GMAT Strategy

Re: some questions about our stats on student results (of our tests), I asked our algorithm guy to address this. Here's what he said [with some additions from me for clarification purposes]:

"Our tests are calibrated so that the mean is equal to the expected GMAT mean—and so that regardless of ability, our test is an unbiased estimator of GMAT performance. Some do better, some do worse, but the expected difference is zero.

I review data about once every 2 months to ensure this – data from practice tests coupled with actual, reported GMAT scores. I do so on an overall score basis and a Verbal and Quant percentile basis. I occasionally tweak the scoring if we get evidence that the difference is far different from zero. Usually, the average difference is between +5 and -5 points on an OVERALL scoring basis.

On the standard error issue:

(1) The 50-point std deviation is a WITHIN-STUDENT st dev. In other words, if one student takes the test 100 times [without significant additional prep in between], the st deviation of his scores will be about 50 points.

The st dev across students is a different story and not relevant (it is about 100, just like the GMAT supposedly is).

(2) The 50-point std deviation is also the st dev of the DIFFERENCE BETWEEN MGMAT (latest practice test) scores and GMAT scores. Some call this a standard error, but effectively it’s the same. The STANDARD ERROR OF MEASUREMENT is mathematically equivalent to the STANDARD DEVIATION OF the error of measurement.

Those two numbers are not [exactly] equal, but both approximately 50."

Also just wanted to add - note that we calculate from the latest practice test to the official test. Most of our students take the last practice test within a week of the real test - so there isn't a significant amount of prep / improvement between the two tests.

We are, of course, comparing apples to oranges (that is, we're comparing two different tests), so that complicates things - but, even so, we can get some useful data.

GMATPrep is still the best general indicator because it is from the same people who make the official test (with all that that entails). The test prep company practice tests have the advantage, though, that you can get explanations and analysis of your performance - so there are good reasons to use both GMATPrep and test prep company practice tests.

FYI - most companies allow you to take one test for free. We don't give everything away, though, because then we couldn't afford to keep making more good stuff!

As far as MGMAT is concerned, our tests are not available for download or distributed on CD; they can only be taken online. This allows us to continually update the test questions and make sure that you're getting the latest and greatest stuff every time you take a test. I'm not sure how other companies do it, but I think some do give a CD with the purchase of a book or something like that.

If you want to take a free test of ours, then you just need to go to our website and sign up .

The scoring algorithm is extremely complicated - it's very tough to explain in just a few sentences. It's primarily based on the difficulty levels of the questions you answered correctly and incorrectly and the order in which you did so. Most people get around 50-60% of the questions right, regardless of scoring level, but everyone has different questions (at different difficulty levels) because the test is adapting to the scoring level you demonstrate as you answer questions.

Our test has a standard deviation of 50 points (compared to a standard deviation of about 30 points on the official test). So it is pretty good, but not as good as the real thing. GMATPrep is the closest test to the real thing and, as such, is the best indicator of your current scoring level.

The free test is built on the same scoring algorithm as the 6 CATs available for sale but does not use the same database of questions. (So the 6 CATs for sale use entirely different questions than the free tes.)

This is the busiest season for study right now, and the week-ends are the most likely times that students will take practice tests. If too many people at once all try to access whatever server is closest to you, that can cause some lag / slowness. Or, if you are very far physically from the server that is closest to you, that can also cause some lag / slowness. (The first explanation is most likely, this time of year.) If possible, you could try taking future tests on weekdays or at times that aren't "peak times" for practice tests.

I'm not quite sure how you can print the answer explanations - I've never tried to do that. I know my students have, though. I'll ask our IT department and get back to you.