Question 548:

Asked on December 10, 2008

Tags: Regression Analysis , Stepwise Regression

1

Answer:

No answer provided yet.A stepwise regression is a method in which the software finds the best possible combination of predictors to maximize the Adjusted R² (or fit of the model). The goal is to maximize the amount of variation explained with as few predictors as possible (called model parsimony). In the above output we see that all predictors have an adjusted R² of .8013 or explaining around 80.13 percent of the variance in BirthRate.
 
We do see however that the p-values for LifeExpectency and Density are both above .05, meaning they are not significantly contributing to the model. As we step-backwards in the output, we see that as soon as we exclude these two variables (Nvar step 3) and have only Literate, GDPCap and InfMortality, we have achieved a higher adjusted R² with only 3 independent variables. Given this output going with those variables would provide the best model to explain birthrates.