‚That’s Not Appropriate!‘ Examining Social Norms as Predictors of Negative Campaigning
- Open Access
- 18-07-2024
- Original Paper
Published in:
Activate our intelligent search to find suitable subject content or patents.
Select sections of text to find matching patents with Artificial Intelligence. powered by
Select sections of text to find additional relevant content using AI-assisted search. powered by (Link opens in a new window)
Abstract
The use of negative campaigning, i.e., “any criticism leveled by one candidate against another during a campaign” (Geer, 2006, p. 23), has substantially increased over the last two decades (Geer, 2012; Klinger et al., 2023). Negative campaigning has been associated with harmful consequences for modern democracies including voters’ disengagement from political processes and institutions (e.g., Ansolabehere et al., 1994; Ansolabehere & Iyengar, 1995) and more recently, rising polarization (Banks et al., 2021; Sood & Iyengar, 2016). Negative campaigning is thus increasingly attracting the attention of social scientists (Haselmayer et al., 2019), who theoretically reflect and empirically test the determinants driving its use.
Many researchers of negative campaigning agree that candidates attack based on rational considerations (e.g., Benoit, 2007; Walter & Nai, 2015). The more perceived benefits (e.g., damaging the opponent’s reputation) exceed perceived costs (e.g., backlash effects), the more likely a candidate is to go negative. This intuitive assumption explains a candidate’s behavior only to some extent. Surveys testing the explanatory power of a candidate’s rational considerations have shown that some candidates attacked their opponents despite a negative benefit-cost ratio; others refrained from attacking even if they expected to increase their own favorability at the expense of their opponents (Maier et al., 2023). This points to further factors that influence the decision to go negative. Previous research explains differences among candidates in the frequency and intensity of negative campaigning by differences in the candidate’s political profile. For example, incumbents have consistently been found to use negative campaigning less than members of the opposition (Nai, 2020). Candidates running for ideologically extreme parties use negative campaigning more often than moderate candidates (Elmelund-Præstekær, 2010). More recent studies also associate a candidate’s positive (Big Five) and negative (Dark Triad) personality traits as mechanisms driving a candidate’s use of negative campaigning beyond rational considerations (Nai, 2019; Nai & Martínez i Coma, 2019).
Advertisement
This study suggests that like rational considerations, perceived social norms drive a candidate’s use of negative campaigning during election campaigns. Social norms are informal rules that guide the behavior of members of a group (Cialdini & Trost, 1998). Perceptions of such social norms, that is, perceptions of what other group members do and approve of, create social pressure on what behavior (not) to comply with. Social norms research differentiates between descriptive and injunctive norms (Cialdini et al., 1990). Descriptive norms refer to the perceived prevalence of a behavior (Cialdini et al., 1990; Rimal & Real, 2005) indicating an effective action (Cialdini et al., 1990). Injunctive norms refer to the perception of others’ approval of a behavior and, thus, indicate whether approval or social sanctions need to be expected (Cialdini et al., 1990; Rimal & Real, 2005; Shulman et al., 2017). Social norms and the distinction between descriptive and injunctive norms are considered behaviorally relevant in established behavioral theories, such as the theory of planned behavior (Ajzen & Fishbein, 2005). Moreover, social norms have also proven to be a fruitful approach for explaining political engagement, such as voting behavior (Hansen & Tyner, 2021) and policy support (Rinscheid et al., 2021).
We apply the concept of social norms to political campaigning and assume that perceptions of descriptive and injunctive campaigning norms, in addition to rational considerations, affect a candidate’s strategic campaign communication for two main reasons. First, campaigning is inherently a public behavior. Public behavior is more likely than nonpublic behavior to be influenced by social norms, as is observable by others (Rimal et al., 2011). Second, the success of negative campaigning depends on the acceptance of others (such as a candidate’s voters) and the behavioral reactions of others (such as a candidate’s political opponents who reward the behavior with social approval and appreciation or sanction it). Thus, the greater the perceived prevalence of negative campaigning (descriptive norms) and the greater the perceived social approval of the behavior (injunctive norms), the more likely a candidate is to engage in negative campaigning because it is perceived as effective and appropriate behavior (Chung & Rimal, 2016; Cialdini et al., 1990; Shulman et al., 2017). Therefore, in this study, we aim to answer the overarching RQ: To what extent do social norms of negative campaigning, i.e., the perception of relevant others’ behavior and approval with respect to negative campaigning, predict a candidate’s use of negative campaigning?
To answer this research question, we build on the theory of normative social behavior (TNSB, Rimal & Real, 2005) and current social norms research to test the direct and indirect effects of social norms on the use of negative campaigning (Chung & Rimal, 2016; Geber & Hefner, 2019; Rimal & Lapinski, 2015). We conducted a preregistered postelection survey among all candidates of the six major German parties (SPD, CDU, Gruene, FDP, AfD, and Die Linke) who ran for office in five subsequent German state elections in 2022 and 2023 in Lower Saxony, Berlin, Bremen, Bavaria, and Hesse (N = 1.087).
Who Sets the Tone for Negative Campaigning?
Individuals align their behavior with the social norms of their most important reference groups to “do the right thing,” in the sense of acting effectively (the descriptive norms-based motivation; Rimal & Lapinski, 2015, p. 397), and to not be aberrant (the injunctive norms-based motivation; see also Cialdini et al., 1990). It is important to note that people are embedded in various reference groups and that social norms can differ across these groups (Geber & Sedlander, 2022). Against this background, the use of negative campaigning presents an interesting case because three different reference groups might be considered in a candidate’s decision whether to go negative: (1) candidates of other parties, (2) members of a candidate’s own party, and (3) potential voters.
