Replicating the classification accuracy of the Verbal Paired Associates and Visual Reproduction recognition trials as embedded performance validity tests.

Abstract

Objective: This study was designed to replicate previous research on the clinical utility of the Verbal Paired Associates (VPA) and Visual Reproduction (VR) subtests of the WMS-IV as embedded performance validity tests (PVTs) and perform a critical item (CR) analysis within the VPA recognition trial. Method: Archival data were collected from a mixed clinical sample of 119 adults (M Age = 42.5, M Education = 13.9). Classification accuracy was computed against psychometrically defined criterion groups based on the outcome of various free-standing and embedded PVTs. Results: Age-corrected scaled scores ≤6 were specific (.89−.98) but had variable sensitivity (.36−.64). A VPA recognition cutoff of ≤34 produced a good combination of sensitivity (.46−.56) and specificity (.92−.93), as did a VR recognition cutoff of ≤4 (.48−.53 sensitivity at .86−.94 specificity). Critical item analysis expanded the VPA's sensitivity by 3.5%-7.0% and specificity by 5%-8%. Negative learning curves (declining output on subsequent encoding trials) were rare but highly specific (.99-1.00) to noncredible responding. Conclusions: Results largely support previous reports on the clinical utility of the VPA and VR as embedded PVTs. Sample-specific fluctuations in their classification accuracy warrant further research into the generalizability of the findings. Critical item analysis offers a cost-effective method for increasing confidence in the interpretation of the VPA recognition trial as a PVT.