@Geir and all, partly it may be due to the way science works out in real life.  An example is Dalton's atomic theory.  His postulations were found to be limited, when more science was discovered.  

More recently, medical scientists thought that canola oil was as good as olive oil, but this is found to be false.  Many ladies were using aspartame sweeteners, until newer findings came into place.  Science is a never ending search for the truth of our environment an ourselves, and we are held back by our imperfect knowledge.  Perhaps the 'half that may be false' is the half where research is still needed to pry into it.  We must exercise patience, except with proven dishonest research.  @George, I agree scientists do jump to conclusions due to present imperfect work.

Geir -

It certainly does seem obvious that too much emphasis is put on isolated p-values and not enough on effect size in many areas, not just medicine. 

See https://www.researchgate.net/publication/262971440_Practical_Interpretation_of_Hypothesis_Tests_-_letter_to_the_editor_-_TAS

It also seems that in many cases there are so many variables that there is room for manipulation, especially disturbing when pharmaceutical companies, the petroleum industry, and others have so much to gain or lose. So science needs to be undertaken with especially great diligence.  When economists can make a mistake in a spreadsheet whose false results can be used to push a political agenda, as has happened fairly recently, diligence is especially important.

Perhaps the answer to your question might best be given in terms you quoted:  "...'Afflicted by studies with small sample sizes, tiny effects, invalid exploratory analyses, and flagrant conflicts of interest, together with an obsession for pursuing fashionable trends of dubious importance ....'" 

This seems to put the problem into three categories: (1) sloppy science, (2) unethical politics, and (3) popularity (which may be under both 1 & 2).  All of the problems listed above are in category (1), except perhaps the last two, and the other two categories are covered by the last two issues, respectively.

In all cases, there seems to be a good deal of concern on ResearchGate that the dubiously conceived "impact factors" of journals, which interact with monetary concerns, may be a big part of the problem. Let's look for a moment, however at category (1):

Well, small sample sizes again could be monetarily related, and conveniently for some, they help to make p-values larger.  (That is a good reason to use confidence intervals instead, when possible, which may be more practically interpretable.)  I have also seen an apparent lack of concern, I think, on ResearchGate questions and answers, regarding the estimated variance of the prediction error or other measures of accuracy for regression results.  It is hard for many to move past the often misused/misinforming p-value and volatile R-square as quick, but inadequate measures.  But this has gone on for a long time, and I think - though I may just be too optimistic here - that there may be some signs this is getting better.  That is very speculative, however, and a good scientific assessment of that would be nice.    ;-) 

Statistics can find correlations in data that may support a given theory, or not.  Causality has to be attributed to subject matter theory.  If the theory is not based on a scientific discipline, but instead on a conflict of interest, then this can be damaging.

So the 'false half' of the literature must be partly a somewhat random part of what falls under category (1), which can be addressed by better 'common sense' with regard to statistics and subject matter science, and a large part of what falls under categories (2) and (3), which can be addressed with better common sense in that regard. It may be something like reading an account of history: Who wrote it? What was their agenda?  Does it make sense in context with other accounts?  Are other views being stiffled?

Cheers - Jim 

Article Practical Interpretation of Hypothesis Tests - letter to the...

Dear All,

This is a very similar situation that John Ioannidis stated as “Why most published research findings are false?” Here there are a more detailed analysis than the very brief summary like paper Geir indicated.

http://www.ncbi.nlm.nih.gov/pmc/articles/PMC1182327/

https://en.wikipedia.org/wiki/File:Ioannidis_(2005)_Why_Most_Published_Research_Findings_Are_False..pdf

https://en.wikipedia.org/wiki/John_Ioannidis

I suggest you to read J. Ioannidis other paper:  “How to make more published research true. „

http://journals.plos.org/plosmedicine/article?id=10.1371/journal.pmed.1001747

Dear All,

By the way, if the mentioned publications are false or unreliable who will ever correct the false impact points of the authors?

@Geir and all, partly it may be due to the way science works out in real life.  An example is Dalton's atomic theory.  His postulations were found to be limited, when more science was discovered.  

More recently, medical scientists thought that canola oil was as good as olive oil, but this is found to be false.  Many ladies were using aspartame sweeteners, until newer findings came into place.  Science is a never ending search for the truth of our environment an ourselves, and we are held back by our imperfect knowledge.  Perhaps the 'half that may be false' is the half where research is still needed to pry into it.  We must exercise patience, except with proven dishonest research.  @George, I agree scientists do jump to conclusions due to present imperfect work.

