There is a polish idiom "the lie has short legs", what means that sooner or later the truth will be revealed. I don't understant these guys, submitiing fake data. Every one can base on they're experimental data and get different results!

On the other hand, I have to agree with Mathias. Policy change is essential here. People work under great pressure.

The greatest inventions were obtained when the policy was totally different. Tesla, Faraday, Smoluchowski, Einstein were not working under so big pressure like nowadays. They had time to think, to do sometimes taime consuming experiments. Currently, If You're not publishing - You don't exist! And You have to chose to start totally new subject, or just start of new branch of research of a well-known topic. Then You will have to wait for greater benefits, or have smaller benefits right now (e.g. publications, awards, grants etc.).

When Maria Skłodowska-Curie was inventing a method of extraction of radium and polonium, she had to repeat her research with other researchers to prove that her method is repetitive and every one can obtain these elements with her method. Honestly, before publication, few manuscripts from current issue should be randomly chosen, and part of the research should be repeated by other researchers. It would generate costs, but the quality of science would be improved. And no one would dare to publish fake results.

I had one terrifying experience myself. I have got one manuscript for reviewing, which was invented from the first to the last sentence. I do not want to go into details, but it should have been obvious to anyone at home in the relevant topic. I was shocked by the fact that the same group has already published a series of similar papers that were accepted in other journals.

One great way to control fale data is changing the entire polices envolved in founding science around the world. Scientists usually fake data becausa they are under pressure ro publish even more and more papers to get their projects approved and obviously get the money for it.

I am not deleting the ethical behavior around this topic, but some scientist just don't have it and than they startto choose some wrong ways to do hw things...

There is a polish idiom "the lie has short legs", what means that sooner or later the truth will be revealed. I don't understant these guys, submitiing fake data. Every one can base on they're experimental data and get different results!

On the other hand, I have to agree with Mathias. Policy change is essential here. People work under great pressure.

The greatest inventions were obtained when the policy was totally different. Tesla, Faraday, Smoluchowski, Einstein were not working under so big pressure like nowadays. They had time to think, to do sometimes taime consuming experiments. Currently, If You're not publishing - You don't exist! And You have to chose to start totally new subject, or just start of new branch of research of a well-known topic. Then You will have to wait for greater benefits, or have smaller benefits right now (e.g. publications, awards, grants etc.).

It's funny, a Hungarian saying says: it is easier to catch a liar than a lame dog ...

There should be efficient institutional policy for publication in any journal. So that author can submit his work after the authentication from his institution. this can reduce the fake data to a great deal.

Rajneesh is right. But on the other hand - publishing would take a long time. I was wondering about ethical and moral issues concerning the publication. In my opinion only ethical people should be researchers. If not - fake data will flod the real, meaningful ones.

@Gyorgy There is also a polish idiom "Poles and Magyars are two brothers - can figth and drink arm in arm" :-))) I was in Budapest last year and I really like it!

I'd like to point out that this situation happens also at patenting, even thought lies and half-trues seem to be well accepted, given the mechanism involved with patents protection. As a matter of fact, USPTO does accept a "perpetual motion machine" patent, only if this iis shown to actually work. As Gregory House (from Dr. House serie) says: "Everybody lies".

Noone has yet mentioned refereeing. Perhaps some attention should be given to improving the quality of refereeing? I know that I reject a large number of the papers I am sent, papers that I know other referees would let through. Is peer review good enough (and I am assuming for these purposes that it is OK in principle)? That is, do the reviewers know their job? I suspect not, in too many cases.

In principle, there is no way to guarantee the avoidance of published fake results. Even the best refereeing will not certainly trap all these. Partly, the reason for this is that science essentially depends on good faith. Of course, important results will be eagerly replicated by many people - for these results liars will surely have short legs! But not-so-important results are not likely to be replicated very quickly. In such cases the liars may well have retired before their short legs are noticed! But in these cases perhaps it doesn't matter very much??

If scientists do not love truth we are all damned!

