Statistics Glossary → Spearman rank correlation
Spearman rank correlationconcept
A correlation coefficient that measures the strength and direction of a monotonic relationship between two variables using the ranks of the values rather than the values themselves.
Spearman's ρ is the Pearson correlation computed on the ranks of the two variables. Without ties it can also be found as ρ = 1 − 6Σdᵢ² / [n(n² − 1)], where dᵢ is the difference between an observation's ranks on the two variables. It lies between −1 and 1. A value of 1 means that whenever one variable increases the other always increases too; the relationship need not be linear.
Pearson's correlation measures linear association and is sensitive to outliers. Because Spearman's is based on ranks, it is less affected by outliers and can also be used with ordinal (e.g. Likert) data. It cannot, however, detect non-monotonic relationships such as a U shape. Kendall's τ serves the same purpose; it usually gives smaller values and may be preferred in small samples and when there are many ties. Correlation, however it is measured, does not establish causation.
An everyday example
Two judges rank 10 contestants in a competition. If their rankings are almost identical, Spearman's ρ is close to 1; if the one ranked first by one judge is ranked last by the other, the value moves toward negative. Since ranks are compared rather than the scores themselves, one judge giving everyone higher marks does not change the result.
In MerQur
In Correlation Analysis it is one of the methods, alongside Pearson and Kendall's τ. In the bootstrap CI it is among the statistics for which an interval can be computed, and in Bayesian correlation a Bayes factor can be computed for it.
Analyses where this concept appears
Bootstrap CICorrelation AnalysisBayesian Correlation