Partisan Ambivalence and Negative Campaigns: A Survey Experiment

Stephen C. Craig Department of Political Science

University of Florida P.O. Box 117325

Gainesville, FL 32611-7325

Jason Gainous Department of Political Science

University of Louisville 203 Ford Hall

Louisville, KY 40292

Paulina S. Rippere Division of Social Sciences

Jacksonville University 2800 University Blvd. North

Jacksonville, FL 32211 Abstract: Studies show that going negative does not always work in political campaigns, and yet candidates and consultants are rational people whose experience has persuaded them that it can be a winning strategy under the right circumstances. As scholars continue to explore what those circumstances might be, recent work by Lavine, Johnston, and Steenbergen (2012), suggests that when a stimulus/cue prompts partisan ambivalence, motivated reasoning should vitiate and a focus on the substance of the frame should increase. Based on this logic, it follows that a campaign attack against one’s opponent will be more effective among voters who express a mix of positive and negative feelings toward the parties because they are more focused on the substance of the attack than those who are less ambivalent. The following study uses experimental data derived from a national Internet survey of registered voters to examine the effectiveness of both campaign attacks and candidates’ responses (rebuttals) to those attacks among subjects with varying levels of partisan ambivalence. Our results show that ambivalence plays an occasionally meaningful but inconsistent moderating role across a range of campaign scenarios, more so with attacks than with responses. Paper presented for delivery at the Annual Meetings of the American Political Science Association, Washington, DC, August 28-31, 2014. We would like to thank Roger Austin for his assistance in designing this experiment, Jim Kitchens for facilitating the Internet survey used in our analysis, and Marissa Silber Grayson for her contributions to two companion papers (Craig, Rippere, and Grayson 2012, 2015).

In January 2014, the Gallup organization released a report purporting to show that a

“Record-High 42% of Americans Identify as Independents” – a figure that by mid-summer had

inched even closer to the 50-percent mark.1 Despite the belief in some circles that the American

electorate has become increasingly less partisan over the past several decades (Dalton 2013),

both academics and pundits often point to heightened partisanship as a defining feature of

contemporary mass politics in the United States. For over half a century, scholars have

understood that most citizens identify (to a greater or lesser degree) with one of the major

parties, and that this attachment provides a powerful cue that shapes their attitudes and especially

their choice of candidates on Election Day (Campbell et al. 1960; Green, Palmquist, and

Schickler 2002). Even during the so-called “dealignment” era of the 1960s and 1970s (Nie,

Verba, and Petrocik 1976; Norpoth and Rusk 1982; Dalton 1984), party remained a central

element in the political identity of many Americans and, indeed, by the 1980s, signs of

“resurgent” partisanship were beginning to emerge (Bartels 2000; Hetherington 2001). What

about now? There is reason to believe that the high number of self-identified Independents

reported by Gallup and others (Pew Research Center 2012, 13) is misleading because it includes

many individuals who fall into the “leaner” category, i.e., after initially claiming to be neither

Democrats nor Republicans, they acknowledge a preference for one or the other. When leaners

are counted as partisans (as perhaps they should be; see Magleby, Nelson, and Westlye 2011;

Magleby and Nelson 2012), the proportion of Independents drops from a plurality of the

electorate to less than 20 percent.2

According to Marc Hetherington (2001, 628), resurgent mass partisanship in the late 20th

century (involving an “impressive increase in party-centric thinking” rather than simply more

identifiers3) was largely driven by changes that occurred at the elite level; specifically, “[g]reater

2

ideological polarization in Congress has clarified public perceptions of party ideology, which has

produced a more partisan electorate” (p. 629). That party leaders, in Washington and throughout

much of the country, have become more ideologically polarized is beyond dispute. Republicans

in Congress, for example, are now more conservative – and more uniformly conservative – than

in the past, while their Democratic counterparts are more consistently liberal (McCarty, Poole,

and Rosenthal 2001; Stonecash, Brewer, and Mariani 2003; Theriault 2008; Noel 2013).4 But

there is a difference of opinion as to whether this sharp division between left and right has been

accompanied by a similar drift toward the ideological extremes among the general public

(Abramowitz 2010), or whether it just looks that way because citizens have increasingly

followed party cues and “sorted” themselves into the correct categories, i.e., with most liberal

Republicans moving to the Democratic side of the aisle, and conservative Democrats (notably in

the South) switching to the GOP (Fiorina with Abrams and Pope 2011; Levendusky 2009; Pew

Research Center 2014a; Fiorina 2014).5

We are more concerned, however, with the nature and consequences of mass partisanship

than with ideology per se. Political conflict in the United States today is frequently portrayed as

being bitterly partisan, us against them, with both sides faulting the other for its failure to address

important national problems. Among Republican and Democratic candidates, officeholders, and

party leaders, such hyperpartisanship (reflecting “a sharply polarized situation in which political

parties are in fierce disagreement with each other”6) and ideological polarization seem clearly to

go hand in hand: Conservative Republicans and liberal Democrats cannot find common ground

on which to base policy action. For the general public, this relationship is not so clear. Despite

evidence of increased issue consistency in recent years (that is, fewer citizens expressing a mix

of liberal, moderate, and conservative opinions; see Abramowitz 2010; Pew Research Center

3

2014a), many rank-and-file partisans do not share the same “extreme” policy views that

predominate among their respective party elites (Ellis and Stimson 2012; RePass 2008; Pope

2012; Pew Research Center 2014a; Fiorina 2014; Diggles and Hatalsky 2014). A lack of

ideological congruency does not, of course, prevent individuals from positively identifying with

a party, nor does it preclude their developing a strong dislike for the opposition. But it does

suggest that partisan attachments at the grassroots may be rooted in something other than shared

values or policy preferences (e.g., Iyengar, Sood, and Lelkes 2012; Pew Research Center 2014a)

– and, as a result, that citizens’ feelings toward one party or both may be more equivocal than

one would expect during a period characterized by elite hyperpartisanship and (in response)

seemingly “resurgent” partisanship among the general public.

In fact, there is a growing body of research indicating that many Americans, including a

fair number who consider themselves to be either Democrats or Republicans, simultaneously

possess both positive and negative feelings about the party with which they identify and/or the

one with which they do not, that is, they are ambivalent partisans (Greene 2005; Mulligan 2011;

Thornton 2011, 2013; Lavine, Johnston, and Steenbergen 2012). Further, this blend of positive

and negative has a variety of consequences; compared to their univalent counterparts, for

example, ambivalent partisans appear more likely to correctly perceive candidate positions

(Lavine, Johnston, and Steenbergen 2012), more likely to use issues (Lavine, Johnston, and

Steenbergen 2012) and ideology (Thornton 2011) when evaluating candidates and making vote

decisions, more likely to cast a split ballot (Mulligan 2011), and less likely to exhibit a variety of

party-oriented attitudes and behaviors (such as being active on behalf of one’s party and voting

for the presidential candidate of that party; see Greene 2005). In this paper, we seek to determine

whether ambivalence also helps to shape voters’ reactions to both campaign attacks and

4

candidates’ responses to those attacks.

Partisan Ambivalence and Negative Campaigning

Scholars have increasingly embraced the idea that people do not always have a single

“true” attitude about a given topic; to the contrary, they often possess multiple and even

contradictory attitudes that they might draw upon, for example, when thinking about a policy

issue or deciding which candidate to vote for in an election (Zaller and Feldman 1992; Zaller

1992). When someone’s evaluations, beliefs, or emotions concerning an attitude object are in

conflict with one another or, more simply, when they simultaneously possess positive and

negative evaluations of an attitude object, that person can be described as being ambivalent

(Alvarez and Brehm 1995; Craig et al. 2005; Eagly and Chaiken 1993). Prior studies have

examined ambivalence across a range of contemporary issues, including abortion (Alvarez and

Brehm 1995; Craig, Kane, and Martinez 2002), gay rights (Craig et al. 2005), and social welfare

(Gainous 2008; Gainous and Martinez 2005; Zaller and Feldman 1992). As noted earlier, there

also is a growing body of research focusing on ambivalence toward the political parties.

While they were not the first to examine partisan ambivalence (e.g., see Greene 2005;

Mulligan 2011; Thornton 2011, 2012), our theoretical framework draws heavily from Lavine,

Johnston, and Steenbergen’s (2012) explanation for how and why voters who are ambivalent

may come to think more deeply and thoroughly than others when making political judgments.

These authors suggest that partisanship functions both as a stable psychological construct based

in affective attachment to the party, and as a temporal summary judgment of party performance.

In line with theories of cognitive consistency (Festinger 1957; Greenwald et al. 2002; Heider

1958), they assert that citizens strongly prefer there to be harmony between attachment and

summary judgment. When the information flow regarding one’s own party trends negative,

5

however, the ability to maintain harmony and to objectively justify their support for that party is

compromised. As a result, the individual is likely to become an ambivalent partisan.

Inherent in this argument is the idea that voters are motivated to have “resolve” when it

comes to their attitudes; in other words, they prefer to possess an objectively justifiable attitude

that avoids ambiguity, uncertainty, and doubt (Frenkel-Brunswik 1949; Kruglanski and Webster

1996) because to experience doubt is unsettling and uncomfortable. The problem here is that if

one’s partisanship, long viewed by scholars (Campbell et al. 1960) as a filter or perceptual screen

through which people evaluate the political world, is riddled with ambivalence, it may become

blurred and the ability to justify one’s attitude through the prism of partisanship will no longer be

effective. Lavine and colleagues (2012) posit that such a development encourages voters to think

more deeply or deliberatively about their political options. This argument is framed around the

idea that citizens are motivated to find a sufficiency threshold, or level of confidence, when

making a decision (Chaiken et al. 1989; Downs 1957; Payne, Bettman, and Johnson 1993).

Specifically, the sufficiency threshold “provides a commonsensical mechanism for predicting

when – and more important, why – decision makers will transition from heuristic to deliberative

thinking” (Lavine, Johnston, and Steenbergen 2012, 35).

The transition supposedly occurs when partisan cues no longer furnish enough confidence to

create a settled or resolved preference in a given judgment option. A high degree of confidence is

not present for ambivalent partisans because partisanship no longer serves as an effective heuristic,

and so they are compelled to look elsewhere to objectively justify their position. Our contention is

that this explains why ambivalent partisans are more likely than univalent partisans to weigh the

actual content of a negative ad and incorporate it into their own position. They are compelled to do

so in order to pass the sufficiency threshold, and thereby to attain the necessary resolve for their

6

personal comfort or satisfaction. Thus, we expect our experimental results to show that ambivalent

partisans are more likely than their univalent counterparts to have both their vote choice and

evaluations of candidates influenced by exposure to an attack against their party’s candidate. We

further anticipate that ambivalence will affect how voters react to the counterframes that usually

follow, that is, to the responses (or rebuttals) made by those candidates when they are attacked. We

develop our argument more fully in the following section.

