Comments by "Mark Pawelek" (@mark4asp) on "Academia is BROKEN! - Harvard Fake Data Scandal Explained" video.

  1. Peer-reviewers do not check calculations or raw data. A peer review takes about half an hour, and is just a reality check. Just a basic read.
    19
  2. Part of the issue here is the difficulty one has finding out exactly what happened: fraud or incompetence. Institutions need to start auditing raw data. They should also keep duplicates of raw data, recorded at source. There should be clear protocols in place such that there's always a copy of raw data available for audit. At least then these "dog ate my homework" excuses would no longer then tenable. Start randomly auditing data - looking for basic mistakes such as the silly mistakes noted above. If cheats thought they could get caught they may think twice. If they had the raw data themselves, they could even employ programs to check everyone's work - looking for duplicate IDs, and applying statistics to look for clear anomalies indicating possible fraud which warrant more detailed investigations. PS: I looked at one dodgy paper in energy policy 7 years ago. The raw data was copied from EU publicly available data, OK. They took a 7-year period and simply recorded the percentage change over 7 years, OK. 6th year maths I can do in my sleep. But the percent changes published were 95% wrong! - when checked. It wasn't me who found this because it would never occur to me that professors with PhDs can't do year-6 maths and that when they do basic maths for a new paper they don't check each others calculations. It took another - a PhD student to catch this.
    3
  3. She's cheated honest people - honest people who didn't get their paper published because it didn't make sexy claims like hers. Honest, boring, people who tell the truth to their students. Honest people who has their careers stymied by cheats.
    1
  4.  @SailingSoWhat  In the same way that hiring security guards, installing CCTV and alarms, putting hidden tracking IDs on wares, making your building burglar proof - all just makes thieves steal more carefully?
    1
  5.  @SailingSoWhat  Q: Why do they make up fake data? A: I don't think they plan to. I - more often than not - something goes wrong with the real data. For example, they plan to survey 50% male and 50% female - but the actual survey doesn't have enough of one sex. or the interviewees don't answer one question sensibly. By the time you find your data is unusable - they paper is almost ready for publication and it's either publish - or lose your paper - lose your grant, and maybe lose your PhD. So in policing this data collection by demanding they record the actual results at the time of collection - one plans to stop this retroactive data fiddling - which is so seductive. LOL - the researchers who wrote that 2016 paper I talked about previously claimed that their (bad) data was independent of their actual argument made in the paper - so - despite the obviously wrong data their paper should NOT be retracted! Maybe other forgers feel that way?
    1
  6. If you think no one will ever look at your data, you won't go to much trouble to fake it.
    1