Fabrication

Fabrication is the most clear-cut form of research misconduct because the evidentiary chain is entirely invented. It is distinct from falsification, where real data exist but are manipulated. Common indicators include data values that lack the random variance expected from genuine measurement, instruments that show no usage logs for the recorded period, and participants who cannot be matched to recruitment records.

Legal definition: 42 CFR § 93.211

In US federally funded research (Public Health Service-funded work overseen by the Office of Research Integrity), fabrication has a precise regulatory meaning. 42 CFR § 93.211 defines it as: “Fabrication means making up data or results and recording or reporting them.” That short definition does most of the work: fabrication requires that no underlying experiment, observation, measurement, or survey response actually occurred. If real data were collected and then altered, that is falsification (42 CFR § 93.212) instead — a related but legally and practically distinct act.

Fabrication’s place in the FFP framework

US federal regulation limits the definition of research misconduct to three acts, commonly abbreviated FFP: Fabrication, Falsification, and Plagiarism. Each has its own numbered definition in 42 CFR Part 93, Subpart B (§ 93.211 Fabrication, § 93.212 Falsification, § 93.227 Plagiarism), and the umbrella term Research Misconduct is defined separately at § 93.234. The regulation deliberately excludes honest error and legitimate differences of scientific opinion — misconduct requires more than being wrong. For a side-by-side comparison of the two most commonly confused categories, see Falsification vs. Fabrication.

Fabrication is generally treated as the most straightforward FFP finding to establish once discovered, because there is no genuine underlying dataset to argue over the interpretation of — the dispute is simply whether the reported observation happened at all. That does not make it easier to detect; well-executed fabrication can pass casual review for years before statistical anomalies, failed independent replication, or a whistleblower complaint surfaces it.

How a fabrication finding is made

A federal fabrication finding is not just “the data looks suspicious.” Under 42 CFR § 93.103, an institution or ORI must establish all three elements before a finding of research misconduct (including fabrication) can be made:

• The act is a significant departure from accepted practices of the relevant research community;
• The misconduct was committed intentionally, knowingly, or recklessly (not through honest error); and
• The allegation is proven by a preponderance of the evidence.

Institutionally, this plays out as a two-stage process — an initial inquiry to decide whether a full investigation is warranted, followed by the investigation itself, both administered by the institution’s Research Integrity Officer. See How a Research Misconduct Investigation Actually Works for the full procedural walkthrough.

Consequences of a fabrication finding

Where PHS funding is involved, ORI can impose administrative actions independent of — and often in addition to — whatever action the employing institution takes, which can include: debarment from applying for or receiving federal funding for a defined period; supervision requirements on any future PHS-supported research; a certification requirement for future publications; and a requirement that the researcher submit corrections or retractions to journals that published the fabricated data. Journals themselves, guided by COPE and ICMJE guidance, treat a confirmed fabrication finding as grounds for retraction independent of any federal action, since the published record itself is no longer reliable.

Documented cases

Fabrication findings are relatively rare compared to the volume of published research, but several closed, publicly adjudicated cases illustrate the pattern: fabricated patient-derived stem cell line results that led to journal retractions and a national research scandal (Hwang Woo-suk, South Korea, 2005-2006), and a PHS-funded researcher found by ORI to have fabricated and falsified data across roughly ten publications and seventeen grant applications tied to several million dollars in federal funding (Eric Poehlman, University of Vermont, ORI finding 2005). For a broader set of adjudicated examples and what made each detectable, see Research Misconduct Case Studies: What Real Investigations Reveal.

Frequently asked questions

Is fabrication the same as falsification?

No. Fabrication means the underlying data never existed at all — it was invented outright. Falsification means real data or results exist but were manipulated, omitted selectively, or altered so the record no longer reflects what actually happened. Both are separately defined under 42 CFR Part 93 and both count as research misconduct, but they are proven differently: a fabrication case typically turns on showing an experiment or observation could not have occurred as reported; a falsification case typically turns on comparing reported results against original raw data, instrument logs, or lab notebooks.

Does the fabrication definition only apply to federally funded research?

The 42 CFR Part 93 definition is scoped to Public Health Service-supported research (including most NIH funding) and to institutions that receive it, since ORI’s jurisdiction derives from that funding relationship. In practice, most US research universities apply the same FFP definition institution-wide in their own misconduct policies regardless of funding source, and journal publishers apply an equivalent standard through COPE and ICMJE guidance. Other countries and funders (for example UKRIO in the UK, or ALLEA’s European Code of Conduct for Research Integrity) define research misconduct somewhat more broadly than the strict US FFP definition — see the Research misconduct entry for that comparison.

Is fabricating a small, “harmless” data point still misconduct?

Yes, if the three-part test above is met. The regulation does not include a materiality threshold based on how much the fabricated element affected the paper’s conclusions — a single invented data point recorded and reported as real can meet the definition, though the scale and impact of the fabrication typically affects the severity of the resulting sanctions.

References

• 42 CFR Part 93, Subpart B § 93.211 (Fabrication), § 93.212 (Falsification), § 93.227 (Plagiarism), § 93.234 (Research misconduct), § 93.103 (Findings)
• US Office of Research Integrity, 42 CFR Part 93 (2005)
• Singapore Statement on Research Integrity (2010)