Content uploaded by Sophia Fu
Author content
All content in this area was uploaded by Sophia Fu on Feb 18, 2018
Content may be subject to copyright.
Full Terms & Conditions of access and use can be found at
http://www.tandfonline.com/action/journalInformation?journalCode=utis20
Download by: [99.32.20.232] Date: 08 March 2017, At: 08:38
The Information Society
An International Journal
ISSN: 0197-2243 (Print) 1087-6537 (Online) Journal homepage: http://www.tandfonline.com/loi/utis20
News media, social media, and hyperlink
networks: An examination of integrated media
effects
Jiawei Sophia Fu & Michelle Shumate
To cite this article: Jiawei Sophia Fu & Michelle Shumate (2017) News media, social media, and
hyperlink networks: An examination of integrated media effects, The Information Society, 33:2,
53-63
To link to this article: http://dx.doi.org/10.1080/01972243.2016.1271379
Published online: 08 Mar 2017.
Submit your article to this journal
View related articles
View Crossmark data
News media, social media, and hyperlink networks: An examination of integrated
media effects
Jiawei Sophia Fu and Michelle Shumate
Department of Communication Studies, Northwestern University, Evanston, Illinois, USA
ARTICLE HISTORY
Received 9 June 2015
Accepted 12 May 2016
ABSTRACT
Research typically focuses on one medium. But in today’s digital media environment, people use
and are influenced by their experience with multiple systems. Building on media ecology research,
we introduce the notion of integrated media effects. We draw on resource dependence and
homophily theories to analyze the mechanisms that connect media systems. To test the integrated
media effects, we examine the relationships between news media visibility and social media
visibility and hyperlinking patterns among 410 nongovernmental organization (NGO) websites in
China. NGOs with greater news media visibility and more social media followers receive significantly
more hyperlinks. Further, NGOs with a similar number of social media followers prefer to hyperlink
to each other. The results suggest that both news media and social media systems are related to
the configuration of hyperlink networks, providing support for the integrated media effects
described. Implications for the study of hyperlink networks, online behaviors of organizations, and
public relations are drawn from the results.
KEYWORDS
Hyperlinks; social network
analysis; nongovernmental
organization (NGO); media
ecology; social media;
homophily; resource
dependence
Introduction
In today’s media environment, organizations communi-
cate with audiences using various offline and online
channels, including news media, their websites, and
social media. However, most research to date examines
these channels as isolated systems (e.g., hyperlink net-
works in Shumate and Lipp 2008; follower networks on
Weibo in Huang, Gui, and Sun 2015), assuming that
organizations connect with other organizations (e.g.,
hyperlinking or following an organization on Twitter)
on media platforms (e.g., websites) independent of their
behavior and resources on other media platforms.
This article argues that hyperlink networks are open
systems that interact with the news media and social
media systems. As spotlighted by media ecology theory,
multiple media construct the environments with which
people interact. As such, different media are additive and
complementary, forming an organic, holistic media eco-
system (McLuhan 1964; McLuhan and Fiore 1967;
McLuhan 1964; Nystrom 1973; Postman 1970; Strate
2006; Scolari 2012). Building on media ecology theory,
this article shows that organizations’communication
behaviors in news media, on websites, and on social
media are closely connected and mutually influential. To
understand the interaction between these media systems,
it introduces the notion of integrated media effects—
organizations’Web behaviors communicate similarity
and resources possession signals to other organizations,
which influences their communication behavior across
media systems. In effect, these communications serve as
extrinsic cues for other organizations to select partners.
Here we draw on resource dependence and homophily
theories to analyze the mechanisms that connect media
systems.
Our empirical analysis focuses on a hyperlink net-
work, or the totality of hyperlinks from a set of organiza-
tions’websites (Park, Barnett, and Nam 2002). Although
integrated media effects might occur bidirectionally or
originate from organizations’websites, we focus on
hyperlink networks for two reasons. First, while Internet
users navigate primarily with local search engines (Wu
and Ackland 2014), the hyperlink network remains an
important element in the determination of search engine
results (for the classic explanation of how this is done see
Kleinberg 1999). As such, the structure of the hyperlink
network influences which organizations are given
salience when publics seek information via search
engines. Second, and perhaps more importantly, there is
CONTACT Jiawei Sophia Fu sophiafu@u.northwestern.edu Department of Communication Studies, Northwestern University, 2240 Campus Drive,
Evanston, IL 60208, USA.
Color versions of one or more of the figures in the article can be found online at www.tandfonline.com/utis
Published with license by Taylor & Francis. © Jiawei Sophia Fu and Michelle Shumate
THE INFORMATION SOCIETY
2017, VOL. 33, NO. 2, 53–63
http://dx.doi.org/10.1080/01972243.2016.1271379
a robust line of scholarly research on hyperlink networks
(e.g., Ackland and O’Neil 2011; Park et al. 2002; Shumate
and Lipp 2008). This research has primarily focused on
hyperlink networks among nongovernmental organiza-
tions (NGOs) and has demonstrated that the inclusion
of a hyperlink is a strategic communication choice—not
a random one. Similar research on the nature of Twitter
follower networks (e.g., Peng, Liu, Wu, and Liu 2015),
Facebook friend networks (e.g., Lewis, Gonzalez, and
Kaufman 2012), or other types of social media networks
(e.g., Huang, Gui, and Sun 2015; Lai, She, and Ye 2015)
has yet to draw conclusive results. We test our hypothe-
ses by analyzing a hyperlink network of 410 Chinese
NGOs. We examine the relationship between similarity
and resources possession signals communicated by
NGOs on news media and social media to the patterns
of this hyperlink network.
The rest of the article is organized as follows. First, we
review the literature on hyperlink networks, especially
among NGOs. Next, drawing on literature from public
relations and media ecology theory, we hypothesize how
hyperlinks networks are related to signals derived from
the news media and social media systems. Then we
describe our methodology and discuss our findings.
Lastly, we discuss the implications of these findings for
Internet studies and hyperlink network research.
