Cross-validation of the WAIS-IV Similarities incidental learning task

The incidental learning index based on Similarities subtest (IL-Sim) from the Wechsler Adult Intelligence Test-Fourth Edition (WAIS-IV) has shown promise as a screening measure for memory impairment. We aimed to examine the reliability, convergent and divergent validity, and the utility of IL-Sim index in classifying at least mild and severe memory impairment. One hundred and fifty-three Veterans completed the WAIS-IV Similarities subtest and Repeatable Battery for the Assessment of Neuropsychological Status (RBANS) as part of their clinical evaluations. IL-Sim had acceptable internal consistency (Cronbach's α = .85). Among RBANS indices, IL-Sim had the largest association with the Delayed Memory Index (r = .69), followed by the Immediate Memory Index (r = .59), and relatively smaller associations with non-memory indices (rs = .26-.38). For both younger (ages 40-69) and older (ages 70-87) adults, IL-Sim had excellent discriminability between patients without memory problems and those with at least mildly or severely impaired RBANS memory (AUC = .83-.87). To enhance the ease of interpretation for IL-Sim scores, we presented logistic regression-derived probabilities across multiple base rates for each age group and impairment severity. IL-Sim showed strong convergent and divergent validity and discriminated along the memory impairment continuum highlighting its promise as an efficient screening measure of memory.