Beyond the Identification Problem: Modeling Age, Period, and Cohort Patterns of Mental Wellbeing in South Korea Using Recent Advances in APC Analysis

Jos van Leeuwen , European University Institute

Background - The life course approach involves three fundamental temporal variables: age, period, and cohort. Conceptually, it is clear that each of these three variables has quite different effects on individual outcomes. However, the proper distinction between their respective contributions to life course patterns and the identification of their separate causal effects has remained a methodological minefield for the last 50 years. Recent advances in the field of age-period-cohort (APC) analysis may provide a way forward. Methods - In this study, I apply the most important classical and contemporary methods of APC analysis to data about mental wellbeing in one dataset, the Korean Welfare Panel Study (KOWEPS), and investigate to what extent they are able to arrive at the same conclusions with regard to age, period, and cohort patterns in this outcome. Results - Findings show that most APC methods actually arrive at similar conclusions. Six of the models find a positive age, zero period, and positive cohort trend, while four find a negative age, positive period, and zero cohort trend. Two further models find a negative age, positive period, and negative cohort trend. If these statistical outcomes are further bounded by theoretical considerations, such as the exclusion of a negative cohort trend during the observation period, the range of possible results can be narrowed down even more. Conclusion - APC methods based on different conceptual and mathematical foundations can actually reach similar substantial findings. Thus, the problems of APC analysis may in reality be less pernicious than is commonly thought.

See extended abstract

 Presented in Session P2. Health, Mortality, Ageing - Aperitivo