Electoral institutions affect many facets of political life. The literature on electoral systems – specifically on their systemic effects – has extensively explored the consequences of electoral mechanisms on proportionality, the number of existing parties, the representation of minorities and women, etc.Footnote 1 Moreover, electoral institutions have been shown to shape voter turnout as well as vote choice.Footnote 2 Finally, a significant body of research suggests that electoral mechanisms also affect the behaviour of elected repres1entatives.Footnote 3 Equally, the same institutions should shape the behaviour of parliamentary candidates. In this article, we extend our understanding of electoral institutions and how they structure political competition by exploring their effect on the campaign activities of parliamentary candidates in the run-up to first-order parliamentary elections. We do so by relying on a unique, pooled dataset of candidate studies from countries that employ a diverse range of electoral mechanisms.
To date, most studies of electoral campaigns during first-order parliamentary elections have been based on single systems, lacking an understanding of the role that electoral institutions play in the process. At the same time, studies of the effects of electoral systems are much more concerned with their systemic effects than their potential impact on the behaviour of political elites. In this study, we bridge these two traditions in order to enhance our understanding of elite behaviour, which fundamentally impacts on the extent to which citizens are exposed to campaign stimuli. Therefore, while directly addressing the scholarly literature on campaign behaviour, our study also contributes to the wider debate on the role of electoral institutions in promoting political engagement and mobilization.
The novel data brought about by an extensive cross-sectional collection of candidate studies – we combine data from the Comparative Candidates Study (CCS) with compatible survey data from other countries – enable us to better grasp the multifaceted phenomena of political campaigns. In particular, we disentangle the extent to which candidates’ individual-level campaign efforts vary under different electoral set-ups. We do so by looking at the intensity and complexity of candidates’ overall campaign efforts in the run-up to first-order parliamentary elections. We find that under candidate-centred electoral institutions – for example, single-member district (SMD) plurality, open-list proportional representation (PR) and single transferable vote PR (PR-STV) systems – candidates’ campaign efforts tend to be more intense and complex than under party-centred mechanisms such as closed-list PR systems. This finding – while intrinsically intuitive – contradicts previous findingsFootnote 4 and indicates that electoral institutions have much more of an impact than previously thought. We also shed further light on the relationship between individual candidates and their parties by assessing how electoral institutions influence the candidates’ campaign focus. Not surprisingly, the extent to which the candidates’ campaign messages focus on themselves versus their party also depends on the structure of the electoral incentives, ceteris paribus. In addition, we find that candidates campaign harder when the district magnitude is smaller.Footnote 5 All in all, the amount and type of campaign stimuli that voters are likely to experience is closely related to the electoral institutions.
The article is organized as follows. In the next section, we survey the literature that we touch upon to guide our investigation. We then outline our expectations, and describe the data and the operationalization of the variables we use. This is followed by a presentation and discussion of the results of our empirical analysis. We conclude by summarizing our findings and evaluating their implications.
ELECTORAL INSTITUTIONS AND ELECTORAL CAMPAIGNS
Partisan dealignment and societal and technological changes have jointly contributed to shaping the nature of electoral campaigns. As the number of floating votersFootnote 6 and late decidersFootnote 7 who can be influenced in the run-up to an election keeps rising, so does the scholarly attention paid to campaign mobilization.Footnote 8
In defining the very concept of an electoral campaign, Farrell and Webb claim that there is an obvious relationship between the nature of electoral campaigns and institutional settings.Footnote 9 Empirical tests to corroborate this claim are very scarce. The available evidence comes mostly from the voters’ perspective: several studies suggest that electoral institutions can play a significant role in shaping the extent to which voters experience campaign stimuli.Footnote 10 While the claim that electoral institutions influence campaign mobilization is essentially uncontested in this literature, there is disagreement over whether PR systems stimulate electioneering more than plurality-based systems, with reasonable arguments provided by both camps. On the one hand, PR systems involve more competitors, which tends to increase overall campaign activity and produce competitive contexts characterized by extensive campaign efforts. On the other hand, lower turnout – typical of plurality systems – boosts the level of mobilization in the run-up to a vote. However, these two divergent theses share the common assumption that electoral institutions play a role in shaping campaign activity, which in turn affects electoral participation. Karp and colleagues systematically tested these competing hypotheses and disentangled the matter by relying on data from the Comparative Study of Electoral Systems as well as national election studies.Footnote 11 Consistently, across both types of data sources, they find that voters are more likely to experience higher degrees of campaign stimuli in candidate-based electoral systems. The representation literature has provided further evidence that electoral systems influence the behaviour of members of parliament (MPs).Footnote 12 It is therefore reasonable to expect that differences across electoral institutions will also shape the behaviour of parliamentary candidates. After all, public representatives have been campaigning before the election and most likely will run again.
