Gender Gaps in Behavior by Race

Suppose you have two populations: A and B. In population A, 2 percent of the boys and 1 percent of the girls get suspended from school. In population B, 20 percent of the boys get suspended compared to 10 percent of the girls.

There is not a difference in the relative rates of trouble caused by the genders in the two populations, in each, a boy is twice as likely as a girl to get suspended. But what if you had a political agenda and wanted to make it seem as if boys in population B were more “disadvantaged” relative to girls compared to those in population A?

What you might do is tell people that the gender gap in population B(10 percentage points) is ten times that of the gender gap in population A.(1 percentage point) That’s exactly what David Autor and four other authors did in “Family Disadvantage and the Gender Gap in Behavioral and Educational Outcomes,” which analyzes achievement and behavior in Florida schoolchildren. This was recently cited in the New York Times in an article which argued that “A Disadvantaged Start Hurts Boys More Than Girls;” this being the New York Times there can be only one explanation for why. This is a good example of how academia and the media can deceive without it being considered lying.

They write that:

(…)Approximately 11 percent of Florida public school children are suspended for at least one day per school year during grades 3 through 8 (Table 5). But suspension rates are more than twice as high for boys as for girls (15.0 versus 7.3 percent), and the boy-girl differential is twice as large among blacks as whites (12.5 vs. 6.3 percent). (…)

david autor table 5

From table 5 you can calculate the real, non-deceptive “differentials” for suspension rates by race and gender. As 51.2 percent of the sample is male, the following can be concluded:

suspension rates by race and gender

It’s the exact opposite of what Autor implies. Why is this? Suppose that the quality that causes kids to cause trouble is distributed by a normal distribution, like IQ or height. Below a certain level this causes suspension. Suppose, also, that this quality is different for males and females. How different is it, in standard deviations? From these “cutoff” numbers the sd difference can be calculated:

bell curve black suspensions by gender

Black males are .42 sd below Black females, Hispanic males are .433 sd below Hispanic females, and White males are .509 sd below White females(all natives).

This, too, contradicts Autor et al. It is true, however, that on the other things cited, kindergarten “readiness” rates, absence rates, test scores, and high school graduation rates, there is a higher gender gap among Blacks than Whites.

I wondered if this pattern of a lower Black gender gap held for the percentage of adults who wind up in prison, I found it doesn’t. By this data on lifetime prevalence of incarceration, the gender gap for Blacks is 1.13 sd and the gender gap for Whites is .803 sd.

Since the data, with the exception of suspension rates, seems to support Autor et al’s hypothesis, can we conclude that “a disadvantaged start hurts boys more than girls?” Raj Chetty found something similar, that the effects of living in one place or another on income are “sharper for boys than for girls.”

But even if we ignore the probability that this might be caused by biology and pretend it’s all about environment, income and education are important but they aren’t everything. If you are raising your son in a “disadvantaged” environment your greatest fear might be that he joins a gang. If you a raising a daughter there, it might be that she ends up pregnant as a teenager. It really is an apples to oranges comparison.


Clair Cain Miller, A Disadvantaged Start Hurts Boys More Than Girls, New York Times, October 22 2015

David Autor et al, Family Disadvantage and the Gender Gap in Behavioral and Educational Outcomes, Institute for Policy Research Northwestern University, October 21 2015

Thomas P. Bonczar, Prevalence of Imprisonment in the U.S. Population, 1974-2001, Bureau of Justice Statistics, August 2003

Raj Chetty, The Best and Worst Places to Grow Up: How Your Area Compares, New York Times, May 3 2015,

This entry was posted in Genetics/HBD. Bookmark the permalink.