From: Rakesh Bhandari (bhandari@BERKELEY.EDU)
Date: Thu Jul 05 2007 - 18:14:50 EDT
Jurriaan, you are mentally disturbed as shown by for example your
violently sexist comments on the old marxism lists (I refer here to
your threat to sodomize which led to a universal desire for your
expulsion) and your rants against phantom orthodox Marxists and your
offensive and wildly inaccurate comments against Cyrus and your
forwarding of virulently anti Islamic comments to this list, and
abusive comments towards me (first you fantasized my balls being held
tightly; now you call me a shopkeeper??!!). It should be an
embarrassment for all those responsible that you are allowed on this
list.
Plus, I have no idea what you are talking about. Perhaps you have
confabulated my opposition to statistics because I think your Bairoch
entry is misleading and poor. Of course you cannot quote anything I
wrote against statistics (as opposed to say e-con-ometrics); indeed I
have been looking at data on value added in commoditiy chains and the
data on wage differentials between skilled and unskilled labor as a
result of OPE-L discussion.
Bairoch's ten pages in support of the broad "thesis" (and it's not
an empirically testable thesis) you quote has been subjected to
exhaustive empirical and statistical critique in the last ten years.
Colonialism was more important to the rise of the West than he
understands (see Findlay, Darity, Barbara Solow, Robin Blackburn,
Kenneth Pomeranz, Amiya Bagchi, John Hobson) or could understand in
ten pages (I refer here to his myth book, and note that you say
nothing of his defense of protectionism or case for the
deindustrializing effects of colonialism so your entry is actually
very poor as a summary), and the third world (and what exactly is
that?) is a functioning part of the global capitalist system
(including through the export of labor). You seem to know of none of
the critique (here you prove yourself as knowledgeable as the
defenders of vulgar sociobiology in regards to slavery and dolphins
on this list).
I think you should remove yourself forthwith from this list to spend
more time alone in the hallucinatory paranoia caused evidently by
cheaply available hashish.
Rakesh
Social Research > Summer, 2001 > Article >
How Real Are Statistics? Four Possible Attitudes - )
Alain Desrosieres
REFERENCE to "reality" is a commonplace among both producers and
users of statistics. This "reality" is understood to be self-evident:
statistics must "reflect reality" or "approximate reality as closely
as possible." But these two expressions are not synonymous. The very
notion of "reflection" implies an intrinsic difference between an
object and its statistics. In contrast, the concept of
"approximation" reduces the issue to the problem of "bias" or
"measurement error." Thus even these two common expressions,
generally used without regard to consequences, tell us something
important: a critical reexamination of this notion of "reality" is,
for statisticians, an efficient way to reconsider the deepest-rooted
but also the most implicit aspects of their daily work--precepts,
tricks of the trade,justifications provided to users, and so on. This
paper argues that the way in which producers and users of statistics
talk about "reality" is informed by the fairly unconscious
intermingling of several attitudes to reality. The mix of these
attitudes and the links between them vary according to the
circumstances--or, rather, according to the specific constraints
prevailing in different situations. As it happens, the field of
business statistics offers a representative spectrum of these various
possible attitudes to reality. However, we cannot say--despite their
differences--that some are better than others since each is so
closely associated with situational constraints specific to
particular phases of the statistician's technical, administrative, or
managerial work.
Four possible attitudes to reality (among others) will be discussed.
First, we will describe them in a "pure" (hence certainly
exaggerated) form. Next, we will see how they are applied to concrete
situations. Each attitude has its language, that is, a register of
words, requirements, and arguments; these are consistent, but
difficult to interlink if one shifts from one attitude to another.
Here is a list of the four attitudes ranked by "obviousness" (at
least for statisticians; for other communities the order would, no
doubt, be different):
* metrological realism;
* pragmatism of accounting (which may be "national" accounting);
* use of material from a database for argumentative purposes in
social life or in quantitative economics and social sciences; and
* the explicit admission that the definition and coding of the
measured variables are "constructed," conventional, and arrived at
through negotiation.
