New Analysis: Speaking Time and Camera Shots in the August 6 Republican Presidential Primary Debate
Patrick Stewart, associate professor of political science at the University of Arkansas, analyzes non-verbal communication in politics with a focus on presidential primary debates.
His latest analysis, with graduate student Austin Eubanks, of the Aug. 6 FOX News presidential primary debate provides context and data useful in coverage of the upcoming debate on Sept. 16. The analysis details who dominated the debate and how camera angle and speaking time played a role in audience perception.
“What is most interesting is our finding that when the debate was analyzed in depth, there was a marked difference not only in how much each of the Republican Party candidates spoke, but the number of speaking turns by the GOP presidential contenders,” Stewart wrote.
In addition to his book Debatable Humor: Laughing Matters on the 2008 Presidential Primary Campaign (2012), Stewart is a certified Facial Action Coding System (FACS) coder and is researching how individuals respond to different types of facial displays by presidential candidates.
Nonverbal analysis of the FOX News Aug. 6, 2015 Republican Party presidential debate: Speaking time and camera shots.
The August 6, 2015 FOX News GOP presidential primary prime time debate was watched by an unprecedented number of viewers. Since its airing, it has been commented upon extensively in the media. To better understand network production choices, which influence how the candidates present themselves, we carried out a frame-by-frame content analysis of speaking time and camera shots concerning the ten candidates. Major findings concerning candidate speaking time given indicate:
- Donald Trump dominated candidate speaking time during the debate with eleven minutes (16%) of the total time and twenty-eight speaking turns (24.5%). In comparison, fellow front runner Jeb Bush had seven minutes (10.2%) speaking time from eight (7%) speaking turns.
- Top tier candidate Scott Walker used only 68.5% of his allotted speaking time.
- Late qualifier John Kasich went over his allotted time by 70.7 seconds
Using five camera shot categories (head-and-shoulders of a single candidate, candidates physically side-by-side, or via a split screen, multiple candidates, and the audience) our major findings concerning camera shots show:
• The majority of camera shots (85.1%) were of a single candidate while nearly one-fifth (19.1%) put candidates side-by-side in a competitive position.
When comparing camera shot choices concerning candidates, three clusters stand out:
- Low competition candidates where nearly 95% or more of camera time shows onecandidate only. Ted Cruz (95.9), Scott Walker (95.8%), and Marco Rubio(94.4%) fall into this category.
- Moderate competition candidates with nearly ten percent of camera time side-by-side with other candidates. Four candidates fit this category: Jeb Bush (11.5%),Mike Huckabee (9%), Ben Carson (9.3%), and John Kasich (9.3%).
- High competition candidates having from nearly one-fifth to one-third of cameratime presented side-by-side with other candidates. Donald Trump (19%), Chris Christie (30%), and Rand Paul (31.3%) fall into this category.
The FOX News prime time Republican Party presidential debate, watched live on television by 24 million viewers, and numerous others through simulcast video streams (2.5 million) and afterwards (8 million video streams), provides ample evidence of public interest in the 2016 presidential election. Not only did the debate surpass all previous presidential primary election debates in terms of viewership, it also approached being the most watched cable event ever, second only to the 33.6 million viewers of the 2014 collegiate football BCS National Championship Game between the Ohio State University and the University of Oregon.
In addition to the debate’s extremely large viewership, second screen involvement during and in the immediate aftermath of the debate was exceptionally high, indicating robust connection between candidate performance and viewer reactions on their mobile devices . Presidential debate co-sponsor Facebook reported 7.5 million people commenting on the debate with twenty million posts. Likewise, Twitter noted 3.3 million tweets and 1.1 billion viewings of these comments on the social media site.3 In summary, not only was the debate a highly anticipated and watched event stimulating extensive general public interest, it has been the focus on ongoing media commentary concerning its effect on candidate prospects.
Due to the persistence of the first prime time debate in influencing the perceived electoral fates of the GOP candidates, this research project considers the factors occurring during debates that influence mass media and public perceptions. While candidate performance through their verbal responses and nonverbal presentation  affects audience response , the choices made by the debate producers and moderators play a major role in public perception. Specifically, how the networks present the candidates affects how the candidates can present themselves. While the questions asked can be subtle, they can act as powerful media primes influencing audience opinion and judgments [4,5]. As part of that, the speaking time given the candidates influences what positions and arguments can be presented and how extensive and thoughtful explanations can be. Therefore, our first research goal is to consider the speaking time given to the GOP candidates during the FOX News debate, taking into account the number and length of speaking turns as well as the proportion of time each candidate received.