Conversely, the few studies that have investigated electoral systems from the candidates’ perspective – in the context of the second-order European elections – have not found such an uncontested link between electoral institutions and campaign efforts. Bowler and Farrell explore the nexus between electoral systems and campaign activity using 2006 MEP survey data,Footnote 13 and find no evidence of electoral institutions affecting levels of campaign effort.Footnote 14 They conclude that electoral institutions affect campaign goals, but fail to play a distinct role in structuring the campaign effort. Moreover, Giebler and Wüst, studying the 2009 European election, find no evidence that electoral systems shape the intensity of candidates’ campaign efforts (measured in terms of money or time), and find only a partial indication that they influence candidates’ choice of campaign tools.Footnote 15
In sum, the claim that electoral institutions shape campaign mobilization efforts is essentially contested and empirically less obvious than what is theorized. If the electoral architecture does indeed play a role in explaining variance in the campaign processes, citizens’ capacity to experience first-hand contact with political elites consequently varies, as does the nature of that contact. This has critical implications for political mobilization and electoral participation.
DISENTANGLING THE RELATIONSHIP BETWEEN ELECTORAL INSTITUTIONS AND CAMPAIGNS
We begin by identifying what might explain the inconsistency between the voter- and elite-side literatures. Conflicting findings on the effects of electoral institutions may be due to three reasons, each of which our empirical strategy directly addresses.
First, comparative studies of the elite perspective have, so far, looked at European elections, which have limited variation in electoral set-ups, as all European Union member states must use some form of PR. Accounting for a greater degree of variation in electoral set-ups might therefore be necessary to uncover the effects of electoral institutions on candidates’ campaign behaviour. The CCS project, complemented by other compatible candidate surveys, covers a wide array of electoral mechanisms, including plurality systems. We implement a threefold strategy to systematically measure variation across the electoral systems and grasp their potential impact on campaign effort. First, we separately examine PR and SMD systems to explore whether candidates behave differently under these two broad electoral system families. This comparison will offer an initial indication of whether electoral institutions affect campaigning and electioneering. We then gauge more nuances of electoral institutions by using the index developed by Farrell and Scully’s seminal study.Footnote 16 This modified version of Carey and Shugart’s indexFootnote 17 simultaneously accounts for the role of (a) ballot access, (b) vote choice and (c) district type in shaping the electoral environment in which candidates compete. It captures more of the nuances of the various electoral set-ups than a simpler PR versus SMD dichotomy. This is particularly relevant given the plethora of PR systems that are being used to elect MPs. Finally, we focus on the effect of district magnitude and address its interplay with vote choice. The assumption here is that candidates’ campaign behaviour is influenced by whether voters can cast personal/preference votes, with the district size acting as a moderator.Footnote 18
Secondly, measurement issues in the dependent variable may be responsible for the lack of findings from studies of second-order European elections. For example, the analysis by Giebler and Wüst is limited to overall campaign effort in terms of the time and money spent by candidates, which does not necessarily account for the complexity of their campaign effort. In order to address this shortfall, we use a wider collection of information on candidates’ campaign efforts. In addition to capturing the intensity of their campaign – the time spent campaigning – we also account for the complexity of their campaign efforts by exploring the range of communication channels used to reach potential voters. In other words, we gauge both the ‘quantity’ and the ‘quality’ of the candidates’ campaign efforts.
Finally, the mismatch between the findings from the voter and elite sides of the electoral equation may be due to the level of election that these studies have focused on. While the former are based on first-order parliamentary elections, the latter are, to date, limited to second-order European elections. Our study allows for the comparison of findings from the supply and demand sides within the same type of election, which is pivotal to resolving the unsettled incongruence.
