Micro-Macro Theoretical Linkages in Social Demography: A Thematic Session

This paper addresses the critical stances of certain sociologists and statisticians in relation to quantitative, comparative, multilevel analysis. The author reaches two main conclusions: (1) Multilevel analysis brings with it a great deal of conceptual baggage in the form of assumptions that need to be checked. More public effort should be expended on assumption checking. (2) The major criticisms of statisticians concerning the use of observational data instead of experimental data apply fully to multilevel analysis -- comparative or intra-societal -- if the multilevel analysis is based on nonexperimental data. But it does not have to be. For example, it is possible to carry out a series of experiments, and model variability in the results across experiments.