ArticlePDF Available

Abstract and Figures

How do nonprofits advocate and shape climate conversations on Twitter? We answer this question by combining computational analyses with thick descriptions of discursive data to analyze message diffusion on Twitter. We first map a temporal message similarity network comprising 298,073 unique tweets sent by climate action and obstruction nonprofits. We then identify four leading nonprofits and trace their message similarity to 2,479 accounts over 2 weeks. Our results suggest that while climate obstruction nonprofits might not be frequent tweeters, their voices are highly reciprocal in the Twitterverse. We also find that messages of either side are most echoed by the public rather than elite audiences. Although diffusion to policymakers is almost absent, we uncover high semantic similarities between messages from climate obstruction nonprofits and bot-like accounts. Our analyses contribute to new theoretical and empirical insights into the roles of nonprofit conversation leaders and their potential message diffusion in climate discourse.
Content may be subject to copyright.
Nonprofit and Voluntary Sector Quarterly
1 –30
© The Author(s) 2023
Article reuse guidelines:
DOI: 10.1177/08997640231174048
Who Leads and Who
Echoes? Tracing Message
Similarity Network of
#ClimateChange Advocacy
on Twitter
Viviana Chiu Sik Wu1 and Weiai Wayne Xu1
How do nonprofits advocate and shape climate conversations on Twitter? We
answer this question by combining computational analyses with thick descriptions
of discursive data to analyze message diffusion on Twitter. We first map a temporal
message similarity network comprising 298,073 unique tweets sent by climate action
and obstruction nonprofits. We then identify four leading nonprofits and trace their
message similarity to 2,479 accounts over 2 weeks. Our results suggest that while
climate obstruction nonprofits might not be frequent tweeters, their voices are highly
reciprocal in the Twitterverse. We also find that messages of either side are most
echoed by the public rather than elite audiences. Although diffusion to policymakers
is almost absent, we uncover high semantic similarities between messages from
climate obstruction nonprofits and bot-like accounts. Our analyses contribute to new
theoretical and empirical insights into the roles of nonprofit conversation leaders and
their potential message diffusion in climate discourse.
climate change, nonprofit advocacy, computational text analysis, network analysis, Twitter
Social media platforms have become a new frontier for climate advocacy whereby
climate obstructionists and nonprofit advocates strive to influence public opinion and
climate policy (Marlow et al., 2021; Pearce et al., 2019). Nevertheless, social media
1University of Massachusetts Amherst, USA
Corresponding Author:
Viviana Chiu Sik Wu, Assistant Professor, School of Public Policy, University of Massachusetts Amherst,
Room 628 Thompson Hall, 200 Hicks Way, Amherst, MA 01003, USA.
1174048NVSXXX10.1177/08997640231174048Nonprofit and Voluntary Sector QuarterlyWu and Xu
2 Nonprofit and Voluntary Sector Quarterly 00(0)
offer a very “noisy information environment” where gaining attention from stakehold-
ers is highly competitive if not challenging (Guo & Saxton, 2018). Furthermore, it
becomes difficult to ascertain who leads the climate discourse and how the messages
might diffuse across stakeholders (Bail, 2016). Although a burgeoning line of research
has begun to study climate advocacy on Twitter, little do we know the extent to which
climate advocacy nonprofits of opposing camps might shape and lead the Twitter con-
versations about their cause.
Adjudicating the question of message diffusion on Twitter—“who leads, who
echoes”—is theoretically important for expanding the nonprofit advocacy literature in
part because social media advocacy is distinct from traditional offline advocacy. On
one hand, social media networks tend to follow a power-law distribution, whereby the
messages sent by a few central players inspire far-ranging conversation, yet most
social media users receive little or no attention (Bail, 2016; Guo & Saxton, 2020).
Research on the “who leads” question will reveal the source of influence in social
media-based advocacy. On the other hand, different from centralized and hierarchical
offline advocacy, decentralized social media form networked publics (Ausserhofer &
Maireder, 2013), a mediated public sphere where diverse stakeholders across sectors,
social spheres, and geographies contribute voices, gain attention, and set agendas.
Social media’s network effect could create cascading message diffusion and mobiliza-
tion, which later translate into political actions and policy changes (Bennett &
Segerberg, 2012; Jackson et al., 2020; Jung & Valero, 2016). Hence, research on “who
echoes” illustrates how such a cascade effect of message diffusion may occur in com-
plex, multistakeholder online environments. In short, identifying the leading actors
and patterns of how their voices might diffuse to potential audiences allows us to
gauge the role of climate nonprofits in social media-based advocacy (Halpin et al.,
2021; Jung & Valero, 2016; Yang & Saffer, 2018). Practically speaking, research on
the “who leads and who echoes” question contributes an analytical tool to quantify,
visualize, and audit an individual nonprofit’s influence in online public arenas, with
the insights informing stakeholder engagement strategies on Twitter.
Although the “who leads, who echoes” question of social media advocacy has
gained scholarly interests across social science disciplines (Freelon, 2018; Jackson &
Welles, 2015), including the issue of climate change (Pearce et al., 2019; Vu et al.,
2019; Wang et al., 2019), there generally lacks a specific focus on nonprofits’ roles in
leading or echoing online public discourse. This coincides with the limited theoriza-
tion of nonprofits’ online advocacy in the context of networked public sphere in the
nonprofit literature (Ausserhofer & Maireder, 2013). Networked public sphere or net-
worked publics is a macrolevel theory describing the interdependency and rivalry that
exist among social media users who form networks of influence for political and dis-
cursive power. Nonprofits’ social media advocacy is hinged upon mobilizing multi-
stakeholder networks for relaying voices and agendas, while simultaneously, nonprofits
compete with many stakeholders for the scarce commodity of online attention, par-
ticularly advocates with opposing ideologies (Farrell, 2015; Guo & Saxton, 2018). On
climate change, proindustry, climate obstruction nonprofits, such as trade associations
and conservative think tanks, are known to be driving and funding environmental
Wu and Xu 3
lobbying. They serve as the key proponents of the climate change counter-movements
(CCCMs) to cast doubts, deny climate issues and advocate for market solutions to
climate issues (Brulle, 2014; Brulle et al., 2021; Farrell, 2016). Proenvironment non-
profits and civil society actors also play defining roles in the U.S. environmental and
climate movements by framing public discourses, strengthening political lobbying,
and promoting public awareness of environmental issues (Brulle, 2018; Brulle et al.,
2012). Although the rivalry between climate action and obstruction nonprofits are
known to scholars, how they coexist and compete in the same advocacy field (particu-
larly on Twitter) remains largely unknown.
To explore nonprofits’ roles in influencing online climate discourse, the current
research connects computational analysis with thick descriptions of discursive data for
studying diffusion in climate discourse. Specifically, we lean on computational text
analysis and network analysis to temporally track and identify influential nonprofits
and tweets for qualitative assessment. In particular, we adopt Bail’s (2016) notion of
cultural network as the blueprint and introduce a different type of diffusion network,
not based on the flow of the same information but on the flow of homogeneous ideas
and agendas. We argue that advocacy organizations create “cultural bridges” in poten-
tially leading and shaping climate discourse when their online messages share similar
themes with other stakeholders. We focus on the advocacy efforts of opposing climate
nonprofits and ask the following question: What do messages from climate nonprofits
of opposing camps reveal about their role in shaping the climate conversations on
Twitter among a broad range of stakeholder groups?
