Information content of survey data : applications of entropy and dissimilarity measures

摘要: This paper evaluates information content of survey data by the means entropy and dissimilarity measures. Similarities between a priori (expectations) posteriori (realizations) are assessed for four variables originating from Polish business tendency analyzed across two ownership sectors size categories. The measures employed include Shannon empirical entropy, Kullback-Leibler relative Chomatowski-Soko³owski coefficient. Results analysis allow to conclude that enterprise does not significantly affect values even though some effects observed in case production general conditions variables. Public private enterprises differentiated interpretation expectations horizon therefore can be studied on aggregated level without loss important forecast patterns. Additionally, previous findings (see E. Tomczyk, 2011) generally confirmed longer sample. Production time series found characterized highest prices – lowest; is also least variable. In public enterprises, concentration answers questions higher more variable than sector.