Experiments for factors with 3 levels

Fractional factorial experiments can be designed with similar design principles for experiments in which the factors all have three levels.

The details of the design are however different since each 2-factor interaction has four degrees of freedom though the main effects only have two degrees of freedom. As a result, main effects will be only confounded with part of an interaction term.

It is best to rely on statistical software to design this type of experiment.

Factors with mixed levels

The design becomes even more complicated if different factors have different numbers of levels. Again, some main effects are confounded with interactions but an explanation is beyond the scope of this e-book.