Statistics Glossary → Pearson correlation coefficient
Pearson correlation coefficientconcept
A coefficient between -1 and +1 that measures the direction and strength of the linear relationship between two numerical variables.
r is the covariance of the two variables divided by the product of their standard deviations: r = cov(x, y) / (s_x · s_y). +1 means a perfect positive and -1 a perfect negative linear relationship, while 0 means no linear relationship; r² is the share of variance in one variable explained linearly by the other. The null hypothesis ρ = 0 is tested with t = r √(n - 2) / √(1 - r²) on a t distribution with n - 2 degrees of freedom; the confidence interval is built with Fisher's z transformation.
r measures only linear relationships: for a strong but curved (e.g. U-shaped) relationship it can be close to zero. It is very sensitive to outliers; a single extreme point can change r substantially. If the range of one variable is restricted, r shrinks. For monotonic but nonlinear relationships or for ordinal data, Spearman or Kendall correlation is more suitable. And, as always, correlation does not imply causation.
An everyday example
For a group of students, the correlation between weekly study hours and exam scores is r = 0.60. Scores tend to rise with study time; r² = 0.36, so about 36% of the variation in scores can be linked linearly to study hours. The rest depends on ability, sleep and other factors.
In MerQur
The 'Method' option of Correlation Analysis (Pearson, Spearman, Kendall τ) and 'Correlation type' in Bayesian Correlation. Pearson correlation distance is one of the choices of 'Distance measure' in Hierarchical Clustering and of 'Distance metric' in t-SNE and UMAP; in Multidimensional Scaling, 'MDS of variables (1 - r)' positions the variables with the 1 - Pearson r distance.
Analyses where this concept appears
Correlation AnalysisHierarchical Clusteringt-SNE (t-Distributed Stochastic Neighbor Embedding)Multidimensional Scaling (MDS)UMAP (Uniform Manifold Approximation and Projection)Bayesian Correlation
Related concepts
Spearman rank correlationKendall's tauFisher z transformation (correlation confidence interval)