Advertisement
Candidates of other parties can be the sponsor and target of a candidate’s negative campaigning. Previous research has shown that another candidate’s attack behavior can trigger negative campaigning designed to retaliate (Dolezal et al., 2016) or to maintain structural balance in multiparty systems (Song et al., 2019). In contrast to attacking a political opponent in a two-party system—from which most research on negative campaigning originates—attacking a political opponent in a multiparty system is not necessarily a zero-sum game since an uninvolved third party can benefit (Song et al., 2019; Walter, 2014). Furthermore, since, unlike in two-party systems, ‘the winner does not take it all,’ candidates need to consider whether the target is a potential future coalition partner. Leaving scorched earth behind is not conducive to future cooperation after the election is decided (Haselmayer & Jenny, 2018). Hence, other parties and their norms of negative campaigning (i.e., descriptive and injunctive) are supposedly important factors in the context of negative campaigning.
Furthermore, members of a politician’s own party might affect a candidate’s campaign behavior. Drawing on social identity theory (Tajfel & Turner, 1979), studies found that party affiliation creates a sense of belonging to ‘us’, not them. Membership can thus stimulate “individuals’ aspirations to emulate referent others” (Rimal & Real, 2005, p. 395). Consequently, politicians might copy the behavior of other party members or conform to the behavior expected from them to generate shared positive emotions. Consistent with this line of reasoning, the use of negative campaigning has been found to differ across parties. For instance, negative campaigning is more widespread among US Republicans than among Democrats (Bekafigo & Ellis, 2017; Benoit, 2007; Stromer-Galley et al., 2018) and among Eurosceptic parties than among Europhile parties (Nai et al., 2022). Overall, this theoretical and empirical evidence suggests that a candidate’s own party and its negative campaigning norms (i.e., descriptive and injunctive) influence a candidate’s campaigning behavior.
Ultimately, a candidate’s voters decide the success or failure of a candidate’s campaign strategy. Candidates use negative campaigning to mobilize their base and to convince undecided voters (e.g., Stuckelberger, 2021). However, the effectiveness of such a strategy is still unclear (Krupnikov, 2011; Nai, 2020). While some studies argue that mobilization effects have the upper hand (e.g., Ceron & d’Adda, 2016; Martin, 2004), others show substantial backlash effects on the sponsor of an attack as voters who dislike negative campaigning withdraw their support (Galasso et al., 2023; Pattie et al., 2011; Walter & van der Eijk, 2019). Thus, it is likely that the voters and their norms concerning campaigning are crucial to a candidate’s decision to engage in negative campaigning. Because voters obviously do not participate in campaigning themselves but hold attitudes toward negative campaigning (e.g., Pew Research Center, 2019), their injunctive norms (and not their descriptive norms) are supposedly influential.
A contrasting observation suggests that these three reference groups differ in their respective social norms and, hence, differentially affect a candidate’s decision to go negative. While we see a general increase in the use of negative campaigning among candidates (Geer, 2012; Klinger et al., 2023), there is a broad consensus that a negative tone is predominantly disliked by voters (e.g., Fridkin & Kenney, 2011; but, also see Nai & Maier, 2021 for opposing results). Considering the distinction between descriptive norms (H1) and injunctive norms (H2) and the potential reference groups for an individual candidate, we derive the following two hypotheses:
H1
The greater the perceived use of negative campaigning (descriptive norms) among (a) candidates of other parties and (b) candidates of a politician’s own party, the more frequently candidates use negative campaigning.
H2
The more positive the perceived attitudes toward negative campaigning (injunctive norms) among (a) candidates of other parties, (b) members of a politician’s own party, and (c) the candidate’s potential voters are, the more frequently candidates use negative campaigning.
Moderators of the Impact of Social Norms on the Use of Negative Campaigning
Based on the social norms literature (e.g., Rimal & Lapinski, 2015; Rimal & Real, 2005), we propose that the influences of negative campaigning norms are moderated by outcome expectations (also see Geber & Hefner, 2019). Specifically, we choose three outcome expectations of negative campaigning that are likely to play a role in strategic campaign behavior: (1) benefits to oneself, (2) benefits gain by others, and (3) anticipatory socialization.
Perceived benefits to oneself have been found to moderate the impact of descriptive norms on behavioral intentions (Rimal et al., 2005). Drawing on social cognitive theory (Bandura, 1986) and subjective utility theory (Sutton, 1982), the theory of normative social behavior (Rimal & Real, 2005) postulates that the greater the perceived benefits are, the greater an individual’s willingness to engage in normative behavior. The literature on negative campaigning has proposed several reasons that drive a candidate’s decision to go negative; these reasons include attracting undecided voters (e.g., Nai, 2020) and receiving media attention (Haselmayer et al., 2019). Thus, we assume that the normative influence on a candidate’s tendency to go negative is accelerated by the magnitude of the benefits assumed for him-/herself.
H3
The effect of perceived (descriptive and injunctive) norms on the use of negative campaigning is moderated by candidates’ perceptions of the benefits of negative campaigning such that the effect of social norms on the use of negative campaigning increases with perceived benefits.
In addition, the perceived benefits of negative campaigning gained by others might moderate normative influences. Rimal and Real (2005) argue that “individuals may become fearful that they will be denied important outcomes that others who engage in the behavior are able to attain (p. 394).” Based on prospect theory (Kahneman & Tversky, 1979), they assume that individuals become risk averse against the backdrop of an increasingly perceived descriptive norm. Consequently, in the context of negative campaigning, candidates increasingly engage in attacks the more they fear foregoing advantages that other candidates derive from going negative. We thus hypothesize that the normative influence is moderated by the fear of missing out on the benefits of negative campaigning:
H4
The effect of perceived (descriptive and injunctive) social norms on the use of negative campaigning is moderated by candidates’ fear of missing out on the benefits of negative campaigning such that the effect of social norms on the use of negative campaigning increases with the perceived fear of missing out on benefits.
