Trends in Tone

It is nearly impossible to talk about political advertising without also discussing advertising tone and negativity in particular. Negative spots have dominated the airwaves in recent elections comprising 65% of all congressional advertisements aired from September 1 through Election Day in 2006 and 66% of all those aired in 2008. Negativity peaked in 2010 congressional races at 75% of general election airings and receded somewhat in subsequent years (to 68% in 2012 and 70% in 2014), but nevertheless remain very high. Presidential campaigns are even more overwhelmingly negative. Nearly three quarters (73%) of general election airings in the 2008 contest between Barack Obama and John McCain were negative, and that figure grew to nearly 87% in the 2012 contest between Obama and Mitt Romney.

As these data suggest, contemporary congressional and presidential campaigns are predominantly negative, especially in competitive contests (Fowler & Ridout, 2013, 2014). In addition, we also know that the use of negativity differs across sponsors. Figure 3 documents different sponsorship trends in the use of negative and positive advertising, in congressional campaigns. The top panel displays the proportion of all negative and contrast (upper left-hand panel) and positive (upper right-hand panel) airings in each general election cycle accounted for by each type of sponsor. The bottom panel of Figure 3 shows the proportion of each sponsor’s airings that are negative.

Focusing first on the top panels, it is clear that candidates, parties, and interest groups play different roles in contributing to overall levels of campaign negativity and positivity. Whereas candidate-sponsored negativity has consistently comprised about half of all negativity on the airwaves between 2006–2014, party-sponsored negativity was more prominent than interest group-sponsored negativity prior to the 2010 midterms. In 2008, for example, party-sponsored negativity comprised 49% of all advertisements aired in congressional races, while interest group negativity accounted for only about 3%. In sharp contrast, only 26% of negative ads aired in the 2010 midterms were put on the airwaves by parties, whereas 17% were aired by interest groups. For the remainder of elections in the series, parties and interest groups have played roughly equal roles in contributing to overall levels of negativity in congressional campaigns.

Open in new tab

Figure 3. Advertisement Tone and Sponsorship Trends in Congressional General Election Cycles.

Note: Trend lines and bar graphs reflect all advertisements aired after 9/1 in each election year. Wesleyan Media Project (2006, 2010, 2012, 2014) and Wisconsin Advertising Project (2008) data. 2006 data reflect airings in the top one hundred media markets only.

Conversely, nearly all positive advertisements aired in congressional general election cycles between 2006–2014 have been candidate-sponsored advertisements. Candidates consistently sponsor more than 85% of all positive ads on the airwaves. However, it is not the case that candidates avoid going on the attack. As the bottom panel of Figure 3 demonstrates, candidate airings in congressional elections between 2006–2014 tend to be more or less balanced in terms of how many negative messages they air (hovering around the 50% mark). Negative messages comprised more than 60% of all candidate advertisements aired in only one election cycle in the series: the highly negative 2010 midterms, which may in part have been driven by slow economic recovery. In contrast, the overwhelming majority of interest group and party advertisements aired are negative. More than 90% of party-sponsored ads were negative in every election cycle displayed here, and interest group airings ranged from a low of 75% negative (in 2006) to a high of 93% negative in 2010.

What these findings suggest is that, while parties and interest groups play prominent roles in contributing to overall levels of advertising in congressional campaigns, the lion’s share of advertisements that they air are negative. Consequently, parties, and interest groups account for roughly half of all negativity in congressional races on the airwaves in most election cycles, but never account for more than a sixth of positive advertisements aired (at best).

Why the discrepancy in the deployment of negativity? As discussed earlier, voters dislike negative advertising and may punish sponsors of negativity. However, a growing line of research suggests that sponsors do not experience this “backlash” equally. In particular, interest groups are not on the ballot and the paid for byline frequently says explicitly that they are not connected to a candidate, which decreases the likelihood that their favored candidate will be punished for the negativity they air (Brooks & Murov, 2012; Dowling & Wichowsky, 2013, 2015). Candidates may therefore have an incentive to allow interest groups to engage in negative campaigning on their behalf—they gain the benefit of the attack without the penalty associated with the backlash. Consistent with this view, Figure 3 demonstrates that interest group-sponsored negativity has become increasingly common over time in congressional contests (the presidential airings data reveal a similar trend).⁷ Of course, these data cannot discern whether or not this trend is demonstrative of candidates’ conscious efforts to let others “do the attacking,” and the figure also makes clear that candidates themselves do not shy away from negativity—even if they employ it less often than their party and interest group counterparts.

