What determines who qualifies?

(1)

What determines who qualifies?

A quantitative study on the presence of first- and second-level agenda setting and issue ownership in the 2020 Democratic primary debates.

Vad avgör vem som går vidare?

En kvantitativ studie av förekomsten av första och andra nivån av

dagordningsteorin samt issue ownership i demokraternas primärdebatter 2020.

Lovisa Boström

Faculty of Humanities and Social Sciences

Media and Communication: Digital Media and Analysis 15 HP

Michael Karlsson 2021-06-03

(2)

Abstract

The purpose of this study is to investigate the presence of first- and second-level agenda setting as well as issue ownership in the 2020 Democratic primary debates and whether there is a relationship between using strategies based on these theories and qualifying for future debates.

The study seeks to answer three research questions:

- What is the relationship, if any, between a candidate whose statements focused primarily on the three issues considered most important by the public according to opinion polls and whether this candidate qualified for future debates?

- How did candidates use frames to redraw the attention of issues?

- What is the relationship, if any, between the extent to which a candidate’s statements discussed

performance issues more than Republican-owned or Democratic-owned issues and whether this candidate qualified for future debates?

The study draws mainly on the first and second level of the agenda setting theory, as well as the theory of issue ownership, and analyzes what issues candidates focus on, what attributes of these issues they emphasize, and whether they discuss performance issues like the economy or foreign policy more than issues owned by either the Republican or the Democratic Party. Through a quantitative content analysis of four candidates’ (Joe Biden, Bernie Sanders, Amy Klobuchar, &

Andrew Yang) statements from three of the eleven primary debates held in the 2020 primary process, the study found no direct relationship between focusing on the public’s three most important issues and qualifying for future debates. Similarly, no such relationship was found between emphasizing certain attributes and qualifying for future debates, although the results suggest that candidates may have benefited from avoiding framing issues economically, which concurs with previous findings (Boydstun, Glazier, & Pietryka, 2013a; Boydstun, Glazier, &

Phillips, 2013) and supports Vavreck’s (2009) theory that insurgent candidates should not

emphasize the economy. Findings also demonstrated the contrasting ways three of the candidates framed the same issues, where Joe Biden and Amy Klobuchar tended to emphasize economic frames when discussing Medicare while Bernie Sanders emphasized effectiveness.

Lastly, the findings support previous research on issue ownership since findings showed that most candidates discussed Democratic-owned issues more than other issues, while the eventual presidential nominee, Joe Biden, overall discussed performance issues more than issues owned by either party. This suggests that focusing on such issues may be beneficial for challenging

candidates during an election cycle where the sitting president has been criticized for not being able to handle the job.

(3)

Thus, no direct relationship could be found in the case of RQ1 or RQ2 but discussing performance issues the most overall may have benefited one candidate, suggesting there is a relationship in the case of RQ3.

Keywords: agenda setting, attribute agenda setting, issue ownership, framing, primary election debates

(4)

Sammanfattning

Syftet med denna studie är att undersöka förekomsten av första och andra nivån av

dagordningsteorin samt av issue ownership i Demokraternas primärdebatter 2020 och huruvida det finns någon relation mellan att använda strategier baserade på dessa teorier och att kvalificera sig för framtida debatter. Studien undersöker tre frågeställningar:

- Vad är relationen, om någon, mellan en kandidat vars uttalanden under debatterna fokuserade främst på de tre frågor som väljarna ansåg var viktigast enligt opinionsundersökningar och huruvida denna kandidat kvalificerade sig för framtida debatter?

- Hur använde kandidaterna ”frames” för att kontrollera diskussionen kring frågor?

- Vad är relationen, om någon, mellan den utsträckning en kandidats uttalanden under debatterna diskuterade så kallade ”performance issues” mer än frågor ägda av det republikanska eller demokratiska partiet och huruvida denna kandidat kvalificerade sig för framtida debatter?

Studien bygger huvudsakligen på den första och andra nivån av dagordningsteorin, liksom teorin om issue ownership, och analyserar vilka frågor kandidaterna fokuserar på, vilka attribut de betonar när de talar om dessa frågor och om de diskuterar performance issues såsom ekonomi eller utrikespolitik mer än frågor som ägs av antingen republikanska eller demokratiska partiet.

Genom en kvantitativ innehållsanalys av fyra kandidaters (Joe Biden, Bernie Sanders, Amy Klobuchar och Andrew Yang) uttalanden från tre av de elva primärdebatterna som hölls under primärprocessen 2020 fann studien ingen direkt relation mellan att fokusera på de tre frågor som väljarna ansåg var viktigast och att kvalificera sig för framtida debatter. Det hittades inte heller någon sådan relation mellan att betona vissa attribut och att kvalificera sig för framtida debatter, även om resultaten tyder på att kandidater kan ha haft nytta av att undvika att betona ekonomiska attribut, vilket överensstämmer med tidigare resultat (Boydstun, Glazier, & Pietryka, 2013a;

Boydstun, Glazier, & Phillips, 2013) och stöttar Vavrecks (2009) teori att så kallade ”insurgent candidates” drar nytta av att inte diskutera ekonomin mer än nödvändigt. Resultaten visade också hur olika kandidaternas inramning av en specifik fråga var, då Joe Biden och Amy Klobuchar hade en tendens att betona ekonomiska attribut när de talade om frågor gällande Medicare medan Bernie Sanders fokuserade mer på effektivitetsattribut.

Slutligen stöder studien tidigare forskning om issue ownership då resultaten visade på att de flesta kandidater diskuterade frågor ägda av det demokratiska partiet mer än andra frågor, medan den kandidat som slutligen skulle få det demokratiska partiets presidentsnominering, Joe Biden, totalt sett diskuterade performance issues mer än frågor ägda av något av partierna. Detta tyder på att

(5)

ett fokus på sådana frågor kan vara till nytta för att utmanande kandidater under en valcykel där den sittande presidenten har kritiserats för sin hantering av arbetet.

Således kunde ingen direkt relation hittas när det gällde RQ1 eller RQ2, men resultaten tyder på att en kandidat kan ha gynnats av att diskutera performance issues mest över lag, vilket i sig tyder på att det finns en relation gällnade RQ3.

Nyckelord: dagordningsteorin, andra nivån av dagordningsteorin, framingteorin, issue ownership, amerikanska primärdebatter

(6)

Acknowledgements

I want to thank my supervisor Susanne Almgren for her advice and feedback as well as valuable discussions throughout the writing process.

