University of Southern California Dissertations and Theses

Feeling polarized: emotional partisan polarization and its reinforcement through campaign advertising

(USC Thesis Other)

Feeling polarized: emotional partisan polarization and its reinforcement through campaign advertising

PDF

Download a page range

Download transcript

Copy asset link

Request this asset

Transcript (if available)

Content FEELING POLARIZED:
EMOTIONAL PARTISAN POLARIZATION AND ITS REINFORCEMENT
THROUGH CAMPAIGN ADVERTISING

by

Jesse John Mills

A Dissertation Presented to the
FACULTY OF THE USC GRADUATE SCHOOL
UNIVERSITY OF SOUTHERN CALIFORNIA
In Partial Fulfillment of the
Requirements for the Degree
DOCTOR OF PHILOSOPHY
(POLITICAL SCIENCE)

August 2012

Copyright 2012 Jesse John Mills

ii

Table of Contents

List of Tables iii

List of Figures iv

Abstract v

Chapter One: Introduction 1

Chapter Two: Emotion and Polarization 26

Chapter Three: Campaign Advertising and Emotional Response 48

Chapter Four: Ad Content and Horserace Framing 65

Chapter Five: Putting the Pieces Together 90

Bibliograpy: 108

iii

List of Tables

Table 2.1: Ideology 28

Table 2.2: Party Identification and Ideology 28

Table 2.3: Ideology Means 29

Tale 2.4: Partisan Differences in Candidate Favorability Means 29

Table 2.5: Partisan Differences in Candidate Trait Means 31

Table 2.6: Partisan Differences in Emotions 36

Table 2.7: Relationship Between Emotion and Favorability 43

Table 2.8: Regression of Candidate Favorability with Emotions 46
and traits

Table 3.1: Impact of Ad Sponsor on Partisans’ Emotions 57

Table 3.2: Ad Tone and Positive Emotions 59

Table 3.3: Ad Tone and Negative Emotions 60

Table 3.4: Video Treatments and All Emotions Felt Towards 63
the candidates

Table 4.1: Issue vs. Trait Ads 71

Table 4.2: Issue vs. Trait Ads by Candidate 73

Table 4.3: Issue vs. Trait Ads by Tone and Candidate 75

Table 4.4: Horserace Frame and Candidate Emotions 83

Table 4.5: Horserace Frame and Ad Sponsor by Tone 85

Table 4.6: Horserace Frame by Candidate and Tone 88

iv

List of Figures

Figure 2.1: Scaled Partisan Differences between Ideology and Traits 38

Figure 2.2: Scaled Partisan Differences between Ideology and Emotions 39

v

Abstract
This project examines the emotional dimension of partisan polarization, and
executes a variety of tests of Intergroup Emotion Theory (IET). Utilizing original survey
and experimental data gathered during the 2008 presidential election season, this work
demonstrates that partisans are significantly polarized over emotional appraisals of the
candidates, and that the emotions matter; they are significant indicators of overall
candidate favorability. The data also demonstrate that emotions can be impacted by
campaign advertising, and that the substance of these ads matters. Ad tone and topic both
had significant effects on the type and strength of emotions aroused amongst partisan
viewers. The data provide support for IET, and for a conception of partisanship
emphasizing social identification with the parties.

1

Chapter 1 – Introduction

Since the ancient Greek philosophers, there has been debate about the role of
emotion in politics (Aristotle 1954, Plato 1974). Democratic governance rests on the idea
that government is responsive to the interests of its citizens. However, this representation
does not come without responsibility. An ideal of the democratic citizen is one who is
informed, and makes political decisions after careful deliberation (Thompson 1970).
Emotion has often been viewed as incompatible with this ideal (Marcus et al. 2000,
Marcus 2002), as it was thought to distort or replace reasoned decision making. Recent
research has demonstrated that the role of emotion in politics is more complex (Neuman
et al. 2007). Emotions can result in behaviors inconsistent with “ideal democratic
citizenship” such as reliance on standing decisions, decreased information search and a
lack of deliberation. However, emotions are also capable of encouraging political
participation, political information seeking and the re-examination of standing decisions
(Brader 2006, Marcus et al. 2000).
This work adds to the longstanding tradition of studying emotion in politics by
connecting emotion to the study of a widely cited, contemporary phenomenon in
American politics; partisan polarization in the electorate (Abramowitz 2010, Jacobson
2007). Studying the role of emotion in polarization seems especially pertinent given
accounts of a lack of civility in American politics (Carter 1998, Putnam 2000) and the
emergence of ideologically extreme Super PAC’s, who have grown at an astounding rate
since the last election, and can spend virtually limitless sums endorsing or opposing
2

candidates (Garrett 2010). This dissertation examines the role of emotion in partisan
polarization, and the impact of political advertising on partisans’ emotions.
This dissertation begins to address these questions using nationally representative
survey and experimental data gathered during the 2008 presidential election season. In
particular, this work demonstrates that partisans in the electorate are polarizing to a large
extent over emotions felt towards the candidates, and that these emotional divisions
matter towards assessments of candidate favorability. This work also finds that political
advertising has significant effects on the emotions partisans feel towards the candidates:
certain forms of advertising have the ability to incite or depress particular emotions.
Finally, the data demonstrate that all ads are not created equally. Some ads are more
emotionally persuasive than others, and their particular content matters in the emotional
appraisal process. These findings raise a number of important questions regarding the
role of emotion in politics. The remainder of this chapter will review the current literature
on polarization in American politics, and the role of emotion in politics, before making
the argument that accounts of partisan polarization are under-theorized and would benefit
from an inclusion of the emotions.

Partisan Polarization

In recent years, a literature has emerged to address concerns over widening
divisions in American politics (Fiorina and Abrams 2008, Jacobson 2007). The roots of
this literature can be found in studies of the U.S. Congress dating back to the late 1970's
and early 1980's. Scholars studying role-call votes in Congress began to note widening
3

divisions between the voting patterns of its partisan members (Poole and Rosenthal 2007,
Sinclair 2006). For years, this trend continued, and currently, scholars (utilizing a variety
of measures) have concluded that members of Congress have polarized to a significant
extent. Specifically, Democrats and Republicans in Congress have grown more
ideologically distinct and more loyal to their respective party agendas (Stonecash et al.
2003). A similar pattern has also been identified in the American electorate, as partisan
identifiers have begun exhibiting these traits (Abramowitz 2010). Many studying this
phenomenon believe it has problematic implications; polarized leaders may not best
represent the majority of Americans’ and polarization in government can hinder
compromise and deliberation (Fiorina et al. 2005). Polarization in the electorate may
engender more contentious politics, lead to the election of extreme politicians, and some
claim it could ultimately result in a “culture war” (Hunter 1994). It should be noted that
polarization has waxed and waned over time. Ideological polarization in Congress was
extremely high during the mid to late 1800’s (Jacobson 2007). This is not unexpected, as
the nation experienced civil war. After a period of relative calm, polarization has been on
the rise again, and critics of polarization are concerned. The study of polarization has
taken place in largely spatial terms. Scholars tend to note polarization when observations
begin to clump at opposite ends of any spatial spectrum. This dissertation continues in
the tradition of studying polarization in spatial terms, but applies the analysis to a new
dimension in polarization; emotional partisan polarization.

4

A good deal of work has emerged, demonstrating the reality of a polarized
Congress (Groseclose et al. 1999, Jacobson 2007, Jacobson 2000, McCarty Poole and
Rosenthal 2006, Sinclair 2006, Stonecash et al. 2003, Theriault 2008). These studies use
a variety of methods to measure polarization, most derived from roll-call votes. These
votes can tell us a great deal about the level of polarization in Congress, and there have
been a number of ways roll-call votes have been utilized. One such measure involves
counting party votes; roll-call votes where a majority of Republicans vote in opposition
to a majority of Democrats. One would expect that the more polarized the parties, the
greater the percentage of party votes. Not surprisingly, scholars found the percentage of
party votes in the House of Representatives to have risen markedly in the last 25 years,
after relatively low numbers in the late 1960's and 1970's (Sinclair 2006, Stonecash et al.
2003).
Party unity scores are another statistic derived from roll-call data which are used
to measure the level of polarization in Congress. They measure the percentage of the
time an individual congressperson votes with his/her party. While party votes can
demonstrate how often the parties oppose one another, party unity scores help gauge the
depth of those disagreements. According to Stonecash et al. (2003), party unity in the
House of Representatives increased steadily throughout the 1980's and 1990's. Using
party unity scores, Sinclair (2006) found the distance between the parties on partisan
votes in the Senate and the House of Representatives had increased dramatically since the
early 1990's. This evidence demonstrates that in recent years, party conflict in Congress
5

has been occurring more frequently, and that the divisions between the parties have
grown deeper.
These conclusions are validated by Poole and Rosenthal (2007), who derive
ideological scores for every member of congress from roll-call voting records; the DW-
Nominate scale. In agreement with the polarization hypothesis, they found the distance
between the median Nominate score for Republicans and Democrats in Congress to have
doubled since 1980. They also note the changing distribution of Nominate scores. While
once somewhat evenly distributed, Democrats and Republicans in Congress now form
two distinctly bimodal blocks; Republicans are almost exclusively located together on the
right, while Democrats are similarly clumped on the left.
While almost all studying congress agree that polarization has occurred, evidence
of a polarized electorate has been more contentious. This literature was jump-started in
the early 1990's by James David Hunter (1991) who claimed that Americans had become
fundamentally divided over issues of sexuality, morality and religion, and were currently
engaged in a “culture war.” Hunter (1994) also argued that American democracy was not
prepared to mediate differences so deep-seeded and fiercely opposed to one another.
According to Hunter, the new lines of conflict are based on differing views of ultimate
moral authority; while one camp exhibits “impulses towards orthodoxy” the other
demonstrates “impulses towards progressivism” (Hunter 1991, p. 43). The orthodox,
believe the source of moral authority to be external, clearly defined, and transcendent.
However, progressives believe moral authority can be defined “by the spirit of the
modern age, a spirit of rationalism and subjectivism...(where) truth tends to be viewed as
6

a process, as a reality that is ever unfolding” (Hunter 1991, p. 44). The outcome of this
divide, is that the “orthodox” are almost always cultural conservatives, while the
progressives tend to favor a more liberal or libertarian social agenda.
However, the scholarly work that followed, found little support for the notion of
an electorate polarized around cultural issues (Demerath and Yang 1997, DiMaggio et al.
1997, Evans 2003, Fiorina and Levendusky 2006, Mouw and Sobel 2001). Scholars
instead found the American electorate to be no more polarized over moral issues than
they were a quarter-century ago (DiMaggio et al. 1997, Evans 2003, Mouw and Sobel
2001), and believe polarization to be an exaggerated description of American public
opinion on these issues (Demerath and Yang 1997, Fiorina and Levendusky 2006). In
fact, Fiorina and Levendusky (2006) even claim that when looking at abortion and same
sex marriage, two issues that supposedly lie at the center of the “culture war,” Americans
as a whole do not exhibit preferences that should be described as polarized.
Although evidence of a culture war never materialized, scholars studying
polarization have unearthed convincing evidence that there are large divisions in the
electorate, falling largely along partisan lines (Abramowitz 2010, Abramowitz 2006,
Galston and Nivola 2006, Jacobson 2007, Jacobson 2006, Jacobson 2005, McCarty et al.
2006, Sinclair 2006). Looking to measures of self-described ideology, issue positions,
voting habits and political evaluations, these scholars make a strong case that partisan
identifiers in the electorate have become significantly polarized. However, not all believe
this phenomena warrants the title “polarization” or even “partisan polarization” (Fiorina
and Abrams 2008, Fiorina et al. 2005). They instead claim that these divisions are the
7

result of ideological “sorting.” This argument will be further examined later in this
section.
One measure on which partisan divisions have been noted is ideology.
Researchers have found the relationship between voter’s self-described ideology and their
party identification to have strengthened significantly since 1972; Democrats are more
likely to be liberal and Republicans are more likely to be conservative than in the past
(Jacobson 2001). McCarty et al. (2006) echo this sentiment and claim that partisans are
more likely to identify themselves ideologically as “conservative” or “liberal” than non-
partisans, and that partisans are becoming less likely to identify as “moderates.” The
distance between Democrats’ and Republicans’ ideological self-placement has also
grown since the late 1970's; partisans report increased ideological distance between their
party and the opposition party (Abramowitz 2006, Jacobson 2007).
Partisans in the electorate have also grown more divided over issue positions.
Examining the answers to questions that have been included in the NES surveys since
1980 or earlier, Abramowitz (2006) finds that “On all...of these questions, there has been
a substantial increase in partisan polarization over time” (Abramowitz 2006, p. 80-82).
Specifically, partisans have grown farther apart in views on defense spending, views on
government aid to blacks, and views on abortion. Since the mid 1970's Jacobson (2007)
reports increased partisan coherence on a variety of issues. He states, “On every issue-
ranging from the government’s economic role, to race, to women’s role in society, to
abortion policy-the overall trend is upward” (Jacobson 2007, p. 31-32). Probing even
deeper into the economic questions, Abramowitz and Saunders (2005) find large
8

disparities in policy liberalism between active Democrats’ and active Republicans’ on
social issues, economic issues and foreign policy issues.
Partisan are also increasingly divided over evaluations of political figures.
According to Jacobson (2007), partisans in the electorate have become more likely to
differ in their evaluations of political figures, specifically President Bush, and his
handling of a variety of issues. Bush’s approval rating amongst Republicans (between
2001-2006) hovered between 80%-90%, while Democrats’ approval has remained
between 10-20%. Ratings of Bush’s handling of the economy, terrorism and Iraq look
similar (Jacobson 2007, p. 10-13).
Finally, the trend of partisan polarization continues when looking to political
behavior. Jacobson (2007) notes that voters’ party loyalty (when voting) has increased,
and that they are less likely to split their tickets than in the past. He points specifically to
the declining number of conservatives voting for democratic candidates for House and
Senate seats since the 1970's. However, it is not only conservatives who have begun to
display such high levels of party loyalty in elections. According to Jacobson, “In the
most recent congressional elections, 80 percent of self-identified liberals voted for
Democrats, while 80 percent of conservatives voted for Republicans...among presidential
voters in 2004...82 percent of conservatives voted for Bush, while 92 percent of the
liberals voted for Kerry (Jacobson 2007, p. 29).”
It must be noted that some scholars (Fiorina et al. 2005, Fiorina and Levendusky
2006, Fiorina and Abrams 2008) do not believe the above-mentioned trends are
exemplary of polarization, but rather “sorting.” Fiorina et al. (2005) demonstrated that
9

the red-blue rift in state-level public opinion was being exaggerated. This, combined
with evidence presented earlier that mass opinions have not polarized (DiMaggio et al.
1997, Mouw and Sobel 2001), lead Fiorina to claim that the differences between
Democrats and Republican identifiers are not the result of mass opinion change, but
rather the result of partisans who have better sorted themselves into the ideologically
“correct” party. Thus, the differences between partisans on all of the dimensions
mentioned earlier (ideological cohesion, issue cohesion, political evaluations and voting
behavior), result not from shifts in public opinion, but from more accurate party sorting.
Fiorina’s argument is quite convincing; the differences between partisans in the electorate
appear not to be the result of mass opinion change. However, party “sorting” has left us
with parties which have been growing apart on a variety of measures for decades.
This body of research has increased understanding of the widening differences
between Democrats and Republicans. However, it has largely ignored the emotional
dimension of partisan polarization. Recent research has demonstrated the significance of
emotions in politics (Brader 2006, Marcus et al. 2000), and suggested that emotion plays
a large role in political evaluations and behaviors. The following section will review the
contributions of those studying emotion in politics, in order to make the case for the
inclusion of emotion in accounts of partisan polarization.

Emotion in Politics
In recent years scholars have found politics to be laden with and significantly
influenced by emotion (Abelson et al. 1982, Belt et al. 2007, Brader 2005, Cassino and
10

Lodge 2007, Gross 2008, Huddy et al. 2007, Huddy et al. 2005, Just et al. 2007, Isbell
and Ottati 2002, Lau and Redlawsk 2006, MacKuen et al. 2007, Marcus et al. 2000,
Marcus and Mackuen 1993, Valentino et al. 2008). In reviewing the contributions of
these scholars it’s important to recognize that while all share a fascination with emotion
and its consequences, they do not adhere to a single theory of affect. These theories vary
along a variety of dimensions, including the relationship posited between affect and
cognition.
In the preface to a current edited volume on this topic, Neuman et al. (2007) note,
“By our count, there are 23 named theories, models, or central concepts used to explicate
the interaction of affect and cognition at various points in this work...” (Neuman et al.
2007, p. 6). Others have even characterized the proliferation rate of these theories as
“horrifying” (Kinder 1994, p. 279). The sheer number of theories of affect make it all the
more important for scholars conducting this type of research to be explicit in their
preferred theory, and the assumptions they make about emotion in their work.
One way is to group these theories by the structure of affect proposed. Doing so
leads to three conceptions of affect’s structure: (1) valenced conceptions of emotion view
emotion as existing along a single dimension (such as a positive-negative continuum); (2)
multidimensional conceptions claim multiple, valenced dimensions exist simultaneously,
and finally; (3) discrete conceptions of emotion view emotions as separate and
qualitatively different both in the way they are triggered and their effects.
Abelson et al. (1982) utilize valenced scaling of discrete emotion in order to study
the affective and cognitive components of judgments of political figures. Using surveys
11

conducted during the 1980 presidential election season, they measure both the traits and
emotions ascribed to the candidates by members of the public. They constructed scales
of both the affect and trait scores, in which each trait and emotion was viewed as either
positive or negative. The authors found affective feelings towards the candidates to be
independent of candidate trait assessments. They also found individuals’ affective scores
of the candidates to be stronger predictors of voting preference than trait scores.
Those who ascribe to multidimensional models of emotion, typically extend
valence models into two-dimensional space. One well-received multidimensional model
created by Marcus et al. (2000) is coined “affective intelligence.” This model views
affect along two separate dimensions, one monitored by the dispositional system
(enthusiasm – depression), and the other monitored by the surveillance system (anxiety –
calmness). Marcus et al. (2000) claim that these two systems perform separate, but
equally important tasks. The dispositional system monitors habitual activity, rendering
emotional feedback to regulate these types of behaviors. When tasks are performed
successfully, this system relays enthusiasm, which encourages the continuation of these
behaviors. When they are not properly executed, this system relays feelings of
depression or sadness, encouraging the person to discontinue these habits. In summary,
“...we sustain those habits about which we feel enthusiastic, and abandon those that cause
us to despair” (Marcus et al. 2000, p. 10).
The surveillance system monitors the environment, scanning for novelty and
threat. According to Marcus, “It serves to warn us when we cannot rely on past learning
to handle what now confronts us and to warn us that some things and some people are
12

powerful and dangerous” (Marcus et al. 2000, p. 10). Particularly, when it recognizes
situations that are different and/or potentially dangerous, it triggers anxiety, which leads
to a rational re-assessment of habitual behavior in light of the new circumstances. In
contrast, when nothing novel or threatening is perceived, this system produces emotions
of calmness or safety. While each dimension resembles valence models of emotion the
two in tandem produce a more complex model which Marcus et al. claim more accurately
describe human emotional response.
Others have incorporated this model into their research (Brader 2005, Redlawsk et
al. 2007, Graber 2007 and Valentino et al. 2008). Brader (2005) studies the impact of
images and music in political ads on emotional response, and the subsequent impact of
the emotions on political behaviors. In order to measure the effects of images and music,
two experiments were conducted during the 1998 Massachusetts Democratic primary
race for governor. One, the enthusiasm experiment, had subjects view positive campaign
ads with and without nonverbal enthusiasm cues. The fear experiment, had subjects view
negative campaign ads with and without nonverbal fear cues. Brader expected the
images and music to increase the affective power of the ads. He also expected, consistent
with Marcus et al’s findings (2000), feelings of enthusiasm to increase interest in the
campaign, the intention to vote, and the reliance on prior candidate preference in the
voting calculus, while feelings of fear would lead voters to seek out new information, and
rely less on their “standing decisions.”
Consistent with expectations, Brader (2005) found nonverbal cues present in
campaign ads to have the ability to (further) elicit emotion: subjects who were shown the
13

ads with manipulated music and images reported higher levels of enthusiasm/hope and
anxiety/fear (Brader views fear and anxiety interchangeably, although scholars from the
“discrete emotions” tradition identify anxiety and fear as qualitatively distinct emotions.
The same is true of the hope/enthusiasm dimension measured in this work). Brader’s
results demonstrated that enthusiasm cues increase interest in the campaign, intention to
vote and to increase reliance on pre-existing beliefs about the candidates, while fear cues
increased the search for new information and were better able to persuade voters to
support the sponsor.
Redlawsk et al. (2007) tested the expectations of affective intelligence theory
using “dynamic process tracing” where they “presented voters with a campaign in which
evaluative expectations were often violated and emotional responses to candidates were
heightened (Redlawsk et al. 2007, p. 152). They hypothesized enthusiasm would result
when voter-candidate agreement (between voters and their initially preferred candidate,
or their initially rejected candidate) on the issues was high, anger would result when such
agreement was low, and anxiety would result when agreement was too distant for
comfort, but too close for anger. They also expected information generating enthusiasm
to take less time to process than information generating anxiety and anger. Finally, they
claimed that while anxiety for a candidate would trigger more information search
regarding that candidate, enthusiasm would have no effect on information search, and
anger would lead to less information search (Redlawsk et al. 2007, p. 156-159). The data
supported all three hypotheses.
14

Valentino et al. (2008) reported similar findings in a study of voter information
seeking. Using an experimental design to induce emotions, these researchers found that
although anger, enthusiasm and anxiety all lead people to claim they would seek out
campaign information, anger actually depressed the search for information. In a separate
experiment these authors found that exposure to realistic threats triggered multiple
emotional reactions, but only anxiety increased learning and the search for new
information.
Finally, Graber (2007) studied attention thresholds and news consumption,
focusing specifically on how certain news stories are able to engage the surveillance
system, and encourage engagement, while others are not. Affective intelligence theory
leads one to believe that stories which elicit anxiety or fear in their viewers are most
likely do this. But what elements are necessary for such stories to break through the
information threshold to relevance for the individual? Graber found a list of factors
contributed: fear arousing elements, stimuli suggesting grave harm is imminent for the
self or significant others, signals of journalistic importance, embedding the story in a
supportive context of political and social events, and finally, interaction of fear-producing
stimuli at the individual and societal levels (Graber 2007).
It is important to note that affective intelligence theory is not the only theory of
emotion with a successful pedigree. Appraisal theories of emotion are also widely
utilized by political researchers, and they offer specific advantages over affective
intelligence theory (Huddy et al. 2005, Just et al. 2007). They are consistent with a
discrete conception of emotion, and outline the steps of the appraisal process determining
15

which emotion is aroused by a particular stimulus (Lazarus 2001, Roseman and Smith
2001). Those utilizing appraisal theories of emotion have found distinct differences
between positive emotions such as hope and enthusiasm (Just et al. 2007), and negative
emotions such as anger and fear (Mackie, Devos and Smith 2000).
There is a longstanding tradition of scholars who have studied emotions as
discrete phenomena (Gross 2008, Huddy et al. 2005, Just 2007). Huddy et al. (2005)
examine the effects of threat and anxiety on support for anti-terrorism policies. While
both are common psychological reactions to terrorism, these authors claim threat and
anxiety are fundamentally different, and lead to different behaviors. Conducting a
national telephone survey, Huddy et al found evidence of evidence of these differences.
As expected, anxiety increased risk aversion, as those anxious over the 9/11 attacks were
less supportive of aggressive military action against terrorists, while those who felt
threatened (but not particularly anxious) were far more supportive of anti-terrorism
policies.
Gross (2008) examined the emotional effects of two common media frames
(episodic and thematic), and the cognitive and behavioral effects that result. She had
subjects read a column arguing against mandatory minimum sentences. The frame
through which the article was written was varied between episodic and thematic. She
found episodic frames to be more emotionally engaging than thematic frames. She also
found the emotions elicited (sympathy and pity) to be correlated with opposition to
mandatory minimum sentences.
16

Just et al. (2007) studied how emotions felt towards the candidate influence vote
choices and expectations about election results. These researchers focus their attention
on hope, rather than the numerous negative emotions commonly studied, and claim it is
“...the key emotion in voting decisions and...essential for the democratic process” (Just et
al. 2007, p. 231). Using original survey data gathered during the 1996 presidential
election season, they found hope for President Clinton to be the emotion reported more
often than any emotion for either candidate. They also compared hope’s influence on
vote choice to party identification. They found feelings of hope for Clinton and Dole in
July to contribute as strongly as party identification to candidate preferences. By
campaign’s end, “hope for Clinton was the strongest contributor to candidate preference”
(Just et al. 2007, p. 243).

