Bioscientists photoshop their cultures to fake results

Jeff sez, "Researchers often use Photoshop to clean up the images they produce in the laboratory. If the experiment didn't go quite right, a bit of tampering can make a gel look like things did work. Editors at Science, Nature, and other journals are turning into detectives, using new tools to hunt for fraudulent images."

And the level of tampering they find is alarming. "The magnitude of the fraud is phenomenal," says Hany Farid, a computer-science professor at Dartmouth College who has been working with journal editors to help them detect image manipulation. Doctored images are troubling because they can mislead scientists and even derail a search for the causes and cures of disease.

Ten to 20 of the articles accepted by The Journal of Clinical Investigation each year show some evidence of tampering, and about five to 10 of those papers warrant a thorough investigation, says Ms. Neill. (The journal publishes about 300 to 350 articles per year.)

Link (Thanks, Jeff!)


  1. Right you are, RotWang! Chain of custody (of the pictures, not just the samples) becomes all-important, and the integrity of the researchers becomes fair game for cross-examination.

    What a nightmare.

  2. Just print and rescan the altered image, right? What little understanding I have of fraudulent image detection seems like it could be rendered ineffective by this method.

  3. I think we’re missing the bigger picture… what the real travesty is here… is how this is all only an issue in the first place because of poor retouching skillz. L2photoshop?

  4. Damnit. Just looking at that I can think “The telltale is the lack of phoresis smear on the tail of the photoshopped area.” – and the next fraud will include a phoresis smear.

  5. Yeah, DNA evidence is great! I usually pose as a plumber and collect enough for the frame-up on my usual jobs.

  6. Rotwang, it’s true..

    I just had a look for a similar article, but for the life of me I can’t find it. It was about the UK forensic lab that handled the IRA cases. A few years ago (5-10ish) they discovered contamination on some piece of equipment (a centrifuge I think) that had been used to convict people in many similar cases.

    So years of evidence was called in to doubt, I’m pretty sure it led to (or came from, and led to more) quashing of cases and immediate releases.

  7. I had a work-study job in college washing flasks in a university research lab. I now refer to it as the ‘Laboratory of the Damned’, because it was a nightmare of iffy results and not-quite-the-whole-truth papers. The researchers were always griping about how they could only submit un-retouched photos, while NASA got to enhance their images. While they whined, I was always thinking, “You guys are EXACTLY why journals have standards. Your fudged results are EXACTLY what should be weeded out.” They were clumsy and sloppy with their research, I imagine that their attempts to fake pictures would be just as easily spotted.

  8. That’s kind of pointless to make up data.

    I never make up my data and I will never will.

    So where is the fun of running experiments then?

    I run experiments because I want to find something new. Most of the time it does not work the way I expect it to work. But hey! I am not trying to confirm others people results because I want to find something really new. Granted it is harder to get funded for that.

    The most boring experiment is an experiment that try to reproduce the result of someone else or try to confirm a consensus.

    Often your experiment does not work and you can not conclude anything. This is ok. It is the way it work.

    When it’s work meaning you get some results it is often not what you expect and it is where it really get exciting. Do not assume necessarely that that they must be something wrong with your experiment. Trust your results and trust yourself. Do not make up results to try to conform and avoid pissing off people.

    Many reviwers are guilty of bias and dishonesty. They are partial and when experiments end up with a conclusion that seem to antagonize their one stuff or a general consensus they reject the paper.

    I know. It make difficult to get your paper accepted but hey the results is the result and Fuck these stupid and /or dishonest reviewers!

    Science is not politic. For those who think it is then may be you should change job!

    Stand by your results!

    And trust me. when two results that seem to be correct also seem to conflict there is usually something new to be found!

  9. The larger issue this illustrates is the widespread and total disregard for the scientific method in the modern scientific community. The purpose of testing a hypothesis is to disprove, not to prove – then if your theory happens to hold up to the rigorous scrutiny you have subjected it to, you may be getting somewhere. But nowadays, scientist are more inclined to persuade, to infer, to fudge, to smear, then to rigorously follow the scientific method as they are professionally and morally bound.

