Professional Documents
Culture Documents
Ads55 1
Ads55 1
A Content Analysis of Donald J. Trump’s and Joe Biden’s Presidential Campaign Ads
2
Abstract
This study adopts a content analysis to compare the persuasive strategies of Facebook
campaign ads placed by Trump and Biden for the 2020 Presidential Election. After analyzing a
total of 917 randomly selected Facebook ads posted during three months prior to the Election,
this study found that Trump adopted more attacking ads, less acclaiming ads, and more ads with
videos and filters, made more references to the candidate's character, and used more emotional
Keywords: Content Analysis, Persuasive Strategies, Social Media Ads, Functional Theory of
In almost every election found in the United States, campaigning is an essential part of
our democracy and political process, which gives the electorate an opportunity to look into the
candidates’ policies, their political stances, and the proposed political agenda for their term
(Nwachukwu, 2020). Political campaigns may take different formats and involve various kinds
of activities. This project chooses to study one critical component commonly found in political
political advertising are defined as “information or materials, other than nonpaid social media
messages, that are mailed, e-mailed, posted, distributed, published, displayed, delivered,
broadcasted or placed in a communication medium and that are for the purpose of influencing an
election” (National Conference of State Legislatures, 2020). They serve the functions of
visualizing and dramatizing political candidates, parties, or ideas, spreading the ideals of
candidates, and providing messages of supporting candidates and attacking the opponents (Kaid,
2012). Over the time, the primary delivery methods of political campaign advertising have
evolved from print brochures and posters in early days, to soundbites and commercials through
radio and television broadcasting, and now to various kinds of online and interactive media, such
The 2020 Presidential Election is by far the most expensive election in history
(Goldmacher, 2020). It also sets the new record for social media ad spending, with $107 million
from Donald J. Trump’s campaign and $94 million from Joe Biden’s campaign on their
respective Facebook ads since January 2019 (Leigh, 2020). Despite the increased use of social
candidates in social media ads are yet to be further studied (Porsmita, 2016). By choosing
2
Facebook as the context, this study compares how Donald J. Trump and Joe Biden adopted
different communication strategies in their campaign ads for the 2020 U.S. Presidential Election.
As one of the most popular social media, Facebook has been used by over 79% of U.S.
adults (Kreiss & McGregor, 2017) and serves as the news source for 45% of Americans (Shearer
& Gottfried, 2017). Given its wide reach, Facebook has increasingly become the focus of
political campaigns and been deemed crucial to winning an election as well as creating a
presence for politicians online (Levy et al., 2020). According to Wu (2019), the majority of both
Democratic and Republican candidates had a presence on Facebook, with 72% of whom
regularly updated their Facebook pages about policy as well as personal anecdotes. In addition,
Facebook has been found to have a significant impact on citizens’ attitudes and behavior via
Political advertising on social media has been part of presidential candidates’ campaign
strategies during the past several election cycles (Swant, 2020). In these campaigns, Facebook
has been playing important roles in facilitating political participation and encouraging
connections between voters and candidates (Bode et al., 2014). For example, since the primaries
of the 2008 presidential election, almost all frontrunners have used Facebook pages showcasing
their candidacy (Porsmita, 2016). Facebook has also been used to Donald J. Trump’s advantage
and deemed as one of the critical factors contributing to his victory in the 2016 Presidential
Due to the ongoing COVID-19 pandemic, the candidates of the 2020 Presidential
Election were not able to do as many in-person events as in the previous election cycles. As a
result, social media, particularly Facebook, has played an even bigger role than ever before.
3
Since the announcement of their candidacies, Donald J. Trump and Joe Biden have paid for over
300,000 and 58,000 ads respectively on Facebook (Facebook Ad Library, 2020), which far
exceeds the number of ads placed through other media channels. In other words, Facebook has
served as a targeted platform for both candidates to reach voters and deliver persuasive
Facebook provides different media modalities for ad creation and placement. A political
campaign ad can therefore be in the format of text-, image-, and video-only, or text + image, or
text + video. When comparing the Facebook campaign ads from Trump and Biden, it is
interesting to examine if and how they deliver persuasive messages through different types of
media modalities afforded by Facebook. Therefore, we propose the following research question.
