Population Census and Truth

The article discusses the quality of population census data – information obtained in the oldest, largest and most famous statistical surveys. In particular, the quality assessment of census data refers to the Polish experience. The article aims to propose an integrated approach to censuses, based on administrative records and sample surveys. Despite a long tradition, as well as a well-developed research methodology, censuses do not provide ‘perfect’ results. First of all, the census as a comprehensive study may be burdened with non-random errors. Taking into account all the different methods for conducting censuses, including the traditional method, using data from administrative records with sample surveys and the mixed method, different sources of error are analyzed. The study also attempts to define the census population in the light of international standards concerning usual residents, permanent residents and actual residents, with the resulting consequences. Sources of errors, methods of their identification and elimination are discussed. Particular attention is paid to coverage errors, which are illustrated by examples from Polish censuses. The results of previous studies assessing the quality of administrative records, which were conducted in preparation for 2011 Population Census, are taken into account. Evaluation of population censuses is traditionally carried out based on the results of post-census surveys. In the UK, the results of the Census Coverage Survey are the basis for estimating the degree of underenumeration in 2001 Census. Polish experiences in this field are rather limited. The discussion also includes harmonization problems, particularly the danger of divergent results and estimates, in terms of several data sources. In this regard, the British One Number Census project is presented. The idea of the project was to reduce the population underestimate and provide outputs which would be consistent in cross-section of territorial units in the regional and at a national level