3. Methodology
3.2. Data Analysis
3.2.2 Intercoder Reliability
國
立 政 治 大 學
‧
N a
tio na
l C h engchi U ni ve rs it y
racism and anti-protest mobilizers), four codes for affective responses (anger, sarcasm,
sympathy, and admiration), and two codes for efficacy (group size and protest activities) have been identified in the preliminary round of analysis. In addition, emerging codes will be identified through the inductive approach if necessary during analysis.
3.2.1 Coding instrument
Appendix A includes examples of the types of pictures that were retrieved from Twitter and how they were categorized into the different concepts within this study’s proposed theoretic framework. Not only the image tweet, but the accompanying Twitter text was also retrieved.
This was because if context could not be gathered from the picture alone, the text then was used to provide further insight. The overarching concepts from the proposed theoretical framework and initial sub-codes that have been identified through an extant review of previous literature and research are included (see Appendix B).
3.2.2 Intercoder Reliability
It is essential to have a measurement of reliability for this research. Therefore, along with the researcher, another coder was recruited and trained. Coders was assigned to code the data based on the coding scheme and compare their results in terms of intercoder reliability. Data were manually coded using the seven steps of qualitative content analysis.
A Cohen’s Kappa (κ) was used to gauge the level of reliability (Burla, et al., 2008;
Lombard et al., 2002). Cohen’s Kappa coefficient is a ‘concordant rating’ in which agreements are split in -1 and +1. Once Cohen’s Kappa (κ) has been calculated, the value needs to be
calculated. Kappa values above .80 are considered the best agreements possible, while above .60 is an adequate, but solid agreement, but a value between .41 and .60 is considered moderate, but not ideal. (McHugh, 2012). During the intercoder pretest, 50 images were coded by the two
‧
國立 政 治 大 學
‧
N a
tio na
l C h engchi U ni ve rs it y
coders. Intercoder reliability was established where the Cohen’s Kappa (κ) was 0.83. Moreover, post-test reliability using 50 different images was achieved with Cohen’s Kappa (κ) being 0.87.
‧
國立 政 治 大 學
‧
N a
tio na
l C h engchi U ni ve rs it y
CHAPTER IV
FINDINGS
This chapter outlines the results of the Twitter visual content analysis. The first four sections present the overall findings of image trends with relation to different social identities, grievances, emotion focused coping (shared and reciprocal affective responses), and instrumental focused coping (efficacy). The next section describes the temporal changes of each of the
aforementioned and provides possible explanations to these noteworthy changes.
4.1 Social Identity
RQ1 attempts to understand what kinds of social identities (minority support, constructive patriotism, sports fandoms, and the emerging sub-code of protest avatars) are present during the 2017 NAPs. Three identities that were thought to be present within image tweets were minority support (Caucasians and people of color, primarily African Americans, engaging in protest activities). Images also account for those who identify as a minority member but are not necessarily protesting).
In response to RQ1, minority support was found to be the most dominant social identity found in the image tweets containing a pronounced social identity (n = 663; 47.36%) (see Table 1). Additionally, the most prominent theme within image tweets that contained minority support was people of color engaging in protest activities (n = 310; 22.14%). Pictures within this sub-code were typically of African Americans taking pictures of themselves or with organizations they are a part of, all engaging in protest activities, specifically kneeling down (see Figure 5).
‧
國立 政 治 大 學
‧
N a
tio na
l C h engchi U ni ve rs it y
Figure 5. Self-portrait of an African American kneeling down for NAPs (2017)
Constructive patriotism, which is defined as ““an attachment to country characterized by critical loyalty” (Schatz et al., 1999, p. 153), was coded with images that showed support of First Amendment and/or identifying as a military member. Surprisingly, these were found only within 8.79% (n = 123) of image tweets. In comparison, images that voiced identification with the military and supported the NAPs was more prevalent (n = 87, 6.21%) than those showing
support for the first amendment (n = 36, 2.57%) such as images showcasing the U.S Constitution or Lady Liberty. The majority of these pictures were of military members in uniform engaging in the protest activity of taking a knee.