RQ1. How did Trump and Biden adopt different media modalities in their 2020 Facebook
campaign ads?
Aside from media modalities, another technical feature examined in this study is the use
of filter in Facebook ads. The use of filters in political campaigns is a recent development that
originated on social media. Filters and the process of image filtering were first seen via social
media on Instagram (Ferwerda & Tkalcic, 2018). Initially used as a feature supported by this
photo sharing app, users are able to edit their photos by altering certain aspects of the photo to
express personal style (Ferwerda & Tkalcic, 2018). With the rise of tools such as photoshop,
(Ricchiardi, 2007). By altering the photos, such as changing the coloring or graining, users were
able to differentiate their images and emphasize certain aspects of the picture. Studies have
shown that image manipulation, including the use of filter via social media, could have an impact
on audience (Kleemans et al., 2018). Researchers have found that the misuse of technology to
4
distort or change images can shake the credibility within photojournalism (Ricchiardi, 2007).
Within social media and photo journalism, there have been more discussions on the ethical
framework that goes into sensationalizing an image and how that will impact the reader
(O’Brian, 1993). Recently, filters and photo distortion have bridged from social media self-
expression into political campaigning, filter is increasingly discussed and used (Kleemans et al.,
2018). The concept of filter has been loosely defined in research about political campaign.
Following Noggle and Kaid (2000), this study defines filter as a notable distortion or change to a
photo or a video. Noggle and Kaid (2000) have also expressed concerns on how visual distortion
and alternation are allowed in political campaigns. Such kind of distortion is said to impact
election by enhancing the evaluation and vote likelihood of the sponsoring candidate while
compromising the evaluations and vote likelihood of the opponent (Noggle & Kaid, 2000). With
filters shifting from just a form of personal social media editing to a tool used in political
RQ2. How did Trump and Biden differ regarding the use of filters in their 2020 Facebook
campaign ads?
In looking at the functional features of media (i.e., use of modality and filter), it is
important to examine the other factors that make up a Facebook campaign ad. One of these
aspects is message characteristics, such as the function and topic of campaign ads. We study
these characteristics through the lens of the Functional Theory of Political Campaign
Election is a choice between competing candidates (Benoit et al., 2000), with voting as a
“comparative act” (Beniot, 2019, p.8). The primary purpose of political campaign ads is to
5
distinguish a candidate from the opponent(s). To achieve this aim, campaign ads may adopt
different message strategies to persuade voters (Benoit, 1999). Following the Functional Theory
candidate through three primary functions: acclaim (i.e., presenting positive statement(s) about
oneself), attack (i.e., criticizing the opponent), and defense (i.e., refuting the attacks from the
opponent) (Beniot et al., 2016; Benoit et al., 2000). Depending on the candidate and the goals of
image building, campaign ads may vary in the specific type of function to use and the degree of
usage, meaning whether they choose to run more ads that praise the candidate, defend the
Earlier research on this theory has predicted that acclaims are like to be more common
than attacks and defenses in campaign discourse due to concerns about backlash from attacks and
defenses (Benoit, 2014), which was later confirmed by a meta-analysis of relevant research on
the Functional Theory of Political Campaign Discourse (Benoit, 2017). Incumbency is said to be
a variable that could influence the primary functions of campaign discourse (Dover, 2006), such
that incumbents are likely to acclaim more and attack less than challengers (Benoit, 2017). This
is due to the benefits incumbents possess when running for reelection, such as greater
recognition, ability to raise campaign funds, and ability to begin campaigning early. Thus,
incumbents are thought to have an easier time campaigning for reelection than their opponents
(Benoit, 2017). According to Benoit, unless an incumbent is deeply unpopular, challengers must
give voters a reason to vote out the incumbent and attacks are usually the basis for that argument
s (Benoit, 2017). Thus, that is why challengers are often attacking more and acclaiming less in
order to prove that point. However, more recent research has discovered that incumbency alone
is not enough to predict the adoption of discourse function. Instead, the candidates who were
6
losing in the polls were found more likely to run ads attacking other candidates instead of
praising themselves (Benoit, 2019). Guided by the Functional Theory, this study aims to answer
the following question about the campaign discourse function of Trump’s and Biden’s Facebook
ads.
