3 ms·
How do we prevent fraud? Is there a way to change the incentives? Statistical analysis of results to detect bogus numbers? Only cite research that has been inde
by liminal 6y ago
How do we prevent fraud? Is there a way to change the incentives? Statistical analysis of results to detect bogus numbers? Only cite research that has been independently replicated and apply a "provisional" label to the first results? Then only cite the duplicating lab and not the first one, so the incentive is there for doing the grunt work?
- leovailati 6y agoI generally like the idea of rewarding replication studies. Replication/validation could be a required process that runs concurrently to peer review. The researchers who run the experiments to replicate results could rewarded by being added as contributors to the original paper, so they also get the citations. And like you suggest, the paper would then be marked as "validated" or something similar. I wonder if any journals out there are already doing something like this. There is, of course, the danger of collusion among original authors and validators. Hopefully the fear of having your results rebuked would prevent people from trying to publish bullshit in the first place. Another problem is logistics. Research labs have their own ideas they want to push forward, so spending time and resources proving or (even worse) disproving some else's idea doesn't sound that great. Also, even if it gives you citations, it probably wouldn't help you with your thesis. Anyways, it's a tough problem to solve.