Research misconduct: how the scientific community is fighting back

0:00You know that feeling when too many things fall through the cracks. Monday.com was built for that gap. The AI work platform where people and agents work side by side to deliver more together. Create your first Monday agent today at Monday.com. Hi, this is Becca Moore of For the Girls Podcast. All right, a quick shout out to our presenting sponsor, Apple Cash. Summer is here, which means the group chats are already spiraling with festival and summer plans. And we all know the worst part of being the one who fronts the money for the passes or housing. That's why this year we're only using Apple Cash.

0:30It's a super easy way to settle up because you can do it right in messages. Just tap the plus sign to get started. And the best part is you can spend the money you get right away anywhere Apple Pay is accepted. Every payment is secure and protected by Face ID, Touch ID, or Passcode, and your transactions are never shared on a public feed. Set up your Apple Cash today in the Wallet app. Apple Cash services are provided by Green.Bank, member FDIC. Terms apply.

1:01A bunch of volunteer data sleuths from around the world started digging into these papers about pizza eating, and they found just within these four papers 150 inconsistencies and errors in the data. This is Stephanie M. Lee, a journalist from the Chronicle of Higher Education who writes about research integrity. These papers came from the lab of Brian Wonsink A high-profile diet researcher at Cornell Who Stephanie wrote about He was in a very influential position And a lot of his research boiled down to the message that You can eat healthier and lose weight Without having to make drastic changes to your diet and exercise Unsurprisingly, such results caught a lot of people's attention landing him a position in the US Department of Agriculture where he was responsible for the 2010 dietary guidelines for Americans.

2:03However, Brian's results were put into question when several of his papers were shown to have signs of data manipulation something that he himself unintentionally revealed in a blog post in 2016. Dr Wansink wrote about how the study had failed in his own words However, it had been saved because he'd given the data to an ambitious visiting scholar from Turkey. He described her as the grad student who never said no. And with his encouragement and help, they kept analyzing the data until they found associations that kept holding up. and he was very proud that they had turned this data set into four studies about pizza eating in various journals.

2:54So Dr. Monsing published this thinking this was a model for academic behavior, but he soon saw that many others saw it as a model for something different, which was a cautionary tale about p-hacking and evidence of extreme manipulating of data in order to get results that seem statistically true but are very likely to be false positives.

3:30What Brian hinted at in that blog post was an example of misusing data to find something that is statistically significant. A misunderstood and somewhat arbitrary threshold that is often used as a marker of a result being interesting or important, also known as the p-value. Often, that threshold is chosen to be less than 0.05. The thinking being that if a result is below that threshold, there's less than a 5 % chance that that was down to chance. But if you throw enough statistical tests at something and cut the data enough ways, then you're bound to find something that ends up below that 0.05, just by random chance.

4:16This is known as p-hacking. In fact, in emails that Stephanie uncovered, Brian said about one result that, for some reason, this is a p-value of 0.06. Seems to me it should be lower. Another collaborator of his talked about running 400 mediation analyses to try to find an association that would work with one data set. So I wrote stories about these conversations after, of course, talking to other experts in the field, specializing in statistics and open science and making sure that what seemed outrageous to me was, in fact, outrageous to them. and Cornell had initially investigated a handful of studies and said nothing to see here we don't consider this misconduct and then after I started writing my stories and they were getting more and more attention online Cornell said I was going to investigate again and then this time after the second investigation they said that they determined that he was in fact guilty of academic misconduct and then Dr Wansig said he was going to retire, which she did a year later.

5:29All in all, this was years of investigation by scientific sleuths who went through Brian's work with fine-toothed combs, along with Stephanie's reporting, at least 12 retractions and two investigations by Cornell before he was eventually found to have committed misconduct as defined by his institution and retired. i'd like to say that this is an isolated case but unfortunately science is not immune to misconduct the system is largely built on trust when you read a paper or get it to review you likely implicitly assume that what people have said they have done is what they've done but manipulation of data, questionable research practices, falsifying of images, and more happens.