Attack, Response, and Ambivalence

Candidates, consultants (especially media specialists), party leaders, and political journalists

believe that the effects of campaign ads are often substantial and occasionally decisive. There is no

shortage of anecdotal evidence to suggest that they are correct. In addition to the well-known Willie

Horton and “prison furlough” ads that the Bush campaign ran against Michael Dukakis in 1988, and

the independently sponsored Swift Boat ads questioning John Kerry’s military service in 2004,

political junkies can name countless races over the years that conventional wisdom insists might

have turned out differently if not for a well-conceived and well-timed ad or series of ads, usually

delivered over the television airwaves (see Diamond and Bates 1992; Johnson-Cartee and Copeland

1997; Mark 2006; Westen 2007; Trent, Friedenberg, and Denton 2011).

It is not a coincidence, by the way, that the campaign ads most remembered for having had a

major impact were negative in tone. While positive appeals are usually a central component in any

candidate’s overall communications package (Johnson-Cartee and Copeland 1997; Franz et al.

2008), attacks are more frequently seen as game-changers or game-deciders – particularly for

challengers, who are thought to have little chance of winning unless they go negative (Mayer, 1996;

Geer, 2006). Nevertheless, while negativity can have its rewards, there is some risk involved.

According to Dennis Johnson (2007, 84), “[v]oters are not fools and understand when campaign

7

commercials have crossed the line of fairness.” Tow that line, though, and negative ads can be the

difference between victory and defeat on Election Day. This, at least, is the common wisdom among

political practitioners.

Academic studies, however, have yielded mixed results regarding the persuasive effects of

campaign ads generally,7 and of negative advertising (or other forms of negative campaigning) in

particular (Lau and Rovner 2009; Fridkin and Kenney 2011; Fernandes 2013). Rather than shaping

candidate preference directly, negative ads are believed to be important primarily because they serve

to activate existing predispositions (often rooted in voters’ partisan attachments), reinforce initial

preferences (based to a large degree on the same attachments), or mobilize/demobilize segments of

the electorate that are more/less supportive of the sponsoring candidate. Still, it seems likely that the

inconsistent findings of this literature are due at least in part to the fact that negativity lies in the eye

of the beholder, that is, “whether a tactic, a candidate, or a campaign is [perceived as] negative

depends on whose ox is being gored” (Sigelman and Kugler 2003, 144). The fact that self-identified

partisans react differently to attacks coming from the other side than they do to accusations made by

candidates of their own party should therefore not come as a surprise (Iyengar, Jackman, and Hahn

2008; Ansolabehere and Iyengar 1995; Stevens et al. 2008; Geer 2006; but see Franz and Ridout

2007). What we seek to determine here is whether voters with ambivalent feelings about their party

are more open to persuasion than those whose feelings are more uniformly positive.

It has long been recognized that partisanship serves to simplify and guide citizens’

political judgments (Campbell et al. 1960; Popkin 1991; Zaller 1992; Lau and Redlawsk 2006;

Lewis-Beck et al. 2008). Theory suggests that these judgments are often the product of motivated

reasoning (Lodge and Taber 2013; Lavine, Johnston, and Steenbergen 2012; Lebo and Cassino

2007; Slothuus and de Vreese 2010), whereby individuals are motivated to mitigate any internal

8

conflict they may be experiencing – as, for example, when a candidate from their preferred party

is criticized by her opponent during a campaign. Partisanship works as a heuristic, or cognitive

shortcut, to facilitate this process, as citizens are exposed to new information prior to reaching a

decision. Little deliberation is required for many identifiers, as “confirming preexisting biases

seems to be a priority” (Lau and Redlawsk 2006, 24). On the other hand, Lavine, Johnston, and

Steenbergen (2012) contend that motivated reasoning is vitiated among those who are

ambivalent about their party because partisanship no longer serves as a sufficient heuristic;

rather, more deliberative thinking is required for the decision maker to find resolve. We suspect

that this theoretical framework can help us to understand how voters respond to negative

campaigning. Specifically, ambivalent partisans should be more likely to focus on the substance

of a negative frame because their own attachment is of less utility in helping them to decide

whether to either accept or dismiss the claim therein. Thus:

H1: An attack made against one’s fellow partisan will be more effective among voters

who express a high degree of ambivalence about their party.

Our focus in this paper is almost entirely on citizens’ reactions when a fellow partisan is

attacked. The reason is simple: All else equal (as it is in our experimental simulation), few

identifiers are likely to be moved by an attack on the opposition candidate because most do not

plan to vote for that candidate in the first place, i.e., there is a ceiling (or floor) effect that leaves

little room for a meaningful erosion of support.8 Similarly, our analysis centers on the effects of

ambivalence about one’s own party rather than ambivalence about the opposition. In the case of

H1, one might normally expect that a high level of out-party ambivalence will render the attack

less credible, hence less effective. As a practical matter, however, our data will show that (a)

most respondents feel relatively little ambivalence about the opposing party; instead (b) their

9

feelings tend to be decidedly negative overall, with the result that (c) if anything, an attack

against one’s fellow partisan might actually be more effective among those who express a high

degree of ambivalence toward the party of the attacker. Given such theoretical uncertainty, our

analysis centers mainly on the moderating effects of ambivalence about one’s own party.

We will, however, look more closely at identification with and ambivalence toward the

attacking party in one instance. As noted above, a potential downside of going negative is that it

can generate backlash against the sponsoring candidate, especially if the attack is seen by voters

as having “crossed the line of fairness” (Johnson 2007, 84).9 In our study, backlash is measured

most directly by change in favorability ratings for the attacker (vote preference and target

favorability are our other two dependent variables) – and it is here that we expect ambivalence

toward that candidate’s party to be important, at least among the attacker’s fellow partisans.

Following the same logic that underlies H1:

H2: An attack delivered by one’s fellow partisan will generate greater backlash (lower

favorability for the attacker) among voters who express a high degree of ambivalence

about their party.

What happens, then, after an attack has been made? One option is for target candidates to

try and stay on message while ignoring the charges made against them. Few practitioners today

consider this to be a prudent strategy (Westen 2007; Craig and Hill 2011) and, as a consequence,

almost any campaign attack is likely to be followed by some sort of rebuttal. The literature

dealing with attacks themselves is quite rich, if also problematic in terms of the clarity of its results,

but there are only a handful of studies that examine the effects of candidate responses to those

attacks.10 In the present context, we are interested in determining whether the role played by

partisan ambivalence is similar on the back end of this two-way flow of campaign communication

10

to what it is on the front.

When candidates respond to an attack, they are in effect offering voters a counterframe, that

is, a way of viewing the original charge differently than the attacker intended (see Jamieson 1992,

106-111). Although recent work by Chong and Druckman (2013) suggests that individuals with

“weak” attitudes are more likely to be influenced by counterframes than those who possess “strong”

attitudes,11 their argument may not apply in the campaign scenario envisioned here. In particular,

does it make sense to predict that a response made by one’s fellow partisan will be more effective

among voters who express a high degree of ambivalence about their party? This outcome would run

counter to the logic underlying H1: that cognitive discomfort caused by ambivalence frequently

makes out-party messages more effective. Accordingly, we hypothesize the following:

H3: A response made by one’s fellow partisan will be less effective among voters who

express a high degree of ambivalence about their party.

Some responses may be more effective than others, of course (Craig, Rippere, and Grayson

2015), but we have no a priori reason to doubt that the moderating effect of partisan ambivalence

will be evident across all five types of responses/counterframes tested in our study.

Finally, we again will consider whether ambivalence plays a role in shaping the impact of

responses on challenger (attacker) favorability among that candidate’s co-partisans. The term

“backlash” might not be appropriate as it implies a negative reaction to a candidate’s own words

or actions – and with responses, the words in question are those of the original target rather than

the original attacker. Nevertheless, when candidates respond to an attack they often do so in the

hopes of accomplishing more than simply restoring the status quo; in addition to improving their

own standing with voters, they (along with their consultants) may attempt to craft a message that

inflicts damage on the opponent as well. This is especially true for the two types of responses in

11

our study that are explicitly negative in tone (see below), but virtually all campaign rebuttals

contain at least implied criticism of the attacker. Our expectation, then, is that impact of such

rebuttals on the attacking candidate’s co-partisans will be similar to the backlash predicted to

follow (H2) as a result of the attacks themselves:

H4: A response made by an opposing partisan will generate lower favorability ratings for

the attacker among voters who express a high degree of ambivalence about their party.

Research Design

The research reported here is part of a larger project in which we assess the overall

effectiveness of two campaign attacks, and of five different types of responses that a candidate

might make in an effort to mitigate the damage inflicted when his opponent decides to go negative

(see Craig, Rippere and Grayson 2015; Craig and Rippere 2013). Findings are based on a within-

subjects controlled experiment in which a national sample of 662 registered voters participated in an

Internet survey that was conducted June 21-24, 2012. Our data were provided by qSample (see

www.qSample.com), a market research firm that has recruited over five million individuals to

participate in research projects related to video gaming, home building/contracting, home

ownership, issues of particular interest to college students and Baby Boomers, as well as politics.

Respondents for our survey were drawn from a national panel of registered voters, geographically

balanced by region, whose members engage in online polling, ad testing, focus groups, and in-depth

interviewing on a range of politically relevant topics. Although the sample is quite diverse, we

make no claim that it is representative of all registered voters nationwide. 12

The structure of our experiment was as follows: Respondents were randomly assigned to

one of twenty treatment groups (or orders) and asked to complete a background questionnaire that

measured basic demographics, political knowledge, party identification, feelings about each of the

12

major parties, and a number of other political orientations. They were then told to imagine that it

was the fall of 2012, and one of the races on their ballot involved a congressional matchup between

an incumbent seeking a third term and an experienced challenger who had served in both local

office and the state legislature. After reading short biographies of the candidates,13 participants

were asked to indicate a vote preference (“Based on the information you currently have, which

candidate would you vote for if the election were held today?”) and to rate each candidate on a 7-

point scale ranging from “very unfavorable” (1) to “very favorable” (7); answers to these three

questions serve as the dependent variables for our analysis. Participants subsequently read what

was described as a direct-mail attack14 by the challenger and once again registered their vote

choice and candidate assessments. Finally, they were given the incumbent’s response and

answered the vote and candidate questions a third time.15

In crafting the attacks, we opted not to employ policy appeals that would likely be viewed

through a partisan/ideological lens by many voters (Iyengar, Jackman, and Hahn 2008; also see

Druckman, Peterson, and Slothuus 2013; Lavine, Johnston, and Steenbergen 2012). Our focus was

instead on performance-based attacks wherein a challenger alleges that “the incumbent has lost his

touch with the people back home, doesn’t work hard, doesn’t stand on principle and changes his

mind to please different people, has used the office for personal gain, [and] will say just about

anything to get reelected” (Johnson 2007, 65). This seems to be an especially good basis for

evaluation when the target is an incumbent, as is the case in our experimental manipulations.

