Campaigning in Poetry, Governing in Prose Pre- and Post-Election Effects of Election Pledge Rhetoric

Elina Lindgren

Department of Political Science

University of Gothenburg

Doctoral Dissertation in Political Science Department of Political Science University of Gothenburg 2017

Distribution: Elina Lindgren Department of Political Science University of Gothenburg Box 711 SE 405 30 Gothenburg Sweden elina.lindgren@gu.se © Author Elina Lindgren Campaigning in Poetry, Governing in Prose: Pre- and Post-Election Effects of Election Pledge Rhetoric Print: BrandFactory AB, Kållered, 2017 ISBN: 978-91-9844-020-1 (PRINT) ISBN: 978-91-9844-021-8 (PDF) ISSN: 0346-5942 This study is included as number 152 in the series Göteborg Studies in Politics, edited by Bo Rothstein, Department of Political Science, University of Gothenburg.

Contents Preface iSammanfattning på svenska v Introduction 1 Paper I. Explaining Framing Effects: Insights from Linguistic Theory on Persuasive Words 55 Paper II. Changing Policy with Words: How Persuasive Words in Election Pledges Influence Voters’ Beliefs about Policies 83 Paper III. Election Pledge Rhetoric: Selling Policy with Words 113 Paper IV. Making a Rod for Their Own Back? Post-Election Effects of Pre-Election Rhetoric 137

Doctoral Dissertation in Political Science Department of Political Science University of Gothenburg 2017

Distribution: Elina Lindgren Department of Political Science University of Gothenburg Box 711 SE 405 30 Gothenburg Sweden elina.lindgren@gu.se © Author Elina Lindgren Campaigning in Poetry, Governing in Prose: Pre- and Post-Election Effects of Election Pledge Rhetoric Print: BrandFactory AB, Kållered, 2017 ISBN: 978-91-9844-020-1 (PRINT) ISBN: 978-91-9844-021-8 (PDF) ISSN: 0346-5942 This study is included as number 152 in the series Göteborg Studies in Politics, edited by Bo Rothstein, Department of Political Science, University of Gothenburg.

Contents Preface iSammanfattning på svenska v Introduction 1 Paper I. Explaining Framing Effects: Insights from Linguistic Theory on Persuasive Words 55 Paper II. Changing Policy with Words: How Persuasive Words in Election Pledges Influence Voters’ Beliefs about Policies 83 Paper III. Election Pledge Rhetoric: Selling Policy with Words 113 Paper IV. Making a Rod for Their Own Back? Post-Election Effects of Pre-Election Rhetoric 137

The following paper is reproduced with permission from the

publisher:

Lindgren, E. & Naurin, E. (2017). Election pledge rhetoric: Selling policy

with words. International Journal of Communication, 11, 22.

In memory of Hans Andersson

i

Preface When someone says “democracy is good,” do you ask what they mean by democracy? Or when someone says that a “humane society must have a safety net for the truly needy,” do you ask: who are the “truly needy?” Some words and concepts seem so obvious that we rarely question their actual meanings. But if you put these words in a new context, you may find that what you accepted as an indisputable fact, is indeed far from undisputed.

I will use an example from a Swedish election campaign (2006), which contributed to the foundation of this dissertation. What confused me in this campaign was when a right-wing party, the Moderate Assembly Party (commonly known as the Moderates), declared that they had changed their name to the Moderates – the New Labor Party. At first, I could make no sense of this. Growing up in a society dominated by a social democratic party, I had a fairly certain understanding of what a labor party entailed. It was a left-of-center political party originating with union movements, formed to represent the common interests of ordinary workers. But instead of this, we now had a right-of-center party, with no roots in the union move-ments, calling itself this name.

Had the Moderates changed their political orientation towards tradi-tional social democracy? Or was it a trick to gain support from the social democratic voters? Over time, it became evident that neither of these was the case; the Moderates were just using a very different definition of labor party. I realized that my interpretation of this concept was only one of many, and that I needed to reconsider my own understanding of what “labor party” could mean in politics.

The use of these types of widely acknowledged, but descriptively vague, words and concepts is not specific to the Swedish Moderate Party; in most political parties’ campaigns you will find examples of words that stand out and are used more frequently. While, however, the use of these types of vague words often leave some room for interpretation in election cam-paigns, things quickly become specific after the election and the room for interpretation of what the parties meant with different words shrinks. For-mer New York Governor Mario Cuomo, summed up the phenomenon very well in an interview with The New Republic (4 April 1985) when he said: “parties campaign in poetry, but they govern in prose.”

Several years after the 2006 election campaign, I took part of research that showed that, while democratic parties across the world fulfill the vast majority of their election pledges, there seems to be a lingering perception among voters that parties are breaking their promises. While thinking about the incongruity between voters’ beliefs about parties’ pledge performance and their actual performance, the 2006 election campaign came to mind. If there are multiple ways to define words such as “labor party,” “democracy,”

ii

or safety net for the “truly needy,” then some individuals will probably be interpreting policies that are framed with these words in ways that differ from what the parties actually mean. Maybe, I thought, this incongruity is occurring somewhere between this “poetry” which is characteristic of elec-tion campaigns, and the clarity that is necessary with “the prose” of gov-erning.

There are many people who have contributed to making this project pos-sible. I am particularly indebted to my two supervisors, Peter Esaiasson and Elin Naurin. Peter, who also supervised my master’s thesis and encouraged me to apply for the doctoral studies program, has the rare talent of being able to give truly tough comments, while still leaving you with the pleasant feeling that everything is possible. Elin, whose dissertation about parties’ pledge fulfillment contributed with inspiration to the subject of this disser-tation, has taught me the importance of grit, and that the more difficult an issue appears to be, the more important it is to go straight at it and fix it. My sincere gratitude to both of you for all the readings and insightful com-ments, for always believing in this project, and for continuously pushing me and supporting me to overcome my own limitations.

I am grateful to the Department of Political Science, for providing an inspiring work environment, and for giving me the opportunity to work with some of the things I find truly exciting: investigating small details of large structures. I am grateful for comments and feedback provided by De-lia Dumitrescu, Ann-Kristin Kölln, Simon Labbé St-Vincent, Johan Mar-tinsson, Maria Oskarson, Mikael Persson, Adam Shehata, and Patrik Öh-berg. I am also grateful to people like Rasmus Broms, Jenny de Fine Licht, Annika Fredén, Henrik Friberg-Fernros, Olof Larsson, Sebastian Lundmark, Niels Markwat, Marina Nistotskaya, Henrik Oscarsson, Marina Povitkina, Jacob Sohlberg, Maria Solevid, Aksel Sundström, Aiysha Var-raich, Erik Vestin, and Karin Zelano for having made the time as a PhD student much more fun. A special thanks to my PhD student cohort, Sofia Jonsson, Fredrik Dybfest-Hjorthen, Mikael Holmgren, and Ole-Martin Laegreid – these years wouldn’t have been the same without you.

To the departmental administration, in particular Lena Caspers, Karin Jorthé, Maria Lilleste, and Christina Pettersson, thanks for always being helpful whenever I have needed assistance. To LORE, Maria Andreasson, Elias Markstedt, and Johan Martinsson, thanks for help with data collection and for the valuable comments on my survey designs. To André Blais and colleagues at the Department of Political Science, Université de Montreal – thanks for providing me with office space, an inspiring research network, and for the critical comments on my project during my visit in 2014.

Finally, but not least, thanks to friends and family, for your patience, and for forcing me to realize the benefits and joy of getting out of the office once in a while. Special thanks to my parents, Gunilla and Stig-Olov, for always supporting me in whatever I have chosen to do, for teaching me to

iii

always ask questions, and to never accept a statement as a fact. Thanks to my siblings Dag, Björn, Jorun, and Anton, for being there. And, to Andreas, thanks for putting up with me during this busiest of year; for supporting me and for questioning me, and for cooking the best dinners ever.

Finally, a special thanks to my grandfather Hans Andersson, who taught me the important lesson: for all the books I choose not to read, I’ll never be able to know their story.

ii

or safety net for the “truly needy,” then some individuals will probably be interpreting policies that are framed with these words in ways that differ from what the parties actually mean. Maybe, I thought, this incongruity is occurring somewhere between this “poetry” which is characteristic of elec-tion campaigns, and the clarity that is necessary with “the prose” of gov-erning.

There are many people who have contributed to making this project pos-sible. I am particularly indebted to my two supervisors, Peter Esaiasson and Elin Naurin. Peter, who also supervised my master’s thesis and encouraged me to apply for the doctoral studies program, has the rare talent of being able to give truly tough comments, while still leaving you with the pleasant feeling that everything is possible. Elin, whose dissertation about parties’ pledge fulfillment contributed with inspiration to the subject of this disser-tation, has taught me the importance of grit, and that the more difficult an issue appears to be, the more important it is to go straight at it and fix it. My sincere gratitude to both of you for all the readings and insightful com-ments, for always believing in this project, and for continuously pushing me and supporting me to overcome my own limitations.

I am grateful to the Department of Political Science, for providing an inspiring work environment, and for giving me the opportunity to work with some of the things I find truly exciting: investigating small details of large structures. I am grateful for comments and feedback provided by De-lia Dumitrescu, Ann-Kristin Kölln, Simon Labbé St-Vincent, Johan Mar-tinsson, Maria Oskarson, Mikael Persson, Adam Shehata, and Patrik Öh-berg. I am also grateful to people like Rasmus Broms, Jenny de Fine Licht, Annika Fredén, Henrik Friberg-Fernros, Olof Larsson, Sebastian Lundmark, Niels Markwat, Marina Nistotskaya, Henrik Oscarsson, Marina Povitkina, Jacob Sohlberg, Maria Solevid, Aksel Sundström, Aiysha Var-raich, Erik Vestin, and Karin Zelano for having made the time as a PhD student much more fun. A special thanks to my PhD student cohort, Sofia Jonsson, Fredrik Dybfest-Hjorthen, Mikael Holmgren, and Ole-Martin Laegreid – these years wouldn’t have been the same without you.

To the departmental administration, in particular Lena Caspers, Karin Jorthé, Maria Lilleste, and Christina Pettersson, thanks for always being helpful whenever I have needed assistance. To LORE, Maria Andreasson, Elias Markstedt, and Johan Martinsson, thanks for help with data collection and for the valuable comments on my survey designs. To André Blais and colleagues at the Department of Political Science, Université de Montreal – thanks for providing me with office space, an inspiring research network, and for the critical comments on my project during my visit in 2014.

Finally, but not least, thanks to friends and family, for your patience, and for forcing me to realize the benefits and joy of getting out of the office once in a while. Special thanks to my parents, Gunilla and Stig-Olov, for always supporting me in whatever I have chosen to do, for teaching me to

iii

always ask questions, and to never accept a statement as a fact. Thanks to my siblings Dag, Björn, Jorun, and Anton, for being there. And, to Andreas, thanks for putting up with me during this busiest of year; for supporting me and for questioning me, and for cooking the best dinners ever.

Finally, a special thanks to my grandfather Hans Andersson, who taught me the important lesson: for all the books I choose not to read, I’ll never be able to know their story.

v

Sammanfattning på svenska I valkampanjer använder partier ofta värdeladdade ord som frihet och kon-kurrenskraft, för att nå ut med sina vallöften och policyförslag till väljarna. Medan det är välkänt att närvaron av sådana ord i vallöften kan påverka väljares stöd för policyförslag, vet vi mindre om vad exakt det är som gör dessa ord effektiva. I den här avhandlingen presenterar och prövar Elina Lindgren en språkfilosofisk förklaring till hur värdeord kan påverka, inte bara om väljare gillar ett politiskt förslag, utan också hur de uppfattar själva innehållet i förslaget. Genom en serie survey experiment visar hon att när-varon av ord som frihet och konkurrenskraft i vallöften kan leda till att väl-jare ser saker i policyförslag som går utanför förslagens faktiska innehåll, vilket i sin tur kan leda till (miss)tolkningar att förslagen och dess framtida effekter är i linje med de egna policypreferenserna. Hon visar också hur dessa tolkningar i förlängningen kan göra det svårare att infria väljares för-väntningar på vallöften efter valet, och även leda till en uppfattning att ett vallöfte inte blivit uppfyllt. Resultaten väcker frågor kring funktionen hos representativa demokratier. Enligt ett representativt demokratiideal ska val-kampanjer förmedla lättillgänglig och tydlig information om partiers policy intentioner, så att väljarna kan göra rimliga förutsägelser av vad som ska hända efter valet. Om värdeord i vallöften kan skapa förväntningar på po-licyförslag bortom vad som faktiskt utlovas, blir det svårare för väljare att förutsäga vilka partier som bäst kommer att representera deras egna policy-preferenser efter valet. Detta skapar frågor kring i vilken grad vi kan an-vända väljares stöd för policyförslag innan val som en indikation på deras faktiska policypreferenser, och därmed också vilka möjligheter partier har att avgöra hur de bäst kan handla för att representera medborgarnas prefe-renser efter valet – något som är centralt för en representativ demokrati.

1

Introduction

Elina Lindgren

There are words, such as “terrorism” and “freedom,” that carry with them more than a simple description of reality. Citing the linguistic philosopher Charles Stevenson, these words have a “magnetic effect,” and an “impera-tive force” that influences the hearers’ decisions (1937, pp. 18–19). This thesis is about the role of such words in politics. My point is that we know that the choice of words matters, but we do not know how, exactly. Previous research shows ample evidence that words affect voters’ policy support. Research on the individual micro-level mechanisms which lead up to the effects of words on voters’ support for policies, are, however, inconclusive (and also sometimes inconsistent).

The aim of this dissertation is to investigate, on an individual micro level, how words influence voters’ evaluations of policy pledges before, and after, elections. Election pledges are crucial to the process of linking parties’ policy intentions to voters’ policy preferences, and they are consid-ered to be essential tools that parties use to reach out to voters with their policies. But they are also vital to mustering electoral support. In the words of Anthony Downs from his classical work on representative democracy: “Parties formulate policies in order to win elections, rather than win elec-tions in order to formulate policies” (Downs, 1957, p. 28).

But can parties really muster electoral support simply by formulating appealing policy pledges? There are currently several factors that indicate that they should not be able to do so. Research on political engagement points to a steadily growing trend of political cynicism in Western democ-racies (e.g., Cappella & Jamieson, 1996; Hooghe, 2004), and it has also been shown that the conventional view of parties’ pledges is that they are constantly being broken (Naurin, 2011; Thomson, 2011). Hence, we could expect that voters would be skeptical of whatever parties say in election campaigns. Yet, empirical studies show otherwise; politicians can both at-tract voters’ attention, and trigger enthusiasm for policies simply by using rhetoric (e.g., Lakoff, 2002; Nelson, Clawson, & Oxley, 1997). In his work on representative democracy, Adam Przeworski (2009, p. 72) summarizes his perspective on these findings: “Democracy incessantly rekindles our hope. We are perennially eager to be lured by promises, to put our stakes on electoral bets.”

2

The former American President Barack Obama said it more positively in his 2008 campaign: “I don’t understand this argument that words don’t matter. The most important thing that we can do right now, is to re-engage the American people […] to get them excited and interested again […] If we cannot inspire the country to believe again, then it doesn’t matter how many policies and plans we have. […] Don’t tell me words don’t matter!”1

Obama’s statement illustrates how policies alone are not enough to at-tract voters. In order to gain support, parties need to inspire and engage the voters, and to make them believe that the policies being proposed could actually make a difference. Since policies are presented to voters via elec-tion pledges—be it through written manifestos, public speeches or televised debates—communicating the pledges in inspiring and persuasive ways be-comes crucial.

Whether it should be regarded as discouraging, or uplifting; the view that words matter to policy support is ubiquitous. The way that parties for-mulate their promises of future action is vital for electoral success (e.g., Riker, 1986; Jacobs & Shapiro, 2000), and political parties have always used rhetoric to draw attention to their policies. A common way to appeal to voters, is to call on societal values (e.g., Clifford & Jerit, 2013; Feldman & Steenbergen, 2001; Jerit, 2004; Lakoff, 2002), such as “democracy,” “se-curity,” “freedom,” and “equality.”

It is easy to accept that values are a central component of political rhet-oric, especially in light of the vast literature which shows how values are essential to individuals’ opinion-formation processes (e.g., Feldman, 1988; Fleming & Petty, 2000; Jacoby, 2006; McCloskey & Zaller, 1984); values simply seem to serve as a benchmark that individuals rely upon when they evaluate political objects (e.g., Rokeach, 1973; Kinder, 1998). Voters sup-port those policies and candidates that they perceive as being closest to their own values (e.g., Feldman, 1982; Katz & Hass, 1988; Kinder & Sanders, 1996), and when political issues are presented in value-laden terms, this framing alone can shape public opinion on these particular issues (e.g., Nel-son & Garst, 2005; Schemer, Wirth, & Matthes, 2012; Shen & Hatfield-Edwards, 2005).

Although it is well documented that value-laden words engage people and muster policy support, less is known about the micro-level mechanisms that make these words persuasive. Previous research on value-laden rheto-ric has focused on so-called framing effects (for an overview, see e.g., Kinder, 1998; Schemer, Wirth, & Matthes, 2012), which refers to changes in the importance that individuals assign to different beliefs they have about something (e.g., their cognitive understanding of an object’s content, causes or consequences, e.g., Glynn et al., 2004; Nelson & Oxley, 1999). In this thesis, I will investigate how value-laden words may also influence citizens 1 Campaign speech by Senator Barack Obama at the Democratic Party of Wisconsin Founders Day Gala, February 16, 2008.

3

by changing, or introducing new content, to those existing beliefs (some-thing which is also known as persuasion, see e.g., Nelson, Oxley, & Claw-son, 1997).

One central part of my thesis is the notion that the presence of value-laden words in election pledges can elicit beliefs on the content of specific policies among voters; beliefs that are not supported by factual information on the exact policy proposals that are outlined in the pledges. George Or-well (1946, p. 5) once argued that value-laden words, such as freedom and equality, each have “several different meanings which cannot be reconciled with one another.” If policy pledges are described by value words—to which individuals attach different meanings—then voters will form differ-ent beliefs about what is being pledged. This means that such words might mislead voters as they struggle to grasp what parties aim to do, as the words can mean one thing for a voter and something significantly different for a pledge-maker. It also means that the wording of pledges can impact voters’ post-election evaluation of policy outcomes, and possibly—if the words cause discrepancies between voters’ beliefs about the policies and parties’ real policy intentions—lead to a perception that the pledges have been broken.

The aim of this thesis is thus to investigate the effects of words on voters’ beliefs about policy proposals outlined in election pledges, and what pre- and post-election effects these have on their evaluations of policies and pol-icy outcomes. To understand the micro-level mechanisms that are involved when certain words influence individuals’ reactions to policy pledges, one must first define the theoretical premises as to why these words might be particularly compelling. In this thesis, I utilize theory from the literature on philosophy of language and linguistic semantics for this purpose (hence-forth linguistic theory). This literature (e.g., Stevenson, 1944; Walton, 2006) provides a theoretical account of specific characteristics that make certain words become so-called “persuasive words”—and thereby useful in political communication (e.g., Schiappa, 2003; Zarefsky, 2004). In short, these are words with value connotations, strong emotive meaning, and a vague and flexible descriptive meaning. Making use of the linguistic theory on persuasive words, this thesis gains a theoretical framework to study the specific mechanisms at play when certain words persuade voters.

This thesis also brings together two political communication theories—framing theory and the linguistic theory on persuasive words—which, alt-hough closely related, operate separately. This thesis contributes to both literatures. It contributes to the framing literature by suggesting and testing a linguistic explanation to framing effects, which adds to current explana-tions. It contributes to the field of linguistics by empirically testing the the-ory on persuasive words, and by developing the theory to account for dif-ferent types of persuasive words—universal, left-leaning and right-leaning words—and how the effects of these types of words in election campaigns

2

The former American President Barack Obama said it more positively in his 2008 campaign: “I don’t understand this argument that words don’t matter. The most important thing that we can do right now, is to re-engage the American people […] to get them excited and interested again […] If we cannot inspire the country to believe again, then it doesn’t matter how many policies and plans we have. […] Don’t tell me words don’t matter!”1

Obama’s statement illustrates how policies alone are not enough to at-tract voters. In order to gain support, parties need to inspire and engage the voters, and to make them believe that the policies being proposed could actually make a difference. Since policies are presented to voters via elec-tion pledges—be it through written manifestos, public speeches or televised debates—communicating the pledges in inspiring and persuasive ways be-comes crucial.

Whether it should be regarded as discouraging, or uplifting; the view that words matter to policy support is ubiquitous. The way that parties for-mulate their promises of future action is vital for electoral success (e.g., Riker, 1986; Jacobs & Shapiro, 2000), and political parties have always used rhetoric to draw attention to their policies. A common way to appeal to voters, is to call on societal values (e.g., Clifford & Jerit, 2013; Feldman & Steenbergen, 2001; Jerit, 2004; Lakoff, 2002), such as “democracy,” “se-curity,” “freedom,” and “equality.”

It is easy to accept that values are a central component of political rhet-oric, especially in light of the vast literature which shows how values are essential to individuals’ opinion-formation processes (e.g., Feldman, 1988; Fleming & Petty, 2000; Jacoby, 2006; McCloskey & Zaller, 1984); values simply seem to serve as a benchmark that individuals rely upon when they evaluate political objects (e.g., Rokeach, 1973; Kinder, 1998). Voters sup-port those policies and candidates that they perceive as being closest to their own values (e.g., Feldman, 1982; Katz & Hass, 1988; Kinder & Sanders, 1996), and when political issues are presented in value-laden terms, this framing alone can shape public opinion on these particular issues (e.g., Nel-son & Garst, 2005; Schemer, Wirth, & Matthes, 2012; Shen & Hatfield-Edwards, 2005).

Although it is well documented that value-laden words engage people and muster policy support, less is known about the micro-level mechanisms that make these words persuasive. Previous research on value-laden rheto-ric has focused on so-called framing effects (for an overview, see e.g., Kinder, 1998; Schemer, Wirth, & Matthes, 2012), which refers to changes in the importance that individuals assign to different beliefs they have about something (e.g., their cognitive understanding of an object’s content, causes or consequences, e.g., Glynn et al., 2004; Nelson & Oxley, 1999). In this thesis, I will investigate how value-laden words may also influence citizens 1 Campaign speech by Senator Barack Obama at the Democratic Party of Wisconsin Founders Day Gala, February 16, 2008.

3

by changing, or introducing new content, to those existing beliefs (some-thing which is also known as persuasion, see e.g., Nelson, Oxley, & Claw-son, 1997).

One central part of my thesis is the notion that the presence of value-laden words in election pledges can elicit beliefs on the content of specific policies among voters; beliefs that are not supported by factual information on the exact policy proposals that are outlined in the pledges. George Or-well (1946, p. 5) once argued that value-laden words, such as freedom and equality, each have “several different meanings which cannot be reconciled with one another.” If policy pledges are described by value words—to which individuals attach different meanings—then voters will form differ-ent beliefs about what is being pledged. This means that such words might mislead voters as they struggle to grasp what parties aim to do, as the words can mean one thing for a voter and something significantly different for a pledge-maker. It also means that the wording of pledges can impact voters’ post-election evaluation of policy outcomes, and possibly—if the words cause discrepancies between voters’ beliefs about the policies and parties’ real policy intentions—lead to a perception that the pledges have been broken.

The aim of this thesis is thus to investigate the effects of words on voters’ beliefs about policy proposals outlined in election pledges, and what pre- and post-election effects these have on their evaluations of policies and pol-icy outcomes. To understand the micro-level mechanisms that are involved when certain words influence individuals’ reactions to policy pledges, one must first define the theoretical premises as to why these words might be particularly compelling. In this thesis, I utilize theory from the literature on philosophy of language and linguistic semantics for this purpose (hence-forth linguistic theory). This literature (e.g., Stevenson, 1944; Walton, 2006) provides a theoretical account of specific characteristics that make certain words become so-called “persuasive words”—and thereby useful in political communication (e.g., Schiappa, 2003; Zarefsky, 2004). In short, these are words with value connotations, strong emotive meaning, and a vague and flexible descriptive meaning. Making use of the linguistic theory on persuasive words, this thesis gains a theoretical framework to study the specific mechanisms at play when certain words persuade voters.

This thesis also brings together two political communication theories—framing theory and the linguistic theory on persuasive words—which, alt-hough closely related, operate separately. This thesis contributes to both literatures. It contributes to the framing literature by suggesting and testing a linguistic explanation to framing effects, which adds to current explana-tions. It contributes to the field of linguistics by empirically testing the the-ory on persuasive words, and by developing the theory to account for dif-ferent types of persuasive words—universal, left-leaning and right-leaning words—and how the effects of these types of words in election campaigns

4

are likely limited by the voters’ ideological predispositions. Finally, by do-ing the above, I intend to comment on the role campaign rhetoric plays in the functioning of mandate models of representation, and how voters and voters’ policy preferences can be viewed in the light of the effects of such rhetoric.

The Dual Role of Election Pledges for Democratic

Representation

In “mandate models” of representation2 (e.g., Klingemann, Hofferbert, & Budge, 1994; Manin, Przeworski & Stokes, 1999; Mansbridge, 2003), elec-tion pledges play a central role in the representative chain when voters’ policy preferences are linked to parties’ policy intentions both before and after elections. Before elections, parties use election pledges to present pol-icy proposals to voters, and to explain the content and consequences of these policies. In this ideal type of democratic representation, citizens do their part by comparing and evaluating the parties’ policy proposals, decid-ing which party to vote for based on this. This means that “elections emu-late a direct assembly and the winning platform becomes the ‘mandate’ that the governments pursue” (Manin, Przeworski, & Stokes, 1999, p. 29).

After elections are held, the winning party implements the policies on which they have been mandated to govern. At this stage, election pledges serve as a tool that citizens use to hold governments accountable for their post-election actions. More precisely, election pledges help citizens com-pare governments’ post-election policy performance to what the govern-ments had pledged to do before the election, and thereby offer a way to evaluate how well governments have performed their mandate (e.g. Artés, 2011; Mansergh & Thomson, 2007; Naurin, 2014; Royed, 1996; Thomson, 2001; Thomson, et al., 2017). In other words, election pledges serve as a “contract” between voters and parties in elections; post-election, they serve as a tool that voters can use to evaluate the extent to which the government has fulfilled its part of this contract. This is why it is important for election pledges to be formulated in a clear and unambiguous way, making it easy for voters to grasp the parties’ policy intentions before they make their elec-toral decisions (e.g., Downs, 1957).

Based on this mandate model of representation, scholars perform sys-tematic comparisons between campaign pledges and subsequent govern-ment behavior. Such empirical studies show that governments (as opposed to a general view among voters, see Naurin, 2011), actually fulfill the vast 2 The mandate model of representation (sometimes also called “promissory representation,” Mansbridge, 2003) is one of the most extensively used models of democratic representation in empirical research (Esaiasson & Holmberg, 1996, p. 3; Pierce, 1999, p. 10).

5

majority of what they pledge in elections (e.g., Naurin, 2011; Thomson, et al., 2017). Such studies also show that parties make a great number of elec-tion pledges in campaigns, and that over time, this number increases (Håkansson & Naurin, 2016). These findings highlight why election pledges are so important to the actual functioning of representative democ-racies around the world, and why studies focused on how voters perceive these pledges are therefore in high demand (Born, van Eck, & Johannesson, 2017; Elinder, Jordahl, & Poutvaara, 2015; Johnson & Ryu, 2010; Naurin, 2011; Naurin & Oscarsson, 2017; Thomson, 2011; Thomson et al., 2017).

Effects of Linking Policies to Values

The mandate model of representation begins with the belief that voters should be capable of forming well-founded political preferences that can be accurately implemented through democratic elections (e.g., Dahl, 1971; Downs, 1957; Manin, Przeworski & Stokes, 1999). What does it mean, then, if citizens’ support for policies changes as a result of changes in elite communication? Some scholars believe that this indicates that voters are politically uninformed, and hold only weak and unfounded policy prefer-ences (e.g., Achen & Bartels, 2016; Entman, 1993; Iyengar, 1991; Riker, 1986; Zaller, 1992). Others believe that the fact that citizens’ policy support changes as the political debate changes instead indicates that they are able to make use of and incorporate new information, adjusting their policy pref-erences based on this updated information (e.g., Page & Shapiro, 1992; Popkin, 1991).3 Regardless, it is a well-established empirical finding that politicians (and other stakeholders) in real-life democracies can influence citizens’ support for policies “simply” by re-wording them (see e.g., Kinder, 1998, for an overview).

It seems that one particularly effective way of influencing citizens’ sup-port for policies is to link policy proposals to social values. The empirical findings supporting this apply to a variety of political objects, for instance: determining whether public rallies by extreme groups (Nelson, Clawson, & Oxley, 1997) or affirmative actions for ethnic minorities (Domke, McCoy, & Torres, 1999) should be allowed, how political candidates are evaluated (Domke, Shah, & Wackman, 1998; Shah, Domke, & Wackman, 1996), and how generous gay rights policies (Brewer, 2002, 2003), welfare policies (Brewer, 2001; Shen & Edwards, 2005; Slothuus, 2008), and government spending (Sniderman & Theriault, 2004) should be. For this reason, some have described value-laden rhetoric as a “powerful and reliable weapon in the persuader’s arsenal” (Nelson & Garst, 2005, p. 490), and it is frequently

3 For an overview of different interpretations of the implications of elite framing effects for citizen competence, see Druckman (2001a).

4

are likely limited by the voters’ ideological predispositions. Finally, by do-ing the above, I intend to comment on the role campaign rhetoric plays in the functioning of mandate models of representation, and how voters and voters’ policy preferences can be viewed in the light of the effects of such rhetoric.

The Dual Role of Election Pledges for Democratic

Representation

In “mandate models” of representation2 (e.g., Klingemann, Hofferbert, & Budge, 1994; Manin, Przeworski & Stokes, 1999; Mansbridge, 2003), elec-tion pledges play a central role in the representative chain when voters’ policy preferences are linked to parties’ policy intentions both before and after elections. Before elections, parties use election pledges to present pol-icy proposals to voters, and to explain the content and consequences of these policies. In this ideal type of democratic representation, citizens do their part by comparing and evaluating the parties’ policy proposals, decid-ing which party to vote for based on this. This means that “elections emu-late a direct assembly and the winning platform becomes the ‘mandate’ that the governments pursue” (Manin, Przeworski, & Stokes, 1999, p. 29).

