“The rush to celebrate “eureka” moments often
overshadows a rather mundane activity on which science depends: repetition.”
After reading several articles on the possible crisis
of reproducibility, I find that these articles seem to be written in a way that
suggests no one currently attempts to replicate experiments within their lab
before publishing an exciting result. To me, the idea of publishing without
replication sounds careless and is not a quality I anticipate in most
scientists. Certainly, the numbers they report on individuals’ inabilities to
repeat various experiments are staggering. Like these individuals, I have
failed to obtain the same result from repeating an experiment. That’s just
science. Shit happens. However, among the seven scientific mentors that I have
had the pleasure of working under, I have never gotten the sense that
scientists tend to publish without verifying results. This may be a small
sample size, but it is these mentors and others that instilled a sense of the
importance of replication, both with the same experiment and with other
methods, in myself.
In my lab, obtaining an exciting result is met with
intense interest… and skepticism. From the moment we find such a result, we begin
working to falsify it. This includes both replication of the initial experiment
as well as finding ways to show the same finding using other methods, all done
with plenty of controls. If these attempts prove futile, then we become truly
excited. We do this because my PI would sooner see us spend more time to be
confident in our findings than have to retract a paper after its publication. Personally,
I am a huge proponent of showing the same result using different techniques. Perhaps
this opinion is a side effect of my own paranoia but, in science, we sometimes
forget that our techniques are often prone to error when an exciting result
appears. Further, this method for improving data reproduction was not mentioned
once in the four articles I read on the subject of the reproducibility crisis.
Yes, this method requires time and resources, but so do all the other methods
proposed in these articles. A few extra resources to offset the supposed $28
billion wasted on non-reproducible results is a small price to pay.
No comments:
Post a Comment