CAT Accuracy

[ch339]

Hello,

I was wondering about CAT accuracy.

I took my 3rd full-length CAT today. It was a free test offered through another company. I have read that, at least in the past, this company's scoring algorithm places higher emphasis on earlier questions. I have read on your blogs that it is a myth that the official GMAT differentially weights questions based on placement in a section. Is this correct?

I'm concerned because I want to know whether I'm improving and by how much. I'm not quite ready to purchase Manhattan CATs and am saving the second GMAT Prep, so for now it's pretty much whatever I can get my hands on.

SCORES:
Manhattan CAT 1: 44 Q, 38 V, ~3.7 IR, 680
GMAT Prep 1: 47 Q, 37V, 5 IR, 690
Other CAT: 43 Q, 44 V, 3 IR, 700

I just began studying again after a few months away.

What is confusing is that I have been focusing on quant and, while I wasn't really meticulous about this latest test, I expected a higher quant score. Earlier today, I took just the quant (49) on another free CAT from the other company's website. I know this can artificially inflate my score, but the 6-point drop seems excessive. I got 15/18 PS correct on both, 15/19 DS earlier, and 16/19 DS tonight. While the questions I got wrong tonight were rated easier, I was also presented with more, easier questions. On Manhattan's CATs, it seems that if you mess up an easier question, you may get 1-2 lower-level questions, and quickly work your way back up. On theirs, if you miss a question earlier in the section, you get a lot of lower-level questions before you can start to climb again.

For example, for the 49Q, I got the first 19 questions correct. When I missed a question, I might get an easier question or 2 and continue with harder questions. I started getting marginally difficult questions by #5. For the 43Q, I foolishly missed the third question (medium). I stayed with medium questions until question 11, which I got wrong. Rather than getting another marginally hard question to make sure, they pushed me straight back into medium territory. Even when I continued to get 8+ questions right in a row, they did not increase the difficulty till the very end.

I know that no one really knows the official GMAT algorithm or process, but would they ever give you such a mix of easier questions for that long if you continue to get questions right? How would you demonstrate that you can do better if you're scarcely presented with the option?

Last, how do I gauge my true performance? For quant, is it closer to low or high 40s? And what about verbal? I haven't studied a lick for verbal since the fall. Could my higher score also be a reflection of getting the first 15 questions right, thus biasing my score (based on the other company's algorithm) upward?

Sorry if this isn't entirely coherent. I wasn't exactly sure what information would be relevant to discern what the trend in my scores means.

Thank you so much!

[StaceyKoprince]

I can't tell you how good another company's CATs are, unfortunately - I don't have access to the underlying algorithm and data that I would need to be able to tell.

It is, indeed, a myth that the earlier questions are worth more. In fact, at the last GMAC conference in December, their Chief Psychometrician (yes, that's a real title!) said that the questions that end up counting less are the outliers. So if, when you're done with the session, you map out the difficulty levels of the questions that you get, any high or low outliers will count less towards your score than the ones that are clustered in the middle.

It would be extremely unusual to answer 19 questions in a row correctly - so... that (along with some of the other things you describe) induces some skepticism in the test. But I'd really have to look at the underlying algorithm to know - it's insanely complex.

If I were you, this is what I'd do. Ignore the scores and just concentrate on what you answered incorrectly and what took you too long (does the test tell you how long you took on each problem?). Use that to figure out what you need to study for the next few weeks. Then, take one of our practice tests. :)

You can use this to analyze your MGMAT CATs:
http://www.manhattangmat.com/blog/index ... ice-tests/

You can also read through that article in order to get some ideas for how you would analyze the other company's test. You might not be able to do all of the pieces - it just depends what data they give you.

I also recommend reading these two articles (and starting to do what they say):
https://www.manhattangmat.com/blog/inde ... lly-tests/
http://www.manhattangmat.com/blog/index ... -the-gmat/

If you want any specific advice after you've analyzed your tests, come back here to ask. If you eventually take one of ours and analyze it, make sure to include that analysis in your post. Figure out your strengths and weaknesses as well as what you think you should do based on that analysis. Tell us and we'll tell you whether we agree and advise you further.

[ch339]

Hi Stacey,

Thank you for the articles!

My pacing is way off.