6:21And it can pollute the literature. I have come to this interview from a morning of looking at a pediatric treatment that I'm almost certain is supported by fake evidence that is probably killing children. And that's just Thursday. And there's a growing problem in science. Such fakery of scientific results is getting industrialised. So the number of paper mill papers grows faster than the rate of publications in general. I'm Nick Petrich Howe, and this is Self Correction, a two-part miniseries from Nature. Last time, we talked about how to tackle more honest errors in science, when researchers were still trying their best but made mistakes.

7:14This time, we're looking at where things are a bit more deliberate, when researchers commit misconduct.

7:33Misconduct in science is a tricky thing to pin down. both in the literal sense it's in principle very easy to get away with it in science as the system is built on trust and in the definitional sense it's hard to pin down what people mean by misconduct different institutions and countries will have different definitions of what misconduct actually is you might think about things like criminal behaviour plagiarism and outright fabrication of results and those are certainly misconduct. But there are also things often referred to as questionable research practices that are harder to pin down. These can be things like cherry-picking certain papers to report on or hypothesizing a result after you already know what the result is.

8:22Here it can be tricky to know someone's intentions. Did they intentionally leave a paper out of their references or did they just not know about it? In fact, many people I've spoken to have told me that questionable research practices were just fairly normal behaviour in many areas of science a few decades ago. I was doing myself specification searching for my master thesis and just piacking to the bone my thesis and at one point I just realised it makes absolutely no sense. This is Abel Boudreau, an economist from the Institute for Replication. You might remember him from the previous episode in this series.

9:03Specification searching, or p-hacking, is a way of trying to inflate the statistical significance of a result. Here you just try different statistical tests or associations on different parts of the data until you find something that is significant. If you do this enough, then you'll end up with a significant difference just by chance. And what can happen is that only those with significance are reported, which can make a result seem a lot more important. But that's kind of the training you get as like an economist or just like as a social science researcher. Like you're just thinking statistical significance is something you want to achieve.

9:46And I remember like people were not telling me to PIAAC, but sort of like that was I was kind of normal to look for a specification that would show something like interesting. The statisticians amongst you are probably groaning in horror at this. It's the behavior that also led to the investigations into Brian Wansink. Now Abel, for his part, recognised that what he was doing was wrong and now heads an institution that tries to improve the quality of science by doing more replications. But it highlights that such behaviours can be quite common. In fact, several studies have shown that across disciplines, there is an odd tendency for results in papers to be statistically significant.

10:31Although this has been debated. Practices like this can undermine scientific research. They could, for instance, make other researchers believe an effect exists that is actually only tenuous or non-existent in reality. People could waste time following up on some research that is flawed, or like in the case of Brian Wansink, it could be used to inform policy. and while the prevalence of such questionable research practices may be concerning when it comes to more direct misconduct plagiarism data manipulation and falsification that too may be quite common in a 2021 survey 80 % of researchers in the netherlands confessed to fabricating data and an off-sited meta-analysis from 2009 also found around 2 % of scientists admitted to fabricating, falsifying or modifying data at least once.

11:30But these estimates have been hotly debated. Not only there is some debate, but it's not really a question that has a, or could ever have a definitive answer. This is Daniele Fennelli, a metascientist and the author of the 2009 study. What most people will have in mind is how much of the literature, for example, is untruthful. But even here, we would have to break down what exactly we mean, because it is certainly not the case that we should expect many, in fact, most probably of published research findings to be any good. That was the process of science all along. Finding out the prevalence of misconduct in science is a complicated endeavour.

12:14It's hard to work out what is misconduct and what is sloppiness or an honest mistake. And it depends on how you define misconduct, which varies from institution to institution and from country to country. The best data that we have comes from a few specific fields, a few specific methodologies in particular. Data, for example, on the prevalence of papers that have duplicated or manipulated images, which is one of the most straightforward forms of misconduct that can be detected and therefore measured. And this suggests something between 1 and 4%. Daniele also says it's hard to know how dishonest scientists are compared to other professions, and that there may even be evidence to suggest they are more honest.

13:02That may be surprising, considering that if you really minded to, it's in theory very easy to get away with misconduct in science, as it's rare for people to dig through other researchers' papers, and even rarer to find clues that point towards misconduct. Others, though, have suggested that the rates of misconduct are much higher. A meta-analysis in 2021 found that 15.5 % of researchers had witnessed someone else committing misconduct at least once. And at the end of last year, James Heathers, a forensic matter scientist, in an un-peer-reviewed preprint, came up with a rough estimate that around one in seven papers is fake, fabricated or falsified to some extent.

