Σ
SigmaPrep
GuaranteeResultsPricingPractice testsFormula sheetFAQLog in
Log in
← All articles
SAT MathProblem-Solving and Data AnalysisTopic Guide

SAT Statistical Claims Questions: Samples, Studies, and the Four Traps

Σ
Sigma Prep
SAT Math Instructor · 10+ Years Experience
September 6, 2026·5 min read

This is the smallest skill on the test, inside Problem-Solving and Data Analysis, and most tests have at most one of these questions. It is also the easiest skill to go from zero to perfect on, because there are only two rules and the answer choices are written around them.

The two rules

A random sample lets you generalize to the population the sample came from, and only that population. Random assignment to groups lets you claim cause and effect. No random sample, no generalizing. No random assignment, no "caused." Every question on this topic is checking one of those two, and the correct answer is the one that claims exactly as much as the design allows and no more.

The four ways the SAT asks it

1. A sample that is not representative

Evaluating Statistical Claims: Observational Studies and Experiments · Easyanswer C
A school board wants to assess the opinions of all parents in the district about a proposal to extend the school day by one hour. The school board surveyed 600600600 parents whose children participate in after-school sports programs. The survey found that the majority of those surveyed opposed extending the school day. Which of the following is true about this survey?
AThe survey sample should have included more parents whose children participate in after-school sports
BThe survey sample should have consisted entirely of parents whose children do not participate in after-school sports
✓The survey sample is biased because it is not representative of all parents in the district
DThe survey shows that the majority of all parents in the district oppose extending the school day
Watch the video solution in the app · free account

Only parents with kids in after-school sports were asked, so the sample is biased and the result says nothing about all parents. Two choices try to fix the bias by biasing it the other way, which is not how sampling works. The fix is a random sample of everyone.

2. Generalize only to the sampled group

Evaluating Statistical Claims: Observational Studies and Experiments · Easyanswer B
A researcher selected 150150150 people at random from a group of people who regularly eat at Italian restaurants. The 150150150 people were asked to try a new Thai restaurant and then give their opinions. Of those surveyed, 808080% said they did not enjoy the Thai restaurant. Which of the following inferences can appropriately be drawn from this survey?
AAt least 808080% of all people who try Thai food will not enjoy it
✓Most people who regularly eat at Italian restaurants will not enjoy this Thai restaurant
CAt least 808080% of all people who eat at restaurants will not enjoy this Thai restaurant
DMost people who do not regularly eat at Italian restaurants will enjoy this Thai restaurant
Watch the video solution in the app · free account

The sample was random, but it was random from people who regularly eat at Italian restaurants. So the conclusion is about that group and that restaurant. The choices about "all people" or "all restaurants" reach further than the sample can.

3. Voluntary response

Evaluating Statistical Claims: Observational Studies and Experiments · Hardanswer B
Near the end of a popular podcast, the host invited listeners to respond to a poll on the podcast's website. The poll asked "Do you support a ban on single-use plastics?" The host reported that 656565% responded "Yes" and 313131% responded "No." Which of the following best explains why the poll results are unlikely to represent the opinions of all adults in the country?
AThe podcast did not allow listeners enough time to respond to the poll
✓Those who responded to the poll were not a random sample of all adults in the country
CThe percentages do not add up to 100100100%, so any possible conclusions from the poll are invalid
DThere were not 505050% "Yes" responses and 505050% "No" responses
Watch the video solution in the app · free account

Listeners who chose to answer a podcast poll are not a random sample of the country. That is the whole reason, and it is the only reason the test cares about. The choice about percentages not adding to 100 is noise; the missing 4% did not answer.

4. A convenience sample

Evaluating Statistical Claims: Observational Studies and Experiments · Hardanswer A
To determine the mean number of hours spent exercising per week for adults in a community, Marcus surveyed 252525 people at a local gym. For the 252525 people surveyed, the mean number of hours of weekly exercise was 6.26.26.2. Which of the following statements must be true?
✓The sampling method is flawed and may produce a biased estimate of the mean number of hours of weekly exercise for adults in the community
BThe mean number of hours of weekly exercise for adults in the community is 6.26.26.2
CA determination about the mean number of hours of weekly exercise should not be made because the sample size is too small
DThe sampling method is not flawed and is likely to produce an unbiased estimate of the mean number of hours of weekly exercise for adults in the community
Watch the video solution in the app · free account

Surveying at a gym to learn how much the community exercises is sampling the people most likely to exercise. Flawed method, biased estimate. The choice about sample size is a distractor: 25 is small, but the size is not the flaw, the location is.

Try one

This is a real question from our Statistical Claims · Medium set. Solve it the fast way, then watch how I do it.

try one · Statistical claims · Medium
A polling agency surveyed 800800800 registered voters selected at random from a large county and asked about their opinion on a proposed property tax increase. 626262% of those surveyed responded that they supported the proposed property tax increase. Which of the following statements must be true?

I. Of all registered voters in a large county, 626262% supported the proposed property tax increase.
II. If another 800800800 registered voters selected at random from a large county were surveyed, 626262% would say they supported the proposed property tax increase.
III. If 800800800 registered voters selected at random from a neighboring county were surveyed, 626262% would say they supported the proposed property tax increase.

Where the points actually go

Picking the answer that claims too much. Every distractor on this topic either generalizes past the sampled group, turns a correlation into a cause, or invents a fix that is not random sampling. The right answer is the modest one.

The app has 70 questions on this skill, 14 easy, 21 medium and 35 hard, every one with a video explanation. The free diagnostic shows whether this is where your points are going.

Common Questions

Random sampling means the participants were chosen at random from a population, which lets you generalize results to that population. Random assignment means participants were randomly split into treatment groups, which lets you conclude the treatment caused the difference. You need both to generalize a cause-and-effect claim.

Because the people who answer chose to answer. A voluntary response sample is not random, so it does not represent everyone, and the results cannot be generalized beyond the people who responded.

Only the population the random sample was drawn from. A random sample of employees at one office building supports conclusions about that building, not about the whole city or company.

Usually zero or one per test. It is the smallest skill in Problem-Solving and Data Analysis, but it is also the one where a single rule gets the question right every time.

Want more SAT tips like this?

One article a week. No spam.

Ready to raise your SAT Math score?

Take the free Diagnostic Quiz and see where you stand. No payment required.

Try the Diagnostic Quiz →

Keep reading

SAT Math

SAT Ratios, Rates, and Unit Conversions: The Five Types and the Fast Way

Ratios, rates, density and unit conversions are the easiest points in SAT Math to get and the easiest to give back. The five ways they are asked, worked.

SAT Math

SAT Percentage Questions: Five Types, One Equation for All of Them

Percent of, percent increase, reverse percent and chained percents. Every SAT percentage question is one equation used three ways. Five types, worked.

SAT Math

SAT Mean, Median, and Standard Deviation Questions: The Five Types

One-variable data on the SAT is five calculations and two ideas. Here are the five ways mean, median, range and standard deviation are asked, worked.

All SAT Math topic guides →

Σ
SigmaPrep
FeaturesPractice TestsPricingBlogAboutTermsPrivacyContactYouTube
© 2026 Sigma Prep. All rights reserved.SAT® is a registered trademark of the College Board, which was not involved in the production of, and does not endorse, this product.

Sigma Prep (sigmaprep.io) is an online SAT Math prep platform: 7,000+ practice questions, 24 full-length adaptive tests, and a video solution on every problem, built by a tutor with 10+ years of SAT teaching. $49 a month, first test and diagnostic free. Reading and Writing is in the works.