Of these four attitudes, the first three may be qualified as
"realistic," but each has different reality tests--that is, ways of
verifying and articulating the substance of that reality and its
independence from observation. The fourth attitude, instead,
emphasizes the conventional and social character of statistical
variables, and may thus be labeled as "constructivist." It mainly
comes into play in situations marked by discontinuity, controversy,
and innovation. We will now examine each attitude individually,
spelling out their languages, origins, and conventions.(1)
Sampling and Confidence Interval
Metrological realism derives from the theory of measurement in the
natural sciences that is complemented, in the social sciences, by the
sampling method. The object to be measured is just as real as a
physical object, such as the height of a mountain. The vocabulary
used is that of reliability: accuracy, precision, bias, measurement
error (which may be broken down into sampling error and observation
error), the law of large numbers, confidence interval, average,
standard deviation, and estimation by the least-squares method
(Stigler, 1986; Hacking, 1990). This terminology and methodology was
developed by eighteenth-century astronomers and mathematicians,
notably Gauss, Laplace, and Legendre. The core assumption is the
existence of a reality that may be invisible but is permanent--even
if its measurement varies over time. Above all, this reality is
independent of the observation apparatus. In a sense, this is the
dream of the statistician and the specialist in quantitative social
sciences: the possibility of making the metrology of these sciences
equivalent to the proven methodologies of the natural sciences. This
may be seen as a benchmark, an ideal to which statisticians aspire,
despite an awareness that their objects do not display all the
properties assumed by the methodology. We could describe this as the
lost paradise of the social sciences, which would have liked to have
been endowed with the same persuasiveness as the natural sciences of
the nineteenth century.
In this endeavor to connect the methodologies of the social sciences
and the natural sciences, one element plays a crucial role: the law
of large numbers, which is the basis for probability formulas and the
resulting convergence theorems. The "law" serves as an operator for
the transformation and transition from the world of observations to
the world of generalization, extrapolation, and forecasting. Its
hybrid nature is summed up in the famous and revealing quip dating
back to the nineteenth century: astronomers and physicists believe
the law to be a theorem demonstrated by mathematicians, whereas
mathematicians think the law has been proved by the results of
repeated testing. In fact, the possibility of conducting multiple and
mutually independent observations of comparable objects is the
foundation of a statistical methodology initially developed to study
populations of human individuals or households.
Official statisticians in charge of business statistics are
specifically exploring the gaps between the methodology of business
statistics and the metrological ideal of the classical age. One of
these gaps is due to the heterogeneity of the population, which is so
great that the largest firms have to be "profiled" individually using
monographic methods quite different from the established statistical
method. There is also the difficulty of defining and classifying the
"statistical units": establishments (local legal units), enterprises,
groups, and so on--in other words, the very units that need to be
"counted" and "measured." Last--and most important for our
discussion--there already exists a quantification system internal to
the world of enterprises: double-entry bookkeeping ledgers recording
receipts and expenditures, along with claims and debts. The system,
which was conceived in the sixteenth century, predates the age of
statistical observation and is thus far older than the metrology of
eighteenth-century astronomers (Hopwood and Miller, 1994).
Double Entries and Balancing the Books
Business accounting is predicated on concepts of reality and proof
that underscore its profound differences with the metrology of
natural science. To begin with, the "equivalence space" is composed
not of physical quantities (space and time), but of a general
equivalent: money. Money allows the circulation of claims and debts
(via bills of exchange); it serves to determine profits by measuring
receipts and expenditures and by assigning a "probable" value to
claims and debts. It should be noted that this "subjective"
probability--used, for example, to assess a doubtful loan--is
different from the "frequentist" or "objective" probability on which
classical metrology bases its computations.(2) As we can see from
this crucial example of the calculation of doubtful loans, business
accounting is a rich and dense social practice that seeks to achieve
consistency and coordination in evaluations, actions, and decisions,
either for a single player over time, or for several players whose
relationships need to be regarded as fair and hence reproducible. In
this test of an accounting reality, double-entry bookkeeping plays a
role similar to the repetition of observations in classical
metrology. The requirements and tests involved in "balancing the
books" are analogous to the regularities and "normal" distributions
of repeated observations of the same object.