(7)

List of tables

Table 1 - A measure of which party the public thought could do a better job at different issues ... 36

Table 2 - A measure of how much trust the public has in Democrats in Congress to handle issues .... 36

Table 3 – Issue salience compared to debate content measured in percentage ... 42

List of figures

Figure 1 - Percentage of candidates’ statements during debate 2 that focused on the public’s three most salient issues (%) ... 43

Figure 4 - Relation between candidates and issues during debate 2 ... 45

Figure 7 - Percentage of frames emphasized by each candidate during debate 2 (%) ... 48

Figure 10 - The relationship between issues and frames used by Biden in debate 2 ... 50

Figure 13 - The relationship between issues and frames used by Sanders in debate 2 ... 51

Figure 16 - The relationship between issues and frames used by Klobuchar in debate 2 ... 52

Figure 19 - The relationship between issues and frames used by Yang in debate 2 ... 53

Figure 20 - The relationship between issues and frames used by Yang in debate 4 ... 53

Figure 21 - Candidates’ statements divided by issue ownership during debate 2 (%) ... 54

Figure 24 - Candidates’ statements divided by issue ownership during all three debates (%) ... 56

(10)

10

1.0 Introduction

Televised debates have become a tradition in American politics since the first one was held in 1960 (McKinney & Carlin, 2004). It gives politicians a strategic opportunity to engage and inform the audience on their policy positions and plans while simultaneously trying to outperform their opponents. However, since these debates are almost always hosted by news or media outlets, it is important to note that media plays a large role in politics, both as an arena for political debate and as political actors (Strömbäck, 2009).

Strömbäck (2009) argues that because people today often rely on media to keep them informed on political and societal issues rather than actively seeking out information on their own, modern politics have become mediated. News media can thus influence what issues become more or less salient in the minds of the public (McCombs & Shaw, 1976) by giving more attention to a certain issue in news broadcasts. In essence, televised debates are no different. Hosted by journalists, televised debates are not simply an arena for candidates to discuss their political platforms, but an event where predetermined questions shape the debate structure, and by extension, the debate agenda (Benoit & Hansen, 2001; Stromer-Galley & Bryant, 2011).

There are several ways in which candidates can still try to control the debate agenda, to some extent, by straying from issues to focus more on topics the public deems important (Boydstun, Glazier, & Pietryka, 2013a), framing issues in a more advantaged way (Boydstun, Glazier, &

Phillips, 2013; Jerit, 2007), or discussing issues where the candidate or the party is at an advantage (Petrocik, 1996; Benoit & Hansen, 2004; Damore, 2004). This study thus aims to analyze how candidates attempt to control the agenda through first- and second-level agenda setting as well as issue ownership.

1.1 Research problem

When analyzing how candidates use agenda-setting strategies and framing to try to control the debate agenda, previous research has mainly focused on presidential debates where one candidate from either party faces one another. This study aims to expand this area to primary debates, as candidates are likely to attempt to shift the discussion to more advantaged topics or frame issues in more advantaged ways here as well. The research on issue ownership has similarly focused more on the general election than on the primary process, and it would therefore be interesting to see how issue ownership appears during these debates where all candidates belong to the same party and thus own the same issues.

(11)

11

1.2 Purpose

The purpose of this study is to examine the presence of first- and second-level agenda setting, as well as issue ownership, in the Democratic primary debates leading up to the 2020 election. There are a lot of factors that come into play during the primary election process, and this study will focus only on a few of those aspects. Namely, the content of candidates’ statements during these debates and how it coincides with the debate qualification process. It is, however, important to note that what is said during debates does not alone determine who qualifies since there are requirements each candidate needs to meet before a debate to even be invited. These

requirements, as well as moderator and question influence, will not be considered in this study but will be explained further in this chapter.

1.3 Research questions

With the research problem and purpose of this study in mind, these are the research questions the study aims to answer.

RQ1: What is the relationship, if any, between a candidate whose statements focused primarily on the three issues considered most important by the public according to opinion polls and whether this candidate qualified for future debates?

RQ2: How did candidates use frames to redraw the attention of issues?

RQ3: What is the relationship, if any, between the extent to which a candidate’s statements discussed performance issues more than Republican-owned or Democratic-owned issues and whether this candidate qualified for future debates?

1.4 Definitions

Presidential debates or general election debates to the debates sponsored by the

Commission on Presidential Debates (CPD) and are held prior to the general election during every election cycle. The CPD is a nonpartisan, independent organization whose mission is to provide the public with structured debates that may help them get a better idea of the candidates’

policy positions (Commission on Presidential Debates, n.d.). These debates will in this study be referred to as presidential debates and general election debates interchangeably.

Primary debates are debates organized by each party and are held during the primary process leading up to the general election. Unlike general election debates, these are not sponsored by the CPD but by either the Democratic National Committee (DNC) or the Republican National

(12)

12 Committee (RNC) and only invite candidates from their respective parties. Since there most often are more candidates in the running during the primary process, there tend to be more primary debates than general election debates to help winnow the field (Azari & Masket, 2019).

First-level agenda setting, perhaps more commonly known simply as agenda setting, was conceptualized by McCombs and Shaw in 1976 and refers to the media’s ability to influence public opinion. Since the first and second level of agenda setting will be examined separately, the study distinguishes between them to avoid any confusion. The first level will be explained further in section 2.2.

Second-level agenda setting, also called attribute agenda setting, was later added as an

extension to the traditional theory. This level examines the media’s ability to influence how people think of issues by emphasizing different attributes (McCombs, 2014). The second level of agenda setting and its convergence with the framing theory will be explained further in section 2.3 and 2.3.1.

Attributes are, in the theory of agenda setting, the characteristics of an object. In other words, they are the way an object is characterized by, for instance, journalists or the media; the aspects that are emphasized when discussing a certain topic, or object (McCombs, 2014). Frames are used to highlight certain aspects of an object that communicates something, thereby making it more salient (Entman, 1993). These terms will be used interchangeably in the study.

Issues and topics will also be used interchangeably in this study and refer to the content of the candidates’ statements, i.e., what issue or topic the candidate is discussing.

1.5 Delimitations

This study will analyze whether there is a relationship between different agenda-setting strategies as well as focusing on performance issues and whether a candidate who uses these qualifies for future debates. There were over twenty candidates from the Democratic Party that ran for president during the 2020 election and a total of eleven primary debates. Since including all of these in the study would not be possible, the sample was limited to four candidates and three debates. A broader analysis that included all candidates and debates could possibly have yielded more detailed results.

(13)

13

1.6 Disposition

Altogether, this study consists of seven chapters. This first chapter contained an introduction of the study, the research problem, research questions, and some background on the primary

election process as well as the candidates included in the study. The second chapter will introduce the theoretical framework and go through previous research on which this study builds. The third chapter contains a description of the methodology used in the study, quantitative content

analysis, as well as methodological problems, the empirical material and data collection, and the variables and the operationalization of these. The fourth chapter provides the results the study yielded, presented by tables and figures to demonstrate the findings in an understandable way.

