Content uploaded by David Matthew Farrell
Author content
All content in this area was uploaded by David Matthew Farrell on Oct 07, 2015
Content may be subject to copyright.
Full Terms & Conditions of access and use can be found at
http://www.tandfonline.com/action/journalInformation?journalCode=fbep20
Download by: [University College Dublin] Date: 07 October 2015, At: 13:26
Journal of Elections, Public Opinion and Parties
ISSN: 1745-7289 (Print) 1745-7297 (Online) Journal homepage: http://www.tandfonline.com/loi/fbep20
Expert opinion on electoral systems: So which
electoral system is “best”?
SHAUN BOWLER , DAVID M. FARRELL & ROBIN T. PETTITT
To cite this article: SHAUN BOWLER , DAVID M. FARRELL & ROBIN T. PETTITT (2005) Expert
opinion on electoral systems: So which electoral system is “best”?, Journal of Elections, Public
Opinion and Parties, 15:1, 3-19, DOI: 10.1080/13689880500064544
To link to this article: http://dx.doi.org/10.1080/13689880500064544
Published online: 16 Aug 2006.
Submit your article to this journal
Article views: 1925
View related articles
Citing articles: 11 View citing articles
Journal of Elections, Public Opinion and Parties
Vol. 15, No. 1, 3–19, April 2005
ISSN 1368-9886 Print/1745-7297 Online/05/010003-17 © 2005 Taylor & Francis Group Ltd
DOI: 10.1080/13689880500064544
Expert Opinion on Electoral Systems:
So Which Electoral System is “Best”?
SHAUN BOWLER*, DAVID M. FARRELL** & ROBIN T. PETTITT**
*Department of Political Science, University of California Riverside, Riverside, CA, USA, **GIPP,
School of Social Sciences, University of Manchester, Manchester, UK
Taylor and Francis LtdFBEP106437.sgm10.1080/13689880500064544Journal of Elections, Public Opinion and Parties0000-0000 (print)/0000-0000 (online)Original Article2005
Taylor & Francis Group Ltd1510000002005DavidFarrellGIPP, School of Social SciencesUniversity of ManchesterManchesterUKdavid.farrell@manchester.ac.uk
A
BSTRACT
Electoral systems are now much more prone to reform than 10–20 years ago,
and as a consequence electoral system specialists are developing a useful sideline in advising
on electoral system design across the globe. Given that in many of the recent cases of
electoral system design some variant of a “mixed-member” electoral system has been
adopted, one might be forgiven for concluding that the electoral system specialists are them-
selves leaning towards this particular electoral system family. So what do they really think?
This paper reports the responses to a recent survey of electoral system specialists (members
of the relevant specialist groups of the Political Studies Association, American Political
Science Association and International Political Science Association) that was funded by the
McDougall Trust. It asks and seeks to answer two questions. First, is there an electoral
system which scholars agree to be the best? Second, if there are disagreements about
electoral systems, are the disagreements about the properties of electoral systems or the
normative values of democracy?
The empirical study of electoral systems and their effects is now well established.
The social choice literature has done much to uncover the early theoretical work on
electoral systems by writers as diverse as Condorcet and Charles (Lewis Carroll)
Dodgson (McLean & Urken, 1995). Alongside this theoretical literature has run a
long-standing interest in the empirical effects of electoral systems, both scholarly
and practitioner. A good example of the latter is the Electoral Reform Society which
was formed in 1884. Good early examples of scholarly interest include Commons
(1907), Humphreys (1911), Hoag and Hallett (1926), or Lakeman and Lambert
(1959). These have subsequently been joined by a succession of studies that embrace
more countries and more sophisticated analyses (for example, Rae, 1971; Taagepera
& Shugart, 1989; Lijphart, 1994; Katz, 1997). This is hardly an exhaustive list of
books on the topic (let alone the fact that the numerous articles have not been
Correspondence Address:
Shaun Bowler, Department of Political Science, University of California
Riverside, Riverside, CA 92521, USA. Email: shaunb@citrus.ucr.edu; David Farrell, GIPP, School of
Social Sciences, University of Manchester, Manchester M13 9PL, UK. Email: david.farrell@manches-
ter.ac.uk; Robin T Petitt, GIPP, School of Social Sciences, University of Manchester, Manchester M13
9PL, UK. Email: RTPettitt@yahoo.co.uk
Downloaded by [University College Dublin] at 13:26 07 October 2015
4
S. Bowler
et al.
included), but it serves to show the long and rich history of the study of electoral
systems. At this point, it might be reasonably considered what kinds of areas of broad
agreement, or indeed disagreement, exist among experts in this field. There are
several reasons why such questions are worth contemplating.
First, the “third wave” of democratization has meant that a series of countries
have expressly chosen electoral systems for their newly democratic political
systems. This means that expert opinions, aided in part by organizations such as
International Institute for Democracy and Electoral Assistance (IDEA), Interna-
tional Foundation for Electoral Sysems (IFES), or the Westminster Foundation have
had a chance to influence policy. And it is not just new democracies that are consid-
ering such questions. After years in which the established argument (apart from a
few exceptional cases such as Australia and France) was that electoral systems were
best left unchanged, suddenly within the last decade or so established democracies
like Italy, New Zealand, Japan and Israel have all made fundamental reforms to their
electoral systems. The issue has been raised in other countries, such as Canada, the
Netherlands and Ireland. And, closer to home, the Labour government, when elected
in 1997, initiated (through the establishment of the Jenkins Commission) a high-
level debate over the system used to elect MPs, a debate which Labour (newly re-
elected in 2001) has promised to revisit about now. In short, then, electoral system
design seems to be one of the more solid areas of institutional engineering (Sartori,
1977). Just how much consensus or disagreement exists among experts is thus
relevant to the kinds of advice people have been giving and countries receiving.
