gre scores for different disciplines
From Discover magazine, a basic analysis of GRE scores from different fields. We do pretty bad in math, but ok, though not great, in writing and verbal scores. The chart makes sense. A few comments:
- The surprise for non-academics is that economics has converted itself into a kind of engineering. The profile of students is pretty much what you get in engineering and most physical sciences. It’s also apparent in their approach to research, mimicry of physical science.
- I’m a little surprised that business does so badly in these comparisons. It’s a competitive field and the b-schools often have high quality economics and statistics programs.
- Academia does break up into big clusters, a la Snow’s two cultures thesis. I think you could argue that the humanities have a more verbal profile than the bio/social sciences/professions cluster (minus econ).
- Philosophy is the only field that does not fit into one of the two (or three) major camps. Probably the only “sui generis” discipline, which kind of makes sense.
Interesting stuff.
orgtheory.net 
I love Open Office.
But they shouldn’t have plotted those useless regression lines except maybe for the Analytic Writing vs Verbal Score graph.
LikeLike
James
December 17, 2010 at 1:28 am
Social work, education, and public administration are all near the bottom, huh? Maybe most government failure isn’t so structural afterall…
LikeLike
joshmccabe
December 17, 2010 at 1:44 am
would be interesting to see the same figures restricted to students entering the top 25 schools for each discipline. i suspect that there’s some selection bias — i.e., likely that folks weak in math don’t bother getting their hopes up in applying to econ phd programs, but not necessarily the case that they’ll get into competitive sociology programs either.
LikeLike
sean
December 17, 2010 at 3:07 am
@sean: When I was in grad school, I heard that the graduate division did an internal study comparing dept GRE scores. The average scores weren’t bad, even in soc, but the ranks were the same. So restricting to top 25 likely shifts the intercept.
LikeLike
fabiorojas
December 17, 2010 at 5:24 am
The ETS scaling is very misleading. From page 13 of the current ETS score guide, http://www.ets.org/s/gre/pdf/gre_guide.pdf . A 70th percentile Math (Quantitative) score is 700. A 70th percentile Verbal is about 530. The mean Math is 590 and the mean Verbal is 456. Equal scores for SAT Verbal and Math have almost equal SAT percentiles across the whole range. GREs do not.
Agriculture is about the point of mean Math and mean Verbal scores. Everything to the right of Agriculture is above average Verbal. Everything above Agriculture is above average Math.
The 80th percentile is 570 Verbal and 740 Math. All disciplines on the chart are plotted below the 80th percentile, Verbal and Quantitative. The best and the brightest in verbal and quantitative reasoning in the US may not be applying to grad schools.
Art history, history, english, anthropology and others in that cluster are as verbally intelligent on a percentile basis as the science, engineering, economics, finance cluster is quantitatively intelligent.
The percentile scaling difference between Math and Verbal raise at least two issues about the test takers and the test.
1. The test takers could be more Math proficient than Verbal proficient and than the general population, which would lead to different mean and percentile scores.
2. The test could be harder on the Verbal section than it is on the Quantitative section.
Of course, both and neither are possibilities.
I would venture a guess that the data needs to be disaggregated into US and International students. US GRE takers might be a more representative sample of US students (remember these are test takers, not program accepted students); and International GRE takers seeking US schools might be more proficient in Math than their country and US counterparts, apply to more Math related grad science programs and are less verbally proficient in English than US students. The International students are less likely to apply to grad schools for subjects in the lower left section of the chart, making those below average Verbal scores less likely to be lowered by international students.
If the test is harder on the Math than Verbal sections, (and reading some blogs about the GRE that is the general impression posted by test takers) then the upper end percentiles are too compressed and not adequately measured to distinguish characteristic differences in Quantitative ability at the upper end Math cluster on the chart. A test to show upper percentile math ability would more accurately measure quantitative ability in that grouping and might spread out that upper cluster to show disparate Quantitative abilities among those disciplines. Engineering might not then plot near Economics.
And of course, there are other possible interpretations of the data.
LikeLike
Milton Recht
December 17, 2010 at 8:16 am
I got 590 in Verbal, 720 in Quantitative, 5 in Analytical Writing. How well did I do for Sociology?
LikeLike
Guillermo
December 17, 2010 at 1:31 pm
also remember that b-schools generally require the GMAT, not the GRE. my guess is that students who do apply to a b-school with a GRE (and that b-schools that accept the GRE) are both at the lower end of rankings, so the results shown here are just an artifact.
LikeLike
audrey
December 17, 2010 at 3:17 pm
“Social work, education, and public administration are all near the bottom, huh? Maybe most government failure isn’t so structural afterall…”
Hey Josh, don’t forget that the social scientists who work for ETS, the company who creates, studies, and administers the gre, by and large, retrench the outliers. Also, social workers and teachers, perhaps unfortunately, are often not trained to think about issues. Rather, they are trained to feel a certain way about particular issues in order to respond empathically. Can you imagine Rothbard attempting to calm down six ADHD kids?
Nevertheless, I am still a bit embarrassed by the averages of social workers and sociologists.
LikeLike
bpitt
December 17, 2010 at 3:26 pm
Milton’s detailed and well-informed remarks needlessly complicate the core “Let’s snicker at the stupid people” story here.
LikeLike
Kieran
December 17, 2010 at 3:29 pm
There are actually very few social workers who take the GRE. For most schools the only people who need to take the GRE to do social work are those who did so poorly in undergrad, they couldn’t get in otherwise. I think this accounts for the low scores from that field in particular. The same may be true to education. I’m not sure.
LikeLike
Josh
December 18, 2010 at 10:36 pm
Kieran, /cough/, doesn’t the graph imply that we are the stupid people?
LikeLike
Benjamin Mako Hill
December 19, 2010 at 6:35 am
Benjamin, we’re just stupid in relative terms.
LikeLike
fabiorojas
December 19, 2010 at 6:37 am
[…] gre scores for different disciplines […]
LikeLike
gre scores for different disciplines ‹ Sultan Alfarhood – "Silence is the true friend that never betrays" (CONFUCIUS)
December 20, 2010 at 5:09 am
[…] gre scores for different disciplines […]
LikeLike
gre scores for different disciplines ‹ Sultan Alfarhood – Freelance software engineer
December 21, 2010 at 6:17 am
[…] There is also evidence about graduate students. Studies of GRE score by major, once again, show that sciences do better than humanities/arts/social sciences in math, and there are many science fields that do better than the humanities & arts in verbal GRE. Once again, education and some types of business, don’t do well. […]
LikeLike
an inconvenient truth about social science and humanities majors « orgtheory.net
July 13, 2012 at 12:01 am