What the Data Say About Police

Ed Comment:I read the paper. It raised (in me) all the concerns I raised to you. Black officers act the same toward blacks and whites as white officers. Disparities in the use of force decline as force increases. And their data comes primarily from stop and frisk, which WAS an extreme form of policing. They try to make an estimate of the dangerousness of the suspect but their estimate is silly—education, income, and employment—none of which would be know to the officer, and probably not very predictable since so few people commit crimes. Ironically, in many instances, especially with stop and frisk, the office may know the suspect. And nowhere where did I see a simple measure of how much more dangerous are whites than blacks on average or more precisely white stopped by the police vs blacks stopped by the police. Since blacks commit 8-times as much crime, it seems likely they are more dangerous on average. By measuring the dangerousness of suspects with variables such as income, education, and employment, that an officer can’t identify, you limit the use of aggressiveness to reactive uses rather than proactive uses. That limitation seems impractical. If I apprehend a frail man or women, i.e., less dangerous to me, I might be much less inclined to use proactive aggressiveness because I know I can win reactively. (SWAT teams come storming in with overwhelming firepower to discourage aggressiveness before it starts. Policemen wear guns to be intimidating.) If I fear that I can’t win reactively, I might try to intimidate you into not trying in the first place, so I can reduce the chances of needing to react. It surely seems odd that black policemen would discriminately treat black suspects the same as white policemen treat them, if there were no hidden variables/reasons, i.e. it was merely racism.

Ed Comment:“We found that when police reported the incidents, they were 53% more likely to use physical force on a black civilian than a white one. We controlled for every variable available in myriad ways. That reduced the racial disparities by 66%, but blacks were still significantly more likely to endure police force.” Does that mean 53% x (100% - 66%) = 17.5% more likely all things equal (other than race)? That seems consistent with his finding that if you don’t resist, blacks have are 21% more likelihood of suffering police aggression than whites. He says aggression includes “officers putting their hands on civilians” to officers shooting unarmed suspects. Presumably, most of the incidents are minor incidences (e.g., hands on). And he finds no disparity at the highest level of aggression (killing). So it may be the case that the disparities range from high for minor acts to none existent for major acts of aggression. So how much of the 17.5% difference stems from minor acts of aggression vs major acts? Despite it being an obvious question, oddly, he doesn’t say. That suggests he is being less than forthcoming about it, and therefore there may be big differences he has chosen not to reveal in the op-ed. Please examine the report. Chetty did the same thing in his teacher study. Teachers had big effects in total. But those effect came predominantly from the most talented students, which was overlooked in the reporting. That doesn’t bode well for improving the scores of low-scores, which was the point of his study and recommendation. PS it would also be revealing to see the disparity between how black and white cops treat black and white suspects, all other things equal. The later is relevant because back cops are likely to be in black neighborhoods dealing with more violent crimes and criminals. So you have to hold those things equal. It may be the case that back cops are as racist as white cops. But if they are, it would be an indication there may be hidden variables at work. Again, he has the data but he doesn't say. That's not a good sign. BTW, what mired of variables do they measure? PSS I suspect the “hidden variable” is the probability of escalation by the suspect. Surely police use aggression to deter aggression/escalation. We know blacks commit more crime and more violent crime. So it seems likely they are more likely to escalate with police than whites. I’m curious to see if he includes this variable. It seems like he has the data to estimate it. And he seems surprisingly reluctant to say that his analysis shows racism.

Steve Comment:Looking through the paper to find out how much of the data is lower level force but lower level force is more frequent:

Ed Comment:“We found that when police reported the incidents, they were 53% more likely to use physical force on a black civilian than a white one. We controlled for every variable available in myriad ways. That reduced the racial disparities by 66%, but blacks were still significantly more likely to endure police force.” Does that mean 53% x (100% - 66%) = 17.5% more likely all things equal (other than race)? That seems consistent with his finding that if you don’t resist, blacks have are 21% more likelihood of suffering police aggression than whites. He says aggression includes “officers putting their hands on civilians” to officers shooting unarmed suspects. Presumably, most of the incidents are minor incidences (e.g., hands on). And he finds no disparity at the highest level of aggression (killing). So it may be the case that the disparities range from high for minor acts to none existent for major acts of aggression. So how much of the 17.5% difference stems from minor acts of aggression vs major acts? Despite it being an obvious question, oddly, he doesn’t say. That suggests he is being less than forthcoming about it, and therefore there may be big differences he has chosen not to reveal in the op-ed. Please examine the report. Chetty did the same thing in his teacher study. Teachers had big effects in total. But those effect came predominantly from the most talented students, which was overlooked in the reporting. That doesn’t bode well for improving the scores of low-scores, which was the point of his study and recommendation.

Ed Comment:“We found that when police reported the incidents, they were 53% more likely to use physical force on a black civilian than a white one. We controlled for every variable available in myriad ways. That reduced the racial disparities by 66%, but blacks were still significantly more likely to endure police force.” Does that mean 53% x (100% - 66%) = 17.5% more likely all things equal (other than race)? That seems consistent with his finding that if you don’t resist, blacks have are 21% more likelihood of suffering police aggression than whites. He says aggression includes “officers putting their hands on civilians” to officers shooting unarmed suspects. Presumably, most of the incidents are minor incidences (e.g., hands on). And he finds no disparity at the highest level of aggression (killing). So it may be the case that the disparities range from high for minor acts to none existent for major acts of aggression. So how much of the 17.5% difference stems from minor acts of aggression vs major acts? Despite it being an obvious question, oddly, he doesn’t say. That suggests he is being less than forthcoming about it, and therefore there may be big differences he has chosen not to reveal in the op-ed. Please examine the report. Chetty did the same thing in his teacher study. Teachers had big effects in total. But those effect came predominantly from the most talented students, which was overlooked in the reporting. That doesn’t bode well for improving the scores of low-scores, which was the point of his study and recommendation. PS it would also be revealing to see the disparity between how black and white cops treat black and white suspects, all other things equal. The later is relevant because back cops are likely to be in black neighborhoods dealing with more violent crimes and criminals. So you have to hold those things equal. It may be the case that back cops are as racist as white cops. But if they are, it would be an indication there may be hidden variables at work. Again, he has the data but he doesn't say. That's not a good sign. BTW, what mired of variables do they measure? PSS I suspect the “hidden variable” is the probability of escalation by the suspect. Surely police use aggression to deter aggression/escalation. We know blacks commit more crime and more violent crime. So it seems likely they are more likely to escalate with police than whites. I’m curious to see if he includes this variable. It seems like he has the data to estimate it. And he seems surprisingly reluctant to say that his analysis shows racism.