Hello RG Community,
I need advice about testing moderation and mediation hypotheses with variables measured at different levels. We have data from 18 countries, and I was curious if I can test the following hypotheses.
Mediation hypothesis: Country-level gender equality leads to higher perpetration of violance against women (measured at individual level) through higher opposition to female sexual freedom (indvidual level).
Moderation hypothesis: Higher opposition to female sexual freedom (measured at the indvidual-level) would relate to higher perpetration of violance against women (indvidual level), especially in higher gender progressive countries (country-level).
I would appreciate your advice on this.
Side note: I don't have any experience with multi-level analysis of large cross-cultural data, but I really would like to learn it. Would you be able to suggest me a good paper/book on this? And is it easier to run it in R or MPlus?
Thank you very much in advance for your time and advice.
Best,
Pelin
Pelin Gul, if I understand your questions correctly, first, your mediation hypothesis can be tested using multilevel regression. It is generally easier to use Mplus than R (i.e. packages) but the latter is free software. Another point is that you less often encounter convergence issues in Mplus. Here are resources for R and Mplus respectively: https://lavaan.ugent.be/tutorial/multilevel.html and https://www.statmodel.com/Mediation.shtml.
When it comes to your second question, you could consider creating a dummy variable (highVSlow.Pathogen) and interact this variable with the homophobic disgust variable and estimate your model using ordinary regression. This analysis can be done using any general statistical software.
Mehmet Mehmetoglu , thank you for your answer! Glad to know that I can test this mediation hypothesis, although I have only 18 countries (i.e., 18 scores on gender inequality). I will give it a try with Mplus as I am familiar with it already, but I have an old version of it.
For the moderation hypothesis, do I not need a multi-level analysis approach to this too?
Thanks
- Badges
- Science topic
- Similar topics
- Mathematics
- Statistics
- Statistical Software
- R