Hypotheses
Bowler and Farrell conclude that, regardless of the electoral incentives in place, candidates work equally hard to get elected.Footnote 19 This casts doubt on whether electoral institutions differ in their capacity to mobilize voters. We treat this claim as a null hypothesis.
Electoral campaigns represent the effort put in by candidates and parties to win votes. From a rational choice theory perspective, it is reasonable to imagine that the rules of the game affect the behaviour of the players: candidate-centred electoral mechanisms should put a stronger onus on individuals to promote their candidacy. The link between one’s campaign activity – effort and focus – and electoral performance is simply more direct under these rules. The need to maximize one’s personal reputation and profile, and to establish personal ties with constituents, is more salient and clear-cut in a system that rewards personal/preference votes.
This should be reflected in the amount of time and the type of resources that candidates invest in their campaigns. For example, the utility of promoting one’s candidacy can reasonably be seen as high for candidates in systems like the Irish PR-STV or the Estonian open-list PR, in which all contenders have, in principle, an equal chance of being chosen by voters. Conversely, candidates in systems such as the Portuguese closed-list PR are undoubtedly less likely to find the motivation to put in an equally high level of campaign effort, as it is ultimately the party vote, rather than the personal vote, that shapes the electoral outcome. The link between an individual candidate’s campaign effort and her electoral performance is simply not as clear-cut in party-centred electoral systems as it is in candidate-centred ones. For example, in Portugal, candidates at the top of their party list can reasonably expect to get elected irrespective of their own campaign effort. At the same time, there are no clear personal benefits associated with stronger campaign efforts for candidates placed at the bottom of their party list. In party-centred systems, the utility of personal campaign efforts is lower than in candidate-centred systems.
In addition to the effect associated with electoral rules, district magnitude is likely to affect campaign behaviour, both on its own and in tandem with the institutional set-up of candidate choice options. Identifying the target voters to contact and making one’s presence visible is simpler in small districts.Footnote 20 Therefore, a lower district magnitude should encourage candidates to seek out direct contact with voters, while a larger district magnitude should push them to rely more on their party image and exert less personal effort. However, the effects of the district size on the campaign effort are likely to be conditioned by whether voters cast their ballot for candidates or parties. If candidates compete for personal votes – that is, ballots are cast for candidates – they simply cannot afford to rely merely on their party image, even when standing in a large constituency where establishing direct contact with voters is difficult. In these contexts, candidates face competition from co-partisans and are under more pressure to distinguish themselves from others. Therefore, the effect of district magnitude on campaign effort should be moderated by whether voters cast their ballot for parties or candidates. Based on these considerations, we formalize the following hypothesis:
Hypothesis 1: Electoral institutions significantly affect campaign effort in the following ways: (a) candidate-centred electoral systems promote greater campaign effort than party-centred systems, (b) district magnitude has a negative effect on campaign effort and (c) this is weaker in systems where voters cast their ballot for candidates as opposed to parties.
Another element of campaigning that is reasonably related to the structure of the electoral institutions is the focus of the candidates’ messages. All candidates seek to get out the vote and win votes, but they can choose to do so by putting more (or less) emphasis on themselves (versus their party). Again, the rules of the game should impact the players’ behaviour. Bowler and Farrell introduce ‘a conceptual distinction, reserving the term “electioneering” for the kinds of acts that get people out to vote and using the term “campaigning” to refer to the strategy at election time with regard to campaign goals (maximise party as opposed to individual vote share)’.Footnote 21 Building on this, we expand our assessment of candidates’ campaign behaviour to explore whether electoral institutions influence the extent to which candidacy is promoted by encompassing or bypassing the party.
