An earlier post about GRE scores for education students suggested that students heading toward graduate study in education look about the same GRE-wise as students in other disciplines on the verbal part of the test, but compare somewhat less favorably on the quantitative part of the test. A commenter suggests a more nuanced view, which is the topic of today’s post.

(Aside on comments: The way my software works comments originating on social media don’t post, although I sometimes get to see them. I don’t know whose poke inspired today’s nuancing, but it pointed to this interesting commentary.)

The suggestion was that when looking at GRE scores, one really ought to separate elementary school teachers from secondary high school teachers. I suspect the underlying argument is that grade school teachers don’t need to know as much mathematics as do teachers at higher levels. In any event, here are the score distributions done separately.

Elementary school teachers look about like the general population of test takers on the verbal part of the test, but compare less well on the quant score. Secondary school teachers perform at the general test taker level on the quant score, and do better than the general set of test takers on the verbal part.

When looking at GRE scores, remember that the universe of test takers is already pretty academically select…they’re all people who are at least thinking of going to grad school. That means it’s hard to say whether we think people aiming for master’s degrees in education should look above average/average/exactly what?, when compared to other test takers.

At the same time, understand that the quant part of the test is not testing high level math skills. In fact, the questions stop before the upper end of high school math. The ETS description reads in part

The content in these areas includes high school mathematics and statistics at a level that is generally no higher than a second course in algebra; it does not include trigonometry, calculus or other higher-level mathematics.

*If* you think that students getting master’s in education have (as a group) too low a level of academic ability, then it makes some sense to ask for higher GRE scores as part of a generic screen for academic talent. But it probably does not make sense to take GRE scores of education students as evidence about the current education students’ academic ability. It’s too hard to make apples-to-apples comparisons.

The two graphs have exactly the same title, but one of them is supposed to be the verbal score, right? I find your comment right below the graph a little puzzling. Hard to make those judgments from the graphs only. What are the means and the medians, and maybe even selected percentiles, or quartiles? Thanks!

You’re right. One is verbal and one quantitative. Sorry for the poor labeling.

“You seem to be asserting, though, that because it does not directly assess more advanced math knowledge that the test is a poor indicator of math skills (otherwise I’m not sure why you included point 2). ”

No. I’m saying that it’s absurd to argue that the GRE is easier than a *real* math test, as he does here “At the same time, understand that the quant part of the test is not testing high level math skills. “–implicitly saying yeah, they did okay, but this is not testing higher level math, so we can’t assume much about their math skills. This is an IQ test, testing extremely well on it says nothing about your higher level math knowledge but says a great deal about your intelligence. It also says a fair amount about your basic numeracy.

“800 quant score beats the average of most math grad programs isn’t surprising”

If you follow his premise, that this is low level math and therefore we should have high expectations, then by definition anyone entering an advanced math program should completely ace it and yawn about it. It should be a no-brainer–how on earth could you miss something on this test and expect to get into the math/engineering doctoral program?

Now, in fact, the expectations for an elite phD program and the GRE Quant are very high–most will tell you that if you don’t get a 750 or higher, don’t bother applying. But 750 is the top 15% not top 1%, and again, if we are to view the GRE solely as an indicator of math skills, that’s ridiculous for everyone not to get a perfect 800. In fact, only 4% of the people who take it (regardless of specialty) get an 800 (the quant. Again, the verbal is much much harder, and I cover why here: http://educationrealist.wordpress.com/2012/01/28/the-gap-in-the-gre/

“Even if the test is purely an IQ test, and IQ is correlated with better math teachers, then raising the bar on GRE scores for educators makes sense on some level, which is all the author is saying. ”

One of the HUGE problems with his entire argument is that there is next to no correlation with IQ/GRE/SAT/teacher credentialing scores. And that, I cover here in a few places:

http://educationrealist.wordpress.com/2013/03/31/its-the-tests-zitbrains/

But from there, you may want to check out the Mumford links.

The entire premise of his argument is completely flawed, although of course he’s just doing what all the reformers do on teacher quality. Nothing unique to him. All that raising teacher requirements will do is completely wipe out black and Hispanic teachers, and there’s a non-trivial amount of research showing that teacher *race*, particularly in the case of black teachers, is correlated with student outcomes.

1) So now that you’ve begun separating by elementary and secondary, you need to start separating secondary by subject. What’s really important is that math and science teachers do well on the math section, and about 20% of secondary teachers scored above 700, and another 27% scored between 600-690. Those are probably the math and science teachers.

2) The GRE is normally distributed. If, as your thinking suggests, low level math means that college graduates should ALL score above 700, a standard you apparently only hold for secondary teachers. But your thinking is wrong. The old GRE (it changed a couple years ago) was largely an IQ test, only incidentally a math test. What the secondary school scores suggest is that high school teachers who get a credential through grad school are roughly distributed by intelligence in the same manner as the rest of college graduates. By the way, I scored an 800 on the GRE quant section, and a 790 on the verbal (only 2% of the population got above 700 on that test). I majored in English, yet got a score on a “low level math test” that beats the average for advanced math grad programs at pretty much any elite school you can think of. Again, it’s an IQ test.

You might want to take a look at the followup, Teacher Quality Pseudofacts II, because it directlyo addresses your misconceptions on the SAT scores of teachers. http://educationrealist.wordpress.com/2012/01/15/teacher-quality-pseudofacts-part-ii/

In general, it’s a really, really bad idea to discuss teacher quality without reading ETS. There’s tons of data on teacher test scores, and you’re trying to reinvent the wheel with advocacy propaganda.

Ed Realist,

I completely agree with your first point: breaking down scores by subject would certainly provide a clearer picture. However, I’m not sure what point you’re trying to make after that.

The quant portion of the GRE does resemble an IQ test; it uses relatively low level math to assess reasoning and problem solving, which is likely why 10% of test-takers achieve the top score. You seem to be asserting, though, that because it does not directly assess more advanced math knowledge that the test is a poor indicator of math skills (otherwise I’m not sure why you included point 2). Even if the test is purely an IQ test, and IQ is correlated with better math teachers, then raising the bar on GRE scores for educators makes sense on some level, which is all the author is saying.

Side note – saying that your 800 quant score beats the average of most math grad programs isn’t surprising, a single 790 admit could cause that. Admission offices weigh other criteria, it would be surprising if their average was 800.