A model-building method in multiple regression that first enters into the regression equation the variable with the highest correlation, followed by the other variables one at a time that increase the multiple R by the greatest amount, until all statistically significant variables are included in the equation.