The natural selection of good science

Alexander J. Stewart, Joshua B. Plotkin

Research output: Contribution to journalArticlepeer-review

8 Citations (Scopus)
13 Downloads (Pure)


Scientists in some fields are concerned that many published results are false. Recent models predict selection for false positives as the inevitable result of pressure to publish, even when scientists are penalized for publications that fail to replicate. We model the cultural evolution of research practices when laboratories are allowed to expend effort on theory, enabling them, at a cost, to identify hypotheses that are more likely to be true, before empirical testing. Theory can restore high effort in research practice and suppress false positives to a technical minimum, even without replication. The mere ability to choose between two sets of hypotheses, one with greater prior chance of being correct, promotes better science than can be achieved with effortless access to the set of stronger hypotheses. Combining theory and replication can have synergistic effects. On the basis of our analysis, we propose four simple recommendations to promote good science.
Original languageEnglish
Number of pages11
JournalNature Human Behaviour
Publication statusPublished - 17 May 2021


Dive into the research topics of 'The natural selection of good science'. Together they form a unique fingerprint.

Cite this