It glitters,but is it gold? Negative campaigning on social media in Portugal

Abstract

In the era of dissonant public spheres, negative campaigning on social media is an increasingly relevant topic. This study enhances the understanding of negative campaigning in the Portuguese political communication context by introducing three advancements: (1) an assessment of which factors influence the use of negative campaigning by political parties on social media; (2) a more detailed characterization of the attacks and their targets; and (3) its impact on the public. To achieve this, we conducted a quantitative content analysis of all the posts by Portuguese political parties on Facebook during the 2022 National Election (n = 1384). Our findings suggest that negative campaigning has increased in Portugal over the last decade. Attacks came mostly from challengers, more extreme and better-resourced parties; they were predominantly based on economic matters and mainly issued through audiovisual formats. The key targets were parties, especially the established ones, who faced both issue-based and trait-based attacks throughout the electoral month. Furthermore, users seem receptive to negativity, as negative campaigning posts performed better across nearly all engagement metrics. The implications of these results are further discussed.

Keywords

political communication negative campaigning dissonant public sphere social media Facebook Portugal

Introduction

In the toolbox of political communication, one of the most commonly used techniques is negative campaigning (Maier & Nai, 2022, p. 25), defined as “any criticism leveled by one candidate against another during a campaign” (Geer, 2006, p. 23).

After decades of research, this topic continues to puzzle researchers (Galasso et al., 2023). On the one hand, there is an effort to understand why and under what circumstances political actors resort to negativity (Duggan & Milazzo, 2023; Maier & Nai, 2022). On the other hand, there is no consensus on whether the effects of negative campaigning are beneficial or detrimental to democracy. Does it increase dissatisfaction with politics and contribute to low voter turnout, or does it dissect political alternatives and help voters make informed decisions (Haselmayer, 2019)? Klinger et al. (2023) conclude that, by resorting to attacks and negativity, political parties play a divisive rather than conciliatory role, contributing to reinforcing dissonance in the public sphere.

The term “dissonant public spheres” was popularized by Pfetsch (2018) and developed by Bennett and Pfetsch (2018) to characterize the new digital public sphere. This term aims to mark a reversal from the idealized version of the public sphere as a place of consensus, conceptualizing it as a space of noise, disruption, and dissonance (Pfetsch, 2020, 2023). This idea aligns with the characterization of the fourth age of political communication proposed by Blumler (2016), who concluded that “[w]here a relative uniformity, coherence and simplicity once prevailed, now everything seems to be laced with complexity, multiplicity, variety and cross-currents” (p. 28). Indeed, social media has renewed the attention to negative campaigning in recent years because these platforms quickly propagate negative messages due to their inherent characteristics and popularity among users (Klinger et al., 2023; Ross & Caldwell, 2020).

Even so, in Portugal, negative campaigning remains a very unexplored topic. Apart from a couple of academic theses and a few mentions in broader political communication studies, the most comprehensive research is by Ramalhete (2015), who analyzed the airtime, election programs, and billboards by parties during the 2011 National Election and concluded that “negative campaigning does not work in Portugal” (p. 145). In this study, we analyze what has changed in the last decade by introducing a new variable—social media—which allows us to explore a crucial part of the political communication puzzle more deeply: the public.

Negative campaigning

While there is a perception that electoral campaigns are becoming uglier (Klinger et al., 2023), it should be remembered that politics has never been a “gentleman’s sport” (Lau & Rovner, 2009, p. 286). In Ancient Rome (64 BC), Quintus Cicero famously advised his brother Marcus to use negative campaigning. The letter he wrote was translated by Freeman (2012), who synthesizes Cicero’s point: “Winning candidates do their best to distract voters from any positive aspects of their opponents possess by emphasizing the negatives. Rumors of corruption are prime fodder. Sexual scandals are even better” (p. xx). In 1800, the US presidential campaign between incumbent John Adams and candidate Thomas Jefferson was marked by vicious personal attacks from both sides. Most remarkably, James Thomson Callender, a journalist hired by Jefferson, publicly described Adams as “a hideous hermaphroditical character which has neither the force and firmness of a man, nor the gentleness and sensibility of a woman” (Haselmayer, 2019, p. 356).

In one way or another, the media has always contributed to negative campaigning. The media’s coverage of electoral campaigns is characterized by a “constant attempt to create conflict” (p. 23) and the reduction of politics to “a spectacle or a game of win/lose” (p. 30) (Salgado, 2007). It is not surprising that there is a large audience for this “two-ring circus” (Patterson, 2010, p. 25), especially considering the human bias toward negativity: negative information has a more significant impact on opinion formation and captures more attention than positive information (Jamieson, 1992; Soroka et al., 2019; Soroka, 2014). In addition, negative campaigning, built on a black-and-white, good-versus-evil logic, satisfies the cognitive need to simplify information (García Beaudoux & D’Adamo, 2013). For these reasons, it is arguable that the media has an incentive to give visibility to negative campaigning (Duggan & Milazzo, 2023).

While not a new phenomenon, negative campaigning thrives on social media, where its implementation is easier, and its effects are amplified (Ross & Caldwell, 2020). First, in politics, as in society, digital technologies tend to reproduce and reinforce existing tendencies (Milstein et al., 2004). Second, while the operation of algorithms is often opaque, it is known, for example, that Facebook values reactions five times more than likes, prioritizing emotionally charged or provocative content (Merrill & Oremus, 2021). According to whistleblower Frances Haugen, speaking to the British Parliament, using anger and hate is “the easiest way to grow on Facebook.”

Despite being an undeniably relevant tool in political chess (Maier & Nai, 2022), negative campaigning is yet to be the subject of extensive research in Portugal, as demonstrated in a recent literature review by Santana-Pereira (2023). The first systematic study, by Ramalhete (2015), analyzed airtime, election programs, and billboards of parties during the 2011 National Election, finding a predominantly attack-free campaign—with only 14% of negative content—and concluding that negative campaigning is not an effective strategy in Portugal.

