Typical White Fact-Based Reasoning (updated)

Senator Barack Obama said in his memoir, The Audacity of Hope, "The arguments of liberals are more often grounded in reason and fact."  Yet he also claimed that to fear a black person on the street more than a white person is a racial "stereotype", "bred" into us. 

To the contrary, such a fear is actually "grounded in reason and fact."  When the numbers are crunched, a black person is almost six times more likely than a white person to be a murderer.Senator Barack Obama in his Race Speech  said his white grandmother "once confessed her fear of black men who passed by her on the street".  In a follow-up interview  he clarified that remark by saying,

"She is a typical white person who, you know, if she sees somebody on the street that she doesn't know, there is a reaction. That has been bred into our experiences that don't go away and that sometimes come out in the wrong way."  He described such feelings as, "racial or ethnic stereotypes that made me cringe."

It is true that Jesse Jackson confessed to the same fear .  More importantly, that fear is a rational one, borne out by "reason and fact". The data are available.

The U.S. Justice Department provides a breakdown of homicides by the race of both the victim and offender.  Looking at the data for 2005 (the latest year available), we find that whites committed 48.0% of all murders and blacks committed 51.2% of all murders.

However, whites outnumber blacks in the population.  In fact, non-Hispanic whites are about 69% of the population and blacks are about 13% .

These statistics alone, that blacks are 13% of the population but commit 51.2% of the murders, indicate that blacks commit a seriously disproportionate number of murders.

What we would like to find specifically is the likelihood that a given person is a murderer.  The Justice department also provides overall murder rates.  In 2004 (last year of available data), 5.9 people were murdered out of every 100,000.  Since some of those were multiple murders, let's assume that only 5 people of every 100,000 were murderers.  That is, the chance that some person you see on the street will murder someone this year, knowing nothing else, is about 5 in 100,000.

From all this data, we can use either Bayes' theorem or a table of relative frequencies to answer our question of "What is the chance that a person is a murderer (or will murder someone this year), given we know that person's race?"  I provide a table of relative frequencies below, since that is usually easier to understand than Bayes' theorem.  (The table is based on a total population of 10 million for illustration purposes only; the total population does not matter for relative frequencies or computing likelihoods.)



White

Black

Other

Total

Murderer

240

256

4

500

Not a murderer

6,899,655

1,299,935

1,799,910

9,999,500

Total

6,899,895

1,300,191

1,799,914

10,000,000


Feel free to satisfy yourself that the frequencies cited above are reflected in this table.  That is, whites are about 69% of the population and 48% of the murderers, etc.

But now look only at columns of that table.  Of 6,899,895 whites, for example, 240 of them are murderers.  That is a rate of 3.5 per 100,000.  (Without knowing a person's race, that rate was 5 per 100,000, which is borne out in the rightmost column of the table.)  Now look at the column for Blacks.  The rate there is 19.7 per 100,000.

That is, the chance of just any person you see on the street murdering someone this year is about 5 of 100,000 (assuming you see a random sample of people).  The same chance of a white person is 3.5 of 100,000.  And the same chance for a black person is 19.7 of 100,000.  That means a black person is 5.6 times more likely than a white person to be a murderer.

It is totally rational for a any person (including Jesse Jackson and Barack Obama's grandmother) to fear a black person more than a white one.  In fact, you should fear them 5.6 times more.

If you think I made any errors in assumptions or calculations, feel free contact me at randall.hoven@gmail.com.

Update by the author (largely in response to the content of received emails):

(1) I believe the FBI crime statistics for homicides include Hispanics under "White".  If we  simply analyze "Blacks" and "Non-Blacks" (instead of White, Black, and Other), with Non-Blacks making up 87% of the population, then the chance of a non-Black being a murderer would be 2.7 (instead of the 3.5 cited in the article for "White"), the figure of 19.7 for Blacks would remain the same, and the ratio of Blacks to Non-Blacks would be 7.0 (instead of the 5.6 cited in the article for Blacks to Whites).