The attack used in the following analysis (taking advantage) charged the incumbent with

“helping himself to other people’s money” in a number of ways, i.e., voting for pay raises, using

party money to finance a family vacation, renting office space from his brother at inflated prices,

overbilling clients for professional services, and channeling no-bid government contracts to

13

campaign donors.16 These allegations, regarding matters that are clearly relevant to the target’s

performance as an elected representative, were made in language that might not be considered

“uncivil” by Fridkin-Kenney (2008) standards but comes very close to that line by using such

emotionally charged terms as “corrupt,” “immoral,” and “deserves to be in jail more than in

Congress.”17

Each participant read the text for taking advantage plus one of five responses:

• denial (claiming that the charges are false);

• counterimaging (a positive message that flips the attack on its head by “laying out for

the voter a counterproposition to the content of the opponent’s negative ad,” e.g., police

officers praising the record of someone accused of being soft on crime; see Johnson-

Cartee and Copeland 1991, 244);

• justification (acknowledging the behavior in question and accepting responsibility, but

attempting to downplay its negative consequences);

• counterattack (the content of which often has little or nothing to do with the original

attack); and

• accusing the opponent of mudslinging (a weak form of counterattack which may suggest

to some voters that the target candidate is merely trying to deflect attention away from

charges that are both damaging and true).

These responses are summarized in Table 1, with complete wording available via our online

appendix. The impact of the ads was tested empirically using the following design:

1 (Attack Type: taking advantage) X 5 (Response Type) X 4 (party affiliation + gender

combinations) = a total of 20 treatment groups, to which participants were randomly

assigned.

14

Because we wanted to know whether the effectiveness of ads and responses varied by candidate

gender, shared partisanship, and/or shared gender, four treatment conditions were created for each

attack-response pair: male Republican incumbent attacked by female Democrat, female Republican

incumbent attacked by male Democrat, male Democratic incumbent attacked by female Republican,

and female Democratic incumbent attacked by male Republican. 18 The role of gender (fairly

modest as it turns out) is explored more fully in Craig and Rippere (2013).

Table 1 about here

Our measures of partisan ambivalence are modeled on prior work by scholars studying

ambivalence toward issues such as abortion (Craig, Kane, and Martinez 2002), gay rights (Craig

et al. 2005), and social welfare (Gainous and Martinez 2005), as well as attitudinal ambivalence

generally (Thompson, Zanna, and Griffin 1995; Martinez, Gainous, and Craig 2012). The

approach is based on the idea that citizens’ overall attitudes are made up of both positive and

negative elements. To measure those two dimensions, respondents are asked to indicate both

how positively and how negatively they view an attitude object. For the present study, we

adapted language used in our earlier studies (Craig, Kane, and Martinez 2002; Craig, Kane,

Martinez, and Gainous 2005; Gainous and Martinez 2005) to accommodate the structure of an

Internet survey:

“We want to know how you feel about the two major political parties in American

politics today. Please indicate how positively you feel about each party in the following

manner: If you do not have any positive feelings about the party, give it the lowest rating

of 1; if you have some positive feelings, rate it a 2; if you have generally positive

feelings, rate it a 3; and if you have extremely positive feelings, rate it a 4. Please rate

each party based solely on how positively you feel about it, while ignoring or setting

15

aside for the moment any negative feelings you may also have.”

Respondents then read the name of each party and were prompted to rate each party separately.

Later in the survey, the above statement was repeated except with the words “positive” and

“positively” changed to “negative” and “negatively” (see Thompson, Zanna, and Griffin 1995;

Osgood, Suci, and Tannenbaum 1957).

While other methods of measuring ambivalence have been used widely within the social

psychology literature (see Kaplan 1972), an advantage of the model employed here is that it

accounts for the presence of polarized beliefs (Thompson, Zanna, and Griffin 1995; Priester and

Petty 1996; Craig, Kane, and Martinez 2002; Craig et al. 2005; Gainous and Martinez 2005). For

example, someone who rates one of the political parties as a 4 on the positive component and as

a 2 on the negative component probably can be viewed as experiencing less ambivalence than

someone who answers 2 on both. Megan Thompson and colleagues (1995) originally accounted

for this by devising a method that includes both similarity and intensity of components, i.e.,

reasoning that increased similarity between positive and negative components reflects greater

ambivalence. In addition, they proposed that while holding similarity constant, increased

intensity should lead to greater ambivalence. Putting it all together, the argument is that

Ambivalence = [(P + N)/2] - |P - N|

where P is the positive reaction score and N is the negative reaction score.19 Individual scores

range from –0.5 ("extremely" positive and no negative feelings, or "extremely" negative and no

positive) to +4.0 ("extremely" positive and negative feelings for the same party). The Thompson

study (also Priester and Petty 1996), found the similarity-intensity (or SIM) model to be superior

to others, including Kaplan’s, at predicting subjective ambivalence, or the degree to which

subjects reported feeling discomfort when asked to provide an evaluation of some attitude object.

16

Results

In the aggregate (including Independents, other-party identifiers, and those who say

they’re not sure),20 our respondents viewed the Republican and Democratic parties in much the

same light: roughly seven in ten expressed “no” or only “some” positive feelings, while a little

over 40 percent said they had either “generally” or “extremely” negative feelings. These findings

are consistent with polls indicating that neither party today is held in particularly high regard by

the American public as a whole.21 The more important question for our purposes is whether

feelings about the parties are one-sided or mixed for most citizens. It turns out that for both the

Republican Party (36.4 percent) and the Democratic Party (36.8 percent), just over one-third of

the sample expressed little or no ambivalence, i.e., they register SIM scores of zero or below. In

sharp contrast, approximately one in four exhibited what we categorize as a “high” degree of

ambivalence (a SIM score of 2 or above): 23.2 percent of all respondents were highly ambivalent

about the Republican Party, 25.3 percent about the Democratic Party, while the rest (40.5 and

37.9 percent, respectively) fell in the middle. These numbers do not prove that most citizens are

profoundly ambivalent about the parties. They demonstrate beyond any doubt, however, that (a)

some feelings of partisan ambivalence are more the norm than the exception; and (b) a sizable

minority of the electorate possess highly conflicted views toward one party, the other, or both.22

A more nuanced picture is provided in Table 2, where ambivalence is broken down by

respondents’ self-identifications. There are two sets of entries, Part I showing how Democrats (N

= 271) and Republicans (N = 226) feel about their own party, and Part II revealing how those

same groups feel about the opposition (at this point, other-party identifiers and nonidentifiers are

excluded23). It is hardly surprising that most people viewed their party much more favorably, and

much less unfavorably, than they did their rivals (Pew Research Center 2014a). For example,

17

65.5 percent of Republicans said they felt “generally” or “extremely” positive about the GOP,

while 85.4 percent felt “generally/extremely” negative (and a mere 1.7 percent “generally/

extremely” positive) about the Democratic Party. Similarly, 65.3 percent of Democrats felt

“generally/extremely” positive about their party, compared with 74.6 percent who felt

“generally/extremely” negative (and 1.5 percent “generally/extremely” positive) about the GOP.

Three points stand out here. The first is that identifiers, in the aggregate, expressed considerably

less positive affect for their own party than negative affect toward the opposition (Iyengar, Sood,

and Lelkes 2012). Second, some equivocation was evident even among the majority who held

their party in relatively high regard, i.e., just 6.6 percent of Republicans and 12.2 percent of

Democrats had “extremely” positive feelings about their respective parties. Third, more than

one-third of both partisan groups expressed only “some” or even “no” positive feelings, and

almost three-quarters (73.9 percent Republicans, 74.5 percent Democrats) said they had “some”

negative feelings, toward their own party.

Table 2 about here

These numbers suggest, and a further examination of the results in Table 2 confirms, that

moderate-to-high ambivalence is not uncommon among voters – though such ambivalence is

more likely to be directed at one’s own party than at the opposition. For example, 70.8 percent of

Republicans and 71.2 percent of Democrats simultaneously expressed “some/generally” positive

feelings and “some” negative feelings about their party; in contrast, 67.7 percent of Republicans

had “generally/extremely” negative along with “no” positive feelings about the Democrats, and

56.1 percent of Democrats exhibited the same configuration of attitudes regarding the GOP. The

predominant pattern, then, was for voters to see the opposition in unambiguously negative terms

while simultaneously feeling at least moderately conflicted about their own party. Something

18

along the lines of: We’re not perfect, but the other guys are worse. And it is ambivalence about

one’s own party that we expect to play a role in shaping citizens’ reactions to negative campaign

communications.

Results support some, but not all, of our expectations regarding the role that ambivalence

plays in moderating the effectiveness of campaign attacks. According to H1, attacks against a

fellow partisan will be most effective among respondents who are highly ambivalent about their

party. Column 1 in Table 3 reveals two things about ambivalent vs. univalent partisans:24 First,

even before the attack, based solely on the biographical information provided, those individuals

who expressed either high or moderate levels of ambivalence were significantly less likely (80.6

and 86.9 percent, respectively, p < .01) than their low-ambivalence counterparts (97.6 percent) to

say they would vote for their party’s candidate. Second, high- and moderate-ambivalence voters

also exhibited larger post-attack declines in support for that candidate (33.3 and 30.8 points,

respectively, compared with just 21.4 points for those who experienced little or no cognitive

conflict regarding their party). While this latter finding, in particular, is consistent with H1, none

of the differences observed between ambivalence groups is significant at conventional levels.

We do, however, see a similar pattern for incumbent favorability, where the mean score

(on a 7-point scale) dropped most among the highly ambivalent and least among univalent

partisans as a result of the attack – and in this instance, differences between high- and moderate-

ambivalence voters on the one hand, and low-ambivalence voters on the other, are significant

(both p < .05). It is important to note that the real contrast revealed in the first two columns of

Table 3 is between the univalent and everybody else, i.e., the effect of what we characterize as

“moderate” and “high” ambivalence is essentially the same. With that caveat, and keeping in

mind the marginal significance of differences between high/moderate and low ambivalence

19

groups on vote preference (p < .15), we might conclude that ambivalence about one’s own party

– a phenomenon which is fairly widespread if not universal – makes voters at least somewhat

more receptive to campaign attacks by the opposition.

