I got this comment from a respected reviewer for one of my paper submitted for publication since I used Baron and Kenny's approach for mediation test.
I just need to defense and justify the Baron and Kenny's approach and I find it late now to use Bootstrapping approach now, what shall I answer him or her??
Plz advice and help & Many thanks
"One of the major limitations (as critiqued by scholars) of Baron and Kenny’s mediation is the issue of normality of the indirect effect (the same is accused of Sobel test too) so it’s better to carry out bootstrap (which is a resampling technique with replacement) along with baron and Kenny, so that the results will get more credibility".
YOU CAN ALWAYS GO FOR BOOTSTRAPPING. THE REVIEWER SUGGESTED 'ALONG WITH BARON AND KENNY'. IT WON'T TAKE MUCH OF YOUR TIME AND EFFORT. THE BOOTSTRAPPED PROCEDURE WOULD GIVE YOU THE CONFIDENCE INTERVAL WHICH NEEDS TO BE INTERPRETED ALONG WITH THE OTHER RESULTS FOR MEDIATION.
Sobel's test involves calculating the significance (i think using the direct sample values of effects) manually, however in contrast, the bootstrapping option calculates the p-value/significance value by creating the random samples with replacement that is more valid (because of sample effects and std deviations (and thus p-value/significance) averages). If you have used the PLS-SEM (SmartPLS) then the bootstrapping options are available without manual calculations (the same is the case with CB-SEM i think so).
Besides, I would suggest you to use the Preaches and Hayes method of testing the mediation particularly if you are using the CB-SEM/PLS-SEM. The method mentioned is more recent than the Baron and Kenny one.
I agree to the reviewer and will suggest you to please consider using bootstrapping. The technique is very simple to use. Please follow James Gaskin he is a great guide in such situations
http://statwiki.kolobkreations.com/index.php?title=Structural_Equation_Modeling
Hope you will find it helpful..
- Badges
- Science topic
- Similar topics
- Linguistics
- Composition Studies
- Publication