However, although the social identity of constructive patriotism was not as evident in the image tweets as minority support was, the top two of the most retweeted images contained both social identities. This may indicate that although the flow of the NAPs’ online dialogue did not dwell too heavily on issues related to constitutional rights and military support of those rights, they regardless still received quite a large amount of attention. Two images tweeted by the same celebrity, Norman Lear, a famous TV sitcom writer, were shared to the public, receiving 252,000 retweets.
Tweet text:
#NewProfilePic #TakeAKnee
‧
國立 政 治 大 學
‧
N a
tio na
l C h engchi U ni ve rs it y
The first picture (Figures 6) shows the Lear as a young WWII Army soldier in 1943 kneeling down, most likely a simple pose for the photo. The second picture (Figure 7) shows the Lear in the present, kneeling down now for a different reason, to show solidarity for minority equality and sympathy for police brutality victims, all of which was gained in context from the tweet text. Although the large amount of retweets may be attributed to the Lear’s celebrity status, other celebrity tweets during the same period of time have not received nearly as much retweets.
Thus, it can be inferred that the popularity of this succession of images is attributed to a WWII veteran
Figure 6. Self-portrait of a white American as a young WWII vet. Picture 1 out of 2. (2017)
Figure 7. Self-portrait of an older white American vet. Picture 2 out of 2. (2017) Tweet text:
As a combat vet, I fought Nazis of WWII. Today I #takeaknee,
once more, in solidarity w/my brothers & sisters still fighting 4
equality & justice
Tweet text:
As a combat vet, I fought Nazis of WWII. Today I #takeaknee,
once more, in solidarity w/my brothers & sisters still fighting 4
equality & justice
‧
國立 政 治 大 學
‧
N a
tio na
l C h engchi U ni ve rs it y
Moreover, sports fandoms, which, defined as “an enthusiastic devotee of some particular sports consumptive object” Hunt et al. (1999, p. 440), was coded with images that depicted people either wearing clothing articles or displaying items pertaining to specific team logos, team mascots, the NFL, football, etc. Sports fandoms was only found in 1.21% (n = 17) of the image tweets of NAPs. Six images show sport teams, their logos and mascots (0.43%) and eleven revealed people wearing team merchandises (0.79%).
Table 1
Frequency and Percentage of Social Identity Image Tweets
Code Code Frequency Code Percentage
Minority Support 663 47.36%
People of color engaging in protest activities 311 22.14%
People of color engaging in protest activities 70 5%
People of color and & non- white people in protest activities
116 8.28%
Portraits of people of color 166 11.86%
Constructive Patriotism 123 8.79%
Support of First Amendment 36 2.57%
Military members 87 6.21%
Sports Fandoms 17 1.21%
Wearing sports team merchandise 6 0.43%
Display of sports team’s symbols 11 0.79%
‧
國立 政 治 大 學
‧
N a
tio na
l C h engchi U ni ve rs it y
*Protest Avatars 64 4.57%
* emerging codes
One emerging code within the image tweet data is “protest avatars” (n = 64, 4.57%), which included various images of fictional characters, such as superheroes or characters found in television ads and shows, and non-human characters, such as cats and dogs, engaging in protest activities (i.e., kneeling, raising a fist). Protest avatars in this sense did not act as images
associated with human profile pictures, but were accompanied by the protest hashtag showing support of the NAPs. These were most likely tweeted out to the public in order to show that support for the NAPs is upheld by some of the most honorable and loveable characters, whether they be real or fictional. If an admirable character supports the social movement, it might make fans of that character think about where they stand on the issue. Figure 8 is the most retweeted example of protest avatars (372 retweets) and depicts a fairly recent fictional hero within the Star Wars universe kneeling in support of the NAPS.
Figure 8. Star Wars the Force Awakens “Rey” kneeling. Text indicates support of NAPs (2017).
