ABSTRACT

Multilevel modeling (MLM), also known as hierarchical linear modeling (HLM), or sometimes as mixed effect modeling, is a method of analysis designed for modeling hierarchical data due to a nested structure or repeated measures. Repeated measures analysis of variance (RMANOVA) is one well-known method for analyzing such repeated measures data. This chapter demonstrates the relative merits of RMANOVA and multilevel modeling using Stata, a popular and highly accessible general purpose statistical packages. RMANOVA investigates the between-subject variability (gender effect), the within-subject variability (time effect) and the interaction between gender and time using the following Stata command: anova locus female / student|female time female/time, repeated. 95 percent Confidence Intervals for the variances associated with the random-effects parameters, which provide estimates of the precision of the point estimates. The chapter demonstrates several ways in which the MLM framework can be extended to include a number of different predictors in the equation in order to model different growth trajectories conditional on these predictors.