Random Sample vs Random Assignment on the SAT: Generalization vs Causation

Random sampling and random assignment solve different statistical problems.

The distinction to memorize conceptually is:

random sample → generalize to a population

random assignment → support a causal conclusion

Random sampling answers:

Who gets into the study?

Random assignment answers:

Once participants are in the study, who gets which treatment?

Those two random processes are easy to confuse because both use chance, but the conclusions they support are different.

Why this distinction is explicitly tested

College Board's current Problem-Solving and Data Analysis specifications include evaluating statistical claims from observational studies and experiments. The Student Question Bank description states that students should understand which population results from random samples can extend to and why random assignment can provide evidence for a causal relationship.

That makes this a claim-strength question as much as a vocabulary question.

For sampling problems, see SAT Sampling Bias. For the study-type distinction, see Observational Study vs Experiment.

Random sampling: how participants are selected

Suppose a city has 100,000 adult residents.

Researchers use a complete resident list and randomly select 1,000 adults.

That is a random sample from:

adult residents of that city

If the sample is appropriately random and the response process is not badly biased, sample statistics can be used to estimate characteristics of that population.

Random sampling mainly addresses representativeness.

Random assignment: how treatments are allocated

Suppose 300 study participants are already enrolled.

Researchers randomly assign:

• 150 to Treatment A;
• 150 to Treatment B.

Random assignment is intended to make treatment groups comparable on both known and unknown background variables, apart from chance.

If the study is well conducted and a meaningful difference emerges, random assignment supports a causal interpretation:

The treatment caused the difference.

Random assignment mainly addresses confounding.

A four-cell framework

There are four broad study-design combinations.

This table captures most SAT claim-evaluation questions.

Worked example 1: random sample, no random assignment

Researchers randomly sample 800 adults in a county and ask how many hours they exercise each week. They also record resting heart rate.

They find people who exercise more tend to have lower resting heart rates.

What conclusions are supported?

Generalization

Because participants were randomly sampled from county adults, the association may reasonably be generalized to that population.

Causation

Exercise was not assigned. People chose their own exercise behavior.

Other variables could differ:

• age;
• diet;
• medication;
• income;
• baseline health.

So the study does not by itself prove that additional exercise caused the lower heart rate.

Generalizable association, not a causal experiment

Worked example 2: random assignment, volunteer sample

A university recruits 120 volunteers for a study of a study app.

The volunteers are randomly assigned:

• app group;
• no-app group.

After four weeks, the app group performs better on the study's outcome.

What can be said?

Causation

Random assignment supports the claim that the treatment caused the observed difference, assuming the study was otherwise well designed.

Generalization

The participants were volunteers, not a random sample of all students.

So broad generalization to every student is less secure.

Causal evidence, limited population generalization

Random sample does not mean random groups

Suppose a survey randomly samples 1,000 people and compares:

• people who already drink coffee;
• people who do not.

The sample is random, but coffee behavior was not assigned.

The groups may differ in many ways.

Therefore random sampling does not create an experiment.

Random assignment does not make a sample representative

Suppose an experiment recruits only competitive runners, then randomly assigns them to two training programs.

Random assignment may support causal comparison of the programs within the studied participant group.

But it does not magically make competitive runners representative of:

• all adults;
• sedentary people;
• children.

Selection and assignment answer separate questions.

Worked example 3: identify what was randomized

A researcher randomly chooses 20 classrooms from a district. Every student in those classrooms receives the same new curriculum. Outcomes are compared with outcomes from schools using the old curriculum.

Is that random assignment of individual students?

No.

The description says classrooms were selected randomly, not necessarily that treatment was randomly assigned.

The word "random" alone is not enough.

Ask:

Randomly selected for what?

The key SAT reading move: identify the verb

Look for verbs such as:

Sampling language

• selected;
• sampled;
• chosen from the population;
• randomly surveyed.

Assignment language

• assigned;
• allocated;
• placed into treatment/control;
• randomly divided into groups.

A question may deliberately include one random process while withholding the other.

Worked example 4: both random sampling and assignment

A company randomly samples 600 customers from its active-customer list. Those customers agree to participate and are randomly assigned to receive one of two website designs.

If enough sampled customers take part and the random assignment is carried out properly, the design supports:

• causal comparison between the website designs;
• generalization to the population from which the random sample was drawn.

This is the strongest basic design in the SAT framework.

Generalization follows the sampling frame

Suppose participants are randomly sampled from:

registered voters in County A

Even excellent random sampling does not directly support inference to:

• all residents of State B;
• all U.S. adults;
• people who are not registered voters.

The population from which the sample was drawn defines the natural scope.

See SAT Sampling Bias.

Causation follows treatment assignment

Suppose two groups differ after a randomized experiment.

Why does random assignment matter?

Because, before treatment, the groups should differ mainly because of chance rather than a systematic selection rule.

That makes treatment status less tangled with confounding variables.

Random assignment does not prove every experimental result is true; poor measurement, attrition, noncompliance, or random chance can still matter.

But it is the key design feature for causal comparison in the SAT framework.

Worked example 5: self-selection destroys causal certainty

Students choose whether to enroll in an optional tutoring program.

Tutoring participants later have higher scores.

Can we conclude tutoring caused the difference?

Not from this design alone.

Students who volunteer for tutoring may differ in:

• motivation;
• prior achievement;
• family support;
• available study time.

The groups were not randomly assigned.

Association, not established causation

Control groups and random assignment

Randomized experiments often compare:

• treatment group;
• control group.

The control group provides a baseline.

