Content uploaded by Jürg Steiner
Author content
All content in this area was uploaded by Jürg Steiner on May 25, 2017
Content may be subject to copyright.
UNCORRECTED PROOF
DEBATE
concept stretching: the case of
deliberation
ju
¨rg steiner
a,b
a
University of North Carolina at Chapel Hill, Chapel Hill, NC, USA
E-mail: jsteiner@email.unc.edu
b
University of Bern, Untere Wart 30, 3600 Thun, Switzerland
doi:10.1057/palgrave.eps.2210186
Abstract
Sartori (1970) warned a long time ago of the danger of concept stretching
for effective and cumulative theory building. Such concept stretching has
happened with regard to deliberation, which has become a very faddish
term. For theoretically well-founded empirical research it is better
conceptually to distinguish clearly between strategic bargaining and
deliberation, although in the empirical political world the two concepts
are usually heavily intertwined.
Keywords deliberation; concept stretching; strategic bargaining
The concept of deliberation has be-
come so faddish that it is in danger
of being ‘stretched’ until it becomes
virtually a synonym for talk of any kind. A
long time ago, Sartori (1970) warned of
the dangers of concept stretching in
political science, and this warning is still
valid. Based on Sartori’s work, Collier and
Levitsky (1997) have also prominently
written about the danger of concept
stretching, in their case specifically in
relation to the concept of democracy.
One encounters the problem not only with
regard to deliberation but also with many
other concepts used in political science.
There are certainly different versions of
the concept of deliberation. But as Goodin
(2005: 183) points out, on the basis of a
review of the literature: ‘As regards
standards for what counts as ‘‘good’’
discourse and deliberation, there seems
to be an impressively broad scholarly
consensus’. Deliberation does not simply
mean talk of any kind; the concept has a
very specific meaning. For fruitful and
rigorous research, this specificity must be
kept. If the concept is stretched too far it
begins to mean everything and therefore
nothing.
A particularly extreme example of con-
cept stretching with regard to deliberation
is a paper by Austen-Smith and Fedder-
sen (2006). I use this example as a useful
illustration to make my point, although
there are other examples where I could
make the same point (Steiner et al, 2004:
43–52). Austen-Smith and Feddersen use
the concept in its very opposite meaning
from the standard usage in the delibera-
tive literature. They present a formal
european political science: 00 2008
(1– 5) &2008 European Consortium for Political Research. 1680-4333/08 $30 www.palgrave-journals.com/eps
1
UNCORRECTED PROOF
mathematical model in which they estab-
lish theoretically under what conditions
members of political committees are will-
ing to share private information with
other committee members. Their exam-
ple is a jury consisting of three members,
who have to choose whether to acquit or
to convict a defendant. Two jury members
are cautious and wish to have at least
three pieces of evidence of guilt to
convict; the third jury member is less
cautious and considers one piece of
evidence of guilt as sufficient to convict.
The question is under what conditions the
jury members are willing to share their
private information on guilt with the other
jury members. The assumption is that
when it serves their individual prefer-
ences, jury members share such informa-
tion, when it does not, they withhold the
information since in these cases ‘there is
no downside risk of lying’ (p. 211). The
specific research question is whether
such lying occurs more often under
majority rule or rule of unanimity. The
model of Austin-Smith and Feddersen
predicts that lying is more likely under
unanimity than under majority rule. We
are not concerned here with the content
of this hypothesis. Our interest is rather in
why they use the concept of deliberation
throughout their paper and even in its
title.
Austen-Smith and Feddersen cite two
works of the deliberative literature, so
evidently they are aware of this literature.
Their usage of the concept of deliberation
is therefore not a harmless error or a
result of inattentiveness. I will now argue
that they stretch the concept in a way that
is ‘beyond-the-pale’. One key element of
the deliberative model is that actors do
not lie but are truthful and authentic in
their statements. According to Habermas
(1983: 98), ‘each speaker may only
assert what he believes himself’. This
criterion of truthfulness means that all
participants are open about their true
preferences and do not try to deceive
and mislead others. Put in another way,
participants really mean what they say.
When they refer to particular values such
as social justice, they truly believe in
these values and do not utter them
merely for tactical reasons to increase
their chances to win with their own
preferences.
Perhaps most politicians are indeed not
truthful and lie whenever it serves their
preferences. Therefore, it can certainly
be a useful exercise to model such situa-
tions, and I have no objection to such
modelling. But Austen-Smith and Fedder-
sen would do better not to use delibera-
tion for such situations, since the concept
has a specific meaning in political
philosophy. Its core is that political actors
should be willing to change their pre-
ferences by the force of better argument
in truly listening to the points of view of
others. To quote Habermas (1996: 305),
actors should be open to ‘the unforced
force of the better argument’. What the
better argument is, is not a priori given,
but must be searched for in common
discourse. It is through such discourse
that participants find out what counts as a
good argument. The model of Austen-
Smith and Feddersen violates this key
assumption of the deliberative model in
an extreme way. During the simulated
jury session, the three members do not
change their preferences on how much
information on guilt is necessary for con-
viction. Their preferences are assumed
to be fixed. No exchange of arguments
takes place.
