Interpret the correlation coefficient of a linear fit and distinguish correlation from causation, recognizing lurking variables (Ohio S-ID.8, S-ID.9).

What this topic is asking

Ohio standards S-ID.8 and S-ID.9 ask you to interpret the correlation coefficient $r$ of a linear fit and to distinguish correlation from causation. The coefficient $r$ measures how well a line fits the data; causation is a separate, stronger claim that data alone rarely proves. This is a frequent, high-yield Statistics item, and a classic reasoning trap.

Reading the correlation coefficient

A single number, $r$, captures both the direction and the tightness of a linear trend.

So $r = 0.9$ is a strong positive fit, $r = -0.2$ a weak negative one, and $r = 0$ no linear relationship. Algebra I computes $r$ with technology and focuses on interpreting it.

Strength versus direction

Keep the two pieces of information separate.

Correlation is not causation

This is the headline idea, and the most tested.

The classic example: ice-cream sales and drowning rise together, but hot weather (the lurking variable) drives both, neither causes the other.

How Ohio examines this topic

• Multiple choice and multiple-select. Interpret the sign and strength of $r$, or compare two values.
• Reasoning items. Choose the best conclusion about a correlation, recognizing lurking variables and the correlation-causation distinction.
• Numeric response. Match an $r$ value to a described scatter plot.

Why magnitude is strength and sign is direction

The correlation coefficient packs two separate facts into one number, and keeping them apart prevents the most common errors. The sign answers "which way does the trend go?", a positive $r$ goes with a line of positive slope (up-right), a negative $r$ with a line of negative slope (down-right). The magnitude $|r|$ answers "how closely do the points follow that line?", values near $1$ mean the points cluster tightly around the line (strong), values near $0$ mean they scatter widely (weak). Because these are independent, a strong relationship can be either positive or negative: $r = -0.9$ is stronger than $r = 0.3$ even though it is negative. Comparing strength means comparing $|r|$, while the sign only tells direction, which is exactly the distinction reasoning items probe.

Why correlation cannot prove causation

The gap between correlation and causation is the deepest idea in the standard, and it rests on what observational data can and cannot rule out. A correlation tells you two variables move together, but movement-together is consistent with several stories: the first variable might cause the second, the second might cause the first, or a hidden third variable might cause both while neither affects the other. Observational data, where you simply record what happens, cannot distinguish these, because it does not control the other factors that might be responsible. Only a controlled experiment, which deliberately changes one variable while holding others fixed, can isolate cause and effect. This is why "correlation does not imply causation" is a rule rather than a slogan: a strong $r$ is genuine evidence of a relationship, but the explanation for that relationship requires more than the correlation itself.

Try this

Q1. Which is a stronger linear relationship, $r = 0.6$ or $r = -0.8$? [1 point]

• Cue. Compare $|r|$: $0.8 > 0.6$, so $r = -0.8$ is stronger.

Q2. Sleep and test scores are positively correlated. Does more sleep cause higher scores? [1 point]

• Cue. Not necessarily; correlation alone does not prove causation (a lurking variable or reverse direction is possible).