The Substance of Advertising

Next, we attempt to extend a few important insights from Geer’s 2006 book In Defense of Negativity. Geer argues that negative advertisements are “asymmetrically constrained” to focus on issues relevant to voters compared to positive advertisements, in part because they make claims requiring external information sources to back up them up. As Mattes and Redlawsk (2014) argue, despite disliking negative campaigning in the abstract, voters are unlikely to punish sponsors of negativity when specific negative ads focus primarily on policy issues and may even report viewing these types of advertisements as “helpful” in decision-making.

Figure 4 plots the proportion of negative (left-hand panel) and positive (right-hand panel) advertisements that “cite” at least one outside source (e.g., a newspaper, website, piece of legislation), as well as the proportion that focus exclusively on policy issues (and not candidates’ personal characteristics or some combination of the two). Because data limitations prevent us from analyzing citation trends in 2008, we carried this analysis out for congressional elections only: extending the trend line across the missing data point in the process for the “source documentation” line.

Open in new tab

Figure 4. Source Documentation and Issue Focus in Congressional General Election Cycles.

Note: Trend lines reflect all advertisements aired after 9/1 in each election year. Wesleyan Media Project (2006, 2010, 2012, 2014). 2006 data reflect airings in the top one hundred media markets only. ^Indicates that the 2008 data did not contain a measure of source documentation, and so the trend line in the left hand pane does not include data from that election cycle.

As shown in the figure, the proportion of advertisements focusing exclusively on policy issues is similar across positive and negative advertisements. However, as expected, that proportion is higher for negative advertisements, for every year in the series. On the other hand, there are stark differences in the extent to which positive and negative advertisements “cite” external sources. More than half of negative advertisements aired in each election in the series (again, excluding 2008 due to data constraints) document at least one source, while less than a quarter of positive advertisements do the same (with 2014 posing the only exception: where 30% of positive advertisements cited an external source), supporting Geer’s “asymmetric constraint” hypothesis.

Partisan and Ideological Cues

We conclude our content analysis by drawing attention to a rapidly growing literature in political science. Recently, scholars have drawn attention to the premise that the mass public’s evaluations of Democrats and Republicans have become polarized on the basis of affect. One mechanism by which affective polarization is thought to occur is through campaign messaging (of which political advertising is one form). Candidates for public office from one political party denigrate and attack members of the other in widely aired and expressly partisan messages, encouraging partisans to dislike members of the other party on a highly personal level (Iyengar, Sood, & Lelkes, 2012; Iyengar & Westwood, 2014). While there is strong reason to believe that partisans’ affect toward opposing partisans has certainly become polarized along affective lines in recent years, our content analyses cast some doubt on the possibility that negative campaigning is the mechanism by which these effects occur.

Figure 5 plots the proportion of all general election congressional advertisements that make mention of candidates’ political parties and ideological affiliations. The left-hand side of the figure displays the proportion of advertisements aired that make mention of the favored candidate’s party (either the sponsor, or the promoted candidate in a party or interest group spot), the opposing candidate’s party (the “target” of negativity in attack or contrast ads), and the proportion of ads making no mention of party at all.⁸ The right-hand side captures whether or not the phrases “liberal” and “conservative” were mentioned at all in the content of all advertisements aired (WMP did not code for whether this word was mentioned specifically to describe the favored or targeted candidate). As was the case in Figure 4, data limitations prevent us from conducting these analyses on liberal and conservative mentions in 2008—so we limit our analyses to congressional races only, and extend the time series line through the missing 2008 data point.

Open in new tab

Figure 5. Mentions of Party Labels and Ideology in Congressional General Election Cycles.

