Although there are established measures to diagnose Autism Spectrum Disorder (ASD), there are no currently comparable measurement tools available to assess outcomes for core and associated features for ASD interventions. One scale, the Aberrant Behavior Checklist-Community (ABC-C; Aman & Singh, 2017), originally developed to assess intervention research outcomes for problematic behavior and associated features in individuals with intellectual disability (ID), appears to be a promising option for this purpose. The 58-item ABC-C rating scale has become a popular choice amongst ASD intervention researchers (Bolte & Diehl, 2013). Many of the core and associated features of ASD, the prime targets of intervention, are represented within the scale. However, ABC-C validity research in the ASD population specifically is still limited. Previously, three exploratory factor analyses (EFA; Brinkley et al., 2007; Kaat, Lecavalier, & Aman, 2014; Mirwis, 2011) and two confirmatory factor analyses (CFA; Brinkley et al., 2007; Kaat et al., 2014) have been performed on the ABC-C in ASD samples. These analyses have yielded inconsistent factor solutions across studies, with marginally fitting models upon testing. This has left questions about the rigor or thoroughness of the analytic strategies, including the range of factor solutions examined, the logic behind the selection of the factor solutions retained, and possible differences due to rater type. Thus, additional thorough and independent factor analyses were warranted for the purpose of determining whether the ABC-C authors' posited five-subscale interpretive structure is the most appropriate, useful, and valid for the ASD population or if an alternative model is more suitable. Present study one involved using EFA to examine the data structure of the ABC-C in an ASD sample (N = 300), age range 3.17 to 21.05 years, based on ratings provided by special education staff. A nine-factor solution was retained following examination of factor models consisting of between three and 11 factors. Study two involved using CFA to test the absolute and relative fit of the derived ABC-C factor solution from the EFA of study one with an ASD validation sample (N = 243), age range 2.95 to 21.15 years, across five fit indices (Chi Square [2], Standard Root Mean Square Residual [SRMR], Root Mean Square Error of Estimation [RMSEA], Comparative Fit Index [CFI], and the Tucker-Lewis Index [TLI]). The fit of the factor model from study one was then directly compared to the fit of the existing models of the ABC-C found in ASD samples (or proposed for use with individuals with ASD) using Akaike's Information Criterion (AIC) and the Bayes Information Criterion (BIC). Results from the CFA revealed the nine-factor model from study one meeting or approximating cut off-values on the SRMR, RMSEA, CFI, and TLI. Results from the AIC and BIC fit tests showed the nine-factor model to be the best fitting model compared to the other existing models of the ABC-C found in ASD samples. Findings from study one and two highlight the possibility that the current five-factor author version of the ABC-C is potentially not the most viable model for the ASD population and the nine-factor version may be a more appropriate choice. Findings also underscored the need for similarly rigorous factor analytic methodology to be employed in future replication studies, and the recommendation for a major scale revision of the ABC-C.
Read
