More on Factorial Treatment Structure
The extension to more than two factors in a factorial experiment is fairly simple in terms of the calculation of the analysis of variance. There is really only one point of difference between the analysis of an experiment with a complete factorial treatment structure in complete blocks and a confounded design in which each block contains an incomplete set of treatment combinations. Because the analysis is almost the same as that for a factorial treatment structure in complete blocks, the information required for the analysis by statistical software is also in the usual format with columns for the yield, blocks, and each treatment factor. The variation between split plots within main plots can be divided in an analysis of variance into variation between split plot treatments, the variation due to the interaction between split plot treatments and main plot treatments, and a second error variation, the split plot error sum of squares.