For the 43Q (evening, full test minus AWA), I answered 14 too fast, 4 too slow, and 2 warning track. For incorrect answers, 4 were too fast, 1 was warning track, and 1 was on target for time. I was off pace immediately, with a virtually uninterrupted string of too fast from questions 3-26 (1 was warning track). Then I had a string of 4 too slow questions. 7 hard questions, of which 3 were experimental: -2/4 counted, -1/3 exp. I finished the section with 14 minutes and 27 seconds to spare--way too early.

I suspect this was partially due to too many questions below my difficulty, but the 49Q (morning, just quant + AWA) still had some issues. I finished about 4 minutes early, so overall time was okay. However, I had 7 too fast, 5 too slow, and 1 warning track. Of my incorrect responses, 2 were too fast, 2 were too slow, 1 was warning track, and 2 were target.

44V. For SC, 4 were too fast (19-28s), but I got them all right. The only SC I got wrong was on target (1m 25s) and a higher difficulty. Four passages, 3 short and 1 long. For the first short passage, I got all questions right. I was "fast" (26s - 1m 37s), but the questions had easy-to-search-for terms, so I think that's okay. For the next 2 passages, I got 1/3 and 2/3 correct, respectively. The second short passage was a science passage I had trouble interpreting correctly. For the third, I got 3/4 after changing my correct answer (I understand the mistake I made). I was on target for all questions, except a general question on the longer passage, which I got correct (2m 51s). For CR, 5 questions were "too fast" but all correct. I really did not feel like I was rushing and fully understood the arguments. -1 for CR (on target).

Although I was not rushing on verbal, I finished 31 minutes early. I was mostly on target, but even then I was on the faster side of the target window. I don't know how to slow down or whether I should if I feel good about my answers. Maybe this verbal was too easy? I actually found the passages to be denser, so I don't know. Twenty-four questions were rated on the harder side. The rest were medium to marginally difficult, with 6 easy questions thrown in.

IR was just a mess. I got 3, 4, 5, 6, 10 partially or completely wrong. I ran out of time and didn't even get to the last question (percents, which I'm pretty good at). Every question that was too fast was correct. I suspect I was just more comfortable with those topics/data. Of the three questions I spent too much time on, 2 were completely right (medium) and 1 was completely wrong (hard).

Looking at my results from last summer for your CAT 1, my pacing issues seem to have reversed...except for IR. I'm probably going to buy the IR Strategy Guide. That way I can get better and take your awesome CATs!

UPDATE

I just finished another CAT. 41Q 42V 4IR 4.5AWA (670). Sad.

Although I got 2 of the first 10 verbal questions wrong, and only 1 of the first quant wrong, I began getting harder verbal questions much sooner (within the first 10). It would seem this company weights the first 10 more heavily for quant only.

Pacing issues persist. Once again, I finished early (~15m for quant and ~22 for verbal). I tried to slow down this time. I suppose I succeeded a bit.

4 quant too slow. 23 too fast. 1 in the danger zone. -1 PS (slow, first question). -5 DS (1 fast, 1 slow, 2 target, 1 danger). I took almost 4 minutes on the first question and 20s on the second, so I was on pace for cumulative time. Then I just kept getting further and further away. Again, not sure how to slow down. No more than 2 wrong answers in a row. I knew I had messed up early because I could tell most questions were medium. Only presented with 2 hard questions, both experimental.

-4/13 RC, -1/17 SC, -1/11 CR. Was on the faster side to target. No slow questions. I reread everything carefully, including several winding sentences written in dryer, arcane language. No more than 1 wrong answer in a row. I know it's bad to finish this early. I had 22 minutes left.

This was only the second test I did without skipping AWA, but I didn't feel tired or burnt out. My score dropped by 30 points and by 2 points each on quant and verbal. Although I found this test more challenging, I was not prepared for that.

UPDATE 2.0

Took another full CAT. Even though this was supposed to be a different test, several math (6) and verbal (12) questions were repeated, including 2 full passages with corresponding questions (6), so I do not know how informative these scores are. All new IR. Still good for stamina.

47Q 44V 6IR (730)

-3 DS, -1 PS. -1 in the first 10 questions, followed by a string of 16 correct answers. Score seemed to not have suffered as much. Many fast. 6 slow, 2 towards end of the section. I was 5-10 minutes ahead by the end of each slow question. Four were correct, and 2 were not. Finished ~8 minutes early. Higher number of harder questions, including ones I'd seen. Since I will probably not finish very early on test day, I would like to minimize slow solving in the future. Thoughts? Practice, practice, practice?

