Step Summary

The initial output contains results for the saturated model. Skip down to the Backward Elimination section. The step summary table presents the results of the model selection.

Step 0. The model generated by the three-way interaction of factors; that is, the saturated model, is considered. This model also includes all lower-order interactions and the main effects. The three-way interaction is tested for significance by deleting it from the model. The change in chi-square from the saturated model to the model without the three-way interaction is tested and found to be not significantly different from 0 (significance value > 0.10). Thus, this interaction term can be dropped from the model.

Step 1. The model generated by all two-way interactions is considered. This model also includes the main effects. Each two-way interaction is tested for significance by deleting it from the model. Since the significance value for the change in chi-square for each effect is less than 0.05, these terms should be kept in the model.

Step 2. Since none of the two-way interactions were removed from the model, there are no more terms to test. Thus, the final model includes the main effects and the two-way interaction terms. This means that Newspaper subscription and Income category both affect the response rate to the mailing.

Next