RC — Inference & Assumption Questions

Inference questions are the highest-difficulty, highest-value RC type in CAT — typically 1–2 per passage. They separate 95+ percentilers because the answer is never stated; it must follow from the text.

Inference vs assumption

• Inference = what must be true given the passage (a valid deduction, one small step beyond the lines).
• Assumption = an unstated premise the author needs for their argument to hold.

Both are invisible in the text — but both are tightly constrained by it.

The "must be true" rule

A correct inference is guaranteed by the passage, not merely plausible. If you can imagine the passage true and the option false at the same time, it is wrong.

Option smells like...	Verdict
A restatement of a stated line	Wrong (that's a detail, not an inference)
A large logical leap	Wrong (needs outside info)
Extreme words: all, never, only, must	Usually wrong (too strong)
A cautious, hedged deduction	Usually right

Method

1. Locate the exact lines the question points to.
2. Ask "what small step is forced by these lines?"
3. Test extremes — reject options stronger than the text warrants.
4. For assumptions, apply the negation test: negate the option; if the argument collapses, it was a required assumption.

Exam Tricks & Tips

• 🎯 Weak is strong. Hedged options ("may", "some", "tends to") survive scrutiny; absolute options ("always", "cannot") usually die.
• 🎯 Negation test for assumptions: negate the choice — if the author's conclusion no longer holds, that choice is the assumption.
• 🎯 One step only. A valid inference is one deduction away; two-step reasoning imports outside assumptions.
• 🎯 Stay inside the passage's world. Never use real-world knowledge the passage didn't give you.
• 🎯 A restated fact is a decoy. If an option merely paraphrases a printed line, it is a detail question's answer, not an inference.
• ❌ Common mistake: choosing the most interesting or sophisticated statement instead of the one the text strictly forces.

Expected exam pattern

"It can be inferred from the passage that...", "The author would most likely agree that...", "Which assumption underlies the argument?". MCQ, −1. Distractors are plausible-but-unproven or too-extreme. 1–2 per passage.

Quick recap

Pick what the passage forces, not what it suggests. Favour hedged wording, take exactly one logical step, and stay inside the text. For assumptions, use the negation test: kill the premise and watch the argument fall.

Cover the answer, recall, then check. 12 cards on inference and assumption questions.

Q1. What defines a correct inference?
A1. It must be true given the passage — a valid deduction, not merely plausible.

Q2. What is an assumption?
A2. An unstated premise the author needs for the argument to hold.

Q3. State the "must be true" rule.
A3. If the passage can be true while the option is false, the option is wrong.

Q4. Why do extreme words usually mark a wrong inference?
A4. "All, never, only, must" claim more certainty than a passage normally warrants.

Q5. What is the negation test?
A5. Negate an option; if the author's conclusion collapses, that option is a required assumption.

Q6. How many logical steps should a valid inference take?
A6. Exactly one — two steps smuggle in outside assumptions.

Q7. Is a restatement of a printed line a good inference?
A7. No — that is a detail, not an inference; it's a decoy.

Q8. "Weak is strong" — meaning?
A8. Hedged options (may, some, tends to) survive; absolute options usually fail.

Q9. Can you use real-world knowledge to answer an inference question?
A9. No — stay strictly inside the passage's world.

Q10. Typical phrasing of an inference question?
A10. "It can be inferred that...", "The author would most likely agree that...".

Q11. How many inference questions appear per passage?
A11. Usually 1–2 — the highest-value, highest-difficulty type.

Q12. The classic trap in inference options?
A12. The most interesting/sophisticated statement, rather than the one strictly forced by the text.

Inference questions are where careful readers over-think and over-reach. "Which of the following can be inferred?" does not mean "what is the cleverest conclusion I can draw?" On the CAT, a valid inference is a small, safe step the author is logically committed to — not a bold leap.

What this tests

Your discipline in staying just one logical inch beyond the text — reading what the author must accept, without importing outside knowledge or dramatic extrapolation.

The method

Beginner — inference = "must be true", not "could be true"

An inference is something that is necessarily true given the passage. If you can imagine the author's world where the option is false, it is not a valid inference. Ask: does the passage force this? If it merely allows it, reject it.