Finally, a candidate might anticipate socializing with other party members when engaging in negative campaigning. Such behavior can then be understood as a “social lubricant” (Rimal & Real, 2005, p. 394) that allows the development and/or strengthening of bonds with like-minded others. Joining the attack of the top candidate on a political opponent might be well received by the candidate and demonstrates closing ranks to voters and other parties. In contrast, reluctance to do so might be perceived as less supportive and potentially affect social and political standing after the election campaign. Hence, we assume that candidates’ anticipated socialization through negative campaigning moderates the influence of negative campaigning norms.
H5
The effect of perceived (descriptive and injunctive) social norms on the use of negative campaigning is moderated by candidates’ anticipated socialization through negative campaigning such that the effect of social norms on the use of negative campaigning increases with perceived anticipated socialization.
Methods
We answered our research question with a survey among candidates of the six major German parties (Die Linke, SPD, Gruene, FDP, CDU/CSU, and AfD) who ran for office in five state elections in 2022 and 2023 in Lower Saxony (n = 220), Bremen (n = 81), Berlin (n = 144), Bavaria (n = 421) and Hesse (n = 221). The data were collected as part of a larger project funded by the German Research Foundation to study “Negative Campaigning in German Election Campaigns” (grant no: 441,574,527). Ethical approval was provided by the GESIS ethics committee (reference number 2020-6) for the entire project prior to data collection. Hypotheses and procedures for this subproject and manuscript were preregistered (https://osf.io/72bs8). While our preregistration stated that we intended to collect data solely for the election in Lower Saxony, we decided to continue the data collection in subsequent state elections to test the robustness of the findings and to increase the generalizability of our conclusions. Fully anonymized data will be shared via the GESIS data archive (https://www.gesis.org/en/home) at the end of the larger DFG project in accordance with the IRB regulations1.
Participants and Sampling Procedure
In total, N = 2,800 invitations were sent to candidates across the five state elections. The survey was conducted in mixed mode over a period of two months, starting one day after the elections (election days: Lower Saxony 9 October 2022; Berlin 12 February 2023; Bremen 14 May 2023; Bavaria and Hesse 8 October 2023). Candidates with a publicly available email address were invited with a personalized link to an online questionnaire. The remaining candidates were invited by regular mail and provided with a return envelope. In addition, these candidates were also provided a personalized link to the online questionnaire should they prefer to participate online. Two reminders were sent to those candidates who did not participate in the study to maximize response rates.
In total, N = 1,103 candidates completed the questionnaire (response rate = 39%). We removed 16 online participants based on quality criteria indicators embedded in the survey tool. First, we identified respondents who rushed through the survey with an adapted “relative speed index” that also considers open questions based on Leiner (2019). We excluded 11 respondents with a value of 2.0 or higher. Moreover, five participants with substantial missing data (> 60%) were excluded. The final sample consists of N = 1,087 candidates across the political spectrum (nLinke= 158, nSPD= 221, nGruene= 225, nFDP= 172, nCDU/CSU= 211, nAfD= 100). The candidates were aged 18 to 86 (M = 45.93, SD = 13.59), and 42% were females. Table 1 summarizes the participation across state elections and parties. Table S1 in the OSF appendix provides further information on the electoral systems and government composition during the election campaign.
Table 1
Candidates participating across state elections and parties from left to right
Party | |||||||
|---|---|---|---|---|---|---|---|
State | Linke n (%) | SPD n (%) | Gruene n (%) | FDP n (%) | CDU/CSU n (%) | AfD n (%) | Total n (100%) |
Lower-Saxony | 27 (12) | 37 (17) | 45 (21) | 36 (16) | 54 (25) | 21 (10) | 220 |
Bremen | 12 (15) | 27 (33) | 17 (21) | 8 (10) | 17 (21) | 0 (0) | 81 |
Berlin | 22 (15) | 21 (15) | 33 (23) | 23 (16) | 25 (17) | 20 (14) | 144 |
Bavaria | 64 (15) | 67 (16) | 102 (24) | 79 (19) | 70 (17) | 39 (9) | 421 |
Hesse | 33 (15) | 69 (31) | 28 (13) | 26 (12) | 45 (20) | 20 (9) | 221 |
N | 158 | 221 | 225 | 172 | 211 | 100 | 1.087 |
Measures
Negative Campaigning (Dependent Variable)
The use of negative campaigning was measured with one question on a 5-point Likert scale: “How often have you attacked your political opponents, i.e., criticized other parties or candidates?” (1 = never, 5 = very often) (M = 2.85, SD = 0.98). Please see the item overview in the online appendix on OSF for all item wordings and their English translation (https://osf.io/4umrn/files/osfstorage). To confirm the validity of the measurement, we correlated the self-reported attack behavior of the candidates with two external data sources. First, we calculated the correlation of self-reported attack behavior with available expert assessments of the candidates’ use of negative campaigning in Lower Saxony, Berlin, and Bavaria (for this approach, see Maier & Nai, 2020).2 Self-reports and expert ratings highly correlated, thereby validating our survey results, r = .71, p = .001 (see Fig. 1).
Fig. 1
Correlation between candidates’ self-reported use of negative campaigning (aggregated by election and party) and expert ratings of parties’ negative campaigning. Note: Candidate self-reports: The scale is from 1 (“never attacked the political opponent”) to 5 (“very often attacked the political opponent”). Expert ratings: “When considering the electoral campaigns of each of the following actors during the most recent state election, would you say that the actor’s campaign was exclusively negative, exclusively positive or somewhere in between? Please provide a score between − 10 (‘exclusively negative’) and 10 (‘exclusively positive’)” (reverse coded). Expert data are available for two of the three elections, with the participation of 4 experts for Lower Saxony (LS), 4 experts for Berlin (Ber), and 17 experts for Bavaria (Bay)
Second, we compared the survey data with the actual attacks that candidates posted on their Facebook accounts during the campaign. The data were collected as part of the larger project (for this approach, see Maier et al., 2023). In line with the expert ratings (see Fig. 2), self-reports and observed attack behavior on Facebook correlated significantly with our survey results (r = .67, p < .001).
