I am determining the association between physical activity and fasting glucose levels amongst South Asians and Europeans using Multivariate linear regression. I want to see the possible intearction between ethniicty (covariate) and physical activity (main independent variable under study). How can I do it in SPSS please?
If I add the product of ethnicity*physical actvity as a covariate, do I add the ethnicity and physcial activity as predictors in the same model too. That is, my predictors list will have : Physical activity, age, gender, ethnicity and ethnicity*physical activity?
regards
saima
It sounds like you have only one dependent variable, so why use multivariate regression; why not just do univariate OLS regression (general linear model)? Testing an interaction is straightforward, you just add the interaction term in along with the other IVs in the model.
Yes calling a multiple (or multivariable) linear regression a multivariate regression is a common mistake seen in many articles. Multivariate regression is something completely different.
When you introduce an interaction term into a model, you should also introduce those terms that form the interaction term. So ethnicity and physical activity alone should be in the model as well.
Thanks Stephen and Mehmet
@Mehmet - you made it so clear to me :). Extremely grateful.
Regards
Saima
- Badges
- Science topic
- Similar topics
- Mathematical Sciences
- Statistics
- Statistical Software
- SPSS