Abstract
Objective
The Boston Naming Test (BNT) has recently been proposed as an embedded performance validity test (PVT) with high specificity/low sensitivity; however, this has not been replicated. This study therefore aimed to cross-validate findings in a mixed clinical neuropsychiatric sample.
Method
This cross-sectional study of 136 primary monolingual English-speaking patients who completed the BNT during outpatient evaluation was 57% female/43% male, 38% Caucasian, 39% African American, 16% Hispanic, and 6% Asian with mean age of 47.7 years (SD = 16.6) and mean education of 14.0 years (SD = 2.7). In total, 109/136 (80%) were classified as valid and 27/136 (20%) as invalid based on 4 independent criterion PVTs.
Results
Respective mean BNT raw/T-scores were 49.5 (SD = 9.2)/45.3 (SD = 10.9) for the valid group and 45.8 (SD = 8.2)/41.1 (SD = 7.8) for the invalid group. Analyses of variance fell just above significance for both BNT raw F(1, 134) = 3.75, p = .05 and T-scores F(1, 134) = 3.55, p = .06. Receiver operator characteristic curve analysis for the raw score was significant, with an area under the curve (AUC) of .67 (p < .01) and an optimal cutoff of ≤ 35 (4% specificity/90% sensitivity). BNT raw scores remained significant after removing bilingual participants, (AUC = .68; p < .01), with identical psychometric properties. In contrast, analysis of BNT T-scores (AUC = .61; p = .08) were nonsignificant.
Conclusions
Overall, results showed that the BNT cannot psychometrically distinguish valid versus invalid performance and therefore has questionable utility as a PVT in a mixed clinical setting. Findings contribute to a growing literature base cautioning against the indiscriminate use of measures of actual cognitive ability as validity indicators, particularly in populations with cognitive impairment.