Identities in order statistics – from moments to arbitrary distributions

Abstract

Linear identities for moments of order statistics from an independent and identically distributed sample are defined. It is shown that such identities are true for all distributions if and only if it is true for Bernoulli distributions. It is also shown that such identities hold for all symmetric distributions if and only if they hold for symmetrized Bernoulli distributions.

Keywords:

• Order statistics
• Linear identities
• Bernoulli distribution
• Symmetrized Bernoulli distribution

Notes

^*E-mail: [email protected]

Additional information

Notes on contributors

K. Balasubramanian*

*E-mail: [email protected]