Prelims arithmetic: guessing, elimination and risk

Two aspirants finish GS Paper I knowing the correct answers to exactly the same 55 questions. One scores 104, the other 92. Neither was lucky or unlucky in any unusual way. One simply understood the arithmetic of the paper better.

Prelims is a qualifying stage with a moving cut-off, and it is decided by margins of a few marks. So it is worth spending an hour understanding how negative marking actually works — not as a trick, but as a way of making sensible decisions under uncertainty.

The rules

In GS Paper I:

• there are 100 questions, each worth 2 marks;
• each wrong answer costs one-third of 2 marks, that is 0.66;
• a question you leave blank costs nothing.

(CSAT follows the same one-third rule, with 2.5 marks per question and a deduction of 0.83 per wrong answer.)

Expected value, in plain language

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

For a guess on a GS Paper I question:

Expected value = (chance of being right × 2) − (chance of being wrong × 0.66)

Now apply it to the situations you actually face in the hall.

Two things follow immediately:

1. A pure blind guess gains nothing on average. It does not lose marks on average either — but it adds randomness.
2. Every option you can genuinely eliminate makes attempting worthwhile. Even eliminating one option turns a zero-value guess into a positive one. Eliminating two makes it clearly profitable.

Why "real confidence" matters

The table assumes your eliminations are correct. They often are not. If you eliminate an option that was actually the right answer, your "fifty-fifty" was really zero-in-two.

This is where many aspirants lose marks: they eliminate on vague feelings, then treat the question as a safe fifty-fifty. Honest elimination means you can give a reason for rejecting an option — a fact that contradicts it, a logical inconsistency, or a wording that makes it impossible.

Elimination techniques that actually work

Extreme words. Statements containing "only", "all", "always", "never" or "exclusively" are more often wrong than statements with qualifiers such as "may", "generally" or "some". This is a tendency, not a rule — UPSC knows the trick too — so use it to support a decision, not to make one.

The statement-count format. Many questions list three or four statements and ask "how many of the above are correct?" with options "only one", "only two", "all three" and "none". These are harder to guess because you cannot eliminate by combination. Your chance of being right depends entirely on judging each statement. Treat them carefully; they are not natural guess candidates unless you are sure of most statements.

The combination format. Where options are combinations ("1 and 2 only", "2 and 3 only"…), one statement you are certain about can eliminate two or three options at once. Always start with the statement you are surest of.

Internal consistency. Sometimes two statements cannot both be true. Spotting that removes combinations containing both.

Current-affairs anchoring. Where a question is about something in recent news, recall why it was in the news. The reason often points to the right option.

How many questions should you attempt?

There is no universal number, but the arithmetic tells you how to think about it.

Suppose you are confident about 50 questions and get 46 of them right. That is 92 − 2.64 = 89.4 marks. Now suppose there are 30 more questions where you can eliminate two options. Attempting all of them, you would expect about 15 right and 15 wrong: 30 − 10 = +20 marks on average. That takes you to roughly 109 — often the difference between clearing and not.

On the other hand, attempting 20 further questions as blind guesses adds nothing on average, but widens the range of your possible score — you might gain 8 marks, you might lose 8.

This is why toppers' attempt numbers vary so widely. The number is not the strategy. The quality of elimination is the strategy.

A three-round approach to the paper

Many candidates find a structured pass through the paper keeps their decisions calm:

1. Round one (about 50 minutes): Answer every question you are sure of. Mark every question where you can eliminate at least one option. Skip the rest without a second look.
2. Round two (about 45 minutes): Return to the marked questions. Work through elimination properly. Attempt those where you are down to two options, and those with three left where you have a real lean.
3. Round three (remaining time): Look again at the questions you skipped. Sometimes a later question jogs a memory. Fill the OMR sheet carefully — do not leave bubbling to the last five minutes.

Practise the decisions, not just the content

Mock tests are the place to calibrate. After each test, sort your attempts into three groups — sure, two options left, guessed — and check your accuracy in each. Many aspirants discover their "sure" answers are only 85 per cent right, or that their fifty-fifty calls are right far less than half the time because their eliminations are faulty.

Once you know your own numbers, you can decide how aggressive to be with far more confidence than any rule of thumb can give.

The last word

Negative marking punishes carelessness, not courage. Attempt where the arithmetic is on your side, skip where it is not, and let your mock tests show you which is which. Knowledge gets you most of the way to the cut-off; disciplined decisions carry you over it.