GRE Scores are an Obstacle to Diversity in Graduate Admissions

by Guest on April 22, 2013

This guest post from Keivan Stassun originally appeared as a message on the aas_panchromatic email list. Keivan is a Professor of Astronomy at Vanderbilt University. In addition to researching star formation, Keivan is actively involved with several initiatives to engage minorities in astronomy and space sciences. The aas_panchromatic list is an online discussion forum sponsored by the AAS Committee on the Status of Minorities in Astronomy and its main purpose is to facilitate networking and mentorship of astronomers who are members of underrepresented minority groups.

A recent article in the APS News, Admissions Criteria and Diversity in Graduate School, is a must-read for anyone who is thinking about improving diversity in our graduate programs:

Please read the article — it is short and excellent — but the figure tells the entire story.

You can see immediately the consequence of lopping off the applicant pool at a Quantitative GRE score of 700: Most Asian Americans, nearly the majority of whites, and nearly the majority of men make the cutoff, whereas almost all Hispanics, Native Americans, and African Americans are cut out, as well as a clear majority of women. These statistics are from ETS, the GRE people themselves, yet amazingly this information has been largely unknown to my knowledge. Consider that it is not uncommon for physics/astronomy PhD programs to use a cutoff score of 650–700 on the quantitative GRE score, either as a matter of official policy, or more commonly as an unofficial consequence of ranking applicants by GRE score and applying a strong weight to that score.

This figure is another illustration of the same data that I find additionally compelling. GRE Quantitative Scores broken out by gender and ethnicityWhile unfortunately lacking range bars, this version shows that the same effect holds when considering only those students who were Physical Science majors in college and who had undergraduate GPAs in the “A” range (defined as > 3.7).

I think these data are a rallying cry for action. The AAS could be among the first professional societies to issue a “position statement” on the proper use of GRE scores in graduate admissions, recognizing that the data should be what drives our approaches, not our desire or belief that these standardized measures are predictive of success (where success in a PhD program can be defined as just about anything, because GRE scores correlate with almost nothing!).

In the Fisk-Vanderbilt Bridge Program, we have done away with GRE scores entirely. Instead, we perform a holistic review of each applicant, and we have developed an interview protocol that probes for what the psychology research literature calls “performance character”, essentially a measure of an individual’s demonstrated ability to persist in the face of challenge (also sometimes referred to as the “grit” factor). As I’ve heard one colleague say, “Persistence is the ‘P’ in PhD!” You can find our interview protocol in the article we published in the American Journal of Physics.

How do you think our grad admissions criteria can be modified to ensure that (1) women and minority applicants are not systematically eliminated, and (2) appropriate metrics are used to identify applicants with promise for success?

{ 5 comments… read them below or add one }

1 Tertullian April 22, 2013 at 5:57 pm

Interesting post! I recall hearing from a professor at my program that the best correlation with graduate school success (for non-foreign students) was the _verbal_ GRE, interestingly enough. It’s amusing to note that most students probably study for the general GRE a tiny fraction of the time they study for the Physics GRE, whereas the latter usually has much more lenient cutoffs than the former.

It would be beneficial (to all, I think) if graduate programs would release these sorts of statistics.

Reply

2 Johanna April 22, 2013 at 9:11 pm

I would like to see similar data, only with the *Physics* GRE score on the y-axis.

Reply

3 anon April 23, 2013 at 10:41 am

As a tangent, what about the $50-100 that each institution charges as an application fee? This must be a huge deterrent for minorities in all graduate programs, including physics and astronomy.

Surely we should be doing away with these, at least for need-based applicants.

Reply

4 Nathan April 23, 2013 at 5:49 pm

It’s my understanding that these fees help fund prospective visits. When I applied, most universities mentioned fee wavers on their applications. I never contacted them about it but I suspect that with a little bit of paperwork it should be possible to avoid paying most fees. Definitely a little bit of a barrier to entry, but not an insurmountable one.

5 Niall April 26, 2013 at 6:13 am

So I read up a bit on the Fisk-Vanderbilt Bridge programme and thought, “this is fantastic, but the idea is based on the education system in the United States and probably more specifically the Southern US”. So how does this apply to other countries? According to this report,

http://raceforopportunity.bitc.org.uk/research-insight/research-articles/1race-higher-education-2the-race-work

The universities with the highest ethnic minority populations in the UK tend to be post-1992 universities situated in cities or towns with large non-white populations. This isn’t surprising. However of the top ten, only one (Queen Mary) seems to offer an undergrad programme in astronomy (although I’m willing to be corrected if I missed another one or two). While some post-1992 institutions have excellent astronomy departments, most concentrate on applied sciences. Hence I’m wondering if a bridge programme in the UK would focus on bringing in students with applied science degrees in to astronomy master programmes. Does anywhere do this? Also how does this problem manifest itself in other education systems?

A quick additional note about the least diverse universities in the UK, these tend to be in Scotland and Northern Ireland. This makes sense as these are areas with very small non-white populations. However both have significant problems with access based on class and sectarianism so have non-visible underprivileged groups.

Reply

Leave a Comment

Previous post:

Next post: