Fraud & How Science Self-Corrects
The line that began with honest flexibility and ran through questionable research practices ends in outright fraud: making the data up. Fraud is rare but does real damage, and science has a growing set of tools for catching it. This lesson covers what fraud is, how one of its most famous cases came to light, and how the record gets corrected. Then you try one of the arithmetic checks that can flag a fabricated result from the published numbers alone.
The two Fs: fabrication and falsification
Research misconduct has a narrow definition, usually with three parts. Fabrication is inventing data or results that were never collected. Falsification is altering real data: changing values, deleting inconvenient cases, doctoring an image. The third is plagiarism, covered in its own lesson. The first two are about the numbers themselves: reporting findings that don't match what actually happened. That puts them in a different category from the QRPs covered earlier. QRPs bend the analysis of real data, while fraud reports data that never existed.
Anatomy of a fraud: the Stapel case
Diederik Stapel was a celebrated Dutch social psychologist with dozens of clean, elegant, widely cited papers. The results were clean because for years he invented his datasets, giving collaborators and students spreadsheets of data that no participant had ever produced. There were warning signs: effects that were suspiciously large and tidy, and studies whose "data" arrived without the messiness of real collection. In the end it was junior colleagues and students who raised the alarm, not an automated system. Investigations led to more than fifty retractions. The case teaches two things. Fabricated data often looks better than real data, because a fabricator produces the result they expect, without the noise of real measurement. And the people best placed to notice are insiders, so it matters that students can speak up safely.
Correcting the record: retractions
When a published result can't be trusted, whether through fraud, a serious honest error, or unreliable data, the formal fix is a retraction: the paper is withdrawn from the scientific record. It usually stays visible online (stamped "RETRACTED", linked to a notice explaining why) so the correction is transparent, but it should no longer be cited as evidence. A retraction does not mean fraud: plenty of retractions are honest mistakes, and a good notice says which. Services like Retraction Watch catalog them. The annual count has climbed steadily, which suggests that detection and post-publication scrutiny are improving, not necessarily that misconduct is rising.
Whistleblowing and its costs
Insiders catch most fraud, so self-correction depends on people who are willing to raise concerns, and raising them is rarely free of cost. Whistleblowers, often junior, risk their relationships, their references, and sometimes their careers to report someone more powerful, often through months of uncertainty before anything is confirmed. The Stapel case came to light because students took that risk. Institutions increasingly provide protected channels for raising integrity concerns, so that reporting misconduct is safe for the person who does it.
How fabricated numbers are detected
Fabricated numbers can often be detected, and anyone can learn the checks. A family of consistency checks tests whether published statistics obey the rules of arithmetic:
- GRIM checks whether a reported mean of whole-number data is even possible for its sample size (you'll try it below).
- statcheck recomputes p-values from the reported test statistic and degrees of freedom, flagging results that don't add up. The APA results formatter runs this check on anything you paste into it and warns when the p you reported and the p implied by your statistic and degrees of freedom disagree.
- Data forensics looks for too-even digit patterns, duplicated rows, or variances that are implausibly small for real measurement.
None of these proves fraud on its own, since an inconsistency can be a typo or a rounding slip. But they narrow the search, and they make fabricated data more likely to be caught than it used to be.
Try the GRIM test
A mean is a total divided by the number of responses. If every response is a whole number (a count, or a single Likert item), then the total is a whole number too. So the only means that can occur are fractions with denominator n. A reported mean that falls in the gap between two of those reachable values cannot have come from that many whole-number responses. Enter a reported mean and its sample size to see which values are reachable.
🔢 The GRIM test
For a mean of whole-number responses (a count, or one Likert item): is that mean even arithmetically reachable?
What GRIM can and cannot tell you. GRIM can't tell you why a mean is impossible, because a typo and a fabrication look identical from the outside. It can tell you that a reported number could not have come from the sample it claims. All the detection tools work this way: they check, cheaply and publicly, whether the numbers are internally consistent, without accusing anyone.
Why it matters: this lesson does not show that science is full of frauds. Fabrication is rare, most researchers are honest, and the system catches more of the fraud there is than it used to. Transparency, open data, protected whistleblowing, and consistency checks keep it that way, because they make dishonest numbers likely to be caught. For your own work, the lesson is to report real numbers and share your data.
Common questions
I think I have found an error in a published paper. What should I do?
Assume a mistake before you assume misconduct, because the great majority of errors are mistakes. Act in a way you would be comfortable defending if you turn out to be the one who is wrong. First write the problem down precisely: which number, on which page, why it cannot be right, and what you did to check. A vague suspicion is no use to anyone. Then contact the corresponding author directly and neutrally, and ask whether you have misread something. Give them a real chance to answer, since typos and mislabeled tables are the usual explanation and authors generally want them fixed. If there is no response, or the answer does not hold up, contact the journal editor next. PubPeer exists for public post-publication comment. Keep the discussion about the arithmetic, not the person. If you are a student, talk to a supervisor you trust before you send anything, both for advice and because raising a concern about a more senior researcher can be costly for you.
What happens when a paper is retracted?
A retraction is the formal withdrawal of a published paper from the scientific record because it can no longer be trusted, whether through fraud, a serious honest error, or unreliable data. The article usually stays online so the record is transparent, but it's stamped 'RETRACTED', linked to a notice explaining why, and should no longer be cited as valid evidence. Retraction is not the same as an accusation of fraud: many retractions are for honest mistakes, and the notice ideally says which. Databases like Retraction Watch track them. The number has grown as detection tools and post-publication review have improved, so the rise reflects more problems being caught, not necessarily more misconduct. If you find a key source has been retracted, don't build on it.
How common is research fraud?
Outright fabrication or falsification is rare. In anonymous surveys pooled by Fanelli (2009), about 2% of scientists admitted to having fabricated, falsified, or modified data at least once, and higher fractions said they had seen colleagues do it. A far more common problem is the gray zone of questionable research practices (selective reporting, optional stopping, HARKing). Many more researchers admit to these, and together they do more damage to the literature than the rare fraud case. Fraud is uncommon and most researchers are honest. The fixes are structural: better incentives, transparency, open data, and detection tools make fabricated results increasingly likely to be caught.