Negativity and Political Behavior: A Theoretical Framework for the Analysis of Negative Voting in Contemporary Democracies

Introduction

While voters in democratic elections are called to the polls to express their preference in favor of a given set of parties and candidates, it is well-established that a consistent proportion of voters considers their electoral choice as a vote against. American research on negative voting dates back to the 1960s. Yet, the more recent literature on the electoral consequences of affective polarization suggests that negativity remains an important feature of American citizens’ political behavior. Survey data from the 2020 US Presidential election find 30% of voters declaring to have cast a negative vote (Garzia and Silva, 2021a).

Moreover, there are indications that negative voting also pertains to multi-party democracies. For instance, in the second round of the 2002 French presidential election, voters from the left rallied under the call to vote for “the crook” (i.e. Jacques Chirac) rather than “the fascist” (i.e. Jean-Marie Le Pen; Medeiros and Nöel, 2014: 1023). In Brazil, the 2018 Presidential election was marked by ferocious negative campaigning against the Workers’ Party. The crystallization of diffuse negativity toward the incumbent led to the spread of Antipetismo, which arguably played a decisive role in the election of Jair Bolsonaro (Samuels and Zucco, 2018). Turning to parliamentary democracies, large proportions of the Italian electorate have been shown to cast a “dishonest vote”—that is, a vote for an admittedly dishonest candidate based only on negative attitudes toward his or her opponents (Garzia, 2014). In each of these examples, the reasons behind voters’ behavior appear purely negative—and yet, highly consequential for election outcomes.

We follow the existing literature and define negative voting as an electoral choice more strongly driven by negative attitudes toward opposed parties and candidates than by positive attitudes toward one’s preferred party and candidate (Fiorina and Shepsle, 1989; Kernell, 1977). On the bases of this characterization, “[a] negative voter [. . .] does not like his or her preferred candidate. Rather, the negative voter is motivated by antipathy toward the opposition; he or she votes ‘against’, not ‘for’” (Gant and Davis, 1984: 272).

In this article, we review the scant (and largely disconnected) scientific literature on negativity and political behavior, merging contributions from social psychology, public opinion, and electoral research, with a view on developing an integrated theoretical framework for the study of negative voting in contemporary democracies. As we shall argue, the tendency toward negative voting is driven by three (partly overlapping) micro-behavioral components, namely, (1) an instrumental–rational component characterized by retrospective performance evaluations and rationalization mechanisms, (2) an ideological component grounded on long-lasting political identities, and (3) an affective component, motivated by (negative) attitudes toward parties and candidates.

Early Analyses of Negative Voting in the United States

Although the first piece of empirical work to explicitly mention the term “negative voting” is an article by Kernell (1977), this notion traces back to V O Key’s (1966: 60) contention that “the people only vote against; never for.” The hypothesis that citizens do not always vote for the candidate they like the most has been put to empirical test by American electoral research in the 1970–1980s, often with confirmatory results (Gant and Davis, 1984; Sigelman and Gant, 1989). Still, from an attitudinal perspective, negative voting remained “an observed regularity with an as-yet uncertain explanation” (Fiorina and Shepsle, 1989: 424).

Rational choice approaches to negative voting understood mainly it as a special case of retrospective voting in elections involving incumbents (Kernell, 1977). Yet, the rational choice paradigm was also instrumental in the development of a political psychology perspective that envisaged negative voting as a rationalization mechanism. The influence of prospect theory in the later development of a vast array of political psychology studies on the positive–negative asymmetry is, perhaps, a good example of such an intersection. The positive–negative asymmetry in human cognition has long been a prominent research topic in social psychology (for an extensive review, see Baumeister et al., 2001), but only in the 1980s did a coherent research core develop around the study of its implications for voting behavior.

