Electoral systems scholarship has focused much attention on the importance of electoral systems in shaping both voting and parties’ entry decisions (Cox Reference Cox1997). However, we have almost no insight into how, if at all, electoral systems shape the strategies adopted by political parties in election campaigns. There is a large body of literature studying the effectiveness of voter mobilization strategies (see Johnston et al. Reference Johnston, Cutts, Pattie and Fisher2012), but few studies have examined how party strategies vary by electoral system (Cox Reference Cox1999, Reference Cox2015; Karp et al. Reference Karp, Banducci and Bowler2007). A parsimonious empirical model explaining how parties define their campaign strategies to influence voters’ decisions in different electoral systems is still lacking. As Rosenstone and Hansen (Reference Rosenstone and Hansen1993: 11) argue, ‘the competitive pressures of the democratic system encourage political leaders to mobilise their fellow citizens, and if we are to understand participation, we must also comprehend their choices’.
The goal of this article is to explore how district magnitude and the number of districts shape campaign strategies (that is, the actions taken by parties to influence voters’ decisions). We use qualitative and quantitative data from a districted electoral system and a system using a single national district in Spain to determine whether visits to districts in the election campaign by the leaders of national parties are driven by district magnitude, the share of the population entitled to vote in each district, the number of districts or the strength of parties’ local organizations.
We show that district magnitude and the number of districts dramatically affect the campaign strategies of large parties (that is, parties which are viable in all or most of the districts within an electoral system) but do not affect the behaviour of small parties (that is, parties which are only viable in some or few districts) in election campaigns.
The structure of the article includes the following sections. The first section discusses the previous literature and our theoretical arguments regarding how electoral systems shape campaign strategies. Qualitative data from interviews with campaign managers are used to formulate our theoretical expectations. The next section describes our case study, data and methods, while the fourth section discusses the quantitative results of the empirical analysis. The last section presents our conclusions and suggestions for further research.
ARGUMENTS
The conventional wisdom argues that election campaign strategies vary across electoral systems and across districts within electoral systems (Cox Reference Cox1999; Katz Reference Katz1980). However, existing empirical research is clearly unbalanced. While differences in campaigning across districts in a single election or electoral system have been carefully examined (see Criado Reference Criado2008; Denver et al. Reference Denver, Hands, Fisher and MacAllister2003; Pattie and Johnston Reference Pattie and Johnston2009), the research agenda is virtually blind to an exploration of how electoral institutions – namely district magnitude – affect the actions parties take to influence voters’ decisions.
A substantial body of literature has shown the efficacy of party canvassing, particularly campaign spending, on mobilizing voters in specific countries, mainly the UK and the US. For instance, Fisher et al. (Reference Fisher, Cutts and Fieldhouse2011: 817–18) show that the impact of district campaigning in British single-member districts is a function of four variables: the competitiveness of the election, whether a significant change is anticipated in an election, the degree of coordination of campaigns (namely a function of the number of target seats), and the ‘national mood’ – that is, the popularity of a given party. Similarly, using data from a three-wave panel survey for the 2010 British general election, Johnston et al. (Reference Johnston, Cutts, Pattie and Fisher2012) show that parties focus their mobilization efforts in the last weeks before an election on those marginal districts that are likely to vote for them, and that such tactics are successful; the more ways in which respondents were contacted by a party, the more likely they were to vote for it. However, as the electoral system is a constant it is not possible to know whether the factors found to influence campaigns are idiosyncratic to that electoral system or can be generalized to other electoral systems.
On the other hand, very few studies have used empirical methods to examine how campaigning at the constituency level is a function of the electoral system. Denemark (Reference Denemark2003) shows that parties adapted their campaigning when the electoral system changed from a single member plurality/plurality system to a mixed member proportional representation (MMP) system in New Zealand in the 1990s. Consequently, parties devoted more of their campaign budgets to the overall campaign, and less to local constituency campaigns.
