Yoshitaka Fujii

[wolf63]

I am unable to sleep so visited Wiki. There is an article about the gentleman named above who has been found to have falsified information on 170ish research papers.

http://en.wikipedia.org/wiki/Yoshitaka_Fujii

If a scientist submits a research paper does nobody check the credibility of the data supplied? The data was queried at an earlier date and nobody seems to have taken the 'allegation' forward.

I don't know much about science but surely this guy is bound to have some sort of admin staff or infrastructure when he conducts research, would one of these people not have noticed an error or two in the figures.

I realise that I am looking at this with an untrained brain/eye but it does make me wonder if there are more scientists out there with a creative approach to research and statistics.

Can we be sure that research programmes are well policed?

[jake-the-peg]

There have been a number

http://en.wikipedia.o...Scientific_misconduct

This is why things have to be peer reviewed

Extraordinary claims tend to have to be reproduced too prior to general acceptance.

Take the Higgs

You had two seperate detectors CMS and Atlas both working in jealous secrecy coming up with the same result.

Not only that but it turned out to be in a low energy range and the Tevatron in the US was able to add weight although it could not independantly confirm.

Alarmingly a disproportionate amount of these seem to be drugs related -so much more to play for - in that case there are also regulatory bodies

[wolf63]

As you say $£€ talks and there is big money to be made, especially in drugs.

I took part in a research programme into finding a genetic link in Bipolar Disorder - it seemed to be well run. I had an MRI, psychology tests and got my hand printed in the same manner as a finger print would be. I am not sure what the final conclusion was - it was over 8 years ago.

Thanks for your answer.

[theprof]

There are a number of issues involved here wolf63 which are conducive to this sort of thing happening.

First of all, you have to bear in mind that in general, research papers are extremely specific to a certain area of research. The scientists who publish these subsequently discovered claims are not amateurs in their field. They are invariably recognised experts in their area of expertise within their own university. Often, very few of their colleagues will challenge their opinion and in general, their colleagues do not have sight of the material before it turns up in a scientific journal. As far as admin staff are concerned, these are far from being experts in the field under question and are not qualified or recognised to challenge the views of senior university scientists. ; after all, if admin staff were that knowledgeable, they wouldn't be working in admin would they?

I am a Science Dean and hold a professorial chair in Biochemistry at one of the top UK universities. I am also a visiting professor at various ivy-league universities in the US. I also hold professorships in canadian and European universities and have lectured in chemistry, botany, zoology and would you belive it, biochemistry during my career. But enough about me - from experience, I can tell you that when a paper is published involving statistics, you have to have very competent admin staff at your side to spot any errors. This is why in my university department, we have statisticians and other mathematicians employed to review such papers before they reach the scientific journal. I'm afraid that many other universities are not so thorough in requesting peer-reviewing or checking articles before submission.

The normal procedure is for the journals to publish these apparently genuine research articles and then wait for criticism regarding the methodology etc from other acknowledged experts in the field. These criticisms are published as "letters to the editor" during subsequent months. Believe me, I've worked extensively on both sides of the coin during my career.

In general, you can assume that most stuff published from scientists working in any of Russell Group universities in the UK or the ivy-league universities in the US has been peer reviewed, although I'm aware of a number of exceptions to this rule.

[wolf63]

prof - thanks for your detailed biography and explanation. I know that every day there are thousands of research projects taking place that are run ethically but people like Yoshitake Fujii give scientists a bad name.

I am Scottish and have extra genes which give me extra amounts of sarcasm and cynicism in my psychological make up. This means that my following assumption may be wrong.

"Surely the end result of a research programme should include lists of numbers/statistics. So if a scientist starts research on a subject that he/she, in their own mind, has already reached a conclusion over then this set of numbers/statistics could be manipulated to agree with the preconceived outcome."

Sorry if that doesn't make much sense.

[theprof]

No, I understand perfectly well what you mean wolf63, and I agree with you.

Statistics have been shown to have been manipulated in very many famous published papers. It is to be seen most often in apparently independent drug trials where a bona-fide group of renowned clinicians return their trial results and data to the drug company, only to have them manipulated or "smoothed-out" by the drug company prior to publication. The conclusions can be turned on their head using such methods but unfortunately, statistics manipulation is not that simple to detect even when you have the figures in front of you.

I have people working in my department who are famous enough in their own field to be recognised by a fair proportion of the public through TV appearances etc. ( including myself, I regret to say) Nevertheless, I fully support my departments view that no paper shall be published without it being reviewed by their colleagues and this policy has worked reliably for years. I've always worked on the basis that no matter what you know about a subject, there's always someone out there who knows more than you. Personally, this canon has always been at the forefront of my mind when I write scientific papers or edit academic tomes etc. I regret to say that not all scientists accept it.

[theprof]

Taking the p!ss mike 11111? Sorry but I don't do that on AB as it would ruin my reputation amongst my followers and I understand my reputation has been in place a little longer than you have been a member of AB.

I regret to say that I have never partaken of such nefarious activities. My initial BSc degree was awarded at a university where plagiarism and similar nefarious activities led to immediate expulsion of the student concerned. No appeal procedure took place and for many of us, this ethos kept us on the straight and narrow for many years.

My subsequent academic career has amply demonstrated how the situation has changed in subsequent years. Nowadays, we have software to detect documentary plagiarism to hand and indeed, it is very effective. Nevertheless, having been blessed with a good memory, I tend to rely on my recall of previously published papers in order to raise my own suspicions. By and large, it works.

If you really wish to provide cogent arguments regarding my comments, you must really try to expand upon the information you post and be more specific.

[jomifl]

I have known some 'scientists' choose a method of statistical analysis that favours the result they expect, though this gets found out at the first draught stage by their colleagues.
I have also seen 'scientists' so intent on massaging the stats. to support their views that they fail to see that the results show something completely unexpected and worthy of further investigation.
The truth will out sooner or later, only a fool doesn't realise that. Fortunately most scientists are pretty ruthless with regards maintaining a high standard of honesty, it is only those who work in small teams or by themselves that are able to deceive but only for a short while usually.

[atalanta]

Unfortunately, the easiest way to manipulate statistics is the hardest to detect. Scientists can simply omit the results they don't like, and keep the ones which suit them. The reviewers then do not know even that any other results ever existed, let alone what the omitted ones look like.
There are a couple of relevant articles in today's edition of the "Times Higher Education".

[theprof]

jomifl, yes indeed simplistic manipulation of statistics is readily detectable. We try at my uni to detect the anomaly prior to publication rather than subject the article author to a barrage of criticism following publication. Statistics remains an important subject for scientists and we teach a fairly advanced module to all science undergraduates at the uni, regardless or not they had a demonstrable ability in the subject at "A" level.

I have known pharmaceutical companies who have selected statistical data to publish and ignored other data in order to show their drug trials in a favourable light. Often, the lead researcher will raise his concerns with the company but the threat of ruination of his or her academic reputation leads to many researches backing off. By and large small research teams and individuals no longer perform major research because of these issues. Furthermore, most reputable universities have in place compliance rules that have to be strictly adhered to to protect the researcher and the university.

Thankfully, I was able to withdraw from sponsored research quite a few years ago as my young family at the time were more important to me than money. It was one of the best decisions I made. I do appreciate that others aren't so fortunate.