Fabrication vs. Falsification: What’s the Difference?

What Separates Fabrication From Falsification?

A researcher invents observations that were never collected. Another researcher collects genuine observations but removes inconvenient values, alters measurements, or changes the research process so that the record presents a misleading account. Both situations may involve false research records, but they are not the same kind of conduct.

The distinction matters because fabrication and falsification describe different ways in which the research record can cease to represent what actually happened. Understanding that difference helps researchers recognize problematic practices, describe concerns more precisely, and distinguish serious allegations from legitimate data processing, correction, or methodological judgment.

The Difference Comes Down to Invention Versus Misrepresentation

How Fabrication and Falsification Differ in Research Practice

Fabrication Introduces Invented Data or Results Into the Research Record

The U.S. Department of Health and Human Services Office of Research Integrity (ORI) defines fabrication as making up data or results and recording or reporting them. The essential feature is invention. The purported data or result has no genuine research event behind it.

Imagine that a study recruited 180 participants, but a researcher enters responses for another 20 participants who were never recruited. Those additional observations are not estimates, corrected values, or alternative interpretations of existing evidence. They represent observations that never occurred.

Fabrication is not limited to an entire dataset. A researcher may potentially fabricate a single observation, measurement, interview response, experimental result, or other piece of research information. The relevant question is whether purported data or results were made up and then entered into the research record. The more detailed question of what counts as fabricating research data therefore depends on what was invented and how it entered that record.

Falsification Alters How Research Is Represented

Falsification begins from a different problem. Rather than simply creating nonexistent data or results, it involves manipulation of research materials, equipment, or processes, or changing or omitting data or results in a way that causes the research record to represent the research inaccurately.

That definition is broader than manually changing numbers in a spreadsheet. Depending on the circumstances, falsification may involve manipulating a research process, selectively altering measurements, improperly omitting observations, or modifying research materials in a way that changes what the record appears to show.

The phrase “such that the research is not accurately represented in the research record” is crucial. Researchers routinely transform data legitimately. They correct documented errors, recode variables, exclude observations according to defensible criteria, process images, calculate derived variables, and conduct alternative analyses. A transformation is not automatically falsification merely because the final dataset differs from the raw dataset.

A Useful Diagnostic Question: Did the Evidence Exist?

When trying to understand the conceptual difference, start by asking what happened to the underlying evidence.

If the observation or result never existed and was nevertheless created and recorded or reported, the conduct points toward fabrication. If genuine research existed but something about its materials, process, data, results, or representation was manipulated in a way that made the research record inaccurate, the conduct points toward falsification.

The Difference Is Not Simply “Fake Data” Versus “Changed Data”

Everyday descriptions can obscure the regulatory distinction. Calling both practices “fake data” may communicate that something is seriously wrong, but it does not explain how the research record became unreliable.

Similarly, “changed data” is too broad to define falsification. Researchers often have legitimate reasons to change a dataset. A transcription error may be corrected against the original record. A variable may be recoded according to a documented analysis plan. An observation may be excluded because it satisfies a valid exclusion criterion. These actions alter data without necessarily misrepresenting the research.

The relevant distinction is therefore not merely whether a file changed. The question is whether the conduct falls within the applicable definition of falsification and, for a formal misconduct finding, satisfies the additional requirements imposed by the governing policy.

Fabrication and Falsification Can Occur Before Publication

Neither concept is confined to a published journal article. Under the current U.S. Public Health Service framework, research misconduct encompasses fabrication, falsification, or plagiarism in proposing, performing, or reviewing research or in reporting research results. The research record can extend well beyond the final paper.

This matters because researchers sometimes imagine misconduct as something that occurs only when false information reaches a journal. An invented result recorded during research may already raise a fabrication issue even if the manuscript is never published. Likewise, manipulation of a research process or research record can raise falsification concerns before publication.

Not Every Incorrect Research Record Establishes Research Misconduct

Classification of an act and a formal finding of research misconduct are related but distinct questions. Under the current PHS regulation, a finding of research misconduct requires a significant departure from accepted practices of the relevant research community, conduct committed intentionally, knowingly, or recklessly, and proof by a preponderance of the evidence.

This is why an anomalous number, an incorrect figure, or an undocumented change should not automatically be described as proven misconduct. Evidence may instead reveal an honest transcription mistake, software error, reasonable methodological disagreement, or another problem that requires correction without satisfying the requirements for a misconduct finding.

Some Cases Sit Near the Boundary

Real cases do not always arrive neatly labeled. Suppose a genuine observation is missing and a researcher types a value that seems plausible. Is that an invented observation, an alteration of an existing record, an improper reconstruction, or something else? The facts matter, including what the value purports to represent and how it was generated and documented.

That is why situations such as filling in a missing value from memory require closer analysis rather than a reflexive label.

Similar boundary questions arise when researchers remove observations, change coding decisions, alter analyses, or modify figures. Those practices can be legitimate. They can also become misleading when the resulting record no longer accurately represents the research. The category cannot be determined simply from the fact that a researcher made a change.

Two Researchers, Two Different Problems With the Research Record

Common Misunderstandings About Fabrication and Falsification

How to Think Through a Possible Fabrication or Falsification Problem

If you encounter a suspicious value, missing observation, changed figure, altered dataset, or unexplained analytical decision, resist the temptation to begin with a misconduct label. Begin with the research record.

Good recordkeeping becomes particularly important here. Raw data, instrument outputs, analysis scripts, audit trails, version histories, laboratory records, coding documentation, and original images may help establish whether a disputed value was genuinely observed, subsequently transformed, incorrectly entered, or invented.

If you discover a genuine error in your own work, hiding it out of fear that a correction will look suspicious can make matters worse. Researchers can correct genuine data errors without treating every correction as manipulation. What matters is preserving the original evidence, documenting the basis for the correction, and ensuring that the corrected record accurately represents the research.

For consequential concerns involving another researcher, follow the relevant institutional process rather than attempting to adjudicate intent from a spreadsheet or figure alone. A discrepancy may justify scrutiny. A formal finding requires evidence.

Before Calling Something Fabrication or Falsification, Check the Record

Frequently Asked Questions About Fabrication and Falsification

Fabrication Invents; Falsification Misrepresents

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.