This article takes a step towards exploring how district magnitude and the number of districts shape party strategies in election campaigns. To analyse the differences between electoral systems we shall consider how changing a (the) single parameter – district magnitude and consequently the number of districts – changes the incentives for campaigning. We examine how party strategies vary when moving from a proportional representation (PR) electoral system using a single national district to a districted PR electoral system, where the seats are awarded by a PR electoral system in geographically defined districts. Our focus is on theory building, and our arguments are formulated by incorporating qualitative data collected through a series of semi-structured face-to-face interviews with members of the campaign teams of the two largest national parties in Spain – the right-wing Partido Popular (PP – People’s Party) and the left-wing Partido Socialista Obrero Español (PSOE – Spanish Socialist Workers’ Party) – and a small national party, Unión Progreso y Democracia (UPyD – Union, Progress and Democracy) in European and national elections.Footnote 1 The Spanish European MPs are elected in a single nationwide district, while the members of the Lower House are elected in 52 districts of varying magnitudes.
According to the decision-theoretic model formulated by Cox (Reference Cox1999), the amount of effort parties devote to mobilization in a particular district depends on three variables: how effort translates into votes (V’); how votes translate into seats (S’); and how seats translate into portfolios (P’). The parameter V’ depends on the quality of parties’ links to social groups – that is, the ties between a party and a church, union or interest group – while S’ and P’ are a function of local and national closeness, respectively.
We focus on how district magnitude affects the parameter S’. The rate at which a party gains seats when it receives more votes (that is, local- or district-level closeness) is determined to a great extent by electoral proportionality or the number of seats to be filled. Our argument is that district magnitude and the number of districts affect large and small parties’ campaign strategies differently. All else being equal, when moving from a PR system using a single national district to a districted PR system, the incentives for mobilization clearly change for large (national) parties, but not for small (national) parties.
First, when there is a single nationwide district, the payoff in seats of any given increment of votes as a consequence of mobilization efforts is the same everywhere. Accordingly, both large and small parties will invest more heavily in the most populated areas/provinces, given that they can win more votes there.
The interviews with campaign managers in Spain strongly support this claim about the importance of district population size when there is single nationwide district. According to one of the campaign architects for the PP, ‘In European elections there is only one single national district and therefore the priority is obtaining the maximum number of votes. The more votes you get, the higher the chances you have of increasing your representation.’ One of the leaders of the PSOE campaign committee offered a similar statement: ‘In European elections the criteria for determining where to go in the campaign is based exclusively on demographical weight, in comparison with national elections, where the weight in terms of the number of seats elected in each district is relevant, but also the risk of that votes will be transferred to another party.’ Unsurprisingly, the respondent from the minor national party UPyD also outlined a similar strategy: ‘In the European election our candidate visited the provinces with the largest populations and others that were not high in the ranking of population, but where the electoral results in other elections had been good.’
When using a districted electoral system, the number of individuals entitled to vote in each geographic unit (that is, the number of seats to be allocated in the district) also matters. However, party mobilization will be driven less by this variable because the rate at which a mobilizing party gains seats when it gains more votes differs between districts: the closer the local race, the higher the chance that mobilization efforts could make a difference. But this logic only applies to large parties which are viable competitors in all (or almost all) districts and face incentives to mobilize everywhere. Given that small parties only have good chances of winning a seat in those districts that allocate a high number of seats, they will continue investing their resources in the most populated areas.Footnote 2
When campaign managers in Spain were asked about their campaign strategy in national elections, the difference between large and small parties is clear. According to one of the campaign architects for the PP: ‘In national elections an increase in the number of votes does not imply an increase in the number of seats, since it is possible that in one district you may have already reached your electoral limit because the next seat to be obtained is far from the votes you are able to get.’
For a small party like UPyD, with good chances of winning seats only in the biggest districts, there are no differences in campaigning between European and national elections. The UPyD campaign manager described the focus of the campaigns as follows: ‘In the national election, the decision about which districts Rosa Díez [the national leader] visited was taken as a function of the number of seats available in each district and the chance the UPyD had of winning seats . . . In the European election our candidate visited the provinces with the largest populations.’
Second, the differences between large and small national parties when campaigning under an electoral system using a single national district and a districted system affect not only the concentration/dispersion of resources across districts, but also where they invest. As parties exert mobilization efforts, the higher the probability of those efforts being decisive (Cox Reference Cox1999); small national parties will target densely populated areas regardless of the electoral system. The reason is that larger districts, where small parties have their best chances of winning seats, are also areas of high population density. Conversely, large parties’ mobilization efforts will be more driven by population when using a single national district than in a districted electoral system.
