Forward Selection (FORWARD)

The forward selection technique begins with just the forced-in covariates and then sequentially adds the effect that most improves the fit. The process terminates when no significant improvement can be obtained by adding any effect. You request this method by specifying SELECTION=FORWARD in the MODEL statement.

If you specify the SELECT=SL method-option, you can use the TEST= method-option to specify a test statistic for gauging improvement in fit. For example, if TEST=LR1, at each step the effect that yields the most significant likelihood ratio statistic is added and the process continues until all effects that are not in the model have LR1 statistics that are not significant at the entry significance level (which is specified in the SLE= option). Because effects can contribute different degrees of freedom to the model, it is necessary to compare the p-values that correspond to these statistics.