Toward an empirical analysis of polarization

Stochastic dominance conditions may be interpreted as a means of defining the degree and nature of separateness of two distributions. By describing polarization states in terms of stochastic dominance conditions this paper presents a taxonomy of, and tests for, polarization both between and within population distributions. Four examples from the poverty, growth and development, wage distribution and assortative pairing literatures are used to illustrate how the techniques may be used in empirical applications. (C) 2003 Elsevier B.V. All rights reserved.