Conferences

Edwin Diday and Monique Noirhomme-Fraiture edited the book “Symbolic Data Analysis and the SODAS Software” that was published in the Wiley
series in 2008. The book opens a new approach to analyse data with complex units, called symbolic objects. Such objects can be of various
types but in this paper, I use countries on one hand, and countries by gender on the other. The variables in symbolic analysis can be more
complex than in classic data analysis. In this paper, I use two types of variables, that is, intervals and modals. The former ones are such
intervals that well describe the phenomenon concerned, e.g. certain quantiles of a micro variable, or confidence intervals, respectively.
The modal variables are in survey context factually the frequency distributions of micro variables. So, symbolic analysis is a certain
type of aggregate approach that is better than the corresponding classic analysis since less information will be lost. The paper
presents this approach in more details but its major part concentrates on the multivariate analysis of 32 European Social Survey countries.
Two types of variables will be used in empirical examples: people’s values and political/trust variables, respectively. Several
illustrations are given, using zoom graphics, classification trees and dissimilarity measures. A compact result is given in the form of
bi-dimensional mapping. This shows nicely how close or far the symbolic objects (32 countries or 64 countries by gender) are from
each other.