RQ3. How did Trump’s and Biden’s Facebook ads differ regarding the primary campaign
used to further differentiate between candidates. Benoit (2017) also argues that the topic of
political campaign discourse needs to be studied along with functions. The two topics commonly
found in election discourse are policy and character (Benoit et al., 2016; Sohal & Kaur, 2018).
The topic of policy within election discourse often bridges between mentioning “a
candidates’ past deeds, future plans, and general goals in offices” (Jaeger, 2010, p. 33). Past deed
references outcomes or effects of action taken by a candidate. Topics in policy can include the
budget surplus, social security, taxation, and the size of government (Beniot, 2003). The
Functional Theory predicts that references to policy will be more frequent than references to
character in presidential campaign discourse (Jaeger, 2010, 33). Although the Functional Theory
does not provide specific reasoning as to why that is the case, researchers theorize that it goes
hand in hand with the preferability tenet, which states that voting is a comparison act and voters
will choose one candidate over the other. Opinion poll data from election cycles since 1976
reveals that more voters indicate policy, rather than character, is the most important determinant
of their vote for president (Benoit, 200). Therefore, campaigns are more likely to focus on
policy.
7
The topic of character in election discourse often focuses specifically on the candidates as
individual, referencing their personal qualities, leadership ability, and ideals (Airne & Benoit,
2005). Personal qualities are seen as descriptions of a candidate's personality traits, such as
management of a country. Ideals focus on the candidate's principle to be aimed at within the
country (Jaeger, 2010). Though the mention of character is also commonly found in political
campaign ads, Benoit found that candidates whose attacks discourse focus on policy emphasize
policy more and character less, and are more likely to win the election than their opponents
(Benoit, 2004 ). When examining the topics of Trump’s and Biden’s Facebook campaign ads, we
RQ4. How did Trump’s and Biden’s Facebook ads differ regarding the mention of a)
Persuasive Appeals
Aside from functions and topics, another communication strategy we examine in this
study is persuasive appeal. Vesnic-Alujevic and Bauwel (2014) referred to persuasive appeal as a
component that intentionally seeks to influence voters. According to texts dating back to
Aristotle, persuasion appeal can fall under logos, pathos, and communicator characteristics, also
referred to as ethos (English, 2011). Logos, also named logical appeal, focuses on persuading
through the use of reasoning, which includes critical cognition, analytical skills, good memory,
and purposeful behavior; pathos, also referred to as emotional appeal, persuades by invoking an
impassioned or moving reaction from the audience; finally, ethos, also called as source appeal or
credibility appeal, convinces the audience by the character (such as importance of credibility) of
a message source (Mshvenieradze, 2013). For ethos, politicians use experts and the source-
8
credibility appeals often in an effort to influence the degree to which the audience finds the
message credible (English, 2011). For logos, politicians often use the presentation of statistics or
campaign ads (English, 2011). For pathos, politicians often use emotional appeals, such as
humor appeals, to help audiences take political information relevant to their campaign (English,
2011). When comparing how Trump and Biden persuaded Facebook users through their ads, we
RQ5. How did Trump and Biden use different persuasive appeals (i.e., logos, pathos, and
Methods
Coding Scheme
The unit of observation in this study is each Facebook campaign ad posted by both
candidates during three months before the Election Day. After deciding the sample, a codebook
Type of media modality used. A typical Facebook ad could be in several possible forms:
text only, video only, image only, text + video, and text + image. In order to capture different
types of media modality, this study coded whether the ad used text (1 = yes, 0 = no), image (1 =
yes, 0 = no), and video (1 = yes, 0 = no). For example, when an ad adopted the form of text +
Use of filters. The ads with images or videos were coded whether they applied filters to
alter shades, tones, texture, and color in a noticeable manner, or add any special effect with the
purpose of changing the appearance of an image or a video (1 = yes, 0 = no). An example of this
could be a color shift in the photo, such as altering the photo from color to black and white. It is
9
important to note that the filter has to be applied for a significant amount of time or a significant
amount of space, meaning it must take a significant role in altering the photo or the video.
