Introduction

By many accounts, the use of negative campaigning in American elections has increased with each and every campaign cycle. During the 1960 campaign only 10% of all televised ads were negative (Mattes & Redlawsk, 2014). In contrast, in 2012 only 14.3% of all campaigns ads aired during 2012 could be classified as positive (Fowler & Ridout, 2012). Not only has the sheer volume of ads increased with each passing campaign, so too has the proportion of ads devoted to criticizing one’s opponent (Fowler & Ridout, 2012).

There is widespread criticism of this deluge of negativity. Polls show that American voters dislike campaign negativity. When asked if discouraging negative advertising would improve the campaign process, for example, 76% of survey respondents answered affirmatively.¹ Generally, politicians and journalists also have criticized negative campaigning. Former Senator Tom Daschle, for example, has described negative advertising as the “crack cocaine of politics.”² To this end, for example, it is not uncommon to hear a candidate criticize his or her opponent for relying on negativity.

The widespread criticism of negativity suggests that negative advertising may have adverse consequences for the political process. As a result, a vast body of work has focused on the empirical relationship between negativity and political outcomes. This scholarship is broad and diverse; most importantly, however, this research suggests that the effects of negativity are much more nuanced and conditional than the widespread criticisms of negativity would suggest. Taken as a whole, the political science literature points to some ambivalence about the role of negativity in the political process. Although some scholars have demonstrated that negative advertising can have negative consequences (see, e.g., Ansolabehere & Iyengar, 1995; Jamieson, 1992), others have suggested that negativity may actually be somewhat beneficial for the democratic process (see, e.g., Geer, 2006; Mattes & Redlawsk, 2014).

In this article we consider scholarship on campaign negativity, addressing both arguments that criticize its use and those that underscore its usefulness. We do so by considering the key questions scholars have asked about the relationship between campaign negativity and American politics:

We conclude with a discussion of the existing research on campaign negativity by turning to the remaining questions and explore avenues for future work.

What Is a Negative Advertisement?

When scholars discuss “negativity” or “negative campaigning,” they are typically referring to negative advertising. Although it is possible for candidates to rely on negative campaigning in their speeches, interviews, and websites (e.g., Druckman, Martin, Kifer, & Parkin, 2010), much of the existing empirical work on negativity focuses on televised advertising. In turn, research on negative advertising has pointed to multiple ways of defining the concept. One of the broadest definitions of negativity comes from Geer (2006), who defines negativity as “any criticism leveled by one candidate against another during a campaign” (p. 23). In contrast, a positive ad is one that “states why a candidate is worthy of your vote” (p. 23). As Geer (2006) argues, this definition leaves no “grey area”—an ad either has a criticism and is negative, or has no criticisms and is positive. What makes Geer’s definition particularly broad is that it does not distinguish between the substantive content of advertisements: under this definition both an advertisement that critiques an opponent’s voting record and one that critique’s the opponent’s personal life would be coded as equivalently negative.

Other definitions suggest a middle category between positive and negative advertising. A campaign advertisement, however brief (typically 30–60 seconds), can have more than one appeal. Within the same ad, for example, a candidate could promote his own issue positions and criticize an opponent’s voting record, which means that one advertisement has two different appeals for voter support. Some scholars have defined these types of ads as “contrast” ads because they contain both a criticism of the opposition and praise of one’s own candidacy (Fowler & Ridout, 2012).

Whereas the above definitions focus solely on tone, other scholars have also focused on the type of negativity and the substance of the criticisms. Although all negative ads have a similar tone (i.e., criticism), the very content of negativity may differ, and research suggests that these differences in content can play an important role on the effect of negativity. Some scholars have distinguished between negative ads that focus on issues and negative ads that focus on character (Freedman, Wood, & Lawton, 1999). Still others have suggested that “mudslinging” is a different type of negativity that may have particularly adverse consequences for politics (Kahn & Kenney, 1999). In defining mudslinging, Kahn and Kenney (1999) distinguish between what they term “‘legitimate’ negative information,” such as a candidate’s issue positions or voting record, and campaign ads that focus on information that voters will deem “irrelevant or inappropriate” (p. 878). These “inappropriate” mudslinging messages, Kahn and Kenney (1999) argue, are damaging to democracy.

