BharatNotes Each question is 2 marks; wrong answer deducts 0.667 marks (1/3 of 2). Attempt if you can eliminate 2 options — expected value turns positive.
Marking scheme (GS Paper I): 100 questions × 2 marks = 200 marks. A wrong answer costs 1/3 of the question's marks, i.e., 0.667 marks. Unattempted questions score 0.
The Expected Value (EV) Framework
Expected value is the single most powerful concept for deciding whether to attempt a doubtful question. It tells you the average marks you will gain (or lose) per question over many such decisions.
EV calculations by elimination level:
| Options eliminated | Options remaining | Probability of correct | EV per question |
|---|---|---|---|
| 3 (know the answer) | 1 | 100% | +2.00 |
| 2 (50/50 toss-up) | 2 | 50% | +0.67 |
| 1 (3 options left) | 3 | 33% | +0.22 |
| 0 (pure random guess) | 4 | 25% | 0.00 |
Derivation (1-elimination case): EV = (1/3 × 2) + (2/3 × −0.667) = 0.667 − 0.444 = +0.22 → worth attempting.
Decision rule: Attempt if you can eliminate at least one option with genuine reasoning. Skip only when all four options are equally unknown and you have zero basis for elimination.
Why This Matters More Than Most Aspirants Think
Consider a common scenario: you skip 25 questions where you could have eliminated one option (3 left, EV = +0.22). Your expected cost of skipping = 25 × 0.22 = 5.5 marks forfeited.
In a year where the General category cut-off is 87.98 (2024) and you scored 83, those 5.5 marks would have cleared you. This is not a hypothetical — it describes the situation of thousands of aspirants who fell just below the cut-off.
Worked Numerical Examples
Scenario A — Attempting 80 questions at 70% accuracy:
- Correct: 56 × 2 = 112 marks
- Wrong: 24 × 0.667 = −16 marks
- Net score: 96 marks → clears General (87.98) and OBC (80.00) comfortably
Scenario B — Attempting 70 questions at 75% accuracy (more conservative):
- Correct: 52.5 × 2 = 105 marks
- Wrong: 17.5 × 0.667 = −11.7 marks
- Net score: 93.3 marks → clears General but tighter
Scenario C — Overly aggressive: 95 questions at 60% accuracy:
- Correct: 57 × 2 = 114 marks
- Wrong: 38 × 0.667 = −25.3 marks
- Net score: 88.7 marks → barely clears; risky strategy
Lesson: The sweet spot is 80–90 attempts at 65–75% accuracy. Pure aggression at low accuracy hurts; pure conservatism at high accuracy leaves marks on the table.
CSAT (Paper II) — Same Rules Apply
CSSAT is qualifying only (33% = 66/200 minimum). Same 2-mark scheme, same 0.667 negative marking. However, since clearing CSAT is binary (pass/fail), the risk calculus changes: if you are at 62–65 marks, attempt anything you can partially reason — a correct answer means you qualify, a wrong one still keeps you in range with one more correct answer elsewhere.
Question-Type Specific Rules
Statement-based questions (40–50% of paper): EV logic is dramatically easier — one known false statement eliminates multiple options simultaneously. A single anchor ("Statement 2 is definitely false") can take you from 4-option random (EV = 0) to 2-option 50/50 (EV = +0.67).
Absolute-language traps: UPSC frequently embeds statements with words like "always", "never", "only", "all", "must". In most cases, these absolute statements are false. If you can identify one such statement as likely false, you gain elimination leverage.
Assertion-Reasoning questions (~13% of paper): Four answer choices involve whether A is true, R is true, and whether R is the correct explanation of A. If you know A is false, options (a) and (b) are eliminated immediately — 50/50 between (c) and (d), EV = +0.67.
Common Mistakes
- Over-skipping: Many aspirants skip 20–30 questions they could have partially reasoned. At +0.22 EV per question with 1-option elimination, skipping 20 such questions costs ~4.5 marks — often the difference between clearing and not.
- Under-skipping: Attempting pure guesses (all 4 options equally unknown) adds zero expected value and increases variance. In a volatile year, this can hurt.
- Failing to apply EV to CSAT: Many candidates treat CSAT as "just qualifying" and don't apply the same disciplined skip/attempt logic — then miss 66 by 2 marks.
Topper Perspective
Shakti Dubey (AIR 1, CSE 2024) practiced CSAT PYQs from 2013–2023 to identify patterns and focused on time management, solving 10 mock papers under timed conditions. Her approach to negative marking was rooted in pattern recognition, not random guessing — she attempted doubtful questions only when structural elimination (from prior PYQ pattern exposure) gave her a clear edge.
With 8 Days to CSE 2026 Prelims (24 May 2026)
Now is not the time to change your gut feel about skip/attempt thresholds. Stick with what your mock test data shows. If your last 5 mocks show that your accuracy on "1-option-eliminated" questions is above 40%, always attempt them. If your accuracy is below 33%, skip — the EV has turned negative for you personally.
