A procedure for enumeration area code imputation based on deterministic and probabilistic methods.
What is it about?
The new Italian Permanent Census is based on the integration of administrative and sample data and plans to provide accurate and consistent statistical data each year at the various spatial levels. For the production of sub-municipal data, the strategy requires the integration of the Base Register of Individuals and the Base Register of Places. The accuracy of the data depends on the quality of the registers and the procedures adopted to integrate and process the input data. One of the critical issues is the presence of non-geocoded units due to problems in the administrative data. Istat has studied a procedure that integrates deterministic and probabilistic approaches to assign the enumeration area code to these critical units. An experimental study was conducted to evaluate the quality of the imputation procedure. This article discusses the approach taken, the evaluation process, the results obtained, and the impact on data quality.
Why is it important?
The geocoding of population individuals by residence address is a task that is highly dependent on the completeness and accuracy of the spatial information contained in administrative records. The presence of a percentage of non-geocoded units does not allow the publication of statistical data at the sub-municipal level. The procedure described in the paper makes it possible to resolve cases of geographical non-coding, thus offering the Permanent Italian Census the possibility of producing census results at the maximum spatial detail coincident with enumeration areas. The article will explain the imputation procedures, experimental tests, the analysis process, and the results obtained, and will evaluate the impact of the procedures on the results. Finally, the validation process of some census results produced at the sub-municipal level will be illustrated, a preliminary operation to the official dissemination of the final data.