P7 (00 421 P7)
The role of social networks to explain political behaviour and attitudes
Form of presentation:
Facebook and revolutions: new evidence from Ukraine.
Tymofii Brik | University of Carlos III | Spain
Online social networks, Facebook in particular, are often regarded as the important tool for mobilization during the 2013-2014 Euromaidan revolution in Ukraine. The present study investigates empirically to what extent this claim is true. A unique data set of online activities was collected from the Euromaidan Facebook page. Overall, 27,458 posts and 1,123,049 comments generated from November 2013 to May 2014 are analyzed by means of social network analysis tools. The present analysis suggests that the role of Facebook communication in mobilizing was limited. However, there are some indications that Facebook played coordination role. Furthermore, the analysis shows that structure of networks of communication affected success of communication measured through likes.
A Network Model of Selective News Consumption-Induced Fragmentation
Pascal Jürgens | Uni Mainz | Germany
For most political issues, the news remain citizens' primary source of information. This privileged position forms the core of mass media's persuasive power. But with the vast range of outlets and the cacophony of opinions available on the net, readers can pick and assemble their preferred information digest more than ever before. And they do: The well-documented phenomenon of selective exposure (Stroud 2011) leads recipients to favor attitude-consistent content over reports that challenge their views. In extreme cases, they might even choose to limit their reading to coverage that, to an outside observer, constitutes a parallel universe. Such biased information behavior represents more than an individual's curious taste: Selective exposure can drive fragmentation of news audiences along partisan lines. A fragmented society, in turn, loses the crucial ability to reconcile different positions and act on a common conception of political issues.
The rise of the internet has also brought a second shift: The chain of brokers involved in delivering news to an internet user grows steadily. Where the evening news are directly broadcasted to their audience, an online article often reaches its readers in the form of a recommendation from social networking sites or a search engine result. Faced with these new intermediary positions, the need for a network analysis of selective exposure and audience fragmentation becomes more and more evident.
Hence, news consumption behavior becomes both a network research topic and a matter of societal cohesion. The methodological fit is convincing: Consumption relations between recipients (nodes) and news sources (nodes) form a two-mode network that conceptualizes news preferences. Global and local structural metrics may then be used to measure fragmentation (for a first application of this approach, see Webster & Ksiazek 2012).
In this paper, we develop a full network theoretic approach to the analysis of recipient-news relations. It introduces two major contributions: (1) in addition to recipients and news sources, we add topics of news coverage as a third node type. Sources may be linked by covering the same topics. A distance-based similarity metric then reflects the degree of differences between any two user vertices. By connecting either directly though shared use of the same source or (with a distance penalty) through a topic shared by two different sources, we can accurately position individuals on a gradual instead of a binary similarity scale. (2) we further introduce intermediary nodes that represent the broker role of news recommenders such as social networking sites and search engines. Their impact upon the similarity metric is computed through an interative removal algorithm. The whole approach is validated using natural extreme states (zero/full fragmentation) and a simulation study of gradually increasing selective exposure.
Measurement of Electoral Behavior in Election Research: Online Experiments within Online Political Stock Markets
Björn Klein | Heinrich Heine Universität Düsseldorf | Germany
Election research studies have asked whether surveys influence the voters' electoral expectations about the election results and which effects have risen from these expectations of voting decisions. The theoretical assumptions (e.g. the bandwagon effect) have been empirically examined with different methodological designs. Experimental designs have been considered particularly relevant. The following points are critically seen in the previous experimental designs: the voters' electoral expectations about election results via questioning, the self-assessment can lead to errors; and from a chronological perspective, the measurement of the voters' electoral expectations takes place only once. Influences of survey dynamics could not be displayed.
Our studie questions whether experimental designs within Online Political Stock Markets (OPSMs) open up further methodological possibilities to investigate the impact of surveys on the expectations about election results. OPSMs operate according to the following principle of market logic: the convergence of many participants leads to realistic prices of objects. The goal of participants within OPSMs is to forecast the election result by exchange stocks of the parties. The exchange creates the price of the parties, which is interpreted as a forecast value.
In order to evaluate our question, in the period of May 1 to July 17, 2013, 413 active participants of the Handelsblatt OPSM on the national election 2013 in Germany were randomized and divided into four groups. These groups were shown different presentations of the parties’ survey results: group 0 (n=117) was shown no surveys; group 1 (n=97) was shown a survey trend over the course of 10 weeks; group 2 (n=94) was shown a survey trend over the course of 100 weeks; and group 3 (n=105) was shown a survey trend from a daily cross section. The values of the stimulus materials were calculated from the mean values of the current daily survey results. The daily number of trading activities, the kind of transaction, and the transaction price generated the dependent variables. The dynamics of the survey results are included as an intervening variable.
The results show that the trading activities of the four groups were different. Participants who were not shown any survey results traded 48.5 percent of all transactions (group 0 mean value=69 trading activities/group member), which was more than any of the other groups that were shown different forms of surveys (group 1 mv=20; group 2 mv=31; group 3 mv=35). In addition, the data-analytic temporal subdivision shows that dynamics in the survey values of the parties had an effect on the trading activities and the price of the stocks of the respective parties (group 1-3). If the survey values were changed, more stocks of the parties were traded, and the price of the stocks became more favorable or more expensive.
Contrary to previous experimental designs, the voters' electoral expectations could be measured in form of real actions as a result of knowledge or non-knowledge about survey results. Temporal dynamics of different expectations could also be mapped during the long-term experimental period.