Factor Structure and Measurement Invariance of the Spence Children's Anxiety Scale - Parent Version Across Anxious and ASD Groups

Background: Anxiety, is a common health concern in children with autism spectrum disorder (ASD) affecting between 11% and 84% (White et al. 2009) compared to 3-24% of typically developing children (Green and Ben-Sasson 2010). The Spence Children’s Anxiety Scale-Parent Version (SCAS-P; Spence 1998) is a commonly used parent-report instrument for assessing anxiety (Nauta et al. 2004) and has been used with children and young people on the autism spectrum (Chalfant et al. 2006; Russell and Sofronoff 2005, Rodgers et al. 2012). However, the SCAS-P was developed with typically developing children and there is very little information available regarding the validity of the SCAS-P for use with ASD samples. The measure is frequently used to compare anxiety in children with ASD to other groups and further investigation is required to be confident that the scale functions in the same way across groups.

 Objectives: The aims of this study were:

• to determine then compare the factor structure for the SCAS-P in a sample of  young people with ASD and a sample of anxious young people,
• to use measurement invariance techniques to determine whether SCAS-P items function in the same way and are on the same metric in children with ASD and children with anxiety in order to suggest whether cross-groups comparisons using the SCAS-P are appropriate and meaningful.

 Methods: Parents of 232 children with ASD aged between 8 and 16 years, and 163 children without ASD with a current anxiety disorder aged between 7 and 15 years, completed the SCAS-P. Confirmatory Factor Analysis (CFA) was undertaken in order to determine the best-fitting factor structure. Measurement invariance was performed in order to establish whether structure, factor loadings, item intercepts and error variance were same across groups.

 Results: A five-factor model with the Generalized Anxiety Disorder subscale excluded was found to be the best-fitting factor structure in both groups. Configural and metric measurement invariance showed that the general model structure was invariant across groups, but non-invariance was found to be present for factor loadings between groups, terminating further analysis. Additional Exploratory Factor Analysis (EFA) indicated that the items loading on the physical injury fears factor differed between groups, with stronger loadings in the anxious group than in the ASD group.

 Conclusions: Factor analyses determined that the five-factor model (excluding GAD subscale) fit the data best. Anxiety as a concept, measured with the SCAS-P, is not identical across the two groups. Cross-group comparisons between children with ASD and children diagnosed with anxiety disorder based on the SCAS-P scores should therefore be undertaken with caution. The SCAS-P norms established for anxiety disordered children may not be appropriate for individuals with ASD. Qualitative work is needed to explore the validity of some SCAS-P items in the ASD population.