-3RC, -1SC, -1CR. RC seems to be my biggest weakness. ~30 minutes early.

7 questions right for IR, but it seemed easier this time. Maybe I'm actually getting better...would be nice.

[StaceyKoprince]

For next time: you gave me a ton of data but you didn't give me much analysis. Tell me what you think the data means and what you think you should do about it - even if you're not sure, make your best analysis. Don't worry: I'll tell you wherever I disagree! But you've got to learn to figure out the analysis yourself.

My pacing is way off.

Yep. :) But you can fix it! Here's where (next time) I want you to push your analysis further. It's off because your mindset is still "I'm trying to get everything right" - but that's not the best mindset for this test. Read the "what the GMAT really tests" article again and tell me how you're going to change your mindset going forward.

Look out for another one on mindset that should be published in the next week or two on our blog. Also, start doing what these two articles say:
http://www.manhattangmat.com/blog/index ... -to-do-it/
http://www.manhattangmat.com/blog/index ... nt-part-1/

For verbal, you may be naturally faster - just make sure that you are not leaving points on the table going that fast. (At your speed, chances are that you are.)

It would seem this company weights the first 10 more heavily for quant only.

Algorithms are complex enough that you can't really deduce anything from just a few tests - you'd have to run thousands of simulations to try to backwards-engineer it.

The earlier questions should not be weighted any more heavily at all - on the real test, the position of the question does not determine its possible weighting. (There isn't really any "weighting" at all - but the way that the algorithm works, it turns out that "outlier" questions affect your score less. "Outlier" questions are the ones that fall farthest from your average difficulty levels - the easiest and hardest ones in the set of questions that you receive.)

Let's talk a little bit about how to slow down. Are you writing down all of your work? If not, start. Don't do stuff in your head; that leads to mistakes. Are you keeping track of your answer eliminations on paper for verbal and DS? Again, if not, start.

Are you finding yourself making careless mistakes on quant? What kinds? If you make mistakes transcribing stuff from screen to paper, then double-check what you've written after you've set up the problem on paper. If you make mistakes in calculations along the way, make sure you're writing everything out and find a way to check your work at the end of the problem. Most people don't have the luxury of doing this... but you've got the time. :)

What about verbal? Lots of times, people are reading so fast that they "skip" words on screen. Skipping even a single word can make a huge difference! Do you ever find this happening?

If so, pick up a hand and start pointing at what you read. Concentrate on looking at every single word; make it conscious. Don't lose points just because you're going so fast that you skim or skip important words / points / language.

This was only the second test I did without skipping AWA, but I didn't feel tired or burnt out.

You may not always realize when you're mentally fatigued. You don't feel physically tired - in fact, you may have adrenalin pumping through you. Mental fatigue feels more like having to read things multiple times before they sink in. Or not caring anymore / just wanting the test to be over. Getting distracted and thinking about things other than the test. Having trouble concentrating. All of those are mental fatigue.

I would like to minimize slow solving in the future. Thoughts? Practice, practice, practice?

Look at the How to Study section of the 2nd level of GMAT study article that I linked to in my last post. That's how you learn to get better (including solving more efficiently)!

[ch339]

Thoughts:

I've read "What the GMAT Really Tests" and "Letting Go," and one thing I struggle with is letting go of harder questions. I am aiming for 51+ in quant (I know 51 is the highest reported). I realize that in the real exam and probably on the other 5 Manhattan CATs I do that I will hit my ceiling and be unable to answer some questions in a reasonable timeframe. I feel like most of the questions I see are doable. Still, I have not been able to break 50. With this knowledge, and the excess time, I cannot let go. I know this will not work long-term, but I'm not sure whatever strategy to pursue to move on from the mid to high 40s plateau. I started with 44 on my first CAT and have only reached 49 at the highest (although I've become more efficient). I'm worried because I do get so engrossed in some tougher problems and don't realize that 4-6 minutes have gone by. That I'm still finishing early only reinforces my bad behavior. I'm interested to see what happens on Manhattan CAT 2.

I always do a grid for DS and have started showing more work, even when I think the answer is obvious. I almost always show work for PS as well, although there are times that I copy incorrectly. I'm getting better at spotting this. I have really not been taking notes for verbal. I did when I was first starting out. Now unless it's an EXCEPT question, I don't bother. Once I know an answer is wrong, I just don't consider it. SC and CR have been pretty easy for my last few tests. I will have to re-commit myself to studying harder incarnations of these questions.

