Minimum disparity estimators for discrete and continuous models. (Q1771843)

The paper presents a concept of minimum disparity estimators. The idea starts with \(D_{\varphi }(p;q)=\sum _{i=1}^nq_i\varphi(p_i/q_i)\), called \(\varphi \)-disparity of \(p\) and \(q\), provided \(\varphi \) is unimodal with zero-minimum in \(1\), two-times differentiable, and convex about \(1\). The authors are proving strong consistency of minimum \(\varphi \)-disparity estimators. The explained theory is supplied with examples, and applications to discrete and continuous models are also discussed.