Second, in a more purely academic sense a cursory glance at the available academic
texts reveals some significant disagreements between electoral system specialists.
Given the likelihood of disagreement within
any
academic community it is likely that
a survey of experts will show some dissent, but the disagreements do seem to go
beyond that to suggest that different academics (and possibly academic communities)
support different kinds of electoral systems. Among the more recent cases of advice
on electoral system design there is the following range of options: Blais and Massi-
cotte (1996), Horowitz (1991), Reilly (1997) and Sartori (1997) all favour some form
of majoritarian system; Reynolds (1999), and Taagepera and Shugart (1989) reveal a
sympathy for single transferable vote (STV) (though this appears more a case for
Taagepera than for Shugart: compare Taagepera, 1998 with Shugart, 2001); Lijphart
(1994) promotes list systems with two-tier districting; Dunleavy and Margetts (1995)
and Shugart and Wattenberg (2001) reveal some sympathy for mixed systems. Katz
(1997) is one of the few to pointedly hedge his bets, stating quite simply that the deci-
sion on which system is best depends on circumstances. In examining the views of
experts in this area some sense can be gained of how broadly shared are these opinions:
do the advocates of systems only speak for themselves or are they expressing a wider
consensus over fundamental features of electoral systems? More to the point, do
disagreements about the nature of electoral systems reflect differences over which
system fulfils a broadly agreed upon set of values or are they more serious disagree-
ments over underlying values of elections? What is the nature of the disagreement
experts may have over electoral systems? After over 100 years of study in this area,
Downloaded by [University College Dublin] at 13:26 07 October 2015
Expert Opinion on Electoral Systems
5
it might be expected to see general agreement that the choice of electoral systems
comes down to a series of two or three broadly agreed criteria, albeit with some
disagreement over which electoral system best fulfils these criteria. Alternatively, and
more fundamentally, there may be disagreement over what the criteria should be by
which electoral systems might be judged.
If disagreements do exist, then, the assumption might be that they are more likely
to be disagreements over underlying values than over which system best meets a set
of defined criteria. The study of electoral systems is a rare subfield within political
science in that it has developed some relatively hard and fast measures of electoral
system properties that are, in the main, broadly accepted. For example, the Gallagher
Index is a widely accepted measure of disproportionality (Gallagher, 1991; for
discussion, see Taagepera & Grofman, 2003). By this measure it is clear which
systems do “well” (large district magnitude [M] and proportional representation
[PR]) and which do poorly (single member plurality). One might therefore plausibly
expect that because the literature – and hence the experts in this area – possess a
series of well-defined measures of electoral systems, then there should be a high
degree of consensus, since there can be very few disagreements of fact over which
systems perform well according to which measure. In contrast, any disagreements
among experts will be over which measures are the most important yardsticks to use
in judging electoral systems. This choice of different measures will reflect differ-
ences in the underlying dimensions of electoral systems.
In some ways, the preceding discussion is not terribly surprising. What has
been said is that in any survey of experts on electoral systems there are reasons
for expecting to find either a great deal of consensus or a great deal of disagree-
ment. Consensus might be expected because the literature on electoral systems is
so well developed and, also, within that literature, there are a set of widely
accepted measures and yardsticks. On the other hand, there may be a great deal of
disagreement partly because the literature reflects a certain amount of this but
partly, too, because electoral systems may well embody varying dimensions of
democracy, with different electoral systems capturing different dimensions (for
further discussion, see Bowler & Grofman, 2000): for instance, electoral system A
may be good on dimension 1 but bad on dimension 2, while for B the situation is
reversed.
Saying that electoral systems are multidimensional and, further, that expert opin-
ions may reflect that multidimensionality and so be divided may not appear terribly
novel. The innovation in this study is in uncovering what those dimensions of
disagreement consist of. To illustrate why, let us take an example from outside this
study. Single member plurality (SMP) is a favourite “whipping boy” of electoral
systems experts (for an exception, see Pinto-Duschinsky, 1999) on the grounds that
it is an “unfair” system. But there are many definitions of the term “unfairness”, and
some definitions show SMP in a much more favourable light than others. However,
developing this kind of argument, made by Blau (2004), really depends on us
discussing electoral systems in terms of their normative, rather than arithmetic,
properties. And, as Blau’s study shows, it is quite surprising that in such a long-
Downloaded by [University College Dublin] at 13:26 07 October 2015
6
S. Bowler
et al.
standing literature as that on electoral systems, there is relatively little discussion of
underlying democratic values of the systems themselves. In other words, stating that
views on electoral systems reflect different values is one thing, presenting evidence
establishing what those values are, and just how many values are important, is quite
another. Furthermore, if this is indeed the nature of the disagreement, then there are
consequences for how confident, or at least how careful, our claims on behalf of
institutional engineering need to be.
Such questions lay behind the rationale for this study. Its purpose was to find out
the views of electoral system experts about the world’s main electoral systems,
which are grouped into seven main families in Table 1 (for further details, see
Farrell, 2001). Obviously, reality is more complex than that summarized in the table
(and our questionnaire was somewhat more nuanced to allow for more degrees of
variation), but this table does at least give a flavour of the set of systems over which
experts were making judgements.