Primary discourse function. Following Benoit (2019, 2020), we coded the primary
campaign discourse: acclaim, attack, and defense. The attack function focuses on criticizing the
opponent, while the acclaim function focuses on portraying oneself in a favorable light, with the
defense function focusing on refuting the attacks from the opponent. When an ad uses more than
one function in the discourse, we coded the primary function based on the most dominant theme
Reference to policy. In addition to primary discourse function, we also coded the topics of
the campaign ads (Benoit et al., 2016; Sohal & Kaur, 2018). The first element we looked at was
the reference to policy. In this study, if an ad mentioned any action (past, current, or future)
“1 = yes”, regardless of which candidate’s policy the ad referred to. Otherwise, it was coded as
“0 = no”. For example, the reference to policy could be using the words “policy, issue, law,
regulation” or referring to a specific issue/policy topic (e.g., abortion, health care, labor
Reference to character. Another topic of the campaign ads coded in this study was
reference to a candidate's character. For each ad, we coded whether it mentioned anything about
the characteristics, traits, abilities, or attributes of the candidates (Benoit et al., 2016), such as
personal qualities, leadership ability, ideals, etc. Some examples of personal qualities included
oriented, etc. Some leadership ability examples included the use of words like “leader”,
10
“leadership”, “run the country”, etc. Some examples of ideals included the mention of the
standard of perfection, the principle to be aimed at, something somebody is pursuing, etc. When
any of these elements was mentioned in the ad, it was assigned “1 = yes”. Otherwise, the ad was
coded “0 = no”.
Use of persuasive appeal. In terms of persuasive appeals, each ad was coded for whether
it used logos, ethos, and/or pathos. When an ad included persuasion with reasoning, such as
using facts, figures, statistics, case studies, scientific evidence, etc. to justify arguments, it was
coded as “1 = yes” for using logos (i.e., logical appeal). Otherwise, it was coded as “0 = no” for
using logos. When an ad included elements that persuade by creating an emotional response,
such as through an impassioned plea or a convincing story with sensory details, it was coded as
“1 = yes” for using pathos (i.e., emotional appeal). Otherwise, it was coded as “0 = no” for using
pathos. For ethos (i.e., source appeal), the ad was coded whether it included any elements that
persuade an audience via the authority or credibility of the persuader who was disclosed as a
notable or expert figure in the field, a popular celebrity, a figure representative of a particular
To code each ad, the student and the faculty mentor first went through several training
and pilot-coding to finalize the operationalization of the coding categories and developed
sufficient examples to guide coding, as well as multiple rounds of revisions to fine tune the
coding schemes. After establishing the coding schemes, the student and the faculty mentor then
went through two rounds of pilot coding to establish inter-coder reliability for all items. We first
randomly selected 10% of the sampled ads, coded them, and made some changes to the existing
coding schemes. Then we randomly selected another 15% of the sampled ads and coded them
11
based on the revised codebook to establish inter-coder reliability. The Krippendorff’s alpha for
all of the coding categories ranged between .83 and 1.0, revealing high degrees of inter-coder
reliability. After establishing the inter-coder reliability, the student independently finished the
Results
With regard to the type of media modality used, Trump never adopted any text-only
Facebook ads during the three-month period whereas Biden placed some. During this period,
Biden placed more ads with images than Trump, X2(1, N = 917) = 4.39, p < .05 (Figure 1),
whereas Trump placed more ads with videos than Biden, X2 (1, N = 917) = 4.39, p < .05 (Figure
2). These differences were mainly attributable to the candidates’ respective choices of image vs.