In Lau’s (1982, 1985) analyses of American mid-term elections, the “negativity effect” refers to the higher weight given to negative information relative to equally extreme and equally positive information in a variety of information-processing tasks. While the previous voting behavior literature offered scant rationale for the effects of positive–negative asymmetry, Lau’s theoretical framework tested two potential explanations. The motivational cost-orientation perspective hinges on people’s inclination to avoid costs rather than to approach gains, which could translate into a stronger motivation to consider negative information over positive cues. The perceptual figure-ground hypothesis postulates that negative information stands out against a positive background. In Gestalt terms, negative information is figural against a positive background, offering a greater perceptual contrast and thus becoming more salient. Also, as negative information is counter-normative, it is potentially perceived as more credible and informative, since disseminating negative information carries higher costs. These studies’ results provide support for both explanations and find a negativity effect on voters’ impression formation of presidential candidates, as well as on vote choice and turnout in American presidential elections.

Gant and Sigelman (1985) also moved from a political psychology background to develop and test the notion of “anti-candidate voting.” They defined this form of voting behavior as one focused more on an opposed candidate one than on a preferred one—or, more specifically, as a form of voting that occurs “whenever the vote decision hinges on negative affect toward one candidate, irrespective of the feelings one has toward the candidate for whom one votes” (Gant and Sigelman, 1985: 329). Influenced by cognitive dissonance theory, these authors conceived negativity as the product of psychological conflicts. By considering their candidate support as a rejection of the opponent, more than as an endorsement of the preferred candidate, negative voting emerges as a “psychological device by which the voter tried to cope with the dissonance engendered by cross-pressures” (Gant and Sigelman, 1985: 331). Based on information-processing theory, Gant and Davis (1984) asserted that negative information is rarer and thus perceived as more valuable than positive information, leading to more consequential behavioral responses.

Eventually, scholarly attention to these aspects faded away as a result of the profound changes in American society over the closing decades of the twentieth century. After a period of electoral dealignment, the realignment of the American electorate into ideologically and socially sorted partisan groups, reduced cross-pressures and motivated a growth in affective polarization (Dalton, 2018; Levendusky, 2013; Mason, 2018). A type of polarization that is less ideological than based on othering, aversion, and moralization has led to rising political sectarianism in American society, whereby out-party hate stands out as a stronger force than in-party love (Finkel et al., 2020). ¹

Examining the electoral consequences of these processes, a recent strand of literature has demonstrated that these underlying divides may lead to the development of negative partisan identities, pervading both Presidential and parliamentary democracies (Bankert, 2020; Caruana et al., 2015; Mayer, 2017; Medeiros and Nöel, 2014). Similarly, other studies have argued that negative attitudes toward out-groups are not limited to political parties but can also transpire to candidates or party leaders (Abramowitz and Webster, 2018; Garzia and Silva, 2021a, 2021b). In the sections that follow, we take stock of these recent contributions, focusing in turn on the negative partisanship literature and what we designate by negative personalization.

Negative Partisanship

The negative partisanship literature finds inspiration in Maggiotto and Piereson’s (1977) original “hostility hypothesis.” These authors insist on Campbell et al.’s (1960: 122) unappreciated notion that negative partisan attitudes are an intrinsic part of the concept of party identification. Along these lines, the hostility hypothesis poses that the likelihood of defection from partisan identification is inversely related to the degree of hostility felt toward the opposition party. Maggiotto and Piereson find evidence of a stable, durable, and independent effect of negative partisan attitudes on vote choice. Party defection is most unlikely when evaluations of the opposing party are negative, and most likely, when they are positive. Most importantly, however, by demonstrating the independent impact of negative party evaluations on vote choice, Maggiotto and Piereson have sown the seeds for the negative partisanship literature to come years later.

As a common denominator, this literature moves from the social-psychological notion that that hostility toward the out-group can independently drive support for the in-group. Utilizing parallelism, the theory of negative partisanship implies that voters do not need to identify with one party (i.e. the in-group) to feel aversion toward the opposing party (i.e. the out-group). Zhong et al. (2008: 794) refer to this type of identity as negational, outlining that “negational categorization refers to the process by which an individual’s identity is defined by outgroups, by what people are not.”

