Using Bayesian methods to test mediators of intervention outcomes in single-case experimental designs

Abstract

Single-Case Experimental Designs (SCEDs) have lately been recognized as a valuable alternative to large-group studies. SCEDs form a great tool for the evaluation of treatment effectiveness in heterogeneous and low-incidence conditions, which are common in the field of communication disorders. Mediation analysis is indispensable in treatment research because it informs researchers about the mechanism through which the intervention leads to changes (e.g., communication skills) in the outcome of interest (e.g., developmental outcomes). Despite the increasing popularity of both SCEDs and mediation analysis, there are currently no methods for estimating mediated effects for a single individual. This paper describes how Bayesian piecewise regression analysis can be used for mediation analysis in SCEDs. A Playskin LiftTM dataset from one infant born preterm who is at risk for cognitive developmental delays is used to illustrate two approaches to mediation analysis in SCEDs: Bayesian computation of the mediated effect and Bayesian informative hypothesis testing. Annotated R code is provided so researchers can easily fit the proposed models to their own SCED data set. Advantages and limitations of the method are discussed.