Manhattan GMAT books- question difficulties + content ?

[jonathanc]

Hello ,

I dont really know if my question should be posted here , so please forgive any misposting (does this word actually exist?) .

I started to work on GMAT since about 3 weeks, by myself since there is not any organisation where I am living (Guangzhou / china). I currently have two books to do that : Official Guide 11th (orange) and Kaplan GMAT premier program 2009 + CD. I have been looking around I didnt find any other material that I can buy directly except online.

I dont really know what to think about Kaplan, I only used the online ressources until know because I feel it is a good way to be in the situation (computer + real timing) and I indeed do much more mistake than with the OG which is certainly the sign that it is harder (or that I am weaker face to a computer).

Anyway I am frustrated about something with both materials : I am totally unable to judge the difficulty of the questions. On the OG it is written on the cover " New Organisation of questions in order of difficulty saves study time" but nevertheless they never write anywhere which one are easy, medium, hard. Maybe they dont provide hard question maybe not easy one ,anyway I dont know. After respectively 156 , 104, 69, 65, 67 questions in PS, DS, RC, CR and SC my proportion of right answers never changed 93 % / 98 % / 90% / 92% / 67%. And with Kaplan the same pattern 85% , 85% , 83% , 83% , 70% for about 150 questions of each. These results are consistent with almost every serie I am doing (always series of 30 /40 questions for maths and 35/45 for verbals with a timing in each case and a rough mix of every type of question).


I think I will need to find more books since I will soon finish every questions I got but I have been unable to find hard copies of Manhattan Gmat Books where I am living ( If you know a retailer around I am interested - Guangzhou / Canton.) . I know it s possible to buy Manhattan GMat books online through your website but I am a bit reluctant to buy books when I can't touch them and have a look first, especially when they weight about 25 Kg and cost 300 bucks. On a first tough I really like the way you are managing this forum, and the possibility to talk about the encoutered difficulties with instructors so here comes my questions in order to help me to know if I want to buy your book set (the one with everything ):

-Does the new 12th edition OG book provide new questions comparatively to the 11th ? if yes is it only 10 news for each kind or more like 80% of new questions ?

-Are the two small independant Verbal and Quant books providing totally different questions or will I find roughly the same than in the OG ? and more important is there any clear ranking of the difficulty for each question ?

-Are the Manhattan GMAT questions in the strategy books and in the online bank providing questions for each level of difficulty and with an equitable repartition ?

-Is this set suitable for self study and can allow somebody to improve significantly a score even without coaching and courses (I dont really have the opportunity here , seems there is not any training) or it will not bring me anything to do more question and I should better keep working on my mistakes (that I will do anyway).

-How can we know how many questions we are supposed to get wrong if we aim for a given score (let say 720) ? and how many questions of each level of difficulty we are likely to encouter to succeed on this score ? (I assume I cannot make more 2/3 mistakes in quant since I cannot make less than 5 mistakes in sentence correction).

Thanks a lot for your answer and sorry if it nos the right place or if it is already answered somewhere else ( I searched a bit but didnt find).

Jonathan

[StaceyKoprince]

Addressing your questions in order:

The OG questions are listed in rough order of difficulty, so as the numbers get higher, the questions get harder. It's true that they don't tell you a specific number or level of difficulty for a question, but they don't tell you that during the test either, so that information isn't really necessary (beyond knowing that, generally, question 1 will be easier than question 50, which will be easier than question 100, and so on).

It is not the case, however, that we can say that question 33 is definitively harder than question 32. Part of the determination of whether something is hard depends upon you and your own strengths and weaknesses. For questions right next to each other in the book, you should generally think of those as "about the same" in difficulty. But you can also plan that questions 1-10 are easier than questions 30-40, for example.

I don't know how Kaplan organizes their materials, so I can't help you with that one.

More importantly, this test is not scored based upon percentage correct, so concentrating on the percentage you get correct while studying does not actually tell you very much. It's important to take practice tests periodically (every 2-3 weeks) under official timing rules in order to see how you are progressing and to plan your study time accordingly.

The new 12th edition book has approximately 300 new questions; the other 600 questions are repeated from the 11th edition.

The Quant and Verbal supplements have zero overlap with OG11 or OG12 - totally different questions. I have heard that the people who make these books are planning to publish new editions of the supplements sometime this fall - FYI.

The difficulty level thing is the same for the supplements: in general, lower-numbered problems are easier than higher-numbered problems.

The MGMAT books do not list difficulty level of questions, but we do offer a varying level of difficulty. The online question banks and the CATs do show difficulty level buckets (eg, a question will be labeled 500-600 or 600-700). You will not be able to see these difficulty levels until after you have finished the online bank or the CAT exam. The question banks include a full range of questions, as do the CATs. The CATs, however, are fully adaptive, which means you will only see the levels that you "earn" - eg, you may not see 700+ questions if you do not perform well enough on lower-level questions to "earn" the right to do the higher-level questions.

