Does Exposure to Televised Debates Change the Weight of Different Criteria for Candidate Assessment? A Quasi-Experiment in the Context of the 2014 Spitzenkandidaten Debate

Abstract

In this article, we show how exposure to debates primes specific candidate assessments as key factors of candidate appraisal. To fulfil this goal, we rely on quasi-experimental data collected in 24 European Union Member States and focus on a debate starred by largely unknown candidates (the 2014 European Spiztenkandidaten) engaged in a remarkably invisible campaign. Our results show that candidate perceptions become much more important factors of general candidate appraisal after the debate in the case of three out of the five lead candidates, namely those whose image benefitted from their participation in the debate. In several cases, personal likeability became more important in the general assessment of the Spitzenkandidaten, but there was also an increased relevance of the perceptions of leadership strength (Keller) and quality of the ideas to stimulate the European economy (Schulz and Tspiras). Moreover, in the cases of Schulz, Keller, and Tsipras, post-exposure candidate perceptions impacted more their general appraisal by participants without previous knowledge of them than by those who claimed to know them before the debate. Interestingly, leadership strength appraisal was more relevant for the former than for the latter participants. In short, by unveiling these patterns, this article not only provides evidence of the priming effects of debate exposure but also illustrates how such effects may vary according to citizens’ previous knowledge and the candidates’ general performance in the debate.

1. Introduction

Both politicians and political communication experts recognize that televised debates are paramount to bring voters and candidates closer together (e.g., Graber 1996; Schill 2012), helping the audience to form an impression of the political actors involved in them. The study of effects of exposure to televised debates, which is almost as old as televised debates themselves (Katz and Feldman 1962; Chaffee 1978; McLeod et al. 1979; Swanson and Swanson 1978), has indeed unveiled impacts on voting intentions and political attitudes (e.g., Maier and Faas 2011; Maier et al. 2018), interest and knowledge about politics (e.g., Pfau 2003), or issue salience (e.g., Benoit et al. 2003). The impact of political debates on candidate perceptions and evaluations has also been studied (e.g., Pfau and Rang 1991; Schill and Kirk 2014; Baboš and Világi 2018; Nina and Santana-Pereira 2021). This latter effect is particularly relevant as an indirect way of impacting voting behavior, since candidates’ or leaders’ assessments are an important driver of vote choices in contemporary democracies (e.g., Lobo and Curtice 2014; Warner and Banwart 2016).

Amongst the possible impacts of televised debate exposure, priming—the fact that the debates make certain dimensions more salient leads viewers to use those dimensions as criteria in the assessment of candidates—is a strong possibility. Nonetheless, the study of priming effects has been mostly restricted to newscasts or newspaper articles (e.g., Iyengar et al. 1982; Iyengar and Kinder 1987). Indeed, apart from a few studies (Druckman 2003; Pattie and Johnston 2011; Nina and Santana-Pereira 2021), the literature has neglected the impact televised debates may have on increasing the weight of specific factors in the assessment of candidates. At the same time, with a few exceptions (e.g., Druckman 2003; Nina and Santana-Pereira 2021), the extant literature has struggled with the need to isolate and disentangle debate effects from those of other campaign events taking place at the same time, in a context of heightened political communication and with citizens holding previous information about the candidates.

This article contributes to the study of debate effects by analyzing the extent to which exposure to a political debate activates different criteria for the assessment of candidates, in a low intensity campaign in which the candidates are unknown by considerable proportions of the electorate. In other words, this article seeks to understand how the exposure to a televised debate might prime certain traits over others when it comes to evaluate the politicians participating in it, dealing with the difficulties mentioned above (other campaign events and previous knowledge of candidates) via the selection of an appropriate electoral context. In order to achieve this goal, we use data collected via a quasi-experiment carried out in 24 European countries, which used as experimental stimulus the 15 May 2014 Eurovision debate between the five lead candidates (Spitzenkandidaten) to the European Commission (EC) presidency: Jean Claude Juncker, nominated by the European People’s Party (EPP); Martin Schulz, from the Progressive Alliance of Socialist and Democrats (PES); Guy Verhoftstadt, from the Alliance of Liberals and Democrats for Europe (ALDE); Ska Keller, nominated by the European Green Party (Greens); and Alexis Tsipras, the candidate from the Group of the European United Left (PEL).

This Spikzenkandidaten debate constitutes an interesting tool to test how exposure to a political debate affects candidate evaluations, both due to the nature of European campaigns in general and the Spitzenkandidaten process in particular. Indeed, European Parliament (EP) election campaigns are marked by a low degree of salience (Van der Eijk and Franklin 1996; De Vreese et al. 2006), being indeed second-order campaigns for second-order elections (e.g., Wilke and Reinemann 2007). In 2014, the existence of Spitzenkandidaten was a complete novelty, resulting from the changes introduced by the Lisbon Treaty in 2009, which aimed to address the so-called democratic deficit in the European Union (EU) (e.g., Follesdal and Hix 2006; Hix 2008), improving citizens’ participation in the EU’s decision-making process and strengthening their connections with the EU institutions and elites. However, empirical research on the 2014 and 2019 contests suggests that the results of this innovation were, to say the least, quite modest. For instance, Hobolt (2014) showed that awareness of the Spitzenkandidaten existence and of specific lead candidates was minimal in most Member States, while also stressing that “presidential candidates did not play a major role in the election campaigns, except in a handful of countries, and thus had a limited impact on voter participation and vote choices” (p. 1528; on this matter, see also Schulze 2016). National parties gave little salience to the Spitzenkandidaten process (Braun and Popa 2018), and while media consumption slightly increases the odds of being able to express a preference for a lead candidate, it has become increasingly unlikely for the average citizen to find information about the lead candidates in the media (Gattermann et al. 2016; Gattermann 2020; Richter and Stier 2022; Fotopoulos and Morganti 2022). Levels of engagements achieved by the lead candidates on Twitter were also remarkably low (e.g., Rivas-de-Roca and García-Gordillo 2020). This feeble salience of the Spitzenkandidaten in the public sphere during the EP electoral campaign, which can be understood as a failure of this intended innovation, is actually a point of strength of our research design, because it allows us to isolate the effects of the Eurovision debate and understand the role of holding previous information about the candidates. Such a context may also be a fertile ground for debate effects at large and priming effects in specific, as some studies have pointed that such effects are greater when candidates are less known (Moy et al. 2006; Blais and Perrella 2008; Baboš and Világi 2018; Nina and Santana-Pereira 2021).

Moreover, the use of the data produced by the 2014 quasi-experiment constitutes an unique opportunity to test the priming effects of debate exposure with a good balance between internal and external validity: it resorted to a real-life stimulus in real time and included participants from 24 different countries, while allowing us to experimentally isolate the effects of debate exposure by comparing the weight of different factors in explaining general candidate assessments before (that is, in a situation which is similar to cross-sectional polls and surveys) and immediately after watching a debate.

Lastly, the focus on European debates constitutes a contribution to expand the literature on debate effects beyond the American context, as debates taking place outside the US have been understudied (McKinney and Carlin 2004; Anstead 2016; for recent exceptions, see Baboš and Világi 2018; Maier et al. 2018, or Nina and Santana-Pereira 2021).

This article is organized in five sections. In the next section, we provide an overview of the research on the effects of televised debates, with a particular emphasis on their specific effects on candidate perceptions and evaluations, and summarize the existent literature on the effects of the 2014 Eurovision Spitzenkandidaten debate. The following section restates the main goals of this article and introduces the hypotheses to be tested. Then, the methodology is described. A fourth section is devoted to the results of the statistical analysis of experimental data, aimed at testing our hypotheses. The article concludes with a brief discussion of the main results and their implications.

