The shape of the trap
An out-of-text answer is a statement the passage never establishes. It is not the opposite of the text (a different, easier trap where the words contradict the choice). It simply has no support. What makes it lethal is that it is usually true anyway: a real fact about bees or the economy, sitting one association away from the topic. Your brain checks it against what you already know instead of against the passage, gets a green light, and moves on.
The College Board builds it the same way every time: take a statement that is adjacent to the topic and reasonable on its face, then offer it as if the passage backed it. In Words in Context questions it wears a different costume, the "could-be-true" definition: a meaning that is plausible for the scenario but that the specific context clue does not point to. Same engine, smaller stage.
Note the family resemblance to its sibling, the true-but-irrelevant answer. That one is accurate and actually in the passage, but it answers a different question than the one asked. Out-of-text is a step further out: not in the passage at all. Both punish the reader who grades choices by "is this true?" instead of "does the text say this?"
Knowledge is the vulnerability
Here is the uncomfortable part. On most tests, knowing more is an advantage. On SAT Reading and Writing, knowing more is the thing being exploited. A student who has read three books about the French Revolution is more likely to wave through an out-of-text claim about it, because the choice matches memory, and memory feels like evidence. It is not. Your outside knowledge is not on the answer sheet. The passage is.
This is why the trap is invisible from the inside. A content gap feels like confusion, and confusion makes you slow down. An out-of-text trap feels like recognition, and recognition makes you fast and confident, which is worse. That frictionless "yes" is the tell. When a choice feels obviously right and you never went back to the text to check it, treat the feeling as a warning, not a green light. It is a close cousin of answer anchoring: you stopped reading the passage and started matching your own knowledge to the choices.
The passage is the only authority
Adopt one rule and hold it without exception: on a reading question, the passage is the only authority in the room. Not you. Not your science class. Not what is verifiably true in the world. The question is never "which of these statements is correct?" It is "which of these statements did this text establish?" Different questions, different answers, and the whole trap lives in the gap between them.
Concretely, a choice can be a flawless, textbook-accurate fact and still be dead wrong here, because the passage did not say it. And the correct answer can feel underwhelming, because all it does is restate what the passage actually supported. Boring-but-supported beats impressive-but-invented every time. Interesting is not a criterion. Supported is.
Convict every choice with words: the Receipt Rule
The method that kills this trap is the Receipt Rule: you may keep an answer only when you can point to the exact words in the passage that force it. A feeling is not a receipt. Your background knowledge is not a receipt. The receipt is a specific clause you could underline and read aloud, after which the answer is undeniable.
Run it as a courtroom, not a vibe check. Each choice is a defendant, and to convict one as the answer you must produce the evidence: the line that makes it true. If you cannot find that line, you cannot convict, no matter how reasonable the choice sounds. "It could be true" is not proof. "The text says it, right here" is.
Two habits make the Receipt Rule automatic:
- Predict before you peek. Answer the question in your own words from the passage first, then look. A choice that matches your text-based prediction is far likelier to have a real receipt than one that merely matches your memory: the same Predict-Then-Peek move that defends against anchoring.
- Point, do not paraphrase. When you defend your pick, put your finger on the words instead of explaining why it is reasonable. If your defense is a paragraph of reasoning and zero quoted text, you are about to get trapped.
Two examples, both invented
Example one, an inference question. Suppose a passage reports that a marine biologist gave octopuses a puzzle box with food inside and found the animals remembered a working solution for several weeks. A choice reads: "Octopuses are more intelligent than most other invertebrates." That is a popular claim, and you may believe it is true. But scan for the receipt. There is none. The text established memory of a solution, not a ranking against other invertebrates. No line convicts the choice, so it goes, however smart octopuses may be. The supported answer is the plain one: the octopuses retained what they learned over an extended period.
Example two, a Words in Context question. A scientist calls her new dataset "rich," and the sentence explains that it held thousands of separate measurements for every subject. Asked what "rich" means here, you see "costly" and "abundantly detailed." In the real world "rich" can absolutely mean expensive, and a huge dataset probably was expensive to collect, so "costly" is could-be-true. But the context clue, "thousands of measurements for every subject," points at quantity of detail, not price. The receipt supports "abundantly detailed." "Costly" is plausible, unsupported, and wrong: the specific clue outranks the plausible association every time. (For the full method, see words in context.)
The filter, and a self-test you can run today
Before you commit to any reading answer, run one silent question: Can I point to the words that make this true? If yes, lock it. If the honest answer is "well, it makes sense" or "everyone knows that," you have caught an out-of-text trap in the act. Reasonable is not the bar. Cited is.
Practice it now: the free simplesat command-of-evidence drill trains exactly this reflex, forcing you to find the line before you can pick, with per-trap feedback so you see the out-of-text pattern by name. Free, no signup, works as a guest.
Here is the self-test. Pull your last practice section. For every miss where you chose a "true-sounding" answer, go back and try to underline the exact words that would have backed your pick. Count how many times you cannot find them. That number is your out-of-text leak, one of the most fixable numbers in the section, because the fix is a rule, not more knowledge. Log each one with a One-Sentence Autopsy: "picked it because I knew it was true, not because the text said it." When that sentence stops showing up, the leak is closed.