13:49If you look at a big bag of aggregated evidence, it's not really consistent with the narrative around this. That's James. Which is, in general, there are singular untrustworthy people, a very low percentage of untrustworthy researchers, who produce a non-specific amount of untrustworthy papers. That is not really what any broad assessment of the literature is going to look for mistakes or inconsistencies or problems that are detectable. That is not really consistent with the narrative that there are a few dishonest people and the system working as it does over time will find them and then everything will be okay essentially people can be trusted mistakes happen everything is fine i don't think they're mutually tenable according to james the idea that there are a few bad apples in science doesn't really track with what he sees in the literature.

14:56When I asked Daniele about this preprint, though, he had this to say. Basically, what that paper does is lump together under a completely sort of made-up, frankly, category of fake papers. It's lumping up all sorts of different things, okay? Apples and oranges, but apples and oranges that, again, if you go and look in the details, have very little even to do with the specific problem we're discussing here, for example. put them all together in order to claim something that, quite frankly, if it was true, it would be absurd, which is that one in seven papers, I think, is claiming are untruthful and real.

15:31So if the implication of that is that we can't trust one-seventh of the literature, we would be beyond disaster. While Daniele has concerns about this method of determining the prevalence of misconduct, James says that this is a very rough estimate. Where he and Daniele agree is that there is not a lot of good data on misconduct. James also points out that past surveys may be underestimating the problem, as they rely on people being honest about their dishonesty. Now, I don't know about you, but if, for instance, you have ever received any official form and then asked to identify yourself as someone who's causing a problem, I have a tendency to tick no regardless of whether or not that problem is applicable to me or not.

16:21Given the sparsity of the data and how difficult it is to find accurate numbers it would be a big challenge to really figure out the prevalence of misconduct. And even if researchers are able to pin down a specific number Daniele thinks it's going to vary by country and by institution. So there are going to be, in other words, all sorts of contextualising factors that modulate that figure. So unless one cares about the global average of something, which I don't think is very useful for any concrete discussion of addressing problems, then, again, there isn't going to be a definitive answer unless we specify a very clear context.

17:00So we may not have a very clear idea of how much misconduct is happening in science, but it's definitely not zero. Even if the true number is at the lower end of the more reliable estimates, around 1%, with millions of papers being published every year, that's still many tens of thousands of papers that include some kind of misconduct. And if these papers are untrustworthy with false results, they could stall life-saving research or send scientists down false alleys. for example in 2022 it was revealed by an investigation in science that a seminal alzheimer's paper had manipulated images suggesting that some of the results were fabricated that 2006 paper published in nature was later retracted and one of the key authors resigned after their institution found data integrity concerns in several of their papers that paper though, is one of the most highly cited to have ever been retracted, showing evidence that a certain molecule, known as amyloid beta star 56, was key to Alzheimer's progression.

18:14Now the amyloid hypothesis that this work formed part of is supported by other evidence, but for this molecule at least, this fabrication could mean that for 16 years, people tried and failed to replicate relevant findings, and millions were spent on drug trials based on the paper and the papers that built on it.

18:47Now, this discussion of how much misconduct exists is important as it determines the best way to tackle it. If it is rare, then perhaps it can be dealt with with more targeted approaches. Whilst if it is common, then it is likely caused by the system of science, and so solutions would involve changing that system. For instance, one of the most oft-cited reasons for misconduct is the pressure on scientists to publish their work. Researchers are said to have to publish or perish, as their careers will often rely on the number of publications they put out, and in what journals they publish in. If they don't do that, then they perish, falling out of science.

19:32This pressure seems to have an obvious relationship to misconduct. If the alternative is perishing, then people will publish by any means necessary, otherwise they risk losing their careers. This is one example of a perverse incentive in science, and is one of the most talked about, along with the intense competition for stable scientific careers, money and notoriety. But according to Daniele, except in a few specific circumstances, we don't have great evidence of these incentives leading to misconduct. Here I can speak primarily, in fact, as someone who's been trying to test these ideas with all intentions to actually verify them, quite honestly.