Along with mudslinging, another potential consideration is incivility. Brooks and Geer (2007), for example, distinguish between negative messages that are civil and those that are delivered in a manner the voters may perceive to be uncivil. Brooks and Geer (2007) argue that campaign ads focusing on issues are fair; it is the negative ads that critique a candidate’s character that are viewed as unfair. Brooks and Geer suggest that voters respond better to negative ads that are about “the opponent’s stands on issues like education or military spending” but generally dislike negative ads that discuss topics like “the extramarital relationships of an opponent”(2007, p. 3).

Although these more fine-grained definitions offer a nuanced perspective to negativity, the umbrella definition of negativity as a criticism of the opponent is still relevant and useful. Even as Kahn and Kenney (1999) distinguish between types of negativity, they still retain the idea that at the most general level negativity is a critique—it is just that some types of critiques are more “appropriate” than others. Brooks and Geer (2007) too begin with the base definition of negativity as a criticism and from that point consider whether voters are more or less responsive to certain types of negativity. More importantly, even these more fine-grained definitions clearly distinguish negative ads from positive ads, and distinguishing “contrast” ads from “negative” ads still leaves the percentage of purely positive ads (currently at an all-time low) unchanged.

Suggesting that the positive–negative dichotomy can be a useful categorization in empirical research on advertising is not the same as suggesting that it is a consistently normatively useful distinction. Often, the positive–negative dichotomy is interpreted as a suggestion that positive ads are inherently “good” while negative ads are inherently “bad” for politics. This need not be the case. As Jamieson (2000) argues, positive ads may be deceptive, while certain negative attacks can be informative and important (see also Lau & Rovner, 2009). Indeed, as Mattes and Redlawsk (2014) note, there is nothing implicitly good about positive ads. Nonetheless, if the goal is to understand the effect of tone—rather than, for example, the effect of deceitfulness—distinguishing between ads that promote the candidate and ads that critique the opponent (or “target”) can be a useful empirical approach.

What Do Ordinary Americans Think of Negativity?

As noted in the introduction to this article, the public is not supportive of negative campaigning. As Geer (2006) writes, “a major part of the public’s unhappiness with elections [is] tied to negativity” (p. 2). Polls reinforce Geer’s point. For example, 55% of survey respondents “strongly agree” with the statement “negative attack-oriented campaigning is unethical,” 46% “strongly agree” that negative campaigning is “undermining and damaging our democracy,” and 57% believe that negativity is making people less likely to vote.³ In contrast, only 13% of people viewed negative campaigning as “very effective.”⁴

The negative response to negativity, however, may be at least in part a function of the questions many polls and surveys use to measure people’s responses to this type of advertising. Using a survey experiment, Mattes and Redlawsk (2014) randomly assign respondents to two different versions of a question about negative advertising. In one version, respondents are asked how they would feel if “a political candidate began to use negative campaigning, that is, began to run ads to give you information about what his opponent had done in office.” In the second version, the authors exclude the words “negative campaigning” but include the description of the type of ad. Mattes and Redlawsk (2014) find that the modal response to the question that describes the ads as “negative campaigning” is “somewhat angry.” In contrast, the modal response to the question that excludes those terms is “not at all angry” (2014, p. 57).

Moreover, in another set of studies Mattes and Redlawsk (2014) find that people respond differently to actual ads when these ads are not explicitly described as negative. People are not universally anti-negativity, and people do not dislike the idea that a candidate might air an ad that describes his opponent’s behavior—especially if this ad focuses on issues. Rather, Mattes and Redlawsk conclude, “only when we actually tell voters that this behavior is in fact negative campaigning do respondents recoil against it” (2014, p. 57).

