News for hackers. Stuff that matters.
Re:50 Years Ago, the Sugar Industry Paid Scientists to Blame... by kovek on Friday January 18, 2019 @10:22PM (#18945125)
Slightly related to research and meta-studies. The other day I was wondering: what stops a researcher from publishing random numbers (following some distribution they pick) as the results of their experiments?
Re:50 Years Ago, the Sugar Industry Paid Scientists to Blame... by MertsA on Friday January 18, 2019 @11:28PM (#18945335)
Reproducibility. What's worrying is the surprisingly large number of papers that are not reproducible. Most of that is likely not because of fraud though.

https://www.sciencedirect.com/science/article/pii/S002228281...

Re:50 Years Ago, the Sugar Industry Paid Scientists to Blame... by jowiar on Friday January 18, 2019 @10:49PM (#18945218)
The general thing that's supposed to stop this is replication -- when an experiment is published, another researcher should be able to replicate the results.

Here's an example of researchers being caught doing exactly this: https://en.wikipedia.org/wiki/When_contact_changes_minds

Re:50 Years Ago, the Sugar Industry Paid Scientists to Blame... by WhompingWindows on Friday January 18, 2019 @11:49PM (#18945389)
Not unheard of, but usually there are credentials and reputations behind researchers' names which other peer-reviewers/editors will know. Many articles are invited articles or follow-ups to previous articles. Seems very unlikely that most reputable scientists would put their careers on the line for provably fraudulent data when they can just p-hack or use biased methods on actual data. It's much easier to get caught fabricating, than cutting corners or using biased approaches.