Being Heard above the Noise: The Role of Incumbent Issue Diversity in Election Campaigns

Historical policy reputations influence voters’ perceptions of parties’ electoral campaigns. In the face of their recent experiences in office, government parties’ thoughtfully crafted electoral messages likely compete for voters’ attention with a wealth of broader information about the government’s policy activities and priorities. For their message to be heard, incumbent parties must offer a focused policy message that draws voters’ attention to the issues they most prioritize. Considering the issue scope of parties’ electoral messages, I hypothesize that incumbency status determines the effect issue appeals have on the votes parties receive. Opposition parties may profit from including more issues, but incumbent parties’ policy reputations limit the potential benefits from diverse appeals. Using evidence from 25 OECD countries over a 60 year period, I find that parties’ incumbent status conditions the effect of issue diversity on parties’ aggregate electoral success. Voters reward incumbents for focusing their platforms, but reward opposition parties for diverse appeals. The results for incumbent parties are robust to extensive sensitivity analyses. The theory and evidence broadly suggest that incumbent parties with more focused policy messages can, at least partially, overcome the weight of their past policy reputations.

1 3 on the issue scope of campaigns and its electoral consequences remains unexplored. For example, it is unclear whether parties such as the French Parti Socialiste in 1993 are punished for discussing so many issues (over 27) that specific policy goals are difficult to discern. Or, do single issue parties only attract limited support because of the narrow scope of their messages? Perspectives deriving from Downs (1957), Schattschneider (1975), or Riker (1982) suggest an electoral dynamic between parties' past electoral success and the issues they discuss, yet few analyses consider how the breadth or scope of parties' campaigns influences their electoral success.
Building on the strengths of issue competition research, I outline a perspective on issue scope and competence. I propose that the relative number and salience of issues parties include in election campaigns, issue scope or issue diversity, conditionally determines voter support. 1 The effect of diverse issue appeals depends on parties' government status. Including a broad range of issues allows parties to appeal to a wide coalition of supporters when they are in the opposition. Incumbent policy reputations and the broader information available about government parties' priorities, however, place constraints on the issues parties can use to mobilize voter support (Green and Jennings 2012a, b;Bawn and Somer-Topcu 2012;Greene 2016a). The immediacy of recent policy activities to voters limits the effect of issue scope by diluting the extent to which the manifesto alone represents the party's message. Parties campaigning on a smaller number of issues in this case can indicate a strong and explicit dedication to those issues, while more diverse appeals become lost. Limited issue scope comes with a potential downside; parties risk being perceived as uncompromising, irresponsible, or in the most extreme case, even single-issue if they ignore salient topics. Therefore, incumbent status determines the effect of issue scope on electoral success. Opposition parties attract more votes when campaign appeals are more diverse, whereas incumbents benefit when they limit their scope to a smaller diversity of issues.
To evaluate predictions on the effect of parties' campaigns, I examine hypotheses predicting that issue scope motivates voter behavior. I demonstrate support consistent with this perspective using data on issue diversity from the Comparative Manifesto Project for 25 OECD countries from 1960 to 2012. 2 In particular, the analysis reveals consistent evidence for the negative effect of issue diversity on incumbent parties' vote shares. Extensive robustness checks including two-stage least squares, a range of governmental and party controls, varying measurement and modeling strategies, and even accounting for the medley of reasons driving parties' issue competition strategies suggest that issue diversity's effect remains even once accounting for issue diversity's contextual causes. A series of sensitivity analyses using an alternate measure of issue diversity from the Chapel Hill Expert Surveys and large number of alternate model specifications reveal consistent support for the effect of incumbent issue diversity.
This perspective holds implications for conceptions of election campaigns, government accountability, and broad democratic theory. The findings show that parties' ability to distract from their policy reputation by increasing the scope of issues may be more independent from their context than past studies suggest (e.g. Greene 2016a, b). Incumbents may increase the diversity of their platforms in response to poor economic conditions, but the results show that a more focused policy platform nearly always benefits incumbents even accounting for the endogenous effects of the economy. An unfocused policy message limits the effectiveness of parties' campaigns and results in decreased voter support for incumbent parties. Furthermore, diverse appeals work differently than those from a broad or vague appeal strategy (e.g. Somer-Topcu 2015), as the distinct groups parties seek to mobilize through their issue emphasis will be unable to project their positions onto the party if issue appeals fail to be heard or are perceived as insincere. Substantively, the scope of parties' campaigns has a similar immediate impact on vote outcomes to other important tactics. While voters may become informed using broad factors such as the economy, the main results and robustness checks indicate that election campaigns likely also matter.