The first studies on negative partisanship emerged in the 2010s, with a few comparative and single-country case studies of both parliamentary and presidential democracies. Despite some ambiguity regarding the conceptual definition of negative partisanship, most studies agree with a minimalist approach, according to which “negative partisanship toward a party is an affective repulsion from that party, one that is more stable than a current dislike and more strongly held than a passing opinion” (Caruana et al., 2015: 772). Indeed, numerous studies have repeatedly demonstrated, through different methodological strategies, that partisans’ feelings about their own and the opposing party are largely independent of one another (Abramowitz and Webster, 2016; Bankert, 2020; McGregor et al., 2015).

If negative identifications are to be conceived differently from mere bipolar opposites of positive ones, they can be thought to have distinct antecedents and consequences. Concerning its causes, the literature on negative partisanship has often turned to social identity theory when attempting to explain its rise in contemporary democracies. The social identity background stands out the most in studies focusing on the American context, frequently exploring the relationship between recent trends in American democracy reinforcing group belongings, such as racial resentment and partisan sorting (Abramowitz and Webster, 2016, 2018). While group identity has been demonstrated to play an important role in the development of negative partisanship in the United States, comparative studies of multi-party systems tend to explain negative partisanship mostly through the lenses of negative ideological identities. For example, Medeiros and Nöel (2014) found that negative partisanship is more rooted in instrumental reasoning with strong anchors in ideology rather than group identity. McGregor et al. (2015) and Mayer (2017) also found that ideological identities strongly influence negative partisanship.

When it comes to its electoral consequences, it has been suggested that “the dark side of partisanship may have a power all its own” (Caruana et al., 2015: 775). Transversal to almost all studies analyzed is the finding that negative partisanship has an effect on turnout (Caruana et al., 2015; Mayer, 2017) and party choice (Bankert, 2020; Caruana et al., 2015; Mayer, 2017; Medeiros and Nöel, 2014). The relevance of these findings is amplified by the magnitude and pervasiveness of negative partisanship even in political systems characterized by relatively low levels of polarization. For example, McGregor et al. (2015) find that about half of the Canadian voting population holds some sort of negative partisanship. Also in Canada, Caruana et al. (2015) demonstrate that negative partisanship is somewhat diffused across the political spectrum, not corresponding to disgust directed at a specific party. Importantly, Abramowitz and Webster (2016, 2018) show that negative partisanship is on the rise and has reached unprecedented levels in recent US presidential elections. Regrettably, no longitudinal analysis of negative partisanship is available beyond the US case so, even if one of the main features of party identification relates to its stability, we still do not have a clear picture of how stable negative partisanship actually is.

Negative Personalization

Following the theoretical coordinates defined by the negative partisanship literature, an emerging strand of electoral research contends that negative attitudes toward out-group political actors concern not only political parties, through negative partisanship, but can also spill over to other actors such as political leaders (Barisione, 2017). If that is the case, evaluations of (out-party) political leaders may also act as a significant determinant of the vote, acting alongside positive (in-party) leader evaluations. Such a contention is supported by recent research on affective polarization, demonstrating that “when people think about the other party, they think primarily about political elites” (Druckman and Levendusky, 2019: 115; see also Kingzette, 2020). Along these lines, a recent study examining the electoral consequences of political sectarianism in the US 2020 Presidential election suggests that comparative candidate assessments associate more strongly with voting behavior than affection toward the respective parties (Garzia and Silva, 2021a).

The negative partisanship literature itself has sometimes made the argument that negative attitudes toward political leaders may represent a parallel dimension to partisan negativity. For example, Abramowitz and McCoy (2019) measure negative partisanship through feeling thermometers of candidates, assuming that negative partisanship could imply negativity toward candidates as well, in a reversed funnel causality logic. Again, putting negative attitudes toward out-party candidates under the umbrella of negative partisanship, Abramowitz and Webster (2018: 132) find that “the most important factor in predicting partisan loyalty is how an individual feels about the opposing party’s presidential candidate.” These studies seem to suggest that negativity can also be strongly directed toward political leaders, associating negative partisanship and negative attitudes toward candidates.