While the expectations here are theoretically intuitive – party-oriented campaigns should prevail in party-centred electoral set-ups like closed-list PR, and candidate-focused campaigning should take place in systems such as open-list PR and SMD – the empirical evidence is still unsystematic. Zittel and Gschwend’s study of the German mixed system indicates that electoral incentives affect the candidates’ campaign objectives,Footnote 22 but the study is limited to one country and therefore has narrow external validity. A series of case studies in Colomer’s collection offers further support for the underlying intuition, but does not provide comparable or comparative assessment.Footnote 23 Counterintuitively, Marsh finds that under the Irish PR-STV system – one of the most candidate-centred electoral mechanisms – campaigns are candidate-centred, as one would expect, but still remain ‘party-wrapped’.Footnote 24 Moreover, we still do not know how district magnitude affects – directly and conditionally – the link between electoral institutions and campaigning. Following the above-mentioned considerations on how larger district magnitudes should push candidates to rely more on their party image, particularly where voters cast party-based (versus candidate-based) votes, we expect district magnitude to shape candidates’ campaign focus. Formally, we test the following hypothesis:
Hypothesis 2: Electoral institutions significantly affect campaign focus in the following ways: (a) candidate-centred electoral systems promote a more candidate-centred campaign focus, (b) district magnitude has a negative effect on candidate-focused campaigning and (c) this is weaker in systems where voters cast their ballot for candidates as opposed to parties.
DATA AND MEASURES
We evaluate our theoretical expectations using a unique collection of information on candidates running in first-order parliamentary elections.Footnote 25 The CCS project, which brings together a wide range of national candidate studies and uses a common core questionnaire to allow for cross-country comparisons, is the main source of this information. To further maximize the array of electoral institutions under study, we combine the CCS data – version 2013AF2 – with additional survey data from countries not included in the project.Footnote 26 As a result, we have information on the campaign activity of candidates from fifteen countries.Footnote 27 With the same key questions asked in a large pool of countries, it offers the first opportunity to extend the analysis of how electoral institutions influence the campaign behaviour of candidates to first-order parliamentary elections.Footnote 28
Dependent Variables
In line with our theoretical approach, we identify survey measures that tap into the different strategic choices associated with electioneering and campaigning. Starting with the former, we implement a dual operationalization. Our ‘quantitative’ measure of Campaign Effort describes how many hours per week each candidate spent on her campaign during the last month before the election, ranging from 0 to 70.Footnote 29 This is preferred to campaign spending – which is extensively used as a proxy of campaign effort – as candidates were considerably less inclined to self-report their expenditure in the surveys. The number of hours – measured on an open scale – reduces the amount of missing values and moderates concerns about misreporting, while still representing a reliable and widely used proxy for the overall intensity of one’s campaign effort.Footnote 30 To better gauge the concept of campaign effort, we complement the ‘quantitative’ measure with an index (ranging from 0 to 5) that captures the complexity of a candidate’s campaign effort, tapping into its ‘quality’. It describes how many campaign activities, from the following options, each candidate used as part of her campaign: (1) canvassing, (2) direct mail, (3) online campaigning, (4) newspaper interviews and (5) TV interviews. These options not only maximize the amount of cases for analysis but also account for the different types of effort, including localized door-to-door direct contact with voters as well as both traditional and new media forms of campaign advertisement. This measure accounts for the complexity and richness of the campaign effort put in by candidates in order to mobilize voters and seeks to provide evidence of whether voters are likely to experience various stimuli under different electoral rules.
The benefits of the dual operationalization are twofold. First, the two measures tap into slightly different aspects of the candidates’ campaign effort, on both theoretical and empirical grounds, correlating at 0.4 (p<0.001). While the ‘quantitative’ measure remains an intuitive way of assessing how much an individual works to win a seat, it does not per se offer any nuances about the complexity of her campaign. In the context of post-modern campaigns – in which several targeting tools and contacting options are available – candidates are compelled to fight on multiple fronts. The extent to which candidates decide to embed multiple communication channels into their campaigns is therefore crucial to defining the type, and scope, of the stimuli that voters experience before the election. One could indeed choose to spend a great amount of time on canvassing, but ignore other forms of contact. Therefore, our ‘qualitative’ measure offers further insights into whether electoral system effects extend to shaping the choice of electioneering techniques. Secondly, the reliance on two indicators offers a robustness check as both measures, while describing different elements of one’s campaign effort, tap into the same underlying concept.
With regard to campaigning, our dependent variable is labelled Campaign Focus. It is based on a survey question that directly addresses the primary aim of one’s campaign. This measure ranges from 0 ‘to attract as much attention as possible to my party’ to 10 ‘to attract as much attention as possible to me as a candidate’, and offers a unique comparative insight into the kind of messages that candidates convey to the electorate through their campaign effort.