Theoretical Basis for the Inquiry
This project utilizes a discrete conception of emotions, and is theoretically
motivated by Intergroup emotion theory; originally described as an offshoot of appraisal
theory. Appraisal theories generally claim that emotions are elicited by evaluations of
events, situations or other stimuli as they relate to an individual’s goals, maintenance or
well-being (Lazarus 1991, Scherer 1999, Smith and Kirby 2001). Lazarus (1991) defines
cognitive appraisals as evaluations (either conscious or sub-conscious) of the significance
of events or stimuli in the world, towards personal well-being. He claims this process
entails both “primary” and “secondary” appraisals; primary appraisals evaluate core
relational meanings (what are the costs and/or benefits of the event/stimuli), while
17

secondary appraisals weigh possible actions, and strategies for coping. The resulting
emotional response reflects both of these evaluations and is intended to help the
individual deal with the world around him. Appraisal theories of emotion recognize a
variety of discrete emotions, which can result from the appraisal process, and specify the
action tendencies associated with them. Appraisal theories of emotions have some key
strengths: they offer insights into the processes through which emotions are triggered,
and the behaviors that are likely to result. Most important to this work, appraisal theories
claim that when individuals appraise events similarly, they are likely to experience the
same emotions.
One weakness of appraisal theory is its inability to account for emotional states
that are not the result of individual appraisals of goal-relevance (Huddy and Mason
2008). Recent work by social psychologists has found that in addition to individual-level
emotions, group-level emotions are important to explain intergroup behavior and the
emotions that individuals report feeling (Costarelli 2003, Garcia et al. 2006, Mackie et al.
2000, Rydell et al. 2008, Smith et al. 2007). Intergroup emotion theory (IET) draws on
social identity theory to expand appraisal theory to groups. Identity scholars (Tajfel
1974, Tajfel and Turner 1979, Tajfel and Turner 1986) have spoken on this issue,
offering a conception of identity in which individuals view themselves not only in terms
of an “individual self,” but also in terms of groups with which they identify. Most
importantly, they find that triggering group identities, often leads to ingroup favoritism
and outgroup discrimination.
18

Intergroup emotions theory combines appraisal theory and social identity theory
to illuminate the relationship between group identity and emotion. It argues that
individuals feel emotions on behalf of groups they identify with, even when they,
themselves, are not directly impacted by the events or stimuli affecting the group. IET
also provides expectations regarding the particular emotions group identifiers are likely
to feel towards in-groups and out-groups when group identity is made salient. For
example, Mackie, Devos and Smith (2000) find that anger and fear are distinct, as are the
tendencies to act against the outgroup, versus moving away from it. Specifically, they
claim that when the ingroup is perceived as “strong,” group members are likely to
respond to threats with anger, and the tendency to act against the outgroup. However,
when the ingroup is perceived as “weak,” fear was a more likely emotional response to
outgroup threat, and group memers were less likely to desire taking action against the
outgroup.
The prototypical sports “fan” provides a good illustration of how intergroup
emotions theory works. The fan identifies strongly with a sports team, and feels
emotions on behalf of the team, even though he/she does not actually participate in the
contest, and cannot “win” or “lose” as the players do (Crisp et al. 2007). Sports are an
appropriate and often used metaphor for election campaigns (Lakoff 1996, Patterson
1993), and partisan identity is central to individuals’ understanding of and reactions to the
campaigns. Recent accounts of partisanship also provide support for the notion that
partisanship is more than an ideological decision, but also a social identity (Green,
Palmquist and Schickler 2002). Partisans during the campaign season socially identify
19

with their group, as the election and horserace makes their partisan group identity salient.
Once group identity is made salient, group members tend to process information about
the campaigns through the lense of partisanship. This leads to ingroup favoritism and
outgroup discrimination (Smith et al. 2007).
For these reasons IET appears ideally suited for the study of partisans’ emotions
during an election campaign. This work utilizes IET as the framework for analyzing the
emotions partisans’ feel during the campaigns. Although IET has been underutilized in
the study of partisans’ emotional reactions to the campaigns, there have been notable
exceptions (Huddy and Mason 2008, Crigler, Mills and Hootstein 2009, Just et al. 2010).
The contributions of those attempting to incorporate the study of emotion into
theories of voter behavior have been sizeable. It is important to take a few lessons from
this body of work. First, emotions matter. Whether looking at vote choice, information
seeking or candidate favorability, emotions play a role. Secondly, there are a variety of
emotions that are being studied, and current research should speak to the variety of
emotions which can influence political thinking and behavior. Finally, no cohesive
theory of emotions has emerged, leaving researchers with the task of deciding which
theory of emotions would be well-suited to the phenomena one is studying.

Data and Methods
Data utilized in this and the following chapters were gathered during the 2008
presidential election, in collaboration with the CCAP study, a multi-wave Internet panel
survey conducted by Polimetrix, under the leadership of Simon Jackman and Lynn
20

Vavreck. The weighted data are drawn from the October wave of the panel, which took
place only weeks before the general election. The sample consisted of 979 participants,
49.9% female and 50.1% male. The participants in the sample are overwhelmingly white
and non-Hispanic (83.7% white, 7.7%% African-American, 5.4% Latino/a, 0.6% Asian-
American, 2.6% other). Residents of battleground states were oversampled in order to
gauge opinions of those most impacted by the campaigns.
The survey contained baseline questions regarding campaign attentiveness, media
consumption, campaign involvement, political knowledge, issue positions and ideology,
candidate preferences, emotions felt towards the candidates and a variety of demographic
variables. A customized video experiment was also embedded into the study, with a
variety of pre-test and post test measures. These included measures of emotional
assessments of the candidates, trait assessments of the candidates, candidate favorability
and other dependent variables. The video experiment was designed to gauge the ways
campaign communication can impact emotional assessments of the candidates. This
experiment seeks to provide data regarding how reassurances and threats conveyed
through campaign ads, affect partisans’ emotions.
The video experiment simulated exposure to actual campaign ads in the context of
the Presidential election campaigns. Subjects were either shown either an actual
campaign ad, sponsored by the candidates’ campaigns, and aired during the 2008 election
season, or a neutral video which explained the importance of the vice-presidential
debates. Each video was preceded with a horserace statement indicating that one of the
candidates had recently moved ahead or fallen behind in the polls in key battleground
21

states. This design attempted to maximize external validity by using actual ads aired by
the candidates in 2008, and preceding them with a horserace message (which candidate is
ahead in the race), the most common media trope utilized in election coverage
(Farnsworth and Lichter 2006, Patterson 1993). The horserace frame was utilized in
order to make group identity salient, and to ensure subjects were reminded of intergroup
conflict aspect of the election. Intergroup emotion theory speaks to contexts (especially
group struggles/conflict) where group identity is salient; campaign ads, contextualized by
a horserace statement indicating one side is winning while the other trails, accurately
convey this.
Specifically, horserace frames were displayed in print on the screen, and read,
“Candidate X (John McCain or Barack Obama) has recently moved ahead/fallen behind
in the polls in key battleground states.” The subject of the statement was varied, so each
candidate was mentioned as moving ahead or falling behind. This allows for
measurement of the impact of positive and negative horserace statements on partisans
from both parties. It was important to limit the horserace statement to “key battleground
states,” because at the time the experiment was conducted, Obama was leading in almost
all national polls. The statement needed to be believable.
The ad component of the experimental stimuli was varied by sponsor (Obama vs.
McCain), tone (attack vs. appeal) and topic (issue concerns vs. character concerns). This
allows for the examination of a variety of variables which may help explain the activation
of emotions. Because ads varied in length, those longer than 30 seconds were edited
down to less than 40 seconds. This ensures that subjects weren’t exposed to ads of
22

drastically different lengths. Edits were made carefully to ensure the key message of
each ad was left intact. Music and symbols in ads can also have a significant impact on
their emotional persuasiveness (Brader 2005), so edits were made with consideration
towards preserving these aspects as well. The particular ads shown were:
- McCain Reassuring Ad (Character): “Love” Released July 7, 2008: recounts
McCain’s traumatic experiences as a Prisoner of War, discussing his heroism,
patriotism and love for country.
- McCain Attack Ad (Character): “Celeb” Released July 30, 2008: McCain attacks
Obama’s celebrity persona (by comparing Obama to notable celebrities of
popular culture) questioning Obama’s ability to lead.
- Obama Reassuring Ad (Character): “Country I Love” Released June 19, 2008:
features Obama himself speaking directly into the camera positively discussing
his background, family history, upbringing and education.
- Obama Attack Ad (Character): “Original” Released August 6, 2008: Obama
attacks McCain’s persona as a self-described “maverick” and compares
McCain’s policies to those of George W. Bush.
- McCain Reassuring Ad (Issue): “Ignite” Released April 14, 2008 - McCain
speaks about creating jobs, simplifying taxes, making energy cleaner and more
efficient, re-training workers, making healthcare affordable, and restructuring
mortgage debt.
- McCain Attack Ad (Issue): “Taxman” Released August 15, 2008: McCain
criticizes Obama’s economic policies and claims they will increase taxes and
cause a rise in gas prices.
- Obama Reassurance Ad (Issue): “Three Bedroom Ranch” Released August 13,
2008: Outlines Obama’s economic growth plan, which includes tax-cuts,
investing in education and energy independence.
- Obama’s Attack Ad (Issue): “Original” Released August 6, 2008: Criticizes
McCain’s economic plans as not benefitting the working-class.
- Neutral Video - Consists of a 30-second excerpt from the Chancellor of
Washington University in St. Louis discussing the importance of presidential and
vice-presidential debates in the 2008 election. Neither candidate’s name was
mentioned.
This experimental design has some key strengths. First, rather than creating a
mock campaign with fictional characters, or manipulating advertisements in an existing
23

race, this study uses actual campaign ads aired by the candidates in 2008. This means
some subjects may have already viewed some of the ads. This fact does not detract from
the validity of this experiment, because in a real-world presidential campaign, partisans
are likely to see a variety of televised ads and are likely to see the same ads aired
repeatedly. Secondly, by preceding each ad with a statement regarding the horserace, this
experiment’s applicability to the real world is supplemented. The horserace frame is
ubiquitous during campaign season (Graber 2000). Preceding each ad with a horserace
frame ensures that these real-world conditions are met.
Another strength of this design is the specificity with which emotions are
measured. This project gauges four discrete emotions (hope, enthusiasm, fear and anger),
as well as the object of those emotions; the candidates running for office. Intergroup
emotions theory claims that partisans feel a mixture of negative and positive emotions
during campaign season, but that the object of those emotions is likely to be determined
by their partisan identity. For this reason, it is necessary to measure which candidate the
emotions are felt towards. This design was created to closely mimic the real-world
campaign. The final strength of this experiment was its timing. Measuring the effects of
campaign ads and horserace messages so close to the general election provides a tough
test of this experiment; by the end of the campaigns, voters are more likely to have their
candidate choices made. They have also been exposed to campaign communication for
months, so viewing one ad online pales in comparison to the barrage of campaign ads
they’ve likely viewed. These conditions provide a robust test of the experimental stimuli.
24

The subsequent chapters proceed as follows. Chapter two will looks at pre-
experimental data regarding the abundance and structure of polarization. It examines
variables measuring partisan differences in ideology, issue positions and emotions felt
towards political objects. Key to this chapter is the finding that while partisans are
deeply divided over ideology and candidate traits, they are more divided over emotional
assessments of the candidates. The chapter concludes by demonstrating the importance
of emotion in determining candidate favorability.
Chapter three examines the effects of common types of political advertising and
media frames on polarization. This chapter envisions polarization as a fluid
phenomenon, consisting of a set of evaluations that are malleable from moment to
moment. This differs from more traditional accounts of partisan polarization. Scholars
agree on the emotionally persuasive nature of campaign ads shown during presidential
elections. This chapter examines the effects of these ads on voters’ emotions and their
evaluations of the candidates. The goal is to discover what types of political ads are most
effective at generating emotion. In order to do this, the effects of positive and negative
ads from each candidate, addressing a variety of topics (particular issues and personal
characteristics) are studied. The ads were also paired with a common media trope; the
horserace. This pairing closely mimics the real-world campaign environment.
Chapter four takes a deeper look into the emotion-inducing effects of campaign
advertising and compares the impact of issue vs. image driven ads, and the interaction
between ad content, ad tone and the emotions elicited. This chapter also examines the
impact of the horserace frame on the emotional effectiveness of the campaign ads and
25

asks, “Are certain ads more emotionally persuasive when they are paired with a particular
horserace message? Finally, chapter five summarizes the key findings, and discusses the
implications of those findings for representation, theories of emotion, accounts of
polarization and literature concerning political advertising.

26

Chapter 2 – Emotion and Polarization

This chapter examines the state of partisan divisions in the electorate, during the
2008 presidential election season. Motivated by previous work on partisan polarization
in the electorate, this work looks to many of the traditional dependent variables such as,
ideology, candidate favorability assessments and candidate trait assessments. However,
this study also examines emotional divisions between partisans. While partisan divisions
exist on all of these variables, partisan differences are most distinct in the arena of
favorability ratings of presidential candidates and emotional assessments of those
candidates. The analysis in this chapter demonstrates that the emotions have distinct and
significant effects on favorability ratings of the candidates.
This chapter also adopts a new measure of polarization on the dependent variables
that allows for comparison of partisans’ relative degree of polarization on each. The
term “polarization” has been used loosely in the literature, and applied to situations where
partisan divisions have increased over time. Specifically, Fiorina (2005), Abramowitz
(2006) and Jacobson (2007) all use it to describe situations where partisans take more
extreme positions, and are less likely to be found in the center of any measure. This work
operationalizes polarization as the distance between Democrats’ and Republicans’ mean
scores on a variety of variables, as a percent of the scale.

How Divided Are Partisans?
According to Intergroup Emotion Theory (IET), group members feel group-level
emotions in intergroup contexts, when group identity is salient (Smith, Seger and Mackie
27

2007). These situations can range from sporting contests, to election contests. This
section compares the degree to which Democrats and Republicans differ on a variety of
dependent variables, including the emotions. The data demonstrate that partisans are
highly divided over emotion.
Typically, dependent variables used to measure polarization have been ideology,
issue positions, political evaluations and voting behavior (Abramowitz 2010, Fiorina
2005, Jacobson 2007). Beginning with an examination of Democrats’ and Republicans’
self-reported ideology during the 2008 campaign season, divisions are noticeable, but not
substantial. As Table 2.1 shows, the percentage of those who self-identified as strong
liberals or strong conservatives was fairly low; only 19% did so. Moderate was the
category most often cited (42%), while about 17% of the sample identified as liberal, and
22% identified as conservative. These numbers suggest an electorate that while divided
over ideology, has more ideological moderates than strong ideologues (see Table 2.1).
This pattern holds examining Democrats and Republicans individually.
Democrats (including including Democratic leaners) were likely to identify as moderates
(51%) or liberals (31%) but not at all likely to identify as very liberal (9%). Democrats
also identified as conservative (7%) and very conservative (2%) occasionally.
Republicans (including leaners) were more likely than Democrats to hold extreme
ideologies, but very conservative still was not the most common response. Republicans
identified themselves as very conservative at a high rate (29%) but were more likely to
identify as conservative (42%) and almost as likely to identify as moderate (27%). They
identified as liberal (1%) and very liberal (<1%) at very low rates. Independents, as
28

expected, were most likely to identify as moderate (64%) and in this sample were slightly
more likely to identify as conservative (15%) and very conservative (9%) than liberal
(8%) and very liberal (4%). While partisans are clearly divided over ideology, the data
presented in table 2.2 demonstrates that these divisions are not as large as some may
expect, and that Republicans are more ideologically extreme (see Table 2.2).

Table 2.1: Ideology
N Very Liberal Liberal Moderate Conservative Very Conservative
914 48 (5%) 150 (17%) 386 (42%) 205 (22%) 125 (14%)

Table 2.2: Party Identification and Ideology
N Very Liberal Liberal Moderate Conservative Very Conservative
Democrats 411 38 (9%) 128 (31%) 209 (51%) 27 (7%) 9 (2%)
Independents 92 4 (4%) 7 (8%) 59 (64%) 14 (15%) 8 (9%)
Republicans 361 2 (<1%) 4 (1%) 96 (27%) 153 (42%) 106 (29%)

Comparing Democrats’ and Republicans’ mean ideological scores, more clearly
demonstrates the magnitude of these differences. On a 1-5 scale of ideology, Democrats’
mean score was 2.61, followed by Independents’ at 3.16 and finally Republicans at 3.99.
These differences were significant. In conclusion, Democrats and Republicans, while
divided over ideology, do not appear to be highly polarized. Democrats’ mean score falls
somewhere between moderate and liberal, while Republicans’ mean score falls on
conservative. The average partisan is not an ideological extremist. In fact, Democrats
are more likely to identify as moderate than liberal or very liberal. Republicans were
more likely to identify as conservatives, but a large number still identified as moderate.

29

Table 2.3: Ideology Means (1-5 Scale)
N Mean F Sig. Paired Sig.
Democrats 411 2.61
268.69

***
D/I ***
I/R ***
D/R ***
Independents 92 3.16
Republicans 361 3.99
Significance levels: * P .05, ** P .01, *** P .001

Candidate Evaluations
While ideology is an often-cited indicator of partisan polarization, it is not the
only one. This section examines how divided partisans are on measures of overall
candidate favorability and candidate traits. Measures of candidate favorability exhibit
the largest partisan differences. On a 5 pt. favorability scale Democrats’ and
Republicans’ mean evaluations of their party’s candidate were 4.31 and 4.13 respectively
(see Table 2.4). However, their evaluations of the opposing candidate were far lower:
Republicans rated Obama 1.52 while Democrats rated McCain 1.89. Clearly, members
of the two parties overwhelmingly favor their own candidate over the partisan opponent.
As expected, independents rated the candidates in the middle range, but gave a slight
advantage to Obama, rating him 3.06 while rating McCain 2.70.

Table 2.4: Partisan Differences in Candidate Favorability Means (1-5 Scale)
Party Mean F Sig. Paired Sig.
Obama
Favorability
Democrats 4.33
592.18

***
D/I ***
I/R ***
D/R ***
Independents 2.99
Republicans 1.52
McCain
Favorability

Democrats 1.87
400.42

***
D/I ***
I/R ***
D/R ***
Independents 2.63
Republicans 4.06
Significance levels: * P .05, ** P .01, *** P .001

30

When examining evaluations of character traits, there is a similar, yet less
substantial partisan divide. These divisions did not appear strong on all traits however.
Traits such as “experience” and “patriotism” focus on particular qualities which may
exist independent of overall favorability. However, even where partisan differences in
partisans’ evaluations weren’t large, evidence of partisan bias existed.
Looking specifically to evaluations of trust, experience and patriotism, it is clear
that partisans in the electorate are significantly divided in their character trait evaluations
of the candidates. Based on a 4pt. scale, Democrats and Republicans rate trust in their
own party’s candidate at 3.24 and 3.26, while only rating the opposing candidate 1.80 and
1.33 respectively. While neither trusts the opponent it is interesting to note that
Republicans trust Obama less than Democrats trust McCain. Independents show virtually
identical levels of trust towards the two candidates, rating Obama 2.30 and McCain 2.32.
Looking to how partisans rate the experience levels of the candidates again demonstrates
a significant partisan divide. Democrats and Republicans rate their own candidate’s
experience at 2.88 and 3.57, far higher than their ratings for the opponent; Democrats
assign McCain 2.41 while Republicans assign Obama 1.28. These results indicate that
partisans are deferential to their party’s candidate even in instances where the facts
appear to be against them; Democrats gave Obama higher experience ratings than
McCain even though Obama was a newly elected Senator and McCain was serving his
fourth term in office. Independents, however rate Obama’s experience as a 2.02 and
McCain’s as a 2.76, indicating that they hold less of a bias than the partisans (see Table
2.5).
31

Finally, when looking to measures of patriotism, there is a similar partisan divide
for Obama, but a smaller divide when looking at McCain. This may be due to the fact
that McCain was a prisoner of war and campaigned explicitly on patriotic themes. It is
still important to note that Democrats assigned McCain significantly lower patriotic
ratings than Republicans. Obama’s ratings were 3.30 for Democrats and 1.63 for
Republicans, while McCain’s ratings were 3.26 for Democrats and 3.84 for Republicans.
Independents gave McCain high patriotism ratings (3.22) and gave Obama much more
favorable patriotism ratings than Republicans did (2.37) (see Table 2.5).

