Inferring Population Dynamics from Imperfect Census Data: An Evaluation Study in Three Health and Demographic Surveillance Sites in Rural Senegal

Bruno Masquelier, Université Catholique de Louvain
Cheikh Tidiane Ndiaye, Agence Nationale de la Statistique et de la Démographie, Sénégal
Clémentine Moerman, Institut National d'Études Démographiques (INED)
Ndèye Binta Diémé Coly, Agence Nationale de la Statistique et de la Démographie, Sénégal
Samba Ndiaye, Agence Nationale de la Statistique et de la Démographie, Sénégal

In low-income countries, censuses remain a fundamental building block for assessing population dynamics, in the absence of a comprehensive system of vital registration. We evaluated key demographic parameters obtained from the 2002 census in Senegal against high-quality data collected prospectively in three Health and Demographic Surveillance Sites (Bandafassi, Niakhar and Mlomp). The age and sex profiles of the population were broadly similar in HDSS and census data, but many long-term HDSS residents were not reported in the census (conducted during the dry season) because they were in temporary migration. Under-five mortality rates obtained from census reports were slightly lower than expected from prospective mortality data. Recent births were correctly reported but fertility rates could be biased by selective migration. Age-specific mortality rates based on recent household deaths were consistent with surveillance data except for young infants.