Factor-Covariate Combinations

Abstract

The examples in the previous chapters all revolve around the effects of one (Chaps. 2 and 5– 10) or two or three (Chap. 11) categorical treatment factors on a categorical or quantitative outcome variable. However, in not so few cases, we have at least one other, non-treatment variable measured along with our primary outcome variable of interest. How to deal with this additional variable depends on whether it is measured before or after the start of treatment and how it relates to the primary outcome variable of interest. When this additional quantitative variable is measured together with the primary outcome variable, we may in some cases treat this additional variable as a second outcome variable that is correlated with our primary outcome variable. When this additional variable is measured after the start of treatment but before our primary outcome variable, the additional variable may mediate treatment effects of interest. Finally, when the additional variable is measured before the start of the treatment, we may under conditions outlined in this chapter include the additional variable as a covariate. Both covariates and mediators may moderate a treatment effect of interest.