I have begun to double check my reading, and this has minimized errors. However, I still notice that I've done this once I'm reviewing my score report. I will miss "per week" in a quant problem, or worse, autocorrect a SC so that I don't spot the lack of subject-verb agreement in a more concise answer choice. Even when I re-read, I'm still correcting or glossing over the same part of the sentence. All I can think of is to pause and look away for a few seconds before checking my answers. I don't know if that is a feasible strategy on test day though. Don't most people report a slight drop in scores and an increase in time it takes to complete a section?

I would also like to score 45+ in verbal consistently in practice so that I may be reasonably confident of hitting mid 40s on test day.
Part of my strategy will be to dust off OG and continue with RC, so that I fix my weaknesses in this area. Other goals are to practice harder rates/work and property of numbers problems.

Last, I think I will take a break from CATs and just focus on time management and accuracy for smaller sets of problems.

Thank you for your help.

[StaceyKoprince]

Does this help? I spoke with a student a couple of weeks ago who followed one of our strategies: give yourself 4 to 7 "freebies" throughout the section - questions on which you just decide to guess and move on.

She did this 4 times throughout her quant section. She almost certainly got other questions wrong, too - nobody's perfect.

Her score on the quant: 51. (This is indeed the highest score, by the way, not just the highest recorded. They do not give scores higher than 51.)

In other words: what we're telling you is actually the way to "win" this game.

If that doesn't convince you, try this analogy. Let's say you're a really good golf player. Your goal is to get the ball in the cup. Unless you find yourself in a completely bizarre location (very rare), you're always going to be hitting the ball towards the cup.

Now, I'm trying to teach you to play a different game, soccer (football everywhere else in the world :). The ultimate goal is to get your ball into the other team's goal, right? But, often, you have to back off and even go in the wrong direction for a bit in order to then continue to make progress towards the other team's goal. That's a standard part of the game.

If you keep your golf mindset and refuse to do anything but move forward... well, then you're not going to do very well as a soccer player! And that's essentially what you're doing when you won't let yourself give up. You're trying to hold to a strategy for a different game - one that doesn't work for the game you're actually playing.

Until you really believe that, you're going to keep struggling with this issue.

I'm worried because I do get so engrossed in some tougher problems and don't realize that 4-6 minutes have gone by. That I'm still finishing early only reinforces my bad behavior.

How often are you making careless mistakes? These are often caused by timing or mental fatigue issues. When you spend 4-6 minutes on a problem, you've just set yourself up for both of those issues.

Also, is something else going on, too? Is the "stubbornness" issue (not letting go) also spilling over into HOW you solve these problems? I work with students all the time who persist in trying to do quant in the "textbook math" way, because they still believe that this is a math test. If you're doing this too... you're not going to get a 51 on quant.

The mis-reading / glossing thing that you're describing is usually a combination of mental fatigue and a feeling that you've got to hurry, hurry! so you look right over stuff. If you fix the mindset stuff that I'm talking about, you'll mostly fix this problem too. You won't be using up so much mental energy on too-hard / too-long questions and you won't be messing up your time.

For verbal, a 45 is the 99th percentile - did you know that? I understand your desire to leave yourself some "wiggle room," but if you want to be confident of hitting "mid 40s" on test day, you're actually talking about hitting the 99th percentile. That takes a huge amount of work and even still (by definition!) most people never get that high.

I'm not telling you not to go for that score, but I want to make sure you understand what you're actually planning here. A Q51, V45 score is like a 780+ on the test.

[ch339]

Hi Stacey,

I'm trying to keep what you said in mind. Although I had resolved to swear off CATs, I could not stay away for as long as I had hoped. Letting go is still tough, but I think I'm learning. Timing is still an issue I fear.

