Social media and personal attacks: A comparative perspective on co-creation and political advertising in presidential campaigns on YouTube
First Monday

Social media and personal attacks: A comparative perspective on co-creation and political advertising in presidential campaigns on YouTube by Noortje de Boer, Hannah Sutfeld, and Jacob Groshek

In this era of politically charged social media uses, broadcast campaign advertising is being transformed in the online environment, not only by candidates but also by citizens. Likewise, negative campaigning, specifically ‘attack’ advertising, has been analyzed and discussed widely concerning possible implications for voters. However, a specific focus on analyzing such broadcast advertisements — especially in online environments across nations — has not been explored in great detail. Thus, this inquiry examines negative campaigning in social media, since the presence of both can not be ignored in contemporary broadcast and political cultures. Specifically, this study quantitatively analyzes and compares the content of political advertising in the 2012 presidential campaigns in France and the United States that were posted on YouTube. In essence, this study considers political–cultural differences between France and America as those are expressed and presented in increasingly interconnected and dynamic forms of political communication, advertising and media co–creation. Findings presented here identify unique dimensions of reciprocity between broadcast cultures and political advertising as those are adapted and negotiated by candidates and audiences in social media.


Influence and aim of political advertising
Theoretical perspectives on political advertising
Discussion and conclusion




An important contributor to political campaigning has long been traditional forms of broadcast mass media. Generally speaking, it of course is well understood that “[m]ass media serve as the vehicles for conveying political messages” [1]. According to McNair (2011), in the field of political communication, political advertising is one of the main ways by which presidential candidates communicate with voters. In other words, political advertising is one of the backbones of political communication, particularly from candidates to citizens.

More generally, advertising has been defined as the “paid placement of organizational messages in the media” [2] and has taken place across a variety of platforms, from commercials in cinema, radio, television and more recently, online. Particularly in this contemporary era of converged media and increasing Internet penetration across nations, political advertising has become more widely accessible, monitored, and crucial to understand and explore cross–nationally (Croteau and Hoynes, 2003; McNair, 2011).

Moreover, political advertising has multiple functions of which informing individuals in society about available choices, and persuasion are the most important ones (McNair, 2011). Notably, the use of negative advertising, and particularly ‘attack’ advertising, has become increasingly common in many presidential elections around the world (McNair, 2011). Over the years, this phenomenon has been researched widely concerning possible implications for voters (see Brader, 2005; Clinton and Lapinski, 2004; Fridkin and Kennedy, 2008, 2004; Franz and Ridout, 2010; Freedman, et al., 2004; Richardson, 2001; Valentino, et al., 2004; Wattenberg and Brians, 1999; Zhao and Chaffee, 1995).

Until recently, however, most commercials have only been broadcasted on television (or radio), which are hierarchical one–to–many media with stricter gatekeepers and regulations. More current political advertisements, including attack videos, however, are now also being spread rapidly and through less centralized media flows through the Internet by means of video hosting platforms such as YouTube, where users and viewers can interact and co–create dynamically.

This study thus aims to further explore ‘attack’ campaign advertising online by presidential candidates and fill a vital gap in the prevailing literature. Furthermore, this study is carried out cross–nationally by examining cases of both American and French 2012 presidential campaigns in order to more fully examine socio–political and cultural differences in high–context and low–context countries concerning ‘attack’ campaign advertisements as they are presented online in social media.



Influence and aim of political advertising

In this line of reasoning, campaign advertising is essential to political communication, because “ads sell candidates/products to voters/consumers in a way that blurs the line between politics and commerce” [3]. Campaign advertising has thus become one of the main ways that presidential candidates interact and communicate with voters, specifically in the United States, but also Britain and other comparable countries (McNair, 2011).

In addition, political advertising serves many functions as a campaign tool such as general media exposure, informing and attracting voters, enhancing name identification (particularly for candidates and parties who are less known), and increasing citizens participation and fund–raising (Croteau and Hoynes, 2003; Johnston and Kaid, 2002; McNair, 2011). In other words, political advertising upholds a determining position within political campaigns and therefore it is worthwhile to further investigate this concept.

Yet, while the main function of campaign advertisements is to inform citizens, they also aim to persuade voters (Franz and Ridout, 2010; McNair, 2011; Richardson, 2001). The reason why advertisements are so extensively used by politicians is that journalists and editors have less interpretative control over paid advertising compared to news articles (Kaid and Johnston, 2001; McNair, 2011). Consequently, politicians themselves are able to directly manage what will and what will not be advertised and communicated to prospective voters.

Political advertising strategies

In addition, McNair (2011) argued that there are several advertising strategies by which political advertisers can add value and meaning to commodities for their prospective audience, thereby ‘purchasers’ of their political actors and their messages. First of all, importing “familiar (to the audience) meanings and signifiers from outside” [4]. One example is when political figures, experts, or otherwise ‘common folk’ justify why a certain political actor would suit the presidential role perfectly. Secondly, “advertisements may be constructed so as to associate their product–signifiers with well–known icons from the wider culture”, which is one of the most commonly used strategies within advertising [5]. For instance, comparable to celebrities being used to sell perfume, celebrity endorsements can also positively impact the popularity of presidential candidates (McNair, 2011).

Similarly, symbols of power and status are very relevant in political advertising and can significantly work in favor of the candidate. Political characters can use their experience and credibility of past (political careers) to create a sense of authority. Opposing candidates or parties with less impressive resumes, images, news articles or networks cannot use this advertising trend (McNair, 2011). Nonetheless, in general, the influence of media on political advertisements may be exerted on different levels across nations due to different media systems (Croteau and Hoynes, 2003). Thus, a cross–national research concerning campaign advertising is of particular and growing relevance to the field of political communication.

