8. Boolean plans

It is not always possible to exploit the results of a design of experiments using conventional tools: multilinear regression for continuous variables or analysis of variance for discrete variables. This is particularly true for Boolean variables. If the study factors are Boolean variables, i.e. variables taking only two values, we need to find the Boolean function, itself taking only two values, as a function of the values of the variables.

Boolean designs resemble two-level factorial designs, but the mathematical interpretation is completely different, since the nature of the variables is different. Each trial yields the value of a minterm (or maxterm). The set of all...