Practical Data Analysis with JMP®

Multi-factor Experiments and Factorial Designs

It might be natural to think that one should experiment by holding all factors constant except for one, and systematically alter the values or levels of that single factor. So, for example, we might want to test varying amounts of sand in concrete mixtures with all other components being held at given quantities. The shortcoming of such an approach is that it can conceal an important combined influence of two or more factors. Perhaps the sand quantity is not nearly as important as the ratio of sand to coarser aggregate. When two variables combine to lead to a result, we say that there is an interaction between the two, and if, in reality, there is an important interaction to be found, we'll never ...