Negative political advertising

Another important trend in political advertising is negative advertising, specifically ‘attack’ advertising, which highlights weaknesses of the opponent in often uncivil and misleading terms (McNair, 2011). Negative ‘attack’ advertisements are used regularly and have “become a staple of many competitive campaigns” [6]. Many researchers have looked at increased mobilization, but also demobilization amongst voters due to political advertising. This scholarly discussion about whether negative political advertising does encourage voters to participate in democracy, discourages voters to participate, or only influences voters with particular characteristics and under certain circumstances remains ongoing and heated (Clinton and Lapinski, 2004; Fridkin and Kennedy, 2008, 2004; Richardson, 2001; Valentino, et al., 2004; Wattenberg and Brians, 1999).

Most prior research on ‘attack’ advertisements has focused on the emotional appeal of political advertisements (Brader, 2005; Valentino, et al., 2004), their persuading focus (Brader, 2005; Franz and Ridout, 2010), extent of political issue information (Freedman, et al., 2004; Zhao and Chaffee, 1995), and vote choices (King and McConnell, 2003; Valentino, et al., 2004). Findings of these studies all state that, in one way or another, political advertising is centrally influential in elections but little attention has been paid to negative advertising online, with scant attention across nations and media models.



Theoretical perspectives on political advertising

Political advertising holds an important position within political communication and there are several strategies political actors can use to successfully reach an audience (McNair, 2011). Crucially, political advertisements focus voter attention on certain issues and therefore help dictate the agenda items of elections by carrying out a well–understood agenda–setting function. Relatedly, and as identified by Devereux (2007), framing extends the concept of agenda–setting because media frames focus on the nature of media attention and salient aspects. Furthermore, framing theory provides the ability to gain insight in the way messages are produced and thus create a form of reality, which in the case of political campaigns is meant to persuade voters toward a specified action (Devereux, 2007).

The particular content and framing of issues and characteristics within political ads are therefore also important considerations. Notably, “political advertisements can make unknown candidates better known by establishing name identification, can connect the candidate with particular demographic groups, can attract new support, stimulate participation in the campaign, and help raise money for candidates and attack opponents” [7]. In addition, the content frame, in this case meaning whether ads are focused on political issues or personal characteristics, can likewise be situate and engage influential perceptual cues (Johnston and Kaid, 2002; Shen, 2004).

According to Johnston and Kaid (2002), deliberate democracy requires the citizens to be fully informed and thus, consider the stands of presidential candidates on campaign issues (see also Croteau and Hoynes, 2003; McNair, 2011; Shen, 2004). Devereux (2007) has further argued that Habermas elaborated on this idea of democracy since Habermas identified a space in which citizens can make decisions based on an open and fair discussion amongst public peers. In other words, deliberative democracy allows individuals to critically assess the government by using discussions with equals regarding political issues (Devereux, 2007).

The ‘public sphere’ that provides citizens with the ability to have such conversations is “a realm characterized by reasoned debate among equals, in which members of a public discuss matter of common concern” [8]. It can be argued then, that any construction of political advertisements include some form of persuasion and therefore do not merely fulfill the role of merely informing citizens about campaign issues (Johston and Kaid, 2002; Shen, 2004).

Put differently, the way an advertisement is framed, based on issues or personal characteristics as well as level of negativity, can have tremendous impacts on an audience and can influence the way individuals think about and evaluate certain issues or political actors. Therefore, some scholars have suggested that political advertisements, and negative ‘attack’ advertisements more specifically, can harm both the public sphere and the carrying out of deliberative democracy (Shen, 2004).

Cross–national comparison of America and France

The majority of research on political advertisements has yet to make extensive cross–national comparisons, particularly when looking at campaign ads online. Still, there has been some cross–national research that has examined the differences and similarities of political communication between Korea and the U.S. (Tak, et al., 1997), the content and appeals of political advertising in Taiwan and the U.S. (Chang, 2000), and direct marketing reactions in Japan and the U.S. (Taylor, et al., 2000). This study extends this vital arena of scholarship both technologically in moving analyses online and geographically by comparing higher–context France and lower–context America, as defined by Hall (as cited in Croteau and Hoynes, 2003; Tak, et al., 1997).

According to Hall, communication, in this case political communication, and culture are interconnected. Hall conceptualizes that communication and culture are inseparable. Furthermore, Hall differentiates between high–context and low–context cultures because sociocultural systems differ in the emphasis they put on context in communication (as cited in Croteau and Hoynes, 2003; Tak, et al., 1997). By comparing France and the U.S., cultural differences between ‘attack’ campaign advertisements, which will reflect their culture, can be highlighted.

This comparative framework is particularly relevant since France and the U.S. have very different communication styles and cultures, and thus can be modeled in using a most different systems design (Dimitrova and Strömbäck, 2005). Whereas France can be understood as a higher–context culture, where people are more reserved, indirect and formal, America is often situated as being a less contextualized cultural system with more verbal explicitness, informality, and spontaneity (Tak, et al., 1997).

Moreover, France and the U.S. are exceptionally unique, namely in that they have disparate media systems and philosophies, which “lie underneath the decisions that structure the operations of media systems” [9]. It is worthwhile to consider that Hallin and Mancini (2004) identify France as having a ‘Polarized pluralist’ media model and the U.S. as being among the countries with a ‘Liberal’ model favorable to market mechanisms and commercial media. As Benson wrote on this important feature, “The French media system offers an instructive comparison with that of the United States since it is much less commercialized and much more statist” [10].

