An Investigation of Gender Bias in the Personality Assessment Inventory in a Sample of Military Veterans of the United States

The purpose of this study was to examine the possible effects of gender bias within the Personality Assessment Inventory (PAI) as applied within the United States Veterans Health Administration (VHA). A quantitative descriptive design was employed using preexisting clinical data obtained from the Corporate Data Warehouse (CDW) as accessed through the VA Informatics and Computing Infrastructure (VINCI) to examine means differences, factor invariance, and predictive validity using a sample of 9,079 PAI profiles completed by veterans within the VA system. Mean differences were examined via t-tests between men and women veterans. Factor invariance was explored by comparing the overall factor structure of this sample with the standardization samples presented in the Personality Assessment Inventory (PAI): Professional Manual, 2nd Edition (Morey, 1991), by comparing the factor structure of men and women veterans, and by factor analyzing three scales of interest separately for the two groups. Finally, hierarchical binary logistic regression was employed to test the predictive validity of the PAI and assess for slope and/or intercept bias. Overall, there was minimal evidence of bias within this instrument based on gender. There were some mean differences on T-scores of the PAI between men and women with small to medium effect sizes. The factor structures were mostly similar in comparison with the standardization samples and between the veteran samples. Some trends included the Mania (MAN) scale performing inconsistently across all samples and the ANT (Antisocial), ALC (Alcohol Problems), and DRG (Drug Problems) scales loading together consistently. The individual factor analyses of the ANX (Anxiety), ARD (Anxiety Related Problems), and DEP (Depression) scales did not provide a meaningful factor structure and when restricted to one factor solutions, nearly twice as much variance was explained in the women veteran sample than the men’s sample. Finally, the logistic regression equations demonstrated intercept bias in prediction of male veterans having a lower likelihood of a depressive diagnosis, relative to female veterans, above and beyond all the control variables and T scores for the DEP scale. No other intercept or slope bias was detected. Although some differences were found, these results are primarily supportive of this being a useful instrument with the veteran population across gender.

Name thesis:degree_name

Ph.D.

Level thesis:degree_level

Doctoral

Discipline thesis:degree_discipline

Counseling Psychology (UMKC)

Grantor dc:publisher

University of Missouri--Kansas City

Year dc:date.issued

2017

Author dc:creator

• Williams, Sarah Elise

Advisors dc:contributor.advisor

• Berkel, LaVerne A.
• Barber, Carolyn E.

Language dc:language.iso

en_US

Handle dc:identifier.uri

https://hdl.handle.net/10355/60578

OAI identifier oai:identifier

oai:mospace.umsystem.edu:10355/60578