Professor Daniel Courgeau is an engineer graduated from l'Ecole Polytechnique de France and PhD in Demography. From 1965 to 1987 he was the head of the Seita commissioned at the Institut Nacional d'Etudes Démographiques (INED). In 1988 he became director of research at INED and from 2002 he was appointed Research Director Emeritus. His work is very extensive both in the field of scientific research and teaching and dissemination which has earned him wide national and international recognition. The topics covered in it focus on four well-defined areas: the use of a space-time perspective; The creation of a new paradigm in demographic studies that transcends the foundations of classical demography by moving from an analysis of groups of individuals to that of individual biographies; Multilevel analysis; And, epistemological reflection. The methodological contribution of all these studies have appeared in numerous books, articles translated into several languages and in the applications, in several countries, of biographical type surveys (Three b). The methods of biographical analysis have allowed us to introduce a contextual and multilevel character that works simultaneously with different levels of information aggregation in order to better understand the behaviours of individuals. At present, his reflections revolve around the main methodological and epistemological questions that arise in the social sciences, behavioural genetics and Bayesian approximations.
Among his main publications are: "Méthode de mesure de la mobilité spatiale" (INED, 1988); "L'analise démographique des biographies", co-authored with Eva Lelièvre (INED, 1989), an English version of "Event history analysis in demography" (Clarendon Press, Oxford, 1992), Spanish version "Demographic analysis of biographies" Colegio de México, 2001); "Methodology and epistemology of multilevel analysis" (Ed., Kluwwe Academy Publishers, 2003); "Du groupe à l'individu: synthèse multiniveau" (Ed., INED, 2004); Avec Eva Lelièvre "A nouveau paradigm in Démographie" (Revue Population, INED 1996-3); And, "Nouvelles approches méthodologiques en sciences sociales" (Ed., Revue Population, INED, 1997-4). And other recent books such as "Multilevel synthesis" in 2007, and "Probability and social science" also published by Springer in 2012.
I will show in this presentation how I went from surveys to paradigms, leading finally to a more general philosophy of social science during the last fifteen years. I undertook during the 1970’s social network surveys in order to understand the migration process, and they were also used for network analysis. During the 1980’s I undertook one of the first event history survey to understand the internal logic of family, professional and migratory events. Lastly during the 1990’s I reconstituted from Norwegian population registers and censuses the life of two exhaustive generations in order to understand how multilevel analysis may extend demography. At the end of the 1990’s after an attack against event history analysis, I tried to give a paradigmatic view of the history of demography. From Graunt to the end of World War II a cross-sectional aggregate paradigm was the norm, introducing regression analysis. At the end of World War II an alternative vision of time permitted a new longitudinal paradigm to take place. However at the beginning of the 1980’s some problems encountered with this last paradigm lead to an event history one permitting to introduce time regression methods. Finally at the end of the 1980’s the necessity to introduce a more complex society with its different levels, leads to multilevel network analysis. During the last fifteen years I went towards a deeper reflexion about epistemology, methodology, and philosophy of social science. Agent-based models even if they lead, by their bottom-up approach to, appealing properties, are shown to be not sufficiently grounded. There is finally a need to introduce a model based program, complementing the four extant paradigms while incorporating insights gained from model-based science. This program must conform the classical scientific research programme which tries to infer the functional structure which rules the process generating a studied property. Such an approach provides a stronger basis for demography.