2. The Effects of Televised Debates: An Overview

Research on televised debate effects is about six decades old. The four head-to-head debates in the US between Richard Nixon and John F. Kennedy in 1960 sparked interest in the effect of debates on public opinion, giving rise to a plethora of studies (McKinney and Warner 2013). Along with the consolidation of television as the preferred means of political communication, research on the effect of televised debates has become a prevalent topic in the field. The academic literature has primarily focused on how debate exposure might influence voting behavior (e.g., Holbert 2005; Maier and Faas 2011) and the perceived importance of certain issues (e.g., Swanson and Swanson 1978; Benoit et al. 2003). Additionally, researchers have analyzed how exposure to debates affects viewers’ interest and knowledge of politics (e.g., Pfau 2003) and candidate appraisals (e.g., Pfau and Rang 1991; Baboš and Világi 2018; Nina and Santana-Pereira 2021).

First, some of the academic literature concerned with the impact of televised debates on voting behavior has shown that exposure has little impact on voting intentions (Katz and Feldman 1962; Benoit et al. 2001, 2003; McKinney and Carlin 2004). However, a few studies suggest that debates can help shape electoral preferences among undecided voters with feeble party loyalties (Chaffee and Choe 1980; Geer 1988). Chaffee (1978), based on an analysis of the 1960 and 1976 US presidential debates, concluded that the dynamics of the electoral campaign as well as the candidates’ performance in the debates might impact the vote, especially when one of the candidates is not particularly well known, there is a large number of undecided voters, and party identification is weak. In addition to this direct impact on voting behavior, televised debates can also play a role in determining the salience of topics that are significant to viewers (e.g., Swanson and Swanson 1978; Benoit et al. 2003). Furthermore, other research has explored how televised debates can increase citizens’ political knowledge and interest in the electoral campaign. For example, Chaffee (1978) and Wald and Lupfer (1978) found that exposure to debates can foster political knowledge, while Lemert (1993) examined how debates lead viewers to seek additional political information. Pfau (2003) argued that debates may be the only televised political event that can attract disengaged young voters and improve their interest in politics. This argument has been supported by subsequent work by McKinney and Chattopadhyay (2007) and McKinney and Rill (2009).

The impact of debates on candidate assessments—the focus of the present article—has also been often studied. The 1960 American presidential debate between Nixon and Kennedy prompted numerous studies on how televised debates can influence the way individuals form opinions on candidates and assess their skills and performance (Weaver 2007). To name just a few, Benoit et al. (2003) found that watching debates significantly impacts the evaluation of candidates’ personalities but did not significantly affect viewers’ perceptions of their competence, especially their leadership skills. Zhu et al. (1994) and Blais and Perrella (2008) showed how debates can improve the perceptions of less known or less popular candidates. In addition, Schill and Kirk (2014) observed that debates can influence the way undecided voters think about presidential candidates, while Warner et al. (2011) reported modest impacts of debate exposure on presidential candidate trait perceptions within a sample mostly composed of people who had already decided who to vote for. Interestingly, Mullinix (2015) reported reinforcement effects of debate exposure on candidate assessments, and Warner et al. (2020) showed how debates might impact positively first and foremost the watcher’s perception of her party’s candidate.

As mentioned above, in this article, we are particularly interested in priming effects. This phenomenon refers to how media content can shape the standards by which people evaluate political actors (Severin and Tankard 1997). The study of media priming effects is based in cognitive psychology research suggesting that exposure to a stimulus can unconsciously activate an idea or concept that will subsequently filter judgment (Pan and Kosicki 1997). The underlying premise is that people tend to rely on easily accessible information to form judgements, as they lack extensive knowledge of political issues and are unable to consider all the relevant information when making decisions (Hastie and Park 1986), and that the media makes some information more accessible than other. Iyengar et al.’s (1982) study showed that priming is subsequent to agenda-setting: once an issue becomes salient in the media, it will also become an important criterion that individuals will use to evaluate candidates (see also Iyengar and Kinder 1987). Media priming effects were observed in studies focusing on newscasts or newspapers (Iyengar et al. 1982; Iyengar and Kinder 1987; Iyengar and Simon 1993; Miller and Krosnick 2000; Scheufele and Tewksbury 2007) and even on late-night shows (Moy et al. 2006), as well as in a couple of studies focusing on debates between political candidates. For instance, Pattie and Johnston (2011), in their analysis of the party leaders’ debates in the 2010 UK election, concluded that the debate performance in valence issues such as prosperity, security, and high-quality services was a key factor of general attitudes towards those leaders. However, the media can also increase the salience of certain personality-related characteristics (Weaver et al. 1981), thus presenting them as relevant criteria for the assessment of public officials. This, of course, paves the way for a series of other interesting priming effects of televised debates, namely in terms of personality features, since personalization seems to be a corollary of televised debates (e.g., Adam and Maier 2010), if not of political communication in TV in general (e.g., Graber 1996). As Lanoue and Schrott (1991) point out, viewers tend to use debates as a source of information about the candidates’ personalities rather than to assess their performance on issues.

In 1962, Katz and Feldman suggested that viewers of the 1960 debate between Nixon and Kennedy placed greater importance on candidates’ character and image than on their performance. However, it was not until Druckman’s (2003) study that one of the most popular anecdotes in political communication regarding the priming effects of televised debates was empirically supported: the evaluation of candidates by individuals who watched the debate relied more on their perception of the candidates’ personalities rather than those who only listened to the debate. This means that watching the televised debate led people to evaluate the candidates according to their personality traits somewhat more than in line with their substantive performance. A few years later, Nina and Santana-Pereira (2021), in an analysis of the 1986 Portuguese presidential debates, found that this effect was greater in the case of the less known candidate—a finding that resonates with that of Moy et al.’s (2006) analysis of priming effects of late-night shows.

As mentioned before, the 2014 Eurovision Spitzenkandidaten debate was different from most debates, which are starred by (at least relatively) known national figures and focus on national subjects. To what extent does this make this debate more or less able to impact viewers? The extant research points to the possibility of observing strong impacts. On the one hand, Baboš and Világi (2018) compared the exposure to the 2014 Spitzenkandidaten debate with the exposure to debates held in the context of the 2016 national elections in Slovakia and found that the effect on viewers’ evaluations of candidates’ leadership skills, credibility, and economic competence was stronger in the pan-European debate. Dinter and Weissenbach (2015) not only corroborated the idea that this debate impacted (positively) candidate assessments by voters, but also reported effects on their perception of being informed about politics. Maier et al. (2016) also reported impacts on attitudes towards the EU amongst young German voters, while Maier (2015) identified amongst the same population positive impacts in the assessment of the political system. Using the very same data that the present article is based on, collected in 24 different EU Member States, Maier et al. (2018) identified effects in terms of knowledge (greater ability to express opinions on the candidates and their positions) and perceived levels of information on the EU (increased), perception of the candidates’ ideological positions (viewers rated all candidates except Juncker as more left-wing and all candidates as more pro-EU after that before the debate), perception of polarization in the candidate pool (increased), interest in the campaign and internal efficacy (increased), and attitudes towards the EU (improved). There is no evidence, however, on whether exposure to this debate had priming effects. As detailed in the next section, this is the key contribution of the present article.

3. Goals and Hypotheses

The main goal of this article is to shed light on the impact of watching a debate on the criteria used to assess those candidates involved in debating, a considerable underdeveloped line of research in the field, as we showed in the previous section. This goal is pursued via the analysis of quasi-experimental data collected in the context of the Eurovision debate broadcast on 15 May 2014, in which the five lead candidates to the presidency of the EC (Spitzenkandidaten) participated. This specific setting allows us to isolate the effects of the debate from that of the EP election campaign (since the lead candidates were quite absent from the campaign; (Hobolt 2014; Braun and Popa 2018)) and to compare its impact in viewers who did or did not hold previous information about the Spitzenkandidaken.