The first objective was to identify our electoral system experts. For the purpose of
this exercise use was made of the membership lists of the relevant specialist groups
of the three main political science associations, namely: the Electoral Systems and
Representation section of the American Political Science Association, the Compara-
tive Representation and Electoral Systems Research Committee of the International
Political Science Association, and the Elections, Public Opinion and Parties special-
ist group of the British Political Studies Association. After some pilot testing to
finalize the questionnaire, it was sent out in two waves in late 2003/early 2004,
resulting in 170 responses.
1
In the light of the preceding discussion, the remainder of this paper is structured
around three questions:
Table 1.
The main families of electoral systems in the 1990s
Electoral systems
Proportion
of countries
(%)
Proportion of
population
(%) Average disprop.
(GI)
Average no.
of parties
(N
s
)
Single member plurality 18.6 52.8 12.3 3.4
Alternative vote 1.7 0.6 10.3 2.5
Two-round 3.4 2.3 21.4 2.3
List 47.5 18.3 3.9 4.5
Single transferable vote 1.7 0.1 5.4 3.2
Mixed member proportional 6.8 3.9 3.8 4.3
Mixed member majoritarian 20.3 21.9 9.9 4.7
Notes:
Only including countries with a population greater than two million (1998 estimates),
and with a Freedom House score of 4.0 or less (1998–99 rankings). GI = Gallagher (Least
Squares) index of disproportionality. N
s
= Laakso/Taagepera index of effective number of
parliamentary parties.
Source:
Farrell (2001).
Downloaded by [University College Dublin] at 13:26 07 October 2015
Expert Opinion on Electoral Systems
7
1) Is there one system favoured by experts on electoral systems, and, if so, which?
2) If there is disagreement, what is the nature of that disagreement?
3) What explains the choice of electoral system by our experts?
Is There One System Favoured by Experts on Electoral Systems and, if so,
Which?
Determining the voting method to use in order to establish which voting system is
preferred turned out to be quite easy since each method produced the same result: in
short, regardless of how you cut it, mixed-member proportional (MMP) is the
preferred choice of electoral system experts.
2
This can be seen in a number of ways
(see Tables 2 and 3). Our experts were asked to rank the systems. MMP won a
plurality (though not a majority) of first preferences and an average ranking higher
than all other systems. Using these rankings, MMP is a Condorcet winner in the
sense that, in a pair-wise comparison, it beats all other systems. It also wins in a
Borda Count. By the same methods, single non-transferable vote (SNTV) is the least
preferred of the electoral systems. It is tied for the smallest number of first prefer-
ences, has the lowest average ranking, and is a Condorcet loser. The second
preferred system is single transferable vote (STV), coming a close second to MMP
on all the means of rating electoral systems and beating all the remaining electoral
systems in a pair-wise comparison. While MMP gets about one-third of all first
preferences, STV receives about one-fifth.
The fact that MMP generates the most support among electoral system experts
begs the question of why that should be the case. Moreover, these results also reveal
that while MMP is extremely popular, so too is STV. It seems that these two
systems seem to generate the two main poles of appeal for election system experts
and they help pose our second question: what is the nature of that disagreement?
Table 2.
Overall rankings of electoral systems
Rank Average Number 1st prefs
MMP 1 2.37 52
STV 2 2.60 38
Open list PR 3 3.26 18
AV 4 4.01 10
Closed list PR 5 4.17 9
SMP 6 4.67 21
Runoff 7 4.9 7
MMM 8 5.18 3
SNTV 9 6.76 3
Notes
: MMP (mixed member proportional); STV (single transferable vote); AV (alternative
vote); SMP (single member plurality); MMM (mixed member majoritarian); SNTV (single
non-transferable vote).
Source
: McDougall Trust expert survey (total N = 169).
Downloaded by [University College Dublin] at 13:26 07 October 2015
8
S. Bowler
et al.
What is the Nature of that Disagreement?
Here two points can be made. First, as will be seen, there is some consensus over
which system does what. Second, there is disagreement over what electoral systems
should actually do.
Some Consensus over Properties of Electoral Systems
The properties of systems are first of all to be examined. As has been noted above, it
would be reasonably expected to see a fair amount of consensus over which system
does what best: and this is indeed the case. Table 4 presents average evaluations of
which systems do best at what and also presents the standard deviations in those
rankings.
These measures give some way of assessing consensus and where it lies: lower
average rankings indicate that our respondents are in greater agreement that a
system fulfils a certain property. The standard deviations give some sense of what
properties people agree on and where they disagree. Since the figures in Table 4 can
seem something of a jumble of numbers and be difficult, or at least tedious, to sort
through, a number of features of interest can be singled out, as is shown in Tables 5
and 6, which, in turn, show striking consensus over the ability of SMP to promote
constituency service, and the ability of list PR to accomplish proportionality.
But there remain areas of disagreement – and the standard deviations presented in
Table 4 are one way to identify which systems our experts disagree most over (list
Table 3.
MMP as Condorcet winner; SNTV as Condorect loser
SMP AV Runoff SNTV STV CLPR OLPR MMP MMM
SMP 43–89 64–62 91–28 43–97 66–69 61–80 43–105 81–36
AV 85–33 99–17 33–101 64–60 55–76 41–100 85–29
Runoff 72–25 31–103 42–74 44–82 23–115 66–33
SNTV 5–115 22–77 18–89 11–122 23–49
STV 115–5 92–37 85–46 58–87 106–21
CLPR 77–22 27–80 29–102 71–28
OLPR 89–18 40–94 85–23
MMP 122–11 121–9
MMM 49–23 9–121
LOSER WINNER
Notes:
Figures are comparison of row system to column system, for example, top left hand
cell, 43–89, is a comparison of the ranking of SMP to AV: 43 people liked SMP, 89 liked AV,
which means AV wins out in this pair-wise comparison. Counting rules: some electoral
systems were not ranked by the experts. Any ranked system ‘beats’ an unranked one. If both
systems were unranked they are counted as missing.
