Thursday, August 9, 2018

The scientific crisis of reproducibility

I first ran across the Journal of Irreproducible Results back in the 1980s, but it was already 30 years old then.  It was a funny science spoof magazine publishing goofy pseudo-scientific papers like The Optimum Number Of Friends In The Information Age.  It ranged from chuckle worthy to belly laugh, but only if you were a nerd.

The premise of the magazine is that some papers could only be published there, as they could never be replicated.  Well, it seems that this may be standard operating procedure for much of the scientific literature:
An ambitious project that set out nearly 5 years ago to replicate experiments from 50 high-impact cancer biology papers, but gradually shrank that number, now expects to complete just 18 studies.
It seems that it's expensive to replicate experiments and so the project is scaling back in the face of lack of funding. This during a period where there is increasing discussion that perhaps a majority of the published papers are spurious, and when scientific paper retraction is at an all time high.


SiGraybeard said...

I beat on this drum like the proverbial rented mule because it's personal.

In '97, my wife had breast cancer and the highlight (low light?) was a bone marrow transplant - actually transplant of her own stem cells - that took six weeks and had us living near a med school campus across state from home. The reason for the transplant is they give such a high dose of chemotherapy that it kills the immune system and bone marrow, so that has to be rebooted.

About two years after that treatment, the paper was retracted. It was academic fraud and the lead experimenter fudged his data.

At the time, my wife was in a support group of 12 women, all of whom were told they had a 75% chance of cure. My wife is the only survivor. 1 out of 12 ain't 75%. It's slightly better than that since "cure" is defined as being alive in five years and a couple of them made it to five years before a recurrence or another cancer that killed them.

The last couple died from leukemia - AML - that's caused by the bone marrow transplant.

For the scientists, and to mangle a Richard Feynman quote, I don't care how beautiful you think your idea is or how smart you think you are for thinking of it: if it doesn't agree with experiment it's wrong.

drjim said...

SiG said it far better than I ever could.

Since he and I are Engineers, we rail against shoddy data gathering, measurement errors, calibration errors, quantization errors, and poor prceedure.

And we see far too much of it by today's "scientists".

Aaron Bell said...

Is the 404 error at the Optimum Number of Friends part of the joke, or am I trying to read too much into a broken link?

Old NFO said...

Sadly Sig is right, and more and more of those 'proven' studies are shams...

Will said...

" It was academic fraud and the lead experimenter fudged his data."

Is this guy taking a dirt nap? If not, why not? Until there is a high price paid for this sort of fraud, there is insufficient deterrent to restrain the idiots.

Ken said...

Another problem (sort of illustrated by the article) is that there are few or no institutional incentives for replication -- journals don't care to publish replications, so in a publish-or-perish environment one focuses on what will advance/preserve one's career.