Table 3 about here

This effect does not extend to favorability ratings for the attacking candidate. Although

those ratings declined sharply and significantly following the attack, backlash was of comparable

magnitude across all three ambivalence categories. Likewise, and contrary to what we expected,

ambivalence toward their own party did not contribute to greater backlash among the attacker’s

co-partisans (column 4), i.e., lower post-attack challenger favorability is evident in roughly equal

measure across the board. H2 is therefore disconfirmed.

To what extent, if at all, does ambivalence come into play when candidates respond to

campaign attacks made by their opponent? H3 posits that a response by one’s fellow partisan

will be less effective among voters who are highly ambivalent about their party. Table 4 displays

in-party shifts in vote preference and incumbent favorability occurring between t2 (post-attack)

and t3 (post-response). With regard to vote choice (top half of the table), our results do not

support H3 in any instance. Among the least ambivalent (whom we believed would therefore be

most responsive to a fellow partisan’s campaign rebuttal), change from t2 to t3 was significant (p

< .10)26 only for a response in the form of a justification. Contrary to expectations, a significant

shift (p < .05) toward the incumbent occurred among high- and moderate-ambivalence voters for

three of the five responses: denials, justifications, and counterattacks. Only in the case of denial,

however, were cross-group differences statistically significant (high- vs. low-ambivalence, p <

.10), i.e., individuals who were highly ambivalent about their party moved in substantially

greater numbers than the univalent (45.5 vs. 14.3 percent) into the incumbent’s column following

20

his challenge to the accuracy of the charges made in taking advantage. While weak overall, these

results are less in line with our own expectations (rooted in the theoretical framework suggested

by Lavine, Johnston, and Steenbergen 2012) than they are with the proposition that people with

weak attitudes will be more influenced by any counterframe than those who possess stronger

attitudes (Chong and Druckman 2013).

Table 4 about here

When we consider incumbent favorability (bottom half of Table 4), H3 fares no better.

Significant change (p ≤ .10) in the form of increased ratings for one’s co-partisan occurred from t2

to t3 among all three ambivalence groups and for each of the five responses. Only for denial was the

magnitude of change less (as hypothesized) for the highly ambivalent than for the least ambivalent,

and this difference cannot be interpreted as having occurred other than by chance (p = .328). A word

of caution is in order here: Because the number of respondents involved in any given comparison is

usually quite small (see note 26), tests of statistical significance may at times cause us to overlook

even large differences between groups. Unfortunately for our argument, there are enough instances

(for both vote choice and incumbent favorability) where the direction of the relationship is opposite

from what we predicted in H3 that significance is not an issue. Indeed, in the one instance where the

degree of change for incumbent favorability is significantly different for high- vs. low-ambivalence

respondents (accusations of mudslinging, p < .10), it is the former who moved the most.

We saw in Table 3 that the original attack generated roughly equivalent levels of backlash,

in the form of lower favorability ratings for the challenger/attacker, among the incumbent/target’s

co-partisans regardless of their ambivalence level. The top portion of Table 5 indicates that a second

wave of backlash occurred following most rebuttals, though a clear pattern is difficult to discern.

For example, challenger favorability actually rose from t2 to t3 among (a) the highly and moderately

21

ambivalent who viewed the counterimaging response and (b) univalent voters who received the

mudslinging treatment.27 Where ratings did fall as expected, sometimes the largest drop was evident

among the highly ambivalent (justifications, mudslinging) and other times among those who were

moderately ambivalent (denials, counterattacks). Also, small N’s notwithstanding, there are three

instances in which the magnitude of post-response backlash was significantly different (p < .10 or

better) across ambivalence groups – but these differences were not always in the same direction (a

sharper decline in favorability among high/moderate-ambivalence voters for justifications and

mudslinging, but among low-ambivalence voters for counterimaging). Overall, the inconsistency

and overall weakness of our findings here raise further doubts about the relationships posited in H3.

Table 5 about here

In the bottom half of Table 5, we consider the impact of responses on challenger/attacker

favorability among her fellow partisans. Our hypothesis (H4) is that the impact of such responses

will be negative (generating lower favorability) in general, but more so among individuals who

express a high degree of ambivalence about their party. Once again, our results are unimpressive.

Just three of five responses tested (denials, justifications, counterattacks) resulted in significantly

lower challenger favorability (p < .10) among that candidate’s co-partisans, and in only a single

instance (justifications) was the decline from t2 to t3 significantly larger (p < .05) among the highly

ambivalent than among the univalent. As with H2, our expectations in H4 are not supported:

Ambivalence had no consistent effect on the attacking candidate’s standing among fellow partisans,

as a result of either the attack itself or the responses that followed.

Finally, we make note of the net effects resulting from our simulated campaign exchange,

from pre-attack to post-response. There are three possibilities: the incumbent/target could have

emerged as (a) weaker than before, (b) stronger than before, or (c) in roughly the same position

22

that existed prior to the attack. Results (not shown; see our online appendix for additional details,

also Craig, Rippere, and Grayson 2015 for a fuller discussion) indicate that, with regard to vote

intention, the effect of all five responses was essentially to restore the status quo; that is, for most

of the incumbent’s fellow partisans, regardless of their level of ambivalence, no significant net

movement occurred from t1 to t3.28

The data have a more interesting story to tell about our two measures of candidate

favorability. First, for most responses, the effect of attack plus response was lower incumbent/

target favorability among all respondents (Craig, Rippere, and Grayson 2015) and, specifically,

among her fellow partisans. In H1 and H3, we predicted opposite reactions by the latter group,

i.e., that an attack (by an opposing partisan) would be more effective, but a response (by a fellow

partisan) less effective, among individuals who felt highly ambivalent about their party – with

the cumulative effect of the exchange being something of a double whammy among the target’s

ambivalent co-partisans. This is more or less what happened for incumbent favorability, where

scores were consistently lower post-response than they had been prior to the attack. For what it is

worth, we note that these t1-t3 changes were statistically significant (p < .10 or better) much more

often among the highly ambivalent (all responses except denials) and moderately ambivalent (all

except counterimaging) than among the univalent (p < .10 only for mudslinging).29 One might

wonder whether lower favorability scores really matter when the distribution of vote intentions is

roughly the same at the end of our experiment as it was at the beginning. We answer this

question by pointing out that candidate evaluations are sometimes believed to be the most

proximal determinants of vote choice (Kelley and Mirer 1974); thus, it is possible that the altered

perceptions documented here could ultimately influence the election outcome even if they do not

have an immediate (or direct) effect on the vote decision itself.

23

Incumbents were not the only ones whose favorability remained lower post-response than

it was prior to the attack. Among both same- and opposite-party identifiers, evaluations of the

challenger/attacker were significantly more negative at t3 than at t1 across (a) all five types of

responses and (b) for the most part, all three ambivalence levels. In other words, the challenger

paid a price in almost every quarter, at least in terms of his personal image, for choosing to go

negative. How this loss of standing might play out on Election Day is a question for which we

obviously have no answer. Within the context of our simulation, however, we can state with

some confidence that the impact of challenger evaluations was not mediated to any meaningful

degree by feelings of partisan ambivalence.

Conclusion

Even political consultants, who are inclined to believe that negative ads “work,” don’t

believe that they always “work” (after all, many candidates who go negative end up losing their

race) or that they “work” equally well among all segments of the electorate. Our focus in this

paper has been on the fact that even in an era of seemingly “resurgent” partisanship, there still

are many Americans who possess a mix of positive and negative feelings about the political

parties, including (especially) their own – and that those mixed feelings may make them more

receptive to campaign attacks against their fellow partisans, but less likely to be influenced by

the rebuttals made by co-partisans following an attack. Similarly, we posited that those who are

ambivalent about their party will be more inclined to “punish” co-partisans (giving them lower

personal evaluations) for going negative in the first place.

In general, our findings indicate that ambivalence may play a less prominent role than

that suggested by Lavine, Johnston, and Steenbergen (2012) in moderating the effectiveness of

both negative campaign ads and candidates’ responses to those ads. This is especially true for

24

responses, where observed changes were frequently in the opposite direction from what we

predicted (and not significantly different across levels of partisan ambivalence in any event).

Ambivalence played a more prominent, if inconsistent, role in shaping voters’ reactions to the

original attack. Not only were highly and moderately ambivalent individuals less supportive of

their party’s candidate to begin with, they also were more likely than the univalent to withdraw

vote support and to view that candidate less favorably following the attack. Not all of these

differences across groups were statistically significant, but the overall pattern suggests that

ambivalence does moderate, to some degree, the reactions of voters to opposition attacks against

their party’s candidates. In our campaign simulation, however, ambivalence played little or no

role in shaping evaluations of the attacking candidate by either fellow or opposing partisans.

Apart from the usual qualifications (especially regarding external validity; see Gadarian

and Lau 2011; but cf. Arceneaux 2010) that must accompany any controlled experiment, we

acknowledge that our study merely touches the surface in terms of exploring the impact of

partisan ambivalence in contemporary political campaigns. We would like to know more, for

example, about how the moderating role of ambivalence might vary with the repetition of ads

(Fernandes 2013); their timing (Chong and Druckman 2013); the degree to which they receive

earned media coverage (Ridout and Franz 2011); candidate vs. party vs. external sponsorship

(Brooks and Murov 2012; Weber, Dunaway, and Johnson 2012); and the specific type of ads

being examined (e.g., positive rather than negative, policy- rather than performance-based). For

now, the jury remains out regarding the ways in which feelings of partisan ambivalence might

affect how voters react to the various kinds of messages they receive at campaign time.

25

Notes

1. See http://www.gallup.com/poll/166763/record-high-americans-identify-independents.

aspx; http://www.gallup.com/poll/15370/party-affiliation.aspx

2. See

(both accessed July 25, 2014).

http://www.gallup.com/poll/15370/party-affiliation.aspx

3. Among the changes noted by Hetherington: citizens were more likely to perceive

“important differences” between the parties, better able to name things they liked and disliked

about each, less neutral (expressing an equal number of likes and dislikes) in their feelings about

either, and more likely to feel positively about one and negatively about the other.

(accessed July 25, 2014).

