Comparison of Parental Stress, Parental Efficacy, and Child Problem Behavior between ASD and Non-ASD Youth: A Multi-Group Confirmatory Factor Analysis

Background: Several studies have examined the role of parental stress and child outcomes in autism spectrum disorder (ASD). However there is little research examining positive parental characteristics, such as self-efficacy, that may reduce the impact of stress in families with a child with ASD. Whether the relationships between parental stress, parental self-efficacy, and child problem behavior outcomes have the same measurement characteristics across groups (ASD/non-ASD), known as measurement invariance, is also unclear.

Objectives: To examine the relationships between parental stress, parental self-efficacy, and child problem behavior using confirmatory factor analysis (CFA). To examine differences in the measurement of these latent variables between ASD and non-ASD samples.

Methods: Participants were 1035 hospitalized children and adolescents, 937 with an Autism Diagnostic Observation Schedule-2 (ADOS-2) confirmed ASD diagnosis, and 98 participants with non-ASD developmental delay, admitted to one of six specialized inpatient psychiatry units and prospectively enrolled in the Autism Inpatient Collection (AIC) study. Parents completed the Aberrant Behavior Checklist Irritability (ABC-I) subscale, Parent Stress Index Short Form (PSI-SF-4) and Difficult Behavior Self-efficacy Scale (DBSS) at admission. The measurement model contained three latent variables – parental stress, parental self-efficacy, and child problem behavior – which were measured using observed variables of each as indicators (see Figure 1). A multi-group CFA was conducted using AMOS version 25.

Results: Individual model results indicated the three latent variables were significantly correlated with each other in a manner that supported the validity of each for the ASD sample only: parental efficacy was negatively correlated with both parental stress and child problem behavior and parental stress was positively correlated with child problem behavior (all p-values ≤ .001). In the non-ASD sample, only parental stress and child problem behavior were positively correlated with each other (p ≤ .001). Group differences were also found among the observed variables used as indicators to measure the child problem behavior latent variable. Specifically, self-injurious behavior items did not contribute to the measurement of child problem behaviors in the non-ASD sample, β= 0.104, p=0.25. Multi-group CFA results indicated differences between group correlations (χ² = 57.5, df = 13, p = 0.001) and means of the observed variables (χ² = 44.5, df = 7, p = 0.001) between ASD/non-ASD groups, however we did not find differences between group factor loadings, that is, how well the observed variables measured the underlying latent variable.

Conclusions: Relationships between parental stress, parental self-efficacy, and child problem behaviors differed between parents of children with ASD compared to those without. Measurement invariance was not found across groups. By applying CFA we’re able to obtain more reliable measures of the underlying constructs of parental stress, parental self-efficacy, and child problem behaviors and the correlations between them. Of clinical importance, while parental stress is correlated with child problem behaviors in both samples, self-efficacy in parents of children with ASD may be a positive characteristic for reducing both parental stress and child problem behavior. Interventions targeting parental self-efficacy could be a novel treatment target with downstream effects on both members of the parent-child dyad.