This study focuses on presidential campaign advertising online, with special attention paid to negativity in video–based political ads, since during the past decade the presence of so-called ‘attack’ advertising has increased in political communication as practiced and therefore, within the ongoing and larger debate on political advertising (Gibson, 2009; Greengard, 2009; Kirk and Schill, 2011; Vaccari, 2008; Williams, et al. 2005).

In essence, this study contributes to the field by investigating cultural differences between France and the U.S. as these relate to and shape the contours of political advertising, campaigning, and negativity in social media channels that are becoming wider reaching and influential.




The U.S. has been reported to have, after China, the highest number of Internet users in the world and an Internet penetration of 78.2 percent (Miniwatts Marketing Group, 2011a). Likewise, France has one of the highest European number of Internet users and an Internet penetration of 69.5 percent (Miniwatts Marketing Group, 2011b). Because both countries have a similar percentage of Internet penetration, it is useful to compare these countries culturally and politically regarding negativity in presidential election campaigning as it takes place online in social media.

In addition, “the growth in online political campaigning has been phenomenal” [11] in the U.S. and the drastic rise of Internet use in political communication has been observed as a turning point in the political process, particularly with the success of Barack Obama in 2008. Furthermore, it has been shown that Sarkozy, in the 2007 French presidential campaign, used extensive video footage on both his political Web site and YouTube (Remenyi, 2007). Given that both the U.S. and France uphold freedom of expression regulations, political campaigning online is not subject to state censorship and can generally be used freely without risk of prosecution.

Although most research is done on political advertisements and ‘attack’ advertising in the U.S. (Brader, 2005; Chang, 2000; Clinton and Lapinski, 2004; Franz and Ridout, 2010; Fridkin and Kennedy, 2008; Kirk and Schill, 2011; Tak, et al., 1997; Vaccari, 2008), it follows from the unique political and media cultures (Benson, 2001) that significant differences should emerge in this comparison. Therefore, the first hypothesis builds from the assumptions implicit to ‘liberal’ and ‘polarized populist’ media systems that there will be more ‘attack’ campaign advertisements in the U.S. than in France.

H1a: American presidential candidates will use a greater number of negative (as opposed to neutral or positive) campaign advertisements online than French presidential candidates.
H1b: The online campaign advertisements of American presidential candidates will feature more negative statements and images than French presidential candidates.

According to Johnston and Kaid (2002), the types of negative attacks in presidential election campaigning can differ due to goal, culture and strategy. This study will make a distinction between two categories, namely (1) attack on personal characteristics; and, (2) attack on issue stands and consistency (Johnson and Kaid, 2002). Furthermore, France is a higher–context culture with more indirect communication and more focus on body language, whereas the U.S. is a lower–context culture with more direct communication.

Therefore, the second set of hypotheses includes the assumption that the purposes of presidential ‘attack’ campaigns will differ. Importantly, due to cultural and media differences it is expected that American presidential candidates use more personal characteristic (direct) attacks, and French presidential candidates use more (indirect) attacks on issues.

H2a: American presidential candidates will use a greater number of campaign advertisements online that attack personal characteristics than French presidential candidates.
H2b: Likewise, French presidential candidates will use a greater number of campaign advertisements online that attack issue stances than American presidential candidates.
H2c: There is an interaction between attack type and country such that the negativity of American personal characteristic ads is greater than French personal characteristic ads whereas the negativity of French issue ads is greater than American issue ads.




A collection of political advertisements posted on official YouTube channels during the 2012 presidential campaigns by both left and right candidates or parties were analyzed as well as a smaller collection obtained using key search terms. For both the U.S. and French sample the aim was to obtain a total sample where N = 100. Sampling began on 9 March 2012 and all coding was completed by 22 May 2012.

Sampling official YouTube channels

For the U.S. specifically, the political advertisements posted on the official YouTube channels during the 2012 presidential campaign of the leading Republican candidates, namely Ron Paul, Newt Gingrich, Rick Santorum and Mitt Romney, and the leading democratic candidate Barack Obama were analyzed. For France, the official page of the 2012 election was used in order to obtain the videos from the official, featured channels of Union for a Popular Movement candidate Nicolas Sarkozy, Socialist Party candidate François Hollande, National Front candidate Marine Le Pen and Christian Democratic Party candidate Christine Boutin.

Due to the abundance of campaign YouTube videos on part of the U.S. candidates, the list was additionally organized according to views, ensuring that the most popular YouTube political advertisements were included in the sample. The official YouTube channel, respectively lists of all videos posted, were then used to obtain a sample (N = 15) of each presidential candidate, comprising 75 percent of the total sample obtained. To ensure no political advertisements of previous presidential campaigns were included, posts older than 12 months at the time of sampling were not included.

In general, the political advertisements posted on the official YouTube channels during the 2012 presidential campaigns were purposefully sampled based on the criteria that one or more negative statement and/or image had to be present which was somehow an attack, comparison or criticism. In addition, the posts had to be sponsored or authorized by the promoted presidential candidate.

With this search strategy applied for the U.S., multiple presidential candidates did not have a total of 15 negative political advertisements on their official YouTube channel. Thus, for Rick Santorum (N = 11), Ron Paul (N = 14), and Barack Obama (N = 9), a slightly smaller sample was used. In addition, for both Newt Gingrich (N = 10) and Mitt Romney (N = 11) the original aimed sample (N = 15) was gathered, but throughout the coding process several links for both candidates were deleted due to withdrawal from the race or due to the deletion of the YouTube video, so these missing videos could not be analyzed here.

