Sunday, March 25, 2012

Interracial murder

The Trayvon Martin case is an opportunity to remind people of the reality of interracial crime in the United States.

For the purposes of this analysis, I'll use the term "whites" to refer to whites plus Hispanics since they get lumped together in the data. According to FBI data for 2010,  218 blacks were murdered by whites. Blacks murdered 447 whites. But this is comparing apples and oranges since whites are 6.4 times the size of the black population. If we calculate murders per 10 million in the race-specific population (interracial murder is rare), the rate is 8.82 for whites and 114.89 for blacks. So blacks kill whites at a rate 13 times higher than that of whites killing blacks.

Another way to look at it is multiply the black population by 6.4 to make it as large as whites. If it were as large, 2861 whites would have been murdered compared to the 218 blacks killed by whites.

Saturday, March 24, 2012

Predictors of racism

As a follow-up to the last post, let's do multiple regression to see which factors predict feeling cool toward blacks (sample size = 958 whites):

Coolness toward blacks (standardized OLS coefficients)

Age .088*
Sex -.096*
Size of place -.036
South .025
Education -.011*
Income -.027
Church attendance -.105*
Political conservatism .082*

*p < .05, two-tailed

So cooler feelings are associated with being: older, male, less educated, less religious, and more conservative. Population size, income, and being from the South do not matter.

And here is the model for blacks feeling cool toward whites (sample size = 117):


Coolness toward whites (standardized OLS coefficients)

Age -.003
Sex .013
Size of place .239*
South .109
Education -.134
Income .077
Church attendance .106
Political conservatism .114

*p < .05, two-tailed


Living in a more populated area is the only significant predictor of coolness among blacks. The sample size is small, so if we ignore significance and focus on the size of the cooefficients, it looks like the following are associated with not liking whites: living in the South, less education, more income, more church attendance, and more political conservatism. Age and sex don't seem to matter. It's interesting that church seems to discourage racism among whites but encourage it among blacks.

What percent of whites don't like blacks?

I have a white friend who is married to a black woman. "Converts make the best zealots" certainly applies to this guy. He rants constantly that George Zimmerman proves that white racism is still alive and well. After reminding him that Zimmerman is Hispanic and that anecdote is the weakest form of  evidence, I bring out the big guns: the General Social Survey.

Respondents were asked in 2002 how warmly or cooly do they feel towards blacks as a group. Answers ranged from "very warm" (1) to very cool (9). Among whites (sample size = 2162), only 11.4% percent gave a cool answer (answers 6-9). Only 2.3% said "very cool." (Blacks were asked the same question, and 5.9% gave a cool answer for fellow blacks.)

Next, I calculated median education and income for whites giving cool answers. It is 12.0 years of school and the income range between $25,000 and $29,999. So whites that don't like blacks are, on average, uneducated, and they have low status jobs. They don't run anything and so are in no position to deny blacks jobs, etc.

While we're at it, 15.6% of blacks (sample size = 405) feel cool toward whites. Four percent answered "very cool." And the share of whites who are cool toward whites: 4.7%.

Wednesday, March 21, 2012

A black-white difference in cortisol level
















A new study from Homones and Behavior reveals a black-white difference among children in cortisol levels throughout the day. A flatter diurnal slope is associated with more chronic stress and health problems.

As shown in the graph, whites wake up with a significantly higher level which then falls more quickly throughout the day. The different slope remains even when the following factors are controlled: age, gender, non-steroid medication usage, sleep, wake time, socioeconomic risk, psychosocial risk, perceived discrimination, harsh parenting, and parental monitoring.

Compared to blacks, the slope for Hispanics is more similar to whites, although Hispanics do have lower cortisol levels than whites by evening time.

The authors suggest that since morning levels of cortisol are known to be influenced by genes, the racial difference might be due, in part, to genetic differences.

Tuesday, March 20, 2012

IQ is more important than social class


























This is a structural equation model of a cohort of Swedish men. It shows that IQ is an important determinant of income and especially occupational position.  Much of its influence is indirect: it is a strong predictor of educational level which, in turn, is a powerful determinant of both income and especially occupational position. Parent's social class, by contrast, is comparatively less important. It has little to no direct influence on income or job position, and predicts educational level only moderately.

A whole academic field--sociology--developed on the view that social class is of critical importance. If disciplines emerged as a result of empirical support, students could major in IQiology.

Sunday, March 11, 2012

Why do wealthy white conservative women have more children?

In the recent post on the fertility of wealthy white women, a reader raised the question of why conservative women have more children. He suggested it was greater religiosity and less involvement in the workplace.

We can use OLS regression analysis to help answer the question. I included white women ages 40-59 in households earning at least $100k in 1986 dollars. First, I regressed the number of offspring onto political orientation. The unstandardized coefficient is .161 (p < .05) which indicates that conservative women have significantly more babies. If we include church attendance, the coefficient drops to .120. If, instead, we include hours worked last week, the political coefficent drops to .132. If, instead, we include years of schooling completed, the political slope falls to .131. (The correlation between education and conservatism among wealthy white women is -.16--liberals tend to be more educated.) If all three controls are included in the model, the results look like this:


Number of offspring (sample size = 231)

Conservatism .081
Church attendance .098*
Hours of work -.007
Education -.081*

*p < .05, two-tail


With the three controls included, the link between poltical views and number of offspring falls to non-significance. In this full model, number of hours worked also has no significant effect on fertility.  To put it in plain English, the data indicate that wealthy white conservative women have more children because they are more religious and less educated than liberal women.

One caution: education's effect is not the influence of IQ. Among this demographic, IQ is unrelated to fertility. The correlation between IQ and number of kids is .02--basically zero. Education's influence appears to be cultural.

Thursday, March 08, 2012

Sexual selection and small female feet

This new study in Evolution and Human Behavior provides evidence that the smaller female foot length to body length ratio is due to intersexual selection. Evidently, men perceive small feet to be more attractive since they are associated with youth and thus greater potential fertility.

Sunday, March 04, 2012

Fertility among wealthy white liberal and conservative women

The issue of fertility among wealthy conservative and liberal white women was raised at Steve Sailer's blog. I calculated the mean number of children for the two groups (ages 40-59, households making at least 100k in 1986 dollars):


Mean number of offspring

Wealthy liberal women (n = 35) 1.60
Wealthy conservative women (n = 43) 2.49

The difference is statistically significant. Wealthy white conservative women average 55 percent more kids than their liberal counterparts.

Are moderates smart?

The media gives the impression that only moderate Republicans are intelligent. The truth is that the dumbest spot on the political continuum is the center.

Using data on the GSS vocabulary test, I created an IQ score for all respondents. I've listed the mean IQ for those who voted for McCain, Obama, and those who describe themselves as moderates (sample size = 944).


Mean IQ

McCain 102.28
Obama 100.06
Moderates 97.72

The difference between people who voted for McCain and self-described moderates is statistically significant.

Thursday, March 01, 2012

Catholics are NOT social conservatives

I scratch my head every time pundits refer to Catholics as social conservatives. As a group, they simply are not. Exit polls show that in the Michigan primary a plurality of Catholics voted for Romney, not Santorum. The church has little influence over parishioners.