After elections are held, the winning party implements the policies on which they have been mandated to govern. At this stage, election pledges serve as a tool that citizens use to hold governments accountable for their post-election actions. More precisely, election pledges help citizens com-pare governments’ post-election policy performance to what the govern-ments had pledged to do before the election, and thereby offer a way to evaluate how well governments have performed their mandate (e.g. Artés, 2011; Mansergh & Thomson, 2007; Naurin, 2014; Royed, 1996; Thomson, 2001; Thomson, et al., 2017). In other words, election pledges serve as a “contract” between voters and parties in elections; post-election, they serve as a tool that voters can use to evaluate the extent to which the government has fulfilled its part of this contract. This is why it is important for election pledges to be formulated in a clear and unambiguous way, making it easy for voters to grasp the parties’ policy intentions before they make their elec-toral decisions (e.g., Downs, 1957).

Based on this mandate model of representation, scholars perform sys-tematic comparisons between campaign pledges and subsequent govern-ment behavior. Such empirical studies show that governments (as opposed to a general view among voters, see Naurin, 2011), actually fulfill the vast 2 The mandate model of representation (sometimes also called “promissory representation,” Mansbridge, 2003) is one of the most extensively used models of democratic representation in empirical research (Esaiasson & Holmberg, 1996, p. 3; Pierce, 1999, p. 10).

5

majority of what they pledge in elections (e.g., Naurin, 2011; Thomson, et al., 2017). Such studies also show that parties make a great number of elec-tion pledges in campaigns, and that over time, this number increases (Håkansson & Naurin, 2016). These findings highlight why election pledges are so important to the actual functioning of representative democ-racies around the world, and why studies focused on how voters perceive these pledges are therefore in high demand (Born, van Eck, & Johannesson, 2017; Elinder, Jordahl, & Poutvaara, 2015; Johnson & Ryu, 2010; Naurin, 2011; Naurin & Oscarsson, 2017; Thomson, 2011; Thomson et al., 2017).

Effects of Linking Policies to Values

The mandate model of representation begins with the belief that voters should be capable of forming well-founded political preferences that can be accurately implemented through democratic elections (e.g., Dahl, 1971; Downs, 1957; Manin, Przeworski & Stokes, 1999). What does it mean, then, if citizens’ support for policies changes as a result of changes in elite communication? Some scholars believe that this indicates that voters are politically uninformed, and hold only weak and unfounded policy prefer-ences (e.g., Achen & Bartels, 2016; Entman, 1993; Iyengar, 1991; Riker, 1986; Zaller, 1992). Others believe that the fact that citizens’ policy support changes as the political debate changes instead indicates that they are able to make use of and incorporate new information, adjusting their policy pref-erences based on this updated information (e.g., Page & Shapiro, 1992; Popkin, 1991).3 Regardless, it is a well-established empirical finding that politicians (and other stakeholders) in real-life democracies can influence citizens’ support for policies “simply” by re-wording them (see e.g., Kinder, 1998, for an overview).

It seems that one particularly effective way of influencing citizens’ sup-port for policies is to link policy proposals to social values. The empirical findings supporting this apply to a variety of political objects, for instance: determining whether public rallies by extreme groups (Nelson, Clawson, & Oxley, 1997) or affirmative actions for ethnic minorities (Domke, McCoy, & Torres, 1999) should be allowed, how political candidates are evaluated (Domke, Shah, & Wackman, 1998; Shah, Domke, & Wackman, 1996), and how generous gay rights policies (Brewer, 2002, 2003), welfare policies (Brewer, 2001; Shen & Edwards, 2005; Slothuus, 2008), and government spending (Sniderman & Theriault, 2004) should be. For this reason, some have described value-laden rhetoric as a “powerful and reliable weapon in the persuader’s arsenal” (Nelson & Garst, 2005, p. 490), and it is frequently

3 For an overview of different interpretations of the implications of elite framing effects for citizen competence, see Druckman (2001a).

6

used by politicians whose aim is to shape citizens’ policy support in ways that fit their goals (e.g., Lakoff, 2002).

The effectiveness of presenting policies with values has been explained in that political attitudes are rooted in individuals’ value preferences (e.g., Lakoff, 2002; Sniderman & Theriault, 2004), and that citizens therefore support policies and candidates that they perceive as promoting their fa-vored values (e.g., Feldman, 1982; Katz & Hass, 1988; Kinder & Sanders, 1996; see Kinder, 1998, for a review). When speakers call on societal val-ues, they activate listeners’ value orientations. This, in turn, triggers “heu-ristic judgments”—a process that describes the causal relationship between evoked values, and individuals’ value predispositions, emotions, and sub-sequent decisions (e.g., Igartua & Cheng, 2009; Kahneman, Tversky, & Slovic, 1982; Schwarz, Bless, Wanke, & Winkielman, 2003; Tversky & Kahneman, 1974). Indications that heuristic processes are at play when cit-izens evaluate political issues that are framed with values are found in stud-ies that show that value-laden rhetoric shapes citizens’ opinions by encour-aging them to view a policy as morally right or wrong (e.g., Clifford & Jerit, 2013; Clifford, Jerit, Rainey, & Motyl, 2015), and to feel a particular way about a policy (to like or dislike it) (e.g., Gross, 2008; Gross & D’Am-brosio, 2004; Lecheler, Bos, & Vliegenthart, 2015).

Moreover, scholars who study the link between value-based rhetoric and policy support point out that the effectiveness of values is dependent on the value orientation of the audience. Political attitudes, according to these scholars, are rooted in “value conflicts” (Brewer, 2008; Sniderman & The-riault, 2004) and “value hierarchies” (Jacoby, 2006; Nelson, Oxley, & Clawson, 1997), which means that when voters make political decisions, they are choosing between different (sometimes mutually exclusive) val-ues. Sniderman and Theriault (2004, pp. 140–141) say that “the substance of political problems, by their very nature, requires choices be made be-tween competing values […]. To have more of one means to have less of the other.” For this reason, both voters and politicians often commit them-selves to either one or another value (typically resulting in ideological di-vides) (e.g., Jackman & Sniderman, 2002). Sniderman and Theriault (2004) give the example of “individual freedom,” which is associated with liberal ideology, as competing with the value “social order,” which is associated with conservative ideals (pp. 140–141).

Accordingly, the effectiveness of value-laden rhetoric depends on whether or not the values align with the individuals’ ideological predispo-sitions. Two examples of these studies are Schemer, Wirth and Matthes (2012), who showed that “individualism” attracted right-oriented individu-als while “fairness” attracted people on the left, and Nelson and Garst (2005) who found that “personal motivation” attracted individuals to the right while “equality” attracted individuals on the left.

7

Despite evidence of a relationship between value-based rhetoric, voters’ value predispositions, and their support for policies, research is not conclu-sive as to how exactly individuals process rhetoric that appeals to their core values. Theory on motivated reasoning, for example, stipulates that people are strongly averse to changing established beliefs, which makes them re-sistant to information that contradicts their prior attitudes (e.g., Festinger, 1957; Kunda, 1990). Therefore, individuals actively seek out and accept information that aligns with their value predispositions (confirmation bias), while they automatically dismiss contrary facts and arguments (disconfir-mation bias) (e.g., Lord, Ross & Lepper, 1979; Taber & Lodge, 2006). In short, this line of research suggests that individuals use cognitive shortcuts rather than employing critical reasoning when they evaluate policies that are presented with value-based rhetoric (in psychology terminology, they engage in peripheral route processing of information, see e.g., Petty & Cacioppo, 1986, 1990).

In contrast to this negative view on voters’ likelihood of rationally eval-uating messages that appeal to their core values, others have stressed that there are factors that make values promote both cognitive and systematic processing of political appeals—the opposite of peripheral-route pro-cessing (see e.g., Eagly & Chaiken, 1993; Nelson & Garst, 2005; Petty & Cacioppo, 1986, 1990; Sniderman & Theriault, 2004).4

Maybe the way that values are used in political appeals—how exactly the messages are formulated—trigger different cognitive processes? Next, I will present two political communication theories which address these questions.

Theoretical Points of Departure

To understand how citizens use value-laden rhetoric to evaluate policy pledges, this thesis connects two communication theories which stress the importance of values in political rhetoric; a theory on (value) framing ef-fects, and a linguistic theory on persuasive words. The framing research has grown dramatically in recent decades (Weaver, 2007, pp. 143–144), and it spans several disciplines, including political science, media and communi-cation, psychology and marketing (for an overview, see Keren, 2011). The literature comprises a vast compilation of empirical studies, which can help explain the effects of rhetoric on political attitudes and policy support. The literature on persuasive words is more limited and less empirical than the framing literature, but it provides a precise theoretical model of the specific semantic characteristics that make value-laden words so compelling.

4 For an overview of the two models, see Brewer (2001).

6

used by politicians whose aim is to shape citizens’ policy support in ways that fit their goals (e.g., Lakoff, 2002).

The effectiveness of presenting policies with values has been explained in that political attitudes are rooted in individuals’ value preferences (e.g., Lakoff, 2002; Sniderman & Theriault, 2004), and that citizens therefore support policies and candidates that they perceive as promoting their fa-vored values (e.g., Feldman, 1982; Katz & Hass, 1988; Kinder & Sanders, 1996; see Kinder, 1998, for a review). When speakers call on societal val-ues, they activate listeners’ value orientations. This, in turn, triggers “heu-ristic judgments”—a process that describes the causal relationship between evoked values, and individuals’ value predispositions, emotions, and sub-sequent decisions (e.g., Igartua & Cheng, 2009; Kahneman, Tversky, & Slovic, 1982; Schwarz, Bless, Wanke, & Winkielman, 2003; Tversky & Kahneman, 1974). Indications that heuristic processes are at play when cit-izens evaluate political issues that are framed with values are found in stud-ies that show that value-laden rhetoric shapes citizens’ opinions by encour-aging them to view a policy as morally right or wrong (e.g., Clifford & Jerit, 2013; Clifford, Jerit, Rainey, & Motyl, 2015), and to feel a particular way about a policy (to like or dislike it) (e.g., Gross, 2008; Gross & D’Am-brosio, 2004; Lecheler, Bos, & Vliegenthart, 2015).

Moreover, scholars who study the link between value-based rhetoric and policy support point out that the effectiveness of values is dependent on the value orientation of the audience. Political attitudes, according to these scholars, are rooted in “value conflicts” (Brewer, 2008; Sniderman & The-riault, 2004) and “value hierarchies” (Jacoby, 2006; Nelson, Oxley, & Clawson, 1997), which means that when voters make political decisions, they are choosing between different (sometimes mutually exclusive) val-ues. Sniderman and Theriault (2004, pp. 140–141) say that “the substance of political problems, by their very nature, requires choices be made be-tween competing values […]. To have more of one means to have less of the other.” For this reason, both voters and politicians often commit them-selves to either one or another value (typically resulting in ideological di-vides) (e.g., Jackman & Sniderman, 2002). Sniderman and Theriault (2004) give the example of “individual freedom,” which is associated with liberal ideology, as competing with the value “social order,” which is associated with conservative ideals (pp. 140–141).

Accordingly, the effectiveness of value-laden rhetoric depends on whether or not the values align with the individuals’ ideological predispo-sitions. Two examples of these studies are Schemer, Wirth and Matthes (2012), who showed that “individualism” attracted right-oriented individu-als while “fairness” attracted people on the left, and Nelson and Garst (2005) who found that “personal motivation” attracted individuals to the right while “equality” attracted individuals on the left.

7

Despite evidence of a relationship between value-based rhetoric, voters’ value predispositions, and their support for policies, research is not conclu-sive as to how exactly individuals process rhetoric that appeals to their core values. Theory on motivated reasoning, for example, stipulates that people are strongly averse to changing established beliefs, which makes them re-sistant to information that contradicts their prior attitudes (e.g., Festinger, 1957; Kunda, 1990). Therefore, individuals actively seek out and accept information that aligns with their value predispositions (confirmation bias), while they automatically dismiss contrary facts and arguments (disconfir-mation bias) (e.g., Lord, Ross & Lepper, 1979; Taber & Lodge, 2006). In short, this line of research suggests that individuals use cognitive shortcuts rather than employing critical reasoning when they evaluate policies that are presented with value-based rhetoric (in psychology terminology, they engage in peripheral route processing of information, see e.g., Petty & Cacioppo, 1986, 1990).

In contrast to this negative view on voters’ likelihood of rationally eval-uating messages that appeal to their core values, others have stressed that there are factors that make values promote both cognitive and systematic processing of political appeals—the opposite of peripheral-route pro-cessing (see e.g., Eagly & Chaiken, 1993; Nelson & Garst, 2005; Petty & Cacioppo, 1986, 1990; Sniderman & Theriault, 2004).4

Maybe the way that values are used in political appeals—how exactly the messages are formulated—trigger different cognitive processes? Next, I will present two political communication theories which address these questions.

Theoretical Points of Departure

To understand how citizens use value-laden rhetoric to evaluate policy pledges, this thesis connects two communication theories which stress the importance of values in political rhetoric; a theory on (value) framing ef-fects, and a linguistic theory on persuasive words. The framing research has grown dramatically in recent decades (Weaver, 2007, pp. 143–144), and it spans several disciplines, including political science, media and communi-cation, psychology and marketing (for an overview, see Keren, 2011). The literature comprises a vast compilation of empirical studies, which can help explain the effects of rhetoric on political attitudes and policy support. The literature on persuasive words is more limited and less empirical than the framing literature, but it provides a precise theoretical model of the specific semantic characteristics that make value-laden words so compelling.

4 For an overview of the two models, see Brewer (2001).

8

Framing has been defined as a communication strategy which “high-lights some bits of information about an item that is the subject of a com-munication, thereby elevating them in salience” (Entman, 1993, p. 53). More precisely, this means that, when presenting a problem or an issue to an audience, an individual makes a piece of information about this issue “more noticeable, meaningful, or memorable” to the audience (Entman, 1993, p. 53). Framing effects, in turn, occur when this selective presentation of certain aspects of an object influences the weight that individuals assign to these aspects when forming their attitudes (see e.g., Nelson, Clawson & Oxley, 1997; Nelson, Oxley, & Clawson, 1997; Price & Tewksbury, 1997).

Apart from these broad definitions, the term “framing” has been used to refer to a number of specific types of strategies over the years, such as “issue framing” (e.g., Slothuus, 2008), “episodic framing” (e.g., Iyengar, 1991), “strategy framing” (e.g., Cappella & Jamieson, 1996) and “value framing” (e.g., Brewer & Gross, 2005; Sniderman & Theriault, 2004). Em-pirical studies of these different types of framing provide specific results on how alternative frames work on different issues (Chong & Druckman, 2007a, p. 117). As many prominent framing scholars have pointed out, however, the “miscellaneous” and “inconsistent” use of the term both within and across disciplines (e.g., de Vreese, 2005, p. 51; Druckman, 2011, p. 279; Entman, 1993, pp. 51–52; Scheufele & Iyengar, 2012, p. 12; Scheufele & Tewksbury, 2007), has led to a conceptual confusion regarding what really counts as a frame (Chong & Druckman, 2007a, p. 114; de Vreese, 2005, p. 53; Entman, 1993, p. 51; Scheufele & Tewksbury, 2007).

With regard to specific sub-types of frames, studies have varied in their definitions of the concepts. Value framing for example, which is the focus of this thesis, is generally described as a strategy in which one specific value dimension of an issue is emphasized (such as a “freedom of speech” dimen-sion of a policy over a “security” dimension). In more precise terms, how-ever, some have described it as a message that “facilitates understanding of a policy conflict” (e.g., Domke, Shah, & Wackman, 1998); while others view it as making “an association between a value and an issue that carries an evaluative implication” (Brewer, 2001, p. 46). Still others have charged value framing with depicting debates about policy issues as “clashes of deep-rooted values” (e.g., Ball-Rokeach, Power, Guthrie, & Waring, 1990; Sniderman & Theirault, 2004). These examples illustrate the richness in substantive meaning of the concept “framing,” only when it comes to the specific sub-type “value framing.”

The linguistic theory on persuasive words differs from framing theory in (at least) two critical respects. In contrast to framing theory, which in-cludes various types of rhetoric, this theory focuses on specific words (e.g., Stevenson, 1944; Walton, 2003, 2006). It presents, more precisely, a lin-guistic account of the exact semantic characteristics which make value-laden words compelling to an audience.

9

The linguistic theory also differs from framing theory when it comes to the proposed effects on the audience. Framing theory targets rhetoric that influences individuals’ attitudes by changing the weight that they assign to certain (existing) beliefs on an issue, when they form their attitudes. The linguistic theory on persuasive words, instead, targets rhetoric that works by changing or introducing new content, to those existing beliefs (conven-tional persuasion)—something that transcends the traditional definition of framing effects (e.g., Chong & Druckman, 2007a, p. 115; Nelson, Oxley, & Clawson, 1997, p. 226; Nelson & Willey, 2001, p. 256). By combining the linguistic theory on persuasive words with the framing theory, this dis-sertation aims to enhance our knowledge about precisely why value-laden rhetoric is so powerful in shaping citizens’ support for policies. Specifi-cally, it investigates whether the “simple” use of certain (value-laden) words can change or introduce new content to individuals’ beliefs about political objects; something that goes beyond the traditional definition of framing effects.

Before discussing the details and implications of this approach, a brief overview of each theory is in order. I will begin with the theory on framing effects.

Theory on Framing Effects

Framing theory builds on an expectancy value model of attitude formation (e.g., Ajzen & Fishbein, 1980). According to this model, (political) objects consist of different dimensions (e.g., a “freedom of speech” dimension and a “security” dimension). Individuals’ attitude to an object is the weighted sum of their evaluative beliefs on the object for each of these dimensions (e.g., Ajzen & Fishbein, 1980). By emphasizing a certain belief dimension, frames make this dimension more salient for individuals, which, in turn, increases the weight they assign to their evaluative beliefs on the object for this dimension when they form their attitudes (e.g., Chong & Druckman, 2007a).

The effectiveness of frames is thought to be conditioned on three factors. First, the frames should be available, meaning that a person is aware of and understands the meaning of a specific belief on an object that is highlighted by the frame (the belief is stored in their long-term memory). Secondly, the frames should be accessible, meaning that an available belief is brought to mind when the individual forms an attitude. Thirdly, the frames should be applicable. This means that an individual should perceive the belief dimen-sion of an object, which is available and accessible, as relevant to account for when they evaluate the object (Chong & Druckman, 2007b, pp. 108–109).

The frame that gains the most power in real-world politics (when citi-zens are likely exposed to competing frames) should, according to Chong

8

Framing has been defined as a communication strategy which “high-lights some bits of information about an item that is the subject of a com-munication, thereby elevating them in salience” (Entman, 1993, p. 53). More precisely, this means that, when presenting a problem or an issue to an audience, an individual makes a piece of information about this issue “more noticeable, meaningful, or memorable” to the audience (Entman, 1993, p. 53). Framing effects, in turn, occur when this selective presentation of certain aspects of an object influences the weight that individuals assign to these aspects when forming their attitudes (see e.g., Nelson, Clawson & Oxley, 1997; Nelson, Oxley, & Clawson, 1997; Price & Tewksbury, 1997).

Apart from these broad definitions, the term “framing” has been used to refer to a number of specific types of strategies over the years, such as “issue framing” (e.g., Slothuus, 2008), “episodic framing” (e.g., Iyengar, 1991), “strategy framing” (e.g., Cappella & Jamieson, 1996) and “value framing” (e.g., Brewer & Gross, 2005; Sniderman & Theriault, 2004). Em-pirical studies of these different types of framing provide specific results on how alternative frames work on different issues (Chong & Druckman, 2007a, p. 117). As many prominent framing scholars have pointed out, however, the “miscellaneous” and “inconsistent” use of the term both within and across disciplines (e.g., de Vreese, 2005, p. 51; Druckman, 2011, p. 279; Entman, 1993, pp. 51–52; Scheufele & Iyengar, 2012, p. 12; Scheufele & Tewksbury, 2007), has led to a conceptual confusion regarding what really counts as a frame (Chong & Druckman, 2007a, p. 114; de Vreese, 2005, p. 53; Entman, 1993, p. 51; Scheufele & Tewksbury, 2007).

With regard to specific sub-types of frames, studies have varied in their definitions of the concepts. Value framing for example, which is the focus of this thesis, is generally described as a strategy in which one specific value dimension of an issue is emphasized (such as a “freedom of speech” dimen-sion of a policy over a “security” dimension). In more precise terms, how-ever, some have described it as a message that “facilitates understanding of a policy conflict” (e.g., Domke, Shah, & Wackman, 1998); while others view it as making “an association between a value and an issue that carries an evaluative implication” (Brewer, 2001, p. 46). Still others have charged value framing with depicting debates about policy issues as “clashes of deep-rooted values” (e.g., Ball-Rokeach, Power, Guthrie, & Waring, 1990; Sniderman & Theirault, 2004). These examples illustrate the richness in substantive meaning of the concept “framing,” only when it comes to the specific sub-type “value framing.”

The linguistic theory on persuasive words differs from framing theory in (at least) two critical respects. In contrast to framing theory, which in-cludes various types of rhetoric, this theory focuses on specific words (e.g., Stevenson, 1944; Walton, 2003, 2006). It presents, more precisely, a lin-guistic account of the exact semantic characteristics which make value-laden words compelling to an audience.

9

The linguistic theory also differs from framing theory when it comes to the proposed effects on the audience. Framing theory targets rhetoric that influences individuals’ attitudes by changing the weight that they assign to certain (existing) beliefs on an issue, when they form their attitudes. The linguistic theory on persuasive words, instead, targets rhetoric that works by changing or introducing new content, to those existing beliefs (conven-tional persuasion)—something that transcends the traditional definition of framing effects (e.g., Chong & Druckman, 2007a, p. 115; Nelson, Oxley, & Clawson, 1997, p. 226; Nelson & Willey, 2001, p. 256). By combining the linguistic theory on persuasive words with the framing theory, this dis-sertation aims to enhance our knowledge about precisely why value-laden rhetoric is so powerful in shaping citizens’ support for policies. Specifi-cally, it investigates whether the “simple” use of certain (value-laden) words can change or introduce new content to individuals’ beliefs about political objects; something that goes beyond the traditional definition of framing effects.

Before discussing the details and implications of this approach, a brief overview of each theory is in order. I will begin with the theory on framing effects.

Theory on Framing Effects

Framing theory builds on an expectancy value model of attitude formation (e.g., Ajzen & Fishbein, 1980). According to this model, (political) objects consist of different dimensions (e.g., a “freedom of speech” dimension and a “security” dimension). Individuals’ attitude to an object is the weighted sum of their evaluative beliefs on the object for each of these dimensions (e.g., Ajzen & Fishbein, 1980). By emphasizing a certain belief dimension, frames make this dimension more salient for individuals, which, in turn, increases the weight they assign to their evaluative beliefs on the object for this dimension when they form their attitudes (e.g., Chong & Druckman, 2007a).

The effectiveness of frames is thought to be conditioned on three factors. First, the frames should be available, meaning that a person is aware of and understands the meaning of a specific belief on an object that is highlighted by the frame (the belief is stored in their long-term memory). Secondly, the frames should be accessible, meaning that an available belief is brought to mind when the individual forms an attitude. Thirdly, the frames should be applicable. This means that an individual should perceive the belief dimen-sion of an object, which is available and accessible, as relevant to account for when they evaluate the object (Chong & Druckman, 2007b, pp. 108–109).

The frame that gains the most power in real-world politics (when citi-zens are likely exposed to competing frames) should, according to Chong

10

and Druckman (2007b), depend on 1) which frame is most prevalent, and 2) which frame is strongest. The prevalence model suggests that the fre-quent exposure to a frame can make certain available beliefs more accessi-ble; hence the frame that is most prevalent will exert the greatest influence on peoples’ attitudes (Chong & Druckman, 2007b, p. 104). The strength model suggests that, regardless of repetition, the strongest5 frame will exert the greatest influence (Chong & Druckman, 2007b, p. 104). Framing effects are then mediated by their applicability on an issue (e.g., Eagly & Chaiken, 1993; Nelson, Clawson, & Oxley, 1997). One study that illustrates this is Shen (2004), who found that the effects of value frames were constrained by whether or not individuals, prior to exposure to the frames, viewed the policies as being of an ethical nature (and thereby compatible with a frame that speaks about the policies in terms of a value). Another example is Brewer (2002), who showed that individuals, when asked to motivate their views on gay rights, referred to values used in the framing of the issue in ways that both reinforced and challenged the frame.

In sum, both the “prevalence,” and the “strength” models suggest that frames work by increasing the weight of specific belief dimensions on ob-jects that already are existing in individuals’ long-term memory (Nelson, Clawson, & Oxley, 1997, pp. 568–569; Nelson, Oxley, & Clawson, 1997; Nelson & Oxley, 1999; Price & Tewksbury, 1997).

More recently, these models have been complemented by a third model to help explain framing effects, the “content change model” (see e.g., Cal-laghan & Schnell, 2005, p. 13; Slothuus, 2008, pp. 5–6). Supported by some empirical evidence (Chong & Druckman, 2007b, p. 104), this model sug-gests that frames can also work by adding to, or changing, the content of individuals’ existing beliefs on an object (e.g., Chong & Druckman, 2007a, pp. 115–116; Nelson & Oxley, 1999; Slothuus, 2008, pp. 4–10). These find-ings indicate that frames sometimes work as conventional persuasive mes-sages, hence challenging the traditional definition of framing effects as changes in the relative importance of certain (already existing) beliefs (e.g., Nelson, Oxley, & Clawson, 1997, pp. 225–227; Nelson & Oxley, 1999).

While the content change model has raised questions as to what should count as a “true” framing effect (see e.g., Chong & Druckman, 2007a, p. 115; Nelson & Oxley, 1999; Nelson & Willey, 2001, p. 256; Slothuus, 2008, p. 22), there are currently few studies on the extent to which media-tors of framing effects and persuasion overlap (Chong & Druckman, 2007a, p. 115; Slothuus, 2008, p. 22). This issue is addressed at length in one of the papers in this dissertation.

5 Some examples of qualities that can increase the strength of frames are that they are value-laden (Chong & Druckman, 2007a, p. 112), come from a credible source (Druckman, 2001b), and trigger emotional reactions (e.g., Gross, 2008).

11

Linguistic Theory on Persuasive Words

As opposed to framing theory, which targets rhetoric that influences atti-tudes by making existing beliefs on an object seem more important, the linguistic theory targets rhetoric that works by changing or adding new con-tent, to those existing beliefs (see e.g., Schiappa, 2003; Zarefsky, 2004). The theory uses three semantic characteristics to define persuasive words.

First, persuasive words have value connotations; they appeal to values and norms that identify what is “good” and “bad” (e.g., Hare, 1952; Ste-venson, 1944; Walton, 2003, pp. 129–131; Walton, 2006, pp. 220–221). This triggers value judgments (Macagno, 2014, p. 104) and indicates for the audience what should, or should not, be done (Stevenson, 1944, pp. 18–26). An example of a positive value-laden word is “freedom;” an example of a negative word is “racism.”

Linguistic theory distinguishes between words’ “emotive” and “descrip-tive” meaning. When a word has an emotive meaning, this means that it has positive (or negative) connotations (Stevenson, 1944, p. 60), whereas de-scriptive meaning refers to words’ substantive connotations (Walton 2001, p. 118). The word “peacekeeping” for example, which typically has posi-tive connotations, also has a descriptive meaning, such as “international enforcement and supervision of a truce between hostile states or communi-ties” (see merriam-webster.com).

The second characteristic of persuasive words is that they have a strong emotive meaning, which is, at least partially, independent of their descrip-tive meaning (e.g., Macagno & Walton, 2010; Macagno, 2014, p. 104; Ste-venson, 1944, p. 88; Walton, 2003). That a word has a strong emotive meaning stipulates that individuals who feel strongly about the word (i.e., those who endorse the values associated with the word) will encounter the same (positive) emotional perceptions, regardless of how exactly the word is descriptively defined (e.g., Stevenson, 1944, p. 141, pp. 210–212; Wal-ton, 2001, pp. 118–122).

The third characteristic of persuasive words is that they have a vague and flexible descriptive meaning. This means that the words can be used when referring to diverse content (Stevenson, 1944, pp. 34–35; Zarefsky, 2006, p. 405), and are applicable to different situations and issues (Steven-son, 1944, p. 208). Consider again the word “peacekeeping:” it has a num-ber of subjective substantive meanings. One person may, for example, say that peacekeepers are military troops who, in critical situations, intervene defensively with a certain amount of violence. Others may say that peace-

10

and Druckman (2007b), depend on 1) which frame is most prevalent, and 2) which frame is strongest. The prevalence model suggests that the fre-quent exposure to a frame can make certain available beliefs more accessi-ble; hence the frame that is most prevalent will exert the greatest influence on peoples’ attitudes (Chong & Druckman, 2007b, p. 104). The strength model suggests that, regardless of repetition, the strongest5 frame will exert the greatest influence (Chong & Druckman, 2007b, p. 104). Framing effects are then mediated by their applicability on an issue (e.g., Eagly & Chaiken, 1993; Nelson, Clawson, & Oxley, 1997). One study that illustrates this is Shen (2004), who found that the effects of value frames were constrained by whether or not individuals, prior to exposure to the frames, viewed the policies as being of an ethical nature (and thereby compatible with a frame that speaks about the policies in terms of a value). Another example is Brewer (2002), who showed that individuals, when asked to motivate their views on gay rights, referred to values used in the framing of the issue in ways that both reinforced and challenged the frame.

In sum, both the “prevalence,” and the “strength” models suggest that frames work by increasing the weight of specific belief dimensions on ob-jects that already are existing in individuals’ long-term memory (Nelson, Clawson, & Oxley, 1997, pp. 568–569; Nelson, Oxley, & Clawson, 1997; Nelson & Oxley, 1999; Price & Tewksbury, 1997).

More recently, these models have been complemented by a third model to help explain framing effects, the “content change model” (see e.g., Cal-laghan & Schnell, 2005, p. 13; Slothuus, 2008, pp. 5–6). Supported by some empirical evidence (Chong & Druckman, 2007b, p. 104), this model sug-gests that frames can also work by adding to, or changing, the content of individuals’ existing beliefs on an object (e.g., Chong & Druckman, 2007a, pp. 115–116; Nelson & Oxley, 1999; Slothuus, 2008, pp. 4–10). These find-ings indicate that frames sometimes work as conventional persuasive mes-sages, hence challenging the traditional definition of framing effects as changes in the relative importance of certain (already existing) beliefs (e.g., Nelson, Oxley, & Clawson, 1997, pp. 225–227; Nelson & Oxley, 1999).

While the content change model has raised questions as to what should count as a “true” framing effect (see e.g., Chong & Druckman, 2007a, p. 115; Nelson & Oxley, 1999; Nelson & Willey, 2001, p. 256; Slothuus, 2008, p. 22), there are currently few studies on the extent to which media-tors of framing effects and persuasion overlap (Chong & Druckman, 2007a, p. 115; Slothuus, 2008, p. 22). This issue is addressed at length in one of the papers in this dissertation.

