Dear Statalist members,

If I regress y on x1 x2 x3, is there a way I can use the estimated values to predict y hat based on a different set of x variables (let's say x1 x4 x3 where I've substituted the values of x2 with x4).