A functioning representative democracy requires that governments react to constituent demands and voters can in turn hold parties electorally accountable for being unresponsive. Groups supporting issues consistently excluded from campaigns will be less substantively represented than groups supporting topics more consistently addressed. Parties, particularly in office, must then make a clear choice over which issues to address, as additional issues over their core message may fail to deliver. Conversely, governing parties minimize the incumbent's curse through a clear, focused message that discusses a narrower range of topics.

Issue Salience and Competition
Research on parties' campaigns focuses on strategic choices such as selective issue emphasis or relative preferences. These theories largely predict that parties emphasize issues or shift positions to attract votes or undercut support for competitors. Theories of issue competition suggest a dynamic role between issue emphasis, electoral conditions and voter support that has become increasingly important (Green-Pedersen 2007), yet provide only limited consideration of the scope of parties' messages.
Studies of issue competition offer conceptual foundations to explain issue scope. Seminal work by scholars such as Schattschneider (1975) and Riker (1982) envisioned political contests being waged over candidates' issue emphasis. Conceptually, issues are topics below the level of broad ideological cleavages (Lipset and Rokkan 1967), but over which groups in society disagree and can be mobilized (Budge 1993). Whereas Schattschneider hypothesizes that electoral loss leads parties to mobilize on new issues (expanding the scope of debate), Riker's perspective predicts that issues disappear from campaigns through a process of natural selection; less successful issues disappear from the public debate. Both perspectives emphasize past elections as a source of information to evaluate the electoral viability of issues and the uncertainty of emphasizing new ones.
Contemporary perspectives also emphasize that parties may use past election results to inform their campaigns, but that their utility is limited across contexts.
Following alternate decision-rules, parties may choose to ignore this information, respond to electoral losses, or even emphasize issues that their competitors may struggle to address for ideological reasons (e.g. Budge 1994). Studies from a dynamic agent-based modeling perspective further add that "Hunter parties," or those that repeat policy changes in response to positive support in the past, outperform others that do not seem to make extensive use of past electoral success (Laver 2005). Yet, parties' ability to respond to this information is likely constrained by parties' internal factions and decision-making process (Budge et al. 2010;Ceron 2012;Schumacher et al. 2013;Greene and Haber 2016).
On both theoretical and empirical grounds scholars have examined the contexts in which parties discuss issues. Broadly, parties' campaigns are designed to manage intra-party support and attract voters. Consequently, these perspectives emphasize the role of parties' historical reputations, public opinion, and parties' strategic positioning for the issues addressed within their campaigns.
Foremost, the issue ownership perspective develops an explanation of parties' issue appeals from their historical behavior in office and policy reputations (Petrocik 1996). Scholars have long shown that parties benefit from emphasizing issues important to historical supporters (Hibbs 1977;Budge and Farlie 1983). By dedicating resources to an issue, parties develop a positive policy reputation (Petrocik 1996). Consequently, parties benefit from addressing these issues when the public perceive them as most competent on those issues.
Empirical analysis indicates a complex relationship between parties' reputations and their campaigns. In particular, issue ownership reputations primarily impact the voters that that hold an issue salient (Bélanger and Meguid 2008). Further, a party's ownership of an issue is rarely complete (Bélanger and Gélineau 2010;Geys 2012). Voters' perceptions of parties' competencies themselves are influenced by partisan identification and ideological proximity (Vegetti 2014; see also Stubager and Slothuus 2013). Ultimately, voter perceptions of ownership provide a weak predictor for what parties do in office (Egan 2013).
Other perspectives contend that parties respond to public opinion. They predict that parties address issues when they become salient to societal groups (Spoon and Klüver 2014;Klüver and Spoon 2016). The logic follows that as issues become prominent to the public, parties discuss them or appear unresponsive (Adams et al. 2011). Exogenous events and economic conditions pressure governmental parties to develop diverse policies (Green-Pedersen and Mortensen 2010;Greene 2016a). The underlying premise is that ignoring salient topics is risky. Disregarding important salient events likely causes parties to appear out of touch and unresponsive, which limits their ability to only focus on issues that voters' perceive them as owning. Political Behavior (2020) 42:487-507 Therefore, when other parties in the system address diverse issues, disregard for important issues perceived to be important could cause parties to lose support or provide opportunities for opposition criticism (Sigelman and Buell 2004;Sulkin 2005;Green-Pedersen and Mortensen 2010).
