Causal conclusion in AP Statistics

A causal conclusion is a statement that one variable directly causes a change in another. In AP Statistics, you can only draw a causal conclusion from a well-designed experiment with random assignment of treatments, never from an observational study, no matter how strong the association.

What is causal conclusion?

A causal conclusion says "X causes Y," not just "X is associated with Y." That's a big claim, and AP Stats is strict about when you're allowed to make it. The key ingredient is random assignment of treatments to experimental units. Random assignment spreads potential confounding variables roughly evenly across the treatment groups, so if the groups end up noticeably different at the end, the treatment is the most plausible explanation.

The CED spells out the logic (VAR-3.E.2 and VAR-3.E.3): random assignment lets researchers conclude that some observed differences are too large to be likely due to chance alone. Those differences are called statistically significant, and statistically significant differences between treatment groups count as evidence that the treatments caused the effect. Observational studies can never get you there, because without random assignment, lurking variables could be driving the relationship instead of the explanatory variable.

Why causal conclusion matters in AP® Statistics

This term lives in Topic 3.7 (Inference and Experiments) in Unit 3: Collecting Data, under learning objective 3.7.A, which asks you to interpret the results of a well-designed experiment. The whole point of Unit 3 is matching the data collection method to the conclusion you're allowed to draw, and "causal conclusion" is the payoff word. Random assignment buys you causation; random sampling buys you generalization to a population (VAR-3.E.4). Mixing those up is one of the most common ways to lose points on scope-of-conclusion questions, which show up constantly in both multiple choice and FRQs.

How causal conclusion connects across the course

Random Assignment (Unit 3)

Random assignment is the engine that makes causal conclusions legal. By randomly assigning treatments, you balance confounding variables across groups, so a significant difference can be pinned on the treatment itself.

Statistical Inference (Units 6-9)

When you run a significance test on experiment data later in the course, a small p-value tells you the difference is unlikely to be chance. Pair that with random assignment from Unit 3 and you've earned the right to say "caused," not just "is associated with."

Sampling Variability (Units 3-5)

Sampling variability is why we need the word "significant" at all. Groups will differ a little just by chance, so a causal conclusion requires showing the observed difference is bigger than what random chance would typically produce.

Experimental Units (Unit 3)

Causation and generalization are separate questions. Random assignment to experimental units supports causation, but you can only generalize the result to a larger population if those units were randomly selected from it (VAR-3.E.4).

Is causal conclusion on the AP® Statistics exam?

Scope-of-conclusion questions are an AP Stats staple. Multiple choice stems ask things like "When can a well-designed experiment support causal conclusions?" or describe a flawed experiment (for example, a fertilizer study where the treatment group accidentally got extra water) and ask whether causation is still justified. The answer hinges on whether random assignment happened and whether a confounding variable slipped through. On FRQs, like 2018 FRQ 4 about ACL surgery recovery, you have to state in words whether a causal conclusion is appropriate and why. Graders want the magic phrases. If treatments were randomly assigned and the difference is statistically significant, say the treatment caused the effect. If it's an observational study, explicitly say a causal conclusion is NOT appropriate and name a possible confounding variable. Vague answers like "correlation isn't causation" without tying it to the design won't earn full credit.

Causal conclusion vs Generalization to a population

These answer two different questions. A causal conclusion ("the treatment caused the effect") is justified by random ASSIGNMENT of treatments. Generalizing results to a larger population is justified by random SELECTION of units from that population. An experiment with volunteers and random assignment can prove causation but only for units like the ones studied; a random sample without random assignment can generalize an association but can't prove cause. Keep the two randoms straight: assignment = causation, selection = generalization.

Key things to remember about causal conclusion

• A causal conclusion claims one variable directly causes a change in another, and it requires a well-designed experiment with random assignment of treatments.

• Observational studies can never support causal conclusions because lurking or confounding variables might explain the association.

• Random assignment balances confounding variables across groups, so statistically significant differences between treatment groups are evidence the treatments caused the effect (VAR-3.E.3).

• Statistically significant means the observed difference is so large it's unlikely to have happened by chance alone (VAR-3.E.2).

• Random assignment justifies causation, but random sampling is what justifies generalizing the result to a larger population. They are separate design features answering separate questions.

• On FRQs, always state explicitly whether a causal conclusion is or isn't appropriate, and tie your reasoning to the study design.

Frequently asked questions about causal conclusion

What is a causal conclusion in AP Stats?

A causal conclusion is a statement that one variable directly causes a change in another, like "the new fertilizer caused the plants to grow taller." In AP Stats it's only valid when treatments were randomly assigned in a well-designed experiment and the results are statistically significant.

Can you ever draw a causal conclusion from an observational study?

No. Without random assignment, confounding variables could explain the observed association, so even a very strong relationship in an observational study only supports a conclusion about association, never causation.

What's the difference between a causal conclusion and generalizing results?

Causation comes from random assignment of treatments; generalization comes from random selection of units from a population. An experiment on volunteers can prove the treatment works for people like those volunteers, but you can't safely extend that claim to everyone.

Does a small p-value alone prove causation?

No. A small p-value only shows the difference is unlikely to be due to chance. You also need random assignment in the design. A significant result from an observational study still only shows association.

How do causal conclusions show up on the AP Stats exam?

You'll see scope-of-conclusion questions in both MCQs and FRQs, like 2018 FRQ 4 on ACL surgery recovery, asking whether causation is justified given the study design. Full credit requires linking your answer to random assignment (or its absence) and naming a confounding variable when causation isn't appropriate.