This is not to suggest that people are supportive of all negative ads. Returning to the differences in the content of negativity, it appears that people can and do distinguish between different types of ads. Brooks and Geer, for example, note that 5% of people view a negative ad focusing on an issue as “unfair,” but 68% view an ad focusing on an “extramarital affair” as “unfair” (2007, p. 3).⁵ Similarly, Mattes and Redlawsk (2014) find that people easily distinguish between negative issue ads and negative ads about character. In turn, people reserve their anger for negative ads about character and are much more supportive when negative ads focus on issues.

In sum, it would be simple to follow polls as evidence that the public “hates” negativity, but public opinion on negativity is more nuanced. The vitriol against negative ads may be at least in part a function of question wording, for example. Moreover, to the extent that the public dislikes when candidates attack each other, the dislike is focused on ads that are considered “unfair.” Negative advertisements that focus their criticisms on issues are much less likely to rouse the ire of voters.

Do Negative Ads Affect Voter Turnout?

No other area of research on negativity has generated as much debate as research on the relationship between negativity and turnout. This work has been marked by three conflicting arguments. The first argument is that negativity demobilizes voters, the second argument is that negativity actually mobilizes voters, while the third argument is that negativity has no effect on voter turnout. Next, we consider the evidence in support of each of these arguments.

Typically termed the demobilization hypothesis, the first argument makes a causal connection between negative ads and declining voter turnout. Using both experimental and observational analyses, Ansolabehere, Iyengar, Simon, and Valentino (1994) and Ansolabehere and Iyengar (1995) demonstrate that increases in exposure to negative ads leave people less willing to vote. Ansolabehere and Iyengar (1995) suggest the mechanism underlying this effect is efficacy. More specifically, exposure to negativity leads people to be less politically efficacious, which makes them less likely to make it to the polls on election day.

Other scholars, however, suggest that the effect of negativity not only does not demobilize voters but actually mobilizes them. Using observational data, scholars have shown that people exposed to campaign negativity are actually more likely to turn out and vote (Goldstein & Freedman, 2002; Wattenberg & Brians, 1999). There are many mechanisms that could explain a possible positive connection between negativity and turnout. Negative ads, for example, can make politics more interesting and exciting, which may lead more people to participate (Goldstein & Freedman, 2002). Negative ads are also more memorable than positive ads (Geer & Geer, 2003). Moreover, research shows that people who are exposed to negative ads know more about politics, which again suggests a greater level of political involvement and interest (Franz, Freedman, Goldstein, & Ridout, 2007).

Finally, other work suggests that the relationship between negativity and turnout is largely null (Finkel & Geer, 1998; Krasno & Green, 2008). The null effect can also be quite logical—it is possible that negativity is simply a proxy for other types of campaign contexts and, on its own, has little effect on individuals’ propensity to turn out and vote.

Is there any way to unify these three conflicting sets of findings? Again, these different sets of results may be pointing to the idea that the effect of negativity on turnout is quite conditional. Brader (2006), for example, demonstrates that ads can affect individual emotions, and these emotions in turn lead to behavioral outcomes. Brader’s findings suggest that whether scholars observe any effects of negativity may be driven by the extent to which an ad evokes an emotional response—a hidden mechanism, given that most research on negativity and turnout does not measure emotional states. Following the same logic, it is also possible that only certain types of negative ads affect turnout. Kahn and Kenney (2004) show that negative issue ads and ads that are considered mudslinging have different effects on turnout. Again, the sets of conflicting findings could be in some part explained by differences in the types of ads used in experiments and the types of ads included in observational studies.

Still other research shows that the conflicting results may be explained by the timing of negativity. Krupnikov (2011, 2014) shows that negative ads aired early in the campaign can have a different effect than negative ads aired at later points. Whereas early ads can mobilize voters by making them more confident in their candidate choices, ads aired late decrease confidence and turn people away from the polls. In short, negativity can have both a mobilizing and demobilizing effect depending on the point in time the negative ads are aired.

Do Negative Ads Change Minds?