Finally, the impact of the number of individuals entitled to vote in geographic units is also (marginally) conditioned by how well apportioned electoral systems are. All else being equal, if the populations of geographic units are exactly the same in electoral systems using a single national district as in districted electoral systems (that is, both electoral systems are perfectly apportioned), the rate at which a mobilizing party gains seats in a given province when it gains more votes is equal. However, when there is a significant disproportion in a districted electoral system, small districts are favoured at the expense of large ones (Samuels and Snyder Reference Samuels and Snyder2001), and the rate will be higher in large geographic units and lower in small geographic units in electoral systems using a single national district than in districted ones.
These rational choice institutionalist arguments are challenged by what can be denominated as the ‘party-centred’ approach: institutional changes are mediated by parties’ internal dynamics at their inception, yet the effects of these changes on party behaviour may be limited due to the inertia of long-standing party organization arrangements. According to the empirical evidence provided by Denemark (Reference Denemark2003: 615) about the impact of electoral reform in New Zealand:
while the party elites were aware well before the campaign began of the need to embrace new tactics to maximise party list votes, the first MMP election in 1996 reflected important residues of the FPP [first-past-the-post] mindset . . . Every party encountered significant resistance from local candidates to priority being given to campaigning for party list votes . . . the patterns of constituency-level activities in the first election campaign under MMP reflected the important influence of inertia amongst the various political actors involved.
Similar evidence can be found in Spain. Indeed, according to the campaign manager for the Socialist Party in the 1996 national election, the party leader had to visit some districts during the campaign due to the pressures of regional committees and therefore could not simply prioritize based on the closeness of district races. The visit of the party leader to a given district is understood to be a sign of commitment to the region and therefore one goal of parties is for the national leader to visit all regions during campaigns (see Criado 2004: 27–9, 72). In a similar vein, Méndez Lago (Reference Méndez Lago2000: 312–13) shows that, apart from local competitiveness, ‘party tradition’ is crucial in the case of the Socialist Party to decide the districts to be visited by the national leader, particularly when deciding where campaigns start and finish.
The evidence from the interviews we have carried out is, however, against the party-centred approach. The statement from the member of the PSOE campaign committee on this topic was: ‘Local organizations do not have any relevant role in determining the allocation of resources during electoral campaigns.’ The leader of the UPyD campaign committee offered a similar assessment: ‘The number of affiliates in the districts or the number of requests from local organizations to receive a visit by Rosa Diez did not have any influence.’
Given these arguments, our hypotheses are the following:
Hypothesis 1A: The campaign efforts of large parties will be more dispersed across geographic areas in districted electoral systems than in electoral systems with a single national district.
Hypothesis 1B: The campaign efforts of small parties will be equally dispersed across geographic areas in districted electoral systems and in electoral systems with a single national district.
Hypothesis 2A: Large parties are more likely to target densely populated areas in electoral systems with a single nationwide district than in districted electoral systems.
Hypothesis 2B: Small parties will target densely populated areas regardless of the electoral system.
Hypothesis 3: According to the party-centred approach, for both large and small parties, the stronger the local organization, the more campaign efforts will be invested in the geographic area.
DATA AND METHODS
The hypotheses are examined through a quantitative analysis of party mobilization in the 2009 European election and the 2011 Lower House election in Spain. There are four reasons for selecting these two European and national elections in Spain as our empirical case. Firstly, there are significant differences between the two electoral systems, mainly in district magnitude, the number of districts and the level of malapportionment, while the party system is largely the same.Footnote 3 The 50 Spanish European MPs are elected in a single nationwide district, while the 350 members of the Lower House are elected in 52 districts, in which magnitude ranges from 1 to 36. The district structure in Lower House elections is particularly appropriate for our purposes. On the one hand, national party leaders cannot visit the 52 districts during the two weeks of the electoral campaign; they have to prioritize some of them. On the other hand, while large national parties are viable competitors in all districts, small national parties have good chances of winning seats only in the biggest districts. Additionally, the electoral system in European elections is by definition perfectly apportioned, but in national elections malapportionment is severe. The largest district, Madrid, has 13.75 per cent of the national population entitled to vote, but only 10.29 per cent of the seats, 36, are elected there; in the two smallest districts, Ceuta and Melilla, the percentage of the national population entitled to vote is 0.18 and 0.16 per cent, respectively, but 0.29 per cent of the seats, 1, is elected in each of them. According to Samuels and Snyder (Reference Samuels and Snyder2001: 661), Spain’s Lower House is among the 20 most malapportioned cases in the world. Finally, in both electoral systems the allocation of seats to parties is proportional to the votes and follows a D’Hondt system of closed party list proportional representation.Footnote 4
Secondly, when studying the impact of electoral systems on party mobilization in cross-national comparisons, there is the possibility that our model omits some important factor that accounts for variations in party mobilization efforts across countries. Since the socio-demographic composition or the characteristics of parties and party systems vary widely across and within countries, the cross-national evidence may not be quite as reliable as we would hope. However, districts in Lower House elections in Spain match with the nineteenth-century administrative provinces, which have never had their borders altered and are perfectly comparable in European and national elections. Accordingly, provinces are our unit of analysis.
