Rank 10 by frequency | 90 questions in corpus (3.6% of all questions)
EXCEPT / NOT / LEAST is not a standalone question type — it is a modifier layered on top of another RC question type. The stem looks like a standard Detail, Inference, Attitude, or Application prompt, but the keyword flips the selection logic: four of the five choices are correct per the passage, and the one "odd one out" is what you're after.
A reversal question asks you to find the one answer that is NOT supported by the passage. Four of the five choices are correct (according to the passage), and your job is to identify the one that stands apart — the one that contradicts the passage, isn't mentioned, isn't inferable, or (in the LEAST variant) is least compatible with the passage.
The skills tested are thoroughness and precision. A reversal question punishes skimming: because four choices are defensible, you cannot eliminate based on gut feel and pick the "best" one. You have to check each choice against the passage, confirm the four that are supported, and isolate the one that isn't. There is no shortcut.
Because the underlying skill can be detail recall, inference, attitude, application, or almost any RC task, this is really a modifier rather than a distinct type. The reversal just inverts the selection logic; the substantive work is determined by the base question underneath.
The single most important thing about reversal questions is to notice the keyword. EXCEPT, NOT, and LEAST are almost always capitalized precisely because missing them is the #1 failure mode. Under time pressure, test-takers read the stem, find the first answer that looks supported, and pick it — forgetting that they're supposed to be looking for the unsupported one. That instantly turns a defensible choice into a wrong answer.
Make this a mechanical habit: every time you read an RC stem, scan for capital-letter keywords first. When you spot EXCEPT, NOT, or LEAST, mentally rephrase the question in non-reversal form ("what does the passage say about X?"), then remember that you're looking for the choice that fails that test, not the one that passes it.
A useful frame: think of the four supported choices as the "passage's team" and the one unsupported choice as the "outsider." The outsider is what wins.
Three keywords drive the subtypes, each with slightly different stakes.
Variation A — EXCEPT (74 questions, 82%). The dominant variant by a wide margin. Four choices are supported; one is not. "According to the passage, all of the following are true EXCEPT:" or "The author mentions each of the following EXCEPT:" EXCEPT typically appears at the end of the stem in capital letters — a hard stop that signals the reversal. The underlying task is usually detail recall, though inference-based EXCEPTs also appear.
Variation B — NOT (7 questions, 8%). Uses "NOT" instead of "EXCEPT." Functionally similar, but often embedded mid-stem, which makes it easier to miss. "Which one of the following is NOT mentioned in the passage as [X]?" or "The passage does NOT provide evidence that [X] exhibits which attitude?" More commonly paired with specific factual claims.
Variation C — LEAST (9 questions, 10%). The most nuanced variant — a matter of degree, not a binary yes/no. "The information provides the LEAST support for which claim?" or "Which one of the following, if true, is LEAST consistent with the hypothesis mentioned in [X]?" All five choices may have some degree of support; the correct answer just has the least. This forces ranking rather than simple classification and is therefore the harder variant.
Variation D — Mixed NOT formulations (6 questions, 7%). Various other "NOT" constructions: "Which is NOT identified by the author as characteristic of [X]?" Functionally equivalent to Variation B.
The most reliable way to handle any reversal question is to work in two explicit steps. Mixing them up is how mistakes happen.
Step 1 — Apply the base question. Strip away the reversal. If the stem says "According to the passage, all of the following are true EXCEPT," re-read it as "According to the passage, which of the following are true?" Treat this as an ordinary detail or inference question and classify each of the five choices as supported or not supported.
Step 2 — Flip. Once you've classified all five, pick the one that fails the base test. In EXCEPT and NOT questions, that's the single choice labeled "not supported." In LEAST questions, it's the choice that is least supported — you may need to compare the weakest two to decide.
Key discipline: check all five choices. A reversal question rewards completeness and punishes pattern-matching. If you find an answer that feels unsupported on choice (A), check (B) through (E) anyway — one of them might be even less supported, and you might have missed textual evidence that does in fact support (A).
Keep a verification trail. For each of the four passage-supported choices, you should be able to point to specific passage text that supports it. If you can only find evidence for three, the fourth is probably your answer.
The correct answer is the one that lacks passage support — but LSAC builds it carefully to look plausible. It will be topical (on theme with the passage), grammatically parallel to the other four choices, and often close to something the passage does say, but not actually asserted or inferable from it. A total non-sequitur would be too easy.
The four "wrong" answers — which are actually the passage-supported choices — are deliberately spread across different parts of the passage, preventing you from finding all four in the same paragraph. This forces you to do a thorough sweep rather than a local check.
Stems average about 17.4 words; answers average about 11.9 words — relatively short, because the choices are typically straightforward factual claims or brief descriptions that can be quickly checked against the passage. All five choices are constructed to look similarly plausible: similar length, similar complexity, similar topic relevance.
The trap shape is inverted relative to normal questions: the "wrong" answers are the four passage-supported choices, and they trap you by looking like winners.
Trap 1 — Missing the keyword. The simplest and most common failure. You read the stem, forget the reversal, and pick the first choice that sounds passage-supported. Defense: scan every stem for capital-letter keywords before evaluating choices.
Trap 2 — Overlooked detail. One of the four supported choices comes from an easily missed sentence in the passage. You don't find the supporting text, conclude the choice is unsupported, and pick it as your answer. Defense: when a choice looks unsupported, do a targeted rescan of the passage for keywords from that choice before committing.
Trap 3 — Paraphrase mismatch. One of the supported choices paraphrases the passage loosely, and you reject it as "not exactly what the passage said." EXCEPT questions often test whether you recognize passage content under paraphrase. Defense: allow reasonable paraphrase — the supported choices won't be verbatim.
Trap 4 — Weakest supporter in a LEAST question. In LEAST variants, two choices may both feel weakly supported. The correct answer is the one with the least support, not just weak support. Defense: when you have two candidates, rank them explicitly rather than picking the first.
Base difficulty is 4. The modifier doesn't inherently make a question hard — but it inflates the verification work.
Difficulty 3 — localized detail. When all four supported choices come from the same paragraph and are easy to verify, the question drops to 3.
Difficulty 4 — spread across the passage. When the four supported choices are distributed across multiple paragraphs, you have to re-read extensively to verify each. Time pressure, not logical difficulty, is what makes these hard.
Difficulty 5 — LEAST or inference-based. LEAST variants require degree-based evaluation rather than binary yes/no. Inference-based EXCEPTs require you to judge what the passage does and does not support, which is inherently harder than checking what it literally says. The correct answer is often crafted to look plausible at first glance, which is what pushes it into 5-territory.
Reversal questions are heavily skewed toward single passages: 86 questions (96%), vs. 4 (4%) on comparative passages. Rare on comparative passages because verifying four supported claims across two passages is extremely time-intensive — LSAC tends to use faster-to-grade formats in that context.
All reversal stems signal the same two-step task: apply the base question, then flip. The critical habit is noticing the capitalized keyword — EXCEPT, NOT, or LEAST — which is the only difference between these and their standard counterparts.