Not sure which lesson to do first? Find your level and get a plan →

Statistics · Grade 9-12

Random Assignment

Random assignment is the process of placing participants into treatment groups by chance.


Quick Answer

Random assignment is the process of placing participants into treatment groups by chance. It helps make the groups similar at the start of an experiment so differences at the end are more likely to be caused by the treatment. Use it whenever you compare treatments: a coin flip, a shuffled list or a random number generator, never the researcher or the participants, decides who goes where. It balances even the differences no one thought to measure, though only on average, so larger groups balance better.

Why This Matters

Random assignment is what lets an experiment say "caused" instead of "is associated with". It protects against the quiet ways groups end up different: keen volunteers choosing the new method, a coach picking the first players to arrive, a doctor giving a new drug to healthier patients. Without it, any gap in outcomes could come from who was in each group rather than from the treatment.

Intuitive Explanation

A sleep researcher wants to compare two homework schedules: finishing homework before dinner or after. If students choose, the ones who pick "before dinner" may be the organized ones who also sleep well, and the comparison is spoiled before it begins. Instead, she lists all 30 students, gives each a random number, sorts the list by that number and puts the first 15 on the before-dinner schedule. Chance, not choice, decides.

Chance balances what you can see and what you cannot. Organized and disorganized students, early birds and night owls, students with noisy homes: all are split roughly evenly, because each student is equally likely to land in either group. After assignment, the groups' average prior grades might be 82.4 and 81.9, close but not identical. Random assignment balances on average, not perfectly; with 15 per group some chance differences remain, and with 500 per group they shrink a great deal.

Random assignment is not random sampling, although both use chance. Random sampling decides who is in the study, so the results can be generalized to a population. Random assignment decides which treatment each participant gets, so differences can be blamed on the treatment. A study of 60 volunteers who were randomly assigned supports a causal claim about people like those volunteers, but says less about everyone else.

Think of it like this. Random assignment is like picking teams for a fair game by drawing names from a hat instead of letting captains choose: strong and weak players end up spread across both teams, so the result tells you more about the strategy than about the roster. With only a few players, one team can still get lucky.

Core idea. Let chance decide which treatment each participant gets; then the groups are alike on average in every respect, known or unknown, and a difference in outcomes can be credited to the treatment rather than to who was in each group.


When to Use

Use random assignment when you will compare treatments in an experiment and want to claim that the treatment caused any difference: two teaching methods, a drug and a placebo, version A and version B of a game level. The signal is that you decide who gets what. If you can only observe groups people chose for themselves, you cannot assign, and confounding must be handled another way; if the goal is choosing who is studied from a population, use random sampling.

Tip. Ask: Is chance being used to decide which treatment each subject already in the study receives?

How to Recognize It

Before calling a study's groups fair, check what decided which subjects got which treatment.

  1. Are the subjects already chosen, so the only decision left is which group each one joins?

    Yes puts you at the assignment step. If the decision is who gets into the study at all, that is Random Sampling.

  2. Did a coin, a shuffled list or a random number generator make that decision?

    Yes is random assignment. If the researcher or the subjects chose, the groups may differ from the start, and Confounding Variables can explain the result.

  3. Is the goal to show that the treatment caused the difference between groups?

    Yes is what random assignment supports. If the goal is to describe a whole population, the tool that matters is Random Sampling instead.

  4. Are the groups small, say four or five subjects each?

    Yes means chance can still leave the groups unbalanced; larger groups, or sorting similar subjects into blocks and randomizing within each block, give more reliable balance.

Random Assignment vs Random Sampling vs Control Group vs Blinding

Random Assignment and Random Sampling both use chance, which is why they get swapped, and a control group and blinding are other parts of a fair experiment. Ask what each one protects: the comparison, the generalization, the baseline, or expectations.

Random Assignment

Meaning
Random assignment uses chance to decide which treatment each subject in a study receives.
Key test
Chance sorts subjects already in the study, which supports cause and effect.
Example
A computer randomly puts 30 of 60 volunteers on the new bedtime routine.

Random Sampling

Meaning
Random sampling uses chance to decide which members of a population get into the study.
Key test
Chance picks who is studied, which supports generalizing to the population.
Example
Drawing 60 names at random from all 1,200 students in the school.

Control Group

Meaning
A control group is the group that gets no new treatment, a placebo or the usual treatment, to serve as a baseline.
Key test
Describes what one group receives, not how its members were placed there.
Example
The 30 volunteers who keep their usual bedtime routine.