Similarly, it became apparent that the original sample sizes per official channel were not possible to meet when considering the French candidates, since some produced more attack videos while others focused more on general issue statements, news broadcasts and candidate profiles. Thus, for the French sample Nicolas Sarkozy (N = 20) and François Hollande (N = 23) were featured more regularly than were Marie Le Pen (N = 8) or Christine Boutin (N = 4).

Sampling non–official YouTube videos

In addition to the sample compiled from the official YouTube channels, a sample was obtained using key search terms, aimed at making up a total of 25 percent of the overall sample for each nation. Similar to the sample of official YouTube channels, after entering the key search terms, both American and French videos were organized according to the amount of views ensuring that only the most prominent negative political advertisements were part of the sample.

As with official YouTube channels, all political advertisements obtained using the key search terms were purposefully sampled based on the criteria that one or more negative statement and/or image that was somehow an attack, comparison or criticism had to be present. Also, duplicates of videos already sampled through the official YouTube channels were excluded, as were videos older than one year.

For the U.S., half of the non–official sample (N = 12) was obtained using the search terms “attack advertisement president 2012.” The other half (N = 11) was comprised using the search terms “elections attack advertisement 2012.” Throughout the coding process several links (N = 5) of the obtained videos were removed from the YouTube platform and could therefore not be included in the final sample.

When drawing the non–official video sample for France, one third (N = 5) of the videos were gathered using the search terms “election presidentielle 2012 France.” For the other remaining two–thirds (N = 10) the search terms “campagne election presidentielle 2012” were applied.

Coding and reliability

By building on, condensing and adapting Geer’s (2006) codebook for attack ads in televised presidential campaigns as well as integrating parts of Kaid and Johnston’s (2006) codebook for style and content of televised political advertising, we coded the YouTube videos for a variety of features. These included general information that required no interpretation (e.g., date of upload, number of views, country of campaign, for and by whom the ad was created) as well as other subjective characteristics, namely overall tone, negative statements and images, the format, the attack (e.g., who and what is being attacked), and the content (e.g., appeals and issues addressed). Operationalizations and reliabilities, as calculated with Cohen’s Kappa, are reported for each of these items as follows.

The overall tone of an ad could either be rated as ‘negative’, ‘neutral’ or ‘positive’, following the categorizations of Geer (2006). These three options can be understood as an initial judgment by the coders after having watched the ad on how it came across generally. In other words, if the ad was perceived as being a full–on attack on the opponent it would be coded as ‘negative’. If the ad was found to be a pro–candidate video, promoting the values and goals of that candidate it would be coded as ‘positive’ and if the ad presented either an equal balance of negative and positive aspects or was a matter–of–fact like video, for instance simply stating that people should vote it would be coded as ‘neutral’. Here, Kappa = 0.89 for American ads and 1.00 for French ones.

The category of negative statements was measured by coders counting the number of negative verbal (spoken) statements per ad. Whenever a statement would recur, it would be counted again, i.e., as an additional negative statement. This number was then inserted directly into SPSS without further judgment of how negative the statements issued were, allowing for this category to be less interpretative on intensity of statements. In this case, Kappa = 0.77 for videos from the U.S. and 0.75 for those originating in France.

Similarly, the category of negative images looked at the number of negative visuals, including images as well as written words that were featured in an ad. Here, coders identified negative imagery from a Western point of view. Accordingly, items were counted and documented by frequency of negativity and not degree of negativity. For U.S. advertisements, Kappa = 0.77 and for French advertisements, Kappa = 0.76.

The ‘attack by whom’ category documented the origin of the video and the ‘attack who’ category referred to the candidate being attacked in an ad. For both categories multiple candidate options were provided that could be chosen. Among these categories were of course the names of the most prominent candidates, such as Obama, Romney, Sarkozy, and Hollande, as well as combined options such as ‘multiple republicans’, ‘multiple democrats’ or ‘Washington in general’. In France it became quite apparent that another important option was ‘Sarkozy and Hollande’ as they were frequently attacked jointly in ads created by or endorsing other candidates. The reliability for attack by whom was Kappa = 1.0 for both American and French units. For the attack who category, Kappa reliability remained perfect for French coding and was 0.91 in American coding decisions.

The ‘category of attack’ variable was introduced based on the work of Geer (2006) to capture how the attack was being made. More specifically, this item differentiate between whether an ad was attacking the personal characteristics of an opponent or the issue stands of an opponent. An additional code was developed to include attacks within ads that utilized both personal characteristics and issues stances jointly. After initial coding, a further ‘General or other criticisms’ allowed for separating videos that were not directly attacking one opponent but rather an entire party, the general state of affairs in the country, or setting the record straight. Reliability for this category was Kappa = 0.84 for U.S. ads and 0.88 for French ads.

The final interpretive category considered in this study distinguished the creative production of the video along professional and amateur lines. The ‘video made by’ category looked at how and by whom the video was produced. Specifically, this category indicated whether an ad was officially sponsored by a candidate and professionally produced, or it an ad was made by an amateur individual or interest group by either using their own footage or dynamically adapting official material to produce a different ad. For both countries, reliability checks produced perfect agreement.

For all categories reported here, the intercoder reliability scores were based on subsamples of 15 videos for each country. Results clearly indicate a reliable level of agreement (above .70) in all instances.




