FRIHOSTFORUMSSEARCHFAQTOSBLOGSCOMPETITIONS
You are invited to Log in or Register a free Frihost Account!


Most published research is probably false?





Ankhanu
A PhD candidate friend of mine posted a link on Facebook to a little report on The Economist; Unlikly Results: Why most published scientific research is probably false. It’s a 1.5 minute infographic video highlighting the statistical likelihood of published results being incorrect due to type 1&2 errors (False positives, false negatives), and the likelihood of rejecting Ho at all. End conclusion is that due to these three factors combining, it’s likely that less than half of published results are likely true. It’s an interesting little expose.

I think the concern is somewhat inflated, however. As presented (I haven’t gone delving deeper), it comes off highly overgeneralized (it is), and seems to be making some rather silly assumptions and ignoring a major element of science - self-correction. It seems to assume that problems or hypotheses are tested once by a single researcher/lab and reported upon, and the truth of it just assumed… it ignores that research is an ongoing process and results are tested, and often the same problems are being tested in multiple labs at the same time. Yes, incorrect results get published… but science is a self-correcting entity. If something is reported incorrectly, it will be discovered and corrected… if false negatives occur, someone WILL catch it… eventually. Yes, some published research is false… but it will be corrected.

There is a certain connotation here in the value of reporting instances in which the null hypothesis is not rejected through experimentation… and a commenter upon the FB post dug up a reference to the International Journal of Negative & Null Results, which is an interesting and valuable concept. I’ll be doing a little investigation around it.
Null results are valuable results!


Thoughts??
Afaceinthematrix
Ankhanu wrote:
Thoughts??


Yes.

Ankhanu wrote:
I think the concern is somewhat inflated, however. As presented (I haven’t gone delving deeper), it comes off highly overgeneralized (it is), and seems to be making some rather silly assumptions and ignoring a major element of science - self-correction. It seems to assume that problems or hypotheses are tested once by a single researcher/lab and reported upon, and the truth of it just assumed… it ignores that research is an ongoing process and results are tested, and often the same problems are being tested in multiple labs at the same time. Yes, incorrect results get published… but science is a self-correcting entity. If something is reported incorrectly, it will be discovered and corrected… if false negatives occur, someone WILL catch it… eventually. Yes, some published research is false… but it will be corrected.


I don't have much incentive to spend too much of my time typing up my thoughts when you have already given an analysis that is pretty similar to mine. I only want to add one more point: so what? Science is about being wrong. You do experiments with the intention of trying to falsify your results or someone else's results. You observe, make a hypothesis, and then try to prove yourself wrong. The important thing is that you modify things that have been falsified.

I wouldn't even care if I was a science teacher and a student got a question that I asked incorrect - assuming that they had a logical reason for saying it. For instance, if I asked, in class, why some tigers are white and a student, not knowing that it was a genetic mutation, raised his/her hand and said that white tigers are probably a subspecies of Siberian tigers that adapted the white fur to blend in with the snow I'd be happy. He/she used previous knowledge and made a well-reasoned hypothesis. It doesn't matter that it is incorrect because the next step is experiment and experiment will self-correct.

Alchemy was an entire scientific field that is now discredited but was once treated as a real science and a good science (as in hypothesis and experiment). That incorrect science laid down some of the foundation for atomic theory/chemistry. Hell, almost everything that Newton did is wrong when velocities approach the speed of light but since we don't generally deal with the speed of light, it's still incredibly useful.
Iceaxe0410
I can agree with that. I highly doubt the majority of research is accurate especially considering that many experiments have to make assumptions or ignore factors to support their initial hypothesis. After all, money isn't made from research that produces no results. That's why experiments need to be replicated and peer reviewed. I feel that experiments and research is most useful when looked at as a collection of interconnected works that support or refute certain claims.

In the grand scheme of things, research simply strives to provide us with a better understanding of how things work or better yet how things don't work. It may not be completely accurate, but the hope is that time corrects and alters the inconsistencies until we have improved upon it to the point where no other conclusions can be drawn. After all, there can't be any progress if we try to make every experiment perfect with no flaws. There are always margins of error in any experiment, some greater or lower than others.
osamaterra
I think one has to distinguish between theoretical and experimental research
Related topics
Stem Cell Research
why i am not so sure about reigion
Philosophy Essays & Philosophy Texts
911
The difference between science and engineering?
How secure are anonymous proxies?
Who decides what is the course of stem cell research?
Have you read the Bible?
do people now a days still like to read?
Obama's Unemployment Numbers Keep Going Up
Similarities between FDR and George W. Bush
10 free antivirus programs for Windows
The Trinity (No.1 in the 'Dogma for beginners' series)
Baptism in the Spirit
Reply to topic    Frihost Forum Index -> Science -> General Science

FRIHOST HOME | FAQ | TOS | ABOUT US | CONTACT US | SITE MAP
© 2005-2011 Frihost, forums powered by phpBB.