Preregistration and Registered Reports: Should You Commit to Your Study Plan Before Seeing the Results?

Should You Really Commit to a Research Plan Before You Know What the Data Will Show?

Research rarely unfolds exactly as imagined. Recruitment may be slower than expected. A measure may perform poorly. An assumption behind an analysis may fail. An unexpected pattern may appear in the data and lead to a genuinely interesting question.

So the idea of committing to hypotheses, methods, outcomes, or analyses before seeing the results can initially sound restrictive. Why constrain yourself when good research sometimes requires adaptation?

That concern sits at the center of preregistration and Registered Reports. Both approaches ask researchers to make important decisions before the results can influence those decisions, but they do so in different ways and with different levels of commitment. The purpose is not to make research inflexible. It is to make the history of the research more transparent, particularly the distinction between decisions made before and after researchers knew what the data looked like.

Yes, Sometimes Committing in Advance Can Make the Evidence Easier to Interpret

What You Are Actually Committing to Before Seeing the Results

Preregistration Creates a Time-Stamped Record of Your Research Intentions

At its simplest, preregistration means specifying important aspects of a study before the relevant outcomes are known and preserving that plan in a time-stamped record. The Open Science Framework describes preregistration as posting a time-stamped, read-only version of a study plan before beginning the relevant data collection or analysis.

The underlying idea is straightforward. Researchers routinely make choices about hypotheses, variables, exclusion criteria, sample sizes, statistical models, transformations, subgroup analyses, and stopping rules. Some of those choices may be made before examining the data. Others may arise because of something observed after data collection or analysis begins.

Both kinds of decisions can be scientifically legitimate. They do not, however, have exactly the same evidential status.

Preregistration provides a record against which readers can distinguish decisions made in advance from decisions influenced by what researchers subsequently learned from the data. This is why the practice is closely connected to the distinction between exploratory and confirmatory research.

The Main Point Is Not Prediction for Prediction's Sake

Suppose a researcher tests ten plausible analyses, discovers that one produces an interesting result, and reports only that analysis as though it had been planned from the beginning. A reader sees a much cleaner inferential story than the one that actually occurred.

The problem is not that the researcher explored the data. Exploration is an important part of research. The problem is that the published account can blur the distinction between a result that was predicted and tested according to an advance plan and a pattern discovered after examining the data.

Nosek and colleagues describe preregistration as a way of distinguishing prediction from postdiction by defining research questions and analysis plans before researchers observe the relevant outcomes. In statistical research, advance specification may also constrain some forms of researcher flexibility that could otherwise allow analytical decisions to depend on the results.

Preregistration therefore does not make exploratory findings illegitimate. It makes their origin easier to understand. Researchers can still investigate unexpected patterns and use them to generate new hypotheses for future studies.

A Useful Preregistration Must Be Specific Enough to Constrain Meaningful Choices

Merely registering a document before data collection does not necessarily accomplish much. A vague statement such as “we will examine whether the intervention affects student performance” leaves numerous consequential decisions unresolved.

What constitutes an adequate plan depends on the research design, but a useful preregistration may address the research questions or hypotheses, design, sampling plan, variables, exclusion criteria, stopping rules, outcome definitions, and intended analyses. The appropriate level of detail is a methodological question in its own right, which is why researchers should think carefully about what needs to be specified in a preregistration rather than treating registration itself as the objective.

Timing also matters. The relevant principle is that the decisions should be documented before the information they are intended to constrain becomes known. For a new experiment, this will often mean preregistering before data collection. With existing datasets, secondary analyses, or other designs in which some information already exists, the appropriate timing and claims require more care.

A Registered Report Adds Peer Review Before the Results Are Known

A Registered Report is not simply a more detailed preregistration. It is a journal publication format built around prospective peer review.

Although procedures vary by journal, the conventional model has two stages. At Stage 1, researchers submit the research question, rationale, hypotheses where applicable, methods, and proposed analyses before the main results are known. Editors and reviewers evaluate the importance or validity of the research question and the strength of the proposed methodology.

If the submission meets the journal's requirements, it may receive in-principle acceptance. This means that publication is provisionally secured before the eventual direction or statistical significance of the results is known, provided that the researchers conduct the study appropriately, satisfy the journal's conditions, and successfully complete Stage 2 review.

After the research is conducted, the completed manuscript returns for Stage 2 review. Reviewers can assess whether the approved procedures were followed, whether deviations were adequately justified and reported, and whether the conclusions are supported by the evidence.

The distinction matters. If your main objective is to create a transparent record of what you intended to do before examining the results, preregistration may be sufficient. If you also want the study design evaluated by reviewers before conducting the research and want the publication process to be less dependent on whether the findings turn out to be striking, a Registered Report may serve a different purpose.

Committing to a Plan Does Not Mean Pretending That Research Is Predictable

A common objection is that researchers cannot anticipate everything that might happen during a study. That is correct, and preregistration does not require them to do so.

A recruitment strategy may fail. Equipment may malfunction. A planned statistical model may not converge. A coding scheme may need refinement. New methodological information may emerge after registration. Researchers may also discover an analysis that is more appropriate than the one originally planned.

