Currently running through some mediation analyses using R with package 'mediation'. So I am running into some results that I am struggling to explain. We love to hear what everyone else thinks.
1) Average causal mediation effect and average direct effects are significant but total effect is not statistically significant. Does this mean that the independent variable is affecting the mediator in the opposite direction to it's direct affect. But this doesn't seem to make any sense because when I run correlations between IV and mediator and DV, it seems to imply that worse IV = Increase mediator = worse DV, so shouldn't the total effect be additive and not counteracting one another?
Looking something like this
Causal Mediation Analysis
Quasi-Bayesian Confidence Intervals
Estimate 95% CI Lower 95% CI Upper p-value
ACME 5.80e-04 2.45e-05 0.00 0.04 *
ADE -3.13e-03 -5.83e-03 0.00 0.04 *
Total Effect -2.55e-03 -5.43e-03 0.00 0.09
Prop. Mediated -1.90e-01 -1.87e+00 0.98 0.13
Signif. codes: 0 ‘’ 0.001 ‘’ 0.01 ‘’ 0.05 ‘.’ 0.1 ‘ ’ 1
Sample Size Used: 142
2) When I add more mediators into the model, the direct effect appears to decrease but the average mediation effect seems to improve slightly. I'm not too sure what this implies. Thanks guys!