The Rise and Fall of Sampling Surveys in Norway, 18751906

Abstract

ArgumentNorwegian statisticians were pioneers in the development of sampling techniques for social and economic investigations in the late nineteenth century. After a few years of extensive use of sampling surveys in large-scale social and economic investigations, the method fell out of use in the early 1900s. This article supports Alain Desrosières’ argument that the emergence of sampling procedures in social investigations must be seen in relation to a modern “individualistic” view of society. But the importance of the institutional setting is also stressed: The way statistical research was connected to the power and resources of the State within the Central Bureau of Statistics (CBS) was a central element in the development and implementation of the new technique. A separate argument is presented to explain why the method suddenly lost ground in Norway and the general director of the CBS stopped promoting his method in the meetings of the International Statistical Institute. The explanation is probably to be found in a large and politically important survey in the 1890s that was attacked by a group of actuaries from private insurance firms. The heated and long-lasting debate turned into a question about trust in the new method and the reputation of the head of CBS as a statistical expert. The necessary trust and confidence was lost when the CBS in 1906 had to admit that important estimates from this survey obviously were erroneous.

Developments in Survey Research over the Past 60 Years: A Personal Perspective

“…Much of the criticism focused on the data collected, but the so‐called ‘representative survey’ and the adequacy of the sample size (80 000 persons) were also attacked. Likely, because of the many criticisms, the Norwegian CBS stood back from sample surveys for a number of years (Lie, ).…”

Eilert Sundt som statistiker

“…Jeg skal gå inn i dette via en omvei: Utvalgsmetoder, der man i stedet for å drive kartlegging gjennom fullstendige tellinger henter ut opplysninger fra bestemte deler av populasjonen, ble tidlig utviklet i Statistisk sentralbyrå, av personer som arbeidet med direkte inspirasjon fra og beundring for Sundt (Lie 2001(Lie , 2002. Norske statistikere var de første som lanserte slike metoder på internasjonale statistikk-kongresser.…”

Bringing Things Together: Developing the Sample Survey as Practice in the Late Nineteenth Century

“…In some countries, it was government agencies (e.g., Norway, cf. Lie, ), and in other countries, it was professional associations whose members were bureaucrats and university professors (e.g., Germany, cf. Oberschall, ).…”

“…Researchers often single out Anders Kiaer (1838‐1919), the founder and first director of Statistics Norway, as the inventor of the “representative method” in the mid‐1890s (e.g., Kruskal & Mosteller, ; Porter, , p. 237; Desrosières, , p. 236; Bethlehem, , p. 10). About a decade later, the technique fell out of use after a bitter academic conflict (Lie, ). However, in the mid‐1920s the International Statistical Institute conducted a number of conferences where two techniques became two accepted methods of sample selection.…”