In Table 1, we show the averages and standard deviations per country for the three dependent variables. This offers an initial indication that electoral institutions might play a part in explaining this variation. For example, if we look at the most party- and candidate-centred PR systems – that is, closed-list PR (Portugal) and PR-STV (Ireland) – we can appreciate some suggestive evidence for it. Candidates in Ireland, on average, engage in greater campaign effort in terms of time (45 versus 28 hours) as well as complexity (4.2 versus 2.5 campaign activities) than those in Portugal, and the former also conduct more personalized campaigns (5.6 versus 1.7).Footnote 31 Yet since these differences may be due to contextual elements, they need to be rigorously assessed by including possible alternative explanations.
Descriptive Statistics on Campaign Effort and Campaign Focus
Note: calculated based on samples used in Models 1, 4 and 7.
KEY EXPLANATORY VARIABLES AND CONTROLS
We rely on survey data from a diverse set of countries that covers a variety of electoral institutions and district magnitudes.Footnote 32 Among these countries, Canada uses SMD to elect all members of the House of Commons, while Australia and Germany use SMD to elect a portion of legislators.Footnote 33 This enables us to start by looking at the differences in the candidates’ campaign activity under SMD versus PR. This initial comparison is straightforward, and we capture it by coding candidates who stood in a SMD system as 1 and those who stood in a PR system as 0. This offers a first, if raw, indication of whether electoral set-ups affect campaigning and electioneering.
To better gauge the nuances, we then classify electoral systems following the seminal study of Farrell and Scully,Footnote 34 which operationalizes the concept of Electoral Incentives as the cumulative score of:
1. Ballot access: the degree of party versus voter control over the ballot placement of candidates, ranging from 1 to 3.
2. Vote choice: the extent to which voters are able to vote for a specific candidate, ranging from 1 to 4.
3. District: the effect of a district type on the importance of personal reputation, ranging from 1 to 2.
Higher scores across these components indicate a candidate-centred electoral system. Therefore, the overall index ranges from 3 to 9, with higher values corresponding to greater incentives for candidates to cultivate a personal vote.Footnote 35
As the final, yet potentially important, element of the electoral set-up, we explore the role of District Magnitude in influencing the candidates’ campaign behaviour.Footnote 36 We do so by looking at whether district magnitude has a direct, independent effect on campaign behaviour, as well as whether its effect is being conditioned by vote choice. The value of personal reputation should decline as district magnitude increases in systems where voters cast party-based votes such as closed-list PR, whereas it should increase as district magnitude increases in systems where voters cast candidate-based votes such as open-list PR. The measure is operationalized as the natural logarithm of the number of seats allocated in the constituency.Footnote 37 The use of a natural logarithm is the conventional practice,Footnote 38 and is particularly useful for correcting the skewed nature of the district magnitude in our data.Footnote 39
Finally, we control for several elements that previous studies have shown to be relevant to explaining campaign effort and focus.Footnote 40 First and foremost, we account for campaign marginality – that is, the chance of being elected. Clearly, the higher the foreseen chance of success, the stronger the incentive for electioneering. Moreover, candidates who believe that they are unlikely to get elected are likely to put less effort into their own campaign and opt for a more party-focused campaign strategy to build a profile within their party, with consequent implications for campaigning. We operationalize the Likelihood of Success as the candidates’ self-perceived likelihood of success before their campaign started, ranging from 1 ‘I could not win’ to 5 ‘I could not lose’.
