Assuring Quality in the Statistics Production of Statistics Sweden
Eva Elvers
Process Department, Statistics Sweden, Stockholm, Sweden

The European Statistics Code of Practice defines three major areas to ensure quality in statistics: institutional environment, statistical processes, and statistical output. Statistics Sweden has approaches in all these areas. The focus here is on the statistical processes and the major approaches in this area; some illustrations are also given. There is a strategy for the structure of data warehouses and coordination of registers; and a long-term action plan is being developed. There is a process support system based on a generic statistical business process model with each sub-process described in a how-to-do way for the processors who use it directly or indirectly, e.g. survey managers. The preparatory sub-processes for needs, design, and building lay the ground for the statistics production and its quality assurance. These sub-processes are still at an early stage and under improvement. A first internal guide for design of statistical surveys has been developed, and a project for quality assurance of the methodological design work has been initiated. Decreasing response rates provide a visible quality concern, and data collection strategies are further developed to meet the concern and to be more cost-effective. A tool called Triton is built as a platform for data collection, using modern architecture. New editing tools are implemented successively with positive effects on quality or costs or both. Standardised methods and tools must be suitable for different types of statistical processes and products; they will be used by many - some specific solutions will still be motivated. Internal communication is vital. There have been efforts to show the needs and merits of standardisation and documentation, but more had been desirable. The communication and support efforts will go on and be renewed.

Keywords: Quality assurance; Statistics production; Standardisation; Survey design

Biography: Eva Elvers has a background in mathematical statistics. She works with methodology and quality at Statistics Sweden since 30 years. Currently she is process owner of the sub-processes Design and plan & Build and test.