Note: Trend lines reflect all advertisements aired after 9/1 in each election year. Wesleyan Media Project (2006, 2010, 2012, 2014). 2006 data reflect airings in the top one hundred media markets only. ^ Indicates that the 2008 data did not contain a measures of mentions of liberals and conservatives, so the trend line in the right hand pane does not include data from that election cycle.

If out-party denigration in political campaigning leads to affective polarization, we might expect to see that a large number of political ad airings make mention of opposing candidates’ political parties. In contrast, for every congressional election cycle in the series, the proportion of advertisements mentioning neither the favored nor opposing candidate’s party vastly outnumber those that do. The proportion of advertisements making no mention of party whatsoever ranges from a low of 83% in 2006 to a high of 95% in 2008. Further, for almost every year in the series (with the exception of 2012), the proportion of favored candidate party mentions was at least as high or higher than the proportion opposing candidate party mentions.⁹ Mentions of ideology tell a similar story. Since the Wesleyan Media Project does not code for whether or not individual candidates are labeled as liberals or conservatives, our analyses are likely to overreport the amount of ideological candidate labeling in ads. Still, the proportion of ads mentioning neither liberals nor conservatives never dips below 90% of all ads aired in the (four) election cycles for which data was available. Mentions of conservatism were highest in 2012 (6% of all ads aired), whereas mentions of liberalism were highest in 2006 (again, 6% of all ads aired). Whatever concerns may be raised about advertising, it is clear that they do not emphasize partisan or ideological loyalties.

Vote Choice Effects

Exposure to political advertising, in general, has been shown to play a role in shaping citizens’ electoral choices, as well as their attitudes toward candidates running for office. At the individual level, scholars have noted that individuals living in media markets where one candidate airs more advertisements than another are ceteris paribus more likely to report voting for (or planning to vote for) the candidate with an advertising advantage (Franz & Ridout 2007; Fridkin & Kenney, 2004; Ridout & Franz, 2011). Ridout and Franz (2011) find, for example, that in the 2000 election a half standard deviation increase in the share of pro-Gore advertisements, when compared to a half standard deviation increase pro-Bush advertisements, resulted in more than a 30 percentage point increase in the likelihood of voting for Gore (from a 30% to over a 60% chance of voting for Gore). These effects were almost perfectly reversed given an identical increase in the share of pro-Bush advertisements, relative to Gore ad exposure. Furthermore, in the aggregate, others have noted that—in the most recent presidential election—tracking poll advantages tended to closely mirror short-term advertisement share advantages (Sides & Vavreck, 2014).

These studies all offer support for the notion that political advertisements can persuade individuals to update their electoral selections. Nevertheless, it could be the case that increases in political advertising are accompanied by other alterations in campaign strategy, such as more intensive efforts designed to recruit voters “on the ground.” How can researchers determine that advertisements alone cause citizens to update their electoral preferences?

Huber and Arceneaux (2007) offer one such approach, by taking advantage of the fact that media markets occasionally cross state boundaries. Their quasi-experimental approach shows that intense advertising campaigns in battleground states (where “ground game” efforts are likely to be quite strong) produced spillover effects in non-battleground states (where “ground game” tactics are presumably weaker) that share a media market. Controlling for citizens’ partisan and previously formed electoral commitments in both panel and cross-sectional surveys, a one standard deviation increase in pro-Bush advertising exposure increased the likelihood that citizens cast their ballot for him by more than 9% (and decreased the likelihood that they would do the same for Gore by about 8%). Gerber and colleagues (2011) offer another solution to the aforementioned causality issue. These researchers teamed up with Rick Perry’s gubernatorial campaign to create and air an advertisement on his behalf—varying the intensity of advertisement exposure across a number of different media markets, and tracking public opinion simultaneously. Consistent with the idea that advertising advantages produce short-term persuasive effects (Hill et al., 2013; Sides & Vavreck, 2014), they found that increased exposure meaningfully improved Perry’s (predicted) vote share, but for no longer than seven days. After one week, the effects largely dissipated.

Overall, this research suggests that exposure to political advertising in general can yield persuasive effects. But what about conditional effects? Might some characteristics of advertisements allow them to be better (or worse) at persuading voters than others?