A total of 144 political advertisements online were analyzed, with 71 regarding the French presidential election and 73 the American presidential campaign. The candidates featured in the French ads were for Nicolas Sarkozy (20 ads), François Hollande (23), Marine Le Pen (8), Christine Boutin (4), and a collection of other candidates (16 ads total). The American advertisements promoted Mitt Romney (13 ads), Rick Santorum (11), Newt Gingrich (10), Ron Paul (18), Barack Obama (10), and a number of other candidates (11 ads total).

Before reporting on hypotheses testing, it is worth noting some additional features and comparisons of these online political media across countries. Perhaps most importantly, there were far more views of the American ads, on average 331,575.93, than of the French ads, with 8,376.48 views on average. American ads, however, were shorter (86.19 seconds) than French ads (119.34 seconds) when comparing overall averages [12]. In terms of production, 80.6 percent of American ads were professionally created and official ads, which was virtually equivalent to the 78.6 percent of French ads online that were produced by media professionals.

In examining Hypothesis 1a, a cross tabulation found statistically significant support with chi–square (χ2 (df: 2) = 51.22, p = .000) that was also indicative of a strong relationship between the variables (Cramer’s V = 0.60, p = .000). It can therefore be reported that, as expected, American presidential candidates did use a greater number of negative (as opposed to neutral or positive) campaign advertisements online than did French presidential candidates. In this sample, 60.3 percent of American ads were negative, as compared to just 7.0 percent of negative French ads. While the number of positive ads were fairly comparable (21.9 percent for the U.S. and 25.4 percent for France), only 17.8 percent of American ads took on a neutral tone but 67.6 percent of French ads portrayed a tone that was neutral overall.

Hypothesis 1b predicted that the online campaign advertisements of American presidential candidates would feature more negative statements and images than French presidential candidates, and was also supported. Here, the average level of negative statements and images per online American ad was 10.74 (SD = 7.42), which was significantly greater than the average of 7.41 (SD = 6.62) negative statements and images in online French campaign ads (t(142) = 2.84, p = .005).

Examining this outcome further, a one–way analysis of variance (ANOVA) compared the average negativity levels of all candidates. This analysis was statistically significant (F(10, 143) = 2.85, p = .003) and Tukey post–hoc tests confirmed the propositions of H1b. All candidates are summarized in Table 1, but it is worth pointing out that the online ads of Ron Paul were most negative, on average, with 15.89 (SD = 9.57) combined negative statements and images. The negativity of Paul’s ads was shown to be significantly greater than that of Nicolas Sarkozy (M = 8.00, SD = 5.80), François Hollande (M = 8.52, SD = 6.29), and other French candidates (M = 4.38, SD = 5.51) in pairwise comparisons with Tukey HSD post–hoc analysis. The only American candidate from whom the negativity of Paul’s online advertising was significantly different was Newt Gingrich (M = 7.80, SD = 5.16), but that difference was only at the p < .10 level.

All other candidate ads were not significantly different from one another in terms of their average level of negative statements and images.


Table 1: Average amount of combined negative statements and images per online advertisement, ranked by candidate.
Note: Paul’s ads were significantly more negative than Sarkozy, Hollande, and other French candidates in pairwise comparisons with Tukey HSD post–hoc analysis.
 NMeanStd. deviationMinimumMaximum
Other French164.385.510.0021.00
Christine Boutin47.757.411.0018.00
Newt Gingrich107.805.162.0018.00
Nicolas Sarkozy208.005.792.0025.00
Barack Obama108.206.292.0024.00
François Hollande238.527.930.0032.00
Marine Le Pen88.635.802.0021.00
Mitt Romney138.775.783.0019.00
Other U.S.119.005.802.0022.00
Rick Santorum1111.366.022.0020.00
Ron Paul1815.899.571.0038.00


Finally, online American and French campaign ads were analyzed for their differences in who was negatively portrayed, or attacked, in their presentation. Here, a univariate ANOVA identified a statistically significant main effect of country (F(1, 138) = 7.06, p = .007, ηp2 = .052, observed power = .782), but not an interaction between attack subject and country (F(2, 138) = 0.63, p = .532, ηp2 = .009, observed power = .154) at the mean level of negatively per online advertisement. What this finding indicates, and is shown in Figure 1, is that American ads online were more negative on average than French ads online whether the subject of the attack was from the political left, right, or on general mainstream politics or government establishments.

Indeed, the lowest level of negativity of American online campaign ads, which were attacks on the political left (M = 9.83, SD = 6.65) was greater than that of the highest level of negativity (M = 8.89, SD = 6.92) observed in French ads that attacked the political right. The results of all these statistical analyses provide robust support for Hypotheses 1a and 1b, and the general idea that American political culture and campaign ads — now online — are more negative than comparable French politics and advertisements.


Average amount of combined negative statements and images per online advertisement across countries and differentiated by attack subject
Figure 1: Average amount of combined negative statements and images per online advertisement across countries and differentiated by attack subject.


Hypotheses 2a predicted that American presidential candidates would use a greater number of campaign advertisements online that attack personal characteristics than French presidential candidates. Relatedly, Hypothesis 2b expected that French presidential candidates would use a greater number of campaign advertisements online that attack issue stances than American presidential candidates. Analysis of H2a and H2b began with a cross tabulation, and chi–square results nearly achieved statistical significance (χ2 (df: 2) = 5.48, p = .065) but the strength of this association was relatively low (Cramer’s V = 0.21, p = .065) and some key figures were at odds with expectations.

