Recent Advances in Small Area Statistics (RRC09)
Importance of the topic
Political and economic decisions are more and more based on indicators and data. Many of these data are gained from surveys conducted mainly on country level. However, nowadays it is rapidly becoming more important to gain adequate and accurate information on regional level. The reliability of this information is an essential input, e.g. for budget allocation between certain regions. Since the data gathering process, in general, is not built on regional level or does not allow for gaining accurate data quality on regional level, sophisticated statistical procedures have to be considered, the so-called small area estimation methods.
Small area estimation methods are becoming an important topic also in Europe since the European Commission is interested in several indicators on small regional level in addition to country level to facilitate an adequate support for less developed regions. Further, small area methods started to be used in European Official Statistics these days, e.g. in the UK, Poland, and Italy.
In Germany as well as in many other countries, small area statistics is still little known and only few researchers are already working in this research area. However, modern applications like the German register-based Census 2011 urge the needs of small area estimates. In contrast to a classical census, where all individuals are surveyed, a registerbased census uses register data, like population of unemployment registers, as well as an additional sample. Within the sample estimates can be gained for all variables which are not in the register but of major importance for political and economic decisions. Since classical methods do not fulfil necessary quality constraints for smaller regions or subpopulations new sophisticated statistical methods have to be used, the so-called small area estimation methods.
In order to adequately overcome this change of paradigm in Official Statistics methodology for these or comparable applications in regional statistical topics, further research has to be promoted. These advances in small area methods include topics like spatial alignments and their effects on the methods, their robustness, adequate model building and more. All these topics are covered by the world leading experts who contribute invited presentations to the RRC09 with enough space for detailed discussions.