Literature review
Nongovernmental organizations and hyperlink
networks
Khagram, Riker, and Sikkink (2002)define NGOs as
“private, voluntary, non-profit groups whose primary
aim is to influence publicly some form of social
change”(6). We adapt their original definition to
define Chinese NGOs as “nonprofit groups whose pri-
mary aim is to influence publicly some form of social
change.”In doing so, we highlight some uniqueness of
the NGO sector in China; for example, some NGOs in
China do not have full autonomy and many of them
are organized by the government (government-
organized NGO, a.k.a. GONGO) and depend on gov-
ernment connections (Ma 2002).
The choice to hyperlink is a choice about symbolic
public affiliation (Ackland and O’Neil 2011; Shumate
and Lipp 2008). Targets that are being hyperlinked to
(i.e., hyperlink recipients) are usually not aware of the
hyperlink directed by others, and receiving many hyper-
links does not incur additional costs for them. For exam-
ple, if organization A hyperlinks to organization B,
organization B is usually not aware of the existence of
such a hyperlink. Therefore, hyperlinks form a
representational communication network that commu-
nicates affiliations related to a third party or to the public
(Shumate and Contractor 2013).
Hyperlinks among NGOs are symbolic relationships
that help them reach like-minded civil actors, facilitate
collective action, and develop collective identity (Ackland
and O’Neil 2011; Shumate and Lipp 2008; Rogers and
Marres 2000). NGO hyperlink networks have been con-
ceptualized as collective action networks, which very
often make more visible certain social issues (Ackland
and O’Neil 2011; Shumate and Lipp 2008). As such,
NGO hyperlinks represent cooperative than competitive
relationships, and we expect to see more links among
NGOs than we would see among a set of for-profit
organizations. This article investigates the inter-
hyperlinking practices of NGOs via websites; linking
practices through social media (Hsu and Park 2011; e.g.,
retweets, hashtags, and mentions) are beyond the scope
of this article.
NGO hyperlinking is a purposive and strategic com-
munication choice (Ackland and O’Neil 2011; Shumate
and Lipp 2008). Since hyperlink networks consist of vari-
ous interdependent ties among actors, they demonstrate
the structural embeddedness of online organizational
communication behavior (Lusher and Ackland 2011;
Park 2003). Previous studies have explored NGO hyper-
link networks from two perspectives: structural signa-
tures (e.g., Lusher and Ackland 2011) and offline
organizational attributes (e.g., Pilny and Shumate 2012).
The structural signatures perspective focuses on the
pattern of ties in hyperlink networks among organiza-
tions. Structural signatures denote structural features
that recur across the network in a characteristic way
(Monge and Contractor 2003). For example, organiza-
tions tend to reciprocate hyperlinks, and hyperlinks
among triads of organizations are widespread
1
(Shumate
and Dewitt 2008; Barnett 2008; Kim 2012; Lusher and
Ackland 2011). In addition, preferential attachment (i.e.,
hyperlinking to websites that already have a large num-
ber of hyperlinks from other websites) is common in the
hyperlink networks, due to the perceived social influence
and legitimacy of the hyperlink recipients (Pennock et al.
2002). In sum, this line of research suggests that the con-
figuration of NGO hyperlinks is conditioned upon other
hyperlinks in the network.
In contrast, the offline attributes perspective focuses
on various characteristics of organizations and their rela-
tionship to hyperlinking behaviors. These attributes
include economic resources (Shumate and Dewitt 2008),
organizational longevity (Gonzalez-Bailon 2009b), orga-
nization type (Kim 2012), organization mission (Ackland
and O’Neil 2011), and geographic location (Shumate and
Lipp 2008). In sum, this line of research suggests that
54 J. S. FU AND M. SHUMATE
NGOs’characteristics and offline relationships are
related to the configuration of hyperlink networks.
This article extends prior research by introducing a
new, third line of research on organizational hyperlink
networks: the influence of other media systems. In the
following section, we introduce a media ecology perspec-
tive that suggests that signals about both similarity and
resources possession communicated by organizations in
news media and social media systems are related to the
configuration of hyperlink networks. First we present an
overview of that argument, and then we specify how
resource dependence and homophily mechanisms con-
nect these media systems, a phenomenon we refer to as
integrated media effects.
Media ecology perspective on organizations’public
presence
Previous research suggests that organizational profiles on
different media systems are interconnected. Kwak et al.
(2010)find that accumulated website hyperlinks are
related to the number of microblog followers. Similarly,
Nah and Saxton (2013)find that website reach is related
to organizational social media use and adoption. Follow-
ing this line of research, we focus on hyperlink networks
as extensions of organizational presence in news media
and social media systems. The signals, indicating similar-
ity or resources possession, communicated in news
media and social media systems serve as cues that shape
hyperlinking practices. Although a media ecology per-
spective suggests that news media and social media sys-
tems should be related to hyperlink networks, it does not
specify how they are related. To understand the types of
integrated media effects, we employ two analytics:
resource dependence and homophily.
In the Web 1.0 and Web 2.0 media environment,
organizations are actively engaged in communicating
about themselves to the publics and other organizations,
generating communicative signals about their identity
and resources possession. These communicative signals,
across different platforms, drive other organizations’
Web behaviors. For example, the amount of news cover-
age (Pilny and Shumate 2012; Gonzalez-Bailon 2009a,
2009b), the number of accounts followed and the num-
ber of posts sent on social media (Fu and Shumate
2015), and the number of followers on social media
(Kwak et al. 2010) communicate that an organization
has accumulated resources such as popularity, reputa-
tion, and visibility.
Further, organizations’Web behaviors communicate
signals to other organizations about the characteristics of
the organization and the similarity between organiza-
tions. Hyperlinks are signals of collective identity
(Ackland and O’Neil 2011), trust (Stewart 2003), and
credibility and legitimacy (Park, Barnett, and Nam
2002). For instance, hyperlinks from a credible website
to another website increase the perceived interaction and
similarity between linked actors (Stewart 2003); the qual-
ity of website functions signals the quality of products
(Wells, Valacich, and Hess 2011); and the use of #hash-
tags in social media signals organizations’shared topical
interests (Rossi and Magnani 2012).
We contend that various communicative signals
induce organizations to select particular targets for Web-
based behavior, such as hyperlinking to a particular orga-
nization. Specifically, signals can convey similarity
between two organizations (e.g., identity) or the resour-
ces accumulated by another organization (e.g., popularity
and visibility). Network selection based on communica-
tive signals (i.e., similarity and resource possession) is
typically explained via resource dependence theory and
homophily theory (Guo and Acar 2005; McPherson,
Smith-Lovin, and Cook 2001; Monge and Contractor
2003).
