CAT 2014: Result marking process decoded; Calculate percentile using GATE formula

Columnist & Author, MBAUniverse.com

Updated on December 15, 2014

The Normalization process to be implemented shall adjust for location and scale differences of score distributions across different forms

Experts with MBAUniverse.com have been working and have decoded the pattern of normalization and calculation of percentile in CAT 2014

The confusion on transparent process on score normalization and calculation of percentile continued even after publication of score normalization by CAT Centre 2014. The process claims to focus on Fairness, Equity and comparison of performances across the 4 test sessions in CAT 2014. However, the Normalization process remains without details.

Claim of Equity

CAT Centre 2014 declared “In order to ensure fairness and equity in comparison of performances of the candidates across different test sessions, the scores of the candidates shall be subjected to a process of Normalization. The Normalization process to be implemented shall adjust for location and scale differences of score distributions across different forms and the scaled scores obtained by this process shall be converted into percentiles for purposes of shortlisting.”

As per GATE pattern

On the question of transparency of normalization process across the multiple sessions CAT Centre 2014 states “The process of Normalization is an established practice for comparing candidate scores across multiple Forms and is similar to those being adopted in other large educational selection tests conducted in India such as Graduate Aptitude Test in Engineering (GATE).”

The problem arose when the candidates had to understand first the GATE process which stands a bit changed in 2014. Those appearing in GATE or have understood the process it is good for them to understand it well but for those who are unaware about the process have to understand the GATE process first. In fact the commerce, science, medical, paramedical and graduates in Humanities who appeared in CAT 2014 still remained oblivious of this process.

Know the GATE normalization process first

During 2014, the CE, CS, EC, EE and ME examinations are conducted in multiple sessions. Suitable normalization process is therefore applied to take into account any variation in the difficulty levels of the question sets across the different sessions. The normalization is done based on the fundamental assumption that "in all multi-session GATE papers, the distribution of abilities of candidates is the same across all the sessions".

According to the GATE committee, this assumption is justified since "the number of candidates appearing in multi-session subjects in GATE 2014 is large and the procedure of allocation of session to candidates is random. Further it is also ensured that for the same multi-session subject, the number of candidates allotted in each session is of the same order of magnitude."

So what`s the GATE formula

Based on the above the committee arrived at the following formula for calculating the normalized marks, for CE, CS, EC, EE and ME subjects. From GATE 2014 onward (and year 2014-15 of the 2-year validity period of GATE 2013 score), a candidate's GATE score is computed by the following new formula.

S= Sq + (St – Sq) M – Mq/Mt - Mq

where, S = Score (normalized) of a candidate; M = Marks obtained by a candidate ("normalized marks" in case of multiple-session subjects CE, CS, EC, EE and ME); Mq = Qualifying marks for general category candidates in that subject (usually 25 or μ + σ, whichever is higher); μ = Average (i.e. arithmetic mean) of marks of all candidates in that subject; σ = Standard deviation of marks of all candidates in that subject; Mt = Average marks of top 0.1 % candidates (for subjects with 10000 or more appeared candidates) or top 10 candidates (for subjects with less than 10000 appeared candidates); St = 900 = Score assigned to Mt; Sq = 350 = Score assigned to Mq.

Percentile:

A candidate's percentile denotes the percentage of candidates scoring lower than that particular candidate. It is calculated as: Percentile = ( 1 - All India rank( No. of candidates in that subject) x 100%

The normalization process for CAT 2014

On the similar note the formula to calculate the percentile in CAT 2015 can be devised. Experts with MBAUniverse.com have been working and have decoded the pattern of normalization and calculation of percentile in CAT 2014 as follows-

Normalized marks (Mij) of jth candidate in ith slot, is given by
Mij = Mgt - MgqMti - Miq ( Mij - Miq ) + Mgq
where,

Mij is the actual marks obtained by the jth candidate in the ith slot,
Mgt is the average marks of the top 0.1 % candidates in all slots,
Mgq is the sum of mean and standard deviation of marks of all candidates in all slots,
Mti is the average of marks of top 0.1 % candidates in the ith slot,
Miq is the sum of mean and standard deviation of marks of all candidates in the ith slot.

Understand with example

Total aspirants = 2,00,000
slots = 4
aspirants in each slot = 50,000 [ .1% of it = 50]

Mij = 202 out of 300 ( are the actual marks obtained by the jth candidate in the 3rd slot)
Mgt = [195 + 202 + 210 + 205 ]/4 = 203 ( is the average marks of the top 0.1 % candidates in all slots)
here, 195, 202, 210 and 205 are the average of the top .1% aspirants of all slots

Mgq = 125 (is the sum of mean and standard deviation of marks of all candidates in all slots)
Mti = 210 (is the average of marks of top 0.1 % candidates in the 3rd slot)
Miq = 128 (is the sum of mean and standard deviation of marks of all candidates in the 3rd slot.)

According to the given formula, normalized marks = (203-125)(202-128)/(210-128) + 125 = 195.4

Expect high percentile with good scores

The above score is supposed to fetch a percentile near about 99.5. On the similar note and going by the relative difficulty level across the sections in different slots of CAT 2014 examination, it is expected that CAT 2014 takers could score a higher percentile in CAT 2014 than they could score in earlier CAT exams.

In CAT 2014 sections were found more balanced and more options to attempt the questions were available. A candidate could crack 60 correct out of 100 in CAT 2014 but it was much difficult to crack 35 correct out of 60 in earlier CAT exams despite the fact that more time per question to solve was available in earlier CAT exams. Accordingly due to negative marking system the percentile got dipped low in earlier CAT exams. In CAT 2014 with more questions available most of the candidates who refrained from attempting doubtful and difficult type of questions can expect a high percentile.