(2) Had I assumed a higher murder rate, namely 6 instead of 5 murderers in 100,000, the chances would have gone up from 3.5 to 4.2 for Whites and from 19.7 to 23.6 for Blacks, but the ratio would have stayed the same value of 5.66.  The ratio is essentially insensitive to this assumed murderer rate.

(3) It is true that other factors should be considered in the scenario of meeting a stranger "on the street": location, time of day, sex of the stranger, race of the stranger, age of the stranger, sex, race and age of yourself, etc.  However, FBI statistics, at least those easily accessible via government web sites, do not break down homicide statistics, much less all crime categories, in all these variations.  I therefore settled for answering the surrogate question of "What are the relative chances that a black or white person, chosen at random, will murder someone in a given year?"  Without crunching numbers, I believe it reasonable to fear a white male, age 22, met on a parking lot of a large city at 2 am more than a black female, age 63, met at a Christian Science reading room at 2 pm.  My calculations were for "everything else equal" except race.

(4) It is true that most whites are killed by other whites and most blacks are killed by other blacks.  It is also true that more whites are killed by blacks than blacks are killed by whites .  It is also true that many murders are committed by family and acquaintances rather than strangers.  But our problem starts with all these factors removed; it is based on the condition of meeting a stranger on the street.  I'm willing to assume most people are killed by members of their own race for the simple reason that that is who they around most and most likely to meet.

(5) I did not say it is reasonable to fear blacks.  I said it is reasonable to fear blacks more than whites.  I believe whether a certain situation on the street deserves "fear" would require many more factors than simply race.  But I was not the one to judge Barack Obama's grandmother.  Barack Obama did that, and he judged his grandmother guilty of racial stereotyping.  He added irony to insult by claiming "a typical white person" stereotypes!  There are many things to call that, but "intellectual" (as many commentators have called his speech) is not the first adjective that comes to my mind.
Senator Barack Obama said in his memoir, The Audacity of Hope, "The arguments of liberals are more often grounded in reason and fact."  Yet he also claimed that to fear a black person on the street more than a white person is a racial "stereotype", "bred" into us. 

To the contrary, such a fear is actually "grounded in reason and fact."  When the numbers are crunched, a black person is almost six times more likely than a white person to be a murderer.Senator Barack Obama in his Race Speech  said his white grandmother "once confessed her fear of black men who passed by her on the street".  In a follow-up interview  he clarified that remark by saying,

"She is a typical white person who, you know, if she sees somebody on the street that she doesn't know, there is a reaction. That has been bred into our experiences that don't go away and that sometimes come out in the wrong way."  He described such feelings as, "racial or ethnic stereotypes that made me cringe."

It is true that Jesse Jackson confessed to the same fear .  More importantly, that fear is a rational one, borne out by "reason and fact". The data are available.

The U.S. Justice Department provides a breakdown of homicides by the race of both the victim and offender.  Looking at the data for 2005 (the latest year available), we find that whites committed 48.0% of all murders and blacks committed 51.2% of all murders.

However, whites outnumber blacks in the population.  In fact, non-Hispanic whites are about 69% of the population and blacks are about 13% .

These statistics alone, that blacks are 13% of the population but commit 51.2% of the murders, indicate that blacks commit a seriously disproportionate number of murders.

What we would like to find specifically is the likelihood that a given person is a murderer.  The Justice department also provides overall murder rates.  In 2004 (last year of available data), 5.9 people were murdered out of every 100,000.  Since some of those were multiple murders, let's assume that only 5 people of every 100,000 were murderers.  That is, the chance that some person you see on the street will murder someone this year, knowing nothing else, is about 5 in 100,000.

