Stepwise Regression

Stepwise regression is an automatic computational procedure that attempts to find the “best” multiple regression model using only statistically significant predictors from a larger set of potential predictive variables. The regression model describes the relationship between a dependent (outcome) variable (Y) and two or more independent (predictor or explanatory) variables (Xj), using a model where $Y=B_0+B_1{X}_1+B_2{X}_2+\dots +B_k{X}_K+e$. Computer programs are used to find the Bj weight for each Xj variable so as to minimize the sum of the squared error (e) for cases used to generate the model. The unique contribution of each Xj variable can be tested with a t test and associated p value, using the null hypothesis that Bj = 0 in the population. Stepwise regression attempts to find [Page 1621]the best regression ...