Three topics within the domain of regression-based analyses of multilevel data from quasi-experiments and field studies in educational research and evaluation are presented. The paper begins with a discussion of the general question of choice of unit of analysis, or what may be the more appropriate question of choice of analytical model. It then provides empirical illustrations of the importance of knowing the question of interest. Two additional topics are considered: the use of within-group slopes as indices in between-group analyses, and the estimation of within-group dependency and its role in analysis of multilevel data. These topics reflect substantive concerns in school-based non-experimental investigations. It is proposed that, in addition to the studies of the variation in analytical properties across approaches with hypothetical data, alternative approaches should be applied to a wide variety and sizeable number of actual data sets, each with a potentially differing set of inadequacies. In this way, more could be learned about both the methods and the influences of data limitations on methods. (Author/PN)