Statistical Journal of the IAOS

Modern technology used in data processing has created a challenge to national statistical services. Two aspects are particularly noteworthy. Firstly, owing to the relative ease of producing data, economic and social information is generated increasingly outside statistical offices. With an accompanying need for new approaches to questions of data processing and consistency within the national information system. Secondly, the emergence of data banks render primary statistical information accessible to outside users; consequently, statistical offices have to reconsider such questions as production lags, the provision of assistance in data interpretation and confidentiality of data.

The potentialities offered by new data processing facilities are as yet unmastered. New data sources have become available, particularly in the area of administrative information kept by public authorities outside the statistical system. This circumstance necessitates the finding of a new balance between the use of administrative files by statistical offices and statistical surveys. The requirement of adapting statistical services to an ‘industrial’ mode of data production implies the conversion of data bases into data banks with the corresponding supply of adequate meta-data information and the re-structuring of time-consuming processes used in the collection of primary data and the compilation of aggregate information.

A second major task relates to the re-examination of the co-ordinate role of statistical offices. The objective of a consistent national information system can only be achieved if statistical offices are given the possibility to co-ordinate basic information effectively. Particular problems arise in exerting sufficient statistical influence on the definitional and other methodological underpinnings of administrative files as well as on their management. It has to be recognized in this task that such files are kept for specific purposes.

While the confidentiality of nominal personal data has to be and will be preserved in statistical services, available social and economic information also has to be seen as public property. The provision of an integrated and transparent body of information is a challenge to society as a whole, for the mere availability of data has to be supplemented with the necessary tools that permit the structuring and understanding of information.

The following problems are of concern to statistical offices. (1) To what extent should the increasing demand for statistics be met through regionalized public statistical activities and to what extent should it preferably be met from a central organization? (2) How is the optimal solution affected by the availability of new technical tools?

The pattern of centralization/regionalization in a statistical production system is discussed in terms of four kinds of subprocesses (collection, preparation, aggregation and dissemination) and three kinds of resources (hardware, methodology and manpower).

Regional users have different statistical demands which may be in conflict with each other and with the demands of central users. Their first claim must be to get access to statistics, particularly for their own region, with a content and a quality adapted to their needs. They may be induced to start own activities on subprocesses earlier in the production chain, if these subprocesses are not designed in accordance with their needs.

The aim of the central statistical organization (CSO) should be (1) to minimize duplication of work as far as possible and (2) to secure a development of statistics which is consistent and comparable between regions and with national statistics.

A tentative norm for a well-balanced solution is seen in the centralization of all three kinds of resources for the collection and preparation activities. For aggregation too the dominating pattern should be centralization. To some extent, however, clean micro-data might be put at the disposal of regional authorities for regionalized aggregation operations, preferably in accordance with a common method. For dissemination the dominating pattern is regional. The hardware used for dissemination is assumed to be either centralized and regionally manipulated through terminals, or regionalized or some combination of both.

EDP technology has introduced new possibilities to satisfy original demands for statistics. One is to transfer clean micro-data to users (e.g. in the form of a tape copy), possibly to be handled by them on their own computer. Another solution is to disseminate statistics to the regional users from a centrally stored data base which they manage interactively from terminals. In Sweden such a system has been developed which consists of a data base management system called AXIS (Auxiliary System for Interactive Statistics) and a data base called RSDB (Regional Statistical Data Base).

Considerable experience is obtained in the Netherlands with statistical co-ordination in the field of economic statistics over the past decade. The basic requirements for co-ordination necessitate specification of the following elements: (i) a system of statistical units; (ii) several types of classifications; (iii) a set of definitions of economic concepts and rules for the practical implementation of these definitions; and (iv) a central register of statistical units.

Two classifications have been developed within the Central Bureau of Statistics, The Standard Activity Classification and the Standard Commodity Nomenclature.

The development of a central register of statistical units is under way. In addition, supplementary measures are taken to enhance co-ordination within the Netherlands Central Bureau of Statistics.

