Measurement and non-measurement influences of test-session behavior on individually administered measures of intelligence

Two nested structural models were developed to determine whether test-sessi on behaviors affect the manner in which intelligence is measured or whether their influence is related to the constructs being measured. Children's te st-session behaviors were assessed using the Guide to the Assessment of Tes t-Session Behaviors for the WISC-III and WIAT (GATSB; Glutting & Oakland, 1 993) and intelligence was measured with the Wechsler Intelligence Scale for Children-Third Edition (WISC-III; Wechsler, 1991). Model 1 investigated re lationships between test-session behaviors and the WISC-III's four-factor s olution by first evaluating the underlying factor structure of the instrume nts. Thereafter, this measurement model served as a baseline against which alternative models were compared. The alternative models considered the in fluence of test-session behaviors on: (a) the subtests used to measure the WISC-III's; constructs of intelligence, and (b) the actual constructs of in telligence. Model 2 explored similar relationships by considering only the WISC-III's Verbal and Performance dimensions. Results indicate that test-se ssion behaviors play a larger role influencing the mechanisms through which intelligence is measured than on the actual constructs of intelligence. Im plications are discussed for clinical practice. (C) 1998 Society for the St udy of School Psychology. Published by Elsevier Science Ltd.