ABSTRACT
This chapter reviews the methods of descriptive and statistical analysis frequently used in age-period-cohort (APC) studies. It discusses their utilities and limitations, and lays out the formal algebra of the APC identification problem and reviews conventional approaches to model identification. The many pitfalls in empirical APC analysis and lack of major breakthroughs in methodology led to the verdict that statistical APC models cannot be relied on to provide accurate estimates of A, P, and C effects. The chapter reevaluates the state of the field and points out the gap in our knowledge. This is followed by a sketch of a generalized linear mixed model framework that unifies the APC models and methods for the analysis of data from three research designs. The rationale for the GLMM approach to APC analysis is provided along with detailed methodological guidelines on how to conduct APC analysis.
