Lee, Mark2022-09-262022-09-262022-07https://hdl.handle.net/11299/241732University of Minnesota Ph.D. dissertation.July 2022. Major: Sociology. Advisors: Jeylan Mortimer, John Warren. 1 computer file (PDF); vi, 197 pages.Age-adjusted dementia prevalence has significantly declined in the United States over the last 25 years, despite little advancement in the biomedical treatment of Alzheimer’s Disease or improvement in proximal dementia risk factors. In this dissertation, I analyze data from the Health and Retirement Study (HRS) to improve current understanding of the descriptive trends and causal mechanisms underlying dementia prevalence decline. In my first study, I rebut the argument that dementia decline in the HRS is an artefact of unmeasured panel conditioning. I show that practice effects do not bias the estimated secular trend in dementia prevalence after accounting for selective panel attrition. In my second study, I argue that cohort trends in early life risk factors offer a more plausible explanation of the observed dementia improvement than period trends, which have been emphasized in previous research. In my third study, I empirically test the contribution of early life risk factors to cohort trends in dementia prevalence. I find that age- and sex-adjusted dementia prevalence declined 2.3 percentage points per 10-year increase in birth year for cohorts born 1892-1952. The majority (72%) of this trend was explained by increases in educational attainment for more recent cohorts. Proximal risk factors had little influence net of education and other early life factors. The trend in dementia decline was steeper for Black than White Americans, and the causal mechanisms also differed by race. In my fourth study, I document cohort trends in midlife cognitive aging. I find that, compared with those born 1942-1947, those born 1954-1959 entered midlife with lower cognitive function, but exhibited greater maintenance of cognition over time. This suggests that dementia prevalence may continue to improve as this latter-born cohort ages. Overall, this research reinforces the importance of social improvement (especially educational expansion) across the 20th century for cognitive health improvements in the 21st century. This work indicates that interventions to reduce or delay dementia and ameliorate racial disparities should be expanded to include social determinants of health across the life course.enagingcognitiondementialife courselongitudinalpopulation healthThesis or Dissertation