Rational Choice
To assess cost‒benefit considerations, we asked the candidates on a 5-point Likert scale “What would you say, all things considered? Have attacks on political opponents produced 1–only disadvantages for your own election campaign, 5–only advantages for your own election campaign” (M = 3.08, SD = 0.76).
Fig. 2
Correlation between candidates’ self-reported use of negative campaigning (aggregated by election and party) and attack behavior on Facebook. Note: Candidate self-reports: scale from 1 (“never attacked the political opponent”) to 5 (“very often attacked the political opponent”). Facebook: Candidates were informed in the informed consent stage before the candidate survey that the project would link their survey responses to their social media posts. Candidates’ Facebook accounts were researched by student assistants, and the candidates’ posts were collected via CrowdTangle. A total of 44,000 posts were coded by a team of five trained student assistants for the presence of attacks, with an interrater agreement of Krippendorff’s alpha = 0.88. Based on the coded social media posts, a machine learning model using word embeddings was trained. The model could classify all social media posts as containing an attack or not with an F1 score of 0.685. The plot is based on the model predictions for the N = 316 candidates who participated in the survey and for whom we could retrieve at least one Facebook post in the last ten weeks prior to the election. The analysis includes all five elections covered in this article (Lower Saxony (LS), Berlin (Ber), Bremen (Bre), Bavaria (Bay), and Hesse (He))
Social Norms of Negative Campaigning
Descriptive and injunctive norms of negative campaigning were measured in an item battery on a 5-point Likert scale: “We are interested in your assessment of how attacks on political opponents are viewed and evaluated in your party, among your political opponents, and among your voters. Please indicate the extent to which you agree or disagree with the following statements”: (1 = strongly disagree, 5 = fully agree). For descriptive norms, we provided statements on the perceived prevalence of negative campaigning among candidates of other parties (“Most candidates of other parties use attacks on political opponents in election campaigns” (M = 3.56, SD = 0.98)) and members of their own party (“Most candidates of my party use attacks on political opponents in election campaigns” (M = 2.88, SD = 1.09)).
To assess injunctive norms, we provided statements on the perceived approval of negative campaigning and considered voters as a relevant reference group (“My voters find that attacks on political opponents are an appropriate strategy in election campaigns” (M = 3.71, SD = 0.90)). Additionally, we provided statements on the perceived approval of negative campaigning among candidates of other parties (“Candidates of other parties find that attacks on political opponents are an appropriate strategy in election campaigns” (M = 3.06, SD = 1.10)) and members of the candidate’s party (“Members of my party find that attacks on political opponents are an appropriate strategy in election campaigns” (M = 2.80, SD = 1.05)).
Outcome Expectations (Moderators)
Benefits to oneself were measured on a separate 5-point Likert scale that asked “To what extent would you consider the following aspects to be distinct advantages?” (1 = strongly disagree, 5 = fully agree). Based on previous research on negative campaigning, we asked for the candidate’s perception of six potential benefits: (1) “You can damage the image of your political opponent,” (2) “You can put yourself in a better light,” (3) “You can emphasize programmatic and personal differences from your opponent,” (4) “You can attract the attention of the media,” (5) “You can mobilize your own voters,” and (6) “You can win over undecided voters.” An explorative factor analysis (PFA, rotation: promax) revealed a two-factor structure (Table 2). Items (1), (2), and (4) loaded on the first factor, which was named image building. Items (3), (5), and (6) loaded on the second factor, which was named vote acquisition (KMO = 0.68).
Table 2
Results of the principal axis factor analysis (PFA) for perceived benefits for oneself: factor loadings for image building and vote acquisition
Image building | Vote acquisition | |
|---|---|---|
You can damage the image of your political opponent | 0.59 | -0.16 |
You can put yourself in a better light | 0.51 | 0.14 |
You can attract the attention of the media | 0.49 | 0.07 |
You can emphasize programmatic and personal differences from your opponent | -0.03 | 0.53 |
You can mobilize your own voters | 0.02 | 0.74 |
You can win over undecided voters | -0.01 | 0.74 |
Fear of missing out on benefits gained by others and anticipatory socialization were assessed for the reference groups. The fear of missing out on benefits gained by others was measured with three items on a 5-point Likert scale (1 = strongly disagree, 5 = fully agree): (1) “If I refrain from attacks on my political opponents, I appear weak to candidates of other parties” (M = 2.48, SD = 1.15); (2) “If I refrain from attacks on my political opponents, I appear weak to members of my party” (M = 2.64, SD = 1.22); and (3) “If I refrain from attacks on my political opponents, I appear weak to my voters” (M = 2.45, SD = 1.11).
Anticipatory socialization of negative campaigning has also been measured with three items: (1) “Attacks on political opponents gain me respect from candidates of other parties” (M = 2.31, SD = 1.02), (2) “Attacks on political opponents gain me respect from members of my party” (M = 3.01, SD = 1.12), and (3) “Attacks on political opponents gain me respect among my voters” (M = 2.69, SD = 1.03).
Control Variables
Research on negative campaigning has identified several variables of a candidate’s personal and political profile that affect negative campaign behavior. We accounted for these variables in our analysis to determine the explanatory contribution of social norms. More specifically, we included (1) the candidate’s gender (0 = male, 1 = female), (2) incumbency (whether the candidate campaigned as an active member of the parliament; 0 = no, 1 = yes), (3) ideology (on an 11-point Likert scale ranging from 1 = left to 11 = right, M = 4.95; SD = 2.46), (4) governmental status (whether the candidate ran for a party represented in the government; 0 = no, 1 = yes), and (5) extremism (measured by folding the ideology scale on itself, M = 2.21; SD = 1.49). Furthermore, we controlled for a candidate’s respective party and state to account for structural differences in negative campaigning.
