Should age-period-cohort studies return to the methodologies of the 1970s?