Hostname: page-component-7b7567c8c-kwt9d Total loading time: 0 Render date: 2026-11-13T12:43:35.824Z Has data issue: false hasContentIssue false
  • English
  • Français

Reply to Dow: Party Positions, Votes and the Mediating Role of Electoral Systems?

Published online by Cambridge University Press:  10 December 2010

Lawrence Ezrow
Rights & Permissions [Opens in a new window]

Abstract

An abstract is not available for this content. As you have access to this content, full HTML content is provided on this page. A PDF of this content is also available in through the ‘Save PDF’ action button.

Information

Type
Notes and Comments
Information
British Journal of Political Science , Volume 41 , Issue 2 , April 2011, pp. 448 - 452
Copyright
Copyright © Cambridge University Press 2010

The proposed relationship between electoral systems and party extremism assumes that electoral systems mediate the relationship between party proximity (to the mean voter position) and vote share. In disproportional systems, parties are expected to gain greater electoral rewards for adopting moderate positions than in proportional systems. Following Jay Dow in this issue, I re-analyse the Comparative Study of Electoral Systems (CSES) data to directly evaluate the hypothesized mediating role of electoral systems in explaining party vote share.Footnote 1 These analyses do not support the finding that electoral systems mediate the relationship between party proximity and vote share. This finding suggests that further exploration is necessary to understand the relationship between electoral systems and party positioning.

To review, in 2001, Jay Dow published spatial mappings of the Dutch, Israeli, Canadian and French party systems and concluded that ‘parties in the majoritarian systems are located significantly closer to the center of the distribution of voters than those in proportional systems’.Footnote 2 Inspired by Dow, I then explored the same relationship between electoral system proportionality and average party extremism in eighteen countries.Footnote 3 Parties competing in proportional electoral systems were expected, on average, to adopt more extreme policy positions than their counterparts competing in disproportional systems. After analysing eighteen democracies, however, I reported that there was no evidence that electoral systems affect parties’ tendencies to propose extreme or moderate policy positions. Jay Dow analyses the same relationship in thirty-one countries using more recent data from the Comparative Study of Electoral Systems (CSES),Footnote 4 and he reconfirms the conclusions he reports in his 2001 article that proportional electoral systems promote party extremism, and disproportional systems are more compact. I do not question the conclusions Dow reports: the empirical findings stand.

If this is the case, the question naturally turns to what accounts for the difference in findings. Earlier in the issue, Dow accurately targets explanations for the differences. These are based on time-period, data and groups of countries. Since I agree that these are the likely candidates, and I do not take issue with Dow’s empirical findings, I wish to push the argument forward by turning the analysis to the central causal mechanism of the expected relationship between electoral systems and party positioning – votes.

The expected relationship between electoral system proportionality and average party extremism is based on vote-seeking incentives for parties. Parties have weaker incentives to maximize votes in proportional systems than they do in disproportional/plurality systems. The logic is persuasive. Disproportional electoral laws tend to punish small parties by awarding them seat shares in parliament that are less than their national vote shares – while correspondingly awarding seat shares to large parties that exceed their vote shares. If maximizing votes means staking out popular positions close to the mean voter position, the expectation is to observe a ‘clustering’ of a small number of competitive parties close to this mean voter position in plurality systems.

By contrast, given that electoral thresholds in proportional systems permit more parties to win seats in the legislature, the competing parties in proportional systems can afford to be less concerned about whether they are occupying moderate, vote-maximizing positions. Thus, parties competing in PR systems are plausibly free to advocate their sincere policy beliefs, even if these preferred policies are distinctly non-centrist.