In surveys and polls, very few respondents believe that negativity will be effective. This is in contrast to politicians and journalists, who argue that the continued use of negativity is its own evidence that these types of ads can work (Lau & Rovner, 2009). What does it mean for a negative ad to “work”? One possible definition of “working” is that negative ads affect relative support for candidates and ultimately voting decisions. The most logical way in which negative ads may affect a candidate’s fate is by lowering evaluations of the candidate critiqued in the ad—or the “target” of the ad. If the entire purpose of a negative ad is to highlight the negative attributes of one’s opponent, then certainly the hoped-for outcome would be a decrease in support for the target among voters exposed to the ad.

There is some evidence to suggest that negative ads can—under certain conditions—lower support for the target. Kahn and Kenney (2004), for example, demonstrate that—compared to positive ads—negative ads do lead people to like the target of the ad less. Their results are not unusual. In a meta-analysis of negativity scholarship, Lau, Sigelman, and Rovner (2007) find that the majority of studies that quantify the effect of negative ads on candidate evaluations do show that negativity decreases support for the target. As Lau and Rovner summarize, “we are reasonably confident that attacking a political opponent is generally likely to result in lower evaluations of that candidate” (2009, p. 296). This is, however, not to suggest that negative ads will always affect the final electoral choices that people make. Indeed, in analyzing numerous studies of negativity and vote choice, Lau, Sigelman, and Rovner (2007) find little evidence that exposure to negativity caused people to vote for some candidates rather than others. In short, although the exposure to ads resulted in lower evaluations, these lower evaluations did not translate into changes in vote choice.

Even if negative ads do not determine the candidate voters choose on election day, it is possible negativity can affect the level of confidence individuals have in their candidate choices. Krupnikov (2012) shows that exposure to negativity early on in the campaign helps people be more certain about their candidate choices. Although people base their candidate preferences on a myriad of factors (the main one being partisanship), exposure to negativity helps people believe that they made better, more informed choices (Krupnikov, 2012). In turn, higher levels of confidence in political choices positively influence people’s level of political participation (Krupnikov, 2012).

Although it may be most logical to assume that negative ads should affect the level of support for the target, scholars have also suggested that negativity may affect support for the sponsor of the ad as well. While Kahn and Kenney (2004) show that negative ads lead to lower evaluations of the target, they demonstrate that negative ads lead to declining evaluations of the sponsor as well. Indeed, there are consistent findings that sponsoring negativity can lead to lower support for the ad’s sponsor—an outcome typically referred to as the “backlash” effect (Lau & Rovner, 2009).⁶

Although the backlash effect has been replicated across numerous studies (Brooks & Murov, 2012; Dowling & Wichowsky, 2015; Garramone, 1985; Weber, Dunaway, & Johnson, 2012; see Lau et al., 2007, for a summary), recent research suggests that there is conditionality to the size of the effect. The topic of the ad, for example, affects the extent to which the sponsor experiences a backlash effect. One might expect that voters are much more likely to punish the sponsor for an ad they deem unfair (Lau & Rovner, 2009), and indeed, Mattes and Redlawsk (2014) find that ad topics that voters deem “inappropriate”—such as a candidate’s family or religious views—are significantly more likely to generate backlash.

Another factor affecting the extent to which a candidate who sponsors negativity may experience a backlash is candidate characteristics. For instance, research suggests that a candidate’s gender and race can condition the way people respond to negative ads sponsored by the candidate. Turning first to gender, research has suggested that female candidates can be somewhat more likely to experience a backlash for going negative (Herrnson, Lay, & Stokes, 2003; Hitchon, Chang, & Harris, 1997). One possible reason for the disproportionate backlash for female candidates may be the fact that sponsoring negative ads breaks with traditional gender stereotypes (Krupnikov & Bauer, 2014). At the same time, there is also some research to suggest that female candidates are actually no more likely to experience a backlash effect than similar male candidates (Gordon, Schafie, & Crigler, 2003). Moreover, as Fridkin, Kenney, and Woodall (2009) demonstrate, women candidates may be better able to withstand negative attacks from male opponents. Most recently, however, experimental evidence suggests that the relationship between candidate gender and the backlash effect is highly conditional and heavily dependent on partisanship: people are more likely to disproportionately punish female candidates only when they are members of the opposing party who are seen as instigators of negativity (Krupnikov & Bauer, 2014). In short, party is more powerful than gender when it comes to negativity.

