Last summer, ETS released a massive file called A Compendium of Studies, a 279-page document of GRE esoterica.
There understandably wasn’t much fanfare surrounding this release, as this is some pretty dry, technical stuff (the GRE reading passages are fun, by contrast). Nonetheless, I’m guessing some of you might have stumbled across it and wondered whether anything in this dense and sprawling compendium might help you on test day.
The short answer: no. The long answer: there are some interesting factoids to “geek out” on…but really it’s not worth your time sorting through the endless graphs, or page-turning (or scrolling) through chapters with such snazzy titles as “The Role of Noncognitive Constructs and Other Background Variables in Graduate Education”.
So I’ve taken the liberty of providing the highlights from the GRE compendium.
1. I always knew there was something fishy about the e-rater
For those who don’t know, both a human grader and an e-rater (basically an algorithm that supposed to correctly judge one’s ability to write) are used to score the GRE essay. If that score differs by more than one point on the six-point scale, a master grader—another human—is brought in to give the final score. And differ the scores do. According to the section entitled “Stumping e-rater”, human graders (who are highly literate types) agreed within one point 92% of the time; the e-rater only agreed within one point of the human grader 65% of the time. Therefore, it is not unlikely to have a human reward your essay a ‘6’ and the e-rater award that same essay a ‘4’.
Given that you can stump the e-rater by spewing illogical arguments laden with ludicrous examples and by repeating an entire paragraph (provided you change the topic sentence) doesn’t make these findings too surprising.
The takeaway is not to be content with arrant nonsense, thinking that as long as you employ large vocabulary and sentence transitions, such as “however”, “granted”, “nonetheless”, etc. (the e-rater eats that stuff up) you’re fine. Rather, do not be frightened by the scepter of a computerized grader. At the end of the day, it is the human graders who have the say.
2. Where a question shows up in a section is important
ETS is able to amass prodigious amounts of data because it has access to the test performance of every student who takes the GRE. As a result, it can discern some pretty interesting patterns. In this case, where a question shows up in a section and the difficulty of the other questions surrounding that question (what ETS calls ”question context”) will determine how well one does on that question.
GRE has tried to neutralize this effect as much as possible by making sure problems in the same section are of similar difficulty. Still, you are likely to get a spread of questions, some of which will be more difficult for you. So skip around. Doing so will switch up the “question context”. When you see a question again, it will appear in a different “light”, since the previous question you’ll have seen will have been different.
3. To calculate or not to calculate
Unless you are an Asian or Hispanic female, then you should use a calculator. Yes, I know a weird finding, but ETS decided to collect data based on the gender and ethnicity of test takers. What they found, for the most part is that using a calculator will boost your score. I should note that the results obtained are pretty murky. First off, you shouldn’t use a calculator for every question. It seems that the higher-level students had a better sense of when to use a calculator.
Another factor confounding the results is that out of students who’ve figured out how to get the answer, many use a calculator. Those students who haven’t figured out what the problem is asking won’t even use a calculator–randomly punching in numbers isn’t going to yield an answer. So it is not that calculator use leads to the correct answer but that if you’ve already figured out all of a question except for the computational part, you are likely to use a calculator.
What does this all mean? Learn to use the calculator discriminately and know that it is no panacea for your GRE math ills.
If you like statistics and are curious about how the current GRE became the way it is, you’ll find lots to enjoy from the GRE compendium. If you have limited time and want to focus on test strategy, then the takeaways from this post should be enough.