While theory may be straightforward, the empirical results reported by Dow and by EzrowFootnote 5 (reviewed earlier) are obviously less so. In this issue, Jay Dow uses identical measures to mine and presents empirical results based on a greater number of countries than mine, which confirms the expected relationship between electoral system proportionality and average party extremism. En route, Dow reiterates the vote-seeking explanation by remarking that disproportional systems encourage ‘parties to compete over a relatively small portion of the electoral space due to the “winner takes all” nature of majoritarian politics. Proportional systems, in contrast, support more parties and increase the range of the viable political space by rewarding parties with seats for relatively modest vote shares.’Footnote 6

Does the relationship hinge on vote-seeking incentives for parties? Are moderate parties rewarded with greater shares of the vote in disproportional systems than in proportional systems? We can explore this empirical question directly, by employing the CSES for the same set of countries to evaluate this proposition that the electoral rewards for centrist positioning are enhanced in disproportional systems.Footnote 7 Accomplishing this requires four pieces of information: parties’ policy positions; the mean voter position; parties’ vote shares; and a measure of electoral system proportionality. Thus, we can re-analyse the CSES data (described by Dow earlier) to evaluate whether electoral systems mediate the relationship between party positioning and vote share.

The party’s distance to the mean voter position is measured by squaring the difference between party position and the mean citizen placement:

(1)
\[--><$$>Party\ policy\ distance = {{({{A}_i} - {{X}_i})}^2}, \eqno<$$><!--\]

where Ai is the position of the mean voter on the Left–Right dimension, and Xi is the mean perceived position of party X.Footnote 8 The following regression model specification evaluates the relationship between policy moderation and vote share across all of the countries in the study:

(2)
\[--><$$>Party\ vote\ share = {{\beta }_1} + {{\beta }_2}\,\left[ {Party\ policy\ distance} \right]. \eqno<$$><!--\]

I have reported a negative relationship between party policy distance and party support, i.e. that parties’ vote shares decrease with the distance between the party’s left–right position and the mean voter position (i.e. β 2 is expected to be negative).Footnote 9 Column 1 of Table 1 presents the results from estimating the parameters of the ‘basic’ model specification in Equation 2. The coefficient on the [party policy distance] variable is negative (−0.55) and statistically significant. The negative estimate corroborates the earlier finding, based on the Eurobarometer data, that parties further from the centre of the voter distribution tend to gain fewer votes than parties closer to the middle of the electorate.

Table 1

Estimating Parties’ Vote Shares

Notes: Parameters are ordinary least squares (OLS) coefficients. Standard errors, clustered by election, are in parentheses. Independent variables are described in the text. Estimates are based on data from countries in the Comparative Study of Electoral Systems (CSES). *p < 0.10, **p < 0.05, ***p < 0.01, two-tailed test; †p = 0.127.

To evaluate the mediating effect of electoral systems on the relationship between policy moderation and vote share, we can rely on the disproportionality measure described earlier by Dow.Footnote 10 Furthermore, to deal with countries with different numbers of competitors, we can similarly rely on the measure of effective number of parties, which is based on the number of parties that are competitive enough to gain seats in the national legislature.Footnote 11 The degree to which centrifugal and centripetal incentives exist across countries featuring different electoral systems can be measured by estimating the parameters of the following regression model specification:

(3)
\[--><$$>\eqalignno{ \qquad \qquad Party\ vote\ share\, = \, {{\beta }_1}+{{\beta }_2}\,\left[ {Party\ policy\ distance} \right] \cr &amp; \quad+ {{\beta }_3}\,\left[ {{\mib{Party}}\ {\mib{policy}}\ {\mib{distance}}\,{\mib{\times }}\,{\mib{disproportionality}}} \right]{} \cr\!\! &amp; \quad+ {{\beta }_4}\,\left[ {Disproportionality} \right]{} \cr &amp; \quad+ {{{\beta }_5}\,\left[ {Effective\ Number\ of\,Parties} \right]\, + \,e. } (3)}<$$><!--\]

If there are differences in electoral incentives across systems that feature different levels of (dis)proportionality, then we would expect the coefficient, β 3, on the interaction term [Party policy distance × disproportionality] to be statistically significant. If party policy distance matters more in disproportional systems, leading to the expectation of clustering of parties at the centre of the voter distribution, then this coefficient is expected to be negative and statistically significant. A negative sign indicates that parties in disproportional systems would experience additional electoral punishment for adopting policies further from the mean voter position.

