Figure 1.

Mediation analysis for the relationship between coffee consumption and meeting physical activity guidelines for (a) 1–2 cups of coffee/day and (b) >3 cups of coffee/day, with correlation coefficients before (τ) and after (τ’) including the mediator in the model. Mediator effect: (τ − τ’)/τ = % difference in the regression coefficient for the independent variable (coffee intake) on the dependent variable (physical activity) between the models. p-value model change = tests the hypothesis that the direct effect of IV on DV is statistically significant after adjusting for the mediator; p-value αβ effect = tests the hypothesis that the indirect effect of the mediator on the IV–DV relationship is significantly different from zero.