One element of ad content that can enhance persuasive effectiveness is advertising tone. Negative advertisements are particularly well-suited to encourage individuals to alter their electoral preferences. In both presidential and senatorial contests, increased exposure to campaign negativity from one side usually rewards the side doing the attacking (Ridout & Franz, 2011), although—at least in Senate races—it may be the case that challengers benefit more than incumbents (Lau & Pomper, 2004). Negative ads have also been shown to be effective at altering citizens’ affective impressions about candidates, such that attack and contrast ads encourage citizens—especially those who pay little attention to politics—to view targeted candidates more negatively (Fridkin & Kenney, 2004). Though recent meta-analyses have not found consistent evidence that increased negativity results in behavioral electoral outcomes that are beneficial to attackers (Lau et al., 1999; Lau & Pomper, 2004; Lau, Sigelman, & Rovner, 2007), they do suggest that more negativity is associated with more negative affective impressions of targeted candidates, which is a first step in behavioral change.

Under certain conditions, however, negative advertisements may not always produce the effects that campaign strategists intend. The “backlash effect” occurs when voters punish candidates at the polls (or in the ballot box) for airing negative advertisements (Brooks & Geer 2007; Kaid, 1997; Mattes & Redlawsk, 2014; Pinkleton, 1997). Overall, Ridout and Franz (2011) find that the impact of negativity on vote choice tends to more closely resemble strategists’ “intended effects.” Consistent with this view, Mattes and Redlawsk (2014) note that voters generally view most negative information (the notable exception being lies and defamation, see Fridkin & Kenney, 2004) as quite helpful in making political decisions, particularly when negative campaigning is referred to without expressly using the term “negative.” In contrast to other work, they find little evidence of backlash toward the sponsors airing negative messages in the laboratory environment.

Other work, however, suggests that some sponsors of negativity may be more likely to experience backlash than others. Though negativity aired by both incumbents and challengers in Senate contests has been shown to decrease affective evaluations of targeted candidates, incumbents—but not challengers—may receive an electoral punishment for doing so (Fridkin & Kenney, 2004; Lau & Pomper, 2004). In the laboratory environment, negative advertisements sponsored by candidates tend to result in more backlash (negative affect toward sponsors) than unfamiliar interest groups (Brooks & Murov, 2012; Dowling & Wichowsky, 2013) and—to some extent—political parties (Dowling & Wichowsky, 2015).

Therefore, candidates may benefit from letting unfamiliar interest groups “do their dirty work” (Dowling & Wichowsky, 2015; Fowler & Ridout, 2010) as the extent to which negative advertisements persuade has been shown to be mediated by perceptions of credibility in the sources of sponsorship (Weber, Dunaway, & Johnson, 2012). That is to say that citizens use the sponsors of advertisements as “source cues” (see also Eagly & Chaiken, 1993; Lupia, 1994)—cognitive shortcuts that aid them in making inferences concerning whether or not they view the claims made in the advertisements as credible. Because citizens are more trusting of groups whom they know little about than they are of candidates or parties, candidates may have an incentive to avoid “going negative” themselves, and let increasingly active, yet unknown to the public, interest groups take that action on their behalf (Ridout, Franz, & Fowler, 2015).

In addition to advertising tone, the emotional appeals featured in political advertisements have been shown to have strong effects with respect to vote choice. Consistent with Affective Intelligence Theory (Markus, Neuman, & MacKuen, 2000), advertisements that appeal to fear (which are almost always negatively toned) can lead individuals to experience anxiety—which has been shown to encourage them to reconsider their prior beliefs and political affiliations (Brader, 2005, 2006). Those experiencing anxiety in the context of negative campaigning are more likely to consider supporting candidates who offer a remedy to a potentially dangerous status quo than those who are exposed to negativity without such emotional appeals.

Political advertising expenditures are often the single-largest expense that contemporary federal political campaigns must account for in their budgets (Fowler, Franz, & Ridout, 2016; West, 2009), and while the academic literature has not always agreed with political strategists’ emphasis on messaging, there is plenty of evidence to suggest that advertising can matter, even if its effects are not always large. Advertising exposure in general has been shown to play a key role in shaping individual vote choice intentions. The tone of advertisements, as well as the emotional appeals they make, can also shape vote choice. However, some of the effects that negativity and emotional appeals exert on citizens’ political behavior are more closely tied to whether or not citizens show up to vote in the first place. We take up those issues in turn.

