Saturday, March 10, 2007

Brand Spanking New GRE

I discovered today that the current version of the GRE will no longer be used after July 31 and a new version will be available starting in September. Of course, even my ideas about the "current" GRE were years out of date. The last time I took the exam, in 1997, it consisted of three parts – Verbal, Quantitative (math), and Analytic (logic) – and it was a standard multiple-choice paper and pencil test. However, by that time, the GRE was also available as an “adaptive” computer-based test; the test would start with a question of medium difficulty and based on whether you got it right or wrong, the test would give you harder or easier questions. The rationale was that by adapting the questions to the level of the test-taker’s ability, it could give a score with fewer questions and thus be more efficient. I’ve always thought this was an iffy idea because it put such undue weight on the first couple questions, which a person could easily botch due to nerves or just not having settled in yet, and forced the person to start with a medium level question rather than having the opportunity to start with something easier. In 2002, the Analytic section was eliminated and a new Analytical Writing section, which required the test-taker to write an essay, was added.

But now the test is being changed quite a bit to make it more closely assess the types of skills required by grad school, to lessen the possibilities of cheating, to lessen the effects of memorization of vocabulary words, and to make it more user-friendly (by ditching the adaptive part of the test and making it an online version of a standard linear test in which everyone gets the same questions and you can skip questions, go back and change answers, and all that). You can read about the details here if you are interested, but the main changes are:

Verbal:
- They are getting rid of the stupid antonym and analogy questions that test vocabulary out of context and instead focus on sentence completion.
- They are asking questions to test inferential reasoning and reading comprehension.
- They are introducing new question types, e.g. that require the test-taker to highlight a sentence out of a paragraph.
- ETS has characterized the changes to this part of the test as substantial.

Example question:

For the following questions, select the two answer choices that, when used to complete the sentence, fit the meaning of the sentence as a whole and produce completed sentences that are alike in meaning.

It is truly paradoxical that the Amazon, the lushest of all rainforests, is rooted in the most [ ] of all soils.
A. acidic
B. coarse
C. stark
D. impoverished
E. infertile
F. austere

Quantitative:
- The test includes fewer geometry questions but more questions that measure data interpretation and rely on real-life scenarios.
- An on-screen calculator with square root capability is available to reduce reliance on simple calculation. (Handy for those of us prone to make stupid math errors.)
- They are introducing new question types here as well, e.g. test-taker will enter the answer via the keyboard rather than selecting from a multiple-choice list.

Example question:

House Prices
Number of Houses
$100,000 - $133,000
12
$134,000 - $166,000
25
$167,000 - $199,000
8

The table shows the distribution of prices of 45 houses for sale in a certain region.
Select two of the following choices and place them in the blanks below so that the resulting statement is true.

$175,000
$185,000
$190,000
at most $42,000
at least $57,000

If the highest price of the 45 houses is [ ], then the range of the prices of the 45 houses is [ ].
Click on a choice, then click on a blank.

Critical Thinking and Analytical Writing:
- Requires specific responses from test-taker in writing the essay so as to lessen the possibility that he is relying on memorization of an essay. I’m not familiar with the current test, but apparently it was open-ended enough that people could memorize their own or somebody else’s example essay and reproduce it on the test.
- The test will include one 30 minute section in which the test-taker writes an essay supporting a particular statement and one 45 minute section in which she will write an essay analyzing and critiquing a passage.

Example question:

"Hospital statistics regarding people who go to the emergency room after roller-skating accidents indicate the need for more protective equipment. Within that group of people, 75 percent of those who had accidents in streets or parking lots had not been wearing any protective clothing (helmets, knee pads, etc.) or any light-reflecting material (clip-on lights, glow-in-the-dark wrist pads, etc.). Clearly, the statistics indicate that by investing in high-quality protective gear and reflective equipment, roller skaters will greatly reduce their risk of being severely injured in an accident."

Write a response in which you examine the argument’s unstated assumptions, making sure to explain how the argument depends on the assumptions and what the implications are if the assumptions prove unwarranted.

The test will no longer be scored on the 200-800 scale that I was familiar with either. The new Verbal and Quantitative sections are designed to have a mean of 150 and be scored in 1-point intervals. The Writing section will continue to be graded on a scale of 0 – 6 with one half-point intervals.

Although I have not looked at the new question examples in detail (and there is as yet a limited amount of information available on the ETS web site, though more is supposed to be coming out soon), my immediate response is (1) the new questions seem likely to be a superior assessment of the kinds of things grad students need to be able to do to succeed and hence a more valid test and (2) I will likely perform better on this kind of test, with its emphasis on drawing conclusions, understanding texts, and interpreting data, than the old one, which relied heavily on memorization of obscure vocabulary and mathematical computation. (I will leave as an exercise for the reader consideration of how much those two thoughts are intertwined.) (Robert said he personally benefited a great deal from the Analytical section's questions based on 5 people sitting around a table eating different flavors of ice cream, etc., but that he doesn't see how that matches up as well against grad school work as analytical writing.) I still plan to study for the test and take the practice exams to become thoroughly familiar with the kinds of questions asked, but I think that there are many areas of the test that I will be very well positioned to answer given that I have, you know, spent the last 10 years since I graduated from college doing quantitative research and writing about it and, one hopes, improving the related skills considerably. If anything, this puts me a leg up on the kids who have just graduated. I may be a full decade further removed from 9th grade geometry, but I have significantly greater experience doing many things this new test is intended to measure, like “analyzing and drawing conclusions from discourse,” “interpreting and analyzing quantitative information,” and “articulating complex ideas clearly and effectively.”

Of course, if I score in like the 60th percentile on the new GRE, be prepared for a long discussion of why it is likely to be totally bogus as a predictive indicator of grad school success. (Yes, this is an empirical question, but the data necessary to answer it are going to be a while in coming, giving me considerable latitude to opine on the subject from a purely theoretical perspective.)

2 comments:

Anonymous said...

What are the answers to the sample test questions?

rvman said...

Since Sally is out of touch for a few days, I'll answer the questions.

The first one is "D" and "E" - the soil is impoverished and infertile.

The math one is $190,000 and at least $57,000 - the lowest price house has to be between 100,000 and $133,000. If it is the top of the range, than if the highest priced place is $190,000 there is a price range of at least $57,000. (It could be larger.)

I won't answer the essay question, but I suspect what it is looking for is stuff like "it is assuming that more safety equipment will reduce injuries, is that true?" "True, 75% of injuries involved unprotected skaters, but what percentage of skaters wear such equipment", "how many injuries are we talking about", that sort of thing.