Establishing Construct Validity for Integrity Tests
Abstract
This dissertation presents extensive evidence related to convergent and discriminant validity of integrity tests in general. The results from both the primary data analyses, and the meta-analyses display significant convergent validity for all integrity tests. The presence of a general factor across integrity tests is confirmed. Of the Big Five dimensions of personality, conscientiousness was found to have the most overlap with integrity tests. However, a more accurate identification of counterproductive individuals is possible by considering scores on agreeableness and emotional stability. It appears that the higher predictive validity of integrity tests arise from the fact that they tap constructs broader than just narrow and construct limited conscientiousness as measured by mainstream personality inventories. With integrity tests, the increased breadth of predictor construct coverage translates itself info better prediction in personnel selection. Both overt and personality-based integrity test scores seem to be correlated with age. This indicates that younger individuals have the potential to more counterproductive employees, possibly due to youthful indiscretions and experimentation. Integrity and gender are also correlated. On average, females score higher than males. Overt integrity test scores may be correlated with race (with minorities scoring higher). However, the same results do not apply to personality-based integrity tests. It is clear that using integrity tests in personnel selection, either overt or personality-based ones, will not cause adverse impact. Integrity test scores and ability test scores are uncorrelated. Using integrity tests in conjunction with measures of ability can lead to substantial incremental validity for all job complexity levels.
Document Details
- Document Type
- Technical Report
- Publication Date
- Jun 01, 1993
- Accession Number
- ADA293940
Entities
People
- Deniz S. Ones
Organizations
- University of Iowa