Analysis Plan
We used descriptive statistics to present the candidates’ reported use of negative campaigning and to examine whether the candidates perceived different descriptive and injunctive norms from the respective reference groups. We used regression models to determine to what extent the perceived social norms of the different reference groups determine the candidates’ decision to go negative beyond the candidates’ rational considerations. In the final step, we tested the assumed moderation effects by adding the respective interaction terms to the regression model.
Results
The vast majority of candidates reported using negative campaigning (M = 2.85, SD = 0.98). Only 7% fully refrained from it. Most candidates (68%) rarely or sometimes attacked their opponents, while approximately 25% often or very often made use of this strategy. Overall, center-left parties were the least likely to engage in negative campaigning, while we observed an increase in its use by candidates toward the extremes of the political spectrum (Fig. 3). In the following, we examine to what extent social norms account for the differences in the use of negative campaigning across the major parties in German state elections.
Fig. 3
Use of negative campaigning during the election campaign across parties. Note: N = 926
Descriptive and Injunctive Norms as Predictors of the Use of Negative Campaigning
Our first two hypotheses referred to the effects of descriptive norms (H1) and injunctive norms (H2) on a candidate’s use of negative campaigning. Overall, candidates perceived greater descriptive norms (that is, more frequent use of negative campaigning) among members of other parties (M = 3.56, SD = 0.98) than among members of their own party (M = 2.88, SD = 1.09). This perceived pattern is particularly pronounced for the center-left SPD and Gruene, whereas the extreme right AfD and extreme left Die Linke do not perceive much difference between their own and other parties’ members (Table 3).
Table 3
Perceived descriptive norm of negative campaigning among other parties and the candidate’s own party
Descriptive norms of other parties | Descriptive norms of own party | |||
|---|---|---|---|---|
Party | M | SD | M | SD |
Die Linke (n = 123) | 3.45 | 0.94 | 3.17 | 1.05 |
SPD (n = 165) | 3.45 | 1.02 | 2.66 | 1.05 |
Gruene (n = 174) | 3.70 | 0.95 | 2.37 | 0.93 |
FDP (n = 143) | 3.52 | 0.98 | 2.94 | 1.11 |
CDU (n = 152) | 3.44 | 0.95 | 3.03 | 1.02 |
AfD (n = 84) | 3.96 | 1.00 | 3.61 | 1.05 |
Mean | 3.56 | 0.98 | 2.88 | 1.09 |
A similar pattern is observable for perceived injunctive norms toward negative campaigning. Candidates perceive attitudes towards negative campaigning among members of other parties to be more positive (M = 3.71, SD = 0.90) than those among members of their own party (M = 3.06, SD = 1.01). Across all reference groups, voters were perceived to have the least positive attitude toward negative campaigning (M = 2.80, SD = 1.05) (Table 4).
Table 4
Perceived injunctive norms of negative campaigning by other parties, the candidate’s own party, and potential voters
Injunctive norms of other parties | Injunctive norms of own party | Injunctive norms of potential voters | ||||
|---|---|---|---|---|---|---|
Party | M | SD | M | SD | M | SD |
Die Linke (n = 123) | 3.61 | 0.94 | 3.48 | 1.00 | 3.04 | 1.07 |
SPD (n = 165) | 3.67 | 0.83 | 2.90 | 1.07 | 2.50 | 0.94 |
Gruene (n = 174) | 3.91 | 0.88 | 2.49 | 1.00 | 2.49 | 0.94 |
FDP (n = 143) | 3.61 | 0.88 | 3.05 | 1.07 | 2.88 | 1.03 |
CDU (n = 152) | 3.54 | 0.86 | 3.19 | 0.99 | 2.81 | 1.00 |
AfD (n = 84) | 4.00 | 0.96 | 3.69 | 1.13 | 3.51 | 1.12 |
Mean | 3.71 | 0.90 | 3.06 | 1.01 | 2.80 | 1.05 |
Next, we conducted a multiple linear regression with the candidate’s reported use of negative campaigning during the election campaign as the dependent variable and the candidate’s rational considerations and the perceived descriptive and injunctive norms of the respective reference groups as independent variables. We further controlled for gender, incumbency, ideology, governmental status, extremism, party, and state. The regression model lends support for our assumption that, in addition to a candidate’s rational considerations, social norms function as a strong distinct mechanism that drives the use of negative campaigning by a candidate (F(20, 757) = 18.19, p < .001, R²adj = 0.307, Table 5). Specifically, the data support H1b. Notwithstanding the highly perceived descriptive norm of negative campaigning by the other parties compared to a candidate’s own, the decisive norm for a candidate to go negative was the perceived descriptive norm of negative campaigning in his or her own party (β = 0.12, p = .008). The greater the perceived descriptive norms of negative campaigning among members of their own party, the more frequently the candidates used negative campaigning. The perceived frequency of other parties’ attack behavior did not affect the candidate’s negative campaigning (β = 0.04, p = .276); thus, H1a was not supported.
The results further show that the perceived injunctive norms of a candidate’s voters are relevant for the decision to engage in attacks during the election campaign (β = 0.23, p < .000), thereby confirming H2c. Voters’ injunctive norms emerged as the strongest predictor of a candidate’s decision to go negative. The more the candidates perceived that their voters approved of negative campaigning, the greater their use of this strategy. The injunctive norms of others (β = -0.05, p = .208) and own party members (β = -0.05, p = .291) did not emerge as predictors of a candidate’s use of negative campaigning. H2a and H2b were thus rejected. Our results are also in line with findings from a broad base of negative campaigning research. Candidates who ran for office as active members of parliament (incumbents) were more engaged in negative campaigning than nonactive members (β = 0.10, p = .001). Moreover, members of the governmental coalition were less likely than members of the opposition to go negative (β = -0.21, p < .000). Members of more extreme parties attacked their opponents more than did members of moderate parties (β = 0.11, p = .009).
