Relative Extremism and Relative Moderation

• Published date: 01 June 2009
• Author: Paul V. Warwick
• Date: 01 June 2009
• DOI: 10.1177/1065912908320663
• Subject Matter: Articles

Political Research Quarterly
Volume 62 Number 2
June 2009 276-288
© 2009 University of Utah
Relative Extremism and
10.1177/1065912908320663
http://prq.sagepub.com
Relative Moderation
hosted at
http://online.sagepub.com
Strategic Party Positioning in Democratic Systems
Paul V. Warwick
Simon Fraser University, Burnaby, B.C., Canada
This article investigates the ways in which parties stake out left–right positions that deviate from the mean positions of
their supporters. Previous research has shown that parties tend to adopt positions that are more extreme than those of their
supporters, but there are at least two arguments that also imply the presence of relative moderation—a tendency for
moderate parties to be more moderate than their supporters. Using surveys covering 34 countries compiled by the
Comparative Study of Electoral Systems, this investigation shows that parties in coalitional systems display both
phenomena.
Keywords:
political representation; party behavior; relative extremism; relative moderation; bilateralism
One of the most common findings in the literature
The investigation takes place using the election sur-
on party behavior is that, even in two-party sys-
veys in Modules 1 and 2 of the Comparative Study of
tems, parties do not converge on the center of the
Electoral Systems (CSES; 2003, 2007) data set, which
left–right ideological spectrum but instead establish
cover a wide range of countries and elections between
and maintain distinct positions along it.1 This finding
1996 and 2006. The main objective is to determine
would seem to follow naturally from the fact that
whether and under which circumstances parties posi-
voters themselves are spread out in left–right terms,
tion themselves either further from or closer to the cen-
but this is true only up to a point. Parties, it turns out,
ter of the left–right spectrum than their supporters. The
do not represent their supporters’ views as accurately
analysis indicates that, notwithstanding the close over-
as they might. Indeed, the available evidence indi-
all alignment of party and supporter positions, there is
cates that they display “relative extremism” (Warwick
significant evidence of a complex pattern combining
2004)—a tendency to adopt positions located further
both relative extremism and relative moderation.
from the ideological center than their supporters, on
Further investigation shows that voters are aware only
average, place themselves.
of the former effect and that the existence of both
The purpose of this article is to attempt to unravel
effects probably has more to do with coalition maneu-
this puzzling result. This task is a formidable one,
vering than with vote maximization.
since there are a number of possible explanations and
not all of them can be assessed adequately using
available survey data. But there is one factor that
The Theoretical State of Play
bifurcates potential explanations and thereby offers
some new purchase on the issue: Although some
We begin with a consideration of the possible
explanations anticipate relative extremism across the
sources of the relative extremism effect. Why would
board, others also imply the presence of both it and
parties adopt positions that are more extreme than
what might be termed “relative moderation.” Whether
those of their own supporters? One apparent answer
they are to maximize votes in elections or to influence
that has received a lot of attention in the past two
coalition outcomes, these latter explanations expect
decades comes from the directional theory of voting,
some parties to adopt more extreme positions and
first proposed by Rabinowitz and Macdonald (1989).
others more moderate positions than their supporters.
In seeking to understand the phenomenon of relative
Paul V. Warwick, Professor of Political Science, Simon Fraser
extremism, we therefore look for its opposite as well.
University; e-mail: warwick@sfu.ca.
276