5 Some examples of qualities that can increase the strength of frames are that they are value-laden (Chong & Druckman, 2007a, p. 112), come from a credible source (Druckman, 2001b), and trigger emotional reactions (e.g., Gross, 2008).

11

Linguistic Theory on Persuasive Words

As opposed to framing theory, which targets rhetoric that influences atti-tudes by making existing beliefs on an object seem more important, the linguistic theory targets rhetoric that works by changing or adding new con-tent, to those existing beliefs (see e.g., Schiappa, 2003; Zarefsky, 2004). The theory uses three semantic characteristics to define persuasive words.

First, persuasive words have value connotations; they appeal to values and norms that identify what is “good” and “bad” (e.g., Hare, 1952; Ste-venson, 1944; Walton, 2003, pp. 129–131; Walton, 2006, pp. 220–221). This triggers value judgments (Macagno, 2014, p. 104) and indicates for the audience what should, or should not, be done (Stevenson, 1944, pp. 18–26). An example of a positive value-laden word is “freedom;” an example of a negative word is “racism.”

Linguistic theory distinguishes between words’ “emotive” and “descrip-tive” meaning. When a word has an emotive meaning, this means that it has positive (or negative) connotations (Stevenson, 1944, p. 60), whereas de-scriptive meaning refers to words’ substantive connotations (Walton 2001, p. 118). The word “peacekeeping” for example, which typically has posi-tive connotations, also has a descriptive meaning, such as “international enforcement and supervision of a truce between hostile states or communi-ties” (see merriam-webster.com).

The second characteristic of persuasive words is that they have a strong emotive meaning, which is, at least partially, independent of their descrip-tive meaning (e.g., Macagno & Walton, 2010; Macagno, 2014, p. 104; Ste-venson, 1944, p. 88; Walton, 2003). That a word has a strong emotive meaning stipulates that individuals who feel strongly about the word (i.e., those who endorse the values associated with the word) will encounter the same (positive) emotional perceptions, regardless of how exactly the word is descriptively defined (e.g., Stevenson, 1944, p. 141, pp. 210–212; Wal-ton, 2001, pp. 118–122).

The third characteristic of persuasive words is that they have a vague and flexible descriptive meaning. This means that the words can be used when referring to diverse content (Stevenson, 1944, pp. 34–35; Zarefsky, 2006, p. 405), and are applicable to different situations and issues (Steven-son, 1944, p. 208). Consider again the word “peacekeeping:” it has a num-ber of subjective substantive meanings. One person may, for example, say that peacekeepers are military troops who, in critical situations, intervene defensively with a certain amount of violence. Others may say that peace-

12

keepers employ strictly non-violent actions such as organizing, and main-taining peace negotiations. Neither of these definitions is necessarily less true than the other.6

Persuasive Words in Argumentation

In linguistic theory, scholars distinguish between rational (reason-driven) and emotion-driven persuasive argumentation. Rational argumentation re-fers to processes when attitude change is driven by factual reasoning. In rational argumentation, a speaker gives a subjective judgment about some-thing— “this is good,” or “I approve of this”—and provides a factual reason (which can be objectively observed and evaluated) as to why the listener should like it, too. The listener can then evaluate the factual reason to decide whether or not she wishes to change her attitudes (Stevenson, 1944, pp. 81–138). But persuasion also occurs through appeals to emotions. As Steven-son (pp. 139–140) puts it: every person, no matter how “reason loving,” should be influenced more strongly when appeals to emotions are involved.

The linguistic theory describes two strategies in which persuasive words can be used to persuade individuals about political objects. The first strat-egy involves the use of persuasive words without defining their descriptive meaning (henceforth Persuasive words). The second strategy involves the use of persuasive words, while wedding them to a specific descriptive meaning (henceforth Persuasive definitions) (Stevenson, 1944, p. 206).

Persuasive words. Since persuasive words have emotive and value-laden connotations, they tell an audience that a speaker finds an object of discussion positive or morally right (Stevenson, 1944, p. 207)—emotion-driven persuasion. Since persuasive words have substantive meanings for individuals, they also appear to be substantively informative (Walton, 2006, p. 218). This gives an incentive to the listeners to use them to interpret what an object (that is described by the word) is about—hence, the words also appear to provide a factual reason as to why this object is good—reason-driven persuasion.

This is problematic for the following reasons. Since persuasive words are linguistically flexible and can have many different meanings, individu-als will likely come to different conclusions about the object. As Stevenson (1944, pp. 87–88) puts it: “the judgement ‘X is good’ may suggest to one hearer that X has certain qualities and to another that X has certain other qualities; whereas the speaker may have had in mind still another set of qualities.” Therefore, unless the speaker defines the persuasive word, it

6 Persuasive words can be compared to words such as “tree,” “toothpaste,” and “hiking.” These words are not particularly value-laden, they are usually not perceived as strongly negative or positive, and their descriptive meaning is fairly straightforward—there is little significant disagreement on how to define their descriptive meaning.

13

should not be seen as a rational argument for an object (Stevenson, 1944, p. 206; Walton, 2006, pp. 225–226).

Persuasive definitions. The second strategy—persuasive definitions—differs from the first, in that the speaker defines the persuasive word. The aim with this strategy is to redirect the listeners’ attitudes by keeping the emotive meaning of the persuasive word, while wedding it to a favored (subjective) descriptive meaning that supports the speakers’ position (Ste-venson, 1944, pp. 210–214; Walton, 2006, pp. 247–248); when I use this word X in relation to this object Y, I mean Z (which differs from the con-ventional meaning Q) (Stevenson, 1944, p. 207). In this way, the speaker provides a factual reason as to why something should be viewed as good—which would otherwise only be indicated by the words—that can be criti-cally evaluated by the listeners (Macagno & Walton, 2010, pp. 2010–2012).

According to the theory however, we can expect people to accept per-suasive definitions without such critical examination for at least three rea-sons. First, since the emotive meaning of persuasive words is stronger than their descriptive, the listeners will “automatically” transfer the emotive meaning from one (often the conventional that you will find in a dictionary) to another (subjectively stipulated) definition of a word, when such is done (Stevenson 1944, pp. 210–212, p. 333; Walton, 2001, p. 119; Walton, 2006, p. 248). As Stevenson (1944, pp. 211–212) puts it: “When people learn to call something by a name, they more readily admire it, and when they learn not to call it by such a name, they less readily admire it.”

Secondly, persuasive definitions sound like “honest” (Walton, 2006, p. 248) and “uncontroversial” (Zarevsky, 2006, p. 404) descriptions of words. Therefore, they look less like stating “a point of view,” and more like “pure intellectual” or “scientific” statements, which can be associated with ra-tional argumentation (Stevenson, 1944, p. 242).7 Moreover, the proposed definitions of persuasive words are often “smuggled in” in messages (Za-revsky, 2006, p. 404), hence the changes in descriptive meaning are per-ceived as trivial. The descriptive meaning of persuasive words can thus be changed and the emotions they evoke directed towards new objects, with-out people even making note of this (e.g., Stevenson, 1944; Walton, 2001, 2006).

A third reason why individuals may accept persuasive definitions is simply that, in order to question them, they would have to get into a “com-plicated discussion about meanings of words and definitions” (Walton, 2006, pp. 264–265).

Zarefsky (2004, p. 615) gives an illustrating example of an instance when persuasive definitions have been used to redirect voters’ attitudes to-ward policies. In his campaign speeches in 1932, the American presidential

7 Words such as “real” and “true” are often used in relation to persuasive definitions, to indicate that the definition is to be understood as the “accurate” one (Stevenson, 1944, pp. 213–215).

12

keepers employ strictly non-violent actions such as organizing, and main-taining peace negotiations. Neither of these definitions is necessarily less true than the other.6

Persuasive Words in Argumentation

In linguistic theory, scholars distinguish between rational (reason-driven) and emotion-driven persuasive argumentation. Rational argumentation re-fers to processes when attitude change is driven by factual reasoning. In rational argumentation, a speaker gives a subjective judgment about some-thing— “this is good,” or “I approve of this”—and provides a factual reason (which can be objectively observed and evaluated) as to why the listener should like it, too. The listener can then evaluate the factual reason to decide whether or not she wishes to change her attitudes (Stevenson, 1944, pp. 81–138). But persuasion also occurs through appeals to emotions. As Steven-son (pp. 139–140) puts it: every person, no matter how “reason loving,” should be influenced more strongly when appeals to emotions are involved.

The linguistic theory describes two strategies in which persuasive words can be used to persuade individuals about political objects. The first strat-egy involves the use of persuasive words without defining their descriptive meaning (henceforth Persuasive words). The second strategy involves the use of persuasive words, while wedding them to a specific descriptive meaning (henceforth Persuasive definitions) (Stevenson, 1944, p. 206).

Persuasive words. Since persuasive words have emotive and value-laden connotations, they tell an audience that a speaker finds an object of discussion positive or morally right (Stevenson, 1944, p. 207)—emotion-driven persuasion. Since persuasive words have substantive meanings for individuals, they also appear to be substantively informative (Walton, 2006, p. 218). This gives an incentive to the listeners to use them to interpret what an object (that is described by the word) is about—hence, the words also appear to provide a factual reason as to why this object is good—reason-driven persuasion.

This is problematic for the following reasons. Since persuasive words are linguistically flexible and can have many different meanings, individu-als will likely come to different conclusions about the object. As Stevenson (1944, pp. 87–88) puts it: “the judgement ‘X is good’ may suggest to one hearer that X has certain qualities and to another that X has certain other qualities; whereas the speaker may have had in mind still another set of qualities.” Therefore, unless the speaker defines the persuasive word, it

6 Persuasive words can be compared to words such as “tree,” “toothpaste,” and “hiking.” These words are not particularly value-laden, they are usually not perceived as strongly negative or positive, and their descriptive meaning is fairly straightforward—there is little significant disagreement on how to define their descriptive meaning.

13

should not be seen as a rational argument for an object (Stevenson, 1944, p. 206; Walton, 2006, pp. 225–226).

Persuasive definitions. The second strategy—persuasive definitions—differs from the first, in that the speaker defines the persuasive word. The aim with this strategy is to redirect the listeners’ attitudes by keeping the emotive meaning of the persuasive word, while wedding it to a favored (subjective) descriptive meaning that supports the speakers’ position (Ste-venson, 1944, pp. 210–214; Walton, 2006, pp. 247–248); when I use this word X in relation to this object Y, I mean Z (which differs from the con-ventional meaning Q) (Stevenson, 1944, p. 207). In this way, the speaker provides a factual reason as to why something should be viewed as good—which would otherwise only be indicated by the words—that can be criti-cally evaluated by the listeners (Macagno & Walton, 2010, pp. 2010–2012).

According to the theory however, we can expect people to accept per-suasive definitions without such critical examination for at least three rea-sons. First, since the emotive meaning of persuasive words is stronger than their descriptive, the listeners will “automatically” transfer the emotive meaning from one (often the conventional that you will find in a dictionary) to another (subjectively stipulated) definition of a word, when such is done (Stevenson 1944, pp. 210–212, p. 333; Walton, 2001, p. 119; Walton, 2006, p. 248). As Stevenson (1944, pp. 211–212) puts it: “When people learn to call something by a name, they more readily admire it, and when they learn not to call it by such a name, they less readily admire it.”

Secondly, persuasive definitions sound like “honest” (Walton, 2006, p. 248) and “uncontroversial” (Zarevsky, 2006, p. 404) descriptions of words. Therefore, they look less like stating “a point of view,” and more like “pure intellectual” or “scientific” statements, which can be associated with ra-tional argumentation (Stevenson, 1944, p. 242).7 Moreover, the proposed definitions of persuasive words are often “smuggled in” in messages (Za-revsky, 2006, p. 404), hence the changes in descriptive meaning are per-ceived as trivial. The descriptive meaning of persuasive words can thus be changed and the emotions they evoke directed towards new objects, with-out people even making note of this (e.g., Stevenson, 1944; Walton, 2001, 2006).

A third reason why individuals may accept persuasive definitions is simply that, in order to question them, they would have to get into a “com-plicated discussion about meanings of words and definitions” (Walton, 2006, pp. 264–265).

Zarefsky (2004, p. 615) gives an illustrating example of an instance when persuasive definitions have been used to redirect voters’ attitudes to-ward policies. In his campaign speeches in 1932, the American presidential

7 Words such as “real” and “true” are often used in relation to persuasive definitions, to indicate that the definition is to be understood as the “accurate” one (Stevenson, 1944, pp. 213–215).

14

candidate Franklin D. Roosevelt used the word “liberalism” when he advo-cated for governmental programs. At the time, liberalism, which was re-garded positively in the US, had the conventional (or “lexical”) descriptive meaning of “employing minimal government intervention in the life of the individual.” By arguing that individual freedom was threatened by the “power of big businesses,” Roosevelt redefined liberalism as meaning ex-actly the opposite. According to him, individuals could not bargain freely in the marketplace when power relations were so unbalanced, hence indi-vidual freedom could be protected only by establishing a counteracting power such as a “big government.” Making use of the strong value, and subsequent emotive meaning of the word liberalism, while redefining it in a way that suited his goals, Roosevelt succeeded to gain popular support for developing governmental programs—something that actually meant major reforms of the liberal policies at the time.

To summarize, the linguistic theory stipulates that persuasive words can be used, both with and without a descriptive definition, to alter individuals’ beliefs on, and thereby their support for policies. In this thesis, I focus on the first strategy, the use of persuasive words without definition. I have chosen to do this because I believe that an empirical understanding of the impact of persuasive words in politics is crucial in order to fully understand the effects that persuasive definitions have as an extension of this strategy. Although this thesis does not cover the effects of persuasive definitions in its empirical studies, this type of persuasive rhetoric, and how it can impact voters’ evaluations of policies pre- and post- elections, deserves attention in future research projects.

Contributions

Theoretical Contributions

This thesis contributes to the broader field of political science, as well as to the specific field of political communication, by connecting two communi-cation theories—framing theory and linguistic theory on persuasive words.

Both theories provide important insights into how political rhetoric can impact citizens’ reactions to policy pledges, for example, and thereby also the representative process as citizens use election pledges to compare their policy preferences to parties’ policy intentions in elections. Framing theory and linguistic theory on persuasive words share the idea that value-based rhetoric can be an effective tool for influencing citizens’ support for poli-cies in politics. However, these theories offer different explanations as to why appealing to core values should be this effective. By bringing together these two theories—which have not yet been compared or used in a cross-

15

disciplinary way—this dissertation provides a possible explanation for framing effects; one that is not often lifted by framing scholars. It also de-velops the theory on persuasive words, by suggesting that persuasive words can be divided into different types, with the effectiveness limited by ideo-logical constrains.

Regarding the framing literature, more precisely, the dissertation sug-gests a complementary explanation of what has recently been acknowl-edged by framing scholars; that framing effects differ in theory and empir-ical studies. These studies have shown that frames, aside from increasing the weight that individuals assign to certain existing beliefs about policies over others (“true” framing effects), sometimes also change, or add new content to those existing beliefs (conventional persuasion) (Nelson & Ox-ley, 1999; Slothuus, 2008). Scholars propose that one explanation for this is that empirical studies of framing effects tend to include both framing (a selective emphasis of certain aspects of the policy), and new policy-relevant information. It is this new information, rather than the frames, which likely changes the content of individuals’ beliefs about policies (see e.g., Calla-ghan & Schnell, 2005, p. 13; Leeper & Slothuus, 2015; Slothuus, 2008, pp. 5–6).

Based on what linguistic theory says about persuasive words, I suggest that certain words can impact individuals’ beliefs on policies, even without presenting any additional factual information. Since persuasive words have characteristics that (falsely) make them appear to be substantively informa-tive, merely the reference to such words in election pledges, without the addition of supportive policy-relevant information, can provoke new be-liefs about the policies and their consequences. This implicates that it is not necessarily framing (i.e., emphasis)—or new information—which pro-duces changes in belief content, but rather specific words that are being used as parts of these frames.

To the field of linguistic semantics, this thesis contributes theoretically by using insights from the framing research; it reinforces that values in real-world politics are often ideologically tainted, and therefore appeal to indi-viduals of different ideological predispositions.

The linguistic theory focuses on persuasive words that appeal to “mass audiences” with diverse principles (e.g., Walton, 2003, pp. 141–142); it rarely differentiates between words that appeal to some, and words that ap-peal to “everyone.” In this thesis, I argue that real-world politics should make these kinds of distinctions.

First, empirical studies of value framing indicate that individuals assign different weight to different values depending on their ideological predis-positions, which in turn affects the impact of value-laden rhetoric (e.g., Schemer, Wirth, & Matthes, 2012; Shen & Edwards, 2005; Sniderman &

14

candidate Franklin D. Roosevelt used the word “liberalism” when he advo-cated for governmental programs. At the time, liberalism, which was re-garded positively in the US, had the conventional (or “lexical”) descriptive meaning of “employing minimal government intervention in the life of the individual.” By arguing that individual freedom was threatened by the “power of big businesses,” Roosevelt redefined liberalism as meaning ex-actly the opposite. According to him, individuals could not bargain freely in the marketplace when power relations were so unbalanced, hence indi-vidual freedom could be protected only by establishing a counteracting power such as a “big government.” Making use of the strong value, and subsequent emotive meaning of the word liberalism, while redefining it in a way that suited his goals, Roosevelt succeeded to gain popular support for developing governmental programs—something that actually meant major reforms of the liberal policies at the time.

To summarize, the linguistic theory stipulates that persuasive words can be used, both with and without a descriptive definition, to alter individuals’ beliefs on, and thereby their support for policies. In this thesis, I focus on the first strategy, the use of persuasive words without definition. I have chosen to do this because I believe that an empirical understanding of the impact of persuasive words in politics is crucial in order to fully understand the effects that persuasive definitions have as an extension of this strategy. Although this thesis does not cover the effects of persuasive definitions in its empirical studies, this type of persuasive rhetoric, and how it can impact voters’ evaluations of policies pre- and post- elections, deserves attention in future research projects.

Contributions

Theoretical Contributions

This thesis contributes to the broader field of political science, as well as to the specific field of political communication, by connecting two communi-cation theories—framing theory and linguistic theory on persuasive words.

Both theories provide important insights into how political rhetoric can impact citizens’ reactions to policy pledges, for example, and thereby also the representative process as citizens use election pledges to compare their policy preferences to parties’ policy intentions in elections. Framing theory and linguistic theory on persuasive words share the idea that value-based rhetoric can be an effective tool for influencing citizens’ support for poli-cies in politics. However, these theories offer different explanations as to why appealing to core values should be this effective. By bringing together these two theories—which have not yet been compared or used in a cross-

15

disciplinary way—this dissertation provides a possible explanation for framing effects; one that is not often lifted by framing scholars. It also de-velops the theory on persuasive words, by suggesting that persuasive words can be divided into different types, with the effectiveness limited by ideo-logical constrains.

Regarding the framing literature, more precisely, the dissertation sug-gests a complementary explanation of what has recently been acknowl-edged by framing scholars; that framing effects differ in theory and empir-ical studies. These studies have shown that frames, aside from increasing the weight that individuals assign to certain existing beliefs about policies over others (“true” framing effects), sometimes also change, or add new content to those existing beliefs (conventional persuasion) (Nelson & Ox-ley, 1999; Slothuus, 2008). Scholars propose that one explanation for this is that empirical studies of framing effects tend to include both framing (a selective emphasis of certain aspects of the policy), and new policy-relevant information. It is this new information, rather than the frames, which likely changes the content of individuals’ beliefs about policies (see e.g., Calla-ghan & Schnell, 2005, p. 13; Leeper & Slothuus, 2015; Slothuus, 2008, pp. 5–6).

Based on what linguistic theory says about persuasive words, I suggest that certain words can impact individuals’ beliefs on policies, even without presenting any additional factual information. Since persuasive words have characteristics that (falsely) make them appear to be substantively informa-tive, merely the reference to such words in election pledges, without the addition of supportive policy-relevant information, can provoke new be-liefs about the policies and their consequences. This implicates that it is not necessarily framing (i.e., emphasis)—or new information—which pro-duces changes in belief content, but rather specific words that are being used as parts of these frames.

To the field of linguistic semantics, this thesis contributes theoretically by using insights from the framing research; it reinforces that values in real-world politics are often ideologically tainted, and therefore appeal to indi-viduals of different ideological predispositions.

The linguistic theory focuses on persuasive words that appeal to “mass audiences” with diverse principles (e.g., Walton, 2003, pp. 141–142); it rarely differentiates between words that appeal to some, and words that ap-peal to “everyone.” In this thesis, I argue that real-world politics should make these kinds of distinctions.

First, empirical studies of value framing indicate that individuals assign different weight to different values depending on their ideological predis-positions, which in turn affects the impact of value-laden rhetoric (e.g., Schemer, Wirth, & Matthes, 2012; Shen & Edwards, 2005; Sniderman &

16

Theirault, 2004). “Individualism,” for example, seems to attract right-ori-ented individuals, whereas “humanitarianism” seems to attract mainly peo-ple on the left.

Second, electoral competitions tend to intensify the ideological left-right conflict (e.g., Jost, Federico, & Napier, 2009) for parties and voters, which should make these types of ideological values particularly important in election campaigns. Political parties in real-life campaigns can, in other words, be expected to use ideological left-right values to appeal to voters who share their views; voters, for their part, can be expected to focus on these types of values when they evaluate parties’ policy pledges. Simply put, persuasive words in election pledges will likely, 1) indicate different types of left- and right-leaning values, and 2) have their effectiveness con-ditioned on the voters’ ideological predispositions.

In this thesis, I therefore propose that persuasive words should be di-vided into Ideological persuasive words (persuasive words that connote left- or right values), and Universal persuasive words (persuasive words that are ideologically neutral and appreciated across the left-right spec-trum). I also propose that the theory on persuasive words should take into account the fact that the type of values being indicated by persuasive words will likely determine the range of their effectiveness in real-world politics.

Empirical Contributions

In its empirical parts, this thesis applies the linguistic theory on persuasive words to an election campaign context, and it tests the effects of such words in election pledges on voters’ evaluations of policies, before and after elec-tions.

This dissertation contributes to the linguistic theory by assessing the im-plications of the theory on persuasive words empirically (which has thus far rarely been done). It also contributes by differentiating between differ-ent types of persuasive words (i.e., words that are associated with universal-, left-, or right-leaning values), and testing whether the effects of such words are limited by individuals’ ideological predispositions.

This thesis also contributes with empirical insights on value framing ef-fects. As these types of frames sometimes include persuasive words (value-laden, linguistically flexible, and emotively charged words), empirical tests of the effects of persuasive words in election pledges will also, indirectly, test whether frames can change citizens’ beliefs about policies (rather than emphasizing pre-existing beliefs)—something that would transcend the tra-ditional definition of framing effects.

Finally, by studying whether election pledge rhetoric can provoke be-liefs about policies that stretch beyond what is actually being pledged, this thesis also reflects on the role of election pledge rhetoric in the functioning of mandate models of representation, which presupposes that election

17

pledges are formulated so that citizens can make accurate predictions for the future policy intentions of parties, pre-election (Downs, 1957; Manin, Przeworski & Stokes, 1999).

Hypotheses

Drawing on the linguistic theory, I have made four predictions about the effects of persuasive words in election pledges, before and after elections. To test these predictions, I have developed corresponding hypotheses.

Pre-Election Effects of Persuasive Words in Election Pledges

I offer three predictions about the effects of using persuasive words in elec-tion pledges, pre-election (detailed in three hypotheses, H1–H3).

First, I suggest that the use of persuasive words in election pledges can elicit beliefs about policies, which relate to the words, but stretch beyond the exact policy proposals outlined in the pledges. The theoretical principle behind this proposition is that persuasive words have substantive meaning for individuals (which can be used to interpret the content of the policies), while also being linguistically flexible and referring to diverse content (e.g., Walton, 2006). Precisely, the first hypothesis stipulates that:

H1) Persuasive words in election pledges can elicit beliefs about policies

that go beyond the exact policy proposals. Secondly, I suggest that the use of persuasive words in election pledges

can help parties muster support for policies, without becoming bound to further policy actions. This prediction takes into account the first hypothe-sis, that persuasive words can elicit substantive beliefs about policies (which appeal to voters’ core values), but which stretch beyond the exact policy proposals. By doing this, parties should be able to raise voters’ ex-pectations on policies and, indirectly, their support for the same, without presenting additional measures upon which their actions can be objectively observed and evaluated, post-election.

The second hypothesis thus shifts attention to voters’ support for poli-cies. It specifically highlights how voters’ beliefs about policies mediate in the evaluative process:

H2) Persuasive words in election pledges can increase citizens’ policy

support by eliciting beliefs about the policies that make individuals per-ceive the policies more positively than what they would otherwise have done based solely on their factual content.

16

Theirault, 2004). “Individualism,” for example, seems to attract right-ori-ented individuals, whereas “humanitarianism” seems to attract mainly peo-ple on the left.

Second, electoral competitions tend to intensify the ideological left-right conflict (e.g., Jost, Federico, & Napier, 2009) for parties and voters, which should make these types of ideological values particularly important in election campaigns. Political parties in real-life campaigns can, in other words, be expected to use ideological left-right values to appeal to voters who share their views; voters, for their part, can be expected to focus on these types of values when they evaluate parties’ policy pledges. Simply put, persuasive words in election pledges will likely, 1) indicate different types of left- and right-leaning values, and 2) have their effectiveness con-ditioned on the voters’ ideological predispositions.

In this thesis, I therefore propose that persuasive words should be di-vided into Ideological persuasive words (persuasive words that connote left- or right values), and Universal persuasive words (persuasive words that are ideologically neutral and appreciated across the left-right spec-trum). I also propose that the theory on persuasive words should take into account the fact that the type of values being indicated by persuasive words will likely determine the range of their effectiveness in real-world politics.

Empirical Contributions

In its empirical parts, this thesis applies the linguistic theory on persuasive words to an election campaign context, and it tests the effects of such words in election pledges on voters’ evaluations of policies, before and after elec-tions.

This dissertation contributes to the linguistic theory by assessing the im-plications of the theory on persuasive words empirically (which has thus far rarely been done). It also contributes by differentiating between differ-ent types of persuasive words (i.e., words that are associated with universal-, left-, or right-leaning values), and testing whether the effects of such words are limited by individuals’ ideological predispositions.

This thesis also contributes with empirical insights on value framing ef-fects. As these types of frames sometimes include persuasive words (value-laden, linguistically flexible, and emotively charged words), empirical tests of the effects of persuasive words in election pledges will also, indirectly, test whether frames can change citizens’ beliefs about policies (rather than emphasizing pre-existing beliefs)—something that would transcend the tra-ditional definition of framing effects.

Finally, by studying whether election pledge rhetoric can provoke be-liefs about policies that stretch beyond what is actually being pledged, this thesis also reflects on the role of election pledge rhetoric in the functioning of mandate models of representation, which presupposes that election

17

pledges are formulated so that citizens can make accurate predictions for the future policy intentions of parties, pre-election (Downs, 1957; Manin, Przeworski & Stokes, 1999).

Hypotheses

Drawing on the linguistic theory, I have made four predictions about the effects of persuasive words in election pledges, before and after elections. To test these predictions, I have developed corresponding hypotheses.

Pre-Election Effects of Persuasive Words in Election Pledges

I offer three predictions about the effects of using persuasive words in elec-tion pledges, pre-election (detailed in three hypotheses, H1–H3).

First, I suggest that the use of persuasive words in election pledges can elicit beliefs about policies, which relate to the words, but stretch beyond the exact policy proposals outlined in the pledges. The theoretical principle behind this proposition is that persuasive words have substantive meaning for individuals (which can be used to interpret the content of the policies), while also being linguistically flexible and referring to diverse content (e.g., Walton, 2006). Precisely, the first hypothesis stipulates that:

H1) Persuasive words in election pledges can elicit beliefs about policies

that go beyond the exact policy proposals. Secondly, I suggest that the use of persuasive words in election pledges

can help parties muster support for policies, without becoming bound to further policy actions. This prediction takes into account the first hypothe-sis, that persuasive words can elicit substantive beliefs about policies (which appeal to voters’ core values), but which stretch beyond the exact policy proposals. By doing this, parties should be able to raise voters’ ex-pectations on policies and, indirectly, their support for the same, without presenting additional measures upon which their actions can be objectively observed and evaluated, post-election.

The second hypothesis thus shifts attention to voters’ support for poli-cies. It specifically highlights how voters’ beliefs about policies mediate in the evaluative process:

H2) Persuasive words in election pledges can increase citizens’ policy

support by eliciting beliefs about the policies that make individuals per-ceive the policies more positively than what they would otherwise have done based solely on their factual content.

18

Thirdly, I propose that the effectiveness of persuasive words on policy support will vary due to the ideological predispositions of the audience.

While linguistic theory focuses on persuasive words that appeal to “mass audiences” of diverse principles (Walton, 2003, pp. 141–142), I argue that there are valid reasons to distinguish between different types of persuasive words (universal persuasive words, and ideological “left“ and “right“ lean-ing words), and to expect that the effectiveness of such words might be limited by individuals’ ideological predispositions (see e.g., Schemer, Wirth, & Matthes, 2012; Shen & Edwards, 2005, for prior findings).

An example of a universal persuasive word (i.e., a word associated with values appreciated beyond the ideological left-right divide), would be “knowledge;” examples of ideological persuasive words would be “equal-ity” (left-leaning), and “competitiveness” (right-leaning). I expect that in-dividuals who place themselves on the left will support policies more (than individuals to the right and center), when they interpret them as a means of achieving a left-leaning value (indicated by the use of a persuasive word like “equality”). Individuals who place themselves to the right will support policies more (than individuals to the left and center), when they interpret them as a means of achieving a right-leaning value (indicated by the use of a persuasive word like “competitiveness”). “Everyone” (i.e., individuals in the center and on both sides of the left-right divide), will support policies more when they interpret them as a means of achieving a universal value (indicated by the use of a persuasive word like “knowledge”).

The third hypothesis about pre-election effects of persuasive words in election pledges thus highlights the moderating effects of voters’ ideologi-cal predispositions in the evaluative process when, by influencing individ-uals’ beliefs about policies, persuasive words indirectly influence their sup-port for the same:

H3) The indirect effects of persuasive words on policy support which

occurs through changes in beliefs about the policies, will be moderated by individuals’ ideological left-right predispositions.

