Measuring absenteeism from the opposite side of the coin.

Arguments for abandoning measures of absenteeism in favor of the criterion, of people who come to work, were supported empirically. Measures of absenteeism and attendance were collected daily from 379 wood harvesting crews for 12 consecutive weeks. Intercorrelations were made between early odd weeks, early even weeks, late odd weeks, and late even weeks. The test-retest reliabilities of worker attendance were superior (.92-.96) to those of absenteeism (.41-.61). These results were replicated on a second sample of 46 wood harvesting crews. Problems in deriving a satisfactory measure of absenteeism have been noted by Huse and Taylor (1962). They found that the intercorrelations between different absenteeism measures can be quite low. Only two such measures, absence frequency—the total number of times absent, and attitudinal absences —the frequency of 1-day absences, could be measured with sufficient reliability to warrant their use as a criterion. Another problem has been pointed out by Latham (Note 1) when dealing with the marginal or hard-core worker. He found that absenteeism measures of these individuals had to be discarded because they were confounded by measures of turnover. A worker who was recorded as terminating his job might reappear 2 days or even 2 weeks later. Similarly, another worker might be recorded as absent for over a month before someone decided by fiat that that worker had quit the job. It is the authors' contention that attempts to use absenteeism as a criterion should be abandoned in favor of measuring an observable job behavior, that is, the number of people who come to work. This is because management is really interested in this measure rather than the number of people who are off the job, and psychologists are interested in recording and understanding behavior (e.g., the response of coming to work) rather than