When looking more closely at the distribution of campaign advertisements online that attacked personal characteristics, just 26.2 percent of American ads focused on personal characteristics, as compared to 46.7 percent of characteristic–based French ads. This outcome is perfectly opposite the expectation of H2a. In addition, there was also a noticeable difference between French (21.7 percent) and American (31.1 percent) online campaign ads that attacked candidates based on issue stances that was incongruent with H2b. A third category attack type ads were also considered that included attacks on both issues and personal characteristics. This combined or dual–focused attack type of online campaign ad was more popular with American candidates, at 42.6 percent, than French candidates, with 31.7 percent, but does not fit neatly into hypothesis testing [13].

Nonetheless, this finding suggests it may be imprudent to reject H2a and H2b outright and fully. While it is clear that these hypotheses did not accurately identify the appropriate location for more frequent issue and personal characteristic attacks, the combined category suggests a need for more nuance in conceptualizing campaign ads, and there may be a greater intensity of negativity in ads with combined attack types.

This proposition was expressly considered in analyzing Hypothesis 2c, which predicted an interaction between attack type and country such that the negativity of American personal characteristic ads is greater than French personal characteristic ads whereas the negativity of French issue ads is greater than American issue ads. In this instance, a univariate ANOVA found a statistically significant interaction between attack type and country (F(2, 115) = 5.55, p = .005, ηp2 = .088, observed power = .846) in average negativity level per campaign ad. Thus, as expected, American personal characteristic ads online were more negative (M = 8.63, SD = 6.33) on average than their French counterpart ads (M = 7.93, SD = 6.67). Likewise, French issue ads met expectations in being more negative, with an average of 10.46 (SD = 9.32) greater than that of American issue–based attack ads online, which had a mean of 8.63 (SD = 5.66).

Most interestingly, however, may have been the average negativity level of ads that combined attacks on both personal characteristics and issues stances. This attack type had the lowest average level of negativity French online campaign ads (M = 6.42, SD = 3.11) but, comparably, by far the highest negativity level of combined attack type American ads (M = 13.92, SD = 6.61). As summarized in Figure 2, this difference is the most drastic across American and French online campaign ads. Moreover, these findings support the key presumptions of H2c and introduce conceptual space for further study of combined personal characteristic and issue–based advertisements.


Average amount of combined negative statements and images per online advertisement across countries and differentiated by attack type
Figure 2: Average amount of combined negative statements and images per online advertisement across countries and differentiated by attack type.




Discussion and conclusion

This study set out to examine presidential election advertising as it transforms in online social media spaces by comparing the French and American campaigns of 2012. Some interesting basic features could be observed at first glance — specifically that American ads were viewed, on average, nearly 39 times more frequently than were French ads, and that American ads were overall about 33 seconds shorter. In addition, there was strong evidence for the notion that American and French political cultures would be reflected in the level of negativity observed in campaign ads online, which follows the conceptual work of Benson (2001).

Hypothesis 1a was supported by findings that showed there were more American ads with a negative overall tone, and Hypothesis 1b found that the average level of negativity in American campaign ads online was greater than that of French ads. This finding coincides with media comparisons of McKenzie (2006) as well as Tak, et al. (1997). Additional analysis confirmed specific American candidates, Ron Paul in particular, had highly negative ads, and that American ads were more negative on average overall than French online campaign ads, regardless of which party or establishment was being targeted in the ads.

Nonetheless, the importance of the findings reported here seem balanced between this support and the unexpected outcomes related to personal characteristic and issue–based attack advertising. Against H2a, and the idea that American ads were be more regularly based on personal attacks, there were fewer American ads focused on personal characteristics than characteristic–based French ads. Likewise, H2b results did not seem to fully support the transfer of a more indirect and deliberative French political culture with fewer French ads that attacked candidates on issue stances than American issue–based ads.

Somewhat unsurprising, though not explicitly hypothesized, a combined or dual–focused attack type of online campaign ad was more popular with American candidates than French candidates. This finding seems to signal a greater need for more sophisticated measurement of complex advertising techniques as well as identifying impact of ads that combined attacks on both personal characteristics and issues stances (see Geer, 2006; Johnston and Kaid, 2002). Here, in contributing to support of H2c, this ad type had the lowest average level of negativity of French online campaign ads but the highest negativity level of American ads.

In addition, though the frequency of negativity was not supported for personal characteristic and issue attacks for H2a and H2b, the intensity (or average level of negativity) by country did match those expectations as further probed by H2c. Altogether, these findings suggest at least limited but reasonable support for French and American online campaign advertising to match expectations derived from their respective political and media cultures. As such, there seems to be good reason to conclude that as broadcasting in general and political advertising in particular continue to transform in the converged co–produced media environment, prevailing political and media cultures will remain relevant to the modes of rhetoric and discourse in media texts (Groshek and Engelbert, 2012).

Tellingly, though not advanced in hypotheses, despite a nearly equivalent split in the production of ads (approximately 80 percent were professionally produced in both America and France), there was a statistically significant interaction on the average level of negativity between nation and production type [14]. As has been well–noted throughout this study, American ads were more negative on average than their French counterpart ads, and that finding remained consistent whether the ads were official, professionally produced or created by amateurs. Interestingly, though, is that online campaign ads created by non–professionals in America were the most negative (M = 13.36, SD = 10.00) whereas non–professionally produced campaign ads online from France (M = 4.53, SD = 5.67) were by far the least negative.

