Strengths and uncertainties of the French annual census surveys

Guy Desplanques  By the same author

Resume

AbstractSince 2004, Insee has implemented a new population census method based on an annual sample survey of nearly nine million people. The first detailed results will be produced after completion of five consecutive surveys. Using data from four surveys, this article shows that in early 2006 the population enumerated using the new method is appreciably larger – by 540,000 – than the estimate based on the 1999 census. The disparity is greatest for the total female population and for the 5-19 and 55-84 age groups. The new census also over-estimates the number of 18-year-olds. For measurement of international migration flows (for which data are incomplete in France), the census supplements information from government sources, in particular thanks to the question on year of arrival in France. This question is subject to a high level of non-response, but comparison of the different surveys with each other and with the 1999 census yields indications on how to adjust the results accordingly. Lastly, because it records household family structure, the new census enables fertility to be measured using the own children method. In spite of slight under-estimation, this method is effective for measuring disparities in fertility resulting from life course changes. The results confirm that immigrant fertility is low in the years before arrival in France but, through a catch-up effect, high in subsequent years. On a different level, the results suggest that the fertility increase of 2005 and 2006 is due partly to second and higher births.