Table 4 shows the associations between each of the temperament factors with the 2 alcohol factors, as well as whether that effect was mediated by the intervening variables (total indirect effect), and whether a direct (residual) effect of the temperament factor on the alcohol factor remained after accounting for the mediation effects (remaining direct effect). The proportion of the total effect that was mediated by the intervening variables is also reported for significant effects. The total indirect effect is an omnibus test for mediation, that is, it provides an overall test of mediation taking into account all the mediating variables in the model, not specific to any one mediator. The specific variables for which there was mediation when the overall indirect effect was significant are shown in the corresponding figures, along with the path coefficients for significantly mediated pathways. The p-value indicated inside the intervening variable boxes indicates the p-value associated with the mediation effect through that variable for significantly mediated effects. Note that the model was fit in a single step (separately for males and females), but the results for each pair of temperament–alcohol factor associations are broken out into a series of figures to make them interpretable.
