In this cross-sectional study, 70 children with ASD (Mage = 3.52, SD = 0.98), 98 TD children (Mage = 3.07, SD = 0.92), and all their mothers were recruited in Foshan, China, between May 2021 and September 2022. The inclusion criteria of the ASD group were diagnosed as ASD by two chief psychiatrists (SJH and HY) using the Diagnostic and Statistical Manual of Mental Disorders, Fifth Edition (DSM-5) criteria, and the total score in the Children Autism Rating Scale (CARS) of all children with ASD was a score above 36. A group of TD children without ASD diagnosis was also recruited from the same child healthcare clinic. Those who had a history of developmental delay were excluded from the TD group. Additional inclusion criteria for selecting both groups were chronological age (between 2 and 5 years old) and all the children accompanied by mothers. Known genetic or chromosomal abnormalities (such as down syndrome and fragile X syndrome), significant organic diseases (such as blindness and deafness), and comorbidity with other mental disorders (such as childhood schizophrenia and attention-deficit hyperactivity disorder) based on the self-report by mothers were excluded for selecting both groups.

First, the selecting ASD group was recruited after the diagnosis by the doctor and the CARS score was greater than 36. Then, all the children accompanied by their mothers were invited to participate in this study. All mothers provided informed consent before they were individually interviewed by health personnel who received standardized training in conducting face-to-face interviews. Mothers of the two groups were asked to answer many structured questionnaires on behavioral problems in the children as well as in maternal parenting styles and autism-like traits, among others. This study was approved by the Ethics Committee of the School of Public Health at Sun Yat-sen University.

Using self-made questionnaires, we collected the demographic information of the subjects, including maternal age, maternal education, income, child age, and child gender.

The Strengths and Difficulties Questionnaire ((SDQ)-parent form) (38) was adapted by Kou and Du from Goodman et al.’s study (39). The SDQ is a 25-item that uses a 3-point Likert scale (“not true,” “somewhat true,” or “certainly true”) and is designed to assess behavioral and emotional problems in children and adolescents ages 3–17 (38). Previous studies also show that the scale is applicable to 2-year-old children in China (40). Thus, mothers were asked to complete the extended parent version of the SDQ for ASD children. The items are divided into five subscales: hyperactivity/inattention, emotional symptoms, peer problems, conduct problems, and prosocial behavior. The composite scores for internalizing problems (emotional and peer items) and externalizing problems (conduct and hyperactivity items) were then analyzed in the current study (41). The SDQ has good reliability and structural validity in Chinese individuals (38).

The PBI is a 20-item, parent-report questionnaire that uses a 6-point Likert scale (never to always) (42). The questionnaire yields two subscales: support/participation and hostility/coercion parenting styles. It is used to assess the parenting style among preschool and school-age children. Mothers were asked to complete the questionnaire. PBI has good reliability and structural validity in Chinese populations, with Cronbach’s α values of 0.807 and 0.652 for support/participation and hostility/coercion, respectively, for Chinese individuals (43).

Maternal autistic traits were evaluated using the AQ (19). The AQ comprises 50 questions and uses a 4-point Likert scale (absolutely agree to absolutely disagree), with 10 questions assessing five different domains: social skills, attention switching, attention to detail, communication skills, and imagination. Each of these items scores 1 point if the respondent records the autistic-like trait either mildly or strongly. The maximum value for each subscale is 10 points. A higher score indicates more severe autistic traits. The AQ has good reliability and structural validity in Chinese adults (44).

This scale is used by trained professionals to assess the interpersonal relationship, imitation, emotional response, language communication, and other aspects of autistic children over a period of 18 months. The scale is composed of 15 items that use a 4-point Likert scale (age-related performance, mild abnormality, moderate abnormality, and severe abnormality). The higher the total score on the scale, the more serious the symptoms; a total score of between 30 and 36 indicates mild–moderate autism, and a score above 36 indicates severe autism (45). The Cronbach’s alpha coefficient is 0.73 in the Chinese version (46). In the current study, autistic children with a score above 36 were included.

