Froggatt, P. (1970).Brit. J. industr. Med.,27, 211-224. Short-term absence from industry. II. Temporal variation and inter-association with other recorded factors. This paper (a) extends the previous analysis (Froggatt, 1970b) of short-term absence from work among groups of male and female industrial personnel and clerks in government service, and (b) studies other recorded variables, particularly lateness, long-term sickness absence, and passes from work both `medical' and `works'.
Multiple regression shows one-day absences to be generally associated with two-day absences, lateness, and medical passes but independent of works passes and long-term sickness absence; and two-day absences to be generally associated with one-day absences and long-term sickness absence but not with lateness or passes from work. Higher order correlations show lateness and works passes, medical passes and works passes, and lateness and age to be (weakly) associated, the last negatively.
Irrespective of season, one-day absences were consistently most prevalent on Monday and least so on Friday, with a subsidiary peak for the male groups on Wednesday; two-day absences - as measured by the day each absence starts - were consistently most prevalent on Monday and least so on Thursday (Friday was omitted) and during the summer months. Medical passes were generally independent of the day of the week and the period of the year; lateness was greatest on Friday and during the winter, though the increase was slight; but works passes were relatively prevalent on Friday.
Correlation and regression show the association between numbers of one-day absences taken in two periods of time (each one year) to be marked (r = 0·5 to 0·7), unaffected by transforming to normal functions, and explicable on a linear hypothesis, but the value of r to increase as the periods of time increase and to decrease as the interval between the periods of time lengthens. Similar analyses for two-day absences show r = 0·25 to 0·55, acceptance of a linear hypothesis, and a suggestion that the value of r may increase as the periods of time increase and as the interval between them shortens. More limited examination shows corresponding values of r (for contiguous years) to be of the order 0·25 for medical passes, 0·60 for works passes, but > 0·80 for lateness.
Values of r between each of these factors in turn for all possible pairs of days of the week are reasonably consistent and show r of the order 0·35 for one-day absences, 0·25 for medical passes, 0·40 for works passes, and 0·80 for lateness (values for two-day absences are irregular and in the range 0 to 0·4). The consistency of lateness experience over days and years is very marked, the correlations being among the highest recorded for any event involving human behaviour.
The importance and application of the findings are briefly discussed; detailed consideration is reserved for the third and last paper.