At the same time, the camera shots – in other words the “image bites”  – influence how the candidates are observed and received by those viewing the debates at home. Specifically, media presentation of candidates often places them virtually face-to-face with viewers in a manner that is artificially intimate [7,8]. This “in your face politics” can lead to either greater connection or antagonism between the viewer and the candidate on screen, depending on pre-existing opinions and the proximity of the image [9,10]. Furthermore, camera shots can heighten the level of conflict discerned by viewers, as is seen with split screen techniques used during presidential general election debates [11- 13]. Having contending candidates side-by-side “presents the debate as a contest between opponents who display their contempt and disagreement for one another with every nonverbal, off-handed gesture, inaudible sigh, and shift in body language.”  (245) This provides us our second research goal, which concerns how the GOP candidates are portrayed on the television screen in terms of the amount, type and proportion of camera shots broadcast over the course of the debate.
Therefore, despite presidential debates being seen as the least mediated and hence one of the most naturalistic televised events, offering glimpses into the “real presidential candidates” of the political parties, the news networks broadcasting these proceedings still exert subtle, yet influential, power over how the candidates are perceived by the viewing public at home. In the research presented below, we analyze the first major debate of the 2016 presidential primary – the August 6, 2015 Republican Party prime time debate on the FOX News cable network. We do so based upon speaking time and camera shots conferred the top ten Republican Party presidential candidates. In addition to aggregate findings, we consider the distribution of coverage between the candidates, providing us insight into the production decisions made, whether consciously or subconsciously, and how these judgments interact with candidate behavior concerning their use of the time allotted.
Candidate speaking time
The most obvious means by which a candidate’s electoral position is communicated during debates is how much speaking time they receive in comparison with their fellow candidates. In the FOX News debate, we coded speaking turns based upon the utterances of the candidates. These utterances ranged from brief interruptions and interjections lasting less than four seconds, and which comprised just over a quarter of the speaking turns (27%), to longer statements of a minute to nearly a minute and a half where the candidates asserted and/or defended their politics and policy positions. While there was a small proportion of overlap, especially during heated encounters between the candidates, most of the 115 speaking turns we coded were relatively long and uninterrupted.
In terms of overall time, nearly sixty-nine minutes (4,135 seconds) of the one hour and forty-nine minutes (6,552 seconds) of debate programming were available for candidates to speak. In other words, 63.1% of the total program time was provided to the ten GOP candidates, with the remainder of the time being taken by either the moderators asking or setting up questions, audience response in terms of applause, laughter, and/or boos, and transition time between debate segments.
What is most interesting is our finding that when the debate was analyzed in depth, there was a marked difference not only in how much each of the Republican Party candidates spoke, but the number of speaking turns by the GOP presidential contenders.
As expected, real estate mogul and television personality Donald Trump dominated discourse during the debate. His eleven minutes speaking time accounted for 16% of total time given the candidates, dwarfing the seven minutes (10.2%) provided his nearest competition, ex-Florida Governor Jeb Bush. Part of this may be due to the contentious nature of Trump’s presence, as his twenty-eight speaking turns was three times as many as Bush’s eight speaking turns, and accounted for nearly a quarter of total candidate speaking turns (24.5%). The third candidate in the top tier, Wisconsin Governor Scott Walker, had more speaking turns (twelve) but spent just under six minutes speaking with the second lowest percent (8.5%) of speaking time. Cursory analysis of the speaking turns suggests Walker was much more succinct with his statements, rarely taking more than forty-five seconds to respond to the FOX News moderators. In other words, Walker used well under three-quarters (68.5%) of this allotted speaking time. For comparison, the other candidates often spent over a minute answering their questions.
