ABSTRACT
The previous chapter indicated some problems with examining only the age distribution, or differences between periods, or differences between cohorts. Many researchers examine two of these variables at a time; for example, an age-by-period table so that they can study how the age distribution of a disease or crime or attitude varies over time. But the shifts in the age distributions across periods can be affected by cohorts. This suggests examining all three of these factors in an age-period-cohort (APC) model. This explicitly raises a new problem: the identification problem that motivated the opening quotation in the article by Mason, Mason, Winsborough, and Poole (1973).
