Hi, I need help deciding which option is best for using mediation model in my thesis
Since all my variables are not normally distributed
Is it enough to report the indirect effects (see picture 1) stating that the indirect effect is significant
or do I need to use the Bootstrap inference for model coefficient option (see picture 2)
Thank you so much,
Keren
Keren,
Process doesn’t assume normality in the distribution of values if your variables. That is why it is a powerful technique. It uses a resampling process called bootstrapping. If zero does not lie within your lower and upper values, then your results are significant.
There are many combinations that are possible and they are contained in the textbook by Andrew Hayes.
If your data are skewed you should have a look at the shape of the distributions and consider transformation and/or removal of outliers. There are plenty of good guides around to doing this. Seeing as you are using SPSS, have a looks at the how2stats channel on YouTube.
- Badges
- Science topic
- Similar topics
- Methodology
- Research Methodology
- Research Papers