Literature Review
Climate Advocacy in Online Networked Publics
Social media have become a key battlefield for shaping public discourse on climate
change (Kirilenko & Stepchenkova, 2014; Pearce et al., 2019). Just as traditional
media create public spheres for policy deliberations and civic participation, social
media afford a virtual public sphere for social exchanges, theorized as networked pub-
lics (Bruns & Highfield, 2015). The defining features of such networked publics
include low barrier to participation, diffusion, and collective actions through decen-
tralized online social networks (Bennett & Segerberg, 2012), as well as participation
from different spheres with distinct interests and often opposing ideologies (Bruns &
Highfield, 2015). These characteristics offer nonprofits ample opportunities previ-
ously unavailable in offline advocacy. Although climate advocacy takes place in vari-
ous social media platforms, Twitter is one of the prominent digital forums to study
networked conversations and climate change publics—whereby publics seek out and
discuss scientific issues, share their opinions, and voluntarily engage in climate con-
versations (Anderson & Huntington, 2017; Pearce et al., 2019).
Climate nonprofit advocacy’s success hinges on the ability to mobilize resources,
rally allies, and build an agenda for the public discourse about climate change
(Pearce et al., 2014; Yang & Saffer, 2018). Social media’s various interactive
4 Nonprofit and Voluntary Sector Quarterly 00(0)
features (e.g., directed messaging, broadcasting, retweeting) can be used to build
alliance networks with other policy actors to pool resources and magnify their col-
lective voices on climate issues (Saffer et al., 2019). Advocacy groups can directly
tap into extensive and diverse networks of stakeholders at a much lower cost.
Because of the openness of the platforms, a broad array of stakeholders participate,
alongside nonprofits, on the same platform. Social media enable personalized
engagement that leads actions and content to be distributed widely across loosely
organized social networks, particularly “diffused stakeholders” and “issue publics”
who might not be otherwise visible, salient, or identifiable in traditional offline set-
tings (Bennett & Segerberg, 2012; Turunen & Weinryb, 2020).
Furthermore, social media’s promise is that it could level the playing field, for
grassroots groups in particular, by enabling networked conversations without the lim-
its of space and time or the resources to sway traditional mass media (Ausserhofer &
Maireder, 2013). Because of the social networking functions, advocates do not need to
first convince journalists and newspaper editors of their value. Instead, they can broad-
cast their messages to their followers, and anyone can join their conversations directly.
It is important to note that social media platforms are not universally adopted (due to
the digital divide) and equally used by different demographics. Twitter users are
younger, more likely to identify as Democrats, more educated, and have higher
incomes than U.S. adults overall (Wojcik & Hughes, 2019). Yet, Twitter, despite not
being a representative platform, is prominently used by elite actors (e.g., journalists,
think tanks, politicians, celebrities, etc.) who likely set media and public agendas.
Evidence from past social movements show how grassroots organizers use Twitter
discourse to draw attention, influence journalists on the ground, and reset media
agenda, which further cascade into broader policy changes (Jackson et al., 2020).
Inspired by these possibilities, a burgeoning line of research examines advocacy tac-
tics and stakeholder engagement on social media (Halpin et al., 2021; Guo & Saxton,
2014). Although such research has generated insights into how social media are relevant
to nonprofit advocacy on climate issues, the literature suffers from three key limitations.
First, social media networks tend to follow a power-law distribution, whereby the con-
tent produced by a few key players receives most of the attention (Guo & Saxton, 2020;
Muchnik et al., 2013). A wealth of studies have examined the role of central players (i.e.,
legislator, influencer, or opinion leaders) in shaping discourse and influencing public
opinion (Barberá et al., 2019; Jackson et al., 2020; Meraz & Papacharissi, 2013).
However, most studies reveal the central players to be activists, journalists, and even
anonymous influencers (Jackson et al., 2020), leaving the role of nonprofit advocacy
groups underdiscussed. It is possible that nonprofits’ digital influence is marginal rela-
tive to other influencers and stakeholders (Vu et al., 2019), yet its role should not be
overlooked as nonprofits actively use social media to shape public opinion and dis-
course. They engage constituencies not only through targeted messaging and connec-
tion-building but also through introducing new topics, agendas, and ideas to the public
(Bail, 2016; Yang & González-Bailón, 2017). If discourse is defined as a dialectic strug-
gle to denote—the social production of meaning—then different types of topical themes
introduced by climate advocates contribute to shaping climate discourse in different
ways (Pond & Lewis, 2019). Bourdieu (1989) has nicely summarized the essence of
Wu and Xu 5
advocacy through the following quote: “To change the world, one has to change the
ways of world-making” (pp. 23). In other words, advocating nonprofits that want to
change the status quo needs to transform “the ways of world-making.” To do so, taking
initiatives in leading and shaping public discourse on Twitter can be an important first
step. In short, extant studies have yet to theorize or examine the central nonprofit players
in the online climate discourse and whom they have influenced through their content.
The second gap relates to how leading and echoing are operationalized and mea-
sured. Extant nonprofit and climate research has often understood and measured stake-
holder engagement or network influence on social media by the number of unique
actors who have “liked,” “retweeted,” “mentioned,” or “hashtagged” (e.g., Agostino &
Arnaboldi, 2016; Guo & Saxton, 2018; Halpin et al., 2021; Lam & Nie, 2020). These
studies discovered topics about climate conversations and consistently found that
users often converse with users holding similar views within polarizing echo chambers
(Jasny et al., 2015; Pearce et al., 2014; Williams et al., 2015). Traditional news sources
and news aggregators were the most referenced ones (Kirilenko & Stepchenkova,
2014). Bots also played a critical role in manipulating and propagating obstruction
narratives (Marlow et al., 2021). Although these engagement measures can serve as
immediate proxies of how the same social media posts are diffused across users, these
engagement metrics cannot reveal how similar ideas, agendas, and topics are diffused
and echoed in continuous streams of social media discourse across users. Increasingly,
leading online discourse means not just attracting likes and shares but also introducing
topics, agendas, and ideas that most resonate with other stakeholders (Bail, 2016). In
particular, advocacy organizations can inspire far-ranging conversation by creating
“cultural bridges” through sharing similar topics that put them into conversation with
each other (Bail, 2016). Through shaping the directions, topics, and salience of public
discourse, they facilitate agenda setting in both Twitterverse and climate policy com-
munities (Farrell, 2015; Feezell, 2018; Yang & Saffer, 2018).
Third, while scholars have paid increased attention to climate debates and discourse
on social media, these studies usually limit their investigation to a single side of the
climate narrative (Almiron et al., 2020; Farrell, 2015), for example, by focusing on
either climate action movement or counter-movement activities (cf. Hoffman, 2015;
Pearce et al., 2014). The literature has not come to a consensus on the degree to which
climate discourse is politicized, with scholars providing conflicting accounts (Wetts,
2020). Advocacy is not an isolated action of a single organization, as other stakehold-
ers in the networked publics can also play a crucial role in influencing one’s advocacy
effort, be it positive or negative. Given that climate discourses are embedded in the
webs of relationships and climate issues, it is crucial to investigate who leads a topic
and who echoes and the relative online influence of opposing climate nonprofits.
A Network Perspective to Map Discursive Influence Through Message
Network perspectives are often applied to understand the complexity and intercon-
nectedness of organization-public relationships (Yang & Saffer, 2019). However, in
studies of networked publics, the same network perspective can be used to map out
6 Nonprofit and Voluntary Sector Quarterly 00(0)
power dynamics among stakeholders in the public sphere based on flows of informa-
tion, influence, attention, and actions (Ausserhofer & Maireder, 2013). This study,
hence, follows the network perspective to reveal the complexities of interactions, col-
laborations, and mobilization that exist in nonprofit advocacy.
