Dependent variables/Outcome measures

a) Have you covered for workers because of their drinking; b) Has your productivity been reduced because of their drinking; c) Has your ability to do your job been negatively affected because of their drinking; d) Were you involved in an accident or close call at work because of others’ drinking; and e) Have you had to work extra hours because of others’ drinking? In Australia, only the third, fourth and fifth questions were asked, and in New Zealand the fourth was omitted. A summary dichotomous (0,1) outcome measure was produced for reporting one or more of these harms from the drinking of co-workers [any work harm]. Responding workers who reported experiencing any of the items were categorised as 1. Respondents who reported no to all items were categorised as 0. Respondents in all countries (apart from Australia) who reported no to all items but who failed to respond on one item (n=60) were coded 0. Twenty-seven respondents who failed to respond on more than one of the work harm items, yet may have answered no to other items, (apart from seven Australian respondents who had one or more missing items) were coded missing for the “any work harm” variable. (In Australia, missing even one response constituted a third of their responses.) Given that we are creating a dichotomous any harm versus no harm variable this procedure was considered adequate.