Stochastic comparisons in multivariate mixed model of proportional reversed hazard rate with applications (Q2267598)

The authors study \(n\)-dimensional random vectors \(\mathbf{T}\) that have distribution functions of the form \[ F(t_1,t_2,\dots,t_n)=\int_{\mathbb{R}_+^n}\prod_{i=1}^nF_{0i}^{v_i}(t_i)h(v_1,v_2,\dots,v_n)dv_1dv_2\cdots dv_n,\quad(t_1,t_2,\dots,t_n)\in\mathbb{R}_+, \] where \(F_{0i}\), \(i=1,2,\dots,n\), are some fixed univariate distributions, and \(h\) is a density function of an \(n\)-dimensional random vector \(\mathbf{V}\). Let \(\mathbf{T}_1\) and \(\mathbf{T}_2\) be two random vectors of the type above, that are associated with the corresponding random vectors \(\mathbf{V}_1\) and \(\mathbf{V}_2\). The authors derive results of the form: If \(\mathbf{V}_1\leq_{st1}\mathbf{V}_2\) then \(\mathbf{T}_1\leq_{st2}\mathbf{T}_2\), where \(\leq_{st1}\) and \(\leq_{st2}\) are two specific stochastic orders that vary from one result to another. The authors also identify conditions that yield that \(\mathbf{T}\) has some multivariate aging properties such as multivariate increasing failure rate (IFR) or new better than used (NBU). Further stochastic monotonicity results are obtained as well.