ABSTRACT
The logic of the split-plot design can best be understood by examining the computation sequence. The assumptions for the split-plot design are the same as those for other designs. The design should be used with caution, however, because statisticians warn that it is not as robust as the regular designs as far as breaking assumptions is concerned. It can be seen that the experimental group means increase on passing from condition A to B to C. However, in the control group, the means do not vary much. Thus we would expect there to be an interaction; the ANOVA will indicate whether this is significant. The overall means increase on passing from A to B to C, while the mean for the experimental group is only slightly higher than for the control group. The ANOVA will indicate whether the treatment means are significantly different.
