http://journals.plos.org/plosone/article?id=10.1371/journal.pone.0141854
Thoughts? I'm admittedly not sophisticated enough in my understand of bayesian statistics to properly interpret this analysis.
The biggest thing that stands out to me is how they use surrounding counties to infer probability where no data exists.
Assuming that the methodology is good, I know I want to see it with at least five more years of data.
>tl;dr is bayesian analysis of cop shootings does indicate racism
>>80330463
bump to save in case anons are reading through and formulating responses
>>80331488
>>80330463
Final bump before I let it die.
>>80331906
Here is a bump, cab you explain a little more Op, seems interesting desu
>>80332758
Honestly, that's kinda why I posted it here. I'm hoping a statistics anon will be able to shed a little more light on it.
Here's the important bit from the abstract:
>The results provide evidence of a significant bias in the killing of unarmed black Americans relative to unarmed white Americans, in that the probability of being {black, unarmed, and shot by police} is about 3.49 times the probability of being {white, unarmed, and shot by police} on average.
This result by way of their analysis was independent of participation in violent crime, but bayesian statistics are a different animal from frequentist statistics...frequentist is (or at least aims to be) purely objective with what they state. Bayesian allows prior knowledge to affect the interpretation of the data (and the correction via new data) because of the way that they interpret the very idea of probability. This isn't at all to say it's not very useful in a ton of fields (especially stuff like game theory), but like I said, I just don't know enough about it to interpret it. I was educated purely as a frequentist.
>>80333624
Is this study taking into consideration the socio-economic status of these individuals?
More blacks live in poverty and inner cities - crime-ridden areas - than whites.
>>80330463
>My subject matter is harder than your subject matter!
>>80332758
Here's some useful reading for a basic understanding:
https://bayesian.org/Bayes-Explained
http://www.kevinboone.net/bayes.html
Oh, and note that Bayesian statistics are logically and mathematically correct. The biggest debates over its use in applicability to reality and a general problem of statistics is using the correct analysis method for the data set, then correctly interpret it.
>>80333807
from the abstract
>Finally, analysis of police shooting data as a function of county-level predictors suggests that racial bias in police shootings is most likely to emerge in police departments in larger metropolitan counties with low median incomes and a sizable portion of black residents, especially when there is high financial inequality in that county. There is no relationship between county-level racial bias in police shootings and crime rates (even race-specific crime rates), meaning that the racial bias observed in police shootings in this data set is not explainable as a response to local-level crime rates.
>>80333912
Nah, not that at all. It's just something I'm ignorant of.
>>80334015
Nah not you, pretty much every one of that guys webcomics follows the exact same pattern.
>>80330463
Or, it could simply mean that blacks chimp out and get themselves shot more often. The bayesian analysis proves exactly nothing.
>>80334928
Ah, yeah, well, it makes easy goto for anything sciency/mathy because there's an xkcd for most every even marginally well known topic in there.
That said, I do love his strips about relationships and about his woman going through cancer.
>>80334996
>The bayesian analysis proves exactly nothing.
And your logic behind this statement is?
>>80334015
>The biggest debates over its use in applicability to reality
>and a general problem of statistics is using the correct analysis method for the data set, then correctly interpret it.
In other words, if you're biased, your results will be biased too, because you set up the math (more or less subconsciously) to produce the results you want.
at the end of the day you a lumping together encounters that aren't the same, separating them by race and then claiming individuals are then shot (at least in part) due to racial bias.
Or are we somehow supposed to believe that a population with a significantly higher proportion of violent career criminals with limited impulse control somehow behaves in the same manner as a population that is (comparitively) docile?
>>80335310
All it shows is that blacks are more likely to be shot by police. It explains exactly nothing as to WHY.
Immediately, the conclusion drawn is- it must be racism!
The possibility that blacks might behave differently in encounters with police than white people is not even considered.
Is it so unreasonable that a demographic group that commits disproportionate amounts of violent crime (at every socioeconomic level), has a culture which glorifies crime and violence and a profound disrespect for authority especially "white" authority, might engage in behaviors more likely to get them shot by the cops?
Someone will need to explain in plain English. I'm not sure I get it. It is saying that blacks are 3X as likely to be shot when unarmed but I'm not sure what the basic circumstance is. Like, blacks overall are 3X times as likely to be shot while unarmed during their lifetime, or blacks during an encounter with police are 3X time as likely to be shot while unarmed....
>>80333624
The study you linked assumes whites and blacks have the same number of raw police interactions per capita, which is how it reaches the 3.49 conclusion with shootings. Black people are arrested/stopped at a higher rate hence the higher rate of shootings.
