All three correlation measures are a standardized (between -1 and 1) of the strength and direction of the relationship between two variables. The three types differ in the type of relationship that they measure.
- Pearson, the most commonly used correlation, measures linear relationships between variables, and thus assumes that each variable is normally distributed.
- Spearman measures monotonic relationships. It is rank-based, and therefore does not assume normality nor linearity and can do a better job at representing non-linear relationships than Pearson correlation.
- Kendall correlation is the more general than Pearson and Spearman correlation in that it does not assume normality, linearity, or monotonicity. It tends to perform better than Spearman if the sample size is small.