, such as a loss of interest in activities, which may hinder active parenting.

We found significant interactions between the children’s group and mothers’ anxiety, as well as parental rearing styles/attitudes. While the direction of the association was the same between the two groups, the association was stronger in urban children. We infer that the weaker associations in migrant children may be due to a less close parent-child relationship compared to local urban children. A study comparing parent-child attachment among migrant and local children found that the quality of parent-child attachment was statistically significantly lower in migrant children.³⁸ The author reasons that this may be due to deficient communication between migrant children and parents, as migrant workers are often busier with work and livelihood than their local counterparts.³⁸ Another possible explanation is the sex imbalance between the two groups: the migrant sample included a higher proportion of boys than the local sample (59.60% vs 51.50%). Sex differences are well-documented in emotional processing; compared with boys/men, girls/women tend to be more adept at decoding affective cues, exhibit stronger emotional reactivity and express sadness and fear more frequently.^{39 40} Consequently, girls are generally more vulnerable to internalising problems such as anxiety. Recent evidence also shows that emotion regulation difficulties mediate the link between psychopathology and mood symptoms more strongly in females,⁴¹ suggesting greater sensitivity to emotional and environmental influences. Therefore, the higher proportion of girls in the urban group may partly account for the stronger associations observed in this subgroup.

This study has several limitations. First, the response rate from parents, particularly fathers, was low (79.1% for mothers and 48.2% for fathers). As a result, the statistical power of analyses including fathers was limited. Second, local and migrant children were from two different schools, which may have introduced confounding factors unrelated to migration experiences, as the two schools could differ in other ways. While it is difficult to avoid this limitation due to migrant children usually attending different schools, we tried to minimise these differences by selecting schools from the same district, which should reduce potential differences. Third, as a cross-sectional study, it is difficult to infer the direction of the relationship between the mental health of parents and children.

Conclusion

In conclusion, this study identified associations among maternal anxiety, parenting styles and children’s anxiety tendencies, particularly in both migrant and local urban families. Although the cross-sectional and exploratory nature precludes causal interpretation, the findings suggest that maternal anxiety and parenting practices may be important factors to consider when supporting children’s emotional well-being. The stronger associations observed in the urban group highlight the potential influence of contextual and relational factors on parent–child dynamics. These findings underscore the value of promoting supportive parenting practices and addressing parental psychological distress as part of broader efforts to enhance children’s mental health. Future longitudinal and multi-informant studies are needed to clarify temporal relationships, evaluate potential mechanisms and better understand how migration-related contexts shape children’s emotional development.

Data availability statement

Data are available upon reasonable request.