Article Text
Abstract
Froggatt, P. (1970).Brit. J. industr. Med.,27, 199-210. Short-term absence from industry. I. Literature, definitions, data, and the effect of age and length of service. This, with two subsequent papers, comprises the first extensive study directed only to short-term absence from industry, an entity common in all branches of organized work and now one of the greatest personnel problems of an industrial society.
This first paper reviews the literature and background of industrial absence, describes the sources of the data and the groups for study, defines terms used throughout, discusses the rationale of the selection criteria, and examines the effect on the numbers of one-day and two-day absences of age and length of service in the organization.
The observations are from two light engineering works and two government departments and cover in all some 2 300 male and female personnel, both salaried and hourly-paid, over periods of up to seven years. Twenty study groups were identified for the analyses, each comprising members of similar `works centre', sex, supervisory grade, and marital status, who neither changed relevant status during the study period nor were absent for more than 65 days in any year. This stringency in delimitation enhanced the validity of the conclusions drawn by (a) ensuring necessary homogeneity for crucial variables, and (b) permitting examination of the consistency of the results over groups and organizations.
Multiple regression analysis for the effect of age and length of service on short-term absence showed that, generally, length of service had no effect but that age was (weakly) negatively linearly associated with the number of one-day absences but independent of the number of two-day absences. Transforming the skewed dependent variates to normal functions for completely valid analysis had no important effect on these results, which were also confirmed by data from a longitudinal study in one company. This association between age and one-day absences was too weak (more than 90% of the variation in the latter was unattributable to linear regression on the former) to be of executive importance but it is relevant to the validity of inferences from curve-fitting analysis presented in a later paper.