The tension between these two forms of quantification--one derived
from scientific metrology, the other from business-accounting
practices--has been sensitively analyzed by Oskar Morgenstern in his
famous work, On the Accuracy of Economic Observations (1963)
(translated into French with the unfortunate title of The Statistical
Illusion: Precision and Uncertainty of Economic Data). Morgenstern is
utterly dedicated to establishing a measurement system for economics
that is just as rigorous as that of the other sciences. For this
purpose, he examines the information provided by business accounts.
He studies the status of "errors"--often regarded as "falsifications"
or "lies"--that are to be found in these documents. Morgenstern
distinguishes--for example, in balance sheets--between the items that
are verified and identified without ambiguity (such as a cash
position) and those that are merely estimated and shrouded in
uncertainty, a practice justified by the need for prudence:
It will, however, be noted that a "lie" is, in this context, not a simple
and obvious concept. It is unmistakable when a false cash position is
willfully given or physical inventories are reported that do not exist. But
when deliberately a more optimistic attitude is taken in interpreting the
success of a year's operation--for example by small amortization--it will
be hard to classify this statement as a "lie." Instead, it may be viewed as
an error in judgment and as such be proved or disproved by later events
(1963).
This analysis effectively indicates the constraints that weigh on the
preparation of "fairly presented" (United States) or "true and fair"
(United Kingdom) accounts, to use the stock phrases. The "reality"
thus described is connected to a set of "wagers on the future" that
do not qualify as "lies" or even as "bias" (in the sense in which the
term is used in classical inferential statistics). In the case of
business accounting, which is a quantitative tool that underwrites
social links, "reality" is inseparable from the trust inspired by the
numbers compiled in the accounts--what Ted Porter calls "trust in
numbers" (1995).
Another aspect of this "accounting" realism is surprising from the
standpoint of scientific metrology: these documents display no
radical discontinuity between the past, present, and future--between
the closed accounts, the current accounts, and the budget forecast.
Rather, we find an incremental shift from one to the other, since all
are informed by the same conceptual frameworks, and are designed as
tools for assessment, action, and decision. A strikingly similar
continuity is found in the daily work of national accountants: they
too concurrently manage economic budget forecasts, followed by
preliminary, semirevised, and revised accounts.
The distinctive density and specificity of accounting practices are
well known to--yet sometimes forgotten by--the business statistician
imbued with a metrological culture. Since the 1970s, business
statistics has gradually imported the sampling method, previously
tested in social statistics. The result has been a juxtapositioning,
followed by a close interlinking, of these two forms of
quantification, despite their different origins and principles. One
of the problems in using business accounts for statistical purposes
is the impossibility of checking the uniformity of accounting
procedures. In particular, the numbers reported in business accounts
are already the result of an initial aggregation of myriad elementary
operations that are entirely beyond the statisticians' grasp. The
verification of this "first-level" accounting work is the job of
professionals such as certified accountants or "auditors"--and they
too, but in another way, have introduced the probabilistic method of
verification on samples of accounting documents (Power, 1992). In
this case, however, the sampled unit is an accounting document among
those of a single enterprise, not an entire enterprise among all
those in an industry.
The issues raised by the linkage between the two methodologies--one
statistical, the other accounting-based--were clearly visible in
other circumstances: the establishment of national accounts, for
example in France in the 1950s and 1960s. National accounting has
partly inherited the reality tests derived from business accounting:
its variables were defined (and, more important, interdefined) a
priori; they were recorded in consistent, comprehensive, and
theoretically balanced tables, where they were arranged in rows
(transactions) and columns (agents). Disparate sources were
reconciled (often for the first time) to compile these tables. The
final--but not least significant--resemblance between national
accounting and business accounting is that both tools were action-
and decision-oriented: the national accounts were intended as
"monitors" of macroeconomic policies, in the same way as the balance
sheet and income statement provide guidance for the company
executive. The accounts form a whole, explaining why the so-called
reliability constraints are not identical to those of a pure
"metrological" measurement of an isolated variable, whatever it may
be.
