Fig. 2From: Bias in estimates of alcohol use among older people: selection effects due to design, health, and cohort replacementThe prevalence of weekly alcohol use in 1992, 2002 and 2011, with and without the estimated effect of non-response and cohort replacementBack to article page