Though there is relatively less research on the intersection between the sponsorship of negativity and candidate race, existing work suggests that there are certain conditions under which minority candidates may experience a greater backlash for sponsoring negative ads. In particular, people with negative racial attitudes are significantly more likely to punish a minority candidate who sponsors negativity than a white candidate who does the same (Krupnikov & Piston, 2015).

Another factor that may either strengthen or weaken the backlash effect is the ordering of negativity. A sponsor who is seen as the instigator of negativity, for example, is more likely to experience a backlash than a sponsor who simply responds to already-existing negative ads (Peterson & Djupe, 2005). A candidate who is seen as going negative “without provocation” is more likely to suffer at the hands of the voters than a candidate who sponsors an ad to “[retaliate] against a negative attack” (Lau & Rovner, 2009, p. 292).

In sum, there is some evidence that negative ads can change the way individuals feel about the candidates either by lowering evaluations of the target, lowering evaluations of the sponsor, or both. Nonetheless, this evidence is highly conditional and, in some cases, quite limited. Negativity, for example, is unlikely to change minds and, as Lau et al. (2007) conclude, somewhat more likely to cause a backlash effect. Even then, the backlash effect is not universal but rather confined to highly specific, almost limited conditions. Taking this argument a step further, as Mattes and Redlawsk (2014) suggest, candidates are quite aware of the potential for a backlash, and for this reason the types of ads that are actually highly likely to lead to backlash effects are very rare or are sponsored by other entities—a topic we consider in the next section.

Does the Sponsor of the Negative Ad Matter?

Much of the research on campaign negativity has focused on negative ads sponsored by the candidates themselves, but recent work has turned its focus to the financial sponsorship of advertising. This shift in focus is in response to changes in campaign law that now allow ad sponsorship by independent groups (an entity that is not a candidate or a party that can legally pay for a political ad, but not in concert with a candidate: see Fowler & Ridout, 2010, 2012, 2014). In 2012 and 2014, for example, these types of groups accounted for approximately 26% of all campaign ads—more than the parties (17%), and more than double the amount groups were responsible for in 2008 and 2010 (Fowler & Ridout, 2014). This trend alone suggests that examining whether the sponsor of ads matters for how they are processed by citizens is worthwhile. Examining whether the sponsor of negative ads matters in particular, however, is especially important given the rate at which groups sponsor negative ads—in 2012, 85% of ads sponsored by groups were negative, whereas approximately 50% of candidate- and party-sponsored ads were negative (Fowler & Ridout, 2012).

Most of the work in this area has used experiments to test whether a negative ad has a different effect depending on who the sponsor is. These experimental investigations have for the most part compared ads sponsored by candidates to ads sponsored by independent groups, but a few studies have also examined whether ads sponsored by political parties have different effects than candidate- or group-sponsored ads. The evidence for differential effects of ads by sponsorship is somewhat mixed, although the preponderance of evidence at this point suggests that negative ads sponsored by groups tend to be the most effective (see Dowling & Wichowsky, 2015, for a summary). What the mechanism is that causes these different effects, however, is still somewhat undetermined.

For instance, Brooks and Murov (2012) conducted an experiment in which they compared the effect of a negative ad sponsored by a candidate to the same ad sponsored by an unfamiliar group. They find that the ad sponsored by the group resulted in less backlash for the candidate than the one in which the candidate sponsored the ad himself, a difference they attribute to the inability of people to link the group-sponsored ad back to the candidate. Weber, Dunaway, and Johnson (2012) performed a similar experiment, but added a comparison of an ad sponsored by a known group (in this case, the National Rifle Association). They find that negative ads sponsored by both unknown and known groups are more effective than the same negative ad sponsored by the candidate, but that the unknown group ad was effective regardless of the subject’s feelings about the NRA whereas the NRA-sponsored ad was only effective for those already favorable to the NRA. As a result, Weber et al. (2012) attribute the differential effect of ad sponsorship to differences in source credibility, as known (and liked) groups and unknown groups were viewed as more credible than candidates.