Recent research consists exclusively of master’s and doctoral theses. In an analysis of the European Elections from 2009 to 2019, Fonseca (2019) concluded that European campaigns are primarily positive, and the percentage of negative campaigning remained stable over the years. In the only study addressing negative campaigning on social media (although it is not the primary goal of the research), Pina (2018) found “few negative and critical messages about other candidates” (p. 251) by political parties during the 2015 National Election, supporting previous findings. However, within the scope of negativity, Pina’s analysis only considers the supply side without including the users’ response to negative content. To the best of our knowledge, no study to date has considered this dimension in Portugal.

Theoretical framework and hypotheses

The first point of analysis to consider is the evolution of negative campaigning over time. Recent research has identified a trend toward growth among European countries (Klinger et al., 2023). In addition, this specific election was characterized by “total uncertainty” (SIC Notícias, 2022), and competitiveness can interact with negative campaigning on social media on two levels. On the one hand, increased competition between parties promotes the professionalization of communication strategies (Lisi et al., 2013), encouraging a more intense and diversified use of social media. On the other hand, high electoral competitiveness, ideological diversity, and uncertain electoral outcomes incentivize the use of negative campaigning (Auter & Fine, 2016; Duggan & Milazzo, 2023; Maier & Nai, 2022). Thus, we propose the following hypothesis:

H1: Negative campaigning has increased between the 2011 and 2022 National Elections.

Second, it is important to assess which factors influence the use of negative campaigning. The literature consistently finds that challengers engage in negative campaigning more intensively than incumbents (Auter & Fine, 2016; Gross & Johnson, 2016; Haselmayer, 2019; Nai, 2020). While incumbents can rely on their governance record, other candidates often resort to negativity to persuade voters of the need for a change in power (Lau & Rovner, 2009). Specifically, parties with fewer resources tend to use more negativity to compensate for their lack of prominence in traditional media and to increase their visibility (Haselmayer, 2019).

Research also suggests that ideological positioning influences the use of negativity. Far-right and far-left candidates, as well as populist parties, tend to employ negative tactics more frequently (Nai, 2020). In particular, far-right populist movements adopt irreverent communication styles that exploit the animosity prevalent on social media (Larsson, 2022; Pfetsch, 2023), using emotions such as fear and anger (Hameleers et al., 2017). Based on these considerations, we define the following hypotheses:

H2: Challengers resort more to negative campaigning than the incumbent.

H3: Parties with fewer resources are more likely to adopt negative campaigning.

H4: Extreme parties are more likely to adopt negative campaigning.

H5: Populist parties are more likely to adopt negative campaigning.

Negativity may also be linked to the content and format of the post. Regarding content, recent research has concluded that immigration and economic issues are usually associated with negative campaigning and populism (Bene et al., 2022; Klinger et al., 2023). In terms of format, given that audiovisual content can overcome space limitations, it is plausible that such content is more prone to spreading negative messages than static content (Ramalhete, 2015). This hypothesis is reinforced by the fact that negativity often relies on the emotionalization of the message (Klinger et al., 2023), and videos are particularly effective at promoting emotional engagement (Brader, 2005; Lau & Rovner, 2009). Based on this evidence, we propose the following hypotheses:

H6: Immigration and economic issues are the most frequent themes of negative campaigning.

H7: Negative campaigning relies more on audiovisual than textual or visual content.

Third, given the potential for negativity to contribute to dissonance in the public sphere, it is relevant to examine the characteristics of the attacks: the targets, the types of attacks, and the evolution of negative campaigning over the electoral month. Regarding the targets, Ramalhete (2015) concluded that, in Portugal, political attacks are issued mainly toward parties rather than candidates. Among parties, since political actors often “punch upward,” mainstream parties, who frequently hold power, become the primary targets of attacks (Gross & Johnson, 2016, p. 748). Thus:

H8: Negative campaigning is more often targeted at parties than at candidates.

H9: Negative campaigning is more often targeted at established parties than at the other parties.

Furthermore, as some authors remind us, negativity is not all the same (Matthes & Redlawsk, 2014; Reiter & Matthes, 2022). A realistic picture of negative campaigning requires a more nuanced analysis of the type of attacks, specifically whether they are trait-based attacks—focused on the political actors’ characteristics—or issue-based attacks, which are centered on the electoral program or political actions and policies (Auter & Fine, 2016; Geer, 2006). Generally, negative political messages tend to focus more on the latter (Auter & Fine, 2016; Geer, 2006), also in Portugal (Ramalhete, 2015).

H10: Issue attacks are more prevalent than trait attacks.

Another relevant line of analysis is the use of negative campaigning throughout the election month. Research in this area suggests that campaigns tend to become more negative near Election Day because candidates and parties prefer first to define their political and ideological positions and attack only after the saturation point of positive information is reached (Damore, 2002). In contrast, an analysis of the prime-time political broadcasts for the 2011 National Election in Portugal indicated that these tended to be more negative at the beginning of the campaign (Ramalhete, 2015). Since the results are not conclusive, we propose the following research question:

RQ1: How did the use of negative campaigning evolve during the election month?

Finally, on the user side, it is essential to analyze how this type of content is received. The effects of negativity are not unanimous, and it is unclear whether voters perceive it as helpful information in a captivating manner (Geer, 2006) or become “disgusted” with politics and withdraw from voting (Haselmayer, 2019, p. 365). Although reactions to negative campaigning depend on many factors, including individual characteristics and the political context, recent studies suggest that negative content is very popular on social media (Bobba, 2019; Klinger et al., 2023; Larsson, 2022). For example, research has demonstrated that high-arousal negative emotions like anxiety can increase the virality of messages (Song & Wu, 2018). This is relevant because it might signal the audiences’ approval of this type of content and contribute to its dissemination. Thus:

H11: Negative campaigning leads to higher levels of user engagement.

