Comparison of exercise models in an elderly population

Background and aims: Two different models of exercise behavior in an older population have been previously published (Schuster et al., 1995 and Resnick et al., 2000). We aimed to validate the models with data from a different sample, and to examine whether any of these models explained a greater proportion of the variance in exercise behavior in this population. Methods: Participants were 324 community-dwelling persons aged 74-85 years who completed a mailed questionnaire. The questionnaire included questions about demographics, current exercise habits, health, activities of daily living, depressed affect, living arrangement, attitudes about exercise, self-efficacy, barriers to exercise, and enjoyment of exercise. Results: While our data explained some of the variance of the two previously published models of exercise behavior, the percentage of variance explained was less impressive than that reported by those authors. A path analysis of demographic and psychological constructs, which have been used in both models, demonstrated that physical health, mental health, gender, and past enjoyment of exercise affect self-efficacy, attitudes/outcome expectancy, and reported barriers to exercise. These psychological constructs in turn affect exercise habits. Conclusions: The relatively low percentage of variance explained by the models may imply that other factors contribute to exercise behavior in older adults, that the constructs are very sensitive to measurement differences, or that these factors are not as important as previously thought. Our results are consistent with the models of both Schuster et al. and Resnick et al., but suggest that a combined model may better represent the data. Future investigations of exercise behavior in older persons should examine this possibility.