Several studies have analyzed negativity in the evaluation of candidates (Bor and Lautsen, 2020; Christenson and Weisberg, 2019; Klein and Ahluwalia, 2005). However, only a handful of studies has explored in a systematic fashion how negative attitudes toward political leaders may affect voting behavior. Aarts and Blais (2011) measure the relative impact of positive and negative thermometer evaluations of party leaders on vote choice in a total of 44 elections in multiple countries, covering circa 40 years. Although they find no evidence of a negativity bias in leader effects on vote choice, negative evaluations of political leaders still exert a significant effect (even if smaller than positive evaluations) on the vote in most elections analyzed. In their comparative analysis of the impact of leaders’ personality traits on vote choice, Ohr and Oscarsson (2011: 204) note that “a trait seems more important the less it is seen as being part of a candidate image.” In this respect, Klein (1996) finds an aggregate-level relationship between below-average trait scores and vote intentions for US presidential elections. Soroka (2014) examines the impact of asymmetric trait assessments in incumbency votes for the US presidential and Australian parliamentary elections. Across both countries, the impact of negative trait evaluations generally appears higher than that of positive trait evaluations. This asymmetric effect also holds when all trait evaluations are subsumed into combined negative versus combined positive trait measures.

Recently, Garzia and Silva (2021b) offered a comparative, longitudinal assessment of the relationship between negativity toward party leaders and vote choice. Their “negative personalization” hypothesis refers to the increasing tendency for voters’ party choice to be shaped by their negative evaluations of out-party leaders. Based on how negative out-party evaluations have been shown to affect party choice, they test the expectation that negative out-party leader evaluations also matter for vote choice, net of the effect of positive party and leader evaluations. Besides confirming the existence of a robust effect of negative party-leader evaluations on vote choice, their analysis of 109 elections held in West European parliamentary democracies in the period 1961–2018 highlights an increasing negative personalization of party choice across time.

Conclusions and Avenues for Further Research

Notwithstanding the merits of the contributions reviewed in this article, this strand of literature remains noticeably underdeveloped. More importantly, current research lacks an integrated approach combining the contributions of studies on negative partisanship and negative personalization in one, unified theoretical framework for negative voting. By pointing to the links between its several components, this review article suggests that upcoming analyses of negative voting consider performance assessments (rational or instrument component), long-standing ideological identities (ideological component), as well as negative attitudes toward political objects, translated into negative partisanship and negative personalization (affective component) in their explanatory models. With this conceptual map at hand, the exploration of the substantive impact of these factors on negative voting becomes the most pressing avenue for further research emerging from this review. In particular, it would be worth assessing the respective extent to which negative partisanship and negative personalization empirically relate to negative voting, as they represent the most novel aspect in relation to the seminal negative voting studies.

As affective polarization intensifies and spreads to multi-party parliamentary democracies, while negative campaign strategies become more intensively used, negative personalization and negative partisanship may become even more common, and negative voting will likely become increasingly salient in explanations of patterns of voting behavior in contemporary Western democracies. As argued by Sigelman and Gant (1989: 91), the fact that “many vote against, rather than for, is a fact of political life [. . .] so basic that students of voting behavior can no longer afford to ignore it.” It is also worth noting that negative voting has been seldom tested in a comparative perspective. Despite studies on negative partisanship having shown that (negative) ideological identities are an important factor in explaining the direction of negativity in multi-party systems (Caruana et al., 2015; Jost et al., 2009; Mayer, 2017; Van der Meer et al., 2015), the existing literature on negative voting is underdeveloped, at best, and for multi-party systems, no longitudinal studies exist at all.

From a theoretical point of view, an updated conceptualization of negative voting is needed in this sense, because in multi-party systems, the choice is not binary—voters rejecting one party have more than one alternative. In such cases, predicting the direction of negative voting would require a more complex explanatory framework and a comparative approach to rule out the possibility that negativity is mechanically driven by competition that is bipartisan (or bipolar) in nature. Such a framework, in turn, can enhance our capacity to answer the crucial question of whether negative voting is, in fact, inherently electorally potent in multi-party systems. The negative partisanship literature has measured the concept in multi-party systems through a simplification of choice-sets (Caruana et al., 2015; Mayer, 2017; Medeiros and Nöel, 2014), a sound strategy but not entirely without problems since in some cases it may involve an artificial reduction of the political supply.

