De gevolgen van nonrespons en retrospectieve vertekeningen voor de uitkomsten van een onderzoek naar ziekteverzuim

Samenvatting

Employees of a Dutch road-building company (N = 205) were sent mail surveys on absenteeism, & responses (77%) were compared with administrative records. Findings support the hypotheses that nonrespondents are on sick leave more frequently & longer than respondents, & underreporting of sick leave duration & frequency is more common than overreporting. It is noted that nonresponse & retrospective bias have a cumulative effect & lead to distortions in variable correlations. No distributional bias was found in sociodemographic variables. 8 Tables, 57 References. Adapted from the source document.