How CATin's questions are checked

Where the questions come from

80+ CAT mocks in all, and 3,000+ questions to practise, each with a worked solution.

How the answers were checked

Official CAT papers

• Every answer key was checked by code against CAT-level data sets that we have used, reviewed and put up.
• Where a data set disagreed with our key, the key was changed only when two of them agreed on the answer. One alone was never enough.
• Some VARC keys were corrected after their explanation was found to argue for a different option than the key. Those were also changed only when two data sets agreed.

CATin mocks and sectionals

Written to the CAT pattern and reviewed by the founder.

CAT Mixers

Built from official CAT questions, with the checks above.

No check catches everything. That is why the app has a report button on its questions, and why corrections are published below.

Report a problem with a question

On a question in your mock results, practice or review, tap Report a problem with this question. Pick what's wrong:

• The answer key looks wrong
• The question is unclear
• The explanation doesn't convince me
• Typo or formatting problem
• Something else

Add a note if you like (up to 500 characters): the option you think is right, and why, helps most. Reports go to the founder, who checks each one against our reviewed data sets. If a key, a question or an explanation changes, it is listed under Corrections.

Not signed in, or prefer email? Write to rd@thrdream.com with the paper and question number.

Corrections

No corrections since launch on 28 Sep 2026.

Each correction will be listed here, newest first, in this format:

[date, e.g. 3 Oct 2026] · [paper] · [section and question number] What changed: [the answer key, the question or the explanation], from [before] to [after] How it was found: [a student report, or our own check]

How the percentile estimate is built

After a mock, CATin shows a percentile. There are two kinds, and the screen always says which one you are looking at.

Measured, once 30 students have taken a paper

When at least 30 students have sat a paper on CATin, your percentile is the share of their first attempts that scored below yours. Only first attempts count, so a retake can't inflate the field.

Estimated, until then

Before that, it is an estimate band (for example "93.7–98.1, estimated · no cohort yet"), read from CAT-level score-to-percentile data sets that we have used, reviewed and put up:

• What they are: CAT-level data sets that we have used, reviewed and put up. They are not official: the IIMs do not publish the mapping.
• What was left out: exam-week predictions, duplicates (counted once), and data with garbled or misaligned columns.
• How the curve is made: at each percentile level, the middle value across the data sets, then across years, and a level is used only when at least 3 years supply it, so one hard or easy year can't pull it alone.
• Why it's a range: the band is wide enough that every data set's value falls inside it. On a full paper that is about ±9.5 marks out of 198. Years really do differ: CAT 2024 needed 20–27% more marks than CAT 2023 for the same percentile. The 2017–2019 papers rest on fewer data sets, so their band is wider.