Secondly, the nature of the candidate’s campaign may be influenced by her relationship with her party. Namely, the further away a candidate’s own positions are from those of her party, the more likely she is to conduct a campaign with a personalized focus and to put in extra effort to get her own political views across. We measure Ideological Distance as the absolute difference between the left-right position of the candidate and that of her party (as perceived by the candidate). It ranges from 0 ‘no difference’ to 10 ‘maximum difference’. Left-right placement is not only the most comparable measure for a study of this sort, but also the best available shortcut for aggregating multiple policy positions.Footnote 41
Thirdly, we look at the candidates’ political experience by accounting for incumbency (Past MP) and their position within the party (Party Hierarchy). Past MP is coded 1 if the candidate has been an MP before and 0 if not, and Party Hierarchy is coded 1 if she has held national party office or been employed by it, and 0 otherwise. We expect candidates with such experience to undertake more intensive and personalized campaigns. Fourthly, we control for the candidates’ proximity to voters by separating those living in the Constituency where they stand for election (coded 1) from the rest (coded 0) on the premise that the former are more invested in constituency matters and, therefore, push harder to get elected through a stronger campaign effort and a more personalized campaign focus.Footnote 42
Empirical Strategy
Given the different data structures of our dependent variables, different estimation techniques are used to analyse variation in the different aspects of campaign behaviour. With regard to the ‘quantitative’ measure of electioneering, time, we use an Ordinary Least Squares (OLS) method, as the dependent variable is measured in hours per week. Similarly, OLS is implemented to empirically address campaign focus. Conversely, the ‘qualitative’ index of campaign effort is analysed using Ordered Probit given the structure of the variable. In line with the key explanatory elements described above, we run three sets of models for each dependent variable: the first set (Models 1, 4 and 7) accounts for electoral mechanisms by using the simple plurality dummy, the second set (Models 2, 5 and 8) uses the electoral incentives index, and the third set (Models 3, 6 and 9) explores the interactive effects of district magnitude and vote choice. Finally, model specifications take into account the country-specific contexts of electoral politics; therefore, we include country dummies as a control.Footnote 43 This accounts for any country-specific effects beyond the electoral institutions.
EMPIRICAL ANALYSIS
We begin by addressing the effects of electoral institutions on electioneering. The first set of models (1–3), reported in Table 2, explains variation in campaign effort (time). In line with our theoretical expectations, candidate-centred electoral systems seem to provide an incentive for candidates to campaign harder. The coefficient of SMD in Model 1 shows a difference of 11 hours per week between plurality and PR systems as the former produce higher-intensity campaign effort. When we look at more nuances of the electoral mechanisms in Model 2, we observe an average increase of 3 hours per week in campaign effort as we move from those who have weaker systemic incentives to cultivate a personal vote to those who have stronger incentives to do so. Consistently across the different models, electoral institutions have a significant effect on the intensity of candidates’ campaign efforts.
Explaining Variation in Campaign Effort (Time)
Note: robust standard errors in parentheses. Personal vote: (1) vote for list only; (2) vote for list or candidate, list vote dominates; (3) vote for list or candidate, candidate vote dominates; (4) vote for candidate only. **p<0.01, *p<0.05
Table 2 also shows that district magnitude has a negative effect on the dependent variable. A shift from the smallest to the largest constituency corresponds to a six-hour-per-week decline in candidates’ predicted campaign effort (based on estimates in Models 1 and 2). As indicated by Model 3, the negative effect of district magnitude is, however, influenced by vote choice. It is strongest in electoral systems where voters are required to cast their ballot for a party list. Clearly, the extent to which electoral mobilization is promoted through candidates’ campaign effort is substantially different across the various electoral systems. The more candidate-centred the electoral set-up, and the smaller the district magnitude, the higher the intensity of candidates’ campaign effort, ceteris paribus.
As expected, however, campaign effort is not just a function of the electoral set-up. At the individual level, we find that candidates who are confident in their electoral chances ahead of their campaign conduct significantly higher-intensity campaigns than those who did not expect to get elected. Predicted campaign effort rises from 18 hours per week for candidates who thought they had no chance of getting elected to 34 hours per week for candidates who felt certain they would get elected prior to campaigning. In addition, small positive effects – that is, around 3 hours per week – are associated with both political experience variables. At the same time, we do not find that ideological distance or proximity to voters has a significant effect on how much time candidates choose to spend on their campaign.