Does that mean there is a market for a new non-subject specific science journal that requires submission of samples along with articles for publication?

This is an interesting way of rephrasing the main subject. Asking for samples along with the to-be paper seems to make sense, even more regarding we all are supposed to daily deal with science, which by the way, must be tested. Nevertheless, it raises more questions, perhaps as important as the written formerly. What about nano-samples? Dark Matter? Radio-active materials? Higgs boson?

Hi Victor, good point on dark matter and Higgs Boson! The other two are doable though. Recently while reviewing an article which consisted largely optical simulation I tried to reproduce the simulation authors reported and could not get the enhancement factor of the amplitude of certain parameters reported in the article. I asked the editor whether s/he could ask the authors to upload the model file. Naivety of me? Editor said it is not the journal policy to ask for model files. We are light years from asking for a sample of Higgs Boson! I am sure.

Dear Ewa, Absolutely! And I think that underscores another point -- my dilemma of not been able to reproduce the same results in two weeks time (within which I have already received a reminder that my report is due) that probably the authors took months, even a year or so to produce that article. I have been reviewing articles for the last 2 -3 years and I get petrified every time I submit one. What if I am wrong - either way?

:) Figure of speech Thomas!! Thanks though. Terayears it is!

Reviewing cannot proceed on the assumption that the authors are lying. Everything cannot be proved! Asking for "model samples" is standard practice in the big journals ("supplementary material") but this is there to give necessary detail to the abbreviated treatments presented in the main article. The devil is always in the detail! But brevity is a virtue, and having a space restriction for the argument is a good writing discipline.

The job of the reviewer is to check that the argument is coherent in its own terms, and that it is reasonably plausible; that it is "timely and significant", and that it represents the state of the art fairly. Authors (not reviewers) take responsibility for their own work.

We cannot always "control fake results", not by reviewing anyway (although some egregious examples ought to have been trapped!), since good faith is basic to the search for truth. Only time will ultimately tell. (But, as John Maynard Keynes famously said, "The long run is a misleading guide to current affairs. In the long run we are all dead" (1923)!)

There are somebody here suggesting the video recording of the results and product in publications. While this may be a good way to prevent the fake date (to some extent at least), this novel approach needs a long time for researchers to accept. Indeed, a new journal called "Journal of Visualized Experiments" is recently coming out, however, not too many researchers are willing to submit their manuscripts in such as a way, and also this new journal requires a high cost of publication fee. But, as mentioned by others, is it enough to prevent data fabrication? The answer it "NO". The first step to revolve this problem relies on the authors themselves. The morality and self-discipline are the fundamentals to this problem.

As we are carrying out the experiments, we may notice that most of the study cannot be actually repeated. Sometimes the data is fake, and sometimes the critical techniques are missed. I think, the whole society is responsible for this problem. People, sometimes, have no choice but to seek for as many publications as they can to get promotions, new hirings, and graduations, so they are becoming more and more mad and finally fake the data and written the manuscripts irresponsibly. The researchers should be given a more relaxed and less pressured working and studying environment, so they can slow down their pace and focus on their research.

Hi everybody good comments about it, sometimes the confidentiality of the results or, method, takes the researcher to invent the experiment and obtain a fake result, and i agree that some researchers have a lot of pressure for write papers. The main problem is not the fake paper is who start if the student or the principal researcher, we don’t forgeting that the student learns from the assessor.

