BharatNotes Generally no — first instinct is more accurate. Change only if you recall a specific fact that contradicts your first answer, not based on vague unease.
The short answer: In approximately 80% of cases, sticking with your first answer is correct. Change only when you have a specific, articulable factual reason — not a feeling.
The Psychology of Answer-Changing
Research in test-taking psychology (including studies published in journals such as Teaching of Psychology and Applied Cognitive Psychology) consistently shows that first instincts are more accurate than changed answers in multiple-choice examinations. Across multiple studies:
- 50–75% of answer changes go from wrong to right
- 20–40% of answer changes go from right to wrong
- 5–15% are wrong-to-wrong
The net effect depends heavily on the aspirant's preparation depth and metacognitive accuracy. For UPSC aspirants with 6+ months of serious preparation, the pattern tilts toward first answers being more reliable — because the first answer draws on deep schema and pattern recognition built over months, while the second answer often draws on exam-day anxiety and confirmation bias.
Why First Instincts Are More Reliable for UPSC Specifically
Schema activation: When you read a question and immediately recognise it ("This is about the 73rd Amendment and Panchayati Raj — Article 243"), your first answer is driven by genuine recall. The second look introduces doubt — not because your memory is wrong, but because the unfamiliarity of the exam environment makes everything feel uncertain.
Misread correction (the legitimate exception): If your first answer was driven by misreading the question stem, a second look will catch the error. This is not changing your answer — it is correcting a reading mistake. UPSC frequently uses "NOT correct", "EXCEPT", and "incorrect" — a misread here is a genuine mechanical error worth correcting.
The anxiety amplification effect: In the exam hall, particularly after 90+ minutes of cognitive load, your critical faculty begins to doubt itself. Questions that seemed clear in Round 1 look ambiguous in Round 3. This apparent ambiguity is almost always a function of fatigue and stress, not genuine new information. Trust Round 1.
Precise Rules — When to Change vs. When to Hold
Change your answer if and only if:
| Situation | Reason to change | Example |
|---|---|---|
| Recalled a specific fact that contradicts your first answer | Genuine new information from memory | "Wait — the 44th Amendment, not the 42nd, restored the right to property as a legal right" |
| Misread the question stem (NOT vs. IS) | Mechanical reading error | You answered "which IS correct" but the question asks "which is NOT correct" |
| Found a calculation error | Arithmetic correction | Spotted a wrong step in an Economy or CSAT Numeracy question |
| Accidentally filled the wrong OMR bubble | Mechanical OMR error | Filled B when you meant C — this is correction, not change |
Do NOT change your answer if:
| Situation | Why to hold |
|---|---|
| Vague unease about the answer without a specific reason | Anxiety is not information |
| The question "seems harder" on second reading | Fatigue effect — not genuine ambiguity |
| A peer mentioned a different answer during Paper I/Paper II break | External noise; ignore completely |
| Another question seems to provide a clue | UPSC papers are designed to be self-contained; cross-question clues are rare and unreliable |
| You "recall something" vaguely without being certain | A vague recollection that contradicts your first answer is more likely wrong than right |
| The option you now favour is longer or sounds more official | Answer-length bias is a known test-taking trap |
OMR-Specific Protocol
Distinguish between two categories:
Category A — Genuine answer change: You are changing which option (A/B/C/D) you believe is correct. Apply the rules above — almost always hold.
Category B — OMR correction: You filled bubble B on the OMR but intended bubble C. This is a mechanical correction, not an answer change. Correct it immediately and verify the correction is legible (no stray marks).
Note: Overly heavy erasing on the OMR can damage the sheet. Use a clean, light eraser and verify the old mark is fully removed and the new mark is crisp.
Building Personal Data on Answer-Changing
This is the most valuable thing you can do in mock tests: track every answer you change and record the outcome.
After 10 mocks, tally:
- Total changes: N
- Changes: right → wrong = X
- Changes: wrong → right = Y
- Net effect: (Y−X) × 2 marks (since each Q is 2 marks, wrong answers −0.667)
Most aspirants discover their changes are net-neutral or net-negative. A small minority find their changes are net-positive — typically those who are genuinely good at catching genuine recall moments versus anxiety moments. Know your personal data before exam day.
If your mock data shows net-negative changes: Commit to a strict "no changes" rule for Round 3, except for stem misreads and OMR mechanical corrections.
If your mock data shows net-positive changes: Allow yourself changes, but only when you can write (on rough paper) the specific fact that drove the change.
The Round 3 Mindset
The primary purpose of Round 3 is OMR hygiene and catching stem misreads — not re-evaluating strategy on content. Enter Round 3 with this mindset:
- Verify OMR is complete and accurate (every attempted question has a filled bubble)
- Re-read question stems of 5–10 questions where you were slightly uncertain — only to catch NOT/EXCEPT misreads
- Resist the pull to reconsider content — the cognitive bandwidth at 90+ minutes of examination is degraded
With 8 Days to CSE 2026 Prelims (24 May 2026)
In your next mock (if you do one), consciously track every answer you change. Note the reason for the change and the outcome. Use this data to set your personal answer-changing rule for 24 May. Do not leave this policy undefined — a clear pre-committed rule outperforms in-the-moment decisions made under exam pressure.
