Looking Back as We Prepare to Move Forward: US Presidential Candidates' Adoption of YouTube

Abstract

As the United States prepares for the 2012 Presidential Election, we reflect on how the 2008 candidates and the voting public adopted social computing technology with specific focus on YouTube. Using the Social Influence Model as a framework, this article sought to examine how candidates adopted YouTube to determine what types of messages were disseminated via YouTube and what kind of response from viewers videos received. We conducted a content analysis of candidate YouTube sites, focusing on video characteristics and content. Findings indicate that the candidates did not use YouTube to its fullest extent and that there were significant differences between the candidates in terms of the types of videos posted and the content posted and that constituents did adopt the medium as demonstrated in viewership numbers and responses.

Introduction

Labeled “the first ‘YouTube election,’” 2008 was the first time US presidential candidates had the opportunity to utilize YouTube since its 2005 inception.¹ For candidates, YouTube is cost-efficient compared to television advertising, which reached $2.2 billion during the 2008 election.² Furthermore, because YouTube is available to anyone with an Internet connection, messages can be easily accessed by not only supporters, but also crucial undecided voters. With America on the verge of its next presidential election, it is informative to examine how candidates and the public adopted this social computing resource.

The Adoption of YouTube in Politics

YouTube provides a means of sharing many types of video content.³ Although the site thrives on the posting of television content, it has become an outlet for nonprofessionals to post video messages as well. A popular form of expression is the videoblog or vlog (monologues delivered facing the camera, sometimes including editing for added interest).⁴

The Social Influence Model is an exemplar of technology adoption applied to social computing.⁵ Four antecedents (social computing action, social computing consensus, social computing cooperation, and social computing authority) combine to create social influence. As more people learn of the social computing action, social influence becomes stronger and more people adopt the technology.⁵ This model can be applied to YouTube; as more people become involved, influence grows, and thus more people adopt the technology. Vannoy and Palvia acknowledge that the area of politics is increasingly affected by online communities and bloggers.⁵ With growing social influence, it is not surprising that political candidates would want to adopt the technology to promote themselves to potential voters and that constituents would want to adopt the technology for insight into political races. What is unknown, though, is how candidates are adopting this medium and whether constituents are participating in the process.

Politicians must carefully control their image on YouTube as it is not forgiving to those who make mistakes.⁶ Hediger uses as an example the 2006 win of Senator Jim Webb of Virginia over incumbent George Allen. A Webb campaign employee taped an Allen speech in which Allen referred to the cameraman using racial slurs, and then posted the video to YouTube. It is speculated that this video helped Webb win this close race.⁶ As Fernandez reported, videos uploaded to YouTube during the 2006 election cycle impacted the results of at least two Senate races: candidates who were portrayed deleteriously (e.g., delivering damaging remarks, falling asleep during a congressional hearing) lost their races.⁷ In light of such instances, Kranish noted that candidates must control their messages as such gaffes are difficult, if not impossible, for campaigns to reconcile.⁸

Likewise, candidates may choose to control positive and negative messages. Benoit found that candidates tend to disseminate positive messages about themselves more than attack messages about their opponents⁹ because attacks can backfire, hurting the attacker and potentially demobilizing voters.¹⁰ Haynes and Rhine noted that candidates are more likely to use negative messages during high-stakes events and delegate-rich contexts, and candidates are more likely to attack when they themselves have been attacked or when losing ground in the election.¹¹

Examining various types of recent elections, researchers have noted that most of the videos posted on YouTube were traditional advertisements or national television program segments.¹² Harp and Tremayne's study of YouTube videos posted during the 2006 US election categorized videos into five genres: originally televised political messages, new political Web messages (i.e., nontelevised content), citizen-produced alternative political content, ambushes and unscripted candidate scenes, and mass media-produced content. New political Web messages were found in merely 1.5 percent of the videos.¹³

Similar to the Harp and Tremayne study, this study examines candidate use of YouTube during the 2008 US presidential election.

RQ1: What types of videos were posted by the candidates?

RQ2: What was the valence of the videos posted by the candidates?

YouTube also provides the online community opportunity to respond to videos by viewership and responses. Online technologies redefine democratic engagement because they contain community building features that generate participation and potential mobilization.¹⁴

RQ3: What kind of response did these videos receive from viewers?