As an Emeritus professor I finally have the time to follow and enjoy the discussions in RG. I admit I published fake date about 20 years ago without knowing it. I even got a patent granted for the described effect: potentiometric immune assay (PIA). According to “my theory” one has to establish a mixed-potential situation on an ion-selective PVC membrane electrode (e.g., potassium electrode) between potassium cations and lipophilic benzoate anions. This is a very unstable electrochemical situation. The binding between an antibody and its larger antigen at the phase boundary of the membrane electrode should result in a reproducible potential change due to its different influence on the two ion-fluxes (potassium and benzoate) into the membrane phase. Two master students were successful and delivered me the proof of my theory. I was happy and filed a patent. My academic competitors started already forcing their students to deliver similar positive results. Then, one year later, a Ph.D. student should broaden-up the theory with other mixed-potential situations, because several publications described an electrochemical sensing of the immune reaction with ion-selective field-effect transistors. After a few months the Ph.D. candidate came in my office and announced that he could not repeat the work of the master candidates! I immediately ask the one still at my university how the results were obtained. The original recordings of the voltage of the electrochemical cell were presented to me. I saw potential changes after the immune reaction was started but with a more or less pronounced drift towards an end potential. Asking what value was chosen for the proof of “my theory” I got the answer: when it fits my expectations……! By chance I had to present a plenary lecture on a world congress on biosensors and I firstly presented the findings of the two master students including the successful analysis of a certified reference material! At the end I asked the audience if they found something less plausible or if they would have doubts about the novel and very simple method. No comment was made. Thereafter I confessed that I was too enthusiastic about the apparent proof of my theory and that obviously pure pressure was done to the students who wanted to please me with in their eyes a little “corrigé la fortune”. Having said this, I encountered several times the phenomenon that even renowned researchers escaped into new fields when their published predictions and/or hopes did not happen.

I find important that experimental science or engineering courses incorporate plenty of laboratory work. There was a clear disinvestment on practical laboratory formation in the last years, at least judging from my personal view centred at my own country (Portugal). On the other hand, particularly after Bologna reform, students are pressed to play as 'researchers', when they clearly lack the necessary laboratorial practice for good quality scientific work. It is very important that students first acquire laboratory proficiency and clearly understand and respect the principles of the scientific method. This is hardly achieved at universities that are much more concerned with quantitatively producing as many 'papers' as possible, rather than forming excellent quality scientists or engineers.

It is also very important that science policies are comprehensively revised, particularly in what deals with research supporting criteria, which, in my opinion, is extensively failing to promote high-quality science and technology research.

A very recent approach to detect fraudulent research has been devised by Uri Simonsohn, and has been applied to data in the social psychology field, which has seen some spectacular cases of fraud in recent years.

'Simonsohn’s method rests on the assumption that if a study measures the same variable in groups that supposedly are independent samples from the same population, chance will produce a certain variance between the means of those groups; if a paper reports means that have too little variance, that may be a sign that the data have been manipulated.'

Reference

Martin Enserink 'Fraud-Detection Tool Could Shake Up Psychology'

SCIENCE VOL 337 6 JULY 2012 , pp21-22.

This is of course a standard method that referees should always apply to all data. Referees should look for plausibility and credibility; for example, my own recent paper (Analytical Chemistry 84, 2012, 6061-6069; http://dx.doi.org/10.1021/ac300904c) had three independent measurements in different labs together with an explicit and carefully constructed Uncertainty Budget including an estimated variance. The measurements had a much smaller variance than the Uncertainty Budget estimated. I simply said that this too-good agreement was "accidental". But I didn't get any grief from the referees on account of it. Basically, I hadn't proved my point!

In mitigation I should add that the paper was not aimed at providing a certified sample, but demonstrating that the method for providing such samples existed. Strictly, there is a calculable probability that the observed variance of a sample from a population will vary by a given amount from the variance of the population, but referees do not habitually look sufficiently critically at these sorts of data.

This may be due to the pressure given by the institute or univ.,

One of my teacher use to say that 'Don't force the students to give data immediately, they may give you fake data with positive results, So, move along with them as like their friend and collect the data from them'.

When I was doing my PhD, my supervisor would pick out a sample or two at random, or a computation model (or a crystal, whatever we were working with) and test himself. Whether its just a melting point or a NMR or the whole shebang of second harmonic generation measurements -- he will do it himself, sometimes working through the night, and yes, I had to be present when he would be doing experiments / checking my samples. At times he would take me at the back of his bike and take me to another lab and do this! I hated it. I fought with him over this, it seemed downright insulting at that time. 10 years on I think it helped, that extra time he put in kept me honest and taught me a lot. He did that for all of 5 PhD students he had at that time. I try to follow a very similar although much gentler approach with students that I am responsible for now. Supervisors has got to put in place check points to make sure the data that is going in the paper is "good"!

