I'm using multiple linear regression to measure both the main effects of SES (income, household size, education level) on a kindness scale, adding an interaction between income (rating between 1-7 depending on amount made per year) and household size (# of people in household). Scale is continuous for all variables. More specifically, both seem to be interval scale.

Should I mean center household size and education level before I create the interaction term. If yes, should I mean center all predictors (as well) included in the the model.

  • $\begingroup$ Do you mean something other than "menter" in your title? (perhaps mean?) $\endgroup$
    – Glen_b
    Nov 16, 2021 at 21:19
  • $\begingroup$ Thats weird it said mean to start with, I think $\endgroup$ Nov 17, 2021 at 4:19
  • $\begingroup$ Why rate income on a 1-7 scale instead of using their income? // I think I know what you mean when you say that your variables are continuous (numerical rather than categorical), but be careful with that. Households don’t usually have $2.5$ people. $\endgroup$
    – Dave
    Nov 17, 2021 at 4:35


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.