Browsing articles tagged with ""
Sep
5
2012

GRE MST Scoring: Adaptivity and Percentiles

Are you confounded by how the GRE is scored? If so, you are not alone. Recently, a student preparing for the GRE asked, “Is the percentile based on how well you do compared to everyone else who took the test on the same day you did? Or, is it just based on how many questions you get right in each section?”

Scoring on the GRE is really complicated.  Neither way she described is the way the percentiles are derived. Remember, this is a multi-stage test (thus: MST). In the first set of 20 questions that count toward your score, the questions are at a mix of various difficulty levels in addition to a mix of different topics and types of questions. The mix of difficulty levels of the second set of questions (in the same area, Math or Verbal) depends on how many of the questions you got correct in the first set.

So, if you got very few correct in the first set, you’ll get a mix of lower difficulty questions on the second set (not all super-low; there will be super-low, medium-low, and barely-low); if you get a medium number right, you’ll get a mix of medium difficulty questions, and if you get lots of them right, you’ll get a mix of high difficulty questions.  But, even before you start that second set, the range of scores you can get as your final score is already predetermined. Lots of questions right on the first section means your final score will be in the upper range, medium number right puts you in the middle range, few right puts you in the lower range.

Then, depending on how many you get right on that second set and also on the difficulty level of the ones you get right on that second set (difficulty level right doesn’t matter on the first set; only how many you get right is what matters on that first set), your final score on the 130-170 scale is determined.  At that point, the score is compared to scores of other people who took the test for all time. Not just on the day you took it, but for every day anyone took it. That’s what sets the percentile. In other words, for instance, 56% of people who took the GRE got a score on the math section that is lower than a score of 151.

I know this is complicated, but I hope it’s clear to you. There are two main takeaways:

• First, you can’t go from the number of questions you get right on the test to figure out what score you got. Someone with lots of questions right on the first section, but few right on the second section might get fewer questions right than someone with few right on the first section and then many right on the second. However, that first person will get a significantly higher score than the second one.
• Second, you really can’t tell how you’re doing while you’re taking the test. The most important thing for you to do is to get as many correct answers as you can, managing your time so that you grab all the questions you’ll do well on first, using the Mark and Review functions to get back to the questions you’re less likely to get right.

Jul
20
2012

GRE Structure: How an MST Works

As a continuation of last week’s blog entry, we are pleased to share a video discussion of the basics of GRE MST scoring.

In our latest video addition to our YouTube channel, Lee Weiss, Kaplan’s Director of Graduate Programs, explains how the new GRE Multi-Stage Test, or MST, works. The test adapts at the section level in both the Verbal and Quantitative section. Each GRE section has 20 questions, and based on how many of those questions you answer correctly (and their difficulty level), you will be moved into one of three levels for your second section. The GRE is scored on a 130-170 scale in 1-point increments.

For more info on the New GRE, visit http://bit.ly/ss3Bph. If your GRE Test Taking skills need a boost, check out Kaplan’s GRE Advantage course, specifically designed to focus on essential tools for success on the GRE: http://bit.ly/KgiNir

Jul
13
2012

How is the New GRE Scored?

As a GRE instructor and tutor, I’ve heard lots of “theories” about how the new GRE is scored.  Because the scoring formula is not terribly well-defined by ETS, it’s completely normal for people to have questions or concerns about how their final GRE scaled score is calculated.

First of all, the test is no longer a Computer Adaptive Test, or CAT, which was the format of the test from 1992 through August 2011.

• One of the frustrating aspects of the Computer Adaptive Test was that whenever you were presented with a question, you had to answer it before moving on.  This was because of the adaptive nature of the test; since the next question in a section relied on your performance of the questions that came before it, there was no opportunity to skip a question.
• Another frustrating element of the CAT was the apparent disconnect between “raw scores” and scaled scores.  The raw score of a test is defined as the number of questions a GRE test-taker answers correctly, while the scaled score is the score that conforms to a fabricated scale.  To demonstrate, let’s take a look at a scenario that happened often on the old GRE.  On that test, the Math and Verbal sections only had 30 questions each, and the scaled scores for each section ranged from 200-800.  And let’s imagine that you and a friend both answered 20 out of 30 questions correctly in the Verbal section, giving you each a raw score of 20.  But your scaled score was 600 and your friend’s scaled score was 650.  What?  How does that work?  Well, remember: because it was an adaptive test, you and your friend were being presented with different questions.  Your friend probably answered more questions correctly at the beginning of the test and was then given more difficult questions later on, while you probably did a little worse at the beginning and were then given less difficult questions at the end of the test.  How exactly the algorithm worked to determine a scaled GRE score from various raw scores was always a little bit of a mystery, but typically the test scored you by how difficult the questions were at the end of your test.

