Open AccessUsing Event-history Analysis: Lessons from Fifteen Years of Practice
Author(s) -
Céline Le Bourdais,
Jean Renaud
Publication year - 2001
Publication title -
canadian studies in population
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 0.157
H-Index - 8
eISSN - 1927-629X
pISSN - 0380-1489
DOI - 10.25336/p6tc7w
Subject(s) - event (particle physics) , test (biology) , data science , immigration , computer science , history , paleontology , physics , archaeology , quantum mechanics , biology
Innovative statistical methods and new longitudinal surveys paved the way tothe widespread use of event-history analysis in social science during the last two decades. This paper does not attempt to provide a comprehensive review of these innovative methods. More modestly, it aims at identifying and describing the problems encountered by two privileged users. Two types of problems are discussed here. The first arises from the design of the surveys, or the way data are collected, and the difficulty to test specific hypotheses with the existing databases; this is the kind of problem that Le Bourdais has faced in analysing family dynamics. The second has to do with the limitations of the survival regression models when the longitudinal phenomena studied can no longer properly be thought of as a small number of unique events; this is the type of problem encountered by Renaud in his ten-year Quebec panel survey of new immigrants.