TE
TechEcho
Home24h TopNewestBestAskShowJobs
GitHubTwitter
Home

TechEcho

A tech news platform built with Next.js, providing global tech news and discussions.

GitHubTwitter

Home

HomeNewestBestAskShowJobs

Resources

HackerNews APIOriginal HackerNewsNext.js

© 2025 TechEcho. All rights reserved.

1 in 4 Statisticians Say They Were Asked to Commit Scientific Fraud

527 pointsby brahmwgover 6 years ago

36 comments

jbogganover 6 years ago
I was going to give a lecture across several departments about my PhD research in bioinformatics. The night before the talk I was generating some new figures and saw something weird, which led to me digging through source code and discovering a bug which invalidated the last 18 months of research and all my conclusions. I went to my advisor with this problem and he told me to present it anyways. I refused. I suffered consequences for that, including getting stonewalled by my advisor whenever I tried to publish a paper. I wish I had actually been in a position to refuse.
评论 #18366830 未加载
评论 #18367047 未加载
评论 #18369073 未加载
评论 #18367584 未加载
评论 #18367349 未加载
评论 #18370484 未加载
评论 #18369281 未加载
评论 #18368179 未加载
评论 #18368815 未加载
评论 #18366433 未加载
评论 #18369142 未加载
评论 #18367531 未加载
评论 #18368693 未加载
评论 #18372273 未加载
danieltillettover 6 years ago
The pressure to get positive results is just too high to be compatible with good science. When you have people&#x27;s whole life and career on the line all the time (publish or perish) then people are going to do what they have to stay employed.<p>Like Gresham&#x27;s law bad science drives out good, because it is much faster to do bad science than good. Those that try to maintain quality can&#x27;t pump out as many papers as the bad and so lose on the grant treadmill.<p>Any solution that does not address the incentives is doomed to fail. Not fixing this problem will kill science.
评论 #18367966 未加载
评论 #18368879 未加载
评论 #18370731 未加载
评论 #18368803 未加载
评论 #18369688 未加载
truantbuickover 6 years ago
Beyond overt fraud, I feel like there&#x27;s so much incentive for imprudent analysis.<p>Several times when I&#x27;ve worked on a project at work that involves data analysis, I get really impatient responses when I don&#x27;t come to a firm conclusion. We&#x27;re not even talking about highly charged subjects. I&#x27;m not aware if my stakeholders are biased toward one conclusion or another -- I think they&#x27;re just very upset that my analysis contains uncertainty. They expect me to be able to massage any kind of data to derive clear and obvious facts.<p>Thankfully, I can refuse this without much consequence, but it&#x27;s really opened my eyes to the potential pressures to corrupt the integrity of fact-finding in even mundane circumstances.
评论 #18366854 未加载
评论 #18367380 未加载
ms013over 6 years ago
Reminds me of a project (not in medicine) I had a couple years ago where my group was in the data science role for another group. We helped out with experimental design, data acquisition, and analysis. After all of our work, when we crunched the numbers, we found that the hypothesis being tested didn’t hold. The people funding the work were not at all happy, and went so far as to claim that “the plots must be wrong”. We showed them everything we had done, and nobody could find any issues with the analysis. But, they were irritated because we didn’t want to play the dishonest game of farting around with plots and numbers to make it look like the experiment worked. So, they angrily broke up with us. On the plus side, when I looked a year or two after, it looks like the grumpy people haven’t made any progress and are either out of business or something.
评论 #18369398 未加载
jlg23over 6 years ago
&quot;Prove our new train signaling system is as secure as the old one. If you can, you&#x27;ll get your degree, if you can&#x27;t, you find someone else to sponsor your research&#x2F;degree and we find someone else to do the proof.&quot; -- Deutsche Bahn (government held national train company in Germany) in around 2001 to a friend of mine (they and him found &quot;someone else&quot;)
评论 #18368072 未加载
评论 #18367840 未加载
2spicy_thrwawayover 6 years ago