Post-Election Effects of Persuasive Words in Election Pledges

Shifting attention to the post-election effects of persuasive words in elec-tion pledges, I suggest the following. While persuasive words in election pledges might be beneficial for parties pre-election, this can have negative post-election effects on voters’ perceptions of a government’s policy per-formance. Because persuasive words are linguistically flexible and can elicit beliefs about policies that stretch beyond the exact policy proposals, mismatches between voters’ substantive expectations on the policies, and the actual policy outcomes post-election, are more likely to occur. This, in turn, will increase the likelihood that voters perceive the pledges as broken.

19

The fourth hypothesis, thus, suggests that using persuasive words in election pledges can come at a long-term cost for parties, in terms of nega-tive evaluations of their policy performance. It highlights, specifically, how voters’ substantive beliefs about policies can play a part in this evaluative process:

H4) Because persuasive words in election pledges can elicit substantive

beliefs about policies that stretch beyond the exact policy proposals, they can, indirectly, lead to a perception that the pledges have been broken when the policy outcomes become known.

Figure 1. Predicted Pre- and Post-Election Effects of Persuasive Words in Election Pledges

Method and Data

Following a larger trend in political communication and public opinion re-search (see e.g., Druckman, Green, Kuklinski, & Lupia, 2006; McDermott, 2002, for overviews), I utilize survey experiments to test the hypotheses.

An important advantage with experimental designs is that participants can be randomly assigned to different treatment and control groups. Through random distribution, it is possible to ensure that no systematic pat-tern exists between the assignment of subjects into groups and any charac-teristics of those subjects, which could potentially intervene, or bias the ef-fects of a manipulation (see Morton & Williams, 2010). Randomized sur-vey experiments, in other words, allow researchers to measure the relation-ship between a specific intervention and an outcome of interest, without having to worry about factors that could confound the analyses, and thereby the possibilities of drawing reliable inferences about the specific relation-ship between two (or more) variables.

19

The fourth hypothesis, thus, suggests that using persuasive words in election pledges can come at a long-term cost for parties, in terms of nega-tive evaluations of their policy performance. It highlights, specifically, how voters substantive beliefs about policies can play a part in this evaluative process:

H4) Because persuasive words in election pledges can elicit substantive

beliefs about policies that stretch beyond the exact policy proposals, they can, indirectly, lead to a perception that the pledges have been broken when the policy outcomes become known.

Figure 1. Predicted Pre- and Post-Election Effects of Persuasive Words in Election Pledges

Method and Data

Following a larger trend in political communication and public opinion re-search (see e.g., Druckman, Green, Kuklinski, & Lupia, 2006; McDermott, 2002, for overviews), I utilize survey experiments to test the hypotheses.

An important advantage with experimental designs is that participants can be randomly assigned to different treatment and control groups. Through random distribution, it is possible to ensure that no systematic pat-tern exists between the assignment of subjects into groups and any charac-teristics of those subjects, which could potentially intervene, or bias the ef-fects of a manipulation (see Morton & Williams, 2010). Randomized sur-vey experiments, in other words, allow researchers to measure the relation-ship between a specific intervention and an outcome of interest, without having to worry about factors that could confound the analyses, and thereby the possibilities of drawing reliable inferences about the specific relation-ship between two (or more) variables.

20

Also, the experimental design allows control over the independent vari-ables and is thus convincing when it comes to determine directions of the causality in the relationships of interest (e.g., Druckman, Green, Kuklinski, & Lupia, 2006; McDermott, 2002; Morton & Williams, 2010). The easiest (and most straightforward) way to assess the effects of exposure to persua-sive words in election pledges, one could argue, would be to ask people in a conventional opinion survey. The extensive volume of political messages that are present during an election campaign, however, makes the accuracy of such self-reported measures of exposure highly questionable—some in-dividuals may simply not remember, or even be aware of what type of in-formation they have been exposed to, or when they were exposed to it (Slater, 2004, p. 169). With experimental designs, I could control respond-ents’ exposure to persuasive words, which helped me to overcome this is-sue.

Design

Using vignettes for treatment, participants in my experiments were pre-sented with fictional election pledges and policy proposals, and in some cases, news articles with post-election information about policy outcomes. The policy proposals varied in content, issue area, as well as salience in the current political debate (which in my case was Sweden), including: school policies (salient), immigration policies (salient), and surveillance cameras in public parks (non-salient). For treatment, I varied the formulations of the policy pledges using universal-, left-, and right-leaning persuasive words. The effects of the treatments were compared to control groups which were presented with the exact same policies, but with persuasive words omitted.

The treatments were designed taking into account the characteristics of persuasive words described in the linguistic theory. Hence, the words I used were emotively charged, value-laden, and linguistically flexible (e.g., Ste-venson, 1944; Walton, 2006). They were also either universal-, left-, or right-leaning, taking into account common interpretations of left-right ide-ology in contemporary Western democracies.8

To be able to draw internally valid inferences about the effects of expo-sure to persuasive words in election pledges, I kept the variation of the treatments (other than persuasive words) at a minimum (for discussion about internal validity in experiments, see Morton & Williams, 2010). I also

8 Note here that persuasive words can be charged with other types of values than those that are placed on a left-right scale—such as ecological, feminist, religious, moral, traditionalist and authoritarian values (see e.g., Giddens, 1994; Inglehart, 2000; Parekh, 1992). Values that are either “universal” or left-right leaning as defined in a western democratic context, are only an example used in this thesis to illustrate value-laden words. The procedures for operationalizing different types of persuasive words are presented in respective paper.

21

kept the treatments free from other variables that could confound the anal-yses of the manipulation effects, such as information about the pledge-giver (see e.g., Slothuus & de Vreese, 2010), and competing values (see e.g., Sniderman & Theirault, 2004). The limitations of, and motivations for omitting these variables in the treatments are discussed at length in the con-cluding sections of the introduction.

Operationalization of Key Variables

When it comes to measures of the dependent and mediating variables, some particular choices should be discussed. These choices were made with the intention to keep the measures of interest clear-cut and unambiguous, as well as to facilitate readily quantifiable analyses.

To operationalize Policy support, I have used a single measure building on the question: “how much do you like the proposed policies?” One could argue that complementary questions would have given more leverage to the analyses (e.g., Saris & Andrews, 2004), such as asking about the extent to which respondents found the policies to be sufficient solutions to a prob-lem, and to what extent they would be willing to support a party that pre-sented such policies. On the other hand, it is possible that such complemen-tary questions would have activated external mechanisms that could have intervened in respondents’ evaluation processes. Asking whether the poli-cies would sufficiently solve a problem could, for example, have activated individuals’ general trust in governments and their capacities to implement policies. Thus, not only does the “simple” question most readily measure individuals’ support for a policy, using this question also helps the analyses to steer clear of the causal effects of interest.

For Substantive beliefs about policies, I have used closed-ended ques-tions: 1) “How do you interpret what the pledge-givers aim at with the pro-posed policies?” 2) “Where would you place the policies on the left-right scale?” and 3) “What were your expectations on the policy outcomes?”

These questions were formulated taking into account the contexts of the specific studies (whether the analyses concerned pre-election effects on the evaluation of policy proposals, or post-election effects on the evaluations of policy outcomes). They were also formulated taking into account whether it was possible to conduct reliable quantitative analyses. The closed-ended format of the questions with fixed items (as opposed to an open-ended format), gave clear-cut measures that were easy to interpret, and it reduced the potential for mistakes due to subjective interpretations on the part of a coder.9 It also enabled me to employ readily quantifiable 9 Since analyses of open-ended questions rely on researcher- and/or theory-driven preconceptions, they can lead to biased classifications and interpretations that may not accurately represent respondents’ intentions (Jackson & Trochim, 2002; Kelle & Laurie, 1998; Weber, 1990). While it is possible that computer-assisted coding could decrease this problem, this

20

Also, the experimental design allows control over the independent vari-ables and is thus convincing when it comes to determine directions of the causality in the relationships of interest (e.g., Druckman, Green, Kuklinski, & Lupia, 2006; McDermott, 2002; Morton & Williams, 2010). The easiest (and most straightforward) way to assess the effects of exposure to persua-sive words in election pledges, one could argue, would be to ask people in a conventional opinion survey. The extensive volume of political messages that are present during an election campaign, however, makes the accuracy of such self-reported measures of exposure highly questionable—some in-dividuals may simply not remember, or even be aware of what type of in-formation they have been exposed to, or when they were exposed to it (Slater, 2004, p. 169). With experimental designs, I could control respond-ents’ exposure to persuasive words, which helped me to overcome this is-sue.

Design

Using vignettes for treatment, participants in my experiments were pre-sented with fictional election pledges and policy proposals, and in some cases, news articles with post-election information about policy outcomes. The policy proposals varied in content, issue area, as well as salience in the current political debate (which in my case was Sweden), including: school policies (salient), immigration policies (salient), and surveillance cameras in public parks (non-salient). For treatment, I varied the formulations of the policy pledges using universal-, left-, and right-leaning persuasive words. The effects of the treatments were compared to control groups which were presented with the exact same policies, but with persuasive words omitted.

The treatments were designed taking into account the characteristics of persuasive words described in the linguistic theory. Hence, the words I used were emotively charged, value-laden, and linguistically flexible (e.g., Ste-venson, 1944; Walton, 2006). They were also either universal-, left-, or right-leaning, taking into account common interpretations of left-right ide-ology in contemporary Western democracies.8

To be able to draw internally valid inferences about the effects of expo-sure to persuasive words in election pledges, I kept the variation of the treatments (other than persuasive words) at a minimum (for discussion about internal validity in experiments, see Morton & Williams, 2010). I also

8 Note here that persuasive words can be charged with other types of values than those that are placed on a left-right scale—such as ecological, feminist, religious, moral, traditionalist and authoritarian values (see e.g., Giddens, 1994; Inglehart, 2000; Parekh, 1992). Values that are either “universal” or left-right leaning as defined in a western democratic context, are only an example used in this thesis to illustrate value-laden words. The procedures for operationalizing different types of persuasive words are presented in respective paper.

21

kept the treatments free from other variables that could confound the anal-yses of the manipulation effects, such as information about the pledge-giver (see e.g., Slothuus & de Vreese, 2010), and competing values (see e.g., Sniderman & Theirault, 2004). The limitations of, and motivations for omitting these variables in the treatments are discussed at length in the con-cluding sections of the introduction.

Operationalization of Key Variables

When it comes to measures of the dependent and mediating variables, some particular choices should be discussed. These choices were made with the intention to keep the measures of interest clear-cut and unambiguous, as well as to facilitate readily quantifiable analyses.

To operationalize Policy support, I have used a single measure building on the question: “how much do you like the proposed policies?” One could argue that complementary questions would have given more leverage to the analyses (e.g., Saris & Andrews, 2004), such as asking about the extent to which respondents found the policies to be sufficient solutions to a prob-lem, and to what extent they would be willing to support a party that pre-sented such policies. On the other hand, it is possible that such complemen-tary questions would have activated external mechanisms that could have intervened in respondents’ evaluation processes. Asking whether the poli-cies would sufficiently solve a problem could, for example, have activated individuals’ general trust in governments and their capacities to implement policies. Thus, not only does the “simple” question most readily measure individuals’ support for a policy, using this question also helps the analyses to steer clear of the causal effects of interest.

For Substantive beliefs about policies, I have used closed-ended ques-tions: 1) “How do you interpret what the pledge-givers aim at with the pro-posed policies?” 2) “Where would you place the policies on the left-right scale?” and 3) “What were your expectations on the policy outcomes?”

These questions were formulated taking into account the contexts of the specific studies (whether the analyses concerned pre-election effects on the evaluation of policy proposals, or post-election effects on the evaluations of policy outcomes). They were also formulated taking into account whether it was possible to conduct reliable quantitative analyses. The closed-ended format of the questions with fixed items (as opposed to an open-ended format), gave clear-cut measures that were easy to interpret, and it reduced the potential for mistakes due to subjective interpretations on the part of a coder.9 It also enabled me to employ readily quantifiable 9 Since analyses of open-ended questions rely on researcher- and/or theory-driven preconceptions, they can lead to biased classifications and interpretations that may not accurately represent respondents’ intentions (Jackson & Trochim, 2002; Kelle & Laurie, 1998; Weber, 1990). While it is possible that computer-assisted coding could decrease this problem, this

22

analyses, and to draw internally valid inferences (see e.g., Brewer & Gross 2005, pp. 931–933), as the answers were already standardized and catego-rized.10

However, the closed-ended questions that were used did not conclu-sively measure the extent to which persuasive words in the election pledges produced different beliefs about the policies for different individuals, due to their linguistic flexibility. To better assess this, open-ended questions, in which the respondents express their beliefs in their own words, can be used, as this would allow more individual variation in the answers (e.g., de Vreese, 2004, pp. 42–43; Shapiro, 1994, pp. 1–14). Choosing to use closed-ended measures in the empirical studies hence represents a limitation of the extent to which conclusions can be drawn about the effects of persuasive words on individuals’ (different) beliefs about policies.

A final limitation that should be mentioned concerns the operationaliza-tion of the moderating variable in the model: the extent to which individuals endorse a value that is associated with a persuasive word. This variable was measured using a standard question of ideological left-right placement: “where would you place yourself on the ideological left-right scale?” One could argue that this variable represents a measure that is too simplistic for the “complex canvas” of value predispositions that individuals hold (e.g., Converse, 1964; Feldman & Zaller, 1992). However, with regard to the context in which the studies were employed—election campaigns in a lib-eral Western democracy, where the left-right dimension represents the main dimension for political controversy (e.g., Budge, Klingemann, Volkens, Bara, & Tannenbaum, 2001; Freire, 2006, 2008; Gunther & Montero, 2001; van der Eijk, Franklin, & van der Brug, 1999)—individuals’ left-right placement should serve as a decent proxy for their value preferences.

Data and Sample

In total, I conducted three survey experiments, with different samples of Swedish citizens. The surveys were carried out on different occasions be-tween October 2014 and January 2016, with respondents from the Swedish method still suffers from validity issues (Kelle & Laurie, 1998) as computers are less capable of interpreting meaning in symbols (Shapiro, 1997) and capturing the meaning of words in different contexts (Jackson & Trochim, 2002). 10 Open-ended questions typically produce a large variety of responses, and can generate a frequent (or infrequent) mentioning of certain topics with varying importance to the respondent (Geer, 1991; Rea & Parker, 2014; Sproull, 1995). This makes standardization and categorizations of such answers difficult (Jackson & Trochim, 2002). Moreover, respondents who are particularly interested in the topic (in my case, individuals who are particularly interested in specific policy cases), are more likely to answer open-ended questions—which typically require more effort from respondents—than those who are less interested (Geer, 1991), which can lead to an over-representation of the interpretations of the interested, thus biasing the results (Jackson & Trochim, 2002).

23

Citizen Panel, an online survey panel administered by the Laboratory of Opinion Research (LORE) at the University of Gothenburg.

LORE is a university-based infrastructure, which provides data collec-tion and data processing through web surveys and web panels, predomi-nantly for academic scholars (from Sweden and elsewhere). The Citizen Panel is LORE’s largest and longest running panel project. At the time of writing, the panel consists of approximately 55 000 active participants, in-cluding both opt-in recruitments (approximately 48 000 individuals), and probability-based recruitments (approximately 7000 individuals).11 (For more information about LORE and the Citizen Panel, see www.lore.gu.se).

Each of my surveys were allotted randomly selected subsamples of the participants in the Citizen Panel. As is true for many contemporary survey experimental studies (see Callegaro et al., 2014), my samples consisted of self-recruited members. In contrast to findings from studies using probabil-ity-recruited samples, therefore, my findings cannot be used to draw gen-eral conclusions about the effects to a larger population (see e.g., Blair, Czaja, & Blair, 2013; Dillman et al., 2009; de Leeuw, 2005).

In terms of representativeness, the samples in my studies were relatively diverse compared with other types of convenience samples, such as student samples (for an overview of different sample types, see Blair, Czaja, & Blair, 2013), but they did have some problematic biases. In particular, my samples had an overrepresentation of politically interested, and highly ed-ucated individuals (see respective paper for details); variables which may condition the relationships studied in my papers.

Both political interest and level of education are often used as indicators of political awareness (for a review, see Highton, 2009), which refers to a person’s degree of political knowledge, interest, and ability to process com-plex political information. These qualities are believed to increase individ-uals’ ability to connect political positions with abstract values, and to form stable and thoughtful political preferences (see e.g., Luskin, 1990, 2002; Zaller, 1990, 1992; Converse, 1964; Carpini & Keeter 1996; Feldman, 1989)—something which, according to these scholars, should make indi-viduals more consistent in their political views, and thus also more resistant to political manipulation and external influence (also see e.g., Iyengar, 1991; Kinder & Sanders, 1990). If these presumptions hold true, then my samples should comprise “hard cases” for testing the hypotheses in my pa-pers, and it should be possible that the effects seen might be greater in a real-election campaign where voters’ level of education and political inter-est are more diverse. However, while there are some empirical findings which support a positive relationship between political awareness and re-sistance to political manipulation (e.g., Kam, 2005; Krosnick & Kinder, 1990), the research is not conclusive (for overviews, see Slothuus, 2008; de 11 Participation in the Citizen Panel is voluntary, and respondents do not receive payment for their participation.

22

analyses, and to draw internally valid inferences (see e.g., Brewer & Gross 2005, pp. 931–933), as the answers were already standardized and catego-rized.10

However, the closed-ended questions that were used did not conclu-sively measure the extent to which persuasive words in the election pledges produced different beliefs about the policies for different individuals, due to their linguistic flexibility. To better assess this, open-ended questions, in which the respondents express their beliefs in their own words, can be used, as this would allow more individual variation in the answers (e.g., de Vreese, 2004, pp. 42–43; Shapiro, 1994, pp. 1–14). Choosing to use closed-ended measures in the empirical studies hence represents a limitation of the extent to which conclusions can be drawn about the effects of persuasive words on individuals’ (different) beliefs about policies.

A final limitation that should be mentioned concerns the operationaliza-tion of the moderating variable in the model: the extent to which individuals endorse a value that is associated with a persuasive word. This variable was measured using a standard question of ideological left-right placement: “where would you place yourself on the ideological left-right scale?” One could argue that this variable represents a measure that is too simplistic for the “complex canvas” of value predispositions that individuals hold (e.g., Converse, 1964; Feldman & Zaller, 1992). However, with regard to the context in which the studies were employed—election campaigns in a lib-eral Western democracy, where the left-right dimension represents the main dimension for political controversy (e.g., Budge, Klingemann, Volkens, Bara, & Tannenbaum, 2001; Freire, 2006, 2008; Gunther & Montero, 2001; van der Eijk, Franklin, & van der Brug, 1999)—individuals’ left-right placement should serve as a decent proxy for their value preferences.

Data and Sample

In total, I conducted three survey experiments, with different samples of Swedish citizens. The surveys were carried out on different occasions be-tween October 2014 and January 2016, with respondents from the Swedish method still suffers from validity issues (Kelle & Laurie, 1998) as computers are less capable of interpreting meaning in symbols (Shapiro, 1997) and capturing the meaning of words in different contexts (Jackson & Trochim, 2002). 10 Open-ended questions typically produce a large variety of responses, and can generate a frequent (or infrequent) mentioning of certain topics with varying importance to the respondent (Geer, 1991; Rea & Parker, 2014; Sproull, 1995). This makes standardization and categorizations of such answers difficult (Jackson & Trochim, 2002). Moreover, respondents who are particularly interested in the topic (in my case, individuals who are particularly interested in specific policy cases), are more likely to answer open-ended questions—which typically require more effort from respondents—than those who are less interested (Geer, 1991), which can lead to an over-representation of the interpretations of the interested, thus biasing the results (Jackson & Trochim, 2002).

23

Citizen Panel, an online survey panel administered by the Laboratory of Opinion Research (LORE) at the University of Gothenburg.

LORE is a university-based infrastructure, which provides data collec-tion and data processing through web surveys and web panels, predomi-nantly for academic scholars (from Sweden and elsewhere). The Citizen Panel is LORE’s largest and longest running panel project. At the time of writing, the panel consists of approximately 55 000 active participants, in-cluding both opt-in recruitments (approximately 48 000 individuals), and probability-based recruitments (approximately 7000 individuals).11 (For more information about LORE and the Citizen Panel, see www.lore.gu.se).

Each of my surveys were allotted randomly selected subsamples of the participants in the Citizen Panel. As is true for many contemporary survey experimental studies (see Callegaro et al., 2014), my samples consisted of self-recruited members. In contrast to findings from studies using probabil-ity-recruited samples, therefore, my findings cannot be used to draw gen-eral conclusions about the effects to a larger population (see e.g., Blair, Czaja, & Blair, 2013; Dillman et al., 2009; de Leeuw, 2005).

In terms of representativeness, the samples in my studies were relatively diverse compared with other types of convenience samples, such as student samples (for an overview of different sample types, see Blair, Czaja, & Blair, 2013), but they did have some problematic biases. In particular, my samples had an overrepresentation of politically interested, and highly ed-ucated individuals (see respective paper for details); variables which may condition the relationships studied in my papers.

Both political interest and level of education are often used as indicators of political awareness (for a review, see Highton, 2009), which refers to a person’s degree of political knowledge, interest, and ability to process com-plex political information. These qualities are believed to increase individ-uals’ ability to connect political positions with abstract values, and to form stable and thoughtful political preferences (see e.g., Luskin, 1990, 2002; Zaller, 1990, 1992; Converse, 1964; Carpini & Keeter 1996; Feldman, 1989)—something which, according to these scholars, should make indi-viduals more consistent in their political views, and thus also more resistant to political manipulation and external influence (also see e.g., Iyengar, 1991; Kinder & Sanders, 1990). If these presumptions hold true, then my samples should comprise “hard cases” for testing the hypotheses in my pa-pers, and it should be possible that the effects seen might be greater in a real-election campaign where voters’ level of education and political inter-est are more diverse. However, while there are some empirical findings which support a positive relationship between political awareness and re-sistance to political manipulation (e.g., Kam, 2005; Krosnick & Kinder, 1990), the research is not conclusive (for overviews, see Slothuus, 2008; de 11 Participation in the Citizen Panel is voluntary, and respondents do not receive payment for their participation.

24

Vreese & Boomgaarden, 2006). Some have found that people in general—regardless of their degree of political awareness—appear to make relatively little use of their own abstract principles when they express political views, and operate more from external influence (Goren, 2004); others show that citizens’ reactions to political information vary more due to the situation, timing and context, as opposed to individual characteristics such as political awareness (Redlawsk, 2002).

To assess whether the effects of my experiments differ in real-election campaigns where the level of political awareness is more varied, would thus require more varied samples—which limits the possibilities of drawing general conclusions about the effects of persuasive words from my studies.

Another limitation of my data is that all studies have been conducted in a Swedish political context, with Swedish citizens. This fact raises ques-tions about the generalizability of the findings, to other contexts and sam-ples. However, since both theory on the impact of values on attitude for-mation processes (Feldman & Steenbergen, 2001; Lakoff, 2002), and the theory on persuasive words (e.g., Stevenson, 1944), are outlined as univer-sal theories applicable over individual and cultural boundaries, I expect the results from my studies to reflect patterns that go beyond the case of Swe-dish citizens. The ways that the findings from my experiments can be inter-preted in broader perspectives are discussed in respective paper.

Analytic Strategies

For analyses of the direct effects of persuasive words on individuals’ beliefs about, and support for policies, I used ANOVA, which is the conventional method for analyzing experimental data. Simply put, the ANOVA com-pares means between treatment groups, and helps assess whether these means differ significantly from control groups (see e.g., Brace, Snelgar, & Kemp, 2012).

For analyses of mediated and moderated relationships between persua-sive words, individuals’ beliefs about policies, and their subsequent policy support, I utilized the SPSS macro “PROCESS,” developed by Andrew Hayes (2013). PROCESS employs bootstrapping (or resampling), to assess the statistical significance of indirect effects in mediated relationships. Tra-ditional methods for assessing indirect effects (e.g., the Sobel test) rely on an assumption of a normal sampling distribution of the indirect effect (something which is not likely to be the case, unless the sample size is very large). By constructing confidence intervals for the indirect effect through bootstrapping (that is, by randomly drawing subsamples from the total sam-ple, and running the analyses of the effect once in each subsample), PRO-CESS sidesteps this assumption. More precisely, it respects the possible irregularity of the sampling distribution by constructing confidence inter-vals based on an empirically derived representation of this distribution.

25

This, in turn, increases the likelihood that statistical inferences being drawn about mediation will be accurate (see Hayes 2013, pp. 102–106).

Papers in Summary

The dissertation comprises four papers. The first paper reviews the framing theory, and suggests a way to integrate this theory into the linguistic theory on persuasive words. More precisely, it uses insights from the linguistic theory to suggest a complementary explanation of how value frames, in particular, can influence citizens’ support for policies.

The subsequent papers draw on the theory of persuasive words to pro-pose, and empirically examine, how the use of value-laden words in elec-tion pledges can impact the representative chain when voters’ policy pref-erences are linked to parties’ policy intentions via election pledges. Two papers present and test hypotheses about the pre-election effects of persua-sive words in election pledges. A final paper presents and tests hypotheses about the post-election effects of such pre-election rhetoric.

Paper I. Explaining Framing Effects: Insights from Linguistic

Theory on Persuasive Words

On the message level, value framing is an umbrella concept used to describe communication that emphasizes certain value dimensions of an object, making these dimensions more salient. The strategy of using persuasive words is, instead, defined as the presentation of an object with the use of words that have specific semantic properties, which makes the object ap-pear more appealing than it is on the basis of its actual content alone. In Paper 1, I show how these two strategies overlap, and argue that it is likely for framing and persuasive words to occur simultaneously in the same mes-sage. Value frames, which emphasize value dimensions of an object, will likely include persuasive words to signify the value dimension which the frame aims to accentuate. Likewise, persuasive words that imply certain values will likely occur as parts of larger frames that emphasize particular value dimensions of an object.

On the individual level (the predicted effects on recipients), however, the two theories diverge. Framing effects are thought to occur when the selective emphasis of certain beliefs about an object (which are already pre-sent in the public’s mind) increases the weight that individuals assign to these beliefs when forming their attitudes (changes in weight components); persuasion occurs when a message changes those existing beliefs about ob-jects (changes in belief content) (see e.g., Chong & Druckman, 2007a, p. 115; Nelson & Willey, 2001, p. 256). Despite this theoretical separation of

24

Vreese & Boomgaarden, 2006). Some have found that people in general—regardless of their degree of political awareness—appear to make relatively little use of their own abstract principles when they express political views, and operate more from external influence (Goren, 2004); others show that citizens’ reactions to political information vary more due to the situation, timing and context, as opposed to individual characteristics such as political awareness (Redlawsk, 2002).

To assess whether the effects of my experiments differ in real-election campaigns where the level of political awareness is more varied, would thus require more varied samples—which limits the possibilities of drawing general conclusions about the effects of persuasive words from my studies.

Another limitation of my data is that all studies have been conducted in a Swedish political context, with Swedish citizens. This fact raises ques-tions about the generalizability of the findings, to other contexts and sam-ples. However, since both theory on the impact of values on attitude for-mation processes (Feldman & Steenbergen, 2001; Lakoff, 2002), and the theory on persuasive words (e.g., Stevenson, 1944), are outlined as univer-sal theories applicable over individual and cultural boundaries, I expect the results from my studies to reflect patterns that go beyond the case of Swe-dish citizens. The ways that the findings from my experiments can be inter-preted in broader perspectives are discussed in respective paper.

Analytic Strategies

For analyses of the direct effects of persuasive words on individuals’ beliefs about, and support for policies, I used ANOVA, which is the conventional method for analyzing experimental data. Simply put, the ANOVA com-pares means between treatment groups, and helps assess whether these means differ significantly from control groups (see e.g., Brace, Snelgar, & Kemp, 2012).

For analyses of mediated and moderated relationships between persua-sive words, individuals’ beliefs about policies, and their subsequent policy support, I utilized the SPSS macro “PROCESS,” developed by Andrew Hayes (2013). PROCESS employs bootstrapping (or resampling), to assess the statistical significance of indirect effects in mediated relationships. Tra-ditional methods for assessing indirect effects (e.g., the Sobel test) rely on an assumption of a normal sampling distribution of the indirect effect (something which is not likely to be the case, unless the sample size is very large). By constructing confidence intervals for the indirect effect through bootstrapping (that is, by randomly drawing subsamples from the total sam-ple, and running the analyses of the effect once in each subsample), PRO-CESS sidesteps this assumption. More precisely, it respects the possible irregularity of the sampling distribution by constructing confidence inter-vals based on an empirically derived representation of this distribution.

25

This, in turn, increases the likelihood that statistical inferences being drawn about mediation will be accurate (see Hayes 2013, pp. 102–106).

Papers in Summary

The dissertation comprises four papers. The first paper reviews the framing theory, and suggests a way to integrate this theory into the linguistic theory on persuasive words. More precisely, it uses insights from the linguistic theory to suggest a complementary explanation of how value frames, in particular, can influence citizens’ support for policies.

The subsequent papers draw on the theory of persuasive words to pro-pose, and empirically examine, how the use of value-laden words in elec-tion pledges can impact the representative chain when voters’ policy pref-erences are linked to parties’ policy intentions via election pledges. Two papers present and test hypotheses about the pre-election effects of persua-sive words in election pledges. A final paper presents and tests hypotheses about the post-election effects of such pre-election rhetoric.

Paper I. Explaining Framing Effects: Insights from Linguistic

Theory on Persuasive Words

On the message level, value framing is an umbrella concept used to describe communication that emphasizes certain value dimensions of an object, making these dimensions more salient. The strategy of using persuasive words is, instead, defined as the presentation of an object with the use of words that have specific semantic properties, which makes the object ap-pear more appealing than it is on the basis of its actual content alone. In Paper 1, I show how these two strategies overlap, and argue that it is likely for framing and persuasive words to occur simultaneously in the same mes-sage. Value frames, which emphasize value dimensions of an object, will likely include persuasive words to signify the value dimension which the frame aims to accentuate. Likewise, persuasive words that imply certain values will likely occur as parts of larger frames that emphasize particular value dimensions of an object.

On the individual level (the predicted effects on recipients), however, the two theories diverge. Framing effects are thought to occur when the selective emphasis of certain beliefs about an object (which are already pre-sent in the public’s mind) increases the weight that individuals assign to these beliefs when forming their attitudes (changes in weight components); persuasion occurs when a message changes those existing beliefs about ob-jects (changes in belief content) (see e.g., Chong & Druckman, 2007a, p. 115; Nelson & Willey, 2001, p. 256). Despite this theoretical separation of

26

framing effects and persuasion, there are empirical findings that suggest otherwise (e.g., Nelson & Oxley, 1999; Slothuus, 2008).