These interrelated specificities of the reality forms, the rationales
used to quantify them, and their applications explain the ritual
debates between statisticians and national accountants in the
statistical institutes in the 1950s and 1960s. The statisticians,
trained in the methodology of fine-tuned sampling, were wary of what
they regarded as the sometimes cursory practices applied by the
national accountants to estimate some of the variables in the
accounts (such as changes in inventories or trade margins). The only
justification offered by the national accountants was pragmatic,
based on purpose: their approach, they argued, was needed for
policymaking. Even low-quality estimates (provided they are contained
within the overall system of constraints created by accounting
balances) are preferable to no estimates at all. This defense was
unacceptable to the methodologist statistician. The differences
between the two camps may seem purely sociological, explained only by
their belonging to distinct socioadministrative networks: national
accountants are closer than statisticians to economic-policy decision
making and implementation. But each group had its own approach to the
orchestration of reality. Statisticians force themselves, sometimes
ascetically, to be nothing other than methodologists specializing in
good metrology; at the opposite end stands the systemic reality of
the accountant, which only makes sense within the framework of policy
guidance and monitoring.
Proof in Use
We can distinguish these two initial forms of realism from a third
form, which is implicit or explicit for the user outside the two
universes where these two concepts of reality reign: the official
statistical departments, and enterprises producing their accounts in
a continuous flow. Typically, this user is a researcher, or a social
player in the administrative, political, or economic sphere (notably
in other enterprises). For users in this third group, "reality" is
nothing more than the database to which they have access. Normally,
such users do not want to (or cannot) know what happened before the
data entered the base. They want to be able to trust the "source"
(here the database) as blindly as possible to make their
arguments--backed by that source--as convincing as possible.
We are confronted here with a "metadata paradox."(3) From a normative
standpoint, users must be given a maximum of detailed information on
the data-production process. It is also true that, from a descriptive
standpoint (i.e., without passing judgment), many users do not
welcome an abundance of metadata: "ideal" information is that which
seems self-sufficient, without footnotes to interfere with the
message. This unfamiliar field--generally mentioned (if at all) with
a touch of irony--would deserve to be studied in terms of cognitive
economics--that is, in terms of the yield (cost-effectiveness) of a
statistical argument. These issues cannot be dealt with in purely
normative terms such as: we must supply metadata (which is, of
course, true). A sociology of the social uses of the statistical
argument remains to be developed. It would be especially helpful in
exploring "quality" issues, often discussed in strictly normative
fashion.
The examination of uses--the argumentative contexts in which
statistical data are introduced--reveals a realism of the third kind,
based on the consistency and plausibility of the results obtained.
This is especially visible in the development of econometric methods.
They provide many tools or arguments for an internal validation of a
data set, without the need (at least in "normal" conditions) to
examine the prior stages (the recording and coding of these "data").
Alas, contrary to what the etymology of the unfortunate term "data"
suggests, very few "data" are actually "given": they come with a high
price tag, in both financial and cognitive terms. Coding always
involves sacrificing something with a view to the subsequent use of a
standardized variable, that is, an investment in form (Thevenot,
1984). This is comparable to the industrial investment needed to
produce the standardized and mutually validating parts of a
machine.(4)
It is striking that the statistical institutes of certain countries
(such as the Netherlands) place such emphasis on the concept of
integration or, in other words, the achievement of consistency among
the statistical data produced by the institute's various departments.
This goal is promoted even if it entails corrections and adjustments
of the "raw data"--of what may appear to be the "reality in the
field." This substitution of a validation internal to the statistical
system for a more external validation--with respect to a putative
field--is relevant to our exploration of the different forms of
realism. It converges toward the realism desired by users: the
internal consistency of their data set. Indeed, it is precisely in
such terms that the most staunchly "integrationist" statisticians
justify their extensive efforts to whittle down any anomalies: "our
users would not tolerate our giving them inconsistent data." This
point of view actually resembles that of national accountants,
who--at least for macroeconomic variables--adopt the same
use-oriented approach in their macroeconometric models, and whose
book-balancing equations represent a major constraint.
