How Should Exploratory Findings Be Reported Without Pretending They Were Predicted?

What Should You Say When the Data Reveal Something You Did Not Predict?

You begin with a clear research question and a hypothesis. Then the data show something unexpected.

Perhaps an effect appears only in one group. Maybe an association emerges between variables you had not linked theoretically. Or the result points in a direction you did not anticipate at all.

The tempting response is to build the paper around the new finding and explain why you supposedly expected it all along. That makes for a wonderfully tidy manuscript. Science, unfortunately, is not always so tidy.

The better approach is to preserve the distinction between what you set out to test and what you discovered while investigating the data. Exploratory findings can be reported, discussed, and even become important contributions. Their evidential status should simply remain visible.

Report the Finding Honestly, Then Tell Readers How It Was Discovered

Exploration Is Part of Research, Not a Reporting Failure

Exploratory Analysis and Confirmatory Analysis Ask Different Questions

The distinction is primarily about the relationship between the analysis and the evidence that informed it.

A confirmatory test asks, in effect, “Does the evidence support this prediction that was specified before seeing these results?”

An exploratory analysis asks something closer to, “What pattern might be present here, and is it worth investigating further?”

Neither question is inherently superior. They serve different purposes.

The First Job Is to Preserve the Chronology

Readers should be able to tell when the hypothesis was formulated relative to the relevant data.

That can be done without awkward methodological confessions. Simple wording is often enough:

“The study originally hypothesized that the intervention would improve examination performance. An exploratory analysis subsequently indicated a possible interaction with prior achievement.”

The sentence does useful work. It tells readers that the interaction was not part of the original prediction.

ICMJE similarly advises authors to state new hypotheses when warranted but to label them clearly. Its current recommendations also distinguish information available when the study plan was written from information obtained during the study, which belongs in the Results.

Do Not Turn Exploration Into HARKing

The problem arises when a post hoc finding is rewritten as a prediction.

Suppose you explore six possible moderators and unexpectedly find that prior experience appears to alter the effect. You then rewrite the introduction so that prior experience sounds like the theoretical focus from the beginning.

The finding is now being presented with a different history from the one that actually occurred.

This is the problem addressed by HARKing, or hypothesizing after the results are known.

The solution is not to delete the exploratory finding. It is to preserve its status.

Exploratory Does Not Mean “Ignore the Statistics”

Labeling an analysis exploratory does not release you from methodological standards.

You still need an appropriate analysis, sensible measurement, accurate numerical reporting, and conclusions proportionate to the evidence. A post hoc result can be spurious, unstable, or driven by analytical flexibility.

Exploratory status changes how strongly the finding should be interpreted. It does not change whether the statistical procedure itself should be defensible.

This is one reason transparent reporting is useful. Readers can interpret an interesting finding differently when they know it emerged from broad exploration rather than from a narrowly specified test.

Tell Readers What Prompted the Exploration

An exploratory analysis can arise in several ways.

You might discover an unexpected pattern while examining descriptive statistics. A diagnostic may reveal a nonlinear relationship. Previous research may suggest a possible subgroup explanation after the primary analysis. A reviewer may ask whether an effect differs across a theoretically meaningful characteristic.

Those origins are not equivalent.

A data-derived subgroup identified after inspecting the results carries a different evidential history from an analysis prompted by a well-established theoretical concern that simply was not specified in the original protocol.

You do not need an elaborate taxonomy every time. But when the origin of the analysis affects how a reader should interpret it, explain the origin.

Do Not Hide the Search That Produced an Interesting Finding

Suppose you explored age, prior achievement, study program, attendance, and socioeconomic background as possible moderators. One produces a compelling interaction.

Reporting only the successful moderator can make it appear as though that variable was the obvious analytical focus.

The relevant issue is not whether every exploratory calculation needs a dedicated paragraph. It is whether the extent of the search matters to the credibility of the reported result.

If dozens of potential patterns were investigated, the successful result should not be interpreted as though it came from a single predetermined test.

This is closely related to the concerns surrounding multiple analyses and result-driven analytical searching.

An Exploratory Finding Can Be Reported Prominently

Exploratory status does not require burying an important discovery in the final paragraph of the paper.

Sometimes the unexpected finding is the most interesting part of the study. It may challenge prevailing assumptions, reveal an important subgroup pattern, or suggest a new mechanism.

You can give it substantial attention while preserving the chronology.

The important distinction is between prominence and confirmatory status. A finding can be prominent without being presented as a test of a prediction established before the data were examined.

Explain What the Exploratory Finding Does and Does Not Establish

A useful Discussion should answer two different questions.

First, what might the finding mean?

Second, how much confidence should readers place in that interpretation?

These should not be collapsed into one enthusiastic claim.

Suppose an unexpected interaction suggests that the intervention works differently depending on baseline experience. You can discuss a plausible mechanism, connect it with previous literature, and explain why it matters. At the same time, you can state that the interaction was exploratory and therefore requires further evaluation.

That is not excessive caution. It is information about how the evidence was generated.

Independent Confirmation Changes the Evidential Situation

A hypothesis generated from one dataset can subsequently be tested with new data.

The new study can specify the hypothesis before observing the relevant results, reducing the particular circularity created when the same observations both suggest the hypothesis and appear to confirm it.

This does not make the original exploratory study worthless. The first study generated the question. The second study can provide a more direct test.

In practical terms, a useful sequence is:

Exploratory Findings Should Not Become More Certain Merely Through Repetition of the Same Narrative

A common reporting trap is narrative escalation.

The Results section says the finding is exploratory. The Discussion calls it “promising.” The Conclusion says it “demonstrates” a mechanism. The abstract describes it as a major contribution. By the end, the exploratory label has technically survived while the actual language has become much stronger.

Readers experience the overall claim, not just one methodological adjective.

Interpretive language should therefore remain proportionate throughout the paper.

Reporting Guidelines Can Help Keep the Distinction Clear

The relevant reporting guideline depends on the study design. EQUATOR Network maintains reporting guidelines for randomized trials, observational studies, systematic reviews, qualitative research, and many other designs. Its purpose is to help researchers identify the information readers need to understand and appraise a study.

This is useful because the amount of detail needed to describe an exploratory analysis is not identical across all methodologies.

The underlying principle remains stable: report enough information for readers to understand the analytical process and distinguish what was planned from what emerged during the research.

How to Turn an Unexpected Result Into a Transparent Research Contribution

Notice what is not required. The researcher does not have to pretend the interaction was predicted. Nor do they have to suppress it simply because it was unexpected.

The finding can be central to the paper's intellectual contribution while retaining the label that tells readers how it was discovered.

A later study can then ask the new question directly. That creates a useful research sequence rather than an artificially polished first study.

Common Mistakes When Reporting Exploratory Findings

How to Write About an Exploratory Finding Without Underselling or Overselling It

A good report gives the finding enough attention to be useful while preserving the uncertainty attached to how it was discovered.

In writing the Discussion, avoid the temptation to make the result sound more certain merely because you can now explain it elegantly. Explanation after discovery is part of science, but it is still explanation after discovery.

ICMJE specifically recommends that new hypotheses be stated when warranted but labeled clearly. It also recommends that conclusions be linked to study objectives and supported by the actual data rather than by unqualified claims.

Before Publishing an Exploratory Finding

Frequently Asked Questions About Exploratory Findings

Discovery Does Not Need to Pretend It Was Prediction

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.