In Paper 1, I use the linguistic theory on persuasive words to offer a plausible explanation for these seemingly contradictory empirical findings within the framing research. My point is, in short, that if frames include persuasive words, which have characteristics that make them appear factu-ally informative (Stevenson, 1944, pp. 87–88; Walton 2006, p. 218), then individuals can mistake them for substantive information about policies, which they can use as a way of understanding the policy’s content and con-sequences. In this way, frames could impact individuals’ attitudes by changing their beliefs about a policy, solely by including certain words (not by providing new information about the policies). In other words, persua-sion can occur via framing.

These insights shed new light on framing effects. They suggest that if frames include persuasive words, then they can impede citizens’ under-standing of political messages, such as election pledges; something which has implications for the functioning of real-world representative democra-cies.

Paper II. Changing Policy with Words: How Persuasive

Words in Election Pledges Influence Voters’ Beliefs about

Policies

The second paper empirically tests the theoretical claims of Paper 1. More precisely, it investigates whether the use of persuasive words in election pledges can elicit substantive beliefs on policies which stretch beyond the actual policy proposals, and how this can indirectly influence individuals’ support for the policies. It also tests the moderating effects of individuals’ ideological predispositions in this process, when the persuasive words in-dicate values that are associated with either left or right ideology.

The hypotheses were tested in a survey experiment where Swedish citi-zens interpreted and evaluated a policy pledge about schools, presented with either left-leaning or right-leaning persuasive words. In support of the hypotheses, the study showed that persuasive words in election pledges in-fluenced respondents’ beliefs about the policies. When respondents were exposed to left-leaning (or right-leaning) persuasive words, they were more likely to adhere to beliefs about the policies that spanned beyond the exact policy proposals, than were respondents in a control group who were ex-posed to the same policies, but without the persuasive words. These beliefs, in turn, indirectly influenced the individuals’ support for the pledged poli-cies. Moreover, the study showed that the indirect effects of persuasive

27

words on individuals’ policy support were conditioned on whether the re-spondents placed the policies on the same side of the left-right divide as they placed themselves.

Paper 2 thus adds to the ongoing discussion in the framing literature, which questions the traditional explanations of framing effects as simple changes in weight components—and suggests that frames can also intro-duce new content to individuals’ existing beliefs (Leeper & Slothuus, 2015). It complements the discussion by showing that this can occur by simply including persuasive words, without providing additional, factual information about the policies that supports these beliefs.

The findings also have normative implications for democratic politics. Since the study reveals that persuasive words in election pledges can elicit substantive beliefs about policies which go beyond the actual policies out-lined in pledges, it shows that the inclusion of such words can hamper vot-ers’ capability to accurately predict the future policy actions of parties in elections—something that is crucial to mandate models of democratic rep-resentation (e.g., Dahls, 1971; Downs, 1957; Manin, Przeworski, & Stokes, 1999; Mansbridge, 2003).

Paper III. Election Pledge Rhetoric: Selling Policies with

Words (Co-Authored with Elin Naurin)

The third paper, which is co-authored with Elin Naurin, tests the pre-elec-tion effects of using universal persuasive words in election pledges. More specifically, it investigates whether parties can “sell” specific policies to a broad electorate—to voters of center, left, and right predispositions—by using universal persuasive words in election pledges.

As electorates are often spread over a left-right continuum that peaks around the center (e.g., Ezrow, 2005), there should be an incentive for mainstream parties to make pledges that appeal not only to voters of out-spoken left or right alignments, but also to broader groups of voters who identify less with such ideologically constrained values. Despite this, there has been little research on rhetoric that is, or can be, used for this purpose. Whereas it has been repeatedly found that parties using value-based rheto-ric can successfully reach out to voters with policies, there are also repeated findings that the effects are limited by ideological predispositions; different values appeal to voters on either one side or the other of the ideological left–right divide (see e.g., Nelson & Garst, 2005; Schemer, Wirth, & Mat-thes, 2012).

Based on the linguistic theory on persuasive words, Paper 3 contributes by theoretically identifying, and empirically testing, a value-based rhetoric that is “universally appealing” and not colored by ideological (left or right)

26

framing effects and persuasion, there are empirical findings that suggest otherwise (e.g., Nelson & Oxley, 1999; Slothuus, 2008).

In Paper 1, I use the linguistic theory on persuasive words to offer a plausible explanation for these seemingly contradictory empirical findings within the framing research. My point is, in short, that if frames include persuasive words, which have characteristics that make them appear factu-ally informative (Stevenson, 1944, pp. 87–88; Walton 2006, p. 218), then individuals can mistake them for substantive information about policies, which they can use as a way of understanding the policy’s content and con-sequences. In this way, frames could impact individuals’ attitudes by changing their beliefs about a policy, solely by including certain words (not by providing new information about the policies). In other words, persua-sion can occur via framing.

These insights shed new light on framing effects. They suggest that if frames include persuasive words, then they can impede citizens’ under-standing of political messages, such as election pledges; something which has implications for the functioning of real-world representative democra-cies.

Paper II. Changing Policy with Words: How Persuasive

Words in Election Pledges Influence Voters’ Beliefs about

Policies

The second paper empirically tests the theoretical claims of Paper 1. More precisely, it investigates whether the use of persuasive words in election pledges can elicit substantive beliefs on policies which stretch beyond the actual policy proposals, and how this can indirectly influence individuals’ support for the policies. It also tests the moderating effects of individuals’ ideological predispositions in this process, when the persuasive words in-dicate values that are associated with either left or right ideology.

The hypotheses were tested in a survey experiment where Swedish citi-zens interpreted and evaluated a policy pledge about schools, presented with either left-leaning or right-leaning persuasive words. In support of the hypotheses, the study showed that persuasive words in election pledges in-fluenced respondents’ beliefs about the policies. When respondents were exposed to left-leaning (or right-leaning) persuasive words, they were more likely to adhere to beliefs about the policies that spanned beyond the exact policy proposals, than were respondents in a control group who were ex-posed to the same policies, but without the persuasive words. These beliefs, in turn, indirectly influenced the individuals’ support for the pledged poli-cies. Moreover, the study showed that the indirect effects of persuasive

27

words on individuals’ policy support were conditioned on whether the re-spondents placed the policies on the same side of the left-right divide as they placed themselves.

Paper 2 thus adds to the ongoing discussion in the framing literature, which questions the traditional explanations of framing effects as simple changes in weight components—and suggests that frames can also intro-duce new content to individuals’ existing beliefs (Leeper & Slothuus, 2015). It complements the discussion by showing that this can occur by simply including persuasive words, without providing additional, factual information about the policies that supports these beliefs.

The findings also have normative implications for democratic politics. Since the study reveals that persuasive words in election pledges can elicit substantive beliefs about policies which go beyond the actual policies out-lined in pledges, it shows that the inclusion of such words can hamper vot-ers’ capability to accurately predict the future policy actions of parties in elections—something that is crucial to mandate models of democratic rep-resentation (e.g., Dahls, 1971; Downs, 1957; Manin, Przeworski, & Stokes, 1999; Mansbridge, 2003).

Paper III. Election Pledge Rhetoric: Selling Policies with

Words (Co-Authored with Elin Naurin)

The third paper, which is co-authored with Elin Naurin, tests the pre-elec-tion effects of using universal persuasive words in election pledges. More specifically, it investigates whether parties can “sell” specific policies to a broad electorate—to voters of center, left, and right predispositions—by using universal persuasive words in election pledges.

As electorates are often spread over a left-right continuum that peaks around the center (e.g., Ezrow, 2005), there should be an incentive for mainstream parties to make pledges that appeal not only to voters of out-spoken left or right alignments, but also to broader groups of voters who identify less with such ideologically constrained values. Despite this, there has been little research on rhetoric that is, or can be, used for this purpose. Whereas it has been repeatedly found that parties using value-based rheto-ric can successfully reach out to voters with policies, there are also repeated findings that the effects are limited by ideological predispositions; different values appeal to voters on either one side or the other of the ideological left–right divide (see e.g., Nelson & Garst, 2005; Schemer, Wirth, & Mat-thes, 2012).

Based on the linguistic theory on persuasive words, Paper 3 contributes by theoretically identifying, and empirically testing, a value-based rhetoric that is “universally appealing” and not colored by ideological (left or right)

28

overtones. More specifically, it applies the concept of universal persuasive words (persuasive words that appeals to “mass audiences,” e.g., Bench-Ca-pon, 2003; Walton, 2003), to an election pledge context, and tests the ef-fects experimentally.

The study revealed that individuals exposed to universal persuasive words in election pledges were more positive toward the pledges than those who saw the same pledges but without persuasive words. The effects were most notably seen among individuals near the center, and less so for indi-viduals leaning to the left or right. The findings thus indicate that universal persuasive words can help parties to sell policies to median voters (individ-uals oriented in or near the center of the left–right scale, see e.g., Ezrow, 2005) without binding themselves to specific actions; but that they would be less effective in attracting individuals of pronounced left and right pre-dispositions.

Given that, in most contexts, voters are moving toward a political center and are becoming more volatile between elections (e.g., Dassonneville, 2012; Mair, 2002; Oscarsson, 2013), mainstream parties have an incentive to find ways to attract broad electorates rather than just the ideological left and right core voters. This will likely be done not only via different policies but also by how the different policies are presented. Parties’ “center” rhet-oric is thus an interesting tool to study, and Paper 3 takes a first step in that direction.

Paper IV. Making a Rod for Their Own Back? Post-Election

Effects of Pre-Election Rhetoric

The fourth paper investigates how the use of persuasive words in election pledges impacts on voters’ retrospective evaluations of the government’s policy performance. In other words, this paper shifts the focus from the short-term pre-election effects, to the long-term post-election effects of per-suasive words in election pledges.

The point of departure is that we know a lot about how election pledge rhetoric influences citizens’ support for policies pre-election—when poli-cies are presented as proposals for future action—but we know much less about the effects of such pre-election rhetoric when the policy outcomes become known.

In mandate models of democratic representation, election pledges serve as a tool to evaluate how well the policy performance of governments matches their pre-election statements (e.g., Mansbridge, 2003). Therefore, rational parties whose aim is to vote-maximize pre-election as well as opti-mize their chances of being re-elected (Downs, 1957), should formulate

29

their election pledges so that they—if they win the election—do not dissat-isfy voters and hamper their chances of being re-elected (e.g., Aragonès & Neeman, 2000, pp. 183–184; Downs, 1957; Manin, Przeworski, & Stokes, 1999, p. 38). Drawing on the theory of persuasive words, Paper 4 argues that while persuasive words in election pledges might be beneficial to pledge-givers in the short term (pre-election) by raising voters’ expecta-tions on policies, this could cost parties in the long-run (post-election) when voters become exposed to policy outcomes that do not match these expec-tations.

Such post-election effects of election pledge rhetoric have not received much scholarly attention. In fact, how the ways that parties actively shape voters’ expectations pre-election impact voters’ evaluations of the govern-ment’s performance post-election, is generally understudied (Malhotra & Margalit, 2014, p. 1000). This is surprising, since there are normative im-plications to unjustly raising voters’ policy expectations with campaign rhetoric. It is also surprising in light of the vast interest shown for the short-term effects of political rhetoric.

In this paper, I specifically hypothesize that, because persuasive words can provoke beliefs about policies that reach beyond the exact policy pro-posals, persuasive words in election pledges will 1) redirect attention from the exact policy proposals to other, substantive expectations, which go be-yond the exact policies outlined in the pledge, and that 2) this will increase the likelihood that the pledges will be perceived as broken post-election (even when the pledges are actually fulfilled).

An experimental study revealed that, in certain cases (when the words were associated with left-leaning, and universal values), persuasive words influenced respondents’ substantive expectations on the policies. The study also showed that the effects the persuasive words had on substantive expec-tations of the policies indirectly led to a small, albeit significant, increased likelihood that the respondents perceived the pledges as broken when they were exposed to the policy outcomes. However, when the experiment was run on persuasive words associated with right-leaning values, which turned out to have little impact on voters’ expectations on the policies, the use of persuasive words did not cost the parties at all in terms of increasing the perception that the pledges had been broken.

Although the effects seen in the experiments were modest, the fourth paper provides one of few empirical confirmations of post-election effects of election pledge rhetoric. The implications are that, when election pledge rhetoric effectively impacts voters’ expectations on policies pre-election, this may cost the party post-election in terms of a slight increase in percep-tions that the pledges have been broken. When, on the other hand, election pledge rhetoric has little impact on voters’ expectations of the policies (in the study, when the words were “right-leaning”), there is no indication that

28

overtones. More specifically, it applies the concept of universal persuasive words (persuasive words that appeals to “mass audiences,” e.g., Bench-Ca-pon, 2003; Walton, 2003), to an election pledge context, and tests the ef-fects experimentally.

The study revealed that individuals exposed to universal persuasive words in election pledges were more positive toward the pledges than those who saw the same pledges but without persuasive words. The effects were most notably seen among individuals near the center, and less so for indi-viduals leaning to the left or right. The findings thus indicate that universal persuasive words can help parties to sell policies to median voters (individ-uals oriented in or near the center of the left–right scale, see e.g., Ezrow, 2005) without binding themselves to specific actions; but that they would be less effective in attracting individuals of pronounced left and right pre-dispositions.

Given that, in most contexts, voters are moving toward a political center and are becoming more volatile between elections (e.g., Dassonneville, 2012; Mair, 2002; Oscarsson, 2013), mainstream parties have an incentive to find ways to attract broad electorates rather than just the ideological left and right core voters. This will likely be done not only via different policies but also by how the different policies are presented. Parties’ “center” rhet-oric is thus an interesting tool to study, and Paper 3 takes a first step in that direction.

Paper IV. Making a Rod for Their Own Back? Post-Election

Effects of Pre-Election Rhetoric

The fourth paper investigates how the use of persuasive words in election pledges impacts on voters’ retrospective evaluations of the government’s policy performance. In other words, this paper shifts the focus from the short-term pre-election effects, to the long-term post-election effects of per-suasive words in election pledges.

The point of departure is that we know a lot about how election pledge rhetoric influences citizens’ support for policies pre-election—when poli-cies are presented as proposals for future action—but we know much less about the effects of such pre-election rhetoric when the policy outcomes become known.

In mandate models of democratic representation, election pledges serve as a tool to evaluate how well the policy performance of governments matches their pre-election statements (e.g., Mansbridge, 2003). Therefore, rational parties whose aim is to vote-maximize pre-election as well as opti-mize their chances of being re-elected (Downs, 1957), should formulate

29

their election pledges so that they—if they win the election—do not dissat-isfy voters and hamper their chances of being re-elected (e.g., Aragonès & Neeman, 2000, pp. 183–184; Downs, 1957; Manin, Przeworski, & Stokes, 1999, p. 38). Drawing on the theory of persuasive words, Paper 4 argues that while persuasive words in election pledges might be beneficial to pledge-givers in the short term (pre-election) by raising voters’ expecta-tions on policies, this could cost parties in the long-run (post-election) when voters become exposed to policy outcomes that do not match these expec-tations.

Such post-election effects of election pledge rhetoric have not received much scholarly attention. In fact, how the ways that parties actively shape voters’ expectations pre-election impact voters’ evaluations of the govern-ment’s performance post-election, is generally understudied (Malhotra & Margalit, 2014, p. 1000). This is surprising, since there are normative im-plications to unjustly raising voters’ policy expectations with campaign rhetoric. It is also surprising in light of the vast interest shown for the short-term effects of political rhetoric.

In this paper, I specifically hypothesize that, because persuasive words can provoke beliefs about policies that reach beyond the exact policy pro-posals, persuasive words in election pledges will 1) redirect attention from the exact policy proposals to other, substantive expectations, which go be-yond the exact policies outlined in the pledge, and that 2) this will increase the likelihood that the pledges will be perceived as broken post-election (even when the pledges are actually fulfilled).

An experimental study revealed that, in certain cases (when the words were associated with left-leaning, and universal values), persuasive words influenced respondents’ substantive expectations on the policies. The study also showed that the effects the persuasive words had on substantive expec-tations of the policies indirectly led to a small, albeit significant, increased likelihood that the respondents perceived the pledges as broken when they were exposed to the policy outcomes. However, when the experiment was run on persuasive words associated with right-leaning values, which turned out to have little impact on voters’ expectations on the policies, the use of persuasive words did not cost the parties at all in terms of increasing the perception that the pledges had been broken.

Although the effects seen in the experiments were modest, the fourth paper provides one of few empirical confirmations of post-election effects of election pledge rhetoric. The implications are that, when election pledge rhetoric effectively impacts voters’ expectations on policies pre-election, this may cost the party post-election in terms of a slight increase in percep-tions that the pledges have been broken. When, on the other hand, election pledge rhetoric has little impact on voters’ expectations of the policies (in the study, when the words were “right-leaning”), there is no indication that

30

persuasive words will have an impact on voters’ post-election evaluations of the government’s policy performance.

Summary of Results

The empirical studies mostly, but not conclusively, yield support for the theoretical predictions. The first prediction (H1 in the theory section) was that persuasive words in election pledges would provoke substantive beliefs about policies beyond what was actually being pledged. The prediction was supported by empirical tests, employed in Papers 2 and 4. The findings were not conclusive in Paper 4, however, where the effects were seen only when the persuasive words were left-leaning. Together, the studies yield the strongest support for H1 in cases where the persuasive words are asso-ciated with left-leaning values, and less so when they are associated with universal, or right-leaning values. Potential explanations to these differ-ences are offered in the concluding section of Paper 4.

The second and third prediction (H2 and H3 in the theory section) shifted attention to voters’ support for policies. H2 stipulated that, because persua-sive words in election pledges would provoke beliefs about policies that would make individuals view the policies more positively than how they would otherwise based only on their factual content, the words would also indirectly increase citizens’ support for the policies. H3 stipulated that these indirect effects of persuasive words on policy support would be moderated by the individuals’ ideological predispositions. These predictions were tested in Papers 2 and 3.

Paper 2, which tested the effects of left-leaning and right-leaning per-suasive words, yielded support for both hypotheses. In support of H2 it showed that, by provoking substantive beliefs about the policies that went beyond the exact policy proposals, left- and right-leaning persuasive words indirectly influenced respondents’ support for the pledged policies. In sup-port of H3, it showed that the indirect effects on policy support were mod-erated by the individuals’ left-right predispositions. Individuals who be-lieved that the policies were aimed at achieving a left-leaning (or right-leaning) value, increased their support for the policies, when compared to a control group that was not exposed to persuasive words. But the effects were seen mainly among individuals of corresponding left (or right) ideo-logical predispositions.

Paper 3 tested the effects of universal persuasive words on citizens’ sup-port for policy pledges, and how these effects were moderated by individ-uals’ ideological predispositions. This paper also rendered support for H3. Exposure to universal persuasive words increased respondents’ support for policies, but the effects were seen mainly among individuals who placed themselves close to the center of the left-right spectrum. Paper 3 thus shows

31

that it is not only the effects of ideological (left- and right-leaning) persua-sive words that are being conditioned by the ideological predispositions of the audience. Also universal persuasive words—which are neutral to left-right ideology—have pronounced effects only on certain individuals (albeit “different types” of individuals; those with an ideological “center” orienta-tion).

The final prediction (H4 in the theory section) was that persuasive words in election pledges could come at a cost for parties in the long run (post-election), in terms of an increased perception that they have broken their pledges. This proposal was tested in Paper 4. The study yielded support for the hypothesis in two cases: when the persuasive words were left-leaning, and when they were universal. But the effects were not seen when the per-suasive words were right-leaning. Paper 4 therefore confirms that persua-sive words in election pledges can have negative post-election effects on whether or not people believe that election pledges have been fulfilled; but it also raises questions about the extent to which different types of persua-sive words affect voters. Does the effectiveness of persuasive words in elec-tion pledges vary depending on the type of persuasive words used? And, if so, what makes certain types of persuasive words (e.g., left-leaning words) more effective than others (e.g., right-leaning words)? These questions are discussed in the concluding section of Paper 4.

Limitations and Ecological Validity

While my studies show that persuasive words in election pledges can have an impact on the way voters evaluate and think about policies, it should be noted that, on average, the effects have been fairly small in substantial terms. Considering that an election campaign has a variety of variables that can influence voters’ political attitudes and behavior, such as personal con-tact with politicians (Putnam, 1966) political participation (Finkel, 1987), and media exposure (Banducci & Karp, 2003; Bennett & Entman, 2000; Cappella & Jamieson, 1996; Norris, Curtis, Sanders, Scammell, & Semetko, 1999; Patterson, 1993), it is possible that the effects seen in my experiments may be even more modest when tested in real world election campaigns.

As was mentioned in the methods section, the experiments did not con-trol for information about the pledge-giver (the party label, or party cue as it is often referred in public opinion research, e.g., Zaller, 1992), or whether voters’ partisanship might condition the effects of persuasive words (e.g., Slothuus & de Vreese, 2010). Also, the experiments did not account for situations in which voters were exposed to competing values (Sniderman & Theriault, 2004), or competing messages (e.g., Chong & Drcukman, 2007c, 2013). Below, I discuss the implications of omitting these potentially rele-vant variables, my motives for doing so, and why I still believe that the

30

persuasive words will have an impact on voters’ post-election evaluations of the government’s policy performance.

Summary of Results

The empirical studies mostly, but not conclusively, yield support for the theoretical predictions. The first prediction (H1 in the theory section) was that persuasive words in election pledges would provoke substantive beliefs about policies beyond what was actually being pledged. The prediction was supported by empirical tests, employed in Papers 2 and 4. The findings were not conclusive in Paper 4, however, where the effects were seen only when the persuasive words were left-leaning. Together, the studies yield the strongest support for H1 in cases where the persuasive words are asso-ciated with left-leaning values, and less so when they are associated with universal, or right-leaning values. Potential explanations to these differ-ences are offered in the concluding section of Paper 4.

The second and third prediction (H2 and H3 in the theory section) shifted attention to voters’ support for policies. H2 stipulated that, because persua-sive words in election pledges would provoke beliefs about policies that would make individuals view the policies more positively than how they would otherwise based only on their factual content, the words would also indirectly increase citizens’ support for the policies. H3 stipulated that these indirect effects of persuasive words on policy support would be moderated by the individuals’ ideological predispositions. These predictions were tested in Papers 2 and 3.

Paper 2, which tested the effects of left-leaning and right-leaning per-suasive words, yielded support for both hypotheses. In support of H2 it showed that, by provoking substantive beliefs about the policies that went beyond the exact policy proposals, left- and right-leaning persuasive words indirectly influenced respondents’ support for the pledged policies. In sup-port of H3, it showed that the indirect effects on policy support were mod-erated by the individuals’ left-right predispositions. Individuals who be-lieved that the policies were aimed at achieving a left-leaning (or right-leaning) value, increased their support for the policies, when compared to a control group that was not exposed to persuasive words. But the effects were seen mainly among individuals of corresponding left (or right) ideo-logical predispositions.

Paper 3 tested the effects of universal persuasive words on citizens’ sup-port for policy pledges, and how these effects were moderated by individ-uals’ ideological predispositions. This paper also rendered support for H3. Exposure to universal persuasive words increased respondents’ support for policies, but the effects were seen mainly among individuals who placed themselves close to the center of the left-right spectrum. Paper 3 thus shows

31

that it is not only the effects of ideological (left- and right-leaning) persua-sive words that are being conditioned by the ideological predispositions of the audience. Also universal persuasive words—which are neutral to left-right ideology—have pronounced effects only on certain individuals (albeit “different types” of individuals; those with an ideological “center” orienta-tion).

The final prediction (H4 in the theory section) was that persuasive words in election pledges could come at a cost for parties in the long run (post-election), in terms of an increased perception that they have broken their pledges. This proposal was tested in Paper 4. The study yielded support for the hypothesis in two cases: when the persuasive words were left-leaning, and when they were universal. But the effects were not seen when the per-suasive words were right-leaning. Paper 4 therefore confirms that persua-sive words in election pledges can have negative post-election effects on whether or not people believe that election pledges have been fulfilled; but it also raises questions about the extent to which different types of persua-sive words affect voters. Does the effectiveness of persuasive words in elec-tion pledges vary depending on the type of persuasive words used? And, if so, what makes certain types of persuasive words (e.g., left-leaning words) more effective than others (e.g., right-leaning words)? These questions are discussed in the concluding section of Paper 4.

Limitations and Ecological Validity

While my studies show that persuasive words in election pledges can have an impact on the way voters evaluate and think about policies, it should be noted that, on average, the effects have been fairly small in substantial terms. Considering that an election campaign has a variety of variables that can influence voters’ political attitudes and behavior, such as personal con-tact with politicians (Putnam, 1966) political participation (Finkel, 1987), and media exposure (Banducci & Karp, 2003; Bennett & Entman, 2000; Cappella & Jamieson, 1996; Norris, Curtis, Sanders, Scammell, & Semetko, 1999; Patterson, 1993), it is possible that the effects seen in my experiments may be even more modest when tested in real world election campaigns.

As was mentioned in the methods section, the experiments did not con-trol for information about the pledge-giver (the party label, or party cue as it is often referred in public opinion research, e.g., Zaller, 1992), or whether voters’ partisanship might condition the effects of persuasive words (e.g., Slothuus & de Vreese, 2010). Also, the experiments did not account for situations in which voters were exposed to competing values (Sniderman & Theriault, 2004), or competing messages (e.g., Chong & Drcukman, 2007c, 2013). Below, I discuss the implications of omitting these potentially rele-vant variables, my motives for doing so, and why I still believe that the

32

experimental studies are a valuable contribution to our understanding of how rhetoric can work in real-world election campaigns.

Absence of Party Label

In the experiments I conducted, I chose not to include party labels. The main reason for not including this variable (as is generally the case when causality in specific relationships is tested) was to clear the empirical tests of factors that could decrease the possibility of assessing the causal effects of the key features that I hoped to capture (see e.g., Morton & Williams, 2010). More precisely, I wanted to increase the likelihood that I could test the causal effects of persuasive words on individuals’ support for specific policies, focusing on one mediating and one moderating factor on the indi-vidual level: the individuals’ beliefs about the policies, and the individuals’ ideological left-right predispositions.

To assess such causal mechanisms, it was important that respondents did not associate the policy pledges with any actual political party. This would have introduced treatment bias, making it difficult to isolate the effects of the manipulation—namely, persuasive words (see e.g., Malhotra & Mar-galit, 2014, p. 1007). Not only could information about the pledge-giver have introduced a second manipulation (“party cue”), which would have made it difficult to infer the specific effects of exposure to persuasive words (see e.g., Morton & Williams, 2010, chapter 4), but including party labels in the treatments could also have introduced a number of other, individual level factors, which could have potentially moderated or mediated the ef-fects, aside from those stressed in the theoretical models.

An example of one such factor, which would be fairly easy to control for by asking respondents, is the subjects’ individual party preferences. But there are also factors that would have been more problematic to control for without interfering with the experimental design or introducing potential pre-treatment biases. Examples of such variables would be individuals’ prior information about the party’s policy positions, or what type of infor-mation processing strategies they utilize when they process political ap-peals (e.g., the extent to which they rely on heuristic cues, such as group identity). Since my theoretical models were already fairly complicated (in-cluding both mediation and moderation), I did not want to introduce more variables that could further complicate the analyses; hence, I chose to pre-sent the pledge-givers in my treatments as generic political parties of un-identified ideological left-right positioning.

While isolating the effects of the manipulations (exposure to persuasive words) from party label gave me the chance to determine causality in the relationships predicted in the theoretical models, using an anonymous party as a pledge-giver created additional questions about the magnitude of the effects of persuasive words when used in real election campaigns. Political

33

parties occupy a central role in most theories of public opinion formation (e.g., Campbell, Converse, Miller, & Stokes, 1960; Popkin, 1991; Zaller, 1992), and research has shown that information about the candidates’ party affiliations can directly impact what citizens prefer (Jacoby, 1988; Mondak, 1993; Squire & Smith, 1988), in addition to conditioning the effects of po-litical appeals on individuals’ subsequent attitudes (Slothuus & de Vreese, 2010). It is thus possible that, when voters are informed about pledge-giv-ers, they will support whatever policies are being proposed merely based on whether or not they like (and/or trust) the particular party. Despite these concerns, there is reason to believe that the effects seen in my experiments should also be seen in real-life election campaigns, where the party is known.

There are several situations in real election campaigns where, because of the specific characteristic of the election or electoral competition, the inclusion of a party label should matter less for voters’ evaluation of policy pledges. One such situation could be when voters are choosing between candidates from the same party, such as in presidential or gubernatorial elections (see e.g., Barker, 2005); another could be when voters feel at-tached to a certain coalition of parties, but not attached to a specific party within said coalition (e.g., Gunther & Montero, 2001).

Second, as a result of the decreasing identification with social groups, which have previously been an important determinant for party affiliation (such as economic class), the alignment between parties and voters has been steadily declining (see e.g., Dalton & Wattenberg, 2002; Mair, 2002; Mair & Biezen, 2001; Thomassen, 2005). Instead of group affiliations and party loyalty, citizens now base their party choice on individual policy prefer-ences (e.g., Holmberg & Oscarsson, 2013; Popkin, 1991), and parties’ issue positions (e.g., Aardal & van Wijnen, 2005; Green-Pedersen, 2006, 2007), to a greater extent than they did before. This trend can be seen as an increase in electoral volatility; voters switch parties more often between elections, and the number of voters that are deciding which party to vote for late, even within the final days of election campaigns, is on the rise (e.g., Dassonne-ville, 2012; Mair, 2008; Oscarsson, 2013. For an overview, see Håkansson & Naurin, 2016). In other words, citizens are becoming less concerned about party identification, and more concerned about parties’ policy posi-tions, which speaks to the growing importance of parties’ election pledges, and campaign strategies, for electoral outcomes.

The trends can also be seen in political communication research. In this research, studies have shown that factors such as shared personal values, or “value identity,” can play a part beyond party effects, with how citizens are forming their opinions. Nelson and Garst (2005), for example, found that, while voters indeed made use of party cues when they evaluated political messages that appealed to their core values, they also made use of personal cues, such as a perceived shared individual value identity with the speaker.

32

experimental studies are a valuable contribution to our understanding of how rhetoric can work in real-world election campaigns.

Absence of Party Label

In the experiments I conducted, I chose not to include party labels. The main reason for not including this variable (as is generally the case when causality in specific relationships is tested) was to clear the empirical tests of factors that could decrease the possibility of assessing the causal effects of the key features that I hoped to capture (see e.g., Morton & Williams, 2010). More precisely, I wanted to increase the likelihood that I could test the causal effects of persuasive words on individuals’ support for specific policies, focusing on one mediating and one moderating factor on the indi-vidual level: the individuals’ beliefs about the policies, and the individuals’ ideological left-right predispositions.

