Bayesian Inference for Single Case Designs

• Activity Number: 332
• Type: Topic Contributed
• Date/Time: Wednesday, August 11, 2004 : 10:30 AM to 12:20 PM
• Sponsor: Section on Bayesian Statistical Science
• Author(s): Richard Evans*+
• Companies: Iowa State University
• Keywords: Bayesian ; changepoint

Abstract

Single case statistical inference has been investigated and implemented in the behavioral sciences for many years. However, there has been little use of this method in other sciences and statistical development has been limited. Single case designs can be useful for research on rare or expensive animals or diseases, where sufficient sample sizes required for traditional statistical methods may be difficult to obtain. Typical single case designs apply and withdraw interventions (e.g., ABAB or ABC) over time. Our approach is to use Bayesian changepoint methods to calculate the posterior probability that the intervention affected the subject and changed the outcome during the course of treatment relative to other treatments or baselines. Note that observations within a treatment time may not be exchangeable and the analysis may require different numbers of cutpoints (e.g., A|BC and A|B|C). We apply this method to data obtained from an emu that was trained to walk on a pressure platform.