Analysis of skew, examination of intercorrelations, and determining the optimal threshold for performance invalidity...

This study cross-validates and expands upon previous research by examining the optimal number of PVT failures necessary to determine invalid performance when 10 PVTs are administered during a neuropsychological evaluation. Additionally, the study assessed the degree of skewness of individual PVTs and PVT intercorrelations for the overall sample and by validity group.

Participants were 283 adult neuropsychology outpatients evaluated at an academic medical center. Participants were initially classified as having valid (≤1 PVT failure; n = 225) or invalid (≥2 PVT failures; n = 58; base rate of 20% performance invalidity) performance based on four independent criterion PVTs. Failure rates of 10 additional PVTs were then compared, and sensitivity and specificity were calculated at different thresholds (e.g. ≥1, ≥2, ≥3, ≥4 PVT failures) to determine the optimal threshold for detecting invalid performance while maintaining ≥ 90% specificity.

Findings indicate that failing ≥ 2 PVTs yielded 86% sensitivity/76% specificity, failing ≥ 3 PVTs yielded 69% sensitivity/92% specificity, failing ≥ 4 PVTs yielded 57% sensitivity/96% specificity, failing ≥ 5 PVTs yielded 29% sensitivity/99% specificity, and failing ≥ 6 PVTs yielded 22% sensitivity/100% specificity. PVT intercorrelations were generally small for the overall sample and by validity group. As expected, data were more highly skewed for patients with valid performance.

Findings were consistent with previous research and demonstrate that the three-failure threshold optimally detects invalid performance when 10 PVTs are administered. These findings inform the use of multiple PVTs in clinical settings and aid in the interpretation of PVT results.