Table 1: Speaking turns and time per candidate
Candidate Speaking turns % Speaking Turns Speaking time (secs) %Speaking time
Bush 8 7.0% 421.08 10.2%
Carson 9 7.8% 395.73 9.6%
Christie 12 10.4% 387.84 9.4%
Cruz 8 7.0% 400.57 9.7%
Huckabee 8 7.0% 399.91 9.7%
Kasich 6 5.2% 400.7 9.7%
Paul 15 13.0% 327.56 7.9%
Rubio 9 7.8% 389.3 9.4%
Trump 28 24.3% 660 16.0%
Walker 12 10.4% 352.75 8.5%
Total 115 100.0% 4135.44 100.0%
With the exception of the disproportionate amount of speaking time given to Donald Trump, there was a relatively equal amount of speaking time dispersed by the moderators over the course of the debate and ranging from nine-to-ten percent of the total. Specifically, only just over a minute and a half (93.52 seconds) separated the second greatest amount of time (Bush) and the lowest amount of time allotted (Paul). In other words, each candidate had between six and a half to seven minutes to make their points and to leave an impression on the voters.
However, the approach taken by the candidates, and abetted by the FOX News moderators, varied. For instance, while hometown candidate and Ohio Governor John Kasich had nearly the same amount of speaking time as Senator Ted Cruz and ex- Arkansas Governor and FOX talk show host Mike Huckabee, he had only six speaking
turns compared with eight for the latter two. When Kasich’s speaking turns were analyzed in greater depth, it was found he went over his allotted speaking time by 70.7 seconds. What this indicates is that he was able to speak for a longer period of time on average (Kasich Average = 66.78 seconds vs. Debate Average = 35.96 seconds), despite his barely breaking into the top ten GOP candidates eligible for this prime time debate.
While most candidates had between six and nine speaking turns, three candidates had proportionally more turns, largely due to interruptions and/or interjections lasting less than four seconds. Senator Rand Paul (fifteen speaking turns) and New Jersey Governor Chris Christie (twelve speaking turns) – who proved to be combative towards each other – had more speaking turns, likely due to their public reputation as being contentious and aggressive individuals. Specifically, of his fifteen turns Paul had five short utterances, while of the twelve turns each for Chris Christie and Scott Walker, Christie had three and Walker had two.
Candidate “image bites”
As discussed in our introduction, even such subtle factors as the choice of camera shot has the potential to influence the audience’s perceptions of the candidate due to the “image bite” being consumed by the audience . As a result, negotiations between candidates, political parties, and the networks broadcasting the debate over camera angles and shots are often tense and highly contested, with debate contracts many pages long .
Specifically, it has been well established that the most valid and reliable sign of group dominance is the attention an individual gets from the other group members [14-17]. In other words, if we are looking at someone, they must be important. Therefore, whether or not a candidate is seen on television, and how often and much we see them, is an indicator of their importance as a potential leader . Thus, the production decisions made by network producers and moderators influence who is perceived as the most likely leader just as much as the behavior by the candidate during the time allotted them influences their audience.
Total candidate “image bites” during the FOX News debate
Our analysis found that of the hour and forty-nine minute broadcast (6,552 seconds), just under an hour and twenty minutes (4,767 seconds), i.e., almost three-quarters of the time available (72.8%) focused the camera on the candidates while they were either listening to or responding to questions and attacks from the moderators and other candidates. A total of 256 shots ranging from a third of a second (0.33s) to nearly a minute and a half (88.9) focused on the candidates in one form or another. It should be noted that while the greatest proportion of time was focused on individual candidates, when two candidates were presented either side-by-side or via split screen, the time was double counted with each candidate accrued time on camera. Specifically, there was an estimated 282 seconds shared (563.91/2).
“Image bites” by candidate
As can be expected, the top two candidates in terms of face time given by FOX News to all the GOP candidates during the debate were front-runners Donald Trump and Jeb Bush. However, Trump arguably received a disproportionate amount of screen time with almost one-fifth of camera time given the candidates (18.2%) followed by Bush receiving just above twelve percent (12.4%). The proportion of time given the candidates from there diminished to just below ten percent for Kasich (9.7%) and Christie (9.2%). Carson and Cruz (both 8.9%) received near equal screen time, as did, albeit at a lower proportion of the total, Walker, Rubio, and Paul (8.3%). Finally, Huckabee received just below eight percent (7.8%) of the camera time available.
