In AP Statistics, a treatment group is the set of experimental units randomly assigned to receive a specific treatment (a level of the explanatory variable), so its responses can be compared to other treatment groups or a control group.
A treatment group is the collection of experimental units (people, plants, plots of land, whatever you're studying) that gets assigned one specific treatment in an experiment. The CED defines treatments as the levels of the explanatory variable that researchers intentionally manipulate. So if you're testing a new blood pressure drug at two doses plus a placebo, you have three treatment groups, one for each level.
Here's the key idea. An experiment doesn't measure a treatment's effect in a vacuum. It measures it by comparison. The whole point of a treatment group is that its responses get stacked up against another group's responses. That's why LO 3.5.B lists "comparisons of at least two treatment groups" as the very first element of a well-designed experiment. A treatment group also needs more than one experimental unit in it. That's replication, and an experiment with one person per group can't separate the treatment's effect from individual quirks.
Treatment groups live in Topic 3.5 (Introduction to Experimental Design) in Unit 3 and support learning objectives 3.5.A (identify the components of an experiment), 3.5.B (describe elements of a well-designed experiment), and 3.5.C (compare experimental designs). On the exam, you're expected to identify treatment groups in a described study, explain why random assignment to them matters, and recognize design flaws like inadequate replication.
The concept also reaches into Unit 8 (Topic 8.7). When an experiment's response variable is categorical, comparing outcomes across treatment groups becomes a chi-square test for homogeneity. Choosing that test correctly requires you to recognize that the groups were created by random assignment of treatments. Treatment groups are where data collection (Unit 3) and inference (Units 6-8) meet, because random assignment to groups is what licenses cause-and-effect conclusions.
Keep studying AP Statistics Unit 8
Control Group (Unit 3)
A control group is just a treatment group that gets no active treatment (or a placebo). It's the baseline you compare against. The CED is explicit that the comparison group "could be" a control group, meaning two active treatments compared head-to-head still count as a valid design.
Random Assignment (Unit 3)
How units end up in a treatment group matters as much as the group itself. Random assignment tends to balance confounding variables across groups, so a difference in responses can be attributed to the treatment and not to who happened to land in which group.
Experimental Unit (Unit 3)
A treatment group is made of experimental units, the individuals actually assigned treatments. Counting units per group is how you check replication. One unit per treatment group is a design flaw the exam loves to test.
Chi-Square Test for Homogeneity (Unit 8)
When the response variable is categorical, comparing the distribution of outcomes across treatment groups calls for a chi-square test for homogeneity. Topic 8.7 tests whether you can match the design (randomly assigned groups) to the right inference procedure.
Multiple-choice questions typically describe an experiment and ask you to identify its components (treatments, treatment groups, experimental units, response variable) or spot what's wrong with the design. A common stem describes a completely randomized design with 100 participants and asks how treatments should be assigned, or asks which scenario violates replication, like putting too few units in each treatment group when testing three pain medications.
On FRQs, treatment groups show up in design and probability contexts. The 2017 FRQ Q6 had two men and two women randomly assigned to a treatment group or a control group of two people each, blending experimental design with probability. The 2022 FRQ Q5 involved a flavonoid study comparing blood pressure reduction across groups. When you write about treatment groups on an FRQ, name the treatments explicitly, describe a concrete random assignment method (random number generator, drawing chips without replacement), and state that comparison between groups is what lets you measure the treatment's effect.
A treatment group receives an active treatment being tested; a control group receives no treatment or a placebo and serves as the baseline. The catch is that a control group is technically one of the treatment groups (the placebo is a "level" of the explanatory variable). Also, you don't always need a control group. Comparing drug A to drug B with no placebo is still a valid experiment under LO 3.5.B, as long as there are at least two groups to compare.
A treatment group is the set of experimental units assigned to one level of the explanatory variable, and the levels themselves are called treatments.
A well-designed experiment compares at least two treatment groups, and one of them can be (but doesn't have to be) a control group.
Random assignment to treatment groups balances confounding variables, which is what allows cause-and-effect conclusions from an experiment.
Replication means each treatment group contains more than one experimental unit; a group of one can't distinguish the treatment effect from individual variation.
In a completely randomized design, you can assign units to treatment groups with a random number generator, a table of random values, or drawing chips without replacement.
When the response across treatment groups is categorical, the comparison becomes a chi-square test for homogeneity in Unit 8.
It's the group of experimental units assigned to receive a specific treatment, where a treatment is a level (or combination of levels) of the explanatory variable. Comparing the response variable across treatment groups is how an experiment measures a treatment's effect.
No. The CED requires at least two treatment groups for comparison, and one of them could be a control group. Comparing two active treatments head-to-head, like two doses of a drug, is still a valid experiment.
The treatment group gets the active intervention being tested, while the control group gets no treatment or a placebo as a baseline. A control group is really just a special kind of treatment group, since the placebo counts as a level of the explanatory variable.
More than one, at minimum. That's the replication requirement in LO 3.5.B. Exam questions often describe a design with one unit per group and ask you to identify it as a replication violation, because with one unit you can't tell the treatment's effect apart from that individual's quirks.
Random assignment is the method of forming treatment groups, not the groups themselves. Using a random number generator or drawing chips without replacement to assign treatments balances confounding variables across groups, which is the whole reason experiments can support causal conclusions.