Resource dependence theory suggests that hyperlinks
are organizations’responses to their external environ-
ments (Pfeffer and Salancik 1978). The more resources
(e.g., popularity and visibility) an organization’s online
signals, the more likely it is to receive hyperlinks. In con-
trast, homophily effects are based upon organizations’
tendency to recognize other organizations with similar
attributes and to affiliate with them. We suggest that
these mechanisms underlie integrated media effects. In
the following sections, we further extrapolate these argu-
ments, specifying a series of hypotheses based upon these
two mechanisms.
Resource dependence and integrated media effects
Resource dependence theory suggests that organizations
use network relationships (e.g., hyperlinks) with other
organizations to manage uncertain environments; in par-
ticular, they attempt to link to organizations that control
critical resources (Pfeffer and Salancik 1978). Corre-
spondingly, we contend that resources accumulated in
one media system will increase the attractiveness of
organizations in another system. More specifically, in the
case of interorganizational hyperlink networks, organiza-
tions with greater resources (e.g., visibility and popular-
ity) in the news media and social media systems are
likely to attract more hyperlinks.
News media visibility is a type of organizational
resource (Gonzalez-Bailon 2009a,2009b) that previous
research suggests shapes the configuration of hyperlink
networks. Specifically, Gonzalez-Bailon (2009b) and
Pilny and Shumate (2012)find that NGOs with greater
media visibility are more likely to receive hyperlinks.
THE INFORMATION SOCIETY 55
Pilny and Shumate (2012), more specifically, found that
for every additional news article mentioning an organi-
zation, there is a 1% increase in the odds that the
organization will receive a hyperlink. Therefore, we
hypothesize that:
H1: NGO news media visibility is positively related to
the propensity to receive hyperlinks.
Social media followers are another type of online sig-
nal that communicates an organization’s resource pos-
session, defined as the number of accounts following or
liking an organization’s profile. The number of social
media followers directly influences search engine result
rankings (Kwak et al. 2010). Thus, it represents an orga-
nizational resource indicating influence, popularity, and
visibility. Therefore, we hypothesize that:
H2: The number of followers an NGO has is positively
related to its propensity to receive hyperlinks.
Homophily and integrated media effects
Homophily is the tendency to select and interact with
similar entities (McPherson, Smith-Lovin, and Cook
2001; Monge and Contractor 2003). Various social sys-
tems create contexts to induce homophilous relations
(McPherson, Smith-Lovin, and Cook 2001), and in this
study, signals on media platforms can communicate sim-
ilarities in scale and serve as hyperlinking cues to organi-
zations. For our analysis, the primary cross-platform
attribute we focus on is status similarity (Lazarsfeld and
Merton 1954; McPherson, Smith-Lovin, and Cook
2001). Organizations with similar status are motivated to
connect to each other because they perceive themselves
to be more compatible and trustworthy and are less
uncertain about the behavior of their partner than organ-
izations of dissimilar statuses (Atouba and Shumate
2015).
News media visibility can act as a communicative sig-
nal that organizations have a similar status. Status simi-
larity signaled in one media system may induce
connections in other systems. Therefore, we hypothesize
that:
H3: NGOs with similar levels of media visibility are
more likely to hyperlink to each other.
Whether an organization has a social media profile
depends at least partly on organizational capacity. Addi-
tionally, an organization needs to have the expertise and
skill to build a successful social media presence (Lovejoy,
Waters, and Saxton 2012; Lovejoy and Saxton 2012). As
such, social media presence may signal organizations’
similar stakeholder engagement strategies (Lovejoy and
Saxton 2012). Because establishing presence requires an
NGO to devote resources, some NGOs are more likely
than other NGOs to adopt and use social media (Nah
and Saxton 2013; Seo, Kim, and Yang 2009). Moreover,
the mutual presence of a social media profile is the mini-
mum needed for homophily integrated media effects to
occur between the social media system and the hyperlink
network system. Therefore, we hypothesize that:
H4: NGOs with social media profiles are more likely to
hyperlink to each other.
Organizations with a similar number of social media
followers have achieved similar organizational status on
social media (Kwak et al. 2010). Homophily suggests
that organizations are more likely to hyperlink to other
organizations with a similar number of followers. There-
fore, we hypothesize that:
H5: NGOs with similar levels of social media followers
are more likely to hyperlink to each other.
Method
Sample and procedures
Data for this study were extracted from the Beijing Civil
Society Development Research Center (CDB) (http://
www.cdb.org.cn), a Beijing, China-based NGO dedicated
to providing nonpartisan information, resources, con-
sulting, and networking services for China’s NGO sector.
CDB cooperates with nine of its regional NGO partners
to identify influential NGOs on various social issues both
at the local and the national level in China. CDB con-
tacted such NGOs, and NGOs that returned a form with
background information on their organizations were
included in its directory of NGOs nationwide. CDB
updates its NGO directory on an annual basis. After
removing duplicates, the directory yielded a total of 814
organizations. These NGOs covered a variety of social
issues, including education, health, child welfare, envi-
ronmental conservation, animal protection, and labor
rights. They served each of China’s national regions.
Among these, 556 organizations (68.3%) had active web-
sites. After removing 14 duplicate websites and 132 inac-
tive NGOs, there were 410 NGOs in the final sample.
Issuecrawler (Rogers 2009) was used to crawl the 410
NGO websites in late January 2014. As this period was
during China’s biggest national holiday, no modification
of NGO websites was expected during the crawling and
downloading period. Before crawling, every NGO web-
site’s homepage, “About Us”page, and “Partner”page
was archived for future reference in case of inconsisten-
cies in the hyperlink data. Issuecrawler crawls each
entered website domain and records all hyperlinks. We
used the inter-actor analysis within Issuecrawler to crawl
NGOs’outlinks and analyze the interlinking patterns
between NGOs. The crawler was run with a page depth
of three and more than three iterations. Issuecrawler was
also used to create the final network of entered domains.