Mobilization Effects

In addition to persuading citizens to change their vote choices, political advertisements also influence whether or not citizens turn out to the polls. At the level of individual voters, frequent television viewers residing in media markets with high levels of advertising across states are more likely to turn out to vote on Election Day (Freedman, Franz, & Goldstein, 2004) and report that they intend to turn out (Hillygus, 2005). However, in the aggregate, validated turnout levels do not seem to vary across levels of advertisement exposure after accounting for differences in campaigning tactics between and within state boundaries (Krasno & Green, 2008). While these differences pose an interesting empirical puzzle, it is important to point out that neither account suggests that overall levels of advertising exposure decrease turnout.

The same cannot be said, however, about negative advertising. The premise that exposure to campaign negativity may decrease citizens’ willingness to turn out to the polls on Election Day has been a rich source of scholarly research over the past two decades. Early studies of the link between negativity and turnout found that—in both the experimental environment and in the aggregate—exposure to attack advertising was associated with lower levels of turnout as well as the propensity to cast ballots on Election Day (Ansolabehere et al., 1994; Ansolabehere & Iyengar, 1995). As Geer (2006) notes, analyses of presidential campaign advertisements dating back to the early 1960s offer limited bivariate support for this hypothesis: more negative campaign cycles did tend to correlate with lower levels of turnout. Many studies have challenged the demobilizing effects of negativity on the grounds that they either increase (e.g., Finkel & Geer, 1998; Franz & Ridout, 2007; Goldstein & Freedman, 2002) or fail to decrease (e.g., Garramone et al., 1990; Green & Krasno, 2008; Wattenberg & Brians, 1999) voter turnout.

Finkel and Geer (1998), for instance, found that while aggregate turnout levels and campaign positivity have decreased over time, they have not always done so simultaneously. For example, the highly negative 1980 presidential campaign actually boasted somewhat higher levels of turnout overall than the more positive 1976 contest. The duo further used individual-level survey data to model voter turnout intentions in relation to negative advertising exposure. They found that increased exposure to campaign negativity between 1960 and 1992 did not depress turnout intentions for most voters, though it slightly decreased it for political independents (in certain models). Franz and Ridout (2007) expanded on this individual-level approach by matching respondents’ self-reported television viewership habits (i.e., whether or not they tune into particular programs at a certain time) to advertisement frequency and content data (for methodological background and validation, see Freedman & Goldstein, 1999; Ridout et al., 2004) in order to estimate citizens’ exposure to negative advertising. Overall, they find that exposure to negativity not only fails to depress turnout, but may actually slightly increase it—a finding consistent with broader meta-analytical assessments of the field (Lau, Sigelman, & Rovner, 2007), and earlier investigations using similar methodological approaches (Goldstein & Freedman, 2002). In the aggregate, Krasno and Green (2008) too do not find evidence that exposure to negativity depresses turnout, although they also do not find evidence of a mobilization effect in response to negative campaigning in the aggregate.¹⁰

It may also be the case that the link between negativity and turnout is moderated by a host of factors pertaining to advertisement content. For example, Kahn and Kenney (1999) find that campaign civility in senatorial campaigns (whether or not negativity focuses on “legitimate” political issues or “inappropriate” and derisive personal characteristics of candidates) can have both mobilizing and demobilizing effects on turnout. Survey analyses of voters’ turnout intentions show that civil negativity can boost electoral participation, whereas uncivil attacks depress it. Brooks and Geer (2007) offer a potential individual-level mechanism by which such effects may occur: as voters tend to dislike uncivil ads, and view their content as less credible than more civil forms of negativity. However, that same experimental trial also showed that incivility in general did not depress voters’ intentions to turn out. The important takeaway from this research is that different forms of advertisement negativity can play an important role in influencing whether or not citizens turn out to vote.

