Pearson and Spearman correlation calculator
Paste two columns. You get both correlations on the same data — and notice of when they disagree, which is the moment the choice matters.
Two correlations, not one
Pearson's r measures a linear relationship. Spearman's ρ measures a monotone one — whether Y tends to rise as X rises, without requiring that it rises by a constant amount. They answer different questions, and the same data can give very different answers to each.
A worked example where one outlier decides it
Twelve points lying almost exactly on a rising line. Then the first value of Y is changed from 2.4 to 40 — a single data point.
| Pearson r | Spearman ρ | |
|---|---|---|
| As entered | 0.997 (p < .001) | 1.000 |
| One Y changed to 40 | 0.188 (p = .559) | 0.538 (p = .071) |
Pearson collapses from a near-perfect relationship to nothing. Spearman falls too, but stays positive because the ordering is largely intact. Neither is wrong: the outlier genuinely destroys the linear relationship, and does not destroy the ordering.
This is the case to check before reporting r. A scatter plot costs ten seconds.
Which one to report
- Pearson when both variables are continuous, the relationship looks linear, and there are no extreme values — or when you need a confidence interval.
- Spearman when the data are ordinal, when there are outliers you cannot justify removing, or when the relationship is clearly monotone but curved.
Reporting both is normal, not hedging. When they agree, the choice of test does not matter and you can say so.
Correlation is not a claim about cause
A significant r says two variables move together. It says nothing about direction, and nothing about a third variable that moves both. The sample size needed to detect a modest correlation is larger than most people expect: around 85 pairs for r = .30 at 80% power.
What to report
r(10) = .997, p < .001, 95% CI [.980, .999] — always with the degrees of freedom or the sample size, and with the confidence interval when it is available.
Frequently asked
My r is significant but rho is not. Which is right?
Both are. They test different null hypotheses. Look at the scatter plot: if the relationship is linear but has outliers, Pearson is fragile; if it is monotone but curved, Spearman is the better description.
Does the confidence interval exist for Spearman?
Yes, but it requires a bootstrap. This page reports the point estimate and p value for ρ only.
What if my variables are ordinal?
Use Spearman. Applying Pearson to Likert items treats the spacing between categories as equal, which is usually not justified.
Is my data uploaded?
No. The calculation runs in JavaScript on this page.
See it in a table
The app computes the same correlations from a data file and exports a three-line table with the confidence interval already formatted.
Which test should I use? · t test calculator · chi-square calculator · ANOVA calculator · correlation calculator · p value calculator