Downloaded by [University College Dublin] at 13:26 07 October 2015
Expert Opinion on Electoral Systems
9
PR and SMP), and also which properties our experts disagree most over relative to
other areas (mostly the governability issues: cohesive parties, effective government
and accountability).
3
Two observations can be made at this point. First, despite a
degree of consensus, there remain disagreements over which system has which
property. One interpretation is that despite the lengthy literature, political science
has yet to develop widely accepted yardsticks and benchmarks equivalent to that of
Table 4.
Average rankings and standard deviation (SD) of system properties (Rankings are
1 = best, 2 = next best and so on)
SMP AV Runoff SNTV STV List Mixed
Representation of
minorities Mean 5.41 4.28 5.22 4.17 2.08 1.34 2.31
SD
1.75 1.35 1.19 1.23 0.93 0.75 0.86
1.15
Accountability of
government Mean 1.92 2.71 2.85 4.89 3.59 3.80 3.19
SD
1.60 1.28 1.42 1.57 1.84 2.22 1.77
1.67
Proportionality Mean 5.85 4.45 5.25 4.21 2.21 1.17 2.34
SD
1.60 1.28 1.42 1.57 1.84 2.22 1.77
1.67
Effective government Mean 2.05 2.86 2.97 5.08 3.48 3.37 2.73
SD
1.66 1.30 1.43 1.44 1.82 2.19 1.68
1.65
Good constituency
representation Mean 1.82 2.53 3.09 4.61 2.79 4.57 3.11
SD
1.33 1.13 1.35 1.69 1.88 2.45 1.61
1.63
Cohesive parties Mean 2.40 3.38 3.54 4.88 4.21 1.99 2.55
SD
1.87 1.67 1.59 2.00 1.69 1.77 1.09
1.67
Average of rankings
2.78 2.89 3.27 3.98 2.62 2.32 2.32
Standard deviation
1.60 1.31 1.35 1.52 1.51 1.64 1.32
Table 5.
Ranking of ability of SMP to provide good constituency service
Rank Freq. Per cent
18062.50
22015.63
3107.81
4129.38
521.56
632.34
710.78
Total 128 100.00
Downloaded by [University College Dublin] at 13:26 07 October 2015
10
S. Bowler
et al.
the Gallagher Index over key concepts involving governability. Given this lack of
benchmarks, the recent work of Arend Lijphart (1999) and G. Bingham Powell
(2000) can be seen as especially important steps in the right direction.
Second, it is somewhat surprising that list PR does not score better in the overall
ranking of electoral systems. On average, list PR and MMP have the same rankings
on these criteria. It might, therefore, have been expected that list systems and not STV
would have been the runner-up to MMP. List systems come just after STV in the
overall rankings and so can be said to “do well” but not, perhaps, quite as well as one
might expect given the various scores reported in these tables. One explanation is that
respondents weigh some of these criteria more heavily than others. This, in turn helps
pose the question of whether scholars value different aspects of representation.
Some Disagreement over What Electoral Systems Should Do
What is it that electoral systems should do? That is, what properties should electoral
systems have? In general, the experts value proportionality of outcome as the most
desirable property of election systems. As Table 7 shows, in terms of average rank-
ing, number of 1st preferences and standard deviation, proportionality of outcome is
clearly the most important criterion for judging electoral systems.
It still remains curious that list PR systems do so poorly in our ranking. Propor-
tionality of outcome is seen as important and the personal role of the representative
as a fairly low priority, yet the proportional systems which allow choice of candi-
dates – open list PR and MMP, not to mention STV – do rather better than closed
list in the rankings. One possible explanation for this is that if the alternatives on
offer are between one which satisfies the main criterion (proportionality) with no
voter choice, and one that satisfies the same main criterion
with
voter choice, then
the tie is broken by choice. This line of inquiry is taken up in the next section.
There’s Something about MMP …
In looking at preference rankings of systems a number of approaches can be
adopted. One tack is to ask whether the preferences of experts cluster or are consis-
tent in any way. That is, if someone ranks system x high, what other systems do they
also rank highly? A second approach is to tie the valuations of different aspects of
democracy to different ratings of electoral systems in an attempt to understand just
what it is that leads our experts (us) to differ and whether these differences explain
Table 6.
Ability of list systems to promote proportionality
Rank Freq. Per cent
1 108 85.04
21612.60
332.36
Downloaded by [University College Dublin] at 13:26 07 October 2015
Expert Opinion on Electoral Systems
11
the choice of MMP. Both strategies are pursued here in order to arrive, by an admit-
tedly circuitous route, at the assessment (in the final section below) that it is not
clear why the experts are so fond of MMP.
Internal consistency among preference rankings is first of all considered. Table 8
presents intercorrelations of rankings. As can be seen, there are patches of consis-
tency. Those who rank SMP high also like the alternative vote (AV) and Runoff
systems, and these people tend to rank the PR systems – including MMP – lower. Or
this can be looked at the other way round: people who like PR – including MMP –
systems do not like SMP or AV (and so forth); those who rank C(losed) L(ist) PR
Table 7.