Blinding

Meaning
Blinding keeps subjects, researchers or both from knowing who got which treatment.
Key test
Guards against expectations after groups are formed, not against unequal groups.
Example
The people scoring sleep quality do not know which volunteers used the new routine.

Worked Examples

Example 1 — Assign by shuffling

Easy

Four of the 30 students in a homework-schedule study are Ana, Ben, Cal and Dee, and the computer gives them random numbers 0.83, 0.12, 0.57 and 0.31. The class is sorted by these numbers, and the 15 students with the lowest numbers go to the before-dinner schedule; the 15th-lowest number in the class is 0.49. Afterwards, the before-dinner group's average prior grade is 82.4 and the after-dinner group's is 81.9. Which schedule does each of the four get, and is the result balanced?
Show the steps and the answer
  1. Sort the four: Ben 0.12, Dee 0.31, Cal 0.57, Ana 0.83.

    Sorting by a random number puts the class in an order that no one chose.

  2. Compare with the cut-off 0.49: Ben (0.12) and Dee (0.31) are below it and go before dinner; Cal (0.57) and Ana (0.83) are above it and go after dinner.

    Each student had the same chance of a low number, so each was equally likely to land in either group.

  3. Check balance: 82.4−81.9=0.5 points in average prior grade.

    A small gap like this is what chance leaves behind; it is not evidence of unfair assignment.

  4. Note what else is balanced: sleep habits, home noise and motivation were never measured, but they are spread by the same chance process.

    This is the advantage over matching by hand: chance balances unmeasured variables too, on average.

Answer: Ben and Dee go before dinner; Cal and Ana go after dinner. The groups' prior grades differ by only 0.5 points, a normal chance difference, so the assignment is fair.

Takeaway: Give every participant the same chance of each group, then expect small, chance-sized differences, not perfect balance.

Example 2 — Random sampling or random assignment?

Standard

Study 1 randomly selects 200 adults from a city's voter list and asks about exercise and mood; those who exercise report an average mood of 7.2 out of 10, others 6.1. Study 2 takes 60 volunteers and randomly assigns 30 to an exercise program and 30 to none; after 8 weeks the groups average 7.0 and 6.2. What can each study conclude?
Show the steps and the answer
  1. Study 1 used random sampling, but participants chose whether to exercise.

    Random selection lets the results describe the city's adults, but no treatment was assigned.

  2. So Study 1 shows that, among the city's adults, exercisers report moods 7.2−6.1=1.1 points higher, an association.

    Exercisers may differ in health, income or free time, which could also lift mood.

  3. Study 2 used random assignment, but its participants were volunteers.

    Chance made the groups comparable, so the treatment is the only systematic difference between them.

  4. So Study 2 gives evidence that the program raised mood by about 7.0−6.2=0.8 points, for people like these volunteers.

    Volunteers may be more willing to exercise than typical adults, which limits how widely the result applies.

Answer: Study 1 generalizes to the city but shows only association (1.1 points). Study 2 supports cause (about 0.8 points) but only for people similar to its volunteers.

Takeaway: Random sampling earns generalization; random assignment earns causation. A study needs both to claim both.

Example 3 — Find the mistake: the first to arrive

Application

A coach wants to test a new sprint drill. She puts the first 20 players to arrive at practice in the new-drill group and the other 20 in the old-drill group. After a month, the new group's 40 m times improved by 0.4 s and the old group's by 0.1 s. She writes: "The new drill improves times by 0.3 s more." Find the mistake and correct it.
Show the steps and the answer
  1. Compute the gap: 0.4−0.1=0.3 s.

    The arithmetic is fine; the question is whether the drill caused the gap.

  2. Ask how the groups were formed: by arrival time, not by chance.

    Players who arrive first may be the most committed, and committed players may improve more with any drill.

  3. Redesign: list all 40 players, assign 20 to each drill by random numbers, and train both groups for the same time.

    With chance deciding, commitment is spread across both groups, so a gap can be credited to the drill.

Answer: The 0.3 s gap cannot be credited to the drill, because early arrivers may differ from late ones. Repeat the test with players assigned to drills at random.

Takeaway: Any non-random rule for forming groups, even one that seems neutral like arrival time, can build a difference into the groups.