Table 2.5: Partisan Differences in Candidate Trait Means (1-4 Scale)
Party N Mean F Sig. Paired Sig.
Trust
Obama
Democrats 405 3.24
519.77

***
D/I ***
I/R ***
D/R ***
Independents 89 2.30
Republicans 358 1.33

Trust
McCain
Democrats 408 1.80
271.69

***
D/I **
I/R ***
D/R ***
Independents 89 2.32
Republicans 362 3.26

Experience
Obama
Democrats 393 2.88
400.64

***
D/I ***
I/R ***
D/R ***
Independents 90 2.10
Republicans 365 1.28

Experience
McCain
Democrats 396 2.41
193.34

***
D/I **
I/R ***
D/R ***
Independents 90 2.76
Republicans 366 3.57

Patriotic
Obama
Democrats 397 3.30
358.50

***
D/I ***
I/R ***
D/R ***
Independents 92 2.37
Republicans 355 1.63

Patriotic
McCain
Democrats 402 3.26
81.05

***
D/I N/S
I/R ***
D/R ***
Independents 93 3.22
Republicans 364 3.84
Significance levels: * P .05, ** P .01, *** P .001
32

The data presented in Table 2.5 demonstrate that partisans exhibit significant
division over trait assessments of the candidates. Partisan bias can even be evidenced on
some of the traits where partisan differences don’t appear all that large (McCain’s
patriotism and experience ratings for example). In summary, partisan divisions remained
high for Obama’s traits, and all of McCain’s traits except patriotism.

Divisions over Emotion
Many researchers studying political evaluations and behavior claim that including
the emotions is essential to understanding politics (Brader 2005, Marcus et al. 2000).
This section reviews the original survey data gathered in the October wave of the CCAP
that measures a variety of emotions felt towards the candidates. Looking to positive
emotions voters feel during election season (hope and enthusiasm), as well as the
common negative emotions (fear and anger), the data demonstrate a large and significant
partisan differential. This is important, because the inclusion of the emotions in the story
of partisan polarization helps indicate the true magnitude of polarization. This is also
important, because as this chapter will demonstrate, emotions have a large and
statistically significant impact on candidate favorability, and they can be activated and
depressed by campaign communication such as political advertising (see chapter 3).
While the literature on polarization provides reason to expect deep partisan
divisions on ideology and candidate trait assessments, IET gives us reason to expect even
deeper emotional divisions. IET is an offshoot of appraisal theory, which claims that
emotional appraisals can be made as a result of social group identities as well as
33

individual considerations (Mackie, Devos and Smith 2000, Smith, Seger and Mackie
2007). According to appraisal theory, individuals appraise goal-relevant stimuli with
reference to the effect of the stimuli, and their ability to cope. The result, is the elicitation
of a particular emotion aimed towards helping them adapt to or cope with the stimuli.
Appraisal theories are consistent with discrete conceptions of emotion, and differentiate
between a variety of positive and negative emotions (Frijda 1987, Lazarus 1991,
Roseman 1996, Roseman and Smith 2001, Scherer 1999).
Appraisal theories of emotion also claim that each emotion has a distinct core
relational meaning, and a set of behavioral predispositions (Lazarus 2001). For example,
anger and anxiety are two negatively valenced emotions, which result from different
appraisal processes and lead to distinct behavioral tendencies. Anger results from
appraisals that there has been an offense against the individual, and produces tendencies
to remove the source of harm from the environment and undo that harm. Anger results
when the individual appraises a threat as manageable. The behavioral tendency is one of
action. Anxiety results from an uncertain or existential danger or threat. As a result, the
behavioral tendency associated with this type of threat is avoidance of potential harm
(Smith and Lazarus 1990).
IET takes appraisal a step further and claims that individuals can make emotional
appraisals, and feel emotions as a result of stimuli, which are goal relevant to groups they
socially identify with (Mackie, Silver and Smith 2004). This theory provides a starting
point for studying the emotions of partisans in the electorate. Much like sports fans,
34

partisans in the electorate cheer their team on, while mourning losses. This leads to the
first hypothesis regarding emotions felt by partisans during election season:

H1: Democrats and Republicans are expected to have distinct emotional profiles;
they are expected to feel positive emotions towards their candidate and negative
emotions towards the opponent.

Looking first to hope for Obama, on a seven-point scale (1=absence of hope,
7=very strong hope) the data demonstrate large and significant partisan differences.
Republicans gave a mean score of 1.51 while Democrats’ mean score was 5.45.
Independents were somewhat hopeful for Obama scoring 3.13. Hope for McCain looked
similar, with the Republican mean equaling 4.59 while Democrats registered only a 1.42.
Again Independents were less hopeful towards McCain in general, registering only a
2.45. This demonstrates that opposing partisans felt virtually all their hope for their
candidate, and virtually none for the opponent.
Levels of enthusiasm also demonstrate this partisan differential, although
Republicans weren’t very enthusiastic about McCain. Levels of enthusiasm for Obama
were 4.51 for Democrats, and only 1.24 for Republicans. Levels of enthusiasm for
McCain were 2.90 for Republicans and 1.22 for Democrats. Independents fell in between
the partisans scoring 2.14 on enthusiasm towards Obama, while registering only a 1.57 in
enthusiasm for McCain. While these differentials appear smaller (particularly
enthusiasm for McCain), what is most striking is the lack of positive emotion felt by
partisans towards the opponent. Democrats were fairly enthusiastic towards Obama, but
35

Republicans felt virtually no enthusiasm at all. While Republicans weren’t very
enthusiastic towards McCain they felt virtually no enthusiasm for Obama.
Although the partisan differential still exists in enthusiasm for McCain it is
interesting to consider some reasons why McCain had difficulty generating enthusiasm
amongst his partisans. First, Republicans may not have been that enthusiastic about
McCain because of his position in the race. McCain was trailing in the polls by a large
amount by the time this survey was conducted in late October. Enthusiasm arises when
things are going well, and is an emotion which encourages continuance of behavior
(Marcus et al. 2000). It is also possible that McCain simply wasn’t good at generating
enthusiasm for his candidacy. Following his unexpected loss in 2000, the argument was
made that Al Gore’s personality should be blamed as he was simply an undesirable
candidate (Fiorina, Abrams and Pope 2003). McCain may have suffered a similar fate in
2008; sources described him as an uninspiring speaker, “ill-suited to lecterns….(and) not
a great orator” (Leibovich 2008). It is possible he was simply unable to arouse
enthusiasm in Republicans due to a lack of campaign acumen.
This trend continues when examining negative emotions felt by partisans. Levels
of fear towards Obama were driven almost entirely by Republicans. Democrats averaged
1.34 while Republicans’ mean fear score was 4.82. Independents fell between the two
again, averaging 2.58. Fear towards McCain was felt almost exclusively by Democrats,
who averaged 3.76, while Republicans averaged only 1.14. Independents again fell
between the two scoring 2.23. Levels of Anger were also highly partisanship dependent.
Anger felt towards Obama averaged 1.31 for Democrats, 2.40 for Independents and 4.60
36

for Republicans. Anger felt towards McCain averaged 4.34 for Democrats, 2.75 for
Independents and 1.54 for Republicans. Again, partisans felt virtually no anger towards
their own party’s candidate (See Table 2.6).

Table 2.6: Partisan Differences in Emotions (1-7 Scale).
Party N Mean F Sig. Paired Sig.
Hope
Obama
Democrats 433 5.45
442.96

***
D/R ***
I/R ***
D/R ***
Independents 101 3.13
Republicans 367 1.51

Hope
McCain
Democrats 433 1.42
275.93

***
D/I ***
I/R ***
D/R ***
Independents 101 2.45
Republicans 368 4.59

Enthusiasm
Obama
Democrats 433 4.51
247.53

***
D/I ***
I/R ***
D/R ***
Independents 101 2.14
Republicans 367 1.24

Enthusiasm
McCain
Democrats 433 1.22
80.43

***
D/I N/S
I/R ***
D/R ***
Independents 101 1.57
Republicans 365 2.90

Fear
Obama
Democrats 432 1.34
264.61

***
D/I ***
I/R ***
D/R ***
Independents 101 2.58
Republicans 363 4.82

Fear
McCain
Democrats 431 3.76
158.94

***
D/I ***
I/R ***
D/R ***
Independents 101 2.23
Republicans 366 1.14

Anger
Obama
Democrats 432 1.31
259.80

***
D/I ***
I/R ***
D/R ***
Independents 100 2.40
Republicans 363 4.60

Anger
McCain
Democrats 431 4.34
160.47

***
D/I ***
I/R ***
D/R ***
Independents 100 2.75
Republicans 366 1.54
Significance levels: * P .05, ** P .01, *** P .001
37

This section paints a picture of partisan members capable of feeling varying
degrees of positive emotions towards their candidate, but virtually no negative emotions.
Virtually all anger and fear were directed towards the opponent. The data provide strong
support for hypothesis #. They also demonstrate that partisans are far more divided over
emotional appraisals of the candidates than they are over ideology and trait evaluations of
the candidates. This is not unexpected. General feelings of favorability are highly
polarized, but as we will explore in the next section, this finding is due in large part to the
role of emotions in determining overall candidate favorability.
In order to compare how divided partisans were on a variety of dimensions, the
scales used to measure each dimension have been standardized, by calculating the mean
difference between Democrats’ and Republicans’ scores, as a percent of the scale. These
scores indicate that Democrats’ and Republicans’ mean ideology scores are 34% apart on
the scale. This difference is substantial, but it is no larger than differences in trait
evaluations of the candidates. Looking to trait evaluations, the largest partisan
differences occur over trust in Obama (48%), patriotism attributed to Obama (42%) and
Obama’s experience (40%). Partisans did not differ by as large an extent over their trait
evaluations of McCain. Trust in McCain was the most divisive (37%), while experience
showed less disagreement (29%) and finally, McCain’s patriotism demonstrated little
division (14%) (See Figure 2.1).

38

Figure 2.1: Scaled Partisan Differences between Ideology and Traits

Partisans did polarize to a significant extent over their emotional appraisals of the
candidates. Again, Obama was the more polarizing figure. Partisans were polarized in
hope felt towards Obama (64%), enthusiasm felt towards Obama (53%), fear of Obama
(58%) and anger felt towards him (53%). However, partisans were still very divided in
the emotions they felt towards McCain, showing large differences in hope (54%), fear
(41%) and anger (44%). There was not as large a partisan difference in enthusiasm for
McCain (29%), mainly because he wasn’t able to generate much enthusiasm amongst
Republicans. An overall comparison of partisan differences on ideology, candidate traits
and emotions shows that partisans are most divided over emotion (See Figure 2.2).

0%
10%
20%
30%
40%
50%
60%
Partisan Differences: Ideology and Traits
% Difference Between
Democrats and Republicans on
Variable Scales
39

Figure 2.2: Scaled Partisan Differences between Ideology and Emotions

Emotion and Favorability
The measurement of candidate evaluations has typically utilized candidate traits
and favorability scales. This chapter has examined these variables, in addition to a less-
used candidate evaluation; emotional appraisals. Overall candidate favorability is
important to any study of polarization, because it is an aggregation of voters’ thoughts
and feelings towards the candidates. This chapter will next examine the impact of voters’
emotions on measures of favorability. Consistent with previous research, this work
expects emotions to impact candidate favorability (Brader 2005, Marcus et al. 2000).
Finally, this work also expects emotional evaluations of the candidates to be statistically
distinct from character trait evaluations (Abelson 1982).
0%
10%
20%
30%
40%
50%
60%
70%
Partisan Differences: Ideology and Emotions
% Difference on Between
Democrats and Republicans on
Variable Scales
40

H2: Emotions are expected to have an impact on candidate favorability

H3: Emotions are expected to be independent from candidate trait evaluations.

As anticipated, candidate favorability and emotional evaluations of the candidates
were strongly correlated. Those who rated a candidate most favorably typically felt
strong favorable emotions towards that candidate and low levels of negative emotion.
Looking first to Obama, the data demonstrate that significant relationships exist between
all four emotions (hope, enthusiasm, fear and anger), and Obama’s favorability. As
expected, hope and enthusiasm were positively related with favorability ratings, while
fear and anger were negatively associated. The relationship between hope and
favorability was particularly striking, as those who felt most favorably about Obama also
felt large amounts of hope (6.42). Those who viewed Obama least favorably felt virtually
no hope (1.10). The relationship between enthusiasm and Obama’s favorability was
similar with those who viewed him most favorably, scoring very high on the enthusiasm
scale (5.61). However, enthusiasm appears to wane for those who only view Obama as
somewhat favorable (2.47), while those who view him as somewhat favorable feel high
amounts of hope towards him (4.54). Those feeling unfavorable towards Obama were
not likely to feel much hope or enthusiasm towards him. Overall, it appears hope was the
more widespread of the positive emotions measured, as even those who rated Obama
“neutral,” felt a degree of hope (2.43), while feeling virtually no enthusiasm (1.69).
Negative emotions felt towards Obama, were inversely related with favorability
ratings. Those who rated Obama most favorably felt almost no negative emotions
41

towards him, while those who rated him least favorably were most likely to feel fear and
anger. Those who felt very favorable or somewhat favorable towards Obama gave very
low fear scores (1.04, 1.18) and very low anger scores (1.04, 1.35). Those feeling
somewhat unfavorable or very unfavorable gave him much higher fear scores (3.52,
5.40), as well as anger scores (3.06, 5.16). This relationship however, does not appear
linear. As favorability decreases, feelings of fear and anger slowly increase, then spike
upwards for those who rate Obama least favorably.
This trend continued for evaluations of McCain; those who felt most favorably
towards him, also felt the most hope (5.74) and enthusiasm (4.50) towards him. However,
while hope was felt broadly by those who favored him even somewhat (4.03), enthusiasm
was not (2.11). This demonstrates that McCain had a more difficult time generating
enthusiasm than hope, and that few people were enthusiastic about him. This is
exemplified by those who gave McCain a neutral favorability rating. While they felt
some hope for him (2.33), they felt almost no enthusiasm towards him (1.36). This may
be due to the fact that McCain as a candidate wasn’t able to generate as much enthusiasm
as Obama. It could also be the case that Republicans weren’t that t enthusiastic about
McCain because they didn’t think he was going to win. They were still able to feel hope,
because hope is the positive emotion likely to be felt when future outcomes are uncertain
(Lazarus 1991, Just et al 2007).
Negative emotions felt towards McCain were also inversely related with his
favorability ratings. Those who rated McCain very or somewhat favorably felt almost no
fear (1.03, 1.10) and little anger (1.14, 1.49). Those who rated McCain very or somewhat
42

unfavorably felt significant fear (4.60, 3.16) and large amounts of anger (5.35, 4.11). This
is to be expected, as he was running behind in the race, and opposing partisans who rated
him unfavorably are likely to feel large amounts of anger (since they feel they are in a
strong position), while being less likely to be fearful. This demonstrates that the
relationship between emotion and candidate favorability is not always linear. Those who
feel somewhat favorable or unfavorable towards a candidate may not feel a large degree
of positive or negative emotions. However, the decision to favor or disfavor one of the
candidates strongly is always fueled by large degrees of emotion (See Table 2.7).

43

Table 2.7: Relationship Between Emotion and Favorability (7 pt. scale)
N Mean F Sig.
Hope
Obama
Very Favorable 314 6.42

814.46

***
Somewhat Favorable 130 4.54
Neutral 67 2.43
Somewhat Unfavorable 65 2.06
Very Unfavorable 326 1.10

Enthusiasm
Obama
Very Favorable 314 5.61

326.13

***
Somewhat Favorable 129 2.47
Neutral 67 1.69
Somewhat Unfavorable 65 1.51
Very Unfavorable 327 1.07

Hope
McCain
Very Favorable 160 5.74

268.08

***
Somewhat Favorable 218 4.03
Neutral 97 2.33
Somewhat Unfavorable 170 1.31
Very Unfavorable 254 1.08

Enthusiasm
McCain
Very Favorable 158 4.50

130.94

***
Somewhat Favorable 218 2.11
Neutral 96 1.36
Somewhat Unfavorable 170 1.08
Very Unfavorable 254 1.05

Fear
Obama
Very Favorable 313 1.04

270.15

***
Somewhat Favorable 130 1.18
Neutral 67 2.25
Somewhat Unfavorable 65 3.52
Very Unfavorable 322 5.40

Anger
Obama
Very Favorable 313 1.04

249.77

***
Somewhat Favorable 130 1.35
Neutral 67 2.00
Somewhat Unfavorable 65 3.06
Very Unfavorable 321 5.16

Fear
McCain
Very Favorable 159 1.03

135.80

***
Somewhat Favorable 218 1.10
Neutral 96 1.81
Somewhat Unfavorable 170 3.16
Very Unfavorable 252 4.60

Anger
McCain
Very Favorable 159 1.14

190.62

***
Somewhat Favorable 218 1.49
Neutral 96 2.01
Somewhat Unfavorable 170 4.11
Very Unfavorable 251 5.35
Significance levels: * P .05, ** P .01, *** P .001
44

It is important to note that the relationship between emotion and favorability
would mean little if the emotions were simply capturing the effects of party identification
and ideology. While the emotions are correlated with these two, regression analysis
indicates that the emotions have a distinct effect on ratings of candidate favorability.
Regressions were conducted on the favorability ratings of both candidates, controlling for
demographics (age, education level, race, income, religiosity, party identification),
ideology and candidate trait evaluations.
Regression analysis was conducted twice, once without the emotions and once
with them in addition to candidate trait evaluations. When the emotions are not included
in the model, traditional predictors of favorability appear most significant. Both McCain
and Obama’s favorability ratings were driven primarily by partisanship. Ideology was
the next strongest predictor, although age and religiosity were both significant as well.
Being older positively impacted McCain’s favorability while lowering Obama’s.
Religiosity was also positively correlated with McCain’s favorability but a negative
predictor of Obama’s. Finally, race had some impact; being African American positively
impacted Obama’s favorability, while being Asian-American was a slight positive
predictor of favorability for McCain.
Regressions conducted with the emotions tell a different story. While party
identification is still a significant predictor of favorability, its relative importance is
greatly diminished. In these models, all of the emotions reach statistical significance,
with hope and anger having the largest impact on favorability. Being hopeful for McCain
had almost as large an impact as one’s party identification on McCain’s favorability.
45

Most interestingly, being angry at McCain was the most powerful emotional predictor,
having a strong negative relationship to favorability for McCain. Hope for Obama had a
far larger impact on Obama’s favorability than party identification, and was about four
times as strong a predictor as any of the other emotions.
Most significantly, this set of regressions indicates that the emotions have an
effect on favorability, independent of candidate trait assessments. This is consistent with
prior work conducted by Abelson et al. (1982). While trust is the strongest determinant
of McCain’s favorability (with party identification, hope and anger towards him close
behind), hope felt towards Obama is the strongest predictor of Obama’s favorability. In
addition to having significant independent effects, the addition of the emotions (and
traits) to the models, produce more robust r-squares. The adjusted r-square for Obama’s
favorability improves from .627 to .878. The model of McCain’s favorability improves
from an adjusted r-square of .531 to .791. This analysis demonstrates how important the
inclusion of emotion is to discussion of candidate favorability and polarization over
candidate favorability. Put simply, emotions are crucial to understanding evaluations of
political candidates. While they are closely tied to party identification and ideology (and
related to candidate trait assessments), they are distinct, and add to the explanatory power
of the models (See Table 2.8).

46

Table 2.8: Regression Candidate Favorability with Emotions and Traits

Favorability
Obama
Favorability
Obama
Favorability
McCain
Favorability
McCain

Demographics

Party ID
Ideology
Age
Religiosity
Education
Income
Gender
Black
Asian
Hispanic

-.571***
-.241***
-.051^
-.052^
.021
.034
.007
.099***
.011
-.033

-.074**
-.042^
.009
-.016
.007
.011
.012
-.009
.036*
.385

.595***
.119**
.080*
.116***
-.045
.013
.013
-.023
.071*
.037

.187***
.000
.006
.039^
-.002
-.012
.019
.024
.068**
-015

Traits

Trust
Experience
Patriotism

-
-
-

.193***
.188***
.113**

-
-
-

.302***
.105**
.023

Emotions

Enthusiasm
Hope
Anger
Fear

-
-

.055*
.269***
-.063*
-.057*

-
-

.053^
.160***
-.213***
-.045^
R2
Adjusted R2
N
.634
.627
536
.883
.878
484
.540
.531
535
.798
.791
495
OLS regression
Significance: ^ p<.1, * p<.05, ** p<.01, ***P<.001
Note: Ordinal logistic regression was performed with consistent results. VIF factors for McCain
regressions were < 3.4. VIF factors for Obama Regressions were <5 except for trust (8.4) and hope
(6.4).

47

Putting the Pieces Together
This chapter has investigated partisan differences across a variety of dimensions
including strength of partisan identification, ideology, candidate favorability ratings,
candidate trait assessments and emotions felt towards the candidates. While those
studying polarization in the electorate often cite ideology or assessments of candidates
and policies, this work looks to emotional appraisals of the candidates. This inclusion
demonstrates that partisans polarize to a greater extent over emotions than all measures
other than candidate favorability. These findings make a strong case for studying this
previously uncharted dimension of polarization. They also are not surprising, given
recent work demonstrating the relationship of a strong relationship between emotion and
candidate choice (Crigler, Just and Belt 2007).
Most importantly, this chapter demonstrates the relative importance of the
emotions to determining favorability ratings of the candidates. Favorability scores can be
seen as a composite measure of overall candidate evaluations; reflecting ideology and
issue positions, evaluations of those candidates particular traits, and the emotions felt
towards the candidates. By modeling candidate favorability with the inclusion of
candidate traits and emotions, we see that emotional assessments of the candidates have
distinct, significant effects on overall candidate favorability. Most importantly, their
relative importance is high with respect to ideology, which is a variable many scholars of
polarization claim to be the driving force behind partisan divisions.