The Three Realisms: Summary and Comparison
To conclude this review of the three ways of being "realistic," we
can try to compare the reality tests that characterize them. The
first is that of the pure statistician, trained in a probabilistic
culture; this test is a remote descendant of the theory of
measurement errors in eighteenth-century astronomy. The latter
science regarded observations, however numerous, as independent of
one another. The object's reality and substance are proved by the
normal distribution of error-ridden observations. A confidence
interval can be presented in probabilistic terms. This metrology was
imported into the social sciences through the sampling method. The
astronomers' basic hypotheses have been transposed to this new
universe: statistical units are "homogeneous" (but the definition of
this term is ambiguous, notably in the case of enterprises); the
distributions of the variables studied do not diverge too
significantly from the normal curve; and the law of large numbers can
be applied. The central notion of this transfer is that--just as with
the distributions of astronomical observations--the computed moments
(averages, variances, correlations) have a substance that reflects an
underlying macrosocial reality, revealed by those computations.
Therein lies the essence of metrological realism.
Accounting realism is altogether different. It is internal to the
enterprise. Accounting is already an aggregation, in monetary terms,
of heterogeneous elements. Some of these are measured with certainty
(cash positions, at least when the monetary unit is reliable and
stable); others are estimated with uncertainty and imply a degree of
subjective probability. The choices of these values are guided by the
two potentially contradictory requirements for prudence and for
communication with other players. The main underpinnings of this
approach are the "interdefinition" of the variables and their
recording in balanced tables. Its overall realism is more
pragmatic--in the sense in which we describe someone as
"realistic"--than metrological. In any event, the two orders of
realism are closely intertwined, and this combination forms the core
of a practice of construction and use of quantitative data that
differs from the statistician's approach.
The reality judgments of users represent yet another category. Here
the technical and social divisions of labor, between the production
and use of statistics, have engendered their social and technical
effects. The dam set is a black box whose input and output sides can
be clearly distinguished, provided that the input side is perceived
as meeting "quality standards." Today these are increasingly explicit
and guaranteed, whereas they used to be more implicit. The user's
trust in the data-production phase is a precondition for the social
efficiency of the statistical argument. The reality test is provided
by the consistency of the results and constructions issued from the
data set.
Let us now stand back from these three forms of realism, which we
have separated for analytical purposes. We can observe that, if
"ultimate reality" is never accessible directly but only through
different perception systems, then the three realisms come together
in a single test--that of the consistency between the various
perceptions. However, the three approaches contrast with a fourth,
which, in its concern to reconstruct the chain of coding and
measurement conventions, effectively challenges the reality of the
objects. This attitude--which we may describe as nominalist or
constructivist--generally does not result from a theoretical
philosophical choice but arises in situations marked by controversy,
crisis, innovation, and changes in the economic, social and
administrative contexts: this is indeed the situation in the 1990s
and 2000s.(5)
From Measurement Conventions to the Languages of Reality
Even more than social statistics, business statistics is a field in
which the measurement conventions for most of the objects studied are
continually being debated and called into question. This trend has
been accelerated and made more visible by the requirements of
European harmonization. Witness the list of topics to which INSEE has
devoted its annual one-day seminar on business statistics between
1995 and 1999: enterprise groups, networks, accounting standards,
frontiers between goods and services, and restructuring. This list
alone enumerates fields where the "reality" of the objects dealt with
by statisticians are constantly eluding their grasp and changing
appearance. This makes the job of the official statistician in charge
of business statistics uncomfortable but stimulating. The statistical
activity for each of these topics is determined and constrained, at
an early stage, by a multitude of negotiations--at the micro- or
macro-level--among a multitude of macro- or micro-players: the
government, trade organizations, the European Commission,
competition-protection agencies, labor unions, and enterprises, from
the largest to the smallest. The seminar topics seem to have been
chosen to cover a succession of fields where all these players
exercise their negotiating skills.