Chapter five consists of the discussion about the previous chapter’s findings and discusses these in light of previous research. The study is concluded in the sixth chapter, followed by suggestions on future research, after which the seventh and last chapter discusses the study’s implications for society and a few of its shortcomings.

(14)

14

2.0 Background

In this chapter, the background of the American primary election process will be explained, followed by a brief description of the four candidates analyzed in this study and their key issues.

2.1 Background on the primary election process

Presidential primary debates are not entirely unlike general election debates, but whereas the latter are organized by the Commission on Presidential Debates (CPD), primary debates are organized by each respective party (Bill of Rights Institute, n.d.). A party’s presidential nominee—i.e., the candidate who eventually runs for office against the opposing party’s candidate—is chosen through primary elections and caucuses. This process is quite complex as there are a lot of steps to go through before a candidate is nominated (Epstein, 2020).

2.1.1 Primaries

Primary elections are organized by state governments around the country, and most often only members of each respective party are allowed to vote in these elections, though some tend to have open primaries (Ballotpedia, n.d.-b). The vote is cast by a secret ballot, similar to the general elections, and the number of votes a candidate gets decides how many delegates the candidate wins. Thus, the winner of the primary election will have the most delegates in that state (BBC News, 2020). Primary elections are held to allow the party’s members to vote for the candidate they would like to see as the presidential nominee in the general election.

2.1.2 Caucuses

A caucus, however, is an event organized by each party where groups of people come together to discuss whom they think should be that party’s presidential nominee (Bill of Rights Institute, n.d.). These are not state-administered, and thus the voting at the end of the event tends to be conducted by a headcount or a raise of hands (Ballotpedia, n.d.-a), although this differs from state to state. Caucuses are less common than primary elections today but are still organized in a few states at the start of every primary process (Ballotpedia, n.d.-a). The Iowa caucuses are known to hold significant power over the primary process as they are the first state to vote, and candidates who do well in this state tend to gain momentum and media attention that can help them win the party nomination (Epstein, 2020). The actual presidential nomination occurs at each party’s convention which is an event organized by the party where the delegates vote according to the primary election results (BBC News, 2020).

(15)

15 2.1.3 2020 primary debate rules and requirements

The DNC released the framework for the 2020 primary debates in late 2018 (Democratic

National Committee, 2018). About two months later, further details on the first two debates were released, along with the announcement of its media partners and the threshold for participation (Democratic National Committee, 2019). To qualify for the first two debates, candidates needed to either get “1% or more support in three polls” (either in the four states that vote early or national polls) or receive “at least (1) 65,000 unique donors; and (2) a minimum of 200 unique donors per state in at least 20 U.S. states” (Democratic National Committee, 2019). These requirements were considered rather low which in turn led to a wide field of candidates at the start of the race, but as the threshold got higher, fewer candidates were able to meet the

requirements. Prior to the tenth debate, the DNC retracted the donation threshold after receiving criticism that the amount of money raised does not signify electability (Burns, 2020).

2.2 Background on the 2020 Democratic candidates

This section will provide a brief description of the four candidates included in this study as chosen through stratified sampling (see section 4.3). Candidates’ key issues are collected from what is featured on their official 2020 campaign websites.

2.2.1 Joe Biden

Joe Biden was Vice President between 2008 and 2016 under Barack Obama’s administration and was elected President in the election of 2020 (The New York Times, n.d.). Biden announced that he was running for president on April 25, 2019 (Biden, 2019; Burns & Martin, 2019).

According to Biden’s official 2020 campaign website, some of his key issues are rebuilding the middle class by ensuring tax cuts for working families while raising taxes for the wealthy and larger corporations, the opposite of what Trump did while in office, Biden argues (Joe Biden for President, n.d.); building the American economy back better which means not simply going back to how things were but emphasizes economic equality; and tackling climate change by building a sustainable, resilient economy and infrastructure that sets America on a course to reach net-zero emission by 2050.

(16)

16 2.2.2 Bernie Sanders

Bernie Sanders has been a United States senator for the state of Vermont since 2007, and prior to that, he served in the House of Representatives for 16 years (Sanders, n.d.). Sanders first

announced he was running for president on a Vermont radio station on February 19, 2019 (Kinzel & VPR News, 2019), following his run against Hillary Clinton in 2016.

Sanders is known for his social-democratic approach to politics (Frizell, 2015). His key issues during his 2020 campaign were, among others, Medicare for All, an act that ensures universal health care and effectively eliminating private health insurance; dealing with the climate crisis by enacting the Green New Deal Act which emphasizes reaching “100 percent renewable energy for electricity and transportation by no later than 2030 and complete decarbonization of the

economy by 2050 at latest” (Friends of Bernie Sanders, n.d.-c); and canceling all student debt as well as making all public universities and colleges tuition-free.

2.2.3 Amy Klobuchar

Amy Klobuchar is a United States senator for the state of Minnesota (Klobuchar, n.d.), and announced her campaign to run for president on February 10, 2019 (Montoya-Galvez & Segers, 2019). Some of her key issues are to ensure affordable and universal health care by expanding Medicare (though not enacting Medicare for All), Medicaid, and the Affordable Care Act, as well as lowering prescription drug costs; combatting climate change, and reaching net-zero emission by 2050 by, among other things, promoting homegrown renewable energy; and protecting

America’s democracy which emphasizes the protection of civil rights, criminal justice reform, and restoring the public’s trust in the American government.

2.2.4 Andrew Yang

Andrew Yang is an American entrepreneur with no prior political experience, though he was named a Champion of Change by Barack Obama’s administration in 2012 (Carson, 2012), and a Presidential Ambassador for Global Entrepreneurship in 2015 (Brayton, 2015). Yang announced he was running for president in late 2017 (Smith, 2019), and is most known for his policy on social security, more specifically, that he wants to implement a universal basic income of $1.000 a month (Friends of Andrew Yang, n.d.-c). Some of Yang’s other key issues are reforming the economic system to what he calls “human-centered capitalism” (Friends of Andrew Yang, n.d.-a), and ensuring universal health care by focusing on the core issue: the high cost of prescription

(17)

17 drugs. Yang has said that he supports “the spirit of Medicare for All” but does not believe that eliminating private health insurance is a realistic strategy (Friends of Andrew Yang, n.d.-b).

(18)

18

3.0 Theoretical Framework

This chapter will introduce the theoretical framework used in this study. It will describe the different theories this study builds, namely agenda setting, attribute agenda setting, framing, and issue ownership.

