Intra-individual Cognitive Variability as a Predictive Sign of Future Cognitive
  
   Lead Investigator:    Alyssa De Vito
   Institution      :    Louisiana State University
   E-Mail           :    adevit1@lsu.edu
   Proposal ID      :    813



   Proposal Description:
   The present study seeks to further contribute to the literature on cognitive variability as an early predictor of future cognitive impairment by conducting a coordinated analysis of three longitudinal studies: the English Longitudinal Study on Aging (ELSA), Rush Memory and Aging Project (MAP), the Einstein Aging Project (EAS), and National Alzheimer?s Coordinating Center (NACC) data set. These data sets were chosen based on criteria which allow for an examination of within-personality variability and later conversion to dementia (e.g., numerous waves of data collection, annual evaluations). The aim of this analysis will be to replicate the findings of Galmado et al., (2012) who, using the Baltimore Longitudinal Study of Aging, found that greater within-person variability in cognition was associated with later diagnosis of dementia. We will follow the data analytic plan presented in Galmado et al., (2012) which used multilevel modeling (MLM) to identify patterns of cognitive trajectories across cognitive tests and tested whether estimates of within-person variability were similar in a group of individuals who never converted to dementia vs. a group of individuals who did convert. By constraining within-in person variability to be equal in each group or freely estimated, nested models can be compared using the Akaike information criterion (AIC). Analyses will be conducted using restricted maximum likelihood estimation. Assessments conducted within 5 years of dementia diagnosis will not be included to allow for a focus on variability as an early predictor.