Our legacy to future generations – using databases for better availability and documentation

Abstract

Our paper takes a long-term perspective, noting that regular Swedish demographic statistics date back to 1749. The question now is: Are we going to leave a documentation of our society that is of as high a quality as they did 250 years ago? Will it be possible to read all our electronic documents in another 250 years? Is our documentation good enough? Are our methods of storage safe enough?

A broad picture is given on the developments of the technical infrastructure within Statistics Sweden from mainframes to PC networks. Given the complicated nature of the new system it is easy to understand that good quality documentation may be difficult to maintain, compared to the rather simple forms and tables that were produced 250 years ago, or even compared with the mainframe production system of 35 years ago.

The new technical platform will form the foundation for our new databases of official statistics. However, all official statistics will not be placed in them, but only such as are of wider interest, primarily those statistics in general demand by users in different sectors of the society. These statistics, at least, will definitely be well documented and saved for future generations. A number of registers (e.g. population, enterprises) and some “observation registers” from important surveys will also be part of the databases. An observation register is the final edited register with the observations from the survey. An important aspect discussed at some length in the paper is the documentation needed for these registers.

Some background is presented in the UN/ECE “Guidelines for the Modelling of Statistical Data and Metadata”. These guidelines, edited by Prof. Bo Sundgren of Statistics Sweden, are a result of the UN/ECE METIS project. Prof. Sundgren is also the project leader for the development of the new Swedish databases. It is therefore natural for our database development to follow these guidelines.

We try to show how a systematic handling of metadata can help to provide present and future users of data with the needed background information. The research type user trying to re-use data from past surveys has special needs that must be catered for. A key issue is having links between different types of metadata.