Trial-Based Functional Analysis: Implications of Criterion-Based Decision-Making

The trial-based functional analysis (TBFA) is commonly used to evaluate environmental variables maintaining challenging behavior in a naturalistic environment. Several criterion-based approaches have been developed to support data-based decision-making during the TBFA, including trial-based ongoing visual inspection (TB-OVI) and a binomial distribution (BD) approach. The current study compared four analytic strategies applied to existing clinical TBFA data as a retrospective concordance analysis: traditional aggregated data visual analysis, TB-OVI, BD, and a modified BD of valid demonstrations of effect (BD-VDE). The BD-VDE model specifies the test–control segment as the primary unit of analysis rather than the individual trial. We additionally identify and correct a computational issue in the published BD procedure by using the cumulative upper-tail probability when evaluating a criterion. Therefore, the BD-VDE is presented as a novel analytical model rather than a computational variant of the existing criteria, and it represents the paired format of the TBFA. Across eight TBFA datasets, we compared function classifications and criterion-referenced decision points. The three criterion-based methods produced identical classifications across all participants, settings, and condition combinations, although the number of segments required to reach a criterion differed across methods. Latency data additionally demonstrated variability in response patterns across participants, conditions, and settings, which was not represented by binary classifications alone. Because the present study did not include an independent criterion for behavioral function or treatment validation, the outcomes of this study do not establish accuracy or clinical validity. Thus, we present the BD-VDE as a proof-of-concept requiring empirical evaluation. The implications of these analyses may provide a better understanding of the different variables to consider when using criterion-based measures for data-based decision-making in the TBFA.