Method

To focus on the candidates as nominees of their respective parties, the sampling frame for video inclusion was from the end of each party's national convention to the evening before the election (Obama, August 29, 2008 to November 3, 2008; McCain, September 5, 2008 to November 3, 2008). This resulted in 622 videos posted by Obama and 72 by McCain. To ensure that the samples were comparable and equal in terms of error estimated, we used an equation that estimates sample sizes for the tolerated “margin of error” for a proportion, with a “finite population correction factor.”¹⁵ As such, 238 Obama videos and 61 McCain videos were randomly selected. As with previous YouTube studies, the unit of analysis was the video.^13,16,17

A codebook was created detailing how to code video type and valence (positive/negative). Video length, number of views, viewer rating, number of raters, number of times the video was favorited, number of text comments, and number of video responses were pulled directly from YouTube.

Video type included such codes as advertisement, news clip, and speech (see Table 1 for a full list). Video valence was coded as positive, negative, both, or neutral. All items were coded as present or absent in the video, but if one of the codes was marked as present, all others had to be coded as absent; therefore, all codes were mutually exclusive. Code categories for these items came from previous literature^18–23 as well as repeated viewing of video postings during the primaries.

Table 1.

YouTube Video Type Frequencies and χ ² Results

Category	Crosstab results	McCain % (n)	Obama % (n)
Ad
Yes	χ²=115.05, p<0.001	77.0 (47)	10.9 (26)
No		23.0 (14)	89.1 (212)
News clip
Yes	χ²=0.78, FET=1.000	0.0 (0)	1.3 (3)
No		100.0 (61)	98.7 (235)
Speech
Yes	χ²=18.60, p<0.001	1.6 (1)	27.3 (65)
No		98.4 (60)	72.7 (173)
Edited speech
Yes	χ²=2.93, FET=0.128	0.0 (0)	4.6 (11)
No		100.0 (61)	95.4 (227)
Press conference
Yes	χ²=0.312, FET=0.497	1.6 (1)	0.8 (2)
No		98.4 (60)	99.2 (236)
Celebrity endorsement
Yes	χ²=8.23, p<0.01	0.0 (0)	12.2 (29)
No		100.0 (61)	87.8 (209)
Testimonial
Yes	χ²=1.20, FET=0.270	3.3 (2)	1.3 (3)
No		96.7 (59)	98.7 (235)
Audio-only Ad
Yes	χ²=36.21, FET=0.000	14.8 (9)	0.0 (0)
No		85.2 (52)	100.0 (238)
Vlog
Yes	χ²=3.48, FET=0.078	0.0 (0)	5.5 (13)
No		100.0 (61)	94.5 (225)
Doc-like video
Yes	χ²=16.22, p<0.001	1.6 (1)	24.8 (59)
No		98.4 (60)	75.2 (179)
Town hall
Yes	χ²=1.04, FET=0.585	0.0 (0)	1.7 (4)
No		100.0 (61)	98.3 (234)
Animation
Yes	χ²=2.65, FET=0.223	0.0 (0)	4.2 (10)
No		100.0 (61)	95.8 (228)
Debate
Yes	χ²=1.57, FET=0.353	0.0 (0)	2.5 (6)
No		100.0 (61)	97.5 (232)
Other
Yes	χ²=1.84, FET=0.351	0.0 (0)	2.9 (7)
No		100.0 (61)	97.1 (231)

df for all analyses is 1. FET is indicated for analyses in which one or two cells had an expected cell count less than 5.

FET, Fisher's exact test.

Two independent coders were trained and tested the codebook. Thirty percent of the sample was used to generate intercoder reliability (Krippendorff's alpha for all codes was 1.00). The remaining videos were coded by these two trained coders.

Results

Cross tabulations were performed to assess significant differences using χ² for frequency data (2×2 analysis, present/absent x campaign; the Fisher's exact test was reported when expected cell totals were less than five). The t-Tests and analysis of variance were used for ratio-level data.

