If independent variable is group level and mediator is individual level and outcome variable is also individual level, I want to ask that the basis concept of mediation is same statistically as given by baron and kenny 1986 (4 steps), or is there any difference for multilevel of analysis?
You can also check Sobel's test (1982), and Preachers and Hayes (2004, 2008), for details.
Generally Sobel and Baron/Kenny approaches are discouraged these days (Including by Baron and Kenny).
Look up work by Preacher and Kelly, Hayes, MacKinnon and others.
Is it possible to analyse a 2-2-1-Model (it means Level 2 Predictor and Mediator and dependent variable as Level 1) with the framework: http://www.page-gould.com/r/indirectmlm/
The (free and online) web application CaviR provides a function to perform (multilevel) mediation analyses up to two levels:
1. Go to www.cavir-statistics.com
2. Upload your datafile
3. Go to 'Mediation'
4. Select predictor, mediator, outcome, covariates, levels.
5. You get your table including all results and the triangle figure which you can adapt to your own preferences.
Enjoy!
Joachim
Take a look at the following:
https://www.youtube.com/watch?v=bAzOekt39fQ
- Badges
- Science method