The existence of enterprise groups--whose boundaries are often
problematic--raises the issue of which statistical unit is relevant
for business statistics, which is an inexhaustible subject of debate
among European statisticians. What better illustration of the
problematic nature of the "reality" sought by business statisticians
than the recurrence of controversies over the statistical unit, the
most basic and central link in the system of business statistics? A
host of related questions ensue: What is an activity? A product? A
"relevant market"? A "dominant position"? As we know, the answers to
these questions are interlinked, in keeping with the famous "chicken
and egg" principle. Scientific and legal issues are enmeshed, and
some of these have major implications. For example, the December 25,
1999, issue of Le Monde carried four pages of dry but fascinating
"legal notices" describing a "Ruling by the Competition Council of
July 20, 1999, on practices observed in the sector of thermal
applications of energy," consecutive to a dispute between the
national power utility (Electricite de France) and the national
library (Bibliotheque Nationale de France). The concept of "relevant
market," which is central to this type of dispute, thus made its way
into Santa Claus's stocking.(6)
In dealing with these issues, business statisticians are faced with a
dilemma that puts their professional identity at stake. On the one
hand they must assert a public-interest objective guided by science
alone, above individual interests and their contingent controversies;
on the other hand, they cannot ignore that their output will serve as
arguments in such controversies. "Reality" will, in that case, be no
more than a rhetorical instrument brandished by business lawyers.
Some statisticians go to great lengths to claim they can
single-handedly define a theory of economic reality and then observe
that so-defined "reality" in compliance with that theory. The chief
exponents of this position are the Dutch business statisticians.
Their "statistical units" are, in principle, entirely distinct from
"legal units." They are the object of painstaking, quasi-monographic
definitions and observations. The French, in contrast, try not to
distance themselves too much from corporate bookkeeping methods and
practices, although it is harder to synthesize this material in a
consistent, comprehensive, and theoretically elegant manner.
The theme of accounting standards and their harmonization at the
European and global levels that was mentioned earlier lies at the
heart of the questions concerning the social, negotiated, and
conventional character of business statistics. The accountant's work
carries within it a language of reality that defies the
statistician's language; the accountant constructs, from the ground
up, a universe to serve as a reference for action. This undertaking
is more or less constricted by rules and standards that vary from one
country to another, or rather between a few broad geographic areas,
since the leading Anglo-Saxon corporations and their auditing and
consulting firms are imposing de facto world standards, which the
European Commission's Fourth and Seventh Directives briefly tried to
anticipate.(7)
Enterprise networks are new production and distribution structures
that also pose a problem for business statistics and make it
difficult to use a confident realistic language. Because of their
lack of a central entity and their high mobility, their boundaries
are hard to draw--in contrast to enterprise groups, which generally
possess a decision-making center whose sphere of influence one can
attempt to define. The frontiers between the goods-producing and
service-producing sectors of the economy are equally blurred by the
practice of subcontracting and outsourcing entire segments of
corporate production activities. Last, the permanent restructuring of
large corporations and even small and medium-size enterprises makes
it particularly difficult to monitor the producing sectors over time
and therefore to build time series--the daily fare of econometric
research.
These major issues in business statistics are well known and have
been extensively studied. How do they affect the "language of
reality" expected of statisticians and economists? This question can
be addressed in two ways. Advocates of a defensive realistic approach
will try to fill the gaps or glue the pieces back together. Ingenious
solutions will be proposed to, for example, track the changes in
industrial corporations over a two-year period. In a more
constructivist approach, however, one can also be attentive to the
way in which the language of reality itself evolves in times of
crisis and rapid change. Viewed from this angle, statistical work not
only reflects reality but, in a certain sense, establishes it by
providing the players with a language to put reality on stage and act
upon it. That is why, beyond the issues of pure statistical
measurement that they raise, the topics of the INSEE seminars listed
previously are a rich terrain for observing the current changes in
vivo.