Desirable properties of electoral systems
Rank Average SD Number of 1st
Preference
Proportionality of outcome 1 2.83 1.89 45
Helps ensure stable government 2 3.36 2.28 34
Transparent relationship between seats and votes 3 3.43 2.02 22
Simple for voters to use 4 3.80 2.13 22
Allows voters to ‘kick the rascals out’ (i.e. unseat a
government) 5 3.94 2.61 28
Allows voters to choose an individual representative 6 4.58 2.39 12
Helps ensure representation for women and
minorities 7 4.68 2.04 4
Helps build up strong cohesive parties 8 4.84 2.42 9
Helps ensure coalition government 9 6.73 2.53 3
Note
: Question = “Electoral systems have many attributes. How would you rank order the
following set of attributes in terms of the criteria we should use for judging across electoral
systems?”
Table 8.
Correlation of ranking of systems (
all
respondents)
SMP AV Runoff SNTV STV CL PR OLPR
SMP 1.0000
AV 0.40* 1.0000
Runoff 0.46* 0.51* 1.0000
SNTV 0.10 0.23 0.28 1.0000
STV 0.03 0.30*
−
0.01 0.29* 1.0000
CL PR
−
0.17
−
0.35*
−
0.24*
−
0.19
−
0.09 1.00
OL PR
−
0.26*
−
0.27*
−
0.19
−
0.13
−
0.00 0.43* 1.0000
MMP
−
0.03
−
0.31*
−
0.23*
−
0.20
−
0.29* 0.08 0.00
*statistically significant
Downloaded by [University College Dublin] at 13:26 07 October 2015
12
S. Bowler
et al.
high also rank Open List PR high. There is
some
kind of consistency in preference
rankings of the preferential systems that group together and there are the PR systems
that have some sort of grouping. It will be returned to below as to whether this
means electoral systems can be placed along one dimension. For the moment the
question is as to whether differences in rankings related to differences in values and
this is done with an eye to trying to understand why it is that MMP is so popular.
One reason why MMP might be popular is because, even as it garners many
supporters in its own right, it is also a fairly high runner-up for those who generally
prefer either SMP or one of the list systems. To some extent this is true. Table 9, for
example, displays the average rating of various systems by those who ranked SMP
or one of the list systems first or second: namely, if SMP and the list systems anchor
the ends of a notional continuum, MMP (and STV) are towards the middle of this
continuum. As can be seen, for list and SMP supporters, MMP does appear to be
placed in the middle in the sense that it is not disliked nearly as much as the other
systems. In a sense, this is to be expected: MMP is a system that is held to be the
“best of both worlds” (Shugart & Wattenberg, 2000) and so supporters of the two
worlds gravitate towards it.
Two points can be made at this juncture. First, and most narrowly, the ranking of
STV is something of a surprise in terms of this “best of both worlds” schema. Its
position seems to suggest that it, too, is de facto seen as another possible best of both
worlds. Second, and more importantly, it could be asked whether the ranking of
MMP is because it realizes at least some of the normative values of democratic
governance and electoral system performance.
One simple way of seeing this is to correlate rankings of the electoral systems
with the rankings of criteria outlined in Table 7 above. This is shown in Table 10.
Responses to the comments in the leftmost column are ranked (1st most important,
Table 9.
Average preference ranking of systems by those who placed SMP or a list system
as their top two
If SMP ranked
1st or 2nd If CLPR ranked
1st or 2nd If OLPR ranked
1st or 2nd
Ranking of
SMP – 6 6.5
AV 3.3 5.2 4.9
Runoff 3.8 5.57 6.0
SNTV 6.4 7.08 6.95
STV 3.2 3.47 2.72
CL PR 5.5 – 3.37
OL PR 5 2.85 –
MMP 2.6 2.88 2.80
N352246
Downloaded by [University College Dublin] at 13:26 07 October 2015
Expert Opinion on Electoral Systems
13
2nd most important, and so forth) and these are correlated with the rankings of the
electoral systems. Thus, a low ranking of proportionality of outcome is associated
with a high ranking of SMP. On the other hand a high ranking of proportionality is
associated with a high ranking of MMP and open list PR. One of the findings of this
table is that expert opinion seems relatively well defined over both SMP and the list
systems: namely, there are clear and consistent correlates of rankings of those
systems grounded in assessments of what criteria should be important about elec-
toral systems. But this does not seem to be the case for STV or MMP. As can be
seen, the correlation coefficients are much smaller for these two systems. One inter-
pretation of this pattern is that there are far fewer – if any – clear normative values
that drive preferences for our “middle placed” electoral systems: STV and MMP.
To take this last thought a step further, some evidence is examined using regres-
sion models to establish more clearly what kinds of values lead people to like one
system, and to prefer it to another. As can be seen from Table 10, the rankings of
criteria are not terribly useful in establishing preferences for the main systems of
interest – MMP and STV. For this reason a set of survey instruments that tap into
values of political outcomes and processes are used. These measures are then used
to predict the rankings of electoral systems.
Table 11 presents results from OLS models of rankings of systems based on
some evidence on preferences. These fairly straightforward models take as their
Table 10.
Correlation coefficients: desirable properties of electoral systems with rankings
of electoral systems
SMP STV MMP Open
list PR Closed
list PR
Proportionality of outcome
−
.57* .05 .22 .23* .10
Helps ensure stable government .20*
−
.03 .03 .16 .13
Transparent relationship between seats and votes
−
.12 .07 .05 .20* .33*
Simple for voters to use
−
.25* .05
−
.02 .15
−
.06
Allows voters to ‘kick the rascals out’ (i.e. unseat a
government)
−
.16 .01
−
.02 .29* .28*
Allows voters to choose an individual
representative .38* .00 .03
−
.14 .05
Helps ensure representation for women and
minorities .31*
−
.09 .02
−
.12 .24*
Helps build up strong cohesive parties .07 .32*
−
.16 .04
−
.40*
Helps ensure coalition government .28* .37
−
.06
−
.21*
−
.18
*statistically significant.
