Wednesday, June 24, 2009

Significant Gender, Ethnic Differences on Math SAT

The 2008 SAT Math scores (see table above, click to enlarge) reveal statistically significant ethnic differences. Difference-of-means tests (not reported here) reveal that Asians score significantly higher on average than Whites, who score significantly higher on average than American Indians, who score significantly higher on average than Mexicans, who score significantly higher than other Hispanics, who score significantly higher than Puerto Ricans, who score significantly higher than Blacks/African Americans. All differences are statistically significant at the 1% level.

Additionally, gender differences on the math SAT exist for all ethnic groups, see table below. For each major ethnic group (Asian, White, American Indian, Other Hispanic, Mexican, Puerto Rican and Black), the mean score for males on the 2008 SAT exam are statistically significantly higher (1% level) than the mean score for females.

In many cases, the ethnic differences on the SAT math exam outweigh the gender differences. For example, Asian females score significantly higher on average than males of any other ethnic group, white females score significantly higher than males of any ethnic group except Asians, American Indian females score significantly higher than black or Puerto Rican males, and females from all ethnic groups except black/African-American score significantly higher than black males. All significant levels are 1%.

Bottom Line: There are statistically significant ethnic and gender differences on the SAT math exam.

12 Comments:

At 6/25/2009 12:35 AM, Anonymous Benny Crap Pants said...

China will rule the world someday. that's what I get from these charts. Our military is like a pop-gun against that intellect and numbers.....

 
At 6/25/2009 1:20 AM, Blogger misterjosh said...

So I'm left wondering... what's your point with these SAT posts? I'd assume it's something for your class.

I would say that if our goal as a nation is to provide every citizen with the same opportunities that we are failing.

The gender gaps are one thing, but these ethnic gaps are another. I think some of the difference arises from the reason these people are Americans. Asian origin Americans are largely new and have come a long way in order to succeed. They want it more.

So what's next Professor? I am curious to see the return on investment for the dollars spent per student. but I'm too lazy to do it myself.

 
At 6/25/2009 5:00 AM, Anonymous Anonymous said...

"I would say that if our goal as a nation is to provide every citizen with the same opportunities that we are failing."

Huh?

All students are provided the same opportunity to learn.

That the outcomes are different is not evidence that we are failing to provide the same opportunity.

"The gender gaps are one thing, but these ethnic gaps are another."

Mighty cavalier of you, misterjosh.

"I think some of the difference arises from the reason these people are Americans."

"I think"? Got any evidence? I think the moon is made of blue cheese, but that doesn't make it so.

 
At 6/25/2009 6:28 AM, Blogger Golfintiger said...

I would say that if our goal as a nation is to provide every citizen with the same opportunities that we are failing.

Equal opportunity does not equal outcome. If you are expecting a goal of all genders and ethnic groups to statistically score the same you will waiting a very long time.

The gender gaps are one thing, but these ethnic gaps are another.

Occam's Razor: just as the gender gaps are expected because boys and girls are different, the simplest explanation is that ethnic groups are also different. It's not a grand conspiracy...

 
At 6/25/2009 8:02 AM, Anonymous Anonymous said...

Do you have the regression results tabled out to display?

 
At 6/25/2009 8:20 AM, Blogger Mark J. Perry said...

The best statistical test would not be a regression, but would be a difference-in-means t-test, and here is an
Online t-test calculator. You can enter the means, standard deviations and sample size from the SAT tables, and verify that all of the differences in mean test scores mentioned in the post are statistically significant at the 1% level.

 
At 6/25/2009 8:43 AM, Blogger Paul said...

Politically incorrect as all get out, but not surprising. I remember
in college we'd be heading to the football game half loaded on a beautiful Saturday morning, and the Asian students passed us in the opposite direction headed towards the math lab.

 
At 6/25/2009 11:42 AM, Blogger misterjosh said...

I knew I shouldn'ta posted that late at night.

This is the land of opportunity, and by god it should be our goal to provide every citizen of the US with the opportunity to succeed.

Does the son of a single black mother in an urban area have the same opportunity as the son of an affluent Asian couple in the suburbs? No. That is the test that we are failing.

I'm not talking identical opportunities. I'm not saying the black kid can't grow up to be a successful entrepreneur, but it's much less likely. I'm primarily talking about education.

Poor black kids are getting the shaft when it comes to education, and I think these SAT statistics show that.

It's also possible that blacks in the US are genetically predisposed to have lower IQ than whites, but god I hope not.

Now WHY are they getting the shaft? Therein lies the rub. I'd say shitty schools that, contrary to what I've always heard, usually aren't actually underfunded. They have the same problem that GM had: restrictive union work rules. I believe more and more that school choice is the only thing that will improve the schools that these children are attending.

The kids whose parents care enough to actually make a choice will be the ones who succeed. Not just the kids whose parents can afford to make the choice.

 
At 6/25/2009 1:31 PM, Blogger DaveinHackensack said...

"Therein lies the rub. I'd say shitty schools that, contrary to what I've always heard, usually aren't actually underfunded. They have the same problem that GM had: restrictive union work rules."

You're half-right there. It's true that "shitty" schools often aren't actually underfunded. For example, poor school districts in NJ (Google "Abbott decision") are given higher funding per student than the state average, thanks to tax transfers from wealthier districts. The problem with blaming the continued poor performance of students in these districts with union rules is that the best-performing districts in our state also have unionized schools.

So if money isn't the differentiating factor, and unions aren't, what's left? The evidence suggests that better students make better schools, not the other way around.

 
At 6/25/2009 7:05 PM, Blogger Hot Sam said...

This comment has been removed by the author.

 
At 6/25/2009 8:14 PM, Blogger Mark J. Perry said...

Sure, you can do a regression if you have a set of independent, explanatory variables. When all you have is the mean and standard deviation of two distributions, the difference-of-means t-test is the only and best option.

 
At 6/25/2009 9:04 PM, Blogger Hot Sam said...

This comment has been removed by the author.

 

Post a Comment

<< Home