Researchers should tick a checkbox “I swear I did not hack the data” before submitting a paper to a peer-reviewed journal to prevent this kind of misconduct.
Isn't this just basically fraud? I'm sure it's already covered by the existing things you sign, but surprisingly that doesn't stop people who are willing to commit fraud.
I assume that in this particular case, it's a joke referring to the subject of the study, which involved similar, if even weaker, assurances.
In general, this kind of thing is oddly common. It's all over government forms. I just interviewed with a Chinese father who wanted me to spend time with his children providing exposure to English. He asked me whether I had a criminal record.
I don't, but if I did, and I chose to lie about it, random Chinese parents would never know the difference. (Though entering China might have been a challenge.) Why ask?
> I don't, but if I did, and I chose to lie about it, random Chinese parents would never know the difference. (Though entering China might have been a challenge.) Why ask?
I think you'd be surprised how many people are really bad at lying; or even bad at acting normal when they think they have something to hide, particularly when asked such a question unexpectedly. Sure, some people with a criminal record may be able to lie plausibly when put on the spot, but there are a reasonable number who would give themselves away even if denying it. No real cost, some benefit, so why not ask?
Well, I sanitized my report of the interview. He actually asked if I had a bad history. I was confused and said I didn't understand. He clarified that he was asking about things like crime.
So the element of getting suddenly put on the spot doesn't really apply. That's already about as awkward as communication gets.
More importantly, though, I don't think the point is correct to begin with. This is fine:
> I think you'd be surprised how many people are really bad at lying; or even bad at acting normal when they think they have something to hide, particularly when asked such a question unexpectedly.
But job applicants with criminal records are only going to match this description once or twice. For the rest of their lives, it's not going to be an unexpected question, and they'll have lots of practice in denying their record if that's the way they choose to go. You're just never going to catch anyone out this way.
> But job applicants with criminal records are only going to match this description once or twice. For the rest of their lives, it's not going to be an unexpected question, and they'll have lots of practice in denying their record if that's the way they choose to go. You're just never going to catch anyone out this way.
You're right, you're only going to catch people out if you're the only one asking this kind of question... but right now that's more or less true of the people in your story. First-mover advantage. :-)
Why do forms require signatures if it's easy to forge (especially if you are submitting it electronically), and evidence of the agreement is gathered from other sources if it comes up in court?
Because it boils down what could have started as a nebulous mix of halfway-unethical actions into a single fraudulent act that can be pointed to and punished later, and will provide a lot of people with the impetus to back out of the fraud once they see themselves about to commit an unambiguously illegal act.
Maybe, but engaging an American company to run a background check on an American would represent a significant commitment for the typical Chinese person. It's not a realistic idea.
They do, but those "Reporting Summaries" are useless makework, IMO.
First, you fill them out after the data has been collected, analyzed, and written up. It's perhaps helpful as a reminder to include a few tidbits in the text (e.g., the ethics approval #), but literally no one is going to fill this form out, realize the sample size is way too small, and....abandon the manuscript.
Second, you don't actually want people to comply with the instructions. For example, it asks for "A description of any assumptions or corrections, such as tests of normality". A decent number of statisticians argue that you shouldn't be using normality tests to choose between parametric and non-parametric stats. On top of that, nobody actually writes out assumptions behind OLS in their paper either.
I am deeply skeptical that this cookie-cutter stuff actually helps in any meaningful way. It feels like rigor-theatre instead.