56 J. S. FU AND M. SHUMATE
The hyperlink network was a 410 by 410 matrix of web-
site domains, where 1 indicated the presence of a hyper-
link and 0 indicated absence.
To assess news media visibility in China, we used the Baidu
(Google’sequivalentinChina)newsarchiveservice(http://
www.news.baidu.com) to determine the amount of online
and offline media coverage NGOs received in 2013. To
ascertain social media visibility, we searched for the pro-
files of each NGO on Sina Weibo (http://www.weibo.
com), the largest microblogging platform in China. Sina
Weibo, China’s Twitter equivalent, is a major new media
platform for various civil actors to engage in advocacy
and mobilization activities in China (Huang and Sun
2014). Sina Weibo offers services in both Chinese and
English. For each posting, users can send up to 140 Chi-
nese characters, plus other types of multimedia content.
As of 2015, Sina Weibo has gained more than 800 million
users worldwide; three-quarters of them are Chinese
Internet users. In addition, Sina Weibo has almost
300 million monthly active users, about a quarter of Chi-
na’s population (Lin 2016). For those NGOs with an offi-
cial social media profile, we recorded the number of
followers from their Weibo front page.
Measures
Network structures
As suggested by Lusher, Koskinen, and Robins (2012), three
types of internal network structures were entered as control
variables in the network analysis to address network inter-
dependence. Edge,orchoice,isthesimplelikelihoodof
hyperlink ties being present in the network (Monge and
Contractor 2003). A mutual hyperlink tie represents the
reciprocal ties between two organizations. Cyclical ties mea-
sure the presence of two-path from organization A to orga-
nization C, where A is connected to B, B is connected to C
and C is connected to A (i.e., a complete cycle).
News media visibility
We performed queries in Baidu news search for the
name of each NGO in 2013. The number of returned
results was used to indicate the amount of online and off-
line news media visibility. News media visibility was
transformed using log method to address L-shaped skew-
ness of the variable. This variable was utilized to test
hypotheses 1 and 3.
Social media profile presence
Social media profile presence is a dummy variable that indi-
cates whether an organization has an official organizational
profile on Sina Weibo. Personal Weibo profiles of executive
officers working in NGOs were not counted as an official
profile. This variable was used to test hypothesis 4.
The number of social media followers
The number of followers an NGO has on Sina Weibo
was accessed from the organization’s front page. For
organizations without an official Weibo presence, the
number of followers was zero. Number of social media
followers was transformed using log method to address
the L-shaped skewness of the variable. As with tradi-
tional parametric statistics, exponential random graph
modeling (ERGM) is sensitive to variables that are highly
skewed. The standard error of skewness was reduced by
0.71 Z-score. However, there remains a slight skewness
after the log transformation largely due to those NGOs
without Weibo accounts. This variable was used to test
hypotheses 2 and 5.
Analysis
We used exponential random graph modeling (ERGM)
with Markov-chain Monte Carlo (MCMC) maximum-
likelihood estimation within the R-project package
(Morris, Handcock, and Hunter 2008) to test the
hypotheses. ERGMs allowed us to determine whether
news media and social media presence were related to
the probability of hyperlinking, while simultaneously
accounting for network structures and other variables. If
the hypothesized network structures occurred more
often than by chance alone, then the parameter was posi-
tive and significant. Similarly, significant negative
parameter estimates are associated with negative propen-
sities. Convergence is achieved when MCMC pvalues are
larger than .80, and a parameter is considered significant
when the estimate is 1.96 times larger than the magni-
tude of the standard error (Goodreau et al. 2008). This
indicates that the probability of receiving a hyperlink
based on the structural or attribute parameter is greater
than by chance alone. In this study, all parameters con-
verged in the reported models. In order to test the
hypotheses and account for correlations between the
independent variables, several nested models were cre-
ated to assess the contribution of each hypothesized
parameter.
Results
This study examined the relationship between news
media visibility and social media visibility and the con-
figuration of a hyperlink network of 410 nongovernmen-
tal organization (NGO) websites. There were 1,218 links
among the 410 websites. The average degree centrality
was 2.97, and the overall density was .007, indicating a
sparse network of hyperlinks compared to the number of
all possible hyperlinks (see Table 1). We created six
nested models to test the hypotheses, and the results of
THE INFORMATION SOCIETY 57
these models are reported in Table 2. Compared to the
nested models, the Akaike information criterion (AIC)
and Bayesian information criterion (BIC) statistics of the
final model suggest that our final model is the “best-fit-
ting”model (see Table 2). Except where a suppressor
effect was found (H4), we used the estimates in the final
model to draw conclusions about the hypotheses.
Figure 1 is a network visualization that highlights the
interdependencies between the three systems.
Hypothesis 1 examined the resource dependence
link between news media and hyperlink network sys-
tems. NGO news media visibility was positively
related to the propensity to receive hyperlinks
(MLE D0.36, SE D0.03), supporting hypothesis 1.
In other words, NGOs with more mentions in online
and offline news media were significantly more likely
to receive hyperlinks.
Hypothesis 2 investigated the resource dependence link
between the social media and hyperlink network systems.
Specifically, we hypothesized that the number of social
media followers was positively related to the propensity to
receive hyperlinks, and thishypothesiswassupported
(MLE D0.21, SE D0.03). In short, NGOs with more
social media followers received more hyperlinks.
Hypothesis 3 examined the homophily mechanism link-
ing news media and hyperlink systems. We did not find
support for hypothesis 3 (MLE D0.01, SE D0.03); NGOs
Table 2. ERG Models for NGO hyperlink network related to news media and social media resources and signals.
Baseline model Media visibility Weibo profile Weibo followers Weibo combined Final model
Edges ¡5.20 ¡5.82 ¡5.39 ¡6.05 ¡5.75 ¡5.66
(0.03)
!!
(0.05)
!!
(0.06)
!!
(0.09)
!!
(0.14)
!!
(0.13)
!!
Mutual 3.26 3.14 3.24 3.22 3.26 3.13
(0.12)
!!
(0.12)
!!
(0.12)
!!
(0.13)
!!
(0.12)
!!
(0.12)
!!
Cyclical ties 0.72 0.63 0.71 0.69 0.72 0.62
(0.05)
!!
(0.05)
!!
(0.05)
!!
(0.05)
!!
(0.05)
!!