To assess such causal mechanisms, it was important that respondents did not associate the policy pledges with any actual political party. This would have introduced treatment bias, making it difficult to isolate the effects of the manipulation—namely, persuasive words (see e.g., Malhotra & Mar-galit, 2014, p. 1007). Not only could information about the pledge-giver have introduced a second manipulation (“party cue”), which would have made it difficult to infer the specific effects of exposure to persuasive words (see e.g., Morton & Williams, 2010, chapter 4), but including party labels in the treatments could also have introduced a number of other, individual level factors, which could have potentially moderated or mediated the ef-fects, aside from those stressed in the theoretical models.

An example of one such factor, which would be fairly easy to control for by asking respondents, is the subjects’ individual party preferences. But there are also factors that would have been more problematic to control for without interfering with the experimental design or introducing potential pre-treatment biases. Examples of such variables would be individuals’ prior information about the party’s policy positions, or what type of infor-mation processing strategies they utilize when they process political ap-peals (e.g., the extent to which they rely on heuristic cues, such as group identity). Since my theoretical models were already fairly complicated (in-cluding both mediation and moderation), I did not want to introduce more variables that could further complicate the analyses; hence, I chose to pre-sent the pledge-givers in my treatments as generic political parties of un-identified ideological left-right positioning.

While isolating the effects of the manipulations (exposure to persuasive words) from party label gave me the chance to determine causality in the relationships predicted in the theoretical models, using an anonymous party as a pledge-giver created additional questions about the magnitude of the effects of persuasive words when used in real election campaigns. Political

33

parties occupy a central role in most theories of public opinion formation (e.g., Campbell, Converse, Miller, & Stokes, 1960; Popkin, 1991; Zaller, 1992), and research has shown that information about the candidates’ party affiliations can directly impact what citizens prefer (Jacoby, 1988; Mondak, 1993; Squire & Smith, 1988), in addition to conditioning the effects of po-litical appeals on individuals’ subsequent attitudes (Slothuus & de Vreese, 2010). It is thus possible that, when voters are informed about pledge-giv-ers, they will support whatever policies are being proposed merely based on whether or not they like (and/or trust) the particular party. Despite these concerns, there is reason to believe that the effects seen in my experiments should also be seen in real-life election campaigns, where the party is known.

There are several situations in real election campaigns where, because of the specific characteristic of the election or electoral competition, the inclusion of a party label should matter less for voters’ evaluation of policy pledges. One such situation could be when voters are choosing between candidates from the same party, such as in presidential or gubernatorial elections (see e.g., Barker, 2005); another could be when voters feel at-tached to a certain coalition of parties, but not attached to a specific party within said coalition (e.g., Gunther & Montero, 2001).

Second, as a result of the decreasing identification with social groups, which have previously been an important determinant for party affiliation (such as economic class), the alignment between parties and voters has been steadily declining (see e.g., Dalton & Wattenberg, 2002; Mair, 2002; Mair & Biezen, 2001; Thomassen, 2005). Instead of group affiliations and party loyalty, citizens now base their party choice on individual policy prefer-ences (e.g., Holmberg & Oscarsson, 2013; Popkin, 1991), and parties’ issue positions (e.g., Aardal & van Wijnen, 2005; Green-Pedersen, 2006, 2007), to a greater extent than they did before. This trend can be seen as an increase in electoral volatility; voters switch parties more often between elections, and the number of voters that are deciding which party to vote for late, even within the final days of election campaigns, is on the rise (e.g., Dassonne-ville, 2012; Mair, 2008; Oscarsson, 2013. For an overview, see Håkansson & Naurin, 2016). In other words, citizens are becoming less concerned about party identification, and more concerned about parties’ policy posi-tions, which speaks to the growing importance of parties’ election pledges, and campaign strategies, for electoral outcomes.

The trends can also be seen in political communication research. In this research, studies have shown that factors such as shared personal values, or “value identity,” can play a part beyond party effects, with how citizens are forming their opinions. Nelson and Garst (2005), for example, found that, while voters indeed made use of party cues when they evaluated political messages that appealed to their core values, they also made use of personal cues, such as a perceived shared individual value identity with the speaker.

34

In fact, contrary to the hypothesis that a party label would act as the main cue (e.g., Zaller, 1992), the respondents in their study seemed more suscep-tible to an appeal when the speaker used values similar to their own, than when the speaker came from a party matching their own party preference (Nelson & Garst, 2005, p. 510). These findings indicate that individual val-ues and shared value identity are becoming more important to voters than categorical party identification (Nelson & Garst, 2005, p. 510. Also see Kam, 2005; Petty & Cacioppo, 1990; Reid & Deaux, 1996; Turner, Oakes, Haslam, & McGarty, 1994). As Nelson and Garst (2005, p. 510) state: val-ues (or value-based rhetoric) may simply serve as “more striking, salient communication markers” for these individuals, which emphasizes how im-portant such value-laden rhetoric is in election pledges.

To summarize, individuals’ policy preferences and the ability of parties to engage voters based on their policy proposals are becoming increasingly important to electoral outcomes, meaning that even small effects of election campaigning can be decisive for the electoral outcomes if the competition is tight (e.g., Oscarsson, 2013). This puts pressure on parties to formulate policy pledges in ways that are appealing to voters, while capturing their attention. As a result, electoral competition, with its increased profession-alization (e.g., Strömbäck, 2007) and use of marketing strategies (e.g., Strömbäck, Mitrook, & Kiousis, 2010), has intensified, and parties have begun to rely more heavily on campaign experts in order to attract prospec-tive voters with their policies (e.g., Perloff, 2010; Strömbäck, 2007, pp. 54–55).

The increasing role of parties’ election campaigns shifts our attention to a second variable which has the potential to condition the effects of persua-sive words in election campaigns; namely, competing messages from op-posing parties, media, and the like.

Absence of Competing Information

As is the case with party labels, presenting respondents with one-sided mes-sages and omitting any competing information, helps to isolate the causal effects of persuasive words on individuals’ beliefs about, and subsequent support for, policy pledges. Nevertheless, excluding competing information from the experimental designs does raise questions about the scope of the effects of persuasive words when they are used in real-life campaigns. Prior studies have shown that competing messages can indeed limit the effects of campaign rhetoric (e.g., Chong & Drcukman, 2007c, 2013; Sniderman & Theriault, 2004), which means that the effects from the treatments in my experiments could be exaggerated.

On the other hand, scholars in this line of research point out that, alt-hough competing information can limit the effects of campaign rhetoric, it does not seem to eliminate them (see Chong & Druckman, 2007c, 2013).

35

Rather, the effects of campaign messages in competing environments likely depend on the type of message, and the circumstances under which it is relayed. One factor that should increase the effect of campaign rhetoric is the prevalence (and thereby the accessibility) of these messages (Higgins, 1996; Iyengar, 1990, 1991; Zaller, 1992), relative to other competing mes-sages. Exposing voters to the same or similar rhetoric over time can thus help to sustain, or even enhance, the effects of rhetoric in election campaigns (Chong & Druckman, 2013; Lecheler & de Vreese, 2013; Lecheler, Keer, Schuck, & Hänggli, 2015). Parties that have a wealth of campaign resources should therefore have an advantage in successfully reaching out to voters with their campaign rhetoric (Chong & Druckman, 2013, p. 13).

Moreover, the effect of campaign messages also depends on their per-ceived strength (Chong & Druckman, 2007b, p. 104; Chong & Druckman, 2007c). One example of a factor that can increase the strength of a political appeal is referring to core values (Chong & Druckman, 2007a, p. 112). Alt-hough being exposed to competing values on a policy issue may decrease the aggregate effects of such value-laden rhetoric, it can reinforce the ef-fects among certain groups of voters, whose prior value predispositions cor-respond with the rhetoric (see e.g., Brewer, 2002, 2008; Chong & Druck-man, 2013; Sniderman & Theriault, 2004). Policy debates that include a value conflict can, in other words, trigger voters’ value predispositions, thereby enhancing the effects of value-laden rhetoric among those who share the same value predispositions. Simply put, the effects of campaign rhetoric are likely a combination of the environment, the type of rhetoric employed, and the predispositions of the audience.

In sum, whereas one could expect competing political messages to play a part in the effects of election pledge rhetoric, such variables do not nec-essarily eliminate, or even weaken, the effects. To assess the relative impact of persuasive words on real-world election campaigns, future studies should test the effects of persuasive words controlling for exposure to com-peting messages, long-term effects of repeated exposure to the same per-suasive words, and the effects of persuasive words when party labels are present. Such future studies should also include a control for individual fac-tors that can potentially moderate voters’ susceptibility to persuasive words, such as political awareness (e.g., Kam, 2005; Slothuus, 2008)—a variable upon which my samples in the studies were unbalanced in com-parison with more representative, probability-recruited samples.

Design of Paper IV

A final limitation that should be mentioned concerns the design of Paper 4, on post-election effects of election pledge rhetoric. Participants in this study were exposed to policy outcomes directly related to the exposure of the

34

In fact, contrary to the hypothesis that a party label would act as the main cue (e.g., Zaller, 1992), the respondents in their study seemed more suscep-tible to an appeal when the speaker used values similar to their own, than when the speaker came from a party matching their own party preference (Nelson & Garst, 2005, p. 510). These findings indicate that individual val-ues and shared value identity are becoming more important to voters than categorical party identification (Nelson & Garst, 2005, p. 510. Also see Kam, 2005; Petty & Cacioppo, 1990; Reid & Deaux, 1996; Turner, Oakes, Haslam, & McGarty, 1994). As Nelson and Garst (2005, p. 510) state: val-ues (or value-based rhetoric) may simply serve as “more striking, salient communication markers” for these individuals, which emphasizes how im-portant such value-laden rhetoric is in election pledges.

To summarize, individuals’ policy preferences and the ability of parties to engage voters based on their policy proposals are becoming increasingly important to electoral outcomes, meaning that even small effects of election campaigning can be decisive for the electoral outcomes if the competition is tight (e.g., Oscarsson, 2013). This puts pressure on parties to formulate policy pledges in ways that are appealing to voters, while capturing their attention. As a result, electoral competition, with its increased profession-alization (e.g., Strömbäck, 2007) and use of marketing strategies (e.g., Strömbäck, Mitrook, & Kiousis, 2010), has intensified, and parties have begun to rely more heavily on campaign experts in order to attract prospec-tive voters with their policies (e.g., Perloff, 2010; Strömbäck, 2007, pp. 54–55).

The increasing role of parties’ election campaigns shifts our attention to a second variable which has the potential to condition the effects of persua-sive words in election campaigns; namely, competing messages from op-posing parties, media, and the like.

Absence of Competing Information

As is the case with party labels, presenting respondents with one-sided mes-sages and omitting any competing information, helps to isolate the causal effects of persuasive words on individuals’ beliefs about, and subsequent support for, policy pledges. Nevertheless, excluding competing information from the experimental designs does raise questions about the scope of the effects of persuasive words when they are used in real-life campaigns. Prior studies have shown that competing messages can indeed limit the effects of campaign rhetoric (e.g., Chong & Drcukman, 2007c, 2013; Sniderman & Theriault, 2004), which means that the effects from the treatments in my experiments could be exaggerated.

On the other hand, scholars in this line of research point out that, alt-hough competing information can limit the effects of campaign rhetoric, it does not seem to eliminate them (see Chong & Druckman, 2007c, 2013).

35

Rather, the effects of campaign messages in competing environments likely depend on the type of message, and the circumstances under which it is relayed. One factor that should increase the effect of campaign rhetoric is the prevalence (and thereby the accessibility) of these messages (Higgins, 1996; Iyengar, 1990, 1991; Zaller, 1992), relative to other competing mes-sages. Exposing voters to the same or similar rhetoric over time can thus help to sustain, or even enhance, the effects of rhetoric in election campaigns (Chong & Druckman, 2013; Lecheler & de Vreese, 2013; Lecheler, Keer, Schuck, & Hänggli, 2015). Parties that have a wealth of campaign resources should therefore have an advantage in successfully reaching out to voters with their campaign rhetoric (Chong & Druckman, 2013, p. 13).

Moreover, the effect of campaign messages also depends on their per-ceived strength (Chong & Druckman, 2007b, p. 104; Chong & Druckman, 2007c). One example of a factor that can increase the strength of a political appeal is referring to core values (Chong & Druckman, 2007a, p. 112). Alt-hough being exposed to competing values on a policy issue may decrease the aggregate effects of such value-laden rhetoric, it can reinforce the ef-fects among certain groups of voters, whose prior value predispositions cor-respond with the rhetoric (see e.g., Brewer, 2002, 2008; Chong & Druck-man, 2013; Sniderman & Theriault, 2004). Policy debates that include a value conflict can, in other words, trigger voters’ value predispositions, thereby enhancing the effects of value-laden rhetoric among those who share the same value predispositions. Simply put, the effects of campaign rhetoric are likely a combination of the environment, the type of rhetoric employed, and the predispositions of the audience.

In sum, whereas one could expect competing political messages to play a part in the effects of election pledge rhetoric, such variables do not nec-essarily eliminate, or even weaken, the effects. To assess the relative impact of persuasive words on real-world election campaigns, future studies should test the effects of persuasive words controlling for exposure to com-peting messages, long-term effects of repeated exposure to the same per-suasive words, and the effects of persuasive words when party labels are present. Such future studies should also include a control for individual fac-tors that can potentially moderate voters’ susceptibility to persuasive words, such as political awareness (e.g., Kam, 2005; Slothuus, 2008)—a variable upon which my samples in the studies were unbalanced in com-parison with more representative, probability-recruited samples.

Design of Paper IV

A final limitation that should be mentioned concerns the design of Paper 4, on post-election effects of election pledge rhetoric. Participants in this study were exposed to policy outcomes directly related to the exposure of the

36

election pledge. It is possible (or even likely) that voters in real world de-mocracies, when exposed to policy outcomes, will not remember the exact policy pledges from the campaign. This could make comparisons between the policy outcomes and the actual election pledges become difficult, and voters could evaluate the policy outcomes mainly on the basis of whether they like them or not.

But it is also possible that voters, in real electoral cycles, will evaluate the policy outcomes based on their emotional expectations (i.e., whether they expect positive or negative outcomes) of the policies to a greater extent than participants in my experiment did. These participants were exposed to outcomes immediately related to the actual pledges. Lodge, Steenbergen and Brau (1995), for example, showed that, while they did not remember the exact information they received in election campaigns, voters stored the emotions they got based on the information in a “running tally,” which then guided them through their electoral decisions, even when the actual pledges had been forgotten (for similar findings, see also e.g., Lodge & Taber, 2005; Morris, Squires, Taber, & Lodge, 2003). Thus, if persuasive words could create these types of emotionally-driven expectations before an election, then the effects of persuasive words on voters’ evaluations of the policy outcomes could actually increase as the time between the exposure to the election pledges, and their experience of the policy outcomes, increases.

To assess the magnitude of the post-election effects of election pledge rhetoric in real-election campaigns, future studies should include a time lapse between the moment when respondents are exposed to election pledges, and when they are exposed to information about the policy out-comes.

Concluding Discussion

The aim of this dissertation is to increase our understanding of how value-laden rhetoric influences voters’ evaluations of election pledges before, and after, elections. I argue that, while contemporary research on framing ef-fects teaches us a lot about how voters’ support for policies is influenced when policies are framed in value-laden terms, the theory on framing ef-fects is inconclusive regarding the individual micro-level mechanisms which lead up to those effects.

Drawing on linguistic theory on persuasive words, I suggest, and test, a linguistic factor which can add to contemporary explanations of framing effects—one which is rarely discussed in the framing literature. In addition to this, I contribute to the linguistic literature by empirically testing the the-ory on persuasive words—something which has rarely been done—and by developing the theory to account for different types of persuasive words, as well as their subsequent effects on individuals with different ideological predispositions.

37

The general implications are that the use of persuasive words in election pledges can influence, not only whether individuals like the policies, but also their cognitive beliefs on the content of these policies. Drawing on the theory on persuasive words, I show that, if frames include persuasive words, then they can provoke beliefs about policies which span beyond the content of the exact policy proposals being outlined in election pledges. This adds a new perspective on framing effects, which have traditionally been restricted to situations where politicians (or other elites, such as the news media) have emphasized certain (already existing) beliefs about a pol-icy, meaning that individuals have assigned a greater weight to those beliefs (relative to other existing beliefs) when forming their own political attitudes (e.g., Nelson, Oxley, & Clawson, 1997; Iyengar, 1991; Price & Tewksbury, 1997).

Moreover, this dissertation shows that different types of persuasive words can impact voters differently, depending on their ideological predis-positions. Persuasive words that denote left- or right-leaning values seem to attract individuals with the same left-right predispositions, whereas uni-versal persuasive words seem to attract individuals close to the center of the left-right spectrum. These findings indicate that parties in real-world election campaigns do, in fact, use different types of persuasive words to target different groups of voters. When parties aim to reach the “median voters,” usually found around the center of the left-right divide, it would be more likely for them to use universal persuasive words. But when parties aim for ideological core voters of pronounced left- or right predispositions, the use of traditional left- and right-leaning persuasive words seems more likely.

This dissertation also provides some insights on the post-election effects of using persuasive words in election pledges. It shows that certain types of persuasive words, left-leaning, and somewhat universal persuasive words, can influence individuals’ substantive expectations on policies in a way that could negatively impact their perceptions of whether the pledges have been fulfilled (post-election). The summary takeaway is thus that, in the short term (pre-election), parties using persuasive words in election pledges may gain support for these policies from certain groups of voters that they choose to target. The drawback is that doing this can leave some of these voters with the feeling that the policy outcomes do not match the pre-elec-tion pledges, which can, in turn, make them perceive that the pledges are slightly less fulfilled.

I end with some reflections on the larger implications of the findings concerning the role of campaign rhetoric for the functioning of mandate models of representation, and how voters and voters’ policy preferences can be viewed in light of the effects of such rhetoric.

Scholars have drawn different conclusions about voters’ political capa-bilities, when their support for policies has been influenced by the way

36

election pledge. It is possible (or even likely) that voters in real world de-mocracies, when exposed to policy outcomes, will not remember the exact policy pledges from the campaign. This could make comparisons between the policy outcomes and the actual election pledges become difficult, and voters could evaluate the policy outcomes mainly on the basis of whether they like them or not.

But it is also possible that voters, in real electoral cycles, will evaluate the policy outcomes based on their emotional expectations (i.e., whether they expect positive or negative outcomes) of the policies to a greater extent than participants in my experiment did. These participants were exposed to outcomes immediately related to the actual pledges. Lodge, Steenbergen and Brau (1995), for example, showed that, while they did not remember the exact information they received in election campaigns, voters stored the emotions they got based on the information in a “running tally,” which then guided them through their electoral decisions, even when the actual pledges had been forgotten (for similar findings, see also e.g., Lodge & Taber, 2005; Morris, Squires, Taber, & Lodge, 2003). Thus, if persuasive words could create these types of emotionally-driven expectations before an election, then the effects of persuasive words on voters’ evaluations of the policy outcomes could actually increase as the time between the exposure to the election pledges, and their experience of the policy outcomes, increases.

To assess the magnitude of the post-election effects of election pledge rhetoric in real-election campaigns, future studies should include a time lapse between the moment when respondents are exposed to election pledges, and when they are exposed to information about the policy out-comes.

Concluding Discussion

The aim of this dissertation is to increase our understanding of how value-laden rhetoric influences voters’ evaluations of election pledges before, and after, elections. I argue that, while contemporary research on framing ef-fects teaches us a lot about how voters’ support for policies is influenced when policies are framed in value-laden terms, the theory on framing ef-fects is inconclusive regarding the individual micro-level mechanisms which lead up to those effects.

Drawing on linguistic theory on persuasive words, I suggest, and test, a linguistic factor which can add to contemporary explanations of framing effects—one which is rarely discussed in the framing literature. In addition to this, I contribute to the linguistic literature by empirically testing the the-ory on persuasive words—something which has rarely been done—and by developing the theory to account for different types of persuasive words, as well as their subsequent effects on individuals with different ideological predispositions.

37

The general implications are that the use of persuasive words in election pledges can influence, not only whether individuals like the policies, but also their cognitive beliefs on the content of these policies. Drawing on the theory on persuasive words, I show that, if frames include persuasive words, then they can provoke beliefs about policies which span beyond the content of the exact policy proposals being outlined in election pledges. This adds a new perspective on framing effects, which have traditionally been restricted to situations where politicians (or other elites, such as the news media) have emphasized certain (already existing) beliefs about a pol-icy, meaning that individuals have assigned a greater weight to those beliefs (relative to other existing beliefs) when forming their own political attitudes (e.g., Nelson, Oxley, & Clawson, 1997; Iyengar, 1991; Price & Tewksbury, 1997).

Moreover, this dissertation shows that different types of persuasive words can impact voters differently, depending on their ideological predis-positions. Persuasive words that denote left- or right-leaning values seem to attract individuals with the same left-right predispositions, whereas uni-versal persuasive words seem to attract individuals close to the center of the left-right spectrum. These findings indicate that parties in real-world election campaigns do, in fact, use different types of persuasive words to target different groups of voters. When parties aim to reach the “median voters,” usually found around the center of the left-right divide, it would be more likely for them to use universal persuasive words. But when parties aim for ideological core voters of pronounced left- or right predispositions, the use of traditional left- and right-leaning persuasive words seems more likely.

This dissertation also provides some insights on the post-election effects of using persuasive words in election pledges. It shows that certain types of persuasive words, left-leaning, and somewhat universal persuasive words, can influence individuals’ substantive expectations on policies in a way that could negatively impact their perceptions of whether the pledges have been fulfilled (post-election). The summary takeaway is thus that, in the short term (pre-election), parties using persuasive words in election pledges may gain support for these policies from certain groups of voters that they choose to target. The drawback is that doing this can leave some of these voters with the feeling that the policy outcomes do not match the pre-elec-tion pledges, which can, in turn, make them perceive that the pledges are slightly less fulfilled.

I end with some reflections on the larger implications of the findings concerning the role of campaign rhetoric for the functioning of mandate models of representation, and how voters and voters’ policy preferences can be viewed in light of the effects of such rhetoric.

Scholars have drawn different conclusions about voters’ political capa-bilities, when their support for policies has been influenced by the way

38

elites have chosen to present them. The negative interpretation of such find-ings is that voters are not able to fully comprehend politics, and that, at best, they possess weak and unstable policy preferences (e.g., Iyengar, 1990, 1991; Riker, 1986; Zaller, 1992). From this perspective, the fact that my studies showed voters to be affected (albeit modestly) by the exposure to persuasive words on policy support, would indicate that voters have ill-founded policy preferences.

But the findings can also be interpreted more positively for voters. First, the fact that the effects seen in my empirical studies are modest indicates that, in general, the respondents actually based their support for the policies on the information they received about the actual policy content. Second, the analyses on moderated mediation showed that, when persuasive words did influence individuals’ support for policies, this occurred mainly when they also 1) impacted their cognitive beliefs about what the policies were aiming for, and 2) coincided with the fact that the ideological left-right lean-ing of these beliefs corresponded with their own left-right predispositions. These findings suggest that the participants in my studies did not arbitrarily change their policy preferences depending on whatever information they received at the time. Instead, the participants seemed to consider the infor-mation given, incorporated it into their evaluative process, and expressed their support for the policies based on 1) whether or not they believed that said policies would actually lead to values indicated by the persuasive words, and 2) whether or not they endorsed these values (as indicated by a correspondence between the left-right leaning of the values, and where they placed themselves on the left-right scale). In other words, the findings sug-gest that the respondents used the persuasive words to interpret the policies, and this made them think that their underlying preferences would be met through these specific policies.

In the well-known novel “Nineteen Eighty-Four” (1949), George Orwell wrote; “say freedom, and people will buy slavery.” While this may be true to a certain extent (although this example is somewhat extreme), the find-ings of this thesis do not suggest that voters buy whatever politicians are saying. Rather it suggests that, when politicians use words such as “free-dom,” they can make voters see things in the policies that are simply not there, based on their factual content. In other words, changes in opinion as a result of exposure to political appeals may not always reflect changes in voters’ actual policy preferences, but instead (mis)interpretations of whether the policies and the consequences of these policies will meet those preferences.

Suggestions for Future Research

I began the introduction by stating that parties often use value-laden words such as freedom and equality. These are words that clearly fit the criteria of

39

persuasive words; hence it should be uncontroversial to say that parties make use of persuasive words in real election pledges. However, the fre-quency and various forms in which such words are used in real election pledges has not yet been studied, and therefore remains something worth investigating in the future.

In my theoretical paper, I used insights from linguistic theory on persua-sive words to suggest an explanation to value framing effects, something which the framing literature has seldom touched on. This would presuppose that frames sometimes also include persuasive words, a proposition that has not been empirically investigated in my dissertation. A promising path for future research would therefore be to examine the extent to which frames include persuasive words in real life communication. A meta study of the use of persuasive words in existing experimental studies on framing effects, as well as the effects of framing studies that include and do not include persuasive words, could also be an interesting continuation of this thesis.

Another question that remains for future research is the relative impact of frames and persuasive words, when these are used simultaneously in the same message. To effectively test the causal relationships between key var-iables proposed in the theoretical model—the relationship between specific semantic characteristics of words, and individuals’ evaluation of messages including these words—I used experimental treatments free from other types of rhetoric that could influence this psychological process. However, it is likely that real election pledges would include combinations of differ-ent types of rhetoric, such as framing and persuasive words. Future research should thus investigate the different combinations of campaign rhetoric, and the extent to which different combinations lead to different outcomes in terms of “framing effects” (changes in the weight individuals assign to existing beliefs about a policy), and “persuasion” (changes in the content of those existing beliefs).

Finally, this dissertation focuses on the effects of persuasive words in election pledges. However, the theory on persuasive words is general, and it extends to other contexts where an actor either has direct incentives to influence the attitudes of an audience, or might do so mistakenly by using “unfortunate” formulations. To test the generality of the linguistic theory on persuasive words, studies could also be employed in other contexts where these kinds of words are likely to be used. One logical extension of this thesis would be to investigate the effects of persuasive words when delivered by politicians in public speeches or televised debates (i.e., beyond written manifestos). Examples of other contexts that are mentioned in the linguistic literature, and which could also be studied further, are journalistic news media reports (Walton, 2007a) and legal communication (e.g., Bon-giovanni et al., 2017; Walton, 2007b).

38

elites have chosen to present them. The negative interpretation of such find-ings is that voters are not able to fully comprehend politics, and that, at best, they possess weak and unstable policy preferences (e.g., Iyengar, 1990, 1991; Riker, 1986; Zaller, 1992). From this perspective, the fact that my studies showed voters to be affected (albeit modestly) by the exposure to persuasive words on policy support, would indicate that voters have ill-founded policy preferences.

But the findings can also be interpreted more positively for voters. First, the fact that the effects seen in my empirical studies are modest indicates that, in general, the respondents actually based their support for the policies on the information they received about the actual policy content. Second, the analyses on moderated mediation showed that, when persuasive words did influence individuals’ support for policies, this occurred mainly when they also 1) impacted their cognitive beliefs about what the policies were aiming for, and 2) coincided with the fact that the ideological left-right lean-ing of these beliefs corresponded with their own left-right predispositions. These findings suggest that the participants in my studies did not arbitrarily change their policy preferences depending on whatever information they received at the time. Instead, the participants seemed to consider the infor-mation given, incorporated it into their evaluative process, and expressed their support for the policies based on 1) whether or not they believed that said policies would actually lead to values indicated by the persuasive words, and 2) whether or not they endorsed these values (as indicated by a correspondence between the left-right leaning of the values, and where they placed themselves on the left-right scale). In other words, the findings sug-gest that the respondents used the persuasive words to interpret the policies, and this made them think that their underlying preferences would be met through these specific policies.

In the well-known novel “Nineteen Eighty-Four” (1949), George Orwell wrote; “say freedom, and people will buy slavery.” While this may be true to a certain extent (although this example is somewhat extreme), the find-ings of this thesis do not suggest that voters buy whatever politicians are saying. Rather it suggests that, when politicians use words such as “free-dom,” they can make voters see things in the policies that are simply not there, based on their factual content. In other words, changes in opinion as a result of exposure to political appeals may not always reflect changes in voters’ actual policy preferences, but instead (mis)interpretations of whether the policies and the consequences of these policies will meet those preferences.

Suggestions for Future Research

I began the introduction by stating that parties often use value-laden words such as freedom and equality. These are words that clearly fit the criteria of

39

persuasive words; hence it should be uncontroversial to say that parties make use of persuasive words in real election pledges. However, the fre-quency and various forms in which such words are used in real election pledges has not yet been studied, and therefore remains something worth investigating in the future.

In my theoretical paper, I used insights from linguistic theory on persua-sive words to suggest an explanation to value framing effects, something which the framing literature has seldom touched on. This would presuppose that frames sometimes also include persuasive words, a proposition that has not been empirically investigated in my dissertation. A promising path for future research would therefore be to examine the extent to which frames include persuasive words in real life communication. A meta study of the use of persuasive words in existing experimental studies on framing effects, as well as the effects of framing studies that include and do not include persuasive words, could also be an interesting continuation of this thesis.

Another question that remains for future research is the relative impact of frames and persuasive words, when these are used simultaneously in the same message. To effectively test the causal relationships between key var-iables proposed in the theoretical model—the relationship between specific semantic characteristics of words, and individuals’ evaluation of messages including these words—I used experimental treatments free from other types of rhetoric that could influence this psychological process. However, it is likely that real election pledges would include combinations of differ-ent types of rhetoric, such as framing and persuasive words. Future research should thus investigate the different combinations of campaign rhetoric, and the extent to which different combinations lead to different outcomes in terms of “framing effects” (changes in the weight individuals assign to existing beliefs about a policy), and “persuasion” (changes in the content of those existing beliefs).

Finally, this dissertation focuses on the effects of persuasive words in election pledges. However, the theory on persuasive words is general, and it extends to other contexts where an actor either has direct incentives to influence the attitudes of an audience, or might do so mistakenly by using “unfortunate” formulations. To test the generality of the linguistic theory on persuasive words, studies could also be employed in other contexts where these kinds of words are likely to be used. One logical extension of this thesis would be to investigate the effects of persuasive words when delivered by politicians in public speeches or televised debates (i.e., beyond written manifestos). Examples of other contexts that are mentioned in the linguistic literature, and which could also be studied further, are journalistic news media reports (Walton, 2007a) and legal communication (e.g., Bon-giovanni et al., 2017; Walton, 2007b).