McCain posted significantly more advertisements and audio-only advertisements than Obama, while Obama posted significantly more speeches, celebrity endorsements, and documentary-like videos than McCain (RQ1; Table 1). Obama tended to post videos that did not fall into traditional types of presentations, but appeared to be created for Web presentation (e.g., documentary-like videos, animations, and vlogs). McCain did not post any vlogs, whereas Obama posted 13. Only 2 of the 13 featured Obama; others featured Obama family members and campaign affiliates. All vlogs focused on volunteerism, knocking on doors, calling neighbors, and getting out to vote.

McCain posted significantly more negative videos than Obama, while Obama posted significantly more positive videos and neutral videos than McCain (RQ2; Table 2). There was no significant difference in the number of videos posted containing both positive and negative messages.

Table 2.

YouTube Video Valence Frequencies and χ ² Results

Category	Crosstab results	McCain % (n)	Obama % (n)
Only candidate-positive messages
Yes	χ²=6.25, p<0.05	19.7 (12)	36.6 (87)
No		80.3 (49)	63.4 (151)
Only opponent-negative messages
Yes	χ²=50.27, p<0.001	54.1 (33)	12.6 (30)
No		45.9 (28)	87.4 (208)
Both positive and negative messages
Yes	χ²=2.77, p=0.096	18.0 (11)	28.6 (68)
No		82.0 (50)	71.4 (170)
Neither positive nor negative messages
Yes	χ²=6.15, p<0.05	8.2 (5)	22.3 (53)
No		91.8 (56)	77.7 (185)

df for all analyses is 1. FET is indicated for analyses in which one or two cells had an expected cell count less than 5.

Significantly more people viewed, rated, made text comments on, and created video responses for the McCain videos than the Obama videos (RQ3). However, the Obama videos were rated significantly higher than the McCain videos. There was no difference in the number of favorite videos (Table 3). Also, significantly more people viewed, rated, made text comments on and created video responses for the negative videos compared to the other three types (positive, both, and neither), yet the negative videos were rated significantly lower than the other three (Table 4).

Table 3.

t-Tests of Viewer Variables

Variables	Campaign	M	SD	t-Test	p
Views				3.64	0.000
	McCain	118,198.93	133,392.33
	Obama	48,124.68	136,977.64
Rating				−26.56	0.000
	McCain	2.97	0.51
	Obama	4.81	0.38
Number of raters				4.75	0.000
	McCain	1,550.21	1,742.40
	Obama	446.68	1,004.20
Favorited				0.30	0.761
	McCain	147.23	162.24
	Obama	126.37	528.36
Text comments				6.06	0.000
	McCain	2,300.80	2,907.19
	Obama	46.22	107.77
Video responses				3.73	0.000
	McCain	8.48	15.84
	Obama	0.62	8.90

Table 4.

Analysis of Variance of Viewer Variables by Video Valence

Variables	Video valence	M	SD	F	p
Views
	Positive	52,445.95	181,968.43	4.41	0.005
	Negative	114,102.16	139,895.44
	Both	58,799.85	105,033.19
	Neither	28,242.02	57,280.49
Rating
	Positive	4.60	0.69	17.58	0.000
	Negative	3.80	1.01
	Both	4.56	0.74
	Neither	4.68	0.71
Number of raters
	Positive	600.76	1,427.37	4.32	0.005
	Negative	1,098.48	1,632.03
	Both	696.35	993.19
	Neither	296.24	567.29
Favorited
	Positive	172.73	788.44	0.84	0.473
	Negative	125.56	115.85
	Both	141.87	243.81
	Neither	48.93	118.33
Text comments
	Positive	193.95	584.39	8.87	0.000
	Negative	1,380.98	2,747.48
	Both	378.62	1,328.27
	Neither	262.69	982.15
Video responses
	Positive	1.71	13.82	3.37	0.019
	Negative	6.05	15.22
	Both	0.63	3.34
	Neither	1.10	5.38

df=3, 298.

Discussion

As Vannoy and Palvia suggest, the political arena is an area that can benefit from being proactive in utilizing trends in social computing.⁵ YouTube is one such trend, and our findings indicate that candidates and constituents adopted YouTube for distributing and viewing political content.