48

Chapter 3 – Campaign Advertising and Emotional Response

The previous chapter charted partisan divisions along a variety of dimensions, and
demonstrated that partisans are as divided, or more divided over emotional assessments
of the candidates, than they are on more traditional measures of polarization such as
ideology and candidate traits. Multiple regression analysis also indicated that the
emotions were reliable and strong predictors of candidate favorability, controlling for
ideology and candidate traits. This leads one to wonder how these emotions arise, and
whether they can be triggered by particular stimuli. This chapter explores the ways in
which common forms of political advertising affect viewers’ emotions.
The data presented in this chapter demonstrate that political advertising has the
ability to both activate and depress emotions in the electorate. The findings also
illuminate the importance of the content of individual ads to their emotional
persuasiveness. Finally, the data show that ad content is processed through the lense of
one’s partisan affiliation. Ads can produce drastically different emotional effects
depending on which party the viewer is aligned with.
This chapter takes a first look at experimental data gathered during the 2008
campaign election season, as part of the Cooperative Campaign Analysis Project
(CCAP), an online public opinion survey. Specifically, it examines how partisans
respond to the threats and reassurances conveyed through campaign ads and their
framing.

49

Campaign Advertising and Emotion
There are multiple, competing theories of emotions in politics. As discussed in
chapter 1, these theories share a common thread; they claim emotions arise during
political contests, and have demonstrable behavioral effects. However, they differ with
respect to the particular mechanisms at work, and the way emotion is conceived (as
valenced, multidimensional or discrete). These nuances were discussed at length in
chapter 1, so we will not spend time rehashing the intricacies of the theoretical debate.
Instead this chapter begins with a review of the central tenets of Intergroup Emotion
Theory (IET), which guides this inquiry. Specifically, IET claims that group identifiers
feel emotions on behalf of the group(s) they identify with, and that these emotions are
aimed towards regulating intergroup behavior (Smith, Seger and Mackie 2007, Mackie,
Devos and Smith 2000). It utilizes a discrete conception of emotions, and is highly
compatible with models of political communication, and emerging views of partisanship
which view partisan membership like a group affiliation or social identity (Green,
Palmquist and Schickler 2002).
The following section reviews contributions of those studying the content and
effects of political campaigns on the emotions. Reviewing this work will help frame
hypotheses regarding campaign advertising and partisan emotions. Previous studies of
campaign effects generally focus on analysis of candidate speaking opportunities, media
coverage of the campaigns, campaign advertisements from the political parties, and most
recently, the use of new media. This body of work examines the premise that campaign
communication matters.
50

Politicians spend millions of dollars annually on political communication because
they believe it will have an impact on their chances for election. However, before the
systematic study of these advertisements, little was known about the way(s) in which
these ads affected voters. This body of work has examined the efficacy of these ads on
political persuasion (Franz and Ridout 2007, Huber and Arceneaux 2007), political
knowledge (Koch 2008, Stevens 2005), candidate impressions (Garramone et al. 1990,
Hichton and Chang 1995), interest in the campaigns, voter turnout (Ansolabehere and
Iyengar 1995, Wattenberg and Brians 1999, Goldstein and Freedman 2002, Lau and
Pomper 2004, Martin 2004, Hillygus 2005), and finally democratic attitudes
(Ansolabehere et al. 1994, Ansolabehere and Iyengar 1995, Jackson, Mondak and
Huckfeldt 2009). Conversely, this project looks to one outcome of political
communication which has received less attention: emotional response.
An emerging literature notes the use of emotion-laden content in political
advertising (in video and print), as well as the ability of this content to elicit emotions in
the electorate. Looking to print advertising, Jerit (2004) utilizes content analysis of
newspaper coverage of the 1988 Canadian federal election campaign to study the
campaign rhetoric employed during the race. She noted that the issues which produced
the most visceral reactions enjoyed the most enduring media coverage. She also found
that supporters and opponents of the FTA relied on anger and fear appeals, with
opponents most heavily reliant. This notes the prevalence of emotional appeals in
advertising, and leads us to expect political ads to elicit emotional responses from
viewers.
51

Valentino et al. (2008) experimentally manipulated political information gathered
from an online “special edition” AP news story. They constructed two versions of the
story; one aimed at generating a high degree of political threat and one geared towards
producing a low degree. They found the more threatening message to produce
significantly more anger, anxiety and disgust than the non-threatening story, while
depressing enthusiasm. This demonstrates that concept that not all political
communication produces equal emotional effects; threats tend to produce more negative
emotions. Similar work by Civettini and Redlawsk (2009) also studies the relationship
between campaign stimuli, emotion and information recall. Simulating a presidential
primary campaign, they found evidence that voters are more likely to remember
information that generates an affective response.
The emotional impact(s) of negative campaigning have also been noted by
scholars utilizing telephone surveys. Fridkin and Kenney (2008) investigate these effects
by having subjects’ listen to a political ad read over the telephone. Their work
differentiates negative ads along two dimensions; relevance and civility. Findings
suggest that relevant and uncivil attack ads produce significantly more negative emotions
than relevant and civil ads, and irrelevant ads. This finding is important because it
demonstrates (along with Valentino’s work) that even without images, political
advertising elicits emotional responses.
Those studying video ads have also made significant contributions. Richardson
(2001) analyzed the effects of two political ads used in the 1990 Michigan U.S. Senate
campaign of Republican Bill Schuette. The first was a comparative attack ad, while the
52

second was a simple assault. He found that the comparative ad produced higher degrees
of negative emotions (anger, sadness and frustration) than the assault ads. This suggests
that attack ads have a larger emotional effect when they compare the sponsor to the
opponent than when they simply attack the opponent. Brader and Corrigan (2007) also
study the effect of negative campaign ads, and claim that distinguishing the particular
emotions the ads are created to elicit is key to gauging their effects. They find that in
many cases, the specific emotional tenor of presidential ad campaigns produces effects on
political participation distinct from those of message tone. They also found that the
observed effects of emotional appeals were overwhelmingly positive with regard to a host
of variables measuring attentiveness to the campaigns, as well as participation in the
campaigns. Brader (2005) also notes the importance of music and imagery to the
emotional effects of televised political advertising. He finds that political ads can change
the way citizens get involved and make choices simply by using images and music to
evoke emotions. Fear and enthusiasm cues in images and music were effective. This is
especially helpful, as all of the ads utilized in this study (and most of the ads aired during
the campaign) were wrought with emotion eliciting images and music. The creates an
even stronger expectation that ads will elicit emotion.
Sullivan and Masters (1998) examined the impact of candidates’ facial gestures
on viewers’ emotions and attitudes towards the candidates. In general, they found both
happy/reassuring displays and neutral displays to increase the strength of positive
emotions felt towards the candidates. These displays also impacted subjects’ attitudes
towards the candidates, even when controlling for party id, issue agreement and
53

assessment of leadership ability. These effects were evident even when the video
excerpts were viewed without audio. Similar work by Bucy (2000) tested the impact of
presidential reactions to images of compelling news events, on viewers’ emotional
reactions and trait evaluations. He found that inappropriate leader displays resulted in
higher negative emotions and lower positive emotions, as well as lower trait scores
than emotionally appropriate leader displays. Interestingly, negative leader displays were
also evaluated as more honest, credible, trustworthy and appropriate than positive
displays. These findings demonstrate that even the gestures and reactions of candidates
are capable of producing emotional reactions amongst viewers.
Finally, researchers have examined the strategic use of emotional appeals in
election campaigns. Ridout and Searles (2011) develop and test hypotheses regarding
when candidates use certain emotional appeals and which candidates are likely to do so.
They examined 631 ads aired in Senate races during 2004. They found trailing
candidates significantly more likely to use fear, and leading candidates to utilize
enthusiasm and pride appeals. They also found that enthusiasm did not spike at the end
of the campaigns to bring people to the polls.
This body of research has provided us with some key insights; political
communication through a variety of mediums is often emotionally charged and produces
a wide range of emotional effects. The particular content of the communication (such as
tone or civility) has an impact on the emotions as well. Finally, the emotions elicited by
these messages are associated with certain attitudinal and behavioral effects. However,
this work has left a variety of avenues of inquiry untested. For example, political
54

scientists have long noted the pervasive effects of partisanship (Campbell et al. 1960,
Converse 1964) on political evaluations, yet little work has looked at the effects of
partisanship on the emotions. This body of research also focuses mainly on emotions felt
in general, without taking into account the object of the emotion. This is an important
distinction, because during an election it is important to know which candidate receives a
voter’s emotional wrath/admiration. Finally, little work has tested the effects of real-
world televised campaign ads on voters’ emotions and subsequent evaluations of the
candidates (Crigler, Mills and Hootstein 2009). This project attempts to fill these gaps in
the literature.

Hypotheses
This chapter’s hypotheses are derived from IET. It serves as a guide to
understanding how partisan identifiers in the electorate react emotionally to campaign
stimuli. IET claims that people feel emotions as members of groups, when social
identities are made salient. This leads to the first set of hypotheses, that partisan
identifiers view these ads through partisan colored glasses and as a result, react distinctly
from independents and opposing partisans.

H1: Partisans will react to ads in emotionally distinct ways: hope and
enthusiasm are expected to be elicited from partisans of the ad sponsor’s party,
and not by opposing partisans.

H2: Partisans will react to ads in emotionally distinct ways: anger and fear are
expected to be elicited from opposing partisans, and not by partisans of the ad
sponsor’s party.

55

As discussed in chapter 1, IET also predicts how partisans will react emotionally
to a variety of stimuli. IET focuses on threat and reassurance, and claims that reassuring
messages (like those found in appeal ads) generally produce positive emotions such as
enthusiasm and hope, while threatening messages (like those found in attack ads from the
opponent) generally elicit negative emotions such as fear and anger (Huddy and Mason
2008). This leads to the expectation that positive ads from the ingroup candidate will
produce more hope and enthusiasm, while attack ads from the outgroup candidate should
produce more anger and fear. This leads to the next hypotheses:

H3: Appeal ads (from the ingroup candidate) are better suited to elicit hope and
enthusiasm, than attack ads.

H4: Attack ads (from the outgroup candidate) are better suited to elicit anger
and fear than appeal ads.

Finally, IET helps generate predictions regarding the particular negative or
positive emotions ads are likely to generate. For example IET claims that in response to
threatening messages, those whose ingroup is in the “strong” position are likely to react
with more anger, while those in the “weak” position are likely to react with more fear
(Mackie, Devos and Smith 2000). Because Democrats were leading in the race they are
expected to feel more anger than fear compared to Republicans. With regard to
reassuring messages, IET predicts members of the party in the stronger position will be
more likely to feel enthusiasm. Members of the party in the weaker position may still feel
positive emotions, but hope is a more likely outcome than enthusiasm. Hope is a positive
emotion individuals often feel in the fact of uncertainty (Just et al. 2007). To quote a
56

commonly used colloquialism, Republicans will likely hope for the best, while expecting
the worst. The final two hypotheses are:

H5: Attack ads from the outgroup candidate will elicit anger and fear in both
Democrats and Republicans, with Democrats expected to feel more anger, and
Republicans more fear.

H6: Appeal ads from the ingroup candidate will elicit more enthusiasm from
Democrats than Republicans. Republicans may still feel positive emotions, but
hope is more likely.

Partisan Emotions

As predicted by hypotheses 1 and 2, partisans’ reactions to the ads were distinct.
Specifically, these hypotheses claim that partisans will feel hope and enthusiasm and not
anger or fear for their ingroup candidate, following ads from that candidate. They are
also predicted to feel anger and fear (and not hope or enthusiasm) for the outgroup
candidate following ads from that candidate. Looking to all Obama ads, we see that
Democrats feel both hope (4.53) and enthusiasm (3.62) to a significant degree, while
Republicans feel virtually no hope for Obama (1.29) or enthusiasm (1.21). Also as
expected, Obama’s ads did not induce Democrats to feel fear or anger towards him, but it
did produce these reactions in Republicans. Democrats felt fear (1.41) and anger (1.32)
in small amounts, while Republicans felt fear towards Obama (3.54) as well as anger
(3.96).
McCain’s ads produced similar results, as Democrats and Republicans reacted
quite differently. Viewing McCain’s ads produced a good deal of hope (3.86) and some
enthusiasm (2.76) for McCain amongst Republicans. However, these same ads produced
57

virtually no hope for McCain (1.51) or enthusiasm for McCain (1.21) amongst
Democrats. Regarding the negative emotions, Democrats and Republicans responded
quite differently again. Republicans were not fearful (1.00) or angry (1.25) towards
McCain but Democrats did feel some fear (2.62) and a fair amount of anger (3.45)
towards McCain following McCain’s ads. These data provide support for hypotheses 1
and 2, and suggest that IET does provide a good framework for understanding how
opposing group members view campaign communication; they tend to polarize
emotionally as a result of these ads (see Table 3.1).
Table 3.1 Impact of Ad Sponsor on Partisans’ Emotions (1-7 scale)
Hope
Obama
Hope
McCain
Enthusiasm
Obama
Enthusiasm
McCain
Fear
Obama
Fear
McCain
Anger
Obama
Anger
McCain
All Obama
Ads

Democrats
Republicans

N
Sig.

4.53
1.29

281
***

1.21
3.06

281
***

3.62
1.21

280
***

1.18
2.31

282
***

1.41
3.54

281
***

2.74
1.08

280
***

1.32
3.96

279
***

3.05
1.17

280
***
All McCain
Ads

Democrats
Republicans

N
Sig.

3.70
1.08

269
***

1.51
3.86

268
***

3.02
1.04

269
***

1.21
2.76

270
***

1.28
3.92

266
***

2.62
1.00

267
***

1.17
3.04

265
***

3.45
1.25

267
***
Significance: ^ p<.1, * p<.05, ** p<.01, ***P<.001

Attacks vs. Appeals
Hypothesis 3 states that positive ads from the ingroup candidate are better geared
towards generating enthusiasm, than attacks from that candidate. In general, the data
support this hypothesis. Comparing the hope and enthusiasm generating abilities of
Obama’s ads, it’s clear that appeal ads are more effective at eliciting these two emotions.
58

Democrats’ levels of hope for Obama were very high (5.34) after viewing an Obama
appeal ad, and significantly lower (3.83) after viewing one of his attack ads. The same
was true for enthusiasm, as Obama’s appeal ads produced a high level of enthusiasm
(4.42), while his attacks produced far less (2.37). These differences are statistically
significant and substantively meaningful (see Table 3.2).
Interestingly, the same trend does not hold for Republicans. Positive emotions
towards McCain were not generated any more effectively through positive advertising.
Republicans’ mean hope levels for McCain were fairly high after viewing a McCain
appeal ad (4.02), but were equally high after viewing one of McCain’s attacks (3.74).
The same thing was true of enthusiasm felt towards McCain; levels were equally high
after viewing one of his positive ads (2.93) as they were after viewing one of the negative
ads (2.62). This suggests that while Democrats are significantly more reassured by
positive messages than negative ones from Obama, Republicans are equally reassured by
both. This is surprising from the standpoint of IET, which suggests that positive
messages should provide more reassurance than negative ones (Huddy and Mason 2008).

59

Table 3.2: Ad Tone and Positive Emotions
Hope for Ingroup
Candidate
Enthusiasm for
Ingroup Candidate
Democrats

Obama Appeal
Obama Attack

N
Sig.

5.34
3.83

158
**

4.42
2.90

157
**
Republicans

McCain Appeal
McCain Attack

N
Sig.

4.02
3.74

124
N/S

2.93
2.62

124
N/S
Significance: ^ p<.1, * p<.05, ** p<.01, ***P<.001

Looking to the negative emotions (hypothesis 4), the data provide evidence that
viewing attack ads from the opponent is more conducive to eliciting anger and fear, than
positive ads from the opponent. Democrats become significantly angrier with McCain
after viewing a McCain attack (4.73), than a McCain appeal (2.42). They do not become
significantly more fearful though after an attack ad (2.98) than an appeal ad (2.42). This
is to be expected as Democrats were generally not all that fearful of McCain at the time
of this survey (see ch.. 2). Republicans were significantly angrier with Obama after
viewing an Obama attack ad (4.61) than one of Obama’s positive ads (3.35). However,
they were more fearful after viewing a positive Obama ad (4.00) than a negative Obama
ad (3.00) (see Table 3.3). This finding is not necessarily unexpected. Republicans were
losing in the polls at the time of this experiment, and as a result, may have viewed a
60

positive Obama ad as even more threatening than an Obama attack ad. Clearly, they
were more sensitive to threat than Democrats, because Democrats were not as threatened
by McCain’s appeal ads. In conclusion, Democrats and Republicans were both angered
by attack ads from the opponent. Both also felt less fear than anger. Most striking
though, is that Obama’s attacks provoked less fear (but more anger) than his positive ads.
When attacked, Republicans, although losing in the race and in the weaker position,
lashed out with anger and diminished fear.

Table 3.3: Ad Tone and Negative Emotions
Fear Towards
Outgroup Candidate
Anger Towards
Outgroup Candidate
Democrats

McCain Appeal
McCain Attack

N
Sig.

2.42
2.98

157
N/S

2.42
4.73

157
*
Republicans

Obama Appeal
Obama Attack

N
Sig.

4.00
3.03

124
N/S

3.35
4.61

122
^
Significance: ^ p<.1, * p<.05, ** p<.01, ***P<.001

Differences between Democrats and Republicans
The electoral context leading up to this experiment, was one in which Obama was
clearly leading in the polls. According to intergroup emotion theory, group identifiers are
likely to have differential reactions to threats and reassurances when one group is in a
61

stronger position vis-a-vi the opposing group. In this case, Democrats were in the
stronger position, and thus are expected to respond to threat (attack ads from the
opponent) with anger more than fear. Republicans, who were in the weak position are
expected to react with more fear and less anger than Democrats (Mackie, Devos and
Smith 2000). The position of the parties in the race is also expected to play a part in the
elicitation of enthusiasm. Reassuring messages are more believable if group members
feel confident in the position of their group. As a result, Democrats are expected to be
more receptive to enthusiasm appeals. Republicans, however are not likely to have the
same degree of confidence in their chances, and enthusiasm appeals made through
reassuring messages are less likely to have their desired effect (Huddy and Mason 2008).
Hope is a unique positive emotion (Just et al. 2007). It often arises as a result of
reassuring stimuli, but it is associated with a degree of doubt over the outcome. Hope is a
positive feeling which arises when people are unsure whether their goal can be attained
(Lazarus 1991). It is often felt when individuals want something to happen, and they are
still holding onto positive feelings that it can happen, but the outcome is uncertain (Just et
al. 2007, Huddy and Mason 2008). This is the position Republicans were in during
October 2008; they were trailing in all of the polls by a substantial margin, but still had a
chance of winning. In this situation, Republicans can be expected to react to positive
campaign messages with more hope than enthusiasm.
Regarding the negative emotions, IET claims that when group-members feel their
group is in a strong position, they are more likely to respond to threatening messages with
anger. Fear is likely to arise when resources are perceived as low and the threat seems
62

overwhelming (Mackie, Devos and Smith 2000, Huddy and Mason 2008). Republicans,
who were in a weaker position are expected to feel more fearful than Democrats and less
angry.
The results of the experiment provide affirmation for hypothesis 6, but only
partial support for hypothesis 5. When exposed to positive ads from the ingroup
candidate, Democrats felt significantly more enthusiasm (4.42) towards Obama, than
Republicans felt towards McCain (2.93). As predicted, Republicans were also more
likely to feel hope (4.02) than enthusiasm (2.93). Regarding hypothesis 5, Democrats
were more likely to respond to attack ads with anger (4.73) than fear (2.98). However,
Republicans were as well. They felt large amounts of anger (4.61) and only moderate
fear (3.03) (see Table 3.4). This indicates that the accuracy of IET’s predictions may rest
on variables other than the tenor of the message and the group’s position in the contest.
Specifically, groups may possess attributes that make them dissimilar from one another.
In order to accurately predict the emotions group members are likely to feel in response
to threat and reassurance, IET may need to take these differences into account.
Scholars such as George Lakoff and John Jost acknowledge these differences,
although they have differing perspectives. Lakoff (1996), identifies core differences
between the moral systems underlying liberalism and conservatism and claims that
conservative morality resembles a “strict father” while liberal morality resembles a
“nurturant parent.” While the nurturant parent morality lends itself to compassion and
understanding, the strict father model appears stern and unforgiving. Jost et al. (2003)
also find psychological differences between liberals and conservatives and claim that
63

several psychological variables predict political conservatism. This list includes
dogmatism and intolerance of ambiguity, lack of openness to experience, uncertainty
avoidance and a personal need for order. It is easy to imagine that individuals possessing
these traits may respond more angrily to threatening political messages than those who do
not. These psychological differences between liberals and conservatives lead us to
question the interchangeability of groups proposed by intergroup emotions theory. While
this assumption lends parsimony to the theory, it may be important to take into account
psychological differences between groups (arising from core beliefs and historical
context) when studying intergroup emotions.

Table 3.4: Video Treatments and All Emotions Felt Towards the Candidates
Hope
Obama
Hope
McCain
Enthusiasm
Obama
Enthusiasm
McCain
Fear
Obama
Fear
McCain
Anger
Obama
Anger
McCain
Democrats

McCain Appeal
McCain Attack
Obama Appeal
Obama Attack

N
Sig.

3.41
4.06
5.34
3.83

301
***

1.65
1.32
1.27
1.17

302
^

3.26
2.71
4.42
2.37

300
**

1.37
1.00
1.30
1.07

302
^

1.31
1.25
1.51
1.31

298
N/S

2.42
2.98
2.58
2.88

298
N/S

1.26
1.05
1.22
1.42

298
N/S

2.42
4.73
2.50
3.54

298
***
Republicans

McCain Appeal
McCain Attack
Obama Appeal
Obama Attack

N
Sig.

