Setting Stepwise Regression Options

  To set appropriate stepwise method options:

1. In the Stepwise Options dialog, select R-Squared and Partial F-Test settings.

   Text boxes, settings, and buttons in the Stepwise Options dialog:

   • R-Squared—Stops the stepwise regression if the difference between a specified statistic (either R-Squared or Adjusted R-Squared) for the previous and new regression solution is below a threshold value. When this happens, Predictor does not use the new regression solution. By default, this stopping criterion is selected and uses R-Squared as the statistic. If this setting and the Partial F-Test Significance are selected, the stepwise regression stops when it reaches either criterion’s threshold value.

   • Threshold—Sets the minimum increment required between the R-Squared or Adjusted R-Squared of the last step and the R-Squared or Adjusted R-Squared of the new step to continue with the stepwise regression. The default is 0.001.

   • Partial F-Test Significance—Stops the stepwise regression if the probability of the F statistic for a new solution is above a maximum value. By default, this stopping criterion is not selected. If this setting and the R-Squared setting are selected, the stepwise regression stops when it reaches either criterion’s threshold value.

   • Probability to add—Sets the maximum probability of the correlation (partial F statistic) of the independent variable required to add the variable to the regression equation. The default is 0.05. When dealing with statistical tests, smaller probabilities indicate more significance.

   • Probability to remove—Sets the minimum probability of the correlation (partial F statistic) of the independent variable required to remove the variable from the regression equation. The default is 0.05. This setting is only available with iterative stepwise regression. The Probability to remove setting must be at least 0.05 higher than the Probability to add setting.

2. Click OK.