Factor Aliasing

This screen shows the aliasing pattern for the design. This means that all possible effects cannot be estimated independently. The effects that are calculated are actually combinations of two or more effects. The bracketed term on the left is the label for the linear combination of effects shown on the right side of the equals sign.

Here’s what to watch for:

Ideally, main effects are aliased with nothing, or at the most, three-factor or higher interactions. If they are aliased with two-factor interactions (2FI), it will be impossible to know if the main effect or the 2FI is the true effect, it could even be both. Two-factor interactions can be aliased with each other, but this can also cause some confusion when trying to determine which interaction is the true effect.

The more runs you do, the less complicated the aliasing structure, but of course it will cost more. Fractional factorials must be used with a consideration of balancing the cost of runs versus the value of gaining process understanding.