A network consists of actors who form connections through interaction, shared
activities, and resource mobilization. Actors and their web of connections form a net-
work, which is a social system where ideas emerge, influence is built, and resource is
exchanged. Particularly relevant are online social networks based on online interac-
tions. The most well-studied online networks in the nonprofit and climate advocacy
literature are the ones based on Twitter retweeting, replies, mentions (Pearce et al.,
2014; Saffer et al., 2019; Vu et al., 2019), or website hyperlinking (Fu & Shumate,
2016; Yang & Saffer, 2018). In such networks, actors (referred to as nodes in network
studies) are social media users, and connections among the users are based on direct
information exchanges in the form of retweeting, sharing, and replies or are based on
association (e.g., hyperlinked on official websites). Applied to such networks, the net-
work approach can encapsulate the broader social structural arrangements in which
nonprofit advocates shape public discourse of other stakeholders (i.e., who said what
and to whom) (Kirilenko & Stepchenkova, 2014). It can also reveal who dominates the
discourse by ranking users by centrality measures, with users located at the center of
the network deemed as more influential than others.
Although nonprofits can build a large online community and garner attention
through viral content, oftentimes, content diffusion is not strictly limited to the diffu-
sion of the same message. What also matters is that advocacy organizations serve as
“cultural bridges.” Bail (2016) used the notion of “cultural bridges” or “cultural
betweenness” to introduce a new type of online social connections: Organizations and
stakeholders all belong to a cultural network in which the interorganizational and
interstakeholder connections are based on shared themes and similar topics in their
online messages. Essentially, nonprofits that introduce entirely new discursive themes
are likely to be ignored because they are perceived as irrelevant to the central concerns
of those within an advocacy field. The ideal message, in other words, is both new and
familiar (Bail, 2016). In the current work, we adopt cultural network as the blueprint
and introduce a different type of diffusion network, not based on the flow of the same
information but on the flow of homogeneous ideas and agendas. In other words, two
actors are connected when they discuss similar ideas. Organizations producing mes-
sages that are most echoed by other actors are centrally located in the network. They
create “cultural bridges” that lead and inspire more online engagement (Bail, 2016).
For instance, a climate advocacy nonprofit may begin by discussing a newly published
study on green economy, and a stakeholder may also discuss the same study subse-
quently, but in a different message. Even though the two users produced two different
messages and, in a conventional social network, will not be connected (because there
is no retweeting or replying relationship), the common theme addressed still connects
the two users and may imply that one user may exert discursive influence in the sense
that their messages are most resonated by other actors in the subsequent discourse. In
short, to understand message diffusion involving climate nonprofits, we must also
Wu and Xu 7
look at a different type of diffusion based on content similarity in understanding the
roles of opposing climate nonprofits in shaping the climate conversations on Twitter.
Data and Empirical Strategy
Twitter Data Collection
In this study, we examine the Twitter discourse on climate change. Big social media
data constitute a second-by-second trace that provides massive observational data on
public discourse and debates on climate change (Manovich, 2012; Schmidt, 2014). In
particular, Twitter provides valuable public data for studying climate advocacy net-
works when the interactions are highly related to interest (Kirilenko & Stepchenkova,
2014; Marlow et al., 2021; Vu et al., 2019), as opposed to Facebook, another highly
popular social network, where data are restricted and not reliable at the time of study
(Freelon, 2018).1 Although different platform cultures exist within social media (Gibbs
et al., 2015), given the discursive nature of climate advocacy on Twitter, a textual
analysis to detect similarities of the message content is an ideal way to investigate the
issue (Farrell, 2015). However, traditional methods of textual analysis are labor-inten-
sive and inadequate in terms of the volume, scale, and the complexity of data they can
accommodate (Yang & Saffer, 2019). To scale up our analysis and address previous
data limitations, we pursue a novel empirical strategy by combining computational
text analysis and network approach to examine both the network structure and discur-
sive similarities of the climate change discourses by opposing advocates.
We drew on a massive public Twitter data set collected by Littman and Wrubel
(2019) to temporarily analyze the message similarity network of Twitter accounts. The
raw data set contains 39,622,026 English language tweets related to climate change,
collected between September 21, 2017, and May 17, 2019, from the Twitter API using
Social Feed Manager. Tweets were collected using the POST statuses/filter method of
the Twitter Stream API, with the following keywords: #climatechange, #climat-
echangeisreal, #actonclimate, #globalwarming, #climatechangehoax, #climatedeniers,
#climatechangeisfalse, #globalwarminghoax, #climatechangenotreal, climate change,
global warming, climate hoax.
Given the massive amount of data that render longitudinal network analysis com-
putationally intensive and impossible, we chose to narrow down to the 2-week period
spanning December 15, 2017, and December 31, 2017 (see Figure 1). The 2 weeks
saw the biggest burst in original tweet volume on the topic (N = 298,073), which
coincided with salient climate-related events, including a major shift in climate policy
when Trump removed climate change as a national security threat, the occurrence of
extreme weather, and the release of climate report on global warming that embarked
heated debates in Twitterverse. We picked the spike period because the heat of the
topic might draw a more diverse and dynamic set of stakeholders to the conversation.
Arguably, this timeframe selection only captured nonprofits’ potential discursive
influence on climate discourse when the issue was hot on public agenda. We do not
8 Nonprofit and Voluntary Sector Quarterly 00(0)
presume that our study findings can directly apply to the periods when the Twitter
public interest in the climate issue is muted.
Purposive Sampling of Climate Action and Climate Obstruction
To zoom in the climate change discourse produced and propagated by nonprofit actors,
we first identified a list of pro-climate change nonprofits and anti-climate change non-
profits and their respective Twitter accounts. For the purpose of our study, climate
action organizations accept man-made climate change as a genuine threat and advo-
cate strong governmental, economic, and social action to address it. Climate obstruc-
tion or countermovement organizations in our study may or may not accept climate
change as real but generally oppose or delay sweeping governmental programs or
economic disruption to address it, preferring instead to find market solutions. The
curation of the list is based on two related literatures (Brulle et al., 2021; Trzyna &
Didion, 2001) and public websites that document the climate change movement and
counter-movement (see Supplemental Appendix 1). The raw list includes 1,000 orga-
nizations, of which 815 organizations are present on Twitter. Such organizations are
climate action and obstruction nonprofits, including think tanks, foundations, trade
associations, and advocacy organizations that are known to be involved in climate
change movement and counter-movement.
Multistage, Multilevel Computational Analyses
To analyze and track the tweet messages at a large scale, we drew on a combination of
computational text analysis and network science to detect content similarities and
Figure 1. Tweet Count Between September 21, 2017, and May 17, 2019.
Note. The straight upward lines correspond to the data collection gaps in the Littman and Wrubel (2019)
data set.
Wu and Xu 9
trace the discourse network in which messages might be propagated in the climate
discourse over time. The analytical process consisted of three stages: creating a mes-
sage similarity network, ranking nonprofit users by centrality, and building an egocen-
tric stakeholder network around nonprofit actors.
