[Coral-List] push for more reliable research in ecology
allison.billiam at gmail.com
Mon Dec 21 21:59:25 UTC 2020
Doesn’t Rupert’s experience reflect how science is supposed to work -
An experiment is conducted and presumably published.
The experiment is replicated,possibly getting different results.
Further research ensues to resolve the issue.
The process becomes inefficient, and a lot of paper, ink, or pixels are
wasted, when the initial experiment or field survey is too limited in scope
or duration or both to support the conclusions, as seems to be implied by
The publish or perish imperative or the ambition to make headlines seem to
have a lot to do with this.
On Mon, Dec 21, 2020 at 8:57 AM David Blakeway via Coral-List <
coral-list at coral.aoml.noaa.gov> wrote:
> I agree this is a major problem for science. Here's a link to an
> influential 2005 article concluding that most published research is false
> <https://www.ncbi.nlm.nih.gov/pmc/articles/PMC1182327/>. The argument is
> based on the probability distribution of false positives in statistical
> tests. We should not expect the outcome to be any better in research that
> does not involve statistical tests. There are just so many ways to be
> wrong. Although science is supposed to be self-correcting, correction will
> only happen with the type of concerted effort that is barely possible in
> today's 'hyper-productive' science (current estimates of >5000 publications
> per day, increasing at 8% per year)*.
> Charles Darwin had a relevant opinion:
> "*False facts are highly injurious to the progress of science, for they
> often endure long; but false views, if supported by some evidence, do
> little harm, for everyone takes a salutary pleasure in proving their
> falseness; and when this is done, one path towards error is closed and the
> road to truth is often at the same time opened*."
> Fair enough, but what happens when science's self-correction mechanisms are
> overwhelmed? That would have to increase the likelihood of false views
> becoming false facts without ever being fully tested.
> The *Acanthaster *example is instructive. A significant result in a test
> with n=30 would be considered pretty solid; write it up and move on. But
> caution revealed something deeper. If the expected result had come first
> though? Well, I don't think I'd be repeating the experiment in that
> situation (especially as it's December 21 already and the lab's only
> published 15 papers this year!!)
> *from a quick search, these numbers are so astounding I'm not sure they're
> Coral-List mailing list
> Coral-List at coral.aoml.noaa.gov
More information about the Coral-List