I am aware that in other formal models,
preferences may be changed on the basis
of new information or a new choice
environment (Gilligan and Krehbiel, 1990).
The assumption in such models is that
only ultimate preferences are fixed. If it
turns out that currently held pre-
ferences are in contradiction with these
ultimate preferences, it is rational to
correct the error and to change the
currently held preferences so that they
european political science: 00 2008 deliberation
2
UNCORRECTED PROOF
become compatible with the ultimate
preferences. In this way, formal models
come closer to the deliberative model
in the sense that actors are willing to
consider information procured by other
actors. The ultimate preferences, how-
ever, are never changed by the force of
the better argument. In the model of
Austen-Smith and Feddersen, no change
of preferences is assumed, not even at a
lower level of abstraction. The three jury
members stick to their preferences with
regard to the amount of evidence of guilt
necessary for conviction. Therefore, the
concept of deliberation seems particularly
inappropriate. Austin-Smith and Fedder-
sen are, of course, entitled to use
concepts however they wish. But it would
be helpful for the scholarly community if
they would label the talk that is going on
in their model as strategic bargaining,
which seems a more appropriate term for
what happens in their model.
There are other important elements of
the deliberative model that are comple-
tely missing in the paper of Austen-Smith
and Feddersen. According to the delib-
erative literature, claims must be exten-
sively justified with logical reasons and
supporting evidence. The merits of an
argument need to be expressed not
merely considering one’s own interests
but also the interests of others. Partici-
pants must treat each other with genuine
respect. Taking all these elements to-
gether, deliberation is a rather complex
concept, and it should be used with care.
In what sense exactly do Austen-Smith
and Feddersen stretch the meaning of
deliberation? When members of a political
committee, in their case a jury, share
information only when it serves their
individual preferences, it is strategic talk
in a pure form. In order to reach one’s
preferences, actors lie whenever it serves
their individual purposes. There is no
concern and respect for the preferences
and arguments of others. What counts
are only one’s own preferences. To attain
these preferences, actors do anything,
including lying. In a nutshell, political
actors are assumed to be individual utility
maximisers.
I grant that such utilities may refer in
some more sophisticated formal models
not only to narrow self-interest such as
being elected or making money but also
to the common good. But what the
common good is, is defined on an indivi-
dual basis. An actor has a personal
preference of how the common good
should look like. How others define the
common good is not relevant. Even in
such sophisticated models there is no
exchange of arguments about the best
possible form of the common good. In the
model of Austen-Smith and Feddersen no
reference at all is made to the common
good. The preferences of the three jury
members with regard to the amount of
evidence of guilt needed for conviction is
simply assumed without any justification
with regard to the common good or other
moral values. The preferences are exter-
nal to the model, and the actors in the
model do not give any arguments as to
why they hold these preferences. In the
deliberative model, by contrast, actors
are assumed to give elaborate reasons for
their preferences.
There is now an increasing literature on
empirical studies of deliberation where
the concept is defined in a manner con-
sistent with its meaning in philosophy.
1
Empirical measures have been developed
with high reliability and validity to estab-
lish the level of deliberation, which has
enabled testing hypotheses about the
antecedents and consequences of varia-
tion in the level of deliberation. We should
be aware that the deliberative model is a
Weberian ideal type. The real world of
politics hardly ever corresponds fully to
‘deliberation is a rather
complex concept, and it
should be used with care’
ju
¨rg steiner european political science: 00 2008
3
UNCORRECTED PROOF
this ideal type. Habermas (1996: 323)
acknowledges that
(e)ven under favorable conditions, no
complex society could ever correspond
to the model of purely communicative
social relations yrational discourses
have an improbable character and are
like islands in the ocean in everyday
praxis.
Given the ideal-typical formulation of
the deliberative model, the empirical
research task is to establish how far away
a particular speech situation is from the
ideal type. Such research has been done
in particular for international negotia-
tions, debates in national parliaments,
discussions among ordinary citizens, local
politics, non-governmental organisations,
internet communication. To establish the
level of deliberation in these different
settings, a wide variety of research meth-
ods has been used: participant observa-
tion, focus groups, experiments, analysis
of transcripts, interviews.
2
The formal
model of Austen-Smith and Feddersen is
as far away from the deliberative ideal
type as one can imagine. Indeed, it
contains not a single deliberative element
but is exclusively built on strategic action.
Therefore, to label the talk taking place in
their model as deliberative is troubling
from a conceptual point of view.