With respect to the key expectation formulated in this section that electoral systems mediate the effects of proximity, there is no statistically significant evidence that moderate parties gain more votes in disproportional systems than in proportional systems. If the coefficient on the interaction term was negative and statistically significant, this would suggest that electoral systems mediate party proximity effects. However, the estimates on the interaction term in Table 1, Column 2 for the ‘electoral systems’ model are not statistically significant.

If there are greater electoral penalties for large parties that adopt distinct positions in disproportional systems, then we would expect the coefficient on the interaction [Party policy distance × disproportionality] variable to be negative and statistically significant, and significantly larger in magnitude than that for all parties. The parameter estimates for the model specifications in Column 3 are based on parties that receive more than 10 per cent of the vote. In addition, an important recurring theme raised in the previous discussions is the importance of accounting for the underlying distribution of voters.Footnote 12 In comparing two countries, party policy distance may be regarded as greater where there is a compact distribution of voters’ political preferences; and less where the voter distribution is dispersed. In Column 4, spatial distances are normalized to the underlying voter distribution, which is calculated as the standard deviation of voter Left–Right self-placements in the CSES. The coefficients in Columns 3–4 do not approach conventional levels of statistical significance (indeed, the coefficient is slightly positive in Column 3). Thus, they continue to support the finding that electoral systems do not mediate the effects of party proximity on electoral success. This conclusion holds under a number of additional robustness checks.Footnote 13

Discussion

It seems obvious that plurality systems promote policy centrism. Candidates gravitate towards the centre of political competition in an attempt to win elections in systems like those in the United States and Britain. They lose if they do otherwise. A casual observation of these systems suggests that this proposed relationship is slightly more complex. While there are periods of convergence in these systems, there are also periods of polarization. British party competition in the mid-1990s, with the arrival of Tony Blair and ‘New Labour’, was noticeably centrist. By contrast, Margaret Thatcher initiated a decade of polarized politics in the 1980s. In the United States, the gap between Democrats and Republicans has been growing steadily since the mid-1970s with McCarty, Poole and Rosenthal neatly summarizing it: ‘partisan differences in congressional voting behavior have grown dramatically to levels not seen since the early twentieth century’.Footnote 14

These examples point to difficulties that arise when generalizing about the effects of electoral systems on party positioning. Electoral rules have remained constant, but average party extremism has changed dramatically.

The discussion of Britain and the United States suggests that one clear area for future exploration is within country analysis over time. One may concentrate on democracies where electoral rules have remained constant, but levels of polarization vary (such as Britain and the United States). Why have party strategies changed? Alternatively, one could focus on political systems where electoral rule change has occurred – like New Zealand, Italy and France – and on whether these rule changes have produced changes in polarization.

A second direction follows the most recent and welcomed study by Dow, which reports that disproportional electoral systems encourage centrist positioning (as one would expect). Based on the empirical analyses I present above, however, the proposed vote-seeking explanation for this relationship is doubtful. Identifying the elusive causal mechanism is then a second direction for exploration.

Footnotes

* Department of Government, University of Essex (email: ). The author thanks Jay Dow, Jim Adams, Spyros Kosmidis and Hugh Ward for comments on earlier drafts.

References

1 Dow, Jay K., ‘Party System Extremism in Majoritarian and Proportional Electoral Systems’, British Journal of Political Science, this issue; see also http://www.journals.cambridge.org/jps, doi:10.1017/S0007123410000360CrossRefGoogle Scholar.

2 Dow, Jay K., ‘A Comparative Spatial Analysis of Majoritarian and Proportional Systems’, Electoral Studies, 20 (2001), 109–125CrossRefGoogle Scholar, p. 111.

3 Ezrow, Lawrence, ‘Parties’ Policy Programmes and the Dog that Didn’t Bark: No Evidence that Proportional Systems Promote Extreme Party Positioning’, British Journal of Political Science, 38 (2008), 479–497CrossRefGoogle Scholar. I note that the expanded country coverage came at the expense of including a second dimension in the analyses.