20:18And yet we never encountered a single conclusive data which would suggest that across the board, any of those factors are actually measurable risk factors for scientific misconduct. The exceptions to this appeared to be largely geographical. Once we narrowed down the analysis in a fairly sophisticated way to a few countries, and in particular China that has a peculiar, obviously, specific academic system that was known, in fact, to generate problems. And we'd run those analysis just there. All of a sudden, most of these factors actually became really important predictors. Long story short, the Chinese system, which again, has been acknowledging this for a while, has a problem with having many low-ranking universities and new universities that basically have a very aggressive policy of marketing and incentivizing with cash incentives, high-ranking publications in international journals because they're trying to promote themselves.

21:21Then all of a sudden, you also see that, for example, researchers that publish a lot and early career researchers were all at higher risk of actually being authors of studies with misconduct. Now China has clamped down on such actions to an extent, but giving researchers payments to publish does seem to have a link to misconduct. Perhaps unsurprisingly, financial gain may be a good motivator to be a bad actor. Daniele also takes the view that we need to consider the specific circumstances in different institutions and countries to understand misconduct fully. In the case of China, there is a link.

22:00But that, in his view, is not enough to make a sweeping change to the whole culture of science to tackle misconduct. That, he says, could even be harmful. So, when it comes down to what we do... Well, that's going to depend, I'm afraid, on the particular country, and even arguably potentially the particular institution. And so to me, where really the discussion at this stage, if you will, of maturity, of research on research integrity and generally metascience, is that we have to take on board and find ways to measure and account for the context. James also points to evidence of paying for publications as being a cause of misconduct.

22:44In his view though, misconduct is a lot more rampant, so his proposed solutions are different. He would also like more research into the causes of misconduct, but he suggests that there is a lot of fake research out there already, and this is something that needs to be tackled quite urgently. and one way he does suggest to tackle misconduct is to make sure that everyone is responsible for what goes into a paper i think that if a paper is shown to be unreliable and especially fabricated or heavily falsified i think all of the authors of that paper should be responsible There is a strong tendency to say, well, there's 12 of us.

23:29And it turns out there was one person in this team who just took advantage of us and hid things from us. And then it's all their fault, basically. And this is an unfortunate incident for me. So I'm the victim here. My response to that would be, no, you're not. Sorry. you're negligent check the work from your co-authors have a research process that involves understanding what you put your name on or don't publish things and put your name on them i'm sorry if that makes things longer and more complicated i too wish my life was easier but you cannot simply say they did it it's got nothing to do with me this is ostensibly already the case, but it's not uncommon in cases of misconduct for co-authors on a paper to say they had no idea what the perpetrator did.

24:30Again, research is built on trust. Generally, people don't suspect their co-authors, who are often their colleagues and friends, of misconduct.

24:44misconduct does happen in science to what extent is hard to say but fake or falsified papers can have far-reaching consequences time money and people's lives can be put on the line by research tarred by misconduct there may be solutions but it's clear that more research needs to be done to understand the problem better. Changing incentives could work in some specific context and perhaps a more general understanding that misconduct is something that every scientist should be on the lookout for could help. But whilst the prevalence of misconduct is debated, one thing that does have broad agreement is that misconduct is getting industrialised.

25:29We're really looking at a literature that has a kind of detritus of fake papers in it.

25:42When it comes to understanding misconduct, one way to try and get a sense of what is going on is to look at retractions of papers. As discussed in the last episode, something between 20-40 % of papers are retracted due to errors, meaning that the rest are likely down to some form of misconduct. But if you dig into the data on retractions, something interesting emerges. They are on the increase. Retractions were extremely rare events. This is Richard Van Norden, a journalist at Nature who has looked extensively at data on retractions. So that was sort of in the mid-2000s. They were running at something like 0.02 % of papers published.

26:29So it just almost never happened. but then really I think to be fair Retraction Watch catalyzed this they started realizing that looking at retractions and why they happen sort of opened up a window as they would put it into the scientific process what happens when things go wrong and it turned out that at that time more and more of science began to come online and this made it much easier for outside sleuths to pick up things that were wrong with papers and it turned out there was an awful lot wrong which had been let slide in earlier years. There wasn't some golden period in the 2000s or the 1990s when science was all wonderful and nothing ever needed to be retracted.

