When Do Close Elections Matter for Higher Turnout? Gauging the Impact of the Interaction Between Electoral Competitiveness and District Magnitude

Abstract

A conventional rational choice perspective assumes a positive relationship between close elections and higher turnout because of the increased probability that few votes might actually matter for the election outcome. However, this hypothesis should only hold in electoral districts where one or few candidates are elected. In contrast, close elections should have less of an impact in districts where many members are elected because these districts are characterized by a proportional translation of votes into seats. Testing the foregoing logic, this article evaluates the interactive influence between the closeness of an election and the district magnitude on turnout. The results of the multivariate regression analysis, which includes data for over 600 elections across the globe, support the idea that close elections boost electoral participation in districts with one or few elected members, but not in districts with many elected representatives.

Notes

1 While the electoral closeness can matter under PR for government formation – under the scenario that a party or a predetermined coalition of parties is at the verge of getting the absolute majority of seats – it matters much less in the distribution of votes into seats.

2 I use electoral closeness and electoral competitiveness interchangeably to denote the closeness of the election.

3 The literature puts forward at least four arguments for why larger districts should have higher turnout: (1) all parties (i.e. rather successful and less successful ones) have an incentive to campaign everywhere in the country if the number of votes proportionally translate into the number of seats (Lijphart, 1994). (2) Voters will have an incentive to vote in districts, where many representatives are elected as well because they know that several hundred additional votes for their party of choice might translate into an additional seat (Siaroff & Merer, 2002). (3) A proportional system with large M offers voters more policy alternatives and gives them a change to see their own policy positions represented in parliament (Banducci & Karp, 2009: 127). (4) By including women, minorities, and different age cohorts on their electoral slates, parties tend to diversify their slates the larger the district (Reynolds et al., 2005: 145).

4 There is one counter-argument to this rational. Selb (2009: 532) claims that because “there is no a priori reason to assume that seats are more beneficial in single-member plurality than in multi-member PR districts, we would not expect any differences in turnout levels between SMD and PR districts, keeping competitiveness constant”.

5 The D'Hondt method allocates seats based on the highest average method. This requires the number of votes for each party to be divided successively by a series of divisors (i.e. 1, 2, 3, 4, … ), and seats are allocated to parties that secure the highest resulting quotient, up to the total number of seats available (Norris, 1997: 302).

6 Alongside Freedom House, Polity IV is the most widely used database to measure the regime type of a country. Based on a rather minimal definition of democracy (i.e. free and fair election and political rights), Polity IV uses a 20-point scale to determine whether a country is a democracy, anocracy, or autocracy. Countries ranked between −10 and −6 are considered autocracies, countries ranked between −5 and +5 are coded as anocracies, and countries with ranking of 6 or higher are included in the category democracy (see Marshall et al., 2011).

7 I used four data sources to retrieve official turnout data: (1) the International Institute for Democracy and Electoral Assistance (IDEA) (2013a), (2) Adam Carr's Election Archive (2013), (3) the Inter Parliamentary Union (IPU) (2012) and, in rare cases, the websites of the respective national parliaments.

8 I retrieved the electoral closeness data from the same sources as the turnout data.

9 In total, I only use this latter measure in about 20% of the cases.

10 Data were collected from the International Institute on Democracy and Electoral Assistance (IDEA 2013b). http://www.idea.int/vt/compulsory_voting.cfm.

11 The GDP per capita data come from the UN: http://data.un.org/Data.aspx?q=GDP+per+capita&d=SNAAMA&f=grID%3a101%3bcurrID%3aUSD%3bpcFlag%3a1.

12 The variable population sizes were retrieved from the United Nations Statistical Division 2010. “Population and Vital Statistics Report.” http://unstats.un.org/unsd/demographic/products/vitstats/default.htm. Missing data were collected from the “CIA World Factbook 2011”, https://www.cia.gov/library/publications/the-world-factbook/.

13 Data were retrieved from the IFES Election Guide: http://www.electionguide.org/country.php?ID=29. Any legislative election is coded 1 if it is held at the same time as a presidential election or if there is no presidential election.

14 The tendency of the error terms to be auto-correlated implies that observations from the same country are not independent from one another. Rather high turnout countries tend to have high turnout throughout and low turnout countries tend to have the same tendency. An autoregressive model controls for these dependencies by removing the influence previous values of the dependent variable have on future values (Kmenta, 1990). Unequal variances or heteroscedasticity entails that the variances are not similar across observation, which implies that values with large variances have a higher influence on the equation than data points with low variances. The Huber White Standard Errors control for these discrepancies in the variance across observations (White, 1980).

15 For example, Luskin and Bullock (2011) report that approximately 50% of the US population does not know the length of a Senate term and around 40% cannot correctly recall the nominating procedure for Supreme Court judges. Given this lack of basic political knowledge, it is more than doubtful whether individuals are informed about the competitiveness of the election. They might even think that an election is competitive, even when it is not, and vice versa.