Dowling and Wichowsky (2015) compared the effect of negative ads across three sponsorship conditions: candidate, party, and group. The results of their experimental studies suggest that negative ads sponsored by groups and sometimes parties (in one of two experiments the effect of party sponsorship is statistically distinguishable from the effect of candidate sponsorship) are more effective, primarily because they produce less backlash (consistent with Brooks & Murov, 2012). They also show that the effect of group sponsorship and party sponsorship varies by the type of ad—in one experiment the group-sponsored ad and the party-sponsored ad were equally effective, in another (using a different ad) the group-sponsored ad was more effective. Much like the other topics we have discussed, a finding from Dowling and Wichowsky (2015) illustrates another way in which the effect of negativity is often conditional. Specifically, they find that when an attack ad is not obviously partisan/ideological, a party-sponsored ad polarizes partisans. However, the same ad sponsored by an unknown group does not, most likely because of voters’ unfamiliarity with the group.

Given the relatively consistent finding that group-sponsored ads, particularly those sponsored by relatively unknown groups, permit (would-be) attacking candidates to escape backlash, recent work has also begun to address the question of whether more information about the groups alters the effectiveness of the ad. Both Dowling and Wichowsky (2013) and Ridout, Franz, and Fowler (2015) find that the disclosure of the largest donors (their names, occupations, and amount donated) to groups after the ad (either in a newspaper write-up or as a list) can limit the influence of group advertising. Although this finding has important implications for campaign strategy and campaign finance law, more work on donor disclosure is needed to determine the generalizability of the experimental findings to the real-world, campaign context.

In sum, there is an accumulating body of evidence that the effect of negative advertising does depend on ad sponsor type. It is less clear, however, what mechanism accounts for this difference. Likely explanations are differences in source credibility across sponsors and the ability of candidates to escape backlash when a group (and, to a lesser extent, party) is the sponsor, but more work is needed to disentangle these mechanisms from one another.

Why and When Do Candidates Turn to Negativity?

Using a negative ad does not guarantee success. Rather, there is a possible backlash to using negativity—a backlash that can be escaped, at least somewhat, if the candidate is not the sponsor—and the effects of negativity are somewhat ambiguous. Why, then, would a candidate ever be willing to sponsor a negative advertisement? Research on the strategic use of negativity suggests that candidates do not turn to negativity when their campaigns are going well. Rather, reliance on negativity is, in some sense, a strategy of desperation.

Using a formal model, Skaperdas and Grofman (1995) show that a candidate who is ahead in the polls is less likely to turn to negativity than a candidate who is trailing. Although some have criticized the Skaperdas and Grofman model (see, e.g., Buell & Sigelman, 2008; Harrington & Hess, 1996), there is some empirical research to suggest that their predictions do bear out (Theilmann & Wilhite, 1998). The idea that a trailing candidate might be more likely to risk going negative in a last-ditch attempt to undermine the front-runner would also follow from the idea that negativity may cause a backlash. The front-runner may not want to risk criticism in return for the possibility that a negative ad may in some way decrease support for his or her opponent. Indeed, Buell and Sigelman (2008) find that the trailing candidate is most likely to go negative when the campaign can be described as a “blowout,” which reinforces at least some desperation in the use of negativity by the losing candidate.

Aside from the Skaperdas and Grofman model, other scholars have attempted to consider the strategic use of negativity through observational data. One of the most consistent findings, for example, is that negativity is more likely to appear in more competitive races (Hale, Fox, & Farmer, 1996; Kahn & Kenney, 1999). Controlling for several other explanatory factors, Lau and Pomper (2001) show that candidates are also much more likely to go negative when they have less money than their opponents and when they have already been attacked. Thus, the more negativity there is early on in a campaign, the more likely it is the race will take on an overall negative tone (Kahn & Kenney, 1999).

