A Caveat on Using Single-Item Measures

ABSTRACT

Methodologists advocate the use of multipleitem versus singleitem measures for relatively complex constructs. However, Wanous and his colleagues recently challenged this stringent view. Using the classical formula for the correction for attenuation and data from over 500 subjects, they demonstrated that meaningful reliability estimates can be calculated for single-item measures. The present study provided an independent examination of this approach by using quality of work life data from a sample of 562 teaching and administrative faculty from two Canadian community colleges. The obtained single-item reliability estimates for Overall Job Satisfaction (.45) and Job Dissatisfaction (.21) were low whereas the single-item reliability estimates for the individual job satisfaction facets of Supervision (.80) and Work Group (.57) were higher. While these results are interesting, they are not strong enough to challenge the long-standing view that multipleitem measures are needed to reliably measure relatively complex constructs such as quality of work life. Practitioners and researchers are warned to be wary of single-item measures.