27:04It just wasn't a phenomenon that happened. With more papers online, it allowed scientific sleuths, people who scour through papers in search of errors or worse, more access. And with this, misconduct and errors were increasingly uncovered, leading to more retractions. And the increased coverage of retractions with organisations like Retraction Watch led to yet further scrutiny. But more recently, looking through retraction data also shows a more pernicious increase. Through the mid-2000s, we started to see the onset of paper mills. As scientists were judged on their productivity and told they needed papers to publish to advance in their careers, they would turn to these third-party firms that would publish fake papers for them.

27:53And this industrialisation of fraud led to waves and waves of papers coming into journals. This fraud is so extensive that even with millions of papers being published each year, it's had an effect on those percentages of retractions. Perhaps listeners might know at Hindawi, an imprint of Wiley, which has now been closed down because it was swamped by so many paper mills. and as a result of all this the retraction rate has gone up to about 0.2 percent in 2022 specifically so that's about one in 500 papers and these paper mills often produce entirely fake papers filled with experiments people didn't actually do it's a big problem for science this is anna abalkina a scientific sleuth who investigates fraudulent research and uncovers paper mills?

28:45Because there is evidence that paper mill papers, they are associated with scientific misconduct, plagiarism, image manipulation, manipulation of results and data. And these kind of papers, especially in medicine, they're included in matter research and systematic analysis, systematic reviews, and this delays the progress and biases the results. For instance, a paper mill paper could ostensibly show results of the efficacy of a drug, which are entirely made up. If that fake research ends up in a meta-analysis that's trying to determine whether or not that drug is useful, it could end up showing that it falsely is or isn't, perhaps preventing an effective treatment for patients being approved, or even hiding that a drug has damaging effects.

29:41And this is big business too. A single paper mill could be reaping in millions. So I have calculated for international publisher and I found that their income could reach six million dollars during several years. How might this work? Well there's a lot we don't know about these organisations. They are, by their nature, somewhat secretive. But some that have been uncovered have business models where they sell authorships to researchers, sometimes on real papers, albeit ones that the person buying the authorship had nothing to do with, and sometimes on entirely fake papers. There's been a lot of investigation into these fake paper practices in China.

30:30As we talked about earlier, there has been a particular pressure on Chinese researchers to publish, and this has led many to turn to paper mills. A recent analysis in Nature showed that some of the leading institutions for retractions were from Chinese hospitals and medical universities where such perverse incentives have dominated. Again, the government of China has been cracking down on such practices, but it's unclear whether this has had the desired effect yet. And while a lot of investigation has focused on China, Anna says that there are paper mills all over the world, but we just don't know a lot about them.

31:08We don't know much about paper mills in Africa or Latin America. Now, we see, for example, a lot of journals in Latin America are flooded with paper mill papers. And this is a challenge because there is not enough sleuths to detect paper mills? Sleuths, like Anna, are generally in short supply. Most scour through papers in their spare time, making it hard for paper mills to be uncovered. Additionally, the tactics of these organisations are evolving. They come up with strategies to specifically avoid the gaze of a watchful sleuth. For instance, AI tools allow paper mills to churn out more plausible looking articles.

31:54One strategy to detect paper mill papers was to look for so-called tortured phrases. Phrases that seem at odds with what people would normally say, such as breast cancer becoming bosom peril. This is because bad actors would plagiarise papers and then use paraphrasing tools that would churn out these odd turns of phrase. But the rise of AI chatbots means that, ironically, it's become easier than ever to make a text seem human. And this rise of AI tools means that paper mills can now create fake papers much more quickly. So AI-ridden papers, it's very difficult to, first of all, to detect them, and then to prove that a paper is not written by a human.

32:42For example, in the case of plagiarism, you have a source of the text and the text, and you can compare these texts and clearly say that these texts are identical. In the case of AI, this is just a probability. In all software that detects artificial intelligence papers, they say the probability of AI-written paper is 75%. What does it mean? so you don't have evidence of an AI written paper. This is further complicated by the legitimate reasons people may use an AI tool to help write their paper such as improving the wording for someone who isn't a native English speaker. As we don't know a lot about paper mills it's hard to say whether the problem is growing.