Methodological approach

We conducted a quantitative content analysis of all posts by political parties on Facebook (n = 1,384) in the month leading up to the electoral event (January 3 to January 28, 2022). In Portugal, where social media is used by 76% of the population (Luz, 2024), Facebook remains the most used social media network, not only in general (69%) but also specifically for news (40%) (Newman et al., 2023). It is followed by WhatsApp (general use: 68%; for news: 24%), YouTube (general use: 64%; for news: 23%), Instagram (general use: 54%; for news: 23%), Facebook Messenger (general use: 49%; for news: 14%) and TikTok (general use: 26%; for news: 9%) (ibid.). Due to its architecture, Facebook is considered the most important social media platform for election campaigns (Bossetta, 2018), a relevance that has been increasing over the years (Larsson, 2022). In Portugal, considering that young people are the least likely to vote (Magalhães, 2022), the decline in Facebook’s popularity among the youth—who prefer YouTube and Instagram (Newman et al., 2023)—has not diminished the platforms’ significance as a central electoral campaign tool.

Data were collected through the Crowdtangle (2022) platform, a Facebook public information tool, and coded by two independent trained coders using a shared codebook. Ethical approval was deemed unnecessary as we exclusively utilized publicly accessible materials posted on the parties’ official Facebook accounts. The unit of analysis is the entire post, including both the copy and (audio)visual elements, such as images (only the first in the case of photo albums) and videos (the first minute). For reliability testing, coders coded a random sample of 192 posts (14% of the total posts), and the results showed a common understanding of the categories and a high level of agreement (Holsti > 0.89 for all indicators). All data collected were analyzed using SPSS software for Windows (version 28). Descriptive statistics, a Student’s t-test, binomial tests, and logistic regression were used, with a p-value of <.05, indicating statistical significance.

Measures

Party characterization

Details on party status, political positioning, and resources are detailed in Table 1. The classification of the parties’ political positioning follows the Chapel Hill Expert Survey (Jolly et al., 2022), the PopuList 3.0 database (Rooduijn et al., 2023), and the current seating arrangement in the Portuguese Parliament. The parties’ resources were measured using the publicly available records of the 2022 campaign expenses (Entidade das Contas e Financiamentos Políticos, 2022).

Table 1.

Party status, political positioning, and resources.

Party	Status	Political positioning^a	Resources (2022 campaign expenses, €)
Unitary Democratic Coalition (CDU)	Challenger	Far-left	551,771
Left Bloc (BE)	Challenger	Far-left	590,428
Free (L)	Challenger	Left	59,650
Socialist Party (PS)	Incumbent / Establishment party	Center-left	3,388,778
People-Animals-Nature (PAN)	Challenger	Center	252,030
Social Democratic Party (PSD)	Challenger / Establishment party	Center-right	1,858,865
Liberal Initiative (IL)	Challenger	Right	599,002
People’s Party (CDS-PP)	Challenger	Right	299,660
Enough (CH)	Challenger	Far-right/populist	615,390

Parties are ordered from top (left) to bottom (right) according to their ideology.

Source: Elaborated by the authors.

Negative campaigning

In line with the terminology proposed by Geer (2006), we conceptualize negative campaigning as a political communication strategy aimed at highlighting the political adversary’s flaws. Therefore, we coded this variable when a single post contains attacks (see type of attack) on a political actor (parties and/or candidates). These attacks can be oral (e.g., in a video) or written.

Type of attack

Attacks were analyzed according to two dimensions, based on the categorization by Brooks and Geer (2007): issue attacks versus traits attacks. Issue attacks include criticisms of the issue stances resulting from the performance of a party/candidate (e.g., when a party was part of a government) or policy standpoints of a party/candidate (e.g., in rallies or the election program). Trait attacks include criticisms of political actors’ professional competence (e.g., experience in public offices, academic degrees, and politics-related job experience), personal competency (e.g., the ability to make sane decisions in demanding situations and to be a good problem solver), rhetorical skills (e.g., the speed, pitch, and understandability of someone’s public speeches), appearance (e.g., describing someone as ugly and fat/thin), and credibility and integrity (e.g., the degree to which a party or a candidate is trustworthy or not and accusations of lying and corruption).

Authors

The authors of negative campaigning were considered the authors of the post (a list of parties is available in Table 1).

Targets

Among the negative campaigning posts, we registered if the targets were parties and/or candidates and their affiliation.

Topics

Posts were categorized based on whether they addressed themes related to economy and finance (e.g., economic crisis, austerity measures, trade agreements, taxation, national debt, and budget) or immigration (e.g., integration policy, naturalization, and refugee distribution).

Format

Posts were coded as audiovisual (videos) or static (images—photos, graphics, illustrations—or text-only).

User engagement

To measure user engagement, we analyzed the metrics adopted by reference studies in this field: reactions, comments, and shares (Bene et al., 2022; Klinger et al., 2023).

Results

Proportion of negative campaigning

The use of negative campaigning on Facebook, in total and by party, is detailed in Table 2. Results show an election campaign predominantly free of attacks on the political opponent, with only 20% of negative campaigning (barely one out of five posts). Nevertheless, there is an increase of six percentage points compared to the proportion documented by Ramalhete (2015), thereby supporting H1.

Table 2.

Frequency of negative campaigning posts by party.

Party	Frequency of posts
	Without attacks		With attacks		Total
	n	%	n	%	n	%
CDU	160	76.9	48	23.1	208	100
BE	99	62.3	60	37.7	159	100
L	87	94.6	5	5.4	92	100
PS	148	87.6	21	12.4	169	100
PAN	158	90.8	16	9.2	174	100
PSD	153	75.7	49	24.3	202	100
IL	62	63.9	35	36.1	97	100
CDS-PP	85	82.5	18	17.5	103	100
CH	149	82.8	31	17.2	180	100
Total	1101	79.6	283	20.4	1384	100

Source: Elaborated by the authors.

Drivers of negative campaigning

A logistic regression analysis was conducted to assess which factors influence the use of negative campaigning. The inferential goodness-of-fit test of Hosmer-Lemeshow was non-significant [Χ²(8) = 15.32, p > .05], suggesting that the model fit the data well. The summary of the estimated parameters for the main effects is shown in Table 3.

Table 3.

Logistic regression model.

