Funded Research
Latent Growth Curve Models of Cognitive Aging
Source: NIH/NIA
Active: 09/30/02 - 07/31/06
Investigator(s):
Duane Alwin
Linda Wray
This research applies strategies of latent growth curve (LGC) modeling within a structural equation framework to examine intra-individual change in trajectories of cognitive performance in old age and its correlates, particularly age, education, health, and sensory functioning. The proposed work focuses explicitly on the following objectives: (1) to confirm patterns of age-related change across a broad spectrum of mental abilities using both cross-sectional and longitudinal assessments of cognitive performance in nationally representative samples of the older population; (2) to ascertain the extent to which cross-sectional age-differences in cognitive functioning are spuriously due to cohort-related factors, such as level of schooling, or other age-related phenomena, e.g., comorbidity and sensory functioning; (3) to assess the age-related intra-individual trajectories of several measures of cognitive functioning as well as age-linked covariates such as comorbidity and sensory functioning using longitudinal data over periods of time of up to seven years; (4) to assess the extent to which age-related declines in health and sensory function explain the link between aging and cognitive functioning in middle-aged and older adults, net of education; and (5) to assess how patterns of age-related cognitive change and its covariates differ by sex and race/ethnicity in older adults.
We investigate these issues using two innovative nationally-representative panel surveys of middle-aged and older adults: the original HRS (Health and Retirement Study) national panel study of pre-retirement men and women aged 51-61 assessed in 1992 (n=9,824) and re-interviewed in 1994, 1996, 1998, 2000, and 2002 and the parallel AHEAD (Study of Asset and Health Dynamics Among the Oldest Old) national panel study which interviewed adults aged 70 and above in 1993 (n=7,443) and re-interviewed in 1995, 1998, 2000, and 2002. The key elements of our design allow us to: (1) generalize to a national population of middle-aged and older individuals sampled using probability methods; (2) assess differences between birth cohorts in the processes studied; (3) assess occasion-based age changes in the population to estimate the nature and shape of intra-individual change; (4) assess inter-individual and inter-cohort differences in intra-individual change; (5) build models of accelerated age-based change using the multiple cohort feature of the panel design; (6) assess the effects of fixed covariates and appropriately lagged time-varying covariates on individual differences in age-related change; and (7) assess the extent of bias in these aging functions introduced by problems of sample attrition and/or mortality.