The broad expectation is that the weight of different factors on the general appraisal of these five candidates will be different before and after the debate was watched. In more concrete terms, before the debate, the candidates’ general assessment will be explained by variables such as ideology or issue positions, key factors of candidate appraisal, along with partisanship ((Huckfeldt et al. 2005; Pattie and Johnston 2011; see also Gattermann et al. 2016), for evidence on the importance of party cues as factors of Spitzenkandidaten preference for knowledgeable citizens in the Netherlands, thus replicating Lyons and Scheb’s 1992’s findings). Of course, candidate assessments, both issue-related and persona-related (Hacker et al. 2000), will matter too (e.g., Lavine and Gschwend 2007), even if their relative weight might vary from candidate to candidate (Funk 1999; Huber 2015). However, based on the literature on effects of debate exposure on candidate assessments (e.g., Druckman 2003; Nina and Santana-Pereira 2021), we expect that the assessment of specific characteristics will matter more in explaining the general assessment of candidates after the debate has been watched, since those characteristics are made more available in memory by the debate. Moreover, based on that same literature, a second expectation is that the explaining weight of personality assessments will be particularly boosted by debate exposure when compared with that of the assessment of the candidates’ ideas. In short, we expect that:

Hypothesis 1.

The assessment of candidates’ characteristics will have a stronger impact on their general assessment after the debate has been watched than before.

Hypothesis 2.

The increase in the impact of personality assessments on the general candidate appraisals after the debate will be larger than the increase in the impact of other candidate assessments.

In this study, we also compare the weight of different factors on the post-exposure assessments of candidates made by viewers who held previous knowledge on them and who were made aware of the candidates only via their participation in the experiment. By doing so, we can test whether candidate awareness is an important moderating factor of priming effects, as the extant research suggests (e.g., Moy et al. 2006; Baboš and Világi 2018; Nina and Santana-Pereira 2021). We therefore expect that:

Hypothesis 3.

Specific candidate appraisals will be more important for those who did not know the candidates before the debate than for those who did know them.

4. Methodology

4.1. Procedure

We tested our hypotheses with data from a quasi-experiment with the participation of young citizens from 24 EU Member States1, who were exposed to the 2014 Eurovision Spitzenkandidaten debate in their respective universities. The reception mode differed slightly across countries: about four in each five participants were able to watch the debate simultaneously translated in their native language, while the remainder watched the original version of the debate broadcasted in English (Maier et al. 2018). Moreover, 91 percent watched the debate live, with the others having watched a recording of the debate the very next day.

The participants were invited to complete a pre-treatment and post-treatment questionnaire immediately prior to and right after watching the debate. Both questionnaires included questions designed to measure attitudes towards the candidates and their debate performance, amongst other items.

In contrast to a classic experiment, the quasi-experiment design used in this study did not include a control group, for reasons of practicality and feasibility. Quasi-experimental designs are useful for studying real-world phenomena and are common in experimental studies on debate effects (e.g., Maier and Faas 2003; König and Waldvogel 2022). In addition, in this specific experimental design, the absence of a control group and randomized participant assignment does not strike us as able to undermine the validity of our findings. On the one hand, the short lapse of time between the pretest and the posttest questionnaires and the restrained lab setting in which the experiment took place (classrooms without access to additional information) minimize the likelihood of history threats to the internal validity of this research. On the other, Lupfer and Wald (1979) point out that in experimental studies in which participants display a high level of interest in politics, questionnaire-induced differences in replies between the pretest and the posttest are rare, so the testing threat to internal validity is also under control. As we show in the next subsection, our sample of participants is indeed characterized by high levels of interest in politics.

4.2. Participants

A total of 870 youngsters (mostly undergraduate students) participated in this experiment, 828 of which were eligible voters in the 2014 EP election. 41 percent were Political Science students and 1 in every 3 were enrolled in Communication or other Social Sciences programs. The average participant age was 23.7 years, and there some female overrepresentation (56 percent). Participants placed themselves, on average, in position 4.3 in a 0 (left) to 10 (right) scale and 2 in every 5 claimed they did not feel close to a party. More than 80 percent of the participants declared to be “very” or “somewhat” interested in politics, and their levels of general knowledge of the EU were remarkably high, with 69 to 81 percent being able to correctly answer factual questions on the number of Member States, on whether specific countries were members or not, and on how MEPs and the president of the EC are elected. In spite of this, less than 20 percent of the participants claimed to be interested in the EP campaign. Significant proportions of our participants declared that they did not know the candidates in the initial stage of the experiment (the pre-treatment questionnaire): the figures were 53 percent for Keller, 52 percent for Verhofstadt, 51 percent for Tsipras, 41 percent for Juncker, and 31 percent for Schulz.

4.3. Experimental Stimulus

The 90-min Eurovision Spitzenkandidaten debate was broadcast live on 15 May 2014, one week before the first EP election day2. For the first time in the history of EU democracy, all five candidates to the EC presidency shared the same stage in a public, televised, debate. In the lead-up to the 2014 EP elections, the lead candidates were invited to participate in several televised debates, but these only included two to four candidates and were broadcast in only a subset of EU countries. In contrast, the Spitzenkandidaten debate was aired 28 countries through multiple national broadcasters and websites and offered simultaneous interpretation in several languages. Although the debate was introduced and moderated in English (by Italian journalist Monica Maggioni) the candidates were free to use their preferred language. This was the case for Tsipras and Juncker, who debated in their native languages: Greek and French, respectively.

The five candidates were arranged from left to right according to their ideology: Alexis Tsipras, then Ska Keller, then Martin Schultz, then Jean-Claude Juncker, and finally Guy Verhoftstadt. The order in which each the candidate spoke was predetermined and alternated according to the topic being discussed. The debate rules were straightforward: each candidate had one minute to respond to each question, and they were all asked the same questions. Each intervention was timed on a giant screen. Additionally, each candidate had the chance to use the “blue joker”—a 30-s rebuttal—up to three times.

The debate was divided into three substantive parts. The first set of questions was devoted to the European financial crisis and its aftermath, the second set of questions discussed the role of EU in the world and its external relations, whereas the third part focused on the priorities for the future of the EU. In addition to addressing these three topics, each candidate had the opportunity to make an opening statement. Verhofstadt, Schultz, and Juncker—all labeled as federalists (e.g., Dinan 2016; Christiansen 2016)—emphasized in their opening statements the need for further European cooperation and solidarity among EU countries to address both old and new challenges. They focused on consolidating the European economy and creating a more responsive EU that prioritizes the needs of its citizens. Keller also emphasized the importance of European solidarity and the need for the EU to address the social needs of citizens. She argued for a new approach that takes into account current problems such as climate change and the future of the younger generations. Tsipras, on the other hand, strongly emphasized the negative impact of austerity measures, particularly those that Greece (his home country) had experienced. He argued that the path of austerity cannot be the solution to the EU’s problems. Furthermore, he was the only candidate who explicitly used ideological arguments in his opening statement, claiming that a left-wing vote is the only way to save Europe from austerity and restore European values. The debate ended with Ska Keller drawing attention to the plight of people suffering from poverty, hardship, war, and persecution. She held up a poster with the message #BringBackOurGirls International Campaign, a gesture that was replicated by the other candidates and received a big round of applause from the audience.

Table A1. Main positions taken by the five candidates during the debate and on each theme addressed.

	Financial Crisis		Europe and the World		Priorities of the Future
Alexis Tsipras	Stopping austerity and pursuing a more solidarity-focused EU. A comprehensive solution to bank regulation. Holding accountable the parties that paved the way to Eurosceptic parties by implementing austerity measures.		Against any military intervention and imposing sanctions on Russia as it will further divide Europe. The EU should respect the self-determination of people, but he does not encourage border changes. It is counterintuitive to discuss immigration when EU is a party in international conflict. EU needs an asylum program for the burden of refugees to be shared by all Member States.		It is anachronist to use religion as a political tool. Corruption is a systemic issue, especially in southern countries, so fighting it is a collective problem that requires effective and transparent EU legislation. The biggest challenge for the EU is the democratic deficit, particularly due to the austerity measures which were imposed by EU elites without regard to national constitutions. Euroscepticism must be combatted, and one way to do it is by respecting the outcomes of the 2014 EP election.
Ska Keller	Instead of an austerity policy, investment in green economy such as renewable energies: they are the future job opportunities to younger generations. The banks and the financial markets should be strictly regulated. It is necessary to listen to the concerns of citizens and prioritize their social rights.		In addition to imposing sanctions to Russia, the EU should re-evaluate its dependence on Russian energy and invest in renewable energy sources. If Catalonia or Scotland decide to become independent and join the EU, they should be welcomed. The EU needs a common asylum law and promote dialogue is each Member States on this issue.		Religious symbols should not be regulated, especially not at the EU level. Individual freedom should be respected, but the public space should remain neutral. People care about the EU, but they do not believe that their vote will produce changes. An example is how lobbyists are able to influence decisions, while the same cannot be said for citizens. It is necessary to strengthen European democracy by giving a voice to EU citizens, so the results of the 2014 EP elections should be respected.
Martin Schulz	The austerity measures should have been combined with budget strategic investment, economic growth, and employment opportunities. It is necessary to regulate the banks and hold them accountable for their actions. The Europeans want a different Europe, one in which the interests of the people come first.		EU is a soft power and will do everything to prevent a war. We must assist eastern EU countries if needed and reinforce the role of each Member State’s foreign minister. Since EU is not a federation, it should not take a stance on the ambitions of independentist movements. The EU must have a legal system for immigration, as this is a competency of the EU, not the national governments.		EU advocates for an anti-discrimination policy, but we must also respect the specific laws of each Member State. Corruption is an EU issue that must be tackled through join efforts at both national and European level. The President of the EC is now selected through a more democratic process and this debate is an example of how the EU is responding to citizens’ calls for a more transparent and accountable Europe.
Jean-Claude Juncker	To achieve sustainable employment, debt should be controlled, because it is important to maintain credibility and economic stability. Young people should be prepared to work in other countries. The banks were saved to protect the small and medium-size companies.		The EU should intensify the sanctions against Russia. It is not wise for the EU to interfere in matters that fall under the jurisdiction of Member States, such as independentist movements. However, regarding illegal immigration, a common EU law and an increase in the budget to address this issue is needed.		It is important to have European legislation against discrimination, but the EU should not intervene at local or regional level: EU defends the principles, but the employment of these principles belongs to Member States. Similarly, a better framework for lobbying activity and fighting corruption is necessary, but interest groups cannot be banned. National governments must stop blaming Europe for certain issues as the EU only exists because they allow it. Youngsters must be convinced that the EU protects them and that their voices are heard.
Guy Verhofstadt	In order to achieve greater economic integration, the EU should consider expanding investment opportunities and attracting foreign companies. The EU requires a new model of banking system. If southern EU countries decide to leave the Euro, how will they survive?		The EU should work together to implement sanctions on Russia and develop a unified immigration law to address issues such as illegal immigration and human trafficking, as well as establish a common policy for refugees and asylum seekers. The EU should respect the decisions of national citizens regarding the independence of regions but refrain from intervening in matters that solely concern Member States.		The EU needs a collective solution on anti-discrimination law to protect minorities and religions. While it may be impossible to completely forbid EU lobbyists, corruption must be fought with new laws. If the President of the EC is not nominated on based on the election results, it will be the end of a democratic Europe. Politicians who defend Europe and engage new generations with new arguments as well as values that cannot be achieved at the national level are necessary.

in Appendix A provides a summary of the main arguments made by the five lead candidates during the debate. The full transcription of the debate can be found in Maier and Jasen (2014), while the debate itself is still available online.3

Interestingly, the candidates’ appearance and behavior were distinct. For instance, Juncker, Schulz, and Verhoftstadt—all veterans of EU politics—wore formal attire (dark suits and ties) while Tsipras—a newcomer to the European level competition and a leftwing politician—decided to wear a dark suit without a tie. Keller, the only female candidate, opted for a vibrant green pantsuit, which emphasized her connection to the Green Party, while also presenting her as a serious politician. In terms of the candidates’ behavior, Verhoftstadt was described by some media4 as highly energetic, even when not speaking, and with an effusive body language. However, he also displayed a tendency to be highly reactive and frequently interrupted the other candidates, often using an accusatory tone and making sarcastic comments directed at Keller in particular. Keller, the youngest among the five candidates, stood out as the most likable and approachable: she frequently addressed the public in the hall and the audience at home, and even when disagreeing with the other candidates, she always maintained a pleasant tone. Her efforts were met with enthusiastic support from the hall, who cheered loudly after each of her interventions. Schulz, on the other hand, adopted a «presidential» posture from the start, speaking carefully and addressing the audience as through he was already the president of the EC. Schulz maintained a serious demeanor and did not smile or acknowledge the opinions of the other candidates. He remained polite throughout the debate but occasionally displayed a condescending attitude towards Verhoftstadt and subliminally scolded Tsipras. Juncker and Tsipras, in contrast, were more neutral. However, they differed in their approach to the debate: Tsipras appeared hesitant and reactive when discussing austerity, while Juncker remained relaxed and focused on his role in the debate.

4.4. Measures

The dependent variables in the regression models presented in the next section correspond to the general assessments of the five lead candidates on a scale ranging from −5 (very unfavorable feeling towards the candidate) to 5 (very favorable feeling). For each candidate, three separate dependent variables are used: a pre-exposure measure (only for those who claimed to know the candidates), a post-exposure measure for the same group of participants, and a post-exposure measure for those who got to know the candidates only by watching the debate.

In what regards independent variables, these are assessments of candidates’ personality traits, dealing with both warmth and competence (see, for instance, Costa and Ferreira da Silva 2015), and political ideas. Concretely speaking, three items dealt with the personality dimension: a general character appraisal (“I like [CANDIDATE] as a person”; 1 = strongly disagree; 5 = strongly agree), an appraisal of the candidates’ trustworthiness (“[CANDIDATE] is a trustworthy politician”; 1 = strongly disagree; 5 = strongly agree), and an assessment of their leadership skills (“[CANDIDATE] is a strong leader”; 1 = strongly disagree; 5 = strongly agree). In turn, the substantive appraisal is based on an item regarding the candidates’ ideas to deal with a key issue in the 2014, post-Great Recession, public sphere (“[CANDIDATE] has good ideas to stimulate the economy in the EU”; 1 = strongly disagree; 5 = strongly agree). All variables were measured pre-exposure and post-exposure, and are used in the models dealing with dependent variables measured at the same time. Our models also include some control variables, namely age (continuous), gender (1 = female), and country of residence (1 = same as that of the candidate’s home country),5 as well as known factors of candidate assessment, such as ideology (measured through the traditional self-positioning question, and varying from 0 = left to 10 = right) and issue positions, namely EU-related issues, measured through two variables: views on European integration (0 = European integration has already gone too far; 10 = European integration should be pushed further) and on solidarity between EU Member States in times of financial crisis (“Do you agree or disagree that in times of crisis countries that are better off should give financial help to another EU Member State facing severe economic and financial difficulties?”; 1 = totally disagree; 4 = totally agree).

5. Results

Before the presentation of the statistical analysis aimed at testing our hypotheses, it is important to show whether exposure to the Spitzenkandidaten debate of 15 May 2014 had an impact on the general candidate assessments by those who already knew them, and if there are differences between the post-exposure assessments made by those who were able to rate the candidates before the debate and by those who claimed not to know them. In fact, if such impacts were to be found absent, priming effects would lose substantive interest. What the experimental data show us is that, in the cases of Jean-Claude Juncker and Alexis Tsipras, general ratings decreased after the debate, going from 0.36 to −0.40 for the former [t(477) = 3.55; p < 0.001] and from 0.60 to −0.36 for the latter [t(395) = 3.44; p < 0.001]. In turn, Martin Schulz’s general image improved, going from 1.3 to 1.9 [t(586) = −5.88; p < 0.001], as did those of Guy Verhofstadt [from 0.23 to 0.85; t(381) = −4.62; p < 0.001] and Ska Keller [from 1.0 to 2.3; t(385) = −9.19; p < 0.001]. Post-exposure ratings from those who had assessed the candidates before the debate and from those who were only able to do so after the debate were not statistically different, except for the case of Alexis Tsipras, whose ratings were a bit more negative in the group of participants who did not know him in advance (−0.74 vs. −0.36) [t(851) = 2.16; p < 0.05]. In short, participation in this debate seems to have benefitted some candidates and harmed others in the eyes of the participants in this study, whereas previous information on the candidates is not (except in the case of Alexis Tsipras) linked to differences in terms of impact of debate exposure on general appraisals of the Spiztenkandidaten.