McCain's site contained significantly more advertisements than the Obama site (RQ1). However, Obama posted several different types of content, including speeches, celebrity endorsements, and documentary-like videos, which is consistent with reports that suggested Obama's success was, in part, due to his campaign's social-networking savvy.²⁴ Of particular interest as it relates to YouTube is the posting of vlogs, a format that is quite familiar to YouTube users, yet neither candidate fully utilized this method of engagement. McCain included no vlogs in his YouTube site, whereas Obama posted 13, and only two featured Obama himself. Vlogs invite feedback, especially in the form of criticism, debate, and discussion.⁴ If candidates wish to utilize this medium to its fullest potential, then the vlog is a format that deserves careful consideration.

Both candidates posted positive and negative messages in an attempt to have voters regard them positively, yet also be motivated to vote for them (RQ2).^25–27 McCain posted significantly more negative videos than Obama, and Obama posted significantly more positive videos than McCain. It has been noted that incumbent candidates use more positive acclaim messages about themselves than attacks on their opponent, while challengers use more attacks on their opponent than positive acclaims about themselves.^9,28–31 However, as Haynes and Rhine noted, candidates are more likely to attack when they feel desperate that they are losing ground in the election.¹¹ With Obama's appeal to younger voters, use of celebrities, and message of “Hope,” “Change we need,” and “Yes we can,” to mobilize voters, McCain may have felt the need to attack, and thus disseminate negative messages. Furthermore, Brody and Sigelman noted that the performance of an incumbent president and his party can affect the results of an election, even when the incumbent president or vice president is not running in the current election.³² It is possible that McCain tried to distance himself from Bush (whose presidency ended relatively low in popularity) by focusing on attacks as positive acclaims could have been perceived as “more of the same.” Yet for Obama to keep his message of “Need for Change” and hope consistent, he had to focus on positive messages.

People turned to YouTube to take part in the online communities created by McCain and Obama (RQ3). Viewership was high for both candidates, though higher for McCain. Participants on the McCain site tended to be more active than those on the Obama site as there were more viewers, video raters, text comments, and video responses than on the Obama site. However, videos on the Obama site were rated higher than those on the McCain site. The numbers demonstrate that candidate YouTube sites provided an opportunity for political connectedness and democratic engagement.¹⁴ Furthermore, viewers were able to actively engage the political process by rating videos and responding with text or video responses (though few created videos). Furthermore, results indicate that although people viewed and responded to the negative videos more than positive, neutral, and mixed videos, they rated the negative videos lower than the other types, suggesting that negative videos may create buzz, but are not necessarily well-liked.

Conclusions

As we look forward to the 2012 US Presidential Election, this study brings to light how candidates of the 2008 Presidential Election utilized YouTube for the same purpose, but in different ways. Although neither of them embraced it fully (especially with regard to vlogs), Obama posted more videos in more varied formats. YouTube allows candidates to directly address their voters in a friendly, simulated “face-to-face” format and to post speeches, town hall meetings, or any type of staged voter contact that the majority of people would not otherwise have the opportunity to see because these events are not always covered by traditional media. Obama's campaign took advantage of this more than McCain's campaign and posted a larger variety of videos, whereas McCain mostly reposted standard television ads. Furthermore, viewers were actively engaged with the community by viewing and responding to videos. Elements of the social influence model can be seen in the use of YouTube during the 2008 election, and as social influence grows with the pending 2012 election, we should see further adoption of the site by both candidates and viewers.

Footnotes

Author Disclosure Statement

No competing financial interests exist for both authors.

References

Center for Communication & Civic Engagement. YouTube election 2008: a study on how YouTube is influencing the 2008 presidential election. 2008. http://ccce.com.washington.edu/projects/youtubeElection2008.html. 2008 Nov. 19.

Atkinson

. 2008 political ads worth $2.5 billion to $2.7 billion. Broadcasting & Cable. Dec. 2, 2008. www.broadcastingcable.com/article/160115-2008_Political_Ads_Worth_2_5_Billion_to_2_7_Billion.php. 2012 Mar. 25.

Levinson

. 2009. New new media. Boston: Allyn & Bacon.

Burgess

, Green

. 2010. The entrepreneurial vlogger: participatory culture beyond the professional-amateur divide. The YouTube Reader. Sweden: National Library of Sweden.

