This paper o§ers a novel approach to solving Bayesian persuasion games based on applying multivariate stochastic orders to ranking signal structures. By using results in theory of positive dependence stochastic orders and the majorization theory, we identify the class of problems, in particular, the weighted Schur convex problems, for which optimal signal structures are monotone partitional. This allows us to characterize optimal signal structures in various applications of Bayesian persuasion games: information provision by a seller; consumers protection; and strategic communication.