3.1 A history of debate research

The first televised US election debate took place in 1960 between the then-incumbent Vice President Richard Nixon and the Democratic nominee John F. Kennedy (McKinney & Carlin, 2004). After 1960, televised debates did not occur again until 1976, when then-incumbent President Gerald Ford met Democrat Jimmy Carter in three debates, and since then, they have become a tradition in every election cycle (McKinney & Carlin, 2004).

Due to the vast number of citizens who tune into debates, researchers have concluded they merit scholarly investigation (Benoit & Hansen, 2001). This has resulted in a plethora of studies and books on the subject, most of which seek to examine debate effects and whether debates matter at all (McKinney & Carlin, 2004). But many studies have investigated different aspects as well.

Candidates and their strategies (Boydstun, Glazier, & Pietryka, 2013a; Boydstun, Glazier, &

Phillips, 2013; Bilmes, 1999), rhetoric (Conway et al., 2012; Demeter, 2017), and character traits (Demeter, 2017) have all been examined. As have moderators and their influence over the debate agenda through the questions they ask (Bechtolt, Hilyard, & Bybee, 1977; Benoit & Hansen, 2001; Stromer-Galley & Bryant, 2011).

According to McKinney and Warner (2013), there is substantial empirical evidence that debates do, more likely than not, matter. Due to this, more recent analyses have moved beyond the question of if to consider, instead, how they matter (McKinney & Warner, 2013). From previous research, we know, for instance, that televised debates have the largest influence on politically interested viewers who have yet to decide whom to vote for, since they are likely watching to learn more about the candidates and their policy positions; less impact on those just somewhat interested in politics and already committed partisans; and least influence on uninterested voters, since they are likely not to tune in at all (McKinney & Carlin, 2004).

Other studies have found that effective use of political strategies like focusing on the public’s issue agenda (Boydstun, Glazier, & Pietryka, 2013a) and drawing attention to issue frames where the candidate is at an advantage (Boydstun, Glazier, & Phillips, 2013) can increase a candidate’s performance. Rhetorical strategies like interruptions and attacks are frequently employed by candidates (Bilmes, 1999; Hinck et al., 2018). Both Bilmes (1999) and Demeter (2017) noted that

(19)

19 the Republican candidate in their respective election process was more aggressive than their Democratic counterpart, and Hinck et al. (2018) even stated that Donald Trump had been the most aggressive candidate when compared with similar studies they had previously conducted.

Researchers from the media and communications field have implemented different theories in such studies, among them agenda setting (McCombs & Shaw, 1972), framing (Entman, 1993;

Scheufele, 1999; Ardèvol-Abreu, 2015), priming (Sheafer, 2007), and issue ownership (Petrocik, 1996; Benoit & Hansen, 2004; Damore, 2004). Below, the theories this study draws upon will be introduced and some of the existing literature in the field will be reviewed further.

3.2 Agenda setting

Agenda setting was conceptualized by McCombs and Shaw in 1972, in their study “The Agenda Setting Function of Mass Media” and has since been revised and implemented by its creators and several other researchers alike (McCombs, 2014). Though not formally named prior to McCombs and Shaw’s study, the concept was already a topic of discussion among some scholars. One of whom was Bernard Cohen, who in his book Press and Foreign Policy wrote the now widely quoted

“[the press] may not be successful much of the time in telling people what to think, but it is stunningly successful in telling its readers what to think about” (Cohen, 1963, p. 13), which is about as close to agenda setting you can get without explicitly naming the theory. But even before McCombs and Shaw’s (1972) study and Cohen’s (1963) book, the notion that public opinion is influenced by what we read or what we are told was considered by Walter Lippman (1922) in the first chapter of his book Public Opinion. Cohen (1963) built on Lippman’s work and noted that the world, based on the pictures in one’s head (Lippman, 1922), will not look the same to all people, but is rather influenced by the news we consume and that that which the news does not cover is essentially nonexistent to us.

McCombs and Shaw’s (1972) study examined whether the mass media influenced not yet fully decided voters and the issues they considered most important. 100 participants in Chapel Hill, North Carolina were interviewed over the course of eighteen days and, during the same time, content from various mass media was collected to see if the participants’ issue salience coincided with media coverage. Results showed a strong relationship between the two. Since then,

numerous studies on agenda setting and agenda control have been conducted, and the general consensus seems to be that media coverage does, indeed, influence issue salience in voters (McCombs & Shaw, 1972; Benoit, McKinney, & Holbert, 2001).

(20)

20

3.3 Second-level agenda setting

Second-level agenda setting—or attribute agenda setting—concerns the attributes that frame a certain issue (McCombs, 2014). While first-level agenda setting emphasizes issue salience, the second level focuses on attribute salience, and, by extension, how we think about issues. If the news focuses primarily on health care, we are more likely to consider this an important issue (first level). If the news focuses primarily on the cost to implement health care plans, however, we are more likely to see the issue from an economic perspective (second level). McCombs (2014) argues that because of this second level of agenda setting, Cohen’s (1963) quote, as mentioned earlier, may need some revising since a consequence of media telling us what attributes to see an issue through (or how to think of about an issue) might lead to news effectively telling us what to think as well.

Moreover, McCombs (2014) argues that certain attributes—or compelling arguments—do not only increase attribute salience but also help increase the salience of an issue. This is to say, traditional agenda setting is influenced both by how frequently it is covered in the news and by which attributes are emphasized. In the case of candidate salience, McCombs (2014) mentions that here, too, are certain attributes more likely to influence public opinion, namely, moral quality and leadership.

Moon (2011) argues that, among voters, cognitive effects are not the only aspects of agenda setting worth scrutiny, but that attitudinal and behavioral effects are just as important to note.

Behavioral effects of agenda setting have so far, according to Moon (2011), been restricted to voting, but there are other ways in which people can be politically engaged.

3.3.1 Second-level agenda setting v. Framing

There is an ongoing discussion within media and communication studies about whether framing should be included under the agenda-setting umbrella or be regarded as its own theory. Framing theory, which has its roots in psychology and sociology (Pan & Kosicki, 1993), is another media effects theory that builds on the presumption that how an issue is characterized, or framed, can influence how audiences understand said issue (Scheufele, 1999). However, framing is, as Entman (1993) puts it, a “scattered conceptualization” (p. 51) with different definitions that leave a lot of room for confusion (Kensicki, 2000). For instance, Entman (1993) considers frames as an aspect of an object, while Tankard et al. (1991, as cited in McCombs, Lopez-Escobar, & Llamas, 2000), defines frames as a central theme. Thus, the definition one employs depends on whether frames

(21)

21 are aspects of something else (or attributes, as within second-level agenda setting) or the general focus.