33:33Certainly the awareness is growing and the retractions due to paper mills are growing too. One analysis suggested that 1.5 to 2 % of papers published in 2022 resembled paper mill productions. For many researchers, though, they highlight the problem of perverse incentives in science. The whole system of research evaluation creates a great pressure to publish more papers. So, a quantitative number of papers is more important than their quality. And this creates a great demand for papers from paper mills. Also, some countries offer financial rewards for publications. And so it can happen that the price for co-authorship is less than a financial reward.

34:27and that creates a kind of a great field for business, for individual researchers, because they can just maximise the number of papers and it will maximise the financial reward from their university. Many nations, and even some institutions in the US, have or continue to offer financial rewards for publications. China ended the policy to pay for papers a few years ago. But even without direct payments, in many countries, publications may be tied directly or indirectly to career progression.

35:09Now, as we talked about before, the problem is complicated. In some specific situations, incentives like these may lead to misconduct, and it certainly seems that the proliferation of paper mills shows that there is a profitable business in rapidly producing shoddy publications for desperate or ethically dubious scientists. And that industrialization of fabricated papers pollutes the scientific literature. They can get cited in legitimate articles, as the authors are unaware they are giving credence to a fake, or they can get scooped into meta-analyses or government reports, distorting policy or our understanding of life-saving drugs.

35:53So what do we do about it? There are no easy answers. A constant topic in tackling misconduct that it doesn't come clean cut. Here's I've already fabricated and that's it, period. More often than not, it's more complex.

36:16whenever i've spoken to researchers about errors and misconduct in science the idea of incentives has come up a lot as danielle pointed out this can actually be tricky to link to misconduct but the prevalence of things like paper mills do show the impacts of such perverse incentives why perish when you can publish but incentives are created by the entire system of science and vary from place to place and institution to institution so often people will point to those that have a bit more power as the places that need to act to solve these issues in science journals and institutions journals are often the gatekeepers of what gets published or not so people argue that they can have a key role in ensuring paper quality.

37:07Additionally, they point to the resources at their disposal. It has been argued, for example, that scientific publishing is among the most profitable businesses in the world. But they too have their constraints. Journals can help lay down guidelines for best practices to avoid errors and misconduct, but they can't really force anyone to follow these rules. they tend to take it on faith that they do and every guideline may come with additional work and consequences that can be hard to predict for example cash incentives for publishing papers was seen as a positive thing by some to encourage good quality research but with hindsight they perhaps weren't always the best idea journals too are taking steps to try and tackle paper mills many are increasing the staff they have to look into research integrity.

38:01Some have removed the ability of researchers to suggest peer reviewers for their papers, as this has been abused in the past to get paper mill papers into legitimate journals. And generally, they are more willing to issue retractions than before. But they are being cautious, as every retraction could be a legal issue. A researcher could sue, as retractions when associated with misconduct are seen as a stain on their reputation. And when it comes to more individualised misconduct, generally it is seen as the role of institutions who employ researchers to investigate. They also have guidelines that researchers must adhere to.

38:43For the example of Brian Wansink that we heard about at the start of this episode, Cornell found that he had, quote, committed academic misconduct under their policy, unquote, and then removed him from his teaching and research positions. They too tend to be cautious as such investigations are very costly and the accused scientists can sue and some already have.

39:17One solution may be to strengthen the law around research integrity. A key issue is that, while the scientific community may shake its head at misconduct, it's often perfectly legal. For example, there are few laws aimed at tackling paper mills. Currently, the legislation in different countries doesn't prohibit scholars to purchase paper mill papers. So it's a legal thing. probably only Peru has an idea to to have a legislation that prohibits scholars to engage in these kind of activities it's legal for paper mills to operate according to legislation of many countries so they are not prohibited it's an ethical problem but not a legal problem in many countries.

40:09And even when paper mills are uncovered, according to Anna, institutions can be reticent to act. The second challenge is the lack of investigations by universities and consequences for scholars that purchase paper mill papers. So journals can retract the paper, but scholars continue to work in their positions and normally they enjoy their foothold careers. If their actions are legal under the letter of the law and the institutions would have to undertake costly investigations and risk being sued, then it may not be surprising that such bad actors aren't removed from their positions. But what if someone else investigates for them?

