On the External Validity of Single-Case Designs: A Bayesian Approach

When selecting interventions for students or clients, practitioners typically need to estimate the probability that it will lead to socially significant changes in behavior. One potential solution to address this issue involves computing the conditional probability of success given the results observed in the research literature. The purpose of the study was to use Bayesian inference, an approach that considers prior probability, to examine how the number of successful experiments may be used to assess the external validity (i.e., generality across participants) of findings from single-case research. Although the analyses show that relying on more experiments tends to produce better estimates, having as few as five successful experiments may result in a success rate with a lower bound of at least .50. Given that research on the topic remains limited, we argue that our proposal should not be viewed as immutable, but rather as a starting point for spurring further discussion and research on the topic.