so if you’re doing a study on people– like, to find out whether a behavior causes a certain health outcome– there’s the challenge of demonstrating whether it’s a cause, or only a correlation
(e.g. does your study show that “people who drink wine are healthier” because wine makes you healthy? Or just because people who drink wine tend to be wealthier and have access to better healthcare?)
And this can probably lead you to feel that the best way to structure a study, is to randomly assign people to do one behavior or another, so you can be sure the behavior isn’t caused by something that also causes the outcome.
But, this has its own challenges.
Like, if the behavior you’re studying becomes a whole different behavior when people are consciously doing it because they’ve been assigned to.
Case in point: That study that showed that “people in general just don’t have any detectable skill at recognizing whether someone is flirting with them or not”
and this was studied by…. putting people with a random other person… and randomly assigning them to “flirt” or “not flirt” with that person?
which to me feels like it… probably wasn’t studying flirting at all?
Or, if it was, it was studying only the type of flirting that’s done as a calculated choice, to achieve an objective that’s unrelated to one’s actual feelings of attraction or lack thereof.
i.e. “the kind of flirting that the sexy woman character in a heist movie does to distract the guard”
Not the kind of flirting that the vast majority of real people in real life worry about whether they can identify… because of how it’s directly connected to the question “does this person like me”