What are evaluating statistical claims questions on the SAT?

Evaluating statistical claims is one of the seven Problem-Solving and Data Analysis testing points on the digital SAT. The question describes how a study was run and gives four conclusions, and you pick the one the study design actually supports. There is no calculation — the work is comparing each conclusion's wording against the design.

When can you claim cause and effect on SAT statistical claims questions?

Only when subjects were randomly assigned to groups, which makes the study an experiment. An observational study, where subjects were watched as they already were, supports association only — wording like 'is associated with' or 'tends to'. A causal verb on an observational study is wrong even when the claim sounds true in real life.

What is the difference between random selection and random assignment on the SAT?

They do different jobs and the SAT tests both. Random selection decides which group a conclusion may name — results from a random sample of one city's residents apply to that city, not to everyone. Random assignment decides whether a conclusion may claim cause at all.

Why do students miss SAT evaluating statistical claims questions?

The three common failures are treating 'randomly selected' as permission to claim cause, picking the conclusion that is true in real life rather than the one this study shows, and accepting a hedged causal verb such as 'may cause' or 'likely caused'. A softener on a causal verb is still a causal verb.

SAT Evaluating Statistical Claims

Last updated: September 17, 2026

Evaluating statistical claims questions hand you a short description of a study and four conclusions, then ask which conclusion the study supports. You are not grading the study — you are grading the sentence. Two things in each answer choice decide it: the verb, which has to match how subjects landed in their groups, and the group the claim names, which has to match who was actually sampled.

Practice SAT skills now →

What does this skill actually test?

Whether you can tell a claim from an overclaim.

This is its own testing point — one of the seven in the Problem-Solving and Data Analysis domain, which supplies about 15% of the math section, roughly 5 to 7 of the 44 questions. It sits next to inference and margin of error, covered in the companion post, but it asks a different thing. Inference questions ask what the numbers estimate. These ask what sentence you are allowed to say out loud. There is nothing to compute.

Which verb is the study allowed to use?

The verb is where most of the points go.

Studies come in two kinds, and one detail decides which: did the researchers put subjects into groups, or did the subjects show up already in them?

Experiment — subjects were randomly assigned to groups. A causal verb is allowed: caused, led to, resulted in.
Observational study — subjects were watched as they already were. Only an associative verb is allowed: is associated with, tends to, is linked to.

Random assignment is the whole license. Without it the groups may differ in a hundred ways nobody measured, and the strongest honest word is "associated."

Which group is the claim allowed to name?

Whoever was sampled, and no one past them.

If the sample was randomly selected from one city's residents, the claim can name that city's residents. Not the state. Not "people." If the sample was not random at all — volunteers, or whoever answered the email — the claim stays with the participants.

So every answer choice gets two checks, in this order: is the verb bigger than the design, and is the group bigger than the sample?

What does that look like on a real question?

Researchers tracked 400 randomly selected adults in one city for a year. Adults who reported walking at least 30 minutes a day had lower resting heart rates than those who did not.

Nobody assigned the walking — the adults chose it. Random selection, no random assignment: associative verb only, and the claim stops at this city. The choices:

Walking 30 minutes a day causes lower resting heart rate in adults. — causal verb on an observational study, and "adults" is everyone. Two failures.
Walking 30 minutes a day causes lower resting heart rate in adults in this city. — group fixed, verb still causal. Out.
Adults everywhere who walk at least 30 minutes a day tend to have lower resting heart rates. — verb fine, group too big. Out.
Among adults in this city, walking at least 30 minutes a day is associated with lower resting heart rate. — verb matches the design, group matches the sample. This one.

Four choices, one paragraph, no arithmetic. That is the whole question type.

What mistakes cost the most points here?

Reading "randomly selected" as permission to claim cause. Selection and assignment are different words doing different jobs. Selection sets the group you can talk about; only assignment lets you say caused.
Picking the answer that's true in real life. Walking probably is good for your heart. Irrelevant. The question is whether this study shows it.
Letting a hedge smuggle cause back in. "May cause," "likely caused," "would reduce" — all still causal. A softener on a causal verb is a causal verb.

Practice routine

Work 8 to 10 of these and make both calls before you read the options:

Write "assigned" or "observed" next to the paragraph before looking at a single answer choice, because that one word eliminates about half the options on sight
Circle the verb in every choice and strike the causal ones the moment the study is observational, so you're choosing between two options instead of four
Circle the group each choice names and compare it to the sampled group, since a correct verb attached to an inflated group is the trap that survives the first cut

Overclaim misses don't look like math misses. They feel like a coin flip and they hide inside a math score that seems fine. HIROSCORE scores this skill on its own instead of folding it into the rest of Problem-Solving and Data Analysis, so a run of them shows up as exactly what it is. The GPS for your SAT score.

Put this into practice.

Practice SAT skills now →

Take the HIROSCORE diagnostic →

References

Problem-Solving and Data Analysis — SAT Suite, College Board
Digital SAT Test Specifications Overview — College Board
Skills Insight for the SAT Suite — College Board
Math Content Alignment — SAT Suite, College Board
Princeton Review — SAT Sections (digital SAT Math section length and module structure)