How Preregistration and Registered Reports Curb Bad Science

Preregistration timestamps a study's hypotheses and analysis plan in a public archive before any data are collected, and Registered Reports go one step further by having a journal accept the study on the strength of its methods alone, before a single result exists. Both tools attack the same weakness: the freedom researchers have to decide what counts as a finding after they have already seen the data. When you can tell that a study's plan preceded its results, you can weigh its claims more fairly. This is a guide to appraising evidence, not medical advice.

The problem these tools were built to solve

Start with two habits that quietly inflate the scientific literature. The first is HARKing, short for "hypothesizing after the results are known." A team runs an analysis, notices an unexpected pattern, and then writes the paper as if that pattern were the prediction all along. The second is p-hacking: trying many analyses, outcomes, or subgroups and reporting only the combination that crosses the threshold for statistical significance. Neither requires fraud. Both can feel like ordinary diligence to the people doing them, which is exactly why they are so common.

Layered on top is publication bias. Journals have historically preferred clean, positive, surprising results, so studies that found nothing often went into a file drawer and never appeared. The reader sees the successes and never counts the silent failures, which makes any single striking finding look more solid than it is.

The cumulative cost of these patterns became hard to ignore in 2015, when the Open Science Collaboration attempted to replicate 100 studies from three psychology journals in a project published in Science. Ninety-seven percent of the original studies had reported statistically significant results; only 36 percent of the replications did, and the replicated effects were on average about half the size of the originals. That gap, widely called the replication crisis, is what preregistration and Registered Reports were designed to narrow.

What preregistration actually does

Preregistration is the simpler of the two. Before collecting data, researchers write down their hypotheses, their sample size and stopping rule, their primary and secondary outcomes, and the exact statistical tests they will run, then deposit that document in a public registry with a timestamp. For clinical trials this has been mandatory for years through registries such as ClinicalTrials.gov, but preregistration now extends to observational and laboratory work as well.

The point is not to forbid discovery. A preregistered study can still run unplanned analyses; it simply has to label them as exploratory rather than confirmatory. That single distinction does most of the work. A confirmatory test, specified in advance, carries its stated statistical meaning. An exploratory test, chosen after seeing the data, is a hypothesis for the future, not a conclusion. Preregistration draws a bright line between the two so a reader is not asked to take the researcher's word for which is which.

Its main limit is enforcement. A preregistration exists, but nothing compels a journal or an author to honor it, and studies have found that published papers sometimes deviate from their registered plans without saying so. Preregistration gives you a document to check against; it does not guarantee anyone checked.

What Registered Reports add

Registered Reports close that gap by moving peer review to before the results exist. The format, coordinated and tracked by the Center for Open Science, splits review into two stages. In Stage 1, authors submit only their introduction, methods, and analysis plan. Reviewers and editors judge whether the question is worth answering and whether the design can answer it, and they can demand changes while changes are still cheap. A protocol that survives this scrutiny earns in-principle acceptance.

In-principle acceptance is the structural innovation. It means the journal commits to publishing the completed study regardless of how the results turn out, so long as the authors follow the approved protocol and the work meets basic quality standards. In Stage 2, after the data are in, authors add their results and discussion, keeping any unregistered analyses in a clearly marked exploratory section. Acceptance cannot be revoked because the findings were null or unsurprising.

That design removes the incentive to p-hack or HARK, because a positive result no longer buys publication that a negative result would be denied. It also front-loads the most useful criticism, since reviewers shape the study when their suggestions can still improve it. More than 300 journals now offer the format.

Does it work, and how to read one

The early evidence is encouraging. In a 2021 study in Nature Human Behaviour, Soderberg and colleagues had 353 researchers blindly rate 29 published Registered Reports against 57 comparison papers across 19 criteria. The Registered Reports scored higher overall, with the largest advantages in rigor of methodology and rigor of analysis, and no penalty in novelty or creativity. In other words, tying researchers to a fixed plan improved quality without dulling the science.

For a reader, the practical move is to check the sequence. Does the paper link to a preregistration or carry a Registered Report badge? Do the confirmatory analyses match what was registered, and are post-hoc analyses labeled as exploratory? A study whose plan demonstrably preceded its results deserves more of your confidence than one where you simply cannot tell. That question, plan before results or results before plan, is one of the sharpest tools you have for judging a claim.