Thirdly, given that only two years elapsed between the 2009 European election and the 2011 national election in Spain and that the electoral supply between the two elections is constant, the availability of social networks for parties’ mobilization efforts should not change significantly. Further, as no executive is selected in European elections, on the one hand, and a clear majority of seats for the winner was predicted by all polls in the 2011 national election (see the results of the election in Table A1 in the appendix below), on the other hand, obtaining more seats did not translate into more government portfolios in either of the two elections: seats are valuable as ends in themselves. The comparison between the two elections allows us to observe the change in campaign strategies that will result from a change in the payoff in seats of any given increment in votes (S’), when both the payoff of votes of any mobilizational effort (V’) and the payoff in portfolios of a given increment in seats (P’) are constant.
Fourthly, as the PSOE was the incumbent party when the 2009 European election and the 2011 national election were held, a possible incumbent effect affecting party strategies is controlled for.
However, the electoral system is not the only incentive for mobilization that changes between the 2009 and 2011 elections; the saliency of European and national elections (Reif and Schmitt Reference Reif and Schmitt1980) was also very different. We think that this difference in saliency has little effect on how efforts are distributed by each party across districts. In fact, if the dependent variable was the total amount of resources invested by parties during campaigns, then the analysis would be biased in favour of national elections: elite effort is higher when the expected return is higher. But in our research we are focused on how resources are distributed across districts and therefore the crucial assumption is that cross-district differences in the propensity of voters to support a given party do not change in each election (V’). In other words, districts are more or less favourable for parties in exactly the same way in the two elections. This assumption is supported when observing the correlation between parties’ vote shares at the district level in the 2009 and 2011 elections in Spain: 0.97 for the PP, 0.95 for the PSOE and 0.87 for the UPyD – the three national parties considered (see Figure A1 in the appendix below).Footnote 5 In sum, even if the ceteris paribus assumption is not entirely met, the comparison between European and national elections in Spain is useful for theory-generating purposes.
The empirical analysis of party strategies will be focused on the three national parties which won seats in both the 2009 European election and the 2011 Lower House election: the two largest parties, the centre-left PSOE and the centre-right PP, and a small party, the centrist UPyD.Footnote 6 The fourth national party, the post-communist Izquierda Unida (IU – United Left), which also won seats in the two elections, and subnational parties that only enter the race in one region in national elections have not been considered.Footnote 7
The dependent variable is the number of visits to each district made by the candidate that topped the list in the 2009 European election − Juan Fernando López Aguilar for the PSOE, Jaime Mayor Oreja for the PP and Francisco Sosa Wagner for the UPyD − and the candidate that topped the list put out by the district of Madrid in the 2011 Lower House elections − Alfredo Pérez Rubalcaba for the PSOE, Mariano Rajoy for the PP and Rosa Díez for the UPyD.Footnote 8 Party leaders’ visits and the public rallies they organize fulfil the condition of being a limited resource for parties because they are activities that consume both money and time. Moreover, time is an equitable variable that is irrespective of party resources – leaders’ schedules are limited to the 15 days of the electoral campaign – so the number of visits during this period is a good measure of the distribution of mobilization efforts. We counted each meeting or event involving the presence of the leader of the party that the party organized in each of the provinces during the two weeks of the official campaign.Footnote 9
Party campaign spending at the district level and party canvassing during the campaign are the conventional measures of strategic mobilization (Rosenstone and Hansen Reference Rosenstone and Hansen1993). However, neither are very effective measures for the Spanish case. On the one hand, apart from leaders’ visits to districts, there are no data for party strategies in election campaigns in Spain. For instance, as can be seen in both Criado (2004) and Méndez Lago (Reference Méndez Lago2000), campaign spending is a measure which is neither reported nor publicly available in Spain. On the other hand, party contact is not a valid measure in Spain either, because parties do not contact voters during electoral campaigns in the way that American parties do, for instance (see Criado 2004: 17–18).
