FACET MEASURES AND GLOBAL JOB SATISFACTION Scott Highhouse
Indiana University-Purdue University at Indianapolis
Alene S. Becker
ABSTRACT: This research was an attempt to find evidence for Scarpello and Campbell's (1983) conclusion that a single-item global measure is more content valid than a composite facet measure for assessing overall job satisfaction. The first study investigated responses from employees in three separate plants who completed a facet measure, a single-item global measure, along with items and facets chosen by an employee committee. Results indicated that the employeechosen facets accounted for some incremental variance in the prediction of the single-item measure. A second study investigated the relationship between a single-item and a facet measure of benefit package satisfaction when the entire benefit package domain was included in the facet measure. The magnitude of the relationship suggested that the single-item measure of global benefit package satisfaction involved consideration of more than merely satisfac¢ion with each element in the package. Limitations and alternative explanations for the findings are discussed.
Although job satisfaction is one of the most widely studied topics in organizational behavior and h u m a n resource management, m a n y questions surrounding the measurement of the construct remain unanswered (Brief & Roberson, 1989; Rice, Gentile, & McFarlin, 1991; Scarpello & Campbell, 1983). One important question is w h y single-item measures of global job satisfaction (e.g.,All things considered, h o w satisfied are you with your job?) fail to correlate highly with composite facet job satisfaction measures. Studies comparing facet measures with single-item measures typically find that one measure does not account The helpfulcomments of Mark E. Tubbs and John W. Joneson an earlierversionof thisarticleare greatlyappreciated. Send allcorrespondenceto ScottHighhouse,IUPUI, Department ofPsychology,Indianapolis,IN 46202-3275. 117
© 1993 HumRn Sciences Press, Inc.
JOURNAL OF BUSINESSAND PSYCHOLOGY
for more t h a n 50 percent of the variance in the other (e.g., Aldag & Brief, 1978; Ferratt, 1981). Weighted models have not substantially improved the relationship in these studies. One possible reason for the lower than expected relationship may be that the single-item measure lacks reliability. Indeed, Stone and his colleagues (Stone, Stone, & Geutal, 1990) cautioned against using attitude measures composed of only a few items. However, Ironson, Smith, Brannick, and Gibson (1989) still found only a moderate relationship between global and composite facet measures using a multiple-item global scale. Scarpello and Campbell (1983) argued that the single-item global measure has temporal reliability and may be, in fact, superior to facet measures in capturing employee job satisfaction. Scarpello and Campbell interviewed incumbents from two plants about the factors that determined their job satisfaction. They found that subjects mentioned m a n y variables that were not included in a commonly used facet scale. Moreover, Scarpello and Campbell found that the variables mentioned by employees were more correlated with a single-item measure than with a composite facet measure and added incremental variance to the prediction of single-item global satisfaction. These authors concluded that facet measures may lack content validity and should not be summed for use as overall measures of job satisfaction. According to Scm~ello and Campbell (1983): "Low correlations between sum of many facet scores and the global measures are probably not due to unreliability of the global measures. Rather, they may be related to failure to measure the range of variables that influence job satisfaction" (p. 597). The authors went on to suggest that, because single-item global measures are more content valid...