In addition to their historical reputations, constituencies and public opinion, interparty competition and the electoral context more broadly drives issue appeals. Studies of the party system agenda indicate that incumbents generally respond to prominent opposition critiques (e.g. Green-Pedersen and Mortensen 2010). Likewise, new party challengers with distinct issue-focused campaigns can decrease support for older parties (Spoon 2011;. New parties' electoral threat and mainstream parties' ability to legitimize new party challengers that undercut their primary competitor parties can lead parties to address new issues (Meguid 2008). Furthermore, when parties hold electorally unfavorable or ideologically distant preferences on the most important dimension of conflict they act as entrepreneurs by discussing new topics to attract electoral support (De Vries and Hobolt 2012;Hobolt and De Vries 2015). Parties with greater internal diversity also increase the breadth of issues they address (Greene and O'Brien 2016). Although voters allow parties greater leeway to change positions on certain issues (Tavits 2007), evidence suggests that emphasizing too many tangential topics may lead to electoral costs. The broader political context constrains parties' ability to choose to address core or more tangential topics (Green 2011).
Broadly, theories of issue competition propose clear reasons why parties address new issues and the effect of these appeals on voter perceptions. Parties add issues as they become salient to the public or seek to mobilize new groups with previously undiscussed topics. These theories address the competing incentives parties encounter as they select and emphasize topics, yet make few predictions about how many appeals a party can concurrently make. Few analyses directly link these perspectives to broad empirical tests predicting voter behavior from issue scope. I extend this research to consider the electoral implications of issue scope.

The Consequences of Diverse Appeals
Parties discuss issues when they expect electoral rewards for emphasizing them. Parties, therefore, should only emphasize their political strengths, assuming full information about the issues that increase support. This logic implies that parties capable of campaigning on diverse topics will appeal to a greater range of issue-motivated voters. The extent to which parties benefit from emphasizing issues depends on voters' priorities. Voters value the relative location of parties' preferences on broad ideological dimensions (de Vries and Hobolt 2012; Hobolt and De Vries 2015), but they also hold strong preferences and differences in perceived salience. Issue ownership theories argue that parties benefit from campaigning on topics voters perceive the party as competent on. From this perspective, parties likely address all topics that would provide them with even a marginal electoral gain.
Competence perceptions are less informed when parties have less relevant or current policy reputations. Consequently, opposition parties are likely able to articulate a more deliberate and crafted policy message, free from ongoing compromise and focused only on issues they identify as most important to traditional electoral constituencies and voters more broadly (Stubager and Slothuus 2013). Relatively less unencumbered by recent policy reputations or the need to square campaign messages with ongoing policy-making, opposition parties should discuss issues on which they hold a relative advantage to build the largest coalition of supporters. 3 Due to past electoral losses, opposition parties will try to benefit from drawing attention away from the issues benefitting the government (Riker 1982) or on which they hold unpopular positions (de Vries and Hobolt 2012). Voters reward parties for discussing their most important issues, as they perceive these parties as responsive, ideologically proximate, and competent on those issues (Bélanger and Meguid 2008). Furthermore, parties addressing diverse issues are more likely to hit on those that resonate most with new voter coalitions (e.g. Riker 1982). Consequently, opposition parties discussing more issues would attract a greater range of voters. As Schattschneider's (1975) boxing match intuits, parties addressing more topics mobilize more diverse constituents. This logic suggests that parties capable of including more issues for the preceding reasons will attract great electoral support. 4 This logic informs the preliminary hypothesis in which parties have little recent policy reputation; H1 Parties with greater issue diversity in their election platforms will attract a greater percentage of the vote.
The initial hypothesis focuses on the scope of issues for parties with only hypothetical policy reputations. But, the effectiveness of parties' strategies likely varies under alternate political arrangements. I argue that incumbency imposes the key constraint limiting the effectiveness of parties' policy messages. From this perspective voters hold stronger perceptions of current government parties' reputations than they do for opposition reputations (this perspective is consistent with Bawn and Somer-Topcu 2012;Fortunato and Stevenson 2013). Consequently, these perceptions limit the effectiveness of incumbents' appeals to focus voter attention on their most preferred issues. This logic leads me to argue that unlike opposition parties, incumbents must focus their policy messages on a small number of issues for their messages to be heard over the noise from their electoral context. Voters' perceptions Political Behavior (2020) 42:487-507 of incumbents' recent policy reputations and current political conditions create a noisy hurdle that incumbent messages struggle to overcome.