We now turn to the test of the three hypotheses presented above, focused on potential shifts in the relative weight of different factors in the general appraisal of these lead candidates. Table 1 allows a comparison of the variance explained by two linear regression models (one only containing ideology, EU issue positions and the controls, and a second one also including the candidate specific appraisals) in three instances (before and after watching the debate for those who knew the candidates before participating in the experiment, after the debate for those who did not). This comparison allows us to observe if the global weight of specific appraisals is stronger after the debate and for those who learned about the candidates only by participating in this study, as hypothesized. In turn, the linear regression models presented in Table 2, Table 3, Table 4, Table 5 and Table 6 address the impact of different criteria on the general assessment of the five lead candidates before and after exposure to the debate, and, in this latter case, for those who did and who did not have previous knowledge about the candidates. In the analysis of this data, standardized coefficients (Betas) are of paramount importance, as they allow us to identify the factors that most strongly impact the general assessment of each Spitzenkandidat.

We start with the test of Hypothesis 1, which postulated that the assessment of candidates’ specific characteristics would be more important after than before the debate. The following analysis therefore only focuses on the participants who were able to express opinions on each candidate before watching the debate. The data displayed on Table 1 leads to the conclusion that the expected pattern was observed for Ska Keller (the difference in the variance explained by the full model vs. a model without assessments was 47 percent points after the debate and only 25 percent points before the debate), Guy Verhosftad (the corresponding figures are 63 and 40 percent points), and, quite modestly, Martin Schulz (47 vs. 44 percent points).

Table 1. Candidates’ general assessments: Difference in the variance explained by a restricted (only controls, ideology, and issue positions) and a full model (including specific assessments).

	Before Exposure			After Exposure			Only Post-Exposure
	Model without Assessments	Full Model	Difference	Model without Assessments	Full Model	Difference	Model without Assessments	Full Model	Difference
Juncker	4.8	61.4	56.6	12.2	69.2	57	7.1	58.3	51.2
Verhofstadt	10	50.3	40.3	7.9	71.3	63.4	7.5	65.7	58.2
Schulz	12.3	56.3	44	9.1	55.9	46.8	3.9	53.6	49.7
Keller	11.2	36.3	25.1	8.5	55.4	46.9	7.3	60.4	53.1
Tsipras	1	52.2	51.2	15.4	66	50.6	4.6	61	56.4

Table 1. Candidates’ general assessments: Difference in the variance explained by a restricted (only controls, ideology, and issue positions) and a full model (including specific assessments).

	Before Exposure			After Exposure			Only Post-Exposure
	Model without Assessments	Full Model	Difference	Model without Assessments	Full Model	Difference	Model without Assessments	Full Model	Difference
Juncker	4.8	61.4	56.6	12.2	69.2	57	7.1	58.3	51.2
Verhofstadt	10	50.3	40.3	7.9	71.3	63.4	7.5	65.7	58.2
Schulz	12.3	56.3	44	9.1	55.9	46.8	3.9	53.6	49.7
Keller	11.2	36.3	25.1	8.5	55.4	46.9	7.3	60.4	53.1
Tsipras	1	52.2	51.2	15.4	66	50.6	4.6	61	56.4

For Verhosftadt, for instance, participants’ ideological self-placement became less important after the debate, whereas their perceptions of this candidate’s likeability became considerably more important, even if this came with a lower relevance of perceived trustworthiness and strength (Table 2). The pattern identified for Ska Keller was somewhat similar: participants’ positions on EU integration became less relevant after the debate, as did ideology (it became insignificant, even though it was only barely significant before debate exposure), whereas personal likeability and strength increased their impact on her general assessment (Table 4). Trustworthiness also became less relevant. Lastly, in the case of Martin Schulz, the assessment of his ideas to stimulate the European economy became much more relevant after the debate, whereas trustworthiness lost weight, as did participant’s positions on EU integration (Table 5).

Table 2. Determinants of the general assessment of Guy Verhofstadt before and after exposure to the debate and for those who did not know the candidates before the debate (OLS Regressions).

	Before		After		Only Post-Exposure
	B (St. Error)	Beta	B (St. Error)	Beta	B (St. Error)	Beta
Intercept	−8.0 *** (0.86)		−7.4 (0.65)		−7.2 *** (0.70)
Age	0.01 (0.02)	0.02	0.01 (0.01)	0.01	−0.01 (0.01)	−0.02
Gender (female)	0.02 (0.23)	0.01	−0.27 (0.18)	−0.04	−0.07 (0.19)	−0.01
Ideology (L/R)	0.15 ** (0.05)	0.14	0.10 * (0.04)	0.08	0.08 (0.04)	0.06
EU integration should be pushed further	0.09 (0.05)	0.10	0.05 (0.04)	0.04	0.01 (0.04)	0.01
Solidarity between Member States during times of crisis	0.04 (0.19)	0.01	−0.01 (0.14)	−0.01	−0.11 (0.15)	−0.02
Verhofstadt has good ideas to stimulate economy	0.77 *** (0.15)	0.30	0.76 *** (0.11)	0.29	0.87 *** (0.12)	0.31
Verhofstadt is trustworthy	0.58 ** (0.18)	0.21	0.28 * (0.14)	0.10	0.25 (0.14)	0.09
I like Verhofstadt as a person	0.49 ** (0.17)	0.19	1.23 *** (0.11)	0.51	1.10 *** (0.11)	0.45
Verhofstadt is a strong leader	0.34 * (0.15)	0.12	0.09 (0.11)	0.03	0.26 * (0.10)	0.09
R2 (adjusted)	50.3%		71.3%		65.7%
N	247		357		415

Notes: This table displays unstandardized coefficients with standard errors in parentheses, as well as standardized coefficients. Significance levels: * p < 0.05, ** p < 0.01; *** p < 0.001.

Table 2. Determinants of the general assessment of Guy Verhofstadt before and after exposure to the debate and for those who did not know the candidates before the debate (OLS Regressions).

	Before		After		Only Post-Exposure
	B (St. Error)	Beta	B (St. Error)	Beta	B (St. Error)	Beta
Intercept	−8.0 *** (0.86)		−7.4 (0.65)		−7.2 *** (0.70)
Age	0.01 (0.02)	0.02	0.01 (0.01)	0.01	−0.01 (0.01)	−0.02
Gender (female)	0.02 (0.23)	0.01	−0.27 (0.18)	−0.04	−0.07 (0.19)	−0.01
Ideology (L/R)	0.15 ** (0.05)	0.14	0.10 * (0.04)	0.08	0.08 (0.04)	0.06
EU integration should be pushed further	0.09 (0.05)	0.10	0.05 (0.04)	0.04	0.01 (0.04)	0.01
Solidarity between Member States during times of crisis	0.04 (0.19)	0.01	−0.01 (0.14)	−0.01	−0.11 (0.15)	−0.02
Verhofstadt has good ideas to stimulate economy	0.77 *** (0.15)	0.30	0.76 *** (0.11)	0.29	0.87 *** (0.12)	0.31
Verhofstadt is trustworthy	0.58 ** (0.18)	0.21	0.28 * (0.14)	0.10	0.25 (0.14)	0.09
I like Verhofstadt as a person	0.49 ** (0.17)	0.19	1.23 *** (0.11)	0.51	1.10 *** (0.11)	0.45
Verhofstadt is a strong leader	0.34 * (0.15)	0.12	0.09 (0.11)	0.03	0.26 * (0.10)	0.09
R2 (adjusted)	50.3%		71.3%		65.7%
N	247		357		415

Notes: This table displays unstandardized coefficients with standard errors in parentheses, as well as standardized coefficients. Significance levels: * p < 0.05, ** p < 0.01; *** p < 0.001.

