Start with your outcome variable, then predictor variable. Arguably, however, other views think one should maintain cause-effect relationship (i.e. temporal order). Therefore, staring with your independent variable, followed by the dependent variable.
