Not sure which lesson to do first? Find your level and get a plan →

Statistics · Grade 6-8

Correlation

Correlation is a statistical relationship between two variables where changes in one are associated with changes in the other.


Quick Answer

Correlation is a statistical relationship between two variables where changes in one are associated with changes in the other. In a positive correlation the two rise together, like height and weight; in a negative one, one rises as the other falls, like screen time and sleep; with no correlation, the dots of a scatter plot show no steady trend. You use it to describe paired data, judge how closely two measurements are linked and make rough predictions, but a correlation alone never shows that one variable causes the other.

Why This Matters

Correlation turns a cloud of paired measurements into a short claim you can check: "students who sleep more tend to score higher". That is how doctors spot risk factors and coaches spot useful training habits. Without the ideas of direction and strength, people either miss real patterns or read a few scattered dots as a rule, and they often mistake a link for a cause.

Intuitive Explanation

Measure every student in a class twice: hours of sleep last night and score on this morning's quiz. Plot each student as a dot. If the dots drift up from left to right, students who slept more tended to score higher: a positive correlation. If they drift down, as with screen time and sleep, one goes up while the other tends to go down: a negative correlation. If the dots form a shapeless blob, there is no correlation.

Direction and strength are separate. Direction is the sign, uphill or downhill. Strength is how tightly the dots hug a straight line. Six students whose sleep drops by about half an hour for every extra hour of screen time make a nearly straight downhill line, a strong negative correlation. A strong negative correlation is exactly as strong as a strong positive one; the minus sign only tells you which way it runs.

Correlation describes a straight-line trend, so a curve can fool it. A ball thrown upward is at 0, 15, 20, 15 and 0 metres after 0, 1, 2, 3 and 4 seconds. Time completely decides the height, yet the dots go up and then down with no single direction, and the correlation is zero. No correlation means no straight-line trend, not no relationship. And even a strong correlation says only that the variables move together, not that one pushes the other.

Think of it like this. Watch two kayaks on a river. If they tend to drift the same way, they are positively correlated; if one tends to go left whenever the other goes right, negatively. Watching, like correlation, shows only that they move together; it cannot tell whether one kayak steers the other or the current moves both.

Core idea. Two measurements are correlated when knowing one helps you guess the other, so their scatter plot slopes up or down. Direction is the sign, strength is how tightly the dots follow a line, and neither one shows a cause.


When to Use

Use correlation when every individual has two numerical measurements, like hours practiced and points scored, and the question asks whether they tend to rise or fall together and how consistently. Phrases like associated with or "the more ..., the more ..." are signals. If one variable is a category (grade, team, yes or no), compare groups with a two-way table or group averages instead; if the question asks whether one thing causes the other, you need an experiment.

Tip. Ask: Do you need to say whether two measured variables tend to rise together, move oppositely or show no trend, without computing a number?

How to Recognize It

Before calling a pattern a correlation, check that each individual was measured twice and decide what kind of statement the question wants.

  1. Was each person or thing measured on two numerical variables, like hours of sleep and quiz score?

    Yes means a correlation can describe them. If one variable is a category, like team or favorite color, compare group averages instead; if both are categories, use Two-Way Tables.

  2. Is an exact value such as r=−0.62 needed?

    No keeps you in Correlation, described as positive, negative or none. Yes means the Correlation Coefficient, which puts a number on the same idea.

  3. Does the question stop at the two tending to go together, with no claim that one makes the other change?

    Yes stays with Correlation. If it claims a cause, such as phones making students sleep less, the issue is Correlation vs Causation, and you should look for a hidden third variable.

  4. Do the dots form a roughly straight band rather than a curve?

    Yes means positive or negative describes them well. If the dots rise and then fall, like a thrown ball's height over time, describe the curve you see on the Scatter Plot, since the correlation can come out near zero.

Correlation vs Correlation Coefficient vs Correlation vs Causation vs Scatter Plot

These four ideas share one picture, a cloud of dots, so students use the names interchangeably. Ask whether you need a description in words, an exact number, a judgment about cause, or the graph itself.

Correlation

Meaning
Correlation describes whether two measured variables tend to rise together, move in opposite directions, or show no steady trend.
Key test
A statement in words about direction and how tight the trend is.
Example
Students who sleep more tend to score higher on the morning quiz: a positive correlation.

