Proportion mediated

[jeremyeorr]

I am performing a mediation analysis with multiple mediators. I am using specific_effect_significance to extract the mediation effect for each mediator, with the bootstrapped CI. I want to determine the proportion mediated by each mediator, which I can get by dividing by total_effect from the model summary, but would also like the confidence interval around this proportion. I could divide the CI from specific_effect_significance by the total_effect but suspect this is not methodologically sound. Any suggestions?