Another element of advertising content that may influence turnout is the extent to which advertisements appeal to some emotions and not others. Exposure to negative advertisements that also attempt to instill anxiety in those watching them, as well as positive advertisements that attempt to do the same with enthusiastic emotional content, is associated with higher levels of reported turnout in experimental studies (Brader, 2005, 2006).

Further, the most recent work on the subject suggests that the timing of negative campaigning may influence whether or not citizens turn out to vote. When negative advertisements aired late in campaigns conflict with citizens’ prior electoral selections, citizens are thought to experience vote choice ambivalence (conflicting negative and positive evaluations about the candidate for whom they wish to vote), and are subsequently less likely to turn out to vote (Krupnikov, 2011). In the 2004 presidential election, for instance, Krupnikov (2011) found that a 60% increase in exposure to negative advertising targeted at voters’ preferred candidates was associated with nearly a 15% reduction in the likelihood that they would turn out to vote. Overall negativity throughout the campaign, however, had no relationship to voter turnout. Conceivably, this could be the case because negativity aired earlier on in campaigns—when voters are still making up their minds about for whom they should cast their vote—does not conflict with their electoral choices.

Advertisements also influence voter turnout when news media outlets pick up and amplify certain advertisements in their coverage of the campaign. Studies suggest that news coverage of advertising is common but also that it much more likely to focus on negative ads or new tactics (Fowler & Ridout, 2009; Ridout & Smith, 2008), which could increase and reinforce the persuasive benefits of the original ad even when the ad is mentioned in a critical way (Pfau & Louden, 1994), but may also increase the risk of backlash for candidates (though see Mattes & Redlawsk, 2014, for a counter perspective on backlash).

Overall, advertising exposure in general may boost turnout levels (or, at the very least, not decrease them), and exposure to campaign negativity in particular has the potential to mobilize as well although these effects may be conditioned upon ad characteristics and the timing of citizen exposure to criticisms. News media amplification of advertising may further boost persuasion and may also drive up citizen perceptions of campaigns being very negative, especially when the coverage is focused on negativity as a strategic tactic (Ridout & Fowler, 2012).

The Downstream Consequences of Political Advertising

In addition to altering vote choice and turnout, exposure to political advertising (both positive and negative) has been shown to have a wide range of consequences pertaining to citizens’ knowledge about electoral politics and views of the political system. Dealing first with knowledge, increased exposure to advertising in general has been shown to boost citizens’ abilities to recall candidates’ names, to correctly guess where candidates stand on highly salient issues, and to increase interest in politics (Freedman, Franz, & Goldstein, 2004; Ridout et al., 2004). Further, as campaigning efforts become more intense, citizens appear to make use of information about political issues, the performance of the incumbent president, and impressions about the economy when formulating attitudes about presidential (Bartels, 1993) and Senate candidates (Kahn & Kenney, 1997).¹¹ In general, then, many scholars have found that intense campaign advertising efforts, and exposure to advertising more generally, can be particularly useful in boosting knowledge about politics.

Consistent with the idea that political advertisers expertly craft spots in order to be informative to even the most unengaged voters—perhaps even serving as information supplements for those who need it most (Freedman, Franz, & Goldstein, 2004)—early studies of learning effects in response to political advertising exposure found that politically inattentive individuals learned more from advertisement exposure than those who are more politically aware (Patterson & McClure, 1976). More recent analyses, though, have casted doubt on this optimistic view of campaign learning effects. In both panel survey (Freedman, Franz, & Goldstein, 2004) and experimental (Valentino, Hutchings, & Williams, 2004) studies, scholars have not consistently found that low-information voters learn more from increased exposure to campaign advertisements than those who already hold high levels of knowledge about politics and current affairs. This is not to say that low-information voters fail to learn from campaign advertisements altogether, as both of the aforementioned studies do find sizable learning effects for even the least attentive citizens. However, despite already possessing a wealth of information about politics and current events, high-information voters tend to learn from campaign advertisements in such a way that is not statistically discernible from that of less attentive voters.