Mapping Diffusion Network Based on Content Similarity Across Time. We operationalized
content homogeneity and traced message diffusion by the topical similarities of tweet
messages across Twitter accounts within a time distance. Part-of-speech tagging (a
common procedure in natural language processing) was used to filter nonessential
words and retain only nouns, proper names, and hashtags in the tweets. These linguis-
tic markers were arguably the most indicative of topics and agenda reflected in the
text. We then gauged longitudinal diffusion of climate discourse by using an R pack-
age called RNewsflow to analyze content homogeneity of tweets across accounts
(Welbers et al., 2018). By using a sliding window approach, RNewsflow compared
messages within a given time distance to calculate their topical similarities. Top orga-
nizations by this metric were deemed as the early accounts that introduced a conversa-
tion topic while later accounts that posted similar content were considered followers
of an organization’s discourse. Upon initial inspection, we followed the documenta-
tion to use the default similarity threshold of 0.4 with an hour window spanning 24
hours. In other words, the algorithm compared each text to all text within the next 24
hours; two messages were deemed similar only when at least 40% of nouns, proper
names, and hashtags used in their tweets are the same (see examples in Supplemental
Appendix 2). These message pairs were considered to have a path of “potential influ-
ence.” The RNewsflow algorithm was applied to 298,073 unique tweets (not including
retweets) sent by 207,268 unique accounts between December 15, 2017, and Decem-
ber 31, 2017.
Identifying “Who Leads” Through Betweenness Centrality Analysis. The first stage of anal-
ysis employed the RNewsflow algorithm to create a diffusion network including
19,518 unique Twitter accounts based on content similarity. We recall that Twitter fol-
lows a power-law distribution whereby a large majority of tweets come from a small
minority of tweeters (Wojcik & Hughes, 2019). The next step involved identifying
leading nonprofit accounts that we defined as central accounts in the message similar-
ity network, whose earlier posted messages shared high topical similarities with the
content from other accounts.
To borrow the terminology from network analysis, we referred to the Twitter users
as network nodes and their connections as ties/edges. In the network, two users are
connected if their tweets have similar content. The strength of their ties is determined
by the level of content similarity. In other words, the tie between two users is stronger
if there are more nouns, proper names, and hashtags commonly used in their tweets.
We then applied the predefined list of 815 climate nonprofit accounts to the network,
that is, we created a subset of the network (a subnetwork) by including only ties that
involved nonprofit accounts on the list. The subnetwork has 26,735 ties and 14,279
unique Twitter accounts. Based on this nonprofit-oriented subnetwork, we calculated
10 Nonprofit and Voluntary Sector Quarterly 00(0)
the betweenness centrality of these organizational accounts. Betweenness centrality
accesses how central the organization (i.e., a node) is in a network, and consequently,
it can be used as a proxy measure of its importance in the message similarity network
(Borgatti et al., 2009). Following Yang and Saffer (2018) and Bail (2016), we asserted
that nonprofits’ network positions on Twitter signaled their advocacy efficacy and
their abilities to have their messages heard in the networked public sphere—the digital
space where politicians, journalists, organizations, groups, and citizens publicly nego-
tiate and frame climate issues.
Tracing “Who Echoes” Through Stakeholder Analysis. The key to our interest is how
specific climate advocacy nonprofits contribute to the online climate discourse.
Thus, our final step selected four leading organizations and examined their egocen-
tric networks. Here, an egocentric network was deemed as a type of network
anchored around specific actors and ties that span from the actors. Twitter accounts
that shared topical similarity with the advocacy organizations are considered stake-
holders in this analysis. Our analysis involved analyzing the type of stakeholders in
the egocentric networks and messages that connected the stakeholders to the key
organizations. To differentiate the stakeholder types, we conducted content analysis
on the Twitter profiles that shared similar content with the nonprofits. We considered
using automatic classification or machine learning techniques to scale up the stake-
holder analysis. However, a preliminary analysis suggested that the Twitter account
profiles were nuanced, and manual content analysis would produce the most reliable
results. To better focus the analysis, we zoomed into the egocentric network of four
nonprofits and analyzed a total of 2,479 user profiles whose 3,150 tweet messages
shared similarity with the nonprofit accounts. We adapted the stakeholder frame-
work proposed by Halpin et al. (2021) to classify Twitter users into three major
groups—elite audiences (policymakers and journalists), peer audiences (nonprofits),
and mass audiences (the public). Research suggests that the CCCM is closely tied
with elite corporate benefactors, in addition to a complex network of think tanks,
foundations, public relations firms, trade associations, and ad hoc groups (Almiron
et al., 2020; Brulle et al., 2021). Following the prior studies and upon initial inspec-
tion of 100 Twitter accounts, we identified two more stakeholder types, namely cor-
porations and others, to reflect an inclusive spectrum of stakeholders who might be
climate sponsors and partake in climate conversations (Brulle, 2014; Elgin & Weible,
2013; Farrell, 2015). Although a stakeholder identity can be multifaceted and might
vary, for this empirical exercise, we manually coded the Twitter profiles based on
one primary stakeholder identity. For instance, we coded official accounts of non-
profit organizations as “nonprofits” (N); registered, mainstream media agencies and
journalists as “media” (M); government officials, policymakers, or politicians as
“government” (G); businesses as “corporations” (C); any individual accounts were
classified as public members (P); and accounts that did not fall into the categories,
including parody and bot-like accounts, were identified as “others” (O). Table 1 lists
the stakeholder codes we used for classify nonprofit stakeholders’ Twitter accounts.
Wu and Xu 11
Who Leads?
After classifying stakeholder types, we analyzed the egocentric networks of four lead-
ing climate nonprofits to provide an in-depth analysis of their message diffusion at the
stakeholder level. We recall that the four nonprofits were selected because of their
central positions in the network (see Supplemental Appendix 3 for the ranked list). We
purposively shortlisted two climate advocates and two climate skeptics based in the
United States to analyze their discourse networks, including Natural Resources
Defense Council (NRDC) (ranks second) and Citizens Climate Lobby (CCL) (ranks
28th), which are established environmental groups that advocate on fighting human-
caused climate change. In terms of the opposing side, the Daily Caller News Foundation
(DCNF) (seventh) and Committee for a Constructive Tomorrow (CFACT) (51st) are
nonprofit groups that obstruct climate action and promote market solutions to environ-
mental problems (Brulle et al., 2021). As shown in Table 2, these organizations vary in
size, nonprofit status, climate position, and size of Twitter followers, providing hetero-
geneity to our network analysis.2 Each organization is discussed in more detail below.
Natural Resources Defense Council (NRDC). NRDC is a 501(c)(3) international environ-
mental group formed in 1970 to “ensure the rights of all people to clean air, clean
water, and healthy communities” (NRDC, 2021). NRDC is the oldest and largest
established nonprofit compared with the other three. In terms of Twitter use, NRDC
joined Twitter in January 2009 and had about 347,900 followers as of October 11,
2022, which is the largest among all. Within the 2-week timeframe, the RNewsflow
algorithm picked up eight climate advocacy messages sent by the NRDC, which were
found to be similar to the subsequent discourse (see Table 3). Two of them (Tweets 4
and 5) related to President Trump’s removal of climate change as a national security
threat and were found to be identical or similar to 143 and 24 tweets, respectively, later
published by stakeholder accounts. A thematic content analysis of these messages
revealed that they tended to reflect and address the same or related events. These
tweets about Trump’s action shared coverage by media such as the Associated Press,
Table 1. Distribution of Stakeholder Types in Climate Discourse.
