A Simple Method for Variable Selection in Regression with Respect to Treatment Selection