I am looking to examine whether BMI (as a continuous variable) at baseline moderates the relationship between mental health symptoms and changes in BMI at follow up. Could you please suggest the best way to examine the interaction effect in multiple regression analysis?
Thank you in advance for taking the time to look at my question!
Hello Neli,
The basic tactic is to create a product variable (mental health symptoms and baseline BMI), and include it (or them, if symptoms represents a categorical variable that has to be dummy coded) in the model with symptoms and baseline BMI as predictors.
Here's a good starting reference for thinking about the set-up: https://www.spss-tutorials.com/spss-regression-with-moderation-interaction-effect/
As well, Andrew Hayes' PROCESS (available for SAS, SPSS, R) easily handles this and many other related applications: http://processmacro.org/index.html
Good luck with your work.
https://www.researchgate.net/profile/Ludmila_Zajac-Lamparska/post/How_can_I_analyze_baseline_measures_as_predictors_of_change_in_longitudinal_designs/attachment/59d61de779197b807797c2a0/AS:273843497701387@1442300786437/download/Hayes+process.pdf
http://dm.darden.virginia.edu/ResearchMethods/Templates.pdf
Well Neli Pavlova, if you are using Stata as your statistical software and as you have said BMI is continuous and Mental Health Symptoms is categorical (let's suppose) then while fitting the regression model in Stata you can combine these variables by #
For details see to the link below -
https://www.stata.com/meeting/germany12/abstracts/desug12_royston.pdf
- Badges
- Science topic
- Similar topics
- Psychology