Statistics (i.e., sensitivity, specificity, hit rates, positive and negativ
e predictive values, odds ratios, and likelihood ratios) that best describe
a diagnostic test's ability to classify persons as either "impaired" or "n
ormal," but that are not commonly reported in neuropsychological research,
are reviewed. These statistics are applied to Mayo Cognitive Factor Scale s
cores (MCFS; Smith et al., 1994) to demonstrate information that can be acq
uired about the diagnostic capabilities of cognitive tests as they are comm
only used in clinical settings. Multivariate analyses then generated a stat
istical model that combines MCFS scores and improves on the diagnostic capa
bilities of the individual MCFS scores. This model enjoys better diagnostic
power than individual scores. It establishes that cognitive testing that u
ses multiple measures is very good at differentiating normal from impaired
cognitive states. Information is also provided that helps clinicians quanti
fy a person's risk for cognitive impairment based on specific cognitive tes
t score(s).