ideasIdea
Random audits of published, funded research, like tax audits
Funders should randomly select a small fraction of the papers they paid for and check the raw data, analysis and records, with public results. The possibility of an audit changes behaviour.
Research misconduct investigations are triggered by complaints, which is slow and rare. Randomly selecting 1 to 2% of funded publications per year for a data and records audit (raw data availability, analysis reproducibility, image integrity, cell-line and reagent documentation), carried out by an independent office with published aggregate results, would create a deterrent and an evidence base on the prevalence of problems.
Hypothesis
Audits will find serious reproducibility or integrity problems in at least 5% of sampled papers, and adopting funders will see this rate fall by half over five years.
Rationale
Random audits deter tax evasion at low audit rates because the expected cost of detection is high; the same logic applies to research where career consequences are severe.
What would test it
One funder audits 100 randomly selected funded oncology papers per year; publish aggregate findings and track the rate.
Maturity
speculative
Who has to act
philanthropy
Cost to try
Medium ($1M to $50M)
Years to first evidence
2
Bottlenecks it attacks
- Preclinical results do not reproduce · Fewer than half of landmark cancer biology findings reproduce when someone else tries.