Proposed stakeholder types Codes N
1. Public members P 1,893
2. Nonprofit and advocacy groups N 60
3. Media and journalists M 139
4. Government, policymakers, and politicians G 17
5. Corporations and companies C 46
6. Others (including parody, bot-like accounts) O 324
Total unique Twitter accounts 2,479
Note. Fifteen accounts became nonexistent or were suspended at the time of profile coding.
12 Nonprofit and Voluntary Sector Quarterly 00(0)
ClimateWire, The Guardian, The New York Times, and other mainstream news outlets.
Other tweets engaged in sharing information, links, and comments on the potential
threats posed by climate change. Another thread to NRDC’s advocacy was to invoke
the scientific evidence on the catastrophic events and extreme weather to justify cli-
mate action, as reflected in tweets 3 and 6 (see Table 3). They were found to be identi-
cal or similar to 44 and 373 later tweets, respectively, from stakeholder accounts.
Citizens Climate Lobby (CCL). CCL is a 501(c)(4) grassroots advocacy organization
formed in 2007 by Marshall Saunders. CCL “is a climate change organization that
exists to create the political will for climate change solutions by enabling individual
breakthroughs in the exercise of personal and political power” (CCL, 2021). CCL
joined Twitter in March 2010 and had 46,200 followers. We found that CCL sent out
10 climate advocacy messages that were similar to 483 tweets produced within 24
hours (Table 4). Among them, one dominant tweet (see Tweet 4 in Table 4) was found
to be similar to 446 tweets in the data set. CCL’s tweets were about a San Francisco
Chronicle editorial that denounced Trump for ignoring climate change (Tweet 3),
visual messages advocating that climate change is real (Tweet 2) and mocking Trump
on the distinction between climate and weather (Tweet 8).
The Daily Caller News Foundation (DCNF). DCNF, a 501(c)(3) nonprofit news organiza-
tion formed in 2009, was founded by Fox News host Tucker Carlson and political
commentator Neil Patel. DCNF’s mission is “to train up-and-coming reporters and
editors, to carry out investigative reporting, and to perform deep policy reporting with
a purpose of consumer awareness and education” (DCNF, 2022). Based on its website
Table 2. Summary Statistics on the Four Climate Nonprofits.
Twitter handle @NRDC @citizensclimate @DailyCaller @CFACT
Climate position Climate action Climate action Climate delay
and denial
Climate delay
and denial
Nonprofit status 501(c)(3) 501(c)(4) 501(c)(3) 501(c)(3)
Total assets (in million) 442.8 1.3 1.1 1.0
Total revenue (in million) 181.8 1.2 2.5 1.6
Employees 763 43 45 10
Volunteers 0 112,059 5 150
Formation 1970 2007 2011 1985
Joined Twitter January 2009 March 2010 May 2009 May 2009
Followers (in 1,000) 347.9 46.2 901.4 14
Following (in 1,000) 3.8 15.2 7.6 4.5
Note. Organizational data retrieved from the IRS form 990 for the fiscal year ending December
2019. Twitter followers and following number as of October 11, 2022. NRDC = Natural Resources
Defense Council; CCL = Citizens Climate Lobby; DCNF = Daily Caller News Foundation; CFACT =
Committee for a Constructive Tomorrow.
Wu and Xu 13
Table 3. Messages sent by NRDC and Similar Tweets Identified by RNewsflow.
No. Originating tweets from NRDC Examples of similar tweets Count
1 “the food recovery and recycling
act would help feed hungry new
yorkers and fight climate change
at the same time. . .”
@pattiemcmahon: how to
feed hungry New Yorkers
and fight climate change via
2 “the gop tax bill would expose
irreplaceable public wilderness in
alaska to industrial ruin for the
sake of oil an. . .”
@nrdc_wild: the gop tax bill
would expose irreplaceable
public wilderness in alaska to
industrial ruin for the sake
of oil an. . .
3 “with climate change causing more
frequent and intense extreme
weather it s becoming more and
more important to fix. . .”
@_tcglobal: extreme
weather and rising seas will
create millions of refugees
4 Under the Trump admin., the
United States’ National Security
Strategy is getting a frightening
edit: it will no longer include
#climatechange as a national
@eljmkt_daily: climate
change no longer a security
threat or so says trump ..
5 “the trump admin. has officially
removed climate change from the
list of national security threats.
its yet another. . .”
@einwarming: trump may
doubt global warming but
most Americans don’t. . .
6 “scientists from around the world
analyzed 27 extreme weather
events from the last year and
found that global warmin. . .”
@friendsoscience: @userid
@ userid you are confusing
weather with climate.
weather is temporal.
#climate change is measured
in. . .
7 “even as president trump and his
toxic minions put the interests
of oil gas and coal ahead of our
children s futu. . .”
@ladybosephus: @
realdonaldtrump look at all
his minions loving this. yeah
we dun told them. it s cold
this week. climate change. . .
Grand total 589
Note. NRDC = Natural Resources Defense Council.
reporting, DCNF owns and creates all news content.3 However, its right-wing corpo-
rate news outlet has a licensing agreement with DCNF; reporting created by the foun-
dation is automatically sent to the for-profit Daily Caller to publish for free. Essentially,
the Daily Caller brand is both the foundation and the corporation. DCNF is classified
as a climate obstruction organization that upholds climate denial and delay. It does not
affirm mainstream climate science as defined by the Intergovernmental Panel on
14 Nonprofit and Voluntary Sector Quarterly 00(0)
Table 4. Messages sent by CCL and Similar Tweets Identified by RNewsflow.
No. Originating tweets from CCL Examples of similar tweets Count
1 #climate consequences: this is
the kind of global chaos that our
world simply cannot adjust to. @
fionaharvey. . .
@slaveryexperts: current
#climatechange trends to drive
1 million of migrant arrivals to
Europe per year @fionaharvey
2 a picture – or graphic – is worth a
thousand words when it comes to
communicating #climatechange.
great to see th. . .
@wwfeu: an image is worth
thousand words; a year of
#climatechange in photos
3 from the @sfchronicles editorial
board: though the trump white
house ignores #climatechange the
rest of the world. . .
@eciu_uk: Trump is wrong to
ignore the security threat of
climate change it’s sinking a key us
naval base—the independent
4 from the editorial board of
Charlestons @postandcourier:
Mr. Trump s decision to downplay
the importance of. . .
@ccl_wyoming: from the editorial
board of Charleston’s post and
courier: Mr. trump s decision to
downplay the importance of. . .
5 millennials will feel the impact of
#climatechange in their lifetimes.
hear how one is dealing with it in
the lates. . .
@cclpalmbeaches: Millennials will
feel the impact of #climatechange
in their lifetimes. Hear how one is
dealing with it in the lates. . .
6 nice op-ed from CCLers about
Oregons attempts to price
carbon and reduce pollution.
#climatechange #putapriceonit. . .
@ccl_wyoming: nice op-ed from
cclers about Oregons attempts to
price carbon and reduce pollution.
#climatechange #putapriceonit. . .
7 Our friend @bobinglis gave a
stellar TEDx talk about being a
conservative and stepping up on
#climatechange. If yo. . .
@path2positive: Conservative
climate courage—a TEDx talk by
8 Thank you @weatherchannel for
setting @realdonaldtrumpstraight
about the difference between
weather and #climate
@steveblankdds: @
realdonaldtrump What an a$$.
You really don’t understand the
difference in climate change vs.
Seasonal weather. . .