Tweet text:
I #TakeAKnee for all who suffer injustice and brutality in our nation
due to fear, anger, and hate this
*Admin incites #TakeAKneeFriday
‧
國立 政 治 大 學
‧
N a
tio na
l C h engchi U ni ve rs it y
4.2 Grievances
RQ2 analyzes the key visual themes for image tweets featuring grievances with two subcodes (i.e. institutional racism and anti-protest mobilizers) during the 2017 NAPS.
Institutional racism is defined as racial discrimination as normal behavior within a social system such as governmental organizations, schools, banks, and courts of law (Definition of institutional racism, 2018). The images under this code primarily related to police brutality such as photos of police members beating, tackling, shooting, or using other types of excessive forces against people of color. In response to RQ2, police brutality as a type of institutional racism accounted for only 1.79% (n = 25) of image tweets. Images of police brutality often included officers beating or tackling African Americans who had visibly painful expressions on their faces. The lack of images that portray police brutality might be due to the NAPs being a peaceful protest, forcing users to actively search for images online that represent violence used by the police on minority members. Although police brutality was the initial issue that sparked the NAPs, this Twitter content analysis also identified other types of institutional racism were account for.
A second type of grievance was anti-protest mobilizers, defined as those that are threatened by change but are also those that possess leadership that is iterative and who define meaning in a movement (Mottl, 1980; Sutherland et al., 2013). These included image tweets consisting of leaders against NAPs such as political figures and NFL organization members.
These images typically were made to look incompetent, immoral, idiotic, arrogant, hypocritical, or unqualified in order to mock or criticize their unreasonable discussions or actions regarding NAPs. While political figures as anti-protest mobilizers were present in 17.14% (n = 240) of the image tweets, the NFL organization as anti-protest mobilizers were depicted in 1.21% (n = 17) of the image tweets (see Table 2).
‧
國立 政 治 大 學
‧
N a
tio na
l C h engchi U ni ve rs it y
Table 2
Frequency and Percentage of Grievances Image Tweets
Code Code Frequency Code Percentage
Anti- Protest Mobilizers
Political Figures 240 17.14%
Donald Trump 202 14.43%
Other political figures 28 2%
Donald Trump and Others 10 0.71%
NFL Organization 17 1.21%
NFL logos or products 7 0.5%
NFL org. members 10 0.71%
Police Brutality 25 1.79%
Beating/tackling 16 1.14%
Shooting 9 0.64%
* emerging codes
With regard to political figures, the Twitter image content analysis showed that President Donald Trump was the most dominant political figure (202 = 14.43%), followed by Vice
President Mike Pence and others (n = 28, 2%). Additionally, there were 10 pictures that included both Donald Trump and other political figures (0.71%). The most dominant theme within images that placed Donald Trump as the focal point of blame is depicting the president as a white
supremacist or as the new Hitler (see Figure 9). In this image tweet, Trump’s face is a mask for Hitler, which implies that the president of the United States is eerily similar to the Nazi
demagogue. Additionally, the tweet text indicates that influences on the new White House
‧
administration, such as from Stephen Miller, a senior policy advisor, are detrimental to race relations as they might favor the white majority. Image tweets featuring Donald Trump as an anti-protest mobilizer were more prominent directly following tweets posted by the president.
Figure 9. Depicting Trump as a hidden Hitler to show (2017)
4.3 Emotional coping responses
RQ3 attempts to understand the key themes in image tweets in terms of emotionally-arousing responses in the 2017 NAPs, which falls within the emotional route to coping with collective disadvantage. Emotionally-arousing images, as explained previously, was divided into two categories (shared and reciprocal affective responses). Shared affective responses are those that are primarily negative and directed toward members of the out-group (anger and sarcasm), while reciprocal affective responses are those feelings held by group members and directed toward each other (sympathy and admiration).
