Put it together, and you’re studying a highly variable phenomenon using a minimal treatment, using a statistically inefficient design.