Pre-registration

Pre-registration means writing down your hypothesis, design and analysis plan, with a timestamp, before you collect or look at the data.

Why it works

It separates what you predicted from what you discovered. Both are valuable, but they're different kinds of evidence. Without a record, it's easy, even honestly, to tell yourself afterwards that you expected what you found.

Striking evidence

In 2000, the US government began requiring large clinical trials to register their main outcome in advance. Kaplan and Irvin looked at large heart and lung trials funded by the US National Heart, Lung, and Blood Institute. Before 2000, 57% reported a benefit. After registration was required, only 8% did. The treatments hadn't got worse; there was simply less room to choose the most flattering outcome.

How to do it

• Use a free registry: OSF Registries or AsPredicted for most fields, ClinicalTrials.gov for medical trials.
• Write: hypothesis, design, sample size, main outcome, analysis, exclusion rules.
• Afterwards, report every deviation from the plan, and label any extra analyses as exploratory.

Registered Reports

Some journals review your plan before the study and promise to publish whatever the result, if you follow it. This removes the pressure to find "positive" results.

RememberWrite your plan down with a timestamp before you see the data. Then report what you planned, and label what you didn't.