Key note 1: Todd Little
Title: On the Merits of Longitudinal Multiple-Group Fixed Effects Modeling Versus Multilevel Modeling for Evaluating Interventions
For the analysis of analysis of multi-cohort longitudinal intervention trials, features of traditional multilevel modeling approaches present numerous obstacles to model estimation and obtaining valid inferences. An alternative approach for testing interventions effects is using longitudinal multiple-group (LMG) modeling. The advantages of LMG include a) robust missing data treatment, b) estimating mean-level information in a non-arbitrary metric, c) incorporating standardization constructs for accurate between group comparisons, d) robust estimation and testing of complex interactions, and e) and testing many critical assumptions that otherwise can be tested. The LGM approach is a guided confirmatory model testing framework that places a premium on avoiding Type II errors particularly when complex interactions are potentially a play. I will provide an example from a large-scale evaluation recently performed by IMMAP.