[Coral-List] push for more reliable research in ecology
fathom5marineresearch at gmail.com
Tue Dec 22 09:54:28 UTC 2020
Sorry Bill, my fault, as my message was a semi-personal response to you but
also sent to the list.
Yes Rupert's example is good science. But in that example he was the one
doing the replication. If he hadn't replicated the experiment, but instead
chosen to publish his initial significant result, who knows whether the
experiment would ever have been replicated?
And Doug is right, I really do believe a lot of published science is likely
false. Ioannidis' conclusions seem credible to me, and it's worth noting
that in the ensuing 15 years nobody has discredited the essence of his
conclusions, just disputed some aspects of the methodology and hence the
derived proportion of false science.
I will stop talking soon, as I think I'm even beginning to annoy myself,
but I would like to explain my perspective. I went from farming to
commercial fishing, marine biology, coral reef geology, then 15 years in
private consulting. But I always wanted to get back into academic research.
However, as I began to understand how modern academia works I realised I
would not be competitive in the employment market, primarily because I am
not productive in terms of research output. That led me to evaluate what
would constitute a genuinely productive scientific career, and the aspects
of science we have previously been discussing: the H-index, the publicity
machine, the forgotten history. I do realise that science needs all types
of scientists and that, while I'm complaining, thousands of people are out
there doing good science. Some scientists *can *produce a high volume of
quality research, and that's impressive. Some create their own path and
that can be impressive and even uplifting (e.g. onepeopleonereef.org/). I
would just hope there is some niche for a scientific tortoise.
By the way I did not provide any evidence for my earlier claim that the
H-index is an unreliable indicator of quality. The reason is that the
H-index is essentially a measure of popularity, and popularity is not
always closely correlated with quality. I would be very interested in
knowing the frequency at which scientist's most highly-cited publications
correspond to their own assessment of their best publications (citations
would have to be corrected for time since publication, methods papers
removed etc.). I'd be surprised if it came anywhere near 50%.
Peaceful Christmas wishes,
More information about the Coral-List