A primary goal of this project is to describe adult age differences and age changes in nonverbal memory performance. Nonverbal memory is measured in the Baltimore Longitudinal Study of Aging with the Benton Visual Retention Test (BVRT), a test of memory for designs involving spatial memory. This year, we analyzed the BVRT data for 277 men who had three measures spanning at least 1 years to address the question of whether cognitive declines are smaller for those who are initially better endowed. Although this question has been addressed in the literature on intelligence and aging, all of those studies share a common methodological problem. It is known that when there are error of measurement, initial level and measures of change are artifactually negatively related. In the current analyses, initial measures were adjusted for errors of measurement, and maximum likelihood procedures were applied to relate these adjusted scores to individual regression measures of change. We found small positive correlation between initial level and change for the BVRT in t two groups over age 55 at first test. Men with high error scores initially tended to increase their errors more than men with lower initial scores. When this procedure was applied to the Vocabulary subtest of the Wechsler Adult Intelligence Scale for the same 277 men, the results were quite different. M with high vocabulary.scores tended to decline more than men with lower initial levels. When the methodological bias is avoided, it appears that the answer the question about the relationship of initial level and change is specific for the kind of cognitive performance measured.
