We survey state-of-the-art approaches to study trajectories in their entirety, adopting a holistic perspective, and discuss their strengths and weaknesses. We begin by considering sequence analysis (SA), one of the most established holistic approaches. We discuss the inherent problems arising in SA, particularly in the study of the relationship between trajectories and covariates. We describe some recent developments combining SA and Event History Analysis, and illustrate how weakening the holistic perspective-focusing on sub-trajectories-might result in a more flexible analysis of life courses. We then move to some model-based approaches (included in the broad classes of multistate and of mixture latent Markov models) that further weaken the holistic perspective, assuming that the difficult task of predicting and explaining trajectories can be simplified by focusing on the collection of observed transitions. Our goal is twofold. On one hand, we aim to provide social scientists with indications for informed methodological choices and to emphasize issues that require consideration for proper application of the described approaches. On the other hand, by identifying relevant and open methodological challenges, we highlight and encourage promising directions for future research.
Keywords: Cluster analysis; Mixed latent Markov models; Multistate models; Sequence analysis; Trajectories.
Copyright © 2018 The Authors. Published by Elsevier Ltd.. All rights reserved.