4.3.1 Shared Affective Responses
Shared affective responses are defined as emotions held by the group participants at the same time (Jasper et al., 1998) and include anger and sarcasm. In respond to RQ3, the twitter
Tweet text:
Stephen Miller’s continued influence in WH should come as
no surprise to anyone. This administration has consistently shown that its true color is white.
#BLM #IndigenousPeoplesDay
#FireStevenMiller #45Resign
#TakeAKnee #Equality
#DefendDACA
‧
國立 政 治 大 學
‧
N a
tio na
l C h engchi U ni ve rs it y
image content analysis identified that anger was the most dominant shared affective response with 16.71% (n = 234), while sarcasm as a shared affective response only accounted for 7.21%
(n = 101) of image tweets (see Table 3). The two dominant types of anger were those that portrayed violence or degrading actions toward the opposition (see Figure 8) and anger against the status quo, or the existing state of affairs in the United States.
Table 3
Volume of Shared Affective Responses Image Tweets
Code Code Frequency Code Percentage
Anger 234 16.71%
Profane language 37 2.64%
Violence/ degradation towards opposition 96 6.86%
Anger expressions 3 0.21%
Anger toward status quo 98 7%
Sarcasm 101 7.21%
Using flag as decoration 36 2.57%
Insulting NAPs 62 4.43%
Other 3 0.21%
*Humor 56 4%
* emerging codes
Images that showcased anger against the status quo, especially in regard to politically and socially accepted stances, were typically images of hypocritical actions in American society. For
‧
國立 政 治 大 學
‧
N a
tio na
l C h engchi U ni ve rs it y
pray after every time he was successful, also protested the anthem in support of the
anti-abortion/pro-life movement. However, Tebow used to be celebrated for his actions but football players now are condemned for kneeling in order to protest police brutality (see Figure 10).
Thus, anger for this hypocritical stance within the United States was evidenced in the image tweets.
Figure 10. Anger for reactions when Time Tebow knelt during anthem (2017)
One emerging sub-code, humor was identified in the category of the shared affective responses. Humorous image tweets are those that show pictures that are meant to assert superiority over out-group members through mocking and ridiculing behaviors (n = 56, 4%).
Image tweets under humor were often political cartoons and memes. Figure 11 is one example of a political cartoon. It mocks not only the newest proposed border wall decision by President Trump between Mexico and the United States, but it also indicates four negative stereotypical character images found within U.S society: Trump himself, a Ku Klux Klan member, a cowboy with a machine gun, and a pedophile (inferred) priest. This political cartoon was meant to indicate that in no way is the United States a perfect country. In relation to the NAPs,
Tweet text:
Everyone send your #TakeAKnee Pics; To @nflcommish
‧
國立 政 治 大 學
‧
N a
tio na
l C h engchi U ni ve rs it y
#TakeAKnee is much more than just a protest against police brutality. It is a protest designed to acknowledge the faults within the United States in order to achieve justice for all.
Figure 11. Political cartoon criticizing the current United States in support of NAPs (2017)
In the 202 image tweets that included Donald Trump, 100 of those images also included some type of anger: profane language (22%), violence or degrading actions towards opposition (69%), and anger expressions (9%). Additionally, images may also have been coded with one or more affective response and typically included anger and the emerging code of humor. This type of image most often expresses anger by violently hurting Donald Trump and by acting out a play on words associated with the NAPs. This is done to invoke a sense of superiority and will be elaborated more in the following section. Figure 12 is an example of this image tweet
combination, including the most common type of anger. The humorous aspect of this picture will always be debatable as humor is subjective, but one can argue that Donald Trump taking a knee to the groin with the caption “Take A Knee” is the intended witticism.
Tweet text:
� � � IM NOT APOLOGIZING.
#TakeAKnee
‧
國立 政 治 大 學
‧
N a
tio na
l C h engchi U ni ve rs it y
Figure 12. Cartoon image of football player and Trump – violent (2017).
4.3.2 Reciprocal Affective Responses
Reciprocal affective responses are defined as participants’ on-going feelings toward each other, such as solidarity and loyalty, and the specific emotions they elicit (Jasper et al., 1998).
