Correlational Studies in AP Psychology

Reading the sign and size of r

Results from a correlational study are usually summarized with a correlation coefficient (r), which ranges from -1 to +1. The sign gives the direction. A positive r means both variables rise together, and a negative r means one rises while the other falls.

The size of the number, ignoring the sign, gives the strength. An r of -0.60 is a stronger relationship than an r of +0.35, because 0.60 is farther from zero. A common slip is reading a negative correlation as a weak one. Negative only describes direction.

On a scatterplot, a tight cluster of dots along a line means a strong correlation, and a shapeless cloud means a weak one or none. An r near 0 means the two variables don't predict each other in any straight-line way.

Why a correlation can't name the cause

Say a researcher surveys teens and finds that more daily screen time goes with higher anxiety scores. It's tempting to conclude that screens cause anxiety, but the design can't support that claim for two reasons.

The first is the third-variable problem. Some other factor, like poor sleep or social isolation, might push both screen time and anxiety up. The second is the directionality problem. Anxious teens might turn to their phones to cope, which would mean anxiety drives screen time, the reverse of the first guess.

The finding is still useful. If you know a teen's screen time, you can make a somewhat better guess about their anxiety score. Prediction is what correlational data allow.

Spotting a correlational design in a study summary

This term doesn't belong to one numbered topic. It belongs to the research-methods skills that run through the whole course, especially skill 2.A (determine the type of research design used in a study) and skill 2.C (evaluate design elements in non-experimental methods).

The quickest test comes down to who decided each participant's level of the variable. If the researcher assigned people to conditions, ideally with random assignment, it's an experiment. If the researcher recorded existing characteristics and examined the relationship between two or more measured variables, it's correlational. A non-experimental study that only describes what people do or are, without relating variables to each other, is descriptive instead.

Two cases catch people. A study that compares groups that already exist, like athletes versus non-athletes, isn't an experiment, because the researcher didn't create the groups. And a survey is a way to collect data, so a survey study whose goal is to relate two measured variables counts as correlational.

What Article Analysis and Evidence-Based answers need

In the Article Analysis question, you read one summarized study and may be asked to identify its research method, explain what its statistics show, or say whether its results support a conclusion. A strong answer for a correlational study does three things.

• It names the design as correlational.
• It explains why, pointing to the fact that the variables were measured and never manipulated.
• It states the limit on causal claims, ideally naming a specific third variable that fits the study.

In the Evidence-Based Question, you build a claim from three summarized studies. If one of them is correlational, use it as evidence that two variables are related, and avoid writing that it proves one causes the other. Stating that limit directly makes your reasoning stronger.

In multiple-choice sets built on a study description, the wrong answer that looks right usually says "X causes Y" when the data only show a relationship. Questions with an r value or scatterplot ask you to read direction and strength, which takes no calculator.