Interaction effect

the effect of one factor depends on another

Effect(AB) = ȳ(AB = +) − ȳ(AB = −)

Two factors interact when the effect of one changes with the setting of the other. On an interaction plot the lines are not parallel; crossing lines mean a strong interaction.

Interactions are the main reason one-factor-at-a-time trials give wrong answers, and the main reason to use a factorial design.

Example

Higher holding pressure cuts shrinkage by 0.45 % at 210 °C but only by 0.08 % at 240 °C: pressure and melt temperature interact.

Common mistake

Picking the best level of each factor from the main effects plot alone when a significant interaction exists.

Calculate it

DOE: 2-Level Factorial Design: 2^k factorial: run plan, effects, Pareto, interaction plots and best settings.

Related terms

• Main effect: average change when one factor goes from low to high
• Design of experiments (DOE): planned trials that change several factors at once