Note
: Responses on the left are ranked (1st most important, 2nd most important, and so forth)
and these are correlated with the rankings of the electoral systems. Thus, a low ranking of
proportionality of outcome is associated with a high ranking of SMP. On the other hand a
high ranking of proportionality is associated with a high ranking of MMP and open list PR.
Downloaded by [University College Dublin] at 13:26 07 October 2015
14
S. Bowler
et al.
dependent variable the preference ranking (1, 2, 3, and so forth) given to the
(named) system and predicts this using responses to a set of questions about demo-
cratic values and processes. (The Appendix provides exact question wording of
these survey instruments.
4
) To keep with the intuition of ranking the most preferred
systems “1” (for 1st), the dependent variable is one where low numbers mean
something is more highly rated. Table 11 provides evidence that some normative
values are tied to evaluations of electoral systems. There is a straightforward
component to the preferences for electoral systems: for example people who prefer
one party government prefer SMP (column 1) and dislike list PR (column 5). In
terms of support for MMP in particular, then, it should be noted that among the
values that generate support for MMP are attitudes towards political parties.
Column 3 of Table 11 shows that higher rankings to MMP are given by people who
think parties should work towards agreement rather than emphasize clear differ-
ences. Oddly, this same opinion generates support for SMP. Seeing parties as
central actors also helps generate support for MMP (just, at the .10 level), while
this depresses support for STV. Ultimately, what can be seen again here, as in
Table 11. Predicting the rankings of systems by different democratic values
(1)
ranking
SMP
(2)
ranking
STV
(3)
ranking
MMP
(4)
ranking
open list PR
(5)
ranking
closed list PR
Clear difference between
parties vs more agreement −0.300*
(2.13) 0.016
(0.16) −0.219*
(2.14) 0.097
(0.67) 0.223
(1.50)
One person vs many to
represent area 0.248+
(1.95) −0.099
(1.09) 0.226*
(2.32) −0.150
(1.18) −0.373*
(2.64)
Single preference vs
preference ranking 0.275*
(2.21) −0.327**
(3.32) 0.113
(1.11) −0.001
(0.01) 0.599**
(4.19)
One party govt vs two or
more parties 0.591**
(3.59) 0.047
(0.39) −0.010
(0.08) −0.159
(0.95) −0.341+
(1.95)
Strong and stable vs
consensus 0.204
(1.15) −0.077
(0.56) −0.262+
(1.92) −0.519*
(2.60) −0.418+
(1.94)
Personal vote vs parties as
central actors 0.161
(1.31) 0.235*
(2.44) −0.191+
(1.87) 0.025
(0.18) −0.348*
(2.47)
Should be delegate vs
trustee 0.269+
(1.95) 0.195+
(1.88) −0.040
(0.37) 0.082
(0.59) −0.182
(1.20)
Constant −1.476
(1.54) 3.163**
(3.87) 3.926**
(4.69) 6.608**
(5.18) 7.985**
(6.14)
Observations 101 103 110 89 80
R2.51 .33 .22 .33 .55
Absolute value of t statistics in parentheses; + significant at 10%; * significant at 5%; **
significant at 1%.
Note: Rankings are 1 = highest ranking or 1st preference … 2 = next highest or 2nd preferred
and so on.
Downloaded by [University College Dublin] at 13:26 07 October 2015
Expert Opinion on Electoral Systems 15
Table 10, is that models predicting the evaluation of MMP perform worse than
other models: the whole parameters are smaller, fewer pass conventional thresholds
of statistical significance, and measures of fit (R2) are lower.
The interpretation of the overall patterns in Tables 10 and 11 is the following:
even though MMP may be relatively highly rated, it is not clear why this is so.
Experts have quite clearly established and fairly well-defined preferences over two
sorts of institutions – SMP and list systems – but have much less well-defined pref-
erences over MMP and STV. The kinds of things that drive them to prefer these
systems are, moreover, not well established in the literature. This can be shown in
the case of MMP, where to the extent that any values associated with support for
this system can actually be identified, it is not entirely clear how these should fit.
For instance, these data reveal that one reason for liking MMP is that consensus is
valued over clear differences. But is consensus a product of MMP; and, if so, how?
It may be true for the German case but is it also true for Russia? Another value that
drives support for MMP seems to be a dislike of asking voters for preference rank-
ings per se and a liking for one representative per district. But these may not be
well-established properties of MMP. It is not clear, for example, how much of a
difference having “a” representative makes to German voters. In short, then, the
preference for MMP among the sample of experts is unmistakable, but the reasons
for that preference, and for preferring MMP over list systems or STV, are not nearly
so clear-cut.
Discussion: Moving Beyond Proportionality
This study makes a number of points. First, while any discussion of this kind of
survey data will tend to focus on areas of uncertainty or disagreement among schol-
ars, one of the major patterns found is a large degree of consensus both over the key
issue of proportionality and also over the capacity of various electoral systems to
produce proportionality. It is hard to think of another subfield of political science
that shows this level of consensus about its topic of study. It may be the case, too,
that it is this kind of certainty that lends itself to confidence in the capacity to engi-
neer electoral institutions. Furthermore, in outlining some areas where there is lower
consensus, this gives a way to see how studies of electoral systems may move
forward beyond the age-old question of proportionality.