With regard to the qualitative measure of candidates’ campaign effort, we find that the same patterns are visible when explaining the complexity of the candidates’ campaign effort (Table 3). In line with our theoretical expectations, candidates in SMD systems tend to use a wider range of campaign activities than their counterparts in PR systems, while the positive effect associated with electoral incentives indicates that candidates undertake more complex campaigns when the electoral set-up offers stronger incentives to cultivate a personal vote. In addition, larger district magnitude coincides with the use of a narrower range of campaign activities, and the largest negative effect is again observed in electoral systems where voters have to cast their ballot for a party list. These findings are in line with what we observed when looking at the ‘quantitative’ aspect of campaign effort, which reaffirms that higher levels of campaign effort are associated with more candidate-centred electoral set-ups and smaller district magnitudes.
Explaining Variation in Campaign Effort (Complexity)
Note: robust standard errors in parentheses. Personal vote: (1) vote for list only; (2) vote for list or candidate, list vote dominates; (3) vote for list or candidate, candidate vote dominates; (4) vote for candidate only. **p<0.01, *p<0.05
Figure 1 illustrates the effects associated with electoral institutions, and shows how candidates’ likelihood of undertaking campaigns of high and low complexity is conditioned by electoral incentives (Figure 1a) and district magnitude (Figure 1b). Figure 1a highlights how the probability of high campaign complexity – that is, maximum score for campaign complexity – increases as we move from a party-centred electoral system to a candidate-centred one, while the opposite is true for candidates’ likelihood of conducting low-complexity campaigns. In a similar vein, Figure 1b shows that the effects of district magnitude go in the opposite direction. While lower district magnitude leads to a higher likelihood of implementing a broad array of campaign tools, larger districts depress the use of multiple campaign tools. These effects are, however, of notably smaller scale. A shift from the smallest to the largest district brings about an 8 per cent increase in candidates’ likelihood of conducting a low-complexity campaign (from 8 per cent to 16 per cent) and a 9 per cent decline in their likelihood of conducting a high-complexity campaign (from 24 per cent to 15 per cent). In sum, candidate-centred electoral set-ups and smaller district magnitudes tend to propel higher mobilization efforts, in terms of campaign intensity as well as complexity.
Effects of electoral institutions on campaign effort (complexity) (a) Effect of electoral incentives on campaign effort (complexity) (b) Effect of district magnitude on campaign effort (complexity)
When district magnitude and electoral incentives are examined separately, we can appreciate their independent effects on campaign effort. When we assess their joint effects in Models 3 and 6, however, we find some evidence to confirm the intuition of Carey and Shugart that the incentives to cultivate a personal vote decline in closed-list systems when district magnitude increases, and that the opposite happens in open-list systems.Footnote 44 Nevertheless, there is insufficient proof that such interactions exist: while the direction of the coefficients is in line with the expectations, the associated confidence intervals overlap in Model 3 and the significance of certain categories in Model 6 has little meaningful value.Footnote 45
Having seen how electoral institutions affect both the ‘quantity’ and ‘quality’ of campaign effort, we move to assessing whether the electoral set-up also has an effect on the extent to which the candidates’ campaign messages focus on themselves versus their party. Table 4 presents the outputs from the analyses of campaign focus. The findings are clear and consistent: candidate-centred electoral set-ups incentivize candidates to opt for a more candidate-focused campaign strategy. The coefficient of SMD in Model 7 shows a 2.2-point difference between SMD and PR systems, with the former producing more personalized electoral campaigns. When looking at electoral systems in a more nuanced manner in Model 8, we observe an average 0.6-point increase in the extent to which candidates prioritize their own (versus party) image when we move from those who have weaker systemic incentives to cultivate a personal vote to those who have stronger incentives to do so. While quite modest in size, the effect is robust across the different specifications of the key independent variable. On the contrary, the evidence that the district magnitude has an effect on campaign focus is limited. We do not observe a significant independent effect for district magnitude in Models 7 and 8, and a significant negative effect is only present when focusing on two categories of vote choice in Model 9. Once again, the evidence of a potential interaction effect between electoral incentives and district magnitude, albeit in line with the expectations, is empirically weak.
Explaining Variation in Campaign Focus
Note: robust standard errors in parentheses. Personal vote: (1) vote for list only; (2) vote for list or candidate, list vote dominates; (3) vote for list or candidate, candidate vote dominates; (4) vote for candidate only. **p<0.01, *p<0.05
At the individual level, we find that candidates who are more confident in their electoral chances conduct more candidate-focused campaigns, as do those who feel ideologically more distant from their party. At the same time, there is no evidence that the candidates’ previous political experience or proximity to voters plays a role in influencing their campaign focus.