Correlation Coefficient

Meaning
The correlation coefficient r is one number from −1 to 1 that measures the direction and strength of a straight-line trend.
Key test
An exact value; its sign gives direction and its size gives strength.
Formula
−1≤r≤1
Example
Sleep hours and quiz scores with r=0.77: a fairly strong positive trend.

Correlation vs Causation

Meaning
Correlation vs causation is the question of whether two variables that move together do so because one actually changes the other.
Key test
Asks why the variables move together, not just how.
Example
Ice cream sales and swimming rescues rise together because hot weather drives both.

Scatter Plot

Meaning
A scatter plot draws each individual as one dot placed by its two measurements, so the trend can be seen.
Key test
It is the picture you read the correlation from.
Example
Twelve students as dots, hours of sleep across and quiz score up.

Formula & Notation

How to read it: r denotes the sample correlation coefficient. r>0 indicates positive correlation, r<0 indicates negative correlation, and ∣r∣ close to 1 indicates a strong linear relationship.

Worked Examples

Example 1 — Screen time and sleep

Easy

Six students report daily screen time and sleep, in hours: (1, 9.5), (2, 9), (3, 8.5), (4, 8), (5, 7), (6, 6.5). Describe the correlation's direction and strength.
Show the steps and the answer
  1. Order by screen time and read the sleep values: 9.5, 9, 8.5, 8, 7, 6.5.

    Reading one variable in order of the other shows which way the second one moves.

  2. Direction: every step up in screen time comes with less sleep, so the correlation is negative.

    One variable rises while the other falls, which is what a negative correlation means.

  3. Strength: the drops are 0.5, 0.5, 0.5, 1 and 0.5 hours, almost the same each time.

    Steady steps mean the dots lie close to a straight line, which makes the correlation strong.

  4. Phrase the conclusion: students with more screen time tended to sleep less.

    The data show an association; they do not show that the screens caused the lost sleep.

Answer: A strong negative correlation: more screen time goes with less sleep, about half an hour less per extra hour.

Takeaway: Read direction from the sign of the trend and strength from how steady the steps are.

Example 2 — A curve with zero correlation

Standard

A ball is thrown upward. Its height in metres after 0, 1, 2, 3 and 4 seconds is 0, 15, 20, 15 and 0. Is the correlation between time and height positive, negative or zero? Are time and height unrelated?
Show the steps and the answer
  1. Split the data: from 0 to 2 s the height rises (0, 15, 20); from 2 to 4 s it falls (20, 15, 0).

    A correlation looks for one straight-line direction, so check whether the trend keeps a single direction.

  2. Compare the halves: the falling half (20, 15, 0) is the rising half (0, 15, 20) in mirror image, around the peak at 2 s.

    An uphill half and an equal downhill half cancel, so there is no overall straight-line trend and the correlation works out to exactly 0.

  3. Note that each time gives exactly one height, following h=20t−5t2.

    Height is completely determined by time, so the two are strongly related, just not along a straight line.

Answer: The correlation is zero, but time and height are strongly related by a curve that rises and then falls.

Takeaway: Zero correlation rules out a straight-line trend only; always look at the plot for curves.

Example 3 — Find the mistake: negative means weak?

Application

Set A, weekly exercise hours and resting heart rate: (0, 80), (2, 76), (4, 71), (6, 68), (8, 63). Set B, shoe size and math score: (5, 70), (6, 85), (7, 62), (8, 90), (9, 74). Priya writes: "A has a negative correlation, so it is the weaker relationship." Find the mistake and correct it.
Show the steps and the answer
  1. Set A: heart rate falls by 4, 5, 3 and 5 as exercise rises in steps of 2 hours.

    Every step goes the same way by a similar amount, so the dots sit close to a downhill line.

  2. Set B: the math score changes by +15,−23,+28,−16 as shoe size rises.

    The score jumps up and down with no steady direction, so there is little or no correlation.

  3. Separate the two ideas: the minus sign gives A's direction; the tight steps give its strength.

    Priya treated the sign as a measure of strength, but strength depends only on how closely the points follow a line.

Answer: Set A has a strong negative correlation and is the stronger relationship; Set B shows little or no correlation.

Takeaway: Negative describes direction, not weakness; a downhill pattern can be the tightest one in the problem.