4. Updates of the McCarty, Poole, and Rosenthal results are available online at http://

voteview.com/political_polarization.asp

5. Some of this movement occurred as the result of generational change rather than

conversion. While there may have been an increase in the proportion of Americans who say there

are “important differences” between the parties (see note 3), it is also true that the public as a

whole perceives the Republicans and the Democrats as being only somewhat further apart

ideologically in the 2010s than was true twenty, thirty, or even forty years ago (Sides 2013).

(accessed July 25, 2014); also see Pew Research Center

(2014b).

6. See http://en.wiktionary.org/wiki/hyperpartisanship

7. To be sure, the impact of campaigns ads does not have to be of great magnitude to be

meaningful in a close election, e.g., see Goldstein and Freedman (2000); Johnston, Hagen, and

Jamieson (2004); Franz and Ridout (2007); Huber and Arceneaux (2007); also Gerber et al. (2011);

Hill et al. (2013).

(accessed July 25, 2014).

8. Although we will note the behavior of non-leaning Independents, our expectations of

how that behavior might be shaped by ambivalence about one or both parties is uncertain.

26

9. Also see Garramone (1985); Roddy and Garramone (1988); Jasperson and Fan (2002);

Fridkin and Kenney (2011); Brooks and Murov (2012); Craig, Rippere, and Grayson (2015).

10 For example, see Garramone (1985); Roddy and Garramone (1988); Pfau and Kenski

(1990); Freedman, Wood, and Lawton (1999); Carraro et al. (2012); Craig, Rippere, and Grayson

(2015).

11. Attitude strength, or extremity, and attitudinal ambivalence are not synonymous, but

there is evidence that the two are negatively related (Kaplan 1972; Thompson, Zanna, and

Griffin 1995). We therefore take ambivalence to be an example of a “weak” attitude.

12. The sample consisted of 51.8% women and 48.2% men; 78.6% whites and 21.5%

nonwhites; 17.1% aged 18-29, 26.6% aged 30-44, 29.6% aged 45-59, and 26.7% aged 60 and over;

10.7% high-school graduates or less, 26.6% with some college, 34.6% college graduates, and 28.1%

with some postgraduate education; 29.3% self-identified liberals, 21.9% moderates, 34.3%

conservatives, and 14.5% other (including 11.8% who said they “haven’t thought much about it”);

40.9% self-identified Democrats, 13.6% “pure” independents, 34.1% Republicans, and 11.3% other

(the latter figure higher than one would normally expect to see in a national survey, in part – or so

we suspect – because our questionnaire listed both “other” and “don’t know/not sure” as explicit

response options; see our online appendix at http://www.clas.ufl.edu/users/sccraig/

apsa14appendix.pdf for question wording).

13. Each candidate’s party affiliation and status as either challenger or incumbent was

specified, but otherwise the biosketches were crafted in such a way as to ensure that the two

portrayals were essentially equivalent.

14. On the potential persuasive effects of negative direct mail, see Gerber, Kessler, and

Meredith (2011).

27

15. We were assisted in writing the candidate biographies, attacks, and responses by an

experienced campaign consultant who has worked on numerous legislative races over the years.

Additional details about this survey can be found in our online appendix.

16. A second attack, accusing the incumbent of being out of touch with constituents (see

Craig, Rippere, and Grayson 2015 for a description), was used in an earlier stage of this project

but dropped here because its impact on respondents’ attitudes and vote was substantially less

clear-cut than that of taking advantage.

17. According to Fridkin and Kenney (2008), relevant (containing information regarding

how a politician has influenced/is likely to influence voters’ lives) and uncivil (uncertain about

where to draw the line, we prefer the term “hard-hitting”) negative ads have the greatest impact on

evaluations of the targeted candidate. Some of the charges made in our ads involved behavior that

was said to have occurred during the target’s prior service as a county commissioner or state

legislator.

18. We should note that our randomization process appears to have been successful. No

statistically significant differences (p < .10) were observed among members of the twenty groups

with regard to demographics, partisanship, ideological self-placement, or baseline candidate

preferences (vote choice, favorability ratings). In addition, there were no statistically significant

differences for these same variables among respondents assigned to the five response types.

Thus, if differences are found across research groups after respondents read the attack and

response messages, we can be confident that these are driven by exposure to the experimental

stimuli.

19. Conceptually, the first part of the equation – [(P + N)/2] – states that with similarity

held constant, greater intensity leads to greater ambivalence; that is, as the average value of

28

positive and negative scores increases, so do feelings of ambivalence. The second part of the

equation – |P - N| – indicates that when similarity increases (e.g., an equal number of positive

and negative reactions), a lesser amount is subtracted from the ambivalence total than if

similarity were reduced; consequently, greater similarity translates into higher scores on

ambivalence.

20. Two respondents did not answer either of the party feelings questions and were coded

as missing for our entire analysis.

21. See http://www.gallup.com/poll/169091/democratic-party-seen-favorably-gop.aspx

22. Additional details can be found on our online appendix.

(accessed July 25, 2014). According to Gallup, not since 2005 have Americans simultaneously

viewed both parties more favorably than unfavorably.

23. As stated earlier, we had no clear expectations for how partisan ambivalence would

affect the behavior of these individuals. For the record (see our online appendix for more),

although non-leaning Independents expressed few positive feelings either way (none were

“extremely” positive, just 17.8 percent were “generally” positive about the Democrats and 10.0

percent about the Republicans), neither were they overwhelmingly negative (43.3 percent were

“generally/extremely” negative about the Republicans, 42.2 percent about the Democrats). As for

ambivalence, the modal response combination for each party was from respondents who said

they had “some” positive and “some” negative feelings (37.8 percent Republicans, 34.4 percent

Democrats).

24. The coding scheme used to classify respondents as either high, medium, or low on

ambivalence toward the parties is discussed above (also see our online appendix for details).

25. Readers will recall that we made no predictions about the impact of ambivalence

29

among those who do not identify as either Democrats or Republicans. A replication of Table 3

for all nonpartisans (pure Independents, other-party identifiers, and respondents who aren’t sure

of their affiliation) suggests that ambivalence may matter in some circumstances, e.g., voters

who felt highly ambivalent about the incumbent/target’s party were more likely than others to (a)

say they would support the incumbent based solely on biographical information (62.1 percent);

(b) abandon that vote intention following the attack (minus 32.8 percentage points); and (c)

evaluate the incumbent less favorably following the attack (a drop of 0.517 on the 7-point scale).

Overall, though, our results for nonpartisans are neither consistent (for example, the post-attack

decline in challenger favorability was greatest among those expressing moderate ambivalence)

nor in most cases statistically significant.

26. In this section, we relax our standard for significance slightly due to the very small

N’s involved.

27. Neither of these changes was statistically significant in and of itself (p > .10), though in

both cases they resulted in cross-group differences that did clear the significance bar (see text for

details).

28. The mudslinging response was notably less effective than the others, as the incumbent

suffered a double-digit loss of support in all three ambivalence categories. With the small N’s

involved, however, only a 20-point drop among the moderately ambivalent was statistically

significant (p < .05).

29. The net decline in incumbent favorability was most pronounced among high- relative to

low-ambivalence voters in the case of counterattacks (p < .01) and counterimaging (p < .10).

30

References

Abramowitz, Alan I. 2010. The Disappearing Center: Engaged Citizens, Polarization, and

American Democracy. New Haven, CT: Yale University Press.

Alvarez, R. Michael, and John Brehm. 1995. “American Ambivalence Towards Abortion Policy:

Development of a Heteroskedastic Probit Model of Competing Values.” American Journal of

Political Science 39 (4): 1055-1082.

Ansolabehere, Stephen, and Shanto Iyengar. 1995. Going Negative: How Political

Advertisements Shrink and Polarize the Electorate. New York, NY: Free Press.

Arceneaux, Kevin. 2010. “The Benefits of Experimental Methods for the Study of Campaign

Effects.” Political Communication 27 (2): 199-215.

Bartels, Larry M. 2000. “Partisanship and Voting Behavior, 1952-1996.” American Journal of

Political Science 44 (1): 35-50.

Brooks, Deborah Jordan, and Michael Murov. 2012. “Assessing Accountability in a Post-Citizens

United Era: The Effects of Attack Ad Sponsorship by Unknown Independent Groups.” American

Politics Research 40 (3): 383-418.

Campbell, Angus, Philip E. Converse, Warren E. Miller, and Donald E. Stokes. 1960. The

American Voter. New York, NY: Wiley.

Carraro, Luciana, Luigi Castelli, Ioana Breazu, Giulia Campomizzi, Antonella Cerruto,

Massimiliano Mariani, and Ivano Toto. 2012. “Just Ignore or Counterattack? On the Effects of

Different Strategies for Dealing with Political Attacks.” European Journal of Social

Psychology 42 (6): 789-797.

Chaiken, Shelly, Akiva Liberman, and Alice H. Eagly. 1989. “Heuristic and Systematic

Information Processing within and beyond the Persuasion Context.” In Unintended Thought,

31

eds. James S. Uleman and John A. Bargh. New York, NY: Guilford Press, pp. 212-252.

Chong, Dennis, and James N. Druckman. 2013. “Counterframing Effects.” Journal of Politics 75

(1): 1-16.

Craig, Stephen C., James G. Kane, Michael D. Martinez, and Jason Gainous. 2005. “Core

Values, Value Conflict, and Citizens’ Ambivalence about Gay Rights.” Political Research

Quarterly 58 (1): 5-17.

Craig, Stephen C., and David B. Hill, eds. 2011. The Electoral Challenge: Theory Meets

Practice, 2nd ed. Washington, DC: CQ Press.

Craig, Stephen C., James G. Kane, and Michael D. Martinez. 2002. “Sometimes you Feel Like a

Nut, Sometimes you Don’t: Citizens' Ambivalence about Abortion.” Political Psychology 23

(2): 285-301.

Craig, Stephen C., and Paulina S. Rippere. 2013. “He Said, She Said: The Impact of Candidate

Gender in Negative Campaigns.” Paper presented at the annual meetings of the American

Political Science Association, Chicago, IL.

Craig, Stephen C., and Paulina S. Rippere. 2014. “Political Trust and Negative Campaigns: Two

Tests of the Figure-Ground Hypothesis.” Politics and Policy (forthcoming).

Craig, Stephen C., Paulina S. Rippere, and Marissa Silber Grayson. 2012. “Attack and Response in

Political Campaigns: An Experimental Study.” Paper presented at the annual meetings of the

Midwest Political Science Association, Chicago, IL.

Craig, Stephen C., Paulina S. Rippere, and Marissa Silber Grayson. 2015. “Attack and Response in

Political Campaigns: An Experimental Study in Two Parts.” Political Communication

(forthcoming).

