Abstract: This paper describes the UK CSO's main computing activities over the past eight years or so. Over that time, the Office has converted the statistical computing systems written in Cobol and Fortran and running on proprietary mainframes, to use Relational Database software running on Open System (Unix) computers. The paper briefly reviews the benefits of adopting this architecture and offers recommendations to others following a similar path, drawn from the accumulated experience. In particular, it discusses the difficulty of assessing the size of Unix system needed and stresses the importance of carrying out realistic benchmarks on the intended hardware/software…combination.
Abstract: Paper is the traditional dissemination medium for statistical information. The advent of micro-computers, however, has led to a growing demand for machine-readable statistical information. To meet user demand, statistical organizations have installed computerized information systems. These systems should not only contain the raw data, but also the relevant meta-data. Only by providing this meta-information as well, is it possible for the users to understand and finally use the statistical information. This paper describes the STATview system for the dissemination of statistical information installed at Statistics Netherlands.
Abstract: In Canada independent administrative data sources are used intensively for the creation and maintenance of the frame for business surveys. The paper argues that administrative data do not usually meet the needs of a statistical agency; typically because these sources utilize concepts, definitions and procedures which are pertinent to carrying out only their own specific mandates. The paper provides insight into the strengths and weaknesses of these administrative data from a “frame” perspective, and, in particular, the guidelines that have been used to offset the weaknesses. The paper concludes that while statistical agencies cannot afford to ignore the contributions that…administrative data offer in maintaining a business frame, the ideal administrative source that has no duplicates, sufficient structural detail, and is capable of identifying deaths quickly does not exist. Instead, administrative sources need to be augmented for coverage, transformed for structure, and require creative methodologies for identifying cessations. Additionally, the frame and supporting processes must be designed not only with the survey user requirements in mind, but also with a thorough knowledge of the administrative sources intended to be used so that techniques may be developed to compensate for weaknesses.
Abstract: There are two surveys which deal with income statistics in Sweden. One employs data exclusively from different administrative registers and covers all citizens in Sweden. The household definition used in this survey is the taxable unit. The second survey which is a sample of 10 000 households combines data from administrative registers and data from tax return forms with questionnaire data. As a result of using a questionnaire, a more common definition of a household can be used and data on employment and housing are included. Income data for both surveys are collected from tax files and other administrative registers…with very good coverage and quality. The administrative registers which were created for other purposes than statistics cover wages, interests, dividends, and different social contributions. One essential drawback is that the first published income statistics will only be available about 18 months after the end of the income year.