Leveraging Time-Varying Covariates to Test Within- and Between-Person Effects and Interactions in the Multilevel Linear Model

Abstract

Multilevel linear modeling (MLM) is a powerful and well-defined tool often used to evaluate time-varying associations between two or more variables measured in longitudinal studies. Such variables carry information about stable, between-person differences as well as information about within-person variability. For emerging adults, this variability figures prominently across a variety of developmental domains. A single variable measured on repeated occasions can be easily summarized into two new variables that represent the unique within- and between-person sources of information contained in the original variable. Well-known procedures for statistically disaggregating time-varying predictors in an MLM are straightforward but often not accessible to a nontechnical readership. Using SAS syntax, this tutorial provides step-by-step instructions to recode a single repeated-measures variable into separate between- and within-person predictor variables. Strategies are suggested for testing and interpreting main effects and interactions in the MLM, drawing on a daily diary example of first-year, first-time college-attending emerging adults.