    My real pet peeve are “studies” which identify valid statistical correlations, and then – seemingly out of the ether – infer some totally unproved common-sense reason to justify the relationship. A recent study of the connection of lead levels in infancy to crime comes to mind, although there are an infinite number of examples. The result of all this is worse than hearsay or old wives-tales because it is all done in the name of empiricism and somehow blindly accepted by society as truth.

    While science may be “objective”, humankind is rather the opposite, and the very honorable principles once developed to counteract their subjectivity are being wholly ignored by some, which is quite disturbing, at least to me. Why? Because the result are often dangerous, ridiculous, and – more importantly – unproven concepts (usually furthering some preexisting sociocultural values) veiled behind the curtain of empiricism.

    Of course, I shouldn’t go overboard on method, should I? In fact, the majority of truly astonishing scientific discoveries are indebted to a single and more elusive phenomenon: serendipity.

    So look here, scientists, either get your bloody act together or move to a more suitable career in image retouching.

  10. Oh, the armchair philosophers of science, with their garden variety adaptations of Popper and Feyerabend. How charming they are.

  11. The larger issue this illustrates is the widespread and total disregard for the scientific method in the modern scientific community.

    Science is always paid for by somebody who’s selling something, even if there’s a government or academic intermediary. Proving your hypothesis increases your chances of getting another job. The fox is guarding the petri dish.

  12. I’m with you Polomoche. Science shouldn’t be about commerce or belief. When I was just hatched I had an instinctive affinity for the scientific method as the most honest thing the adult world presented.
    What a disappointment later when I ran into priest-scientists and shop-keeper scientists.

  13. I wonder if this could hamper the use of DNA at trial. Plus what about Barry Scheck’s Innocence Project?

  14. A career in research science is very much “publish or perish” – one must churn out interesting new data for publication and persuade a journal to pick it up, otherwise you’ll struggle to apply for research funding and struggle harder to get your next job.

    The scientific method is a noble ideal, but the simple fact is that a year’s worth of beautifully executed negative results simply won’t get published. And if you don’t publish, you don’t eat. The pressure to find new and exciting “stories” about what nature is up to is tremendous.

    So while scientific fraud is inexcusable and generally career-ending — because who will trust any new data you produce after you’re found out? — but I can understand the pressures that lead people to do it.

    If anything, in the current economic climate scientific funding is set to dry up so this pressure will only increase over time. I think it’s terrible that such safeguards are needed but they are essential. I actually know a scientist who lost over a year of research because they’d based their project on a collaborator’s data that turned out to be faked. The innocent party has lost a lot of time, missed out on some funding and gained quite a few grey hairs over the whole thing.

  15. I suspect it’s a type of laziness and dread and ego-protection; like Sisyphus rolling the boulder up the hill to have it tumble back down, the scientist might spend years working on an idea, have it proven wrong and rather than learn from his mistake and rule out a scientific dead end, he instead fakes results so it at least looks like that boulder is still up there. Reminds me of those human footprints cut into the rock in TX to make it appear that humans walked alongside dinosaurs.

  16. When I was working on a doctorate in organic chemistry in the early 80s, the professional tradition they drummed into us was that what counted in the long run was the data we generated from research. Conclusions come and conclusions go, but if you record and report your experimental procedure and your results carefully, and someone can go back and do the same thing and get the same results (within experimental error), you’re good for the ages.

    And it was true! I used to have to make chemicals where the only published syntheses were in German journals from the mid-1800s. Those guys didn’t know about a lot of things modern chemists take for granted, like quantum mechanics or isotopes or the periodic table, but they sure knew how to write up experiments so you could duplicate them. I was very grateful to those old German chemists. And over a century later, we were able to use their research results to head in directions they wouldn’t have imagined.

    The biochemists had a different tradition, though. Their papers were very focussed on conclusions and slapdash about experimental technique. For instance, for a long time they’d say, “take a neuron and put it in a buffered nutrient solution and measure the nerve impulse”, but they didn’t say how they made the buffer. Some labs used sodium-based buffers, and some used potassium-based. Different labs would do the same experiment and get different results, and they’d just argue about it instead of trying to figure out the source of the discrepancy. Then people figured out that sodium and potassium were involved in neural transport, and that caused the variant results. And because the prior papers didn’t report the chemicals used to make the buffers, decades worth of neurochemistry journals were just toilet paper, and all that work had to be done over.

