There are papers which use only Sobel test and other apply both bootstrapping as well as Sobel.
In which situations Sobel test is preferred over bootstrapping? Or in which situations bootstrapping is not required?
Is it possible to get different results using these two approaches?
Thom S Baguley
Put simply, probably never. The sampling distribution of the indirect effect is skewed and so the normal approximation used in the Sobel test is poor and, in particular, leads to low power. Asymptotically I suppose that it may get as good as bootstrapping. Yes - they can (and often do) give different results.
2 votes
2 thanks
- Badges
- Science topic
- Similar topics
- Papers
More Shweta Jaiswal Thakur's questions See All
Similar questions and discussions