Intermediate — assumptions are the unstated bridge

An assumption question ("the argument assumes that…") asks for the missing premise the author needs for the conclusion to hold. Find the gap between the stated evidence and the conclusion; the assumption is the plank that bridges it. Test with negation: if negating the option destroys the argument, it is the assumption.

Advanced — the FAST CAT approach: the "safe step" filter

Rank every option on a spectrum from restated to reckless:

• Too safe — merely paraphrases a stated fact (that's a detail, not an inference — but on CAT, "closest to stated" often wins over "boldest").
• Just right — a modest, unavoidable consequence.
• Too far — adds cause, prediction, quantity, or value the text never supports.

CAT inference answers hug the text. When two options both seem inferable, pick the weaker, more cautious one. Words like may, some, tends to, is likely survive; will, always, proves, guarantees usually don't.

Worked example

Passage line: "While the vaccine reduced infection rates in trials, its cost has kept it out of reach in low-income regions."

Question: It can be inferred that…
(A) The vaccine is ineffective. → contradicts "reduced infection rates". Out.
(B) Cost is the only barrier to distribution. → "only" over-reaches; the passage names one barrier, not the sole one. Out.
(C) Price is a factor limiting the vaccine's reach in poorer areas. → a modest restatement of the causal link the author states. Correct.
(D) Lowering the price would eliminate the disease. → wild prediction; unsupported. Out.

(C) takes the one safe step; the rest leap.

CAT relevance

Inference and assumption questions are the majority of hard RC items and the primary way CAT separates the 95th from the 99th percentile. They also power the Verbal Ability sub-section (Critical Reasoning). Mastering the "safe step" reflex pays across the whole section.

Speed tricks & shortcuts

• Eliminate on extremeness first. Scan for only, all, never, prove, will — these kill most wrong inference options in seconds.
• For assumptions, use the negation test: negate the option; if the argument collapses, it's the answer.
• Mnemonic — "One inch, not one mile." Valid inferences are short, cautious steps.
• If an option requires knowledge from outside the passage, it's wrong by definition.

Worked Example

Problem: Draw a valid inference — something that MUST be true given the text, not merely plausible.

The museum's new wing was funded entirely by a single donor, on the condition that no admission fee ever be charged for it. Attendance in that wing has tripled since it opened, even as the paid main galleries have seen visitor numbers fall.

Which can be inferred?
(A) The free wing has better exhibits than the main galleries.
(B) Removing admission fees always increases attendance.
(C) At least some visitors' decision to enter the free wing is influenced by factors other than the exhibits in the paid galleries.
(D) The donor is wealthy.

Solution:
An inference must be logically supported by the passage — true in every scenario consistent with the text. It is a small, safe step, NOT the most interesting or likely-sounding claim. Distinguish it from an assumption (an unstated premise the argument needs).

Test each option:
(A) "Better exhibits" — the passage never compares exhibit quality; the attendance gap could be due to the free price, novelty, or location. Not supported → too far.
(B) "Always" — a universal claim from one museum's single case. RC inferences are almost never sweeping universals; one example cannot prove "always". Overreach.
(C) The two wings moved in OPPOSITE directions (free tripled, paid fell). Whatever is driving the free wing's rise, it is not simply "people wanting the paid galleries", since those declined. So some visitors are influenced by factors beyond the paid galleries' exhibits — e.g. the free price or the new wing's novelty. This is a cautious, must-be-true reading of the divergence. Correct.
(D) "Wealthy" — funding one wing suggests means, but the passage gives no financial detail and "wealthy" is an outside assumption, not a forced conclusion. Not strictly entailed.

The right inference (C) hugs the text and only states what the divergent attendance numbers guarantee; the wrong ones add unstated comparisons (A, D) or overgeneralise (B).

Answer: (C). Because the free wing's attendance rose while the paid galleries' fell, some visitors are driven by factors other than the paid exhibits (e.g. the free price/novelty) — the only claim the text forces. (A) and (D) add unsupported outside information; (B) overgeneralises from one case.