Construct Validity Of The Immediate Post-Concussion Assessment and Cognitive Testing (ImPACT ) With Traditional Neuropsychological Measures

J Int Neuropsychol Soc -


Allen, B. J., J. D. Gfeller, A. A. Capps, P. D. Ruppert and M. J. Ross.



Objective: Computerized test batteries (e.g. ImPACT) have several advantages, relative to traditional “paper-pencil” tests for assessing the neuropsychological effects of sport-related concussion. Research indicates that some ImPACT scores correlate moderately with traditional measures (e.g. BVMT-R, SDMT). The present study further contributed to this literature by examining the construct validity of ImPACT. Participants and Methods: A sample of undergraduate participants completed ImPACT and the “NFL” battery of tests (i.e., HVLT-R, BVMTR, Trail Making Test, COWA, and WAIS-III Symbol Search, Digit Symbol – Coding, and Digit Span) in a counter-balanced order of administration. The sample (N=100) was 56% female, 76% Caucasian, and 93% right-handed, with an average age of 19.69 (SD=1.19), and 13.21 (SD=1.09) years of education. Results: Analyses revealed a wide range of correlations that demonstrated ImPACT indicators frequently correlated, as expected, with traditional tasks measuring similar constructs. However, the specific ImPACT subtests that comprise the various composite scores did not correlate highly with each other. In light of this finding, factor analyses were conducted to examine the factor structure of ImPACT. A five-factor solution accounted for a total of 69% of the variance. The factors were determined to measure the following constructs: forced choice efficiency, primarily memory, inhibitory cognitive abilities, visual processing abilities with an element of memory, and errors on a Stroop-like task. Conclusions: Notably, the subtest composition of these factors differs considerably from the standard ImPACT composites and suggests the test’s composite measures merit further examination. Clinicians should exercise caution in interpreting ImPACT composites as they currently stand.

Links to full article: