Alright, here's my CAT thread. I'll start with my credentials to pre-empt the "what do you know" pushback. Engineer-MBA, cleared CAT first attempt (lucky!), 760 GMAT, led IMS' GMAT package revamp, PhD in marketing, decent training in statistics and econometrics, tenured prof. https://twitter.com/gauravsabnis/status/1312739167164657664

CAT the exam is at least clearer on the skills it is testing for, unlike GD's which are just absurd and absurdist. Modeled loosely on the GMAT, it tests for math, logic, english grammar, and comprehension. Worldwide consensus on entrance criteria for MBAs.

Where the two diverge is that you can take GMAT any day of the year. And re-take it if you think you had a bad day. Its questions are randomly chosen from a large question bank (my biggest contribution to IMS' GMAT reboot 15 years ago). Questions which are tested first.

Like any standardized test in the US, GMAT includes several questions that DON'T count towards your score. They are there just for testing the difficulty level of the question for the general population. They may eventually become questions that count towards the score.

So you look at the distribution of people's responses from when it was a "test" question, and based on that, decide the "difficulty level". GMAT is an adaptive test. The testing is how they calibrate the inputs for that adaptive algorithm.

Two people sitting next to each other might not get a single common question on a GMAT. What they will get though, are questions from the same large bank, depending on how they were answered in the test phase and how the candidate is doing on the test.

It's not a flawless test by any means and still has some inbuilt inequalities, but at least it's based on a scientific data-driven approach to testing. And does whatever it can to take away the chance element. And gives you opportunities to retake it whenever you want.

CAT ostensibly tests the same skills, but in a way that seems almost designed to amplify the chance element and inequalities. It ends up more like a budget version of the Math Olympiad. You get only one shot a year. That makes CAT "percentile" fundamentally different.

A GMAT (or GRE etc) percentile places you on a distribution of all people who have taken the test in recent years at any time. Statistically, that's a decent measure of the "expected value" of your score. Also why retaking GMAT hardly ever leads to more than a 5% bump.

A CAT percentile is based on just one unique test each year. Statistically speaking, your percentile there is not an estimate based on a population distribution. But just one sample taken in November/December. Which means the chance element is EXTREMELY high.

So I was very lucky in clearing CAT at first attempt. It was not all luck. I prepared a lot and had societal and systemic advantages (more on that later). But luck played a big role. I had a good day, as did some 2K others who made it to IIMs that year.

To give a sports analogy, players are selected not based on their best match/game but their stats over a season or more. To give a cricket analogy, a CAT score is like highest score. GMAT score is like average or career strike rate. Is Karun Nair a better batsman or Virat Kohli?

Then there is the exam itself. Unless it's changed recently, setting the exam was rotated among the IIMs. A group of professors creates the test each year. So they obviously bring in their own biases in terms of "difficulty level" there. Why CAT fluctuates wildly year to year.

There has also been a tendency to glamorize the "toughness" of tests (be it CAT or IITJEE). The more questions that go unanswered by almost everyone, the more pride said IIM takes. Treating it like an Olympiad rather than a reasonably stable measure of relevant skills.

So not only is your rank/percentile on CAT just a single data point on a largely untested set of questions, it is also extra vulnerable to societal and structural inequalities. Quant questions in CAT are often modeled on math Olympiad questions. Unnecessary and bizarre.

CAT is also skewed towards engineers and/or those with an intuitive grasp of math. I was lucky to be above average at math. Went through the math Olympiad circuit. And was an engineer. Parents and grandparents with advanced degrees to teach me. So I already had a head start.

CAT's one-shot panel-created unpredictable test results in a process that's a bit like Slumdog Millionaire. The difference is often made by a few "tough" questions that some can answer because they coincidentally came across something like those before.

The last question on GMAT for someone who scores a perfect 800 will be based on mountains of data that decide it is super tough. The CAT topper in a given year is decided by their ability to answer what some prof sitting in an IIM decided were tough questions.

I'm not saying that the 2000 people selected with me didn't "deserve" to get into IIMs. I'm saying that on another Sunday, a different set of 1500 deserving people might have gotten in, with maybe 500 staying the same (spitballing). I was lucky it was my day.

CAT (or JEE or UPSC) takes preparation, but still leaves a huge element to chance and amplifies flawed systemic hierarchies. And its flaws, by conscious or subconscious design, are also skewed towards helping the elite, affluent, typically savarna male engineer.

You can practice GMAT questions with focus for a few months and you'll get a score that is extremely unlikely to change. GMAT is also a very small component of admission decisions at US schools compared to how much weightage the CAT test has in Indian business schools.

Most US schools will have an approximate lower cutoff. For example, a school admissions committee might decide, first round, we only look at those who scored 700+. But that's it. A 760 isn't treated as higher than 740 or lower than 780. It's not a contest. It's an estimate.

So TLDR of CAT issues (with apologies to twitter)
- ironically negligible use of data, rigor, stats in setting a test supposed to be testing students on those very things
- Question setting very ad hoc and highly skewed
- exclusionary and elitist underlying mentality

That last point is a larger malaise in indian society. Gate-keeping higher education in line with archaic brahminical norms. Why US has dozens of reputed business schools, but in 4x population india, adding more than 6 IIMs was treated like some "brand dilution" apocalypse.

The best example of CAT exam amplifying societal inequalities is a lot of Number Theory questions. It's a fascinating branch of Math I only learned in Math Olympiad classes, which I had the privilege to attend. Relevance to MBAs? Negligible. Just gate-keeping.

This
What they think are proud features of CAT are actually flaws. In this bizarre quest to not be "predictable", it is more a contest than a rigorous and scientifically designed test of abilities and fit. Acting like quizmasters not professors.

Give me predictable over capricious any day when it comes to tests that measure acceptable levels of skill for getting an education in a public institution. But it took me a few years in US academia and seeing the relatively more rigorous approach to admissions to get that point.