	Odds ratio	95% CI		p-value
	Odds ratio	Lower	Upper	p-value
Intercept	0.056			<.001
H2 Challenger party*	2.562	1.406	4.667	.002
H3 Resources*	1.55	1.003	2.412	.049
H4 Extreme party*	1.822	1.267	2.619	.001
H5 Populist party	0.655	0.407	1.054	.081
H6 Immigration topic	1.020	0.334	3.116	.973
H6 Economy topic*	4.663	3.425	6.348	<.001
H7 Audiovisual format*	2.357	1.766	3.146	<.001

p < .05. Nagelkerke R² = .20. Source: Elaborated by the authors.

At the party level, results reveal significant association between challengers, extreme ideology, and party resources with the presence of negative campaigning (p < .05). Challenger parties were over two times more prone to use negative campaigning than the incumbent (OR = 2.56; 95% CI, 1.41–4.67) and extreme parties were 1.82 times more likely to resort to negative campaigning than moderate political actors (OR = 1.82; 95% CI, 1.27–2.62). Parties with more resources were about 1.5 times more likely to attack adversaries than parties with fewer resources (OR = 1.55; 95% CI, 1.00–2.41), although this result should be interpreted with caution, since the significance level is very tight (p = .049). On the contrary, we do not find a significant association between populism and negativity. These results confirm H2 and H4 but do not support H3 and H5.

At the post level, economic issues and audiovisual content show a statistically significant relationship with negative campaigning. The odds of using negative campaigning were 4.66 times higher when the post mentions economic issues (OR = 4.66; 95% CI, 3.42–6.35) and 2.36 times higher in posts with an audiovisual format (OR = 2.36; 95% CI, 1.77–3.15). In contrast, we do not observe a significant relationship between immigration themes and negativity. Thus, H7 was validated, while H6 was partially supported.

Characteristics of negative campaigning

Next, we focus only on the negative campaigning posts (n = 283) regarding the targets, the types of attacks, and the evolution of negative campaigning over the electoral month.

Targets

The targets of negative campaigning are available in Tables 4 –6. It should be stressed that the categories are not mutually exclusive since a post can target multiple parties and candidates at the same time.

Table 4.

Targets of negative campaigning (parties and candidates).

Parties				Candidates
No		Yes		No		Yes
n	%	n	%	n	%	n	%
62	21.9	221	78.1	202	71.4	81	28.6

Total number of posts = 283. A post may target parties and candidates at the same time. Source: Elaborated by the authors.

Table 5.

Targets of negative campaigning (by party).

Parties	n	%
CDU	10	3.5
BE	20	7.1
L	4	1.4
PS	168	59.4
PAN	14	4.9
PSD	72	25.4
IL	23	8.1
CDS-PP	24	8.5
CH	29	10.2

Total number of posts = 283. A post may target more than one party at the same time. Source: elaborated by the authors.

Table 6.

Targets of negative campaigning (by type of party).

Established parties				Other parties
No		Yes		No		Yes
n	%	n	%	n	%	n	%
84	29.7	199	70.3	192	67.8	91	32.2

Total number of posts = 283. A post may target established parties and other parties at the same time. Source: Elaborated by the authors.

The analysis of Table 4 allows us to conclude that, as expected, there is a clear predominance of negative campaigning posts targeting parties (78%), with the proportion of attacks against candidates being much less expressive (29%). The results of a binomial test confirm that the proportion of posts targeting parties is significantly higher (0.78), and the proportion targeting candidates is significantly lower (0.29) than expected (test proportion: 0.50, p < .001). H8 was confirmed.

Table 5 shows that establishment parties (also leading the polls) are by far the most attacked. Almost two-thirds of the negative campaigning posts target the incumbent (PS), while a quarter mentions the main challenger (PSD). The distance to the rest of the parties is noteworthy.

Table 6 demonstrates that established parties are attacked in 70% of the negative campaigning posts, while the rest of the parties, collectively, are only mentioned in 32% of the posts. The binomial test results confirm that established parties are targeted significantly more (0.70) and that the other parties are targeted significantly less (0.32) than expected (test proportion: 0.50, p < .001). These findings confirm H9.

Type of attack

Next, we analyze the type of attack, as presented in Table 7.

Table 7.

Type of attack.

Issue				Traits
No		Yes		No		Yes
n	%	n	%	n	%	n	%
123	43.5	160	56.5	129	45.6	154	54.4

Total number of posts = 283. A post may contain more than one type of attack. Source: Elaborated by the authors.

The results reveal that the two types of negative campaigning are used similarly, with a slight predilection for issue attacks (57%) over trait attacks (54%). Among the issue attacks, we find, for example, CDU blaming PSD, CDS-PP, and PS for the “wrong policies” on regionalization (Coligação Democrática Unitária [CDU], 2022), or PSD claiming that the Socialist Party’s policies have led Portugal to the “Europe’s tail” (Partido Social Democrata, 2022a), as seen in Figure 1.

Figure 1.

Example of an issue attack.

An extreme example of a trait attack is Enough’s attacks on all candidates: “Costa is worn out and smells of mold (. . .). Catarina Martins is a broken record. A weak figure. A hypocrite elevated to her maximum exponent. Jerónimo de Sousa can no longer keep up. Chiquinho makes videos thinking we are in a circus but always ends up being the clown on duty. Cotrim de Figueiredo talks a lot but has not yet explained, after all, whether or not he had anything to do with João Rendeiro’s bank. Rui Rio is bewildered, unable to admit that he will have to govern with Enough, and Inês Sousa Real is already moldy, rotten fruit from the greenhouse she came from” (Chega, 2022). A more typical strategy is questioning the parties or candidates’ integrity, as seen in Figure 2, where the Liberal Initiative claims that the Socialist Party candidate has been lying in the TV debates (Iniciativa Liberal, 2022).

Figure 2.

Example of a trait attack.

It should be noted that both types of attacks can be used simultaneously. For example, BE criticizes the Socialist Party’s pension policies while also eroding the party’s credibility: “PS has been manipulating the Social Security figures” (Bloco de Esquerda, 2022). Another example is PSD’s claims that “PS was unable to take advantage of a period of economic growth in Europe,” while using “socialist propaganda” to “create the illusion of helping many [people]” (Partido Social Democrata, 2022b).