Where You'll See It

  • Study time and test scores

    As hours studied go up, test scores tend to go up too — a positive correlation, since both variables move in the same direction across students.

  • Screen time and sleep

    More hours of late-night screen time tends to go with fewer hours of sleep — a negative correlation, where one rises as the other falls.

  • Shoe size and favorite color

    A kid's shoe size has no consistent connection to their favorite color, so a scatter plot would show a shapeless cloud — an example of no correlation.

Common Mistakes

  1. The slip: A correlation means one variable causes the other.

    Why it's tempting: Two things moving together looks like one is driving the other.

    Instead: Correlation only describes that variables move together; causation requires a controlled study and ruling out confounders. State 'associated,' not 'causes.'

  2. The slip: A negative correlation means there is little or no relationship.

    Why it's tempting: The word 'negative' sounds like 'weak' or 'bad.'

    Instead: Negative means the variables move in opposite directions, and that relationship can be very strong. Strength is about how tightly points cluster, not the sign.

  3. The slip: No straight-line pattern means the two variables are unrelated.

    Why it's tempting: Students equate 'correlation' with 'any relationship at all.'

    Instead: Correlation measures the straight-line trend. Variables can have a strong curved (e.g., U-shaped) relationship while showing near-zero linear correlation.


Mini Practice

Try these on your own. Tap Reveal when you want to check.

  1. Is each pair positively correlated, negatively correlated or not correlated? (a) hours of practice and free-throw percentage, (b) outdoor temperature and heating bill, (c) shoe size of adults and their test scores.

    Hint: For each pair, ask what tends to happen to the second variable as the first goes up.

  2. Hours studied 1, 2, 3, 4, 5 go with scores 60, 65, 72, 78, 85. Describe the correlation, and find the average gain in score per extra hour.

    Hint: Compare the change in score with the change in hours from the first to the last pair.

  3. Car ages of 1, 3, 5, 7 and 9 years go with values of $28k, $22k, $17k, $12k and $8k. What kind of correlation is this? Estimate the value of a 4-year-old car.

    Hint: 4 years is halfway between two of the given ages.

  4. Five students' (phone minutes per night, hours of sleep) are (30, 9), (60, 8.5), (90, 8), (120, 7), (150, 6.5). Describe the direction, and say how much sleep drops from 30 to 150 minutes.

    Hint: As phone time goes up, what happens to sleep?

  5. A scientist wonders whether students who sleep more get better grades. She has no numbers yet. Is her next step to find the correlation, or to do data collection?

    Hint: A correlation describes paired data that already exist.

  6. On a scatter plot, 20 points fall steeply in a tight, nearly straight line from upper left to lower right. Max says, "The correlation is negative, so the relationship is weak." Correct him.

    Hint: Direction and strength are two separate features.

Want the full set?

50 practice questions for this concept — free to try, every one with a complete worked solution showing the why, not just the answer.

Frequently Asked Questions

Does a strong correlation mean the dots make a steep line?

No. Strength is about how close the dots stay to a line, not how steep that line is. Dots hugging a gentle slope show a strong correlation; dots spread widely around a steep slope show a weak one. You can even change the steepness without changing the strength: if height is on the vertical axis and you record it in meters instead of centimeters, the slope becomes 100 times smaller, yet the dots hug the line just as tightly.

Can one unusual point change the correlation?

Yes, a lot, especially when there are only a few points. Imagine eight students whose study time and scores show no pattern, plus one student who studied 10 hours and scored 100%. That single dot, far up and to the right, can make the whole cloud look like a positive trend, and removing it makes the trend vanish. Look at the scatter plot and ask whether a far-off point is real before describing the correlation.

If two things are correlated, can I use one to predict the other?

Roughly, yes. If taller students tend to have bigger feet, knowing a student's height gives you a better guess for shoe size than knowing nothing. The stronger the correlation, the closer your guesses will tend to be, but single people still surprise you, since some tall students have small feet. Prediction works even without a cause, which is how insurers use links such as age and accident rates.

Learning Path

You should know Scatter Plot (how paired data become dots) and Data Collection (how the pairs were measured). Correlation names what the dot cloud shows: a direction (positive, negative or none) and a strength (tight or loose). Next, the Correlation Coefficient puts a number r between −1 and 1 on that pattern, and Correlation vs Causation explains why a link is not proof of a cause.

See Also