„Trust is good but control is better“: this sounds good; but in many cases the advisor has often at the same time more than 10 Ph.D. students, and her/his time is being also heavily used for teaching and writing proposals to pay them a decent salary. Finding new and interesting topics and helping them when technical problems have arisen will also take a lot of time. Thus, working her/himself to check the experiments is very time-consuming in a “publish or perish” environment. One way to prevent false claims would be to wait with the publication until several other students, which should extend this specific research direction have gotten the same results as a starting point. That several Ph.D. candidates at different times all will produce fake data seems to me unlikely.

Felices fiestas y feliz año 2013, lo mejor para el siguiente año

That's the problem science is facing but nobody cares, you need publication for jobs, Degrees, Promotions, Increments and a many more...... so how can you expect new research outputs from every one every year? If competent authorities forces you then they are to be blamed not individuals.

Say in INDIA alone 700-800 thesis are submitted each year for masters and Ph.D involving New Drug Delivery System. All of them have published at least one research paper. Now tell me how many new products came to the market?

What I am trying to say is " Today's research works are publication oriented". So, this problem will be there until we change our Reason for Research.

This recent event shows that this discussion is still relevant.

http://www.theverge.com/2013/10/3/4798840/fake-research-paper-exposes-poor-standards-open-access-journals

See also:

http://www.nature.com/news/2009/090615/full/news.2009.571.html

Constantijn Seys : very interesting article.

In 2009 very senior and highly reputable professor in Nanotechnology told in a forum, I was also there, that one time he attached a wrong manuscript accidentally which contained an undergraduate third class conference paper for submission to an good impact factor journal. Strangely it was accepted without minor revision. Later on he himself rejected his paper for the sake of his reputation,

Years ago, just before I quit academia to do honest research and and just after I was inducted into Sigma Xi, I wrote On Honor in Science . I haven't looked at that paper in years and I am not sure what I wrote there; but, you can bet that the problem will never be fixed until capitalism and the profit motive is a thing of the past . If we wait a little longer, capitalism will collapse of its own weight; but, it is not clear that there will be anything left of civilization by then.

http://www.dematerialism.net/On%20Honor%20in%20Science.html

Until publish or Perish mentality of employers and research institutes are changed, science always will face threats.... Whom to blame ??? The researcher who is just trying to survive???? I don't think that will be fare. Again by these publications the scientific fraternity is on the loosing end. We need to understand this simple fact and set targets for researchers that are achievable.

Fake data in our research journals is not so much a great threat as it is an added cost and inefficiency. Research publication is about teaching people methods and results they can reproduce and use. So science publication is self-correcting for any fake data. Of course this comes at the cost of directing effort and funds from real advances to discover that something published is wrong. It would be useful therefore to reduce even the small amount of fake data in current publications if this could be done at low added cost in time and effort. In contrast, if reducing fake data generates too much added cost for publications, reviewers, or editors then it would not be an improvement over the current self-correcting system. Experts in any area know by reputations whose work to take more seriously and where questions might be appropriate so some papers are largely ignored unless the controls and data are strong. No fake data can stand the test of time so it is a short-term problem and something worth keeping in mind but not something worth changing scientific publication over.

''In the wake of a report released last week (August 28) in Science, which showed that just 39 percent of 100 studies published in 2008 in the top three psychology journals could be successfully reproduced, researchers are ready to clean up the field’s literature. The results come from a collaborative effort of 270 scientists working for The Reproducibility Project based at the Center for Open Science in Charlottesville, Virginia.''

http://www.the-scientist.com/?articles.view/articleNo/43875/title/Psychology-s-Failure-to-Replicate/

Totally disagree John, Fake data  per se is a short term standing in the scientific society, but the person making the fake data may rob other scientist's opportunity in both career development and grant application. If they are successful in either career development or grant application, it will be a very bad demo to other people.