Negative marking in CLAT

CLAT's negative marking is milder than in many exams, and that tempts some candidates to attempt everything. But a quarter mark lost on each wrong answer adds up quickly across 120 questions, and hurried attempts usually come with lower accuracy.

The good news is that the maths is simple. Once you understand it, you will know exactly when a guess makes sense, and you will stop worrying about it in the hall.

The rules

Each question has four options. Four wrong answers cancel exactly one right answer, since 4 × 0.25 = 1.

Expected value, in plain language

"Expected value" means: if you faced the same situation many times and answered the same way, what would you gain on average per question?

Apply it to the situations you actually face:

Even a pure guess is very slightly positive on average, but the gain is tiny and adds a lot of randomness. Once you have eliminated even one option honestly, attempting is clearly worth it.

The break-even point

When does a guess stop paying? Set the expected value to zero:

1. Let p be your chance of being right.
2. Expected value = p × 1 − (1 − p) × 0.25.
3. Set it to zero: p − 0.25 + 0.25p = 0, so 1.25p = 0.25.
4. p = 0.25 ÷ 1.25 = 0.2, or 20%.

So in marks alone, attempting is worthwhile whenever your chance of being right is above 20%. With four options, even a blind guess is at 25%. That is why the marks maths almost always favours attempting.

So why not attempt everything?

Because marks are not the only cost. Time is.

A guess itself takes seconds. The problem is the time you spend before you guess: reading a passage you do not understand, wrestling with a hard question, then guessing anyway. That time is taken from questions you could have answered properly.

There is a second cost too. The table assumes your eliminations are right. If you "eliminate" the correct option, your fifty-fifty was really zero. Candidates who guess a lot usually eliminate on vague feelings, and their real accuracy on guesses is lower than the table suggests.

Accuracy beats attempts

(Made-up figures for practice.)

Candidate A attempted the most and scored the least. Candidate B attempted fifteen fewer questions but read carefully and scored 12.5 marks more. Candidate C shows that attempting too few can also cost you: similar accuracy to B, but ten fewer attempts and 5 marks lower.

The lesson is not "attempt less". It is "attempt what you have actually read and reasoned through".

A quick way to see your accuracy's value

Each attempt at accuracy a is worth, on average, a − (1 − a) × 0.25 = 1.25a − 0.25 marks.

• At 80% accuracy: 1.25 × 0.8 − 0.25 = 0.75 marks per attempt.
• At 60% accuracy: 1.25 × 0.6 − 0.25 = 0.50 marks per attempt.

Raising accuracy from 60% to 80% on 100 attempts is worth 25 marks. Few strategies give that much.

A practical rule for the hall

1. Answer when you are confident.
2. Attempt when you can honestly eliminate at least one option, and you have a reason you could say aloud.
3. Skip questions on a passage you have not really read. Spend that time on a passage you can reason through.
4. Return to marked questions only if you have buffer time at the end.

Calibrate yourself in mocks

In your next three mocks, write a small S, T or G beside each attempted question: Sure, Two options left, or Guess. Afterwards, check your accuracy in each group.

• If your "sure" answers are below about 85% right, you are misreading. Slow down.
• If your "two left" answers are right well under half the time, your eliminations are faulty. Practise giving a reason for each elimination.
• If your "guesses" are right about a quarter of the time, they are truly random. Consider skipping more of them and using the time elsewhere.

That table is your personal negative-marking strategy. It is more reliable than any general rule.

Practice

1. A candidate gets 88 right and 20 wrong. What is the score?
2. What is the expected value of a guess when two options have been eliminated?
3. A candidate's accuracy is 70% on 110 attempts. About what score should they expect?
4. At what chance of being right does a guess stop being profitable in marks?

Answers:

1. 88 − (20 × 0.25) = 88 − 5 = 83.
2. (1 − 0.25) ÷ 2 = +0.375.
3. Per attempt: 1.25 × 0.7 − 0.25 = 0.875 − 0.25 = 0.625. Over 110 attempts: 110 × 0.625 = 68.75. (Check: 77 right and 33 wrong gives 77 − 8.25 = 68.75.)
4. Below 20%.

What to do next

• Use the S, T and G marking in your next three mocks and work out your accuracy in each group.
• Decide your personal rule for when to guess, and write it on the first page of your mock booklets.
• Plan your time so there is a buffer for marked questions, using the CLAT time strategy.
• Review your error log with the method in mock tests and analysis.