The existence of a preregistration does not make such changes inherently improper. The important question is how those changes are handled. A useful principle is to preserve the original record, identify consequential departures from it, explain why they occurred, and avoid presenting revised decisions as though they had always been planned.

This is why preregistration should not be interpreted as a prohibition on change. Transparency about adaptation is generally more informative than rigid adherence to a plan that has become methodologically inappropriate.

Preregistration and Registered Reports Address Different Sources of Bias

The two practices overlap, but the problems they address are not identical.

Preregistration can expose the temporal relationship between research decisions and observed results. This may help readers evaluate selective reporting, outcome switching, undisclosed analytical flexibility, and hypotheses presented as predictions even though they were developed after examining the results.

Registered Reports add another mechanism: peer review and an editorial commitment before the final outcomes are known. This is intended to reduce the influence of result direction or statistical significance on publication decisions while allowing reviewers to identify design problems at a point when researchers may still be able to correct them.

That difference can be consequential. Traditional peer review often evaluates methods after data collection is complete, when even an excellent methodological suggestion may arrive several months too late. Registered Reports move part of that scrutiny upstream.

Not Every Study Needs the Same Degree of Advance Commitment

The value of preregistration depends partly on what kind of inference a study is intended to support. It can be particularly informative when researchers are testing prespecified hypotheses and when substantial analytical flexibility could affect the reported conclusions.

That does not mean that preregistration belongs exclusively to conventional quantitative experiments. Researchers have developed preregistration approaches for other designs, including qualitative and explicitly exploratory work. The objectives and appropriate content may differ substantially, however.

A highly emergent qualitative design, an early-stage exploratory investigation, and a tightly specified confirmatory experiment should not be forced into identical preregistration templates. The more useful question is whether advance specification adds meaningful transparency for the particular research process.

Preregistration Is Not the Same as Having a Research Protocol

Researchers sometimes already have a protocol, ethics application, dissertation proposal, grant methodology, or internal study plan and wonder whether that means the study has effectively been preregistered.

These documents can overlap substantially in content, but they serve different purposes. A protocol primarily specifies how research is to be conducted. Preregistration adds a time-stamped record intended to establish what was specified before the relevant outcomes were known. Depending on the platform and research context, the registration may also be preserved independently and made accessible to readers.

The practical overlap can be substantial, but the concepts should not automatically be treated as interchangeable. The distinction between a protocol and a preregistration depends on the function and timing of the record, not simply on how detailed the document happens to be.

The Empirical Evidence About Preregistration Deserves Some Caution

The rationale for preregistration is strong enough that it is tempting to assume preregistered studies must consistently produce better or more trustworthy findings. The empirical picture is less simple.

For example, a study by van den Akker and colleagues compared 193 preregistered psychology studies with 193 related non-preregistered studies. The preregistered studies more often reported power analyses and generally had larger sample sizes. However, the researchers did not find robust evidence that the preregistered studies had fewer positive findings, smaller effect sizes, or fewer statistical errors. Because researchers self-select into preregistration and the groups may differ in other ways, causal interpretation is difficult.

This is an important qualification. Preregistration is a research practice with particular functions, not a guarantee that every preregistered study is superior to every non-preregistered one. Its value should be assessed in terms of what it makes transparent and which inferential or publication problems it can plausibly address.

How Advance Commitment Changes the Interpretation of the Same Study

If the researchers had instead submitted the design as a Registered Report, reviewers might have evaluated the research question, outcome measures, sample-size justification, and proposed analyses before the study proceeded. Successful Stage 1 review could lead to in-principle acceptance, meaning that an unexciting or null primary result would not by itself make the completed study unpublishable in that journal.

Common Misunderstandings About Committing to a Study Plan

How to Decide Whether Preregistration or a Registered Report Fits Your Study

Do not begin with the question, “Is preregistration good research practice?” That is too broad to be useful. Begin with what you need advance commitment to accomplish in your particular study.

If you choose preregistration, treat the exercise as part of research design rather than an administrative form to complete. Writing down your decisions can expose ambiguities that are easy to overlook when they remain only in your head: What exactly is the primary outcome? When does recruitment stop? Which observations will be excluded? Which model answers the primary question? What happens if an assumption fails?

If you choose a Registered Report, plan even earlier. Journal requirements vary, and Stage 1 review may require substantial methodological detail before data collection proceeds. Check the target journal's current instructions rather than assuming that every Registered Report follows exactly the same workflow.

Most importantly, do not confuse commitment with inflexibility. The strongest plan is not necessarily the one that never changes. It is the one that makes clear what was decided when, preserves a defensible record of consequential changes, and allows readers to understand how much the observed results could have influenced the analytical story eventually reported.

Before You Decide to Preregister or Submit a Registered Report

Questions Researchers Ask About Preregistration and Registered Reports

Commit in Advance When the Commitment Makes the Evidence More Informative

How to Cite This Guide

This guide is intended to be read, shared, and used in research, teaching, and academic work. If you draw on its ideas, explanations, or other content, please acknowledge the source by citing the guide. Doing so gives appropriate credit and helps your readers locate the original resource.