So I was progressing through quant on Monday, not paying too much attention to the time. Figured I was ahead, but I was certainly not rushing. There was this killer problem for #29 I just knew I could get. Didn't finish my rephrase, thought I was done, but if I'd kept going, I would've seen that Statement 1 was also sufficient. Plugged in numbers for Statement 2, but ran out of steam when I went back to 1. Just had to let it go because I was exhausted. Question 37 was also tough, but I recognized the concept tested once I'd read Statement 1. Statement 2 I figured out the equation and reasoned that it was sufficient, did not check it because Combinations + Algebra = Sadness. I know DS is only about sufficiency, but sometimes on harder DS, I find that I think an equation is sufficient when it isn't and vice versa. I wanted to be 100% sure about this question, but I knew it just wasn't worth it, so I clicked submit and ended up getting it right...with 59s left in the section. Holy cow that was close! My stubbornness almost cost me dearly. It turns out I'd really screwed up on 29. I spent a record 9m 25s and still got it wrong, of course. I didn't think it had been that long. Prior to that, I was "ahead" by ~ 12m. Afterward, I was ahead by 4. Because I was doing well, however, I needed the time. I got some tough questions at the end and had to spend over two minutes each on 5 of the last 8.

I also noticed more signs of fatigue on verbal, that feeling of "I cannot take this." On 2 questions, I found myself just settling for an answer I eliminated because it looked attractive again. I knew these answers were wrong--they were crossed off on my scrap paper--but I was just done.

Anyway, I ended up getting 740: 51Q, 42V, 4 IR, 3 AWA (terrible essay score). TPR.

A few days prior, I received 720: 49Q, 44V, 4 IR. Kaplan.
I ran out of time on IR and Quant. I think I was overcautious and wasted too much time mentally slowing myself down at the start of quant, which probably hurt me (didn't get to 36 or 37). I took a CAT as part of a marketing session, and we were instructed to skip the essay. I intend to continue doing full tests from now on.

I think I've officially crossed the 700 barrier. I have scored 700+ on four tests total (2 if you only count tests with the essay). I realize that I do worse on Verbal when I'm tired. As you've pointed out, it's all the way at the end of a 4-hour block. I took full advantage of my breaks on the latest test, which definitely helps to mitigate the fatigue if not overcome it. I need to be at peak performance for verbal as well, so I will have to improve my stamina.

I will be doing the full breaks from now on. I will also institute the 4 freebie rule. For IR, I need to figure out which 2 questions to let go. Big table questions with multiple questions (T/F or Y/N) take a lot of time. I'm torn because I can do them, but then I run out of time at the end. I end up actually skipping 10/11-12 with no time to input answers, which doubtless suppresses my IR score.

Do you have any advice on:

1) Which questions to skip for IR?
I'm hesitant to let go of a question I know I know how to do for fear of getting stumped by a question I have to do.

2) How to know when I'm done rephrasing for DS?
For example, after some algebraic manipulation on a geometry problem (TPR free online CAT, Level 10 difficulty, looked like the highest diffiulty, tested algebra and overlapping shapes), I quickly got this: " (a-b)^2 > [(a+b)^2]/2 ?" My rephrase should actually have been " a^2+b^2 > 6ab ?" The problem started with recognizing I had to square a binomial and subtract. I then expanded and combined like terms, and recognized that my new polynomial was actually the square of another binomial. However, I was supposed to recognize that I could expand and rearrange my inequality to get the final, simpler rephrase. Do I just do that from now on? Say "Oh, I can't possibly be done yet because this isn't pretty enough. Make it prettier."

3) How to streamline essays?
I've read the blog on this, but I find it hard to cull details from my reasoning. I tend to build up to my point, and I think it would be more effective if I started with my conclusion straight away, with 1-2 pithy sentences to elaborate on the underlying reasoning. I need more time to edit or re-orient my entire approach to writing.

Thanks for all your help so far!

[StaceyKoprince]

Keep pushing yourself. You still don't really "believe" that the test is not actually trying to grade you on getting everything right - if you did believe / understand / internalize that, then you wouldn't keep trying to do something that is against your best interests!

But you'll get there - just keep working on it. :)

From now on, whenever you want to spend a bunch of extra time on a problem, remind yourself that you spent 9 minutes on one problem and got it wrong anyway. Imagine you're a salesperson and you spent 12%* of your sales time going after a single client who never returned your calls, didn't seem interested in the product, and never bought anything from you. During your annual review with your boss, would that be considered one of your good moments? Not so much...

* 12% = 9 min / 75 min.

Not to mention, all that time you spent on that one client meant not spending time on other clients. On the real test, this is compounded, because you used a bunch of mental energy on something and you can't use it elsewhere later on. That causes mistakes - as you experienced!