Specifically, in September, Biden placed more ads with images than Trump, X2 (1, N =
299) = 4.68, p < .05. In October, Trump’s ads also used significantly fewer images than Biden’s,
X2 (1, N = 310) = 8.92, p < .01. It was the opposite situation for the use of videos. In both
September, X2 (1, N = 299) = 4.68, p < .05 and October, X2 (1, N = 310) = 8.92, p < .01, there
were more ads with videos from Trump than from Biden. There was no significant difference
Use of Filter
Trump’s ads used filters more frequently than Biden’s during the three-month period, X2
(1, N = 917) = 13.34, p < .01. This difference is particularly attributable to the more frequent use
of filters by Trump in the months of August, X2 (1, N = 308) = 11.16, p < .01, and October, X2 (1,
12
N = 310) = 15.09, p < .01 (Figure 3). There was no significant difference between the two
Overall, Trump placed more attacking ads whereas Biden adopted more acclaiming ads
on Facebook across this three-month period, X2(1, N = 917) = 51.55, p < .01(Figure 4). After a
closer examination by month, the same pattern was found for the ads placed in August, X2(1, N =
308) = 43.51, p < .01., and October, X2 (1, N = 310) = 28.40, p < .01, but not for the ads placed in
September. In both August and October, Trump placed more attacking ads than Biden, whereas
Biden placed more acclaiming ads than Trump. It is interesting that neither candidate adopted
defense function in their Facebook campaign ads at all during the examined three-month period.
Reference to Policy
During the entire period of the three months, there was no significant difference between
the two candidates regarding the mention of policy in the ads. However, when delving into the
specific month prior to the election, Trump made more references to policy in his ads then Biden,
X2 (1, N = 310) = 6.35, p < .05, in October (Figure 5). There was no significant difference
between the two candidates in terms of mentioning policy either in August or September 2020.
Overall, Trump had more references to the characters of either candidate more frequently
than Biden in Facebook campaign ads, X2(1, N = 917) =22.89, p < .01 (Figure 6). After having a
close examination of ads by month, the same pattern was found in their August ads, X2 (1, N =
308) =14.94, p < .01, and October ads, X2 (1, N = 310) =28.16, p < .01, but not the ads placed in
September. In both August and October, Trump made more references to characters of either
candidate than Biden. Some prevalent themes observed in Trump’s attack ads were "Sleepy Joe'',
13
"Weak Leadership", “Corrupt Politician”. Biden’s ads did not include that many references to
character, although he did use things like “Irrational Trump'' in his attack ad.
Although both candidates did not use logos (i.e., logical appeal) that much, the analysis
result showed that Biden’s ads used this appeal more frequently than Trump’s over the entire
three months, X2(1, N = 917) = 6.37 p < .05 (Figure 7). The same pattern was discovered in
August, X2(1, N =308) = 6.68, p < .05, and September, X2(1, N = 299) = 4.32 p < .05, whereas in
October, there was no significant difference between the two candidates regarding the use of
logos.
Biden used pathos (i.e., emotional appeals) less frequently in his ads placed during the
entirely three-month period than Trump, X2 (1, N = 917) = 31.65, p < .01 (Figure 8). This pattern
was consistently demonstrated in the months of August, X2 (1, N = 308) = 37.57, p < .01, and
October, X2 (1, N = 310) = 12.82, p < .01. But the difference between the two was not significant
in September. Some examples of the emotional appeals used in their Facebook campaign ads
include “We need you in this election”; “Our nation depends on you”; “We cannot lose our
In terms of source appeals (i.e., ethos), there was no significant difference between the
two candidates either for the three months as a whole or by individual month.
Discussion
Through a content analysis, this study compares how Trump and Biden adopted different
persuasion strategies in their Facebook campaign ads for the 2020 Presidential Election. The
findings of this project inform us how the two candidates made strategic choices of media
features and message features to persuade voters on Facebook. Overall, Trump placed more ads
14
with videos, filters, attacks, references to candidate’s characters, and emotional appeals than
Biden, whereas Biden placed more ads with images, acclaims, and logical appeals than Trump
Unlike some early findings of the incumbent's reliance on acclaims rather than attacks in
political campaigns (Benoit et al., 2000; Benoit & Harthcock, 1999), the incumbent in this
election, Donald J. Trump adopted attacks a lot more than the challenger, Joe Biden, in Facebook
ads. Trump's choice of this campaign style could have been influenced by his overall personal
branding as a politician – a person who is incredibly outspoken and not conforming to societal
standards of what a president should say and not say (Abhad-Santos, 2021). Another possible
explanation for the more aggressive attacks from Trump could be associated with the poll losses.
