Alonzo, William and Starr, Paul, editors. The Politics of Numbers. New York: Russell Sage,

 

Alonzo, William and Starr, Paul, editors. The Politics of Numbers. New York: Russell Sage,
1987.

This anthology consists mainly of papers prepared for a conference on "The political economy of
national statistics" held in Washington, D.C., 13-15 October 1983. It is a very important book
and merits inclusion on this list and attention from everyone interested in fraud in science. Its
focus is on the "background," the assumptions, and values, which surround the choice of
"numbers" which purport to describe, in this case, the population of the country.

As Starr's insightful opening essay suggests in its very title, "The Sociology of Official
Statistics," numbers are not value-free but are the result of more or less conscious decisions.
"Official statistics do not merely hold a mirror to reality. They reflect presuppositions and
theories about the nature of society. They are products of social, political, and economic interests
that are often in conflict with each other. And they are sensitive to methodological decisions
made by complex organizations with limited resources." (p. 1)

"The opportunity for strategic behavior in data reporting is greatly increased by ambiguities of
classification and the sheer complexity of some calculations. Without actually lying, respondents
such as corporations being questioned about their financial condition may choose to report to the
government figures conveniently different from those they use internally. The story is told of a
series of interviews for a job an auditor in a Soviet factory. ‘How much is two and two,' asks the
manager. "Four,' says the first candidate, who is promptly dismissed. Asked the same question,
the second job-seeker says ‘Five' - and obviously he won't do either. Finally, the third candidate
is asked the question, and he responds,'How much do you need?' That man gets the job. The
same story might easily be told of creative accounting in the West. It is a commonplace of
accounting that numbers are not absolute; they are a policy decision. As Morganstern argues, any
corporate balance sheet consists of values ranging from precisely known, highly liquid assets to
others that are purely speculative estimates, such as the value of trademarks. Even within the
limits of standard accounting practices, the opportunity for strategic data reporting is vast." (pp.
34-35)

"In an authoritarian regime, the problem of falsification arises not merely because the regime
may lie but because subordinate officials in industry and regional government have good reason
to lie to their superiors." (p. 37)

"Data processing is a form of work - and in modern societies an increasingly important form. The
production of statistical data is, among other things, an industrial process, or labor process, in
which large numbers of clerical workers perform routine tasks and are supervised by several
levels of administrators and professionals. The distribution of knowledge and control over this
work repeats the more general industrial pattern - the concentration of conceptual and planning
functions among a relatively small class. In the organization of statistical work, a minority of
professionals understands the system and can spot errors. For the majority, on the other hand, the
work itself is virtually meaningless. Consequently, they are likely to have neither the interest nor
the capacity to identify mistakes." (pp. 37-38)

"The professional and administrators who design official statistics must, therefore, not only apply
their technical competence but also interpret political and social objectives for the data." (p. 40)

"The rational for sociological inquiry into statistics is partly to avoid being deceived by official
statistics. Studies of the development and organization of statistical systems are a journey
upstream toward the sources of everyday facts. In that sense, they serve purposes fundamental to
sociology and help illuminate a sphere of knowledge that has taken on central importance in our
political and economic life." (p. 57)

In another contribution, Judith Innes De Neufville, "Federal Statistics in Local Governments,"
pp. 343-362, much is made of the "ritualistic" value of statistics. "While the statistical
requirements (for grant-getting) did provide some rough controls on who received funding, by far
their most significant consequence was to assure that local governments hired professional staff
capable of manipulating data, who in turn would influence local decision processes." (p. 345)

"...(S)tatistical requirements were used to professional local government, influence its policies,
increase the accessibility of the data and promote a norm that statistical information ought to be
used in decisions." (p, 346)

"The required statistics seldom genuinely informed or directly affected program decisions by
what they showed. Rather they were assemblages of numbers tacked onto proposals." (p. 346)

"Thus the required statistics became merely window dressing - part of the ritual or grant getting.
As such they were not particularly accurate but they were accepted. Few bothered to point out
their limitations. It simply did not matter." (p. 346)

"The ostensible object was to determine the exact nature of the housing need and to shape local
programs to match this need. Since funding was never adequate, however, the objective was
meaningless." (pp. 346-347)

"Though most data managers did their best to provide honest estimates, some discovered that
invented statistics were neater and therefore more acceptable to HUD. Though none of the actors
believed in these statistics, there was no incentive for anyone to explode the myth. The
standardized numbers were essential to HUD's reports to Congress, and the HAPs provided the
appearance that the programs were in accord with national goals. From the communities' view,
statistics acceptable to HUD allowed them to conduct their programs without federal
interference. The use of statistics helped absolve each level of the bureaucracy from substantive
responsibility for judging the program and reduced the need for conflict of negotiation. The
numbers were not just without meaning, the actually operated as a smoke screen." (p, 347)