Due to this inconsistency in definition, McCombs (2014) means that it is at times difficult to distinguish framing from agenda setting. Sometimes the two are used synonymously; other times they are considered at odds with one another. McCombs and Ghanem (2001) have, therefore, argued that both theories could benefit from converging with one another.

To Entman (1993), who describes these aspects (frames) as promoting “a particular problem definition, causal interpretation, moral evaluation, and/or treatment recommendation” (p. 52), framing is a process that involves selection and salience. Whether a newsagent, for instance, decides to report on an event as well as how they report on it are both parts of this process and can, as stated earlier, influence the way an audience thinks about this event. Though this is undeniably similar to agenda setting, particularly the second level, other scholars argue that the two theories are conceptually different and should therefore be kept apart (Scheufele & Tewksbury, 2007), so as not to “further complicate the distinction between loosely defined concepts” (Ardèvol-Abreu, 2015, p. 426).

A common reasoning behind this is that while framing concerns applicability, agenda setting is grounded in accessibility (Scheufele, 1999; Scheufele & Tewksbury, 2007; Ardèvol-Abreu, 2015).

Accessibility, here, means that more salient issues—and, by extension, more accessible issues—

are more likely to influence public opinion than issues given less attention. This goes for attributes as well, as Scheufele (1999) argues “the frames that are most accessible are the ones that are most easily available and retrievable from memory” (p. 116). In other words, the more accessible an issue or an attribute is, the more likely is it to be remembered by the public which, in turn, means it is more likely to have an influence on the public agenda (Scheufele &

Tewksbury, 2007).

Applicability, on the other hand, refers to “the outcome of a message that suggests a connection between two concepts such that, after exposure to the message, audiences accept that they are connected” (Scheufele & Tewksbury, 2007, p. 15). Framing is, unlike agenda setting, not

concerned with the amount of attention an issue or its attributes receives (Ardèvol-Abreu, 2015), but how an issue is reported or discussed and focuses on the perceived importance of a frame, which entails “a more conscious process of information gathering and processing” (Scheufele, 1999, p. 116).

(22)

22 Due to these conceptual differences, Scheufele and Tewksbury (2007) mean that second-level agenda setting is only effective if the audience is already familiar with the issues or attributes at hand whereas framing need not rely on this memory-based requirement. However, McCombs (2014) argues that this line of reasoning has little empirical evidence.

Since this study seeks to examine neither the effects nor the perceived importance of frames, but rather how different frames are employed by different candidates, second-level agenda setting seems more apt. To be clear, the study acknowledges the conceptual differences laid forth by scholars but will use the terms attributes and frames interchangeably since the agenda-setting effects are not analyzed, but rather the ways in which issues are emphasized by the candidates.

3.4 The role of agenda setting in political debates

As mentioned earlier, the general consensus among scholars seems to be that agenda setting does indeed influence public opinion (McCombs & Shaw, 1972; Benoit, McKinney, & Holbert, 2001).

It has also been concluded by McKinney and Carlin (2004) that citizens want moderators to ask questions on topics deemed most important by the public during debates. Considering this, one would assume that debate moderators and presidential candidates alike would strive to inform voters of their positions on the most salient issues according to the public. However, scholars have found that this is not always the case (Benoit & Hansen, 2001; Stromer-Galley & Bryant, 2011).

3.4.1 Agenda control

Benoit and Hansen’s (2001) analysis of moderator questions throughout the years, for instance, found that the questions did not reflect the public’s issue agenda. Similarly, Stromer-Galley and Bryant (2011) found that journalists’ questions during the 2008 debates hosted by CNN and YouTube focused more on issues like foreign policy and the economy which did not reflect the public’s interest. During these debates, citizens were able to submit their own questions by uploading them to YouTube with the possibility that they would be posed to the participating candidates. However, as the study also showed, CNN chose questions that focused

disproportionally on issues as submitted by voters. This concurs with Benoit and Hansen’s (2001) findings that moderators do not reflect public opinion but rather focus on their own (or the news) agenda. It is, therefore, important to note that moderators are equally—if not more—likely to try to control the debate agenda as the participating candidates are (Bechtolt, Hilyard, &

Bybee, 1977; Benoit & Hansen, 2001; Stromer-Galley & Bryant, 2011).

(23)

23 In their study, Stromer-Galley and Bryant (2011) also make the distinction between agenda setting and agenda control that seems to often be overlooked in research. Agenda control is, by their definition, “the struggle over [the agenda] between multiple actors on the political stage” (p.

534), rather than a synonym for agenda setting, which refers more broadly to how issues that get media and elite attention become more salient to the public than issues that receive less or no attention (McCombs & Shaw, 1972; Boydstun, Glazier, & Pietryka, 2013a). While Bechtolt, Hilyard, and Bybee’s (1977) study on the 1976 presidential debates made no explicit distinction between agenda setting and agenda control, it focused more on whether the candidates (Gerald Ford and Jimmy Carter) or the questioners controlled the content of the three debates. Verbatim transcripts were analyzed and coded as “topic initiation” and “time spent on topic” (p. 676) to find who initiated the most topics and how much uninterrupted time was subsequently spent on that topic. The results showed that, overall, Carter exerted the most control over debate content as the time spent on his initiated topics outperformed Ford. However, an interesting finding in this study is the fact that out of 253 minutes of discussion in total, questioners initiated 130 minutes which amounts to just slightly over 50%, suggesting that moderators actually controlled debate content more than either candidate did.

Agenda control during the 1992, 2004, and 2008 presidential debates was examined by Boydstun, Glazier, and Pietryka (2013a) who analyzed two rhetorical strategies—playing to the crowd and heresthetics—and how candidates used them. Playing to the crowd means, quite understandably, to focus on issues that the public considers important whereas heresthetics indicates focusing on issues where the candidate is at an advantage. Agenda setting, here, functions as a political tool candidates can use to shift the attention to more favorable issues. By using these strategies and tools effectively, candidates can gain more power over the debate agenda than they previously had since the ones controlling debate content, in large part, are moderators (Benoit & Hansen, 2001; Stromer-Galley & Bryant, 2011).

Boydstun, Glazier, and Phillips (2013) refer to Vavreck’s (2009) theory of the importance of the economy during the election cycle. According to Vavreck (2009), the current state of the

economy can dictate whether a candidate will, or rather should, emphasize the issue in their campaigns. The theory suggests that an incumbent candidate in a good economy or a challenging candidate in a bad economy will benefit from discussing the topic in their campaign messages or during debates. These are clarifying candidates, i.e., ones that only need to clarify their relation to the good economy, or their lack of involvement in creating bad economic times to be favored by the public. Insurgent candidates are the opposite of clarifying candidates, and they benefit from avoiding the topic of economy in their campaign messages. These candidates are either

(24)

24 incumbents in a bad economy or challenging candidates in a good economy. Boydstun, Glazier, and Phillips’ (2013) analysis of the 2008 general election debates supported Vavreck’s (2009) theory when the study demonstrated that the insurgent candidate, John McCain, tended to frame issues economically even though this did not benefit him but rather his opponent, Barack

Obama.

