|
|
| Message/Author |
|
|
|
|
Hi,
I have a path analysis I've done using complex survey data, FIML, a dichotomous IV, a continuous M, and a dichotomous DV. But I have two questions about it.
1) I was able to get betas and their 95%CIs for the a-path (the linear regression) main effects. I was also able to get odds ratios and their 95%CIs for the b-path (the logistic regression) main effects. However, when I look at the decomposition of effects (the direct, indirect and total effects), I am only seeing log odds and the related 95%CIs of the log odds. Is there any way to get odds ratios and their 95%CIs for the direct, indirect and total effects? I can calculate the OR from the log odds but I don't know what to do about the 95%CI. What do you recommend?
2) Because my mediator is continuous and my dependent variable is dichotomous, and they are therefore on different scales, I know Mplus is not using the Preacher and Hayes (a*b) method to calculate the mediation effects. But what is it using? I've read about a counterfactual, but I don't think I fully understand what that is and if it is what Mplus is using to appropriately calculate the mediated effect. Any clarity you can provide on this would be greatly appreciated.
Thank you!
Allison |
|
|
|
|
1) You can compute the odds ration CIs using the approach of the FAQ on our web site:
Odds ratio confidence interval from logOR estimate and SE
2)See our Mediation page
http://www.statmodel.com/Mediation.shtml
where you will find a 2015 article with me as first author - that's how we do it and that's the appropriate indirect effect. |
|
|
|
|
| This is very helpful. Thank you so much for your quick response! |
|
|
| Back to top |
|
|
