A computer-based reading test for use as an index of premorbid general intellectual level in North American English-speaking adults.

Brief tests of vocabulary or reading ability are often used as an index of general intellectual level to hypothesize deficits in patients, to establish comparability of comparison groups, and as a covariate of performance on other neuropsychological tests. We evaluated a new variant of a reading test for potential use in estimating premorbid general intellectual ability and implementation as a computerized-adaptive test. A total of 319 outpatients who participated in an epilepsy self-management study were tested on three occasions, completing the Wide-Range Achievement Test-3 (WRAT-3) Reading test at baseline and the NES3 Adult Reading Test (ART) at the 3-month follow-up. At the 6-month follow-up, some participants completed the ART and some the NES2 Vocabulary test. ART and WRAT-3 Reading scores obtained 3 months apart were highly correlated (r=0.87, n=179), while ART and NES2 Vocabulary scores were somewhat less correlated (r=0.80, n=61). The ART test-retest correlation was high (r=0.95, n=79). The ART distribution was spread out more evenly at the high end than that of the WRAT-3 Reading. ART score was significantly related to education, race, income, and examiner category, but was unrelated to age category and depression status. Non-significant gender differences were observed. An item-response theory (Rasch model) analysis showed good fit of the model to the observed data and yielded an improved ranking of the ART items in order of difficulty. The NES3 ART exhibited high reliability and strong association with WRAT-3 Reading. These results suggest that the NES3-ART may be useful as an index of general intellectual ability for comparison of groups and for use as a covariate in data analyses among English-speaking, North American adults. The Rasch model analysis provides a basis for implementing the ART as a computerized-adaptive test, which should improve its time efficiency and reduce participant burden in taking the test.