Title: Compositional analysis in tourism
Authors: Berta Ferrer-Rosell - Universitat de Lleida (Spain) [presenting]
Germa Coenders - University of Girona (Spain)
Abstract: Compositional Data analysis (CoDa) is the standard statistical method when data contain information about the relative importance of parts of a whole, typically but not necessarily with a fixed sum. Many research questions in the field of tourism have to do either with distribution of a whole (e.g., share or allocation), or with relative importance (e.g., dominance, concentration, profile, etc.). Some examples of such research questions might be: What are the determinants of market share of a given destination or tourism product? Or, which origins and destinations concentrate the most tourist flows, per season or tourist segment? The main purpose is to present the manner in which CoDa solves statistical problems that arise (e.g. spurious correlations) when treating compositional data with classical statistical methods, as well as, to show how to apply the most common exploratory tools to analyse compositions (data transformations, distances, CoDa biplot and cluster) by means of real examples.