The experiments were always elegant. A railway station in Utrecht, its stairwells strewn with trash during a cleaners’ strike, where white commuters chose to sit farther from a Black stranger when the platform was dirty — disorder, the study concluded, promotes discrimination. A questionnaire showing that meat eaters were more selfish than vegetarians. A finding that a messy desk makes the mind hungry for structure and therefore for stereotype. The studies of Diederik Stapel, professor of cognitive social psychology and dean of the School of Social and Behavioral Sciences at Tilburg University, had a quality his colleagues envied and journalists adored: the world’s moral texture, confirmed by data, packaged in a headline. The station study ran in Science in April 2011, and newspapers around the world wrote it up. By December it was retracted, because the cleaners’ strike may have happened but the experiment never did. Stapel had invented the data, as he had been inventing data for at least fifteen years — at his kitchen table, at night, alone.
The Stapel affair holds a particular place in the history of scientific fraud, and not because of its scale, though the scale was extraordinary: fifty-eight retracted publications, a record in psychology, and ten doctoral dissertations built on data their authors never knew were counterfeit. What distinguishes the case is its diagnostic value. When three official committees spent a year excavating Stapel’s career, they concluded that the fraud had succeeded not despite the norms of his field but, in uncomfortable measure, because of them — and their report remains the most unsparing audit any scientific discipline has published about itself.
The Ascent
Stapel was, by every conventional signal, a superb scientist. Born in 1966, trained at the University of Amsterdam, he took his doctorate cum laude in 1997, became a full professor at Groningen by his mid-thirties, and moved to Tilburg in 2006, where he founded a behavioral-economics research institute and rose to dean. He was charming, generous with students, skilled at the conference performance, and staggeringly productive; in 2009 the Society of Experimental Social Psychology gave him its Career Trajectory Award. His specialty was the crisp, counterintuitive effect — priming studies in which some incidental feature of the environment reorganized judgment and behavior. His data had a signature quality that colleagues would recall, later, with the particular bitterness of hindsight: they almost always worked. Hypotheses were confirmed on the first attempt. Effect sizes were healthy. Outliers were rare. Students called it the Midas touch, and the field’s journals, hungry for exactly such results, published them at a pace that made the touch self-financing: each clean finding bought the grants, the graduate students, and the editorial goodwill that the next clean finding would repay.
The method, as Stapel eventually confessed and the committees documented, was a fraud of unusual purity. In the early years he massaged real datasets — an adjusted number here, a deleted subject there. Then he stopped bothering with subjects altogether. He would design an experiment in full with a student or colleague — hypotheses, conditions, questionnaires, the whole apparatus — and then announce that he would handle the data collection himself, through his network of schools and testing sites that collaborators were never invited to visit. Weeks later a dataset would arrive, typed by Stapel into a spreadsheet at home, its numbers built to deliver the agreed hypothesis with publishable cleanliness. Doctoral students analyzed the files gratefully; they had no reason to imagine that a dean was manufacturing their dissertations, and the division of labor — senior man controls the data, junior people write it up — struck no one as irregular. That was the detail with the longest shadow: it struck no one as irregular.
The counterfeits were good, but they were counterfeits, and they carried the die marks of their manufacture. When investigators later subjected the datasets to forensic statistics, the tells were everywhere: effect sizes implausibly large and implausibly uniform; variances too similar across independent samples; means that repeated to the decimal across studies that supposedly involved different subjects in different years; and, in places, entire columns of numbers duplicated from one experiment to another. Real data are ragged in ways that are hard to fake, because the human mind, asked to produce randomness, produces pattern. Stapel, by his own later account, understood the aesthetics of data better than their arithmetic — he built numbers that looked persuasive to a reader, not numbers that could survive an auditor, because in fifteen years no auditor ever came.
The Whistleblowers
He was caught from below. In the late summer of 2011, three young researchers in Stapel’s Tilburg group — junior enough that exposure could have ended their careers, and kept anonymous in the official record for that reason — assembled their suspicions: datasets too clean, experiments no one had witnessed, a schools network no one could contact, and, fatally, patterns in the numbers themselves, including rows of data that repeated across supposedly independent studies. They took the case to the head of the department, who took it to the university’s rector. On September 7, 2011, Tilburg suspended Stapel and announced, with a bluntness universities rarely manage, that it had established serious research fraud. Stapel confessed within weeks. “I have failed as a scientist,” he said in a statement that October, admitting that he had fabricated data for years and expressing shame; he returned his doctorate to the University of Amsterdam voluntarily, before anyone formally asked.
