2. This case uses a manual backward model reduction approach, in which the full model (i.e., the one using all of the predictor variables) is reduced one predictor at a time based on statistical significance. Let's consider a different approach - forward model selection. Begin with a one-predictor model, perhaps based on the predictor with the highest correlation with Price or the lowest simple regression p-value, and add new predictors one at a time. As you add predictors, examine the results under Summary of Fit, Analysis of Variance and the Parameter Estimates table