Abstract

BackgroundRandomized Controlled Trials are the gold standard for assessing whether an intervention is effective; however, they require large sample sizes in order to detect small effects. For rare or complex populations, we advocate a case series approach as a more realistic and useful first step for intervention evaluation. We consider the importance of randomization to such designs, and advocate for the use of Randomization Tests and Between Case Effect Sizes to provide a robust and statistically powerful evaluation of outcomes. In this tutorial, we describe the method, procedures, and analysis code necessary to conduct robust single case series, using an empirical example with minimally verbal autistic children.MethodWe applied a pre-registered randomized baseline design with between-case effect size to a case series, to test the efficacy of a novel, parent-mediated, app-based speech production intervention for minimally verbal autistic children. Parent-rated probe scores were used to densely sample performance accuracy over time.ResultsParents were able to reliably code their children’s speech productions using BabbleBooster. A non-significant Randomization Test and small Between-Case Effect Size, suggested there was no evidence that BabbleBooster improved speech production in minimally verbal autistic children, relative to baseline scores, during this brief period of intervention.ConclusionThe current analyses exemplify a more robust approach to examining treatment effects in rare or complex populations, where RCT may be difficult or premature to implement. To facilitate adoption of this method by researchers and practitioners, we provide analysis code that can be adapted using open source R packages. Future studies could use this case series design to evaluate interventions aiming to improve speech and language outcomes for minimally verbal autistic children, and other heterogeneous and hard to reach populations.