How our CAT questions are made

Start from what the CAT actually asks

We study recent CAT papers and write a note, in our own words, for each kind of question: what it gives, what it asks, the idea that solves it, what makes it hard and the usual traps.

No text is copied from a past paper. Every reading passage is new.

Write it harder than the source

Each question is written from one of those notes, to a written guide for its section, and pushed harder than the original. It comes with a full worked solution.

Check it by script

A script checks the format, the options and the marking. For a DILR set or a countable QA question, a program works through every possible case and proves the keyed answer is the only one.

Solve it blind

Five independent solvers work the question without seeing the answer key. Their answers must agree with the key. If they do not, the question is repaired or dropped.

Attack it

A reviewer tries to defend every wrong option. If a wrong option can be argued for, or the right one can be argued against, the question fails.

Hold it to the hardness floor

A weaker solver is run on every question, and the floor counts its steps. A QA question is rejected if the weaker solver reaches the answer in three steps or fewer. A verbal question is rejected if it gets there in two steps or fewer. A reading question is rejected if it can be answered without the passage. Every question in every paper is held to this floor.

Compare it with past papers

Every question is compared with past CAT papers. Anything too close to one is rejected.

A question that fails a check gets at most one repair. If it fails again, it is dropped. One too close to a past paper is dropped at once.