=Paper=
{{Paper
|id=Vol-1691/paper_08
|storemode=property
|title=Studying the Role of Elites in U.S. Political Twitter Debates
|pdfUrl=https://ceur-ws.org/Vol-1691/paper_08.pdf
|volume=Vol-1691
|authors=Sebastian Stier
|dblpUrl=https://dblp.org/rec/conf/msm/Stier16
}}
==Studying the Role of Elites in U.S. Political Twitter Debates==
https://ceur-ws.org/Vol-1691/paper_08.pdf
Studying the Role of Elites in
U.S. Political Twitter Debates
Sebastian Stier
GESIS Leibniz Institute for the Social
Sciences
Cologne, Germany
sebastian.stier@gesis.org
ABSTRACT 2. THEORY
Because of their ever-growing importance, elite actors from the
political sphere and news media have integrated social network 2.1 Conceptualization
sites and especially Twitter into their communication strategies. The present study concentrates on political and news media actors
However, the extent of these adaptation processes is not yet fully who traditionally occupy the most powerful positions in
understood. This article presents lists of U.S. actors from politics, representative democracy. The definition of actor groups is the
news media and government. As an exploratory analysis, the most critical question when estimating the impact of elites on
influence of elites in U.S. political Twitter debates is investigated Twitter. As elites from political parties, I defined the sitting
by applying basic measures of Twitter influence to two test members of U.S. congress, incumbent governors, national party
datasets. accounts and presidential candidates. In light of the constitutional
separation of powers in the U.S., the group of government actors
Categories and Subject Descriptors is classified separately and is comprised of the President and his
• Applied Computing • Law, social and behavioral sciences social media accounts, the government departments as well as
• Sociology. their respective secretaries.
The definition of news media actors is the most intricate
Keywords conceptual challenge, since the production and dissemination of
online political communication; Twitter; politics; news media;
news is becoming increasingly fuzzy on the social web. Clay
government
Shirky, who is an often-cited source with regard to collaborative
1. INTRODUCTION news production, proposed definitional boundaries between “a
mass amateurization” of news production on the social web and a
Oftentimes it is assumed that the power and agenda-setting role of
“professional class” that “implies specialized functions, minimum
established political and media elites are severely weakened on
tests for competence, and a minority of members” [8]. The
the less hierarchical social web that promotes a collaborative
production of content [8]. Yet, elites still have a relative practical application of this definition here therefore includes the
various forms of professionalized online media like Mashable or
advantage on the web in terms of political and economic
VOX that can be regarded as “online social elites” [4].
resources, thus profiting from economies of scale in the
production and dissemination of web contents [4]. 2.2 Expectations
Previous studies analyzed the influence of elites in political Studies have shown that elites still have considerable influence in
Twitter debates ex post, based on the most retweeted messages or political debates on Twitter [3, 7]. However, it became apparent
centrality metrics [e.g. 2, 3, 7]. However, such inductive that their influence varies according to different metrics [2, 3, 6,
procedures miss important communication in the long tail of elites 7]. First, a large number of followers does not guarantee an
on Twitter that can only be captured by defining actors ex ante. influential role in topic specific Twitter debates [2]. Second,
The contribution of this article is twofold: First, it presents lists of @-mentions display the perceived importance of actors in debates
twitter handles of actors from the U.S. government, news media without necessarily signaling an intention of endorsement,
and politics [9] and compares these elites in terms of their whereas retweets can be regarded as the best available predictor of
follower count, which is the most basic metric of importance. ideological homophily [6]. We should be able to observe similar
Second, the actor lists are applied to the political Twitter debates patterns when differentiating these metrics in the political sphere.
on net neutrality and the State of the Union Address 2015 to
The structural characteristics of debates and the specific roles of
estimate the influence of elites in these discussions. The article
political actors should also be reflected online. Actors who are
concentrates on the U.S., since Twitter use of political actors is
involved in the policies or political events to which Twitter
most advanced in this illustrative case.
debates relate should be referenced most often, since expectations
are directed towards them, also from users who do not agree with
Placeholder textbox the political positions taken by elites. From a reversed logic,
retweet shares to some extent also reflect elites’ own level of
Copyright c 2016 held by author(s)/owner(s); copying permitted activity and the importance of the medium Twitter as perceived by
only for private and academic purposes. them. Previous research indicates that especially the U.S.
