Bayesian Methods in Single-Case Experimental Designs

Session Type: Symposium

Abstract

Single-case experimental design (SCED) data present unique quantitative data analytic challenges such as small samples, autocorrelations, and often count or ratio data. This renders parametric tests, inferential statistics, and maximum likelihood solutions inappropriate for SCEDs. Bayesian estimation is an elegant solution. There are many criteria to show strong evidence of a causal relation in SCEDs including consistency of data within phases, difference in patterns across phases, and immediacy. The Bayesian unknown change-point model (BUCP) is the only inferential model that estimates all model parameters that are required to establish causality in SCEDs. This symposium will illustrate how BUCP complements visual analysis, compare BUCP with simulation modeling analysis, and extend BUCP to multiphasic designs using Variational Bayesian.

Sub Unit

• SIG-Educational Statisticians

Chair

• Anthony J. Onwuegbuzie, Sam Houston State University

Discussant

• John H. Hitchcock, Indiana University - Bloomington

Papers

• Illustrating How the Bayesian Unknown Change-Point Model Supplements Visual Analysis for Single Case Designs - Ratna Nandakumar, University of Delaware; Prathiba Natesan, University of North Texas; Pragya Shrestha, University of Delaware

• Investigating Immediacy in Multiphasic Single-Case Experimental Designs Using a Bayesian Unknown Change-Points Model - Prathiba Natesan, University of North Texas; Tom Minka, Microsoft Research; Larry V. Hedges, Northwestern University

• Comparing Bayesian Unknown Change-Point Model and Simulation Modeling Analysis of Single-Case Experimental Designs - Ratna Nandakumar, University of Delaware; Prathiba Natesan, University of North Texas; Jayme M Palka, University of North Texas; Pragya Shrestha, University of Delaware