- Prepare an assessable plan, prior-access disclosure and deviation log.
Why record decisions in advance?
Preregistration makes planned decisions distinguishable from choices informed by results. It supports transparency rather than certifying study quality. Exploratory work remains valuable when it is clearly identified instead of presented as a prespecified confirmation.
Specify actionable details
Include questions, hypotheses, primary outcomes, measurement, population, sampling, sample-size justification, stopping rules, exclusions, missing-data handling and the main model. Explain multiple-testing decisions when appropriate. Saying that a suitable test will be selected leaves the critical choice undocumented.
Be candid about existing data
For secondary data or a late registration, record which information and outcomes were already known and when they were accessed. Do not change the original date or describe a plan made after inspecting outcomes as the initial plan.
Changes and registry selection
A design problem or a better method may justify a change. Preserve the original version, date, rationale and implications; label additional analyses. Choose a registry appropriate to the discipline and institutional requirements, and check its current service conditions. Review protocols and trial registrations have different purposes.
Prepare an auditable package
Deliver the readable analysis plan, registration identifier, amendments and a planned-versus-executed table. Adapt the downloadable toolkit template to the study. Preregistration does not replace ethics review, consent or other mandatory study registration.
Practical research checklist
- Define primary outcomes and models in advance.
- Disclose prior access to data.
- Retain dated and justified amendments.
- Label exploratory analyses clearly.
Worked case and implementation decisions
The following is a fictional teaching case. Do not use its numbers or wording as actual study findings.
The teaching plan identifies score as the primary outcome, one semester as the window and a defined planned model. A later-discovered variable is exploratory and described in a dated amendment, not silently inserted into the old plan. New, secondary and qualitative data need different registration choices. Registered Reports are not equivalent to uploading a form. Specify decisions sufficiently to distinguish prospective commitments from choices made after viewing data.
| Stage | Teaching example | Verification question |
|---|---|---|
| Question | Defined association, not desired result | Are hypothesis and units clear? |
| Prior access | Which data/outputs were seen? | Is this disclosed before registration? |
| Plan | Outcome, exclusions, missingness and model | Are ambiguous choices reduced? |
| Amendment | Change after a new variable is seen | Are date, reason and consequence recorded? |
| Report | Planned and exploratory separated | Does it match registration history? |
Exercise output: Prepare an assessable plan, prior-access disclosure and deviation log.
Sources and further reading
Official sources for verification and further reading