Utility of the D-KEFS Color Word Interference Test as an embedded measure of performance validity

Abstract Objective: The D-KEFS Color Word Interference Test (CWIT) is among the most commonly administered measures of verbally mediated processing speed and executive functioning. Previous research suggests that this test may also be sensitive to performance invalidity. We sought to develop new embedded measures of performance invalidity based on multi-condition performance on the CWIT and to evaluate previously proposed embedded measures for performance invalidity on this test. Method: Newly developed and previously proposed embedded measures were evaluated in three samples: a traumatic brain injury (TBI) sample (N = 224) classified into psychometrically-defined performance-valid and performance-invalid subgroups; the D-KEFS normative sample (N = 1,735); and a Parkinson’s disease sample (N = 166). To explore the resilience of CWIT cutoffs to the effects of age and neurocognitive impairment, the D-KEFS normative sample was stratified into age groups of 8–19, 20–69, and 70–89 and the Parkinson’s disease sample into normal cognition, mild cognitive impairment, and dementia groups. Results: Our newly developed embedded measures derived from within the TBI sample were largely resilient to the effects of age and neurocognitive impairment other than dementia. Overall, a cutoff of ≤18 on the sum of age-corrected scaled scores on all four conditions exhibited the strongest classification accuracy. In contrast, previously proposed cutoffs (Erdodi et al., 2018) produced unacceptable false-positive rates within the TBI sample and generally a high rate of potential false positives in the normative age and clinical groups. Conclusion: Scoring ≤18 on the sum of all conditions on the CWIT appears to be an optimal cutoff for performance invalidity detection in non-demented samples.