1.07
1.09
1.41
1.17

249
N/S

4.02
3.74
3.25
2.85

247
^

1.00
1.07
1.09
1.00

249
N/S

2.93
2.62
2.49
2.10

250
N/S

3.23
4.50
4.00
3.03

249
*

1.00
1.00
1.08
1.08

249
N/S

2.48
3.50
3.35
4.61

246
**

1.17
1.32
1.09
1.25

249
N/S
Significance: ^ p<.1, * p<.05, ** p<.01, ***P<.001

64

In conclusion, this chapter examined ways in which political emotions arise.
Analyzing experimental data demonstrated the emotional impact(s) of campaign
advertising on partisans’ emotions. In particular, it showed that political advertisements
had significant effects on the emotions viewers felt towards the candidates, and that the
tenor of the ads mattered; appeal ads were better at generating hope and enthusiasm
amongst Democrats, while attack ads more effectively elicited anger and fear. However,
not all of IET’s predictions regarding the particular emotions likely to be aroused by the
ads were met; while Democrats did feel more enthusiasm than Republicans, they did not
feel more anger. Republicans, when threatened, responded with more anger than
predicted.

65

Chapter 4 – Ad Content and Horserace Framing

This chapter takes a closer look at how the context and content of political
campaigns affect voters’ emotional appraisals. First, the effects of candidate ads focused
on issue concerns are compared with those focused on personal traits. These are the two
most common topics of political advertising (Just et al. 2009, Kaid and Johnstone 2001),
but it is unclear whether they have an equivalent impact on viewers’ emotions. Building
on findings from chapter three, this chapter also asks how ad topic interacts with ad tone
to produce specific emotions in the electorate. Are there certain combinations that work
most effectively to generate particular emotions? In addition, this chapter takes into
account that in a real-world election campaign, voters process ads within the greater
electoral context. This context is usually described by the horserace metaphor; the media
trope telling voters who’s ahead and who’s behind. Because of the prevalence of this
metaphor in media coverage of the campaigns, it is crucial to include an indicator of the
horserace in this experiment. As a result, this chapter also examines the interaction
between the content of the ads, and the horserace context of the daily campaign news.

Issue and Trait Appeals in Political Advertising
There is a longstanding tradition of scholars who examine the content of
campaign advertising. Their work has examined both print media, internet and television
media (Crigler et al. 2011, Goldstein 2002, Jameison 1996, Just et al. 2009), and
measured variables such as topic, tone and subject matter. This scholarship has also
spawned a body of work which explores the differences between positive and negative
66

advertising, both in content (Geer 2006, Johnston and Kaid 2001) and effect
(Ansolabehere et al. 1994, Ansolabehere and Iyengar 1995, Lau et al. 1999, Wattenberg
and Brians 1999). This section looks specifically to examples which study the ways issue
positions and personal traits are utilized in the ads. While a fair amount of scholarship
focuses on this concern, little has been written discussing the impact of these variables on
the emotions.
Those studying differences between issue and trait appeals have examined a
variety of topics. Some have investigated the types of ads that are likely to focus on issue
or trait concerns (Valentino et al. 2006), while others have focused on party differences
(Benoit 2004, Spiliotes and Vavreck 2002) and gender differences in the use of these
appeals (Kahn 1993). Researchers have compared issue and image ads on a variety of
dimensions, including tone (Johnston and Kaid 2002) and candidate impressions (Kahn
and Geer 1994). Finally, political strategists have illuminated the importance of
conveying candidate trait information, even in communication speaking to the issues
(Medvic 2008). This focus on traits appears to be echoed by newsmedia, as voters learn
more about candidate traits than they do about candidate issue stances from media
(Weaver 1996). However, little work has focused on whether campaign communication
focused on candidate traits is more likely to have an emotional impact than messaging
appealing to the issues? Gross (2008) gives us reason to suspect this may be true, as she
finds episodic frames generate more emotional interest than thematic frames. Episodic
frames focus on personal characteristics to a greater extent than thematic frames (which
focus on general themes). However, other scholars find issue ads have a greater impact
67

than trait ads on a variety of measures (Kahn and Geer 1994). This section will review
the current literature examining the use of issue and trait themes in campaign advertising,
in order to generate hypotheses regarding the emotional impact(s) of these ads.
Geer (2006) examines when candidates are likely to use issue and trait ads. His
findings are supported by a massive data collection project, which involved conducting a
thorough content analysis on virtually every televised ad from the major presidential
candidates between 1960 and 2000. The data lead Geer to conclude that negative ads
contain more information than positive ads. He argues that occurs because negative
charges require more support than positive claims; attacks must be substantiated, or they
are judged as unfair and can lead to backlash against the sponsor. One piece of evidence
that helps him substantiate this claim was the fact that negative ads were more likely to
discuss issues, while positive ads are more likely to be about the candidates’ personal
qualities. While this helps explain when candidates are likely to use issue and trait ads,
but does not help predict the emotional effects of issue and trait ads.
Benoit (2004) investigates party differences in campaign communication, and
explains which issues and traits Democrats and Republicans utilize in the campaigns.
Looking to primary and general election television ads, primary debates, general debates
and acceptance addresses ranging from 1952 to 2000, he claims that significant
differences exist between Democrats and Republicans. First, while both parties
emphasize issues, they don’t reference all issues with the same frequency; Republicans
are more likely to stress “Republican” issues such as defense more often than Democrats.
68

Regarding traits, Republicans emphasized sincerity and morality more than Democrats,
and empathy and drive less.
Spiliotes and Vavreck (2002) examine differences in campaign advertising
between Democrats and Republicans as well, and cited the frequency with which the
parties utilized issue and trait ads. Data were gathered during the 1998 midterm
elections, and content analysis was conducted on over 1,000 ads aired by 290 candidates
across 37 states. The data demonstrate that while most ads contained some sort of issue
appeal, only about half were predominately issue driven (p. 253). Even though issue
appeals were widespread, very few of these ads adopted a specific position on the issues.
Interestingly, ads that were issue-dominant were more numerous than trait-dominant, but
a majority of ads contained a mixture of the two.
Kahn (1993) explores the differences between male and female candidates’
televised political advertisements, and notes that while they stress different issues, they
tend to stress the same personality traits, and in fact do so more often. Johnson and Kaid
(2002) directly compare issue and image ads, and find distinct stylistic differences
between them. However, they do not test the outcomes of these differences on voting
behavior. Based on a sample of 1,365 televised ads from the 1952 to 2000 U.S.
presidential election campaigns, they found that issue ads tend to have candidates speak
for themselves using emotional language, while image ads tend to rely on anonymous
commentators as the dominant speaker and utilize source credibility appeals. They also
found issue ads to be more likely to be negative than image ads.
69

Directly related to Johnson and Kaid’s work, is that by Kahn and Geer (1994)
which utilizes an experimental design to compare the effectiveness of multiple types of
political ads at creating positive impressions among voters. They look specifically at the
impact of ads that focus on issues and those that stress traits on candidate favorability.
They found no difference in the impacts of issue and trait ads when those ads were
positive, but in negative ads, attacks based on traits were less effective than issue-based
attacks. This provides some evidence that contrary to Gross’s (2008) findings, issue
based ads may be more effective under certain circumstances.
This past research demonstrates that trait and issue appeals are both utilized, but
in different ways by the differing political parties, and male and female candidates. This
literature also demonstrates clear differences between issue and trait ads. However, this
research does not reach a consensus regarding the effectiveness of issue and trait ads, or
the emotional impacts of these types of ads. Kahn and Geer (1994) found issue attacks to
have a greater impact on candidate favorability than character attacks (when ads were
negative), while Gross (2008) found episodic frames (which focus on individualized
stories rather than general issues) to be more emotionally interesting. This leaves the
question, “are trait or issue ads more emotionally evocative?” However, we are also left
to wonder how issue and trait advertising interacts with ad tone. Kahn and Geer’s (1994)
findings lead us to believe that issue ads may be more emotionally persuasive than trait
ads, when those ads are negative. While this chapter compares issue and trait ads, it is
important to note that issue ads almost always make some reference to candidate traits
(Spiliotes and Vavreck 2002). The issue ads examined may make reference to candidate
70

traits, but they are clearly different from the trait ads, which do not make any mention of
issue positions. This leads to the first set of hypotheses:

H1: Candidate issue ads and trait ads will impact subjects’ emotions in different
ways; trait appeals will be more emotionally evocative in terms of the amount of
emotion generated.

H2: Issue ads and trait ads are not expected to interact equally with ad tone;
attacks are expected to be more emotionally evocative when focused on the issues
than on candidates’ personal traits.

Results: Issue vs. Trait Ads
Hypotheses 1 and 2 refer to the impact of ad topic on emotional persuasiveness.
The dependent variable in this analysis is the amount of various emotions (hope,
enthusiasm, anger, fear) the ads are able to generate. This is measured on a 1-7 scale, 1
being absence of emotion and 7 being strong emotion. Results are self-reported. The
first level of analysis compares the impact of issue and trait ads on partisans’ emotions.
Ignoring ad sponsor and ad tone, issue and character ads appear to influence Democrats
differently than Republicans.
The positive emotions Democrats felt towards Obama appear related to the type
of ad they viewed. Issue ads produced significantly more hope and enthusiasm towards
Obama than character ads. Democrats mean hope score was 4.47 after viewing any issue
ad, and only 3.81 after viewing a character ad. Their mean enthusiasm score towards
Obama was 3.68 after viewing any issue ad, while only 2.99 after viewing a character ad.
Both differences were significant at the p < .05 level. Democrats’ negative emotions
71

were unaffected, and there were no significant differences in Republicans’ emotions (see
Table 4.1).
Regardless of ad sponsor, or tone, Democrats became significantly more hopeful
and enthusiastic towards Obama whenever the issues were mentioned. This indicates that
talk of the issues makes them feel good about Obama, regardless of who was doing the
talking, or how. This is interesting and leads one to speculate that Democrats felt very
good about the issue positions of their candidate, and that these good feelings were even
reinforced when McCain attempted to speak on the issues. These findings suggest that
Democrats felt they had ownership of the issues in 2008. This relationship does appear
when we look at Republicans’ feelings of hope and enthusiasm towards McCain, but the
differences aren’t as large, and they do not reach statistical significance. These data
provide partial evidence that issue and character ads impact viewers’ emotions to
differing degrees.

Table 4.1: Issue vs. Trait Ads
Hope
Obama
Hope
McCain
Enthusiasm
Obama
Enthusiasm
McCain
Fear
Obama
Fear
McCain
Anger
Obama
Anger
McCain
Democrats

Issue Ad
Trait Ad

N
Sig.

4.47
3.81

301
*

1.40
1.30

302
N/s

3.68
2.99

300
*

1.24
1.15

302
N/S

1.40
1.30

302
N/S

2.69
2.68

298
N/S

1.20
1.30

298
N/S

3.23
3.25

298
N/S
Republicans

Issue Ad
Trait Ad

N
Sig.

1.11
1.26

249
N/S

3.70
3.21

247
N/S

1.04
1.04

249
N/S

2.68
2.39

250
N/S

3.84
3.61

249
N/S

1.04
1.04

249
N/S

3.57
3.42

249
N/S

1.17
1.26

249
N/S
Significance: ^ p<.1, * p<.05, ** p<.01, ***P<.001
72

Deeper examination into hypothesis 1 demonstrates that the topic of political ads
still matters when the sponsoring candidate is taken into account. IET leads us to expect
reassuring ads from the ingroup candidate to produce enthusiasm and hope, while
threatening ads from the outgroup candidate would produce anger and fear. However,
IET does not speak to the differences between the content of these ads and their
effectiveness at inducing threat and reassurance. The data indicate that Democrats
responded with more hope to Obama’s issue ads than his character ads. As evidenced
earlier, they also responded with more enthusiasm. Democrats’ mean Obama hope score
was 5.02 following an Obama issue ad, and 4.03 after an Obama character ad. These
differences were significant at the p < .1 level. Democrats’ mean Obama enthusiasm
scores were 4.09 after an Obama issue ad, and 3.13 after an Obama character ad. This
difference did not reach significance, but should still be noted, as the directionality is
consistent with hope. The topic of ads didn’t impact the levels of negative emotions
Democrats felt towards McCain regardless of ad sponsor.
Looking to Republicans, yields a similar trend. McCain’s positive issue ads
generated more hope and enthusiasm for McCain than positive ads about character traits.
Republicans shown McCain’s positive issue ads scored 4.41 and 3.08 on hope and
enthusiasm respectively for McCain, while those who viewed the trait ads scored 3.28
and 2.44. The differences in hope were significant at the p < .1 level, while the
differences in enthusiasm did not reach statistical significance (see Table 4.2). This
mirrors the findings on the Democratic side. Negative emotions towards Obama were
unaffected by the difference in character and trait ads. These findings indicate support
73

for hypothesis 1, and demonstrate that ads focusing on the issues are more emotionally
persuasive than trait ads. These findings also reinforce literature which suggests
differences between Democrats and Republicans run deeper than ideology.

Table 4.2: Issue vs. Trait Ads by Candidate
Hope
Obama
Hope
McCain
Enthusiasm
Obama
Enthusiasm
McCain
Fear
Obama
Fear
McCain
Anger
Obama
Anger
McCain
Democrats

Obama Issue Ad
Obama Trait Ad

McCain Issue Ad
McCain Trait Ad

N
Overall Sig.

5.02
4.03^

3.83
3.58

301
**

1.21
1.22

1.63
1.39

302
N/S

4.09
3.13

3.22
2.84

300
*

1.22
1.13

1.26
1.16

302
N/S

1.51
1.30

1.26
1.31

302
N/S

2.70
2.78

2.68
2.57

298
N/S

1.28
1.38

1.13
1.21

298
N/S

3.04
3.06

3.46
3.44

298
N/S
Republicans

Obama Issue Ad
Obama Trait Ad

McCain Issue Ad
McCain Trait Ad

N
Overall Sig.

1.13
1.47

1.09
1.06

249
*

2.98
3.14

4.41
3.28^

247
**

1.00
1.10

1.08
1.00

249
N/S

2.28
2.33

3.08
2.44

250
N/S

3.70
3.37

3.98
3.85

249
N/S

1.08
1.08

1.00
1.00

249
N/S

3.95
3.97

3.18
2.90

246
^

1.00
1.36

1.34
1.16

249
N/S
Significance: ^ p<.1, * p<.05, ** p<.01, ***P<.001
Note: Pairs that are significantly different have a significance marker next to them.

Finally, this section examines how ad substance interacts with ad tone. Are
appeal ads more emotionally persuasive when they focus on the issues rather than
character traits? Do attack ads have a greater emotional impact when they focus on the
issues rather than character traits? Although few findings held statistical significance
(likely due to small sample size), the data demonstrate that appeals may generate more
hope and enthusiasm when issue based. Attacks do not appear more emotionally
arousing when issue based versus trait based. Democrats mean hope scores were 5.73
74

for appeal issue ads, and 4.77 for appeal trait ads. Their mean enthusiasm scores were
4.97 for appeal issue ads and 3.60 for appeal trait ads. While differences in hope and
enthusiasm for Obama between issue and trait appeal ads were not statistically
significant, the differences in means are consistent with previous findings. Differences in
emotional response to issue and trait attacks were smaller.
The impact of issue vs. trait focus in McCain’s ads on positive emotions was also
noticeable. McCain’s issue appeal generated a mean hope score of 5.07, while the
character appeal generated a mean of only 2.81. This difference was significant at the p <
.01 level. Enthusiasm for McCain also appeared to be more readily generated by
McCain’s issue appeal than his character appeal. Subjects exposed to the issue appeal
reported a mean enthusiasm score of 3.67, while those who viewed the character appeal
averaged 2.14. This difference did not reach statistical significance, but it reinforces the
trend in the data demonstrating the increased efficacy of issue appeal ads in positive
emotional arousal (see Table 4.3). Again, differences were less substantial for negative
emotions elicited by Obama’s issue and trait attacks.

75

Table 4.3: Issue vs. Trait Ads by Tone and Candidate
Hope
Obama
Hope
McCain
Enthusiasm
Obama
Enthusiasm
McCain
Fear
Obama
Fear
McCain
Anger
Obama
Anger
McCain
Democrats

Obama Appeal Issue
Obama Appeal Trait

McCain Attack Issue
McCain Attack Trait

5.73
4.77

4.03
4.10

1.27
1.27

1.18
1.46

4.97
3.60

2.90
2.50

1.41
1.13

1.00
1.00

1.61
1.37

1.18
1.33

2.23
3.10

3.00
2.97

1.36
1.00

1.00
1.10

2.50
2.50

5.12
4.30
Republicans

McCain Appeal Issue
McCain Appeal Trait

Obama Attack Issue
Obama Attack Trait

1.00
1.14

1.00
1.38

5.07**
2.81

2.94
2.73

1.00
1.00

1.00
1.00

3.67
2.14

2.42
1.69

3.52
2.93

3.24
2.77

1.00
1.00

1.00
1.19

2.61
2.36

4.90
4.23

1.00
1.36

1.00
1.57

Significance: ^ p<.1, * p<.05, ** p<.01, ***P<.001
Note: Hypotheses only refer to issue vs. trait paired comparisons so overall significance is not
provided.

The data presented in this section support both hypothesis 1 and hypothesis 2. In
support of hypothesis 1, the data indicate that issue ads are more emotionally persuasive
than ads focusing on candidate traits. For both candidates, issue ads were more effective
at generating positive emotions than trait ads. These differences were most distinct
amongst partisans when the ad was positive in tone, and originated from the candidate of
their party. The data also provided limited support for hypothesis 2; that content (issue
vs. trait) interacts in a distinct way with ad tone (attack vs. appeal). Issue ads were more
likely to generate higher positive emotion scores (Republicans’ hope was highly
significant), but few differences reached statistical significance. Differences between
issue and trait ads had a smaller (and insignificant) impact on anger and fear felt towards
the opponent.
76

These findings are not surprising. Geer (2006) provides reason to believe that
issue ads have more factual content, and that they are more useful than trait ads,
especially when going negative. While Geer does not study the emotions, it is believable
that more factual content in ads would result in a stronger emotional response. However,
as Brader (2005) notes, factual content isn’t the only source of emotional stimuli. Brader
finds music and images to have a significant effect on viewers’ emotions. Gross (2008)
also claims that news stories were more emotionally evocative when focusing on episodic
frames (personalized stories) than thematic frames (factual, general accounts of an issue).
Trait ads clearly have more personalized information, rather than facts about the issues of
the election. This set up competing hypotheses to test, as there was reason to believe
each type of ad may be more emotionally effective. While this analysis demonstrates that
issue ads were better at generating hope and enthusiasm, further examination is
warranted.

The Horserace Frame
Political contests do not occur in a vacuum. In fact, news coverage of the
campaigns is so extensive that it is often referred to as unpaid advertising (Ansolabehere
and Iyengar 1994). The bulk of this coverage however, tends to focus on the horserace;
who’s winning and who’s losing (Farnsworth and Lichter 2006, Graber 2000, Patterson
1993), even in local news broadcasts (Stevens et al. 2006), and in local coverage of
Canadian elections (Mendelsohn 1993). The horserace frame is especially prevalent in
coverage of primary elections (King 1990), and is also used frequently by political
77

scientists and other political experts in election commentary (Brewer and Sigelman
2002). While scholars often bemoan this frame for crowding out other types of political
reporting (Broh 1980, Iyengar, Norpoth and Hahn 2004, Lawrence 2000), its impact has
been gauged infrequently, although notable exceptions exist (Baum et al. 2010, Mutz
1995, Rhee 1997).
Before examining the scholarship on horserace framing, it is useful to first review
the concept of framing in general and the way it has been defined and studied by
researchers. Framing theory begins from the premise that an issue can be viewed from a
variety of perspectives, which impact one’s understanding of the implications and
considerations that issue brings to mind. Framing specifically refers to, “the process by
which people develop a particular conceptualization of an issue or reorient their thinking
about an issue” (Chong and Druckman 2007, p. 104). Political scientists have studied the
effects framing can have on individual opinions on issues (Jacoby 2000, Grant and
Rudolph 2003, Jasperson et al 1998), and how various frames can affect the importance
individuals give to various considerations (Druckman 2001). Some have even provided a
conceptual framework for understanding different types of frames (Druckman 2006).
Out of this literature comes the lesson that framing has an impact on individual opinions,
and they way people think about issues. This is significant, because it gives reason to
suspect that horserace prompts have an impact on voters.
This section examines the literature demonstrating the prevalence of the horserace
frame, as well as some of its’ effects on the public, before asking the question, “how does
the horserace frame interact with campaign advertising to elicit emotions in voters?”
78

While the horserace trope has been the topic of much inquiry, its impact on voters’
emotions has been left largely undocumented (Baum et al. 2010). The prevalence of the
horserace frame has been noted for decades. Patterson (1993) documents the increasing
role the press has played in elections since the late 1960’s and criticizes them on a variety
of fronts including their reliance on the game frame, at the expense of issue-based
reporting.
Broh (1980) gauges the degree to which the horserace frame was used during the
1976 Presidential election, and voices the general concern that while this framing
mechanism can heighten citizen involvement, it may also do so at the cost of ignoring
serious issues. He finds that the horserace frame was used extensively during the 1976
election, but takes a somewhat rosy view of its use; he claims it serves valuable functions
such as enhancing public interest, induces reporters to focus on the polls, and that it
mitigates the danger of media sources crowning a winner early, and having undue
influence on the election results. While many others have found evidence of the
prevalence of the horserace frame, not all have such an optimistic take on its ubiquitous
use.
The prevalence of the horserace frame in primary elections is noted by King
(1990). Conducting a content analysis of two major newspapers during the 1998
primaries, King finds the horserace frame was by far the most widely used frame. In
contrast, policy frames were only utilized in only about a tenth of the coverage. Stevens
et. al (2006) report similar findings, even when looking at local news coverage. Utilizing
content analysis on all evening newscasts of the four non-cable channels in the
79

