A foldover on all factors produces a second block of experiments in which the levels of each factor are reversed from the original block. A resolution III design, will improve to a resolution IV design. The main effects will be cleanly estimated. Do not use an all-factor foldover for a resolution IV design; no new terms will be de-aliased.
A foldover on a single factor is used for a resolution IV design. Turn off all but one factor from the list that is part of an aliased interaction. This will produce a second block where only the selected factor’s levels will be reversed. All aliased interactions involving that factor will be de-aliased.
A semi-foldover is similar to a one factor foldover but another factor is chosen to remain constant. The secondary factor should be one that is not part of a significant main effect or interaction. Select the level (low or high) of this factor that will not change in the foldover block. The resulting foldover block will be half the size of the first block. This is the default augment for a resolution IV design.
Both of these options can be used from the Design Layout screen by choosing Design Tools, Augment Design.
These methods will produce some possibly unexpected results if any of the runs contain botched (altered) levels. The foldover will give the opposite level as in the original runs, so for instance if a coded level is +.73 (rather than +1), the foldover will provide a coded level of -.73. The semifold procedure will ignore rows where the subset factor has botched levels, resulting in fewer added runs.