search

kosukeimai / mediation

R package mediation
58 stars 29 forks source link

How much variation should we expect in the CIs based on nonparametric bootstrapping? #66

Closed adcascone closed 5 months ago

adcascone commented 5 months ago

Hello,

I'm working on recreating the outputs in the 'mediation' vignette (i.e., not the vignette labeled "mediation-old"). Specifically, I ran the code included on pg. 22, attached in this screenshot: image

The estimates I've derived match the vignette's outputs, but there is some variation in the upper and lower bounds of the confidence intervals. My output is:

image

What degree of variability should we expect in the CI outputs when running mediate.sed() (and other functions in the mediation package that return CI intervals)?

Thanks, Arianna

kosukeimai commented 5 months ago

increasing the number of simulations will stabilize these estimates.