The relationship of scale reliability and validity to respondent inconsistency

The generalizability of respondent inconsistency and of its relationship to estimates of the reliability and validity of personality measures was examined. Two hundred and six undergraduates completed two IPIP, the Thompson Mini-Markers measures of the Big Five and three other questionnaires. Respondent inconsistency was calculated based on average within-dimension standard deviations from one of the HIP questionnaires and used to form three groups of respondents differing in mean inconsistency on that questionnaire. When groups formed from the first questionnaire were compared, mean reliabilities of the 13 other scales were largest when estimated using the most consistent respondents. Convergent validities between the other two scales of all Big Five measures were larger, and validity of conscientiousness computed from those scales as a predictor of GPA was higher in groups composed of more consistent respondents. Implications for consideration of inconsistency as a stable personality characteristic are considered. (C) 2011 Elsevier Ltd. All rights reserved.