3.4.2 Framing the issue

Despite the scholarly disagreement concerning second-level agenda setting and framing, many scholars agree that political actors frame issues to their advantage in order to redraw the issue focus and win policy debates (Jerit, 2007; Boydstun, Glazier, & Phillips, 2013). One such study, conducted by Boydstun, Glazier, and Phillips (2013), found that both candidates during the 2008 presidential debates used framing to emphasize their advantaged issues, although one of them, John McCain, was less strategic in his framing.

On the whole, issue framing is a way for political actors to convey information to the public in a more understandable way (Lachapelle, Montpetit, & Gauvin, 2014). Issue framing is particularly imperative during primary campaigns as candidates from the same party are competing over frames (Rich, 2018). In 2016, Bernie Sanders became an unexpectedly tough competitor to front runner Hillary Clinton, and by framing issues like health care and education in a more social- democratic way, Sanders may have affected the party establishment’s issue framing to lean more left despite not winning the presidential nomination (Rich, 2018).

In a study on the policy debate surrounding Bill Clinton’s health reform of 1993, Jerit (2007) found that supporters of the bill were more effective at engaging the public than they were at framing the issue. Jerit (2007) reasons that this was likely due to there already being a lot of preexisting opinions about health care as an issue and that framing is more effective on issues where preexisting opinions are less prevalent among people. Similarly, some members of the public may be more easily persuaded by a certain issue frame due to their already existing worldview, while people with a different worldview may be more difficult to influence

(Lachapelle, Montpetit, & Gauvin, 2014). People are often more likely to rely on their own beliefs and opinions than they are on experts in that field, even when factual information is provided (Druckman & Bolsen, 2011). In other words, it is difficult to change public opinion where preexisting beliefs are strong and widespread, even with substantial facts.

(25)

25

3.5 Issue ownership

The term issue ownership was introduced in 1996 by John R. Petrocik but had already been theorized by Budge and Farlie in 1983. Petrocik (1996) tested the theory in a case study about issue ownership in the 1980 election campaigns of then-incumbent Jimmy Carter and challenger Ronald Reagan. Petrocik (1996) builds the theory on the basis that the two major parties in the United States, the Republican Party and the Democratic Party, “own” different issues and, therefore, tend to favor these more than the opposing party’s owned issues. The Republican Party owns issues such as national defense and crime, while the Democratic Party owns issues like health care, education, and social security (Benoit & Hansen, 2004; Petrocik, 1996; Petrocik, Benoit, & Hansen, 2003). Because of these ownerships, presidential candidates have to strategize to appeal not only to their own supporters but to the electorate as a whole. The strategies differ depending on the current standing a candidate holds, for instance, a candidate who is trailing behind in public opinion polls might trespass onto issues owned by the opposing party (though with different frames) to showcase their own credibility on this issue (Damore, 2004). Not unlike issue trespassing, Banda (2013) means that candidates can emphasize their opponents’ issues to create a dialogue and appeal to median voters while making certain to emphasize the issue, not the opposing party’s opinions. This is, however, done at some risk as diverging from the party’s platform can be viewed negatively by partisan voters and can potentially only garner more

attention to the opposing party and thus reinforce already existing voter perceptions of their issue ownership (Banda, 2013; Lefevere, Tresch, & Walgrave, 2015). A candidate who is leading in public opinion polls, however, will be less inclined to change their message and discuss the opposing party’s issues (Damore, 2004). It should be noted that both Damore (2004) and Banda (2013) studied issue ownership and issue convergence in general election debates where a

candidate from either party is involved. Since this study looks only at one party, the tendencies to trespass onto or converge with the opposing party’s issue agenda may look different.

As pointed out by Lefevere, Tresch, and Walgrave (2015), issue ownership remains rather ambiguous in terms of its definition and conceptualization. They mean that there are two different dimensions—associative issue ownership and competence issue ownership—to the theory that need to be distinguished from one another. Associative issue ownership refers to when a party’s ownership of an issue stems from a history of attention being devoted to the issue, whereas competence issue ownership refers to which party is considered to “best handle” an issue.

However, Lefevere, Tresch, and Walgrave (2015) also note that these dimensions are more likely to overlap in countries with a two-party system than nations with multi-party systems.

(26)

26 Although the respective parties own different issues, there are times where a candidate (or a party) can lease an issue, often due to the sitting President’s inability to “handle the job” (Petrocik, p. 827). These issues are called performance issues, and include, for instance, the economy and foreign policy as these can be subject to change quite rapidly. If inflation occurs when a Democrat is holding office, a Republican candidate can temporarily gain ownership over the economy. A good example of when such issues were widely considered salient among the public and thus favored the challenging candidate is Petrocik’s 1980 case study where then-incumbent Jimmy Carter eventually lost to Republican candidate Ronald Reagan, who managed to appeal to these voters by focusing on performance issues.

Scholars have also demonstrated that issue ownership takes a slightly different shape in primary campaigns compared to general election campaigns (Petrocik, 1996; Benoit & Hansen, 2004). In a study conducted by Benoit and Hansen (2004), findings showed that Democratic candidates were likelier to discuss their own issues more in primary debates than in general election debates, possibly to try to appeal to the broader audience that normally tunes in to these debates.

Republican candidates, however, discussed their own issues about as much during both primary and general election debates. Benoit and Hansen (2004) suggest that this might be because Republican-owned issues tend to be more national than issues owned by the Democratic Party.

This shows that Democratic candidates seem more inclined—if not required—to adapt the issues they discuss during debates to correspond with the agenda of the electorate as a whole.

3.6 Summary of the theoretical framework

As has been demonstrated by previous research, agenda setting is not only a process where issues become more salient through extensive news and media attention (McCombs & Shaw, 1972), but can also be a powerful tool that political actors can use to influence the public agenda (Boydstun, Glazier, & Pietryka, 2013a; Boydstun, Glazier, & Phillips, 2013). Boydstun, Glazier, and Pietryka (2013a) concluded that candidates benefited both from playing to the crowd by focusing on issues the public deems salient, and from focusing on issues where the candidate is at an

advantage. However, while candidates tend to hold some power over debate content, the agenda is most often set by the ones asking the questions—moderators (Benoit & Hansen, 2001)—and as previous research has shown, these questions do not always reflect the public’s issue agenda (Stromer-Galley & Bryant, 2011). Candidates can stray from topics, but they do so at the risk of seeming evasive, and thus, as Boydstun, Glazier, and Pietryka (2013a) note, this tool to control the agenda needs to be employed with good judgment. Therefore, this study looks at how

(27)

27 candidates played to the crowd (first-level agenda setting) during the 2020 primary debates and aims to examine whether there was a relationship between doing so and qualifying for future debates.