40

Clearly, the preceding discussions indicate multiple avenues for future research, in order to further assess the impact of persuasive words on indi-viduals’ understanding of, and support for, political issues

References

Aardal, B., van Wijnen, P (2005). Issue Voting. In J. Thomassen (Ed.), The European Voter (pp. 192–212). Oxford, UK: Oxford University Press.

Ajzen, I., & Fishbein, M. (1980). Understanding attitudes and predicting social behaviour. Englewood Cliffs, NJ: Prentice-Hall.

Aragonès, E., & Neeman, Z. (2000). Strategic ambiguity in electoral com-petition. Journal of Theoretical Politics, 12(2), 183–204.

Artés, J. (2011). Do Spanish politicians keep their promises? Party Politics, 19(1), 143–158.

Ball-Rokeach, S., Power, G., Guthrie, K., & Waring, H. (1990). Value-framing abortion in the United States: An application of media system dependency theory. International Journal of Public Opinion Research, 2(3), 249–273.

Banducci, S., & Karp, J. (2003). How elections change the way citizens view the political system: Campaigns, media effects and electoral out-comes in comparative perspective. British Journal of Political Science, 33(03), 443–467.

Barker, D. (2005). Values, frames, and persuasion in presidential nomina-tion campaigns. Political Behavior, 27(4), 375–394.

Bench-Capon, T. (2003). Persuasion in practical argument using value-based argumentation frameworks. Journal of Logic and Computation, 13(3), 429–448.

Bennett, W., & Entman, R. (2000). Mediated politics: Communication in the future of democracy. New York, NY: Cambridge University Press.

Blair, J., Czaja, R., & Blair, E. (2013). Designing surveys: A guide to deci-sions and

procedures. Thousand Oaks, CA: SAGE Publications. Bongiovanni, G., Postema, G., Rotolo, A., Sartor, G., & Walton, D. (2017).

Handbook of legal reasoning and argumentation. Berlin, Germany: Springer.

Born, A., van Eck, P., & Johannesson, M (2017). An experimental investi-gation of election promises. Political Psychology, n/a-n/a.

Jost, J., Federico, C., & Napier, J. (2009). Political ideology: Its structure, functions, and elective affinities. Annual Review of Psychology, 60(1), 307–337.

Achen, C., & Bartels, L. (1016). Democracy for realists: Why elections do not produce responsive government. Princeton, NJ: Princeton University Press.

41

Brace, N., Snelgar, R., & Kemp, R. (2012). SPSS for Psychologists (5th edition). Basingstoke, UK: Palgrave Macmillan.

Brewer, P. (2001). Value words and lizard brains: Do citizens deliberate about appeals to their core values? Political Psychology, 22(1), 45–64.

Brewer, P. (2002). Framing, value words, and citizens’ explanations of their issue opinions. Political Communication, 19(3), 303–316.

Brewer, P. (2003). Values, political knowledge, and public opinion about gay rights: A framing-based account. Public Opinion Quarterly, 67(2), 173–201.

Brewer, P. (2008). Value war: Public opinion and the politics of gay rights. Lanham, MD: Rowman & Littlefield Publishers.

Brewer, P., & Gross, K. (2005). Values, framing, and citizens’ thoughts about policy issues: Effects on content and quantity. Political Psycho-logy, 26(6), 929–948.

Budge, I, Klingemann, H-D., Volkens, A., Bara, J., & Tannenbaum, E. (2001). Mapping policy preferences: Estimates for parties, electors, and governments 1945–1998. Oxford, UK: Oxford University Press.

Callaghan, K., & Schnell, F. (2005). Introduction: Framing political issues in American politics. In K. Callaghan & F. Schnell (Eds.), Framing American politics (pp. 1–17). Pittsburgh, PA: University of Pittsburgh Press.

Callegaro, M., Baker, R., Bethlehem, J., Göritz, A., Krosnick, J., & Lav-rakas, P. (2014). Online panel research: A data quality perspective. West Sussex, UK: John Wiley & Sons.

Campbell, A., Converse, P., Miller, W., & Stokes, D. (1960). The American voter. Ann Arbor, MI: University of Michigan Press.

Cappella, J., & Jamieson, K. (1996). News frames, political cynicism, and media cynicism. The Annals of the American Academy of Political and Social Science, 546, 71–84.

Carpini, D., & Keeter, S. (1996). What Americans know about politics and why it matters. New Haven, CT: Yale University Press.

Chong, D., & Druckman, J. (2007a). Framing theory. Annual Review of Political Science, 10, 103–126.

Chong, D., & Druckman, J. (2007b). A Theory of framing and opinion for-mation in competitive elite environments. Journal of Communication, 57(1), 99–118.

Chong, D., & Druckman, J. (2007c). Framing public opinion in competitive democracies. The American Political Science Review, 101(4), 637–655. doi:10.2307/27644476

Chong, D., & Druckman, J. (2013). Counterframing effects. The Journal of Politics, 75(01), 1–16.

Clifford, S., & Jerit, J. (2013). How words do the work of politics: Moral foundations theory and the debate over stem cell research. The Journal of Politics, 75(03), 659–671.

40

Clearly, the preceding discussions indicate multiple avenues for future research, in order to further assess the impact of persuasive words on indi-viduals’ understanding of, and support for, political issues

References

Aardal, B., van Wijnen, P (2005). Issue Voting. In J. Thomassen (Ed.), The European Voter (pp. 192–212). Oxford, UK: Oxford University Press.

Ajzen, I., & Fishbein, M. (1980). Understanding attitudes and predicting social behaviour. Englewood Cliffs, NJ: Prentice-Hall.

Aragonès, E., & Neeman, Z. (2000). Strategic ambiguity in electoral com-petition. Journal of Theoretical Politics, 12(2), 183–204.

Artés, J. (2011). Do Spanish politicians keep their promises? Party Politics, 19(1), 143–158.

Ball-Rokeach, S., Power, G., Guthrie, K., & Waring, H. (1990). Value-framing abortion in the United States: An application of media system dependency theory. International Journal of Public Opinion Research, 2(3), 249–273.

Banducci, S., & Karp, J. (2003). How elections change the way citizens view the political system: Campaigns, media effects and electoral out-comes in comparative perspective. British Journal of Political Science, 33(03), 443–467.

Barker, D. (2005). Values, frames, and persuasion in presidential nomina-tion campaigns. Political Behavior, 27(4), 375–394.

Bench-Capon, T. (2003). Persuasion in practical argument using value-based argumentation frameworks. Journal of Logic and Computation, 13(3), 429–448.

Bennett, W., & Entman, R. (2000). Mediated politics: Communication in the future of democracy. New York, NY: Cambridge University Press.

Blair, J., Czaja, R., & Blair, E. (2013). Designing surveys: A guide to deci-sions and

procedures. Thousand Oaks, CA: SAGE Publications. Bongiovanni, G., Postema, G., Rotolo, A., Sartor, G., & Walton, D. (2017).

Handbook of legal reasoning and argumentation. Berlin, Germany: Springer.

Born, A., van Eck, P., & Johannesson, M (2017). An experimental investi-gation of election promises. Political Psychology, n/a-n/a.

Jost, J., Federico, C., & Napier, J. (2009). Political ideology: Its structure, functions, and elective affinities. Annual Review of Psychology, 60(1), 307–337.

Achen, C., & Bartels, L. (1016). Democracy for realists: Why elections do not produce responsive government. Princeton, NJ: Princeton University Press.

41

Brace, N., Snelgar, R., & Kemp, R. (2012). SPSS for Psychologists (5th edition). Basingstoke, UK: Palgrave Macmillan.

Brewer, P. (2001). Value words and lizard brains: Do citizens deliberate about appeals to their core values? Political Psychology, 22(1), 45–64.

Brewer, P. (2002). Framing, value words, and citizens’ explanations of their issue opinions. Political Communication, 19(3), 303–316.

Brewer, P. (2003). Values, political knowledge, and public opinion about gay rights: A framing-based account. Public Opinion Quarterly, 67(2), 173–201.

Brewer, P. (2008). Value war: Public opinion and the politics of gay rights. Lanham, MD: Rowman & Littlefield Publishers.

Brewer, P., & Gross, K. (2005). Values, framing, and citizens’ thoughts about policy issues: Effects on content and quantity. Political Psycho-logy, 26(6), 929–948.

Budge, I, Klingemann, H-D., Volkens, A., Bara, J., & Tannenbaum, E. (2001). Mapping policy preferences: Estimates for parties, electors, and governments 1945–1998. Oxford, UK: Oxford University Press.

Callaghan, K., & Schnell, F. (2005). Introduction: Framing political issues in American politics. In K. Callaghan & F. Schnell (Eds.), Framing American politics (pp. 1–17). Pittsburgh, PA: University of Pittsburgh Press.

Callegaro, M., Baker, R., Bethlehem, J., Göritz, A., Krosnick, J., & Lav-rakas, P. (2014). Online panel research: A data quality perspective. West Sussex, UK: John Wiley & Sons.

Campbell, A., Converse, P., Miller, W., & Stokes, D. (1960). The American voter. Ann Arbor, MI: University of Michigan Press.

Cappella, J., & Jamieson, K. (1996). News frames, political cynicism, and media cynicism. The Annals of the American Academy of Political and Social Science, 546, 71–84.

Carpini, D., & Keeter, S. (1996). What Americans know about politics and why it matters. New Haven, CT: Yale University Press.

Chong, D., & Druckman, J. (2007a). Framing theory. Annual Review of Political Science, 10, 103–126.

Chong, D., & Druckman, J. (2007b). A Theory of framing and opinion for-mation in competitive elite environments. Journal of Communication, 57(1), 99–118.

Chong, D., & Druckman, J. (2007c). Framing public opinion in competitive democracies. The American Political Science Review, 101(4), 637–655. doi:10.2307/27644476

Chong, D., & Druckman, J. (2013). Counterframing effects. The Journal of Politics, 75(01), 1–16.

Clifford, S., & Jerit, J. (2013). How words do the work of politics: Moral foundations theory and the debate over stem cell research. The Journal of Politics, 75(03), 659–671.

42

Clifford, S., Jerit, J., Rainey, C., & Motyl, M. (2015). Moral concerns and policy attitudes: Investigating the influence of elite rhetoric. Political Communication, 32(2), 229–248.

Converse, P. (1964). The nature of belief systems in mass publics. In D. Apter (Ed.), Ideology and discontent (pp. 206–261). New York, NY: Free Press.

Dahl, R. A. (1971). Polyarchy: Participation and opposition. New Haven, CT: Yale

University Press. Dalton, R., & Wattenberg, M. (2002). Parties without partisans: Political

change in advanced industrial democracies. Oxford, NY: Oxford Uni-versity Press.

Dassonneville, R. (2012). Electoral volatility, political sophistication, trust and efficacy: A study on changes in voter preferences during the Belgian regional elections of 2009. Acta Politica, 47(1), 18–41.

De Leeuw, D. (2005). To mix or not to mix data collection modes in sur-veys. Journal of official statistics, 21(2), 233–255.

De Vreese, C., & Boomgaarden, H. (2006). Media message flows and in-terpersonal communication: The conditional nature of effects on public opinion. Communication Research, 33(1), 19–37.

De Vreese, C. (2005). News framing: Theory and typology. Information Design Journal, 13(1), 51–62.

Dillman, D., Phelps, G., Tortora, R., Swift, K., Kohrell, J., Berck, J., & Messer, B.

(2009). Response rate and measurement differences in mixed-mode sur-veys using mail, telephone, interactive voice response (IVR) and the In-ternet. Social science research, 38(1), 1–18.

Domke, D., Shah, D., & Wackman, D. (1998). Moral referendums: Values, news media, and the process of candidate choice. Political Communica-tion, 15(3), 301–321.

Domke, D., McCoy, K., & Torres, M. (1999). News media, racial percep-tions, and political cognition. Communication Research, 26(5), 570–607.

Downs, A. (1957). An economic theory of democracy. New York, NY: Harper and Row.

Druckman, J. (2001a). The implications of framing effects for citizen com-petence. Political Behavior, 23(3), 225–256.

Druckman, J. (2001b). On the limits of framing effects: Who can frame? Journal of Politics, 63(4), 1041–1066.

Druckman, J. (2011). What’s it all about? Framing in political science. In G. Keren (Ed.), Perspectives on framing (pp. 279–302). New York, NY: Psychology press.

43

Druckman, J., Green, D., Kuklinski, J., & Lupia, A. (2006). The growth and development of experimental research in political science. American Po-litical Science Review, 100(04), 627–635.

Eagly, A., & Chaiken, S. (1993). The psychology of attitudes. Orlando, FL: Harcourt Brace Jovanovich College Publishers.

Elinder, M., Jordahl, H., & Poutvaara, P. (2015). Promises, policies and pocketbook voting. European Economic Review, 75(April), 177–194.

Entman, R. (1993). Framing: Toward clarification of a fractured paradigm. Journal of Communication, 43(4), 51–58.

Esaiasson, P., & Holmberg, S. (1996). Representation from above: Mem-bers of parliament and representative democracy in Sweden. Aldershot, UK: Dartmouth Publishing Co.

Ezrow, L. (2005). Are moderate parties rewarded in multiparty systems? A pooled analysis of Western European elections, 1984–1998. European Journal of Political Research, 44(6), 881–898.

Feldman, S. (1982). Economic self-interest and political behavior. Ameri-can Journal of Political Science, 26(3), 446–466.

Feldman, S. (1988). Structure and consistency in public opinion: The role of core beliefs and values. American Journal of Political Science, 32(2), 416–440.

Feldman, S. (1989). Reliability and stability of policy positions: Evidence from a five-wave panel. Political Analysis, 1, 25–60.

Feldman, S., & Zaller, J. (1992). The political culture of ambivalence: Ide-ological responses to the welfare state. American Journal of Political Science, 36(1), 268–307.

Feldman, S., & Steenbergen, M. (2001). The humanitarian foundation of public support for social welfare. American Journal of Political Science, 45(3), 658–677.

Festinger, L. (1957). A theory of cognitive dissonance. Stanford, CA: Stan-ford University Press.

Finkel, S. (1987). The effects of participation on political efficacy and po-litical support: Evidence from a West German panel. The Journal of Pol-itics, 49(2), 441–464.

Fleming, M., & Petty, R. (2000). Identity and persuasion: An elaboration likelihood approach. In D. Terry & M. Hogg (Eds.), Attitudes, behavior, and social context: The role of norms and group membership (pp. 171–199). Mahwah, NJ: Lawrence Erlbaum Associates Publishers.

Freire, A. (2006). Bringing social identities back in: The social anchors of left-right orientation in Western Europe. International Political Science Review, 27(4), 359–378.

Freire, A. (2008). Party Polarization and Citizens’ Left—Right Orienta-tions. Party Politics, 14(2), 189–209.

Geer, J. (1991). Do open-ended questions measure “salient” issues? Public Opinion Quarterly, 55(3), 360–370.

42

Clifford, S., Jerit, J., Rainey, C., & Motyl, M. (2015). Moral concerns and policy attitudes: Investigating the influence of elite rhetoric. Political Communication, 32(2), 229–248.

Converse, P. (1964). The nature of belief systems in mass publics. In D. Apter (Ed.), Ideology and discontent (pp. 206–261). New York, NY: Free Press.

Dahl, R. A. (1971). Polyarchy: Participation and opposition. New Haven, CT: Yale

University Press. Dalton, R., & Wattenberg, M. (2002). Parties without partisans: Political

change in advanced industrial democracies. Oxford, NY: Oxford Uni-versity Press.

Dassonneville, R. (2012). Electoral volatility, political sophistication, trust and efficacy: A study on changes in voter preferences during the Belgian regional elections of 2009. Acta Politica, 47(1), 18–41.

De Leeuw, D. (2005). To mix or not to mix data collection modes in sur-veys. Journal of official statistics, 21(2), 233–255.

De Vreese, C., & Boomgaarden, H. (2006). Media message flows and in-terpersonal communication: The conditional nature of effects on public opinion. Communication Research, 33(1), 19–37.

De Vreese, C. (2005). News framing: Theory and typology. Information Design Journal, 13(1), 51–62.

Dillman, D., Phelps, G., Tortora, R., Swift, K., Kohrell, J., Berck, J., & Messer, B.

(2009). Response rate and measurement differences in mixed-mode sur-veys using mail, telephone, interactive voice response (IVR) and the In-ternet. Social science research, 38(1), 1–18.

Domke, D., Shah, D., & Wackman, D. (1998). Moral referendums: Values, news media, and the process of candidate choice. Political Communica-tion, 15(3), 301–321.

Domke, D., McCoy, K., & Torres, M. (1999). News media, racial percep-tions, and political cognition. Communication Research, 26(5), 570–607.

Downs, A. (1957). An economic theory of democracy. New York, NY: Harper and Row.

Druckman, J. (2001a). The implications of framing effects for citizen com-petence. Political Behavior, 23(3), 225–256.

Druckman, J. (2001b). On the limits of framing effects: Who can frame? Journal of Politics, 63(4), 1041–1066.

Druckman, J. (2011). What’s it all about? Framing in political science. In G. Keren (Ed.), Perspectives on framing (pp. 279–302). New York, NY: Psychology press.

43

Druckman, J., Green, D., Kuklinski, J., & Lupia, A. (2006). The growth and development of experimental research in political science. American Po-litical Science Review, 100(04), 627–635.

Eagly, A., & Chaiken, S. (1993). The psychology of attitudes. Orlando, FL: Harcourt Brace Jovanovich College Publishers.

Elinder, M., Jordahl, H., & Poutvaara, P. (2015). Promises, policies and pocketbook voting. European Economic Review, 75(April), 177–194.

Entman, R. (1993). Framing: Toward clarification of a fractured paradigm. Journal of Communication, 43(4), 51–58.

Esaiasson, P., & Holmberg, S. (1996). Representation from above: Mem-bers of parliament and representative democracy in Sweden. Aldershot, UK: Dartmouth Publishing Co.

Ezrow, L. (2005). Are moderate parties rewarded in multiparty systems? A pooled analysis of Western European elections, 1984–1998. European Journal of Political Research, 44(6), 881–898.

Feldman, S. (1982). Economic self-interest and political behavior. Ameri-can Journal of Political Science, 26(3), 446–466.

Feldman, S. (1988). Structure and consistency in public opinion: The role of core beliefs and values. American Journal of Political Science, 32(2), 416–440.

Feldman, S. (1989). Reliability and stability of policy positions: Evidence from a five-wave panel. Political Analysis, 1, 25–60.

Feldman, S., & Zaller, J. (1992). The political culture of ambivalence: Ide-ological responses to the welfare state. American Journal of Political Science, 36(1), 268–307.

Feldman, S., & Steenbergen, M. (2001). The humanitarian foundation of public support for social welfare. American Journal of Political Science, 45(3), 658–677.

Festinger, L. (1957). A theory of cognitive dissonance. Stanford, CA: Stan-ford University Press.

Finkel, S. (1987). The effects of participation on political efficacy and po-litical support: Evidence from a West German panel. The Journal of Pol-itics, 49(2), 441–464.

Fleming, M., & Petty, R. (2000). Identity and persuasion: An elaboration likelihood approach. In D. Terry & M. Hogg (Eds.), Attitudes, behavior, and social context: The role of norms and group membership (pp. 171–199). Mahwah, NJ: Lawrence Erlbaum Associates Publishers.

Freire, A. (2006). Bringing social identities back in: The social anchors of left-right orientation in Western Europe. International Political Science Review, 27(4), 359–378.

Freire, A. (2008). Party Polarization and Citizens’ Left—Right Orienta-tions. Party Politics, 14(2), 189–209.

Geer, J. (1991). Do open-ended questions measure “salient” issues? Public Opinion Quarterly, 55(3), 360–370.

44

Giddens, A. (1994). Beyond left and right: The future of radical politics. Stanford, CA: Stanford University Press.

Glynn, C., Herbst, S., O’Keefe, G., Shapiro, R., & Lindeman, M. (2004). Public opinion (2nd edition). Boulder, CO: Westview Press.

Goren, P. (2004). Political sophistication and policy reasoning: A reconsid-eration. American Journal of Political Science, 48(3), 462–478.

Green-Pedersen, C. (2006). Long-term Changes in Danish Party Politics: The Rise and Importance of Issue Competition. Scandinavian Political Studies, 29(3), 219–235.

Green-Pedersen, C. (2007). The Growing Importance of Issue Competition: The Changing Nature of Party Competition in Western Europe. Political Studies, 55(3), 607–628.

Gross, K. (2008). Framing Persuasive Appeals: Episodic and Thematic Framing, Emotional Response, and Policy Opinion. Political Psychol-ogy, 29(2), 169–192.

Gross, K., & D’ambrosio, L. (2004). Framing emotional response. Political Psychology, 25(1), 1–29.

Gunther, R., & Montero, J. (2001). The anchors of partisanship: A compar-ative analysis of voting behaviour in four Southern European countries’. In N. Diamandouros, and R. Gunther (Eds.), Parties, politics, and de-mocracy in new Southern Europe (pp. 83–152). Baltimore, MD: Johns Hopkins University Press.

Hayes, A. F. (2013). Introduction to mediation, moderation, and conditional process analysis: A regression-based approach. New York, NY: Guil-ford Press.

Hare, R. (1952). The language of morals. Oxford, UK: Oxford University Press.

Higgins, E. (1996). Knowledge activation: Accessibility, applicability, and salience. In E. Higgins & A. Kruglanski (Eds.), Social psychology: Handbook of basic principles (pp. 133–168). New York, NY: Guilford Press.

Holmberg, S., & Oscarsson, H. (2013). Nya svenska väljare [New Swedish voters]. Stockholm, SE: Norstedts juridik.

Hooghe, M. (2004). Political socialization and the future of politics. Acta Politica, 39(4), 331–341.

Håkansson, N., & Naurin, E. (2016). Promising ever more: An empirical account of Swedish parties’ pledge making during 20 years. Party Poli-tics, 22(3), 393–404.

Igartua, J-J., & Cheng, L. (2009). Moderating effect of group cue while processing news on immigration: Is the framing effect a heuristic pro-cess? Journal of Communication, 59(4), 726–749.

Inglehart, R. (2000). Globalization and postmodern values. The Washing-ton Quarterly, 23(1), 215–228.

45

Iyengar, S. (1990). The accessibility bias in politics: Television news and public opinion. International Journal of Public Opinion Research, 2(1), 1–15.

Iyengar, S. (1991). Is anyone responsible? How television frames political issues. Chicago, IL: University of Chicago Press.

Jackman, S., & Sniderman, P. (2001). The institutional organization of choice spaces: A political conception of political psychology. In K. Ren-wick Monroe (Ed.), Political Psychology (pp. 209–224). Malwah, NJ: Lawrence Erlbaum.

Jackson, K., & Trochim, W. (2002). Concept mapping as an alternative ap-proach for the analysis of open-ended survey responses. Organizational Research Methods, 5(4), 307–336.

Jacobs, L., & Shapiro, R. (2000). Politicians don’t pander: Political manip-ulation and the loss of democratic responsiveness. Chicago, IL: Univer-sity of Chicago Press.

Jacoby, W. (1988). The impact of party identification on issue attitudes. American Journal of Political Science, 32(3), 643–661.

Jacoby, W. (2006). Value choices and American public opinion. American Journal of Political Science, 50(3), 706–723.

Jerit, J. (2004). Survival of the fittest: Rhetoric during the course of an elec-tion campaign. Political Psychology, 25(4), 563–575.

Johnson, G., & Ryu, S-R. (2010). Repudiating or rewarding neoliberalism? How broken campaign promises condition economic voting in Latin America. Latin American Politics and Society, 52(4), 1–24.

Kahneman, D., Tversky, A., & Slovic, P. (1982). Judgment under uncer-tainty: Heuristics & biases. Cambridge, UK: Cambridge University Press.

Kam, C. (2005). Who toes the party line? Cues, values, and individual dif-ferences. Political Behavior, 27(2), 163–182. doi:10.1007/s11109-005-1764-y

Katz, I., & Hass, R. (1988). Racial ambivalence and American value con-flict: Correlational and priming studies of dual cognitive structures. Journal of personality and social psychology, 55(6), 893–905.

Kelle, U., & Laurie, H. (1998). Computer use in qualitative research and issues of validity. In U. Kelle (Ed.), Computer aided qualitative data analysis (pp. 19–28). Thousand Oaks, CA: SAGE Publications.

Keren, G. (2011). Perspectives on framing. New York, NY: Psychology press.

Kinder, D., & Sanders, L. (1990). Mimicking political debate with survey questions: The case of white opinion on affirmative action for blacks. Social Cognition, 8(1), 73–103.

Kinder, D., & Sanders, L. (1996). Divided by color: Racial politics and democratic ideals. Chicago, IL: University of Chicago Press.

44

Giddens, A. (1994). Beyond left and right: The future of radical politics. Stanford, CA: Stanford University Press.

Glynn, C., Herbst, S., O’Keefe, G., Shapiro, R., & Lindeman, M. (2004). Public opinion (2nd edition). Boulder, CO: Westview Press.

Goren, P. (2004). Political sophistication and policy reasoning: A reconsid-eration. American Journal of Political Science, 48(3), 462–478.

Green-Pedersen, C. (2006). Long-term Changes in Danish Party Politics: The Rise and Importance of Issue Competition. Scandinavian Political Studies, 29(3), 219–235.

Green-Pedersen, C. (2007). The Growing Importance of Issue Competition: The Changing Nature of Party Competition in Western Europe. Political Studies, 55(3), 607–628.

Gross, K. (2008). Framing Persuasive Appeals: Episodic and Thematic Framing, Emotional Response, and Policy Opinion. Political Psychol-ogy, 29(2), 169–192.

Gross, K., & D’ambrosio, L. (2004). Framing emotional response. Political Psychology, 25(1), 1–29.

Gunther, R., & Montero, J. (2001). The anchors of partisanship: A compar-ative analysis of voting behaviour in four Southern European countries’. In N. Diamandouros, and R. Gunther (Eds.), Parties, politics, and de-mocracy in new Southern Europe (pp. 83–152). Baltimore, MD: Johns Hopkins University Press.

Hayes, A. F. (2013). Introduction to mediation, moderation, and conditional process analysis: A regression-based approach. New York, NY: Guil-ford Press.

Hare, R. (1952). The language of morals. Oxford, UK: Oxford University Press.

Higgins, E. (1996). Knowledge activation: Accessibility, applicability, and salience. In E. Higgins & A. Kruglanski (Eds.), Social psychology: Handbook of basic principles (pp. 133–168). New York, NY: Guilford Press.

Holmberg, S., & Oscarsson, H. (2013). Nya svenska väljare [New Swedish voters]. Stockholm, SE: Norstedts juridik.

Hooghe, M. (2004). Political socialization and the future of politics. Acta Politica, 39(4), 331–341.

Håkansson, N., & Naurin, E. (2016). Promising ever more: An empirical account of Swedish parties’ pledge making during 20 years. Party Poli-tics, 22(3), 393–404.

Igartua, J-J., & Cheng, L. (2009). Moderating effect of group cue while processing news on immigration: Is the framing effect a heuristic pro-cess? Journal of Communication, 59(4), 726–749.

Inglehart, R. (2000). Globalization and postmodern values. The Washing-ton Quarterly, 23(1), 215–228.

45

Iyengar, S. (1990). The accessibility bias in politics: Television news and public opinion. International Journal of Public Opinion Research, 2(1), 1–15.

Iyengar, S. (1991). Is anyone responsible? How television frames political issues. Chicago, IL: University of Chicago Press.

Jackman, S., & Sniderman, P. (2001). The institutional organization of choice spaces: A political conception of political psychology. In K. Ren-wick Monroe (Ed.), Political Psychology (pp. 209–224). Malwah, NJ: Lawrence Erlbaum.

Jackson, K., & Trochim, W. (2002). Concept mapping as an alternative ap-proach for the analysis of open-ended survey responses. Organizational Research Methods, 5(4), 307–336.

Jacobs, L., & Shapiro, R. (2000). Politicians don’t pander: Political manip-ulation and the loss of democratic responsiveness. Chicago, IL: Univer-sity of Chicago Press.

Jacoby, W. (1988). The impact of party identification on issue attitudes. American Journal of Political Science, 32(3), 643–661.

Jacoby, W. (2006). Value choices and American public opinion. American Journal of Political Science, 50(3), 706–723.

Jerit, J. (2004). Survival of the fittest: Rhetoric during the course of an elec-tion campaign. Political Psychology, 25(4), 563–575.

Johnson, G., & Ryu, S-R. (2010). Repudiating or rewarding neoliberalism? How broken campaign promises condition economic voting in Latin America. Latin American Politics and Society, 52(4), 1–24.

Kahneman, D., Tversky, A., & Slovic, P. (1982). Judgment under uncer-tainty: Heuristics & biases. Cambridge, UK: Cambridge University Press.

Kam, C. (2005). Who toes the party line? Cues, values, and individual dif-ferences. Political Behavior, 27(2), 163–182. doi:10.1007/s11109-005-1764-y

Katz, I., & Hass, R. (1988). Racial ambivalence and American value con-flict: Correlational and priming studies of dual cognitive structures. Journal of personality and social psychology, 55(6), 893–905.

Kelle, U., & Laurie, H. (1998). Computer use in qualitative research and issues of validity. In U. Kelle (Ed.), Computer aided qualitative data analysis (pp. 19–28). Thousand Oaks, CA: SAGE Publications.

Keren, G. (2011). Perspectives on framing. New York, NY: Psychology press.

Kinder, D., & Sanders, L. (1990). Mimicking political debate with survey questions: The case of white opinion on affirmative action for blacks. Social Cognition, 8(1), 73–103.

Kinder, D., & Sanders, L. (1996). Divided by color: Racial politics and democratic ideals. Chicago, IL: University of Chicago Press.

46

Kinder, D. (1998). Communication and opinion. Annual Review of Political Science, 1(1), 167–197.

Klingemann, H-D., Hofferbert, R., & Budge, I. (1994). Parties, policies and democracy. Boulder, CO: Westview Press.

Krosnick, J., & Kinder, D. (1990). Altering the foundations of support for the president through priming. American Political Science Review, 84(2), 497–512.

Kunda, Z. (1990). The case for motivated reasoning. Psychological Bulle-tin, 108(3), 480–498.

Lakoff, G. (2002). Moral politics: How liberals and conservatives think. Chicago, IL: University of Chicago Press.

Lecheler, S., Bos, L, & Vliegenthart, R. (2015). The mediating role of emo-tions: News framing effects on opinions about immigration. Journalism & Mass Communication Quarterly, 92(4), 812–838.

Lecheler, S., & de Vreese, C. (2013). What a difference a day makes? The effects of repetitive and competitive news framing over time. Commu-nication Research, 40(2), 147–175.

