ABSTRACT
This chapter focuses on the exposition of the basic cross-classified random effects modeling specifications of Hierarchical age-period-cohert (HAPC) models through algebraic analysis and empirical analyses of examples of the research design (repeated cross-sectional sample surveys), including the General Social Survey and National Health and Nutrition Examination Survey to show how HAPC models for such designs avoid the identification problem and offer more opportunities for testing explanatory hypotheses. The simplest form of model specification of the HAPC model is a member of the class of linear mixed models. An alternative model specification for HAPC analysis of repeated cross-sectional survey data would specify the time period and cohort effects as fixed rather than random. The development of HAPC models may provide a useful apparatus for modeling and estimating distinct age, period, and cohort effects in repeated cross-sectional survey designs. The HAPC approach can be applied to dichotomous and multiple categorical outcome variables.
