Skip to content
AP Psychology · Mental and Physical Health

Correlation causation

Correlation causation appears in 14 reviewed AP Psychology questions and sits in Mental and Physical Health. Below are worked examples showing exactly how the exam tests it, with the answer marked and every distractor explained.

Correlation does not prove causation. Two variables can rise and fall together because a third variable drives both, or because the cause runs in the opposite direction from the one you assumed.

Why it matters

Many health claims, like "more sleep raises grades" or "social media causes anxiety", come from surveys rather than experiments. The course wants you to spot when a causal claim goes further than the data allow.

How the exam tests it

A stem gives a survey result and a researcher's causal conclusion, then asks what is wrong with it. The best answer names either a specific third variable, such as after-school work hours affecting both sleep and grades, or the directionality problem.

Don’t confuse it with

Students confuse the third-variable problem with the directionality problem; a third variable is a separate factor causing both, while directionality means the supposed effect may actually be the cause, as when aggressive children provoke more spanking.

Correlation causation in real questions

Example 1Medium
A high school surveys 600 students and finds that those who report sleeping at least 8 hours on school nights have higher grade point averages than those who report sleeping less. The principal concludes that getting more sleep causes students to earn better grades. Which thinking error does this conclusion MOST clearly illustrate?
  1. AIllusory correlation, because the principal perceives a link between sleep and grades that does not actually exist
  2. BInferring causation from correlation, because a third variable such as after-school job hours could affect both sleep and gradesCorrect
  3. COperational definition error, because hours of sleep and grades were never stated in precise, observable, measurable terms before the survey began
  4. DSampling bias, because students who choose to report their sleep habits may not represent the school's whole student body

Why B is right

The survey is correlational: it measured sleep duration (Unit 1, Topic 1.5 Sleep) and GPA as they naturally occurred, without randomly assigning students to sleep more or less. Correlation does not establish causation. A third variable, such as long after-school work hours or heavy family responsibilities, could both shorten sleep and lower grades, and the direction could also run the other way (students struggling academically may stay up late studying or worrying). Only an experiment that manipulates sleep and randomly assigns students to conditions could support the principal's causal claim.
Why the other 3 choices miss
A
Incorrect because an illusory correlation is perceiving a relationship that is not actually present in the data; here the survey did find a real relationship between sleep duration and GPA, and the error is the causal conclusion drawn from it.
C
Incorrect because self-reported hours of sleep and GPA are concrete, measurable variables; the flaw is not how the variables were defined but the leap from a correlational survey to a cause-and-effect claim.
D
Incorrect because even a perfectly representative sample would still yield only correlational data; the conclusion's error is inferring causation, not a problem with who was surveyed.
Example 2Medium
A developmental psychologist surveys parents of 8-year-olds and finds that parents who report using physical punishment more often also have children who are rated as more aggressive by their teachers. She concludes that physical punishment causes children to become aggressive. Which of the following concepts BEST explains the flaw in her reasoning?
  1. AIllusory correlation, because no statistical relationship exists between physical punishment and aggression
  2. BThe directionality problem, because children's aggressive behavior may lead parents to use more physical punishmentCorrect
  3. CNegative correlation, because as physical punishment decreases, children's aggression also decreases
  4. DSampling bias, because the study included only families that reported high levels of both physical punishment and aggression

Why B is right

This shows the directionality problem: a correlation between two variables does not reveal which one causes the other. In this Unit 3 operant-conditioning scenario, frequent physical punishment might increase aggression, for example through modeling, but the opposite is equally plausible. Children who are already aggressive may provoke parents into punishing them more often. Because the study is correlational and the variables were not manipulated, the researcher cannot conclude that punishment causes aggression. An experiment or longitudinal design would be needed to establish the causal direction.
Why the other 3 choices miss
A
Incorrect because an illusory correlation is a perceived relationship that does not actually exist in the data; here the survey found a real positive correlation, and the error lies in inferring which variable causes which.
C
Incorrect because punishment and aggression rise together, which makes this a positive correlation; variables that decrease together are also positively correlated, and in any case the flaw concerns causal direction, not the sign of the correlation.
D
Incorrect because nothing in the study suggests that only extreme families were sampled; the parents varied in how often they used punishment, and the error is drawing a one-way causal conclusion from correlational data.
Example 3Medium
A social psychologist studying aggression reviews one city's records for a single summer. She finds that on days when stores sell more cold drinks, police receive more reports of physical fights. She concludes that drinking cold beverages makes people more aggressive. Which concept best explains the error in her conclusion?
  1. AIllusory correlation, because the link between drink sales and fights in the records does not actually exist statistically
  2. BThe directionality problem, because fights in the city may lead residents to buy more cold drinks afterward
  3. CObserver bias, because the researcher's expectations about aggression led her to see a spurious association in the data
  4. DThe third variable problem, because hot weather may raise both cold-drink sales and irritability that fuels aggressionCorrect

Why D is right

This is the third variable problem: a correlation alone cannot show causation, because an unmeasured variable may cause both measured variables. Here that variable is hot weather. Heat increases cold-drink sales. In social psychology, heat is also a known situational factor that raises irritability and aggression. The correlation between drink sales and fights is real, but concluding that one causes the other ignores the temperature that drives both.
Why the other 3 choices miss
A
Incorrect because an illusory correlation means perceiving a relationship that is not in the data. Here the sales and fight records really do vary together; the mistake is the causal conclusion, not the existence of the correlation.
B
Incorrect because the directionality problem asks whether B caused A instead of A causing B. Fights causing drink purchases is implausible, and the real issue is an outside variable driving both.
C
Incorrect because observer bias means a researcher's expectations distort how data are recorded. Here the data come from store sales and police reports, so the error lies in interpreting a correlation, not in collecting the data.

Studied alongside

Other concepts from Mental and Physical Health.