Foremost, parties' campaigns reflect electoral context. For example, incumbents divert attention from economic issues by campaigning on diverse topics when they expect to perform poorly (Vavreck 2009;Greene 2016a). Research on party organizational change hypothesizes that electoral losses impact parties' internal politics and strategy (Kitschelt 1989;Harmel and Janda 1994;Harmel and Tan 2003). Parties pursue alternate tactics, focusing on core or broad constituency appeals, dependent on their position in office Green 2011). Incumbent parties often deviate from their primary messages by taking up issues important to competitors to avoid future criticism (Sulkin 2005) and in response to their competitors or the broader party system agenda. This issue uptake results in incumbent parties discussing issues that overlap with their competitors or may even benefit their electoral competitors (e.g. Sigelman and Buell 2004;Green-Pedersen and Mortensen 2010). Public opinion (Ezrow 2007;Spoon 2011;Spoon and Klüver 2014;Wagner and Meyer 2014) and government status regularly influence the topics parties campaign on (Sulkin 2005;De Vries and Hobolt 2012;Schumacher et al. 2013).
I add that incumbents face vastly different incentives in how they construct their policy messages and the structure of their campaigns than opposition parties. Controlling office comes with many policy and resource benefits, but also reputational and informational liabilities. Because they organize the structures of government and consequently bare policy-making responsibility for the state of the country, incumbents have salient and ongoing activities voters use to evaluate the content of their messages (Bawn and Somer-Topcu 2012). This complements research finding that voters use substantial information about incumbents' policy activities such as the state of the economy that is not directly influenced by parties' policy messages. This effect is more unique to incumbent parties as voters use other sources of information (like their perceptions of incumbent behavior) to evaluate opposition parties' strengths (Green and Jennings 2012a, b). The disconnect between the content contained in incumbent parties' messages and voter perceptions of their campaigns and broader priorities is likely exacerbated as media reports often offer limited coverage of the priorities included in candidates' and parties' policy messages (e.g. Brandenburg 2006; Lühiste and Banducci 2016; Greene and Lühiste 2017;Merz 2017).
This logic implies that incumbency structures incentives differently than for opposition parties; government parties have tangible, temporally relevant, policy records. Their ongoing policy activities and association with government means that voters regularly receive information about the party's revealed priorities through their activities in government. Although incumbents are hypothetically free to set the content of their agendas, their policy reputations, electoral context, and even opposition critiques of their policy performance mean they must address a range of issues they would normally seek to avoid emphasizing or appear unresponsive (e.g. Green-Pedersen and Mortensen 2010). 5 Incumbent candidates must account for their strengths as well as the immediately relevant recent policies. 6 In the opposition, parties' reputations are either hypothetical or based on longer term historical experiences. To impact voter perceptions, government party strategies must account for these reputations. Incumbency induces policy reputations to take on a contemporary or recent importance with voters (see also Green and Jennings' 2012a, b discussion of government versus opposition competence perceptions).
Even if incumbents ignore their records, voters likely form opinions on incumbent performance based on media and opposition messages (Walgrave et al. 2008;Walgrave et al. 2009). Indeed, reputations for competence across issues relate. The likelihood that voters perceive a party as competent on a range of topics increases when their competence on one issue increases (Green and Jennings 2012a, b). Assuming a party possesses a positive reputation for some issues, they might narrow their focus to only those issues. 7 By emphasizing positive reputations on these select issues, parties can increase the likelihood that voters become aware of this priority and broadly foster perceptions of competence.
From this perspective, a focused policy message might mobilize voters' cognitive bias to create a competence 'halo.' 8 For potentially supportive, but issue motivated voters, appeals targeted toward their most salient issues encourages the perception that the party is competent on a range of issues. A clear policy message then provides voters with a comparison point (similar to what Bawn and Somer-Topcu 2012 label as a discount) and limits the need to campaign on all issues. Consequently, voters likely place themselves closer to those parties than their actual positions warrant on less salient issues (Tomz and van Houweling 2009;Rovny 2012Rovny , 2013Somer-Topcu 2015). This discussion implies that voter evaluations of incumbents will be more positive when presented with a focused policy message on a small issue scope.
