Need advice. Scored 680-700 in mocks but 560 in actual test.

Some things that come to mind immediately:

1. The actual GMAT algorithm is a closely guarded secret. No algorithm can simulate it perfectly (not even the GMAT prep algorithm). Hence it is very inadvisable to make a strategy based on other algorithms and that too one which deviates a lot from the strategy GMAC advises in its Official Guides.

2. According to GMAC, first 10 questions are no more important than the other questions. If giving extra time to first few questions and stealing time away from the others gives you better results, it could be a weakness of the simulated algorithm. Don't you think that if there was such a bias in the actual GMAT algorithm, GMAC would have fixed it since it knows that GMAT takers believe in this? Perhaps it was somewhat true in the earlier versions and perhaps the GMAT prep software still uses the earlier versions of the GMAT algorithm - we will never know and can just speculate but GMAC insists that every question is equally important. Listen to them. There is one advantage in giving a little bit of extra time to the first few questions - it helps you get in the zone and increases your confidence because you know when you are doing well. But if you don't have time to spare, it is still not advisable to pace yourself in a biased manner.

3. We know that the algorithm is adaptive. So it decides on the next question based on your response to the previous question and adjusts your score depending on your response too. In that case, it is logical (though still speculative) that random guesses in a row would have a much more negative impact than those distributed evenly around. So you guess on a question, the software will deduct a point from your score and give you an easier question. You answer it correctly and you back to your original score. You answer the next one correctly and your score goes up. What happens if you guess on 10 questions in a row - the software will keep reducing your score till there may not be enough questions left to get back on track. Mind you, it is still a speculation but logically derived from whatever GMAC is willing to tell about the algorithm. We can be certain that there is certainly no reason why guessing in a row would be better than spreading the guesses around but there might be reasons why spreading the guesses around would be better.

4. To be honest, I am pleasantly surprised that you scored Q44 even with so many random guesses. It means you are fundamentally strong and answered almost every question that you did, correctly. Now, all you need to do is increase your speed so that you can do whatever you do in 2 mins rather than 3.

Well. I don't know what might have happened to reduce ur score so abruptly, but have you considered the possibility of experimental questions?

The only thing which is different between mocks and actual is experimental Q.

I guess 25% of total Q are experimental. So if I look at extremes, you spend all your time on experimental Qs, which might be top in the order and ignored all marked Qs, probably coming in the end.

I guess the safest strategy will be equal time allocations to each Qs since experimental Qs are random ones.

I am not very sure though, but just guessing out the odds.

- Soumen

To add to what Soumen said, the best strategy regarding experimental questions in the exam is to forget that they exist. Since there is no way to know whether the question is experimental or not, the info that there are experimental questions doesn't help us at all - save for giving us some hope during the test that the previous question we lost our mind on was hopefully, an experimental!

Soumen,
Thanks for your insights.

Karisma,
I think your points are making things clear about my dismal performance. Its best to adhere to common well known strategies rather than sticking to some particular mock results.

My realistic level for quant is 48(same as last GMAT attempt). Although I have scored 49 nd 50 in some mocks, but those were either too easy on quant or repeat attempts. If I had attempted all questions in methodical manner, I would have got at least 47-48. Tricky DS always trips me.

Regarding my verbal, the most unpredictable section is RC. Sometimes I get 75% correct, but on bad days I just get 10% correct. In spite of trying hard, it still remains tricky. So in the real exam, I gave extra time on RC questions, leading to the downfall (last 12 questions random guesses). CR also consumed lot of time, but my historical accuracy rate in CR is 65-75%. Same for SC.

I am planning to take the GMAT again in 40 days after focusing on my weak areas with a simple strategy of attempting all questions with roughly equal time allocations as advised by you.

Thanks a lot for your help!!

Kudos to both of you..
Sam

Hi Sam,

It looks like you have already identified the main problems. You need to fix timing and get your RC more consistent. You don't seem to have timed out as badly on your MGMAT tests as you did on the real thing, but I do see you falling behind really far and then sliding at the end. The trick is to practice increasing your accuracy *within* the time limits. Practice RC realistically: if you're reading a short passage (40 or fewer lines), spend 6 minutes total to read the passage and answer 3 of the questions, and if you're doing a longer passage, expand that to 8 minutes and 4 questions. Don't give yourself any more time than that! Later on, you can go back and review untimed, focusing on understanding the subtle differences between answer choices and finding support for one over the others. But the goal is still to improve your timed performance. Taking time to review is a way of training yourself to see the right things before diving in and doing more timed work.

Make sure you are working with clear benchmarks. Basically, you need to get through about 8 problems every 15 minutes. You can adjust this slightly depending on how much SC you've seen, as those should go a little faster. But make sure you're hitting those benchmarks. If you're 30 minutes in and you're on problem 12, you're probably going to have to guess randomly on a few hard-looking problems (not in a row) to get back on track!

Here's a more detailed version of my RC advice:

how-to-get-good-score-in-rc-140302.html?hilit=passage

Hi Dmitry,

Loads of thanks taking a look at my stats. Really appreciate it. Kudos 2 u!!

As you have already indicated, RC is my pain area in verbal. I get tripped by both main point and inference questions. Basically, I can zero down to 2 choices, but always end up picking the wrong one. I will follow your advice while prepping up for RC by following realistic time allocations in RC questions. Will focus mostly on MGMAT & OG questions for practice.

I am preparing a 30 day study plan, starting with diagnostic. I don't think my aptitude will change in a month, but the real test score has just destroyed my confidence. Need to build it again...

I did not worry about the 44-45 in quant as MGMAT is harder in quant, but it is great to practice and the overall score is more or less the same. Although I do want to increase my quant score, even if for a single point.

Thanks again!!
Sam

Hi Dmitry,

I am doing RC practice under timed conditions as advised by you. I am changing my core approach of reading RCs quite a bit.

Earlier I read the questions after reading whole passage, created para summaries and then looked at the question stem.
Now, I am reading the question stem just to deduce that it is broad based question like main point or an inference question.

For main point questions, I am reading the passage fast without worrying much about specific details. (just to get the basic structure of the passage). I am thinking about the main idea in my own words before looking at the choices.
For inference type questions, I am going thorough the passage with a objective of finding the components of the question stem.

With this approach, my overall time\passage has reduced. It's quite amazing that even though I am referring the passage again and again.

My accuracy by following this approach, my accuracy for easy and medium Qs has increased to 80%, but still need to see the good accuracy on hard Qs though.

Am I going on right track? I will keep practicing on the same lines then.

Although I do understand that each individual has his own optimal RC approach, still will be great to know your perspective.

Thanks,
Sam

If it's working for you, then great. That approach makes me a little nervous because your initial read-through of the passage should be something that helps you on all 3 or 4 questions you get, not just the first one. I find it more efficient to read the passage well once and get a feel for the overall. Then I can answer main point questions without much trouble, but my mental map of the passage allows me to tackle more specific problems, such as inferences, more easily than if I were going into the passage blind.

Still, if you find that the approach works for you, go with it. I'm just curious what happens if the first question is a specific one? Are you able to handle that well even if you haven't given the passage a good initial read? Also, do you ever miss main point questions because your read-through was too sketchy?