Introduction
From Events to Trajectories
Texte intégral
1In recent decades, there has been growing interest in analyzing the biographical trajectories of individuals, driven by changes in residential mobility, family structures, educational pathways, and careers, along with the interdependence of these different spheres of life. In this time, life course analysis has progressively become a major approach in the social sciences and has spurred a shift, from structure to process and from the macro to the micro level (Willekens, 1999, p. 26). This development has been linked both to theoretical considerations and to technical advances in the collection and statistical analysis of longitudinal data. On the collection side, longitudinal data sources have proliferated, in the form of panels and biographical surveys (GRAB, 2009a [1999]). On the methodological side, the development and spread of new statistical techniques for life course analysis has been slow but cumulative and the body of available methods is now very substantial. Since the early 1980s, event history analysis, in particular, has become central in the study of longitudinal data in the social sciences (Allison, 1984; Courgeau & Lelièvre, 1986, 1989; Kalbfleisch & Prentice, 1980; Mayer & Tuma, 1990). These techniques—like Cox’s (1972) famous model—are generalizations of life tables. Ever more sophisticated econometric models (Wu, 2003) offer ways to explore and account for interactions, unobserved heterogeneity, and selection biases (Lillard, 1993; Heckman et al., 1998), as well as different levels of aggregation (Courgeau & Baccaini, 1997; Steele, 2008).
2Event history analysis is an attractive choice, notably because of how well the associated sense of causality matches our experience of life in terms of events unfolding over time (Halpin, 2003): “How long will it take me to get promoted? How would it be different if I were a woman? How much quicker if I get my PhD? if I publish in the AJS?” and so on. Regression models also offer such a homology, but event history analysis has the advantage of incorporating time in a realistic way. Event history models—with their grounding in probabilistic statistics—allow us to avoid a simplistic determinism (Courgeau & Lelièvre, 1992). They allow us to study the interdependencies between the trajectories of the same individual in different domains, such as work and intimate relationships (Courgeau & Lelièvre, 1986), or between potentially linked individuals, such as the members of a couple. And they allow us to combine different levels of aggregation—from micro to macro—in analyzing individual behaviors (Blossfeld & Rohwer, 2002).
3Event history analysis centers on the occurrence (or nonoccurrence) of specific events in the lives of individuals. It models transition (or duration) probabilities, making the assumption that the life course is the result of a complex stochastic process. It is, thus, a parametric approach, the aims of which are mainly explanatory and causal. It is built around events. But the ambition of life course research is also to understand trajectories as a whole. The importance of the trajectory as a theoretical concept has been emphasized since its beginnings (Sackmann & Wingens, 2003). The premise is that rather than studying events independently of each other, we must examine their sequential unfolding. In practice, much of the empirical work on life courses in the social sciences uses methods that focus on transitions. But a different approach is possible, with the whole trajectory taken as a unit of analysis.
4Billari (2001) identifies two viewpoints to motivate the adoption of a “‘holistic’ perspective that sees life courses as one meaningful conceptual unit” (p. 440). According to the first, “strong” viewpoint, trajectories result from the life projects of individuals, understood, for example, as utility maximizers. In this type of view, individuals themselves adopt a holistic perspective as they plan their future life. In doing so, they consider their life trajectory as a whole.
5The second, “pragmatic” viewpoint is based instead on the idea that the life course as a conceptual unit is the contingent result of a sequence of events. Here, a holistic approach to life courses is intended to allow us to describe and summarize the timing and sequence of events, as well as the amount of time that individuals spend in different states and that separates different events (Settersten & Mayer, 1997). The importance of the order of events has also long been highlighted in the literature (Hogan, 1978; Marini, 1984; Rindfuss et al., 1987). For example, the likelihood that an individual will get married and its determinants will differ depending on whether the potential marriage is before or after they have left formal education. Moreover, many events mark reversible transitions (e.g., marriage), while others can be hard to precisely define and have “fuzzy” timing (GRAB, 2009b). Moving out from the parental home, for example, is better understood as a complex process than as a sharp, momentary transition (Diagne, 2006; Villeneuve-Gokalp, 1997).
6A holistic approach to life courses can facilitate the exploration of individual longitudinal data, the discovery of “hidden structures” (Roux, 1993), and the reduction of complexity, in particular through typologies. Unlike event history analysis, the methods that take this approach are most often nonparametric: They do not assume that the life course is a stochastic process and are, instead, part of an “algorithmic” statistical culture (Breiman, 2001). Their main objective is to explore and describe life courses, identifying regularities and differences among them.
7In practice, there are numerous ways of describing trajectories. They can be applied to varied research contexts and topics—urban integration in West Africa, residential mobility in South America, marital and reproductive histories, the use of time in everyday life, and more—including at nonindividual levels (e.g., the trajectories of countries, businesses).
8This book, which is partly derived from a first edition published in 2011 (Robette, 2011), is largely devoted to the construction of typologies of trajectories—the most common approach to dealing with trajectories. After an overview of this approach and the series of decisions that it requires, from how to code data to choosing the number of clusters in the typology, we go on to explore how the results can be characterized. Then, we survey methods for measuring the similarity between trajectories—a crucial step on the way to obtaining a typology. The construction of a typology is illustrated using an application to employment trajectories. We also look at a few complementary approaches.1
9While the methods presented are not trivial to implement, they are more accessible than it might seem, and they can prove highly useful—particularly in cases in which the richness of the data or the complexity of the trajectories to be studied prevent organization “by hand” or based on simple criteria.
Tutorial
10A tutorial is available for implementing many of these methods in the R software environment. It is built into the vignette for the seqhandbook package, which can be consulted on the Nakala platform in English and French using this permanent link:
Notes de bas de page
1 This review of methods is not intended to be exhaustive. The hope is to offer the reader who is interested in using them a broad overview of existing possibilities that are (relatively) easy to put into practice.
Le texte seul est utilisable sous licence Licence OpenEdition Books. Les autres éléments (illustrations, fichiers annexes importés) sont « Tous droits réservés », sauf mention contraire.
États flous et trajectoires complexes
Observation, modélisation, interprétation
Philippe Antoine et Éva Lelièvre (dir.)
2006
Biographies d’enquêtes
Bilan de 14 collectes biographiques
GRAB (Groupe de réflexion sur l'approche biographique) (dir.)
2009
Méthodes de mesure de la mobilité spatiale
Migrations internes, mobilité temporaire, navettes
Daniel Courgeau
2021
L’analyse statistique des trajectoires
Typologies de séquences et autres approches
Nicolas Robette
2021
Fuzzy States and Complex Trajectories
Observation, modelisation and interpretation of life histories
GRAB Harriet Coleman (trad.)
2009
Minorités de genre et de sexualité
Objectivation, catégorisations et pratiques d’enquête
Wilfried Rault et Mathieu Trachman (dir.)
2023
The Statistical Analysis of Trajectories
Typologies of Sequences and Other Approaches
Nicolas Robette Paul Reeve (trad.)
2025