Camera angle composition
Although just having the camera trained on a candidate affects how he (or she) is perceived, the camera shot tells different stories that influence viewer perceptions. During this debate we noted five distinct camera shots: 1) head-and-shoulders shots of a single candidate; 2) two candidates side-by-side physically or; 3) via a split screen; 4) multiple (if not all) candidates in view, and; 5) an audience reaction shot involving either a segment or a panning shot of the entire audience.
A candidate shown from a perspective of just their head-and-shoulders (Figure 1a) allows for that individual to dominate viewers’ perceptions, allowing them to make their point without any diverting of attention [6,18]. When placed in juxtaposition with other candidates, whether side-by-side or as a group, the emphasis becomes one of visual comparison. Camera shots with multiple candidates in view put any one given candidate in the position of being perceived by viewers as solely one contender amongst many (Figure 1b). As noted earlier, camera shots that place the candidates side-by-side, either through the juxtaposition of them physically (Figure 1c), or via split-screen (Figure 1d), places them in direct competition with each other . In the case of the former, having both candidates in frame allows for direct comparisons to be made in terms of their physical stature as well as their body language in response to each other [19,20]. In the case of split screen analysis, astute minute comparisons may be made concerning their facial displays in response to each other .
Finally, audience shots – depending on the reaction of the audience itself – during the candidate’s response may be used to reflect support or antagonism towards the candidate, depending on whether they are accompanied by laughter and applause with the former, or boos in the latter case.
“Image bite” composition during the FOX News debate
While there were many different camera shots during the FOX News debate that placed multiple candidates in the same shot, the great majority (85.1%) focused on head-and- shoulders shots of single candidates. The more comparative camera angles, including screens split with both candidates in view, and those with the contenders side-by-side, accounted for just over eleven percent of the shots (11.2%). More specifically, split screen shots comprised nearly nine percent of the camera shots (8.6%) and just over two and a half percent (2.6%) of the shots occurred with the candidates presented side-by- side. The remaining 1.4% focused on showing the audience.
|Type of camera shot||Camera shots||% Camera shots||Total seconds||% Face time|
|Bush||Candidate’s face & shoulders only||22||64.71||551.49||88.27|
|Candidates split screen||1||2.94||13.15||2.10|
|Carson||Candidate’s face & shoulders only||15||62.50||412.88||91.41|
|Candidates split screen||1||4.17||3.87||0.86|
|Christie||Candidate’s face & shoulders only||11||47.83||337.78||72.35|
|Candidates split screen||9||39.13||120.74||25.86|
|Cruz||Candidate’s face & shoulders only||13||81.25||432.5||95.94|
|Huckabee||Candidate’s face & shoulders only||10||55.56||350.61||89.37|
|Candidates split screen||4||22.22||27.26||6.95|
|Kasich||Candidate’s face & shoulders only||13||54.17||400.46||82.12|
|Candidates split screen||4||16.67||39.33||8.06|
|Paul||Candidate’s face & shoulders only||11||57.89||283.99||67.63|
|Candidates split screen||7||36.84||126.42||30.10|
|Rubio||Candidate’s face & shoulders only||12||75.00||393.25||94.37|
|Candidates split screen||2||12.50||17.95||4.31|
|Trump||Candidate’s face & shoulders only||33||55.00||735.48||79.82|
|Candidates split screen||10||16.67||79.61||8.64|
|Walker||Candidate’s face & shoulders only||13||68.42||399.01||95.75|
|Candidates split screen||2||10.53||4.37||1.05|
|Total||Candidate’s face & shoulders only||153||59.77||4297.45||85.12|
|Candidates split screen||40||15.63||432.7||8.57|
“Image bite” composition by candidate
Analysis of how the ten Republican Party candidates were presented during the FOX News debate reveals differences not just in how long they were on camera, but also how they were presented to viewers. Specifically, analysis of how candidates were visually framed, whether in competitive camera shots with one or more contenders sharing the screen, or in a “leadership” visual frame whereas the candidate was the only person in the shot, suggests there exist three distinct categories by which the ten candidates may be classified.
The first category consists of those candidates presented with a visual mix of “low competition” camera shots. Here, we see Scott Walker (3.62%), Ted Cruz (4.06%), and Marco Rubio (5.01%) spending proportionally less time being featured in competitive camera shots that have one or more competitors present on screen. Instead, the greatest portion of their screen time, at nearly ninety-five percent or higher, consists of head-and- shoulders shots (Figure 1a).