Moreover, studies like Boydstun, Glazier, and Phillips’ (2013) are interesting as they demonstrate how candidates use framing as a tool to control the debate agenda. It suggests that controlling the issues of discussion is not always sufficient, but that controlling the way these issues are discussed is necessary as well. In general election debates, this can become rather apparent as the respective party has “ownership” of different issues (Petrocik, 1996; Petrocik, Benoit, & Hansen, 2003;

Damore, 2004). The Republican Party, for instance, tends to own issues related to national defense and the budget deficit while the Democratic Party tends to own issues related to health care and the environment. It might therefore be interesting to consider the second level of agenda setting during primary debates where all candidates belong to the same party. Since a lot of candidates tend to run on similar policy platforms, the question arises: how are issues framed to only their own advantage? To answer this question, a broader, more thorough analysis than this study provides of the candidates as well as their campaigns would be necessary. Instead, this study aims to look at how four different candidates emphasized different attributes (second-level agenda setting) to draw or redraw the attention of an issue to a more beneficial issue frame.

Furthermore, previous research has established that issue ownership takes a slightly different shape in the primary process compared to the general election since each party tends to focus more on their own issues (Benoit & Hansen, 2004). However, it has also been demonstrated that challenging candidates can benefit from directing more attention to performance issues if the sitting President is generally considered to be doing a poor job (Petrocik, 1996). As Vavreck (2009) theorized, however, and as Boydstun, Glazier, and Phillips (2009) noted in their study, a challenging candidate in a good economy would not benefit from discussing the economy, which is most often considered a performance issue (Petrocik, 1996; Petrocik, Benoit, & Hansen, 2003).

This puts the 2020 Democratic candidates in a disadvantaged position as they were the insurgents during a relatively stable economy (BBC News, 2019; Kurtz & Yellin, 2020)—even though some have argued that Trump inherited this economy from Obama’s administration (Thorbecke, 2021;

Kurtz & Yellin, 2020)—since all but one primary debate took place before the economic crash following the coronavirus outbreak occurred (Amadeo, 2021). On the other hand, many of Trump’s economic policies were criticized for not benefiting the working class as he had previously promised but rather favored the already wealthy and large corporations (Thorbecke, 2021).

(28)

28 Furthermore, according to Gallup’s Presidential Job Approval Center that has measured and reported the public’s approval of presidents going back all the way to Harry S. Truman in the 1940s, Donald Trump is the only president who has consistently stayed below the 50% mark (Gallup, n.d.). This might have given the Democratic Party a temporary lease on both

Republican-owned issues and performance issues (Petrocik, 1996). Therefore, this study also looks at the presence of issue ownership in the 2020 primary debates, and whether a relationship between discussing performance issues more than issues owned by either party and qualifying for future debates could be found.

(29)

29

4.0 Method and material

This chapter will introduce the chosen methodology for this study, quantitative content analysis, and will discuss some methodological problems. It will further describe how the empirical material was collected and the operationalization process to decide the variables for the study.

4.1 Quantitative content analysis

Quantitative content analysis is a data collection method developed by researchers in the social science field (Östbye, Knapskog, Helland, & Larsen, 2004). Though media content has been analyzed for a long time, the groundwork for quantitative content analysis was laid down by Bernard Berelson in his 1952 book Content analysis in communication research (Karlsson & Johansson, 2019). The method seeks to quantify content in order to make statistical observations and

generalizations whereas qualitative content analysis focuses more on the meaning behind the content. Similarly, while qualitative content analysis tends to closely examine smaller samples, quantitative analysis is used to analyze large datasets to identify correlations, similarities, and differences between different variables (Karlsson & Johansson, 2019).

To answer the research questions, this study uses quantitative content analysis to examine the presence of agenda setting, attribute agenda setting (through issue framing), and issue ownership in the 2020 Democratic primary debates. A quantitative content analysis suits this study since it seeks to identify similarities and differences between candidates, and to investigate whether any relationships between different candidate strategies and qualifying for future debates exist. If the study, instead, was interested in analyzing the meaning behind debate discussions or contextual aspects, a qualitative content analysis might have been more apt.

4.2 Methodological problems

One of the biggest issues with quantitative content analysis is that, in many cases, the coding is done manually instead of by a computer. This increases the risk of coder bias which can affect the outcome of a study. It is therefore important to be as open and transparent about how and why the data was collected as well as processed, for instance, why some things were included while others were left out (Karlsson & Johansson, 2019).

(30)

30

4.3 Population, sample, and units of analysis

The population of this study will be comprised of three Democratic primary debates leading up to the 2020 US election. The debates were chosen using stratified sampling after being divided into three smaller, separated subgroups (Karlsson & Johansson, 2019). The first subgroup includes debates 1 and 2; the second, 3-5; the third, 6-11. This, because during the first and second debates, twenty candidates were invited to participate. During the third through fifth, the number had decreased to 10-12 candidates, and by the sixth debate, they were down to 6-7.

Structuring the subgroups like this is relevant since the study aims to analyze possible factors that make candidates more or less likely to qualify for future debates. Thus, it would be interesting to compare a debate from, say, subgroup 1, where twenty candidates had qualified, to subgroup 2, where several of those candidates had either dropped out or failed to meet qualification

requirements. The debates were then given different numbers and a random number generator (random.org) was used to decide which debates were going to be included in the study. These were debates 2, 4, and 10.

The study also analyzes four Democratic candidates—Joe Biden, Bernie Sanders, Amy

Klobuchar, and Andrew Yang. These were, like the debates, chosen through stratified sampling where all candidates were divided into four subgroups. The first subgroup included only Joe Biden as he went on to become the eventual presidential nominee. Similarly, the second

subgroup included only Bernie Sanders since he was the only other candidate who qualified for and participated in all eleven debates. Subgroup 3 consisted of candidates who qualified for ten of the eleven debates (Elizabeth Warren, Pete Buttigieg, and Amy Klobuchar), and subgroup 4 contained candidates who qualified for 4-7 debates (Andrew Yang, Tom Steyer, Kamala Harris, Cory Booker, Tulsi Gabbard, Julian Castro, and Beto O’Rourke). Candidates who qualified for less than four debates were excluded as it would be difficult to get enough data from their performances. The reason Biden and Sanders were separated is that it would be interesting to compare their performances since, although they were both frontrunners of the primary process, Biden eventually “won” the nomination. Because subgroups 1 and 2 thus only consisted of one candidate each, they did not need to be put through the number generator like subgroups 3 and 4 did. The candidates in these subgroups were given different numbers that were then chosen at random using the same number generator as was used to pick out the debates. In subgroup 3, the number landed on Amy Klobuchar (qualified for debates 1-10), and in subgroup 4, on Andrew Yang (qualified for debates 1-6 and 8).

