Factor Analysis of the Parent Stress Index in Children with Autism Spectrum Disorder and Serious Behavioral Problems

Background: Children with autism spectrum disorder (ASD) may exhibit a range of disruptive behaviors including tantrums, noncompliance, aggression and self-injury. These behaviors often pose extraordinary strain on parents. The Parent Stress Index-Short Form (PSI-SF) is a common measure of parental stress that has been used across a wide range of children with psychiatric and developmental disorders including those with ASD. In most studies, investigators used the original three factors developed by Abidin (Abidin, 1995). However, a recent factor analysis of the PSI-SF by Zaidman-Zait and colleagues (2011) raised questions about the relevance of the original three factor model in children with ASD.

Objectives: The aim of this study was to examine the factor structure of the PSI-SF in a large sample of parents of children with ASD and disruptive behaviors.

Methods: We used data from two federally-funded, multisite, randomized clinical trials conducted by the Research Units on Pediatric Psychopharmacology (RUPP) and Research Units on Behavioral Intervention (RUBI) Autism Networks. To be included in these trials children had to have: 1. ASD with moderate or greater disruptive behavioral problems as measured by the Aberrant Behavior Checklist-Irritability subscale and the Severity scale of the Clinical Global Impression; 2. IQ of 35 or greater as assessed by the Abbreviated Stanford-Binet 5, the Leiter-R or the Mullen Scales.  Parents of 304 medication-free children with ASD completed the PSI-SF at baseline. Preliminary factor analysis with promax rotation was performed using maximum likelihood estimation on all 36 questionnaire items of the PSI-SF. Three factors were retained and items with factor loadings greater than 0.3 were kept. Cronbach’s α was calculated to verify each factor as measuring in the direction of a single construct, using a cutoff of 0.7.

Results: The sample of 304 children (263 males; 41 females) had a mean age of 5.8 ± 2.2 years (range 3.1-13.8); 81.3% were white, 10.9% were African American, 10.9% were listed as “Other” (e.g., Asian, Alaskan, Pacific Islander or multi-racial). The mean score on the ABC Irritability subscale was 26.1 ± 7.0. Preliminary factor analysis revealed that 3 items did not load on any factor. The remaining 33 items loaded on factors that were similar to the original three factor structure of the 36-item PSI-SF. The deletion of these three items improved the alpha coefficient for each of the three factors with no change in the alpha value for the total score (Figure 1).

Conclusions: To our knowledge no previous study has examined the factor structure of the PSI-SF in a large sample of children with ASD and serious behavioral problems.  In this sample, our preliminary factor analysis suggests that a three factor solution with 33 items may have advantages over the original 36 item measure in this clinical population. Additional exploratory factor analyses may provide further support for this new factor structure of the PSI-SF.  We will also examine correlations of these new factors with other measures (e.g., Vineland, ABC, IQ, Home Situations Questionnaire) in this well-characterized population.