Evaluating social-cognitive measures of motivation in a longitudinal study of people completing New Year's resolutions to exercise

Abstract

Overweight and obesity are major problems in developed countries around the world, and exercise is strongly associated with weight loss and weight control. Social‐cognitive models have been used to study self‐regulation of health behaviors, but they suffer from several limitations, including the use of intentions as a sole measure of motivation. There are validated motivational constructs from literature outside the health domain that could potentially combine to form a latent variable of motivation that would benefit social‐cognitive models of self‐regulation. It would provide a more accurate assessment of motivation and provide direct avenues of intervention that have yet to be fully explored in health research. On an adjusted longitudinal sample of 713 participants attempting to increase their vigorous and/or moderate exercise as part of a New Year’s Resolution, the present study had four aims. First, it utilized a randomized controlled design to test a novel intervention designed to increase exercise behavior based on the motivational concept goal‐structure. Second, it evaluated the Chulef, Read, and Walsh (2001) goal taxonomy for assessing participant goal‐structure in regards to exercise. Third, it evaluated whether three motivational constructs—goal autonomy, goal commitment, and goal‐structure—form a robust latent variable. Finally, it tested whether the latent variable of motivation predicted exercise behavior over and above intentions. There were three time‐points of data collection that started in January and occurred at 4‐week intervals. Intervention effects immediately post intervention were tested via independent samples t‐tests, and changes over time were tested via multi‐group latent change score models. The intervention successfully impacted participant goal‐structure post‐intervention, however these effects failed to spill over into other constructs. Additionally, both Intervention and Control groups significantly increased their physical activity throughout the course of the study, as well as their perceptions of their intentions for exercise, goal commitment, goal autonomy, and a few goal‐structure scores, however there were few significant differences observed across groups. The utility of the Chulef, Read, and Walsh (2001) goal taxonomy for measuring goal structure in relation to exercise was tested using correlations, regressions, and exploratory factor analyses on five sub‐samples of data. These analyses revealed that two goal‐structure scores—(1) the degree exercise facilitates achieving other life goals and (2) the rank of a heath goal‐cluster in comparison to other goal‐clusters—stood out as potential useful additions to a latent variable measuring motivation. The robustness of the motivation latent variable derived from the motivation constructs was evaluated using longitudinal measurement invariance modeling testing for stable latent variable factor loadings, indicator means, and error variances over time and across gender and study group. The tests of measurement invariance showed that a single latent variable comprised of four others (goal commitment, goal autonomy, facilitation of exercise, and the rank of the heath goal-cluster) exhibited a good fit to the data across three time‐points of measurement. Finally, a latent variable path analysis was used to test if the motivation latent variable predicted exercise behavior over and above intentions. After adjustment for missing data, the motivation latent variable at Time 1 predicted vigorous exercise at Time 2, vigorous/moderate exercise at Time 2, and success in accomplishing vigorous/moderate exercise intentions at Time 2, and the motivation latent variable at Time 2 significantly predicted vigorous exercise at Time 3, vigorous/moderate exercise at Time 3, and success in accomplishing vigorous/moderate exercise intentions at Time 3. Although the intervention failed to create significant differences in exercise across study groups, the findings support using at least some of the motivation constructs within the traditional social‐cognitive frameworks. Future research should continue exploring intervention options based on the motivation constructs, update the method for measuring goal‐structure using the Chulef, Read, and Walsh (2001) goal taxonomy, and continue testing the relationships under study.