The Extended Generalized Linear Model for Counts or Free Proportion Outcomes in Single-Case Designs

• In Event: Poster Session 2
  In Poster Session: Division D Section 2 Poster Session 1

Abstract

In a comprehensive review of the single case design (SCD) literature, Shadish and Sullivan (2011) indicate that over 90\% of the outcome measures used were some form of count. Neither linear regression nor rank-based nonparametric statistical analyses are well-adapted to the analysis of counts. We propose the use of the extended Generalized Linear Model for the analysis of data from SCDs. GLMs can now be fit using most mainstream statistical packages in some form, and mixed and nonlinear models also exist. The approach is illustrated using data from Rantz et al. (2009). In addition to the usual regression output, we emphasize a graphical approach that compares model predictions to empirical data.

Authors

• Jay Verkuilen, City University of New York

• Mariola Moeyaert, State University of New York