Feigning or forgetfulness: The effect of memory impairment severity on word choice test performance

Abstract

Objective

This study cross-validated the word choice test (WCT) in a diverse neuropsychiatric sample and examined the effect of increasing verbal memory impairment severity on WCT performance.

Method

Data from 147 clinically referred patients (113 valid/34 invalid) who completed the WCT, Rey Auditory Verbal Learning Test (RAVLT), and four independent criterion PVTs were analyzed. RAVLT memory impairment bands used were: ≥37T (normal memory); 30T–36T (below average scores/mild impairment); and ≤29T (extremely low scores/severe impairment).

Results

WCT and RAVLT were moderately correlated. The invalid group had significantly worse performance on the WCT and RAVLT. For the overall sample, the WCT yielded an area under the curve (AUC) = .79, with 62% sensitivity/93% specificity at a cut-score of ≤41. When the sample was subdivided by memory impairment severity, the severe impairment group had significantly lower WCT scores than the normal group. Moreover, the WCT retained moderate classification accuracy among the normal memory (AUC = .85) and mild memory impairment (AUC = .76) groups, with sensitivities of 65% and 62% (≥91% specificity) at their respective optimal cut-scores of ≤44 and ≤42. In contrast, the WCT had low classification accuracy among those with severe memory impairment (AUC = .66), with only 15% sensitivity/95% specificity at the optimal cut-score of ≤30.

Conclusion

The WCT is generally useful for detecting invalid neuropsychological test performance, although, its classification accuracy was diminished among patients with severe memory impairment. Therefore, while the WCT remains a viable option for performance validity assessment, neuropsychologists should carefully consider its use when this level of severe memory impairment is known or suspected.

Keywords:

• Performance validity
• assessment
• neuropsychology
• memory
• psychometrics

Disclosure statement

The authors have no conflicts of interest to report, and none have any financial interest with the subject matter discussed in the manuscript.