Second, one of the more striking findings of this research is that while MMP is
liked – and, as was seen in the first section, that many people like it a lot and most
people like it a little – the evidence of this survey suggests that there is no clear basis
for doing so beyond the sense that it is neither wholly list nor wholly SMP, but
“something else”. To a lesser extent, this seems to hold for STV as well. The
reasons why people seem to like these systems (namely, the values that are associ-
ated with a high preference ranking of them) are not ones that are well established
within the literature on electoral systems. It is not known, for example, how these
systems contribute to consensus politics or whether voters – if given a choice – care
whether there are one or several representatives from their district.5 There are
Downloaded by [University College Dublin] at 13:26 07 October 2015
16 S. Bowler et al.
well-established properties of electoral systems such as SMP and list PR but much
less well-established properties of other systems and this suggests an area for future
research (Bowler & Grofman, 2000).
Relatedly, it is not clear that there are commonly accepted measures of key
concepts – concepts over which there are disagreements. True, the lack of clear
measures may reflect and not cause that disagreement and so clear measures may
only come about after disagreements have been sorted out but, to repeat a point
made above, the recent work of Powell (2000) and Lijphart (1999) can be seen as
especially important in developing new measures for the important concepts of
governance and representation.
One much simpler issue of measurement is that there seem to be no well-
established or concise ways to represent preference orderings or schedules. Two of
the authors have worked on this issue, on and off, for a number of years – first in
the Irish context (Bowler & Farrell, 1991), then in the context of Euro-elections
(Bowler & Farrell, 1996), and now with this survey. There seems to be no ready
way to represent preferences other than by using lots of tables. Even with just a
few parties or candidates located within an obvious ideological space, preference
orderings can be hard to represent simply. Here one question was whether the
continuum (or space) can be reconstructed in which to locate attitudes to electoral
systems, given preference orderings. It was not clear that is possible beyond locat-
ing SMP and list systems fairly clearly at opposite ends of a single continuum.
Systems such as MMP and STV seem to be located in the middle of that contin-
uum not for positive reasons but for negative ones: namely, not so much because
they realize important normative values but because they do not seem to be one of
the end-points.
The end result of this is that this survey has pointed up some areas where future
work might usefully take place in electoral studies research. It would seem to be
time to move on from questions about proportionality and examine other issues that
are far less settled. These areas where our knowledge and understanding are less
certain are important, suggesting that we should be much more sceptical of our abil-
ity to engineer electoral institutions.
Acknowledgement
This article has equal co-authorship. An earlier version of this paper was
presented at the annual EPOP conference, Oxford, in September 2004. The
authors are grateful to the participants and to the editors of this journal for their
comments and suggestions. Support for the expert survey came from the
McDougall Trust, which is not responsible for errors or interpretation of the data
presented here. The authors are grateful to those colleagues who provided useful
comments and feedback on earlier drafts of the survey. They would also like to
thank all those colleagues and friends who responded to the survey and who
volunteered to be the subjects of study. Errors and interpretations remain the
authors’ responsibility.
Downloaded by [University College Dublin] at 13:26 07 October 2015
Expert Opinion on Electoral Systems 17
Appendix. Wording of questions used as independent variables in Table 11
(coded 1 to 7 from left to right)
Different electoral systems often imply different kinds of trade-offs, or provide for
the possibility of different kinds of political outcomes. We’re interested in knowing
which kinds of priorities matter more for you in an electoral system. This question
poses pairs of statements. If you greatly prefer or very strongly agree with the option
Strongly
agree/
prefer Neutral Strongly
agree/prefer
Clear difference
between parties [] [] [] [] [] [] [] More agreement
and working
together between
parties
One representative
to represent the
area you live in
[] [] [] [] [] [] [] Several members of
a legislature to
represent a larger
area – possibly
from different
parties
Marking a ballot to
indicate one
preference (for a
single candidate in
US or UK, or for a
single party as
under list PR)
[] [] [] [] [] [] [] Marking a ballot so
you could indicate
your first, second,
third choice, etc.
One party in
government [] [] [] [] [] [] [] Two or more parties
in government
It is important for a
government to be
strong and stable
even if it means it
sometimes rides
roughshod over all
opposition
[] [] [] [] [] [] [] It is important for a
government to gain
stability through
consensus even if
this means
sometimes
problems take a
very long time to
solve
Allow individual
candidates to run
personal campaigns
and cultivate a
personal vote
[] [] [] [] [] [] [] Ensure that parties
are central actors in
the campaign
Downloaded by [University College Dublin] at 13:26 07 October 2015
18 S. Bowler et al.
on the left mark the box closest to it. If you greatly prefer or very strongly agree
with the option on the right, mark the box closest to it. You may, of course, choose a
box anywhere between the two.
Notes
1. Given the nature of the survey “population”, it is not possible to be sure about how representative a
sample this is. All that can be said on this matter is the following. The merged address list consisted
of 731 individuals. Once non-relevant respondents, such as publishers or polling companies, were
cleaned out, this brought the first wave population down to 643, of which a number proved to be
incorrect addresses, dead, or people who stated categorically that they were not suitable respondents
for a survey of this type. By the time of the second wave, the total population had declined to 547. If
this is accepted as an approximation of the total population of electoral systems experts, the response
rate (N=170) of 31% is pretty reasonable. But, of course, this is not an entirely accurate picture of the
representativeness of the survey, for a number of reasons, among them: 1) many of those who did
complete the survey stated that they did not feel themselves to be sufficiently expert on electoral
systems (and for that reason often a number of the questions were left unanswered); 2) not all
electoral system experts are members of these organizations and so will not have been included in
the survey (this is particularly the case for non-English speaking political scientists).
