Home / Learn / Exam prep / SAT prep / Lesson 4.6
All lessons

Module 4: Problem-solving and data analysis · Lesson 4.6 · about 15 min

Samples, margins of error and what a study can prove

Random samples, scaling a sample up to a population, margins of error, bias, and when a study can claim cause and effect: the SAT's reasoning-about-data questions.

Key points, meanings and formulas on one page: choose "Save as PDF" in the print box.

Module 4 · about 15 minutes · read first: 4.5 Probability. By the end you will judge what a survey or experiment can and cannot conclude.

Two questions decide what a study can claim. Who was studied (a random sample lets you generalise to the population)? How were they split into groups (random assignment lets you claim cause)?

1. From a sample to a population

If 18 of 120 randomly chosen students cycle to school, the best estimate for a school of 1,500 is 18/120 × 1,500 = 225.

2. Margin of error

A sample estimate is rarely exactly right. A margin of error says how far off it plausibly is: 52% ± 3 means the true population value is plausibly between 49% and 55%. Bigger random samples give smaller margins.

Population
everyone the question is about
Random sample
chosen so everyone in the population has the same chance of being picked
Bias
a sampling method that leans one way, like surveying gym-goers about exercise
Random assignment
the experimenter decides by chance who gets the treatment, so the groups match in everything else
Physical meaning
A poll of 1,000 random people can describe a country of millions, while a poll of a million volunteers can still be badly wrong if they are not random.

3. Cause and effect

Observational study: people who drink tea have lower blood pressure. Maybe tea drinkers also exercise more. No cause can be claimed.
Experiment: volunteers are randomly assigned to tea or water. Chance balances exercise, diet and everything else between the groups.
Now a difference can be credited to the tea, but only for people like the volunteers, since they were not a random sample of everyone.
Try it: a school surveys the first 50 students to arrive in the morning about sleep. What is wrong?

Early arrivers may sleep differently from others (they may go to bed earlier, or have long commutes). The sample is not random, so its average may be biased.

“A bigger sample fixes everything.” Size cuts random error, not bias. A huge sample of the wrong people gives a precise answer to the wrong question.

Practice set

Questions on this lesson, from warm-up to hard, in the SAT’s own formats. Every question has a full worked solution.

  • Random sampling lets you generalise to the population; scale by the sample proportion.
  • A margin of error gives a plausible range for the population value.
  • Random assignment lets you claim cause, for people like those studied.
  • Larger samples reduce random error but never fix bias.

Check yourself

1An estimate is 52% with a margin of error of 3 points. A reasonable conclusion is that the true value is:
2What allows a study to conclude cause and effect?
3A larger random sample usually gives:

Sources

Your ticks stay in this browser. Save them as a file on the course home.

Updated 2026-11-22. SAT is a trademark of the College Board, which is not involved with and does not endorse fibub. Every question here is original. Test details were checked on the College Board's website on 22 November 2026; check the official source before test day.