Fillmore K M, Hartka E, Johnstone B M, Leino E V, Motoyoshi M, Temple M T
Department of Social and Behavioral Sciences, University of California, San Francisco 94133-3203.
Br J Addict. 1991 Oct;86(10):1221-67. doi: 10.1111/j.1360-0443.1991.tb01702.x.
Meta-analysis combines results from multiple longitudinal studies to describe life course variation by age and sex for quantity of drinking per typical occasion (20 studies) and frequency of drinking during one month (27 studies). There is cross-study heterogeneity for the Time 1 means of the drinking variables blocking for age and sex. Age distributions for the Time 1 means are similar by gender within nations; in the aggregate, males exceed females in the magnitude of drinking. Dramatic shifts in the standardized mean difference (M2-M1) occur among the young; greater homogeneity and moderate change (declines) occur later in life. Implicated in improving cross-study homogeneity for M2-M1 among the young are interval between measurements, nation, Time 1 per capita consumption (PCC), difference in PCC and the Time 1 mean. Lower unstandardized regression coefficients are found for quantity among youth, but are not consistently homogeneous within nations; the association for frequency becomes increasingly stable with increasing age. Nation and interval are implicated in improving homogeneity. Decline in quantity occurs among the old. M2-M1 produces homogeneously higher regressions for groups of the young who increase quantity v. those who do not. Linkage of the group-level and individual-level findings is discussed.
