This application was initially submitted as an R29 FIRST Award, and is being re-submitted as an R01 in response to Program Announcement, PA-97-065 (NIA) entitled, """"""""Social Cognition and Aging."""""""" The goal of this project is to determine how individual risk perceptions influence health behaviors among the near-elderly (51-61) and decisions regarding wealth holdings among the elderly (70+). Among the near-elderly, the following health behaviors will be studied: the use of preventive services (mammography, prostate screening, and cholesterol screening), weight loss and exercise. The preventive services are designed to reduce mortality from cancer and heart disease. Weight loss and exercise are both preventive measures as well as mitigators of harm that results from health shocks. Expanding the use of these preventive and mitigating measures are important to reduce morbidity and mortality from heart disease and cancer. Among the elderly, the behavior of interest relates to asset accumulation/dissaving, specifically focusing on housing decisions. The specific decisions will be selling a house, and changes in housing wealth holdings. The two decisions will be analyzed because housing assets likely have fewer measurement errors, and because housing is the major component of wealth for most elderly persons. Housing decisions are important from a policy perspective because they may represent precautionary savings to finance long-term care, a substantial risk the elderly face. The proposed conceptual model is an extension of the common Bayesian learning model whereby prior risk perceptions are updated on the basis of information. Our model holds that updated risk perceptions, in turn, influence behavior. Risk perceptions will be estimated as enodogenous explanatory variables simultaneously with health behaviors among the near-elderly and housing decisions among the elderly. The panel structure of the data bases to be used (4 waves of Health and Retirement Study; 3 waves of Asset and Health Dynamics Among the Oldest Old) will allow for the modeling of behaviors in time 3 or 4 as a function of changes in risk perceptions regarding longevity and of needing nursing home care between waves 1 and 2 and 1 and 3.