Minneapolis media market for 12 weeks leading up to the election, they found most
stories to emphasize the horserace, with few issue stories. With the prevalence of the
horserace frame firmly established, researchers may still wish to know how it impacts
voters.
Mendelsohn (1993) conducted a content analysis on the dominant frames utilized
in television reporting of the 1988 Canadian election. Using the CBC national news each
night during the election campaign, Mendelsohn found the campaign to be generally
framed through the lens of the horserace. He also found leaders were interpreted through
this frame as well. Finally, he found issues were interpreted through other frames, but
that the horserace was used as well. Brewer and Sigelman (2002) even note the
frequency with which political experts report on the campaign through the lens of the
horserace. Their data derives from a content analysis of every story in which the term
“political scientist” appeared during 1997 in the LEXIS-NEXIS general news database
(which includes the top 50 newspapers in the U.S.). They found the horserace frame was
the most widely used frame. They also found it to be used alone in over 50% of the
quotes. This contrasts sharply with issue frames, which stood alone only about 10% of
the time. While political scientists frequently criticize the use of the game frame, it turns
out they use it more often than any other.
Finally, Iyengar et al. (2004) attempt to explain why it is that the horserace frame
is so widely used. Explanations frequently cited include the length of the campaigns,
built-in conflicts between journalists and campaign operatives, and market forces. They
attempt to test the market hypotheses; that the horserace frame is ubiquitous because it
80

draws the most attention from viewers. They do so by monitoring potential voters during
the 2000 election campaign. They found that consumers of news were drawn to reports
on the horserace and strategy (even when there were equal amounts of more substantive
information available). Thus, they find support for the market hypothesis; the viewing
audience prefers these types of stories.
While this body of work notes the prevalence of the horserace frame, and cites
some reasons for this phenomena, it is important to also review studies which note the
impact this frame has on voters. Mutz (1995) studies the consequences of the horserace
frame on the dynamics of the campaign. Particularly, she is interested in how horserace
messages affect campaign contributions in presidential primary elections. She finds that
the horserace spin does help determine the frequency of campaign contributions.
However, not all contributors are motivated by the same horserace signal; messages that a
favored candidate is losing ground tend to increase donations, while less-favored
candidates benefit from coverage suggesting increased viability.
Rhee (1997) examines how news frames affect individuals’ interpretation of
campaigns. She finds (as many others do) that the dominant frames given to campaign
coverage are issue frames and strategy frames (horserace frames). Using an experimental
design, she also finds that both are effective in influencing interpretations of the
campaign; those exposed to strategy-framed campaign coverage were more inclined to
describe the campaigns in terms of the horserace, while those who were exposed to the
issue frame were more likely to describe the campaign in terms of the salient issues.
81

Finally, Baum et al. (2010) compare the effects of horserace messages and
campaign advertising on voters’ favorability during the 2008 presidential primary and
general elections. As hypothesized, we find the horserace metaphor to be more
influential as a determinant of candidate support in the primary election. The horserace
metaphor did not have a significant effect on partisans during the general election, even
amongst those who were loosely tied to the parties.
This body of literature demonstrates the prevalence and ubiquity of the horserace
frame during election coverage. It also identifies some key effects associated with
exposure to the common media trope. This study recognizes that the horserace frame is
not digested in isolation. In the real-world campaign, the horserace is reported in
conjunction with campaign events and advertisements. For this reason, the goal of this
portion of the paper is to see how the horserace frame interacts with campaign
advertising. In order to do this, the experiment paired each campaign ad with a
horserace statement preceding it. This section will demonstrate how the horserace frame
affects emotional reactions to campaign ads.
While the literature suggests that positive horserace messages increase favorability, and
willingness to donate to candidates who are not perceived as frontrunners, it is largely
silent on the horserace trope’s emotional impact(s).
The literature on the horserace trope does suggest that this framing device
conveys reassurance and threat. This fits perfectly into the IET conception of campaign
communication. While substantively different from campaign ads (which provide
reassurances and threats), the horserace trope works similarly, providing information that
82

one’s partisan “team” has moved ahead or fallen behind in the electoral race. In this
experiment the horserace trope has been paired with political ads, and it is expected to
work in tandem with those ads in the conveyance of threats and reassurances.

H3: “Ahead” horserace prompts during the general election amplify the levels of
hope and enthusiasm generated by appeals amongst ingroup members (increased
reassurance).

H4: “Behind” horserace prompts during the general election amplify the levels
of anger and fear generated by attacks amongst outroup members (increased
threat).

Results: Horserace Framing of Ads

Hypotheses 3 and 4 claim that horserace framing messages will have an effect on
the emotional impact of campaign ads during the general election season. In order to
measure the emotional impact of the horserace frames, the experiment randomly
preceded each ad shown to subjects with a horserace message. These messages claimed
that one of the two candidates had recently “moved ahead” in key battleground states, or
“fallen behind” in key battleground states. This wording was carefully chosen due to the
fact that by October (when the survey was completed), Obama had built a sizeable lead in
the nationwide polls. In order to craft a credible horserace statement, scope was limited
to “key battleground states.”
Analysis begins with comparisons between partisans told their candidate was
“winning” or “losing.” As Table 4.4 shows, differences were generally small, even when
statistically significant. Mean Democrat hope scores were 4.03 for those told Obama was
winning and 3.91 for those told Obama was losing. Enthusiasm scores for Obama
83

averaged 3.08 for those told he was winning, and 3.25 for those who thought he was
losing. Negative emotions felt towards McCain were also largely unaffected as those in
the Obama ahead group averaged 2.70 in fear and 2.96 in anger, while those in the
Obama behind group averaged 2.56 and 3.11 respectively.
Republicans responded similarly. Those told McCain was winning scored 1.23
on hope, while the McCain losing group scored 1.16. The McCain ahead group scored
2.24 on enthusiasm, while the McCain behind group scored 2.76. Regarding the negative
emotions, no significant differences emerged. Republicans in the ahead category had
mean fear scores of 3.50, while those given the behind prompt averaged 3.79. Regarding
anger, those in the ahead category scored 3.11 while those in the behind category scored
3.53. This seems to indicate that the horserace frame had little effect on the emotional
impacts of the ads. But there may be an interaction effect between the horserace prompt
and the particular advertising message.

Table 4.4: Horserace Frame and Candidate Emotions
Hope
Obama
Hope
McCain
Enthusiasm
Obama
Enthusiasm
McCain
Fear
Obama
Fear
McCain
Anger
Obama
Anger
McCain
Democrats

Obama Winning
Obama Losing

N
Sig.

4.03
3.91

387
N/S

1.35
1.33

387
N/S

3.08
3.25

386
N/S

1.23
1.17

389
N/S

1.46
1.15

383
**

2.70
2.56

383
N/S

1.34
1.12

383
*

2.96
3.11

383
N/S
Republicans

McCain Winning
McCain Losing

N
Sig.

1.23
1.16

320
N/S

3.28
3.55

318
N/S

1.04
1.03

320
N/S

2.24
2.76

321
^

3.50
3.79

320
N/S

1.06
1.01

320
N/S

3.11
3.53

317
N/S

1.17
1.16

320
N/S
Significance: ^ p<.1, * p<.05, ** p<.01, ***P<.001

84

Taking a slightly deeper cut at the data yields similar results. Regardless of ad
sponsor, the horserace prompts do not have significant effects on the emotional
persuasiveness of the ads. Obama ads, appear slightly more emotionally persuasive with
regards to hope and enthusiasm felt towards Obama, but these differences did not reach
statistical significance. Hope averaged 4.53 when ahead and 3.94 when behind, and
enthusiasm averaged 3.51 when ahead and 3.14 when behind. Negative emotions
generated by the McCain ads were not impacted by the horserace prompt. Fear averaged
2.53 when McCain was ahead, and 2.69 when he was behind. Anger averaged 3.25 when
McCain was ahead and 3.25 when he was behind.
Republicans appear even less effected by the horserace prompts than Democrats.
Republicans averaged 3.68 in hope for McCain regardless of the tenor of the horserace
prompt. They averaged 2.36 in enthusiasm when he was presented as winning, and 2.85
when losing. Neither of these differences were statistically significant. Republicans also
did not differ significantly in their negative emotional assessments of Obama as a result
of the horserace. Mean fear scores towards Obama were 3.56 for the Obama ahead
prompt, and 3.66 for the Obama behind prompt. Anger averaged 4.01 when Obama was
presented as ahead, while 3.47 while he was presented as behind. These findings do not
support hypotheses 3 and 4 (see Table 4.5).

85

Table 4.5: Horserace frame and Ad Sponsor and Tone
Hope
Oba
ma
Hope
McCain
Enthusiasm
Obama
Enthusiasm
McCain
Fear
Obama
Fear
McCain
Anger
Obama
Anger
McCain
Democrats

Obama Ad Ahead
Obama Ad Behind

McCain Ad Ahead
McCain Ad Behind

N
Overall Sig.

4.53
3.94

3.87
3.48

387
^

1.20
1.19

1.51
1.51

389
^

3.51
3.14

3.39
2.63

386
N/S

1.12
1.20

1.12
1.34

389
N/S

1.45
1.18

1.11
1.47

383
N/S

2.70
2.59

2.53
2.69

383
N/S

1.32
1.19

1.03
1.37

383
N/S

2.70
2.99

3.25
3.24

383
N/S
Republicans

Obama Ad Ahead
Obama Ad Behind

McCain Ad Ahead
McCain Ad Behind

N
Overall Sig.

1.20
1.40

1.04
1.12

320
*

3.41
2.92

3.68
3.68

318
N/S

1.00
1.07

1.00
1.06

320
N/S

2.64
2.14

2.36
2.85

321
N/S

3.56
3.66

3.32
4.00

320
N/S

1.03
1.11

1.00
1.00

320
N/S

4.01
3.47

2.68
3.11

317
*

1.07
1.18

1.15
1.24

320
N/S
Significance: ^ p<.1, * p<.05, ** p<.01, ***P<.001
Note: No pairs were significantly different.

Finally, a more in-depth look at the relationship between ad sponsor, ad tone, and
horserace prompt is necessary to further specify the effects of the horserace prompt on
partisans’ emotions. Here, the data suggest that horserace frames may influence the
emotional effectiveness of particular campaign ads during the general election season.
However, the results do not reach statistical significance.
This chapter hypothesized that appeal ads would be more emotionally persuasive
when preceded by a positive horserace statement. The data suggests this may be true for
Democrats, but not Republicans. Democrats’ mean hope scores for Obama were 5.76
when appeal ads were preceded by an ahead horserace prompt, and 4.89 when preceded
by a behind prompt. Democrats’ mean enthusiasm scores were 5.00 when appeal ads
86

were preceded by an ahead prompt, while only 3.81 when preceded by a behind prompt.
As expected, a horserace prompt declaring Obama was ahead provided reassurance to
Democrats, and led to increased hope and enthusiasm. However, these differences did
not reach statistical significance.
Ahead horserace prompts did not have the same effect on positive emotions felt
by Republicans. Republicans shown an ahead prompt before McCain’s appeal ads
averaged 3.81 and 2.51, in hope and enthusiasm, while Republicans shown McCain’s
appeal ads preceded by a behind prompt averaged 4.21 and 3.29. In both cases scores
went up following the behind prompt. While neither of these differences was statistically
significant (and the difference in hope was minimal), it is interesting that Republicans’
enthusiasm may be piqued by a negative horserace frame. Informed by IET, one would
expect positive emotions to arise in proportion to how reassuring campaign messages are.
A negative horserace prompt preceding a positive ad would seem to suggest that the
candidate has a lessened chance of winning, potentially lessening the effect of the
positive advertisement (by lowering the reassurance it can provide). These data suggest
that Republicans may be motivated towards increased positive affect when threatened.
These findings provide partial support for hypothesis 3, but demonstrate that significant
partisan differences may exist in the emotional appraisal process. This is not surprising
in light of recent research suggesting intrinsic differences between Democrats and
Republicans (Jost 2001, Lakoff 2002).
Looking to negative emotions, attack ads from the opponent are expected to
generate greater negative affect when preceded with an ahead horserace statement. This
87

is because the threat of the attack should be increased when the opposing partisan is
presented as “ahead in the polls.” However, while this combination may be expected to
boost fear, anger may not be higher when the attacker is presented as ahead in the polls.
This is because fear is aroused by attacks when the individual is unsure whether they
have the resources to respond, while anger results from attacks when the individual
believes they have the resources to confront the attack. The horserace message indicates
that the candidate being attacked may or may not have the resources to respond (win the
election) (Lazarus 1991, Mackie, Devos and Smith 2000).
The data provide little support for hypothesis 4. For Democrats, average fear
scores towards McCain were 2.42 when the attack was preceded by an ahead horserace
statement, and 3.53 when preceded by a behind prompt (not statistically significant).
Anger towards McCain remained constant; it averaged 4.74 with an ahead prompt, and
4.72 with the behind prompt.
Republicans didn’t react to differences in the horserace prompt they were given.
Mean fear towards Obama was 2.96 when attack ads were preceded by an ahead prompt
and 3.09 when preceded by a behind prompt. Mean anger was equally unaffected.
Republican anger averaged 4.63 when attack ads were preceded with an ahead prompt
and 4.59 when preceded by a behind prompt (see table 4.6). This indicates that
Republians were exceptionally angry, regardless of how the attack ads were framed.
These findings suggest that the horserace frame may matter, but the impact of this trope
on the emotions during the general election appears minimal.

88

Table 4.6: Horserace Frame by Candidate and Tone
Hope
Obama
Hope
McCai
n
Enthusiasm
Obama
Enthusiasm
McCain
Fear
Obama
Fear
McCain
Anger
Obama
Anger
McCain
Democrats

McCain Appeal Ahead
McCain Appeal Behind
N

McCain Attack Ahead
McCain Attack Behind
N

Obama Appeal Ahead
Obama Appeal Behind
N

Obama Attack Ahead
Obama Attack Behind
N

Total
Overall Sig.

3.64
3.15
(80)

4.00
4.12
(63)

5.76
4.89
(74)

4.26
3.54
(84)

(301)
**

1.67
1.64
(81)

1.32
1.31
(63)

1.16
1.39
(75)

1.41
1.00
(84)

(302)
N/S

3.69
2.79
(80)

2.87
2.56
(63)

5.00
3.81
(74)

2.94
2.88
(83)

(300)
**

1.26
1.49
(81)

1.00
1.00
(63)

1.16
1.44
(74)

1.18
1.00
(84)

(302)
N/S

1.15
1.49
(78)

1.13
1.38
(63)

1.50
1.53
(74)

1.76
1.00
(83)

(298)
N/S

2.49
2.16
(78)

2.42
3.53
(63)

2.63
2.53
(74)

3.09
2.73
(83)

(298)
N/S

1.07
1.49
(78)

1.00
1.09
(63)

1.26
1.17
(74)

1.62
1.29
(83)

(298
N/S)

2.41
2.43
(78)

4.74
4.72
(63)

2.26
2.75
(74)

3.53
3.55
(83)

(298)
888
Republicans

McCain Appeal Ahead
McCain Appeal Behind
N

McCain Attack Ahead
McCain Attack Behind
N

Obama Appeal Ahead
Obama Appeal Behind
N

Obama Attack Ahead
Obama Attack Behind
N

Total
Overall Sig.

1.00
1.13
(58)

1.00
1.18
(68)

1.20
1.59
(64)

1.00
1.31
(59)

(249)
N/S

3.81
4.21
(56)

3.74
3.73
(68)

3.57
2.97
(64)

2.78
2.91
(59)

(247)
N/S

1.00
1.00
(58)

1.00
1.15
(68)

1.00
1.18
(64)

1.00
1.00
(59)

(249)
N/S

2.51
3.29
(58)

2.17
3.10
(68)

2.90
2.14
(65)

1.85
2.31
(59)

(250)
N/S

2.41
3.97
(57)

4.26
4.76
(68)

4.00
4.00
(62)

2.96
3.09
(59)

(249)
N/S

1.00
1.00
(58)

1.00
1.00
(68)

1.00
1.15
(64)

1.00
1.15
(59)

(249)
N/S

1.92
2.97
(56)

3.34
3.67
(68)

3.80
2.94
(63)

4.63
4.59
(59)

(246)
N/S

1.22
1.13
(58)

1.17
1.48
(68)

1.00
1.18
(64)

1.18
1.31
(59)

(249)
N/S
Significance: ^ p<.1, * p<.05, ** p<.01, ***P<.001
Note: No pairs were significant.

This chapter demonstrates that factors such as the topic of political ads and the
horserace narrative may impact the emotion inducing qualities of those ads. However,
consistent with previous research, the “game frame” was found to have little impact on
partisans’ emotions in October (Baum et al. 2010). This is likely due to the fact that
voters are mostly decided by that time in the campaign, and even when trailing in the
89

polls, they tend to hope for the best outcome (Just, Crigler and Belt 2007). The findings
regarding the impact of issue vs. trait advertising are more robust and suggest that ads
with an issue focus tend to elicit more emotion amongst partisans. This is significant to
the study of emotional partisan polarization because it suggests that ads focusing on the
issues can be used to fuel emotions partisans’ feel during the campaign.
A key argument made in this chapter about emotional polarization, is that it is not
constant; certain types of political communication can move partisans emotionally, far
more than others. This chapter contributes to the overall argument of the dissertation by
demonstrating that political ads can be constructed to be more or less emotionally
arousing. Ads discussing the issues tend to arouse stronger emotions than ads focused on
character traits. This may be good news for scholars who bemoan the use of trait ads, in
favor of ads which speak to the issues; issue ads appear more emotionally persuasive, and
campaign strategists may have good reason to prefer them.

90

Chapter 5 – Putting the Pieces Together

The study of polarization has taken place in largely spatial terms. That is,
polarization is said to be evident when observations are clumped at opposite ends of any
spatial spectrum. The study of partisan polarization has followed this tradition, and large
differences between Democrats and Republicans have been noted both in government
and in the electorate (Jacobson 2007, Stonecash 2003). These differences have been
measured in Congress using roll-call votes. The literature has found that; party-unity
scores have gone up, and the ideological distance between Democrats’ and Republicans’
has steadily increased over the last 30 years (McCarty, Poole and Rosenthal 2006,
Sinclair 2006). A similar divide has also been mapped in the electorate using various
measures of issue stances, ideology and political evaluations (Abramowitz 2010,
Jacobson 2007). This study adopts the notion of polarization as spatial distance, but
points to previously uncharted “emotional space.” This research constitutes that space as
(self-reported) strength of emotions felt by partisans during the 2008 Presidential election
season.
The decision to explore the emotional dimension of polarization was not made on
a whim. It arose from literature on political behavior and political communication. The
study of the American voter has undergone many stages, as voting was first believed to
be influenced primarily by social and demographic factors (Berelson et al. 1954).
Campbell et al. (1960) utilized survey data to demonstrate the primacy of partisanship in
political decision making, and argued that voters were neither ideological nor informed.
Still others made the argument that voters were rational (Downs 1957, Fiorina 1981) and
91

were influenced largely by issue and governmental outcome concerns. While
demographics, partisanship and ideology are clearly related to political behavior, they do
not tell the entire story. Recent work has highlighted the importance of emotions to
political behavior and political decision making (Brader 2005, Marcus et al. 2000). This
work has demonstrated that emotional assessments are distinct from personal trait
evaluations of candidates (Abelson et al. 1982), and that they have a measurable impact
on a variety of political behaviors.
The inclusion of emotion in a discussion of partisan polarization also makes sense
given recent insights on the nature of partisanship. Scholarship has indicated that
partisan attachments are not merely ideological or habitual, but also the result of social
identification with the political parties (Green and Schickler 2002). This conception of
partisanship claims that, attachments to the parties are stable and enduring to an extent
that can’t be explained by ideology alone. Instead, the power of party identification
comes from a social attachment to the parties (Mackie, Devos and Smith 2000, Mackie,
Silver and Smith 2004). The analogy of partisans as sports fans resonates here, as party
members often cheer their “team” to victory in the same manner as fans of any other
team.
This work argues that the study of partisan polarization must include emotion, not
only because of its role in political behavior and motivation, but because it is
fundamental to the bond which unites partisans. Partisanship is not a “cold” decision; the
party or candidate one identifies with is often viewed as part of the social self.
92

\When that party or candidate is victorious, the partisan cheers as if he/she had won, and
when victory eludes the party or candidate, the partisan identifier often feels the sting of
loss as their own. The data collected support this notion of partisanship and of partisan
polarization.

