Analysis of Standardized Single-Case Experimental Data in Presence of Autocorrelation: A Monte Carlo Simulation Study (Poster 28)

• In Event: AERA Poster Session Thursday 8:00 a.m.
  In Poster Session: Advancements in Educational Data Analysis: Machine Learning, Multilevel Models, and Beyond

Abstract

Two-stage individual participant data meta-analysis is recommended for single-case experimental design studies. However, challenges (e.g., limited measurement occasions, autocorrelation, and standardization) arise in estimating intervention effects and autocorrelation at the case level (Stage 1) before synthesizing the intervention effects across cases and studies (Stage 2). This large-scale simulation study examined statistical properties of estimating the standardized intervention effect size and autocorrelation at Stage 1. Results showed generally unbiased estimates with small mean squared errors when there were at least 20 measurement occasions and lag-1 autocorrelation ranged between -0.4 and 0.4. However, the standard errors of the intervention effect consistently underestimated with nonzero intervention effect. The significance of this study for preparing to synthesize effect sizes at Stage 2 was discussed.

Authors

• Yukang Xue, University at Albany - SUNY

• Mariola Moeyaert, University at Albany - SUNY