| README.md | ||
Preregistration for Beginners: Why and How to Register Your Study Before You Start
Preregistration means publicly recording your research questions, hypotheses, methods and analysis plan before you collect or analyse data. Once a niche practice in clinical trials, it has spread across psychology, economics, ecology, education and many other fields. Its purpose is simple: to make clear which decisions were made before seeing the results and which were made afterwards. That distinction helps readers judge how much confidence to place in your findings. This guide explains why preregistration matters, what to include, where to register and how to handle the inevitable deviations, and shows how a source-linked research assistant such as Yila can help you ground your plan in the existing literature.
The problem preregistration solves
Researchers make many decisions during a study: which participants to exclude, which outcome measure to treat as primary, which covariates to include, how to handle outliers and which statistical test to use. Each decision is often defensible on its own. But when decisions are made after seeing the data, it becomes easy, even unintentionally, to choose the options that produce clean or significant results. This flexibility inflates false-positive rates and makes published findings less reliable.
Preregistration does not ban flexibility. It makes flexibility visible. By committing to a plan in advance, researchers separate confirmatory tests, which can be interpreted at face value, from exploratory analyses, which generate hypotheses for future testing.
Benefits beyond credibility
Preregistration also improves the research itself:
- Better planning: writing a detailed plan forces you to think through design problems before they become expensive.
- Clearer communication: a registered plan can be shared with collaborators and supervisors, reducing misunderstandings.
- Protection against bias: it helps guard against your own wishful thinking.
- Publication options: some journals offer registered reports, in which the study is peer reviewed and accepted in principle before results are known.
Many researchers find that the time spent preregistering is recovered later, because analysis and writing go faster when decisions have already been made.
What to include
The level of detail depends on the field and study type, but a useful preregistration usually covers:
- Research questions and hypotheses. State them precisely, including the direction of expected effects.
- Design. Describe the study design, conditions, randomisation and blinding.
- Participants or sampling units. Define the population, recruitment method, inclusion and exclusion criteria.
- Sample size and justification. Explain how the sample size was determined, for example through a power analysis or practical constraints.
- Measures. Specify each variable, how it is measured and how it is scored.
- Analysis plan. Name the statistical models, covariates, handling of missing data, outlier rules and criteria for inference.
- Stopping rules. Say when data collection will end.
- Exploratory analyses. Mention planned exploratory work so readers know it was anticipated.
Be specific. "We will use regression" leaves too much room. "We will fit a linear mixed model with condition as a fixed effect and participant as a random intercept, using a two-sided test at alpha 0.05" is much clearer.
Use the literature to justify choices
Many preregistration decisions depend on previous research. Expected effect sizes for power analysis, validated measures, typical exclusion rates and accepted analysis methods all come from the literature. Spending time on a focused review before writing your plan makes it stronger and easier to defend.
Search for previous studies with similar designs and populations, recent meta-analyses for effect size estimates and methodological papers on your chosen measures. Research assistants that link statements to specific sources can speed up this search; open each source and check the details, especially numbers you will use in calculations. Cite the sources that informed your choices in the preregistration itself.
Where to register
Several platforms host preregistrations:
- Clinical trial registries are mandatory for many clinical studies and follow international standards.
- The Open Science Framework (OSF) offers templates for many study types and allows embargoes.
- AsPredicted uses a short, structured form that is quick to complete.
- Field-specific registries exist in economics, education, ecology and other disciplines.
Choose a platform that issues a timestamp and a persistent link, and that is accepted in your field. If you plan to publish a registered report, follow the journal's specific requirements.
Preregistering different types of research
Preregistration began with experiments, but it can be adapted to other designs.
- Secondary data analysis: state what you already know about the data, whether you have accessed them, and how you will avoid being influenced by prior exposure.
- Qualitative research: register the research questions, sampling approach, data collection procedures and analytic approach, while acknowledging that qualitative work is often iterative.
- Systematic reviews: register the protocol, including search strategy, eligibility criteria and synthesis methods.
- Replication studies: specify how you will judge whether the replication succeeded.
Templates exist for many of these cases; use them as a checklist.
Handling deviations
Research rarely goes exactly to plan. Recruitment may fall short, a measure may perform badly or an assumption of the planned model may be violated. Deviations are normal and do not invalidate preregistration. What matters is transparency.
When you deviate, record what changed, when and why. In the final paper, include a section or table listing deviations from the registered plan. Where possible, report both the planned and the revised analyses so readers can see whether the change affected the results. Hiding deviations defeats the purpose of preregistration and can damage trust more than the deviation itself.
Common misconceptions
- "Preregistration prevents exploration." It does not. It simply labels exploration so it is not mistaken for confirmation.
- "It takes too long." A basic preregistration can take a few hours, much of which would be spent planning anyway.
- "Reviewers will penalise deviations." Transparent, well-justified deviations are generally accepted.
- "It only applies to experiments." Many study types can benefit, with adapted templates.
- "Someone will steal my idea." Embargo options allow you to keep the registration private until publication.
Practical tips for a first preregistration
- Start with a template appropriate to your design.
- Draft the analysis code using simulated data before registering; it reveals gaps in the plan.
- Ask a colleague to read the plan and identify ambiguities.
- Keep the registration and the final paper consistent in terminology.
- Link the registration in your manuscript and cite it in the methods.
Conclusion
Preregistration is a simple practice with large benefits. It improves planning, protects against bias and helps readers distinguish confirmatory from exploratory findings. Start by grounding your plan in the existing literature, write a specific and complete registration, choose an appropriate platform and report deviations honestly. Tools such as Yila can help you find and verify the studies that justify your choices, but the commitment to transparency is yours. Each preregistered study contributes to a research culture in which findings can be trusted and built upon.