Table 5
Joint model with rational considerations and perceived descriptive and injunctive norms as determinants of candidates’ use of negative campaigning
b | beta | SE | p | ||
|---|---|---|---|---|---|
Rational Choice | 0.27 | 0.21 | 0.04 | 0.000 | *** |
Descriptive Norms | |||||
Other parties | 0.04 | 0.04 | 0.04 | 0.276 | |
Own party | 0.11 | 0.12 | 0.04 | 0.008 | ** |
Injunctive norms | |||||
Other parties | -0.05 | -0.05 | 0.04 | 0.208 | |
Own party | -0.05 | -0.05 | 0.04 | 0.291 | |
My voters | 0.22 | 0.23 | 0.04 | 0.000 | *** |
Control variables | |||||
Gender | -0.09 | -0.04 | 0.06 | 0.161 | |
Incumbency | 0.31 | 0.10 | 0.09 | 0.001 | ** |
Ideology | -0.02 | -0.05 | 0.02 | 0.338 | |
Governmental status | -0.45 | -0.21 | 0.09 | 0.000 | *** |
Extremism | 0.07 | 0.11 | 0.02 | 0.009 | ** |
Party (AfD = 0) | |||||
CDU | -0.11 | -0.04 | 0.14 | 0.430 | |
FDP | -0.37 | -0.14 | 0.13 | 0.005 | ** |
SPD | -0.42 | -0.17 | 0.16 | 0.008 | ** |
Gruene | -0.30 | -0.12 | 0.16 | 0.065 | |
Die Linke | -0.29 | -0.10 | 0.19 | 0.132 | |
State (Lower Saxony = 0) | |||||
Bremen | 0.29 | 0.08 | 0.13 | 0.027 | * |
Berlin | 0.12 | 0.04 | 0.11 | 0.260 | |
Bavaria | 0.20 | 0.10 | 0.09 | 0.018 | * |
Hesse | 0.37 | 0.15 | 0.09 | 0.000 | *** |
Moderators of the Impact of Social Norms: Candidate’s Use of Negative Campaigning
We proposed three moderators of the relationship between social norms and the use of negative campaigning: perceived benefits of negative campaigning, fear of missing out on benefits gained by others, and anticipated socialization. To test the assumed moderation effects, we subsequently added the respective interaction terms to the joint regression model (= baseline model M0). Table 6 summarizes the results of all interaction terms.
Hypothesis H3 assumed that the effect of the perceived social norms of negative campaigning on the use of negative campaigning is moderated by candidates’ perceptions of the benefits of negative campaigning such that the effect of the social norms of negative campaigning increases with the perceived benefits. No interaction effects were observed for the benefits of negative campaigning (neither for factor image building nor vote acquisition) on the impact of descriptive and injunctive social norms on the candidate’s use of negative campaigning (see Table 6), rejecting Hypothesis H3.
A surprising pattern emerged for the candidates’ fear of missing out on the benefits of negative campaigning (H4). Adding the respective interaction terms subsequently to the baseline model revealed one significant interaction effect between perceived injunctive norms of members of a candidate’s own parties and the fear of missing out on benefits (β = -0.27, p = .029, F(22, 752) = 17.57, p < .001, R²adj = 0.320). Counterintuitively, the more positive the perceived attitudes of one’s own party members toward negative campaigning and the greater the fear of appearing weak to them if one refrains from doing so, the less often a candidate attacks a political opponent. Such an effect must be interpreted with care, as repeating multiple tests with the same data runs the risk of generating random significant results. However, adding the fear of missing out to our joint regression model resulted in the highest explained variance. We thus explored this pattern in greater depth. Following recommendations of the more recent literature (Hayes, 2022; Warner, 2013), we excluded the interaction terms from the regression and repeated the analysis by adding the fear of missing out to the three reference groups as direct predictors to the model. In this model (Table 7), the fear of appearing weak to one’s voters emerged as an additional relevant predictor of a candidate’s use of negative campaigning (β = 0.11, p = .044) and explained the observed variance in the use of negative campaigning to a greater degree (F(23, 748) = 16.56, p < .001, R²adj = 0.317).
Finally, Hypothesis H5 proposed that the effect of perceived (descriptive and injunctive) social norms on the use of negative campaigning is moderated by candidates’ anticipated socialization through negative campaigning such that the effect of social norms of negative campaigning will increase with perceived anticipated socialization. In the multiple regression models, a similar pattern was observed as in the previous analyses. Anticipated socialization with any of the three reference groups did not moderate the impact of social norms on a candidate’s use of negative campaigning (see Table 6).
