A Sensitivity Analysis of Three Nonparametric Treatment Effect Scores for Single-Case Research for Participants with Autism

Abstract

The widely used percentage of nonoverlapping data (PND) treatment effect calculation was compared to more recently developed methods which, it has been argued, better account for outlying variables and trend in single-case design (SCD) intervention studies. Percentage of all nonoverlapping data (PAND) and nonoverlap of all pairs (NAP) were selected for comparison as both are amenable to hand calculation, making them widely accessible to clinicians and teachers as well as researchers. A data set was developed through a systematic search of peer-reviewed literature on self-management interventions conducted with participants with autism spectrum disorder (ASD). Treatment effect sizes derived from each method were compared for studies that provided sufficient data. Results indicated that PND provided a conservative measure of strength of treatment effect when compared to PAND and NAP scores. Interpretation scales for treatment effect scores derived from each method were reviewed. Implications for selecting a calculation method for participants with ASD are discussed.