In summary, the empirical analysis clearly indicates that electoral institutions are significant predictors of electioneering practices after all, leading to the rejection of the claim that candidates work equally hard irrespective of the electoral system type. However, when it comes to assessing the effects of electoral institutions on campaigning, the evidence is more multifaceted. More candidate-centred electoral systems do lead to more personalized campaign messages, as expected, but district magnitude does not have an independent effect on the campaign focus. A significant negative effect for the district magnitude is only found if voters have to (or tend to) cast their ballot for a party. Regardless of how large (or small) the district, candidates concentrate on promoting their own personal image if the electoral set-up encourages or requires voters to cast personal/preference votes.
The disjuncture between previous findings from the elite-side and voter-side literatures is resolved when looking at the first-order parliamentary elections, and when the complexity of electioneering and campaigning are taken into account.
CONCLUSIONS
In this article we have explored the effects of electoral institutions on the campaign behaviour of candidates running for office in first-order parliamentary elections. While studies of the European elections have provided us with valuable insights into the impact of electoral institutions on individual-level campaigning during second-order elections, we have taken a first step towards extending this type of analysis to the arena that remains the most crucial for political competition among candidates and parties, as well as for voters. Our study contributes to the wider debate on the role of electoral institutions in promoting political engagement and voter mobilization by showing that smaller districts in candidate-centred electoral systems maximize the likelihood that voters will experience higher-intensity campaigns.
Campaigns provide voters with the knowledge they need to make an informed vote choice and increase the likelihood of electoral participation.Footnote 46 The engineering of electoral institutions is often inspired by considerations of what systemic effects electoral rules will produce. Here, we have shed some light on the strategic effects of electoral institutions by showing that electoral systems have an independent impact on elite-level behaviour. Where electoral rules allow for candidate-based vote choice(s), candidates tend to put forward more intense and complex campaign efforts, and campaign along less partisan lines. This goes hand in hand with what is found in studies of electoral institutions and styles of representation,Footnote 47 but contradicts what we know from studies of second-order elections, as the latter have pointed to a weak link between electoral institutions and the candidates’ campaign behaviour.
Our study can reconcile the supply, elite side with what is observed in voter studies, suggesting that the second-order European elections may not be the best venue for exploring the effects of electoral institutions, as they are likely to be masked by little variation in the independent variables and limited by the sample of candidates. The cross-national CCS project, taken together with other compatible candidate surveys, can overcome the limits of our current understanding of parliamentary candidates’ campaign behaviour, intentions and attitudes. This collection of candidate studies enables us to extend the analysis of how electoral institutions influence candidates’ campaign behaviour to first-order parliamentary elections. In addition, it offers nuances on how candidates mobilize voters by going beyond voters’ reported contact. Voter studies usually rely on questions about door-to-door and/or telephone contact,Footnote 48 leaving aside the more recent and upcoming forms of campaign tools like contact via the internet. Our qualitative measure of electioneering encompasses a wide array of means that candidates may use to reach the voters.
We find that electoral institutions do shape candidates’ campaign behaviour in the run-up to first-order parliamentary elections, both in terms of campaign effort and campaign focus. The extent to which voters are likely to experience campaign stimuli is closely related to electoral institutions, as candidates’ campaign effort tends to be more intense, and complex, under candidate-centred electoral institutions than under party-centred ones. Equally, candidates seem to tailor their campaign messages to the electoral context, as messages tend to be more candidate focused in candidate-centred electoral systems. While the debate on the personalization of electoral campaigns has, to date, focused mostly on the role of party leaders, our findings indicate that it should also feature the candidate side more prominently. All in all, the findings suggest that the impact of electoral institutions on campaigns is more far-reaching than previously thought. Both the amount and type of campaign stimuli that voters tend to experience is closely related to electoral institutions.
In sum, these findings broaden our understanding of how electoral institutions affect campaign practices in the run-up to first-order parliamentary elections. It is generally accepted that electoral campaigns are shaped by country-specific dynamics as well as party- and candidate-specific characteristics. The empirical evidence offered here adds electoral institutions to this list.