Hi Miranda,

You made a good point. “We must exercise patience.” Yes, fully agreed. Edible oil is a very complicated (and highly political, subjective) and convoluted issue, and conclusions cannot be hastily and randomly drawn. Not black and white! It appears that canola (rapeseed) oil cannot be currently recommended as an appropriate substitute for extra-virgin olive oil. There was an issue with the isomers in canola (rapeseed) oil. There are three canola oils that are being studied: conventional canola oil (oleic acid-rich), high oleic canola oil (highest oleic acid content) and DHA-enriched high oleic canola oil (oleic acid-rich and DHA-rich). And, human studies are continuing to be published.  Jacobson excoriated tropical oils (palm and coconut oils). This continued for a number of years. Now the potential health benefits of medium chain fatty acids (from coconut and palm oil) are being increasingly appreciated. 

Dear Geir,

The following (though not directly germane to your question) is very profound, and may be related to the exponential proliferation of papers in one facet of  biomedical research!

What scientific idea is ready for retirement?

According to Dr Azra Raza (see link below), the mouse models of cancer.

http://www.theguardian.com/science/2014/jan/12/what-scientific-idea-is-ready-for-retirement-edge-org 

 Horton (2015): "The case against science is straightforward: much of the scientific literature, perhaps half, may simply be untrue." Does the other "half" also include "inaccuracies"? Theories are what they are; they are subject to intensive revision, refinement and redefining, and even outright rebuttal. Theories beget theories. Groundbreaking, revolutionary theories and hypotheses, which dramatically change the entire direction of science (to the magnitude of Dalton’s epochal atomic theory, although not directly related to the current question by itself) do not evolve and surface, and make their imprint and impact that often. As far as inaccuracies, one may reminisce the “Raman Effect” (not directly related to the current question either). C. V. Raman (Sir C.V. Raman), during his voyage from London, was incessantly fascinated with the deep blue color of the Mediterranean sea, and  his restless mind was not simply able to accept Rayleigh’s theory that the color of the sea was the mere reflection of the sky.

Some of the “other half” perhaps reflects the exponential, fierce and maddening pace of biomedical publication and the exuberance created by “publication begets publication”.  

I am a neurosurgeon, and have rescued many patients even in midnight operations. The good results of operation are highly supported by careful research works such as for brain edema, vascular obstruction, vasospasm due to SAH, etc. I am now retired from the operation due to the age, and contrarily I am now concentrating on the research works as mentioned before to support the good results of the various neurosurgical operations. It is not necessary to become dishonest. I am still looking forwards to usher the way to rescue patients.     

Good result for patient by using the new found method is the answer.  

Forms of DNA!

http://scripts.iucr.org/cgi-bin/paper?S0907444903003251

Acta Cryst. (2003). D59, 620-626    [ doi:10.1107/S0907444903003251 ]

A glossary of DNA structures from A to Z

A. Ghosh and M. Bansal

Abstract: The right-handed double-helical Watson-Crick model for B-­form DNA is the most commonly known DNA structure. In addition to this classic structure, several other forms of DNA have been observed and it is clear that the DNA molecule can assume different structures depending on the base sequence and environment. The various forms of DNA have been identified as A, B, C etc. In fact, a detailed inspection of the literature reveals that only the letters F, Q, U, V and Y are now available to describe any new DNA structure that may appear in the future. It is also apparent that it may be more relevant to talk about the A, B or C type dinucleotide steps, since several recent structures show mixtures of various different geometries and a careful analysis is essential before identifying it as a `new structure'. This review provides a glossary of currently identified DNA structures and is quite timely as it outlines the present understanding of DNA structure exactly 50 years after the original discovery of DNA structure by Watson and Crick.

http://nucleix.mbu.iisc.ernet.in/MBPublications/DNA_structure_yet_another_avatar.pdf

Every time the story of DNA structure seems to reach a conclusion, it bounces back to centre stage by appearing in yet another incarnation. The latest avatar to manifest itself is a stretched and overwound form of DNA reported recently by a French group1, working with single DNA molecules. When a moderately large stretching force (of about 3 pico Newtons) is applied, the DNA molecule apparently becomes highly twisted and extended, but even more amazingly it takes up an ‘inside-out structure’ in which the phosphodiester chain is on the inside and the bases are exposed (Figure 1 a).

Thus, while the basic principles and essential structural elements for protein structure, elucidated during the early 1950s, viz. the a -helix and b -structure from Pauling’s group and the coiled-coil triple-helical structure for collagen from Ramachandran’s laboratory have remained virtually unchanged and unchallenged even today, with only their permutations, combinations and linking regions varying, in the more than 300 unique sequence protein structures currently known, the structure for a single DNA molecule seems to be able to adapt itself to its environment by twisting, turning and stretching into completely different ‘avatars’. The last word in the DNA story has probably still to be written, but in the meantime Linus Pauling, whose biggest disappointment was that he did not discover ‘the DNA structure’, can now rest in peace, his structure has at last found a niche in the pantheon of DNA structures!!   

سوال مزخرف cheap question