A significant example of the evolution of languages--naturally linked
to that of social relationships--is the rapid emergence of the
expression "value creation" (or "economic value-added") used in the
financial world to describe the change in share prices and the gains
that shareholders expect from restructuring. This reflects the rise
of large pension funds and the use of the stock market to finance
investment. In the process, however, the classic language of "value
added"--employed in national accounting as well as in a tax
context--appears to have been forgotten. One example is the economic
press, which makes extensive use of the new vocabulary. In short, a
competition appears to have arisen between several languages of
reality, each used by different social players.
Amid the uncertainties over the status of the measurements proposed
in statistical products, various expressions are often used. Each, in
its distinctive way, shows that the metrology of the social sciences
is not of the same kind as the metrology of the natural sciences. The
terms "index" and "indicator" suggest that the reported measures are
like the visible symptoms of a hidden reality that is impossible to
reach directly. The expression "latent variable"--used by some--has
identical implications. Occasionally, statisticians or academics
prudently take refuge in the notion that their quantitative data aim
to "encircle reality." Thus surrounded, as in the game of Go, reality
will have no choice but to surrender. More seriously, this apparently
military metaphor evokes the notion--which is quite familiar to
philosophers--that a reality is known only through external, socially
constructed, and historically rooted "points of view." By multiplying
the points of view from different positions, we can always dream of
"encircling reality." But reality will slip away, for new
systems--and languages to make them real--are born every day.
(*) This paper, which has been translated by Jonathan Mandelbaum, is
a revised version of an article originally published in French in La
Lettre du Systeme Statistique d'Entreprise, the in-house bulletin of
INSEE (the French national statistical institute). The "business
statistics" discussed are those supplied by national statistical
institutes to describe the activity of enterprises. The accounting
and statistical data produced and used in firms are discussed here
only with regard to their use by statisticians and, subsequently, by
economists.
Notes
(1) The issue of the realism of statistical production is discussed
in greater detail in Desrosieres (1998).
(2) On the origins and meanings of the distinction between
"objective" and "subjective" probabilities, see Daston (1994).
(3) In the language of statisticians, "metadata" are the information
on the definitions, construction methods, classifications, recording
procedures, and processing of disseminated data.
(4) This key concept first appeared in the nineteenth century in
rifle manufacturing.
(5) Ian Hacking, in The Social Construction of What? (1999), offers a
refined analysis of the social uses of constructivist and realist
arguments without locking himself into either position.
(6) Santa, in turn, seems to threaten the dominant position of the
Magi in Spain (Le Monde, December 26, 1999). Is the Christmas cake a
relevant market?
(7) European accounting standardization, which regulates the
preparation, auditing, and publication of accounts, is governed by
the Fourth Directive (1978) for annual accounts and the Seventh
Directive (1983) for consolidated accounts of limited-liability
corporations.
References
Daston, L. "How Probabilities Came to be Objective and Subjective."
Historia Mathematica 21 (1994): 330-44.
Desrosieres, A. The Politics of Large Numbers: A History of
Statistical Reasoning. Cambridge: Harvard University Press, 1998.
Hacking, I. The Taming of Chance. Cambridge: Cambridge University Press, 1990.
--. The Social Construction of What? Cambridge: Harvard University Press, 1999.
Hopwood, A. G., and P. Miller, eds. Accounting as Social and
Institutional Practice. Cambridge: Cambridge University Press, 1994.
Morgenstern, O. On the Accuracy of Economic Observations. Princeton:
Princeton University Press, 1963.
Porter, T. M. Trust in Numbers: The Pursuit of Objectivity in Science
and Public Life. Princeton: Princeton University Press, 1995.
Power, M. "From Common Sense to Expertise: Reflections on the
Prehistory of Audit Sampling." Accounting, Organizations and Society
17:1 (1992): 37-62.
Stigler, S. The History of Statistics: The Measurement of Uncertainty
before 1900. Cambridge: Harvard University Press, 1986.
Thevenot, L. "Rules and Implements: Investment in Forms." Social
Science Information 23:1 (1984): 1-45.
COPYRIGHT 2001 New School for Social Research
COPYRIGHT 2001 Gale Group
This archive was generated by hypermail 2.1.5 : Tue Jul 31 2007 - 00:00:06 EDT