Altogether, this logic indicates that incumbent parties profit when they focus their attention on a limited set of issues for a number of reasons. When incumbent parties' election campaigns are forced to become too diverse they attract less support than incumbents with a clear, direct message because of the broader information and perceptions voters hold about incumbents. As voters focus more on revealed behaviors and discount parties' messages, the contents of an overly diverse message likely become lost or fail to reach the intended audience. Essentially, the large amount of information about their policy behaviors and activities overpowers the incumbent Footnote 5 (continued) issues, the incumbent de-emphasizes their owned issues and may even make the negative conditions more salient. 6 The 1993 French Parti Socialiste platform stated they "managed great reforms, but also suffered failures" (p. 1). The platform continued that the PS celebrated "10 Years of Reforms" under the guidance of President Mitterand, later discussing policy successes related to the economy, social cohesion, liberty and equality, institutional reforms, and the European Parliament. The platform also addressed "dark points" explaining their failure to reduce high unemployment (p. 2) (Volkens et al. 2011). 7 Empirically, Greene (2016a) finds that parties often focus attention on quality of life issues such as Welfare and the Environment when their platform issue diversity is low. 8 In psychological terms, the halo effect explains that physical attractiveness leads respondents to rate candidates more positively on unrelated scales, such as intellect or height. Political scientists demonstrate similar bias in evaluations of candidate attractiveness to competence evaluations (Verhulst et al. 2010). party's diverse, but weak appeals. Further, incumbents that distribute attention across too many issues (even if they are past achievements) risk being labeled as uninterested in any individual policy as the limited salience signals the party's indifference. Likewise, addressing issues owned by other parties' also likely shifts voters to view opposition parties more favorably (e.g. Green-Pedersen and Mortensen 2010). Finally, an overly diverse message might encourage the perception that incumbents only opportunistically seek to maintain the status quo with no clear policy interests. Ultimately, an inability to reach distinct voter groups, an unclear or distracted message, and evidence of only a weak policy commitment are unlikely to motivate many issue focused voters to turnout.
Instead, a more cohesive, less diverse platform will indicate a clear and forceful policy message to overcome voters' discounts of the incumbent and signals government competence more broadly. A narrow policy message allows incumbents to overcome past policy reputations by focusing directly on future reforms rather than a range of successes and failures (that might also remind voters of other parties' strengths). Proposed reforms will be clearer and their importance articulated to a wider audience. Ultimately, if voters discount government parties' statements because of their policy records (Bawn and Somer-Topcu 2012), incumbents will need to develop a clearer, cohesive message to attract the most voter support. In the second hypothesis, I propose that incumbency status moderates the effect of issue diversity on electoral success.

H2
Incumbency moderates the effect of issue diversity. Greater issue diversity increases voter support for opposition parties, but decreases support for incumbent government parties.
In summary, the scope of issues in parties' campaigns predicts party electoral success. Building on past studies of issue competition, I argue that as opposition parties increase issue scope they attract greater electoral support. The logic follows that their more abstract policy reputations and fewer competing narratives about their policy priorities allow these parties to create more purposeful election campaigns. Incumbent parties, however, benefit from an alternate approach. Including a smaller number of issues enables them to overcome the noise generated by their broader policy reputations and information voters have about the governments' policy priorities from the electoral context. Essentially, a more focused message allows incumbents to overcome voters' discounts of their policy statements. Voters punish incumbents when they are weighed down by the details and compromises of their mundane policy records and incumbents fail to 'stay on message.'

