Methods

Preregistration

Preregistration records a research plan before the relevant outcomes are known. It helps readers distinguish decisions made in advance from decisions prompted by observing results, while keeping exploration a legitimate part of research.

Which hypotheses, measurements, exclusions, and analysis decisions can be specified before learning the outcomes that will test them?

How it works

  1. State the research question, primary outcomes, sampling or stopping rule, exclusions, and planned analysis with enough detail to expose meaningful choices. A vague prediction such as “the groups will differ” leaves many result-dependent decisions unresolved.
  2. Submit the plan to a registry at the relevant point in the workflow. Existing-data projects need a clear account of prior access and knowledge; a timestamp alone cannot establish that the analysis was independent of the outcomes.
  3. Report the plan alongside what actually happened. Explain deviations and distinguish additional exploratory analyses. A useful plan makes changes auditable; it should not encourage researchers to conceal a justified repair or continue an analysis that has become inappropriate.

Evidence: (Center for Open Science, n.d.-a); (Nosek et al., 2018)

Two outcomes become one primary question

Teaching scenario

Synthetic teaching scenario: A team investigating background music proposes both recall accuracy and response time. Before data collection, it names accuracy as its primary outcome and specifies the comparison, recruitment target, and exclusions. Later, an unexpected response-time pattern suggests a new hypothesis. The report retains the accuracy analysis and describes the timing pattern as exploratory. Students then draft a second study that could test the new hypothesis with fresh observations. No claim is made that the first registration validates either measure or proves that the music caused an effect.

This example is constructed for learning. It does not report an observed research result.

What can you conclude?

It makes the relationship between a prior plan and a final analysis inspectable, reducing ambiguity about when research decisions were made.

Evidence: (Center for Open Science, n.d.-a); (Nosek et al., 2018)

Limits and common mistakes

  • A preregistered study can still have poor measurements, insufficient information, or a design that cannot support its intended inference.
  • Registration does not prevent selective reporting, undisclosed deviations, fraud, or accidental outcome knowledge by itself.

A mistake to catch

Presenting a registration badge as evidence of validity without comparing the registered plan, timing, and published analysis.

Evidence: (Center for Open Science, n.d.-a); (Nosek et al., 2018)

Sources and inspection notes

Links checked October 21, 2026. Inspection scope is stated for each source; an access date is not a publication date.

  1. Center for Open Science. (n.d.-a). Preregistration. https://www.cos.io/initiatives/prereg

    Center for Open Science methodological guidance Read definition, existing-data timing, and transparent changes guidance. Checked 21 October 2026.

  2. Nosek, B. A., Ebersole, C. R., DeHaven, A. C., & Mellor, D. T. (2018). The preregistration revolution. Proceedings of the National Academy of Sciences, 115(11), 2600–2606. https://doi.org/10.1073/pnas.1708274114Consulted source

    Nosek, Ebersole, DeHaven, and Mellor, 2018; methodological article Read abstract and prediction versus postdiction discussion. Checked 21 October 2026.

Keep reading

Spielman, R. M., Jenkins, W. J., & Lovett, M. D. (2020). Psychology 2e. OpenStax. https://openstax.org/books/psychology-2e/pages/2-1-why-is-research-important

Nearest related OpenStax Psychology 2e reading. This is a course connection, not evidence for every specialist detail above.

Specialist reading: Center for Open Science. (n.d.-a). Preregistration. https://www.cos.io/initiatives/prereg

Related methods

Try the methods · Read a study · Explore psychological measures