Part 2 reviews the impact of clustering on the analysis of a group-randomized trial (GRT), with extensions to individually randomized group-treatment (IRGT) trials and stepped wedge group-randomized trials (SWGRTs). It also reviews the additional threats to the validity of SWGRTs, including time-varying intervention effects, effect heterogeneity, and complex correlations. It provides guidance on preferred analytic approaches to GRTs, IRGT trials, and SWGRTs. 

Part 2 also includes a discussion of cross-classification and dynamic group models. It closes with a review of alternative approaches to the analysis of data from these designs, including randomization tests, generalized estimating equations, and two-stage fixed-effect approaches.