Random assignment is the use of chance (like a random number generator or drawing chips) to allocate treatments to experimental units, which tends to balance the effects of confounding variables so that observed differences in the response can be attributed to the treatments themselves.
Random assignment means you use a chance process to decide which experimental units get which treatment. The CED lists acceptable methods like a random number generator, a table of random values, or drawing chips without replacement. The point is that no human judgment, volunteering, or pre-existing pattern decides who lands in which group.
Here's why it works. Every group of people (or plants, or knees) comes loaded with lurking variables you can't see, things like fitness level, motivation, or genetics. Random assignment spreads those hidden differences roughly evenly across the treatment groups. So if the groups end up with noticeably different responses, the treatment is the most reasonable explanation. That's why random assignment is one of the four pillars of a well-designed experiment in the CED: comparison, random assignment, replication, and control. It's the single ingredient that earns you the right to say "caused" instead of "is associated with."
Random assignment lives in Unit 3 (Collecting Data), mainly Topics 3.5 and 3.7. Learning objective 3.5.B requires you to describe the elements of a well-designed experiment, and random assignment/allocation of treatments is explicitly one of them. LO 3.5.C asks you to compare designs, where the essential knowledge states that random assignment tends to balance the effects of uncontrolled (confounding) variables. LO 3.7.A is where it pays off: random assignment lets researchers conclude that an observed difference is too large to be due to chance alone, which is the basis for calling a result statistically significant and the treatment causal. It also connects back to LO 3.2.B, which says observational studies can never determine causation, precisely because they lack random assignment. This idea echoes through every inference unit later in the course, since the scope of your conclusion (causal or not) always traces back to how the data were collected.
Keep studying AP Statistics Unit 3
Confounding Variables (Unit 3)
Random assignment exists to defeat confounding. A confounding variable is mixed up with the treatment so you can't tell which one caused the effect. Randomizing spreads confounders roughly evenly across groups, so they can't systematically favor one treatment.
Random Sampling and Generalization (Unit 3)
These are two separate chance processes doing two separate jobs. Random selection of the sample lets you generalize to a population; random assignment of treatments lets you conclude cause and effect. An experiment on volunteers can prove causation but only for people like those volunteers.
Control Group (Unit 3)
A control group only works if units are randomly assigned to it. If people choose their own group, the comparison is contaminated by self-selection. Random assignment is what makes the control group a fair baseline.
Statistical Significance and Inference (Units 3 and 6-7)
Per the CED, random assignment is what allows you to say an observed difference is so large it's unlikely to have occurred by chance. Every two-sample test you run later rests on this. Without random assignment, a significant p-value shows association, not causation.
Random assignment shows up constantly on both sections. In multiple choice, you'll see study scenarios and have to spot the design flaw, like a study where patients already chose their own medication (that's an observational study, no causation allowed) or pick which design is least appropriate for establishing causality. On FRQs, it's a design workhorse. The 2019 fungus-concentration question and the 2022 flavonoid blood pressure study both required describing or evaluating a randomized experiment, and the 2022 acne question with 36 pairs of identical twins tested whether you could randomize within matched pairs. When an FRQ says "describe a method for randomly assigning treatments," graders want a concrete, repeatable process, such as numbering subjects 1 to 40 and using a random number generator without repeats, not just the word "randomly." You also need to interpret results, stating that because treatments were randomly assigned, a statistically significant difference is evidence the treatment caused the effect.
This is the most common Unit 3 mix-up. Random selection is how units get INTO the study from a population, and it lets you generalize results to that population. Random assignment is how units already in the study get split into treatment groups, and it lets you conclude cause and effect. A study can have one, both, or neither, and each chance process buys you a different conclusion. Saying "random sampling lets us conclude causation" loses points; only random assignment does that.
Random assignment uses a chance process, like a random number generator or drawing chips without replacement, to allocate treatments to experimental units.
Random assignment tends to balance the effects of confounding variables across groups, so differences in the response variable can be attributed to the treatments.
Random assignment is what justifies cause-and-effect conclusions; observational studies lack it, which is why they can only show association.
Random assignment is one of the four elements of a well-designed experiment, along with comparison of at least two groups, replication, and control.
Random assignment supports causal conclusions, while random selection of the sample supports generalizing to a population. These are different and the exam tests the difference.
On an FRQ, describing random assignment means giving a specific, repeatable method, not just writing the word 'randomly.'
Random assignment is using a chance process, such as a random number generator or drawing labeled chips, to allocate treatments to experimental units. It balances confounding variables across groups so observed differences can be attributed to the treatments.
No, and the AP exam loves this distinction. Random sampling chooses who is in the study and lets you generalize to a population; random assignment splits participants into treatment groups and lets you conclude causation.
No. It tends to balance confounding variables so groups are similar on average, but small differences can still occur by chance. That's exactly what statistical significance measures, whether a difference is too big to blame on chance alone.
Because treatments aren't imposed, subjects sort themselves into groups, so confounding variables travel with their choices. For example, patients who already chose medication A versus B may differ in age, health, or habits, and no analysis can untangle that without random assignment.
Give a concrete process the graders could replicate, like numbering the 40 subjects 1 through 40, generating 20 unique random integers, and assigning those subjects to treatment A with the rest getting treatment B. Released FRQs like 2019 Q2 and 2022 Q2 rewarded exactly this kind of specificity.