The second category consists of those candidates presented at moderately high rates of competitive television shots, which in this debate consisted of approximately ten percent of camera shots. Four candidates fit this category: Jeb Bush (11.47%), Mike Huckabee (9.72%), Ben Carson (9.36%), and John Kasich (9.25%). Bush is notable for having nearly eight percent of these competitive camera shots in the side-by-side position (i.e., fifty seconds), as he was pitted against Donald Trump for a substantial portion of this time (Figure 1c). Carson’s visual treatment was unique as he had proportionally more multiple candidate shots than all other candidates, and at a greater ratio than shots pitting him against individual candidates. Finally, while Huckabee and Kasich received similar visual treatment, a much greater proportion of Kasich’s time, compared with any of the other candidates, was comprised of audience shots. This is likely due to Kasich’s home state status, and the welcome given him by his fellow Buckeyes.
The third candidate category of “high competition visuals” consists of those candidates portrayed in a much larger percentage of their camera shots alongside other candidates. The proportions range from nearly one-fifth of camera shots including one or more competitors for Donald Trump (19.0%), to nearly of third of all visual shots having a competitor present for Chris Christie (29.95%) and Rand Paul (31.32%). While both Paul and Christies’ belonging to this category are due to their highly contentious exchange concerning privacy rights and the Constitution, Paul distinguished himself with a highly contentious performance almost from the very start of the debate as he attacked Trump’s unwillingness to swear fealty to the Republican Party.
However, the moderators did play a substantial role in asking questions that pitted both candidates against each other. (Figures 1c & 1d) Likewise, front-runner Donald Trump was placed in verbally and, as a result, visually competitive contexts to a much greater extent than the front-running candidates Jeb Bush and Scott Walker. Indeed, Trump was visually matched against all other candidates, with the exception of Ben Carson, Ted Cruz, and Marco Rubio. Trump was either presented in side-by-side shots with fellow front-runners and center stage holders Bush and Walker, or in split screen shots with the remaining lower-tier candidates (Christie, Huckabee, Kasich, and Paul – as well as Walker).
Debates have long been derided for being side-by-side press conferences in which the candidates present positions yet never quite confront each other directly, thus masking their intellectual shortcomings [21-23]. This was not necessarily the case with this debate, especially as FOX News moderators often emphasized political and policy differences between the candidates, pitting them against each other verbally and visually. As a result, viewers had ample opportunities to make direct comparisons between specific candidates not only in terms of their enunciated policy positions, but also their nonverbal style. However, while the policy positions matter, the limited time given each candidate and limited scope of discussion likely did not play as large a role as the automatic, visceral “thin slice” judgments  made concerning the capacity to lead and the intent signaled to Republican Party members .
Additionally, and more subtly, how the media visually presents the candidates affects how the candidates present themselves. While it is up to the candidate to best use their time to connect with the audience, the questions asked and the camera shots taken by the networks producing the debates can artfully define a candidate and his or her role in the debate, and ultimately, the campaign. Whether placing a candidate in the visual and verbal center of contentiousness, as was the case with Donald Trump throughout the FOX News debate, emphasizing audience applause for hometown representative John Kasich, or limiting camera framing to mainly head-and-shoulder shots for Ted Cruz, Marco Rubio, and Scott Walker, the choices made influence public perceptions. While it is doubtful that these choices are consciously malicious towards specific candidates and benign towards others, there is still the potential for visual and verbal bias affecting audience response and media reporting of the event. Further, it should be noted that primary debates are organized and implemented by for-profit private corporations. While public service certainly does play a role in the commission and execution of debates, ratings and network promotion can still be expected to play a major role in production decisions.
Future research is needed to better understand not just the production decisions made concerning speaking time and camera shot choice, but also the effect these decisions have on viewer perceptions and action. In the case of the former, it can be expected that production decisions will vary across the news networks hosting the debates, the political parties and the number of contenders possessed, and the times during the electoral season (pre-primary, primaries, and general election) . While response to nonverbal display behavior by presidential candidates during general election has been studied with some success, both regarding physiological and emotional response [11-13], as well as through second screen interactions with fellow mobile device users , research considering how viewers respond to multiple candidates presented in varying ways has yet to be considered.