From all this data, we can use either Bayes' theorem or a table of relative frequencies to answer our question of "What is the chance that a person is a murderer (or will murder someone this year), given we know that person's race?"  I provide a table of relative frequencies below, since that is usually easier to understand than Bayes' theorem.  (The table is based on a total population of 10 million for illustration purposes only; the total population does not matter for relative frequencies or computing likelihoods.)



White

Black

Other

Total

Murderer

240

256

4

500

Not a murderer

6,899,655

1,299,935

1,799,910

9,999,500

Total

6,899,895

1,300,191

1,799,914

10,000,000


Feel free to satisfy yourself that the frequencies cited above are reflected in this table.  That is, whites are about 69% of the population and 48% of the murderers, etc.

But now look only at columns of that table.  Of 6,899,895 whites, for example, 240 of them are murderers.  That is a rate of 3.5 per 100,000.  (Without knowing a person's race, that rate was 5 per 100,000, which is borne out in the rightmost column of the table.)  Now look at the column for Blacks.  The rate there is 19.7 per 100,000.

That is, the chance of just any person you see on the street murdering someone this year is about 5 of 100,000 (assuming you see a random sample of people).  The same chance of a white person is 3.5 of 100,000.  And the same chance for a black person is 19.7 of 100,000.  That means a black person is 5.6 times more likely than a white person to be a murderer.

It is totally rational for a any person (including Jesse Jackson and Barack Obama's grandmother) to fear a black person more than a white one.  In fact, you should fear them 5.6 times more.

If you think I made any errors in assumptions or calculations, feel free contact me at randall.hoven@gmail.com.

Update by the author (largely in response to the content of received emails):

(1) I believe the FBI crime statistics for homicides include Hispanics under "White".  If we  simply analyze "Blacks" and "Non-Blacks" (instead of White, Black, and Other), with Non-Blacks making up 87% of the population, then the chance of a non-Black being a murderer would be 2.7 (instead of the 3.5 cited in the article for "White"), the figure of 19.7 for Blacks would remain the same, and the ratio of Blacks to Non-Blacks would be 7.0 (instead of the 5.6 cited in the article for Blacks to Whites).

(2) Had I assumed a higher murder rate, namely 6 instead of 5 murderers in 100,000, the chances would have gone up from 3.5 to 4.2 for Whites and from 19.7 to 23.6 for Blacks, but the ratio would have stayed the same value of 5.66.  The ratio is essentially insensitive to this assumed murderer rate.

(3) It is true that other factors should be considered in the scenario of meeting a stranger "on the street": location, time of day, sex of the stranger, race of the stranger, age of the stranger, sex, race and age of yourself, etc.  However, FBI statistics, at least those easily accessible via government web sites, do not break down homicide statistics, much less all crime categories, in all these variations.  I therefore settled for answering the surrogate question of "What are the relative chances that a black or white person, chosen at random, will murder someone in a given year?"  Without crunching numbers, I believe it reasonable to fear a white male, age 22, met on a parking lot of a large city at 2 am more than a black female, age 63, met at a Christian Science reading room at 2 pm.  My calculations were for "everything else equal" except race.

(4) It is true that most whites are killed by other whites and most blacks are killed by other blacks.  It is also true that more whites are killed by blacks than blacks are killed by whites .  It is also true that many murders are committed by family and acquaintances rather than strangers.  But our problem starts with all these factors removed; it is based on the condition of meeting a stranger on the street.  I'm willing to assume most people are killed by members of their own race for the simple reason that that is who they around most and most likely to meet.

(5) I did not say it is reasonable to fear blacks.  I said it is reasonable to fear blacks more than whites.  I believe whether a certain situation on the street deserves "fear" would require many more factors than simply race.  But I was not the one to judge Barack Obama's grandmother.  Barack Obama did that, and he judged his grandmother guilty of racial stereotyping.  He added irony to insult by claiming "a typical white person" stereotypes!  There are many things to call that, but "intellectual" (as many commentators have called his speech) is not the first adjective that comes to my mind.