This research seeks to elucidate significant factors involved in the process of caring for a dementia patient over time. The stress of care giving may be lessened with more clarification of the process. Therefore, this longitudinal study will examine the relation between the components of controllability, social support and health, and how they each predict care giver happiness and depression over four years. Controllability refers to individuals' beliefs that they can control a situation, in contrast to their beliefs that the situation is controlled by factors outside themselves. Two models of controllability will be compared: Rotter's internal-external concept and Rothman's two-process model. Social support will be measured as both the amount and perception of support (overall satisfaction, helpfulness and upset). An additional objective of this study is to encourage the entry of minority students into research on the psychosocial factors of aging. Student involvement in each phase of the project will provide an expanded opportunity for training future social science researchers. Participants will be 50 care giving spouses and 50 non care giving spouses. They will be recruited through community organizations and will be interviewed in their homes. After signing a consent form, they will complete measures of controllability, social support, health, happiness, depression. Three regression analyses will be conducted. One multiple regression analysis will assess the relation between controllability and social support, and another will assess the relation between controllability, support, and health. Finally, a regression analyses will determine the amount of variance attributable to controllability, social support, and health in predicting happiness and depression. Longitudinal analyses will also utilize multiple regression analyses, adding past scores to the equations.