I have a basic doubt on when to use country code as dummy coded control variables in regression and when to model it as a higher level variable within which other independent variables are nested. In my specific study, I have borrowers from several countries using an online P2P portal for seeking finance. I am interested in understanding how various borrowers' characteristics (e.g., age, education, work experience) predict their success at raising finance, beyond their country of origin. Should I use country as dummy codes in the linear model, or use a mixed effects model with borrowers nested in countries? What is the difference? What are the pros/cons of the two approaches. Much appreciate your help.