I am looking for direction on testing simple slopes for a 3-way interaction in a binary logistic regression model. The three predictors are continous varables and the outcome variable is dichotomous. This is multilevel data, if that makes a difference.
I would do 3 things:
1. create an interaction plot and look for lack of parallelism.
2. look at the p-value for the 3-way interaction term in the model
3. Check the assumptions.
I don't know if the multilevel character makes a difference or not. My favorite book in this area is:
https://www.amazon.com/Analysis-Regression-Multilevel-Hierarchical-Models/dp/052168689X/ref=sr_1_1?s=books&ie=UTF8&qid=1532129939&sr=1-1&keywords=Andrew+Gelman+Regression
I have attached some notes that might also be helpful.
Best wishes, David Booth
Hi Tess I am providing links and Reviews for your help:
https://www.youtube.com/watch?v=TcAudtrHTKE
https://www.youtube.com/watch?v=SUYr-FYGRNw
https://www.youtube.com/watch?v=sj2XWfZaxek
http://web.pdx.edu/~newsomj/cdaclass/ho_interactions.pdf
https://stats.idre.ucla.edu/stata/seminars/deciphering-interactions-in-logistic-regression/
https://stats.idre.ucla.edu/sas/seminars/analyzing-and-visualizing-interactions/
I hope it would help you Regards
- Badges
- Science topic