41:02in 2008 austria was rocked by a misconduct scandal an experimental trial of stem cell therapy to treat urinary incontinence led by han strasser was deemed illegal and invalid by the austrian agency for health and food safety hands reportedly had carried out clinical trials without ethical approval, failed to provide raw data and even forged legal documents. Hopeful patients paid for treatment without knowing that they were experimental and some even claimed that their blathers sealed over. In the wake of this, it was clear that something needed to change. So the research community came together and created the Austrian Agency for Research Integrity, known as WAVI.

41:54The research community came together and agreed that they need a better infrastructure, that they need one agency that is separate from the individual research institutions to both improve the protection of good scientific practice and the investigation of potential cases. This is Sabine Chai, the managing director of the agency. Since its establishment in 2008, it has investigated hundreds of cases of alleged misconduct, in addition to providing training and support to researchers and institutions to bolster research integrity. It works by allowing whistleblowers to anonymously report concerns and either ask for advice or request an investigation.

42:39If the agency investigates, it collects statements and documents from all involved and gathers expert opinions. And then finally, it delivers a report to all parties and recommendations to the institution. The institution ultimately decides on how to follow up on those recommendations. The agency doesn't have the power to sanction researchers itself. That was a balance struck when the agency was formed to protect the autonomy of the research institutions. I think Uyavi is a good middle path also in regard to the autonomy issue because we do not sanction anybody. We conduct the investigation, we provide the statement, the outcome of the investigation to the institution and the institution then has to take action and impose any sanctions that work for their regulations.

43:36Sabine thinks this independence though is important for their investigations. There are always cases that cannot really well be investigated at the institution itself. For example, if there would be the suspicion of bias at the institution, or if several institutions are involved, or if the case is very public, and then suspicions of improper processes come up easily, even if that is not the case. So it is good to have an independent institution that can investigate in such cases. More and more agencies like this one are being established all over the world. And while they can support institutions by investigating purported cases of misconduct and providing training to researchers, Sabine acknowledges that their job is not an easy one.

44:24The waters are murky. What it means is that very often cases like inquiries that we get, cases that are investigated, are not single issue. More often than not, there is a part of the issue that has to do with possible research misconduct. And then there are other things that happened which may have to do with discrimination, that may have to do with mobbing, that may have to do with personal issues, or that may have to very often have to do with personal conflicts. And that cannot be resolved simply by looking at the research misconduct issue. The law could potentially help by clarifying what research misconduct is and empowering the growing number of research integrity agencies with legal backing.

45:13In Austria, a new law includes a section on research integrity requiring adherence to good scientific practice. It's hard to say what effects this will have yet, but Sabine is optimistic. It's significant to have that in the law. And it has a list of issues that are clearly named as misconduct now also including falsification, fabrication, sabotage of other researchers, using AI without reference and other things that were also previously included. And it also requires institutions to have an infrastructure, to have regulations related to both protecting good scientific practice and processing misconduct.

45:58So we are now at this transition from having a few issues addressed in law to having a lot. And it will be very interesting what kind of difference that makes. That's something we'll be able to observe in the next few years. Any such changes to the law will need to be approached carefully, as it can be a double-edged sword. For instance, in the US, President Trump signed an executive order in May mandating an overhaul of research integrity policies. Whilst the order purports to promote gold standard science that is transparent, rigorous and impactful, thousands of researchers signed an open letter arguing that this order consolidates political control over science as it implements a system where direct presidential appointees are given broad latitude to designate many common and important scientific activities as scientific misconduct.

46:56Any efforts to legally back research integrity and form agencies to tackle misconduct will need to ensure their independence and insulation from political control, lest they become a tool for politicians on any side of the political divide. Independence and freedom to act is also important for everyone who wants to help tackle misconduct. That's true for institutions, but also for journalists and sleuths. At the start, we heard from Stephanie M. Lee, who helped uncover the actions of Brian Wonsink. Here's Stephanie on how journalists can help tackle misconduct. So what journalists can do effectively is we can try to get all sides of the story.

