On usual multivariate stochastic ordering of order statistics from heterogeneous beta variables (Q2443262)

For positive parameters \(\alpha_1\), \(\alpha_2\), \(\gamma_1\), \(\gamma_2\), let \(X_i\sim\text{beta}(\alpha_i,1)\) and \(Y_i\sim\text{beta}(\gamma_i,1)\), \(i=1,2\), be independent random variables. Denote the corresponding order statistics by \(X_{1:2}\leq X_{2:2}\) and \(Y_{1:2}\leq Y_{2:2}\). Suppose that \((\alpha_1,\alpha_2)\succeq(\gamma_1,\gamma_2)\), where \(\succeq\) is the Schur-majorization relation. The authors then show that \((X_{1:2},X_{2:2})\leq_{\mathrm{st}}(Y_{1:2},Y_{2:2})\), where \(\leq_{\mathrm{st}}\) denotes the bivariate ordinary stochastic order. An extension of this result to the case of proportional reversed hazard rate is given.