This does nothing to prove any sort of racial bias in shootings, other than those at the individual level.
>>80336142
If that's true it doesn't show anything we aren't already familiar with. Blacks have 50% of murders with 13% of population, and a proportional amount of arrests and encounters with police. The task then for the "anti-racist" side would be to prove that the higher rate of murders is false or that it is due to racism somehow that they commit crime, but then the fact that they get shot still wouldn't mean anything.
If you accept that blacks commit more crime but blame it on white racism, you wouldn't blame the cops for performing their duties after a crime has occurred. That's what makes me angry about it, they still go after the cops like just becuase it's the most immediate consequence of their poor actions, so it must be the cops' fault and cops' behavior must be modified.
Somehow in the cognition of that idea they get lost and end up in a dead end where both "blacks may commit more crime technically but it's whites' fault" and "cops are racist and kill blacks for no reason" are both still true.
>>80335700
Not necessarily that. It's more if you pick the wrong analysis methodology, you're just wrong. Interpretation is another matter entirely. I encourage all of you who haven't to take a statistics class (if no longer in school, khan academy it). Every statistical analysis carries with it strict limits on what can and cannot be said from its results. Like a chi square will only ever tell you if there is a difference between the labels you chose. A t-test can tell you if there is a difference between two sets of data -- it doesn't describe that difference, just says whether or not there is one. A pearson's r just tells if if two factors are related -- again, it doesn't say anything about the nature of the relationship (e.g. causality and the direction of the causality), and whether or not these even can say that much depends on whether or not the data sets themselves conform to certain assumptions.
>>80336086
Check out the methods section, but ignore the formulae. It's pretty straight forward. They calculated the probability of various permutations like white, with a gun of being shot by police. They then used census data to create a heatmap to group the data into probabilities by location down to the county level. They used a couple of techniques to extrapolate what the probability is in places where there is no data (a nearest neighbor thing).
>>80330463
I gave it a quick glance
into the trash it goes
it literally proves nothing
>>80336142
>There is no relationship between county-level racial bias in police shootings and crime rates (even race-specific crime rates), meaning that the racial bias observed in police shootings in this data set is not explainable as a response to local-level crime rates.
Eh...while that's not explicitly encounter rates I don't think you can point at what you're saying as the weakness. I think the biggest weakness is in the inferring probability by rates in neighboring counties along with not being granular enough (i.e. looking at Cook County Illinois doesn't tell us what looking at smaller demographic areas would -- honestly, I'd like to see a section breakdown). That said I do think some other factors need to be implemented as quickly as possible, especially since this does not take into account at all whether or not the shootings are justified or not.
The data doesn't lie. Blacks are far more likely to be shot while unarmed than white people are.
Correlation is not causation however.
>>80330463
>Finally, analysis of police shooting data as a function of county-level predictors suggests that racial bias in police shootings is most likely to emerge in police departments in larger metropolitan counties with low median incomes and a sizable portion of black residents,
Well now.
See this image? This is how you skew statistics.
This image doesn't actually mean anything. It has no linkage to the population at large, and instead just colors counties on a chart.
Meanwhile, you can go look at the actual data by numbers and realize that it's a huge misnomer.
>>80330463
Pic related is how liberals define "unarmed black man"
During risk assessment blacks will undoubtedly be considered of a far higher threat level than whites, and the officer will respond accordingly.
Racist? It is, but it's based on statistics, and the police would not be able to function properly without racial profiling. After all, a pitbull is not a poodle, even though they are both dogs. It does mean more unarmed blacks get shot, but it also means more armed and dangerous blacks are neutralized. If they really want to stop police violence, they should first look in the mirror and address the violent crime issues within their own community.
US cops do need better training, and be less trigger happy, but the police violence against blacks is merely a symptom of a different and much, much bigger problem. After all, blacks mostly die to other blacks, not the police.
>>80339441
This is actually pretty on spot with how I think things are.
I'd also add in that militarization and lack of transparency are major issues with law enforcement, but I think those are fundamentally different problems from this.
>>80338021
>The data doesn't lie
The data lies a lot depending on how you look at it.
http://www.copinthehood.com/2015/04/killed-by-police-2-of-3-race.html
You can skew the numbers fairly easily.
>>80338778
>mostly true
What the literal fuck
One last bump, still looking for actual arguments against/criticisms.
Unarmed does not mean nonviolent; if a big guy charges a female cop, she should shoot in self-defense.
Blacks are around 5x as likely to kill cops, so it makes sense for officers to be jumpy.