Partisan Divisions Over Emotion
Chapter two examines survey data which gauges partisan differences on a variety
of variables, including ideology, candidate evaluations (trait assessments and overall
favorability), and emotions. The findings were somewhat surprising; partisans were
divided over emotion felt towards the candidates as much as they were over trait
evaluations of the candidates, and much more than they were over their ideological
leanings. Partisans showed extreme emotional deference towards the candidate of their
party, directing virtually all of their positive emotions in that direction, while feeling
virtually all of their negative emotions towards the opponent.
These findings are consistent with studies of partisan polarization in the electorate
(Abramowitz 2010), yet they suggest that the extent of partisan divisions may be larger,
and run deeper than previous studies indicate. The data also indicated that consistent
with previous work (Abelson et al. 1982), emotional assessments of the candidates were
statistically distinct from trait evaluations and ideology. Finally, regression analysis
demonstrated that inclusion of the emotions in models of candidate favorability
significantly improved the models’ performance, and indicated that the emotions play a
large role in determining candidate favorability.
93

These findings make two significant contributions to the literature on partisan
polarization: (1) partisan polarization exists on an emotional dimension, (2) emotions
serve to fuel partisan polarization on other dimensions. The first contribution is
supported strongly by the data presented in chapter two. Most strikingly, partisans were
more divided over their emotional assessments of the candidates than they were over
ideology (a traditional indicator of polarization). The second contribution points to the
significance of emotions in general, and specifically to the polarization narrative
constructed in this work. So why is it important that the emotional dimension of
polarization be captured?
First, adding an emotional component helps to make the case that partisan
divisions run deep and can be described as polarized. This illuminates another distinct
and relevant dimension along which partisans are deeply divided. While scholars have
identified partisans’ differences for decades (Jacobson 2007), others have tried to make
the case that only the most active party members (activists, etc.) are truly divided from
one another, with the majority of the public falling somewhere in the middle (Fiorina
2006). As a result, Fiorina argues that partisan polarization is not indicative of a larger
culture war, or even widespread voter polarization. Looking to measures of public
opinion, and issue positions, Fiorina (2006, 2010) claims that differences aren’t all that
substantial. Even where party differences are more noticeable, they are not the result of
individual partisans’ becoming more deferential towards their party, but rather
ideological sorting. According to this argument, the rise in partisan polarization by some
measures is really attributable to people more accurately choosing the ideologically
94

“correct” party, and this has only started happening as the parties have become more
ideologically distinct.
While Fiorina attempts to downplay the extent of partisan polarization, others
utilizing different measures have made the case that partisans are deeply divided. While
Fiorina’s account is largely issue and ideology centric, Jacobson (2007) and Abramowitz
(2010) study partisan differences on political evaluations of candidates and parties
(amongst other things). This conception of polarization leads to measures which
demonstrate a deeper partisan divide. These researchers note the exceptionally wide rifts
between partisan assessments of the candidates and partisan evaluations of the parties in
government to make the claim that while partisans may not be deeply divided in their
issue stances, they are in their evaluations of politics.
This begs the question, “if partisans are only moderately divided over issue
positions and ideology, why are they so deeply divided over political evaluations?” The
contention of this work is that partisan polarization isn’t merely occurring over issues and
ideology, but also over social identification with the political parties and the political
emotions that arise as a result. This dimension of polarization helps explain the large
partisan divide which emerges whenever partisans are asked to evaluate candidates,
politicians, and policies in partisan terms. Chapter two demonstrates that candidate
favorability is influenced in large part by the emotions. These emotions are elicited by
the election context and the social identification partisans have with the parties and the
candidates running for office. These (emotional) differences cannot be entirely explained
by underlying ideological differences amongst partisans. While many partisans may
95

choose to identify with the party that is the best ideological fit, once that affiliation has
been forged, the successes and failures of that party take on great emotional significance.
Most importantly, the study of emotion helps bridge the gap between the disparate
strains in the polarization literature. The emotional connection felt by partisans to their
party, and the emotions they feel towards the candidates help explain why candidate
evaluations are so polarized during election season, while partisans’ ideology and issue
positions remain less polarized. When Fiorina (2010, 2006) notes relatively small
differences in issue positions between partisans, and claims this is the end of the
polarization story, emotion can help explain why those same partisans are so deeply split
in their evaluations of the candidates. The data presented in this work indicate that
emotion is fueling the fires of partisan polarization, in addition to individual ideology and
issue positions.

The Elicitation of Emotions
Emotions have a significant impact on political evaluations and behavior, yet less
is known about their stability and arousal. For example, how strong are partisans’
emotions towards the candidates at any given time? How does political advertising
impact these emotions? Chapter three discussed the findings of an Internet video
experiment which gauged the effects of real-world campaign advertising on partisans’
emotions. The data indicate that different forms of political advertising produce varying
degrees of positive and negative emotions. The data also demonstrate that campaign
advertising was filtered through the lens of partisanship, as partisans were unlikely to feel
96

positive emotions towards the opponent, or negative emotions towards their candidate,
regardless of the message. Finally, this data demonstrates that Democrats and
Republicans do not have identical emotional profiles; the same stimulus is not guaranteed
to have congruent effects. This last comment requires further discussion. One of IET’s
contributions is that it makes the case for interchangeable groups; regardless of the
individual makeup of social groups, we can expect them to respond emotionally as a
result of their perceived strength, and the type of stimulus they are exposed to. These
data suggest that groups may not be interchangeable.
Regarding the specific effects of campaign advertising, we found that appeal ads
from one’s in-group candidate produced significantly more hope and enthusiasm amongst
Democrats but not Republicans. This suggests that while partisan affiliation matters in
emotional response, Democratic and Republican partisan identities do not operate in
exactly the same manner. In particular, it appears that hope and enthusiasm can be
aroused successfully amongst Republicans using positive or negative ads.
Looking to negative emotions, Democrats became significantly angrier with
McCain after viewing a McCain attack ad, than an appeal ad. Democrats were not likely
to feel fear regardless of which ad they viewed (this is likely due to the fact that they
were confident about Obama’s chances by late October). Republicans were also angrier
with Obama after viewing an Obama attack than an Obama appeal ad. However, they
were more fearful after viewing an Obama appeal ad than an Obama attack. This
suggests that negative ads are not always more effective at generating fear. When groups
are trailing in the polls, appeals from the opponent may be more threatening (and likely to
97

induce fear) than attacks. This could be due to the fact that the opposing partisans are
primed to feel fearful because they think they may lose the election. In this context, an
appeal from the opponent may reinforce that fear.
These findings suggest that campaign communication (and particularly television
ads) has a significant impact on the emotions. This is important because it demonstrates
that emotion is fluid; it is not held constant, but can be influenced on a minute to minute
basis by different forms of political communication. This is not unexpected, as emotions
are adaptive and reactive. These findings point towards an understanding of how
particular emotions can be most effectively elicited. If a candidate or party is attempting
to build positive emotions amongst their constituents, they should run appeal ads rather
than attack ads. If they are attempting to de-mobilize opposing partisans by increasing
fear, attacks are generally more effective.

Discrete Emotions
A key theme that this work reinforces is that specific emotions matter. The
theoretical landscape of emotions in politics is still diverse. While there are many
competing accounts of emotional elicitation and the substance of emotions themselves, a
unifying theory of emotion has yet to emerge. This project provides theoretical support
for a discrete conception of emotion.
Researchers who adhere to a discrete conception of emotions often base their
research in appraisal theories of emotion, which claim emotion is a reaction to stimuli
intended to help individuals cope. Enthusiasm is a positive emotion that arises to
98

reinforce positive stimuli (Lazarus 1991). In essence, enthusiasm is the good feeling one
gets when things are going well. Hope is also a positive emotion, but it is often felt in the
face of uncertainty; where good outcomes are wished for, but may be in doubt (Crigler,
Just and Belt 2007). This project measures hope, enthusiasm, fear and anger. In chapter
two, when emotional polarization was documented, it was clear that hope and enthusiasm
were distinct. While Democrats felt large sums of both (more hope than enthusiasm),
Republicans felt lower amounts of both, and felt very little enthusiasm for McCain.
Further support for a discrete conception of emotion came when examining the
impact of the emotions on candidate favorability, one indicator of partisan polarization.
For Obama’s favorability, each discrete emotion had a statistically significant effect. In
fact, hope for Obama was the single best predictor of his favorability rating. Finally, this
model demonstrated that hope’s impact on Obama’s favorability was roughly five times
greater than enthusiasm’s. The emotions felt towards McCain were also all statistically
significant, although anger was the strongest predictor of McCain’s favorability.
This dissertation employs IET to make predictions about the emotions partisans
are likely to feel as a result of the ads they view. It predicted that partisans would be
more likely to feel hope and enthusiasm after viewing reassuring messages than attacks
from the ingroup candidate. This was confirmed, along with the prediction that attacks
would produce more fear and anger. IET highlights the importance of a discrete
conception of emotions, as it claims that varying stimuli should be expected to produce
different emotions.
99

This research also allows for some qualifications to be made to IET. In general
IET was correct in predicting when partisans were likely to feel positive and negative
emotions. It also helped us correctly predict the types of positive emotions felt by
partisans. However, there was some difficulty in predicting the negative emotions.
Specifically, Republicans, although behind in the race still felt a high degree of anger
when threatened. This finding isn’t really all that surprising; one shouldn’t expect all
social groups to react to the same stimuli in the same way. Group characteristics clearly
matter, and this thesis hypothesizes that Republicans may think differently from
Democrats and be more prone to anger regardless of the strength of their position in any
given race (Jost et al. 2003, Lakoff 1996).
This finding (Republicans were prone to anger even when in the weaker position),
was only possible due to the study of emotions as discrete phenomena, and may point to
the true power of partisanship. Republicans, although they were clearly trailing in the
polls, were still very angry; they were reacting as if they were still in the strong position.
These findings demonstrate that partisanship can blind members to political reality. It
also has a deep emotional connection, as partisans identify socially with their party and
feel strong emotions as a member of the group.

Impact of Ad Content and the Horserace Narrative
This project also examined whether the content of positive and negative ads had
an impact on their emotional persuasiveness. Literature on campaign advertising has
long focused on ad tone (Geer 2006, Ansolabehere and Iyengar 1995), and has also
100

looked at the effectiveness of ads which focus on traits versus those focusing on issues.
Effectiveness, is generally operationalized in terms of candidate evaluations (Kahn and
Geer 1994). This project found that in October, holding other factors constant, partisans
felt the highest levels of hope and enthusiasm for their candidate after viewing issue ads.
This finding held regardless of ad tone. Negative emotions were not affected by the
content of the ads.
The next slice of the data demonstrated that controlling for tone, content of the
ads still mattered. Democrats became more hopeful and enthusiastic towards Obama
after viewing his issue appeals, than his trait appeals. Republicans demonstrated the
same tendency, and the differences were more significant, as they became far more
hopeful for McCain following McCain’s issue appeals, and somewhat more enthusiastic.
These findings suggest that the content of political ads matters in the emotional
appraisal process. While scholars have noted a variety of differences between issue and
trait ads (Johnson and Kaid 2002), even finding attacks were more effective when paired
with issues rather than traits (Kahn and Geer 1994), little was known about the emotional
effects of issue and trait ads. Gross (2008) gave reason to expect trait ads to be more
emotionally persuasive, as she noted the increased emotional effects associated with
episodic vs. thematic framing of news stories. However, this work demonstrates that
issue ads are more emotionally evocative than trait ads. Political scientists often bemoan
the state of campaign advertising claiming it is based on personal attacks and devoid of
issue content. This research provides support for the notion that emotionally effective
campaigning should focus on the issues rather than personal traits.
101

This project also recognizes that political ads are not viewed in isolation, but
rather in a larger electoral context. This context is generally reported and framed in terms
of the horserace; who’s winning and who’s losing. The horserace narrative is ubiquitous,
and is repeated throughout various forms of broadcast media (Farnsworth and Lichter
2006, Graber 2000). While some of the effects of the horserace frame have been noted
(Mutz 1995, Rhee 1997), its impact on the emotions has received less attention (Baum et
al. 2010). This experiment combined campaign ads with various horserace messages.
The goal was to present the ad within the overall electoral context, in order to maximize
the applicability of the results to the real world. An advantage to preceding each
campaign ad with a horserace message was that it allowed the opportunity to
experimentally manipulate the content of the prompt.
The findings of this project indicated that in the aggregate, the horserace frame
had little effect on partisans’ emotions in October, with the exception of enthusiasm for
McCain which actually increased slightly when he was presented as losing. Further
analysis controlling for ad indicated that the horserace frame had a minor impact on the
emotions, as none of the differences reached statistical significance.
Analysis of these findings suggests that the horserace message had little influence
on partisans’ emotional appraisals of the candidates. The findings in this project do not
rule out the potential effects of the horserace narrative on the emotions, and in fact
previous work (Baum et al. 2010, Mutz 1995) has demonstrated that horserace framing
does have significant effects during the primaries. So why did the horserace frame
appear to have a less significant impact in the 2008 general election?
102

First, this experiment was conducted in late October, only a few weeks before the
general election. It is possible that the horserace manipulation had little impact because
people were aware of the election context; Obama was ahead by a large margin. It is also
possible that the horserace frame is most important during a primary election, as people
are trying to get to know the candidates, and when frontrunner status provides name
recognition and bandwagon effects. By the time the general election arrives, voters are
aware of the candidates, and partisans are there to stand behind their candidates,
regardless of their status in the polls (Baum et al. 2010).
Finally, it is possible that horserace messages do not carry the same degree of
emotional appeal that accompanies other forms of campaign communication. While
horserace messages do convey some degree of threat or reassurance, they may not be
emotion-laden themselves. Campaign speeches and advertisements are carefully crafted
to convey both factual and emotional content (the two are often intertwined). Through
the use of intricately detailed scripts, accompanied by powerful music and images,
campaign ads are built to generate emotion (Brader 2006). In fact, some would argue
that the only purpose of campaign advertising is to incite the emotions. For these reasons
it is not surprising that campaign advertising had a greater impact on partisans’ emotions
in this experiment.

Avenues for Further Research
While this project begins to map the emotional landscape of polarization, it is
only a starting point. It is still important that partisans’ emotions be captured in other
103

electoral contexts. The 2008 Presidential election was unique in a variety of ways; it was
the first time an African-American candidate was in the race, the nation was in the
beginning stages of one of the worst economic downturns in recent history and there were
two ongoing wars in the Middle East. It is important that these factors be taken into
account. While this work contends that the study of emotion is essential to the study of
partisan polarization and electoral politics, its’ findings are limited to the study of only
one election, which needs to be taken in context.
This was also an exceptionally “hard test” of the effectiveness of political
advertising and horserace messages on viewers’ emotions. The experiment took place in
late October, only weeks before the general election. As the election nears, vote choices
become largely solidified, and voters have a clearer picture of who the frontrunner is.
Under these conditions it would be difficult to image a single political ad and horserace
prompt effecting voters’ thoughts or feelings. The fact that these treatments still had
significant effects on partisans speaks to the robustness of the results.
Future research on emotional elicitation may also wish to separate horserace
effects from advertising effects on partisans’ emotions. This work attempted to combine
the two in order to closely mimic the real-world experience partisans have during the
campaigns, but greater separation of experimental treatments would aid in comparing the
impact of the two. This work also gauged the emotions felt towards the candidates, as
they were the primary representatives of the parties during the presidential election.
However, it may be fruitful to gauge the emotions felt towards other political figures, or
towards the parties themselves.
104

While campaign ads were the focus of this study, similar work should be
conducted to examine the effects of other types of political communication on the
emotions. Do candidate speeches, debates, or short films have similar effects? Also, the
ads utilized in this study were nationally televised ads from the political parties. It would
be interesting to see whether ads from third party groups have equivalent emotional
effects. Political advertising is also not currently confined to television spots.
Advertising is taking place on the internet, as the campaigns have begun producing
internet only ads. What effects do these ads have on partisans’ emotions? Are they more
or less emotionally charged than ads made for national viewing audiences?

Implications for Representation
This project speaks to the concept of democratic representation, which is at the
foundation of normative arguments that find partisan polarization troubling (Fiorina
2006). At the heart of representation is the concept of political linkage; elected officials
should be in some way responsive to the wants and needs of their electorate (Luttbeg
1974). While there has been normative debate regarding how much leeway elected
officials should have to pursue the public good (Burke 1774, Eulau et al. 1957), the
purpose of this discussion is not to engage this debate, but rather to engage current
arguments regarding the consequences of partisan polarization.
Fiorina (2006) makes the argument that partisan polarization is really only
occurring in government, not in the electorate. In his work “Culture War,” he utilizes
survey data to demonstrate that aggregate opinion in “red” and “blue” states isn’t that
105

different, and argues that polarization isn’t occurring in the electorate. While he notes
that Democrats and Republicans in the electorate have become more ideologically
distinct over time, he still doesn’t believe the average American is aligned with the more
extreme positions being adopted by the political parties. He claims that polarized parties
in government, combined with a largely moderate public has created a fundamental
disconnect in representation; one which goes against tradition spatial expectations
(Downs 1957). If the majority of Americans hold moderate opinions on a variety of
issues, it is problematic that lawmakers are growing increasingly polarized. This
contrasts with scholars of the 1950’s who advocated that parties be “responsible,”
adopting platforms that distinguished themselves from one another and offered voters
choice (Ranney 1951).
This work finds some support for Fiorina’s argument, but not all evidence points
in this direction. In particular, past research demonstrates that partisans differ to a
significant degree on a number of measures. Based on the data gathered for the current
project, it would be difficult to argue that the electorate is comprised of moderates, while
polarization occurs only in government. However, the data presented in this work
demonstrates that partisans are least divided over ideology, and most deeply divided over
favorability assessments and emotional assessments of the candidates.
Most importantly, this work argues that partisans socially identify with the
parties and feel emotions as a result of their affiliation. This explains why partisans can
be moderately divided over ideology, yet feel so strongly about the candidates they are
presented with at election time. In a sense, Fiorina’s argument may still resonate
106

throughout this work; while these data contest the notion of a largely moderate public,
they demonstrate that partisans do appear to favor or disfavor the candidates more than
would be expected given their ideological differences. This project also demonstrated the
ability of political advertisements to amplify or moderate the emotions partisans felt
towards the candidates. This suggests that Fiorina’s concerns are merited; polarized
representatives may receive more support than expected due to the emotional attachments
partisans feel towards them, and their acumen at disseminating forms of political
communication which arouse these feelings.

Conclusion
This work provides encouragement to those who wish to include emotion in
studies of partisanship, polarization, and political behavior in general. This work finds
that emotions matter. Partisans in the electorate polarize over emotions felt towards the
candidates to a significant extent. These emotions were shown to act independently of
trait evaluations and ideology. Finally, statistical analyses indicated that the emotions
were strong predictors of candidate favorability. These findings demonstrate the power
of emotions in political evaluations of the candidates.
The data presented in this dissertation also suggest that political advertisements
have a significant impact on partisans’ emotions. Specifically, the sponsor, tone and
particular subject matter all work together to produce a variety of emotional responses.
These findings speak to literature on political communication and reinforce the notion
107

that campaigns and campaign communication matter; they have the ability to incite or
depress emotions.
Finally, this project provides support for IET, and the notion of partisanship
defined as a social identity, rather than merely an ideological, or sociological
phenomenon. The data provide evidence, as partisan identifiers were highly polarized in
their emotional assessments of the candidates. They also processed political
advertisements from the candidates through the lenses of partisanship, exhibiting
deferential emotional reactions to the ingroup candidate. However, while this
dissertation found strong support for IET, it also identified a key weakness. IET treats
groups as interchangeable, and claims that group emotions can be predicted by looking to
non-group centric variables. This analysis demonstrated that Democrats and Republicans
did not always react emotionally in the manner predicted by these variables. Republicans
were angrier than expected. These findings indicate that the inclusion of group-centric
variables may improve the predictive power of IET.

108

Bibliography

Abelson et al. 1982. “Affective and Semantic Components in Political Perception.”
Journal of Personality and Social Psychology, 42(4): 619-630.

Abramson, Paul et al. 2002. Change and Continuity and the 2000 Elections.
Washington D.C.: CQ Press.

Abramowitz, Alan. 2010. The Disappearing Center: Engaged Citizens, Polarization,
and American Democracy. New Haven: Yale University Press.

Abramowitz, Alan. 2006. “Comments: Disconnected or Joined at the Hip?” In Red and
Blue Nation, ed. Nivola, Pietro and Brady, David. Washington D.C.: Brookings
Institution Press.

Abramowitz, Alan and Saunders, Kyle. 2005. “Why Can’t We All Just Get Along: The
Reality of a Polarized America.” The Forum 3(2).

Ansolabehere, Stephen, Shanto Iyengar, Adam Simon, and Nicholas Valentino. 1994.
“Does Attack Advertising Demobilize the Electorate?” The American Political Science
Review 88(4): 829-838.

Ansolabehere, Stephen, and Shanto Iyengar. 1995. Going Negative: How Political
Advertising Shrinks and Polarizes the Electorate. New York: Free Press.

Ansolabehere, Stephen, Shanto Iyengar, and Adam Simon. 1999. “Replicating
Experiments Using Agggregate and Survey Data: The Case of Negative Advertising and
Turnout.” The American Political Science Review 93(4): 901-909.

Aristotle. 1954. Rhetoric. Translated by R. Rhys Roberts. New York: Modern Library.

Baum, Matthew et al. 2010. Emotions, the Horserace Metaphor and the 2008
PresidentialCampaign. Paper prepared for the Annual Meeting of the American Political
Science Association, Washington, DC.

Belt et al. 2007. “Affective Priming in the 1996 Presidential Campaign.” Prepared for
delivery at the 2007 Annual Meeting of the American Political Science Association,
Chicago, Il.

Benoit, William. 2004. “Political Party Affiliation and Presidential Campaign
Discourse.” Communication Quarterly 52(2): 81-97.

109

Berelson, Bernard et al. 1954. Voting; a Study of Opinion Formation in a Presidential
Campaign. Chicago: University of Chicago Press.

Binder, Sara. 1999. “The Dynamics of Legislative Gidlock, 1947-96.” The American
Political Science Review, 93(3).

Brader, Ted and Valentino, Nicholas. 2007. “Identities, Interests and Emotions:
Symbolic versus Material Wellsprings of Fear, Anger and Enthusiasm.” In The Affect
Effect, ed. Neuman et al. Chicago: University of Chicago Press.

Brader, Ted, and Bryce Corrigan. 2007. “How the Emotional Tenor of Ad Campaigns
Affects Political Participation.” Paper presented at the annual meeting of the American
Political Science Association, Philadelphia, PA.

Brader, Ted. 2006. Campaigning For Hearts and Minds: How Emotional Appeals in
Political ads Work. Chicago: University of Chicago Press.

Brader, Ted. 2005. “Striking a Responsive Chord: How political Ads Motivate and
Persuade Voters by Appealing to Emotions.” American Journal of Political Science
49(2): 388-405.

Brewer, Paul and Lee Sigelman. 2002. Political Scientists as Color Commentators:
Framing and Expert Commentary in Media Campaign Coverage. The Harvard
International Journal of Press/Politics 7(1): 23-35.

Broh, Anthony. 1980. Horse-Race Journalism: Reporting the Polls in the 1976
Presidential Election. Public Opinion Quarterly 44(4): 514-529.

Bucy, Erik. 2000. Emotional and Evaluative Consequences of Inappropriate Leader
Displays. Communication Research 27(2): 194-226.

Burke, Edmund. 1774. Speech to the Electors at Bristol.

Campbell, Angus. 1954. The Voter Decides. Evanston, Illinois: Row and Peterson.

Campbell, Angus, Phillip Converse, Warren Miller, and Donald Stokes. 1960. The
American Voter. Chicago: University of Chicago Press.

Carter, Stephen. 1988. Civility: Manners, Morals and the Etiquette of Democracy. New
York: Basic Books.

Cassino, Dan and Lodge, Milton. 2007. “The Primacy of Affect in Political
Evaluations,” In The Affect Effect, ed. Neuman et al. Chicago: University of Chicago
Press.
110

Chong, Dennis and James Druckman. 2007. Framing Theory. Annual Review of
Political Science 10: 103-126.

Civettini, Andrew, and David Redlawsk. 2009. Voters, Emotions and Memory.
Political Psychology 30(1): 125-151.

