Thomas on the Functions of Statistics

The articles published in 1984 were the output of a research project supported by the Social Science Research Council (Grant No HR 7195) on the impact of statistics on policy making. The research found that assumption implicit in the original proposal was not justified and concluded that the principal function of statistics is to contribute to organizational coordination. The `Statistical Information Systems and Management' paper published in Organization Studies (1984, 5/4, pp. 345-358) argued that if the principal function of published statistics is to contribute to inter- organizational contribution, then it is reasonable to hypothesize that the principal function of statistics within organizations is to contribute to intra-organizational co- ordination. This idea is contrary to management literature which portray information systems as hierarchical in character - with statistical information from individuals and subunits of the organization flowing up the hierarchy and commands in the form of statistical information flowing down the hierarchy.

The paper `Statistical Information Systems and Management' also draws attention to the limited significance of organizational boundaries. The formal function of boundaries is to define the limits of the responsibilities and powers of organizations. But these formal functions do not preclude the flow of statistical information across organizational boundaries. Where these flows of information occur, as with the use of published statistics, coordination can be achieved independently of the delineation of organizational boundaries. Thus the free flow of statistical information may soften the differences in performance sometimes ascribed to apparently different organizational structures - such as hierarchical and non-hierarchical, capitalist and planned, publicly owned and private, etc.

The paper `Why have government statistics? (And how to cut their cost)' published in the Journal of Public Policy (1984, vol. 4 Part 2, pp. 805-102) reported the organizational coordination conclusion and discussed the implications in terms of the government of statistics by means of a comparison of the findings of the research with those of the Rayner Review of the Government Statistical Services which had just been completed. Rayner concluded, for example, that information should not be collected primarily for publication (but) because government needs it for its own business. The `Why have government statistics paper argued that this conclusion misses the purpose of publication - which contributes to coordination between central government, local government, industry, etc.