Pre-registration means writing down what you intend to test, and how, before you look at the results — then registering that plan somewhere timestamped and public. It is the main structural defence against analytic flexibility: with enough preprocessing choices and model specifications, a large neuroimaging dataset will produce a significant result for almost any hypothesis.
It also draws a clear line between confirmatory and exploratory work, which lets you explore freely without that exploration quietly being reported as confirmation.
Related¶
Reporting guidelines — the reporting-time counterpart
Working with imbalanced data — one place analytic flexibility bites