Published as part of the #Microposts2016 Workshop proceedings,
available online as CEUR Vol-1691 (http://ceur-ws.org/Vol-1691)
executive tries to achieve political goals by influencing public
opinion [5]. Since the legislative competencies of the Presidency
#Microposts2016, Apr 11th, 2016, Montréal, Canada. are limited, its actors aim to set the political agenda via direct
communication and “going public” strategies.
· #Microposts2016 · 6th Workshop on Making Sense of Microposts · @WWW2016
3. METHODOLOGY
3.1 Operationalization of actor lists
I extracted publicly available Twitter lists from @gov, @cspan
and a collection of influential news media accounts from Daniel
Romero1 as a starting point for generating the actor lists [9]. These
were significantly cleaned and expanded by hand in order to cover
every member of U.S. congress with a Twitter account, all
governors, party accounts as well as accounts belonging to the
U.S. government and its officials. To add additional data of
lawmakers, the politics list was matched with a database of
GovTrack. To restrict the category news media to elites, only
media accounts officially verified by Twitter were included, while
individual journalists were excluded.2
3.2 Measures of influence on Twitter Figure 1. Follower distributions by elite actor group
The estimation of actor importance and the identification of
influential users are recurring topics in web science [1, 2]. For the This finding needs to be taken into account when assessing the
exploratory analysis, I use follower counts as the most basic role of elites, especially when studying the diffusion of political
measure of influence. The subsequent empirical application of the information. In terms of aggregate follower counts, especially the
actor lists relies on @-mentions and retweets. government and news media should be able to distribute their
contents to a significant portion of the U.S. political Twitter
3.3 Datasets sphere, either first-hand or via two-step-flow processes. But, as
In order to generate empirical test datasets, I extracted tweets the next section highlights, this potential depends on their own
containing the main debate hashtags on two political topics. propensity to utilize their potential outreach on Twitter [2].
#SOTU was the main hashtag accompanying Barack Obama’s
State of the Union Address in front of U.S. Congress in January 4.2 Case studies: #SOTU and #NetNeutrality
2015. #NetNeutrality refers to a policy discussion on the The present chapter applies the actor lists to the two test datasets.
regulation of data traffic on the internet. The debates differ with Figure 2 shows the influence of elite actors according to the two
regard to the time period covered and the tweet volume generated. most important conversation practices on Twitter. Displayed is the
share of elites in the aggregate number of retweets and
Table 1. Test datasets @-mentions. The most evident pattern is the influential role of the
Number of Number government with a mention share of 29% in #SOTU and 56% in
Topic Period #NetNeutrality. In both debates, government actors were the
tweets of users
1/20/2015– central political figures. President Obama as the speaker attracted
#SOTU 1,271,474 413,832 the highest share of attention during the #SOTU, whereas in the
1/21/2015
1/14/2015– #NetNeutrality debate the FCC and its chairman Tom Wheeler
#NetNeutrality 503,839 174,371 were the accounts to which most users referred.4 The high number
3/6/2015
of mentions reflects their perceived importance in the two debates.
As the data mining was based on Twitter hashtags (Table 1), However, this does not equal an endorsement. Government actors
issue-related tweets without these particular hashtags were not get retweeted, which is a stronger signal of political homophily
captured. This could be problematic if the use of hashtags varies [6], with a lower frequency than @-mentioned.
systematically across actor groups. These limitations are discussed Deviations from the preceding analysis that focused on follower
in the empirical section. counts are evident when looking at the relatively sparse attention
share of news media and actors from politics. When inspecting
4. RESULTS AND DISCUSSION tweet contents and URLs, it becomes apparent that this is mostly
due to their own restrained tweeting patterns. The main event
4.1 Follower counts of elites hashtags were rarely used by these groups, although they regularly
Figure 1 is a log-log plot displaying the number of followers commented on the events. Further qualitative research should
plotted by the rank of an actor in its respective group.3 In all three investigate the social media strategies of those in charge of the
groups, the follower distribution is heavily skewed. The decay of Twitter accounts in America’s newsrooms and politics.
the tails resembles a linear pattern until the curve drops steeply at
the lower ends of the distribution. The unequal distribution of Figure 3 amplifies the skewed distribution of attention on the web
attention is a typical phenomenon on the web in general and by showing the retweet per tweet ratios of actor groups in the two
especially in the political sphere, as Hindman, among others, debates. The relative outreach that elites generated per tweet is
showed in his study of power laws in the political blogosphere [4]. considerable when compared to non-elite users. Non-elite users
had a retweet outreach per tweet below zero, even though this
category also includes a multitude of influential accounts like
1 NGOs. This urgently points towards the need to further
Available online: http://memeburn.com/2010/09/the-100-most-
disaggregate the residual category of others. By adding more lists,
influential-news-media-twitter-accounts.
