Field note · July 16, 2024
Reader question — Designing an experiment that can succeed
Most experiments fail before launch, because they were never large enough to detect the effect anyone hoped for. The fix costs ten minutes and happens before you write code.
Reader question on Designing an experiment that can succeed, and the answer belongs somewhere more findable than an email.
Most experiments fail before launch, because they were never large enough to detect the effect anyone hoped for. The fix costs ten minutes and happens before you write code.
The question was, roughly, "when does this stop applying?" — which is the right question and the one lessons routinely fail to answer. Every technique has a range of validity, and stating it is what separates a lesson from a recipe.
It sits in the Experiments and causal claims module. The exercise is a thinking one rather than a query one — some of this material is about deciding what to compute, and that does not need a keyboard.
Ordering matters here more than in most courses. It follows Correlation, confounding, and Simpson's paradox and leads into Reading a result without fooling yourself, and reading it out of sequence mostly works but costs you the setup.
Free means free, and it also means we can rewrite it whenever it is wrong. No edition, no errata PDF, nothing to repurchase. The whole course is 36 lessons and the fixes land the day we find them.