Accurate Model Versus Accurate Estimates: A Study of Bayesian Single-Case Experimental Designs

• In Event: 47.072 - Model Contemplations: Identification, Detection, Effects, and Errors

Abstract

Although SCEDs have gained importance statistical practices to evaluate intervention effects are yet to incorporate all their analytic complexities. We investigate whether in data patterns that show trend, there is an indeterminacy between pattern due to slope and pattern due to autocorrelation. We compared the performance of four Bayesian change-point models: (a) intercepts only (IO, no slopes or autocorrelations), (b) slopes but no autocorrelations (SI), (c) autocorrelations but no slopes (NS), and (d) both autocorrelations and slopes (SA). Coverage rates showed that for the NS model either the slope effect size or the autocorrelation CI almost always erroneously contained 0. Therefore, it is recommended that researchers estimate either SI or NA models instead of SA models.

Authors

• Prathiba Natesan, University of North Texas

• Larry V. Hedges, Northwestern University