So remember that when you choose to spend extra time / mental effort, that WILL affect your performance later in the test, including verbal. Sometimes, it does make sense to spend an extra 30 seconds, but it pretty much never makes sense to spend an extra 1 min +, because something's already wrong if you need THAT much extra time.

For IR, it depends entirely on your strengths and weaknesses. Know what you're NOT good at and, when those come up, skip them. You may decide, for example, that you dislike the table problems (which have those three either/or questions). A table might pop up that looks easy to you, in which case you can decide to do it - but remember that this will mean skipping something else instead. It needs to be a very conscious choice. It's not about getting everything right! It's about making the best decisions about how to spend limited time and mental energy when you can't do it all.

Plus, you're trying to save mental energy for Q and V. Even when I CAN do something, I sometimes choose not to - because it's not a good use of my mental energy. (Roman numeral questions that are anything beyond super basic? Three questions for the credit of one? No thanks.)

For DS, part of this is just practice. A rephrase should be decently simpler than the starting point. If it's still ugly, there's probably more to do. Now - you might not know what else to do to make it even simpler, in which case, you just have to go with it. It's not always going to work out perfectly. So sometimes you do just have to think, "I suspect there's a way to make it prettier, but I don't see what it is. Oh well." Then see what you can do with what you've got, answer, move on.

Re: essay, I *suspect* this is coming from the same place as your reluctance to let go on multiple choice. You're trying to show how well you write, right? See, I can write a great essay! I've got really clear, logical thought processes! I'm proud of my writing!

Don't think that way. :) Think, instead: nobody is ever going to read this, except the person who grades it, and s/he spends about 1.5 to 2 minutes TOTAL reading this thing. So I need to make it very easy for him/her to skim, get my points, and give me a good score.

How do you read / skim when doing RC? You pay most attention to the first sentence or two of each paragraph, and then you speed up for the rest, just confirming that it does indeed offer support for that main idea in the first sentence.

So do the same when you're writing - make it very easy for the person to read and grade you because there's not a lot of superfluous stuff around to make anything confusing.

And remind yourself that this is NOT a test of how good a writer you are. You can write way better than you want to on the GMAT. You just want to give them what they need to give you a decent-enough score, and you want to save most of your energy for the more-important later sections.

[ch339]

Thanks, Stacey.

Do you think early May is an appropriate time to sit for the GMAT? I am indeed aiming for 780+ (7+ IR). I will be prepping on my own but already find myself losing motivation. I haven't really studied in days and don't want another 3 months to go by again before I resume my studies. Do you have any advice to maintain momentum?

To clarify, in your experience, would I be ready by May? I've always done well on standardized tests and feel I should be progressing more quickly. I worry that I'll somehow never be ready.

Exhibit A:
1/3/14: Just took Kaplan CAT 2 (700) 50Q 41V 4 IR AWA unscored

1 quant question was repeated. I remembered it, so I was able to get it correct. I think I would be able to tackle a similar problem I hadn't seen. Two verbal passages were repeated. I started to guess, and time ran out on quant (-13, 6 omitted, so -19 total). -8 DS, -5 PS. I'm not sure my quant score could be that high on the real test with that performance. ~ 1.5 min behind by question 10--decent. Questions 13-15 slow, but 2/3 right--6 min behind. Then danger zone for a few questions, most of which I got right. Then 8 wrong answers in a row before the clock ran out. I think it was a combination of hard problems and rushing to try to finish. That hadn't really happened to me before, but I think as I'm gaining more competence, I'm setting myself back. There are definitely several problems I would not have been able to figure out 2 weeks ago. I also bombed CR (-6) and RC (-8), only -2 SC (Should it be -0?). Aiming for 1 test a week, so it'll be awhile before I know whether I'm finally getting my timing under control.

1) Is it normal for timing problems to become exacerbated before they get better when one is transitioning to a new level of proficiency?

2) On track for a mid-may test date?

3) I read on Kaplan's site that the GMAT determines your ceiling by the question level at which you're roughly 50% accurate. Do you agree with Kaplan's claim? That would (kind of) explain how I scored 50 with ~50% of questions wrong or omitted. Many of them were rated "high." Actually, no. Of the 21 questions rated "high" that were presented to me, only 9 were correct. Even if we assume that every question I did not get to would have been "high," I would have had only 33.3% accuracy.

[StaceyKoprince]

I can't give a timeframe for anyone aiming for 780+. That's just way too high - there are even some people who want to teach for us who've been trying for a *really* long time.