4 Dow, ‘Party System Extremism’.

5 Dow, ‘Party System Extremism’. See also Dow, ‘A Comparative Spatial Analysis’; Ezrow, ‘Parties’ Policy Programmes and the Dog that Didn’t Bark’.

6 Dow, ‘Party System Extremism in Majoritarian and Proportional Electoral Systems’, p. 5.

7 The Comparative Study of Electoral Systems (www.cses.org), CSES Module 1, Full Release [dataset] (Ann Arbor: University of Michigan, Center for Political Studies, 4 August 2003); The Comparative Study of Electoral Systems. CSES Module 2, Full Release [dataset] (Ann Arbor: University of Michigan, Center for Political Studies, 27 June 2007).

8 An alternative set of analyses was performed based on the parties’ linear proximities to the mean voter position. These analyses supported substantive conclusions that were identical to the ones reported below, although the statistical fit of these models was not as strong as the fit for squared proximity, suggesting that this latter measure is the appropriate metric for evaluating the electoral effects of party positioning. This empirical finding suggests that the parties’ vote shares are concave functions of their policy positions, i.e. that parties’ vote shares drop off slowly at first as they diverge from their vote-maximizing positions, but then drop off more rapidly as the parties move further away. For theoretical arguments about why parties’ vote shares can be expected to be concave functions of their positions, see Adams, James and Merrill, Samuel III, ‘Parties’ Policy Platforms and Election Outcomes: The Three Faces of Policy Representation’, European Journal of Political Research, 44 (2005), 899–918CrossRefGoogle Scholar. This can be interpreted to mean that parties are penalized more for each marginal unit further away they are from the mean voter position.

9 Ezrow, Lawrence, ‘Are Moderate Parties Rewarded in Multiparty Systems? A Pooled Analysis of Western European Elections, 1984–98’, European Journal of Political Research, 44 (2005), 881–898CrossRefGoogle Scholar.

10 Gallagher, Michael, ‘Proportionality, Disproportionality and Electoral Systems’, Electoral Studies, 10 (1991), 33–51CrossRefGoogle Scholar. The equation for the Index of Disproportionality is )(--><$>\sqrt {\tfrac{1}{2}\,{\mathop{\sum}\limits_{i}{{{{({{v}_i}\,{\rm{ - }}}\,{{s}_i})}}^2} } } <$><!--, where vi and si are the vote shares and subsequent seat shares for party i.

11 The equation for the Effective Number of Parties is based on parliamentary seats: N = 1/Σsi 2, where si is the proportion of seats of the ith party (see Laakso, Markku and Taagepera, Rein, ‘ “Effective” Number of Parties: A Measure with Application to West Europe’, Comparative Political Studies, 12 (1979), 3–27CrossRefGoogle Scholar).

12 See also Golder, Matt and Stramski, Jacek, ‘Ideological Congruence and Electoral Institutions’, American Journal of Political Science, 54 (2010), 90–106CrossRefGoogle Scholar.

13 Additional empirical analyses have been conducted that measure party policy distance based on linear losses. The mediating effect of electoral systems has been estimated for countries and elections included in the Eurobarometer surveys for the 1980s and 1990s (see Ezrow, Lawrence, Linking Citizens and Parties: How Electoral Systems Matter for Political Representation (Oxford: Oxford University Press, 2010), p. 60CrossRefGoogle Scholar). The effective number of parties has been measured based on parties’ proportions of votes. Vote share has been measured relative to the sum of all of the parties’ vote shares in a given election. The parameter estimates based on these model specifications support identical substantive conclusions.

14 McCarty, Nolan, Poole, Keith T. and Rosenthal, Howard, ‘Does Gerrymandering Cause Polarization?’ American Journal of Political Science, 53 (2009), 666–680CrossRefGoogle Scholar, p. 666. See also McCarty, Nolan, Poole, Keith T. and Rosenthal, Howard, Polarized America: The Dance of Ideology and Unequal Riches (Cambridge, Mass.: MIT Press, 2006)Google Scholar.

Figure 0

Table 1Estimating Parties’ Vote Shares