Engagement in activity has been central to theoretical discussions and empirical analysis of well-being in later life. Despite its important theoretical role, there has not been very much attention paid to the measurement and use of multiple activity variables in statistical models. Determining a parsimonious means of investigating multiple activities is important to understanding engagement of the "whole person." The aims of this study are: 1) to explore ways to conceptually and empirically consolidate discrete activity measures into testable activity domains and engagement patterns and 2) to explore the performance of the derived activity domains and engagement patterns as intermediate variables in the study of antecedents and outcomes of activities. An Expert Panel will assist in determining optimal measures, analytical methods, and data set(s) to support our team's future research and in making recommendations to the field about use of the activity items in these specific data sets. We will accomplish this exploratory work using three national data sets that each includes discrete activity measures (Health and Retirement Study, Americans'Changing Lives, and Midlife in the US). We repeat one set of methods on each data set to explore possible approaches. We use a variable-centered approach (factor analysis) and a person-centered approach (latent class analyses) to consolidate activity measures. In each of the three data sets, we will identify potential antecedents and outcomes of activities and explore how the measurement approaches work in statistical models. We will use structural equation modeling and general mixture modeling to test factors associated with the derived measures of activity domains and activity patterns, as well as the relationships of domains and patterns with well-being outcomes. Products of the proposed work include: 1) an identified set of activity items that empirically factor together in activity domains in MIDUS, HRS, and ACL data sets;2) an identified set of engagement patterns that exist in each of these data sets;and 3) knowledge about how activity domains and engagement patterns operate in statistical models specifying antecedents and outcomes of activity. These products will be used in the research team's subsequent work. Dr. Putnam will use this development work to study activity engagement of individuals aging with disability, and Dr. Morrow-Howell will use it to study the unique role of productive activities and well-being outcomes. Also, these products will be shared with other researchers through peer-reviewed articles and conference presentations. Thus, our findings will benefit other researchers who use the MIDUS, HRS, and ACL, all of which are funded in part by the NIA, are publicly available, are widely used, and include large, representative samples of older adults. Ultimately, this work will advance the study of activities and well-being outcomes in later life in a more holistic and systematic way.
Improvement in measurement and analytic approaches to the study of activity engagement is critical to knowledge development about the patterns of activities that may be most beneficial for adults in later life. The development work of this R21 is necessary for a more systematic use of NIA-supported public data sets in the study of the antecedents and outcomes of activities in later life and development of interventions to promote active aging. The aim of this research is to promote active aging, which is consistent with initiatives by the World Health Organization, the Centers for Disease Control and Prevention, the Administration on Aging, and NIA's newly released RFA-AG-11-003 (Subjective Well-being: Advances in Measurement and Applications in Aging).
|Morrow-Howell, Nancy; Putnam, Michelle; Lee, Yung Soo et al. (2014) An investigation of activity profiles of older adults. J Gerontol B Psychol Sci Soc Sci 69:809-21|