Second, the cross-sectional design restricted the researchers from testing a predictive relationship between intention and actual behavior as the TpB initially suggested; actual behavior was not included in the hypothesized models, although we examined a bivariate relationship between intention and actual behavior. As a result, the relationships between intention and actual behavior in the models are still questionable among this population. A longitudinal study with all TpB variables should be considered before applying the TpB to designing an intervention study.