Presenteeism (reduced productivity at work) is thought to be responsible for large economic costs. Nevertheless, much of the research supporting this is based on self-report questionnaires that have not been adequately evaluated.
To examine the level of agreement among leading tests of presenteeism and to determine the inter-relationship of the two productivity subcategories, amount and quality, within the context of construct validity and method variance.
Just under 500 health care workers from an urban health area were asked to complete a questionnaire containing the productivity items from eight presenteeism instruments. The analysis included an examination of test intercorrelations, separately for amount and quality, supplemented by principal-component analyses to determine whether either construct could be described by a single factor. A multitest, multiconstruct analysis was performed on the four tests that assessed both amount and quality to test for the relative contributions of construct and method variance.
A total of 137 questionnaires were completed. Agreement among tests was positive, but modest. Pearson r ranges were 0 to 0.64 (mean = 0.32) for Amount and 0.03 to 0.38 (mean = 0.25) for Quality. Further analysis suggested that agreement was influenced more by method variance than by the productivity constructs the tests were designed to measure.
The results suggest that presenteeism tests do not accurately assess work performance. Given their importance in the determination of policy-relevant conclusions, attention needs to be given to test improvement in the context of criterion validity assessment.