In conclusion, while previous research suggests that early primary debates tend to have a large initial effect that diminishes over time , as candidates with greater resource bases are able to endure [28,29], the electoral landscape is evolving with new technologies. Due to the role of social media, with its emphasis on visual learning and political processing , candidate performance is now more important in providing both immediately and lasting impressions with the media and general public [2,31]. As a result, greater understanding and monitoring of media framing techniques, both verbal and visual, should be a continuing part of the political process.
Content analysis methods
The approach taken in this study consisted of multiple steps. The first step was to download the complete video from the FOX News website using clipconverter.cc. The complete video was then loaded into the editing software Adobe Premier Pro and cut into thirteen clips ranging from three and half minutes to almost eleven minutes in length based upon natural stopping points during moderate question set-ups. Markers were set at in and out points in Premier Pro before the files were exported. The video CODEC used was CINEPAK (.MOV) whereas the audio CODEC was IMA4. The CODEC applied allowed for use with the content analysis software ANVIL .
ANVIL data analysis utilized the video clips, which were content coded based upon types of variables put into the specification file. This allowed for output into an annotation file. We considered two primary variables. We first considered speaking time for each candidate/moderators. The second variable considered was face time for each candidate, which consisted of six different types of shots (1 = candidate’s face & shoulders only; 2 = candidate full shot [knee up]; 3 = candidate side-by-side face [split screen]; 4 = candidate side-by-side body; 5 = multiple candidate shot [3+ candidates]; 6 = audience). As a result, we considered 20 variables, which consisted of the ten GOP presidential candidates subdivided into the speaking time and face time for each of the candidates.
Analysis was carried out by exporting each variable separately from each of the thirteen clips, which were then combined into files for each candidate, which were then aggregated in a single Excel spreadsheet, and then transported into SPSS for data analysis. In summary, two data files were analyzed: Candidate speaking time and candidate face time.
Due to the interpretation inherent to coding media, we assessed the internal validity of our findings through inter-coder agreement (or inter-rater reliability). ANVIL offers a built-in function to assess Cohen’s kappa (κ). This measures the level of agreement (for binary or nominal ratings) between two coders’ corresponding annotation files . ANVIL considers time slices (set to 0.04 seconds by default, and used for this research) and compares categories on each time slice. This resulting kappa is used as a measure of how consistent the annotation is. While there is no single benchmark by which the “goodness” of a kappa measure is assessed, Fleiss considers a kappa between 0.40 and 0.60 fair, between 0.60 and 0.75 as good, and over 0.75 as excellent .
For present research, our second coder coded candidate speaking time of two randomly selected video clips (total time 20:14) with each variable considered (e.g., Donald Trump speaking time) having an overall corrected kappa calculated. The results of each variable were then aggregated onto a single table where the range and average of scores could be assessed. The value of kappa for both video clips are well into the excellent range (Clip 4: M(κ)=0.9750; range = 0.9422 to 0.9929 across six speakers; clip 7: (M(κ)=0.9592; range = 0.8446 to 0.9989 across four speakers). Given this level of agreement, we feel internal validity of our measures have been satisfactorily demonstrated.
1. Shah DV, Hanna A, Bucy EP, Wells C, Quevedo V. Examining Social Media Influence: The Power of Television Images in a Social Media Age: Linking Biobehavioral and Computational Approaches via the Second Screen. 2015;659: 225- 307.
2. Stewart PA. Debatable Humor: Laughing Matters on the 2008 Presidential Primary Campaign: Lexington Books; 2012.
3. Stewart PA. Polls and Elections: Do the Presidential Primary Debates Matter? Measuring Candidate Speaking Time and Audience Response during the 2012 Primaries. 2015;45: 361-381.
4. Iyengar S, Kinder DR. News that matters: agenda setting and priming in a television age. 1987.
5. Bucy EP. Nonverbal communication, emotion, and political evaluation. In: Doveling K, von Scheve C, Konijn EA, editors. The Routledge handbook of emotions and mass media. New York, NY: Taylor & Francis; 2011. pp. 195-220.
6. Grabe ME, Bucy EP. Image bite politics : news and the visual framing of elections. Oxford ; New York: Oxford University Press; 2009.
