Data fabrication vs falsification

The distinction is between making it up and doctoring it. Fabrication means reporting measurements, patients, or experiments that do not exist. Falsification means altering or selectively omitting genuine data, such as trimming inconvenient data points, manipulating images, or misrepresenting methods, so the findings look cleaner or stronger than the truth. Both intentionally corrupt the scientific record.

These are the F and F of FFP, the fabrication, falsification, and plagiarism triad that funding bodies define as misconduct, distinct from honest error or questionable but non-fraudulent practices like p-hacking. Detection often comes from statistical anomalies, impossible values, duplicated or spliced images, or whistleblowers, and forensic tools increasingly flag image manipulation that once slipped through.

Peer review catches fabrication and falsification only sometimes, since reviewers generally cannot access raw data. Data availability requirements, statistical checks, and image-integrity screening strengthen the net, and confirmed cases typically lead to retraction and institutional investigation.