You use the UNITEFFECT statement to specify constraints on how the factor levels can change across the runs of the experiment. Such constraints are known as randomization restrictions. UNITEFFECT statements are used in conjunction with a UNITS specification in the BLOCKS statement, where unit factors are defined that index the runs of the experiment.

The uniteffect is an interaction between unitfactors defined in the BLOCKS UNITS=() specification:

unitfactor * …* unitfactor

It defines a partition of the runs on which to apply whole-unit and subunit effects of the factors named in the FACTORS statement.

The wholeuniteffects specified by the WHOLE=() option typically define a necessary feature of how the experiment must be designed, and are thus known as “design constraints.” In contrast, the SUB=() option indicates which unit mean contrasts will be used to compute the subuniteffects and which random error terms will be used to test them, and are thus known as “model constraints.” For both wholeuniteffects and subuniteffects, the effects listed must be enclosed within parentheses, as in the MODEL statement. See the section Specifying Effects in the MODEL Statement for details.

If you have specified units in the BLOCK statement as

blocks units=(WholePlot=4);

then the following statement illustrates how to specify unit effects corresponding to these units:

uniteffect WholePlot / whole=(x1-x3) sub=(x4-x6);

See the section Split-Plot Designs for details about how to use the UNITS= option and the UNITEFFECT statement to construct split-plot designs.