Age-period-cohort identification
Age-period-cohort (APC) analysis tries to separate three explanations for an observed difference over time:
- an age effect caused by a person’s stage in the life course;
- a period effect caused by an event or condition affecting many ages at the same historical time;
- a cohort effect caused by the durable experience of people born in a particular period.
The components are exactly dependent: calendar period equals birth cohort plus age. Observed data therefore cannot identify all three trends without additional assumptions, constraints, or research design. Different modelling choices can assign the same pattern to different causes.
This matters whenever a claim moves from “older voters now prefer X” to “this generation changed its preferences as it aged.” A cross-sectional table can establish the first statement but not the second. Repeated cross-sections with birth year improve the evidence, yet a credible analysis must state how it resolves the APC dependency and test whether conclusions survive alternative specifications.
The distinction is the central methodological limit on Boomers and democracy and other claims about political generations.