Table 6
Moderators of perceived social norms on a candidate’s use of negative campaigning (interaction effects only, separate models)
Model | Interaction term | b | beta | SE | p | R² |
|---|---|---|---|---|---|---|
M0 | 0.307 | |||||
Perceived benefits to oneself as moderator | ||||||
M1 | Descriptive norms (other parties)*Benefits_image | -0.00 | -0.00 | 0.03 | 0.995 | 0.299 |
M2 | Descriptive norms (own party)*Benefits_image | -0.02 | -0.11 | 0.04 | 0.527 | 0.299 |
M3 | Injunctive norms (other parties)*Benefits_image | -0.03 | -0.15 | 0.04 | 0.454 | 0.299 |
M4 | Injunctive norms (own party)*Benefits_image | -0.05 | -0.25 | 0.03 | 0.137 | 0.301 |
M5 | Injunctive norms (own voters)*Benefits_image | 0.06 | 0.29 | 0.04 | 0.098 | 0.302 |
M6 | Descriptive norms (other parties)*Benefits_votes | 0.05 | 0.29 | 0.04 | 0.219 | 0.302 |
M7 | Descriptive norms (own party)*Benefits_votes | 0.01 | 0.07 | 0.04 | 0.752 | 0.301 |
M8 | Injunctive norms (other parties)*Benefits_votes | 0.02 | 0.08 | 0.05 | 0.740 | 0.301 |
M9 | Injunctive norms (own party)*Benefits_votes | -0.02 | -0.13 | 0.04 | 0.517 | 0.301 |
M10 | Injunctive norms (own voters)*Benefits_votes | 0.04 | 0.19 | 0.04 | 0.382 | 0.301 |
Fear of missing out on benefits as moderator | ||||||
M11 | Descriptive norms (other parties)*FoMoBenefits | -0.02 | -0.13 | 0.03 | 0.356 | 0.308 |
M12 | Injunctive norms (other parties)*FoMoBenefits | -0.04 | -0.24 | 0.03 | 0.119 | 0.310 |
M13 | Descriptive norms (own party)*FoMoBenefits | -0.02 | -0.12 | 0.02 | 0.320 | 0.316 |
M14 | Injunctive norms (own party)*FoMoBenefits | -0.05 | -0.27 | 0.02 | 0.029* | 0.320 |
M15 | Injunctive norms (own voters)*FoMoBenefits | -0.02 | -0.11 | 0.02 | 0.365 | 0.317 |
Anticipated socialization as moderator | ||||||
M16 | Descriptive norms (other parties)*AntSocialization | -0.00 | -0.01 | 0.03 | 0.928 | 0.307 |
M17 | Injunctive norms (other parties)* AntSocialization | 0.03 | -0.13 | 0.03 | 0.391 | 0.308 |
M18 | Descriptive norms (own party)* AntSocialization | -0.01 | -0.07 | 0.02 | 0.609 | 0.316 |
M19 | Injunctive norms (own party)* AntSocialization | -0.01 | -0.06 | 0.02 | 0.642 | 0.316 |
M20 | Injunctive norms (own voters)* AntSocialization | 0.02 | 0.08 | 0.03 | 0.544 | 0.319 |
Table 7
Joint model with rational considerations, descriptive norms, injunctive norms and fear of missing out on benefits as determinants of candidates’ use of negative campaigning
b | beta | SE | p | ||
|---|---|---|---|---|---|
Rational Choice | 0.24 | 0.19 | 0.04 | 0.000 | *** |
Descriptive Norms | |||||
Other parties | 0.04 | 0.04 | 0.04 | 0.311 | |
Own party | 0.10 | 0.11 | 0.04 | 0.020 | * |
Injunctive norms | |||||
Other parties | -0.05 | -0.06 | 0.04 | 0.163 | |
Own party | -0.05 | -0.05 | 0.04 | 0.259 | |
My voters | 0.18 | 0.19 | 0.04 | 0.000 | *** |
Fear of missing out on benefits | |||||
Other parties | -0.03 | -0.03 | 0.04 | 0.435 | |
Own party | 0.05 | 0.06 | 0.04 | 0.252 | |
My voters | 0.10 | 0.11 | 0.05 | 0.44 | * |
Control variables | |||||
Gender | -0.08 | -0.04 | 0.06 | 0.187 | |
Incumbency | 0.31 | 0.10 | 0.09 | 0.000 | *** |
Ideology | -0.02 | -0.07 | 0.02 | 0.256 | |
Governmental status | -0.41 | -0.19 | 0.09 | 0.000 | *** |
Extremism | 0.07 | 0.11 | 0.02 | 0.008 | ** |
Party (AfD = 0) | |||||
CDU | -0.12 | -0.05 | 0.14 | 0.389 | |
FDP | -0.34 | -0.13 | 0.13 | 0.009 | ** |
SPD | -0.43 | -0.17 | 0.16 | 0.006 | ** |
Gruene | -0.31 | -0.13 | 0.16 | 0.056 | |
Die Linke | -0.30 | -0.11 | 0.19 | 0.125 | |
State (Lower Saxony = 0) | |||||
Bremen | 0.28 | 0.07 | 0.13 | 0.032 | * |
Berlin | 0.13 | 0.04 | 0.11 | 0.254 | |
Bavaria | 0.20 | 0.10 | 0.09 | 0.018 | * |
Hesse | 0.35 | 0.14 | 0.10 | 0.000 | *** |
Discussion
This study investigated whether the perceived descriptive and injunctive norms of a candidate’s opponents, party members, or voters affect the candidate’s use of negative campaigning during an election campaign. To answer this research question, we conducted a preregistered postelection survey among candidates who ran for office for one of the six major German parties in five state elections in 2022 and 2023.
Most importantly, our results confirm our initial assumption that social norms, in addition to rational considerations, function as a mechanism that affects a candidate’s decision to go negative. Our first set of hypotheses assumed the main effects of a candidate’s most important reference group’s perceived descriptive and injunctive norms on the decision to go negative. The results show a remarkably clear picture. Although candidates of other parties are the essential sponsor and target of negative campaign messages, their behavior does not drive a candidate’s use of negative campaigning. The perceived frequency of using negative campaigning among one’s party members (= descriptive norms) and the perceived acceptance of negative campaigning among one’s voters (= injunctive norms) emerged as the only decisive predictors.
Our findings add to the limited but increasing literature that proves that studying the effect of social norms is a fruitful approach to explaining political behavior (e.g., Hansen & Tyner, 2021; Hassell & Wyler, 2019; Rinscheid et al., 2021). Negative campaigning is usually discussed as an individual campaign strategy, but candidates do not campaign in a vacuum. Social norms allow us to consider fine personal interactions with a candidate’s professional network. On the one hand, a candidate adapts his or her behavior to the perceived party line—as an agreed upon strategy or mutual understanding of how far one could go in election campaigns. On the other hand, a candidate might refrain from using negative campaigning in the first place because he or she is concerned that it could alienate potential voters.