There is increasing collaboration on
deliberation between philosophers and
empirically oriented scholars. As Haber-
mas (2005: 385) puts it, ‘philosophical
analysis (on deliberation) assumes more
and more features of an empirical re-
search’. At the philosophical level, there
are many conflicting hypotheses about
the antecedents and consequences of a
high level of deliberation, for example
whether public meetings or close-door
meetings are more conducive to a high
level of deliberation or whether rules of
unanimity or majority rules make delib-
eration more likely. Empirical studies help
to sort out such hypotheses. Rosenberg
(2007) has edited a volume where
empirical scholars present their research
on deliberation and then philosophers
discuss the impact of this research on
the future philosophical debate on delib-
eration. This debate about the interplay of
empirical and philosophical research on
the deliberative model is confounded if
scholars liked Austen-Smith and Fedder-
sen use the concept of deliberation in a
way that has nothing to do with the way it
is used in democratic theory. Deliberative
theorists are very careful as to how they
define their concepts. Stretching the con-
cept of deliberation as Austen-Smith and
Feddersen do, leads to theoretical vacuity
and practical confusion.
Finally, I wish to argue that it is not
merely a question of terminology as to
how we use the concept of deliberation. It
also matters substantively as to how we
approach at a very fundamental level our
research enterprise in political science.
We may try to integrate the rational
choice model and the deliberative model
into an overall super model. Although
such a super model may be a long-term
‘Empirical measures
have been developed
with a high reliability and
validity to establish the
level of deliberation’
‘the rational choice
model can certainly
much better predict how
the price of oil on
international markets
is set; but perhaps
the deliberative model
is better when it comes
to Aids policies on a
world scale’
european political science: 00 2008 deliberation
4
UNCORRECTED PROOF
goal, in my view, it is better to keep the
two models separate since they are based
on diametrically opposed views of basic
human nature. Both models allow the
formulation of predictions about the poli-
tical world, and one can then establish
which model is a better fit with empirical
reality. If we take the international field,
the rational choice model can certainly
much better predict how the price of oil on
international markets is set; but perhaps
the deliberative model is better when it
comes to Aids policies on a world scale. In
the domestic field, the deliberative model
may be more appropriate to countries
with power sharing institutions than those
based on the Westminster model (Steiner
et al, 2004). Furthermore, the delibera-
tive model may do better in the early
stages of a decision process whereas the
rational choice model may be superior in
the later stages. It is an exciting research
agenda to compare in a systematic way
how the predictions derived from both the
rational choice model and the deliberative
model perform in the real world of poli-
tics. Such a research agenda will also
help to build bridges in the discipline. A
precondition is that we are careful of
how we use our concepts. Sartori was
correct that good research begins with
clear conceptualisations.
Notes
1 See the two special issues of Acta Politica on Empirical Approaches to Deliberative Politics, Volume 40,
numbers 2 and 3, 2005. See also the papers in the workshop on deliberation, ECPR Joint Workshops,
Helsinki, 7–12 May 2007.
2 See literature in note 1.
References
Austen-Smith, D. and Feddersen, T.J. (2006) ‘Deliberation, preference uncertainty, and voting rules’,
American Political Science Review 100(2): 209–218.
Collier, D. and Levitsky, S. (1997) ‘Democracy with adjectives: conceptual innovation in comparative
research’, World Politics 49(3): 430–451.
Gilligan, T. and Krehbiel, K. (1990) ‘Organization of informative committees by a rational legislature’,
American Journal of Political Science 34: 531–564.
Goodin, R.E. (2005) ‘Sequencing deliberative moments’, Acta Politica 4(2): 182–196.
Habermas, J. (1983) Moralbewusstsein und kommunikatives Handeln
Q1 , Frankfurt a. M.
Habermas, J. (1996) Between Facts and Norms. Contributions to a Theory of Law and Democracy,
Cambridge, MA: MIT Press.
Habermas, J. (2005) ‘Concluding comments on empirical approaches to deliberative politics’, Acta Politica
40(3): 384–392.
Rosenberg, S.W. (ed.) (2007) Deliberation, Participation, and Democracy, London: Palgrave Macmillan.
Sartori, G. (1970) ‘Concept misformation in comparative politics’, American Political Science Review
64(4): 1033–1053.
Steiner, J., Ba¨chtiger, A., Spo¨rndli, M. and Steenbergen, M.R. (2004) Deliberative Politics in Action.
Analyzing Parliamentary Discourse, Cambridge: Cambridge University Press.
About the Author
Ju
¨rg Steiner is professor emeritus at the University of North Carolina at Chapel Hill and the
University of Bern. In 2003-04 he held the Swiss Chair at the European University Institute in
Florence. Recent book include Deliberative Politics in Action (Cambridge University Press,
2004) and European Democracies (Longman, 2007).
ju
¨rg steiner european political science: 00 2008
5