Comparative validity of measures of implicit exercise associations
Researchers in the broad domain of public health are becoming increasingly interested in the study of implicit exercise associations (IEA), which are theorized to be predictive of exercise behavior. Interventions designed to improve IEA may enhance exercise promotion, and accounting for differences in IEA may help explain exercise behavior. Despite increasing interest in IEA, researchers are unable to justify their measurement choices because the comparative validity is unknown. No investigation has compared the validity of measures of IEA within the same sample. Purpose: The primary purpose of this dissertation was to compare the validity and reliability of nine measures of IEA for the first time. The secondary purpose was to explore if rational thought-processing styles moderate the relation between IEA and exercise behavior or explicit attitudes. Methods: University community members (N = 95) completed three laboratory visits. First, participants underwent exercise testing to identify their ventilatory threshold (VT). One week later, the participants returned for a session that consisted of 15 minutes on a recumbent cycle ergometer with the last 10 minutes at the intensity corresponding to their VT, in an effort to create maximum variability in affective responses. Experienced pleasure-displeasure during exercise was measured using the Feeling Scale (Hardy & Rejeski, 1989). Remembered pleasure of the exercise session (“recalled affect”) was measured 5 min postexercise using the Empirical Valence Scale (Lishner, Cooter, & Zald, 2008). One week later, participants returned for an Implicit Measurement Session. Participants completed nine measures of IEA in random order. Participants also responded to questions about their explicit affective attitudes and exercise behavior, and completed tasks to assess their rational-thought processing styles. Experienced pleasure-displeasure, recalled affect, affective attitudes, and exercise behavior served as validation criteria. Validity was assessed with a series of bivariate correlation analyses between each implicit measure and the validity criteria.
Results: Eight tasks had acceptable-to-high internal consistency. Only the Approach-Avoidance Task was significantly related to any of the validity criteria after adjustment for multiple comparisons. Exploratory analyses revealed the possibility that validity estimates deteriorated as tasks were performed later, meaning that the most valid measure could not be identified conclusively. Implications and recommendations based on these results are discussed. Moderation analyses indicated that no measure of rational thought-processing capacity or tendency moderated (a) the relation between IEA and exercise behavior, or (b) the relation between IEA and explicit attitudes. Conclusions: Based on the entire sample, these data suggest the Approach-Avoidance Task is the most valid measure of IEA. This must be interpreted judiciously, however, due to the unexpected importance of task order. A research agenda to improve upon the limitations of this dissertation is proposed.