Given that the dependent variable is far from having an unbounded, symmetric, bell-shaped distribution, Poisson regression is usually regarded as an appropriate approach for analysing variables measuring the number of times a particular event occurs (King Reference King1989). As Poisson regression is more demanding than OLS in terms of the number of observations, observations for the PP and the PSOE and/or European and national elections have been merged in some models. Standard errors are clustered by province to account for the non-independence in the data structure.
The key independent variable is Population. It captures the national share of individuals entitled to vote in each province. The expected sign of the variable on the number of visits is positive and the coefficient should be larger in European elections than in national elections. As the expected effect of population should be stronger as districts are more populated, the square of the variable is included in the models. Given that the population of some districts is below 1 per cent, the variable has been defined as (1+Population)2. To see whether the variable plays a different role in European and national elections, we will run an interaction between Population and a dummy variable, European Election, that equals 1 for provinces in the 2009 European election and 0 for provinces in the 2011 Lower House election. To test the party-centred approach, we have created a variable, Delegates, that captures the strength of local (province) party organizations. It has been measured as the national share of delegates that each province sent to the 2008 PP National Congress, the 2011 PSOE National Congress and the 2009 UPyD National Congress. In order to avoid an almost perfect correlation with the population of each province, the variable has been standardized. That is, the share of delegates sent by each province to the respective national congress has been divided by the national share of individuals entitled to vote in each province. Among many other functions, the national leaders of the three parties are elected by the delegates attending these congresses.Footnote 10 The expected sign of the variable is positive. The sources are parties’ websites and the national newspaper, El País.
Finally, two controls have been added in our regression models. Firstly, as there is a single national district in European elections in Spain, we only control for the degree of district-level competitiveness in national elections. Following Blais and Lago (Reference Blais and Lago2009), district-level Competitiveness in the 2011 Lower House election has been defined as the minimum number of votes required by each party to gain or lose an extra seat in a given district, divided by the number of individuals entitled to vote in the districts. Three shares, one for each party, have been computed in each district. We have calculated the inverse of the variable to facilitate interpretation: the higher the value, the higher the competitiveness. The variable has been calculated according to the results of the 2008 Lower House election. The source is www.infoelectoral.mir.es.Footnote 11 Secondly, to test whether differences in mobilization efforts across districts and electoral systems are largely an effect of Madrid, the largest and most populated province in Spain, we have created a dummy variable, Madrid, that equals 1 for Madrid and 0 for all the remaining provinces.
The model we estimate separately for large and small parties to test our hypotheses is as follows:
$$\eqalignno{ Visits_{{it}}\, {\equals} \,& \beta _{0} {\plus}\beta _{1} Delegates_{{it}} {\plus}\beta _{2} Population^{2} _{{it}} {\plus}\beta _{3} European \ Election_{{it}} \cr & {\plus}\beta _{4} Population^{2} _{{it}} \, {\asterisk}\, European\ Election_{{it}} {\plus}\beta _{5} Madrid_{{it}} {\plus}{\varepsilon}_{{it}} $$
To control for competitiveness in national elections, a second model is estimated:
$$Visits_{{it}} \, {\equals}\, \beta _{0} {\plus}\beta _{1} Delegates_{{it}} {\plus}\beta _{2} Population^{2} _{{it}} {\plus}\beta _{3} Competitiveness_{{it}} {\plus}{\varepsilon}_{{it}} $$
EMPIRICAL ANALYSIS
The first piece of evidence supporting our argument is displayed in Table 1. As can be seen in Table 1, the number of visits to provinces in the 2011 Lower House election (the PSOE, 26; the PP, 22; the UPyD, 19) is higher than in the 2009 European election for the PSOE and the UPyD (19 and 14, respectively), but lower for the PP (23). The range of the variable is higher in the Lower House election than in the European election for the two largest parties (the maximum number of visits to a province is two for each party in the former and five for the PSOE and eight for the PP in the latter), while for the UPyD the maximum is five visits in the Lower House election and four in the European election. This initial evidence is clearly in line with our Hypotheses 1A and 1B – that is, that large parties adapt their campaign strategies to electoral systems, while small parties do not.
