Forward Stepwise Method
Let p be the number of total effects considered for the model. Let k be the number of effects currently in the model. Let p1, enter and p2, remove be specified by the user. The Forward Stepwise Method starts with no effects in the model and consists of two alternating steps.
- Entry step: For each of the remaining p-k effects not in the model, construct the p-k score statistics. See Testing Global Null Hypothesis, for a description of the score statistic. Enter the effect with the smallest p-value less than p1, enter.
- Removal step: For all the effects in the model compute the Wald statistic. See Testing Global Null Hypothesis, for a description of the score statistic. The effect with the largest p-value greater than p2, is removed from the model.
The procedure continues alternating between both steps until no further effects can either be added or removed or until the maximum number of steps (see the Max. steps option on the Advanced tab of the GLZ General Custom Design dialog) has been reached.