Lecheler, S., Keer, M., Schuck, A, & Hänggli, R. (2015). The effects of repetitive news framing on political opinions over time. Communication Monographs, 82(3), 339–358.

Leeper, T., & Slothuus, R. (2015). Can citizens be framed? How infor-mation, not emphasis, changes opinions. Paper presented at the Work-shop Lore experiment, University of Gothenburg, Gothenburg, Sweden. http://lore.gu.se/digitalAssets/1525/1525943_leeper-and-slothuus---can-citizens-be-framed.pdf

Lodge, M., & Taber, C. (2005). The automaticity of affect for political lead-ers, groups, and issues: An experimental test of the hot cognition hy-pothesis. Political Psychology, 26(3), 455–482.

Lodge, M., Steenbergen, M., & Brau, S. (1995). The responsive voter: Campaign information and the dynamics of candidate evaluation. The American Political Science Review, 89(2), 309–326.

Lord, C., Ross, L., & Lepper, M. (1979). Biased assimilation and attitude polarization: The effects of prior theories on subsequently considered evidence. Journal of personality and social psychology, 37(11), 2098–2109.

Luskin, R. (1990). Explaining political sophistication. Political Behavior, 12(4), 331–361.

Luskin, R. (2002). From denial to extenuation (and finally beyond): Politi-cal sophistication and citizen performance. In J. Kuklinski (Ed.), Think-ing about political psychology (p. 305). New York, NY: Cambridge Uni-versity Press.

Macagno, F., & Walton, D. (2010). What we hide in words: Emotive words and persuasive definitions. Journal of Pragmatics, 42(7), 1997–2013.

47

Macagno, F. (2014). Manipulating emotions: value-based reasoning and emotive language. Argumentation and Advocacy, 51(2), 103–123.

Mair, P. (2008). Electoral volatility and the Dutch party system: A compar-ative perspective. Acta Politica, 43(2), 235–253.

Mair, P. (2002). In the aggregate: Mass electoral behaviour in Western Eu-rope, 1950–2000. In H. Keman (Ed.), Comparative democratic politics: A guide to contemporary theory and research (pp. 122–141). London, UK: SAGE Publications.

Mair, P., & Biezen, I. (2001). Party membership in twenty European de-mocracies, 1980–2000. Party Politics, 7(1), 5–21.

Malhotra, N., & Margalit, Y. (2014). Expectation setting and retrospective voting. The Journal of Politics, 76(04), 1000–1016.

Manin, B., Przeworski, A., & Stokes, S. (1999). Elections and representa-tion. In A. Przeworski, S. Stokes, & B. Manin (Eds.), Democracy, ac-countability, and representation (pp. 29–54). Cambridge, UK: Cam-bridge University Press.

Mansbridge, J. (2003). Rethinking representation. American Political Sci-ence Review, 97(04), 515–528.

Mansergh, L., & Thomson, R. (2007). Election pledges, party competition, and policymaking. Comparative Politics, 39(3), 311–329.

McCloskey, H., & Zaller, J. (1984). The American ethos: Public attitudes toward capitalism and democracy. Cambridge, MA: Harvard University Press.

McDermott, R. (2002). Experimental methods in political science. Annual Review of Political Science, 5(1), 31–61.

Mondak, J. (1993). Source cues and policy approval: The cognitive dynam-ics of public support for the Reagan agenda. American Journal of Polit-ical Science, 186–212.

Morris, J., Squires, N., Taber, C., & Lodge, M. (2003). Activation of polit-ical attitudes: A psychophysiological examination of the hot cognition hypothesis. Political Psychology, 24(4), 727–745.

Morton, R., & Williams, K. (2010). Experimental political science and the study of causality: From nature to the lab. New York, NY: Cambridge University Press.

Naurin, E. (2011). Election promises, party behaviour and voter percep-tions. Basingstoke, UK: Palgrave Macmillan.

Naurin, E. (2014). Is a promise a promise? Election pledge fulfilment in comparative perspective using Sweden as an example. West European Politics, 37(5), 1046–1064.

Naurin, E., & Oscarsson, H. (2017). When and why are voters correct in their evaluations of specific government performance? Political Studies, n/a-n/a.

Nelson, T., Oxley, Z., & Clawson, R. (1997). Toward a psychology of fram-ing effects. Political Behavior, 19(3), 221–246.

46

Kinder, D. (1998). Communication and opinion. Annual Review of Political Science, 1(1), 167–197.

Klingemann, H-D., Hofferbert, R., & Budge, I. (1994). Parties, policies and democracy. Boulder, CO: Westview Press.

Krosnick, J., & Kinder, D. (1990). Altering the foundations of support for the president through priming. American Political Science Review, 84(2), 497–512.

Kunda, Z. (1990). The case for motivated reasoning. Psychological Bulle-tin, 108(3), 480–498.

Lakoff, G. (2002). Moral politics: How liberals and conservatives think. Chicago, IL: University of Chicago Press.

Lecheler, S., Bos, L, & Vliegenthart, R. (2015). The mediating role of emo-tions: News framing effects on opinions about immigration. Journalism & Mass Communication Quarterly, 92(4), 812–838.

Lecheler, S., & de Vreese, C. (2013). What a difference a day makes? The effects of repetitive and competitive news framing over time. Commu-nication Research, 40(2), 147–175.

Lecheler, S., Keer, M., Schuck, A, & Hänggli, R. (2015). The effects of repetitive news framing on political opinions over time. Communication Monographs, 82(3), 339–358.

Leeper, T., & Slothuus, R. (2015). Can citizens be framed? How infor-mation, not emphasis, changes opinions. Paper presented at the Work-shop Lore experiment, University of Gothenburg, Gothenburg, Sweden. http://lore.gu.se/digitalAssets/1525/1525943_leeper-and-slothuus---can-citizens-be-framed.pdf

Lodge, M., & Taber, C. (2005). The automaticity of affect for political lead-ers, groups, and issues: An experimental test of the hot cognition hy-pothesis. Political Psychology, 26(3), 455–482.

Lodge, M., Steenbergen, M., & Brau, S. (1995). The responsive voter: Campaign information and the dynamics of candidate evaluation. The American Political Science Review, 89(2), 309–326.

Lord, C., Ross, L., & Lepper, M. (1979). Biased assimilation and attitude polarization: The effects of prior theories on subsequently considered evidence. Journal of personality and social psychology, 37(11), 2098–2109.

Luskin, R. (1990). Explaining political sophistication. Political Behavior, 12(4), 331–361.

Luskin, R. (2002). From denial to extenuation (and finally beyond): Politi-cal sophistication and citizen performance. In J. Kuklinski (Ed.), Think-ing about political psychology (p. 305). New York, NY: Cambridge Uni-versity Press.

Macagno, F., & Walton, D. (2010). What we hide in words: Emotive words and persuasive definitions. Journal of Pragmatics, 42(7), 1997–2013.

47

Macagno, F. (2014). Manipulating emotions: value-based reasoning and emotive language. Argumentation and Advocacy, 51(2), 103–123.

Mair, P. (2008). Electoral volatility and the Dutch party system: A compar-ative perspective. Acta Politica, 43(2), 235–253.

Mair, P. (2002). In the aggregate: Mass electoral behaviour in Western Eu-rope, 1950–2000. In H. Keman (Ed.), Comparative democratic politics: A guide to contemporary theory and research (pp. 122–141). London, UK: SAGE Publications.

Mair, P., & Biezen, I. (2001). Party membership in twenty European de-mocracies, 1980–2000. Party Politics, 7(1), 5–21.

Malhotra, N., & Margalit, Y. (2014). Expectation setting and retrospective voting. The Journal of Politics, 76(04), 1000–1016.

Manin, B., Przeworski, A., & Stokes, S. (1999). Elections and representa-tion. In A. Przeworski, S. Stokes, & B. Manin (Eds.), Democracy, ac-countability, and representation (pp. 29–54). Cambridge, UK: Cam-bridge University Press.

Mansbridge, J. (2003). Rethinking representation. American Political Sci-ence Review, 97(04), 515–528.

Mansergh, L., & Thomson, R. (2007). Election pledges, party competition, and policymaking. Comparative Politics, 39(3), 311–329.

McCloskey, H., & Zaller, J. (1984). The American ethos: Public attitudes toward capitalism and democracy. Cambridge, MA: Harvard University Press.

McDermott, R. (2002). Experimental methods in political science. Annual Review of Political Science, 5(1), 31–61.

Mondak, J. (1993). Source cues and policy approval: The cognitive dynam-ics of public support for the Reagan agenda. American Journal of Polit-ical Science, 186–212.

Morris, J., Squires, N., Taber, C., & Lodge, M. (2003). Activation of polit-ical attitudes: A psychophysiological examination of the hot cognition hypothesis. Political Psychology, 24(4), 727–745.

Morton, R., & Williams, K. (2010). Experimental political science and the study of causality: From nature to the lab. New York, NY: Cambridge University Press.

Naurin, E. (2011). Election promises, party behaviour and voter percep-tions. Basingstoke, UK: Palgrave Macmillan.

Naurin, E. (2014). Is a promise a promise? Election pledge fulfilment in comparative perspective using Sweden as an example. West European Politics, 37(5), 1046–1064.

Naurin, E., & Oscarsson, H. (2017). When and why are voters correct in their evaluations of specific government performance? Political Studies, n/a-n/a.

Nelson, T., Oxley, Z., & Clawson, R. (1997). Toward a psychology of fram-ing effects. Political Behavior, 19(3), 221–246.

48

Nelson, T., Clawson, R., & Oxley, Z. (1997). Media framing of a civil lib-erties conflict and its effect on tolerance. The American Political Science Review, 91(3), 567–583.

Nelson, T., & Oxley, Z. (1999). Issue framing effects on belief importance and opinion. The Journal of Politics, 61(4), 1040–1067.

Nelson, T., & Willey, E. (2001). Issue frames that strike a value balance: A political psychology perspective. In S. Reese, O. Gandy Jr, & A. Grant (Eds.), Framing public life: Perspectives on media and our understand-ing of the social world (pp. 245–266). Mahwah, NJ: Lawrence Erlbaum.

Nelson, T., & Garst, J. (2005). Values-based political messages and persua-sion: Relationships among speaker, recipient, and evoked values. Polit-ical Psychology, 26(4), 489–515.

Norris, P., Curtis, J., Sanders, D., Scammell, M., & Semetko, H. (1999). On message: Communicating the campaign. London, UK: SAGE Publica-tions.

Orwell, G. (1946). Politics and the English language. London, UK: Pen-guin.

Orwell, G. (1949). Nineteen eighty-four: A novel. London, UK: Secker & Warburg.

Oscarsson, H. (2013). Väljarnas valrörelser. In J. Strömbäck & Nord, L (Eds.), Kampen om opinionen. Politisk kommunikation under svenska valrörelser. Stockholm, SE: SNS Förlag.

Page, B., & Shapiro, R. (1992). The rational public: Fifty years of trends in Americans’ policy preferences. Chicago, IL: University of Chicago Press.

Parekh, B. (1992). The Cultural Particularity of Liberal Democracy. Polit-ical Studies, 40, 160–175.

Patterson, T. (1993). Out of order: An incisive and boldly original critique of the news media's domination of America’s political process. New York, NY: Alfred A. Knopf, Inc.

Perloff, R. (2010). The dynamics of persuasion: Communication and atti-tudes in the twenty-first century. New York, NY: Routledge.

Petty, R., & Cacioppo, J. (1986). Communication and persuasion: Central and peripheral routes to attitude change. New York, NY: Springer-Ver-lag.

Petty, R., & Cacioppo, J. (1990). Involvement and persuasion: Tradition versus integration. Psychological Bulletin, 107(3), 367–374.

Pierce, R. (1999). Mass-elite issue linkages and the responsible party model of representation. In W. Miller, R. Pierce, J. Thomassen, R. Herrera, S. Holmberg, P. Esaiasson, B. Wessels (Eds.), Policy representation in Western democracies (pp. 9–32). Oxford, UK: Oxford University Press.

Popkin, S. (1991). The Reasoning voter: Communication and persuasion in presidential campaigns. Chicago, IL: The University of Chicago Press.

49

Price, V., & Tewksbury, D. (1997). News values and public opinion: A theoretical account of media priming and framing. In G. Barett & F. Boster (Eds.), Progress in communication sciences: Advances in per-suasion (pp. 173–212). Greenwich, CT: Ablex Publishing.

Przeworski, A. (2009). Self-government in our times. Annual Review of Political Science, 12, 71–92.

Putnam, R. (1966). Political attitudes and the local community. American Political Science Review, 60(3), 640–654.

Rea, L., & Parker, R. (2014). Designing and conducting survey research: A comprehensive guide. San Fransisco, CA: John Wiley & Sons.

Redlawsk, D. (2002). Hot cognition or cool consideration? Testing the ef-fects of motivated reasoning on political decision making. The Journal of Politics, 64(04), 1021–1044.

Reid, A., & Deaux, K. (1996). Relationship between social and personal identities: Segregation or integration. Journal of personality and social psychology, 71(6), 1084–1091.

Riker, W. (1986). The art of political manipulation. New Haven, CT: Yale University Press.

Rokeach, M. (1973). The nature of human values. New York, NY: Free Press.

Royed, T. (1996). Testing the mandate model in Britain and the United States: Evidence from the Reagan and Thatcher eras. British Journal of Political Science, 26(1), 45–80.

Saris, W., & Andrews, F (2004). Evaluation of measurement instruments using a structural modeling approach. In R. Biemer, L. Lyberg, N. Mathiowetz, & S. Sudman (Eds.), Measurement errors in surveys (pp. 575–597). Hoboken, NJ: John Wiley & Sons.

Schemer, C., Wirth, W., & Matthes, J. (2012). Value resonance and value framing effects on voting intentions in direct-democratic campaigns. American Behavioral Scientist, 56(3), 334–352.

Scheufele, D., & Tewksbury, D. (2007). Framing, agenda setting, and prim-ing: The evolution of three media effects models. Journal of Communi-cation, 57(1), 9–20.

Scheufele, D., & Iyengar, S. (2012). The state of framing research: A call for new directions. In K. Kenski & K. Hall Jamieson (Eds.), The Oxford handbook of political communication theories (pp. 1–26). New York, NY: Oxford University Press.

Schiappa, E. (2003). Defining reality: Definitions and the politics of mean-ing. Carbondale, IL: Southern Illinois University Press.

Schwarz, N., Bless, H., Wänke, M., & Winkielman, P. (2003). Accessibility revisited. In G. Bodenhausen & A. Lambert (Eds.), Foundations of so-cial cognition: A Festschrift in honor of Robert S. Wyer, Jr (pp. 51–77). Mahwah, NJ: Lawrence Erlbaum.

48

Nelson, T., Clawson, R., & Oxley, Z. (1997). Media framing of a civil lib-erties conflict and its effect on tolerance. The American Political Science Review, 91(3), 567–583.

Nelson, T., & Oxley, Z. (1999). Issue framing effects on belief importance and opinion. The Journal of Politics, 61(4), 1040–1067.

Nelson, T., & Willey, E. (2001). Issue frames that strike a value balance: A political psychology perspective. In S. Reese, O. Gandy Jr, & A. Grant (Eds.), Framing public life: Perspectives on media and our understand-ing of the social world (pp. 245–266). Mahwah, NJ: Lawrence Erlbaum.

Nelson, T., & Garst, J. (2005). Values-based political messages and persua-sion: Relationships among speaker, recipient, and evoked values. Polit-ical Psychology, 26(4), 489–515.

Norris, P., Curtis, J., Sanders, D., Scammell, M., & Semetko, H. (1999). On message: Communicating the campaign. London, UK: SAGE Publica-tions.

Orwell, G. (1946). Politics and the English language. London, UK: Pen-guin.

Orwell, G. (1949). Nineteen eighty-four: A novel. London, UK: Secker & Warburg.

Oscarsson, H. (2013). Väljarnas valrörelser. In J. Strömbäck & Nord, L (Eds.), Kampen om opinionen. Politisk kommunikation under svenska valrörelser. Stockholm, SE: SNS Förlag.

Page, B., & Shapiro, R. (1992). The rational public: Fifty years of trends in Americans’ policy preferences. Chicago, IL: University of Chicago Press.

Parekh, B. (1992). The Cultural Particularity of Liberal Democracy. Polit-ical Studies, 40, 160–175.

Patterson, T. (1993). Out of order: An incisive and boldly original critique of the news media's domination of America’s political process. New York, NY: Alfred A. Knopf, Inc.

Perloff, R. (2010). The dynamics of persuasion: Communication and atti-tudes in the twenty-first century. New York, NY: Routledge.

Petty, R., & Cacioppo, J. (1986). Communication and persuasion: Central and peripheral routes to attitude change. New York, NY: Springer-Ver-lag.

Petty, R., & Cacioppo, J. (1990). Involvement and persuasion: Tradition versus integration. Psychological Bulletin, 107(3), 367–374.

Pierce, R. (1999). Mass-elite issue linkages and the responsible party model of representation. In W. Miller, R. Pierce, J. Thomassen, R. Herrera, S. Holmberg, P. Esaiasson, B. Wessels (Eds.), Policy representation in Western democracies (pp. 9–32). Oxford, UK: Oxford University Press.

Popkin, S. (1991). The Reasoning voter: Communication and persuasion in presidential campaigns. Chicago, IL: The University of Chicago Press.

49

Price, V., & Tewksbury, D. (1997). News values and public opinion: A theoretical account of media priming and framing. In G. Barett & F. Boster (Eds.), Progress in communication sciences: Advances in per-suasion (pp. 173–212). Greenwich, CT: Ablex Publishing.

Przeworski, A. (2009). Self-government in our times. Annual Review of Political Science, 12, 71–92.

Putnam, R. (1966). Political attitudes and the local community. American Political Science Review, 60(3), 640–654.

Rea, L., & Parker, R. (2014). Designing and conducting survey research: A comprehensive guide. San Fransisco, CA: John Wiley & Sons.

Redlawsk, D. (2002). Hot cognition or cool consideration? Testing the ef-fects of motivated reasoning on political decision making. The Journal of Politics, 64(04), 1021–1044.

Reid, A., & Deaux, K. (1996). Relationship between social and personal identities: Segregation or integration. Journal of personality and social psychology, 71(6), 1084–1091.

Riker, W. (1986). The art of political manipulation. New Haven, CT: Yale University Press.

Rokeach, M. (1973). The nature of human values. New York, NY: Free Press.

Royed, T. (1996). Testing the mandate model in Britain and the United States: Evidence from the Reagan and Thatcher eras. British Journal of Political Science, 26(1), 45–80.

Saris, W., & Andrews, F (2004). Evaluation of measurement instruments using a structural modeling approach. In R. Biemer, L. Lyberg, N. Mathiowetz, & S. Sudman (Eds.), Measurement errors in surveys (pp. 575–597). Hoboken, NJ: John Wiley & Sons.

Schemer, C., Wirth, W., & Matthes, J. (2012). Value resonance and value framing effects on voting intentions in direct-democratic campaigns. American Behavioral Scientist, 56(3), 334–352.

Scheufele, D., & Tewksbury, D. (2007). Framing, agenda setting, and prim-ing: The evolution of three media effects models. Journal of Communi-cation, 57(1), 9–20.

Scheufele, D., & Iyengar, S. (2012). The state of framing research: A call for new directions. In K. Kenski & K. Hall Jamieson (Eds.), The Oxford handbook of political communication theories (pp. 1–26). New York, NY: Oxford University Press.

Schiappa, E. (2003). Defining reality: Definitions and the politics of mean-ing. Carbondale, IL: Southern Illinois University Press.

Schwarz, N., Bless, H., Wänke, M., & Winkielman, P. (2003). Accessibility revisited. In G. Bodenhausen & A. Lambert (Eds.), Foundations of so-cial cognition: A Festschrift in honor of Robert S. Wyer, Jr (pp. 51–77). Mahwah, NJ: Lawrence Erlbaum.

50

Shah, D., Domke, D., & Wackman, D. (1996). To thine own self be true: Values, framing, and voter decision-making strategies. Communication Research, 23(5), 509–560.

Shapiro, M. (1994). Think-aloud and thought-list procedures in investigat-ing mental processes. In A. Lang (Ed.), Measuring psychological re-sponses to media messages (pp. 1–14). New York, NY: Routledge.

Shapiro, G. (1997). The future of coders: Human judgments in a world of sophisticated software. In C. Roberts (Ed.), Text analysis for the social sciences: Methods for drawing statistical inferences from texts and tran-scripts (pp. 225–238). Mahwah, NJ: Lawrence Erlbaum.

Shen, F. (2004). Effects of news frames and schemas on individuals’ issue interpretations and attitudes. Journalism & Mass Communication Quar-terly, 81(2), 400–416.

Shen, F., & Edwards-Hatfield, H. (2005). Economic individualism, human-itarianism, and welfare reform: A value-based account of framing ef-fects. Journal of Communication, 55(4), 795–809.

Slater, M. (2004). Operationalizing and analyzing exposure: The founda-tion of media effects research. Journalism & Mass Communication Quarterly, 81(1), 168–183.

Slothuus, R. (2008). More than weighting cognitive importance: A dual-process model of issue framing effects. Political Psychology, 29(1), 1–28.

Slothuus, R., & de Vreese, C. (2010). Political parties, motivated reasoning, and issue framing effects. The Journal of Politics, 72(03), 630–645.

Sniderman, P., Brody, R., & Tetlock, P. (1993). Reasoning and choice: Ex-plorations in political psychology. Cambridge, MA: Cambridge Univer-sity Press.

Sniderman, P., & Theriault, S. (2004). The structure of political argument and the logic of issue framing. In W. Saris, & P. Sniderman (Eds.), Stud-ies in public opinion: Attitudes, nonattitudes, measurement error, and change (pp. 133–165). Princeton, NJ: Princeton University Press.

Sproull, N. (1995). Handbook of research methods: A guide for practition-ers and students in the social sciences (2nd edition). Metuchen, NJ: Scare-crow Press.

Squire, P., & Smith, E. (1988). The effect of partisan information on voters in nonpartisan elections. The Journal of Politics, 50(1), 169–179.

Stevenson, C. (1937). The emotive meaning of ethical terms. Mind, 46(181), 14–31.

Stevenson, C. (1944). Ethics and language. New Haven, CT: Yale Univer-sity Press.

Strömbäck, J. (2007). Political marketing and professionalized campaign-ing. Journal of Political Marketing, 6(2-3), 49–67.

51

Strömbäck, J., Mitrook, M., & Kiousis, S. (2010). Bridging two schools of thought: Applications of public relations theory to political marketing. Journal of Political Marketing, 9(1-2), 73–92.

Taber, C., & Lodge, M. (2006). Motivated Skepticism in the evaluation of political beliefs. American Journal of Political Science, 50(3), 755–769.

Thomassen, J. (2005). The European Voter. Oxford, UK: Oxford Univer-sity Press.

Thomson, R. (2001). The programme to policy linkage: The fulfilment of election pledges on socio–economic policy in The Netherlands, 1986–1998. European Journal of Political Research, 40(2), 171–197.

Thomson, R. (2011). Citizens’ evaluations of the fulfillment of election pledges: Evidence from Ireland. The Journal of Politics, 73(01), 187–201.

Thomson, R., Royed, T., Naurin, E., Artés, J., Costello, R., Ennser-Jede-nastik, L., Ferguson, M., Kostadinova, P., Moury, C., Pétry, F., & Prap-rotnik, K. (2017). The fulfillment of parties’ election pledges: A com-parative study on the impact of power sharing. American Journal of Po-litical Science, 61(3), 527–542.

Turner, J., Oakes, P., Haslam, S., & McGarty, C. (1994). Self and collec-tive: Cognition and social context. Personality and Social Psychology Bulletin, 20(5), 454–463.

Tversky, A., & Kahneman, D. (1974). Judgment under uncertainty: Heu-ristics and biases. Science, 185(4157), 1124–1131.

Van der Eijk, C., Franklin, M., & van der Brug, W. (1999). Policy prefer-ences and party choice. In H. Schmitt, & J. Thomassen (Eds.), Political Representation and Legitimacy in the European Union. Oxford, UK: Oxford University Press.

Walton, D. (2001). Persuasive definitions and public policy arguments. Ar-gumentation and Advocacy, 37(3), 117–132.

Walton, D. (2003). Ethical argumentation. Lanham, MD: Lexington Books. Walton, D. (2006). Fundamentals of critical argumentation. New York,

NY: Cambridge University Press. Walton, D. (2007a). Media argumentation: dialectic, persuasion and rheto-

ric. New York, NY: Cambridge University Press. Walton, D. (2007b). Witness testimony evidence: Argumentation and the

law. Cambridge, UK: Cambridge University Press. Weaver, D. (2007). Thoughts on agenda setting, framing, and priming.

Journal of Communication, 57(1), 142–147. Weber, R. (1990). Basic content analysis. Newbury Park, CA: SAGE Pub-

lications. Zaller, J. (1990). Political awareness, elite opinion leadership, and the mass

survey response. Social Cognition, 8(1), 125–153. Zaller, J. (1992). The nature and origins of mass opinion. New York, NY:

Cambridge University Press.

50

Shah, D., Domke, D., & Wackman, D. (1996). To thine own self be true: Values, framing, and voter decision-making strategies. Communication Research, 23(5), 509–560.

Shapiro, M. (1994). Think-aloud and thought-list procedures in investigat-ing mental processes. In A. Lang (Ed.), Measuring psychological re-sponses to media messages (pp. 1–14). New York, NY: Routledge.

Shapiro, G. (1997). The future of coders: Human judgments in a world of sophisticated software. In C. Roberts (Ed.), Text analysis for the social sciences: Methods for drawing statistical inferences from texts and tran-scripts (pp. 225–238). Mahwah, NJ: Lawrence Erlbaum.

Shen, F. (2004). Effects of news frames and schemas on individuals’ issue interpretations and attitudes. Journalism & Mass Communication Quar-terly, 81(2), 400–416.

Shen, F., & Edwards-Hatfield, H. (2005). Economic individualism, human-itarianism, and welfare reform: A value-based account of framing ef-fects. Journal of Communication, 55(4), 795–809.

Slater, M. (2004). Operationalizing and analyzing exposure: The founda-tion of media effects research. Journalism & Mass Communication Quarterly, 81(1), 168–183.

Slothuus, R. (2008). More than weighting cognitive importance: A dual-process model of issue framing effects. Political Psychology, 29(1), 1–28.

Slothuus, R., & de Vreese, C. (2010). Political parties, motivated reasoning, and issue framing effects. The Journal of Politics, 72(03), 630–645.

Sniderman, P., Brody, R., & Tetlock, P. (1993). Reasoning and choice: Ex-plorations in political psychology. Cambridge, MA: Cambridge Univer-sity Press.

Sniderman, P., & Theriault, S. (2004). The structure of political argument and the logic of issue framing. In W. Saris, & P. Sniderman (Eds.), Stud-ies in public opinion: Attitudes, nonattitudes, measurement error, and change (pp. 133–165). Princeton, NJ: Princeton University Press.

Sproull, N. (1995). Handbook of research methods: A guide for practition-ers and students in the social sciences (2nd edition). Metuchen, NJ: Scare-crow Press.

Squire, P., & Smith, E. (1988). The effect of partisan information on voters in nonpartisan elections. The Journal of Politics, 50(1), 169–179.

Stevenson, C. (1937). The emotive meaning of ethical terms. Mind, 46(181), 14–31.

Stevenson, C. (1944). Ethics and language. New Haven, CT: Yale Univer-sity Press.

Strömbäck, J. (2007). Political marketing and professionalized campaign-ing. Journal of Political Marketing, 6(2-3), 49–67.

51

Strömbäck, J., Mitrook, M., & Kiousis, S. (2010). Bridging two schools of thought: Applications of public relations theory to political marketing. Journal of Political Marketing, 9(1-2), 73–92.

Taber, C., & Lodge, M. (2006). Motivated Skepticism in the evaluation of political beliefs. American Journal of Political Science, 50(3), 755–769.

Thomassen, J. (2005). The European Voter. Oxford, UK: Oxford Univer-sity Press.

Thomson, R. (2001). The programme to policy linkage: The fulfilment of election pledges on socio–economic policy in The Netherlands, 1986–1998. European Journal of Political Research, 40(2), 171–197.

Thomson, R. (2011). Citizens’ evaluations of the fulfillment of election pledges: Evidence from Ireland. The Journal of Politics, 73(01), 187–201.

Thomson, R., Royed, T., Naurin, E., Artés, J., Costello, R., Ennser-Jede-nastik, L., Ferguson, M., Kostadinova, P., Moury, C., Pétry, F., & Prap-rotnik, K. (2017). The fulfillment of parties’ election pledges: A com-parative study on the impact of power sharing. American Journal of Po-litical Science, 61(3), 527–542.

Turner, J., Oakes, P., Haslam, S., & McGarty, C. (1994). Self and collec-tive: Cognition and social context. Personality and Social Psychology Bulletin, 20(5), 454–463.

Tversky, A., & Kahneman, D. (1974). Judgment under uncertainty: Heu-ristics and biases. Science, 185(4157), 1124–1131.

Van der Eijk, C., Franklin, M., & van der Brug, W. (1999). Policy prefer-ences and party choice. In H. Schmitt, & J. Thomassen (Eds.), Political Representation and Legitimacy in the European Union. Oxford, UK: Oxford University Press.

Walton, D. (2001). Persuasive definitions and public policy arguments. Ar-gumentation and Advocacy, 37(3), 117–132.

Walton, D. (2003). Ethical argumentation. Lanham, MD: Lexington Books. Walton, D. (2006). Fundamentals of critical argumentation. New York,

NY: Cambridge University Press. Walton, D. (2007a). Media argumentation: dialectic, persuasion and rheto-

ric. New York, NY: Cambridge University Press. Walton, D. (2007b). Witness testimony evidence: Argumentation and the

law. Cambridge, UK: Cambridge University Press. Weaver, D. (2007). Thoughts on agenda setting, framing, and priming.

Journal of Communication, 57(1), 142–147. Weber, R. (1990). Basic content analysis. Newbury Park, CA: SAGE Pub-

lications. Zaller, J. (1990). Political awareness, elite opinion leadership, and the mass

survey response. Social Cognition, 8(1), 125–153. Zaller, J. (1992). The nature and origins of mass opinion. New York, NY:

Cambridge University Press.

52

Zarefsky, D. (2004). Presidential rhetoric and the power of definition. Pres-idential Studies Quarterly, 34(3), 607–619.

Zarefsky, D. (2006). Strategic maneuvering through persuasive definitions: Implications for dialectic and rhetoric. Argumentation, 20(4), 399–416.

53

Paper I. Explaining Framing Effects:

Insights from Linguistic Theory on

Persuasive Words

Elina Lindgren

Paper I