Intraindividual variability dispersion (IIV-d) measures variation in a patient's test performance throughout a neuropsychological evaluation. Higher IIV-d has been identified as a potentially useful indicator of both cognitive impairment and noncredible test performance, but further exploration of the utility of these metrics for group classification is warranted. This study assessed measures of IIV-d-individual standard deviation and coefficient of variation (CoV)-for classifying levels of neurocognitive impairment and invalid test performance. Adult patients referred for outpatient neuropsychological evaluation (N = 421, Mage = 44.49, Meducation = 13.71) were administered four freestanding performance validity tests and a neuropsychological test battery yielding 18 indicator scores. Patients were classified into four groups: no neurocognitive disorder (n = 125), mild neurocognitive disorder (n = 163), major neurocognitive disorder (n = 45), and invalid performance (invalid; n = 88). Analyses of covariance revealed that both individual standard deviation (F = 13.75, p F = 64.14, p < .001, η² = 0.31) significantly differed across groups. Results showed that high levels of variability followed increasing severity of cognitive impairment (i.e., no neurocognitive disorder < mild neurocognitive disorder < major neurocognitive disorder); performance invalidity was also associated with high variability. Additional analyses highlighted CoV as a more robust indicator of prediction utility than individual standard deviation. Overall, findings suggest that use of IIV-d metrics, particularly CoV, differentiates level of cognitive impairment. Performance invalidity can also be identified in some cases, although differentiating invalid responding from major neurocognitive disorder is more challenging using IIV-d metrics alone. Ongoing research is warranted for better understanding the use of IIV-d in facilitating diagnostic clarification and determining whether this method provides incremental value beyond other methods. (PsycInfo Database Record (c) 2026 APA, all rights reserved).
Leese et al. (Thu,) studied this question.