Even within the statistics community, there&#x27;s a spectrum of quick-and-dirty vs fully rigorous. People with the ability and inclination to be fully rigorous often get treated as pedants and perfectionists (in the bad sense).<p>I often get that with business partners. &quot;The data says &lt;likely X, but with caveats&#x2F;nuance&#x2F;uncertainty&#x2F;under certain assumptions we can&#x27;t justify&gt;&quot; to which they respond &quot;Can we just say X?&quot; Or &quot;can we get numbers on Y to support a presentation on Z?&quot; when Y <i>seems</i> to support Z, but actually you can&#x27;t draw that connection, so it&#x27;s misleading.<p>Stuff like this happens because people treat extra rigor as pedantry and are comfortable making supporting assumptions that aren&#x27;t supported by data. The people making fraudulent requests aren&#x27;t aware that they&#x27;re fraudulent (usually). In my experience, they just think they&#x27;re being practical.
评论 #18367905 未加载
LeonBover 6 years ago
It’s “nearly 1 in 4” in the text, but I guess they altered the statistics slightly to make their case look better.
postitover 6 years ago
That happens, a lot. Not only statisticians.<p>My friend was half way over her 5 years vesting with a startup as when the CFO asked her to help then improve their numbers due a foreseen investment round. The idea was basically bump revenue basing it solely on the GMV masking returns and not calculating discounts and shipping. They also wanted to hide running costs by forcing vendors to agree on a 90+ days billing cycle, they also pushed the CTO on turning off parts of the system during weekends and holidays and forbid PTO until the deal was closed.<p>She refused doing the number masking and was asked to leave.
评论 #18367691 未加载
dev_dullover 6 years ago
The troubling part of this article (and <i>especially</i> the comments) is how easy it is to be labeled “anti science” today when questioning certain studies. I’d hate to think our culture is causing us to lose the required academic rigor to make data-based decisions.
评论 #18368688 未加载
评论 #18371503 未加载
wwhuangover 6 years ago
Very interesting study. I&#x27;d love to see a follow up on 2 things:<p>1) Did the statisticians refuse or comply with the request?<p>2) When they refused, how did the requester react? Were the requesters actually malicious, or just bad at statistics? If everything was fine once the statisticians explained that removing &quot;just an outlier&quot; wasn&#x27;t a valid option, then this report isn&#x27;t quite as concerning, and is maybe just an indication that more researchers need to hire statisticians to help them out.
IronWolveover 6 years ago
I was working on the Cingular and ATT merger, my upper management asked me to fudge the numbers on roaming. I would gave the numbers in blue and orange, and blue (att) had better data handoffs on 3g. It made orange look so bad, (shitty network), they made me combine them, to which tanked ATT&#x27;s network stats to the c-level.<p>One of many times, stats made someone look bad, and they made me change them.<p>Most the time, support metrics are altered to make it look like support contracts are hitting everything contractual.<p>This is daily business everywhere in tech. Mostly fudging to downright lying, just depends on the importance of the data if some mone is tied to it.<p>I don&#x27;t approve of this shit, but lucky, I never been asked to commit fraud, like reporting sales that don&#x27;t exist...
评论 #18367850 未加载
DrNukeover 6 years ago
Sensitivity &#x2F; ablation studies plus datasets opensourcing should be made mandatory for every computational proposition and research asap, so that results can be checked by peers and by the community at large. With publish-or-perish pressure, democratisation of tools, rogue parties and hyperspecialisation, is it getting to the point almost nothing would sustain any stringent scrutiny? Science cannot afford opacity any more, more than that in these blurred times.
DoreenMicheleover 6 years ago
<i>Of the less serious offenses, 55% of biostatisticians said that they received requests to underreport non-significant results.</i><p>Sigh.
raincomover 6 years ago
That&#x27;s why one should big consulting companies like McKinsey, Bain, BCG, etc, who don&#x27;t need to use hardcore statistics to prove whatever the board&#x2F;CEO wanted in order to execute some policy.
评论 #18367477 未加载
评论 #18369004 未加载
phaemonover 6 years ago
...in the USA, in medical research...
评论 #18366995 未加载
BenoitEssiambreover 6 years ago
