二维方差分析中，每组只有一个数据，pure error为零的情况下，如何计算F和P值呢？文献中说用别的方法可以算，但是不会，高手看看

As there is only an observation per cell it is not possible to evaluate the pure error and the technique given by Tukey [34] has been
applied.Computational details can be also consulted in page 45 and next from Anderson and McLean [35] book. With this approach, the mean square estimated for the interaction with one degree of freedom is 8947.92.



[35] V.L. Anderson, R.A. MacLean, Design of Experiments. A Realistic Approach, Marcel Dekker, New York, 1974.