The extent to which citizens learn from advertising is shaped by certain ad content features, such as tone and the emotional appeals they make. Meta-analyses of the negative advertising literature (Lau, Sigelman, & Rovner, 2007) have shown that increased exposure to negative advertisements increases the extent to which citizens remember the information to which they were exposed (e.g., Brader, 2005; Geer & Geer, 2003; Lang, 1991) and boosts knowledge about political campaigns (e.g., Craig, Kane, & Gainous, 2005; Stevens, 2005). This may be due to the idea that humans are more attentive to negatively valenced information than positively valenced information (Lang, Dhillon, & Dong, 1995; Soroka & McAdams, 2015). Consequently, negative advertisements are more likely to feature appeals to politicians’ stances on issues that are highly salient in public discourse, and are more likely to provide documentation of the sources pertaining to the claims that they make (Geer, 2006; see also Figure 4 in this chapter). When negative advertisements do this, citizens tend to view negativity as being helpful in making political decisions. Perceived helpfulness increases the extent to which citizens view negative ads as believable and appropriate (and decreases the extent to which they are angered by negativity),¹² which then play a role in shaping vote choice (Mattes & Redlawsk, 2014). However, this is not to say that positive ads do not have the potential to play a role in shaping citizens’ abilities to learn about politics. Because negative ads are more likely than positive ads to include discussions of salient political issues, it could be the case that positive advertisements that feature as much information as negative spots are similarly able to foster learning effects (see the argument and experimental evidence in Stevens, 2012).

Psychologically, Affective Intelligence Theory offers additional insight into why exposure to negatively toned advertising may be more memorable and informative than other forms of campaigning. Negative messages, especially those that also make emotional appeals to fear, can be thought about as stimulating the extent to which citizens search for additional information about a threatening stimulus (e.g., a politician’s actions in office, or issue position), pay attention to information that runs contrary to their prior beliefs, and render evaluations of candidates that eschew prior partisan commitments and focus instead on issue stances (Brader, 2005, 2006; Markus, Neuman, & MacKuen, 2000). When evaluating hypothetical candidates in the laboratory environment, Brader (2005, 2006) shows that exposure to negative advertisements that also featured appeals to fear (relative to those that do not make appeals to fear) may stimulate the willingness to search for more information, and (slightly) improve the recall of relevant information in the ad.

However, the emotional content of political advertising is not without potential harmful effects on public opinion and policy outcomes. For example, citizens’ negative racial attitudes are thought to be recalled into memory more readily when individuals experience anger (Banks & Valentino, 2012; Banks, 2015). Consequently, the presence of angry emotional cues in advertisements that make subtle (implicit) mentions of race can encourage citizens to rely on their negative attitudes toward blacks when evaluating political issues and candidates for office (Banks, 2015; Banks & Bell, 2013). In general, angry emotional cues are quite common in political advertising—present in nearly 75% of advertisements aired in the 2012 general election (Fowler & Ridout, 2013). Interestingly, it is important to point out that exposure to negative advertisements in general (i.e., irrespective of the emotional appeals that they might make) does not necessarily make respondents experience anger (Mattes & Redlawsk, 2014). Though individuals are slightly more likely to feel angry after viewing negative ads than positive ads, only those who are exposed to particularly defamatory attacks experience high levels of anger.

In addition to concerns that negative advertising exposure may depress citizens’ willingness to turn out to vote, scholars have also considered whether or not campaign negativity affects political efficacy, trust in government, and views of the system more broadly with mixed results. In an experiment, Brooks and Geer (2007) note that while citizens view trait-based negative attacks as “unfair,” incivility does not decrease perceptions of political efficacy and may even increase the extent to which citizens are interested in politics and turn out to vote. Other research supports the view that negativity does not harmfully affect efficacy and trust (Geer, 2006; Jackson, Mondak, & Huckfeldt, 2008). In their meta-analysis of the relationship between negative advertising and feelings of political efficacy, Lau and colleagues (2007) noted small, yet consistently negative, relationships between exposure to negative advertising and feelings of trust in government, affect toward the government, and perceptions of efficacy (e.g., Stevens, 2002; Thorson et al., 2000). Notably, although the literature is divided, the evidence supporting a more harmful view of negativity in U.S. elections does not suggest that the effects are large.