Descriptive Statistics
The frequency of visits to each province by party leaders in the 2009 European election and the 2011 Lower House election is displayed in Figures 1 and 2. Some significant differences between large and small parties are immediately apparent. Firstly, the largest province, Madrid, was visited eight times by the leader of the PP (35 per cent of the total number of visits), five times by the leader of the PSOE (26 per cent of the total number of visits) and four times by the leader of the UPyD (31 per cent of the total number of visits) in the European election campaign. However, while Madrid was visited only twice by the leaders of the PP and the PSOE in the national election campaign (9 per cent for the PP and 7 per cent for the PSOE of the total number of visits), the leader of the UPyD went five times (31 per cent of the total number of visits). This is clearly in line with our Hypotheses 2A and 2B, emphasizing the crucial role of party size when explaining campaigning. As expected, the variation in the visits of the party leader is higher in the districted electoral system (the national election) than in the electoral system with a single national district (the European election), but only for large parties. That is, while 24 provinces were visited (26 visits) by the leader of the PSOE in the 2011 Lower House election, only 14 provinces (19 visits) were visited in the 2009 European election. Similarly, the PP leader visited 21 provinces (22 visits) in the 2011 Lower House election and 16 provinces (23 visits) in the 2009 European election. However, the UPyD leader visited the same number of provinces, 10, in both elections.
Frequency of Visits to Each Province in the 2009 European Election
Frequency of Visits to Each Province in the 2011 Lower House Election
In the analysis of the determinants of the number of visits to provinces in the 2009 European election and the 2011 national election in Spain, we have considered large and small parties separately. Table 2 estimates the models for the two main parties together.
The Determinants of the Number of Visits to Provinces in European and National Elections in Spain, PSOE and PP
Note: Standard errors in parentheses. * p<0.05; ** p<0.01; *** p<0.001.
In the first model, the observations for the PSOE and the PP in both the European and national elections are merged. The variable Population (squared) is the only statistically significant variable (at the 0.001 level) and has the expected positive sign: the higher the number of eligible voters in a province, the higher the probability of the province being visited by the national leader. The variable Delegates, however, is not statistically significant at the conventional levels. Hypothesis 3 is then rejected. Finally, the dummy variable identifying provinces in the European election is not statistically significant. That is, controlling for the province population and the number of delegates, the number of visits does not depend on the type of election.
In the second model, an interactive term between provinces in the European election and the number of entitled voters is added to the previous model. The two constitutive terms, as well as the interaction term, are statistically significant at least at the 0.05 level and show that the impact of population is greater in the 2009 European election than in the 2011 national election. That is, as our Hypothesis 2A predicted, the number of visits of the leader of a large party to the most populated provinces is higher in electoral systems with a single national district than in districted electoral systems.
We have explored whether differences in mobilization efforts were largely due to a Madrid effect. When the dummy Madrid is added in the third model, the results remain qualitatively the same and, again, provide compelling evidence in favour of the stronger impact of population in the European election than in the national one. The variable Madrid has a positive and statistically significant effect at the 0.05 level.Footnote 12
In Model 4, given that we include the level of competitiveness as an independent variable, we test only provinces in the 2011 Lower House elections. In this case Population (squared) is again statistically significant, at the 0.001 level, and has the expected positive sign. Conversely, Competitiveness is not statistically significant, although it has the expected positive sign. This non-significant impact of competitiveness is in line with Criado’s findings (2004: 70–5).
To determine whether the adaptation of mobilization strategies to the incentives provided by different electoral systems depends on the size of parties, in Table 3 we have run the same models for the small national party UPyD. In the first model, when merging the observations for European and national elections, the (square of the) number of individuals entitled to vote in each province is the only statistically significant (at the 0.001 level) variable and has the expected positive sign. There are no differences between large and small parties here. However, in Model 2 the interaction between provinces in the European election and the number of eligible voters is not statistically significant. Contrary to the evidence shown for large parties in Table 2 (Models 2 and 3), the impact of population when explaining the visits of the party leader to provinces is similar in the 2009 European election and the 2011 national election for the small party UPyD. As predicted in our hypotheses, only large parties face different incentives for campaigning when the electoral system changes. The inclusion in Model 3 of a dummy variable identifying the capital city shows that differences in UPyD’s mobilization strategies in national and European elections are not a result of a Madrid effect. The variable has the expected positive sign, but it is not statistically significant.