2. The survey distinguished between mixed-member proportional (MMP; as used in Germany) and
mixed-member majoritarian (MMM; as used in Russia) systems. It was apparent that not all the
respondents were able/willing to follow this distinction and so, in this paper, this generic form of
electoral system (namely, the “mixed-member” system) is referred to simply as MMP.
3. It is harder to find examples of qualities that electoral systems do not have. That is, it is hard to find
many examples analogous to Tables 5 and 6 where a system is seen as consensually bad at doing
something. Since part of this is probably due to respondents not completing all rankings, this means
that examining the consensus at the top of the rankings is a more reasonable approach. One example
of consensus at the bottom is that 82% of the respondents give SMP a ranking of “5” or lower on its
ability to produce proportional outcomes. On average the experts expressed just over five prefer-
ences, namely ranked five systems over the nine possibilities plus a tenth “other” option. In terms of
the preference profiles (namely the combination of possible rankings of the 169 respondents consid-
ered here) 145 preference rankings were found. The preference profiles for the vast bulk of the
respondents were therefore unique.
4. Preliminary factor analysis suggests these items scale along one dimension: multidimensionality is
not, therefore, an issue.
5. Although, focus group research suggests that, for British voters at least, the preference is for a single
representative per constituency (Farrell & Gallagher, 1999).
References
Blais, A. & Massicotte, L. (1996) Electoral rules, in: L. LeDuc, R.G. Niemi & P. Norris (Eds) Compara-
tive Democratic Elections (Thousand Oaks, CA: Sage).
Blau, A. (2004) Fairness and electoral reform, British Journal of Politics & International Relations, 6.
Bowler, S. & Farrell, D. (1991) Voter Behavior under STV-PR: Solving the Puzzle of the Irish Party
System, Political Behavior, 13, pp. 303–320.
Bowler, S. & Farrell, D. (1996) Voter strategies under preferential electoral systems: a single stransferable
vote mock ballot survey of London voters, in: C. Rallings, D. Farrell, D. Broughton & D. Denver
(Eds), British Elections and Parties Yearbook, 1995 (London: Frank Cass).
Bowler, S. & Grofman, B. (Eds) 2000 Elections in Australia, Ireland and Malta under the Single Trans-
ferable Vote: reflections on an Embedded Institution (Michigan: University of Michigan Press).
Downloaded by [University College Dublin] at 13:26 07 October 2015
Expert Opinion on Electoral Systems 19
Commons, J. (1907) Proportional Representation, 3rd edition (New York: Macmillan).
Dunleavy, P. & Margetts, H. (1995) Understanding the dynamics of electoral reform, International
Political Science Review, 16, pp. 9–29.
Farrell, D. (2001) Electoral Systems: A Comparative Introduction (London: Palgrave).
Farrell, D. & Gallagher, M. (1999) British voters and their criteria for evaluating electoral systems, British
Journal of Politics and International Relations, 1, pp. 293–316.
Gallagher, M. (1991) Proportionality, disproportionality and electoral systems, Electoral Studies, 10.
Hoag, C. & Hallett, G. (1926) Proportional Representation (New York: Macmillan).
Horowitz, D.L. (1991) A Democratic South Africa? Constitutional Engineering in a Divided Society
(Berkeley, CA: University of California Press).
Humphreys, J. (1911) Proportional Representation (London: Methuen and Co).
Katz, R. (1997) Democracy and Elections (Oxford: Oxford University Press).
Lakeman, E. & Lambert, J. (1959) Voting in Democracies (London: Faber and Faber).
Lijphart, A. (1994) Electoral Systems and Party Systems: A Study of Twenty-Seven Democracies,
1945–1990 (New York: Oxford University Press).
Lijphart, A. (1999) Patterns of Democracy: Government Forms and Performance in Thirty-six Countries
(New Haven, CT: Yale University Press).
McLean, I. & Urken, D. (Eds) (1995) Classics of Social Choice (Michigan: University of Michigan
Press).
Pinto-Duschinsky, M. (1999) Send the rascals packing: defects of proportional representation and the
virtues of the Westminster Model, Representation, 36, pp. 117–126.
Powell, G.B. (2000) Elections as Instruments of Democracy (New Haven, CT: Yale University Press).
Rae, D. (1971) The Political Consequences of Electoral Laws (New Haven, CT: Yale University Press).
Reilly, B. (1997) The alternative vote and ethnic accommodation: new evidence from Papua New
Guinea, Electoral Studies, 16, pp. 1–12.
Reynolds, A. (1999) Electoral Systems and Democratization in Southern Africa (Oxford: Oxford
University Press).
Sartori, G. (1997) Comparative Constitutional Engineering: An Inquiry into Structures, Incentives and
Outcomes, Second Edition (London: Macmillan).
Shugart, M. (2001) Electoral “efficiency” and the move to mixed-member systems, Electoral Studies, 20,
pp. 173–193.
Shugart, M. & Wattenberg, M.P. (Eds) (2000) Mixed-Member Electoral Systems: The Best of Both
Worlds? (Oxford: Oxford University Press).
Taagepera, R. (1998) How Electoral Systems Matter for Democratization, Democratization, 5, pp. 68–
91.
Taagepera, R. & Grofman, B. (2003) Mapping the Indices of Seats-Votes Disproportionality and Inter-
Election Volatility, Party Politics, 9, pp. 659–677.
Taagepera, R. & Shugart, M. (1989) Seats and Votes: The Effects and Determinants of Electoral Systems
(New Haven, CT: Yale University Prsess).
Downloaded by [University College Dublin] at 13:26 07 October 2015