Throughout his presidency, Trump had an average polling of 41 percent approval rate, which is
about 7 points lower than his predecessor (Jones, 2021). With the low polls, Trump could have
intentionally chosen the more aggressive attacks as a direct response to combat them.
Consistent with the Functional Theory (Benoit, 2014), this study found that defense was
the least used function in both candidates’ Facebook campaign ads. In fact, neither candidate
adopted this function at all during the examined three-month period. One possible explanation
for this choice is that neither candidate considered the opponent’s statements even worth
responding to. In this highly emotion-charged election (as indicated by the findings related to the
use of pathos), both sides and their supporters were polarized about their perspectives and were
not going to concede their points. Convincing the opposing side thus became very difficult.
Instead of spending time and resources to defend themselves, which could also lead to backfires,
both candidates focused on rallying their supporters on getting out to the ballots.
15
It is worth noting that there was no significant difference between the two candidates for
the mention of policy in their Facebook campaign ads. In fact, neither candidate discussed policy
more than each other’s characters across the three months. This might be largely due to the deep-
seated political divide the nation is experiencing. Many of the issues faced by both candidates are
attract and repel votes with different beliefs and values. Both candidates might only hope for
persuading enough voters to win the election and avoid emphasizing policies.
Using edited imageries in political campaign is nothing new. However, applying filters to
images and videos on social media is still a more recent development. As our election cycles
progress, there will likely be a growing reliance on social media with even more frequent use of
filters in images and videos. If unregulated, filters could significantly impact the integrity of our
elections. As some scholars have already raised concerns about use of filters on social media,
this emerging practice deserves more attention for future research on visual presentations in
political campaigns. One possible idea is to compare how filters are used to present political
There are several limitations with the current study. First, future studies should consider
expanding the sampling period to explore how the candidates’ campaign ad strategies evolved
over time. Given that there were quite some major social and political events co-occurred along
with the ongoing campaigns of both candidates, it will be interesting to see how these concurrent
events influence the candidates’ social media ad strategies respectively. In this study, we only
coded the topics covered in the ads by the general categories of policy and character. Future
research could fine tune the coding by adding more detailed subcategories under each of them,
such as past, current, and future policies, and characters as reflected through leadership, personal
16
qualities, ideals, etc. In addition to coding the type of media modality used in the ad, future
research may conduct further examination on the specific visual elements used in ads.
References
Abad-Santos, A. (2021, January 22). The future of Donald Trump’s brand. Vox.
https://www.vox.com/the-goods/22243022/donald-trump-lawsuits-future-brand
Airne, D., & Benoit, W. L. (2005). Political television advertising in campaign 2000.
Benoit, W. L. (2003). Topic of presidential campaign discourse and election outcome. Western
Benoit, W. L. (2004). Election outcome and topic of political campaign attacks. Southern
Benoit, W. L., Blaney, J. R., & Pier, P. M. (2000). Acclaiming, attacking, and defending: A
Benoit, W. L., Glantz, M., & Rill, L. (2016). Campaigning on the Internet: 2008 Presidential
Benoit, W. L., & Harthcock, A. (1999). Functions of the great debates: Acclaims, attacks, and
Bode, L., Vraga, E. K., Borah, P., & Shah, D. V. (2014). A new space for political behavior:
Brym, R., Slavina, A., Todosijevic, M., Cowan, D. (2018). Social movement horizontality in the
internet age? A critique of Castells in light of the Trump victory. Canadian Review of
Demirdöğen, Ü. D. (2010). The roots of research in (political) persuasion: Ethos, pathos, logos
Dover, E. D. (2006). Images, issues, and attacks: Television advertising by incumbents and
English, K., Sweetser, K. D., & Ancu, M. (2011). YouTube-ification of political talk: An
733-748.