The GDS is currently a widely used neurodevelopmental scale that consists of five domains, namely adaptability, gross motor, fine motor, language, and personal-social (47). The developmental quotient (DQ) of the five domains is used to evaluate the level of neurodevelopment. The higher the DQ, the better the neurodevelopment. The GDS has good reliability and validity in Chinese children (48). In the current study, a GDS score greater than 85 in all children is considered normal.

SPSS v23.0 statistical software was used for the statistical analysis. Descriptive statistics for continuous variables are presented as the mean (M) and standard deviation (SD), and count data are described by prevalence (%). Differences in demographic information and behavioral outcomes between the ASD and TD groups were compared by the t-test or chi-square test, with a significance level of p < 0.05. Hierarchical moderated regression analyses were used to determine whether or not maternal autistic traits moderate the association between maternal parenting styles and behavioral problems in the children. In the first step of the regression models, the independent variables were maternal parenting style (supportive/engaged vs. hostile/coercive) and maternal AQ domains separately across groups, controlling for child age, gender, DQ, family income, family income meeting demand, and maternal education level. The second step of the regression models included interaction terms between different parenting styles and each AQ domain. All five interaction terms were included in each regression model. For all moderated regression analyses, standardized variables (Z-scores) were used to compute the product terms. Simple slope analysis was conducted using the Process 2.16 macro plug-in of SPSS 23.0. Standardized regression coefficients were presented on all betas, and the significance level was p < 0.05.

Compared to the TD group, the ASD group was significantly older (t = −3.08, p < 0.01) and had significantly lower DQ (t = 23.13, p < 0.01) and maternal education levels (χ² = 30.88, p < 0.01). No significant differences were found in the maternal age, child gender, or family income (ps > 0.05) of the two groups. The results are shown in Table 1.

The children with ASD exhibited significantly more externalizing problems (t = 4.09, p < 0.01) and internalizing problems (t = 4.85, p < 0.01) than the TD children. Moreover, compared to the mothers of the TD children, those of children with ASD scored lower in supportive/engaged parenting style (t = 3.20, p < 0.01). However, no significant differences were found in the maternal AQ domains of the ASD and TD groups (ps > 0.05). The results are shown in Table 1.

Multiple regression analysis showed that hostile/coercive parenting style significantly increased externalizing problems (β = 0.30, p = 0.02) only in the children in the ASD group after controlling for child age, gender, DQ, family income, family income meeting demand, maternal age, and maternal education level. In addition, the attention switching domain of maternal AQ was negatively correlated with externalizing problems (β = −0.35, p = 0.02) in ASD children. In the TD group, the maternal AQ attention switching domain was negatively associated with internalizing problems in the children (β = −0.30, p = 0.03). The results are shown in Tables 2, 3.

In the moderation model, the main effect of maternal hostile/coercive parenting style on externalizing problems in the children remained significant (β = 0.45, p < 0.01). Moreover, this association was moderated by maternal attention switching (β = 0.33, p = 0.04) in the ASD group. In the model, we controlled for child age, gender, DQ, family income, family income meeting demand, maternal age, and maternal education level. The results are shown in Tables 2, 3.

More importantly, the significant interaction between hostile/coercive parenting style and maternal AQ attention switching domain for child externalizing problems was probed and plotted at one standard deviation above and below the mean of the moderator (AQ attention switching severity; see Figure 1). Simple slope tests demonstrated that the association between maternal hostile/coercive parenting style and externalizing problems in the children with ASD was significant at high levels of maternal attention switching problems, such that greater use of hostile/coercive parenting style was associated with more externalizing problems in the children (b = 0.22, t = 3.43, p < 0.01). Similar results were found at medium levels of maternal attention switching problems (b = 0.14, t = 3.64, p < 0.01). By contrast, at low levels of attention switching problems, the correlation between maternal hostile/coercive parenting style and externalizing problems was non-significant (b = 0.06, t = 1.46, p = 0.15; Figure 1).

