What Is an SAT Inference Question?
An SAT inference question hands you a short passage that builds toward a point, then stops cold right before the sentence lands. Your job: pick the ending that has to be true given what the passage already said, not the ending that merely fits.
These live inside Information and Ideas, one of the two Reading and Writing domains, and they usually show up three to five times out of the 54 reading and writing questions. The setup barely changes from one to the next: a couple of sentences of fact or findings, then a blank where the final clause should be. Four choices try to fill it. Only one of them is actually earned by the text sitting above it.
A city ordinance requires that any building taller than four stories include an elevator. A permit file for a new building of that size runs to several hundred pages: architectural plan sets, a mechanical schedule listing every motor and shaft, and a signed report from the final inspection. Building inspector Yolanda Cruz reviewed the permit file for a newly completed six-story apartment block and found no elevator listed anywhere in the approved plans, the mechanical schedule, or the final inspection report. Assuming the records are complete and the ordinance was in force, it follows that the building ______
Which choice most logically completes the text?
Must Follow, Not Could Follow
That's the whole test, really: could follow versus must follow. A choice can sound smart, even sound like something a scientist would say, and still be wrong, because the passage never actually gets you there. The right answer is boring by comparison. It just restates what the evidence demands. Nothing more.
Picture this setup. A city library tracks how many teens check out physical books each month for two years straight. Checkouts fall steadily from September through May, then spike every single June, the one month the library gives free pizza to anyone who returns ten books. Based on this pattern, the library can reasonably conclude that ______. A choice claiming teens only read in summer reaches past what the data shows. A choice recommending pizza nights all year answers a different question, what should happen, instead of the one actually asked, what follows from the pattern. The choice that survives says something closer to this: the pizza incentive, not the calendar alone, is likely driving at least part of the June jump. That's the only claim two years of data actually back up.
Catch the Turn: However, Although, But
Watch for a contrast word sitting right before the blank. However, although, but, yet. These flip the direction the sentence was heading, and missing one is the fastest way to pick an answer that runs logically backwards. If a passage spends two sentences building up a claim and then drops a however, the ending has to push back against that claim, not agree with it.
Say a passage describes a moth that pollinates only one desert flower, a flower that blooms for a single night each year. The dependence sounds total. Then: however, researchers have since observed this flower self-pollinating when no moths are present, so ______. Anything that keeps insisting the relationship is exclusive ignores the word that just told you otherwise. The ending has to soften the earlier claim, something like: the flower's survival doesn't depend on the moth as completely as it first appeared.
Why the Wrong Answers Look So Good
Wrong choices on inference questions almost never look dumb. They're built to feel right, and two patterns explain most of them. First, outside knowledge: a choice brings in something true in general, plants need sunlight, teenagers like free food, that the passage itself never said. It feels true because it's probably true somewhere. Just not in the text sitting in front of you.
Second, absolute language. Always, never, completely, entirely, everyone. A passage that supports a limited, hedged claim almost never supports the all-or-nothing version of it. If three of four choices use careful words like likely, some or appears, and one uses always, that one's probably the trap. Scale your pick to match how strong the passage's own evidence actually is, and most of these stop fooling you.
How to Get Faster at This
Before you even glance at the choices, read up to the blank and guess the ending yourself, in plain words, the way you'd finish a friend's sentence out loud. Then check each choice against your guess instead of against a vague gut feeling. A choice close to your prediction and traceable to the text wins. One that needs a fact you supplied from outside, or stretches the passage's claim further than it actually went, loses, no matter how reasonable it sounds.
Getting fast at this takes reps on real examples, not rules memorized off a list. StudyPebble's practice by skill drills inference questions on their own, with an explanation after every one, so you can see exactly which part of the text each right answer traces back to. Got a pile of missed inference questions from a Bluebook test already? Screenshot them and upload them at /mistakes/upload, which sorts your misses by skill, so this one rises to the top of the list instead of hiding inside a general score.