Data and Methods
I examine these hypotheses using aggregate level data on party competition. The primary analysis predicts the percentage of votes for parties using the ParlGov (Döring and Manow 2012) dataset. I focus this analysis on 25 OECD countries in the post-war era. 9 This results in 1854 observations that include 274 parties in 327 elections. As the main test of the proposed effect of issue diversity on election outcomes, I start with a broad test of the theory's validity using the Comparative Manifestos Project (CMP) (version 2015a). This design parallels a broad range of research on the effects of campaign strategies on election outcomes (e.g. Tavits (2016a), I combine the naturally opposing pairs in the CMP to create issue level categories. The resulting 42 categories make up the base issues used to construct the ENMI. 11 Alternate approaches to constructing ENMI using the Herfindahl Index, limiting the index to only economic policies, or collapsing all economic issues into a single category leads to substantively similar results. Like previous indices of issue diversity, 12 this measure includes information about the number of total issues discussed in the manifesto and the relative emphasis each party receives. I present summary statistics in Table 1. 13 At face value, the ENMI measure seems captures a number of important dynamics going on within and across parties. Foremost, estimates of ENMI summarize the parties' issue scope. For example, Fig. 1 illustrates the values of ENMI for the major parties in the United Kingdom. Generally, both parties had higher ENMI in the 1980s, although the parties' platforms seem to change frequently between elections. ENMI drops again in the early 2000s. In both cases, it seems that important changes to party leadership led to fairly large shifts in ENMI such as Blair's selection as Labour leader in 1997 or Cameron's rise to leadership for the Conservatives 9 See the Appendix for sample countries and years. 10 See Boydstun et al. (2014) for a discussion of diversity's importance. 11 I also remove all text that cannot be coded so that the indicator refers to the diversity of policy statements irrespective of non-policy goals. 12 Similar measures have been used in research on political communication and policy agendas (Jennings et al. 2011;Boydstun et al. 2014). For greater discussion on the construction and validation of the ENMI measure, see Greene (2016a). 13 There is little theoretical expectation that parties' left-right positions closely relate to their ENMI. Descriptively, I find that ENMI weakly correlates (− 0.1743) with the CMP's left-right scale (RILE). Increased ENMI is associated with more leftward positions. Opposition party ENMI are much more closely associated with their position (− 0.2266) than incumbents (0.0399), although neither correlation is particularly strong. Intriguingly, there appears to be a stronger non-linear relationship. More extreme positions associate with less diverse (lower ENMI) platforms (RILE squared has a correlation of − 0.372). in 2005. On average, parties in the sample include 16.6 issues in their platforms. ENMI ranges from 1.7 (the ultra-Orthodox religious Israeli Shahs party in 1999) to 31.5 (the Gaullist party in 1962). These indicators match perceptions of these parties' campaigns. In particular, Hazan and Diskin (2000) note that Shas ran "an anti-law-and-order theme" in response to the recent arrest of their leader (p. 632). Linking most issues to the broader international context, the extremely diverse 1962 Gaullist manifesto, on the other hand, included brief discussion on a range of issues such as the recent constitutional reforms in France, the then finished war in Algeria, broader European politics and economics, and even the Soviet Union. More generally, ENMI increases over the course of the sample for the majority of parties,  (Meguid 2008). Mainstream parties tend to have more issues than smaller, niche parties such as green and radical right parties.
To test the second hypothesis, I include an interaction of ENMI with the party's government status. I use an indicator variable equal to one when the party is in the last, non-caretaker government prior to the election as measured by ParlGov (Döring and Manow 2012). I expect that the effect of ENMI will be positive for opposition parties and negative for incumbents.
In the main analysis, I include control variables to account for parties' electoral strategies. In particular, I create a dummy variable labelled Moderated Position to account for relative shifts in parties' positions using the logged RILE scale from the CMP (Adams and Somer-Topcu 2009;Somer-Topcu 2009;Lowe et al. 2011). The dummy variable equals one when the party shifted its position closer to the mean position of all parties in an election. This variable is interacted with the incumbency variable to account for different strategies for incumbent and opposition parties (e.g. Bawn and Somer-Topcu 2012). 14 In addition to parties' strategic maneuvering, I also account for economic voting arguments by including the GDP growth rate from the OECD and its interaction with government incumbency (Lewis-Beck 2006). I multiply the incumbency variable with economic growth to distinguish the effects of the economy for incumbent and opposition parties. Furthermore, I use a dummy variable to distinguish majoritarian electoral systems (Somer-Topcu 2009). 15 To test the primary hypotheses, I present fixed effects models with a lagged dependent variable in the main analysis to account for heterogeneity caused by the inclusion of the same parties in multiple elections. Furthermore, past studies of issue diversity and issue salience indicate there is a complex relationship between economic conditions, incumbency and issue diversity (Vavreck 2009;Hellwig 2012;Greene 2016a;Williams et al. 2016). To account for these arguments, I perform an additional two stage least squares analyses treating ENMI as endogenous to the electoral environment (see Online Appendix).

Analysis
In this section, I describe the results from a fixed effects model predicting the percentage vote for parties accounting for the party and number of elections that the party competed in. In Table 2, I present an analysis of the full sample as well as subsample analyses for incumbent and opposition parties to reduce the chance that negative correlations drive the results. The results from the analysis provide evidence consistent with a story in which incumbent ENMI influences' the party's electoral success.
The first hypothesis (H1) predicts that parties with higher ENMI attract broader electoral support. Consistent with the hypothesis, the coefficients for ENMI in the opposition party sample (Model 1) and the full sample (Model 3) in Table 2 are positive. However, the coefficients fail to reach standard levels of statistical significance. Although the coefficients are often in the correct direction, the effect of ENMI for opposition parties is inconsistent across both the primary and secondary analyses and should therefore be interpreted as inconsistent with H1.
