Abstract
Analyses of absence from work have traditionally treated all types of absence days as identical. It is argued here that such an approach has led researchers to impose unnecessary restrictions on their models of absence behaviour. This hypothesis is tested here using data collected from over 1400 employees from 61 different Australian workplaces. While the absence days measure distinguishes between 25 different reasons for taking time off work, specification testing indicates that it is appropriate to cluster these reasons into ten different types of absence. Dependent variables representing each of these absence types were then created and regressed against a large range of explanatory variables representing demographic, job and workplace characteristics, and worker attitudes, and generally thought to influence attendance levels. The results indicate that the effects of these variables vary, in terms of both size and significance, with the type of absence being considered.