Given that you clearly don't have any desire to spend years (nor should you!!), figure out what you're actually willing to do.

Are you willing to GO HARD for... 1 month? 2? 3? Pick a timeframe, go for it, and take the test. If you get there, great. If you don't, you'll know that you put everything out there, and you can make an informed decision about whether to keep trying.

Re: your Kaplan test, I agree. I don't think the real test will give you a 50 if you omit the last 6 questions. You've got to fix the timing issues or you're never going to get close to 780.

Note that this doesn't mean you have to get everything right. You DO, though, have to figure out which ones are way too hard / will take way too long, so that you can skip those ones, instead. (In this case, skip = guess quickly and move on.)

Questions 13-15 slow, but 2/3 right--6 min behind.

Do not try to qualify / make this better by saying "but I got 2 out of 3 right." You should have gotten 2 out of 3 wrong (if not all 3!) and been no more than a minute or two behind max - that would have been far better for your overall performance in the end.

Is it normal for timing problems to become exacerbated before they get better when one is transitioning to a new level of proficiency?

Yes. Every time you get better, you get sucked back into thinking "Oh, I can do this!" and then BAM there goes your timing. :) Fix your mindset. Your goal is NOT to get everything right. Your goal is to recognize and let go of the too-hard problems before they mess up your performance in the section.

The test doesn't set a "ceiling" for you and the algorithm is extremely complex. You can think of it as figuring out your scoring level by zeroing in on your "60% correct" level (though this is a simplification). Think of it this way: if I give you 10 questions and you get them all right, have I found your ability level? Maybe, maybe not. I don't know whether you can do even harder stuff.

But if I give you 10 questions of varying level - questions that have been very carefully tested so that I know the ability levels of who tends to get them right vs. wrong - and you get 6 of them right, then I know not only what you can do but what you can't do. So I'm more confident that I'm calculating the right scoring level for you.

What you want to do is make sure that the ones you get wrong aren't easier / lower level problems - because if you get those wrong, I'm going to lower the score I calculate for you. If you run out of time with 6 questions to go, you're going to get too-easy questions "wrong," because the test is adaptive, and it continues to calculate as though you are answering each of those omitted questions and getting each one wrong. There goes your score.

When you do hit a crazy-hard question, you want to be able to say, "Oh, look, this is ridiculous! I can get this wrong without hurting my score. Yay! Moving on."

The test makers do figure that you're going to have a few careless mistakes over 4 hours - so they build in the idea that you don't have to be perfect. But you can't get too many lower-level ones wrong without really hurting your score.

[ch339]

Hi Stacey,

I've been trying to stay motivated and have studied every day for at least an hour (usually min. 2).

Although I continue to do well in OG Verbal (-2 for last 4 completed verbal passages (27 question), -4 for last 42 CR, 2 pairs of RC/CR sets), I can't break 45. Similarly, I think I am getting better at quant, but I keep getting 49. I've had one 51 on a good day (TPR) and one 50 (Kaplan) that seems inflated because I did NOT finish quant that day. And IR is still all over the place. Last week, I got a 7 on Kaplan CAT 3 (710, 49Q, 44V), had no trouble finishing the section, and just "got" the questions (8/12 correct). This week, I didn't answer the last question and only got 4 correct, for a total score of 3. I hadn't done that badly in ages.

On Kaplan CAT 4 (3/11/14) I received:

Actual: 720 49Q 44V 3IR

Goal: 780 51Q 47V 7IR

Need: +60 +2 +3 +4

Avg.: 706 47Q 42V 4.37IR (all 10 CATs)

Avg.: 720 ~49Q 43.5V 4.6IR (last 6 CATs)

Although I thought quant was going well (it actually was before this), I received a string of wrong answers: 6 (W)rong, 1 (R)ight, 2 W, 1 R, 1 W, the last 5 R. I definitely think fatigue and rushing toward the middle are the culprits. Prior to that, I had: First 5 R, then 1 W, 1 R, 1 W, 1 R, 1 W, 2 R, 1 W, 2 R, 1 W, 5 R. Overall, I was -14 for quant. I've gotten somewhat better at rates, but some number properties questions are annoying. Still working on timing, but a more comfortable pace than last week. I should not have spent extra time on #23 (~6m), screwed me up and threw everything off. Prior to that I had spent ~2.5m on 3 questions and ~3m on another. All other questions were within 2 minutes. I should've just guessed because I was never going to spend "only another minute" on #23. Screwed up my pacing and cost myself 2 more questions I definitely would have gotten.