Converse, Phillip. 1964. The Nature of Belief Systems in Mass Publics. In Ideology and
Discontent, ed. David Apter.

Corrigan, Costarelli, Sandro and Augusto Palmonari. 2003. “Ingroup Ambivalence and
Experienced Affect: The Moderating Role of Social Identification.” European Journal of
Social Psychology 33(66): 813-21.

Crigler, Ann, Jesse Mills and Jenna Hootstein. 2009. “Intergroup Emotions: Studying
the Effects of Threat and Reassurance on Partisan Identifiers.” Presented at the Annual
Meeting of the American Political Science Association. Toronto.

Crisp, Richard, Sarah Heuston, Matthew Farr and Rhiannon Turner. 2007. “Seeing Red
or Feeling Blue: Differentiated Intergroup Emotions and Ingroup Identification in Soccer
Fans.” Group Processes & Intergroup Relations 10(1): 9-26.
Demerath, N.J. and Yang, Yonghe. 1997. “What American Culture War? A View from
the Trenches as Opposed to the Command Posts and the Press Corps.” In Cultural Wars
in American Politics: Critical Reviews of a Popular Myth, ed. Williams, Rhys. New
York: Aldine de Gruyter.

Dimaggio, Paul et al. 1997. “Have Americans’ Social Attitudes Become More
Polarized?” In Cultural Wars in American Politics: Critical Reviews of a Popular Myth,
ed. Williams, Rhys. New York: Aldine de Gruyter.

Downs, Anthony. 1957. An Economic Theory of Democracy. New York: Harper and
Row.

Evans, J.H. 2003. “Have Americans’ Attitudes Become More Polarized? An Update.”
Social Science Quarterly 84(1): 71-90.

Farnsworth, Stephen and S. Robert Lichter. 2006. The Nightly News Nightmare:
Network Television’s Coverage of U.S. Presidential Elections, 1988-2004, 2
nd
edition.
New York: Rowman and Littlefield.

111

Feldman, Paula M. Niedenthall, and Piotr Winkielman, 312-34. New York: Guilford.
Excerpt in: The Affect Effect, ed. Neuman et al. Chicago: University of Chicago Press.

Fiorina, Morris and Abrams, Samuel. 2008. “Political Polarization in the American
Public.” The Annual Review of Political Science 11:563-588.

Fiorina, Morris and Levendusky, Matthew. 2006. “Disconnected: The Political Class
Versus the People” In Red and Blue Nation, ed. Nivola, Pietro and Brady, David.
Washington D.C.: Brookings Institution Press.

Fiorina, Morris et al. 2005. Culture War? The Myth of a Polarized America. New
York: Pearson Longman.

Fiorina, Morris, Samuel Abrams and Jeremy Pope. 2003. “The 2000 US Presidential
Election: Can Retrospective Voting Be Saved?” British Journal of Political Science, 33:
163-187.

Frank, Thomas. 2004. What’s the Matter With Kansas? How Conservatives Won the
Heart of America. New York: Metropolitan Books.

Franz, Michael, and Travis Ridout. 2007. “Does Political Advertising Persuade?”
Political Behavior 29(4): 465-491.

Firdja, Nico. 1987. “Emotion, Cognitive Structure, and Action Tendency.” Cognition
and Emotion 1(2): 115-143

Fridkin, Kim, and Patrick Kenney. 2008. “The Dimensions of Negative Messages.”
American Politics Research 36(5): 694-723.

Galston, William and Nivola, Pietro. 2006. “Delineating the Problem.” In Red and Blue
Nation, ed. Nivola, Pietro and Brady, David. Washington D.C.: Brookings Institution
Press.

Garcia, Amber., Daniel Miller, Eliot Smith and Diane Mackie. 2006. “Thanks for the
Compliment? Emotional Reactions to Group-Level Versus Indiividual Level
Compliments and Insults. Group Processes & Intergroup Relations 9(3): 307-24.

Garramone, Gina, Charles Atkin, Bruce Pinkelton, and Richard Cole. 1990. “Effects of
Negative Political Advertising on the Political Process.” Journal of Broadcasting and
Electronic Media 34(3): 299-311.

Garrett, Sam. 2010. State of Campaign Finance Policy: Recent Developments and
Issues for Congress. Congressional Research Service.
112

Geer, John. 2006. In Defense of Negativity: Attack Ads in Presidential Campaigns.
Chicago: University of Chicago Press.

Gilmour, John. 1995. Strategies of Disagreement. Pittsburgh: University of Pittsburgh
Press.

Goldstein, Ken and Paul Freedman. 2002. “Campaign Advertising and Voter Turnout:
New Evidence for a Stimulation Effect.” Journal of Politics 64(3): 721-740.

Graber, Doris. 2007. “The Road to Public Surveillance: Breeching Attention
Thresholds.” InThe Affect Effect, ed. Neuman et al. Chicago: University of Chicago
Press.

Green, Donald, Bradley Palmquist and Eric Schickler. 2002. Partisan Hearts and
Minds: Political Parties and the Social Identities of Voters. New Haven: Yale
University Press.

Gross, Kimberly. (2008). “Framing Persuasive Appeals: Episodic and Thematic
Framing, Emotional Response, and Policy Opinion.” Political Psychology 29(2): 169-
192

Groseclose, Tim et al. 1999. “Comparing Interest Group Scores across Time and
Chambers: Adjusted ADA Scores for Congress.” American Political Science Review,
93(1): 33-50.

Hacker, Jacob and Pierson, Paul. 2005. Off Center: The Republican Revolution and the
Erosion of American Democracy. New Haven: Yale University Press.

Hilygus, Sunshine. 2005. “Campaign Effects and the Dynamics of Turnout Intention in
Election 2000.” The Journal of Politics 67(1): 50-68.

Hitchton, Jacqueline, and Chingching Chang. 1995. “Effects of Gender Schematic
Processing on the Reception of Political Commercials for Men and Women Candidates.”
Communication Research 22(4): 430-458.

Huber, Gregory and Kevin Arceneaux. 2007. “Identifying the Persuasive Effects of
Presidential Advertising.” American Journal of Political Science 51(4): 957-977.

Huddy, Leonie Lilliana Mason. 2008. “Heated campaign politics: An intergroup conflict
model of partisan emotions.” Paper presented at the annual meeting of the American
Political Science Association, August 28, 2008, Boston, Massachusetts.

113

Huddy, Leonie, et al. (2007). “On the Distinct Political Effects of Anxiety and Anger.”
In The Affect Effect, ed. Neuman et al. Chicago: University of Chicago Press.

Huddy, Stanley et al. 2005. “Threat, Anxiety and Support of Anti Terrorism Policies.”
American Journal of Political Science 3:610-25.

Hunter, James David. 1991. Culture Wars, The Struggle to Define America. New York:
Basic Books.

Hunter, James David. 1994. Before the Shooting Begins: Searching for Democracy in
America’s Culture Wars. New York: Free Press.

Isbell, Linda and Ottati, Victor. 2002. “The Emotional Voter: Effects of Episodic
Affective Reactions on Candidate Evaluations.” In The Social Psychology of Politics, ed.
Victor C. Ottati et al. New York: Kluwer.

Iyengar, Shanto, Helmut Norpoth and Kyu Hahn. 2004. Consumer Demand for Election
News: The Horserace Sells. Journal of Politics 66(1): 157-175.

Jackson, Robert, Jeffrey Mondak, and Robert Huckfeldt. 2009. “Examining the Possible
Corrosive Impact of Negative Advertising on Citizens’ Attitudes toward Politics.”
Political Research Quarterly 62(1): 55-69.

Jacobson, Gary. 2007. A Divider, Not a Uniter. New York: Pearson Longman.

Jacobson, Gary. 2006. “Comments: Disconnected or Joined at the Hip?” In Red and
Blue Nation, ed. Nivola, Pietro and Brady, David. Washington D.C.: Brookings
Institution Press.

Jacobson, Gary 2005. “Polarized Politics and the 2004 Congressional and Presidential
Elections.” Political Science Quarterly 120(2): 199-218

Jacobson, Gary. 2001. The Politics of Congressional Elections. New York: Longman.

Jacobson, Gary. 2000. “The Electoral Basis of Partisan Polarization in Congress,”
presented atthe Annual Meeting of the American Political Science Association,
Washington, D.C.

Jameison, Kathleen Hall. 1996. Packaging the Presidency: A History and Criticism of
Presidential Campaign Advertising. New York: Oxford University Press.

Jerit, Jennifer. 2004. “Survival of the Fittest: Rhetoric during the Course of an Election
Campaign.” Political Psychology 25(4): 563-575.
114

Johnston, Anne and Lynda Kaid. 2002. Image Ads and Issue Ads in U.S. Presidential
Advertising: Using Videostyle to Explore Stylistic Differences in Televised Political Ads
from 1952 to 2000. Journal of Communication. 52(2): 281-300.

Jost, John, Jack Glaser, Arie Kruglanski and Frank Sulloway. 2003. “Political
Conservativism as Motivated Social Cognition.” Psychological Bulletin 129(3): 339-
375.

Just, Marion, Ann N. Crigler, and Todd L. Belt. 2007. “Don’t give up on hope:
Emotions,Candidate appraisals, and votes.” In The Affect Effect: Dynamics of Emotion in
Thinking and Behavior, eds. W. R. Neuman, G.E., A. Crigler, and M. MacKuen, 231-60.
Chicago: University of Chicago Press

Just, Marion, Matthew Baum, Ann Crigler and Jesse Mills (2010). “Emotions, the
Horserace Metaphor and the 2008 Presidential Campaign.” Presented at the Annual
Meeting of the American Political Science Association. Washington, D.C.
Kaid, Linda and Anne Johnston. 2001. Videostyle in Presidential Campaigns: Style and
Content of Televised Political Advertising. Westport: Prager.

Kahn, Kim and John Geer. 1994. “Creating Impressions: An Experimental
Investigation of Political Advertising on Television.” Political Behavior 16(1): 93-116.

Kahn, Kim. 1993. “Gender Differences in Campaign Messages: The Political
Advertisements of Men and Women Candidates for U.S. Senate.” Political Research
Quarterly. 46(3): 481-502.

Kinder, Donald. 1994. “Reason and Emotion in American Political Life,” In Beliefs,
Reasoning and Decision Making, ed. Roger Shank and Ellen Langer. Mahwah, NJ:
Erlbaum.

King, Erika. 1990. “Thematic Coverage of the 1988 Presidential Primaries: A
Comparison of USA Today and the New York Times.” Journalism Quarterly 67(1): 83-
87.

Koch, Jeffrey. 2008. “Campaign Advertisements’ Impact on Voter Certainty and
Knowledge of House Candidates’ Ideological Positions.” Political Research Quarterly
61(4): 609-621.

Lakoff, George. 1996. Moral Politics: What Conservatives Know that Liberals
Don’t.Chicago: University of Chicago Press.

Lau, Richard and Redlawsk, David. 2006. How Voters Decide: Information Processing
in a Political Campaign. New York: Cambridge University Press.
115

Lau, Richard, and Gerald Pomper. 2004. Negative Campaigning: An Analysis of U.S.
Senate Elections. Lanham, Maryland: Rowman and Littlefield Press.

Lazarsfeld, P. et al. 1944. The People’s Choice. New York: Columbia University Press.

Lazarus, Richard. 1991. Emotion and Adaptation. New York: Oxford University Press.

Leibovich, Mark. 2008. “Between Obama and the Press.” New York Times Magazine.

Luttbeg, N.R. 1974. “Political Linkage in a Large Society.” In ed. Luttbeg Public
Opinion and Public Policy: Models of Political Linkage. P. 1-10. Homewood: Irwin.
Mackie, Diane, Thierry Devos and Eliot Smith. 2000. “Intergroup emotions: Explaining
offensive action tendencies in an intergroup context.” Journal of Personality and Social
Psychology 79(4): 602-616.
Mackie, Diane, Silver, L., and Smith, E.R. 2004. “Emotion as an intergroup
phenomenon”. In C.W. Leach & L.A. Tiedens (Eds.), The Social life of emotions (pp.
227-245). New York: Cambridge University Press.

Marcus, George and MacKuen M. 1993. “Anxiety, Enthusiasm, and the Vote: The
Emotional Underpinnings of Learning and Involvement During Presidential Campaigns.”
American Political Science Review, 87(3): 672-685.

Marcus et al. 2000. Affective Intelligence and Political Judgement. Chicago: University
ofChicago Press.

Marcus, George. 2002. The Sentimental Citizen: Emotion in Democratic Politics.
University Park: Penn State University Press.

MacKuen, Michael et al. 2007. “The Third Way: The Theory of Affective Intelligence
and American Democracy,” In The Affect Effect, ed. Neuman et al. Chicago: University
of Chicago Press.

Martin, Paul. 2004. Inside the Black Box of Negative Campaign Effects: Three Reasons
Why Negative Campaigns Mobilize. Political Psychology 25(4): 545-562.

McCarty, Nolan, Keith Poole and Howard Rosenthal. 2006. Polarized America, The
Dance of Ideology and Equal Riches. Cambridge: Cambridge University Press.

Medvic, Stephen. 2008. “Understanding Campaign Strategy.” Journal of Political
Marketing5(1): 11-32.

116

Mendelsohn, Matthw. 1993. “Television’s Frames in the 1988 Canadian Election.”
Canadian Journal of Communication. 18(2): 149-170.

Mouw, Ted and Sobel, Michael. 2001. “Culture Wars and Opinion Polarization: the
Case of Abortion.” American Journal of Sociology 106(4): 913-943.

Mutz, Diana. 1995. “Effects of Horse-Race Coverage on Campaign Coffers: Strategic
Contributing in Presidential Primaries.” Journal of Politics 57(4): 1015-1042.

Neuman et al. 2007. “Theorizing Affect’s Effects,” In The Affect Effect, ed. Neuman et
al.Chicago: University of Chicago Press.

Patterson, Thomas. 1993. Out of Order. New York: Knopf.

Plato. 1992. The Republic. Translated by G.M.A. Grube. Indianapolis: Hackett.

Poole, Keith and Rosenthal, Howard. 2007. Ideology and Congress. New Brunswick:
Transaction Publishers.

Putnam, Robert. 2000. Bowling Alone: The Collapse and Revival of American
Community. New York: Simon and Schuster.

Rhee, June. 1997. “Strategy and Issue Frames in Election Campaign Coverage: A
Social Cognitive Account of Framing Effects.” Journal of Communication 47(3): 27-48.

Richardson, Glenn. 2001. “Looking for Meaning in all the Wrong Places: Why Negative
Advertising is a Suspect Category.” Journal of Communication 51(4): 775-800.

Ridout, Travis, and Kathleen Searles. 2011. “ It’s My Campaign I’ll Cry if I Want to:
How and When Campaigns Use Emotional Appeals.” Political Psychology 32(1): 1-20.

Roseman, Ira and Smith, Craig. 2001. “Appraisal Theory: Overview, Assumptions,
Varieties,Controversies,” In Appraisal Processes in Emotion: Thory, Methods, Research,
ed. Klaus Scherer et al. Oxford: Oxford University Press.

Roseman, Ira. 1996. “Appraisal Determinants of Emotions: Constructing a More
Accurate and Comprehensive Theory.” Cognition and Emotion 10(3): 241-278.

Rydell, Robert, Diane Mackie, Angela Maitner, Heather Claypool, Melissa Ryan, and
Eliot Smith. 2008. “Arousal Processes and Risk Taking: Consequences of Intergroup
Anger. Personality and Social Psychology Bulletin. 34(8): 1141-52.

117

Scherer, Klaus. 1999. “Appraisal Theory”. In Handbook of Cognition and Emotion,
eds. Tim Dalgleish and Mick Power. New York: Wiley.

Schefufele, Dietram and David Tewksbury. 2007. “Framing, Agenda Setting and
Priming: The Evolution of Three Media Effects Models.” Journal of Communication
57(1): 9-20.

Sinclair, Barbara. 2006. Party Wars: Polarization and the Politics of National Policy
Making. Norman: University of Oklahoma Press.

Smith, Craig and Richard Lazarus. 1990. “Emotion and Adaptation.” From ed. Pervin,
Lawrence, Handbook of Personality: Theory and Research. New York: Guilford Press.
P. 609-637

Smith, Craig and Kirby, Leslie. 2001. “Toward Delivering on the Promise of Appraisal
Theory,” In Appraisal Processes and Emotion: Theory, Methods and Research, ed.
Klaus Scherer et al. Oxford: Oxford University Press.

Smith, Craig, Seger, C. R., & Mackie, D. M. 2007. “Can emotions be truly group level?
Evidence for four conceptual criteria.” Journal of Personality and Social Psychology, 93:
431-46.
Spiliotes, Constantine and Lynn Vavreck. 2002. “Campaign Advertising: Partisan
Convergence or Divergence?” Journal of Politics 64(1): 249-261.

Stevens, Daniel. 2005. “Separate and Unequal Effects: Information, Political
Sophistication and Negative Advertising in American Elections.” Political Research
Quarterly 58(3): 413-425.

Stevens, Daniel et al. 2006. “Local News Coverage in a Social Capital Capital:
Election 2000 on Minnesota’s Local News Stations.” Political Communication 23(1):
61-83.

Stonecash, Jeffrey. 2003. Diverging Parties: Social Change, Realignment and Party
Polarization. Boulder: Westview Press.

Sullivan, Denis, and Roger Masters. 1988. “Happy Warriors: Leaders’ Facial Displays,
Viewers’ Emotions, and Political Support.” American Journal of Political Science 32(2):
345-368.

Tajfel, H. 1974. “Social Identity and Intergroup Behavior.” Social Science Information.
13(2): 65-93.
118

Tajfel, H. and Turner J.C. 1979. “An Integrative Theory of Intergroup Conflict.” In The
Social Psychology of Intergroup Relations, eds. William G. Austin and Stephen Worchel.
Monterey: Brooks/Cole Publishing Co.
Tajfel, H. and Turner, J. C. 1986. “The Social Identity Theory of Inter-Group Behavior.
In Psychology of Intergroup Relations, eds. S. Worchel and L. W. Austin. Chicago:
Nelson-Hall.

Theriault, Sean. 2008. Party Polarization in Congress. New York: Cambridge
University Press.

Thompson, Dennis. 1970. The Democratic Citizen: Social Science and Democratic
Theory in the Twentieth Century. London: Cambridge University Press.

Valentino, Nicholas, Ted Brader, Eric Groenendyk, Krysha Gregorowicz, Vincent
Hutchings. 2011. “Election Night’s Alright for Fighting: The Role of Emotions in
Political Participation.” Journal of Politics 73: 156-170.

Valentino et al. 2008. “Is a Worried Citizen a Good Citizen? Emotions, Political
Information Seeking, and Learning Via the Internet.” Political Psychology, 29(2): 247-
273.

Wattenberg, Martin, and Craig Brians. 1999. “Negative Campaign Advertising:
Demobilizer or Mobilizer.” The American Political Science Review 93(4): 891-899.

Weaver, David. 1996. “What Voters Learn from Media.” American Academy of
Political and Social Science, Vol. 546.

Abstract (if available)

Linked assets

University of Southern California Dissertations and Theses

Conceptually similar

PDF

Partisan media bias vs. structural media bias: which one dominates in a polarized environment?

PDF

The affective framing of tort reform: toward a theory of the mediating effects of emotion on attitude formation

PDF

The interpersonal effect of emotion in decision-making and social dilemmas

PDF

Signaling identity: how race and gender shape what representatives say online

PDF

Neo-liberal hegemony and re-construction of Islamic community discourses in Turkey: the cases of the Gülen and Erenköy communities

PDF

Expressing values and group identity through behavior and language

PDF

The emotional arousal effects on visual processing

PDF

Powellism, Thatcherism and the Conservative Party, 1945-87: the party as the site of ideological transition

PDF

Facilitating children's use of emotional language

PDF

Nihilism's conscience: grounding human rights after Darwin, Marx, and Nietzsche

PDF

The erosion of democracy in the post–Cold War world: understanding the phenomenon and its causes

PDF

The politics of transformation: mass media and the Northern Ireland peace process

PDF

Age differences in jealousy: an ATSS examination of the role of attributions in the emotional reactions of older and younger adults in relationship-threatening situations

PDF

Role of emotion and empathy in moral judgments of real-life situations

PDF

Changing political attitudes and behavior in a diverse America: incorporating individual and contextual determinants

PDF

Emotion, attention and cognitive aging: the effects of emotional arousal on subsequent visual processing

PDF

The boundaries of the intentional

PDF

Modeling emotional effects on decision-making by agents in game-based simulations

PDF

Falun Gong’s evolving definitions through stages and disputes

PDF

Realitics: Speculative economies and the transformation of American politics

Asset Metadata

Creator Mills, Jesse John (author)

Core Title Feeling polarized: emotional partisan polarization and its reinforcement through campaign advertising

School College of Letters, Arts and Sciences

Degree Doctor of Philosophy

Degree Program Politics and International Relations

Publication Date 08/01/2012

Defense Date 05/28/2012

Publisher University of Southern California (original), University of Southern California. Libraries (digital)

Tag American politics,campaign advertising,Elections,emotion in politics,Emotions,OAI-PMH Harvest,partisan polarization,polarization,Political communication

Language English

Contributor Electronically uploaded by the author (provenance)

Advisor Crigler, Ann N. (committee chair), Barnes, John E. (committee member), Goodnight, Gerald Thomas (committee member)

Creator Email jesse.mills@lmu.edu,jessemil@usc.edu

Permanent Link (DOI) https://doi.org/10.25549/usctheses-c3-80833

Unique identifier UC11290299

Identifier usctheses-c3-80833 (legacy record id)

Legacy Identifier etd-MillsJesse-1093.pdf

Dmrecord 80833

Document Type Dissertation

Rights Mills, Jesse John

Type texts

Source University of Southern California (contributing entity), University of Southern California Dissertations and Theses (collection)

Access Conditions The author retains rights to his/her dissertation, thesis or other graduate work according to U.S. copyright law. Electronic access is being provided by the USC Libraries in agreement with the a...

Repository Name University of Southern California Digital Library

Repository Location USC Digital Library, University of Southern California, University Park Campus MC 2810, 3434 South Grand Avenue, 2nd Floor, Los Angeles, California 90089-2810, USA