Three committees were convened, one for each university where he had worked — the Levelt Committee at Tilburg, the Noort Committee at Groningen, the Drenth Committee at Amsterdam — and their joint final report, published in November 2012 under the title “Flawed Science,” is the central document of the affair. The committees examined his entire output, interviewed dozens of collaborators, and subjected the datasets to statistical forensics. Their conclusion: fraud was established in at least fifty-five publications, with more suspected but unprovable, and in material used in ten dissertations. The retraction count in the journals eventually reached fifty-eight. The committees were explicit that the doctoral students and nearly all co-authors were victims, not participants — Stapel had deceived his collaborators with the same care he applied to journals.
The Indictment of a Field
Had the report stopped there, the case would be a story about one man. It did not stop there. The committees asked the obvious question — how does a researcher publish invented data for fifteen years, in the field’s best journals, without detection? — and their answer indicted the ecosystem. The failure, they wrote, was not Stapel’s alone: “the critical function of science has failed on all levels.” Reviewers at leading journals had accepted results too good to be true without asking to see raw data, and had sometimes encouraged authors to drop inconvenient conditions and tidy narratives. Co-authors had signed papers whose data they had never touched. The field’s incentive structure — publish striking effects, fast, in quantity — rewarded exactly the properties that fabricated data possess and real data stubbornly lack. Around the central fraud, the committees documented a gray zone of what they called sloppy science: selective reporting, flexible analyses, results-driven tinkering — practices that were not fraud but that made fraud statistically invisible, because everyone’s published data looked a little too clean.
The timing gave the report unusual force. It landed in the middle of what psychology was already calling its replication crisis — the same year as the Bem precognition controversy and the exposure of fraud by Dirk Smeesters, another Dutch social psychologist, and shortly before large-scale replication projects began reporting that a substantial fraction of the field’s textbook effects could not be reproduced. Stapel became the crisis’s emblem: not its cause, but its limiting case — proof that the field’s quality-control machinery could not distinguish invented data from real data for a decade and a half. The reform movement that followed — preregistration of hypotheses, open data and materials, registered reports, the routine forensic statistics that now scan for impossible means and duplicated numbers — drew explicitly on the committees’ recommendations. It is a fair summary of the last decade of methodological reform in psychology that much of it is infrastructure designed to make the next Stapel arithmetically impossible.
The report’s prescriptions were as concrete as its diagnosis. Raw data, the committees insisted, must be archived and available — to co-authors first of all, and to other researchers after publication; a professor whose datasets no collaborator may inspect is not a data collector but a black box. Journals should require authors to state who gathered the data and where. Universities should maintain functioning channels through which junior researchers can raise integrity concerns without staking their careers. And the field should treat its methodological gray zone — the flexible stopping rules, the unreported failed studies, the hypotheses quietly reverse-engineered from results — not as venial habit but as the medium in which fraud becomes undetectable. Much of this, within a few years, hardened into the reform infrastructure now standard in psychology; the Stapel report is cited in the founding documents of most of it.
The Reckoning
The formal consequences were, by the standards of white-collar accountability, modest and swift. Stapel was dismissed from his chair and his deanship. He surrendered his doctorate. Dutch prosecutors opened a criminal inquiry — the fabricated research had been funded in part by public grant money — and in June 2013 announced a settlement: 120 hours of community service and the forfeiture of benefits equivalent to a year and a half of salary, in exchange for no prosecution. The prosecutors noted, in explaining the deal, that Stapel had confessed, cooperated fully, returned his degree, and been destroyed professionally in full public view — a punishment no court needed to supplement. The ten doctoral students whose dissertations rested on his numbers were permitted to keep their degrees; the committees judged them blameless, though their published work evaporated beneath them, and several left science.
