In this paper, Guillaume Le Roux, Matthias Studer, Arnaud Bringé and Catherine Bonvalet propose a sequence analysis-based method for selecting qualitative cases depending on quantitative results. Inspired by tools developed for cross-sectional analyses, they propose indicators suitable for longitudinal study of the life-course in a holistic perspective as well as a set of corresponding analysis guidelines. Two complementary indicators are introduced, marginality and gain, that allows labelling observations according to both their typicality within their own group and their illustrativeness of a given statistical relationship. They allow selecting a diversity of cases depending on their contribution to a given statistical relationship between trajectories and a covariate or a typology.
The method and its advantages are illustrated through an original study of the relationships between residential trajectories in the Paris region and residential socialization during childhood. Using the Biographies et Entourage [Event history and entourage] survey and qualitative interviews conducted with a subsample of the respondents, the analysis shows the contributions of the method not only to improve the understanding of statistical associations, but also to identify their limitations. Extension and generalization of the method are finally proposed to cover a wider scope of situations.