Deciphering the Output of Anova Programs for Unequal-Subclass-Numbers Data Using Benchmark Data Sets
S. R. Searle · 1979
A series of small, hypothetical data sets of increasing complexity regarding characteristics such as numbers of observations and empty cells are being used to ascertain the precise nature of output values (which are often not labeled unequivocally) from computer routines designed for analysis of variance of data having unequal numbers of observations in the cells. All possible analyses and calculations are known for the data sets and comparison with those of the computer output values reveals what those values exactly are. The different computing features of a variety of routines are easily identified by this means.