Dalton, Russell J. 1984. “Cognitive Mobilization and Partisan Dealignment in Advanced Industrial

32

Democracies.” Journal of Politics 46 (1): 264-284.

Dalton, Russell J. 2013. The Apartisan American: Dealignment and Changing Electoral Politics.

Washington, DC: CQ Press.

Diamond, Edwin, and Stephen Bates. 1992. The Spot: The Rise of Political Advertising on

Television, 3rd ed. Cambridge, MA: MIT Press.

Diggles, Michelle, and Lanie Erickson Hatalsky. 2014. “The State of the Center.” Third Way.

Available online at http://www.thirdway.org/subjects/152/publications/819 (accessed July 25, 2014).

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

Druckman, James N., Jacobs, Lawrence R., and Eric Ostermeier. 2004. “Candidate Strategies to

Prime Issues and Image.” Journal of Politics 66 (4): 1180-1202.

Eagly, Alice H., and Shelly Chaiken. 1993. The Psychology of Attitudes. Fort Worth, TX: Harcourt

Brace.

Ellis, Christopher, and James A. Stimson. 2012. Ideology in America. New York, NY: Cambridge

University Press.

Fernandes, Juliana. 2013. “Effects of Political Advertising and Message Repetition on Candidate

Evaluation.” Mass Communication and Society 16 (2): 268-291.

Festinger, Leon. 1957. A Theory of Cognitive Dissonance. Stanford, CA: Stanford University

Press.

Fiorina, Morris P. 1981. Retrospective Voting in American National Elections. New Haven, CT:

Yale University Press.

Fiorina, Morris P. 2014. “Americans Have Not Become More Politically Polarized.” The

Washington Post/Monkey Cage. Available online at http://www.washingtonpost.com/

blogs/monkey-cage/wp/2014/06/23/americans-have-not-become-more-politically-polarized

33

(accessed July 25, 2014).

Fiorina, Morris P., with Samuel J. Abrams and Jeremy C. Pope. 2011. Culture War? The Myth of

a Polarized America, 3rd ed. Boston, MA: Longman.

Franz, Michael M., Paul B. Freedman, Kenneth M. Goldstein, and Travis N. Ridout. 2008.

Campaign Advertising and American Democracy. Philadelphia. PA: Temple University Press.

Franz, Michael M., and Travis N. Ridout. 2007. “Does Political Advertising Persuade?” Political

Behavior 29 (4): 465-491.

Frenkel-Brunswik, Else. 1949. “Intolerance of Ambiguity as an Emotional and Perceptual

Personality Variable.” Journal of Personality 18 (1): 108-143.

Freedman, Paul, Wood, William, and Dale Lawton. 1999. “Do’s and Don’t’s of Negative Ads:

What Voters Say.” Campaigns and Elections 20 (October/November): 20-25.

Fridkin, Kim L., and Patrick J. Kenney. 2008. “The Dimensions of Negative Messages.” American

Politics Research 36 (5): 694-723.

Fridkin, Kim Leslie, and Patrick J. Kenney. 2011. “Variability in Citizens’ Reactions to Different

Types of Negative Campaigns.” American Journal of Political Science 55: 307-25.

Gadarian, Shana Kushner, and Richard R. Lau. 2011. “Candidate Advertisements.” In Cambridge

Handbook of Experimental Political Science, eds. James N. Druckman, Donald P. Green, James

H. Kuklinski, and Arthur Lupia. New York, NY: Cambridge University Press, pp. 214-227.

Gainous, Jason. 2008. “Who’s Ambivalent and Who’s Not? Ideology and Ambivalence about

Social Welfare.” American Politics Research 36 (2): 210-235.

Gainous, Jason, and Michael D. Martinez. 2005. “What Happens When We Simultaneously

Want Opposite Things? Ambivalence about Social Welfare.” In Ambivalence, Politics, and

Public Policy, eds. Stephen C. Craig and Michael D. Martinez. New York, NY: Palgrave, pp.

34

47-62.

Garramone, Gina M. 1985. “Effects of Negative Political Advertising: The Roles of Sponsor and

Rebuttal.” Journal of Broadcasting and Electronic Media 29 (2): 147-159.

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

IL: University of Chicago Press.

Gerber, Alan S., Daniel P. Kessler, and Marc Meredith. 2011. “The Persuasive Effects of Direct

Mail: A Regression Discontinuity Based Approach.” Journal of Politics 73 (1): 140-155.

Gerber, Alan S., James G. Gimpel, Donald P. Green, and Daron R. Shaw. 2011. “How Large and

Long-Lasting are the Persuasive Effects of Televised Campaign Ads? Results from a

Randomized Field Experiment.” American Political Science Review 105: 135-150.

Goldstein, Ken, and Paul Freedman. 2000. “New Evidence for New Arguments: Money and

Advertising in the 1996 Senate Elections.” Journal of Politics 62: 1087-1108.

Green, Donald, Bradley Palmquist, and Eric Schickler. 2002. Partisan Hearts and Minds:

Political Parties and the Social Identities of Voters. New Haven, CT: Yale University Press.

Greene, Steven H. 2005. “The Structure of Partisan Attitudes: Reexamining Partisan

Dimensionality and Ambivalence.” Political Psychology 26 (5): 809–822.

Greenwald, Anthony G., Mahzarin R. Banaji, Laurie A. Rudman, Shelly D. Farnham, Brian A.

Nosek, and Deborah S. Mellot. 2002. “A Unified Theory of Implicit Attitudes, Stereotypes,

Self-Esteem, and Self-Concept.” Psychological Review 109 (1): 3-25.

Heider, Fritz. 1958. The Psychology of Interpersonal Relations. New York, NY: Wiley.

Hetherington, Marc J. 2001. “Resurgent Mass Partisanship: The Role of Elite Polarization.”

American Political Science Review 95 (3): 619-31.

Hill, Seth J., James Lo, Lynn Vavreck, and John Zaller. 2013. “How Quickly We Forget: The

35

Duration of Persuasion Effects from Mass Communication.” Political Communication 30 (4):

521-547.

Huber, Gregory A., and Kevin Arceneaux. 2007. “Identifying the Persuasive Effects of Presidential

Advertising.” American Journal of Political Science 51: 957-977.

Iyengar, Shanto, Jackman, Simon, and Kyu Hahn. 2008. “Polarization in Less than Thirty Seconds:

Continuous Monitoring of Voter Response to Campaign Advertising.” Paper presented at the

annual meetings of the Midwest Political Science Association, Chicago, IL.

Iyengar, Shanto, Gaurav Sood, and Yphtach Lelkes. 2012. “Affect, Not Ideology: A Social

Identity Perspective on Polarization.” Public Opinion Quarterly 76 (3): 405-431.

Jamieson, Kathleen Hall. 1992. Dirty Politics: Deception, Distraction, and Democracy. New

York, NY: Oxford University Press.

Jasperson, Amy E., and David P. Fan. 2002. "An Aggregate Examination of the Backlash Effect in

Political Advertising: The Case of the 1996 U. S. Senate Race in Minnesota." Journal of

Advertising 31 (1): 1-12.

Johnson, Dennis W. 2007. No Place for Amateurs: How Political Consultants are Reshaping

American Democracy, 2nd ed. New York, NY: Routledge.

Johnson-Cartee, Karen S., and Gary A. Copeland. 1997. Manipulation of the American Voter:

Political Campaign Commercials. Westport, CT: Praeger.

Johnston, Richard, Michael G. Hagen, and Kathleen Hall Jamieson. 2004. The 2000 Presidential

Election and the Foundations of Party Politics. New York, NY: Cambridge University Press.

Kaplan, Kalman J. 1972. “On the Ambivalence-Indifference Problem in Attitude Theory and

Measurement: A Suggested Modification of the Semantic Differential Technique.”

Psychological Bulletin 77 (5): 361-372.

36

Kelley, Stanley., Jr., and Thad W. Mirer. 1974. “The Simple Act of Voting.” American Political

Science Review 68 (2): 572-591.

Kruglanski, Arie W., and Donna M. Webster. 1996. “Motivated Closing of the Mind: ‘Seizing’

and ‘Freezing’.” Psychological Review 103 (2): 263-283.

Lau, Richard R., and David P. Redlawsk. 2006. How Voters Decide: Information Processing

during Election Campaigns. New York, NY: Cambridge University Press.

Lau, Richard R., and Ivy Brown Rovner. 2009. “Negative Campaigning.” Annual Review of

Political Science 12: 285-306.

Lavine, Howard G., Christopher D. Johnston, and Marco R. Steenbergen. 2012. The Ambivalent

Partisan: How Critical Loyalty Promotes Democracy. New York, NY: Oxford University

Press.

Layman, Geoffrey C., and Thomas M. Carsey. 2002. “Party Polarization and ‘Conflict

Extension’ in the American Electorate.” American Journal of Political Science 46 (4): 786-

802.

Lebo, Matthew J., and Dan Cassino. 2007. “The Aggregated Consequences of Motivated

Reasoning and the Dynamics of Partisan Presidential Approval.” Political Psychology 28 (6):

719-746.

Levendusky, Matthew. 2009. The Partisan Sort: How Liberals Became Democrats and

Conservatives Became Republicans. Chicago, IL: University of Chicago Press.

Lewis-Beck, Michael S., William G. Jacoby, Helmut Norpoth, and Herbert F. Weisberg. 2008.

The American Voter Revisited. Ann Arbor, MI: University of Michigan Press.

Lodge, Milton, and Charles S. Taber. 2013. The Rationalizing Voter. New York, NY: Cambridge

University Press.

37

Magleby, David B., and Candice J. Nelson. 2012. “Independent Leaners as Policy Partisans: An

Examination of Party Identification and Policy Views.” The Forum 10 (3), originally

published online by Berkeley Electronic Press.

Magleby, David B., Candice J. Nelson, and Mark C. Westlye. 2011. “The Myth of the Independent

Voter.” In Facing the Challenge of Democracy: Explorations in the Analysis of Public Opinion

and Political Participation, eds. Paul M. Sniderman and Benjamin Highton. Princeton, NJ:

Princeton University Press, pp. 238-263.

Mark, David. 2006. Going Dirty: The Art of Negative Campaigning. Lanham, MD: Rowman and

Littlefield.

Martinez, Michael D., Jason Gainous, and Stephen C. Craig. 2012. “Measuring Ambivalence about

Government.” In Improving Public Opinion Surveys: Interdisciplinary Innovation and the

American National Election Studies, eds. Kathleen M. McGraw and John Aldrich. Princeton, NJ:

Princeton University Press, pp. 238-259

Mayer, William G. 1996. “In Defense of Negative Campaigning.” Political Science Quarterly 111

(3): 437-455.