Focusing on another candidate characteristic, Peterson and Djupe (2005) argue that incumbency status underlies the decision to turn to negativity, with challengers being more likely to rely on negative ads than incumbents. Although not all research has found similar support for the idea that a challenger is more likely to rely on negativity than an incumbent (Lau & Pomper, 2001), negativity may still be a more critical component of a challenger’s strategy than an incumbent’s strategy. As Mattes and Redlawsk note:

The most likely effect of banning negativity would be eliminating the ability of challengers to effectively campaign against incumbents. If a challenger cannot talk about her opponent, cannot describe what a miserable failure he is, and cannot explain to voters why they should not vote for him again, what is left for the challenger to do? (2014, p. 203)

In sum, several factors emerge as pivotal to the tone of the campaign: (1) closeness of the race, (2) incumbency status, (3) funding, and (4) being attacked. This is not to argue that the sheer presence of these four factors will produce a negative campaign. Certainly, each campaign has its own nuances and particularities, and as a result, “a multitude of factors … figure importantly in candidates’ decisions to go negative” (Buell & Sigelman, 2008, p. 442). Nonetheless, these four factors at the very least increase the chance that a candidate becomes more likely to consider the possibility of going negative.

While the above research focuses on the American case, Walter, van den Brug, and van Praag (2014) consider the conditions under which candidates turn to negativity in a comparative prospective. Focusing on Britain, Germany, and the Netherlands, the authors consider what leads different parties to rely on negative ads. Looking beyond America, the study considers both electoral institutions and party characteristics, suggesting that it is party characteristics—such as governmental experience—that matter most in the emergence of negativity. This research provides an important extension to the factors that lead political actors to negative campaigning.

Is Negativity Different When It Is on the Internet?

The previous sections have largely considered televised negativity. More and more, however, candidates have turned to the Internet as an additional medium by which to reach voters. Indeed, most modern campaigns include a large web-based component, utilizing not only candidate websites but also social media sites such as Twitter, Facebook, and YouTube.

There are a number of benefits to web-based campaigning. Internet campaigning is lower cost, allows candidates to respond much more quickly, allows candidates to interact with voters, and provides candidates with unlimited space to transmit their message (Kaid, 2002, 2006). This last point is important: since televised ads are time-limited, candidates have to choose precisely which messages they want to transmit in the short amount of time they have. On the Internet, candidates can launch negative attacks on opponents without limiting the amount of time they have to promote themselves (Druckman et al., 2010).

Pinpointing exactly how the use of web-based advertising techniques has affected the role of negativity in politics is difficult. Some suggest the effects of web and televised advertising will naturally differ due to powerful differences in the media. This effect has generally been termed the “innovation hypothesis.” Others suggest that we are unlikely to observe strong differences—over time, web campaigning should become quite similar to televised campaigning. This effect falls under the “normalization hypothesis.”

Recent research suggests more evidence for the normalization hypothesis. Druckman et al. (2010), for example, conducted a thorough study of candidate websites, comparing the rates of negativity on websites to the rates of televised negativity. They find that the rates are largely equivalent: candidates are just as likely to go negative on websites as they are to go negative on television. Such a similarity suggests that we have not seen the marked changes in campaign styles that the innovation hypothesis would predict.

What is more, the forces that appear to drive candidates to go negative in televised campaigns are the same forces that appear to lead candidates to go negative on the web. Challengers, candidates in competitive races, and candidates in open seat races are more likely to rely on televised negative ads, and the very same types of candidates are also more likely to engage in negativity online (Druckman et al., 2010).

If candidates rely on similar negative campaigning strategies online as they do on television, is online negative campaigning having a different effect when exposure is via the Internet? To date, there have been few studies that have empirically tested the potential differences in the power of online and televised negative ads.

