‘Excited delirium’, used by police to justify brutality, is junk science

Post by AnthonyMouse » Sun, Jul 19, 2020, 6:39 AM UTC

mistermann wrote: post #60

AnthonyMouse wrote: post #51

Earlier quoted context omitted.

I'm not sure that type of analysis is even possible. The more factors you want to take into account, the larger the sample size you need. But there are so many factors here that you could need a sample size which is larger than the entire US population, which is obviously not available. It seems to me the better solution is to just solve the problems. Stop spending all day arguing about whether the criminal justice s…

> I'm not sure that type of analysis is even possible. The more factors you want to take into account, the larger the sample size you need. But there are so many factors here that you could need a sample size which is larger than the entire US population, which is obviously not available. Mostly agree. It certainly couldn't be done with perfection, as there are essentially an infinite number of discreet variables in…

> But what do we do instead? The majority of the time we look at, say, 1 to 3 variables, and then form highly confident conclusions based on this

Which is bad enough in general, but in this case there are political implications, so no matter what conclusion you reach from doing that, the people who want the opposite conclusion can accuse you of choosing the wrong factors and add enough of their own to lead where they prefer. If you only add factors that all point in a single direction you can reach any conclusion you want, but if you let each side keep adding factors in their favor indefinitely then you exceed the limits of statistical significance before either side runs out of factors in their favor.

But it's also important to notice that this is only a statistical problem. It means that statistics can't really tell you what happens in the aggregate, but that doesn't mean you can't investigate individual cases. If you have an email from the hiring manager saying not to hire you because you're black, you shouldn't need statistics to win that lawsuit. So if that sort of thing is prevalent then prove the individual cases, and keep doing that until it's not.

Meanwhile you solve the underlying problems which takes care of the cases you can't prove. The cops can't be murdering disproportionately many black people if the cops aren't murdering anybody whatsoever, which they shouldn't be doing for any reason racism or otherwise.

Post by mistermann » Sun, Jul 19, 2020, 12:52 PM UTC

AnthonyMouse wrote: post #61

mistermann wrote: post #60

Earlier quoted context omitted.

> I'm not sure that type of analysis is even possible. The more factors you want to take into account, the larger the sample size you need. But there are so many factors here that you could need a sample size which is larger than the entire US population, which is obviously not available. Mostly agree. It certainly couldn't be done with perfection, as there are essentially an infinite number of discreet variables in…

> But what do we do instead? The majority of the time we look at, say, 1 to 3 variables, and then form highly confident conclusions based on this Which is bad enough in general, but in this case there are political implications, so no matter what conclusion you reach from doing that, the people who want the opposite conclusion can accuse you of choosing the wrong factors and add enough of their own to lead where they…

> Which is bad enough in general, but in this case there are political implications, so no matter what conclusion you reach from doing that, the people who want the opposite conclusion can accuse you of choosing the wrong factors and add enough of their own to lead where they prefer. If you only add factors that all point in a single direction you can reach any conclusion you want, but if you let each side keep adding factors in their favor indefinitely then you exceed the limits of statistical significance before either side runs out of factors in their favor.

This is my point. When people cherry pick statistics to "prove" their point, it's easy to cherry pick other statistics that disprove that point. Reality is incredibly complex, but even relatively smarter people seem unable to discuss it accordingly, and also seem unable to realize they are debating in an unintelligent/dishonest manner, *while calling their adversary in the argument "stupid". It's interesting to observe human behavior, particularly self-perception.

> But it's also important to notice that this is only a statistical problem. It means that statistics can't really tell you what happens in the aggregate, but that doesn't mean you can't investigate individual cases. If you have an email from the hiring manager saying not to hire you because you're black, you shouldn't need statistics to win that lawsuit. So if that sort of thing is prevalent then prove the individual cases, and keep doing that until it's not.

I agree - dealing with precise details of individual cases is superior. But people on both sides of "the argument" seem intent on broadcasting mistruths based on cherry picked statistics (dimensions of reality, etc), which is how we end up in these silly, never ending arguments, which increases polarization, which begets more silly argument, and around and around we go.

Or, we could discuss reality as it is, to the best of our ability. The way I think about things, I believe more intelligent people should take that burden upon themselves, because they have greater ability to do so. If their argument is just and accurate, they should be able to win the argument, should they not? And yet, they seem to insist on sticking to a disingenuous rhetoric approach, and finding holes in these arguments is very easy so the fight never ends.

Re: ‘Excited delirium’, used by police to justify brutality, is junk science

Re: ‘Excited delirium’, used by police to justify brutality, is junk science