Experimental Methods

Experimental methods define how you structure your designed experiment. The method you choose determines how efficiently you learn, how clearly you can interpret results, and how well the experiment supports improvement decisions. In the Improve phase, selecting the right method ensures that your experiment delivers high‑quality, actionable insight. 

The most common experimental methods include full factorial, fractional factorial, response surface, randomized block, and one‑factor‑at‑a‑time (OFAT) approaches. OFAT is generally discouraged because it cannot detect interactions and requires many more runs. DOE methods, by contrast, vary multiple factors simultaneously, providing richer insight with fewer runs. 

Full factorial designs test all combinations of factor levels. They provide the most complete information and allow you to estimate main effects, interactions, and curvature (with additional levels). They are ideal when the number of factors is small and when you need a comprehensive understanding of the process. 

Fractional factorial designs test only a subset of combinations. They are efficient for screening many factors but introduce confounding, which blends certain effects together. These designs are ideal when you need to identify key drivers quickly. 

Response surface methods—such as central composite and Box‑Behnken designs—are used for optimization. They allow you to model curvature and identify optimal settings. 

Randomized block designs control for nuisance variables by grouping similar experimental units together. This reduces noise and increases the precision of effect estimates. 

In the Improve phase, selecting the right experimental method ensures that your experiment is efficient, interpretable, and aligned with your objectives. It is a strategic decision that shapes the quality and usefulness of your results.