Analyzing discontinuities in longitudinal count data: A multilevel generalized linear mixed model.

Abstract

Numerous tutorial publications are available to researchers seeking the procedures needed to analyze longitudinal count response variable data. However, most of the available tutorial publications have drawbacks that limit their usefulness to applied researchers, and to the best of our knowledge, very few publications make both the sample data and the data analysis syntax scripts available to readers to allow an interactive replication of analyses. The purpose of this article is to provide readers a systematic tutorial for analyzing longitudinal count data that involves a discontinuity, or an intervening event that alters the count change trajectory, using multilevel generalized linear mixed models. The longitudinal count data analysis model options and their assumptions, how the linear model equations for each can be used to correctly specify and analyze each model using Mplus or R, how to select the best-fitting longitudinal count model, and how to interpret and present results, are all described. The example data, analysis syntax scripts, and additional files are all available to readers as online supplemental materials. (PsycInfo Database Record (c) 2021 APA, all rights reserved).