Stochastic monotonicity of dependent variables given their sum (Q2161026)

Given random variables \(X_1,\ldots,X_n\) and their sum \(S=X_1+\cdots+X_n\), the authors establish conditions for stochastic monotonicity of the \(X_i\) conditional on the value of \(S\). Monotonicty results are expressed in terms of the univariate and multivariate usual stochastic order, and likelihood ratio order. Special attention is given to the bivariate case \(n=2\). These conditions are given in terms of the joint density or distribution function of the \(X_i\), and importantly do not assume that the \(X_i\) are independent. The authors also use their results to establish analogous stochastic monotonicity properties of \(S\) conditioned on \(X_1\).