However, the second hypothesis predicts a more complicated relationship. In particular, I hypothesize that incumbent ENMI has the inverse effect of opposition Table 2 Fixed effects OLS with ENMI from the CMP The dependent variable is percentage vote for the party. Models include fixed effects to account for country and election levels. Standard errors are robust. ENMI is constructed using Shannon's H from the CMP Significance tests are two-tailed: + p < 0.10; *p < 0.05; **p < 0.01; ***p < 0.001 parties' ENMI. The subsample analysis for incumbent parties (Model 2) and the interaction of incumbency status with ENMI tests this hypothesis (H2) in Model 3. The negative and statistically significant coefficients for incumbent ENMI show evidence consistent with the hypothesis in both models. Government parties with higher ENMI attract a smaller percentage of the vote than government parties with fewer issues based on these results. 16 I present the predicted effect of ENMI on the percentage votes that parties receive for opposition and government parties in Fig. 2 (from Model 3). Although it appears that as opposition parties discuss more issues they attract greater electoral support, the effect is not statistically significant from zero. The effect reverses for incumbent government parties, as predicted in the second hypothesis (H2) and is statistically different from zero at the 0.1 level and jointly different from the independent effect of ENMI at the 0.05 level. 17 An increase of one standard deviation (5.9) in ENMI

Robustness Checks and Sensitivity Analysis
The primary results evidence a consistent relationship between issue diversity and electoral support. Yet, critical scholars may argue that these results are driven by the operationalization of the independent variable, model specification, case selection, or alternative processes. Therefore, I present results of extensive robustness checks in Online Appendix accounting for differing logics that show evidence largely consistent with the main results. These models account for a range of model specifications to explore the results' validity.
In particular, I use a measure of issue diversity, ENMI, derived using Shannon's H to measure issue entropy using the CMP. This measure incorporates information about small categories that a measure of diversity from the Herfindahl index would devalue. Results from this approach are robust to the operationalization of issue diversity using an alternate index. Likewise, I construct measures of ENMI from the CMP that include only the economy categories and treat all economic categories as a single issue. Analyses including these operationalizations lead to substantively similar results. Transforming ENMI using the natural log (+ 0.5) confirms the effect of incumbent ENMI as well.
Alternatively, scholars have noted numerous challenges that arise from the CMP's coding scheme. To ensure that the results are not driven by solely by coding choices made by the CMP, I create a measure derived from the Chapel Hill Expert Surveys (CHES; Steenbergen and Marks 2007;Hooghe et al. 2010;Bakker et al. 2015). The availability of expert survey data massively reduces the sample size and comes with alternate assumptions about the resulting measures of issue priorities. For example, experts likely use cues such as parties' position in government to evaluate their issue attention (Adams et al. 2014;Lindstädt et al. 2015). Yet, the results using the CHES support the second hypothesis (the sub-sample for incumbents does not reach standard levels of significance, however, when controls are included. Although both measures lead to supportive results, the actual measures only weakly correlate (Pearson Coefficient 0.079).
Other modeling choices such as the exact time series specification, the assumption of a normally distributed dependent variable or the use of fixed effects may also lead to alternate results. However, I find that models accounting more extensively for the time series nature of the data (McDonald and Best 2006), the truncated dependent variable (0 and 100), and models including a range of controls with varying intercepts reveal additional evidence consistent with the second hypothesis. Consistent with both hypotheses, Error Correction and Equilibrium Response Models both find a short and long term negative effect of incumbent ENMI as well as a positive short and long term effect for the opposition's ENMI. Estimates from a Tobit model with random effects that accounts for the truncation of observations with low vote totals also provides similar evidence for the theory.
Although the inclusion of fixed effects for country and election levels in the main analysis account for a number of alternative sources of heterogeneity, they also limit the ability to include measures that differ only at the country or electoral level. Instead a mixed effects model with varying intercepts at the country level enables the analysis to include these controls. In particular, I find that mixed effects models that allow for additional controls such as niche party status, the effective number of parties, federalism, a directly elected president, prime minister's party, cabinet party, and coalition governance lead to similar results.
