Plackett-Burman Designs, Explained
Part of our complete guide to DOE for biologists.
A Plackett-Burman design is a specific, very efficient class of screening design — a dozen runs can screen up to eleven factors, twenty runs up to nineteen. If you have a long list of candidate variables and need to rule most of them out cheaply, that efficiency is genuinely useful. The trade-off is in how it gets there.
Where the efficiency comes from
A fractional factorial built to Resolution IV gives you a clean guarantee: main effects come out clear of two-factor interactions. A Plackett-Burman design doesn't offer that same clean structure — its aliasing pattern is complex and partial, spreading each two-factor interaction's influence thinly across several main effects rather than confounding it cleanly with just one. In exchange for that messier structure, it reaches lower run counts, especially for factor counts that don't land neatly on a power of two.
Why that matters in practice
If your factors genuinely don't interact with each other, this doesn't cost you anything — you get your screening answer in fewer runs. The risk shows up when they do interact: a real two-factor interaction can nudge several main effect estimates at once in a Plackett-Burman design, in a way that's hard to untangle after the fact, where a Resolution IV design would have kept that interaction from touching your main effects at all.
Why BioDOE mostly avoids defaulting to one
For most factor counts, BioDOE builds a proper Resolution IV fractional factorial instead — same screening job, cleaner aliasing structure, and for a good range of factor counts, a Definitive Screening Design that also catches curvature. A Plackett-Burman screen only comes into play as a fallback, for the factor counts where neither of those is constructible. You never have to ask for one by name; if BioDOE uses one, it's because it's the best available option for that specific factor count, not the default.
Start free — screening designs, including the Plackett-Burman fallback when needed, are on the free Starter plan.