Many people have used our books for self-study and have acheived their score goals. Not everyone achieves their score goals, of course. So the materials can be sufficient, but obviously that depends on the quality of your study. We also offer other packages that include tapes of our classes (viewable over a broadband Internet connection) and we even offer live online classes - all you need is a computer and a broadband Internet connection. In my live online classes, I have had students who were physically in Australia, Greece, South Africa, and elsewhere. (I don't think I've had anyone from China yet!) Anyway, the classes are obviously more expensive than just the books, but there are potential options for classes - live or taped. Browse our web site if you want more details.

As I mentioned earlier, CATs are not scored based on percentage correct. Everyone scoring between about 400 and 700 gets around 60% of the questions right. At a 720 level, you'd maybe get 65-70% of the questions right. You will not be offered a standard or set mix of questions at different difficulty levels. The test will merely adjust to your performance on each question, and where you're at when time runs out will be your result. For instance, someone could answer the first 20 questions correct and be scoring 800, then run out of time and leave the next 17 questions blank. That person would score below the 50th percentile despite the fact that s/he answered lots of 700+ questions correctly. (That wouldn't really happen in the real world, of course, because nobody can get 20 questions in a row correct - this is just an example to illustrate that the scoring does not work in the way you are used to paper tests being scored.)

So if you want to aim for low 700s, plan to get about 60 to 70% of the questions right. In other words, plan to get 30-40% of the questions wrong, which is a much larger number than people generally realize. Also, the levels of the questions that you get right vs. wrong will make the big difference in your score. For example, if you want to score a 700, and you answer all of the sub-700 level questions correctly but all of the 710+ questions incorrectly, you can still get your 700-level score. In other words, you can get all of the "too hard" questions wrong and still get the score you want.

The tricky part is: you'll never know the difficulty level of the question that you're answering, because the test won't tell you and it is very difficult to figure out just by doing the question. So the default is: if you can answer it within the expected timeframe for questions of that type, then you answer it, and if you can't, then you make an educated guess and move on.

[jonathanc]

Stacey,

Thank you very much for your long, precise and interesting answer (especially the point about score which change a lot my perception of the situation).

So I conclude the following (I dont mean to generalize it , it s just for my situation).

-Doing new problems again and again will maybe not help me so new version of the OG is not necessary, I still can do all the question of the OG o get familiar with every kind of question (including difficulty + type) but I can take the two small books as a stock to apply the work and learning I will try to get from my mistakes in the first one.

-The Manhattan Cat could be helpful to get an "instant" knowing of oneself situation but is not a training purpose (anyway you say in,many other post that it is not useful to do a test per day that I believe easily and on the other hand I have neither the time nor the energy to do it)

-In math the main problem is not whether it is possible to answer a question or not (I guess anybody should be able to answer everything if you have 2 hours per question) but to be able to let the question go. The reverse is : It is difficult to not let too much question go and to find the stamina and the time to answer to the right one corresponding to your own level of difficulty in order to get your score.


-The way of calculating the score is quite complicate to apprehend and is in anyway not related to the way of calculating the score for paper exams in my Kaplan book , so I should forget about my % wrong-right and instead take a stopwatch and check which question i can answer in 1 min, which in 2 , and which in 3 min. THen try to see what I can change and how I can arrange myself with these results.


=> Here I need a new advice : so do you suggest the following strategy Use 2 min per question whatever is the question and at the end of the 2 minutes ideally always choose an answer whether you really solved the problem or not ? Until know I tried to speed myself up on the easy question to save time to answer the hard one.

-In Verbal , I dont really think I will face time problem maybe because I dont really find necessary to find the answer while in Maths it s very hard for me to let a question go since I want to find the solution.


Still I dont really understand how you are supposed to train yourself if you dont really no what is your objective (I mean not knowing what mistake you are allowed to do). I guess that is the purpose of the test,really problematic.

Thanks again for your previous answer.

John

[StaceyKoprince]

Your objective is to get right everything that you can within the general per-question time limit (not the same as "everything") and to recognize when you can't and let those problems go. If you do that, you'll generally get close to the best score that you are capable of getting.

For quant, I generally follow a 1-minute rule. If I'm going to answer this thing by the 2-min mark, then I should be on track and totally know what I'm doing by the halfway mark (1 min). If I'm not on track at that point, then I forget trying to find the right answer and switch to trying to eliminate wrong answers. I spend another 30-60 seconds narrowing the answers, and then I pick something and move on.

Now, it's okay to go a bit over sometimes, because we are going to answer some other questions more quickly. I generally assume it's okay to go to about 2.5m but not much beyond that. And I ONLY spend longer than 2m if I KNOW exactly what I'm doing but the question is long and complicated for some reason, so it takes a little extra time to do the work. I do NOT spend extra time if I'm thinking, "Hmm, I'm not totally positive, but I know I can do this if I just spend some more time..." That's not good enough. There has to be no question in my mind - I absolutely know what I'm doing if I'm going to spend any extra time.