Putting this finding into perspective with the fact that French campaign ads had drastically less views online in addition to fewer comments, likes, and favorites as compared to American ads [15] seems to suggest something about the maturation and behavior of national online audiences. In the sample of online American campaign ads considered here, non–professional media creators have pushed the level of negativity beyond what was observed from candidates themselves, and videos that generally were broadcast on television. This audience is also very large in number of views and highly active in making contributions (such as likes and comments) in social media spaces, and as might be expected, their content is pushing the boundaries of negativity seen traditional broadcast video norms (Baum and Groeling, 2008).

Given that the data analyzed in this study is effectively cross–sectional in nature, it is not possible to track the trend of negativity in online campaign videos, which makes it difficult to determine what, exactly, these findings suggest for French advertising or political culture. While it is possible that the online French audience will grows in views and online activities, and concurrently non–professional online media producers might well come to morph into and emulate the negativity of non–professional American media (co–) creators. Alternatively, these non–professional media producers in France could equally extend the tone of their political and media culture, but to a drastically less negative degree — in this manner they could be seen to be patterning the same type of boundary negotiation as their American counterparts here in the opposite direction.

Though this study can unfortunately not accurately predict the future of online campaign advertising in either France or the U.S., the work reported here is situated at an important juncture where the online audience can surely be expected to grow in both countries. As that vital trend progresses in both countries, online ad negativity should be examined further and in greater detail with data from professional and non–professional ad producers in the future to develop a better sense of context and change over time (see also Gurevitch, et al., 2009; Hermida, 2010).

While it was not possible to accomplish that objective here, this study still provides key comparative insights as to the negativity of presidential campaign advertising as it transitions in the online social media environment. Evidence presented here suggests that political and media cultures remain central considerations in the production of political ads (Sayre, et al., 2010), and their influence has not yet been overtaken by the shift towards converged media environments and deprofessionalized media creation. Agendas and frames advanced within these ads still seem to flow largely from the candidates and the cultures, but there are at least indications that this direction may evolve (Williams, et al., 2011) but uniquely across nations and with audience–producers. End of article


About the authors

Noortje de Boer, Erasmus School of History, Culture and Communication, Erasmus University, Rotterdam, the Netherlands.
E–mail: 345807nb [at] student [dot] eur [dot] nl

Hannah Sütfeld, Erasmus School of History, Culture and Communication, Erasmus University, Rotterdam, the Netherlands.
E–mail: 339381hs [at] student [dot] eur [dot] nl

Jacob Groshek is Lecturer in Quantitative Methods and Communication at the University of Melbourne, Australia.
E–mail: jgroshek [at] gmail [dot] com



1. Croteau and Hoynes, 2003, p. 238.

2. McNair, 2011, p. 87.

3. Croteau and Hoynes, 2003, p. 227.

4. McNair, 2011, p. 89.

5. Op.cit.

6. Shen, 2004, p. 124.

7. Johnston and Kaid, 2002, p. 281.

8. Devereux, 2007, p. 342.

9. McKenzie, 2006, p. 71.

10. Benson, 2001, p. 4.

11. Schwartz and Pharr, 2003, p. 251.

12. Differences between French and America ads in terms of both the mean number of views (t(72.06) = -2.81, p = .006) and length (t(142) = 2.30, p = .023) were statistically significant.

13. A separate crosstab excluded this category and was statistically significant ((χ2 (df: 1) = 3.95, p = .047) but was relatively weak (Cramer’s V = 0.23, p = .047).

14. Univariate ANOVA returned the following results: (F(1, 138) = 5.67, p = .019, ηp2 = .039, observed power = .657).

15. Differences between French and America ads in terms of the mean number of comments (t(72.00) = -3.00, p = .004), likes (t(72.04) = -4.21, p = .000), and favorites (t(72.12) = -4.00, p = .000) were all statistically significant.



M.A. Baum and T. Groeling, 2008. “New media and the polarization of American political discourse,” Political Communication, volume 25, number 4, pp. 345–365.

R. Benson, 2001. “The mediated public sphere: A model for cross–national research,” Working Paper 2001 Series, Center for Culture, Organizations, and Politics, University of California, Berkeley, at, accessed 6 July 2012.

T. Brader, 2005. “Striking a responsive chord: How political ads motivate and persuade voters by appealing to emotions,” American Journal of Political Science, volume 49, number 2, pp. 338–405.

C. Chang, 2000. “Political advertising in Taiwan and the US: Across–cultural comparison of the 1996 presidential election campaigns,” Asian Journal of Communication, volume 10, number 1, pp. 1–17.

J.D. Clinton and J.S. Lapinski, 2004. “‘Targeted’ advertising and voter turnout: An experimental study of the 2000 presidential election,” Journal of Politics, volume 66, number 1, pp. 69–96.

D. Croteau and W. Hoynes, 2003. Media society: Industries, images, and audiences. Third edition. Thousand Oaks, Calif.: Pine Forge Press.

E. Devereux (editor), 2007. Media studies: Key issues and debates. London: Sage.

D.V. Dimitrova and J. Strömbäck, 2005. “Mission accomplished? Framing of the Iraq War in the elite newspapers in Sweden and the United States,” International Communication Gazette, volume 67, number 5, pp. 399–417.

M.M. Franz and T.N. Ridout, 2010. “Political advertising and persuasion in the 2004 and 2008 presidential elections,” American Politics Research, volume 38, number 2, pp. 303–329.

P. Freedman, M. Franz, and K. Goldstein, 2004. “Campaign advertising and democratic citizenship,” American Journal of Political Science, volume 48, number 4, pp. 723–741.

K.L. Fridkin and P.J. Kennedy, 2008. “The dimensions of negative messages,” American Politics Research, volume 36, number 5, pp. 694–723.