Stapel’s own accounting arrived quickly, in a memoir written during the year of his exposure. “Ontsporing” — “Derailment” — published in Dutch in late 2012 and later translated into English by a volunteer researcher, is a strange, self-lacerating document, part confession and part case study, in which Stapel describes the fraud’s beginnings in small adjustments to disappointing data and its progression into wholesale invention, driven, he writes, by an addiction to elegance and applause: the beautiful result, the confirming world, the next talk. “I wanted too much, too fast,” he told an interviewer for a 2013 New York Times Magazine profile, in which he also demonstrated, on request, how he had typed his datasets into being at the kitchen table while his family slept. Researchers who study misconduct treat the memoir warily — con men writing about their cons remain con men writing — but it is nearly unique in the literature: a full technical confession by a major fraud, describing the mechanics from inside.
The affair also invited the grim exercise of ranking. Science had lately survived Jan Hendrik Schön, the Bell Labs physicist whose fabricated breakthroughs produced dozens of retractions, and Hwang Woo-suk, the Korean cloning researcher whose fraud reached the covers of the world’s journals; Harvard’s Marc Hauser had resigned in 2011 after a federal misconduct finding. Stapel’s fifty-eight retractions placed him among the most retracted scientists ever recorded, and gave psychology the unwelcome distinction of housing one of research fraud’s largest bodies of work. But the comparisons clarified something the raw count obscured: Schön and Hwang had claimed discoveries that other laboratories would inevitably try and fail to replicate, which is how both fell. Stapel worked in a field where exact replication was rare, unglamorous, and nearly unpublishable — a field whose economy of novelty meant that no one’s career depended on rebuilding his experiments. His longevity was not cunning. It was ecology.
His life afterward became its own odd document. He gave interviews with a candor that unsettled interviewers, collaborated with the Dutch novelist A. H. J. Dautzenberg on a book of letters, and appeared before student audiences to explain, patiently, how he had deceived people exactly like them. Some in the field found the second act unseemly — the arsonist lecturing on fire safety, monetizing the notoriety his victims paid for. Others noted, grudgingly, that almost no major scientific fraud has ever explained itself at all, and that Stapel’s testimony, whatever its motives, is among the only inside accounts the research-integrity literature has. Both things are true, which is the usual condition of the case: it resists the comfort of a simple villain almost as stubbornly as it resisted detection.
What the Case Settled
The Stapel affair settled, as well as any single case can, a set of arguments about scientific fraud that had persisted since the fabrications of the physicist Jan Hendrik Schön a decade earlier. It established that peer review offers no protection whatsoever against invented data — reviewers audit inference, not existence. It established that co-authorship without data access is a signature on a document one has not read. It established that the most reliable detection instrument in science is the junior researcher with statistical literacy and a conscience, and that institutions therefore get the integrity they make safe for the powerless to defend. Tilburg’s whistleblowers succeeded because a department head and a rector acted on their word against the most decorated man on the faculty; the committees pointedly observed how easily it could have gone the other way, and in most such cases — the record of research-integrity retaliation suggests — it does.
What the case could not settle is the question Stapel himself embodies: what, exactly, the fraud was for. There was no patent, no company, no expert-witness fee. He counterfeited the only currency his profession recognizes — the confirmed hypothesis — and spent it on the profession’s own rewards: chairs, awards, keynotes, the deanship, the newspaper write-ups of the station study. The committees, searching for a motive, found only the incentive structure itself, operating on a man who wanted its prizes more than its purpose. That is why “Flawed Science” reads less like the autopsy of a criminal than like the environmental-impact statement of a reward system — and why, a decade of reform later, the case is still taught not as the story of a uniquely bad man, but as the story of a field that had built a machine for believing beautiful numbers, and of the dean who understood the machine well enough to feed it.
Sources: “Flawed Science: The Fraudulent Research Practices of Social Psychologist Diederik Stapel,” final joint report of the Levelt, Noort, and Drenth Committees (Nov. 28, 2012); Tilburg University statements of Sept.-Oct. 2011; Science retraction of Stapel & Lindenberg, “Coping with Chaos” (Dec. 2011); Retraction Watch’s running tally of Stapel retractions (58); Dutch Public Prosecution Service settlement announcement (June 2013); Diederik Stapel, “Ontsporing” (Prometheus, 2012), translated by Nicholas J. L. Brown as “Faking Science”; Yudhijit Bhattacharjee, “The Mind of a Con Man,” New York Times Magazine (Apr. 26, 2013); contemporaneous coverage by Nature, Science News, the Guardian, and NRC Handelsblad.