ABSTRACT
Means-end analysis has greatly developed in recent years, notably under the influence of the work of Reynolds and Gutman (1988) in the United States, and the researches of Valette-Florence and his colleagues in France. These recent studies have paid particular attention to the proposition of methodological improvements in the obtaining of resultant chains from the synthesis of individual ladders. In this connection, methods using multiple correspondence analysis and graph theory have ceded to more pertinent analyses, based on constrained clustering of chains. In this research proposes an additional approach in order to define latent chains with a structural equations model based on the method of partial least squares (PLS) performed on categorical variables. Based on an empirical study related to cigarette consumption, this work describes the different methods currently used, and particularly emphasizes the proposed methodology. Finally this research explores the various results of the analysis and puts the stress on the related implications in advertising strategies.
