Preregistration records what a study planned to test before the relevant data collection or analysis. Learners draft a time-stamped plan that covers hypotheses, design, sampling, variables, exclusions, stopping rules, and analysis. They also learn that plans can change: honest deviation reporting keeps confirmatory and exploratory work visible.
By the end of this module, you will be able to:
Draft a time-stamped study plan that distinguishes planned decisions from open questions.
Specify hypotheses or research questions, design, sampling, variables or data sources, exclusions, stopping rules, and analysis decisions when they fit the study.
Classify reported work as confirmatory, exploratory, or a combination of both.
Write a transparent manuscript note that identifies a departure from a plan, explains its reason, and states when it occurred.
A reader opens a paper and cannot tell which analyses were planned before the researchers saw the data and which were developed after patterns appeared. The paper presents every result in the same voice, so a hypothesis generated from the data may look like a prediction that was tested against the data. That ambiguity makes it harder to judge what the study was designed to test and what it discovered along the way.
Preregistration creates a record against which readers can compare the eventual report. It helps authors remember and communicate their original decisions, while leaving room for discoveries, revised procedures, and new questions. Its purpose is not to rank studies as sound or unsound, nor to guarantee that a result will reproduce. It is to make the route from plan to reported findings visible.
A preregistration is a time-stamped record of a study plan made before the relevant data collection or analysis. It records research intentions before researchers observe the outcomes that the plan is intended to address.
The level of detail should fit the design. For a quantitative study, the record may state hypotheses, design, eligibility or sampling rules, target sample or stopping rule, variables and how they will be constructed, exclusion rules, models or tests, covariates, and decision rules for contingencies.
A useful plan also identifies what will be reported and what work, if any, is expected to remain unplanned. Where a design requires choices after inspecting non-outcome features of the data, specify an if–then decision rule: a rule that states what action will follow a stated condition.
Confirmatory work addresses a question or hypothesis using decisions specified before the relevant outcomes were observed. Exploratory work uses observed data or materials to identify patterns, interpretations, questions, or hypotheses worth further study. Both are valuable, but they answer different purposes and should be labelled accordingly.
Preregistration is not limited to experiments or to quantitative research. In qualitative research, a plan can record the research aim, initial research questions, methodological tradition, initial sampling approach, data-collection procedures, intended analytic approach, and the researcher’s theoretical perspective or interpretive lens. Flexibility may be essential; recording the initial plan makes later development easier to inspect rather than treating change as failure.
For studies using existing data, be explicit about what was known before registration: who had seen the data, which summaries or findings were available, and whether the analyst had prior exposure to relevant results. Pre-existing data can limit the separation between prediction and postdiction, but transparent reporting of prior knowledge still helps readers assess that limit.
Studies depart from plans for legitimate reasons. Recruitment may not proceed as expected, an instrument may fail, a planned analysis may violate its assumptions, or fieldwork may reveal that a sampling or data-collection decision needs revision. In qualitative work especially, iterative engagement with data can be integral to the design rather than an error to eliminate.
When a change is necessary, do not silently rewrite the history of the study. Report what changed, why it changed, and when it changed, including whether the relevant outcomes had already been observed. This gives readers a basis for assessing the likely effect of the departure on the interpretation.
Before (planned): “We will stop recruitment after reaching the planned sample and will exclude participants only when the learning-platform log contains no usable activity record.”
After (reported): “Before the primary outcome was analysed, we extended recruitment because a scheduled course offering was cancelled. We retained the planned exclusion rule and report the revised stopping point and reason here.”
The before-and-after note does not turn the revised work into a flaw or erase its confirmatory value automatically. It tells readers where the plan remained intact and where it changed. A change made after outcome inspection deserves especially clear reporting because observed results can influence later decisions.
For qualitative studies, report changes such as revised interview prompts, added participant groups, altered sampling logic, or a developing analytic focus. State how the change arose and how it shaped the materials or interpretation, rather than claiming that an emergent result had been fully specified from the start.
A reader should be able to find a persistent registration reference and see which reported study it concerns. The manuscript should name the registration, describe its relationship to the paper, and identify which analyses, questions, or phases of work were planned under that record. Preserved and accessible registrations allow readers to compare the plan with the reported results.
The paper should also provide a clear account of amendments or deviations. This can be a brief methods note, a dedicated section, or a table that pairs each material change with its reason and timing. The aim is not to imply that every difference invalidates the study; it is to make the difference inspectable.
A registration link alone cannot show that the reported analyses followed the plan, that all planned outcomes were reported, or that later deviations were adequately explained. Readers must compare the record with the manuscript and reported outputs.
Study: A Learning Analytics study of whether weekly feedback messages are associated with subsequent course activity.
Research question: Do students assigned to receive weekly feedback messages show higher activity in the following week than students assigned to receive the usual course communication?
Planned confirmatory work:
Design and sampling: Compare two course sections using the same learning platform. Include enrolled students who have consented to research use of their course records.
Variables: Define subsequent activity as the number of recorded course interactions during the week after the message. Define the group indicator from the assigned communication condition.
Exclusions: Exclude students who withdraw before the feedback week and records with no usable activity data.
Stopping rule: Analyse records after the final planned feedback week; do not add weeks because an interim result appears favourable or unfavourable.
Analysis: Estimate the difference in subsequent activity between conditions using the specified model. Report the estimate and uncertainty for the planned comparison.
Exploratory work: Examine whether patterns differ across types of course activity and use these observations to generate questions for later research.
Planned confirmatory analysis. We conducted the registered comparison of subsequent activity between communication conditions using the preregistered inclusion, exclusion, and outcome definitions.
Justified change. Before analysing the planned outcome, the course timetable changed and one planned feedback week did not occur. We therefore analysed the remaining planned feedback weeks only; this change was made before outcome analysis and is reported as a deviation from the original stopping schedule.
Exploratory analysis. After completing the planned analysis, we examined activity categories separately to identify patterns for future study. These category-specific analyses were not part of the preregistered confirmatory plan and are interpreted as exploratory.