Battlepanda: Nothing personal -- it's statistical racism


Always trying to figure things out with the minimum of bullshit and the maximum of belligerence.

Wednesday, February 01, 2006

Nothing personal -- it's statistical racism

Honestly, I don't know why I read Jane Galt anymore. And if anything her co-blogger Winterspeaks is twice as bad. Behold his latest spiel trying to discount the results of "Project Implicit"
If you do not have detailed information on an individual, it is rational to evaluate them on group averages. If you are walking down a dark street at night, and a small, elderly female approaches from the other side, you would feel safe because small elderly females are statistically non-criminal. On the other hand, if a large, young man approaches, you might be more worried. If the man is black, given higher crime rates among blacks, you may be more concerned still.

Judging people on group averages when you have no additional information is called "statistical discrimination", although given how loaded a term "discrimination" is a better phrase may be "statistical differentiation". Statistical differentiation is not racist in that it is not a preference for one race or another, it is simply a decision based on group averages when individual information is not available.
I love how the term 'racist' and even 'discrimination' has become so prejorative that even people who are defending practices that are racist and discriminatory by definition seek to distance themselves from those terms.

By the way, I took that Implicit Assumptions test. It shows that I strongly associate blacks with violence. Heck, Malcolm Gladwell talked about taking this test in Blink, and he was disturbed to find that he associated blacks with negative terms, even though his mother is black. What those disturbing windows into our subconscious shows is that though we might not be aware of it, there is a background hum of racial prejudice that permeates our society; even the most liberal members aren't exempt. For me, that's the big story about Project Implicit. Not that Republicans have stronger associations between blacks and negative terms, which, to be honest, is not exactly unexpected given the 'Southern Strategy'.

So, Project Implicit exposed racist thinking in people like Gladwell and me. It doesn't mean that we are bad people or hypocritical people. It's something we need to be aware of and acknowledge. It's certainly not something one should smooth over by hiding behind statistical arguments. The fact that the crime rate among blacks is higher is not a justification for associating individual blacks with violence because, unless you want to believe that blacks are naturally more violent, the root cause of that higher crime rate is societal and environmental, and thus breaking the cycle requires a rejection of racism, even when it comes at a cost to the individual of abandoning a statistically useful tool. I wrote semi-jokingly about the racial heirachy of builders in London, with the Eastern Europeans considered the most desirable and the Chinese at the bottom. I considered that form of racism fairly innocuous because it seemed natural to me that recent immigrants would strongly reflect the traits of the population in their home countries and it seemed unreasonable to expect people not to use that information. But what if the children of Ukranians are still being stereotyped as great builders and the descendants of Chinese immigrants written off as shoddy workers and the whole thing becomes one big self-fulfilling prophesy? I would say that it becomes a social problem at that point and right-thinking people should refuse to discriminate between builders based on their racial profile, even if it is still statistically "justified" to use that rubric from a self-interest point of view. Commenter Mycroft said it very well at the Asymmetrical Information thread [emphasis mine]:

I think, barring those with truly crazy ideologies, statistical differentiation with respect to race is the leading cause of racism in practice.

If, when searching for an English tutor, I automatically exclude anybody with a last name of Chang, on the grounds that people named Chang are statistically less likely to have good English, is that racism?

What about if, after asking for writing samples, I still decide to exclude people named Chang, regardless of the quality of the writing sample, on the grounds that, all other things being equal, somebody named Chang is still less likely to be good at English than somebody with an equally good writing sample who's named Johnson?

Under your logic, these are all defensible decisions, given that time is limited (and time is always limited in the real world), and all-encompassing metrics are difficult to devise. But they would be recognized as racism nonetheless.

I think this all ties in with the "post-racism" concept that Publius articulates here, somehow. People get very huffy over suggestions that they are possibly racist. Yet inequality somehow continues even as self-identified racists disappears. I suppose the easiest thing to do for everyone is to pretend that there is no problem anymore and it's all down to individual responsibility and minorities should just buck up. It's always easier to pretend there is no problem when one is not at the sharp end of the stick.