(0.05)
!!
H1: Media visibility (RDT) 0.43 0.36
(0.02)
!!
(0.03)
!!
H2: Weibo followers (RDT) 0.28 0.37 0.21
(0.02)
!!
(0.03)
!!
(0.03)
!!
H3: Media visibility (HOM) ¡0.01 0.01
(0.03) (0.03)
H4: Weibo PROFILE (Yes) 0.29 ¡0.57 ¡0.61
(0.06)
!!
(0.13)
!!
(0.13)
!!
H4: Weibo profile (No) ¡0.09 0.27 ¡0.23
(0.18) (0.22) (0.22)
H5: Weibo followers (HOM)
A
¡0.01 ¡0.13 ¡0.19
(0.02) (0.04)
!!
(0.04)
!!
AIC 13918 13490 13894 13726 13698 13416
BIC 13948 13540 13944 13777 13768 13507
Note. Double asterisk indicates p<.001, which indicates significant parameter estimates. RDT Dresource dependence theory; HOM Dhomophily. A, Social media
follower homophily was measured as the difference between two organizations’social media followers, so a negative parameter estimate suggests homophily.
AIC (Akaike information criterion) and BIC (Bayesian information criterion) are two standard measures of model fit based on likelihood—smaller values are
better.
Table 1. Descriptive statistics of NGO hyperlink networks (410 £
410).
Network measures Figure
Density .007
Average degree 2.971
Number of ties 1218
Number of mutual ties 87
Number of asymmetric ties 1,044
Number of isolates 16
Number of active NGOs 410
Number of domains before extracting NGO network 542
Figure 1. Chinese NGO hyperlink network. The gradations of the
red color indicate high level of social media followership,
medium level of social media followership, low level of social
media followership, and no social media profile, respectively. The
size of the node represents the level of news media visibility.
58 J. S. FU AND M. SHUMATE
with similar level of news media visibility were not more
likely to hyperlink to each other than by chance alone.
Hypotheses 4 and 5 examined the homophily link
between social media and hyperlink network systems. In
the final model, NGOs with social media profiles were
less likely to hyperlink to each other (MLE D¡0.61, SE
D0.13). However, in the Weibo profile model (see col-
umn 3), NGOs with social media profiles were more
likely to hyperlink to each other than by chance alone
(MLE D0.29, SE D0.06). The difference between the
two models was likely due to the suppression effect that
the social media follower variable had on the social
media profile variable (see Conger 1974; MacKinnon,
Krull, and Lockwood 2000). These variables are logically
related: All NGOs without a social media profile had
zero social media followers. Thus, hypothesis 4 was sup-
ported. Similarly, we found support for hypothesis 5,
that NGOs with similar volumes of followers were
more likely to hyperlink to each other (MLE D¡0.19,
SE D0.04). Homophily mechanisms link the social
media and hyperlink network systems.
After assessing the significance of the individual parame-
ters, the overall goodness of fitofthefinal model was fur-
ther tested. Table 3 presents the results of the goodness of
fitforestimatedparametersincludedinthefinal model. It
shows how well the theoretical model provided a good
representation of the observed network, manifested in the
high MCMC pvalue. Figure 2 depicts the results of 100
Table 3. Goodness-of-fit (GOF) results for model statistics for
“best fitting”model.
Observed Minimum Mean Maximum
MCMC
p-value
Edges 1218 1100 1218.77 1332 0.94
Mutual 87 62 88.95 121 0.84
Cyclical ties 177 122 178.42 258 0.96
Weibo profile
(yes)
885 788 886.30 994 0.90
Weibo profile
(no)
32 19 31.47 46 0.96
Media visibility
(HOM)
1888.28 1669.26 1893.32 2186.51 0.84
Media visibility
(RDT)
2367.79 2068.59 2373.97 2749.16 0.92
Weibo followers
(HOM)
1904.86 1700.05 1906.02 2119.63 0.96
Weibo followers
(RDT)
4103.05 3670.81 4110.70 4561.08 0.82
Note. A good GOF of estimated parameters is indicated by MCMC pvalue
equal or greater than 0.80.
Figure 2. Goodness of fit results for additional indegree and edge-wise shared partners parameters related to patterns in the NGO
hyperlink network. Simulation results for the final model. In the plot, the vertical axis is the log-odds of relative frequency. The solid line
represents the observed statistics from the sample network; the boxplots include the median and interquartile range; and the light gray
lines display the range in which 95% of simulated observations fall (see Hunter, Goodreau, and Handcock 2008).
THE INFORMATION SOCIETY 59
simulations for the observedChineseNGOhyperlinknet-
work final model in Table 2.Figure 2 visually shows the
goodness of fitforadditionalnetworkdimensions:indegree
distribution and edge-wise shared partners. They indicate
that the final model performs relatively well for the indegree
distribution and edge-wise shared partners distribution (see
Goodreau 2007;Hunter,Goodreau,andHandcock2008).
Thus, the final model represented a good explanation for
the indegree distribution and edge-wise shared partners.
That means the theoretical model based on resource depen-
dence theory and homophily theory adequately accounted
for the variability in the number of inward connections that
each organization in the network had and the propensity of
organizations within the network to have homogeneous
shared partners (Morris, Handcock, and Hunter 2008).
Discussion
This study develops the idea that hyperlink networks are
open systems that interact with the news media and social
media systems. Although some prior studies have implicitly
assumed that social media are related to websites (e.g.,
Kwak et al. 2010;NahandSaxton2013), this study breaks
new ground by spotlighting how different media systems
are connected to each other. Drawing on research on
resource dependence and homophily mechanisms, it intro-
duces the notion of integrated media effects. It also empiri-
cally investigates the integrated media effects by
interrogating the relationship between social media and
news media resources on NGO hyperlink network configu-
ration. In doing so, it presents some preliminary evidence
for the theoretical perspective introduced.
