Squared skewness minus kurtosis bounded by 186/125 for unimodal distributions (Q1590559): Difference between revisions

Let \(F\) be a non-degenerate distribution with finite fourth moment. The coefficient of kurtosis of \(F\) is denoted by \(\kappa_F\), while the coefficient of \(F\) skewness is denoted by \(\tau_F\). The Pearson inequality states that: \[ \tau^2_F\leq \kappa_F+2.\tag{P} \] The purpose of this mathematical note is to improve the Pearson inequality (P). A distribution function \(F\) is unimodal if it is convex-concave. If \(F\) is linear on its support with an atom at one of its boundary points, \(F\) is called a one-sided boundary-inflated uniform distribution. The main result of the paper can be formulated as follows: If \(F\) has a non-degenerate unimodal distribution with finite fourth moment, then the Pearson-type inequality \[ \tau^2_F\leq \kappa_F+{186\over 125}\tag{P}\('\) \] holds, with equality iff \(F\) is a one-sided boundary-inflated uniform distribution with mass \(1/2\) at the atom.