9 The transition to #cleanenergy is
already happening and we can
speed it up by enacting carbon fee
& dividend.
@cclpalmbeaches: The transition to
#cleanenergy is already happening
and we can speed it up by
enacting carbon fee & dividend
10 Were beyond how #climatechange
will impact us in the future. It’s
happening now. @bradplumer @
@benitezja: Is global warming
responsible for the extreme
weather events? Check it out @
popovichn @bradplumer @noaa
Grand total 483
Note. CCL = Citizens Climate Lobby.
Wu and Xu 15
Climate Change (IPCC) and is skeptical toward climate change mitigation policies or
regulations (Brulle et al., 2021). It joined Twitter in May 2009 and had 901.4 follow-
ers. Within the 2-week timeframe, the RNewsflow algorithm detected three tweets
being similar to subsequent messages by stakeholder accounts. These tweets are
ascribed to the climate denial narrative that casts doubt or refutes the science of cli-
mate change. For instance, two messages spread skepticism toward the effect of global
warming in causing record snowfall and more frequent volcanic eruptions. It ampli-
fied the denialist message from Trump to insinuate that global warming was not hap-
pening given the coldest New Year’s Eve on record (Table 5).
Committee for a Constructive Tomorrow (CFACT). CFACT was founded in 1985 to pro-
vide market and technological perspectives on environmental and development issues
(CFACT, 2021). CFACT is a climate obstruction organization that upholds climate
denial and delay. It does not affirm mainstream climate science as defined by the IPCC
and is skeptical toward climate change mitigation policies or regulations (Brulle et al.,
2021). CFACT joined Twitter in May 2009 and had 14,000 followers. We found two
messages from CFACT to be similar to subsequent messages by other accounts (Table
6). The first message defended Trump’s policy to remove climate change as a national
security threat, bearing resemblance to 820 tweets sent within 24 hours. The second
message averred that climate change was not real given the record snow on Alaska
mountains and blamed the left for making it up. These narratives stem from climate
denialism, which remains an active part of the framing strategy of leading climate
opponents (Cann & Raymond, 2018).
Who Echoes?
We built an egocentric network based on the four nonprofit organizations which had
4,572 similar message pairs (ties) and 3,478 unique accounts (including the four
selected organizations and stakeholders who sent similar tweets). As explained in the
Data and Empirical Strategy section, we used the default similarity threshold of 0.4
Table 5. Messages sent by Daily Caller and Similar Tweets Identified by RNewsflow.
No. Originating tweet from the Daily Caller Example of similar tweets in a day Count
1 Record lake effect snowfall blamed on
global warming but it’s not that simple
Does record snowfall disprove
global warming? exactly the
opposite scientist says.
2 Troll in chief: Trump pokes fun at global
warming critics tells people to bundle
Trump started today by trolling
how climate change is fake. . ..
3 Study claims global warming will cause
more frequent volcanic eruptions
Climate change likely to increase
volcanic eruptions scientists say
Grand total 1,384
16 Nonprofit and Voluntary Sector Quarterly 00(0)
with an hour window spanning 24 hours to trace topical similarity following the docu-
mentation. It follows that the 4,572 message pairs were algorithmically tagged as simi-
lar because at least 40% of nouns, proper names, and hashtags used in their tweets are
the same. We considered that these message pairs exhibit a path of “possible influ-
ence.” We then manually checked the message pairs to confirm the machine-generated
results, removing those that did not share topical similarity (i.e., likely no influence).
Additional message pairs were deleted after the manual screening, resulting in a total
of 3,166 message pairs and 2,499 unique accounts for final analysis. We manually
coded stakeholders by their profiles and categorized them into seven groups (see the
groupings in Table 1). Table 7 shows the number of ties (i.e., message pairs) and the
similarity weight across the four organizations and their stakeholders. Another intui-
tive way to present the aggregated discourse network is by visualizing it at the stake-
holder level (Figure 2). Each node represents a nonprofit actor, or a stakeholder group
engaged in public conversation on climate issues, and the ties (edges) between the
nodes represent those who discussed similar themes within the Twitter advocacy field
within the 2-week period.
In the following, we discuss and interpret the results in terms of network size, tie
types, and tie strength in general and identify differing patterns across opposing camps.
The weight of the ties represents the percentage of messages from the selected four
organizations that later appeared in tweets sent by a specific type of stakeholders. In
other words, it measures an aggregated message-level similarity between the four non-
profits and the stakeholder accounts. Looking at the overall size and reach of the ego-
centric networks, Daily Caller had the most extensive message similarity network with
Twitter stakeholders (1,371 ties), followed by CFACT (710 pairs) and NRDC (588
pairs), while CCL had the smallest network (481 pairs). Similarly, Daily Caller yielded
the largest average similarity score (0.57), followed by CFACT (0.56), CCL (0.19),
and NRDC (0.13). This implies that on average, 57% of Daily Caller messages were
similar to or identical to later published messages in any stakeholder accounts within
Table 6. Messages sent by CFACT and Similar Tweets Identified by RNewsflow.
No. Originating tweets from CFACT Examples of similar tweets Count
1 Trump removes climate change as
a national security threat. Our
military should do what they do
best: protecting
@insideclimate: Despite concerns
of military leaders the Trump
administration is expected to drop
climate change from a list of gl. . .
2 Wed say you can’t make this stuff
up but making this stuff up is all
the left does record snow on
Alaska mount
@sottnet: Earth changes: #iceage
cometh: Researchers find
#climatechange is triggering
record #snow in #alaska
Grand total 710
Note. CFACT = Committee for a Constructive Tomorrow.
Table 7. The Adjacency Matrix Table on the Number of Ties to Stakeholder Types and Tie Strength.
From tweets
NRDC (@NRDC) CCL (@citizensclimate) DCNF (@DailyCaller) CFACT (@CFACT) Grand total
7 10 3 2 22
Stakeholders Pairs Weight Pairs Weight Pairs Weight Pairs Weight
Public (P) 456 0.29 289 0.08 1,047 0.83 458 0.67 2,250
Nonprofits (N) 16 0.13 35 0.42 9 0.33 36 0.67 96
Media (M) 19 0.04 51 0 74 0.50 61 0.67 205
Government (G) 4 0.04 6 0 4 0 8 0.33 22
Corporation (C) 15 0.13 14 0 24 0.33 19 0.33 72
Others (O) 78 0.13 86 0.08 213 0.83 128 0.67 505
Total ties and
average weight
588 0.13 481 0.19 1,371 0.57 710 0.56 3,150
Note. Message pairs refer to the individual messages that were similar. Figure 2 and Table 7 represent the aggregated network results. NRDC = Natural
Resources Defense Council; CCL = Citizens Climate Lobby; DCNF = Daily Caller News Foundation; CFACT = Committee for a Constructive Tomorrow.
18 Nonprofit and Voluntary Sector Quarterly 00(0)
24 hours, compared with 56% of CFACT messages, 19% of CCL messages, and 13%
of NRDC messages that were similar to or identical to later published messages in any
stakeholder accounts within 24 hours. On balance, the analysis suggests that Daily
Caller and CFACT resonated with more stakeholder accounts not only in terms of the
number of message pairs but also in terms of message similarity across a vast diversity
of stakeholders, including the mass audience (public), peer audience (nonprofits), elite
audience (media, public officials/politicians), potential sponsors (corporations), and
other stakeholders.