K.L. Fridkin and P.J. Kennedy, 2004. “Do negative messages work? The impact of negativity on citizens’ evaluation of candidates,” American Politics Research, volume 32, number 5, pp. 570–605.

J.G. Geer, 2006. In defense of negativity: Attack ads in presidential campaigns. Chicago: University of Chicago Press.

R.K. Gibson, 2009. “New media and the revitalisation of politics,” Representation, volume 45, number 3, pp. 289–299.

S. Greengard, 2009. “The first Internet president,” Communications of the ACM, volume 52, number 2, pp. 16–18.

J. Groshek and J. Engelbert, 2012. “A cross–national comparison of populist political movements and media uses in the United States and the Netherlands,” New Media & Society, published online 24 July, doi:

M. Gurevitch, S. Coleman, and J.G. Blumler, 2009. “Political communication — Old and new media relationships,” Annals of the American Academy of Political and Social Science, volume 625, number 1, pp. 164–181.

D.C. Hallin and P. Mancini, 2004. Comparing media systems: Three models of media and politics. Cambridge: Cambridge University Press.

A. Hermida, 2010. “From TV to Twitter: How ambient news became ambient journalism,” M/C Journal, volume 13, number 2, at, accessed 18 November 2012.

A. Johnston and L.L. Kaid, 2002. “Image ads and issue ads in U.S. presidential advertising: Using video style to explore stylistic differences in televised political ads from 1952 to 2000,” Journal of Communication, volume 52, number 2, pp. 281–300.

L.L. Kaid and A. Johnston, 2001. Videostyle in presidential campaigns: Style and content of televised political advertising. Westport, Conn.: Praeger.

J.D. King and J.B. McConnell, 2003. “The effect of negative campaign advertising on vote choice: The mediating influence of gender,” Social Science Quarterly, volume 84, number 4, pp. 843–857.

R. Kirk and D. Schill, 2011. “A digital agora: Citizen participation in the 2008 presidential debates,” American Behavioral Scientist, volume 55, number 3, pp. 325–347.

R. McKenzie, 2006. Comparing media from around the world. Boston: Pearson/Allyn and Bacon.

Miniwatts Marketing Group, 2011a. “Top 20 countries with the highest number of internet users,” at, accessed 2 October 2011.

Miniwatts Martketing Group, 2011b. “Europe Internet users statistics,” at, accessed 30 October 2011.

B. McNair, 2011. An introduction to political communication. Fifth edition. London: Routledge.

D. Remenyi (editor), 2007. Proceedings of the 3rd International Conference on E–government. Reading: Academic Conferences.

G.W. Richardson, 2001. “Looking for meaning in all the wrong places: Why negative advertising is a suspect category,” Journal of Communication, volume 54, number 4, pp. 775–800.

B. Sayre, L. Bode, D. Shah, D. Wilcox, and C. Shah, 2010. “Agenda setting in a digital age: Tracking attention to California Proposition 8 in social media, online news, and conventional news,” Policy & Internet, volume 2, number 2, pp. 7–32.

F. Shen, 2004. “Chronic accessibility and individual cognition: Examining the effects of message frames in political advertisements,” Journal of Communication, volume 54, number 1, pp. 123–137.

J. Tak, L.L. Kaid, and S. Lee, 1997. “A cross–cultural study of political advertising in the United States and Korea,” Communication Research, volume 24, number 4, pp. 413–430.

C.R. Taylor, G.R. Franke, and L. Maynard, 2000. “Attitudes towards direct marketing and its regulation: A comparison of the United States and Japan,” Journal of Public Policy & Marketing, volume 19, number 2, pp. 228–237.

C. Vaccari, 2008. “From the air to the ground: The Internet in the 2004 U.S. presidential campaign,” New Media & Society, volume 10, number 4, pp. 647–665.

N.A. Valentino, V.L. Hutchings, and D. Williams, 2004. “The impact of political advertising on knowledge, Internet information seeking, and candidate preference,” Journal of Communication, volume 54, number 2, pp. 337–354.

M.P. Wattenberg and C.L. Brians, 1999. “Negative campaign advertising: Demobilizer or mobilizer?” American Political Science Review, volume 93, number 4, pp. 891–899.

A. Williams, C. Wardle, and K. Wahl–Jorgenson, 2011. “‘Have they got news for us?’ Audience revolution or business as usual?” Journalism Practice, volume 5, number 1, pp. 85–99.

A.P. Williams, K.D. Trammell, M. Postelnice, K.D. Landreville, and J.D. Martin, 2005. “Blogging and hyperlinking: Use of the Web to enhance viability during the 2004 US campaign,” Journalism Studies, volume 6, number 2, pp. 177–186.

X. Zhao and S.H. Chaffee, 1995. “Campaign advertisements versus television news as sources of political issue information,” Public Opinion Quarterly, volume 59, number 1, pp. 41–65.


Editorial history

Received 19 August 2012; accepted 17 November 2012.

Creative Commons Licence
“Social media and personal attacks: A comparative perspective on co–creation and political advertising in presidential campaigns on YouTube” is licensed under a Creative Commons Attribution–NonCommercial–NoDerivs 3.0 Unported License.

Social media and personal attacks: A comparative perspective on co–creation and political advertising in presidential campaigns on YouTube
by Noortje de Boer, Hannah Sütfeld, and Jacob Groshek
First Monday, Volume 17, Number 12 - 3 December 2012

A Great Cities Initiative of the University of Illinois at Chicago University Library.

© First Monday, 1995-2015.