In contrast, there were no substantial differences in the variances explained by specific candidate assessments before and after debate exposure for the cases of Jean-Claude Juncker and Alexis Tsipras (Table 1). In the case of the former, however, we did observe a lower importance of participants’ ideological self-placement and positions of EU integration and an increased importance of perceived likeability after the debate, but this was accompanied by a lower relevance of trustworthiness (Table 3). For the latter, participants’ ideology was also less relevant after the debate, but what the debate did was to increase the relevance of their positions on solidarity between Member States in times of crisis and the assessment of Tspiras’ ideas to stimulate the European economy (Table 6). In sum, Hypothesis 1 is only partially confirmed, receiving empirical support in the cases of Ska Keller, Guy Verhofstadt, and, to a certain extent, Martin Schulz.

Table 3. Determinants of the general assessment of Jean-Claude Juncker before and after exposure to the debate and for those who did not know the candidates before the debate (OLS Regressions).

	Before		After		Only Post-Exposure
	B (St. Error)	Beta	B (St. Error)	Beta	B (St. Error)	Beta
Intercept	−8.7 *** (0.67)	--	−9.7 *** (0.58)	--	−6.7 *** (0.82)	--
Age	−0.01 (0.01)	−0.03	0.01 (0.01)	0.02	−0.02 (0.02)	−0.04
Gender (female)	−0.01 (0.19)	−0.01	0.30 (0.16)	0.05	−0.35 (0.21)	−0.06
Ideology (L/R)	0.23 *** (0.05)	0.19	0.10 ** (0.04)	0.08	0.07 (0.04)	0.06
EU integration should be pushed further	0.12 ** (0.04)	0.12	0.07 * (0.03)	0.06	−0.01 (0.04)	−0.01
Solidarity between Member States during times of crisis	−0.04 (0.16)	−0.01	0.14 (0.12)	0.03	−0.21 (0.17)	−0.05
Juncker has good ideas to stimulate economy	0.80 *** (0.12)	0.28	0.85 *** (0.10)	0.30	0.76 *** (0.14)	0.26
Juncker is trustworthy	0.54 *** (0.13)	0.19	0.38 *** (0.12)	0.13	0.44 ** (0.15)	0.15
I like Juncker as a person	0.81 *** (0.12)	0.31	1.21 *** (0.11)	0.46	0.86 *** (0.14)	0.34
Juncker is a strong leader	0.21 (0.11)	0.07	0.13 (0.09)	0.05	0.48 *** (0.11)	0.19
R2 (adjusted)	61.4%		69.2%		58.3%
N	342		443		326

Notes: This table displays unstandardized coefficients with standard errors in parentheses, as well as standardized coefficients. Significance levels: * p < 0.05, ** p < 0.01; *** p < 0.001.

Table 3. Determinants of the general assessment of Jean-Claude Juncker before and after exposure to the debate and for those who did not know the candidates before the debate (OLS Regressions).

	Before		After		Only Post-Exposure
	B (St. Error)	Beta	B (St. Error)	Beta	B (St. Error)	Beta
Intercept	−8.7 *** (0.67)	--	−9.7 *** (0.58)	--	−6.7 *** (0.82)	--
Age	−0.01 (0.01)	−0.03	0.01 (0.01)	0.02	−0.02 (0.02)	−0.04
Gender (female)	−0.01 (0.19)	−0.01	0.30 (0.16)	0.05	−0.35 (0.21)	−0.06
Ideology (L/R)	0.23 *** (0.05)	0.19	0.10 ** (0.04)	0.08	0.07 (0.04)	0.06
EU integration should be pushed further	0.12 ** (0.04)	0.12	0.07 * (0.03)	0.06	−0.01 (0.04)	−0.01
Solidarity between Member States during times of crisis	−0.04 (0.16)	−0.01	0.14 (0.12)	0.03	−0.21 (0.17)	−0.05
Juncker has good ideas to stimulate economy	0.80 *** (0.12)	0.28	0.85 *** (0.10)	0.30	0.76 *** (0.14)	0.26
Juncker is trustworthy	0.54 *** (0.13)	0.19	0.38 *** (0.12)	0.13	0.44 ** (0.15)	0.15
I like Juncker as a person	0.81 *** (0.12)	0.31	1.21 *** (0.11)	0.46	0.86 *** (0.14)	0.34
Juncker is a strong leader	0.21 (0.11)	0.07	0.13 (0.09)	0.05	0.48 *** (0.11)	0.19
R2 (adjusted)	61.4%		69.2%		58.3%
N	342		443		326

Notes: This table displays unstandardized coefficients with standard errors in parentheses, as well as standardized coefficients. Significance levels: * p < 0.05, ** p < 0.01; *** p < 0.001.

Table 4. Determinants of the general assessment of Ska Keller before and after exposure to the debate and for those who did not know the candidates before the debate (OLS Regressions).

	Before		After		Only Post-Exposure
	B (St. Error)	Beta	B (St. Error)	Beta	B (St. Error)	Beta
Intercept	−4.0 *** (1.0)		−6.4 *** (0.72)		−6.0 *** (0.65)
Age	−0.01 (0.02)	−0.03	0.01 (0.01)	0.01	0.01 (0.01)	0.01
Gender (female)	0.45 (0.25)	0.09	0.17 (0.18)	0.04	0.06 (0.16)	0.01
Country of Residence (1 = candidate)	−0.23 (0.31)	−0.04	0.10 (0.24)	0.02	−0.33 (0.22)	−0.05
Ideology (L/R)	−0.11 * (0.06)	−0.11	−0.11 (0.04)	−0.10	−11 ** (0.04)	−0.10
EU integration should be pushed further	0.10 * (0.05)	0.12	−0.01 (0.04)	−0.01	−0.04 (0.03)	−0.04
Solidarity between Member States during times of crisis	−0.21 (0.21)	−0.06	0.15 (0.14)	0.04	0.18 (0.13)	0.04
Keller has good ideas to stimulate economy	0.58 ** (0.18)	0.20	0.45 *** (0.13)	0.18	0.49 *** (0.11)	0.19
Keller is trustworthy	0.81 *** (0.20)	0.31	0.34 * (0.16)	0.12	0.28 * (0.13)	0.10
I like Keller as a person	0.39 * (0.18)	0.16	1.08 *** (0.14)	0.40	0.90 *** (0.12)	0.37
Keller is a strong leader	−0.18 (0.17)	−0.07	0.36 *** (0.10)	0.15	0.54 *** (0.10)	0.24
R2 (adjusted)	36.3%		55.4%		60.4%
N	254		362		420

Notes: This table displays unstandardized coefficients with standard errors in parentheses, as well as standardized coefficients. Significance levels: * p < 0.05, ** p < 0.01; *** p < 0.001.

Table 4. Determinants of the general assessment of Ska Keller before and after exposure to the debate and for those who did not know the candidates before the debate (OLS Regressions).

	Before		After		Only Post-Exposure
	B (St. Error)	Beta	B (St. Error)	Beta	B (St. Error)	Beta
Intercept	−4.0 *** (1.0)		−6.4 *** (0.72)		−6.0 *** (0.65)
Age	−0.01 (0.02)	−0.03	0.01 (0.01)	0.01	0.01 (0.01)	0.01
Gender (female)	0.45 (0.25)	0.09	0.17 (0.18)	0.04	0.06 (0.16)	0.01
Country of Residence (1 = candidate)	−0.23 (0.31)	−0.04	0.10 (0.24)	0.02	−0.33 (0.22)	−0.05
Ideology (L/R)	−0.11 * (0.06)	−0.11	−0.11 (0.04)	−0.10	−11 ** (0.04)	−0.10
EU integration should be pushed further	0.10 * (0.05)	0.12	−0.01 (0.04)	−0.01	−0.04 (0.03)	−0.04
Solidarity between Member States during times of crisis	−0.21 (0.21)	−0.06	0.15 (0.14)	0.04	0.18 (0.13)	0.04
Keller has good ideas to stimulate economy	0.58 ** (0.18)	0.20	0.45 *** (0.13)	0.18	0.49 *** (0.11)	0.19
Keller is trustworthy	0.81 *** (0.20)	0.31	0.34 * (0.16)	0.12	0.28 * (0.13)	0.10
I like Keller as a person	0.39 * (0.18)	0.16	1.08 *** (0.14)	0.40	0.90 *** (0.12)	0.37
Keller is a strong leader	−0.18 (0.17)	−0.07	0.36 *** (0.10)	0.15	0.54 *** (0.10)	0.24
R2 (adjusted)	36.3%		55.4%		60.4%
N	254		362		420

Notes: This table displays unstandardized coefficients with standard errors in parentheses, as well as standardized coefficients. Significance levels: * p < 0.05, ** p < 0.01; *** p < 0.001.

