Vedadots
All insights
GS Paper 1CSATStrategyQuestion Format

Negative marking — the exact decision rule for every question format

Most aspirants know the rule. Few apply it under pressure. Here is the precise expected-value calculation for every format — Statement-based, How-many counting, CSAT — so exam day is execution, not decision-making.

18 May 2026·4 min read·Vedadots Compass

The negative marking rule in UPSC Prelims is simple: every wrong answer costs one-third of the marks a correct answer earns. In GS Paper 1 — where each question is worth 2 marks — a wrong answer costs 0.67 marks. In CSAT — where each question is worth 2.5 marks — a wrong answer costs 0.83 marks.

Most aspirants know this. Far fewer apply it consistently under exam pressure. This article gives you the exact decision rule for every question format so you don't have to think about it on exam day — you just execute.

0.67marks lost per wrong answer in GS Paper 1 — the number that should govern every attempt decision

The core calculation

Before the format-by-format rules, understand the underlying logic:

Expected value of attempting = (probability of correct × +2) − (probability of wrong × 0.67)

An attempt has positive expected value when your probability of being correct exceeds 25% — i.e. when you can genuinely eliminate at least one option. Below 25%, skipping is the correct decision.

The problem is that under exam pressure, aspirants systematically overestimate their probability of being correct. The rules below are designed to correct for that overconfidence.


The decision rule for every format

Factual single — attempt aggressively

"Which of the following is/are correct?" with no statement list.

This is the most straightforward format. If you can eliminate even one option with confidence, your probability of being correct is at least 33% — above the threshold. Attempt it.

Rule: If you can eliminate one option → attempt. If you recognise nothing and genuinely cannot eliminate any option → skip.

Strategy note

Factual single questions skew easier than other formats. If you have studied the topic at all, you can almost always eliminate at least one option. The default for this format should be attempt, with the skip reserved for topics you have genuinely not covered.


Statement-based — attempt with partial knowledge

"Consider the following statements... Which of the above is/are correct?"

This is the most common format in GS Paper 1. The elimination logic here is powerful: if you can confidently evaluate even one or two statements, you can often eliminate two or three options, raising your probability well above 25%.

The protocol:

  1. Evaluate each statement independently — write T/F before looking at options
  2. Use your confirmed statements to eliminate options
  3. If you have eliminated two options → attempt (50% probability = strongly positive expected value)
  4. If you have eliminated one option → attempt (33% = still positive)
  5. If you cannot eliminate any option → skip
Statements you can confirmOptions eliminatedRemaining optionsExpected valueDecision
2 of 3 confidently2–3 options1–2+1.0 to +2.0Attempt
1 of 3 confidently1–2 options2–3+0.3 to +1.0Attempt
None confidently04−0.17Skip

How-many counting — the hardest call

"How many of the following statements are correct? (a) Only one (b) Only two (c) All three (d) None"

This is the format where elimination logic breaks down. Because every option (one, two, three, none) remains possible regardless of which individual statement you have confirmed, partial knowledge gives you no elimination shortcut. You must evaluate every statement.

The protocol:

  1. Evaluate each statement independently with a confidence level: High (certain) or Low (guessing)
  2. Count your High-confidence True statements
  3. If your High-confidence count matches one option, and you have genuine uncertainty only between two adjacent counts → attempt
  4. If you are uncertain across three or four options → skip
Your situationDecisionWhy
Certain of exact countAttemptYou know the answer
Uncertain between 2 adjacent counts (e.g. 1 or 2)Attempt50% probability = +0.67 expected value
Uncertain across 3–4 optionsSkip25% probability = −0.17 expected value
Cannot evaluate any statementSkipPure guess, negative expected value

Key insight

How-many counting is the format where confident aspirants lose the most marks. The question looks solvable — you recognise the topic, you have a feeling about the statements — but a wrong answer here costs the same 0.67 marks as any other wrong answer. If you cannot evaluate at least two statements with High confidence, skip.

Read the full How-many counting analysis → The hardest question format — and how to beat it


Assertion-Reason — attempt if you can evaluate both parts

Statement 1 (Assertion) and Statement 2 (Reason). Four options ask whether each is true and whether R correctly explains A.

The protocol:

  1. Evaluate A independently — true or false?
  2. Evaluate R independently — true or false?
  3. If both A and R are clearly true, assess whether R explains A
  4. If you are confident on both A and R → attempt
  5. If you are uncertain about either → the four-option structure makes elimination difficult → skip unless you are confident on at least one

Rule: Attempt if you are confident about both statements. Skip if uncertain about either — the four-option structure does not give you easy elimination.


Matching pairs — use confirmed pairs to constrain options

Match items in Column A with Column B.

The protocol:

  1. Fill in the pairs you are certain about first
  2. Use confirmed pairs to eliminate options — if you know A matches 2, any option that shows A matching 1, 3, or 4 is eliminated
  3. If you can confirm two pairs → typically eliminates 2–3 options → attempt
  4. If you can confirm zero pairs → skip

CSAT — the separate calculation

CSAT questions are worth 2.5 marks each. A wrong answer costs 0.83 marks. The qualifying threshold is 33% — approximately 67 marks out of 200.

The critical mistake aspirants make is optimising for attempting as many questions as possible. Accuracy beats volume. Here is the arithmetic:

StrategyAttemptedCorrectWrongScoreResult
High accuracy5042898.4✓ Qualifies comfortably
High volume70422881.7✓ Qualifies, but barely
High volume, lower accuracy70383268.6✗ At risk
Overconfident guessing70343661.8✗ Fails

The difference between the "qualifies comfortably" and "fails" scenarios is not the number of questions attempted — it is the number of wrong answers. The aspirant who attempts 70 questions with 34 correct is likely guessing on Hard questions they cannot solve. The aspirant who attempts 50 with 42 correct is not attempting the Hard questions they cannot solve.

Strategy note

Before entering a Hard CSAT section (Numerical Ability, DI), do a quick calculation: how many marks do I need from this section to be safe, given my score so far? If your Reading Comprehension and Reasoning scores already put you above 80 marks, the risk-reward of Hard Numerical questions may not be worth it. A blank costs nothing. A wrong Hard Numerical question costs 0.83 marks.


The one rule to remember on exam day

If the exam pressure is high and you cannot run the full calculation, use this single heuristic:

Two options left → attempt. Four options left → skip.

It is not perfectly calibrated, but it is directionally correct for every format and will prevent the worst outcomes — the confident wrong answers on questions you were essentially guessing on.

Key insight

Print or screenshot this rule and read it the morning of the exam. "Two options left → attempt. Four options left → skip." Simple enough to remember under pressure. Calibrated enough to protect your score.