A binomial test was conducted to determine if the frequencies followed the probable distribution. The results indicate that both types are slightly more prevalent than expected but not significantly (issue-based: 0.55; p = .191; traits-based: 0.54; p = .096; test proportion: 0.50). Thus, we reject H10.

Timing

Graph 1 presents the number of negative campaigning posts per day of the campaign.

Graph 1.

Frequency of negative campaigning posts per day.

Regarding RQ1, we find an irregular distribution, not showing a specific trend in employing this strategy at a particular point in the campaign. If we divide the analysis period in half, we find 151 negative-tone posts in the first half and 132 in the second half. No connections between campaign moments and spikes in negative campaigning can be drawn from the content analysis. For example, the peak on January 20 coincides with the last electoral debate among all parties with parliamentary seats on the radio. However, less than half of that day’s posts mention that event. In addition, a debate with the same parties had previously happened on national television on January 17 and did not generate the negativity peak. These data reflect the flexibility and immediacy offered by social media platforms like Facebook, which were not provided by other tools.

User engagement

The user engagement metrics (average reactions, comments, and shares) are presented in Table 8.

Table 8.

User engagement.

Metrics	Posts	n	Mean	Std deviation	Std error mean
Reactions	Without attacks	1101	409.97	450.354	13.573
Reactions	With attacks	283	556.48*	780.732	46.410
Comments	Without attacks	1101	67.79	206.280	6.217
Comments	With attacks	283	68.30	119.785	7.120
Shares	Without attacks	1101	63.43	96.214	2.900
Shares	With attacks	283	141.44*	402.232	23.910

p < .05.

Source: Elaborated by the authors.

The analysis of Table 8 reveals that negative campaigning posts generated, on average, more user engagement across all metrics: more reactions (556.5 vs 410.0), comments (68.3 vs 67.8), and shares (141.4 vs 63.4). Although the difference is almost nonexistent in terms of comments, it is pronounced especially in shares, with negative campaigning posts being shared more than twice as much.

To confirm if the mean difference was statistically significant, we conducted a series of independent t-tests. The results show that negative campaigning posts gathered significantly more reactions (t₃₃₂ = −3.030, p < .05; Cohen’s d = −.27), and shares (t₂₉ = −3.239, p = < .05; Cohen’s d = −.39), although the effect sizes indicate the association is small in both cases. The number of comments did not differ significantly between the two groups (t₁₃₈₂ = −.040, p = .968). Thus, H11 is partially confirmed.

Further analysis of the 10 posts that received the most reactions shows that 7 contain negative campaigning. A similar trend is observed among the 10 more shared posts, with half containing negative campaigning. The most popular post in terms of reactions (8519) and shares (6325) is a video fragment of a TV debate where the PSD candidate criticizes the socialist government for the mismanagement of TAP airline (Partido Social Democrata, 2022c).

Discussion

This study aimed to deepen our understanding of parties’ negative campaigning on social media in Portugal and its impact on the public. Consistent with previous research (Pina, 2018; Ramalhete, 2015), the campaign is predominantly attack-free, with negative campaigning accounting for just a quintile of the content. However, this study finds a six percentage point increase compared to what Ramalhete (2015) reported in the 2011 National Election (H1). This trend aligns with broader research indicating a growth in negative campaigning among European countries (Klinger et al., 2023). Although the heterogeneity among the research objects does not allow for a strictly direct comparison, it is plausible that the increase in the number of negative messages is due to the characteristics of social media platforms, which not only make attacking easier (Ross & Caldwell, 2020), but also value it algorithmically (Merrill & Oremus, 2021).

Next, we analyzed the factors that drive the use of negative campaigning. First, in line with other studies (Auter & Fine, 2016; Gross & Johnson, 2016; Haselmayer, 2019; Nai, 2020), we find that challengers resort more to negative campaigning than the incumbent (H2), who can rely on the results of past performance (Lau & Rovner, 2009). Second, the idea that parties with fewer resources resort more to negative campaigning to increase their visibility (Haselmayer, 2019) was not supported (H3). On the contrary—although the association is weak and should be interpreted cautiously—generally speaking, parties with more resources were more likely to attack adversaries than parties with fewer resources. It is plausible that richer challenger parties have a greater ability to implement more sophisticated political strategies, including negative campaigning. They may also use this tactic to divert attention from internal problems or shortcomings. Third, the expectation that extreme and populist parties are more negative (Nai, 2020) was only partly met: we found an association between extremism and negativity (H4) but not between populism and negativity (H5). This divergence suggests that the association between extremism, populism, and negativity is not linear and might vary from country to country.

Fourth, while economic topics were a driver of negativity, immigration played an insignificant role (H6). Negative campaigning themes appear to mirror the pressing issues of the day in Portugal and at the time of the 2022 National Election, immigration had not yet emerged as a central topic. Given that this landscape has since changed, it would be interesting to repeat this analysis in subsequent elections. Finally, we found an association between negativity and audiovisual formats: videos were the primary format for disseminating criticism during the election campaign under analysis (H7). This finding supports the idea that platforms like Facebook also allow for more rapid and efficient leveraging of videos’ potential to emotionalize the message (Brader, 2005; Lau & Rovner, 2009).

This study also aimed to produce a more detailed characterization of the negative campaigning posts. The findings confirm that parties—rather than candidates—are the main targets of negative campaigning (H8) (Ramalhete, 2015) and that parties predominantly “punch upwards” (Gross & Johnson, 2016, p. 748): established parties are the main targets of attack (H9), especially the incumbent. Contrary to previous national and international research (Auter & Fine, 2016; Geer, 2006; Ramalhete, 2015), issue-based attacks do not stand out, occurring in almost the same proportion as trait-based attacks (H10). Regarding the timing, the dissemination of negative messages throughout the electoral month was irregular (RQ1), demonstrating the flexibility of social media that provides immediate issuance and response to attacks (Auter & Fine, 2016). This is a new paradigm compared to TV airtime, which takes time to produce and must be submitted in advance (Ramalhete, 2015).

