A “Global Interdependence” Approach to Multidimensional Sequence Analysis

Abstract

Although sequence analysis has now become a widespread approach in the social sciences, several strategies have been developed to handle the specific issue of multidimensional sequences. These strategies have distinct characteristics related to the way they explicitly emphasize multidimensionality, interdependence, and parsimony. In this context, the authors introduce an original approach based on structural links between the dimensions, combining optimal matching analysis, multidimensional scaling, canonical partial least squares, and clustering, an approach the authors call globally interdependent multiple sequence analysis (GIMSA). The authors then apply GIMSA to mother-daughter employment histories in France and discuss the value of this method.