The current study showed that children with ASD exhibited more serious externalizing and internalizing problems than TD children. Consistent with our results, the literature showed that children with ASD had more severe internalizing problems such as anxiety/depressive symptoms than their TD peers (49–51). Moreover, ASD children displayed more externalizing problems such as peer conflict (49) and attention-deficit and hyperactivity symptoms (52, 53). We also found that the ASD group scored lower in maternal supportive/engaged parenting style than the TD group. Consistent with our findings, the literature indicated that children with ASD received less maternal affection and warmth than TD children (11, 54). It is possible that children with ASD lack reciprocal relationships, have communication impairments, and decreased response to social stimulation, which may result in a decrease in parents’ affectionate interactions with these children (55, 56). Of course, some studies showed that no significant differences were found in the maternal parenting styles in the ASD and TD groups (57, 58). This inconsistency may be the sample size, sample type, or limitation of the self-report instrument. Further studies are needed to determine the reasons for differing results.

Notably, the current study showed that maternal hostile/coercive parenting style was associated with externalizing behaviors in children with ASD. Similar to our results, previous studies suggested that children with ASD aged 6–10 years exhibited more externalizing behaviors under conditions of higher levels of parental critical comments and harsh-discipline parenting style (59), and higher levels of maternal discipline and harsh punishment parenting style were related to greater externalizing problems in ASD (12, 60). According to the coercion theory (15), we speculate that raising a child with ASD is quite stressful for parents because of the child’s social communication difficulties, narrow interests or stereotyped behaviors, and internalizing and externalizing behaviors, which can lead to hostile/coercive parenting styles. In turn, such a negative parenting style can increase the frequency and intensity of sadness, anger, and bad behaviors further in children with ASD (15). Thus, the relationship between hostile/coercive parenting and externalizing behavior highlights the importance of interventions that may break this negative cycle. By contrast, maternal hostile/coercive parenting style was not related to externalizing behaviors in TD children. The reason is not clear, and we suspect that it may be due to autistic traits in children with ASD such as persistent difficulties in social communication and interaction (1) and deficits in emotional functioning (49).

Attention switching is defined as diverting attention between tasks, which is considered a core cognitive ability that underlies the executive control of thought and action (61). Attention switching problems are expressed that people frequently get so strongly absorbed in one thing that loses sight of other things (19). Our multilevel analyses indicated that maternal attention switching was negatively associated with externalizing problems in the ASD group. We speculated that maternal autistic traits may affect children’s behavior either as a genetic predisposition (62) or together with other factors such as parenting style.

In the current study, we found that more likely hostile/coercive parenting styles were associated with more children’s externalizing problems at medium and high levels of maternal attention switching problems. Neuropsychological testing shows that parents of children with ASD have impaired executive function skills in the area of attentional flexibility (63). Moreover, previous evidence suggested that parents of children with ASD were found to have a “generativity deficit” in patterns meaning tasks (64). As mentioned previously, mothers may have a more negative parenting style (32) and potentially further impact parent–child interactions (65) in the context of high attention switching difficulty. Similarly, mothers with attention conversion difficulties may affect the development of children’s emotional and behavioral regulation abilities (25). One of the possible explanations for the moderate effect of attention switching difficulty on the association between hostile/coercive parenting style and children’s externalizing problems is that children’s needs and behaviors will change with different situations (66); however, the mother with poor attention switching will lack flexibility in parenting style, resulting in more hostility or coercion (32, 67). Based on the coercion theory (15), a more hostile/coercive parenting style may affect the interaction with the child and impact the child’s social–emotional engagement, communication, and social interaction abilities (68, 69), which then leads to poorer child developmental outcomes. Thus, maternal parenting style and autistic traits should be assessed early and should be included in early family-level interventions to help mothers to develop strategies for interaction with ASD children.

Several limitations of the study should be acknowledged. First, our study design was cross-sectional and thus could not prove the causal relationship between maternal parenting style and autistic traits with behavioral problems in the children. Second, this study only included female caregivers. Further research is needed to clarify the influence of both maternal and paternal characteristics on children’s developing behavioral problems. Third, all indicators in the study were assessed by a questionnaire. In future, cognitive assessments can be used to improve the objectivity of the results. Fourth, only children with severe autism were selected for the current study, and future studies will be considered the effect of the severity of autism on parenting style. Finally, the sample size is small; thus, future studies should consider these questions in a larger sample.