---

Warwick / Party Positioning in Democratic Systems
277
In this theory, voters prefer parties to the extent that
out any force than a party at the opposite side of the
they adopt their side of an issue and express it with
force with respect to the voter” (p. 187). Her analyses
great (but not unacceptably great) emotional intensity.
of survey data from a range of parliamentary systems
On closer examination, however, this proves to be a
generally support this expectation, with the more
false lead. It is not just that various tests of directional
consensual systems showing the larger compensa-
theory have failed to yield clear evidence in its favor;2
tional voting effects (Kedar 2005a, 2005b).
the more fundamental problem is that, strictly speak-
Kedar’s reasoning and analysis extended only to
ing, the theory does not address positional extremeness
voters, but there is no reason why it should stop there.
at all. Although tests have repeatedly used extremeness
If party leaders anticipate that policy making will
as a proxy for emotional intensity, Rabinowitz and
involve compromise across party positions, it is rea-
Macdonald make very clear that the two are separate
sonable to expect that they, too, will be driven by a
concepts that bear no necessary relationship to each
compensational logic. Accordingly, left-wing parties
other.3 Thus, even if voters are attracted by emotional
will exaggerate their leftness in order to move the
intensity, it does not follow that they prefer parties that
eventual policy outcome closer to their true prefer-
adopt relatively extreme positions.
ences, and right-wing parties will exaggerate their
It is, of course, not very logical that they would;
rightness for the same reason. This clearly could
other things being equal, voters ought to prefer their
account for the relative extremism phenomenon—
own positions foremost. But parties need not accept the
provided, of course, that policy outcomes tend to
positions of their target audiences as immutable; they
gravitate toward the middle of the left–right spec-
may stake out somewhat different positions in hopes of
trum. Kedar (2005b, 188) explicitly endorses this
molding or influencing them.4 Holmberg (1989) and
assumption. She notes that minority governments
Iversen (1994) believe that this is particularly likely to
depend on external parliamentary support and often
be the case for leaders of parties intent on radical
negotiate support arrangements with noncoalition
change, who may adopt positions that are more extreme
parties, that committees and other consultative
than those of their (potential) voters with this goal in
arrangements may bring an element of opposition
mind. Iversen’s analysis of elite and voter opinion in 34
influence on policy, and that opposition parties can
West European parties supports this view. Since parties
influence the policy agenda by raising issues and
with this type of goal tend to occupy positions at or near
forcing governments to address them.6 Additional
the political extremes, the result might be a pattern in
support for this position comes from McDonald and
which relative extremism becomes more prevalent the
Budge (2005). Using data on 21 democracies span-
further a party is from the political center.
ning the postwar era to 1995, they find a near one-to-
Another perspective on relative extremism is sug-
one correspondence between the weighted mean
gested by two recent analyses of voting behavior con-
left–right position of government parties (using par-
ducted by Kedar (2005a, 2005b). Kedar’s point of
liamentary seat shares as weights) and the left–right
departure is a suggestion by Downs (1957) that “the
position of the median parliamentary party
most important part of a voter’s decision is the size of
(McDonald and Budge 2005, 194). If the latter accu-
his current party differential, i.e. the difference
rately reflects the center of the left–right spectrum in
between the utility income he actually received in
the legislature and the former accurately reflects gov-
period t and the one he would have received if the
ernment policy, the two would appear to be in close
opposition had been in power” (p. 40).5 If voters com-
alignment, perhaps for the reasons Kedar suggests.
pare (anticipated) outcomes in this way, Kedar rea-
Attempts to move either the beliefs of party sup-
sons, the result may not be a decision to vote for the
porters or the policy outputs of governments away from
party that is ideologically the closest. What also mat-
the center might explain a tendency toward relative
ters is the institutional environment and, in particular,
party extremism, but these are not the only considera-
whether policy formulation is likely to involve some
tions that could cause party and supporter positions to
degree of power sharing. Where a minority and/or
diverge. Indeed, there are considerations that might
coalition government is the likely outcome, for
induce parties to move toward, rather than away from,
instance, Kedar (2005b) suggests that “outcome-
the political center. One such consideration—the
oriented voters may prefer a party to their right (left)
tendency for the political center to be relatively vote
if the center-of-gravity in the parliament is to their
rich—comes to the fore in Schofield and Sened’s
left (right) to a party whose position is identical to
(2006) recent model of multiparty politics.
their own. Other things equal, a party placed exactly at
In the Schofield–Sened model, voters make their
the voter’s position will be less effective in balancing
...