01 / 20

Correlation vs.
Causation

Two variables that move together are correlated — but that does not mean one causes the other.

PRAXIS Core Math (5733)

What Is Correlation?

Correlation measures how strongly two variables move together. When one variable changes, the other tends to change in a predictable way.

Variable X Variable Y trend line

Types of Correlation

The correlation coefficient \( r \) ranges from \( -1 \) to \( +1 \) and describes both the direction and strength of a linear relationship.

Positive \( (r > 0) \)
Both variables increase together. Closer to \( +1 \) means stronger.
Negative \( (r < 0) \)
One increases as the other decreases. Closer to \( -1 \) means stronger.
None \( (r \approx 0) \)
No consistent linear pattern between the variables.

What Is Causation?

Causation means a change in one variable directly produces a change in another. The first variable is the cause; the second is the effect.

CAUSE (change here) produces EFFECT (result there)

The Core Rule

Two variables can be strongly correlated without one causing the other. A high \( |r| \) tells you the variables move together — it says nothing about why.

Why Correlation Isn't Causation

When two variables \( X \) and \( Y \) are correlated, there are several possible explanations:

1
\( X \) causes \( Y \). The straightforward case.
2
\( Y \) causes \( X \). The direction is reversed (reverse causation).
3
A third variable \( Z \) causes both (a confounding variable).
4
Coincidence. The correlation is just chance with no real link.

The Confounding Variable

A confounding variable \( Z \) influences both \( X \) and \( Y \), creating a correlation between them even though neither causes the other.

Z (heat) confounder X: ice cream sales Y: drowning incidents correlated (but no direct cause)

Correlation vs. Causation

Correlation Causation
MeaningVariables move togetherOne produces the other
DirectionSymmetric (\(X\)–\(Y\))One-directional (\(X\rightarrow Y\))
EvidenceObservational data, \( r \)Controlled experiment
Implies the other?NoYes

How to Evaluate a Claim

1
Identify the variables. What is being measured, and how strongly are they correlated?
2
Ask: could it be reversed? Maybe \( Y \) causes \( X \) instead.
3
Look for a third variable that might cause both.
4
Check the source. Was it a controlled experiment or just observed data?

Example 1: Shoe Size & Reading

In elementary school students, shoe size is strongly correlated with reading ability. Does bigger shoe size cause better reading?

Age Shoe size Reading skill correlated

Example 2: Exercise & Health

A randomized controlled trial assigns people to either an exercise group or a control group. The exercise group shows lower blood pressure, \( r = -0.6 \).

Example 3: Reverse Causation

A study finds that people who use more painkillers report more chronic pain. A headline claims: "Painkillers cause chronic pain."

Chronic pain Painkiller use causes

Common Mistakes

✕ Assuming Cause
Jumping from "correlated" to "one causes the other" without evidence.
✕ Ignoring Confounders
Overlooking a hidden third variable that drives both.
✕ Wrong Direction
Mixing up cause and effect (reverse causation).
✕ Strong \( r \) = Cause
Treating a high \( |r| \) as proof of causation. It isn't.

Quick Summary

Situation Conclusion
High \( |r| \), observational dataCorrelation only
Randomized controlled experimentCausation possible
Possible third variable existsConfounding likely
Effect could precede "cause"Reverse causation

Question 1 of 5

Which statement is always true?

Question 2 of 5

Cities with more firefighters at a fire tend to have more damage. The best explanation is:

Question 3 of 5

What study design best supports a claim of causation?

Question 4 of 5

Students who sleep more tend to get higher grades, \( r = 0.5 \). A valid conclusion is:

Question 5 of 5

A report claims "owning more books causes higher income." If instead higher income lets people buy more books, this is an example of:

You've Got This

Remember: correlation is a clue, not proof. Always check for confounders, reversed direction, and study design before claiming causation.