Evaluate whether a study design supports a stated statistical claim.
A valid conclusion must stay within what the study design allows.
Core Idea
To evaluate a statistical claim, check two things: does the design support causation (random assignment?), and does the design support generalization (random sampling?). The valid conclusion must not exceed what the design allows.
Understanding
This concept pulls everything together. The SAT gives you a study description and a conclusion, and you decide if the conclusion is valid.
Use this
| Random sampling | No random sampling | |
|---|---|---|
| Random assignment | Causal claim, generalizable to population | Causal claim, applies only to participants |
| No random assignment | Association only, generalizable to population | Association only, applies only to participants |
Most wrong answers make one of two errors:
- Overclaiming causation — saying the treatment "caused" a result when there was no random assignment.
- Overgeneralizing — applying results to a broader population than the sampling method supports.
When you see a question like "Which of the following is an appropriate conclusion?", find the strongest claim that doesn't cross either line. The SAT rewards careful, conservative reasoning — pick the conclusion that fits the design, even if it feels underwhelming.
Step by Step
- Read the study description. Note: Was there random assignment? Was there random sampling? From what population?
- Determine the study type: experiment (random assignment) or observational (no random assignment).
- Determine the scope: generalizable (random sampling from a defined population) or limited to participants.
- Match the stated claim to what the design supports. Reject any claim that asserts causation without random assignment or generalizes without random sampling.
Misconceptions
- Thinking that a well-designed study must support both causation and generalization. Many real studies only support one or the other.
- Choosing the most dramatic or interesting conclusion instead of the one that fits the design.
- Missing that a study can be an experiment (random assignment) without using random sampling — and vice versa.
Worked Example
A company asked for volunteers from its 3,000 employees. Of the 200 who volunteered, 100 were randomly assigned to use standing desks for 3 months while the other 100 used traditional desks. At the end of the study, the standing desk group reported fewer back pain episodes on average. Which of the following conclusions is best supported by this study?
Select an answer to see the explanation