ABSTRACT
In this chapter we present a method for obtaining an approximate analysis of balancedtreatment-structure experiments that have an unequal number of observations in each cell.
We generally recommend using a general computing package such as one of those discussed in Section 10.7 to analyze unbalanced data sets in which every treatment combination is observed at least once. However, many situations arise in practice where it may not be feasible to do so, especially in developing countries where one may have several treatment factors with each factor having several different levels. In these cases, the exact procedures require several matrix inversions, and the size of the matrices to be inverted may exceed the capabilities of both the researcher and the computer that is available for use.
