Bizarre multicollinearity (inferred from VIF) in regression

I'm really confused about why I have such high VIFs in a multiple regression I've run. My predictors are country (2 different countries, so -1 and 1), religiosity, and political engagement, and I'm looking at how my three predictors and their various interactions predict an outcome variable. The problem is, I have super high VIF values (20-30), even though when I correlate each of the continuous variables with each other, there's no relationship and when I do t-tests by country on the continuous variables, there's no relationship with country. WHY ARE MY VIF VALUES SO HIGH??? My sample is decent by psych standards (N = 200), so I don't think that merely adding in more data will help. I've tried only using two predictors at a time, and VIF values are still high. All variables except the country dummies are centered. Please help--thank you!!!


TS Contributor
you could try to use one indicator variable for the countries: 1 if it is country A and 0 if it is county B. that would be a cleaner way of coding the countries and you get rid of the multicollinearity as well.