The Determinants of the Number of Visits to Provinces in European and National Elections in Spain, UPyD
Note: Standard errors in parentheses. * p<0.05; ** p<0.01; *** p<0.001.
Finally, in Models 4 and 5, provinces in the 2011 Lower House elections only are considered. The expected high correlation between Population and Competitiveness (small parties only have good chances of winning seats in the most populated districts in which many seats are allocated) creates a severe problem of multicollinearity when they are included in the same specification. Therefore we have run two models instead of one as in the case of the PP and the PSOE. The variable Population (squared) in Model 4 is positive and statistically significant at the 0.001 level, and it shows a coefficient virtually identical to Models 1 to 3.
Based on Model 2 in Tables 2 and 3, Figures 3 and 4 simulate the impact of the share of the (square of the) population entitled to vote in a province on the number of visits by the party leader in the 2009 European election and the 2011 Lower House election. All the remaining independent variables are set at their values. The gap between the Lower House election and the European election in the number of visits is statistically significant at the 0.05 level for the large parties in the most populated districts (that is, those with more than 9.5 per cent of the population (Figure 3). In contrast, for small parties there are virtually no differences and the number of visits is even slightly higher in the national election (Figure 4).
Population and Visits to Districts in European and National Elections for Large Parties Note: The bars represent the 95 per cent confidence intervals.
Population and Visits to Districts in European and National Elections for Small Parties
CONCLUSIONS
Political parties have a crucial role in elections, particularly when it comes to the mobilization of voters. Yet the current literature has not fully explained how party strategies themselves vary depending on electoral systems. In this article we have aimed to fill this gap by showing that district magnitude and the number of districts shape campaign strategies by creating the strategic conditions under which political parties must manoeuvre.
Specifically, in analysing two elections with different district structures, the 2009 European election and the 2011 Lower House election in Spain, and using interviews with campaign managers to build our theory, and the number of visits to districts by the party leaders as a proxy for the districts targeted by parties, we have found that varying district magnitudes create different incentives for campaigning. But this effect on parties is not universal: smaller parties – given that their viability is less pervasive – always target population size or districts with the highest number of seats to be awarded, where their chances of winning are much greater. We find that larger parties alter their strategies depending on the number of districts, mainly by targeting populous provinces in an electoral system with a single national district. When there are multiple districts, large parties will visit more provinces than when there is only one district. Lastly, we have shown that the strength of the local organization of the party does not matter for the number of visits by the party leaders.
Our results make an important theoretical contribution. We show that there is not always a change in party mobilization strategies when there is a change in the payoff of votes to seats. According to our findings, a small party does not respond to the incentives produced by an electoral system that has a high variance in district magnitude because they are unlikely to be viable in small districts. Therefore, they will concentrate their efforts in districts where there are more seats and provinces with larger populations to receive votes from. Thus, they do not change their behaviour depending on the electoral system. In contrast, we find that the electoral system matters to large parties that are viable in all or almost all districts. Hence, the effect of district magnitude and the number of districts on party mobilization strategies depends on the size of the political party.
APPENDIX
Correlation between Vote Shares in the 2011 Lower House Election and the 2009 European Election
2009 European Election and 2011 Lower House Elections Results in Spain
Acknowledgements
We thank the members of the campaign teams of the PP, the PSOE and the UPyD for their willingness to be interviewed and for giving us permission to quote their answers in this article. Agustí Bosch, Romain Lachat, Ferran Martínez, Tània Verge and two anonymous referees offered helpful comments. This study is part of the Making Electoral Democracy Work project; see Blais (Reference Blais2010) and http://electoraldemocracy.com. The authors acknowledge financial support from the Spanish Ministry of Science and Innovation, research project CSO2010-1639. Preliminary versions of this article were presented at the Elections, Public Opinion and Parties (EPOP) Conference 2012, Oxford, 7–9 September 2012, at the 71st Midwest Political Science Association (MPSA) Conference, Chicago, 10–14 April 2013, and at the 3rd European Political Science Association (EPSA) Conference, Barcelona, 20–22 June 2013.