Remaining Questions

Research on campaign negativity has produced a set of rich, empirically diverse findings. What this broad set of results suggests is that negative campaigning has a highly nuanced effect on American politics. Although criticisms of negativity by the mainstream media, politicians, and even voters may suggest that this type of campaigning has consistent deleterious effects on the political process, empirical research shows this is not the case. Although negativity can influence political outcomes, it only does so for certain candidates and under specific campaign conditions. Moreover, while the overall negative tone is in itself important, the very content of the ad—i.e., what, in particular, is being criticized—also plays a key role in the extent to which negativity is influential. Voters, for example, do not dislike all negative ads, just as not all negative ads produce a backlash against the sponsor.

Although existing research has tackled numerous questions about the role of negativity in politics, additional questions remain, and these questions can set the foundation for a fruitful future research agenda on negativity. People, for example, do not receive negative ads in a vacuum. Negativity comes within the context of a particular campaign or even within the broader context of other American political events. Do people respond differently to negativity during a campaign that has been particularly contentious? Moreover, research on framing suggests that people take cues from elites and respond to frames differently under conditions of strong elite polarization (Druckman, Peterson, & Slothuus, 2013). Can polarization also affect responses to negativity? In particular, could polarization increase the likelihood of backlash against the sponsor and thereby decrease the effectiveness of negative advertising?

Existing empirical research suggests that candidate characteristics affect the extent to which they benefit from relying on negativity. Candidate race and gender, for example, can limit their advertising strategy. Although there is a good deal of work on the intersection of gender and campaign advertising, there is less work on candidate race. Future work, then, can further explore how the race of the candidate affects people’s responsiveness to negative advertising sponsored by this candidate and whether minority candidates are differentially effected by campaign negativity.

Next, existing research shows that the sponsor of the ad does matter. People respond differently to negative ads sponsored by the candidate than to negative ads sponsored by independent groups. Less clear, however, are the mechanisms underlying these differences. Future research, then, may focus on understanding why differences in sponsorship produce this variation in response, exploring individual understanding of ad sponsorship in greater depth.

The bulk of the existing research on campaign negativity focuses on advertising. At the same time, however, a good deal of negativity stems from sources such as candidate speeches, debates, and media discussions. People, research suggests, respond quite strongly to conflict and disagreements they see on news programs (Mutz, 2015), and future research could consider whether criticisms levied in speeches and interviews have a more powerful effect than criticisms presented through sponsored ads. Alternatively, criticisms may be viewed as more “fair” and “honest” when they are brought up as part of a public speech by a candidate than when they are part of an advertisement.

Finally, people are now receiving more and more negativity via the web. Although Druckman et al. (2010) document the use of negativity on candidate websites, research has yet to consider whether campaign negativity transmitted via the web affects people differently than televised campaign advertising. On the one hand, Internet negativity may have no distinguishable effect from more “traditional” forms of negative advertising. As Druckman et al. (2010) find, candidates turn to negativity on their websites under conditions strikingly similar to those under which candidates turn to negative advertising on television. This overlap in strategy suggests that although the mode may be different, the effects of web-based negativity may mimic those of televised negativity. On the other hand, negativity on the web may come in forms that are different from traditional televised negativity. Negative web ads may be embedded in websites that have little to do with politics, and it is possible that (at least for the next several election cycles) web ads may take more people by surprise than televised ads. Furthermore, it is possible that web-based negative ads may come with commentary and could be shared by people via social media. As a result, the change in mode from television to the Internet may also result in a change of effect, and web-based negative ads may lead to different outcomes than televised ads. This is, of course, an empirical question and one that scholars have yet to explore fully.

In sum, existing literature on campaign negativity has highlighted important debates (e.g., the relationship between negativity and turnout) and important areas for consideration (e.g., the role of ad sponsorship). In doing so, this research has set a strong foundation for future work. As of this writing, modern campaigns show no signs of growing more positive, more groups are sponsoring advertising, and campaigns are becoming more and more focused on the web. If the campaign process continues in this direction, it is likely that campaign negativity will continue to be a crucial factor and scholarship on negativity will continue apace.