Finally, we assessed how users received negative posts on social media. In line with recent research (Klinger et al., 2023), negative campaigning posts were more popular on Facebook, receiving significantly more reactions and shares than posts without attacks (H11). This finding contradicts the preexisting idea that “Portuguese voters are not interested in campaigns based on attacks” (Ramalhete, 2015, p. 177). A relevant avenue for future research would be to analyze which negative campaigning topics generate the highest levels of user engagement.

Conclusion

Studying the relationship between democracy and social media always requires examining both faces of Janus: social media as a potential space for social cohesion and democratic discussion and as a potential destroyer of social fabric (Craig et al., 2023).

On the bright side, negative campaigning in Portugal still represents a small part (not even a quarter) of the whole campaign. However, there are some causes for concern, particularly regarding the potential contribution of negativity to the dissonant public sphere. First, the documented increase in negative campaigning may signal the rise of fragmentation and polarization in society. When there is an attack on a candidate or party, it creates an inevitable division between attackers and defenders. Also, negative campaigning can promote uncivil discourse, reinforcing polarization (Anderson et al., 2016).

Second, the predominance of attacks directed at parties—particularly the significant proportion that criticizes their traits and questions their integrity—fosters distrust in these political structures and exacerbates the legitimacy crisis of institutions. It should be noted that, in a recent survey, just 37% of Portuguese citizens claimed to trust the Government (Belchior & Heyne, 2024). In this sense, we challenge the prevailing idea that personal attacks are more harmful than attacks directed at parties (Ramalhete, 2015).

Third, it appears that resorting to negativity is an effective way to engage Portuguese users on Facebook, which in turn incentivizes the use of this strategy by parties (Klinger et al., 2023) and legacy media (Duggan & Milazzo, 2023) in an increasingly data-driven political communication (Gibson, 2023). As Pfetsch (2023) reminds us, “as parties become increasingly movement-oriented, their boundaries become porous, and their identities may be subject to change,” and “they can easily jump on the populist bandwagon to satisfy the requests of grassroots and new voter segments (p. 348).”

For this reason, the intersection between negativity and dissonance deserves continuous and careful scrutiny. Political actors should remember that not everything that glitters is gold: today’s social media success may have detrimental consequences for democracy as a political system in the long run. The promotion and safeguarding of democratic principles should be a shared and enduring commitment (Assis, 2023).

This study is not without limitations. First, it is a single-platform research. It would be interesting to see if the same patterns hold on other social media platforms, such as X, a network focused on personal interaction (Gross & Johnson, 2016) and that promotes a “simple, impulsive, and uncivil discourse” (Ott, 2017, p. 59). Second, the study does not consider candidates’ pages, which could be a source of different and more personalized types of attacks. Finally, recent research highlights the need to avoid studying the internet as a monolithic body, suggesting the implementation of cross-demographic analysis (Liu, 2024). In this context, we acknowledge that generic metrics might hide differences between population segments, and it is essential to develop qualitative studies to understand how different audiences perceive negative communication.

Footnotes

Acknowledgements

This publication is part of the project DigiWorld—“Digital Election Campaigning Worldwide” (). We would like to thank all collaboration partners who contributed to the infrastructure of the project, the coding scheme, and the creation of the dataset used in this research.

Declaration of conflicting interests

The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The author(s) disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: This research was funded by the Foundation for Science and Technology (FCT, Portugal), through the PhD grant number 2020.05202.BD.

ORCID iDs

Sara Machado

Vicente Fenoll

Author biographies

Sara Machado holds a PhD in Information and Communication in Digital Platforms from the Faculty of Arts and Humanities of the University of Porto, Portugal. Her research focuses on political communication, participation, youth, and social media. She integrates the CITCEM—Transdisciplinary Research Center “Culture, Space, and Memory” and is a member of the Digi-World—Digital Election Campaigning Worldwide and CamforS—Campaigning for Strasbourg research networks. She has been teaching communication modules at the University of Porto since 2021.

Vicente Fenoll is an Associate Professor of Audiovisual Communication at the Department of Language Theory and Communication Science at the University of Valencia (Spain). He has broad professional experience in broadcast television news and in election campaigns. He is a member of the research groups MediaFlows—Study of communication flows in political mobilization processes, CamforS–Campaigning for Strasbourg and DigiWorld–Digital Election Campaigning Worldwide. His research interests are in political communication, social media, populism, and disinformation. He has been visiting researcher at various American and European universities.

References

Anderson

A. A.

Yeo

S. K.

Brossard

Scheufele

D. A.

Xenos

M. A.

(2016). Toxic talk: How online incivility can undermine perceptions of media. International Journal of Public Opinion Research, 30(1), 156–168. https://doi.org/10.1093/ijpor/edw022

Assis

C. C.

(2023). O paradoxo da esfera pública digital. Cadernos De Dereito Actual, 21, 101–129. https://doi.org/10.5281/zenodo.8043803

Auter

Z. J.

Fine

J. A.

(2016). Negative campaigning in the social media age: Attack advertising on Facebook. Political Behavior, 38(4), 999–1020. https://doi.org/10.1007/s11109-016-9346-8

Belchior

A. M.

Heyne

(2024). Barómetro da Política Europeia. https://ffms.pt/pt-pt/estudos/barometros/barometro-da-politica-europeia

Bene

Magin

Jackson

Lilleker

Balaban

Baranowski

Haßler

Kruschinski

Russmann

(2022). The polyphonic sounds of Europe: Users’ engagement with parties’ European-focused Facebook posts. Politics and Governance, 10(1), 13. https://doi.org/10.17645/pag.v10i1.4700

Bennett

W. L.

Pfetsch

(2018). Rethinking political communication in a time of disrupted public spheres. Journal of Communication, 68(2), 243–253. https://doi.org/10.1093/joc/jqx017

Bloco de Esquerda. (2022, January, 20). O PS tem manipulado os números da Segurança Social para defender a penalização das pensões. Facebook. https://www.facebook.com/watch/?v=609730990101878

Blumler

J. G.

(2016). The fourth age of political communication. Politiques de communication, 6(1), 19–30. https://doi.org/10.3917/pdc.006.0019

Bobba

(2019). Social media populism: Features and “likeability” of Lega Nord communication on Facebook. European Political Science, 18(1), 11–23. https://doi.org/10.1057/s41304-017-0141-8

10.

