---

If P( C | B ) is 0.05, that means roughly %5 of blacks are criminal. This is really not that low if P( C | W ) is much lower, such as 0.005, and would reinforce profiling even further. Considering blacks are roughly 12% of the population and commit 50% of the violent crimes, I imagine P( C | B ) is much higher than P( C | W ).

---

Irrespective of what  P(C|W) is, if P(C|B) is 0.05, that means that 95% of the times a police officer picks a person based on racial profiling, he would be wrong 95% of the time. In other words, horribly wasteful of resources. The key here is that you need to evaluate P(C|B) in isolation if you want a reasonable picture on how common black criminals are.

-------
"The larger the island of knowledge, the longer the
shoreline of wonder" (Ralph W. Sockman)

---

Quote: from Silico at 11:47 pm on July 9, 2012

---

If P( C | B ) is 0.05, that means roughly %5 of blacks are criminal. This is really not that low if P( C | W ) is much lower, such as 0.005, and would reinforce profiling even further. Considering blacks are roughly 12% of the population and commit 50% of the violent crimes, I imagine P( C | B ) is much higher than P( C | W ).

---

Irrespective of what P(C|W) is, if P(C|B) is 0.05, that means that 95% of the times a police officer picks a person based on racial profiling, he would be wrong 95% of the time. In other words, horribly wasteful of resources. The key here is that you need to evaluate P(C|B) in isolation if you want a reasonable picture on how common black criminals are.

---

Post edited at 2:40 pm on July 9, 2012 by kidd rune

-------
"The Jewish problem is one of the greatest problems in the world, and
no man, be he writer, politician or diplomatist, can be considered mature
until he has striven to face it squarely on its merits." - HW Steed

---

It's not confusing anything at all. Blacks are more likely than Whites to be criminals. This is a proven fact, they have very significantly higher crime rates.

---

You still don't get it, do you? The ratio P(C|B) / P(C|W) is irrelevant. What matters is how high P(C|B) is in an absolute sense. If it is low to very low (say, 0.05 or some such figure), then your position falters under its own dead-weight.

In fact, paying more attention to blacks than whites will be a statistical mistake, because even if P(C|B) > P(C|W), the number of white criminals are much higher than the number of black criminals, thus misdirecting valuable police resources.

Let's take an example. Individuals with schizophrenia with a concurrent substance abuse problem is about 30 times over-represented in the crime statistics. According to your reasoning, the police should profile individuals with schizophrenia. Unfortunately, this group only represents a minority of all individuals with schizophrenia  and the proportion of crime committed by individuals with schizophrenia with a concurrent substance abuse problem is negligible. Congratulations, you just wasted 100 million USD that could have been better spent on actual crime prevention / mitigation.

---

That is subjective, it's still higher than any given White person.

---

No, the probability that a person is a criminal, given that he or she is black, is not subjective (if you want, we can go with the arbitrary 5% figure). Also, blacks are much more likely to be convinced on drug charges than whites, despite the fact that the majority of users are whites. Your argument does not control for this, which, incidentally, is a result of racial profiling together with racism in the criminal justice system.

---

I'm just going out on a limb here but it seems like the majority of suspects based solely on observable characteristics will be false positives. Also what are better predictors of crime?

---

Yes, hence why you should not use observable characteristics of people when trying to look for suspects (during e. g. random screens etc.)!

I am glad that you asked for better predictors of crime. This is what criminology has been doing for the past, say, 60 years. The best predictors of crime are prior delinquency, substance abuse, low socio-economic status, economic inequality, certain mental conditions etc.

In fact, the strongest and most consistent predictor coming out of criminology for homicide is economic inequality.

-------
"The larger the island of knowledge, the longer the
shoreline of wonder" (Ralph W. Sockman)

---

According to research on the MAOA gene, 13.5% of Blacks are genetically predisposed to very extreme aggression, while only 1% of Whites are. 61% of Whites have the low-aggression version of the gene, while an almost equal 59% of Blacks have another aggression-inducing version.

---

That is not how it works. Having the shorter allele of the MAOA gene itself is not a predictor for aggression. It is having the shorter allele of the MAOA gene and being subjected to child abuse that is predictor of aggression (or more precisely, juvenile delinquency). The interaction is necessary.

-------
"The larger the island of knowledge, the longer the
shoreline of wonder" (Ralph W. Sockman)

---

You still don't get it, do you? The ratio P(C|B) / P(C|W) is irrelevant. What matters is how high P(C|B) is in an absolute sense. If it is low to very low (say, 0.05 or some such figure), then your position falters under its own dead-weight.

---

Obviously they don't do that, they simply pay more attention to Blacks because they commit a higher portion of the crimes, just as it makes sense to do the same for men instead of women.

When looking for violent offenders police usually focus on males instead of females because males are significantly likely to have committed the crimes than females. And nobody questions this or calls it "Sexist" because it's true. And Blacks commit more crime than Whites at around the same proportion as males vs females.

Although most men don't commit violent crimes, it's irrational to ignore sex entirely and focus equally on both men and women when that's obviously going to be a waste of time.

---

In fact, paying more attention to blacks than whites will be a statistical mistake, because even if P(C|B) > P(C|W), the number of white criminals are much higher than the number of black criminals, thus misdirecting valuable police resources.

---

---

Let's take an example. Individuals with schizophrenia with a concurrent substance abuse problem is about 30 times over-represented in the crime statistics. According to your reasoning, the police should profile individuals with schizophrenia. Unfortunately, this group only represents a minority of all individuals with schizophrenia  and the proportion of crime committed by individuals with schizophrenia with a concurrent substance abuse problem is negligible. Congratulations, you just wasted 100 million USD that could have been better spent on actual crime prevention / mitigation.

---

---

No, the probability that a person is a criminal, given that he or she is black, is not subjective (if you want, we can go with the arbitrary 5% figure). Also, blacks are much more likely to be convinced on drug charges than whites, despite the fact that the majority of users are whites. Your argument does not control for this, which, incidentally, is a result of racial profiling together with racism in the criminal justice system.

---

---

Yes, hence why you should not use observable characteristics of people when trying to look for suspects (during e. g. random screens etc.)!

---

---

I am glad that you asked for better predictors of crime. This is what criminology has been doing for the past, say, 60 years. The best predictors of crime are prior delinquency, substance abuse, low socio-economic status, economic inequality, certain mental conditions etc.

In fact, the strongest and most consistent predictor coming out of criminology for homicide is economic inequality.

---

---

Quote: from kidd rune at 12:38 am on July 10, 2012

---

According to research on the MAOA gene, 13.5% of Blacks are genetically predisposed to very extreme aggression, while only 1% of Whites are. 61% of Whites have the low-aggression version of the gene, while an almost equal 59% of Blacks have another aggression-inducing version.

---

That is not how it works. Having the shorter allele of the MAOA gene itself is not a predictor for aggression. It is having the shorter allele of the MAOA gene and being subjected to child abuse that is predictor of aggression (or more precisely, juvenile delinquency). The interaction is necessary.

---