The latter finding also adds insights to the heterogeneous discussion on the effectiveness of negative campaigning during election campaigns. While some authors have argued that negative campaigning is an effective strategy for mobilizing and convincing voters (e.g., Ceron & d’Adda, 2016), others have found that candidates who engage in negative campaigning suffer from substantial backlash effects, as voters dislike negative campaigning and withdraw their support for the candidate (e.g., Galasso et al., 2023; Walter & van der Eijk, 2019). Our results also show that voters’ perceptions of negative campaigning differ. We find that candidates of the far-right populist AfD perceive their voters’ injunctive norms on negative campaigning as far more positive than candidates of other parties perceive their voters’ injunctive norms. Consequently, the far-right candidates have more leeway to use negative campaigning and, potentially, to succeed in damaging their opponents and mobilizing their base. This finding also offers an explanation for why negative campaigning is more widespread among US Republicans than among Democrats (Bekafigo & Ellis, 2017; Benoit, 2007; Stromer-Galley et al., 2018). A previous study also found that the provocative tone of the right-wing populist AfD is particularly successful in gaining the media’s attention (Maurer et al., 2022), thereby bringing journalists’ responsibility into play. In line with this, other research indicates that voters’ attitudes toward negative campaigning depend on their personality (Nai & Maier, 2021). When voters themselves have a more confrontational personality, they like negative campaigning more and reward the sponsor of negative campaigning. One practical consequence of this is that it should pay off for candidates to target microsegments of the electorate with specially tailored campaign communications. Based on the current literature on the impact of social norms on individual behavior, we further examined three potential moderators that we deemed important for accelerating the impact of social norms on the use of negative campaigning: the perceived benefits of negative campaigning, the fear of missing out on benefits gained by others, and anticipated socialization. However, the data do not support the assumed moderation effects. Instead, we discovered a direct effect of the fear of missing out on benefits; i.e., the more a candidate appears weak when refraining from negative campaigning, the greater the candidate’s use of negative campaigning during an election campaign.
Limitations and Directions for Future Research
There are some limitations of the study; however, they also open potential for future studies. First, although our sample consists of a rarely surveyed target group, the sample remains limited to German state elections. Future studies should replicate this study in other multi- and dual-party electoral contexts to fully test the robustness of the impact of social norms. Moreover, replicating the study with other German state elections can prove fruitful. While our sample is based mainly on Berlin and four former West German states, the newer East German states still differ to some extent in terms of the socialization of political candidates and voter behavior; these differences might affect the role of social norms. For example, we see stronger support for parties at the extremes of the political spectrum. The support for the left party Die Linke can be historically explained as being the successor of the former GDR party SED. However, Die Linke is increasingly losing its core electorate, while we observe a stronger increase in support for right-wing parties compared to West German states.
Given their strong impact while controlling for other relevant factors of a candidate’s political profile, the role of social norms should be explored more thoroughly. First, based on our cross-sectional data, we cannot fully rule out the possibility that some candidates ex post rationalized their attack behavior by considering the perceived use of negative campaigning by other candidates or the expectations of their voters. Relying on the theory of normative social behavior (Rimal & Real, 2005) as an established approach in the social sciences, we postulate social norms as predictors of individual behavior. Future studies should employ longitudinal studies to test whether social norms are predictors or consequences of attack behavior.
Furthermore, social norms might provide insight for the debate on the role of gender in negative campaigning. Building on role congruity theory (Eagly & Karau, 2002) and gender stereotypes (Fiske, 2018), female politicians are assumed to make less frequent use of attacks during campaigning. However, some studies on negative campaigning show that women use negative campaigning as frequently or even more so than men (Walter, 2013). The current theoretical reasoning overlooks an essential aspect of a candidate’s attack behavior—namely, the gender of the target. Male politicians might refrain from attacking female opponents simply because attacking a woman is perceived within their party, by their voters, or by themselves as inappropriate. Females do not have such restrictions. Adding a social-normative perspective allows us to study the fine social interactions between (male and female) sponsors and (male and female) targets of an attack.
Finally, guided by empirically well-grounded theories (such as the theory of planned behavior (TPB, Ajzen & Fishbein, 2005)), future studies should explore further drivers of the use of negative campaigning in depth. For example, as suggested by the TPB, it seems insightful to consider attitudes toward negative campaigning in addition to social norms. A relevant question could be, for instance, how attitudes and perceived injunctive norms—that is, the perceived attitudes of others—interact in their influence on negative campaigning.
Conclusion
Our research opens new avenues to examine the drivers of attack behavior—both theoretically and empirically. To explain individual attack behavior, we apply established theories, such as the theory of planned behavior (Ajzen & Fishbein, 2005) and the theory of normative social behavior (Rimal & Real, 2005). This opens the potential to explore mechanisms that go beyond typically relying on rational considerations that explain a candidate’s decision to go negative. The key to such explorations is to reach out to rarely surveyed political candidates to examine their inner motives and reasoning. We show that political candidates at the state level are interested in participating in such studies and are willing to do so. Essentially, this study contributes to the endeavor of social scientists to theoretically and empirically understand the increasingly observed use of negative campaigning in election campaigns worldwide (Maier et al., 2023; Nai, 2019, 2020).
Declarations
Ethical approval
IRB approval was obtained from the GESIS ethics committee (reference number 2020-6) for the entire project prior to data collection provided. All participants provided informed consent prior to data collection.
Conflict of Interest
We have no conflicts of interest to disclose.
Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article’s Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by/4.0/.
Publisher’s Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
