A Network Analysis of the Online Audience Behavior

30
Sílvia Majó-Vázquez Internet Interdisciplinary Institute (UOC) [email protected] @silviamajo Ana S. Cardenal, PhD Internet Interdisciplinary Institute (UOC) [email protected] @ana_cardenal A Network Analysis of the Online Audience Behavior: Towards a Better Comprehension of the Public Agenda

Transcript of A Network Analysis of the Online Audience Behavior

Sílvia Majó-Vázquez Internet Interdisciplinary Institute (UOC)

[email protected]

@silviamajo

Ana S. Cardenal, PhD Internet Interdisciplinary Institute (UOC)

[email protected]

@ana_cardenal

A Network Analysis of the Online Audience Behavior:

Towards a Better Comprehension of the

Public Agenda

Broader ongoing project jointly with Professor S. González-Bailón, PhD, Annenberg UPenn. Mapping the two levels of the online public sphere: ­  The online news media network ­  The online audience network Project funded by the Spanish Ministry of Economy and Competitiveness (Ref: CSO2013-4782-P) More information www.opinionet.net  

The Research Project

1)  The Research Question

2)  Justification of Interest

3)  Data and Methods

4)  Results

5)  Discussion and future research

1)  The Research Question

2)  Justification of Interest

3)  Data and Methods

4)  Results

5)  Discussion and future research

1)  The Research Question

2)  Justification of Interest

3)  Data and Methods

4)  Results

5)  Discussion and future research

1)  The Research Question

2)  Justification of Interest

3)  Data and Methods

4)  Results

5)  Discussion and future research

1)  The Research Question

2)  Justification of Interest

3)  Data and Methods

4)  Results

5)  Discussion and future research

1)  The Research Question

2)  Justification of Interest

3)  Data and Methods

4)  Results

5)  Discussion and future research

#1 Does the web guarantee a digital public sphere?   ­  #1.1 Which are the patterns of online audience behaviour?

­  #1.2 Hubs of public attention are the necessary conditions for an online public agenda. Can we find high levels of audience overlapping in the online sphere?

­  #1.3 To what extend do people enjoy shared informational

experiences in the Internet?

The Research Question

1)  The Research Question

2)  Justification of Interest

3)  Data and Methods

4)  Results

5)  Discussion and future research

­  Political information is a vehicle for engagement in the democratic process (Gutmann & Thompson, 2009; Sunstein, 2009)

­  Democracy depends extensively on an informed citizenry about the

most important problems in their societies (Converse, 1964; Habermas, 1994; Katz, 1996; Rawls, 2009)

­  Shared informational experiences are necessary for the public debate in a deliberative democracy

Justification of the interest

­  Decentralized and high-choice media environment on the web has fostered a theoretically ambiguous debate

­  Earliest approaches contended that the web would democratize the

public sphere (Benkler, 2006; Negroponte, 1995; Rheingold, 1994)

­  The Internet fosters the end of the mass audiences (Castells, 2009;

Napoli, 2011) and the unlimited audience fragmentation (Baum & Groeling, 2008; Baum, 2012; Sunstein, 2009; Turow, 1998)

Justification of the interest

­  In place of a collective shared public agenda, fragmented and competing media agendas would have emerged (Althaus & Tewksbury, 2002; Shaw & Hamm, 1997; Tewksbury, 2005)

­  The sheer number of news sources available on the web does not

determine the extend of the audience fragmentation (Ksiazek, 2011)

­  Having more media outlets to get information does not imply that

people might not have a common public agenda

Justification of the interest

1)  The Research Question

2)  Justification of Interest

3)  Data and Methods

4)  Results

5)  Discussion and future research

­  Networks are at the heart of the online communication structure

­  Network Science to map network of audience behavior and identify patterns of audience overlapping

­  Aggregated data from online browsing tracking behavior

Data and Methods

­  ComScore representative panel of 30,000 individuals in Spain

­  First time that this data is used for the study of modes of exposure to political information in Spain

­  Collected in December 2014

Data and Methods

­  Audience data for 165 Spanish news outlets

­  New media and traditional media

­  We measure audience overlapping for each pair of nodes

Data and Methods

­  Identify statistical significant ties (levels of audience overlapping) by building on the previous work by Ronen et. Al., (2014)

­  A link is defined within a pair of nodes when the probability that audience from outlet i also attend outlet j is higher than we would expect based on the overall reach of the outlet alone

­  2,510 total number of ties among 105 nodes

­  Our network is directed and weighted

Data and Methods

1)  The Research Question

2)  Justification of Interest

3)  Data and Methods

4)  Results

5)  Discussion and future research

­  To assess the fragmentation of the online audience network we computed the centralization degree (Freeman, 1979) à 0.22

­  The network is not totally decentralized (0) thus some nodes still concentrate high levels of audience overlapping

­  The bigger the node the higher the audience overlapping it receives

­  The larger levels of audience overlapping flow towards a small group of nodes

Results

­  To identify groups of nodes that share a similar structure we run a structural equivalence analysis (Burt, 1987)

­  The Dendrogram depicts similarities in the network of audience overlapping

­  The majority of the news outlets, at the centre of the figure, are pretty similar

­  Nodes at the left and right extremes are the same media outlets with higher levels of audience overlapping in the visualization

Results

Structural Equivalence Analysis of the Audience Network

­  Audience concentration still remains in the web (Gentzkow & Shapiro, 2010; González-Bailón, 2009; Webster & Ksiazek, 2012)

­  Traditional media driving the public agenda offline are still at the core

of the online audience flow

­  The structure of audience overlapping brings evidence that the web

can provide people with similar bits of information

­  A significant share of citizens gets news from multiple online media outlets

Results

1)  The Research Question

2)  Justification of Interest

3)  Data and Methods

4)  Results

5)  Discussion and future research

­  The extend to what people are aware of the information provided by hubs of audience is not proved by this network analysis

­  The boundary specifications of our network might influence the results

­  Further research across time and platforms to obtain comparative results is necessary

­  We cannot generalize to the entire population in the web. Our study is focus on people politically interested which seek for online news

­  The end of information scarcity does not equate to the ceasing of the online public sphere

Discussion and future research

Sílvia Majó-Vázquez Internet Interdisciplinary Institute (UOC)

[email protected]

@silviamajo

Ana S. Cardenal, PhD Internet Interdisciplinary Institute (UOC)

[email protected]

@ana_cardenal

Thank you!

Questions and feedback greatly appreciated

Distribution of the total node strength