You already know how to read a study. Here's what this site has on weighing evidence properly, from a real randomised trial down to two numbers that happened to move together.
A ranked ladder of causal evidence, from a genuine randomised trial down to two numbers that happened to move together on a dashboard.
Why it matters: a quick way to tell a real result apart from a coincidence dressed up as one.
Read more ›Thirteen real experiments that misled without lying: correct statistics, wrong inference.
Why it matters: a p-value below 0.05 doesn't mean the reasoning built on top of it is sound.
Read more ›Why comparing a metric before and after a change isn't measurement, even when the change looks like it worked.
Why it matters: the pre/post comparison most teams already run is the exact structure this piece takes apart.
Read more ›Why a successful pilot so often fails to survive being rolled out at scale.
Why it matters: the gap between a pilot and a rollout is a real methodological problem, not bad luck.
Read more ›Tversky & Kahneman's real 1991 job-choice study, with its actual methodology critique: what the design got right, and where it's limited.
Why it matters: a worked example of reading a study's strengths and weaknesses together, not just its headline finding.
Read more ›