History matters: the statistical modelling of the life course

Life course studies have the ambitious goal of relating one portion of the life history to subsequent outcomes, which may themselves be later fragments of the overall history. These studies may also wish to relate certain domains, or dimensions of the life course to others, as they each evolve over time. How can these partial histories be operationalised and incorporated into statistical models? Does the process of doing this require additional assumptions or even limit the types of statements we can make about individuals, their life choices and subsequent prospects? From what social theory can we draw when making these choices? Beginning with an abstract framing of this fundamental problem in social science research, we connect three very different statistical models for life course outcomes to theoretical models based in the social sciences. We show how the choices surrounding historical context have deep implications for interpretation through their connection to theoretical frameworks in life course research. We demonstrate that these models inform individual-specific and population-average interpretations of correlates of change and their corresponding life course pathways. Each approach contributes a unique perspective upon which a more comprehensive narrative may be constructed. We illustrate the models and their interpretation using co-residence information in the transition to adulthood using the Swiss Household Panel.