Item Reversals and Response Validity in the Job Diagnostic Survey

Abstract

Experts disagree on whether questionnaires should have items worded in both positive and negative directions or items all worded in the same direction. This study examined this controversy in a sample of 921 employees of a state agency who completed two versions of a widely used questionnaire, Hackman and Oldham's Job Diagnostic Survey. One version had both positively and negatively worded items, and the other version had only positively worded items. Subsamples of respondents high and low on inconsistency and inaccuracy scales (response validity scales developed by the author) were studied to determine whether invalid responding might account for problems with item wording. Correlations between job characteristics (skill variety, task identity, task significance, autonomy, and feedback) and work outcomes (general job satisfaction, internal motivation, and growth satisfaction) were compared between subsamples using the Fisher r to z transformation. Out of 18 comparisons, only 3 correlations were significantly different for the subsamples of respondents. The conclusion was that results were similar regardless of item wording and regardless of invalid responding. However, further research may be warranted to assess whether inflated correlations for respondents high on the inaccuracy scale suggests that they are prone to common method variance.