The units of analysis will be each candidate’s statements divided by sentence, i.e., one sentence is one unit of analysis. Initially, the units were going to be the candidates’ speaking turns as

(31)

31 provided by the transcripts (the collection of which will be described in the next section).

However, since these speaking turns varied quite significantly in length, they appeared to be rather imbalanced which could potentially skew the results. Due to this, it seemed more

appropriate to divide speaking turns by sentence and consider these the units of analysis instead.

The sentences may still vary in length but not to the same extent that the transcript-provided paragraphs did. Once collected and processed, the total amount of units was 1770 (683 for Biden, 454 for Sanders, 436 for Klobuchar, and 197 for Yang).

4.4 Data collection and procedure

The data, transcripts of the debates, were collected from rev.com, a San Francisco and Austin- based company that provides speech-to-text services (Rev, n.d.). The transcripts in question are all available for free on their website.

The material, thus, consisted of three debate transcripts. These transcripts were stored in their entirety, but separate documents for each candidate and debate were also compiled. Since three of the candidates—Biden, Sanders, and Klobuchar—participated in all three debates, they each had three separate documents, and Yang, who participated in two debates, thus had two documents. This was to make it easier to distinguish between the candidates when coding their separate units while having the full transcript to go back to if context was needed to understand a statement. As a result, there were eleven separate candidate transcripts and three full debate transcripts. The total amount of units was, as mentioned earlier, 1770 (683 for Biden, 454 for Sanders, 436 for Klobuchar, and 197 for Yang).

It should be noted that the transcripts used in this study contained a few errors and instances where the dialogue was inaudible, increasing the risk of misunderstandings during coding.

However, there were few and often easy to resolve (for instance, in one instance the word “won”

was transcribed as “one”, but the context made it clear that this was a misspelling). Even so, other transcriptions may have contained similar error as well, since there tend to be some instances where candidates and moderators talk over one another.

Some additional data was collected from public opinion polls conducted before and after each debate by The Economist and YouGov (2019a; 2019b; 2020), in order to determine which issues the public considered to be most important. These polls were conducted very frequently during the entire election cycle and included questions about issue importance that can help answer RQ1.

(32)

32 Using secondary data as the basis of a study comes with certain issues, though, namely that the data was likely collected for different purposes (Karlsson & Johansson, 2019). However, it also comes with certain perks; it saves a lot of time for the researcher. This study needs transcripts of the debates in order to work and creating those transcripts from scratch would simply take too long in this case. Similarly, opinion poll data is necessary to investigate issue salience among voters during the entire election cycle. Such surveys would not be possible to conduct at a later date. Secondary data is therefore the best and most viable option for this study.

4.5 Operationalization of variables

The variables in this study were chosen to answer the research questions presented in section 1.3 and the operationalization process will be described below.

4.5.1 RQ1

“What is the relationship, if any, between a candidate whose statements focused primarily on the three issues considered most important by the public according to opinion polls and whether this candidate qualified for future debates?”

RQ1 concerns the most important issues both during the debates and to the public in general during the 2020 election cycle and seeks to answer whether candidates benefited more from focusing on issues considered most salient by the public than candidates who did not. To answer this question, the coding scheme needs to distinguish different topics from each other. For instance, health care and education. To do this, the transcripts were divided into different units of analysis where each unit consisted of one sentence. So, for instance, “500,000 people go bankrupt because of medically-related issues, they come down with cancer, and that’s a reason to go

bankrupt?” is an example of a unit that would be coded as “health care”.

To compare the amount of time dedicated to a certain topic and how important it was to the public, the study also needs opinion poll results that provide this data. The poll data (collected from The Economist & YouGov, 2019a; 2019b; 2020) was coded separately, and as it already provides issue salience by percentage, most of the work here was already done. The most important issues to the public were then compared to how much time the different candidates spent on each topic to see whether they seemed to play to the crowd or focus on advantaged issues (Boydstun, Glazier, & Pietryka, 2013a). The poll data also allows for comparisons of the public’s issue agenda over time to see which issues increased and decreased in importance as the election process moved forward.

(33)

33 There is, however, a slight problem with implementing the poll data. Namely, because the topics listed in these opinion polls changed over time. The source this study uses (The Economist &

YouGov, 2019a; 2019b; 2020) provided the same list of topics to respondents up until March 2020, where the fifteen topics became ten, at least in the reports. Some of them seemed to have been grouped together, for instance, the new topic “national security and foreign policy” could include both the former topics “terrorism” and “foreign policy”, while others remained the same, for instance, “health care”. It is also difficult to know which topics have been removed entirely from the newer surveys and which have been grouped together with others. For instance, the former topic “Medicare”, which one could assume would have been grouped with “health care”

since, after all, it is a federal health insurance program (Medicare.gov, n.d.). However, it could also go under “Taxes and government spending” since Medicare for All is a reform bill (S. 1129, 2019) proposed by candidate Bernie Sanders and supported by a few fellow Democrats that would be funded by implementing a wealth tax (Friends of Bernie Sanders, n.d.-a).

Due to this slight inconsistency, this study’s issue coding scheme is developed from The Policy Agendas Project’s master topics codebook (2019) where a wide range of issues are listed with coding instructions. However, since the poll data from YouGov/Economist provided

respondents with fifteen issues (that were reduced to ten by March 2020), the categories had to be slightly altered. For instance, Policy Agendas Project’s codebook lists “health” as a broad topic including several subtopics, one of which covers issues related to Medicare. Since

YouGov/Economist listed Medicare as its own issue, a separate variable value was added to ensure Medicare and health care would not be grouped together. The same goes for topics like

“the budget deficit” which, according to Policy Agendas Project would be included under

“macroeconomics” but due to being distinguished from “the economy” in the YouGov/Economist polls, these topics were separated as well.

This, of course, comes with certain issues since it can be difficult to align categorizations that differ like this. However, thanks to clear coding instructions in the Policy Agendas Project codebook (2019), it was somewhat simpler to deduce what subtopics dealt with issues that this study needs to separate.

Furthermore, while performing a test of the coding scheme it became evident that a lot of attention and questions directed at the candidates concerned Trump’s impeachment trial, which does not entirely fit any of the issues as put forth by the poll data. Because of this, a variable value that deals with this issue and Trump’s presidency in general was added to variable 6.

Moreover, previous research has established that challenging candidates can benefit from

What determines who qualifies?