Additivity and conditional information for parametric measures of information (Q1089691)

Several authors (Fisher, 1925; Mathai, 1967; Vajda, 1973; Boekee, 1977) have suggested parametric measures of information. Papaioannou and Kempthorne (1971) and the author and \textit{T. Papaioannou}, Inf. Control 51, 193-208 (1981; Zbl 0524.62005), have further defined some one- dimensional parameter measures of information when the considered parameter is vector-valued. This paper deals with the study of certain interesting properties for such information measures: i) the additivity properties (strong additivity, weak additivity and subadditivity) and their connection with the independence inequality [as intended by \textit{A. El-Sayed}, ibid. 35, 229-245 (1977; Zbl 0367.94042)]; ii) the conditional information inequality, along with the assumptions under which it holds, and some connections with the additivity properties. The relevance of the last inequality in statistical inference, mathematical statistics and communication theory is then pointed out.