A correlation coefficient is a number between -1.00 and 1.00 that measures how strongly two variables are related to each other. In psychology, researchers use it to quantify patterns they observe, like whether people who sleep less tend to have lower grades, or whether higher stress levels go hand in hand with more health complaints. The number itself tells you two things: the direction of the relationship (positive or negative) and how strong it is.
How the Number Works
The scale runs from -1.00 to +1.00, with zero in the middle. A correlation of zero means there is no linear relationship between two variables. A correlation of 1.00 or -1.00 means the relationship is perfect: if you know one variable, you can predict the other exactly. In real psychological research, perfect correlations essentially never happen because human behavior is influenced by too many factors at once.
The sign tells you the direction. A positive correlation means the two variables move together: as one increases, the other tends to increase too. Height and weight, for instance, are positively correlated. A negative correlation means they move in opposite directions: as one goes up, the other tends to go down. Sleep and daytime tiredness are negatively correlated, because the less someone sleeps, the more tired they tend to feel during the day.
Some pairs of variables have no meaningful relationship at all. Hours of sleep and shoe size, for example, would produce a correlation near zero. Knowing someone’s shoe size tells you nothing useful about their sleep habits.
What Counts as Small, Medium, or Large
Psychologists commonly use benchmarks proposed by the statistician Jacob Cohen to describe the strength of a correlation. These are rough guidelines, not hard rules, but they give you a shared vocabulary for interpreting results:
- Small: r = 0.10. There is a real relationship, but it is slight. You would need a large sample to detect it reliably, and it explains only about 1% of the variation between individuals.
- Medium: r = 0.30. The relationship is moderate and noticeable. In practical terms, this is the kind of correlation that a careful observer might pick up on even without running the numbers.
- Large: r = 0.50. The relationship is strong. The two variables share a substantial amount of overlap, and knowing one gives you a genuinely useful prediction of the other.
These same thresholds apply to negative correlations. An r of -0.50 is just as strong as +0.50; the only difference is the direction. To put a real study in context, researchers at the University of Minnesota found a correlation of r = -0.29 between the number of days students slept fewer than five hours per week and their GPA. By Cohen’s benchmarks, that is close to a medium-sized relationship: students who regularly got less sleep tended to have somewhat lower grades, but sleep alone didn’t determine academic performance.
Pearson’s r vs. Spearman’s rho
When psychologists say “correlation coefficient,” they usually mean Pearson’s r, which measures the strength of a straight-line (linear) relationship between two variables. Pearson’s r works well when both variables are measured on a continuous scale (like test scores or reaction times) and the data doesn’t have extreme outliers pulling the values in odd directions.
Spearman’s rho is the main alternative. Instead of working with the raw numbers, it ranks all the data points from lowest to highest and then calculates the correlation on those ranks. This makes it more resistant to outliers and better suited for data with skewed distributions, which is common in psychological research. Survey data, for example, often clusters at one end of the scale and has a long tail of extreme responses. In those cases, Spearman’s rho tends to give a more stable and accurate picture of the relationship than Pearson’s r.
If your data is roughly symmetrical and free of extreme values, Pearson’s r will generally be slightly more precise. If your data has heavy tails or you’re working with ranked categories (like “strongly disagree” through “strongly agree”), Spearman’s rho is the better choice.
Why Correlation Does Not Mean Causation
This is the single most important thing to understand about correlation in psychology. When two variables are correlated, there are three possible explanations, and the correlation alone cannot tell you which one is correct:
- A causes B. Changes in the first variable directly produce changes in the second.
- B causes A. The causal arrow runs in the opposite direction from what you might assume.
- A third variable causes both. Something you haven’t measured is driving both variables to move together, creating the appearance of a direct link.
That third option is what researchers call the “third variable problem” or a confounding variable. Consider the finding that people with more tattoos tend to have different income levels. It would be a mistake to conclude that tattoos cause income differences or that income drives tattoo decisions. A more likely explanation is that other factors, like peer influence, personality traits, or career path selection, independently affect both how many tattoos a person gets and how much money they earn.
Here’s another example: louder bars sell more beer. That’s a real positive correlation. But turning up the music doesn’t cause people to drink more (at least, not directly in the way the numbers suggest). The confounding variable is crowd size. A packed bar is both louder and sells more drinks, so the two variables rise together without one causing the other.
The only way to establish causation is through a controlled experiment where the researcher manipulates one variable and holds everything else constant. Correlational studies, which simply observe variables as they naturally occur, are powerful for identifying patterns and generating hypotheses, but they can’t prove that one thing causes another.
Where You’ll See It in Psychology
Correlation coefficients show up constantly in psychological research because many of the questions psychologists care about can’t be studied through experiments. You can’t randomly assign people to experience childhood trauma to see if it correlates with adult anxiety. You can’t force half your participants to sleep three hours a night for a year to study the effects on cognition. In these cases, correlational research is the ethical and practical approach: measure both variables in a group of people and calculate the relationship.
Common examples include the relationship between self-esteem and academic performance, screen time and attention span, socioeconomic status and mental health outcomes, and parental involvement and children’s test scores. In each case, the correlation coefficient gives researchers a precise, standardized number to describe what they found, making it easy to compare results across different studies and populations.
When you encounter a correlation coefficient in a research summary or a psychology class, focus on three things: the sign (positive or negative), the size relative to Cohen’s benchmarks, and whether the authors are careful to avoid claiming causation from correlational data. Those three checks will tell you most of what you need to know about what the finding actually means.