In a postal questionnaire administered at age 18 years 8 months the young people completed the self-report version of the ten-item alcohol use disorders identification test (AUDIT) [20]. Almost 40 % of individuals in this sample are consuming alcohol at a ‘hazardous’ level (total AUDIT score 8+)—which may be considered as “normative.” We derived two alternative outcome measures. Firstly we used the standard cut point for ‘harmful’ alcohol use (total score 16+) to derive a binary outcome. Respondents answering ‘No’ to the stem question on recent alcohol use were assigned a score of zero and retained in further analyses as members of the non-harmful group. We refer to this measure as “harmful alcohol use”. Secondly, the ten ordinal responses were used to estimate a unidimensional latent variable using confirmatory factor analysis. Whilst these ten items refer to both the use of alcohol and the consequences thereof, for the sake of simplicity we refer to this latent measure as “alcohol use”. We believe that, for this age group in particular, our measure is driven primarily by the initial alcohol-use AUDIT items
