Proposals for Incorporating the Reversal in Single-Case ABAB Designs Data Analysis

Abstract

One of the commonly used single-case experimental designs (SCEDs) in education research is the ABAB design, commonly referred to as reversal design. This design is applicable when the studied behavior is reversible (i.e., reverses to baseline levels when the intervention is withdrawn) and it allows for a within-unit replication of the “basic effect”, that is, a comparison between two adjacent conditions (with and without an active intervention). Methodologically, one of its strengths is to provide three phase change moments to demonstrate experimental control. However, the withdrawal of the intervention (i.e., the comparison between the first intervention phase and the second baseline phase) has not received sufficient attention in the SCED literature. The aim of the current text is to propose quantifications that enable using the information from the withdrawal. Specifically, the quantifications allow for assessing the degree to which withdrawing the intervention leads to initial baseline levels as well as incorporating the reversal in an overall effect size. The proposals are illustrated with several real educational science data sets, including both within-unit analysis and replication of the ABAB design across units, and representing different degrees of reversal of the target behavior when the intervention is withdrawn.