Construct validity of ImPACT in athletes with concussions

Archives of Clinical Neuropsychology. 2004 Nov;


Iverson, G., M. D. Franzen, M. R. Lovell and M. W. Collins.



Objective: The purpose of this study was to examine the convergent and discriminant validity of computerized neuropsychological testing in a sample of athletes with concussions. A multitrait-multimethod matrix (Campbell & Fiske, 1959) was computed. Method: Twenty-five amateur athletes completed ImPACT Version 2.0 and several traditional neuropsychological measures within 20 days post-injury (mean = 9.6, S.D. = 5.5). The traditional measures were the Trailmaking Test, Symbol Digit Modalities Test (SDMT), and the Brief Visuospatial Memory Test (BVMT). Results: The two ImPACT memory composites were highly correlated (r = .75). The two ImPACT memory composites had medium correlations with the BVMT total score (r = .50 for both), and high correlations with the BVMT delayed recall score (r = .85 for both). There were medium correlations between the ImPACT processing speed composite and Trails B (r = −.60) and the SDMT (r = .68). There was a medium correlation between the ImPACT visual memory and reaction time composites (e.g., r = −.50). The two ImPACT memory composites were not significantly correlated with Trails A or Trails B, but they were significantly correlated with the SDMT (r = .42–.47). The ImPACT reaction time composite was not significantly correlated with the total score from the BVMT. Conclusion: In general, the ImPACT composites showed the expected relationships with other tests. The memory composites were most highly correlated with the traditional memory measures and the Processing Speed composite was correlated most highly with Trails B and the SDMT.

Links to full article: