Shifting Positions: Party Positions and Political Manifestos in Costa Rica

ABSTRACT This article analyzes how niche parties may utilize a strategy of policy shifting to garner additional voters. It leverages a unique opportunity in which a Costa Rican political party released two different versions of its party manifesto at different moments during a single election cycle. This rare opportunity uncovers how the party shifted from having a hard conservative stance on social issues, such as abortion, to moderating its stance and centering its focus on less contentious issues in a runoff election campaign. Understanding how a single political party may alter its strategy is important because it allows us to better gauge the effectiveness of shifting policy positions, especially for niche parties, for which a particular issue area is dominant. Moreover, this analysis opens additional avenues of research on political parties in the Latin American context, since research utilizing manifesto data in this context has been limited.

or regionalist parties that dominate particular issue domains. Even then, though, the concept of niche parties is widely applicable. For example, in Latin American systems we can find neo-Pentecostal or Evangelical parties, indigenous parties, and others that arise from the activities of social movements or other activist groups. This article aims to encourage future research on niche parties in the Latin American context, and it therefore seeks to expand knowledge of niche parties and their electoral strategies by considering a Latin American example: Costa Rica.
This research examines how niche political parties alter their policy positions not only between elections, as existing literature has shown, but also within a single election cycle. It does this by exploring the shift in the information that a niche party conveys to voters via its manifesto to attract a wider number of voters at different moments of an electoral process. For this we analyzed party policy positions in their manifestos during the first and second rounds of a single runoff election. The changed manifestos provide a clear view of how a party alters its strategy in the midst of a campaign. This is an improvement over previous research, which has often relied on analyzing policy shifts from prospective or retrospective electoral outcomes and not shifts that may occur within one election cycle (Meguid 2005;Adams et al. 2006;Adams and Somer-Topcu 2009;Rovny 2012;Meyer and Wagner 2013;Bräuninger and Giger 2018).
We examine this phenomenon by focusing on the case of the Costa Rican National Restoration Party (Partido Restauración Nacional, PRN) in the 2018 national election. We take advantage of the unusual occurrence that this party released two versions of its manifesto at different moments of the election cycle.
Previous literature, primarily in the European context, has found mixed results as to how voters may adjust their preferences when parties choose to shift. Some scholars have found that voters are not likely to adjust their perceptions of parties' positions in response to shifts in parties' policy statements (Adams et al. 2011), while others find that party election statements do have an influence on voters' perceptions of party positions (Fernández-Vázquez 2014).
From the mid-1960s until 1994, Costa Rica had a stable political party system, with the number of effective parties hovering around two. However, the party system in Costa Rica began to drastically change, and by the 2002 election it had shifted from the typical two-party system to one of multiple competitive parties. This article analyzes how political parties adapt their strategies in a newly competitive electoral environment to target different sectors of the electorate.
Downsian theory would predict that the parties would moderate their stances, but other work analyzing party positions has found that parties often have an incentive to take up extreme positions in order to show issue ownership and to allow policy differentiation between parties (Kitschelt 1994;Spoon 2009;Wagner 2012). Yet we know from recent work that parties can also blur messages or present ambiguous positions in order to attract votes from different voters who typically would not vote for the party (Rovny 2012;Bräuninger and Giger 2018). This article aims to explore how parties can utilize both extreme positioning and blurring or downplaying an issue position to garner more votes.
We analyze the direction and degree of the shift, finding in favor of the Downsian view that the PRN moderated its conservative stance on specific party-owned policies before the runoff election as a strategic attempt to attract voters from the parties that were left out of the runoff. Until the runoff, the PRN ran on salient conservative issue areas, such as abortion and gay marriage. However, the party made a clear switch after getting to the second round, when it downplayed the social issues so as to broaden its voter base.
This example suggests that while niche political parties in multiparty systems with a two-round election have an incentive to take extreme stances in salient issue areas to differentiate themselves in the first round of an election, after the field is reduced to fewer competitors, political parties will moderate their position to garner more votes.
To explore how the PRN chose to navigate the political environment in Costa Rica, we use party manifesto data taken from the 2018 presidential election, specifically the Base de Datos de Programas de Gobierno by the Programa Estado de la Nación en Desarrollo Humano Sostenible (2018a). The use of these data is a unique research endeavor, since, to the best of our knowledge, these data have not been used in academic research studies for peer-reviewed journals. The database has, however, previously been analyzed in a report prepared for the Programa Estado de la Nación (Gómez Campos and Chavarría Mora 2018). The examination of party manifesto data in the Costa Rican context is important to explore, since it provides a clear picture of how a party wants its image to appear to the electorate.
The use of manifesto data has been thoroughly examined in the European context in cross-national studies using the Comparative Manifestos Project (Meyer and Wagner 2014;Spoon and Klüver 2014). Yet research exploring party manifesto data in Latin America has been scant, due to problems of data availability. A further contrast to the typically studied European cases of niche party behavior is that this case is centered in the Latin American context, where there are vast differences in party systems, electoral systems, executive-legislative relations, and cultural references. We believe that understanding party behavior through exploring how manifestos can change will provide a better comprehension of how strategies such as moderating may function in a newly developed multiparty system such as that of Costa Rica.

PRECEDING LITERATURE ON NICHE PARTIES
While a cursory look over Downs's 1957 spatial competition argument tends to see it as leading to a median voter theorem equilibrium between two parties, we would like to emphasize that the expectation of convergence in the Downsian model rests on a series of assumptions. Indeed, Downs proposes that when voters' preferences are not single-peaked, new political parties will emerge around peaks, where current political parties are not meeting the needs of the electorate. Furthermore, as Grofman (2004) explains, Downs's model relies on about a dozen assumptions necessary for the convergence on the median voter, and violation of a single assumption leads to nonconvergence. For the purpose of our theory, the most relevant assumptions are i) competition between two parties, ii) a single round of elections, iii) one dimension of competition, and iv) single-peaked voter preferences. We will not discuss assumptions that we accept and that do not change between the two rounds of the election, including those related to the rationality of actors or beliefs about their goals, such as the office-seeking assumption for parties.
In this case, we have a two-round electoral system (with two finishers advancing) plus a large number of competitors in the first round. It is clear that this violates assumptions 1 and 2 and should not lead to convergence, but it is also worth noting that assumption 1, the number of parties, is fairly relevant for considering variation between presidential systems. Following the so-called Duverger hypothesis (Duverger 1954;Riker 1982), we would expect presidentialism with a runoff to lead to a multiparty system, a result more similar to the proportional representation electoral system common in parliamentary systems.
On the issue of a single dimension of competition and single-peaked preferences for voters, scholars have noted that parties can rarely be subsumed into the left-right scale, since parties can take positions on various issue and policy dimensions, and in multiparty systems it is essential for newer parties to create spaces to position themselves uniquely among different issues.
Scholarship has sought to address the issue of multidimensional competition in several different ways. One alternative theoretical model is salience theory. Instead of competing by positioning themselves against each other in the same dimension, salience theory suggests that parties emphasize issues that are important to them or will highlight issues that they believe will increase electoral support (Budge and Farley 1983;Meguid 2005).
Relatedly, scholars have also focused on the presence of postmaterialist values and new politics, which, in turn, leads to the emergence of niche parties. Such parties attract voters by concentrating on a narrow range of issues that larger political parties often neglect, and this development has helped to better explain the multidimensional aspect of party competition (Kitschelt 1988;Inglehart and Welzel 2005;Spoon 2011).
The role of public opinion in formulating new theories that analyze party strategy has also been incorporated. Previous literature studying policy shifting has argued that citizens vote prospectively; that is, that voters would examine what politicians are currently proposing. However, some studies have found that political parties and their candidates will adjust their policy promises in the current election in response to the outcome of a previous election (Budge 1994;Fowler 2005;Somer-Topcu 2009).
The lack of research on niche parties in the Latin American context has stemmed from several factors. Programmatic competition, a prerequisite of spatial and issue competition, historically was common in only some countries in the region, and even then, until the 1990s, only one particular ideological dimension seemed relevant for competition: an economic one, focused on property rights and redistribution. Progressive and conservative cultural issues and religion existed as a separate dimension, as did a dimension focused on regime type; however, these did not seem to structure party competition. Ultimately, parties divided around the economic dimension (Kitschelt et al. 2010).
The appearance of niche parties in Latin America logically expanded the number of potential issues on which competition could develop, while forming around cleavages-some common, some very different from the ones traditionally found in Western European countries. As Kernecker and Wagner (2019) document, we can see the emergence of niche parties that focus either on what they refer to as postmaterialist issues or on traditionalist topics.
For some examples of the postmaterialist type, we can think of the emergence of Green parties in Latin American countries, such as Mexico, Colombia, and Brazil, similar to their European counterparts, or of the different ethnic political parties that have emerged out of the unique circumstance of Latin America's historical colonial legacy-extremely different from the context of ethnic or regionalist parties in Europe, Africa, or Southeast Asia. Examples include the Ecuadorian party Pachakutik and arguably Bolivia's Movement Toward Socialism (MAS). On the other hand, the increased importance of religious parties, particularly of the neo-Pentecostal type, is an example of the traditionalist category of the niche party, which focuses on traditional views on women's rights, same-sex marriage, and similar issue areas.

PARTY STRATEGY AND PARTY SHIFTS
Given the fact that Latin American countries have been dominated by a presidential form of government and weakly institutionalized party systems, theories of party strategy stemming from the European context are difficult to apply and assess in terms of party strategies in the Latin American context. Nevertheless, scholarship that has focused on niche parties in Latin America has found that despite being defined on the basis of issue-based characteristics, they are less programmatic in their linkage strategies than mainstream parties. Instead, niche parties in the Latin American context are slightly more likely to draw on charismatic strategies and are more likely to establish strong links to ethnic and religious organizations (Kernecker and Wagner 2019).
This research project allows for a deep understanding of how niche parties respond to particular electoral strategies, especially given that the study looks at two different time periods in a single election. If a niche party is practicing programmatic linkage strategies, then we will see that the party is repeatedly and consistently taking singular stances on issues that do not contradict one another (Kernecker and Wagner 2019). On the other hand, we may see that niche parties rely on the attractiveness or salience of their primary policy stances to strategize and garner more votes from the broader electorate (Meguid 2005).
Scholars studying issue ownership have examined how strategies such as strategic ambiguity can succeed by blurring policy platforms and messages in order to attract voters who would not otherwise vote for a party with a clear-cut policy position (Bräuninger and Giger 2018). The evolving literature on strategic ambiguity has shown that parties can selectively target clear-cut messages toward specific groups but also take ambiguous positions to attract other voters from different groups in order to help win votes on both sides (Rovny 2012;Somer-Topcu 2015). In addition to strategic ambiguity, scholars have analyzed how parties can choose to selectively emphasize certain issue areas and downplay others (Klüver and Sagarzazu 2016). Theories that examine issue ownership claim that political parties will choose to emphasize policy issues in which they are considered to be proficient and on which they can form a competitive platform (Petrocik 1996).
In an attempt to bridge different theories of party competition, especially spatial and salience theories, Elias et al. (2015) propose four potential strategies that can be used to compete in two-dimensional spaces: unidimensionality, blurring, subsuming, and two-dimensionality. Unidimensionality consists in focusing on a single, main dimension of competition while ignoring the second dimension. The blurring strategy maintains the focus on the main dimension of competition, but the party adopts ambiguous positions on the second dimension. Subsuming refers to framing the issues of the second dimension with regard to the issues of the main dimension. In two-dimensionality, the two dimensions of competition are treated as independent.
These studies altogether have important implications for understanding the strategies utilized by political parties and enable scholars to link how different types of political parties, such as niche parties, differ from their mainstream competitors; yet more research is warranted in the Latin American context to better understand how these theories operate outside the European framework.

THEORY
Previous research has studied the shifts in party strategy between one election and the following one. A key difference in this article is that we are looking at the shift in a single election, between two rounds of it. In short, while other studies have focused on the aspect of elections as a repeated game with a single set of conditions, this study looks at a single game with two stages, a single election with two different sets of conditions, one for the first round and one for the second. By capturing these two moments, we can look at how parties move from one strategic behavior to another.
We first presume, following the literature, that in the first round of a multiparty election, a niche party would focus on making salient the issues it "owns" and would attempt to differentiate itself to attract voters supportive of those issues. As it is a tworound election, in the first stage the competition is multiparty. Since there are larger, more established moderate parties at the center of the ideological spectrum, as well as a myriad of other smaller niche parties competing for the extremist voters, the best strategy for the niche party at this stage is to position itself at the extreme end of the relevant ideological dimension as a credible defender of the issue it owns. We also expect it to ignore any other dimension of competition, including the economic one.
In the second stage, the runoff, the number of players is reduced to two. This has two effects. First, there is now a huge number of moderate voters at the ideological center to be captured, as they have been left without their first party of choice. Second, the extreme voters the niche party captured in the first stage cannot credibly hope to defect to the other party, as its position on the issue owned by the niche party cannot be closer to the preferences of those extremist voters.
Given the change in circumstances in the second stage, we theorize a change in the niche party's strategy. The party will now focus on making salient a dimension that is relevant to the voters who supported the parties that lost in the first round, to attract those voters, while at the same time blurring its stance on the dimension that includes the issues it owns. This is done to avoid scaring off potential voters who might perceive the party as being too extremist. Furthermore, because there are now just two parties competing in that new salient dimension, we expect the niche party to moderate its platform toward the center and try to position itself with the median voter to obtain a majority of votes in the second round.
Based on the previously specified conditions for each stage of the election and the expected strategy we have delineated, we propose the two following hypotheses: H1. Moderation on salience: Niche parties will increase the salience of the mainstream issues and decrease the salience of their own issues in the second stage of the election.
H2. Moderation on spatial competition: Niche parties will move closer to the median voter in both the mainstream and their own issues in the second stage of the election.

THE COSTA RICAN ELECTORAL AND PARTY SYSTEM
The differences between the two stages of an election apply in the Costa Rican case. Costa Rica has simultaneous presidential and legislative elections, allowing for split ticket voting. The presidential election uses a two-round system in which a candidate must have over 40 percent of the vote to win in the first round.
Historically, the Costa Rican party system has taken a two-party configuration, despite proportional representation for the legislative election, two rounds for the presidential one, and the allocation formula, all of which would suggest a multiparty system (Duverger 1954;Riker 1982). The configuration follows the results of the Civil War of 1948; the different political actors that played a role in the conflict slowly gravitated into two groups, first as coalitions and ultimately as two parties, the Social Christian Union Party (Partido Unidad Social Cristiana, PUSC) and the National Liberation Party (Partido Liberación Nacional, PLN).
These two factions dominated the country's political arena between 1966 and 1994, with the effective number of electoral parties (ENEP) remaining stable between 1.99 and 2.23, save for one election (Calculated with data from Tribunal Supremo de Elecciones 2014). Because of the electoral control of these parties, there was, historically, little space for other parties to achieve electoral success. Between the 1990s until the mid-2000s, party competition seemed to fit comfortably into a single, left-right dimension that focused on economic-redistributive issues, with little competition happening in other dimensions (Kitschelt et al. 2010) However, factors such as programmatic convergence of the PLN and PUSC, as well as corruption scandals involving former national presidents from both parties, led to the 2002 election ending the two-party system. The shift toward a multiparty system has been described as a realignment favoring a new party (Carreras et al. 2015), and in more detail first, in 2002, as an abrupt realignment into a new equilibrium, followed in 2006 and subsequent elections as a different, gradual realignment over time (Perelló and Navia 2021). Regardless of the more specific details of the realignment process, it is clear that the big winner out of all the new parties was the Citizens' Action Party (Partido Acción Ciudadana, PAC), which, since the first realignment, managed to capture the presidency twice.
The increase in support for new parties, however, has developed together with an unprecedented amount of electoral volatility (Sánchez 2002), leading practically unknown parties to rise and gain a sizable portion of the vote share in one election and then to disappear in the next one. Possibly the most dramatic increase yet was that of the PRN, an Evangelical (specifically, neo-Pentecostal) party that had never obtained more than 2 percent of the vote before, yet in 2018 ended up on top in the first electoral round with 24.99 percent of all valid votes. The PRN illustrates our theory. To better understand how the strategy was pursued, we now turn to the history of the party and its actions during the campaign.

NATIONAL RESTORATION AND THE 2018 ELECTIONS
The 2018 election in Costa Rica has been widely considered atypical (Rojas and Treminio 2019; Alfaro and Alpízar 2020), due in no small part to the surprising rise in popularity that the PRN experienced during the process. Founded in 2005 by Carlos Avendaño Calvo and composed mostly of defectors from other Costa Rican Evangelical parties, the PRN primarily sought voters from the Evangelical community and voters who prized a conservative view of politics, which has earned it classification as a populist radical right party (Pignataro and Treminio 2019;Treminio and Pignataro 2021). This classification, of course, does not preclude considering the PRN as a niche party; indeed, it can be seen as a party whose niche is precisely a subsection of the radical right.
The PRN participated in a presidential election for the first time in 2014, at which time its presidential candidate received only 1.35 percent of the vote and the party elected only one deputy to the Legislative Assembly. In stark contrast, four years later the PRN had gained monumental strength in Costa Rican politics and went on to win 14 parliamentary seats and to place its candidate, Fabricio Alvarado, in the presidential runoff.
The success of the PRN can be understood only in the context of the ruling of the Inter-American Court of Human Rights with regard to same-sex marriage in Costa Rica. On January 9, 2018, in the middle of the political campaign, the Inter-American Court of Human Rights ruled, in response to a consultation from the Costa Rican government, that the law forbidding same-sex marriage was unconstitutional. While most candidates avoided taking a position, four politicians took a stance. Of these, Fabricio Alvarado of the PRN was the only one who clearly opposed the policy, going as far as proposing that Costa Rica should leave the jurisdiction of the court.
According to many scholars and political journalists, Alvarado's popularity surged on the campaign trail, due primarily to his hard stance and opposition to same-sex marriage. For example, Pignataro and Treminio (2019) divide the 2018 electoral campaign into three stages, the second stage dominated by the polarization surrounding the same-sex marriage ruling, while Perelló and Navia (2021) attribute the increase in support for the PRN's candidate to the ruling. An important difference in support can be observed when comparing other religious parties that did not take a stance: Costa Rican Renovation (Renovación Costarricense) and the Catholic party Christian Democrat Alliance (Alianza Demócrata Cristiana) ended up with less than 1 percent of the vote share in the first round, whereas the PRN obtained 24.99 percent.
During the presidential race, the PRN was the third party to release its manifesto, which it did on October 27, 2017. After five months, and just four days before the runoff election, on March 28, 2018 the party presented a second manifesto. According to this manifesto, the changes were orchestrated to reflect the influence of members of other parties who had given their support to the PRN in the runoff (PRN 2018, 4). Parties do not usually present two manifestos in an election, but this case presents a unique opportunity to compare the changes in communication strategy between the first and second rounds.
The parties that placed third and fourth in the first round of the election were the ones that had dominated politics during the two-party system era: the center-leftturned-center PLN (18.63 percent of the vote) and the center-right PUSC (15.99 percent). Both parties campaigned on the expertise and experience of their economic teams; this, alongside the growing fear that a looming fiscal crisis might befall the country, meant that the almost 35 percent of voters who supported these two parties were primarily concerned with economic issues. If the PRN wished to attract these voters, it would have to go beyond being the single-issue party that it was negatively portrayed as by all its rivals during the first part of the campaign, and it would have to build an image as a party with serious economic plans for the country.
Following our theory, we posit that since the PRN is a party that owns the issues of social conservatism, in the first stage of the election it will give salience to these issues. However, once the party advances to the second stage of the election, it will moderate by downplaying the salience of social issues and increasing the salience of the more mainstream issues, economic ones, to gain broader support. We also posit the hypothesis of moderation on spatial competition. During the second stage, the PRN will move closer to the middle position on both economic and social issues, following a standard median voter argument, which predicts that parties will try to move toward the middle range of their potential voters to avoid appearing extremist.

DATA AND METHODS
This study presents a quantitative and qualitative comparison between the two manifestos released by the PRN for the 2018 election in Costa Rica. The first manifesto was released before the first round of the election, and the second manifesto was published between the first round and the runoff. We begin by explaining our logic for focusing on manifestos and detail how the data were coded for the quantitative analysis. Following this is a brief description of the two manifestos by their thematic sections. Then we present the dependent variables to be used and the method for testing the difference between the manifestos.
Based on the timing of their release, we see the changes between manifestos as signals to the electorate of the party's commitment to improving its program. Before the runoff, the PRN was criticized as a single-issue party with no agenda outside of defending traditional family values, and therefore not fit to rule. A manifesto that was seen as serious and with clear policy proposals would signal to the electorate that the party had a plan for what to do if it were elected. The candidates considered the manifestos relevant enough to discuss during televised debates: on March 22, during a debate on one of the two most important channels in the country, the PAC and PRN candidates argued over the quality of their manifestos, accusing each other of not having a clear plan on what to do in case they won the election. 1 There was also a clear interest from the citizenry in the quality of the manifestos. The online reports analyzing the manifestos by the Programa Estado de la Nación were visited about four hundred thousand times during the campaign (Programa Estado de la Nación en Desarrollo Humano Sostenible 2018b). Furthermore, the candidates themselves also mentioned these analyses during the debates, indicating at the very least that the candidates thought the manifestos mattered to citizens when deciding their vote.
The Base de Datos de Programas de Gobierno, created by the Programa Estado de la Nación en Desarrollo Humano Sostenible (2018a) is a database compiled from the party manifestos of the 13 political parties that competed in the presidential election of 2018, including the 2 party manifestos by the PRN. Most of the variables and the methodology were inspired by those used in the Manifesto Research on Political Representation (MARPOR). Each of the questions inspired by MARPOR belongs to one of eight thematic sections: freedom and democracy, political system, economy, well-being and quality of life, social fabric, social class, and foreign policy and international relations. (Programa Estado de la Nación en Desarrollo Humano Sostenible 2018b). Table 1 shows which questions fall into each of these categories. 2 With regard to the two PRN manifestos, the first was 34 pages long and the second 95 pages, once standardized into the same letter font and size. Additionally, the first manifesto presented by the party had 505 sentences, while the second version had 1,386 sentences. Figure 1 shows the salience of each of the thematic sections for both manifestos released by the PRN. The salience of each of the thematic sections corresponds to the percentage of each respective manifesto dedicated to sentences classified under that thematic section. As can be seen, in both versions, little emphasis is placed on foreign policy and international relations or on freedom and democracy. The most important section is well-being in both cases, about 20 percent of both manifestos. The focus on well-being should not come as a surprise, as the party claims to have a Christian-inspired ideology, and its policies will be a way of expanding the type of social work in favor of low-income groups that neo-Pentecostal churches are already doing in peripheral communities in Costa Rica.
The biggest changes occur in four sections: freedom and democracy, political system, economy, and social fabric. The freedom and democracy section presents a drop of 3 percentage points between the first and second manifestos, although, as we have seen, it is one of the sections with the least salience to begin with. Political system also drops between manifestos by about 2 percentage points.
Economy is the one category that increases between the first and the second round manifestos, by 4 percentage points, and, as seen in table 1, it has a broad Figure 1. Salience of the Thematic Category per Manifesto range of topics related to the economy of the country. Social fabric, which is the section related to traditional morality and multiculturalism-that is, the one that includes the issues owned by the PRN-presents the biggest drop in salience: 8 percentage points. This means that in the first round, 15 percent of the manifesto was concerned with these topics, but it dropped to 7 percent for the second round. This decrease by half could be an attempt to de-emphasize the topics. Combined with the increase of interest in economic matters, this can be an attempt at moderation.
To further test our two hypotheses, we calculated measurements for both the salience of economic and social issues in the manifestos and the ideological position of the party as it is reflected in each manifesto, and also for both the economic and the social dimension. In order to calculate the ideological positions of the party on both economic and social dimensions, we followed the formulas used by the manifesto project (Manifesto Project n.d.; Volkens et al. 2018). The scores were calculated with the following formulas, using the questions from table 1: The resulting scores move theoretically between 0 and 1, but in practice, the scale is between 0.3 and 0.7, which is to be expected; to get the score of 0 in the social dimension, a single sentence would have to include positive references to human and civil rights, multiculturalism, and internationalism, while at the same time make negative references to patriotism and traditional morality, which would be hard to do in a coherent sentence in a manifesto.
With regard to the salience of each of the thematic sections, it can be obtained by dividing the number of tokens (that is, the lexical unit of analysis, in this case, each unique phrase in the programs) in which it is mentioned by the total of tokens in the program. We used the same questions that we used to calculate the economic and social positions of the party to calculate the salience in both dimensions, to maintain comparability. Table 2 shows the results of t-tests comparing salience and ideological position on both the economic and social dimensions. Looking at salience, the difference is statistically significant for both the economic section and the social fabric section. The differences indicate an increase in the salience of economic issues and a decrease in the salience of social issues in the runoff. As for ideological positioning, the change in the economic dimension is not statistically significant. On the other hand, the change in the social dimension is significant and in the expected direction. Summarizing, between rounds one and two of the election, we see a decrease in salience of social issues, an increase in salience of economic issues, and a moderation in the social dimension.
To complement these quantitative results, we took a deeper look at specific changes between both manifesto documents. Focusing first on the texts, we see that certain phrases have been eliminated from the second manifesto presented by the PRN. For example, the following phrases appear only in the first version of the manifesto. The phrases are presented in English, as translated by the authors.
• It is in the sphere of the family where cultural change begins and, therefore, sociopolitical [change]; for this reason, the national restoration must start in the family, then to spread through society. (PRN 2017, 9) • Our political project believes in a separation between Church and State. We are a political party of Christian foundation and not religious. (PRN 2017, 12) • Christianity has been, in Western history, the basic moral point of reference for human development. (PRN 2017, 17) • Gender equality implies a restoration of the family as the nuclear axis of society, a restored family in which father and mother interact in conditions of equality. (PRN 2017, 40) • [A proposed policy is to] Eliminate the in vitro fertilization regulation in Costa Rica for being illegal and immoral and for coming from a clear violation of national sovereignty, and [the party proposes] leaving the subject in the hands of the Legislative Assembly, where it should be solved to begin with. (PRN 2017, 50) In other instances, phrases are slightly tweaked; for example, when talking about gender equality policies, the first version of the manifesto refers to some as "radical and gynocentric," while the second version eliminates the second word (PRN 2017(PRN , 40, 2018. At the same time, several new paragraphs were included in the second manifesto, touching on the economic situation of the country with regard to issues like the fiscal crisis, value added tax reform, public budget and spending control via austerity, accountability and evaluation, digital government, simplification of bureaucratic processes and the fight against tax evasion, unemployment, policies in support of small business, and other economic topics (PRN 2018, 27-30, 34-44). An even more telling sign of the party's attempt to moderate can be found in the images used in both manifestos. While the first manifesto includes many different types of purely illustrative images, like images of a beach (PRN 2017, 4), children studying (PRN 2017, 7), the presidential candidate with his wife and daughters (PRN 2017, 10), or two hands shaking (PRN 2017, 36), the second manifesto exclusively uses shots of Costa Rican flags in important historical buildings like the National Theatre or the National Museum, visually becoming both more neutral and more somber. The most interesting differences are, however, the absence from the second manifesto of two images. The first is on page 47 of the first manifesto: a photo of Fabricio Alvarado, the presidential candidate, holding his wife and young daughter, with the following quotation: "We defend life, marriage between man and woman, and integral protection for the family" (PRN 2017, 47). A few pages later (PRN 2017, 50), the manifesto shows a newborn baby in its mother's arms, with the caption "our government will not tolerate, under any circumstance, human life being harmed in any way, and it will oppose aggressively, the legalization of abortion or interruption of pregnancy." The images found in the manifestos relay an important message to voters about whether the political party's platform is congruent with their views. Ultimately, the goal of placing such imagery in manifestos is to attract potential voters who share the same policy platform and may vote for an alternative party but have strong preferences for specific salient issues, such as abortion. By underlining its position with these images, the party signals to voters the importance of the issues to the party, presenting itself as more focused on these issues than similarly positioned parties, in hopes of syphoning off their voters. Moreover, when the party chooses to change its strategy in the runoff and alters (or withdraws) the images in its manifesto, it can have a powerful impact for moderates on social issues. For example, if voters' preferences are not congruent with a political party's social policies but find congruency on other topics, such as economic issues, then the party can downplay its social positions, as exemplified by the disappearance of the PRN's anti-abortion campaign.
The party faces no danger of losing its extremist voters on social issues by doing this, as they are a captive electorate. As there were only two parties in the second round of this election, the voters who valued issues like gay marriage and abortion the most had to choose between the conservative PRN and the progressive PAC. Given these two options, the choice was clear for conservative voters: they should support the PRN. The other competing conservative parties were all out of the race, and no matter how much the PRN moderated both on salience and on spatial, it would remain a more viable option for those captive voters than the PAC with its stances in favor of women's reproductive rights, gay candidates for the Legislative Assembly, and atheist proposed ministers. Downplaying or emphasizing salient issues could then be used to attract the voters who had not yet decided which party to vote for: those most preoccupied with the economic performance of the country. Therefore the PRN replaced its anti-abortion message with more nationalistic images to appear more neutral.

DISCUSSION
This article set out to investigate how niche parties navigate two stages in runoff multiparty elections. In particular, we proposed that in the first round, niche parties strategically try to differentiate themselves from other parties by adopting more extreme positions in salient issue areas, particularly those they own. We also put forward the idea that once a niche party has advanced to the second round of the election, it would then moderate its position to attract the more moderate median voters.
When the theory was applied to the case of the PRN in Costa Rica for the 2018 election, the quantitative analysis showed that in the second round, the niche party increased the salience of economic issues while it moderated its position and moved closer to the center. Regarding the social dimension, the social fabric section of the manifesto data showed the greatest decrease in salience, by 7 percent. Combined with this, the social position, which includes questions from both the social fabric section and others, increased, which means that the party's position in the second round was more moderate than in the first.
At the same time, qualitative evidence obtained from the specific changes between the first and second versions of the manifestos shows that socially conservative and controversial phrases were erased from the manifesto and replaced with detailed economic policy proposals. The tone of the manifesto shifted to a more serious one by replacing pictures of the candidate's family or babies with solemn images of the country's flag.
Based on this, we can say that the PRN effectively increased economic issue salience in the second manifesto it produced, decreased the references to social issues, and moderated its positions on these issues, both obfuscating the issues and moderating its position, following what would be expected on our model. The party chose an issue (abortion) on which it would stand out on the first round, polarized around it, and obtained the conservative votes; then in the second round (once it already had captured voters), it moderated on salience and spatial, trying to capture moderate voters on a second issue, the economy.
In this sense, while previous research on the region has found mostly an economic dimension of ideological competition between parties, to which other issues were mostly subservient, the appearance of niche parties has given such single-issuefocused parties a competitive edge by being able to attract loyal voters in the first round of the election, then reverting to the traditional arena of programmatic competition for the region, the economy, in the runoff. The existence of the second round of the election, combined with the current configuration of the party system as having multiple parties and low partisanship, could potentially affect the number of dimensions of party competition by allowing for strategies such as this one, strategies that could also improve the electoral prospects of niche parties.
The strategy of salience and spatial moderation can be seen as one side of a twosided strategy; the other side is the party's attempt to show competence in the economic issue area. The soberer tone of the PRN's second manifesto points to this: it is not only about showing an economic position closer to the center, but also about showcasing the capability to wrestle with complicated economic policy and demonstrate party competence in this issue area. A constant criticism of this religious niche party throughout the campaign was precisely that it would have no idea what to do in power, and the party engaged in displays of competence to disprove this, not only in the manifesto but also by recruiting respected economists during the runoff as members of Alvarado's potential government, as well as recruiting politicians from the PLN. All this evidence suggests that at least the PRN worked to attract promarket groups (see, e.g., Cascante 2018).
Did the strategy succeed? Well, for starters, the PRN lost the second round, so in the broadest way possible, it did not work. The PRN obtained only 39 percent of the valid votes in the runoff, while the PAC obtained 61 percent. Of course, this does not prove that the strategy of moderating for the second round is not a successful one. It could have also been the case that had the PRN not moderated but instead released a more extreme second manifesto, its vote share would have been even smaller. The problem here, of course, is that we are dealing with unobservable counterfactuals. Not only that; precisely the uniqueness of studying a party that released two manifestos in the same election is, to a degree, a handicap, as there are no other cases in the election to compare it to.
We would like to mention two caveats to the results of this study. The first of these is that, unfortunately, we cannot compare the PRN to the other party in the runoff, the PAC, as the latter did not release a second manifesto during the runoff and had a very short window of time (four days) to react between the release of the PRN's second manifesto and the day of the second round of the election. That said, it should be mentioned that the focus for this article is the strategy of niche parties, meaning that the PAC would be beyond the scope as a mainstream party, considering both its individual strategy and its interactions with the PRN.
The second important caveat is that an exogenous factor was fundamental for the increase of vote intention for the PRN and the success, in our reading, of its strategy. We refer, of course, to the ruling of the Inter-American Court of Human Rights deeming same-sex marriage prohibition as unconstitutional in Costa Rica, an example of judicialization of controversial political issues in the country through the IACHR. (A precedent for this was the court's ruling legalizing in vitro fertilization on November 28, 2012.) The IACHR ruling, which was outside the PRN's control, without a doubt increased the importance of this issue on the public agenda, and, in a sense, what the PRN did was ride the wave of public opinion by skillfully reminding voters that it owned that issue, using the strategy described for the first round of the election: increasing the issue's salience and positioning itself in extreme opposition to it. In other words, while the PRN took advantage of the momentum to increase its support during the first round of the election; it did not create this momentum.
These caveats notwithstanding, we believe that this study is a unique case to examine how party moderating occurs in an attempt to garner more support from median voters or from voters defecting from other parties that did not make the first round cutoff. We would like to reemphasize that the Downsian expectation of convergence to the median voter's position as the best spatial strategy for parties to follow rests on a series of assumptions, several of which were violated in the case of the first round of the Costa Rican election: the number of parties, the presence of a runoff, the number of ideological dimensions relevant to the competition, and the number of peaks in voters' preferences. Failing these assumptions, it is not the best strategy for parties to converge at the median voter, as the first round of the election shows. However, once the conditions change and a party moves into a runoff, we can observe how the party can change its strategy, hoping to adapt to the new conditions it finds itself in.
The unique data that we have utilized clearly depict how moderating is occurring with this particular political party in the Costa Rican context. It is worth further study to examine how other political parties in Costa Rica and elsewhere in the world navigate multiparty systems by downplaying or playing up issue salience. Furthermore, this research agenda also calls into question the speed with which voters adapt their perception of party positions. While previous studies have found considerable lags in the effects of party shifts for European cases, in those studies there is only one signal of the party's position (the single manifesto) per election. More recurrent signals, like the two manifestos studied here, might be more efficient at indicating the voter base where the party wishes to locate itself, making ideological positioning more relevant to the synchronous election.
Our contribution also hinges on the fact that we are studying a very different institutional and structural context than that of Western Europe, the case on which most studies about niche parties focus. In this sense, we are able to study how spatial competition works under very different sets of rules, including presidentialism, as opposed to parliamentarism.
We engage, then, with one of the potential criticisms of the use of the concept of niche party in the Latin American context, particularly highlighting how, even though presidential, the characteristics of the electoral system in Costa Rica would lead to expectations about the number of parties, and therefore the strategic behavior of parties closer to those of proportional representation systems like those found in parliamentary systems in Western Europe. To reiterate, by this we mean that criticisms of the concept of niche parties in Latin America should consider that presidential countries can have electoral systems that would create incentives that lead to multiparty systems, as per Duverger's hypothesis. In that case, Downsian theory would expect more ideological differentiation (rather than convergence) between the parties than in the case of a two-party system, and the strategy that we delineate in our theory allows niche parties to effectively accomplish this ideological differentiation in the first round of the election, and then to become serious contenders in the second round via the strategies of moderation.