This didn&#x27;t even cover the most frequent type of fraud which is to redo the experiment, or reanalyze the data until a sample of noise looks like a signal and then pretend you found something (as illustrated here: <a href="https:&#x2F;&#x2F;www.xkcd.com&#x2F;882&#x2F;" rel="nofollow">https:&#x2F;&#x2F;www.xkcd.com&#x2F;882&#x2F;</a>).<p>I got out of academia and stopped trusting most university research because I observed too much of this culture of fraud.
weberc2over 6 years ago
I wonder how prevalent this sort of thing is in other fields. I also wonder how prevalent fraud is generally, and the extent to which it is driving our decreasing faith in the academy. Also, how much (if any) of this fraud is motivated by political (instead of strictly personal) outcomes.
Dowwieover 6 years ago
What is the &quot;American Council on Science and Health&quot;? See here: <a href="https:&#x2F;&#x2F;www.sourcewatch.org&#x2F;index.php&#x2F;American_Council_on_Science_and_Health" rel="nofollow">https:&#x2F;&#x2F;www.sourcewatch.org&#x2F;index.php&#x2F;American_Council_on_Sc...</a><p>1 in 4 Statisticians weren&#x27;t asked to commit scientific fraud. Within the article:<p>&quot;Researchers often make &quot;inappropriate requests&quot; to statisticians. And by &quot;inappropriate,&quot; the authors aren&#x27;t referring to accidental requests for incorrect statistical analyses; instead, they&#x27;re referring to requests for unscrupulous data manipulation or even fraud.&quot;<p>This isn&#x27;t even remotely close to what the title of the article claims.
DoreenMicheleover 6 years ago
I have serious health issues. I&#x27;ve gradually gotten healthier, in part because I am skeptical of a lot of studies. I know how borked so many of them are, not a thing most people want to hear me assert. I get a metric fuck ton of flack over my skepticism.<p>I kind of have a mental box where I squirrel away little tidbits that meet two criteria: 1. They seem to come from rigorous studies and 2. They also fit with my general understanding of how life works.<p>Over time, I mentally group things -- a la A and B seem related -- without assuming that I know how they relate. I seem to have an inordinately high tolerance for ambiguity.<p>Most people seem to need An Answer even if it&#x27;s wrong. They have two categories -- black and white -- and when presented with purple or pink or blue, they force fit it to one of their existing categories and don&#x27;t confuse them with the facts.<p>I wasn&#x27;t trying to prove anything to anyone. I was just trying to deal with my life. But having gotten substantially healthier, I now wonder how to talk about such. It seems a wasted opportunity for the world for me to not share, but the world has treated me pretty horribly and done all in its power to tell me to STFU, I&#x27;m just crazy and spouting nonsense.<p>So I sometimes think I should write what I understand to be true and then carefully back it up with citations to try to support it. Then I read articles like this, throw my hands in the air and go &quot;Why bother?&quot;<p>The way I have been treated seems particularly unfair when you learn how much &quot;real scientists&quot; cook the books. Like why? It feels like pure prejudice when I read things like this.
评论 #18371547 未加载
jimnotgymover 6 years ago
1 in 4? How do we know that number is not made up?<p>In fact there is a 25% chance that the person involved in working that out has been asked to commit fraud at some point.
评论 #18369938 未加载
usgroupover 6 years ago
Another statistic just in: 3&#x2F;4 statisticians lie about not committing scientific fraud
alurenover 6 years ago
To be honest I&#x27;m surprised it&#x27;s not higher. Relevant discussion: <a href="https:&#x2F;&#x2F;news.ycombinator.com&#x2F;item?id=17789308" rel="nofollow">https:&#x2F;&#x2F;news.ycombinator.com&#x2F;item?id=17789308</a><p>The bottom line is that there&#x27;s not much incentive for doing the boring statistical validations (I can tell you that <i>no one</i> likes doing statistics apart from statisticians, and not even all of them) and verifying that everything is reproductible, and a <i>huge</i> incentive for, let&#x27;s say, &#x27;arrange&#x27; a thing or two so that the paper looks better. So many people just kinda do it, and it slips through the cracks because:<p>-In many fields, the reviewers are not statisticians themselves<p>-No one really bothers to download the data, setup the environment and libraries, and reproduce the exact steps taken to obtain the very same figures in a paper. Which is understandable given how doing all of that can be such a chore. Anyway, most of the time the exact steps aren&#x27;t even described. No, jupyter notebooks aren&#x27;t a solution either (it would take too long to explain why and the post is already long).