First, this study finds resources developed in the news
media and social media systems are related to the pattern
of hyperlinking. Specifically, resources (i.e., visibility and
popularity) generated in the news media and social
media systems induce hyperlinking to these organiza-
tions’websites. Consistent with previous studies (Pilny
and Shumate 2012; Gonzalez-Bailon 2009a;2009b), our
study finds not only that organizations are aware of how
often other organizations are mentioned in offline and
online news media, but also that the number of those
mentions increases the likelihood that organizations will
receive hyperlinks. Further, the number of social media
followers an organization has is positively related to the
propensity to attract hyperlinks. Our findings suggest
that hyperlinking is aspirational across news and social
media systems, such that higher news media visibility
and more social media followers drive hyperlinks above
and beyond structural effects. Other organizations in the
social network aspire to be affiliated with organizations
with higher news media visibility and social media
visibility.
Second, the homophily integrated media effect, how-
ever, did not uniformly link the news media and social
media systems to the hyperlink network systems.
Namely, we failed to find support for the homophily
integrated media effect for similar news media visibility
on hyperlinks. To identify visible NGOs from news
media, NGOs only need to develop a general sense of
whether the NGOs were mentioned and how frequently,
if they were mentioned. In contrast, an organization’s
perception of another organization’s similar level of
news media visibility requires efforts to keep track of the
concrete mentions of both NGOs and to constantly
make comparisons. Thus, news media visibility serves as
an indicator of organizational resource possession more
than a general organizational attribute across which
organizations signal similarities.
In contrast, we found support for the homophily inte-
grated media effect linking social media and hyperlink
network systems. Organizations that had both a social
media profile and a similar number of social media fol-
lowers are significantly more likely to hyperlink to each
other. In contrast with news media visibility, determin-
ing whether an organization has a social media profile
and its number of followers is relatively easy for organi-
zations. One simply needs to search for the organization
and look at the front page of its profile on social media
(if it has one). As such, homophily integrated media
effects across systems seem to depend on the accessibility
of the similarity signals in that media system. If those sig-
nals are easily quantified in one media system, status
similarity seems to play a larger role than when commu-
nicative signals are more difficult to gauge. Compared to
news media visibility, social media provides technologi-
cal affordances (boyd and Ellison 2008) for organizations
to navigate outcomes, evaluate impact, and make organi-
zational status and signals not previously visible in news
media accessible and quantifiable.
In sum, the magnitude of resources organizations pos-
sess in the news media system affects the pattern of
hyperlinks among organizational websites through
resource dependence integrated media effects. The vol-
ume of organizational resources and signals of organiza-
tional status on social media is related to the pattern of
hyperlinks among organizational websites through both
resource dependence and homophily integrated media
effects.
This article makes three contributions to Internet
studies research generally and hyperlink network
research in particular. First, in the age of Web 2.0, this
article introduces a media ecology perspective and inte-
grated media effects on hyperlink networks specifically
and organizational presence across online and offline
media more generally. The integrated media effects
60 J. S. FU AND M. SHUMATE
introduced have the potential to further researchers’
understanding of individuals’and organizations’pres-
ence across various platforms on the Internet. Building
on the media ecology literature (McLuhan 1964;
Nystrom 1973), this article opens up a new theoretical
perspective on the various ways different media systems,
offline and online, are interconnected. Departing from
the structural signatures perspective, the media ecology
perspective suggests that hyperlink networks are not self-
organizing systems driven primarily by iterative net-
works interdependencies and evolution. Similarly, in
contrast to the offline organizational attribute perspec-
tive, the media ecology perspective proposes that organi-
zations socially construct resources and similarity signals
on the Internet (e.g., website and social media) that do
not exist offline.
Second, this article specifies how media systems in the
digital age are intertwined through the resource depen-
dence (Pfeffer and Salancik 1978) and homophily
(McPherson, Smith-Lovin, and Cook 2001) integrated
media effects. Resource dependence connects both news
media and social media systems with the hyperlink net-
work system. In contrast, the homophily mechanism
only connects social media systems to the hyperlink net-
work system. Moreover, the resource dependence effects
are stronger than the homophily effects in our models.
This suggests resource dependence integrated media
effects may be more influential than homophily inte-
grated media effects for organizations’public presence.
Therefore, resources possessed by organizations in media
systems are stronger cues for organizations’Web behav-
iors than similar status signals.
Finally, this article uniquely identifies similarities and
differences in the ways that social media and news media
are related to hyperlink networks. To be more specific,
the news media system is only connected to the hyper-
link network system via the resource dependence inte-
grated media effect. In contrast, both resource
dependence and homophily integrated media effects
connect the social media and hyperlink network systems.
We attribute this difference to the affordances offered by
social media (boyd and Ellison 2008). In particular,
organizations can more easily detect resources accumu-
lated in the social media system than in the news media
system. Status similarity, the underlying homophily
effect investigated here, is more easily signaled and ascer-
tained in the social media system because the number of
followers is prominently quantified.
Limitations
This study has three notable limitations that may be seen
as opportunities for future research. First, it focuses on
one sample of organizations in one country. As such,
future research should verify the conclusions of this
study among other populations of organizations, in other
countries, and across other news media and social media
platforms. In addition, we only investigated the inte-
grated media effects among NGOs, thus limiting the gen-
eralizability to other types of organizations such as
businesses and governments.
Second, this study adopts a rough measure of online
and offline media visibility based on the Baidu news
archive. As Pilny and Shumate (2012) suggest, measures
that only count the number of articles that mention an
organization fail to take into account the valence of the
coverage (whether or not selected NGOs were the main
subject of the coverage) and the level of detail provided.
Future research might consider these factors when con-
sidering the impact of news media coverage on social
media and hyperlink network systems.
Third, our measure of social media visibility only
focuses on the official NGO profile and only captures
one relatively basic measure of popularity. An organiza-
tion’s holistic media presence also includes the public
presence of its high-ranking executive officers, and such
accounts should likely be investigated. Further, zombie
and bogus microblog accounts are present on Weibo
(Liu et al. 2013); thus, organizations’actual level of popu-
larity measured by the number of followers may be
biased. Moreover, future research should consider other
types of social media resources including total number of
posts and friends, #hashtags, and @mentions, as well as
retweets and other types of social media activities.
Fourth, this study empirically examines the relation-
ship between resources developed in one media system
and hyperlinking patterns. We use a cross-sectional
design to understand how these resources are related to
the probability of hyperlinks between two NGOs.