Hypothesis 2, in turn, deals with the expectation that personality traits would be those in which we would observe a considerable increase in relevance after the debate. We already made reference to specific results of personality-related assessments before, but the test of this hypothesis calls for a more systematic comparison. We therefore start with trustworthiness. For this specific variable, the results do not meet our expectations: assessments on whether the candidates are trustworthy politicians did not change their weight in the case of Alexis Tsipras and became even less relevant after exposure to the debate for the other four candidates (Table 2, Table 3, Table 4, Table 5 and Table 6). Regarding leadership strength, there was an increased impact of this specific assessment only in the case of Keller and, to some extent, Schulz; indeed, this variable was nonsignificant for the general assessment of Juncker and Tsipras both before and after debate exposure and became irrelevant for Verhofstadt after debate exposure (Table 2, Table 3, Table 4, Table 5 and Table 6). Results regarding likeability as a person were much more promising—indeed, after the debate was watched, this variable became much more important for the general assessments of Guy Verhosftadt, Ska Keller, and Jean-Claude Juncker, and it kept its (very important) stance in the models for Alexis Tsipras and Martin Schulz (Table 2, Table 3, Table 4, Table 5 and Table 6).

Table 5. Determinants of the general assessment of Martin Schulz before and after exposure to the debate and for those who did not know the candidates before the debate (OLS Regressions).

	Before		After		Only Post-Exposure
	B (St. Error)	Beta	B (St. Error)	Beta	B (St. Error)	Beta
Intercept	−7.4 *** (0.61)		−7.3 *** (0.55)		−4.5 *** (1.0)
Age	−0.01 (0.01)	−0.04	0.01 (0.01)	0.02	0.03 (0.02)	0.05
Gender (female)	0.25 (0.16)	0.05	0.06 (0.14)	0.01	−0.08 (0.22)	−0.02
Country of Residence (1 = candidate)	0.22 (0.22)	0.03	0.45 * (0.18)	0.07	0.34 (0.33)	0.05
Ideology (L/R)	0.04 (0.04)	0.04	−0.05 (0.03)	−0.04	−0.01 (0.04)	−0.01
EU integration should be pushed further	0.07 * (0.03)	0.07	0.04 (0.03)	0.05	−0.03 (0.04)	−0.03
Solidarity between Member States during times of crisis	0.29 * (0.14)	0.08	0.16 (0.11)	0.05	−0.37 * (0.17)	−0.10
Schulz has good ideas to stimulate economy	0.71 *** (0.12)	0.26	0.96 *** (0.10)	0.35	0.38 * (0.16)	0.14
Schulz is trustworthy	0.41 *** (0.12)	0.16	0.22 * (0.11)	0.08	0.48 ** (0.16)	0.19
I like Schulz as a person	0.87 *** (0.12)	0.35	0.80 *** (0.10)	0.33	1.09 *** (0.15)	0.48
Schulz is a strong leader	0.22 * (0.09)	0.09	0.32 *** (0.09)	0.12	0.05 (0.15)	0.02
R2 (adjusted)	56.3%		55.9%		53.6%
N	433		538		241

Notes: This table displays unstandardized coefficients with standard errors in parentheses, as well as standardized coefficients. Significance levels: * p < 0.05, ** p < 0.01; *** p < 0.001.

Table 5. Determinants of the general assessment of Martin Schulz before and after exposure to the debate and for those who did not know the candidates before the debate (OLS Regressions).

	Before		After		Only Post-Exposure
	B (St. Error)	Beta	B (St. Error)	Beta	B (St. Error)	Beta
Intercept	−7.4 *** (0.61)		−7.3 *** (0.55)		−4.5 *** (1.0)
Age	−0.01 (0.01)	−0.04	0.01 (0.01)	0.02	0.03 (0.02)	0.05
Gender (female)	0.25 (0.16)	0.05	0.06 (0.14)	0.01	−0.08 (0.22)	−0.02
Country of Residence (1 = candidate)	0.22 (0.22)	0.03	0.45 * (0.18)	0.07	0.34 (0.33)	0.05
Ideology (L/R)	0.04 (0.04)	0.04	−0.05 (0.03)	−0.04	−0.01 (0.04)	−0.01
EU integration should be pushed further	0.07 * (0.03)	0.07	0.04 (0.03)	0.05	−0.03 (0.04)	−0.03
Solidarity between Member States during times of crisis	0.29 * (0.14)	0.08	0.16 (0.11)	0.05	−0.37 * (0.17)	−0.10
Schulz has good ideas to stimulate economy	0.71 *** (0.12)	0.26	0.96 *** (0.10)	0.35	0.38 * (0.16)	0.14
Schulz is trustworthy	0.41 *** (0.12)	0.16	0.22 * (0.11)	0.08	0.48 ** (0.16)	0.19
I like Schulz as a person	0.87 *** (0.12)	0.35	0.80 *** (0.10)	0.33	1.09 *** (0.15)	0.48
Schulz is a strong leader	0.22 * (0.09)	0.09	0.32 *** (0.09)	0.12	0.05 (0.15)	0.02
R2 (adjusted)	56.3%		55.9%		53.6%
N	433		538		241

Notes: This table displays unstandardized coefficients with standard errors in parentheses, as well as standardized coefficients. Significance levels: * p < 0.05, ** p < 0.01; *** p < 0.001.

Lastly, Hypothesis 2 implied no relevant change in the impact of the ideas to deal with a key issue in the EU such as the need to stimulate the economy after the Great Recession, on the general assessment of candidates. That is indeed what happened for Juncker, Verhofstadt, and Keller: this variable was equally important both before and after exposure to the debate. In turn, it became more important after the debate for the general appraisals of Martin Schulz and, to some extent, Alexis Tsipras (Table 2, Table 3, Table 4, Table 5 and Table 6).

In sum, Hypothesis 2 cannot be completely accepted. It seems that, more than boosting the relevant of personality-related assessments in general, debates might be more effective in increasing the relevance of perceived warmth or sympathy, but not strength or trustworthiness, for the general impressions people have of political candidates. In some cases, exposure to the debate might also increase the importance of substantive, non-personality related, appraisals.

Table 6. Determinants of the general assessment of Alexis Tsipras before and after exposure to the debate and for those who did not know the candidates before the debate (OLS Regressions).

