Optimal per test cutoff scores and combinations of failure on multiple embedded performance validity tests in detecting performance invalidity in a mixed clinical sample

Abstract

We tested the usefulness of six embedded performance validity tests (EPVTs) in identifying performance invalidity in a mixed clinical sample. Using a retrospective design, 181 adults were classified as valid (n = 146) or invalid (n = 35) performance based upon their performance on one of three standalone PVTs (Test of Memory Malingering, Victoria Symptom Validity Test, Dot Counting Test). Multiple cutoffs were identified corresponding to predetermined false positive rates of 0, 5, 10, and 15% for each of six EPVTs. EPVT cutoffs corresponding to the predetermined false positive benchmarks were generally more conservative than currently established scores. Sensitivity was low (.0%–42.9%) for individual EPVTs across these cutoffs and was moderately improved by the combination of multiple EPVT failures. The optimal number of EPVT failures using the 10% false positive rate was ≥ 2. Although the overall classification accuracy of 80.7% and specificity of 89.0% were comparable to prior research, the sensitivity of 45.7% was more modest than previous estimates. Low sensitivities indicate that this combination of EPVTs failed to detect a majority of invalid performers.

Keywords:

• Boston naming test
• controlled oral word association test
• embedded measures
• effort
• embedded performance validity test
• failure to maintain set
• noncredible performance
• performance invalidity
• reliable digit span
• trail making test

Disclosure statement

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