Distribution of Self-Reported Alcohol Consumption in New Zealand, 1978–1979

Abstract
The problem of estimating how much alcohol individuals actually drink is discussed with particular reference to the log-normal frequency distribution of alcohol consumption. It is then presented as being a problem of inference from the recent history of a time series. Following directly from this presentation, data obtained in a national survey of 10,000 New Zealand residents are analyzed (a) to yield the frequency distribution of estimated drinking rates; (b) to test whether this distribution is log-normal in form; and (c) to provide further information about the distribution of alcohol consumption. This analysis enabled us to account for almost the whole of the actual total alcohol consumption in New Zealand. The results are different from those expected on the basis of the uncorrected log-normal description of consumption.