Page 1 of 1

Hi Guys I have uploaded Rank versus percentile data on the based on the assumption out of 2.15 lacs CAT Applicants 2.0 Lacs appeared

and last year out of 2.05 there where 1.85 appeared

Looking on the trend for top B School 95%tile or more is required that mean you need to be in top 5000 to 7000 to be in top school

and last year out of 2.05 there where 1.85 appeared

Looking on the trend for top B School 95%tile or more is required that mean you need to be in top 5000 to 7000 to be in top school

@mathaddict said:@mathaddict@rahul_cattakerRegarding your flaws..1) If Mr.X can solve 25 easy qns in 70 mins or 18 difficult questions in the same, according to the normalization process X would be getting the same score. As you said to try to to solve as many quant qns which many would not, one would eventually end up getting the same percentile in either case , as they are called difficult for a reason and you cant solve equal number of easy and difficult qns in a give time . So you cant actually see it as a flaw.2) If everyone strictly follows the NDA and dont discllose any qn, every one irrespective of their slot will see the qn first time(Unless the qn is from previous cat papers ofcourse.. ). What is the advantage of a well prepared person keeping his exam on the last day? He would eventually face from the same set of qns. He would definitely perform the same way if he had written the exam a day earlier or even a week earlier.If he was so keen to prepare till the last day , I am sure he would have prepared for cat since a good amount of time and the 3 weeks window isnt going to add some enormous advantage for him . His correct answer will not effect anyone who wrote cat in previous slots.P.S: I am in no way connected to prometric or cat organizers and the above explanation is just my interpretation. Just an optimistic point of view

you are right on the point that tough qts would be solved by prepared cat taker

any they deserve extra marks for that

For the Second part- most people take cat dates to buy time to prepare them self

let me give u an example

if a person X plans to take CAT in JUN and thinks that 3 months would be sufficient to prepare

but by the time cat forms are declare he feels some extra time would be beneficial for him

This type of thinking is found in many cat taker and they try to take a later date as possible to prepare

Now If the end slots are all packed with people which are seriously prepare cat taker then

according to my theory they will be attempting a easy set of qts as per IRT (which are not actually easy)

When you try booking a slot for cat you will observe that last few days are packed earlier than those unpacked also cat is about practice the more you practice and get feedback from forums like this the more u accustomed to cat

I appreciate your concern for a ideal student but in cat it is about managing the qts properly rather than solving each of them :D

Also the above mentioned analyses is just an inference based some assumptions

I wanted someone to put more light on the CAT scoring mechanism

@rahul_cattaker

Regarding your flaws..

1) If Mr.X can solve 25 easy qns in 70 mins or 18 difficult questions in the same, according to the normalization process X would be getting the same score.

As you said to try to to solve as many quant qns which many would not, one would eventually end up getting the same percentile in either case , as they are called difficult for a reason and you cant solve equal number of easy and difficult qns in a give time .

So you cant actually see it as a flaw.

2) If everyone strictly follows the NDA and dont discllose any qn, every one irrespective of their slot will see the qn first time(Unless the qn is from previous cat papers ofcourse.. ). What is the advantage of a well prepared person keeping his exam on the last day? He would eventually face from the same set of qns. He would definitely perform the same way if he had written the exam a day earlier or even a week earlier.If he was so keen to prepare till the last day , I am sure he would have prepared for cat since a good amount of time and the 3 weeks window isnt going to add some enormous advantage for him . His correct answer will not effect anyone who wrote cat in previous slots.

P.S: I am in no way connected to prometric or cat organizers and the above explanation is just my interpretation. Just an optimistic point of view

Regarding your flaws..

1) If Mr.X can solve 25 easy qns in 70 mins or 18 difficult questions in the same, according to the normalization process X would be getting the same score.

As you said to try to to solve as many quant qns which many would not, one would eventually end up getting the same percentile in either case , as they are called difficult for a reason and you cant solve equal number of easy and difficult qns in a give time .

So you cant actually see it as a flaw.

2) If everyone strictly follows the NDA and dont discllose any qn, every one irrespective of their slot will see the qn first time(Unless the qn is from previous cat papers ofcourse.. ). What is the advantage of a well prepared person keeping his exam on the last day? He would eventually face from the same set of qns. He would definitely perform the same way if he had written the exam a day earlier or even a week earlier.If he was so keen to prepare till the last day , I am sure he would have prepared for cat since a good amount of time and the 3 weeks window isnt going to add some enormous advantage for him . His correct answer will not effect anyone who wrote cat in previous slots.

P.S: I am in no way connected to prometric or cat organizers and the above explanation is just my interpretation. Just an optimistic point of view

Hi Guys

Would like to share some points which i analyzed for CAT evaluation

first a raw score is calculated as everyone know and after that the normalization which every one knows but

before normalization there is Item response theory applied which first plots a graph with respect to a question

I'll explain in detail

Prometric has a qts bank but the dont know that the qts which the grade easy are really easy or not and tough ones are really tough or not

For this for each qts a graph is plot (i have assume that qts in any slots are not repeated in other slots and with in a slots there a mini slots of papers)

So the graph which is plot will depend on how many people were offered with the question (eg 1000) How many attempted it (Eg 500) How many attempted right (eg 100)

so out of 500 only 100 could solve it right (20%) hence the qts would be matched as tough even if they have marked it as easy earlier

(above is the item response theory IRT as mentioned on CAT site u can go to youtube to understand IRT)

Now once all the question are classified the scores would be scaled or normalised based

Flaw in this step which may be possible

1)eg.

If a Qts is at the end which is easy the one would normally attempt it at the end as time is less

some may just guess it by elimination in this process if the right attempts are low the qts which is easy would be marked as tough

On the other side If you try to solve as many tough qts which many would not then you may get higher scaled score

This can be useful for those whom cat is yet to come

2)eg.

As All of us know that the one who take the slot at the end are more prepared for CAT and have planned it well before hence the get the form first and get the last date

So the competition in last slots would be more hence rights attempts to a qts are more hence IRT would mark most of the questions as easy :(

Above is just my analyses which may go wrong would like is some good statistics background person put some light on IRT and is what I guess is right or wrong

Would like to share some points which i analyzed for CAT evaluation

first a raw score is calculated as everyone know and after that the normalization which every one knows but

before normalization there is Item response theory applied which first plots a graph with respect to a question

I'll explain in detail

Prometric has a qts bank but the dont know that the qts which the grade easy are really easy or not and tough ones are really tough or not

For this for each qts a graph is plot (i have assume that qts in any slots are not repeated in other slots and with in a slots there a mini slots of papers)

So the graph which is plot will depend on how many people were offered with the question (eg 1000) How many attempted it (Eg 500) How many attempted right (eg 100)

so out of 500 only 100 could solve it right (20%) hence the qts would be matched as tough even if they have marked it as easy earlier

(above is the item response theory IRT as mentioned on CAT site u can go to youtube to understand IRT)

Now once all the question are classified the scores would be scaled or normalised based

Flaw in this step which may be possible

1)eg.

If a Qts is at the end which is easy the one would normally attempt it at the end as time is less

some may just guess it by elimination in this process if the right attempts are low the qts which is easy would be marked as tough

On the other side If you try to solve as many tough qts which many would not then you may get higher scaled score

This can be useful for those whom cat is yet to come

2)eg.

As All of us know that the one who take the slot at the end are more prepared for CAT and have planned it well before hence the get the form first and get the last date

So the competition in last slots would be more hence rights attempts to a qts are more hence IRT would mark most of the questions as easy :(

Above is just my analyses which may go wrong would like is some good statistics background person put some light on IRT and is what I guess is right or wrong