A measure of the peakedness of a distribution. The measure (usually denoted by β2 or κ), was introduced by Karl Pearson before 1905. The measure is given by ![kurtosis](Images/oree/doc/10.1093/acref/9780199679188.001.0001/acref-9780199679188-math-0321-full.gif)
where μ4 is the fourth central moment of the distribution and μ2 is the variance. It is invariant under a change of scale or origin.
An alternative definition uses (β2−3) to give a value of 0 for the normal distribution. With this definition, a distribution having negative kurtosis is described as being platykurtic (flatter), and a distribution having positive kurtosis is described as being leptokurtic (more peaked). A distribution having kurtosis zero is described as mesokurtic.