Why 100 'Landmark' Cancer Studies Couldn't Be Replicated | NIH & NCI Directors
What this video covers
A short panel exchange featuring Bhattacharya alongside the director of the National Cancer Institute, in which the two are asked which landmark studies in cancer and chronic disease they would most like to see replicated — and turn the question into an argument about how replication efforts should be chosen.
Key points
- Bhattacharya cites a replication effort covering about 100 landmark studies in cancer biology, which found that most of the randomly chosen papers did not even describe their methods in sufficient detail for a replication to be attempted — “a major problem,” he says.
- He argues the replication revolution should not mean the government picking and choosing which studies to go after, saying he has seen that kind of power used to target disfavored people.
- He says the underlying lesson is that a result should not be believed simply because it was published; replication, not publication, should be the standard of truth in science.
- Instead, he wants the scientific community itself to identify the “rate-limiting step” ideas — findings where, if true, the field goes one way and, if false, another — and focus replication efforts there.
- The NCI director agrees on targeting the highly leveraged studies, adding that the replication crisis applies less to clinical trials, which are hard to overturn absent outright fraud, and more to the preclinical studies that drive those trials.