After having aced Common Admission Test (99.97 %ile) \m/ and
having completed almost all my interviews, I have decided to consolidate my
analysis of CAT 2012 result and post it for future aspirants. As pointless as
this analysis is for improving one’s prospects, it surely helps the ‘analytical’,
‘inquisitive’ minds to understand the beast called CAT a little better. The
following research has been done with the help of information from the
internet, especially pagalguy.com.
To start with, let us look at a simple graph between CAT
score and percentile. The maximum marks for CAT 2012 and all previous online
CATs have been 450. One of the 100 %iler scored 367. I scored 348 for my
percentile of 99.97.
It is also interesting to see how the scores vary
section-wise. From the below graph it is obvious that Quantitative Analysis and
Data Interpretation (QA & DI) section is more scoring that Verbal Analysis and
Logical Reasoning (VA & LR) section. The
scores are known to rise very steeply as the % ile touches 100 for each
section. However, I don't have the data to show it.
Since most of the mock tests that we take for CAT are out of
180 (90 + 90), I have scaled the scores down to give a total out of 180. I will call this the raw score.
Assumption Alert – In scaling down, I have totally ignored
the devil called normalization. As
you read further, you will realize that normalization is really a game changer.
Normalization is a black box and the only way to deal with it is to ignore it.
And pray.
The following table is something I am the most proud of. It
will try to tell you how many questions you need to get correct to get a
certain percentile. I have continued with my assumption that Normalisation is
like unicorns -> it doesn’t exist. And I will repeat again -> My
assumption isn’t all that true.
For Eg – If you have attempted 55 questions and want to get
99.8 percentile, you should get at least 46 questions correct. That means an accuracy of ~84%.
As you see in the table above, couple of correct questions separate a 99.0 from a 99.7. Also, for the same number of correct questions, your
percentile can lie in a big bracket. This means that where you lie in the
bracket depends upon how your scores get normalized.
Feel free to comment and discuss this further. Happy to
help.
Other Relevant Posts:
Useful stuff! Particularly the last table. Thanks for posting... Seems that speed without accuracy is practically useless.
ReplyDeleteit means that percentile largely depends on no. of attempts.
ReplyDelete@Shrish Rai - Percentile does depend largely on number of attempts but it is not directly proportional mathematically. As you must have seen, I have assumed that normalisation is not playing a role. My assumption is not correct.
ReplyDeletewell i believe normalization plays a really important part here as stated by the writer himself and the table prepared does not consider it , so its kind of not true i would say.
ReplyDeletei have observed in cat percentiles that if in one section a person has good percentile then the OA %ile gets increased way too much...eg my friend had 99.89 in VA but only 80 in QA...his OA was 98.xx
similarly accuracy plays an extremely important role eg i did only 11 Qs in quant out of which 1 was definitely wrong...though my quant slot was one of the toughest i still ended up with 81 %ile...similarly in VA i did 23 Q's and got 97.xx %ile...my OA stood at 94...
SO for all future aspirants I'd say try to be at least 80 % accurate...