Experience shows that questions of the quality, timeliness and reliability of statistical data cannot be examined without regard to the cost of the system. However, the use of specific methods better adapted to the general characteristics of a system can given the same effect with little or not difference in cost.

Measures to ensure the quality of statistical information aim at improving the quality of the data from the point of view of accuracy and suitablity for the requirements of users.

Among the most important measures to ensure quality have been the organization of a unified system for improved professional training of statisticians, the provision of assistance to computer specialist training, preparation of the public and of non-specialist personnel participating in censuses and various sample surveys, the perfection of statistical methodology and of the system of indicators, the unification of classifications and nomenclatures, etc. Particular attention has been paid to the use of technology and specific automated processing techniques suitable for ensuring data quality.

Since efficient management presupposes speedy decisions, a special system known as the ‘fast information system’ has been organized in Romania at all levels to collect, transmit and process data.

The timeliness with which statistical information can be supplied has been influenced by the new technology used in transmission and processing, and also by the introduction of new methods of processing statistics: pre-printing of processing codes on input forms; input of data with the aid of ‘on-line’ display terminals; interactive correction of stored data; use of remote or decentralized data processing; standardization of software; development of specialized computer languages for processing statistics; introduction of appropriate data bases and registers.

The measures taken to ensure the reliability of statistical data have aimed at (i) extending the period of usefulness of the data; (ii) ensuring the reliability of equipment used as well as some stability in the use of methodologies, definitions, classifications and nomenclatures, due regard being given to the flexibility of the system; and (iii) using particular methods of statistical mathematics to update aggregate data and of storing historical card indices in data bases which can be processed on demand.

With the rapid expansion of the economic and social life of the country, with growing demands of policy-making and planning bodies on the system of Soviet State statistics great attention is being paid to the training and upgrading of specialists necessary for a successful functioning of the Automated System of Nationwide Statistics of the Central Statistical Board, USSR (CSB).

State statistical bodies of the USSR are staffed with graduates of higher educational establishments and of colleges (vocational specialized schools) who are employed according to manpower utilization plans. Statistical personnel of the computer network of the CSB are also trained in the field of automated data processing of economic and statistical information, based on latest developments in this field both in the USSR and abroad.

Graduates of educational establishments joining the CSB, alongside with fundamental education in subject-matter, participate in practical activities at the computer centres of the CSB where those graduates undergo processes of adaptation to the real production environment. A production base is assigned to educational establishments for a long period of time, which contributes to a better educational level and to practical orientation of the education.

Intensive efforts are made in order to improve the methods of education for some disciplines. Text-books for fundamental cycles of applied disciplines are published.

The scientific base for training and upgrading of statistical personnel is ensured through permanent contact with organizations and institutes of the CSB and of the USSR Academy of Sciences.

Training is based on proper modem computer and data processing equipment. The Main Department for Training and Upgrading of Record-keeping (accounting) Staff of the CSB is making a great effort in publishing text-books on software maintenance and use of modern computing equipment. These measures ensure proper training of specialists according to the growing demands, at present and in the near future, of both the CSB and the national economy of the country.

The possibility of using data from administrative registers varies from one type of statistics to another. For census type statistics there is a special need for the use of such sources. This is due to the high costs involved in the preparation of census type statistics.

The possibilities for using administrative data for statistical purposes depend on the general principles for and organization of the general official statistics in the country in question. Also the administrative conditions, especially as regards the use of EDP and register techniques for administrative purposes are important.

One principal aim for the development of register-based statistics in Denmark has been to compile information of largely the same contents as a traditional general census, but the utilization of register data has wider perspectives, whose realization, however, presupposes the existence of a systematic framework in the form of a general register-statistical system.

In the methodological field, a number of new problems have been encountered. Some of these problems have been solved. Others will require further consideration, for example, error checking and missing data. Other problems are how to integrate sample data in the system, and the ways to cope with the vulnerability of statistics to changes in the administrative systems from which the data are obtained. It is to be expected that the use of the register-statistical system will be considerably extended, notably towards longitudinal surveys. It may finally be mentioned that a new methodology will be required also in the field of statistical production techniques.