7. Sullivan DG, Masters RD. Biopolitics, the media, and leadership: Nonverbal cues, emotions, and trait attributions in the evaluation of leaders. In: Anonymous; 1994. pp. 237-273.
8. Sullivan DG, Masters RD. ‘Happy Warriors’: Leaders’ Facial Displays, Viewers’ Emotions, and Political Support. Am J Polit Sci. 1988; 32: 345-368.
9. Mutz DC. Effects of “in-your-face” television discourse on perceptions of a legitimate opposition. 2007;101: 621-635.
10. Mutz DC, Reeves B. The new videomalaise: Effects of televised incivility on political trust. 2005;99: 1-15.
11. Gong ZH, Bucy EP. When style obscures substance: Visual attention to display appropriateness in the 2012 presidential debates.
12. Gong ZH, Bucy EP. Image bite analysis of presidential debates. In: Browning: Robert X., editor. The C-SPAN Archives: Advancing the Research Agenda. : Purdue University Press; 2015.
13. Cho J, Shah DV, Nah S, Brossard D. “Split screens” and “spin rooms”: Debate modality, post-debate coverage, and the new videomalaise. 2009;53: 242-261.
14. Chance MRA. Attention Structure as the Basis of Primate Rank Orders. . 1967;2: 503-518.
15. Eibl-Eibesfeldt I. Human ethology. New York: Aldine De Gruyter; 1989.
16. Mazur A. Biosociology of dominance and deference. Lanham, MD US: Rowman & Littlefield; 2005.
17. Salter FK. Emotions in command: Biology, bureaucracy, and cultural evolution. New Brunswick, NJ: Transaction Pub; 2007.
18. Masters RD. The nature of politics. New Haven: Yale University Press; 1989.
19. Murray GR, Murray SM. Caveman Executive Leadership: Evolved Leadership Preferences and Biological Sex. In: Anonymous Evolutionary Psychology in the Business Sciences. Springer; 2011. pp. 135-163.
20. Murray GR, Schmitz JD. Caveman politics: Evolutionary leadership preferences and physical stature. . 2011;92: 1215-1235.
21. Racine Group. White paper on televised political campaign debates. 2002; 38: 199- 218.
22. Benoit WL, Stein KA, Hansen GJ. Newspaper Coverage of Presidential Debates. . 2004; 41: 17-27.
23. Jamieson KH, Waldman P. The press effect : politicians, journalists, and the stories that shape the political world. Oxford ; New York: Oxford University Press; 2004.
24. Borkenau P, Mauer N, Riemann R, Spinath FM, Angleitner A. Thin slices of behavior as cues of personality and intelligence. J Pers Soc Psychol. 2004; 86: 599.
25. Spoor JR, Kelly JR. The Evolutionary Significance of Affect in Groups: Communication and Group Bonding. 2004; 7: 398-412.
26. Dowdle A, Limbocker S, Yang S, Sebold K, Stewart PA. The Invisible Hands of Political Parties in Presidential Elections: Party Activists and Political Aggregation from 2004 to 2012: Palgrave Macmillan; 2013.
27. Damore DF. A dynamic model of candidate fundraising: The case of presidential nomination campaigns. 1997; 50: 343-364.
28. Adkins RE, Dowdle AJ. Continuity and Change in the Presidential Money Primary. 2008; 28: 319-341.
29. Steger WP, Dowdle AJ, Adkins RE. The New Hampshire effect in presidential nominations. 2004;57: 375.
30. Prior M. Visual Political Knowledge: A Different Road to Competence? 2014;76: 41- 57.
31. Donovan T, Hunsaker R. Beyond Expectations: Effects of Early Elections in US Presidential Nomination Contests. 2009;42: 45-52.
32. Kipp M. Multimedia Annotation, Querying and Analysis in ANVIL, chapter 19, Multimedia Information Extraction. In: Maybury MT, editor. Video, Audio, and Imagery Analysis for Search, Data Mining, Surveillance and Authoring. Wiley, IEEE Computer Society Press; 2012. pp. 351-386.
33. Cohen J. A coefficient of agreement for nominal scales. 1960; 20: 37-46.
The authors would like to thank Andrew Dowdle, Markus Koppensteiner, Gregg Murray, Jennifer Stewart, and Launa Stewart for comments on previous drafts. We also would like to thank William Cigainero for his untiring and precise coding.