Design and Analysis of Experiments by Douglas Montgomery: A Supplement for Using JMP

4Randomized Blocks, Latin Squares, and Related Designs

Section 4.2 Creating a Latin Square Design in JMP

Chapter 2 introduced the paired t-test for situations where each experimental unit is subjected to both of two treatments under consideration. The difference of the responses from the two treatments is taken within each experimental unit, and the collection of differences is then analyzed with a standard t-test. This process eliminates the unit-to-unit variability from the experimental error. More generally, the unit-to-unit variation may be thought of as coming from a nuisance factor. Nuisance factors are factors that likely affect the response, but are of no interest to the experimenter per se.

A nuisance factor that is known and controllable ...