How to Answer Inference Questions on the Digital SAT

The student who knew too much

I build UnlimitedTests, and we've just opened it to our first students — which means right now I spend an unreasonable amount of time inside their practice data: every missed question, every review session. Almost immediately, one of our sharpest early users showed a miss pattern that stopped me cold. Not grammar. Not vocabulary. She was losing points on exactly one kind of Reading & Writing question: the ones that end with a blank. "Which choice most logically completes the text?"

It took me two evenings with her review history to see it: she was missing them because she was smart. On a passage about an archaeological site, the answer she picked wasn't careless — it was the most interesting of the four, the one a curious person who'd seen a documentary on the topic would reach for. Her wrong answers were consistently the cleverest ones on the screen.

The SAT does not reward interesting. 🙂 On inference questions, the SAT rewards the smallest, dullest, most defensible step you can take from the words on the screen. That's the whole game. Once she started treating these as "smallest step" questions instead of "smartest thought" questions, the miss pattern in her data flatlined — almost overnight.

Let me show you what that looks like in practice.

"Inference" doesn't mean what it means in English class

In a classroom discussion, inferring means reading between the lines — bringing your knowledge and imagination to a text. On the Digital SAT, the word means something narrower, and the difference is worth a real score jump if these questions cost you points today.

An SAT inference must satisfy one test: if the passage is true, the answer must be true. Not "could be true." Not "is probably true in the real world." Must follow, from the passage alone, with nothing added.

That single standard explains every wrong answer you'll see on these questions. Wrong answers are almost always reasonable. They sound like things a thoughtful person might conclude. They just require one extra assumption — a fact from outside, a motive nobody stated, a "must" where the text only supports a "might." The test writers are betting you'll supply that extra piece yourself without noticing.

Watch one get solved (slowly, on purpose)

Here's an original example in the exact shape the test uses. Read it the way you would on test day:

In 2023, researchers described a remarkable find at Zambia's Kalambo Falls: two interlocking logs, deliberately notched and shaped, forming part of a wooden structure. Dating placed the wood at roughly 476,000 years old — hundreds of thousands of years before Homo sapiens appeared. Therefore, ______

Which choice most logically completes the text?

A) the structure must have been built by a species other than Homo sapiens. B) Homo sapiens appeared far earlier than scientists currently believe. C) Kalambo Falls is the most important archaeological site in Africa. D) wooden structures were more common in prehistory than stone ones.

Before the answer, look at how a trained reader marks this passage up — two premises, one tiny step:

Now the choices. B is the trap for strong readers: it feels bold and scientific, but it contradicts a premise instead of following from it — the passage states as fact that the wood predates Homo sapiens; you don't get to overturn that. C brings in a value judgment ("most important") the passage never makes about any other site. D compares wood to stone; the passage mentions no stone at all.

A is correct — and notice how it almost feels too obvious. If the wood predates our species, and somebody shaped it, then somebody who wasn't us shaped it. If the passage is true, A must be true. That "almost too obvious" feeling is not a warning sign. On these questions, it's confirmation.

The leash rule

Here's the mental model I keep coming back to whenever I walk a student through one of these misses:

The right answer is on a short leash staked to the passage. Wrong answers float off toward three places, and after you've done twenty of these you'll recognize all three on sight: your own knowledge (true things the passage never said), a better story (motives, causes, and drama the author never wrote), and extreme language (the passage supports "some" and the choice says "all").

Try a second one — different subject, same leash

City neighborhoods with dense tree canopy stay measurably cooler in summer than nearby blocks with few trees. Yet a recent survey of twelve U.S. cities found that tree canopy tends to be thinnest in the neighborhoods where residents' summer energy costs are highest.

Based on the texts, it can most reasonably be inferred that ______

A) planting more trees would eliminate high energy costs in those neighborhoods. B) the neighborhoods paying the most to stay cool are, on average, the ones receiving the least cooling help from trees. C) city planners deliberately avoided planting trees in lower-income neighborhoods. D) tree canopy has little effect on energy costs outside the summer months.

Run the leash test. A turns a cooling effect into a total cure — "eliminate" is doing work the passage never authorized. C invents a villain; the survey describes a pattern, not a decision, and nobody in the text has a motive. D wanders off to the other nine months of the year, which the passage never discusses.

B just puts the two premises in one sentence: trees cool; the thinnest canopy overlaps with the highest cooling bills; therefore the people paying most get the least of that help. No new actors, no new causes, no absolutes. If the passage is true, B is true. Boring — and correct.

A 60-second drill for test day

When the blank appears, run the same four beats every time — the whole routine takes less than a minute:

1. Name the premises. Usually two. Say them plainly to yourself, the way the margin notes above do.
2. Predict before you peek. Force one dull sentence of your own into the blank. Dull is the goal.
3. Kill the drama. Cross off any choice with outside facts, unnamed motives, or absolute words the passage didn't earn.
4. Run the must-test on whatever's left: if the passage is true, is this forced to be true? Only one choice survives.

I've turned this into a one-page printable you can keep next to your practice sets: download the Inference Questions 60-Second Checklist (PDF) — tape it above your desk until the habit sticks.

Where this fits in your prep

Inference questions live inside the Information and Ideas domain of the Reading & Writing section (each R&W module runs 32 questions in 32 minutes, so a minute-long routine fits the budget). If passages themselves are the struggle — if you finish reading and can't say what the author claimed — start one step earlier with the Central Ideas and Details lesson; the leash only works if you can find the stake.

If you're solid here, the natural next step is Command of Evidence, which is this skill's mirror image: instead of completing a conclusion, you're choosing the fact that props one up. And for the skill on this page, the Inferences lesson has a five-question drill at the end — do it under time, then run the topic practice until wrong answers start looking loud and obvious.

The student from the top of this piece? Every practice set she's run since we walked through the leash rule: zero inference misses. Not because she got smarter — she was always smart. She just stopped letting the test pay her to prove it.

It's one student, and it's early — I won't pretend that's a study. But spotting exactly this kind of pattern is why I built the platform, and as more students come through, I'll keep writing up what their data actually shows. Consider that a promise.