2
political Twitter debates can be analyzed in even more detail.
The updated version of the group news media also includes
influential individual journalists [9]. The category politics
features information on political offices and party affiliations.
3 4
The ranks of actors are standardized by the diverging lengths of The FCC is a regulatory agency. Its five commissioners are
actor lists and thus reported as cumulative percent of accounts. appointed by the President.
44
· #Microposts2016 · 6th Workshop on Making Sense of Microposts · @WWW2016
Figure 2. Share of Twitter conversation practices by elite actor group
The varying results for government actors in the #NetNeutrality Party politicians in particular played a minor part in all of the
debate underline the diverging political and social meanings of analyses. Their role needs to be contextualized by analyzing
Twitter metrics. The total share of retweets originating from Twitter communication on additional political topics.
government accounts is relatively marginal (Figure 2). However, Methodologically, the analysis remained at a basic level of
when the government used the hashtag #NetNeutrality, it analysis by focusing on distributions of Twitter metrics. There is a
generated 492 retweets per tweet (Figure 3). When central figures great potential to further elaborate on these preliminary results by
in political events intensify their efforts to influence hashtag applying the toolkit of network science. Subsequent applications
communities, they have success in doing so. of the actor lists should include topically relevant hashtag
Figure 3 confirms that the most influential actors in the two populations and keywords as well as information diffusion via
debates were affiliated with the government, while tweets from URLs [1]. Lastly, the typology could be expanded by widening
news media and politics diffused less widely on the web. This the conceptualizations of the existing groups and by classifying
finding depicts “going public” strategies by the government [5], politically influential actors like celebrities or NGOs.
but is also a result of the political characteristics of the two case
studies. Future studies should investigate these exploratory 6. REFERENCES
findings by including structurally heterogeneous Twitter debates. [1] Bakshy, E., Hofman, J. M., Mason, W. A., and Watts, D. J.
2011. Everyone’s an Influencer: Quantifying Influence on
5. CONCLUSION Twitter. In Proceedings of the Fourth ACM International
This paper presented lists of elite actors enriched with additional Conference on Web Search and Data Mining. ACM.
information that enable researchers to investigate a broad range of [2] Cha, M., Haddadi, H., Benevenuto, F., and Gummadi, K. P.
research questions related to the activity of U.S. elites and non- 2010. Measuring User Influence in Twitter: The Million
elites in online political communication, such as: Are politicians Follower Fallacy. In 4th International AAAI Conference on
from the Democratic Party more successful in framing political Weblogs and Social Media. AAAI Press.
debates on Twitter than Republicans? What can we learn about [3] Dubois, E. and Gaffney, D. 2014. The Multiple Facets of
the dynamics of communication between elites and between elites Influence: Identifying Political Influentials and Opinion
and their audience when looking at conversation patterns? Leaders on Twitter. Am. Behav. Sci. 58, 10, 1260–1277.
In an empirical application of the actor lists, government and [4] Hindman, M. S. 2009. The Myth of Digital Democracy.
news media were most influential according to their number of Princeton University Press, Princeton, NJ.
followers and also when assessing retweets and @-mentions. Yet, [5] Kernell, S. 1986. Going Public: New Strategies of
within these groups and throughout the study, the skewed Presidential Leadership. CQ Press, Washington, DC.
distribution of influence on the political web became apparent. [6] Lietz, H., Wagner, C., Bleier, A., and Strohmaier, M. 2014.
When Politicians Talk: Assessing Online Conversational
Practices of Political Parties on Twitter. In Proceedings of
the Eighth International AAAI Conference on Weblogs and
Social Media. AAAI Press.
[7] Lin, Y.-R., Keegan, B., Margolin, D., Lazer, D., and
Holme, P. 2014. Rising Tides or Rising Stars?: Dynamics
of Shared Attention on Twitter during Media Events. PLoS
ONE 9, 5, e94093.
[8] Shirky, C. 2008. Here Comes Everybody. The Power of
Organizing Without Organizations. Penguin Press, New
York, NY.
[9] Stier, S. 2016. Elite Actors in the U.S. Political
Twittersphere. Gesis datorium,
http://dx.doi.org/10.7802/1178
Figure 3. Retweet per tweet ratios by user groups
45
· #Microposts2016 · 6th Workshop on Making Sense of Microposts · @WWW2016