Tuesday, July 19, 2016

We need to talk about the bad science being funded

Bad science
http://phys.org/news/2016-07-bad-science-funded.html

 on the issue, published in Nature this May, found that about 90% of some 1,576 researchers surveyed now believe there is a reproducibility crisis in science.
While this rightly tarnishes the public belief in science, it also has serious consequences for governments and philanthropic agencies that fund research, as well as the pharmaceutical and biotechnology sectors. It means they could be wasting billions of dollars on research each year.
One contributing factor is easily identified. It is the high rate of so-called false discoveries in the literature. They are false-positive findings and lead to the erroneous perception that a definitive scientific discovery has been made.
This high rate occurs because the studies that are published often have low statistical power to identify a genuine discovery when it is there, and the effects being sought are often small.


Read more at: http://phys.org/news/2016-07-bad-science-funded.html#jCp

I know a guy, who teaches statistics to users of  one of the high-end statistics programs.  He says, the program is excellent, but that he does not trust any of the results produced by his students (mostly scientists and statisticians working for large research organizations.)   My friend would add poor sampling protocols to the issues identified by Simon Gandevia.  However, my friends job is to teach the mechanics of using the program, rather than the theory of statistics and sampling.

My view is that any significant result can be identified with rather simple statistics - if sampling was unbiased and adequate samples were collected.

If different appropriate statistical protocols can yield different results, then sampling was inadequate or there is no effect.   If there is a real effect then all the appropriate statistical protocols when used correctly will yield the same answer.

No comments: