Neuropsychologists have a problem: Their measures indicate elderly African Americans and Hispanics have more cognitive impairment than whites, even when that isn't the case, said Jennifer Manly, PhD, a psychologist at the Columbia University Medical Center, at APA's 2005 Annual Convention.

One proposed fix is for cognitive tests to factor in more ethnic-group norms, or information about racial classification. For Manly, though, using such norms may just foster stereotyping across racial groups because the norms assume such groups share educational, socioeconomic and behavioral characteristics.

She instead wants cognitive research to "deconstruct race" by giving more weight to intercultural factors like educational quality and literacy--which better predict older adults' cognitive performance than race, her research suggests.

"We need to consider how these factors vary within and between ethnic groups, determine their relationship to cognitive test performance and adjust for these factors before interpreting scores," she said.

After all, said Manly, "When we collect ethnic-group norms, we ignore diversity within groups having to do with education, where you're from, where you grew up, socioeconomic issues, and, perhaps most importantly when it comes to cognitive tests, exposure to mainstream culture."

Using a different approach, Manly and her Columbia University research team carefully measure diversity across and within groups, revealing some stereotype-busting patterns: In one series of studies, for example, they're considering the influence of educational quality on older white and African-American adults' cognitive performance. To do that, they examine factors like per-pupil expenditures and student-teacher ratio across different states where participants attended school between 1917 and 1943.

They've found that both African Americans and whites schooled in districts with higher expenditures and lower student-teacher ratios tend to score higher on the Wide Range Achievement Test-3.

Also telling, Manly said, are her team's results for literacy effects. In a longitudinal sample of 136 African-American, Hispanic and white elders, ages 67 to 94, Manly found that, across ethnic groups, word-list recall declined most sharply over time in those with lower literacy levels. The results, published in the Journal of Clinical & Experimental Neuropsychology (Vol. 25, No. 5, pages 680-690), suggest that, across ethnic groups, higher literacy more effectively protects against cognitive decline than years of education.

Overall, said Manly, "Deconstructing race in education really pays off in terms of looking at aging and cognition and risk for dementia, and it ought to be considered in future research on diverse elders."