Legalized judicial racism…

So if you reject the idea of sentencing disparity due to things like racism, socioeconomic status, etc you're basically making a case that certain types of people are more innately criminal based on their race or socioeconomic status, good luck with that, and you might want to get fitted for your white hood in advance.

 

Yeah, everybody is acting in bad faith but you with your claim that everybody is acting in bad faith.

 

What a gigantic erroneously made leap you’ve made.
How is your groin?

 

Nasty as always. If you wish to provide something other than either personal or group insults, lete know. But that seems to be what you traffic in here.

 

Sorry you are too busy reflexively rejecting things because certain people say them to not think through the implications

 

And you reflexively defend the criminals. Weird how you consistently worry about the authority but not the offenders.

 

And yet, literally in your next post, you deny that there is a "global" disparity, which is about the system and would be studied statistically. So you are jumping back and forth on that point. Try to keep your own arguments straight next time.

 

Frankly, you don't have the ability to examine the data based on your inability to read the study that was posted here. You just argued the data was flawed without even realizing that it was a meta-analysis or understanding what that means.

 

I'm not a stats guy by any means, but the complexity of subconscious bias seems to me to be a reason that statistical analysis is helpful. If there appears to be a 20% discrepancy in sentencing of Black men relative to White men in Federal courts, what are the odds over a large sample size that the discrepancy can be attributable to variables other than race, particularly if they've accounted for things such as score sheets and past violent crimes? Also, I think there have been various studies establishing subconscious racial bias in other contexts. Seems unlikely to me that subconscious bias exists in society at large but somehow criminal sentencing is insulated from it. That doesn't mean judges are doing it on purpose or with a racist motive. Even Black judges could have have such biases against Black men, for example.

 

It really comes down to the idea of do social conditions including racism explain crime and how justice is dispensed, or if you refuse to believe that, you really dont have much else where to go other than racial or class pathologies, that's why I said you should get fitted for a hood, because that's the path you are walking. Because what else would explain why prison populations are what they are? Doesn't mean you are there already, but if you believe in some kind of pathological criminal class that belongs in prison or needs special justice, you might as well get measured.

 

Against my better judgment, especially after the OP tried to argue against the existence of the Law of Large Numbers on another thread, I've decided to explain how regression analysis works in this context and why you don't need perfectly matching cases to discover bias. Let's say, for the sake of argument and simplicity, that there are two variables that we know affect sentencing, the crime committed and prior felony convictions (it should be noted that this is obviously not true in the real world and is not a claim by researchers in the field given the complicated models that they have used, but it is just easier to understand simpler models). A researcher hypothesizes that racial bias also plays a role and finds data that contains the variables for sentence length, race of defendant, the specific crime, and prior felony convictions. To study this, you have one DV (sentence length) and three IVs, race, crime, and prior felonies. Let's say that based on our data, we discover that a prior felony conviction is responsible for a 12.5% increase in a sentence, on average, across our data without regard to race, and a specific crime has, on average, with no prior convictions, a 4 year sentence. Let's assume, for simplicity, that prior convictions are not related to race and are, thus, not biased (a big assumption here, but, again, done for simplicity).

In the data we see two observations, both convicted of the same crime with that four year expected sentence. One observation is with a white defendant the other has a black defendant. The black defendant has 2 prior felony convictions and the white defendant has zero.

"A-ha" says lawyer at this point, "The observations don't match and that explains different sentences." However, in a regression analysis, you create a conditional expectation that allows you to compare these two observations. Since we know a prior convictions is related to a 12.5% increase in sentence, our model expects a sentence of 5 years for the black defendant and a 4 year sentence for the white defendant, before we consider race.

Now, we consider race. We compare the observed sentence for each defendant with our expected sentence for each. That comparison provides us with measure of bias, even in the absence of perfectly matching cases that only differed on race.

This logic still works even if you include hundreds or even thousands of variables to help create this conditional expectation.

Beyond regression models, there is something called a matching model. In this type of model, you would compare matching observations. You could do this only when there are perfectly (or, depending on the researcher's judgment, near perfect) matched observations or you could use an algorithm like propensity score matching to create synthetic observations utilizing other observations to approximate a perfect match with a different race. These synthetic observations are then used as a comparison for our real observations.

Matching and regression models both have advantages and disadvantages. However, it is not necessary to use only a matching model in which you have two perfectly matches cases to uncover bias.

 

I mean its not like poverty and racism are unrelated, look what happens when we try to address poverty