Health literacy, exercise self-efficacy, and physical activity in university students: a cross-sectional study

Background Evidence of the association between health literacy and physical activity is inconsistent. Health literacy may support healthy behavior, but social cognitive theory suggests that acting on health information also depends on confidence to initiate and persist despite barriers. Exercise self-efficacy may therefore help explain when health literacy is associated with greater physical activity. We tested whether adding exercise self-efficacy improved models for weekly physical activity and estimated the cross-sectional indirect association between health literacy and physical activity via exercise self-efficacy. Methods We conducted a cross-sectional study of 592 students at a large Midwestern public university in 2025 (median age = 18 years; 70.8% female). We assessed weekly physical activity using the International Physical Activity Questionnaire Short Form, alongside perceived digital and general health literacy, performance-based functional health literacy, and exercise self-efficacy. We compared a health literacy-centered model and a self-efficacy-extended model using structural equation modeling, 5,000-resample bootstrap confidence intervals, and repeated 10-fold cross-validation with five repeats using held-out predictive R 2 . Results Adding exercise self-efficacy improved total-activity models across all three health literacy measures (bootstrapped Δ R 2 = 0.112–0.128; mean pooled out-of-fold predictive Δ R 2 = 0.110–0.127). Estimation of standardized indirect associations via exercise self-efficacy supported digital health literacy ( β = 0.051, 95% CI = 0.017 to 0.090) and general health literacy ( β = 0.067, 95% CI = 0.031 to 0.107), but not functional health literacy ( β = 0.002, 95% CI = −0.029 to 0.032). Exploratory analyses by activity intensity provided evidence of associations for vigorous and moderate activity, whereas estimates for walking did not survive false discovery rate control. Sensitivity analyses using an alternate specification of total activity yielded the same qualitative pattern. Conclusion Exercise self-efficacy improved models of physical activity across all three health literacy measures, while evidence supported indirect associations for perceived digital and general health literacy but not for performance-based functional health literacy. Exploratory analyses provided evidence of associations for moderate and vigorous activity, but the cross-sectional design precludes causal or temporal interpretation of the indirect associations.