Stakeholder ties of message similarity were not universally found nor equally
strong. The strongest ties appeared among Daily Caller-public (0.83), Daily Caller-
others (0.83), CFACT-public (0.67), CFACT-nonprofit (0.67), CFACT-media (0.67),
and CFACT-others (0.67). The weakest ties were NRDC-government (0.04), NRDC-
media (0.04), and Daily Caller-government (0). Although we identified a dozen of
similar message pairs between CCL and media, government, and corporation, the
weight became negligible in the aggregated egocentric network. The similarity score
of 0 indicates that the algorithm did not detect meaningful diffusion between a stake-
holder type and a nonprofit in the given time. We found that all four nonprofits had
diffusion ties with public members, nonprofits, and others, but not with government
Figure 2. The Discourse Network Centered Around the Four Selected Nonprofit Organizations.
Wu and Xu 19
officials/politicians, media, and corporations. Notably, we coded a total of 505
accounts as others, many of which were bot-like or parody accounts that repost and
curate climate news and reports, and they were neither registered commercial media
nor registered nonprofits. These accounts usually featured analogous postings, small
follower counts, and low audience reactions, such as favorites and retweets. As the
results show, a larger proportion of Daily Caller and CFACT messages were similar to
contents produced by these accounts. CCL, on the contrary, featured a stronger diffu-
sion to nonprofits (0.42) than other stakeholders. Being a federated nonprofit, the
strong diffusion tie to nonprofits might in part be attributed to its work in grassroots
advocacy and that its local chapters and peers readily reciprocate their messages as
shown in Table 4.
Comparing the opposing groups, we found a consistent pattern that both groups
featured the greatest number of message pairs with public accounts, followed by oth-
ers, media, nonprofits, or corporations in a descending order. The differing patterns
and message similarity across opposing camps warrant further investigation. We
noticed greater message similarity from the two obstructionists to all the stakeholders
than CCL and NRDC in general. A strong similarity score suggests that the content
originated by Daily Caller and CFACT highly resonated with these stakeholder
accounts. Substantially, the similarity weight indicated that 83% of Daily Caller mes-
sages and 67% of CFACT messages were similar to or identical to later published
messages in public accounts, compared with only 29% of NRDC messages and 8% of
CCL messages.
A close examination of the discursive data suggests that nonprofits may serve as
initiators, enablers, and brokers of Twitter climate discourse (Bail, 2016; Bennett &
Segerberg, 2012; Jackson & Welles, 2015). We discovered that the sampled nonprofits
frequently initiated the conversations by bringing up news topics and opinions that
likely prompt others to follow up. For instance, when the Trump administration offi-
cially removed climate change from the list of national security threats, we observed
distinct framings across the two camps. NRDC shared an expert commentary to con-
demn the policy change—“. . .It’s yet another step in [sic] towards abandoning
America’s role as a global leader in tackling climate change”—whereas CFACT
expressed its endorsement—“. . .Our military should do what they do best: Protecting
our freedoms and killing terrorists.”
Furthermore, both sides of climate nonprofits shared and republished high volumes
of news, opinions, and information on climate issues to amplify their narratives. News-
sharing might prove to be an easy way to stimulate common interest or provoke com-
mon outrages from likeminded stakeholders (Chinn et al., 2020; Kirilenko &
Stepchenkova, 2014). By being the earliest to talk about a piece of news, the organiza-
tion has the capacity to introduce the wider Twitter community to events that matter to
climate change. This indicates that leading nonprofits stayed relevant to the central
concerns of the advocacy field, engaging stakeholders in shared narratives of breaking
news and current events. For instance, Daily Caller recreated and relayed Trump’s
tweet message: “TROLL IN CHIEF: Trump Pokes Fun At Global Warming Critics,
Tells People To Bundle Up. . .” We found 971 Twitter accounts tweeted the same or
20 Nonprofit and Voluntary Sector Quarterly 00(0)
similar content within the next 24 hours. Similarly, CCL shared the San Francisco
Chronicles editorial to criticize the policy action of the Trump government: “. .
.Though the Trump White House ignores #climatechange, the rest of the world isn’t
neglecting the doomsday issue. #OnePlanetSummit.” In total, 540 accounts tweeted
the same or similar content within the next 24 hours.
Our analyses reveal how similar topics are possibly diffused and echoed in continuous
streams of social media discourse across various stakeholders. Several patterns arising
from the data warrant further discussions. First, the top organizations in the diffusion
network based on content similarity are established and well-resourced 501(c) non-
profits active in climate-related affairs. This means that organizations’ offline status
may be attributed to their online influence. Larger organizations tend to have resources
and comparative advantage in traversing the online social media capital (Saxton &
Guo, 2020; Xu & Saxton, 2019). Furthermore, these organizations are not standalone
nonprofits; they have affiliated ties with other nonprofit groups or corporations. For
instance, CCL is a federated nonprofit with 537 local chapters nationwide. NRDC
partners with E2, a national, nonpartisan group that advocates for policies and operates
a 501(c)(4) subsidiary, the NRDC Action Fund. DCNF, registered as a 501(c)(3) non-
profit news organization, partners with its corporate arm to distribute and republish its
news content.
Second, our results echo the findings from the study by Halpin et al. (2021) that
climate advocacy groups target a variety of audiences online because of their strate-
gic objectives and the extent to which particular audiences engage with them.
Different from the early studies that looked at how nonprofits initiated the connec-
tion with stakeholders (through strategic and preemptive targeted messaging), the
current study looks at the potential outcome of discursive influence through non-
profits’ organic participation in online climate conversations. That is, our data points
show what kind of stakeholders most resonate with the messages from nonprofit
advocacy groups. Our findings show prevalent diffusion ties to public stakeholders,
media, and nonprofits, parallel to previous findings (Halpin et al., 2021). CCL dis-
played a higher degree of message similarity with nonprofit stakeholders, particu-
larly through their local chapters that echoed their messages closely. Such network
embeddedness helps establish an advocacy alliance and amplify their messages
within the Twitterverse. Taken together, Halpin et al. (2021) revealed the kind of
stakeholders nonprofits aim to reach out to in social media engagement. Our current
work shows that nonprofits’ messages are indeed most echoed by these commonly
targeted stakeholders.
More importantly, while the sample of the four nonprofits is rather small to general-
ize the phenomenon, our findings show that potential diffusion from nonprofits to
policymakers are almost absent in the current climate discourse. This is in part due to
the short period of observation and data constraints as discussed in the limitations. Yet,
this raises an important efficacy question for nonprofit advocates. It is possible that the
Wu and Xu 21
sampled nonprofits mainly use social media to engage with public constituents, and
they reach out to policymakers or lobby behind the scene. Hence, their potential online
influence on policymakers is not salient. Policymakers, due to the nature of their work,
may be restrained from producing activist-sounding narratives that may alienate their
constituencies. This might coincide with the finding that legislators are more likely to
follow, than to lead, discussion of public issue (Barberá et al., 2019). Policymakers
work on a variety of issues and might post limited tweets on climate change during the
time of observation. Overall, this finding casts doubt over the efficacy of climate advo-
cacy groups’ engagement with government officials and policymakers on Twitter, war-
ranting investigation in future research.
We observed that, during the 2-week study period, frequent posting did not neces-
sarily make one a central leader on the climate Twitterverse. Putting the number of
originating messages into perspective, the analysis constituted three Daily Caller mes-
sages, two CFACT messages, seven NRDC messages, and ten CCL messages.
