Strengthening Sequence Analysis

Sequence analysis is one of the key approaches to study processes and trajectories from a life-course perspective. It provides a holistic view of trajectories by creating a typology that can be then used in subsequent analyses or simply to describe these trajectories. Despite its increasing uses in several disciplines, sequence analysis still faces several long-standing issues and limitations. This research project aims to address them to consolidate social science research making use of the methodology. More precisely, we aim to:•Develop a robust clustering and validation framework for sequence analysis that can properly handle weakly structured data or atypical trajectories and avoid sample dependence of the results.•Extend sequence analysis to handle large databases, which are increasingly common, by adapting typology creation and validation methods.•Conduct a critical theoretical and empirical simulation-based review of available cluster algorithms grounded on life-course relevant aspects before issuing clear recommendations to sequence analysis users.•Develop a proper methodological framework to study the relationship between trajectories and covariates to avoid drawing wrong conclusions linked to a) simplification implied by the use of a typology instead of individual sequences and b) estimation errors of a typology that can be expected when working with sample data.•Develop a proper framework to handle missing data in sequence analysis in conjunction with multiple imputation.•Review missing data-handling methods and document their respective strengths and weaknesses for missing data patterns commonly encountered in life-course research, such as in panel or retrospective data.•Demonstrate the added values of each developed method through convincing studies on school-towork transitions in Switzerland, first before, and then during, the COVID-19 crisis using the large LABB administrative database.•Diffuse all the reviewed and developed methods by making them available in widely used R libraries and by writing user-oriented documentation.