    It sounds like the bioscientists still haven’t learned. My research director would have had my guts for garters if he’d thought I’d fudged data.

  17. Could have seen this coming, perhaps I was part of the problem.

    Back in 1997 (or so), I got a job as a medical photographer/illustrator, and when I started the job, one would shoot a gel like this on a copy stand using a film camera, with the gel floating in liquid in its pyrex tray. Then the negative of the gel would be printed (in a darkroom!), then cropped manually, pasted up with generated labels, then reshot to negative again and printed.

    Of course, I pretty much immediately instituted a new procedure, using digital scanning and image processing and computer layout. Of course, suddenly there were scientists asking if I could “make this area” a little darker… I suppose this was bound to lead eventually to cloning and replacing areas…

    However, I think it has always been possible to “correct” an image of a gel. I could have done the same thing in the darkroom with dodging and burning and double-exposure and in the reshooting of the final image. (I didn’t.) Now it is just way easier to do, and a lot harder to catch.

  18. Hey, Polomoche: Since when does the fact that 1.4-3.3% of published papers contain doctored images signify “widespread and total disregard for the scientific method in the modern scientific community”? You’re taking a preconceived notion of yours and justifying it using data that doesn’t hold up under the barest scrutiny of the actual numbers – how unscientific of you.

    Also, heard of negative and positive controls? Those are the things that try to disprove your hypothesis, and are standard in all good scientific studies.

    Kindly keep your anti-science choadery to yourself, thanks.

  19. I think the broader and more concerning issue is not the presentation in the paper, but that many if not most papers do not publish the data their analysis is based on. In general people will send you the original data on request (sometimes only after an argument), but since there is no central archiving, as time goes on the data are lost. Publication will generally place the research in the best possible light (not necessarily because of dishonesty).

    The blame can not be shifted from scientists who falsify data. They should know better. However we should design a system which does not foster dishonesty. If data collection were valued at the same rate as analysis, we would be in a far better place to both (i) freely disseminate data and (ii) provide incentives for people to provide the unaltered record of their research.

  20. Just pointing out bias wherever it occurs, my good fellow.

    I’m not condoning the actions of the few scientists who falsify their data through image tampering, just pointing out that the minority of scientists who do engage in such behavior do not represent the general practices of the scientific community as a whole.

  21. Since when does the fact that 1.4-3.3% of published papers contain doctored images signify “widespread and total disregard for the scientific method in the modern scientific community”?

    Because that’s a polite way of saying that 1.4-3.3% have been caught.

  22. Actually, Antinous, that’s not what the article says.

    It says that 5-10 of the 10-20 articles showing evidence of tampering warrant a thorough investigation. My 1.4-3.3% estimate was calculated by finding the lowest and highest possible percentages of tampered papers requiring further investigation (5/350*100 at the lowest, 10/300*100 at the highest). That’s not people who’ve been caught – presumably some of the 5 to 10 cases a year were not actually fraudulent.

    But you’re right, there could be a larger percentage of scientists who tamper with their photographs and don’t get spotted at all – but how many more? Are we talking about a few percent more, which is reasonable, or a “widespread and total disregard for the scientific process” amount more, which is paranoid and delusional?

  23. I think you guys are seeing this all wrong…think of the diseases that can now be cured with Photoshop!

  24. The NSF, probably other funding agencies, are making tiny steps towards dealing with this — there are already rules requiring that raw data be eventually published, e.g.

    and I think there are tiny, tiny steps moving towards having the not-published-yet data be signed and stored while the original researchers are getting first crack at it. There is a long history of traditional controls to make it harder to fake results (or their dates).

  25. Kindly keep your anti-science choadery to yourself, thanks.

    Jackalopemonger, before responding to what are otherwise perfectly valid questions, let me at least object to the suggestion that I keep my opinions to myself. It seems sort of contrary to the nature of submitting comments. I’m quite open to criticisms of my view and promise to consider yours thoughtfully.