Where You'll See It

  • Splitting a class for a teaching method

    Flipping a coin (or using a random generator) to decide which students try a new teaching method makes the two groups similar at the start, so a later difference in scores can be credited to the method.

  • Drug vs placebo trial

    Patients are randomly placed into the drug group or placebo group so that age, health, and habits are roughly balanced between them — the balance that lets the experiment support a causal claim.

  • Two versions of a video game level

    A studio randomly assigns players to level version A or B; because assignment is by chance, differences in completion time point to the level design rather than to who happened to play which version.

Common Mistakes

  1. The slip: Random assignment and random sampling are the same thing.

    Why it's tempting: Both use the word 'random' and both involve chance.

    Instead: Random SAMPLING (choosing who is in the study) supports generalizing to a population; random ASSIGNMENT (sorting subjects into groups) supports causal comparison. They serve different goals.

  2. The slip: Putting people into groups however is convenient still gives a fair experiment.

    Why it's tempting: Any two groups look like a valid comparison.

    Instead: Non-random assignment lets the groups differ before treatment, reintroducing confounding. Only chance-based assignment makes the groups comparable so cause can be inferred.

  3. The slip: Random assignment guarantees the groups are perfectly equal.

    Why it's tempting: If it balances groups, students assume the balance is exact.

    Instead: Random assignment makes groups similar ON AVERAGE and balances confounders probabilistically, but small chance differences remain — larger group sizes make that balance more reliable.


Mini Practice

Try these on your own. Tap Reveal when you want to check.

  1. Describe how to use a random number generator to assign 40 volunteers to a new reading program or the usual program, with 20 in each.

    Hint: Give each volunteer a number first.

  2. For a reading study, 40 volunteers are randomly split into two groups of 20. What is the chance a particular volunteer gets the new program? If 12 volunteers are strong readers, about how many would you expect in each group?

    Hint: Count the spots in the new-program group and divide by the total number of volunteers.

  3. A teacher lets the first 15 students to arrive use a new calculator app; the last 15 use paper. The app group scores 9 points higher. Why is this a weak basis for crediting the app?

    Hint: Could early arrivers differ from late arrivers in ways that affect scores?

  4. Ten volunteers, 4 of them women, are randomly split into two groups of 5. What is the probability that all 4 women end up in the same group?

    Hint: Count all ways to choose group A, then the ways group A holds all 4 women or none of them.

  5. A researcher picks 500 adults at random from a national voter list and asks how many hours they sleep. Is this random assignment or random sampling, and what conclusion can it support?

    Hint: Was any treatment given to the adults?

  6. Sam writes: "We randomly assigned 50 students from our school to a new diet or the usual diet, so our result applies to all teenagers in the country." Correct his conclusion.

    Hint: Random assignment and random sampling each support a different kind of conclusion.

Want the full set?

50 practice questions for this concept — free to try, every one with a complete worked solution showing the why, not just the answer.

Frequently Asked Questions

Is it random assignment if I alternate, putting every other person who walks in into the treatment group?

Not really. Alternating follows a rule someone can predict or bend: a researcher who knows the next person goes to treatment might steer a healthier-looking person into that slot, and people who arrive together may share traits. True random assignment leaves nothing to predict. Systematic schemes are sometimes used in practice, but they give weaker protection against hidden differences between groups.

Why do drug trials keep the assignment list secret from the doctors enrolling patients?

Because a doctor who knows the next assignment might, even without meaning to, enroll a sicker or healthier patient when the matching group comes up. Hiding the list until each patient is enrolled, called allocation concealment, stops the chance process from being quietly bent. Reviews of past trials found that studies without it tended to report bigger treatment effects, a sign that hidden choices had crept back in.

Does random assignment need equal-sized groups?

No. Some trials put two patients on the new treatment for every one on the old, often so more people get a promising drug. What matters is that chance, not anyone's judgment, makes each placement with known probabilities. Equal groups do give the most precise comparison for a fixed total number of subjects, so a 50-50 split is the usual choice when there is no reason to do otherwise.

Learning Path

You need Experimental Design (treatments and comparison groups) and Population vs Sample (who is in a study). Random assignment adds the step that makes groups comparable before treatment, so a difference afterwards can be credited to the treatment. Next, Control Group covers what the comparison group should receive, Placebo Effect shows why it often gets a dummy treatment, and Hypothesis Testing checks whether a gap is bigger than chance assignment alone would produce.

See Also