Vannoy

, Palvia

. The social influence model of technology adoption. Communications of the ACM, 2010; 53:149–153.

Hediger

. 2010. Aesthetics of political accountability. The YouTube Reader. Sweden: National Library of Sweden.

Fernandez

. Uploading American politics. Washington Post, Dec. 9, 2006; A19.

Kranish

. Internet and politics an uneasy fit: campaigns strain to control message. Boston Globe, May 10, 2007; A1.

Benoit

. A functional analysis of political advertising across media, 1998. Communication Studies, 2000; 51:274–295.

10.

Ansolabehere

, Iyengar

, Simon

et al.

Does attack advertising demobilize the electorate?

American Political Science Review, 1994; 88:829–838.

11.

Haynes

, Rhine

. Attack politics in presidential nomination campaigns: an examination of the frequency and determinants of intermediated negative messages against opponents. Political Research Quarterly, 1998; 51:691–721.

12.

Bryant

. The 2006 midterms according to YouTube. 2006. www.reelpopblog.com/2006/10/the_2006_midter.html. 2008 Nov. 19.

13.

Harp

, Tremayne

. Programmed by the people: the intersection of political communication and the YouTube generation. Paper presented at 57th International Communication Association Conference, 2007 San Francisco, CA.

14.

Ricke

. A new opportunity for democratic engagement: the CNN-YouTube presidential candidate debates. Journal of Information Technology and Politics, 2010; 7:202–215.

15.

Triola

. 2001. Elementary statistics, 8th. Boston, MA: Addison Wesley Longman.

16.

Carlson

, Strandberg

. Riding the Web 2.0 wave: candidates on YouTube in the 2007 Finnish national elections. Journal of Information Technology and Politics, 2008; 5:159–174.

17.

Milliken

, Gibson

, O'Donnell

et al. User-generated online video and the Atlantic Canadian public sphere: a YouTube study. Paper presented at 58th International Communication Association Conference, 2008 Montreal, Quebec.

18.

Johnston

, Kaid

. Advertising: using videostyle to explore stylistic differences in televised political ads from 1952 to 2000. Journal of Communication, 2002; 52:281–300.

19.

Petrocik

. Issue ownership in presidential elections, with a 1980 case study. American Journal of Political Science, 1996; 40:825–850.

20.

Petrocik

, Benoit

, Hansen

. Issue ownership and presidential campaigning, 1952–2000. Political Science Quarterly, 2003; 118:599–626.

21.

Prior

. Weighted content analysis of political advertisements. Political Communication, 2001; 18:335–345.

22.

Spiliotes

, Vavreck

. Campaign advertising: partisan convergence or divergence? The Journal of Politics, 2002; 64:249–261.

23.

Wadsworth

. Incumbent and challenger strategies in presidential communication: a content analysis of television campaign ads from 1952–1984 [dissertation] University of Oklahoma, 1986.

24.

Talbot

. How Obama really did it. Technology Review, 2008; 111:78–83.

25.

Benoit

, Pier

, Blaney

. A functional approach to televised political spots: acclaiming, attacking, defending. Communication Quarterly, 1997; 45:1–20.

26.

Kahn

, Kennedy

. Do negative campaigns mobilize or suppress turnout? clarifying the relationship between negativity and participation. American Political Science Review, 1999; 93:877–889.

27.

Martin

. Inside the black box of negative campaign effects: three reasons why negative campaigns mobilize. Political Psychology, 2004; 25:545–562.

28.

Benoit

. The functional approach to presidential television spots: acclaiming, attacking, defending 1952–2000. Communication Studies, 2001; 52:109–126.

29.

Benoit

, Brazeal

. A functional analysis of the 1988 Bush-Dukakis presidential debates. Argumentation and Advocacy, 2002; 38:219–233.

30.

Benoit

, Harthcock

. Functions of the great debates: acclaims, attacks, and defenses in the 1960 presidential debates. Communication Monographs, 1999; 66:341–357.

31.

Brazeal

, Benoit

. A functional analysis of congressional television spots, 1986–2000. Communication Quarterly, 2001; 49:436–454.

32.

Brody

, Sigelman

. Presidential popularity and presidential elections: an update and extension. The Public Opinion Quarterly, 1983; 47:325–328.