	Before		After		Only Post-Exposure
	B (St. Error)	Beta	B (St. Error)	Beta	B (St. Error)	Beta
Intercept	−4.7 *** (0.96)		−7.3 *** (0.73)		−7.6 *** (0.69)
Age	−0.02 (0.02)	−0.05	−0.01 (0.01)	−0.01	−0.01 (0.01)
Gender (female)	−0.03 (0.24)	−0.01	0.09 (0.19)	0.01	0.23 (0.18)	0.04
Country of Residence (1 = candidate)	−0.10 (0.38)	−0.01	0.56 (0.33)	0.06	No cases
Ideology (L/R)	−0.16 ** (0.06)	−0.14	−0.09 * (0.04)	−0.07	−0.02 (0.04)	−0.01
EU integration should be pushed further	−0.01 (0.05)	−0.01	−0.06 (0.04)	−0.05	−10 ** (0.04)	−0.09
Solidarity between Member States during times of crisis	−0.01 (0.20)	−0.01	0.34 * (0.15)	0.08	0.32 * (0.14)	0.07
Tsipras has good ideas to stimulate economy	0.78 *** (0.16)	0.29	0.95 *** (0.13)	0.34	0.73 *** (0.12)	0.27
Tsipras is trustworthy	0.33 (0.19)	0.12	0.33 * (0.16)	0.10	0.36 ** (0.13)	0.12
I like Tsipras as a person	0.96 *** (0.17)	0.36	1.03 *** (0.14)	0.38	0.85 *** (0.11)	0.34
Tsipras is a strong leader	0.05 (0.14)	0.02	0.10 (0.12)	0.03	0.50 *** (0.11)	0.18
R2 (adjusted)	52.2%		66.0%		61.0%
N	273		366		409

Notes: This table displays unstandardized coefficients with standard errors in parentheses, as well as standardized coefficients. Significance levels: * p < 0.05, ** p < 0.01; *** p < 0.001.

Table 6. Determinants of the general assessment of Alexis Tsipras before and after exposure to the debate and for those who did not know the candidates before the debate (OLS Regressions).

	Before		After		Only Post-Exposure
	B (St. Error)	Beta	B (St. Error)	Beta	B (St. Error)	Beta
Intercept	−4.7 *** (0.96)		−7.3 *** (0.73)		−7.6 *** (0.69)
Age	−0.02 (0.02)	−0.05	−0.01 (0.01)	−0.01	−0.01 (0.01)
Gender (female)	−0.03 (0.24)	−0.01	0.09 (0.19)	0.01	0.23 (0.18)	0.04
Country of Residence (1 = candidate)	−0.10 (0.38)	−0.01	0.56 (0.33)	0.06	No cases
Ideology (L/R)	−0.16 ** (0.06)	−0.14	−0.09 * (0.04)	−0.07	−0.02 (0.04)	−0.01
EU integration should be pushed further	−0.01 (0.05)	−0.01	−0.06 (0.04)	−0.05	−10 ** (0.04)	−0.09
Solidarity between Member States during times of crisis	−0.01 (0.20)	−0.01	0.34 * (0.15)	0.08	0.32 * (0.14)	0.07
Tsipras has good ideas to stimulate economy	0.78 *** (0.16)	0.29	0.95 *** (0.13)	0.34	0.73 *** (0.12)	0.27
Tsipras is trustworthy	0.33 (0.19)	0.12	0.33 * (0.16)	0.10	0.36 ** (0.13)	0.12
I like Tsipras as a person	0.96 *** (0.17)	0.36	1.03 *** (0.14)	0.38	0.85 *** (0.11)	0.34
Tsipras is a strong leader	0.05 (0.14)	0.02	0.10 (0.12)	0.03	0.50 *** (0.11)	0.18
R2 (adjusted)	52.2%		66.0%		61.0%
N	273		366		409

Notes: This table displays unstandardized coefficients with standard errors in parentheses, as well as standardized coefficients. Significance levels: * p < 0.05, ** p < 0.01; *** p < 0.001.

We now shift our focus to those who were only able to assess the candidates after watching the debate. (Table 1). Likeability was the most important factor of general candidate appraisal for those who learned about the candidates via the debate, and strength, trustworthiness, and the candidates’ ideas to stimulate the European economy were in the top 4 factors impacting the general assessment of the five candidates for this group of participants (except for the perception of leadership strength in the case of Martin Schulz) (Table 2, Table 3, Table 4, Table 5 and Table 6).

Hypothesis 3 claimed that post-exposure specific assessments of candidates would be more important for those who did not know them before the debate than for those who did know them. Taken as a whole, these variables were indeed more important in explaining the general assessments uttered by the former group of participants than by the latter in the cases of Schulz, Keller, and Tsipras, but not in the case of Juncker and Verhosftadt (Table 1), which partially confirms this hypothesis. Looking now at specific variables, likeability was more relevant for the post-exposure assessments by those who did not know Martin Schulz before the debate than for those who did know him, but less important for all the other four candidates in this comparison (Table 2, Table 3, Table 4, Table 5 and Table 6). Regarding trustworthiness, again in the case of Schulz, we can see that this variable was more important for this group than for those who knew Schulz beforehand, while for the other candidates, its weight was not considerably different. Leader strength appraisals were, in turn, much more important for those without previous knowledge before the debate than for those who already knew the candidates, except in the case of Martin Schulz. The appraisal of the ideas to stimulate the European economy were, however, equally relevant for both groups in the case of Keller and Verhofstadt and less relevant for those without previous knowledge of the candidates in the case of the other three candidates. At the same time, and corroborating Hypothesis 3, the control variables were less important for the post-exposure perceptions of this group than for those of the participants who knew the candidates before the debate (Table 1).

6. Conclusions

By focusing on a low-intensity campaign and on candidates who were unknown to a sizable percentage of the voters (e.g., Hobolt 2014), this article aimed to contribute to the discussion on televised political debate impacts by assessing the degree to which exposure to a debate activates different criteria for the judgment of candidates, favoring some features over others. This goal was fulfilled by using data from a quasi-experiment conducted in 24 European countries, in which the 15 May 2014 Eurovision debate between the five leading candidates for the EC presidency served the purpose of experimental stimulus.

A first noteworthy result is that the general assessment of the five lead candidates changed considerably after the debate, with some being benefitted (Ska Keller, Martin Schulz, Guy Verhofstadt—interestingly, the three candidates who chose to use English as debating language, instead of their native languages), and others harmed by their performance (Jean-Claude Juncker and Alexis Tsipras). This result stresses that debates do impact candidate assessments. Moreover, except for the Greek candidate, there were no differences between the post-exposure general assessments of candidates from those who knew and who did not know the candidates before the debate. This result suggests that holding previous information does not necessarily shield viewers from debate effects on how positively or negatively they perceive political candidates. The exception is, as mentioned before, Alexis Tsipras—in his case, holding previous information about him arguably buffered the impact of a negative debate performance.

More importantly, were the expected priming effects—a stronger weight of specific candidate assessments on the general image of the candidates after the debate—observed? Generally speaking, such effects seem to be dependent on whether the candidates fared well or not in the debate. For those who were more negatively assessed after the debate (Juncker and Tsipras), there was an increase of the weight of participant characteristics and predispositions, whereas the relative weight of specific candidate characteristics, taken as a whole, remained stable. In contrast, for the candidates whose general appraisal improved after the debate (Ska Keller, Guy Verhofstadt, and Martin Schulz), the pattern observed was that of a modest decrease in the relative weight of participant features and predispositions and an increased role of specific candidate assessments. The main implication of this finding is that positive and negative impacts of watching televised debates on the assessment of political candidates may be linked to different criteria being primed by their performance.

We also expected personality assessments to matter more after the debate than before. This was found to be true first and foremost in the case of likeability assessments, but not in the case of traits such as leadership strength or trustworthiness. An interesting exception is Ska Keller, whose perceptions of strength mattered more after the debate. Since her general appraisal improved considerably after the debate, it is safe to assume that Keller was able to reinstate her leadership strength skills during her performance, effectively combatting possibly harmful gender and age stereotypes (e.g., Johnson et al. 2008; Schubert 1988).

Lastly, for those participants with no previous knowledge on the candidates, specific candidate trait assessments were key factors of their general appraisals of the Spitzenkandidaten, with personal likeability being the most important factor. Interestingly, leadership strength appraisals were more relevant for these participants than for those who already knew the candidates before participating in this study, while personal likeability was slightly less important. This suggests that knowing or not knowing a specific candidate might lead us to value different aspects of his/her personality while watching her/his performance in a debate.

In sum, other than confirming that debate exposure can change the way political candidates are viewed (in both ways), this article reports convincing evidence of priming effects, even though dependent on valence attributed to the candidates’ performance and, to some extent, previous knowledge on the candidates. It therefore shows that debates are important political communication events due to their impacts on how political figures are assessed, but that their specific effects are far from straightforward.