Abstract
Using the Mayo Older Adult Normative sample (Ivnik et al., 1992) as our database, we developed regression models for estimating premorbid Full Scale (FSIQs), Verbal (VIQs), and Performance (PIQs) IQs for elderly adults. Age, years of education, and sex were the only demographic variables that showed sufficient variability; therefore, they were used as predictor variables in three stepwise procedures. The Mayo Normative FSIQs, VIQs, and PIQs served as the dependent variables. Both education and sex added significantly to the accounting of variance of both FSIQ and VIQ ( p < .001), whereas education ( p < .001) and age ( p < .05) were significant predictors of PIQ. These models produced statistically significant multiple __R__s of .54, .58, and .35 ( p < .0001), with standard errors of estimate of 9.02, 8.28, and 10.77 for FSIQ, VIQ, and PIQ, respectively. Estimated FSIQs generated with the present model and the model developed by Barona and colleagues (Barona, Reynolds, & Chastain, 1984) were compared. The correlation between estimated IQs was large, the mean difference between IQs was very small, the standard deviations were nearly equal, and the categorical distributions of the two were similar. Because the Barona model is likely to be familiar to most clinicians, these findings argue in favor of the continued use of the Barona model, even when assessing people older than the WAISβR normative sample. Extensions of these models to the WAISβIII also are discussed. Β© 2003 Wiley Periodicals, Inc. J Clin Psychol 59: 457β463, 2003.