    That having been said, for the record I’m very much pro-science. What I am against are the implications of examples like the one in this post. Honestly, I do believe that in those fields where a “controlled environment” is possible, science by and large is conducted rigorously and according to its founding principles, despite the realities of external structural/funding issues – enough to tempt even the most stalwart empiricist. If “fudging” is rampant even in this area, the problem is worse than even I, the ostensible anti-scientist, imagined. Although I do believe that the pressure to validate one’s data – rather than to enthusiastically invalidate – is a pervasive problem, if not a rampant one.

    But where I object more strongly is to the application of “science” to environments where the control of variables is simply not possible; there are too many of them. Unfortunately this usually involves sociocultural subject matter – that is, the sort of thing we “garden-variety” types really care about, and the sort of thing that tends to dominate the news.

    Thus my larger point is that doctoring slides pales in comparison to concluding a “study” by elevating inference to fact. We read about this all the time – “a report has come out showing that such and such causes blah blah blah”… When you look closely, nothing causal has been established at all; on the contrary, a valid correlation between two phenomena has been connected by speculation, usually some “common-sense” explanation that appeals to well established and often tacit cultural beliefs. This, to me, is worse than hearsay.

    Example: “We have found that levels of lead in the brains of individuals from very poor, inner city areas corresponds quite directly with rates of incarceration! Therefore, (this is the part I object to) lead causes crime! Lead causes crime! Soylent Green is people!” I mean, I read this just the other day. Anything, I would suggest, but GOOD SCIENCE.

    Now I find science to be a sacred thing, and because of this I will be its worst critic, particularly when there are many examples, far beyond this one, of its modern misapplication. Such things are dangerous, and you’d better damn well believe I’m going to object to them.

    And to MITTZNZ – I don’t believe my charming opinion (while without question provided from a very common armchair) corresponds at all to the ideas of your friends Popper and Feyerabend. When it comes to a philosophy of science, I’m a disciple of Turgenev.

  26. Oof. The problem with most philosophy of science is that it is done by people who aren’t close enough to science, who have a sense about what sience ought to be like (the prescriptive method) to speak to all of us. So you get logical positivists, falsifiabilists, &c.

    They never say “Sience *obviously* works, bitches… now why does it?” (the descriptive method).

    For example, look at the discovery of the electron by J.J. Thomson. We’ve been told at different points throughout History that somehow a scientist is supposed to disprove the alternatives before leaping to conclusions, but if you review this case there is nothing about it that would even hint at the fact that that’s the way this happened (what alternatives)? This of course doesn’t mean that the discovery of the electron was a happy accident that just happened to be confirmed by a cascade of evidence afterwards, it just means the metatheory is crap.

    That said, I think good philosophy of science exists. Ian Hacking and Pierre Duhem come to mind.

    I’ll also agree that there’s a lot of “scienticianism” in popular culture. But most of the spurious correlations you’ll see reported in the media as causation originate from Social Science. My rule of thumb is that if you have to call it ‘science’, it is not.

    Of course it’s problematic that people see this information as authoritative, but that’s a general problem. Most people in the world believe they have an imaginary friend who can help them with their problems.

  27. Molecular Geneticist here and boy have I seen this before! Worst case I know of was our lab was asked for some fingerprinting data on some bacterial strains. The one tech who replied ended up photoshopping a blot, taking a band (like those in the illustration), copy and pasting it into a facsimile of the RFLP pattern. He then sent it to the requesting doc.

    Months later we see it published in a journal! Zoinks! We got dinked on the work and there’s a *completely* fabricated (but based on real data) figure in a peer reviewed journal.

  28. @#19 Bobert,

    I’m with you 100% there – methods reporting in modern biology is frequently terrible. A couple of weeks ago I decided to use a reagent that’s in common use for some areas, but I’d never come across before.

    It arrived with all sorts of lovely data about its chemical properties and storage, more than I could ever need. Then I went back to the literature to check the method. Of the dozen or so papers I read, none mentioned which of four possible solvents they’d used. Only about half even mentioned the concentration; none said why they’d chosen that concentration.

    I eventually found the answer in a paper from 1968, the first time this reagent was used. While all the papers I read were, presumably, relying on the techniques established here, not one of them bothered to cite it. I wasted a whole bloody afternoon on that.