For reciprocal affective responses, admiration was expressed the most with 31.71% (N = 444) and sympathy was shown in 20.57% of the image tweets (n = 274) (see Table 4). 18.56% of image tweets that included admiration were those directed toward those engaging in protest activities associated with the NAPs. This type of image tweet had used members of family in addition to strangers and celebrities as a focal point of praise, such as Lebron James (professional basketball player) or Megan Rapinoe (professional soccer player). Meaning, that if family
members, celebrities, or even complete strangers engaged in protest activities, and the user found those actions admirable, images of them were shared via image tweets (praising the protestors) to the user’s followers through posts that could be publicly viewed (see Figure 15). Concerning image tweets that referenced sympathy, more than half of them were those of white people engaging in protest activities (n = 186, 13.29%).
Tweet text:
Tick tock.... #PenceStunt
#Pencewalkout #TakeAKnee
#TakeTheKnee #TakeAKneeNFL
#nfl TrumpIsAWhiteSupremacist
#Trumpsuckspence
‧
國立 政 治 大 學
‧
N a
tio na
l C h engchi U ni ve rs it y
Table 4
Frequency and Percentage of Reciprocal Affective Responses Image Tweets
Code Code Frequency Code Percentage
Admiration 444 31.71%
Noted NAP supporters 121 8.64%
Noted NAP protesters 264 18.56%
Historical figures 59 4.21%
Sympathy 274 20.57%
Minority in pain 29 2.07%
Murder victims/ calls to end unjustified killings 89 6.36%
Bystanders in protest (non-P.O.C) 186 13.29%
Other 6 0.43%
Out of the top most retweeted image tweets, three of those included the sub-code of admiration as its focal point (see Figures 13 - 15). The first image is of Colin Kaepernick on the cover of Time magazine (Figure 13), thus the subject of admiration, which had 99,000 retweets.
Time magazine is quite an influential as it has the world’s largest circulation for a weekly magazine with a readership of 25 million, 20 million within the United States alone (Time Magazine, 2018). Thus, the large amount of retweets that the image of Kaepernick on the cover of Time magazine received, might be attributed to the magazines own popularity in the U.S.
With 35,000 retweets, the second picture of protesters is from the 1960’s in the United States during the Civil Rights Movement (Figure 14). This image tweet shows an appreciation of
‧
國立 政 治 大 學
‧
N a
tio na
l C h engchi U ni ve rs it y
peaceful means. This is especially important within the context of the NAPs as they are also non-violent protests used to gain awareness for issues that minority members face in modern day American.
The last image (Figure 15), with 4,800 retweets, is of a modern day volleyball player who is the only individual on her team engaging in protest activities during the playing of the national anthem. Support for the NAPs and admiration for the volleyball player who kneeled when no one else did, was gained through the context and framing of the image tweet. However, it should be noted that “Those engaging in protest activities being recognized,” as a type of admiration was depicted in the image tweets more often than “Noted NAP supporters” or “Historical figures involved in social movements or social justice (i.e., Martin Luther King Jr.).”
Figure 13. Praise for Colin Kaepernick on the cover of Time magazine (2017) Tweet text:
#Kaepernick started kneeling in protest during Obama's presidency.
This wasn't abt Trump until Trump tried to make it abt Trump.
#TakeAKnee
‧
國立 政 治 大 學
‧
N a
tio na
l C h engchi U ni ve rs it y
Figure 14. Admiration for historical 1960’s Civil Rights Movement protesters (2017)
Figure 15. Volleyball player praised for kneeling during national anthem. (2017)
4.4 Efficacy coping approach: instrumental Route
RQ4 attempts to understand the key visual themes in image tweets terms of efficacy-eliciting responses (i.e., group size, protest activities) during the 2017 NAPs. Images that included group size were coded as either “football players” or “non-football players.” This was to account for different types of groups present within the NAPs as organizations and unions
Tweet text:
Tweet text: