Blog · Why We Pre-Register
Pre-registration means publicly locking your hypothesis, your method and your analysis plan before you collect any data, on an open registry such as the Open Science Framework. You publish null results because hiding them is how a field quietly fools itself. Ashta applies both rules to every contested claim it tests — telephone telepathy, applied kinesiology, focused intention, the sealed-box forced-choice procedure — with the standing policy that every outcome, including clear nulls, appears in the same public record.
Most of the ways a study can go wrong are invisible after the fact. A team can run twenty small analyses, find the one that crossed the line, and write it up as if that had been the plan all along. A vague prediction can be quietly reshaped to fit whatever the data happened to show. A surprising result that appeared in the data can be presented as the hypothesis the study set out to confirm. None of this requires dishonesty — it is what motivated, hopeful people do by default. The damage is that the published claim looks far stronger than the evidence actually was, and nobody reading it can tell the difference.
Pre-registration closes those doors by moving the decisions in front of the data. Before a single measurement is taken, you write down — in public, with a timestamp — the exact hypothesis, the sample size, how participants are assigned, how the test is blinded, who is excluded and why, the scoring rule, and the specific analysis that will decide the question. The standard home for this is the Open Science Framework, run by the non-profit Center for Open Science, where a frozen, dated record cannot be edited after the fact. Once that record exists, the result is whatever the pre-specified analysis returns. Other patterns can still be explored, but they have to be labelled as secondary or exploratory — interesting leads for a future study, not evidence dressed up as confirmation.
Publishing nulls is the other half, and it fixes a problem the field has carried for decades: the file-drawer problem. If only the exciting results get written up while the flat ones go in a drawer, the published record becomes a highlight reel — and a highlight reel of a coin that mostly came up tails will still look like magic. This is the heart of the modern reproducibility reckoning, the same concern John Ioannidis raised in arguing that a large fraction of published findings fail to replicate, and the same reason large replication projects have spent years re-running famous results. The correction is unglamorous: report what you pre-registered, whether it worked or not, so that the nulls sit in the record next to the hits and the overall picture stays honest.
For a project that puts contested claims under test, these two rules are not optional polish — they are the whole credibility of the exercise. When Ashta runs a sealed-box forced-choice test, a telephone-telepathy or applied-kinesiology protocol, or a focused-intention study, the temptation to keep only the striking sessions and forget the rest is exactly the temptation that produced a century of disputed results. Pre-registration plus mandatory null reporting is the minimum standard that lets a positive finding mean something and a negative finding be useful. It is also why a researcher like Stephan Schwartz — who has published both a striking positive and a clean null on the same kind of question, in the same journal — is the model rather than the exception: a result is credible precisely because the same person was willing to publish it either way.
The practical commitment is plain. Every test Ashta runs is pre-registered before data collection, and every outcome is reported in the same place — the wins, the nulls, and the studies that had to be stopped or excluded. A claim that can only ever come back positive is not being tested; it is being marketed. The point of building the work this way is that a claim can lose clearly, in public, on the record. The research lineage Ashta draws on is genuinely contested, and the only way to add anything to it honestly is to test under conditions where the answer is allowed to disappoint us — and to publish it when it does.