The omission in existing analyses extends beyond the absence of a comparative perspective to include an almost exclusive focus on a single dependent variable of interest, that is, party choice. However, there are strong normative and empirical grounds to justify equal attention to the decision to participate in elections in the first place. Anecdotic accounts in this respect are divided between those who contend that negativity sours many citizens toward politics, leading to a significant drop in voter turnout (Ansolabehere and Iyengar, 1995) and those who argue that the effect on turnout is curvilinear, with most observed levels of negativity actually stimulating turnout (Lau and Pomper, 2001). A definitive answer to this question would undoubtedly be a valuable contribution to the long-standing debate on turnout decline in advanced post-industrial democracies when it comes to its normative implications. According to Levendusky (2013: 576), “polarized politics draws more citizens into the political arena . . . Such increased participation, however, need not be a normative good.” If negativity is indeed driving participation, should we instead be concerned with the potential reversal of the turnout decline trend due to the “dark side of civic engagement”? (Fiorina, 1999)

From the point of view of political communication, no study to date has explored the potential connection between varying patterns of exposure to political information in the media and negative voting. If we are right in hypothesizing citizens’ responsiveness to political communication, then individual patterns of media consumption should be taken into account in models of negative voting. According to Ohr and Oscarsson (2011: 13):

political broadcasting in a competitive media system increasingly follows news values such as conflict, negativism, and personalization. If there is in fact a change in journalists’ coverage of politics and politicians towards negativism and conflict [. . .] it may have an effect on the public perception of parties and political leaders.

Soroka (2014) has demonstrated that there is a negative bias in reported media content and that also individuals tend to subsequently choose negative over positive media content. Moreover, different media could affect public perceptions differently by prompting different degrees of negativity toward parties, vis-à-vis candidates, and party leaders. Tackling these issues appears increasingly urgent in today’s hybrid media system in which traditional media sources have been partly replaced by relatively more negative sources of information such as partisan channels and social media.

In turn, micro-level research into the socio-political causes and electoral consequences of negative voting could engage with the issue of increasing polarization of democratic politics. While political elites are recurrently blamed for increasing mass polarization, such research could offer a complimentary, bottom-up explanation whereby voters’ negative views of opponents may support further confrontational tendencies among the elites (Abramowitz and Webster, 2016). The authors of early works already raised some important implications premonitory of more recent political developments in American democracy. Given the potential of negative attitudes in guiding voting behavior, they suggest that it may provide an incentive to “polarize the polity by making compromise with the opposition unthinkable” (Maggiotto and Piereson, 1977: 766).

Along these lines, further analyses of negative voting could also address the relationship between changes in campaign strategies, negativity and the upsurge of anti-establishment populist parties and leaders across the Western world. The increased professionalization of politics is argued to have created “a fertile breeding ground for ‘us’ versus ‘them’ populist stealth angst” (Stoker and Hay, 2017: 7). In this sense, we concur with Mudde and Kaltwasser (2018: 1685) that “future studies will have to entangle the close, but undoubtedly complex, relationship between negative partisanship and populist support in both two-party and multiparty systems,” but contend that such framework should be widened as to encompass what we describe as a broader phenomenon of negative voting.

Gaining a better understanding of negative voting is all the more important considering its potential consequences for democracy and for responsible government. When despise for the alternatives emerges as the single most important factor in electing a ruler, competence-related leadership qualities become secondary. The relevance of political leadership tends to be proportionate to the demands of the socio-political momentum. In ordinary times, unqualified leaders may get by relatively unscathed, but the consequences of poor leadership in times of crisis could hardly become more evident than it is in the 2020s. In this respect, the election of political figures like Donald Trump and Jair Bolsonaro illuminates the importance of considering the normative consequences of this increasingly pervasive phenomenon in Western democracies.