47:40So we can try to get in touch with everybody who may be accountable for a problematic study or may know something about how a study got made, got published. So we can make phone calls, we can send emails. If needed, we can show up at people's houses and knock on their doors. And we also have more access to legal resources and lawyers who can help us sue over records if institutions are not producing them. We can give voice to people who may need to speak or share information anonymously because they see a retaliation for whatever reason. They're a junior scientist. Their job is on the line, etc.

48:24etc. We can write stories that are vetted by lawyers to minimise the risk of defamation lawsuits. And, you know, most importantly, I think we can draw public attention to a problem and use the spectre of public pressure to try to get people to answer things on the record. Stephanie also told me how critical scientific sleuths were to her work. They were the researchers whom she asked to check the details of misconduct Sleuths are often well-practised at spotting oddities in papers and checking the statistics And while sleuths have proliferated in recent years with the increasing spotlight on misconduct According to Anna, one such sleuth, there really isn't enough support for them First of all, many sleuths, they work pro bono So they are not paid for this job.

49:18And this is a very time-consuming job, first of all. So mostly slurs are self-trained, and it's based on a lot of experience and papers that they looked and fraud they found. So the first problem is the support of slurs, like financial support of slurs. But second challenge is the protection, because many sleuths are sued by fraudulent actors. If the work of sleuths is helping to clean up science from this detritus of fake and fraudulent papers, then they need protection and support. Sleuths do get sued. And it's not just purported perpetrators of fraud that sleuths need protecting from. Recently, concerns have been raised that Sleuth's work is being used by the current US administration to undermine good quality science and as an argument to grant more political control over research.

50:19If science is truly the pursuit of truth, then institutions and people who work to promote research integrity and uncover fraud should be protected and allowed to be independent of any political machinations that could seek to bend science to their ends. If we want people to trust in science, we need to show it's worth trusting.

50:45There are problems with science, but there is a growing focus and exposure of misconduct, which may be a sign that it's starting to self-correct. Misconduct cases are flagged by journalists and sleuths. There's a growing number of institutions and journals that are doing more to tackle misconduct, but we need to protect those who are working to find fraud and promote research integrity. Public trust in science and the reputation of journals and institutions relies on it. There are worrying trends too. Paper mills are a growing problem but there's also a growing recognition of them as an issue and the incentives that allow them to proliferate.

51:25Science is a process that has some perverse incentives and there aren't easy solutions for how to solve them but is everyone's responsibility journals journalists sleuths institutions and every individual scientist is responsible for ensuring that research is of the highest quality but together maybe we can come up for a way to help science self-correct every part of the producing knowledge chain participates in the act of making new knowledge about the world. We cannot pick one of them in particular and say they need to change in isolation. There is a collective mindset. There is a collective set of values that are obviously deficient.

52:17And it is an everyone problem.

52:30This episode was written and produced by me, Nick Petrich-Hall. Dan Fox was my editor. Thanks to Stephanie M. Lee of the Chronicle of Higher Education, Abel Boudreau from the University of Ottawa, Daniele Fanelli from Heriot-Watt University, James Heathers from Linnaeus University, Richard Van Norden of Nature, Anna Abalkiner from Fryer University, and Sabine Chai from the Austrian Agency for Research Integrity for talking to me for this story.

53:21The CMO asked for already in there. Three angles, no competitors using right now. Just need your approval. Approved. Thanks. Oh, agents, where would we be without you? Hmm. Somewhere with a lot more tabs open. Create your first Monday agent in minutes at Monday.com. Hannah, I just Venmo'd you for dinner. Obsessed. I'm spending it right now in the lip gloss that's been sitting in my cart. What do you mean spending it right now? You can instantly spend your balance with the Venmo debit card or when you pay with Venmo at checkout. Stop. Say more. More exactly. The more you do with Venmo, the more you get.

53:59Like earning up to 5 % cash back with Venmo Stash. Get the Venmo debit card or checkout online. Venmo Stash bundle terms and exclusions apply. Max$100 per month. See terms at venmo.me slash stashterms. Venmo checkout not available at all merchants. The Venmo MasterCard is issued by the Bancorp Bank N.A. Thank you.