Purpose: This article examines the usefulness of epidemiologic indices in furthering the understanding of the intention-behavior relationship in the field of physical activity.
Design: Six prospective data sets of physical activity were used.
Setting: The United Kingdom and Canada in various settings (school, workplace).
Subjects: Different segments of the population (students, employees).
Measures: Intention at baseline and behavior at follow-up, both assessed by means of questionnaires.
Analysis: Intention and behavior were dichotomized to create a 2 x 2 table; this allowed us to compute four standard epidemiologic indices: sensitivity, specificity, positive predictive value (PV+), and negative predictive value (PV-).
Results: Sensitivity was 86.3%, which reflected the high sensitivity of intention for exercising, i.e., active individuals were very likely to hold a positive intention. Specificity was 49.5%, which suggested that a significant number of inactive individuals held a positive intention. With respect to predictive values, a low intention was a very good predictor of being inactive (PV- = 88.1%), whereas a positive intention was a moderate predictor of being active (PV+ = 45.5%).
Conclusion: These results indicate that intention is a moderate predictor of behavior and that the gap between intention and behavior is caused by high intenders not taking action. Health promotion programs would benefit to target factors that moderate the intention-behavior relationship.