The new GRE a Multi-Stage Test, or MST.  Here are some MST facts:

• In this format, test-takers are evaluated on their performance on two separate Quantitative and Verbal sections, or stages.
• The test adapts from section to section.
• The scaled score on the new GRE ranges 40 points (from 130 to 170)

While each individual stage of the GRE is comprised of 20 questions that are not adaptive, the test is not the same for all test-takers.  That’s because depending on a test taker’s performance in the first stage, the second stage (also 20 questions) will be either low difficulty, average difficulty, or high difficulty.  This presents us with a similar problem to the old CAT – if test-takers are being presented with questions of varying difficulty, then how can we reliably convert raw scores to scaled scores?  The answer is: okay, well, to be honest, the answer is that it’s pretty tough.

Now, it may be tempting to infer (and lots of students do) that because the scaled score on the new GRE ranges 40 points (from 130 to 170), and because there are 40 total questions in each content area, the scaled score tacks up one point for every correct answer chosen.  Although that’s close to what’s going on, it’s not entirely accurate.  Because the second section varies in difficulty, raw scores are not directly tied to scaled scores.  Instead, the test looks at not only the number of questions you answered correctly, but also at how difficult those questions were.  The test-makers call this equating, and they briefly describe it on their website: “The raw score is converted to a scaled score through a process known as equating. The equating process accounts for minor variations in difficulty among the different test editions as well as the differences in difficulty introduced by the section-level adaptation. Thus a given scale score reflects approximately the same level of performance regardless of which second section was selected and when the test was taken.” (Source: http://www.ets.org/gre/revised_general/scores/how)

So, just like with old GRE test format, there is an element of unknowing here.  The ETS has their own special sauce that they use to evaluate raw scores, and because of that, there is a not a completely linear 1:1 fit between raw and scaled scores.  But that doesn’t mean we don’t know anything about how the new test is scored.  For one thing, we do know that it is very important to score well in the first stage of both the Quantitative and Verbal sections.  This is necessary in order to bump you up to the medium or high difficulty second section, where you can begin to separate yourself from the bottom third of test-takers. Here’s a visual breakdown of how it works:

Additionally, because each section is only 40 questions, answering just two or three more questions correctly can mean a big jump in percentile rank.  Here is a concordance table released by the test-makers that shows the relationships between old scaled scores, new scaled scores, and percentile ranks: http://www.ets.org/s/gre/pdf/concordance_information.pdf .  If you look at Verbal scaled scores on the new test, a 150 is the 48th percentile while a 153 is the 62nd percentile.  By answering (roughly) three more questions correctly, you can jump over 14% of test-takers.

At the end of the day, worrying about how the test is scored only gets you so far.  While it is definitely helpful to know how the test is scaled, what raw score targets you should be aiming for, and how your scaled score compares to other test-takers, the most important thing you should be focusing on in your prep is how to improve your current score.  The more practice MSTs you take, the more of a feel you’ll get for how many questions you need to answer correctly to hit your target score.  And once you’re consistently answering enough questions to hit that target score, my advice would remain the same – keep practicing and working to raise your score even more.  In all my years of teaching and tutoring, I’ve yet to have a student complain because they improved so much they surpassed their target score.

Now that you have a better idea of how the Multi-Stage GRE is scored and scaled, it’s time to get back to studying.  If you have any questions, please don’t hesitate to ask in the comments below!

May
21
2012

What is a Good GRE Score?

Almost every student I have, in every class that I teach, asks, “What is a good score on the GRE?”

Most students don’t understand just how complex that answer can be. Sure, I could tell everyone, “You know, you really want to aim for a 165 or higher on both sections of the GRE.” That would be a completely truthful answer—anything at 165 or higher is an excellent score—but it may not be an appropriate goal for every student. Quite simply, it’s just not that simple.

For starters, here is what you need to know:

The GRE is scored on, essentially, a 41-point scale. That means that small improvements in performance can increase your score quite a bit. It also means that small improvements in your score can make a big difference in your percentile ranking (sometimes, a one point increase in your score can boost your percentile ranking by 5 points—check it out here).