To do this, you need to know about how long one minute is without looking at a watch or stopwatch. If you don't have one already, buy yourself a stopwatch with lap timing capability. When you go to do a set of problems, start the stopwatch but turn it over so you can't see the time. Every time you think one minute has gone by, push the lap button. When you're done, see how good you were - and whether you tend to over or underestimate. Get yourself to the point where you're within 15 seconds either way on a regular basis (that is, you can generally predict between 45 sec and 1min 15 sec).

You'll also have to learn how to make educated guesses. How do I actually identify wrong answers so I can cross them off? It's different depending upon the type of problem. On geometry, you can often estimate. If it's a theory question, or one with variables in the answers, you can often try numbers. You could also test some of the numbers in the answer choices. Maybe you know the wrong way to do a problem even if you don't know the right way. (I once saw a combinatorics problem which asked us to pick a team of 5 from 9 players. One of the answers was 45. A lot of people who didn't know how to do the problem probably still did know that 9*5 was NOT the right way to do it.) Etc.

[jonathanc]

Thanks a lot Stacey.

I Will try to follow your method.

John

[jonathanc]

Stacey,

I apologise to go back and forth with this question again and again, but this is an important point for me to understand the calculation, especially because I use this to know where I concentrate my time and efforts.

I did order the MGMat guide (by the way it would be a good idea to let people use their own DHL account for far away people because the shipping fees area little bit expensive) and with it I can use the Online CAT (I could have done it before actually because there is a free one)

So I did use the 1 st one right away to know where I am standing right now and to understand the question difficulties and the scoring .

With following results :

Quant 47 (81%), Verb 35 (77%) , Global 680 (90%).

Ok until here nothing unsual it is consistent with my previous results in the way that : I am not doing really better in one of the field (the score are quite different but I consider 4% difference in the percentile as being pretty close even if the score difference is 12, Am I wrong on this ?) and my score is a bit better than with GMAt prepr certainly because I didnt hurry for the last 7 in quant but not that different either (I think I had 660 with GMAT prep last week, both case I am writing seriouslyn the essays before, and I take it at the same time of the day to be in real situation).

Then I looked the details of the test report (which is indeed very precise and well made, I have all the informations I want there.) .

I have around 60% of good answers , very consistent with what you said ( I used your method for the timing , i felt good to not need to speed up for the last 7 questions without any time to read and having 5 wrong anwsers like the time I used the GMAT prep software, even if i had to give up on harder questions that i could have solve earlier).

Thanks to the nice informations provided on the report I can see that my Right/wrong are as follow :

Quantitative R W
300/500 1
500/600 5 2
600/700 11 8
700/800 5 5
Tot 22 15
Verbal
500/600 4 0
600/700 9 1
700/800 10 16
Tot 23 17


=> Here is what I found very surprising, in your previous post you told that it is possible to have a 700 score by having every question below right and every question above wrong (or something like that). I assume that Verbal is not more important than quant. But my score and percentile are both lower for my verbal section than for my quantitative when I only got 1 wrong answer below 700 in verbal and 10 in quant (Including 2 below 600 !) .

I checked in both section I have 4 times 2 wrong answer in a row and not more, so that cannot explain.

Where do I miss something to understand the calculation ? because if i work the quant in a way that I dont have any below 700 question wrong but still a low score it will not help me much. and in the same way if i can have more verbal question right but below 700 and if it can improve my score I would be totally happy with that...


Thanks a lot for your light,

John

[StaceyKoprince]

the score are quite different but I consider 4% difference in the percentile as being pretty close even if the score difference is 12, Am I wrong on this ?

Nope, you're totally right! The percentile ranking is what you care about, not the raw score. Essentially, your math and verbal skills are pretty similar.

The first thing to note: although we show you a classification for each question based upon the 3-digit scale (600-700 or 700-800), that's not actually how the test works. The 3-digit scale is a combined score for both quant and verbal; there's technically no such thing as a 700-question. (You noticed, I'm sure, that your overall percentile is actually higher than your individual percentiles for each section, yet another indication that we can't really directly compare these different numbers. They mean different things.) We use the "700-800" classification just to make it easier to talk about: this is a harder question, this is an easier question. The terminology is useful as a general rubric: I need to get these kinds of questions right, in general, to get this kind of score. But it's not a precise match. (And, if you note, you actually did get "this kind of score" - a 680 is very nearly a 700! The standard deviation on our test is 50 points and on the real test, 30 points.)

Also, part of the algorithm depends upon the order in which things happen as well. If you have a higher mix of questions wrong towards the end, then that will be worse than having a steady mix, even if you have the same number right and wrong overall. You can think of your score as the level you're at when the timer runs out - so you might, for example, have been at a higher level in the middle but lost steam (and points) towards the end.

Finally, I should clarify: while it is possible to get a 700 by getting everything below right and everything above wrong, it's not necessarily the case that you will definitely get a 700 if you get everything below right and everything above wrong. The order in which things happen matters too. (And, of course, on the real test, around 25% of the questions will be experimental and not count toward your score, so that can change things around too.)