Furthermore, alternate explanations of incumbent parties' electoral success such as the cost of ruling (e.g. Fortunato and Adams 2015; Wlezien 2017) or the party system agenda (e.g. Green-Pedersen and Mortensen 2010) may question the strength of the main results. However, controls for coalition participation, the prime minister's party, the effective number of parties, the number of coalition parties, if the party has never been in government before (van Heck 2016), time in office, the opposition's ENMI and the party system ENMI, and even the years in the opposition or government show a robust effect of incumbent diversity (H2) in the full models. Other controls accounting for the electoral context by including measures of electoral volatility to account for voters' willingness to change parties (e.g. their "availability", Bartolini 1999), real levels of immigration, or unemployment yield results comparable to those presented in the main analysis. These results imply that the effect of incumbent ENMI is not solely driven by the costs of ruling, the electoral context, or the party system agenda.
Furthermore, the theory may be driven by the differences between small and larger parties if size also correlates with ENMI. However, the results suggest this is not the case. Excluding parties that did not receive at least one, five or ten percent of the vote in the previous election (similar to Somer-Topcu 2015) shows consistent evidence for the second hypothesis.
Finally, the results may be driven by endogenous processes that drive both the party system agenda and opposition criticisms such as the economy (e.g. Green-Pedersen and Mortensen 2010). Results from a Two Stage Least Squares Model accounting for the potential of endogeneity to the economic context alone lead to substantively similar conclusions. Two Stage Models that also include measures of the largest opposition party's ENMI, the party system agenda, and the incumbent's overlap with the party system agenda illustrate a consistent effect for incumbent ENMI. Likewise, fixed effects models that include interactions of GDP growth with ENMI and incumbency lead to similar results. In the full sample, Wald tests show that the coefficients for ENMI, its interaction with incumbency and GDP growth and the four-way interaction with the GDP growth just barely fail to reach statistical significance at standard levels, although the effect for incumbents is statistically different from the effect for opposition parties. None of the coefficients for interactions including GDP growth and ENMI are statistically significant. Like the 2SLS results, the effect of ENMI is independent from economic conditions and the broader party system agenda.
Altogether, the main results illustrate consistent support for the incumbent hypothesis. These additional models further demonstrate a robust effect of incumbent issue diversity from a range of modeling choices. Incumbent ENMI has a robust negative correlation with the votes these parties receive.

Conclusions
Scholars have long theorized that issues play important roles during parties' election campaigns and impact parties' behavior in government. However, they have yet to evaluate whether the scope of parties' election campaigns matters. Do voters punish or reward parties for discussing a broad range of topics? The results from this analysis provide a clear answer to this question using data from elections in 25 OECD countries over five decades. 20 The results show strong support for the hypothesis that voters punish incumbents for overly disperse campaign messages, although there is little evidence for opposition parties' issue scope. This result holds under a wide range of modeling choices. As the French Parti Socialiste found in the early 1990s, an overly disperse message can lead to electoral failure.
These findings provide support for a theory of issue competition that complements Schattschneider's scope of competition or Riker's natural selection approach. Electoral losers benefit from reaching out to broader groups, but government parties struggle with their realized policy reputations. Economic conditions and government participation may shape parties' strategies (Vavreck 2009;Green-Pedersen and Mortensen 2010;Hellwig 2012;Greene 2016a;Williams et al. 2016), yet these results suggest that an independent effect of parties' messages remains. Further evidence is necessary to determine the extent to which additional aspects of the electoral context such as the media or longer term policy reputations condition voters' perceptions on the scope of parties' messages.
The evidence in this paper indicates that issue diversity should be taken seriously as an important characteristic of election campaigns. Popular media discourse often discusses campaigns as if they are too dominated by a single issue or if they are too disperse. Issue diversity likely matters for a range of political outcomes (e.g. Jennings et al. 2011;Boydstun et al. 2014;Greene 2016b). Incorporating research on issue diversity provides a means of characterizing political campaigns that may result in new approaches to the study of elections. Indeed, if parties' time in office determines the diversity of their platforms, it is not only the increasing distance from the median voter or coalition position that impose the well-established costs of ruling (e.g. Fortunato and Adams 2015;Wlezien 2017), but also the clarity and diversity of the message itself.
Additional consideration of issue scope will benefit broad theories of democracy. The quality of democratic competition and citizens' satisfaction with democracy (e.g. Reher 2015Reher , 2016 likely depends on the extent to which parties campaign on more or less diverse platforms. If all parties campaign on too few issues, voters' ability to select between platforms offering competing solutions might disappear (e.g. Sigelman and Buell 2004). Contexts where parties distribute their attention too widely could suggest that parties provide only limited attention to the most important issues. In this case, deciding which party best represents a citizen's priorities would be difficult. The scope of parties' campaigns, therefore, might provide an approach to consider the breadth and focus of representation in modern democracies.