GRE scores by race (US citizens only) (best, accepted, test)

[Bettafish]

The data is from 1996-1997 so it's a bit old.

Race, Verbal, Quantitative:
Asian, 487, 598
Black, 391, 416
Mexican American, 435, 475
Native American, 456, 479
White, 495, 540


PS: Contrary to the stereotype, males consistently do better than females in verbal

Asian male: 490, 638
White male: 512, 586
Black female: 388, 404

http://www.csua.berkeley.edu/~stephy...erfstatgre.pdf

[compSciGuy]

In all fairness, there are noticeably more women than men taking the tests for each "racial/ethnic" demographic.

[UNC4Me]

This data may or may not even be relevant anymore. The GRE (and the way it's scored) was substantially changed in August 2011. Looking at data based on the current test might be more helpful.

[mizzourah2006]

What is the point of this thread?

[fitzy24]

Originally Posted by mizzourah2006

What is the point of this thread?

I'd like to know too.

[Old Tired Man]

Originally Posted by mizzourah2006

What is the point of this thread?

I liked it and it gave out good information. OP: don't let the politically correct police on this board stop you from writing posts about this interesting topic!

[mizzourah2006]

Originally Posted by Old Tired Man

I liked it and it gave out good information. OP: don't let the politically correct police on this board stop you from writing posts about this interesting topic!

The data is almost 20 years old. The GRE has changed twice since the data was collected. I'm just really confused as to what the OPs point was.

How do these descriptive statistics help us?

The more telling part of the data is the standard deviations, which are extremely large.

[Bettafish]

Originally Posted by mizzourah2006

The data is almost 20 years old. The GRE has changed twice since the data was collected. I'm just really confused as to what the OPs point was.

How do these descriptive statistics help us?

The more telling part of the data is the standard deviations, which are extremely large.

Here is the new data (2009-2010). Still not the most recent, but new enough.
http://www.ets.org/s/gre/pdf/gre_gen..._2009_2010.pdf

The racial gap has become LARGER.

Race, Verbal, Quantitative:
Asian, 486, 622
Black, 398, 431
Mexican American, 436, 501
Native American, 451, 509
White, 493, 569

[HiFi]

I wish there were more data-points so we could narrow down which cultural differences are the ones to best predict the outcomes on these tests. The tests should come with data on everything from favorite tv show, to number of siblings.

I suspect the single greatest cultural indicator of performance is expectation on that individual. What does the family, peers, or community as a whole expect of the child. Children will put in 110% in order to be accepted. Being ok with or understanding of morons is what allows them to turn out as morons.

[mizzourah2006]

Originally Posted by Bettafish

Here is the new data (2009-2010). Still not the most recent, but new enough.
http://www.ets.org/s/gre/pdf/gre_gen..._2009_2010.pdf

The racial gap has become LARGER.

Race, Verbal, Quantitative:
Asian, 486, 622
Black, 398, 431
Mexican American, 436, 501
Native American, 451, 509
White, 493, 569

Again the standard deviations are the more important data point.

Look at it in terms of effect sizes.

Most researchers suggest an effect size of .20 is a small effect size, .50 is a moderate effect size and .80 is a large effect sizes.

Let's look at white and asian quant for example.

569 vs. 622. with a standard deviation of 131 means that the effect size is just over .50, so it is a moderate effect size, or the numbers are moderately different.

Just because means are different doesn't necessarily mean the numbers are drastically different. Again, the standard deviations (which are huge) are the more important data point.

The only really huge differences are between AA and White and Asian in Quant. Other than that all are basically within 1 SD of each other.