    That’s just one example – I spend a horrifying amount of time writing to authors about their methods, or trying to deduce or outright guess how they performed their experiments. I know journals insist that articles need to be kept concise, but what’s the point of publishing this stuff if noone actually knows wht experiments they performed?


  29. “Now I find science to be a sacred thing, and because of this I will be its worst critic,”


  30. Polomoche,

    Sorry for the outburst. I’ve been spending a large part of the past few weeks arguing with creationists on various forums and some will do concern trolling and say things very similar to the first part of your comment. I overreacted as a result and I shouldn’t have, so I’m sorry for that.

    My main objection to what you said earlier is the claim that the few instances of tampering that do occur are indicative of a widespread disregard of scientific principles among the scientific community. At least that was the message that I read; if you had something else in mind, please let me know. I don’t know the exact amounts of tampering that are going on right now, but if the information given above is representative, it’s around 2%. That to me is unsurprising, even if it’s undesirable – ideally we’d have no fudging of data and images, but scientists are human beings too, as you noted.

    I definitely agree that sociocultural scientific studies often contain many more variables and often cannot conclusively draw a causative link between two factors. I don’t know whether scientists conducting such studies tend to draw conclusions where nothing definitive can be stated – my field of research is in the physical sciences. But it’s very likely that many studies conducted in the social sciences, once picked up by the mainstream media, get distorted by the non-scientifically literate news outlets. And so you go from “Higher lead levels in the central nervous system found in demographic areas with high incarceration rates” to “Lead in the brain causes crime”. In the rush to dumb down the science and create a newsworthy article, the actual analysis of the data gets left out.

  31. jackalopemonger,

    FYI, I get accused of being anti-science all the time, when nothing could be farther from the truth. There’s just so much crap science, particularly in the food and drug industries, it seems like science is just a commodity for sale.

  32. ahhhh, another BB triumph. The ground smells good here. There’s something in the place that makes reasonable people reason with each other. A land worth defending.

  33. Bugs wrote:

    A career in research science is very much “publish or perish” – one must churn out interesting new data for publication and persuade a journal to pick it up, otherwise you’ll struggle to apply for research funding and struggle harder to get your next job.

    I dunno… Fermilab researchers publish negative results all the time. Nope, haven’t found the Higgs, yet …more research is indicated!.

    Maybe it’s different in the goop sciences…

  34. Antinous – never declared you, in particular, to be anti-science.

    There is indeed a lifetime’s worth of bad science floating around out there. I think it’s good to distinguish between a scientific community that’s generally good but has a few bad eggs, and a community that doesn’t practice good science and is more interested in turning a profit than creating new knowledge. My impression is that the photo editing mentioned here is indicative of a small group of individuals who feel pressured into publishing new results, even if that means fabricating some of the information. I don’t think that necessarily reflects on the integrity of the community as a whole.

  35. I know a scientist who works for a business that sells analyzed data to various concerns. Some of the raw data that she receives in obviously incorrect. She has to try to write analysis programs that somehow work around this. She’s not doing it for any malicious or monetary reason, but it’s still a slight fudge. At her end, it’s perfectly clear that there’s a margin of error. By the time that it’s gone through twelve other people and is being looked at by someone in Congress, it’s science fact. Part of the problem is that non-scientists don’t quite get concepts like ‘margin of error’ or ‘maybe’. At that end of the equation, money talks and there aren’t many ethics to get in the way of reality.

  36. I’ve also heard of someone just messing with the contrast or exposure to get a spot to show up before the image was even taken when looking for low levels of activity.

  37. Jackalopemonger – no worries, an understandable reaction. Thanks, a good, lively debate…

  38. “Laws are like sausages. It is best not to see how they are made”
    -Otto Von Bishmark

    What Bishmark said about laws applies also to science. At least result Photoshopping can be revealed and published numbers reanalyzed, but what about all the undocumented “features” that are left out, just because they would show how imperfect the research process actually was. Scientists, participants, labs and methods are never as perfect as paper writers like to pretend they were.

    Science is not a candle in the dark. Its more like glow of just another TV shopping channel in the cacophony of self marketing.

  39. Yes. If you cant do proper DNA blott, than please get a job at a Cheesecake factory, not in research!

Comments are closed.