Although any direction of relationship would support
our argument of integrated media effects (i.e., from
another media system to hyperlinks, from hyperlinks to
another media system, bidirectional effects), we cannot
rule out the possibility of a spurious relationship between
the variables. We have controlled for interdependencies
in the hyperlink network as one possibility, but as with
most field research, cross-sectional designs do not allow
for the elimination of all other explanations.
Conclusion
This article introduces a media ecology perspective on an
organization’s online presence, investigating the factors
across media systems that are mutually influential. The
media ecology perspective put forth in this article sug-
gests that news media, social media, and websites should
THE INFORMATION SOCIETY 61
not be seen or researched as independent systems. They
form an integral ecosystem and have integrated media
effects. More generally, this article suggests that research-
ers have much to learn about the interdependencies
among media systems beyond the effects of resources
and similarity signals in news media and social media
systems on hyperlink networks, which are the focus of
this particular article. We suggest that hyperlink net-
works and news media systems may influence both
behaviors and outcomes in social media systems, and
Web presence and social media variables may have an
ever-increasing influence on news media visibility.
Future research should attend to these other types of
integrated media effects.
In a multiplex media environment, organizations inter-
act with the publics using a variety of media. Public rela-
tions professionals have long argued that an integrated
media plan that connects the organization’simageacross
earned, paid, social, and owned media platforms is good
practice (Hallahan 2001). We hope that this article ini-
tiates a new line of research that examines the interdepen-
dencies of organizational behavior as they interact with
audiences using different channels, including websites, off-
line and online news media, and social media.
Note
1. This means that if organization A hyperlinks to organiza-
tion B, organization B will also hyperlink to organization
A (i.e., reciprocal tie). Triad describes the presence of at
least one hyperlink for each pair of organizations for a
group of three organizations.
References
Ackland, R., and M. O’Neil. 2011. Online collective identity:
The case of the environmental movement. Social Networks
33 (3):177–90.
Atouba, Y. C., and M. Shumate. 2015. International nonprofit
collaboration: Examining the role of homophily. Nonprofit
and Voluntary Sector Quarterly 44 (3):587–608.
Barnett, G. A. 2008. The role of the Internet in cultural iden-
tity. In Embedding into our lives: New opportunities and
challenges of the internet, ed. L. W. Leung, A. Y. H. Fung,
and P. S. N. Lee, 347–68. Hong Kong: Chinese University
Press.
Boyd, D. M., and N. B. Ellison. 2008. Social network sites: Defi-
nition, history, and scholarship. Journal of Computer-Medi-
ated Communication 13:210–230.
Conger, A. J. 1974. A revised definition for suppressor varia-
bles: A guide to their identification and interpretation. Edu-
cational and Psychological Measurement 34:35–46.
Fu, J. S., and M. Shumate. 2015. Social media activity and
hyperlink network analysis: A holistic media ecology per-
spective. Proceedings of the 48th Hawaii International
Conference on System Sciences (HICSS), 1808–17. New
York, NY: IEEE.
Gonzalez-Bailon, S. 2009a. Opening the black box of link for-
mation: Social factors underlying the structure of the web.
Social Networks 31:271–280.
Gonzalez-Bailon, S. 2009b. Traps on the Web: The impact of
economic resources and traditional news media on online
trafficflow. Information, Communication and Society 12
(8):1149–73.
Goodreau, S. M. 2007. Advances in exponential random graph
(p
!
) models applied to a large social network. Social Net-
works 29:231–48.
Goodreau, S. M., M. S. Handcok, D. R. Hunter, C. T. Butts, and
M. Morris. 2008. A statnet tutorial. Journal of Statistic Soft-
ware 24 (9):1–26.
Guo, C., and M. Acar. 2005. Understanding collaboration
among nonprofit organizations: Combining resource
dependency, institutional, and network perspectives. Non-
profit and Voluntary Sector Quarterly 34 (3):340–61.
Hallahan, K. 2001. Strategic media planning: Toward an inte-
grated public relations media model. In Handbook of public
relations, ed. R. Heath and G. Vasquez, 461–470. Thousand
Oaks, CA: Sage.
Huang, R., Y. Gui, and X. Sun. 2015. Inter-organizational net-
work structure and formation mechanisms in Weibo space:
A study of environmental NGOs. Chinese Journal of Sociol-
ogy 1 (2):254–78.
Huang, R., and X. Sun. 2014. Weibo network, information dif-
fusion and implications for collective action in China. Infor-
mation, Communication & Society 17 (1):86–104.
Hunter, D. R., S. M. Goodreau, and M. S. Handcock. 2008.
Goodness of fit of social network models. Journal of the
American Statistical Association 103:248–58.
Hsu, C., and H. W. Park. 2011. Sociology of hyperlink net-
works of Web 1.0, Web 2.0, and Twitter: A case study of
South Korea. Social Science Computer Review 29 (3):354–68.
Khagram, S., J. V. Riker, and K. Sikkink. 2002. Restructuring
world politics: Transnational social movements, networks,
and norms. Minneapolis, MN: University of Minnesota
Press.
Kim, J. H. 2012. A hyperlink and semantic network analysis of
the triple helix (University–government–industry): The
interorganizational communication structure of nanotech-
nology. Journal of Computer-Mediated Communication
17:152–70.
Kleinberg, J. 1999. Authoritative sources in a hyperlinked envi-
ronment. Journal of the ACM 46:604–32.
Kwak, H., C. Lee, H. Park, and S. Moon. 2010. What is Twitter?
A social network or a news media? In WWW’10: Proceed-
ings of the 19th International Conference on World Wide
Web, 591–600. Washington, DC: ACM.
Lai, C.-H., She, B., and Ye, X. 2015. Unpacking the network
processes and outcomes of online and offline humanitarian
collaboration. Communication Research, prepublished
online December 21, 2015. doi:10.1177/0093650215616862
Lazarsfeld, P. F., and R. K. Merton. 1954. Friendship as a social
process: A substantive and methodological analysis. In Free-
dom and control in modern society, ed. M. Berger, T. Abel,
and C. Page, 18–66. New York, NY: Van Nostrand.
Lewis, K., Gonzalez, M., and Kaufman, J. 2012. Social selection
and peer influence in an online social network. Proceedings
of the National Academy of Sciences 109 (1):68–72.