<p>-In many cases, the results turn out to be true anyway so people don&#x27;t notice they were initially put forward with fraudulent validations<p>-When they turn out to be false, people just shrug and move on with what&#x27;s actually true. Bogus results often fail to stand the test of time and get forgotten quickly despite initial hype. No one bothers to say: &quot;Hey, that paper from 8 years ago is bullshit and their authors are hacks!&quot; because nobody cares.<p>-There&#x27;s a huge psychological barrier to actually call out one of your peers and affirm that they&#x27;re an impostor and their work is bogus. Especially when said impostor happens to be a big name in the field, that part of you still doesn&#x27;t believe they would commit fraud, not to mention the social repercussions and backlash of doing such an accusation. We scientists aren&#x27;t a very confrontational bunch.<p>So most of the time it kinda works and we&#x27;re all bumbling along hoping to find some modicum of truth at the end of the road with minimal harm done. But of course sometimes you get these guys who kick off their entire career on a high profile, much hyped fabricated result (Woo-Suk) or even a <i>series</i> of bogus papers (Sato), and <i>that</i> may lead to long-term harm. The good news is that hyped papers or very prominent figures quickly attract scrutiny from their peers, and sooner or later reality catches up with them as labs around the world fail to replicate their &#x27;breakthroughs&#x27;.<p>All in all, I&#x27;d say we&#x27;re doing fine. We&#x27;re just not the ethereal source of truth that some people hold us to be, the very same people who, after claming that &#x27;God is dead&#x27;, are very quick to replace Him with His sillicon-based equivalent around which we would act like priests, except with lab coats in lieu of clerical garments.
评论 #18367383 未加载
评论 #18368840 未加载
评论 #18367445 未加载
tomahuntover 6 years ago
&quot;Can you make the error bars a bit smaller please?&quot;.... &quot;um no, I&#x27;m affraid they are as I&#x27;ve calculated &quot;
sjg007over 6 years ago
Is this result statistically significant?
tomlockover 6 years ago
What&#x27;s odd to me is that some people seem to, even in the face of this, think it&#x27;s fair to criticize liberal arts journals for sometimes publishing bad papers. People are just people, in every field.
评论 #18367154 未加载
nathan_longover 6 years ago
&quot;1 in 4 Statisticians Say They Were Asked to Commit Scientific Fraud&quot;<p>It would be hilarious if this report were debunked on the grounds that they inflated the number of statisticians surveyed
ACow_Adonisover 6 years ago
... and 1 in 4 not prepared to answer truthfully, 1 in 4 not having worked in the field long enough, and 1 in 4 not competent enough to know the difference.<p>&#x2F;haha, only serious.
agentofoblivionover 6 years ago
How do I know this study isn’t part of the 1&#x2F;4?
rwilson4over 6 years ago
Most statisticians work for people who are trying to sell something. Statistics is not always the best way of selling something.
anarchopover 6 years ago
I think that this should prompt scrutiny in fields that are highly dependent upon statistical analysis, such as climate change. However, this very suggestion is intolerable to many and in a way equivalent to blasphemy against the church science. Scrutiny will not be tolerated by academe with regard to ideas held as sacrosanct. I find this culture in academia pretty disgusting.
TwoBitover 6 years ago
I wonder what the stats are in China.
评论 #18367132 未加载
jiveturkeyover 6 years ago
&#x27;25%&#x27; would have been a more appealing headline ... :D
jamesb93over 6 years ago
Is this fraudulent stats?
coldteaover 6 years ago
The other 3 in 4 just lie about it.
评论 #18367008 未加载
edooover 6 years ago
I can&#x27;t help but wonder what the rate is that didn&#x27;t admit to being asked to commit scientific fraud because they did.
orthrosover 6 years ago
And the other 3 lied
评论 #18367602 未加载