4 ms·
Yes because you don't have evidence to support your theory. You're generalizing with insufficient data.
by CardenB 11y ago
Yes because you don't have evidence to support your theory. You're generalizing with insufficient data.
- tbrownaw 11y agoI disagree. It's not insufficient data, it's uncharacterized data. If more-or-less everyone with exposure to both cultures is in broad agreement about the differences, that's not "insufficient". It's probably actually more data than any formal study would generate. But it's also lower-quality data, with a greater chance of being distorted by various cognitive biases. Which means that it doesn't meet current regulatory standards[1] for use on large protected-class[2] groups of humans. . [1] as loosely and collectively defined by the morality police, loudest / most extreme non-ignorable activists, Internet Lynch Mob, etc [2] which includes pretty much anything these days
- CardenB 11y agoHow do you know if it's "more-or-less everyone" without a formal study to prove it? You only know anecdotally, which, as you said, is subject to various cognitive biases.
- michael_fine 11y agoBut there are plenty of things that we only know anecdotally -- that one restaurant has bad service, or that line cooks smoke more than usual -- that no one gets up in arms about when we generalize in those cases. Formal studies are great when they exist, but in the absence of them it's only practical to extrapolate from incomplete data.
- icebraining 11y agoDoes being practical preclude it being racist as well? Also, is generalizing about whole continents ever "practical"?
- michael_fine 11y agoI was more addressing the implication that because it's racist, we shouldn't do it. And as to generalizing about entire continents, if your goal is to avoid getting mugged on a late night, and you know that on average black people commit more muggings than anyone else, is it not practical to walk to the other side of the street if you see a black man (and that's your only information?)
- lightcatcher 11y agoIf you can generalize on similar problems and be correct more often than random chance, then you have a useful model regardless of the data backing that model.