Bossetta

(2018). The digital architectures of social media: Comparing political campaigning on Facebook, Twitter, Instagram, and Snapchat in the 2016 U.S. election. Journalism & Mass Communication Quarterly, 95(2), 471–496. https://doi.org/10.1177/1077699018763307

11.

Brader

(2005). Striking a responsive chord: How political ads motivate and persuade voters by appealing to emotions. American Journal of Political Science, 49(2), 388–405. https://doi.org/10.2307/3647684

12.

Brooks

D. J.

Geer

J. G.

(2007). Beyond negativity: The effects of incivility on the electorate. American Journal of Political Science, 51(1), 1–16. https://doi.org/10.1111/j.1540-5907.2007.00233.x

13.

Chega. (2022, January 7). Costa está gasto e cheira a mofo. Facebook. https://www.facebook.com/100044191881575/posts/4847134125353091/

14.

Coligação Democrática Unitária. (2022, January 16). Neste domingo junto ao Mercado Municipal de Aljustrel, estiveram militantes do PCP, do PEV, da JCP. Facebook. https://www.facebook.com/100050462741013/posts/4948134185244906/

15.

Craig

B. B.

Rahko

S. E.

Carpenter

(2023). Reconstituting digital counterpublics in times of crisis: The case of the United States and the Republic of Georgia. Communication and the Public, 9(0), 52–68. https://doi.org/10.1177/20570473231217122

16.

Crowdtangle. (2022). A tool from Meta to help follow, analyze, and report on what’s happening across social media. https://www.crowdtangle.com/

17.

Damore

D. F.

(2002). Candidate strategy and the decision to go negative. Political Research Quarterly, 55(3), 669–685. https://doi.org/10.1177/106591290205500309

18.

Duggan

Milazzo

(2023). Going on the offensive: Negative messaging in British general elections. Electoral Studies, 83, 102600. https://doi.org/10.1016/j.electstud.2023.102600

19.

Entidade das Contas e Financiamentos Políticos. (2022). Contas da campanha eleitoral—Partidos Políticos. https://www.tribunalconstitucional.pt/tc/contas_eleicoes-ar-2022.html

20.

Fonseca

(2019). Negative Campaigning in European Parliamentary Elections: The Case of Portugal (2009-2019) [Master thesis]. Iscte—Instituto Universitário de Lisboa, Lisbon.

21.

Freeman

(2012). How to win an election: An ancient guide for modern politicians. Princeton Unievrsity Press.

22.

Galasso

Nannicini

Nunnari

(2023). Positive spillovers from negative campaigning. American Journal of Political Science, 67(1), 5–21. https://doi.org/10.1111/ajps.12610

23.

García Beaudoux

D’Adamo

(2013). Propuesta de una matriz de codificación para el análisis de las campañas negativas. Opera, 13(13), 7–23. https://revistas.uexternado.edu.co/index.php/opera/article/view/3762

24.

Geer

J. G.

(2006). In defence of negativity. Attack ads in presidential campaigns. University of Chicago Press.

25.

Gibson

(2023). Data-driven campaigning as a disruptive force. Political Communication, 40(3), 351–355. https://doi.org/10.1080/10584609.2023.2207486

26.

Gross

Johnson

(2016). Twitter taunts and tirades: Negative campaigning in the age of Trump. PS: Political Science & Politics, 49, 748–754. https://doi.org/10.1017/S1049096516001700

27.

Hameleers

Bos

de Vreese

C. H.

(2017). “They did it”: The effects of emotionalized blame attribution in populist communication. Communication Research, 44(6), 870–900. https://doi.org/10.1177/0093650216644026

28.

Haselmayer

(2019). Negative campaigning and its consequences: A review and a look ahead. French Politics, 17(3), 355–372. https://doi.org/10.1057/s41253-019-00084-8

29.

Iniciativa Liberal. (2022, January, 21). ESCOLHA BEM QUEM QUER QUE O REPRESENTE. Facebook. https://www.facebook.com/100044320273145/posts/987400268515679/

30.

Jamieson

K. H.

(1992). Dirty politics: Deception, distraction and democracy. Oxford University Press.

31.

Jolly

Bakker

Hooghe

Marks

Polk

Rovny

Steenbergen

Vachudova

M. A.

(2022). Chapel Hill Expert Survey trend file, 1999–2019. Electoral Studies, 75, 102420. https://doi.org/10.1016/j.electstud.2021.102420

32.

Klinger

Koc-Michalska

Russmann

(2023). Are campaigns getting uglier, and who is to blame? Negativity, dramatization and populism on Facebook in the 2014 and 2019 EP election campaigns. Political Communication, 40(3), 263–282. https://doi.org/10.1080/10584609.2022.2133198

33.

Larsson

A. O.

(2022). Picture-perfect populism: Tracing the rise of European populist parties on Facebook. New Media and Society, 24(1), 227–245. https://doi.org/10.1177/1461444820963777

34.

Lau

R. R.

Rovner

I. B.

(2009). Negative campaigning. Annual Review of Political Science, 12, 285–306. https://doi.org/10.1146/annurev.polisci.10.071905.101448

35.

Lisi

Marchi

Evans

A. M.

(2013). Participação política e qualidade da democracia. In Pinto

A. C.

Sousa

L. d.

Magalhães

(Eds.), A Qualidade da Democracia em Portugal: A Visão dos Cidadãos (pp. 51–78). Imprensa de Ciências Sociais.

36.

Liu

(2024). The same old same old? Three proposals for advancing the study of internet and contentious politics in China. Communication and the Public, 20570473241270619. https://doi.org/10.1177/20570473241270619

37.

