In my book, I spend a lot of time talking about the Bayer Reproducibility Study. In this study, Bayer scientists were assigned papers describing important scientific discoveries and were challenged to try to reproduce the results of those experiments. Most of the studies were found to be non-reproducible. As a follow-up to that startling work, another group recently decided to see how reproducible are studies that specifically deal with cancer.
Unfortunately, the results were even worse. Dramatically worse.
Of the 53 studies they picked, 47 were not reproducible. That is 89%. Now, this is a small sample size. And we don't know what the specific scientific backgrounds of the researchers who attempted to do replicate the experiments. And I believe that after the results of the first study, there has to be some level of bias in the minds of the researcher. I know that I am much more skeptical than I was before that study -- but 89%? That figure is simply frightening.
As scientists, a big part of our job is keeping up on the newest progress in our field of study. Most senior researchers spend several hours a day reading and being presented findings from other labs. What if this is true? Then the majority of information we're presented every day is not true?
If you've read my book, you know that I believe this is an inevitable result of how the cancer research system currently works. I believe that so much pressure is placed on scientists in this field to publish that junior researchers would rather publish work that is shoddy, or outright false, in order to continue their careers. To tell the truth, I am not surprised that cancer studies are less reproducible than scientific studies as a whole. I am shocked to think that it might really be this bad. You can read more about this work at Reuters.
No comments:
Post a Comment