Indirect estimation in population census 2011 in Poland

Elzbieta Golata, The Poznan University of Economics
Jan Paradysz

As Small Area Statistics (SAS) provide estimates which are of great demand for local authorities, it becomes more and more popular nowadays. This methodology originate from demographic research. Special methods were developed to provide population estimates for detailed geographic division of a country or for domains defined by age, sex, marital status and other demographic variables in the noncensal years. They are based on exploration of administrative files. Traditional demographic SAS methods employ indirect estimators based on models utilizing symptomatic variables which are strongly related to changes in local population. The expansion of sophisticated small area estimation methodology results also in the development and implementation of alternative methods for conducting censuses. The method proposed for population census 2011 in Poland is register – based census with sample surveys. In the paper we discuss assumptions and methodology concerning population census in Poland. That is why a comparison analysis between estimates based on census and population register is conducted and a deep discussion evaluating both data sources including omissions and false enumerations is provided. Introducing register – based census involves also application of a large sample survey providing long form census statistics. This study is also aimed at taking into account and applying different administrative data sources and SAS methodology for effective population estimates for small areas. Standard indirect estimation techniques like synthetic and empirical bayes are implemented for different types of small domains. The study comprise overall evaluation of the results obtained from domain specific point of view and combining all domains. Poland belongs to countries which are developing, and testing alternative techniques for collecting, processing and disseminating statistics generated by the population and housing censuses. Before new methods are implemented an overall evaluation is needed to meet the crucial demand - to provide population characteristics at the lowest geographical level.

  See extended abstract

Presented in Poster Session 3