A correlation measures the strength of a statistical link between two variables.
Given a positive correlation, the following statement applies: 'the more of variable A….the more of variable B', or vice versa. Given a negative correlation, that statement would be 'the more of variable A…. the less of variable B' or vice versa.
As an example, there is a negative correlation between the variable 'actual age' and 'remaining life expectancy'. The higher the actual age, the lower the average remaining life expectancy will be.
Correlations are always nondirectional, i.e. they contain no information on which variable caused another – both variables are equal. The strength of the statistical relation is expressed by the correlation coefficient, which lies between -1 and +1. The type of a directional relationship is described in a regression.
Correlations, however, are an indication for, but not a prove of causalities, i.e. proven relationships of cause and effect.
An example: The fact that older people more often possess expensive jewelry than younger people do, is not necessarily due to the number of years of their lives, nor to their tastes or interests. The reason might just as well simply be their context of a higher income, which older people usually have.
See also autocorrelation and spurious correlation.
Tenga en cuenta que las entradas de nuestro glosario son explicaciones simplificadas de términos estadísticos. Nuestro objetivo es hacerlo accesible para un público amplio, así que puede que algunas definiciones no cumplan los estándares científicos.