But the presence of a control group alone does not guarantee causal validity.

If people choose their groups, confounding may remain.

The crucial question is:

Was treatment assignment random?

Placebos and blinding: useful but separate

A placebo can help isolate treatment effects from expectations.

Blinding can reduce behavioral or measurement effects.

These are useful experimental-design tools, but the SAT's core causal distinction is still random assignment.

Do not confuse:

• random assignment;
• placebo;
• blinding;
• random sampling.

They solve different design issues.

Worked example 6: random sample plus observational exposure

Researchers randomly sample 2,000 adults and compare people who use standing desks with people who do not.

Because desk use is observed rather than assigned:

• generalization to sampled population can be supported;
• causation cannot automatically be established.

Even if the sample is huge and random, standing-desk users may differ from nonusers in occupation, activity level, workplace culture, and other variables.

Worked example 7: experiment with narrow sample

A study recruits 80 healthy adults ages 18–25 from one university and randomly assigns a supplement or placebo.

The experiment finds a difference.

Strongest conclusion:

There is causal evidence for the treatment effect in the study context/participants.

More caution is needed before claiming the same effect for:

• older adults;
• children;
• people with chronic illness;
• the entire national population.

Random assignment strengthens causality; it does not broaden the sampling frame.

Random sampling and margin of error

When a random sample estimates a population proportion or mean, sample size affects precision.

Larger random samples generally reduce margin of error.

But margin of error describes sampling variability, not all bias.

For that topic, see SAT Margin of Error and Sampling.

Random assignment and sample size

A randomized experiment can still be small.

Random assignment addresses confounding, but a tiny sample may produce unstable estimates and low precision.

Again, design quality has multiple dimensions:

• selection;
• assignment;
• sample size;
• measurement;
• execution.

The two-question test

For almost every SAT study-design prompt, ask:

Question 1

Was the sample randomly selected from the population?

If yes, population generalization is better supported.

Question 2

Were treatments randomly assigned?

If yes, causal inference is better supported.

Do not merge these into one "was it random?" question.

Common trap 1: random sample means causal

No.

It supports representativeness/generalization.

Common trap 2: random assignment means representative

No.

It supports treatment comparability/causation.

Common trap 3: "random" appears, so both claims are supported

Identify exactly what was randomized.

Common trap 4: volunteer participants plus random assignment generalize to everyone

Causal design can be strong while external generalization remains limited.

Common trap 5: representative observational study proves cause

A representative association is still an association unless treatment/exposure was experimentally assigned or stronger causal evidence exists.

Common trap 6: control group automatically means randomized experiment

The groups may have been self-selected.

A claim-evaluation routine

1. Identify target population.
2. Ask how participants entered the study.
3. Was sampling random?
4. Identify treatment/exposure.
5. Was treatment randomly assigned?
6. Decide whether the supported claim is:
   • population estimate;
   • association;
   • causal effect;
   • both population and causal.

SAT-style practice

Question 1

A random sample of 700 city residents reports average commuting time.

What is most directly supported?

Estimate/generalization about city residents

Question 2

Volunteers are randomly assigned to two diets.

What design strength is present?

Random assignment supports causal comparison

Question 3

A random sample is surveyed about naturally occurring sleep habits and grades.

Can it establish sleep causes grade differences?

No

No random assignment.

Question 4

A convenience sample is randomly assigned to treatment/control.

Can causal inference be supported?

Yes, more than population-wide generalization

assuming the experiment is otherwise well designed.

Question 5

Researchers randomly sample 1,000 adults and randomly assign them to two treatments.

Which claims are best supported?

Both causal inference and generalization to the sampled population

FAQ: random sample vs random assignment

What does random sampling do?

It supports representativeness and generalization from a sample to the population from which it was randomly drawn.

What does random assignment do?

It supports causal comparison by making treatment groups comparable apart from chance.

Can a study have random assignment without random sampling?

Yes. Many experiments use volunteers but randomly assign those volunteers to treatments.

Can a study have random sampling without random assignment?

Yes. A representative survey or observational study can randomly sample participants without assigning exposures.

Which one proves causation on the SAT?

Random assignment is the key design feature supporting causal inference.

Which one lets me generalize to a population?

Appropriate random sampling.

Does a big sample replace random sampling?

No. Large biased samples can still be unrepresentative.

Does a control group guarantee causation?

Not by itself. Random assignment is the crucial comparison feature.

How to practice this on MastaPrep

The quickest way to lock in a skill like this is to practice it on its own, then again a few days later, mixed in with other questions. That is how MastaPrep is built. A placement check finds your weakest SAT Math areas, the study plan targets them, every question has a worked explanation, and missed questions come back until you get them right without help.

See how MastaPrep works or try free SAT practice questions. The free plan needs no credit card.

Related MastaPrep guides

• SAT Sampling Bias
• Observational Study vs Experiment
• SAT Margin of Error and Sampling
• SAT Probability and Statistics
• SAT Two-Way Tables

Final takeaway

Keep two questions separate:

Who got selected?

That is random sampling and determines how far you can generalize.

Who got which treatment?

That is random assignment and determines whether a causal conclusion is supported.

Once you stop treating "random" as one generic feature, SAT study-design questions become much easier.

---

Official SAT references:
https://satsuite.collegeboard.org/sat/whats-on-the-test/math/types/problem-solving
https://satsuite.collegeboard.org/practice/student-question-bank/math

Trademark note: SAT® is a registered trademark of College Board. MastaPrep is independent and is not affiliated with or endorsed by College Board.