62 J. S. FU AND M. SHUMATE
Lin, P. 2016. China digital marketing predictions. We Are
Social.http://wearesocial.cn (accessed February 4, 2016).
Liu, H., Y. Zhang, H. Lin, et al. 2013. How many zombies
around you? In Data Mining (ICDM):2013 IEEE 13th Inter-
national Conference on Data Mining, 1133–38. New York,
NY: IEEE.
Lovejoy, K., and G. D. Saxton. 2012. Information, community,
and action: How nonprofitorganizationsusesocialmedia.
Journal of Computer-Mediated Communication 17 (3):337–53.
Lovejoy, K., R. D. Waters, and G. D. Saxton. 2012. Engaging
stakeholders through Twitter: How nonprofit organizations
are getting more out of 140 characters or less. Public Rela-
tions Review 38 (2):313–18.
Lusher, D., and R. Ackland. 2011. A relational hyperlink analy-
sis of an online social movement. Journal of Social Structure
12. http://www.cmu.edu/joss/content/articles/volume12/
Lusher (accessed December 3, 2016).
Lusher, D., J. Koskinen, and G. Robins. 2012. Exponential ran-
dom graph models for social networks: Theory, methods, and
applications. Cambridge, UK: Cambridge University Press.
Ma, Q. 2002. Defining Chinese nongovernmental organiza-
tions. Voluntas: International Journal of Voluntary and
Nonprofit Organizations 13 (2):113–30.
MacKinnon, D. P., J. L. Krull, and C. M. Lockwood. 2000.
Equivalence of the mediation, confounding and suppression
effect. Prevention Science 1 (4):173–81.
McLuhan, M. 1964. Understanding media: The extensions of
man. Toronto, ON, Canada: University of Toronto Press.
McLuhan, M., and Q. Fiore. 1967. The medium is the message.
New York, NY: Random House.
McPherson, M., L. Smith-Lovin, and J. M. Cook. 2001. Birds of
a feather: Homophily in social networks. Annual Review of
Sociology 27:415–44.
Monge, P. R., and N. S. Contractor. 2003. Theories of commu-
nication networks. New York, NY: Oxford University Press.
Morris, M., M. S. Handcock, and D. R. Hunter. 2008. Specifica-
tion of exponential-family random graph models: Terms
and computational aspects. Journal of Statistical Software
24 (4). https://www.jstatsoft.org/search/search?simpleQueryD
handcock&searchFieldDquery (accessed December 3, 2016).
Nah, S., and G. D. Saxton. 2013. Modeling the adoption and
use of social media by nonprofit organizations. New Media
& Society 15 (2):294–313.
Nystrom, C. 1973. Towards a science of media ecology: The for-
mulation of integrated conceptual paradigms for the study of
human communication systems. Doctoral dissertation, New
York University, New York, NY.
Park, H. W. 2003. Hyperlink network analysis: A new method
for the study of social structure on the web. Connections 25
(1):49–61.
Park, H. W., G. A. Barnett, and I. Y. Nam. 2002. Interorganiza-
tional hyperlink networks among websites in South Korea.
Networks and Communication Studies 16 (3–4):155–74.
Peng, T. Q., Liu, M., Wu, Y., and Liu, S. 2015. Follower–
followee network, communication networks, and vote
agreement of the U. S. members of congress. Communica-
tion Research 43 (7):996–1024.
Pennock, D. M., G. W. Flake, S. Lawrence, S, et al. 2002. Win-
ners don’t take all: Characterizing the competition for links
on the web. Proceedings of the National Academy of Sciences
99 (8):5207–11.
Pfeffer, J., and G. R. Salancik. 1978. The external control of
organizations: A resource dependence perspective. New
York, NY: Harper Row.
Pilny, A., and M. Shumate. 2012. Hyperlinks as extensions of
offline instrumental collective action. Information, Commu-
nication and Society 15 (2):260–86.
Postman, N. 1970. The reformed English curriculum. In High
school 1980: The shape of the future in American secondary
education, ed. A. C. Eurich, pp. 160–68. New York, NY:
Pitman.
Rogers, R. 2009. Mapping of the public space on the Web using
the Issuecrawler. In Digital cognitive technologies: Episte-
mology and knowledge society, ed. C. Brossard and B. Reber,
pp. 115–26. London, UK: John Wiley.
Rogers, R., and N. Marres. 2000. Landscaping climate change:
A mapping technique for understanding science and tech-
nology debates on the World Wide Web. Public Under-
standing of Science 9:141–63.
Rossi, L., and M. Magnani. 2012. Conversation practices and
network structure in Twitter. Proceedings of the Sixth Inter-
national AAAI Conference on Weblogs and Social Media,
563–66. Palo Alto, CA: AAAI Publications.
Scolari, C. A. 2012. Media ecology: Exploring the metaphor to
expand the theory. Communication Theory 22:204–25.
Seo, H., J. Y. Kim, and S. Yang. 2009. Global activism and new
media: A study of transnational NGOs’online public rela-
tions. Public Relations Review 35 (2):123–26.
Shumate, M., and N. Contractor. 2013. The emergence of mul-
tidimensional social networks. In The SAGE handbook of
organizational communication, ed. L. L. Putnam, and D. K.
Mumby, 3rd ed., 449–74. Thousand Oaks, CA: Sage.
Shumate, M., and L. Dewitt. 2008. The North/South divide in
NGO hyperlink networks. Journal of Computer-Mediated
Communication 13 (2):405–28.
Shumate, M., and J. Lipp. 2008. Connective collective action
online: An examination of the hyperlink network structure
of an NGO issue network. Journal of Computer-Mediated
Communication 14 (1):178–201.
Stewart, K. J. 2003. Trust transfer on the World Wide Web.
Organization Science 14 (1):5–17.
Strate, L. 2006. Echoes and reflections: On media ecology as a
field of study. Cresskill, NJ: Hampton.
Wells, J. D., J. S. Valacich, and T. J. Hess. 2011. What signal are
you sending? How website quality influences perceptions of
product quality and purchase intentions. MIS Quarterly 35
(2):373–96.
Wu, L., and R. Ackland. 2014. How Web 1.0 fails: The mis-
match between hyperlinks and clickstreams. Social Network
Analysis and Mining 4 (1):1–7.
THE INFORMATION SOCIETY 63