McCarty, Nolan, Keith T. Poole, and Howard Rosenthal. 2006. Polarized America: The Dance of

Ideology and Unequal Riches. Cambridge, MA: MIT Press.

Mulligan, Kenneth. 2011. “Partisan Ambivalence, Split-Ticket Voting, and Divided

Government.” Political Psychology 32 (3): 505-530.

Nie, Norman H., Sidney Verba, and John R. Petrocik. 1976. The Changing American Voter.

Cambridge, MA: Harvard University Press.

Noel, Hans. 2013. Political Ideologies and Political Parties in America. New York, NY:

Cambridge University Press.

38

Norpoth, Helmut, and Jerrold G. Rusk. 1982. “Partisan Dealignment in the American Electorate.”

American Political Science Review 76 (3): 522-537.

Osgood, Charles E., George J. Suci, and Percy H. Tannenbaum. 1957. The Measurement of

Meaning. Urbana, IL: University of Illinois Press.

Payne, John W., James R. Bettman, and Eric J. Johnson. 1993. The Adaptive Decision Maker.

New York, NY: Cambridge University Press.

Pew Research Center for the People and the Press. 2012. “Partisan Polarization Surges in Bush,

Obama Years.” Available online at http://www.people-press.org/files/legacy-pdf/06-04-

12%20Values%20Release.pdf (accessed on July 25, 2014).

Pew Research Center for the People and the Press. 2014a. “Political Polarization in the American

Public: How Increasing Ideological Uniformity and Partisan Antipathy Affect Politics,

Compromise and Everyday Life.” Available online at http://www.people-

press.org/2014/06/12/political-polarization-in-the-american-public/ (accessed July 25, 2014).

Pew Research Center for the People and the Press. 2014b. “The Polarized Congress of Today

Has Its Roots in the 1970s.” Available online at http://www.pewresearch.org/fact-

tank/2014/06/12/polarized-politics-in-congress-began-in-the-1970s-and-has-been-getting-

worse-ever-since/ (accessed July 25, 2014).

Pfau, Michael, and Henry C. Kenski. 1990. Attack Politics: Strategy and Defense. New York, NY:

Praeger.

Pope, Jeremy C. 2012. “Voting vs. Thinking: Unified Partisan Voting Does Not Imply Unified

Partisan Beliefs.” The Forum 10 (3), originally published online by Berkeley Electronic Press.

Popkin, Samuel L. 1991. The Reasoning Voter: Communication and Persuasion in Presidential

Campaigns. Chicago, IL: University of Chicago Press.

39

Priester, Joseph R., and Richard E. Petty. 1996. "The Gradual Threshold Model of Ambivalence:

Relating the Positive and Negative Bases of Attitudes to Subjective Ambivalence." Journal of

Personality and Social Psychology 71 (3): 431-49.

RePass, David E. 2008. “Searching for Voters along the Liberal-Conservative Continuum: The

Infrequent Ideologue and the Missing Middle.” The Forum 6 (2), originally published online by

Berkeley Electronic Press.

Ridout, Travis N., and Michael Franz. 2011. The Persuasive Power of Campaign Advertising.

Philadelphia, PA: Temple University Press.

Roddy, Brian L., and Gina M. Garramone. 1988. “Appeals and Strategies of Negative Political

Advertising.” Journal of Broadcasting and Electronic Media 32 (4): 415-427.

Sides, John. 2013. “What If a Party Re-branded Itself, and Americans Never Noticed?” The

Washington Post/Monkey Cage. Available online at http://themonkeycage.org/2013/06/

13/what-if-a-party-re-branded-itself-and-americans-never-noticed/ (accessed July 25, 2014).

Sigelman, Lee, and Mark Kugler. 2003. “Why Is Research on the Effects of Negative Campaigning

So Inconclusive? Understanding Citizens' Perceptions of Negativity.” Journal of Politics 65:

142-160.

Slothuus, Rune, and Claes H. de Vreese. 2010. “Political Parties, Motivated Reasoning, and

Issue Framing Effects.” Journal of Politics 72 (3): 630-645.

Stevens, Daniel, John Sullivan, Barbara Allen, and Dean Alger. 2008. “What’s Good for the Goose

is Bad for the Gander: Negative Political Advertising, Partisanship, and Turnout.” Journal of

Politics 70 (2): 527-541.

Stonecash, Jeffrey M., Mark D. Brewer, and Mack D. Marianai. 2003. Diverging Parties: Social

Change, Realignment, and Party Polarization. Boulder, CO: Westview Press.

40

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

Press.

Thompson, Megan, Mark P. Zanna, and Dale W. Griffin. 1995. “Let’s Not Be Indifferent About

(Attitudinal) Ambivalence.” In Attitude Strength: Antecedents and Consequences, eds.

Richard E. Petty and Jon A. Krosnick. Mahwah, NJ: Lawrence Erlbaum, pp. 361-386.

Thornton, Judd R. 2011. “Ambivalent or Indifferent:? Examining the Validity of an Objective

Measure of Partisan Ambivalence.” Political Psychology 32 (5): 863-884.

Thornton, Judd R. 2013. “The Impact of Elite Polarization on Partisan Ambivalence and

Indifference.” Political Behavior 35 (2): 409-428.

Trent, Judith S., Robert V. Friedenberg, and Robert E. Denton, Jr. 2011. Political Campaign

Communication: Principles and Practices, 7th ed. Lanham, NJ: Rowman and Littlefield.

Weber, Christopher, Johanna Dunaway, and Tyler Johnson. 2012. “It’s All in the Name: Cue

Ambiguity and the Persuasive Appeal of Campaign Ads.” Political Behavior 34 (3): 561-584.

Westen, Drew. 2007. The Political Brain: The Role of Emotion in Deciding the Fate of the

Nation. New York, NY: Public Affairs.

Zaller, John R. 1992. The Nature and Origins of Mass Opinion. Cambridge, UK: Cambridge

University Press.

Zaller, John, and Stanley Feldman. 1992. “A Simple Theory of the Survey Response: Answering

Questions versus Revealing Preferences.” American Journal of Political Science 36 (3): 579-

616.

41

Table 1: Five Responses to Attacks on Incumbent

Response Type Attack: Taking Advantage

Denial

Candidate did nothing wrong: • pay raises enacted before s/he took office; • only personal money used to pay for family vacation; • office complex where district office is located no longer owned by

candidate’s brother; • overbilling was due to a clerical error and quickly corrected; • investigation by Attorney General found no evidence of wrongdoing

Counterattack

Opponent is the one deserving of criticism because s/he: • filed false business tax return; • accepted illegal campaign contributions; • steered government contracts to business clients and campaign donors; • opposed stricter ethics laws for state officials

Counterimaging

Candidate is a (wo)man of character as s/he: • helped support family after father died; • put him/herself through college and earned scholarship to graduate school; • started own business that creates many jobs; • has served country (active duty/military reserves), community (volunteer

work), those less fortunate (established college scholarship fund), and church (charitable activities, missions)

Justification

Candidate’s actions were reasonable/warranted: • pay raises did not apply to anyone currently in office; • vacation followed official trade meetings, with the party reimbursed for

personal expenses; • district office is the same used by predecessor, and rent has not increased

since then; • dispute over fee charged for professional services settled amicably; • no personal or close political connections to recipients of state contracts

Mudslinging

Opponent is desperate and has become “relentlessly negative,” thereby: • failing to discuss the issues people really care about; • ignoring the fact that voters are sick of campaign mudslinging; • contributing to voter pessimism and low turnout; • exhibiting weak leadership skills, and a lack of class and integrity

42

Table 2: The Distribution of Partisan Ambivalence, by Party Identification Part I: Feelings about One’s Own Party

Repu

blica

ns

No Negative Some Negative Generally Negative Extremely Negative Total No

Positive 0.0% 0.9% 1.3% 2.2% 4.4% (10)

Some Positive 1.8% 22.1% 4.9% 1.3% 30.1%

(68) Generally Positive 8.9% 48.7% 1.3% 0.0% 58.9%

(133) Extremely Positive 4.4% 2.2% 0.0% 0.0% 6.6%

(15)

Total 15.0% (34)

73.9% (167)

7.5% (17)

3.5% (8)

100% (226)

Dem

ocra

ts

No Negative Some Negative Generally Negative Extremely Negative Total No

Positive 0.0% 0.4% 1.1% 0.0% 1.5% (4)

Some Positive 1.1% 27.7% 4.4% 0.0% 33.2%

(90) Generally Positive 8.5% 43.5% 1.1% 0.0% 53.1%

(144) Extremely Positive 8.1% 3.0% 0.0% 1.1% 12.2%

(33)

Total 17.7% (48)

74.5% (202)

6.6% (18)

1.1% (3)

100% (271)

Note: Low, moderate, and high levels of ambivalence (see text for coding) are indicated by cell shadings of light, medium, and dark gray, respectively. Partisans include strong and weak identifiers as well as leaners. Part II: Feelings about the Opposing Party

Repu

blica

ns’ F

eelin

gs ab

out

the D

emoc

ratic

Par

ty

No Negative Some Negative Generally Negative Extremely Negative Total No

Positive 0.4% 0.9% 22.6% 45.1% 69.0% (156)

Some Positive 0.4% 11.5% 14.2% 3.1% 29.2%

(66) Generally Positive 0.0% 0.9% 0.4% 0.0% 1.3%

(3) Extremely Positive 0.0% 0.4% 0.0% 0.0% 0.4%

(1)

Total 0.9% (2)

13.7% (31)

37.2% (84)

48.2% (109)

100% (226)

Dem

ocra

ts’ F

eelin

gs ab

out

the R

epub

lican

Par

ty

No Negative Some Negative Generally Negative Extremely Negative Total No

Positive 1.9% 5.5% 21.8% 34.3% 63.5% (172)

Some Positive 0.4% 16.6% 14.0% 4.1% 35.1%

(95) Generally Positive 0.4% 0.7% 0.4% 0.0% 1.5%

(4) Extremely Positive 0.0% 0.0% 0.0% 0.0% 0.0%

(0)

Total 2.6% (7)

22.9% (62)

36.2% (98)

38.4% (104)

100% (271)

Note: Low, moderate, and high levels of ambivalence (see text for coding) are indicated by cell shadings of light, medium, and dark gray, respectively. Partisans include strong and weak identifiers as well as leaners.

