Score function of distribution and revival of the moment method. Issue 4 (16th February 2016)

• Record Type: Journal Article
• Authors: Fabián, Zdeněk
• Is Part Of: Communications in statistics. Volume 45:Issue 4(2016)
• Page Start: 1118
• Page End: 1136
• Publication Date: 2016-02-16
• Subjects: Characteristics of distributions -- Data characteristics -- General moment method -- Huber moment estimator -- Parametric methods -- Score function.; 62A01, 62F10; Mathematical statistics -- Periodicals; Mathematics; Statistics; 519.2
• DOI: 10.1080/03610926.2013.857688
• Languages: English
• ISSNs: 0361-0926

Abstract

Abstract: The article deals with the scalar-valued score function SF defined for a regular unimodal and continuous probability distribution F with arbitrary interval support, recently introduced by the author. The concept of the central characteristic that describes a relative influence of is described in a much more general way in order to improve its usefulness in parametric estimation by means of the general moment method. In particular, the whole approach is elucidated by describing it in a more suitable framework than before. Further, we show that the inference function to be used in the moment estimating equations is either the score function of distribution (bounded for heavy-tailed parametric distribution models) or its modification based on the Huber's approach.