The percentile ranking on the GRE forms a classic Bell curve. Here is a generic Bell curve, in case you’ve forgotten what one looks like:

Because it falls on a Bell curve, your percentile tells the friendly admissions folks how well you did on the GRE compared with the students who have taken it over the last three years. (You can find that information here – But–fair warning–it’s pretty dry reading.)

Also, your GRE score does not stand alone. Whether or not you are admitted into a graduate program (and whether or not you receive scholarship money) depends on several factors, not your GRE score alone.  Do not put all of your eggs in the GRE basket. You can put 4-6 eggs there, but divide the remaining 6-8 between obtaining the best GPA possible, writing a spectacular personal statement, flattering professors and professionals into writing outstanding letters of recommendation, and rounding out your resume.

I imagine you’re a bit frustrated with me. Because you still want me to give you a number and say, “THIS is a good GRE score.” (I did that already, remember? 165….look back to the beginning of this blog entry if you’ve forgotten.)

The absolute best way I can help you is to provide some general guidelines on how to set a good GRE score goal for YOU:

1. Do your research! This is important. What is the average GRE score of accepted students at the schools you’re interested in? What are the average scores for your specific programs? What do the admissions departments have to say about required minimum scores? Once you’ve done your research, use these numbers in your goal-setting process.  Remember also the meaning of an average score—students are accepted with higher AND lower scores than the average. Your entire application is important. Don’t become so focused on one number that you fail to present yourself in the best manner possible.
2. Know that a good GRE score for YOU is the highest score you can possibly achieve after a reasonable amount of prep time (about 100 hours). Take a diagnostic test as you begin your studies (http://www.ets.org/gre/revised_general/prepare/powerprep2). If that diagnostic test places you in the 80th percentile or higher, you may be good to continue studying on your own. If you’re below the 80th percentile on either the Quantitative or Verbal section, consider signing up for a prep course. Kaplan has some great ones!
3. When you begin studying, set your goal score at 20 points above your diagnostic score. Try to break that 20 point barrier in your studies. Raise your goal again—depending on how easy the first 20-point improvement was, raise your next goal by 5-20 points. Continue to raise your goal until about 2 weeks before your GRE test date. For those last 2 weeks, focus on holding steady and not losing ground.
4. You want to get a score that places you in the 50th-99th percentile range (higher is better, of course). That means that your goal score should be somewhere between 151 and 170 on both portions of the test. (Notice how you could bomb the diagnostic, raise your score by 20 points as you study, and still get into the 50th percentile? Pretty cool, huh?)

Above all, study hard. Learn the test strategies. And walk out of Test Day knowing that there was no way you could have done any better!

Apr
11
2012

Back here, I made the general point that expectations have a major impact on your GRE prep process. Now I’d like to talk about a specific expectation that handicaps far too many GRE students:

“My diagnostic score should have been higher.”

I’ve taught over 650 students, and I can count on one hand the number who were happy with their baseline test score. If you’ve just begun the GRE prep process and aren’t happy with your starting score, let me tell you a secret: nobody is. After all, the GRE diagnostic is just that: a diagnostic measure of your strengths and opportunities. What it is NOT is a prognostic, or an indication of your future success.

“Oh, ho!” I can hear you thinking. “You don’t know me, Mr. Kaplan Man! My diagnostic was the 40th/50th/60th percentile, but if it had been the 50th/60th/70th percentile, I’d be totally content.”

Nope! You wouldn’t. That feeling you have is a mirage shared by thousands and thousands of students, whose scores range from the 30th percentile all the way to the 90th.

And yes, I’ve seen students score in the 90th percentile on their first test — and they were the most anxious and unhappy students I have ever seen, despite starting with a score that most students would be thrilled to end with. How can that be? Well, students who start at the 90th percentile typically have their sights set on the 95th or 99th, and rising in those final few percentile ranges is incredibly difficult. Furthermore, a large degree of natural GRE talent is often paired with a large degree of “I should already know this,” and the trap for misery is set.

The bottom line is, nobody likes their first GRE score. After you establish your baseline, you have two options:

1) Be frustrated.

2) Start doing things to raise your score.

One of these things will raise your GRE score. The other will not. Try to guess which one’s which, and then choose wisely.

Do you have a story about stellar GRE success despite a diagnostic score you were unhappy with? Do you have questions about this post? Please comment here and we’ll respond.