Construct validity of eight tests of attention: Comparison of normal and closed head injured samples

Abstract

Attentional problems are common symptoms of brain impairment and are generally assessed by a number of psychological tests. However, clinicians do not always agree on the processes measured by these tests and validation of the tests is often inadequate. The present study used factor-analytic techniques to examine the construct validity of eight attention tests (Letter Cancellation, Serial Subtraction, Digit Span, Digit Symbol, Stroop Colour-Word, Trail Making, Symbol Digit Modality, and Knox Cube). These tests were administered to 125 university controls, 45 normal controls from the community, and 37 closed-head-injured patients. Each of the 13 measures from the eight tests were found to load on one of three components/factors (identified as visuo-motor scanning, sustained selective processing, and visual/auditory spanning) for the normal as well as the patient group. Comparison of the mean performances of the patients and their matched controls suggested that: (a) severe short-term patients were impaired on the visuo-motor scanning and visual/auditory spanning components; (b) severe long-term patients were impaired only on the visuo-motor scanning component; and (c) mild short-term patients were not impaired on any of the components. The implications of these findings for the measurement of attention are discussed.