Facebook Ad Library. (2020). Advertising data by date range: Spending by advertiser May 7,
https://www.facebook.com/ads/library/report/?country=US&source=archive-landing-
page&campaign_tracker_active_toggle=bar_graph
Ferwerda, B., & Tkalcic, M. (2018). You are what you post: What the content of Instagram
pictures tells about users’ personality. Paper presented at the 23rd International on
18
urn=urn:nbn:se:hj:diva-38866
Goldmacher, S. (2020, October 28). The 2020 Campaign is the most expensive ever (by a lot.
money.html
Jaeger, P. T., Paquette, S., & Simmons, S. N. (2010). Information policy in national political
campaigns: A comparison of the 2008 campaigns for President of the United States and
Prime Minister of Canada. Journal of Information Technology & Politics, 7(1), 67-82.
Jones, J. (2021, January 18). Last Trump job approval 34%; average is record-low 41%. Gallup.
https://news.gallup.com/poll/328637/last-trump-job-approval-average-record-low.aspx
Kreiss, D., & McGregor, S. C. (2017). Technology firms shape political communication: The
work of Microsoft, Facebook, Twitter, and Google with campaigns during the 2016 U.S.
Kleemans, M., Daalmans, S., Carbaat, I., Anschütz, D. (2018). Picture perfect: The direct
Leigh, D. (2020, October 29). Presidential campaigns set new records for social media ad
trump-political-ads-biden/7452228/
19
Levy, A., Rodriguez, S., Graham, M. (2020, October 8). Why political campaigns are flooding
spend-big-on-facebook-as-election-nears.html
Mshvenieradze, T. (2013). Logos, ethos and pathos in political discourse. Theory &
National Conference of State Legislatures. (2020, August 31). Digital political ads.
https://www.ncsl.org/research/elections-and-campaigns/digital-political-ads.aspx
Noggle, G., & Kaid, L. L. (2000). The effects of visual images in political ads: Experimental
testing of distortions and visual literacy. Social Science Quarterly, 81(4), 913–927.
2016 presidential campaign speeches, tweets and news stories (Doctoral dissertation).
https://aquila.usm.edu/dissertations/1749/
O'Brien, S. (1993). Eye on Soweto: A study of factors in news photo use. Journal of
Porsmita, B. (2016). Political Facebook use: Campaign strategies used in 2008 and 2012
Shearer, E., & Gottfried, J. (2017, September 7). News used across social media platforms in
https://www.journalism.org/2017/09/07/news-use-across-social-media-platforms-2017/
Sohal, S., & Kaur, H. (2018). A content analysis of YouTube political advertisements: Evidence
156.
20
Swant, M. (2020, November 2). What Joe Biden’s and Donald Trump’s Facebook ads reveal
https://www.forbes.com/sites/martyswant/2020/11/02/what-joe-bidens-and-donald-
trumps-facebook-ads-reveal-about-their-campaign-strategies/
Vendemia, M. A., & DeAndrea, D. C. (2018). The effects of viewing thin, sexualized
selfies on Instagram: Investigating the role of image source and awareness of photo
Vesnic-Alujevic, L., & Bauwel, S. V. (2014). YouTube: A political advertising tool? A case
study of the use of YouTube in the campaign for the European Parliament elections 2009.
11(1). http://ac-journal.org/journal/2009/Spring/Articles/110108%20Facebook
%20Politics.pdf
21
120
111 109
102
98
100 95
79
80 76
60 55 57
53
44
38
40
20
0
Biden Trump Biden Trump Biden Trump
August September October
79 76
80
60 55 57
53
44
38
40
20
0
Biden Trump Biden Trump Biden Trump
August September October
160
140 136
127
120 109 108
103 100
100
80
60 52
47 47
43
40
26
19
20
0
Biden Trump Biden Trump Biden Trump
August September October
Figure 4. Number of Ads with Primary Discourse Functions by Candidate and Month of
Placement
140
130 130
120
80 87
78 77
68
60
50
40 46
20 23 25
0
Biden Trump Biden Trump Biden Trump
August September October
Attack Acclaim
23
140
121
120
101
100
80
60 54
40 34
20
0
Biden Trump
Refence to Policy
No Yes
140
129
120 110
114
101 99
100
87
80
68
60 54
48
42 41
40
24
20
0
Biden Trump Biden Trump Biden Trump
August September October
160 149
143
140 135
128 125 127
120
100
80
60
40 30 30
18 19
20 9
6
0
Biden Trump Biden Trump Biden Trump
August September October
160 152