Publication bias is bad for science if not necessarily scientists

Area of Science:

• Philosophy of Science
• Scientific Integrity
• Bayesian Reasoning

Background:

• Publication bias is a significant concern, often linked to the replication crisis in science.
• The scientific process aims for unbiased, reliable inference.

Purpose of the Study:

• To investigate the impact of publication bias on scientific reasoning and the broader scientific community.
• To explore whether individual scientists can mitigate publication bias and the societal implications.

Main Methods:

• Utilized a Bayesian model of scientific reasoning.
• Analyzed the theoretical capacity of scientists to interpret literature despite publication bias.

Main Results:

• Aware scientists can theoretically adjust their beliefs to counteract publication bias.
• Individual scientists might not see personal benefit in policies mitigating publication bias.
• Publication bias negatively affects science as a social endeavor, particularly for non-expert intermediaries.

Conclusions:

• While individual scientists may adapt, publication bias undermines the societal benefits of science.
• Policies like mandatory pre-registration are justified by the societal benefits they enable, outweighing individual costs.
• The perspective of non-expert intermediaries should be prioritized when governing scientific institutions.