Causation in AP Statistics

In AP Statistics, causation means a change in one variable directly produces a change in another. You can only conclude causation from a well-designed experiment with random assignment. A correlation, no matter how strong, shows association, not cause and effect.

What is Causation?

Causation is a cause-and-effect relationship. Changing the explanatory variable actually makes the response variable change. It's a much stronger claim than saying two variables move together.

Here's the catch AP Stats keeps hammering. Data can show you that two variables are associated, but association by itself can't tell you why. A lurking or confounding variable might be driving both, the direction might be backwards, or the pattern might be pure coincidence. The only way to back up a causal claim is a randomized experiment. Random assignment spreads those hidden variables evenly across treatment groups, so a difference in the response can be traced to the treatment. Think of it this way: correlation tells you two things are dancing together, and an experiment is how you find out who's leading.

Why Causation matters in AP® Statistics

Causation comes up in Topic 5.2 (Correlation). The core skill there is interpreting the correlation coefficient r correctly, and that includes knowing what it can't tell you. A strong r describes the direction and strength of a linear relationship. It says nothing about cause and effect. The idea reaches well past one topic, though. It's the reason experimental design exists, and it determines which conclusions you can write in any study-design or inference FRQ. Making a causal claim the data doesn't support is one of the most common ways to lose points on the AP Stats exam.

How Causation connects across the course

Correlation (Topic 5.2)

Correlation measures how tightly two quantitative variables follow a straight-line pattern. It never measures whether one causes the other. Even r = -1 is just a perfect linear pattern, not proof that x drives y.

Confounding Variable (Unit 3)

A confounding variable is the usual reason a correlation fakes causation. When a third variable is linked to both x and y, you can't tell whose effect you're seeing. Pirates and global temperature both fell over 200 years because time and industrialization changed everything, not because pirates cool the planet.

Experimental Design (Unit 3)

Random assignment to treatments is the tool that earns you the right to say "causes." It balances out confounders across groups, so the treatment is the only systematic difference left. That's why experiments can show causation and observational studies can't.

Is Causation on the AP® Statistics exam?

Causation mostly shows up in multiple-choice interpretation questions. You get a correlation and a tempting causal story, and you have to reject the story. Typical stems look like this:

• A researcher finds r = -0.82 (or -0.85) between study time and anxiety. Which interpretation is most accurate? The right answer describes a strong negative linear association. It does not say studying reduces anxiety.
• A strong negative correlation between the number of pirates and global temperatures. Which principle is violated by claiming pirates prevent global warming? The answer is that correlation does not imply causation.
• r = -1. Which interpretation is correct? A perfect negative linear relationship, still with no causal conclusion.

On FRQs, the skill is matching your conclusion to the study design. If the data came from an observational study, say "associated with," not "caused." If treatments were randomly assigned, you can make a cause-and-effect conclusion. No released FRQ has used the word "causation" verbatim, but the distinction is built into study-design and inference questions.

Causation vs Correlation

Correlation is a number (r) that describes the direction and strength of a linear relationship between two quantitative variables. Causation is a claim that one variable actually produces change in the other. Correlation comes from any paired data. Causation needs a randomized experiment. A strong r makes causation worth investigating, but it never proves it.

Key things to remember about Causation

• Causation means a change in one variable directly produces a change in another, which is a stronger claim than an association.

• A correlation coefficient, even r = 1 or r = -1, describes a linear pattern and never proves cause and effect on its own.

• Confounding variables, reversed direction, and coincidence can all create a strong correlation with no causal link.

• Only a well-designed experiment with random assignment to treatments supports a cause-and-effect conclusion.

• In an observational study, write that the variables are associated, and avoid saying one causes the other.

• On the exam, any answer choice that turns a correlation into a causal claim is almost always a trap.

Frequently asked questions about Causation

What is causation in AP Statistics?

Causation is a cause-and-effect relationship where changing one variable directly changes another. In AP Stats, you can only conclude it from a well-designed experiment that randomly assigns treatments.

Does a strong correlation mean causation?

No. Even r = -1 only means the points fall on a perfect negative line. A lurking variable or plain coincidence could explain the pattern, like the strong negative correlation between pirate numbers and global temperatures.

What is the difference between correlation and causation in AP Stats?

Correlation is a measured association (r) between two quantitative variables. Causation is the claim that one variable actually produces change in the other. You can calculate correlation from any data, but causation requires random assignment in an experiment.

Can an observational study prove causation?

No. Without random assignment, confounding variables can't be ruled out, so an observational study can only show association. If study time and anxiety have r = -0.85 in survey data, you can't conclude that studying lowers anxiety.

How do you show causation on the AP Stats exam?

Point to random assignment of treatments in a well-designed experiment. That's what balances confounding variables across groups. If the study wasn't a randomized experiment, phrase your conclusion as an association instead.