Relationship of Personality Assessment Inventory (PAI) over-reporting scales to performance validity testing in a military neuropsychological sample

ABSTRACT

This study evaluated the Personality Assessment Inventory’s (PAI) symptom validity-based over-reporting scales with concurrently administered performance validity testing in a sample of active-duty military personnel seen within a neuropsychology clinic. We utilize two measures of performance validity to identify problematic performance validity (pass all/fail any) in 468 participants. Scale means, sensitivity, specificity, predictive value, and risk ratios were contrasted across symptom validity-based over-reporting scales. Results indicate that the Negative Impression Management (NIM), Malingering Index (MAL), and Multiscale Feigning Index (MFI) scales are the best at classifying failed performance validity testing with medium to large effects (d = .61–.73). In general, these scales demonstrated high specificity and low sensitivity. Roger’s Discriminant Function (RDF) had negligible group differences and poor classification. The Feigned Adult ADHD index (FAA) performed inconsistently. This study provides support for the use of several PAI over-reporting scales at detecting probable patterns of performance-based invalid responses within a military sample. Military clinicians using NIM, MAL, or MFI are confident that those who elevate these scales at recommended cut scores are likely to fail concurrent performance validity testing. Use of the Feigned Adult FAA and RDF scales is discouraged due to their poor or mixed performance.

KEYWORDS:

• Personality Assessment Inventory
• PAI
• Feigning
• Military
• Over-reporting

Disclosure statement

No potential conflict of interest was reported by the author(s).

Data availability statement

Due to the nature of this research and IRB stipulation, participants of this study did not agree for their data to be shared publicly, so supporting data is not publicly available.

Notes

1. AUC and classification accuracies range from 0 (completely inaccurate classification) to 1.00 (completely accurate classification), with a value of .50 indicating classification at random chance levels. AUC values were interpreted as having small (.57), medium (.64), and large (.71) effects sizes (Rice & Harris, 2005).