Luz

(2024). Social network user penetration in Portugal from 2019 to 2028. Retrieved 13th June 2024, from https://www.statista.com/statistics/567352/predicted-social-network-user-penetration-rate-in-portugal/

38.

Magalhães

(2022). Um retrato comparativo e longitudinal, 2002-2019 (A Participação Política da Juventude em Portugal, Issue 1). Fundação Calouste Gulbenkian. https://gulbenkian.pt/forum-futuro/participacao-politica-dos-jovens-em-portugal/

39.

Maier

Nai

(2022). When conflict fuels negativity. A large-scale comparative investigation of the contextual drivers of negative campaigning in elections worldwide. The Leadership Quarterly, 33(2), 101564. https://doi.org/https://doi.org/10.1016/j.leaqua.2021.101564

40.

Matthes

Redlawsk

D. P.

(2014). The positive case for negative campaigning. The University of Chicago Press.

41.

Merrill

J. B.

Oremus

(2021). Five points for anger, one for a “like”: How Facebook’s formula fostered rage and misinformation. The Washington Post. https://www.washingtonpost.com/technology/2021/10/26/facebook-angry-emoji-algorithm/

42.

Milstein

Howard

Carr

(2004). Spiders, spam, and spyware: New media and the market for political information. In Consalvo

Baym

Hunsinger

Jensen

K. B.

Logie

Murero

Shade

L. R.

(Eds.), Selected papers from the Association of Internet Researchers Conference 2000-2002 (pp. 18–26).

43.

Nai

(2020). Going negative, worldwide: Towards a general understanding of determinants and targets of negative campaigning. Government and Opposition, 55(3), 430–455. https://doi.org/10.1017/gov.2018.32

44.

Newman

Fletcher

Eddy

Robertson

C. T.

Nielsen

R. K.

(2023). Reuters Institute Digital News Report 2023. https://reutersinstitute.politics.ox.ac.uk/sites/default/files/2023-06/Digital_News_Report_2023.pdf

45.

Ott

B. L.

(2017). The age of Twitter: Donald J. Trump and the politics of debasement. Critical Studies in Media Communication, 34(1), 59–68. https://doi.org/10.1080/15295036.2016.1266686

46.

Partido Social Democrata. (2022a, January 18). “A História Explica”—António Costa. Facebook. https://www.facebook.com/100044373911578/posts/5373356822693174/

47.

Partido Social Democrata. (2022b, January, 5). O PS não foi capaz de aproveitar um período de crescimento económico na Europa e fez aquilo que melhor sabe fazer. Facebook. https://www.facebook.com/123051634390412/posts/5324365727592284

48.

Partido Social Democrata. (2022c, January 14). A TAP é uma empresa que não só não é estratégica, como não serve os interesses de Portugal. Facebook. https://www.facebook.com/watch/?v=468688191323770

49.

Patterson

T. E.

(2010). Media abundance and democracy. Media & Jornalismo, 17, 13–29.

50.

Pfetsch

(2018). Dissonant and disconnected public spheres as challenge for political communication research. Javnost–The Public, 25(1–2), 59–65. https://doi.org/10.1080/13183222.2018.1423942

51.

Pfetsch

(2020). Democracy and digital dissonance: The co-occurrence of the transformation of political culture and communication infrastructure. Central European Journal of Communication, 13, 96–110. https://doi.org/10.19195/1899-5101.13.1(25).7

52.

Pfetsch

(2023). Conditions of campaigning in dissonant public spheres and crisis of democracy. Political Communication, 40(3), 346–350. https://doi.org/10.1080/10584609.2023.2193554

53.

Pina

M. S. F. F.

(2018). O uso da internet pelos políticos em campanhas eleitorais: Portugal Legislativas 2015 [Doctoral thesis]. Instituto de Ciências Sociais, Universidade de Lisboa.

54.

Ramalhete

(2015). Os culpados da crise: ataques e campanha negativa nas Eleições Legislativas de 2011. In Santo

P. d. E.

Lisi

(Eds.), Campanhas eleitorais, debates televisivos e propaganda. Comunicação política e as eleições legislativas de 2011 (pp. 145–186). Livros LabCom.

55.

Reiter

Matthes

(2022). “The good, the bad, and the ugly”: A panel study on the reciprocal effects of negative, dirty, and positive campaigning on political distrust. Mass Communication and Society, 25(5), 649–672. https://doi.org/10.1080/15205436.2021.1934702

56.

Rooduijn

Pirro

A. L. P.

Halikiopoulou

Froio

van Kessel

de Lange

S. L.

Mudde

Taggart

(2023). The PopuList 3.0: An overview of populist, far-left and far-right parties in Europe. www.popu-list.org

57.

Ross

A. S.

Caldwell

(2020). “Going negative”: An APPRAISAL analysis of the rhetoric of Donald Trump on Twitter. Language & Communication, 70, 13–27. https://doi.org/10.1016/j.langcom.2019.09.003

58.

Salgado

(2007). Os veículos da mensagem política: estudo de uma campanha eleitoral nos media Livros Horizonte.

59.

Santana-Pereira

(2023). Election campaigns. In Fernandes

J. M.

Magalhães

P. C.

Pinto

A. C.

(Eds.), The Oxford handbook of Portuguese politics (pp. 262–275). Oxford University Press.

60.

SIC Notícias. (2022, January 27). “Incerteza é total” para as eleições legislativas. https://sicnoticias.pt/especiais/eleicoes-legislativas/2022-01-27-incerteza-e-total-para-as-eleicoes-legislativas

61.

Song

(2018). Tracking the viral spread of incivility on social networking sites: The case of cursing in online discussions of Hong Kong–Mainland China conflict. Communication and the Public, 3(1), 46–61. https://doi.org/10.1177/2057047318756408

62.

Soroka

S. N.

(2014). Negativity in democratic politics: Causes and consequences. Cambridge University Press.

63.

Soroka

S. N.

Fournier

Nir

(2019). Cross-national evidence of a negativity bias in psychophysiological reactions to news. Proceedings of the National Academy of Sciences, 116(38), 18888–18892. https://doi.org/doi:10.1073/pnas.1908369116