Looking over my Kaplan tests, I always have a string of 5+ wrong answers starting around question 20-25. I did not seem to have this pattern with TPR (few weeks ago) or Manhattan/GMATPrep software (few months ago). However, I was also at somewhat different points in my studying and performance level, so it may be more my adjustment and less a function of the tests.

Verbal was okay, but I would like to crack 99th percentile. -1/3 on Passage I, -1/4 on II, -0/3 on III, and -3/4 (all detail) on IV. Still very hit or miss depending on the passage. IV was exhausting--420w over 5 paragraphs, pretty short for real life--but it was hella boring, so I just wanted it to be over. I have realized that I also tend to do better on passages that reference subjects I like or am already interested in. Quelle suprise! I've taken to reading the entire passage since I can't go back in a section. It was a longer passage, so I felt myself losing focus and not retaining as much. This always seems to happen in verbal. Full tests should be doable by now, but I get tired. -5/14 RC, -4/14 CR, -2/13 SC. -11 verbal. First 8 right. From 18-24, W, R, W, R, W, R, W. 3 W in a row on 34-36 (Passage IV). Decent enough pacing.

For quant I just need more practice, but I'm still not sure how to study for verbal. I don't seem to be getting any better on that front, and I'm not sure what to do. I've decided on an early June test date (first week).

Also, I remember reading (I think on your blog post), that most people will get ~60% of questions in each section right, except at the tails. For 51q and 45+v, how accurate should you be on average? 70%? 75%?

Thank you.

[StaceyKoprince]

I should not have spent extra time on #23 (~6m), screwed me up and threw everything off.

Not to mention all of the mental energy it cost you. You might not have had that string of 6 wrong if you'd made better decisions earlier in the section.

Now, don't feel badly that you didn't do things perfectly on that test. It's much more important, in my opinion, that you can come back and point that out to me now. You're analyzing and figuring out what you need to do differently next time!

I always have a string of 5+ wrong answers starting around question 20-25.

That can signal mental fatigue - especially if you're spending too long on a subset of questions. It's going to catch up with you, both the timing pressure and the mental fatigue. So, again, I'm glad you're noticing this. All of this should help you to cut yourself off the next time you're tempted to spend 3, 4, 5 minutes on a single question.

On verbal, you're really close to 99th percentile - it starts at 45.

I have realized that I also tend to do better on passages that reference subjects I like or am already interested in.

Yeah, we're all like this. :) Think of someone you know who likes certain kinds of topics - science or business or humanities or whatever. When you see a topic you don't like, think of that person and say to yourself, "Oh, I want to remember a bit of this so I can tell _______ to read this - I know she'd like this." (You can use me as your default if you like - I tend to like topics that others think are boring!)

That changes your focus - you actually do have an interest in telling your friend about it because you know s/he will actually care. Also, if it's really long, feel free to speed up even more on the details. Just get the big picture, so you can tell me about it later, and tell yourself that I can read the details myself.

Then, when you get a question, you can pretend I (or your friend) just asked you that. And you say, 'Oh, hang on, let me check..." and you go back to read that part.

Try this for verbal:
When you're reviewing, review everything. Identify ALL of the questions on which you narrowed to two and guessed, even when you guessed right. And answer these questions:

1) why was the wrong answer so tempting? why did it look like it might be right? (be as explicit as possible; also, now you know this is not a good reason to pick an answer)
2) why was it actually wrong? what specific words indicate that it is wrong and how did I overlook those clues the first time?
3) why did the right answer seem wrong? what made it so tempting to cross off the right answer? why were those things actually okay; what was my error in thinking that they were wrong? (also, now you know that this is not a good reason to eliminate an answer)
4) why was it actually right?

For example, I've noticed that, on CR, a really tempting wrong answer will have exact word matches with the passage but the meaning won't be quite right. A less-tempting correct answer will use synonyms instead of exact word matches... but the underlying meaning WILL actually be correct. So they're tempting people to cross off the right one because it doesn't 'look' as good, even though synonyms are perfectly fine (as long as they really are synonyms for what you want!).

For percentage correct, they don't disclose, but my sense is that its in the 70% to 80% range at the top-top levels. Maybe even higher for someone who's pushing 51 on verbal!