ABSTRACT
The last chapter showed how to assess the variance uniquely associated with ages, periods, and cohorts, even though it is not possible to estimate the age, period, and cohort coefficients uniquely. It also strongly recommended not taking the fixed effects in such models as unbiased estimates of the parameters that generated the outcome values. These conclusions hold for APC ANOVA (age-period-cohort analysis of variance) and APCMM (age-period-cohort mixed model) approaches and extend to other hierarchical/ mixed-model approaches.