43

Table 3: Effectiveness of Campaign Attack, by Levels of Ambivalence (Partisans Only)

Ambivalence about Incumbent’s Party, Incumbent Co-Partisans

Ambiv. about Challenger’s Party,

Challenger Co-Partisans

Vote for Incumbent

Favorability, Incumbent

Favorability, Challenger

Favorability, Challenger

N Prop N Mean N Mean N Mean

High

Am

biva

lence

Baseline eval. 72 0.806

(0.047) 72 5.125 (0.131) 72 4.333

(0.158) 62 4.935 (0.143)

Post-attack eval. 72 0.472

(0.059) 72 3.194 (0.154) 72 3.694

(0.166) 62 4.500 (0.170)

diff -0.333 (0.075) -1.931

(0.192) -0.639 (0.139) -0.435

(0.125) p = 0.000 p = 0.000 p = 0.000 p = 0.001

Mode

rate

Am

biva

lence

Baseline eval. 130 0.869

(0.030) 130 5.308 (0.083) 130 4.277

(0.118) 147 5.408 (0.091)

Post-attack eval. 130 0.562

(0.044) 130 3.523 (0.122) 130 3.631

(0.118) 147 5.000 (0.110)

diff -0.308 (0.053) -1.785

(0.136) -0.646 (0.117) -0.408

(0.084) p = 0.000 p = 0.000 p = 0.000 p = 0.000

Low

Ambi

valen

ce Baseline

eval. 42 0.976 (0.024) 42 5.405

(0.207) 42 3.429 (0.229) 44 5.727

(0.196) Post-attack

eval. 42 0.762 (0.066) 42 4.238

(0.266) 42 2.786 (0.214) 44 5.318

(0.238) diff -0.214

(0.070) -1.167 (0.231) -0.643

(0.243) -0.409 (0.104)

p = 0.002 p = 0.000 p = 0.006 p = 0.000 Note: The analysis was conducted using paired t-tests. For vote choice, difference is calculated as proportion (post-attack vote) – proportion (baseline vote). For favorability, difference is calculated as mean (post-attack) – mean (baseline). Significance tests are 1-tailed.

44

Table 4: Effectiveness of Campaign Responses, Vote Choice and Incumbent Favorability (Partisans Only)

Denial Counterimaging Justification Counterattack Mudslinging N Mean N Mean N Mean N Mean N Mean

Vote

for I

ncum

bent

(Inc

umbe

nt C

o-Pa

rtisa

ns)

High

Am

biva

lence

Post-attack evaluation 11 0.545

(0.150) 16 0.500 (0.125) 14 0.571

(0.132) 14 0.429 (0.132) 17 0.353

(0.116) Post-response

evaluation 11 1.000 (0.000) 16 0.688

(0.116) 14 0.929 (0.069) 14 0.857

(0.094) 17 0.529 (0.121)

diff 0.455 (0.150) 0.188

(0.170) 0.357 (0.149) 0.429

(0.162) 0.176 (0.168)

p = 0.006 p = 0.140 p = 0.015 p = 0.009 p = 0.150

Mode

rate

Am

biva

lence

Post-attack evaluation 21 0.476

(0.109) 28 0.750 (0.082) 21 0.476

(0.109) 25 0.480 (0.100) 35 0.571

(0.084) Post-response

evaluation 21 0.810 (0.086) 28 0.857

(0.066) 21 0.857 (0.076) 25 0.880

(0.065) 35 0.714 (0.076)

diff 0.333 (0.139) 0.107

(0.105) 0.381 (0.133) 0.400

(0.119) 0.143 (0.113)

p = 0.012 p = 0.157 p = 0.004 p = 0.001 p = 0.106

Low

Ambi

valen

ce Post-attack

evaluation 7 0.857 (0.132) 7 1.000

(0.000) 11 0.636 (0.145) 7 0.571

(0.187) 10 0.800 (0.126)

Post-response evaluation 7 1.000

(0.000) 7 1.000 (0.000) 11 1.000

(0.000) 7 0.857 (0.132) 10 0.900

(0.095)

diff 0.143 (0.132) 0.000

(0.000) 0.364 (0.145) 0.286

(0.229) 0.100 (0.158)

p = 0.150 --- p = 0.014 p = 0.118 p = 0.266

Favo

rabi

lity R

atin

g, In

cum

bent

(Inc

umbe

nt C

o-Pa

rtisa

ns)

High

Am

biva

lence

Post-attack evaluation 11 3.545

(0.413) 16 3.438 (0.316) 14 3.214

(0.350) 14 3.000 (0.363) 17 2.882

(0.319) Post-response

evaluation 11 4.545 (0.390) 16 4.438

(0.258) 14 4.786 (0.408) 14 3.929

(0.355) 17 3.647 (0.331)

diff 1.000 (0.426) 1.000

(0.408) 1.571 (0.500) 0.929

(0.450) 0.765 (0.349)

p = 0.021 p = 0.014 p = 0.004 p = 0.030 p = 0.022

Mode

rate

Am

biva

lence

Post-attack evaluation 21 3.143

(0.295) 28 4.143 (0.261) 21 3.238

(0.330) 25 3.440 (0.300) 35 3.486

(0.198) Post-response

evaluation 21 4.905 (0.275) 28 4.929

(0.241) 21 5.048 (0.271) 25 3.720

(0.308) 35 3.886 (0.208)

diff 1.762 (0.316) 0.786

(0.238) 1.810 (0.400) 0.280

(0.212) 0.400 (0.143)

p = 0.000 p = 0.001 p = 0.000 p = 0.100 p = 0.004

Low

Ambi

valen

ce Post-attack

evaluation 7 4.429 (0.896) 7 4.714

(0.778) 11 3.727 (0.449) 7 3.857

(0.705) 10 4.600 (0.371)

Post-response evaluation 7 5.857

(0.404) 7 5.571 (0.685) 11 5.273

(0.304) 7 4.286 (0.680) 10 4.800

(0.416)

diff 1.429 (0.841) 0.857

(0.459) 1.545 (0.340) 0.429

(0.297) 0.200 (0.133)

p = 0.070 p = 0.056 p = 0.001 p = 0.100 p = 0.084 Note: The analysis was conducted using paired t-tests. For vote choice, difference is calculated as proportion (post-response vote) – proportion (post-attack vote). For favorability, difference is calculated as mean (post-response) – mean (post-attack). Significance tests are 1-tailed.

45

Table 5: Effectiveness of Campaign Responses, Challenger Favorability (Partisans Only)

Denial Counterimaging Justification Counterattack Mudslinging N Mean N Mean N Mean N Mean N Mean

Favo

rabi

lity R

atin

g, C

halle

nger

(Inc

umbe

nt C

o-Pa

rtisa

ns)

High

Am

biva

lence

Post-attack evaluation 11 3.364

(0.338) 16 3.625 (0.364) 14 4.000

(0.457) 14 3.571 (0.388) 17 3.824

(0.324) Post-response

evaluation 11 2.818 (0.325) 16 3.813

(0.390) 14 2.714 (0.339) 14 2.500

(0.344) 17 3.235 (0.265)

diff -0.545 (0.282) 0.188

(0.164) -1.286 (0.485) -1.071

(0.425) -0.588 (0.211)

p = 0.041 p = 0.135 p = 0.010 p = 0.013 p = 0.007

Mode

rate

Am

biva

lence

Post-attack evaluation 21 3.762

(0.248) 28 3.036 (0.249) 21 3.905

(0.330) 25 3.680 (0.256) 35 3.829

(0.230) Post-response

evaluation 21 2.429 (0.254) 28 3.250

(0.234) 21 2.667 (0.340) 25 2.520

(0.193) 35 3.486 (0.176)

diff -1.333 (0.252) 0.214

(0.188) -1.238 (0.292) -1.160

(0.243) -0.343 (0.169)

p = 0.000 p = 0.132 p = 0.000 p = 0.000 p = 0.025

Low

Ambi

valen

ce Post-attack

evaluation 7 2.429 (0.429) 7 2.286

(0.421) 11 2.909 (0.251) 7 3.571

(0.685) 10 2.700 (0.578)

Post-response evaluation 7 2.143

(0.340) 7 2.143 (0.459) 11 2.545

(0.312) 7 2.857 (0.670) 10 3.000

(0.447)

diff -0.286 (0.522) -0.143

(0.143) -0.364 (0.279) -0.714

(0.565) 0.300 (0.367)

p = 0.302 p = 0.178 p = 0.111 p = 0.127 p = 0.217

Favo

rabi

lity R

atin

g, C

halle

nger

(Cha

lleng

er C

o-Pa

rtisa

ns)

High

Am

biva

lence

Post-attack evaluation 18 4.222

(0.348) 8 4.250 (0.412) 10 4.600

(0.306) 11 4.636 (0.432) 15 4.800

(0.380) Post-response

evaluation 18 3.833 (0.364) 8 4.250

(0.412) 10 3.900 (0.348) 11 3.818

(0.585) 15 4.600 (0.400)

diff -0.389 (0.244) 0.000

(0.000) -0.700 (0.260) -0.818

(0.483) -0.200 (0.175)

p = 0.065 --- p = 0.012 p = 0.061 p = 0.136

Mode

rate

Am

biva

lence

Post-attack evaluation 31 5.290

(0.228) 28 4.893 (0.264) 27 4.926

(0.256) 29 4.724 (0.289) 32 5.125

(0.205) Post-response

evaluation 31 4.549 (0.245) 28 5.107

(0.238) 27 4.481 (0.216) 29 2.862

(0.203) 32 5.031 (0.198)

diff -0.742 (0.236) 0.214

(0.140) -0.444 (0.252) -1.862

(0.301) -0.094 (0.094)

p = 0.002 p = 0.068 p = 0.045 p = 0.000 p = 0.163

Low

Ambi

valen

ce Post-attack

evaluation 11 5.364 (0.544) 7 6.000

(0.436) 7 4.143 (0.670) 10 5.400

(0.452) 9 5.556 (0.475)

Post-response evaluation 11 5.091

(0.530) 7 6.000 (0.535) 7 4.143

(0.670) 10 4.100 (0.690) 9 5.444

(0.503)

diff -0.273 (0.195) 0.000

(0.218) 0.000 (0.218) -1.300

(0.396) -0.111 (0.111)

p = 0.096 p = 0.500 p = 0.500 p = 0.005 p = 0.173 Note: The analysis was conducted using paired t-tests. For vote choice, difference is calculated as proportion (post-response vote) – proportion (post-attack vote). For favorability, difference is calculated as mean (post-response) – mean (post-attack). Significance tests are 1-tailed.