There's a big research scandal that's been coming on for a while in the Netherlands. I haven't covered it, since it's in academic psychology, and thus pretty far out of my field, but some of the behavior in it is well worth noting. Here's a rundown from Science on Diederik Stapel, who appears to have forged scores of studies over the years. It seems that he rarely bothered to collect actual data, preferring to just condense it out of thin air to save time and effort:
Many of Stapel's students graduated without having ever run an experiment, the report says. Stapel told them that their time was better spent analyzing data and writing. The commission writes that Stapel was "lord of the data" in his collaborations. It says colleagues or students who asked to see raw data were given excuses or even threatened and insulted.
Here's more from Nature, via Scientific American. At least two earlier groups of whistleblowers had raised questions about Stapel's work, the commission found. No one followed up on their concerns, however. . .His colleagues, when they failed to replicate the results, tended to blame themselves, the report says. Among Stapel's colleagues, the description of data as too good to be true "was a heartfelt compliment to his skill and creativity," the report says.
Well, he did have skill and creativity - I mean, we're talking about someone who's published over 150 papers in the last seven years. And that means that he wasn't lazy, either, because keeping that many balls in the air is no small job. No, what we have here is an industrious, committed, fraud with a real talent for his chosen line of work: fakery. I'd like to think that it's somewhat easier to get away with this (for this long) in a social science field, but then, there's Jan-Hendrik Schön to think about. So I'm not sure that my high ground is all that high.
The same old lessons apply (". . .And the Gods of the Copybook Headings / limped up to explain it once more"): if something looks too good to be true, it probably is. If multiple people can't reproduce what's supposed to be a scientific result, then there may well be something wrong with it. And if someone that you're working for or with won't show you raw data, then head for the door, because nothing but trouble can ensue. The alarm bells have at that point gone off, whether you can hear them or not.
1. KK on November 2, 2011 8:04 AM writes...
Where's the raw data in global warming?
Permalink to Comment2. 4merchemist on November 2, 2011 8:23 AM writes...
This is covered in Reuters:
http://www.reuters.com/article/2011/11/02/us-dutch-scientist-fraud-idUSTRE7A12PL20111102
Permalink to Comment3. Kevin on November 2, 2011 9:00 AM writes...
Large amounts of processing and analysis also make me worry. If the data requires "meta-analysis" or extreme modeling to tease an answer out, I worry. Of course, as a chemist, that also implies poorly designed experiments.
Permalink to Comment4. John Wayne on November 2, 2011 9:07 AM writes...
Whoa! I've been making molecules, collecting data and reporting it without distortion like a chump.
Permalink to Comment5. sara on November 2, 2011 9:13 AM writes...
what about some jail time for this fraud?
Permalink to Comment6. Mutant Dragon on November 2, 2011 9:41 AM writes...
I guess "skill and creativity" would be one way to put it. It's amazing it took this long to catch him. I mean, telling his students their time was "better spent" analyzing and writing rather than doing experiments? insulting colleagues who asked to see raw data? You'd think that someone would have smelt a rat long before now.
Permalink to Comment7. hanguphile on November 2, 2011 9:51 AM writes...
Very easy to do. If I WANTED to be a dishonest spectroscopist, I could manufacture spectra, claim to have measured them, put the measured frequencies just that LITTLE bit away from predictions enough to make everyone wonder what was wrong in the theory... And trust me, I know enough about error analysis and statistics to make the data impervious to reanalysis. The only defence against fraud is to offer to re-obtain data under observation. Possible in the case I just postulated (although bloody irritating), but probably not in the social sciences.
Permalink to Comment8. nitrosonium on November 2, 2011 10:35 AM writes...
i will forever be an advocate for Org. Synth. type of "submitted by" and "checked by" protocols....at least for our field. i know that would affect the bandwidth of published methods but would not the quality go up?? also the reliability and reproducibility??
Permalink to Comment9. NJBiologist on November 2, 2011 11:11 AM writes...
@Derek: "I'd like to think that it's somewhat easier to get away with this (for this long) in a social science field, but then, there's Jan-Hendrik Schön to think about."
How about Joachim Boldt? The guy invented and published clinical data, and nobody at the hospital where he worked seems to have figured it out.
While reproducing work is always the best check (@8), it's less and less practical as the experiments get more and more expensive.
Permalink to Comment10. MTK on November 2, 2011 11:35 AM writes...
I honestly don't get why people go through so much trouble to spin a believable yarn when they could actually do the experiments with the same effort.
That aside, scientific fraud happens. Luckily for us there are usually one of two outcomes of any fraud of this nature:
a) the results are important enough that others try to repeat it, can't, and the fraud is revealed,
b) the results are unimportant enough that nobody cares.
Only rarely do we have cases like Cyril Burt's where studies based on most likely fraudalent data end up effecting in policy change.
Permalink to Comment11. pete on November 2, 2011 12:42 PM writes...
This motif, of Stapel saying he'd gotten "the DATA" from his network of outside collaborators, and then brought it in-house for his students to analyze, is pretty nifty. Less likely that anyone in his group would smell a rat, especially if the numbers were suggesting interesting patterns.
I see parallel hazards in all the out-sourced research in our business. Tempting for some CRO to "give the Pharmas what they want".
Permalink to Comment12. anonymous on November 2, 2011 12:44 PM writes...
a colleague of Marc Hauser?
Permalink to Comment13. Dave on November 2, 2011 1:23 PM writes...
Cold fusion?
Permalink to Comment14. Sisyphus on November 2, 2011 6:14 PM writes...
Seems like he would be a perfect fit for Wall Street or Washington D.C.
Permalink to Comment15. gippgig on November 2, 2011 10:29 PM writes...
Scientific fakery would be a good subject for psychologists to study.
Permalink to Comment16. eugene on November 3, 2011 4:40 AM writes...
"Scientific fakery would be a good subject for psychologists to study."
This talk is from another social scientist and it come pretty close to explaining the motives if you change the subject matter.
hxxp://www.ted.com/talks/dan_ariely_on_our_buggy_moral_code.html
(change the xx to tt)
Permalink to Comment17. dearieme on November 3, 2011 6:07 AM writes...
"While reproducing work is always the best check..", it won't win you tenure or promotion.
Perhaps what's needed is a Federal Office of Reproducibility to do the repeats. That'll fix it.
Permalink to Comment18. Spiny Norman on November 3, 2011 9:07 PM writes...
@KK: You can start here. That's a vastly less stupid answer than your deeply stupid question deserves.
Permalink to Comment19. Vigilantchemist on November 6, 2011 9:22 AM writes...
For an open case of forgery (and plagiarism) see the note at the end of this paper DOI: 10.1021/ol202053m). HOW IN THE HELL CAN SOMEBODY MISTAKENLY USE THE SPECTRA FROM PREVIOUS A PREVIOUS PAPER?
Permalink to CommentAnd now the best part....the paper still has a couple of spectras which are copied from 'another' previous paper. My guess is that the graduate student did not really make the molecule and used the material from a previous synthesis from their group.