Considering the small number of originating tweets from the climate obstruction non-
profits, the higher similarity weight might suggest a higher efficacy in propagating
their messages to stakeholders in the Twitterverse compared to the climate action non-
profits. In fact, the RNewsflow analysis suggested that CFACT might propagate cli-
mate obstructionism loudly with merely two tweets. It might be a sign that fewer
messages correspond to less heterogeneity in message diffusion, thus a high resem-
blance to the subsequent discourse.
The very content of the message could also matter for explaining rivalry and diffu-
sion. We observed more topical variety among climate action nonprofits than among
climate obstruction nonprofits. To attest to the urgency and gravity of climate issues,
climate action groups’ messaging encompasses a variety of topical themes that spoke
to the scientific evidence on climate impact and energy transition and advocated for
climate action. In contrast, the obstruction nonprofits’ messages fit homogenously into
the conservative, countermovement talking points that the priority of climate change
is misplaced and that the impact is inconsistent. They defended the policy change on
national security issues based on a critique of climate science and principled objec-
tions to state intervention and planning (Climate Social Science Network, 2021).
Furthermore, messages that contain inflammatory content and misinformation are
commonly found among climate obstruction groups. For instance, Daily Caller
reshared the trolling message from Trump and misinformation on global warming
effect. As scholars suggested, these messages not only sparked heated ideological
debates from both camps but also are more likely to spread and resurface on social
media (Benkler et al., 2018; Shin et al., 2018).
Although this could be a coincidence—and hence further studies should cross-
examine this pattern—messages from the opposing camps have different appeals.
According to a recent Pew Research Center study, a two-third majority of the American
public supports more actions on climate change (Tyson & Kennedy, 2020).
Conservative advocacy groups, due to their relatively unpopular opinions on climate
change, may be more motivated to share content that affirms their minority opinion,
hence more influence within the network; being the minority in the online public
22 Nonprofit and Voluntary Sector Quarterly 00(0)
sphere, climate obstruction communities might readily follow and echo their narra-
tives for similar reasons, creating echo chambers among obstructionists and contrari-
ans in the Twitterverse (Cann & Raymond, 2018; Farrell, 2015; Jasny et al., 2015).
Besides human agency of a mobilized minority, the “black box” algorithms of social
media are in part culpable for breeding echo chambers or filter bubbles (Benkler et al.,
Relatedly, our investigation found relatively high semantic similarities between
messages originating from climate obstruction nonprofits and suspicious bot-like
accounts. Social media are known to be rife with the use of fake user accounts to
amplify fringe messages, promote disinformation, and coordinate fabricated advocacy
campaigns (Tucker et al., 2018). Nonprofits often have established relationships and
trust with their target audiences, which could make them trusted intermediaries in dis-
seminating information and countering misinformation (Finn et al., 2006; Te’eni &
Young, 2003). Nevertheless, recent studies suggest that shadowy political organiza-
tions, notably 501(c)(3) and 501(c)(4) nonprofits, have enabled the spread of misinfor-
mation and divisive and disrupted rhetoric on social media (Kim et al., 2018). Although
nonprofits may serve as trusted intermediaries, some might have ties to bot-like or
fake user accounts to amplify fringe messages and promote disinformation and
In particular, bots are found to be effective at propagating a message—such as cli-
mate obstruction—much readily than human “trolls” (Marlow et al., 2021). The politi-
cal implications of Twitter bots’ ability to sway climate discourse have been well
established, but attributing their actions to any group with a distinct political intent is
difficult because the source of bots is almost untraceable (Marlow et al., 2021). The
presence of bots is noteworthy for two reasons: First, almost all existent studies on
bots in online activism define bots as something that coordinately shares the same
piece of information over a short span of time (Giglietto et al., 2020). It is also possible
that bots may produce different messages yet still echo the same talking points.
Overall, this observation corroborates the findings that bots are prevalent in online
climate discourse, particularly in climate obstructionism (Marlow et al., 2021) and
right-wing activism (Freelon et al., 2020).
Limitations and Future Directions
The findings must be interpreted with the following limitations in mind. First, the
study is based on 2-week observational data involving a few organizations. It remains
to be empirically tested to draw a generalized conclusion on why some groups might
have a stronger influence over others in the longitudinal discourse. Owing to the labo-
rious task of manually coding stakeholder profiles and the limitations of the data set,
we were able to focus on four key opposing organizations to make case comparisons,
overlooking a large number of other nonprofit actors who were also active and influ-
ential in shaping the climate discourse.4 Future computational endeavors can automate
some of the stakeholder coding so that it is possible to analyze more climate organiza-
tions and their interactions with stakeholders. In addition to account profiles, future
Wu and Xu 23
work that traces the diffusion networks of policy narratives disseminated by opposing
climate advocacy groups would also be valuable (Gupta et al., 2018). It is worth noting
that the data were 5 years old at the time of publication and did not necessarily reflect
recent climate conversations. Some organizations decided to leave the platform
because of the change of leadership in Twitter in 2022, concerns about the changing
API access, privacy policies, algorithms, and the rise of misinformation and disinfor-
mation. Although this study cannot account for these changes, future work may evalu-
ate their impact on climate advocacy on Twitter.
Online climate discourse occurs in the complex, multistakeholder networked pub-
lic sphere where tracking message diffusion from one source to another is challeng-
ing, if not impossible. The computational analysis of tracing message similarity thus
mapped out potential influence instead of actual influence. The analysis cannot
explain diffusion or rule out confounding and unobservable factors. To account for
the confounding factors, interviews with nonprofits’ social media officers might help
understand specific considerations behind messaging, which is beyond the scope of
the current investigation. Nevertheless, we underscore the complementary and neces-
sary combination of network data and discursive data in assessing the influence of
climate nonprofits. The patterns of the four organizations and their stakeholders
might be shaped by contextual factors (such as news cycle), but it is more likely their
patterns could be indicative of a broader trend of how messages diffuse from non-
profits to stakeholders. In addition to yielding exploratory evidence, the current anal-
ysis lays out the blueprint for future analyses involving richer and more social media
data points. The selection of the four organizations may also be biased toward those
organizations that happened to be influential or active during the study period.
Determining diffusion based on content similarity, albeit a novel idea, can be prob-
lematic when applied to short social media texts that are notoriously noisy: Their
brevity may prevent furthering more nuanced deliberations on the issue. Hence, we
incorporated manual verification to ensure the reliability of the machine coding
results (Grimmer & Stewart, 2013). Moreover, the tweets in the original data set
(Littman & Wrubel, 2019) were truncated, rendering it difficult to analyze a complete
tweet. Although the raw data set filtered out noises that were irrelevant to climate
discourse, we might have missed some nonprofits and stakeholders that used loaded
ideological terms but not our specific search terms.
It is well known that climate change critics and advocates strive to influence pub-
lic opinion and climate policy on social media, but the question of who leads and
who echoes remains inadequately answered. Previous studies have focused on the
most obvious content diffusion through social media sharing (e.g., retweeting)
(Agostino & Arnaboldi, 2016; Guo & Saxton, 2018; Halpin et al., 2021; Lam &
Nie, 2020). The less obvious but equally important aspect of diffusion, that is, dif-
fusing similar ideas and topics in different social media texts, has been understud-
ied. In this article, we have helped answer this question by empirically investigating
24 Nonprofit and Voluntary Sector Quarterly 00(0)
climate discourse on Twitter with a particular focus on opposing climate nonprof-
its, examining the extent to which their messages might be diffused and echoed by
a broad range of stakeholders in climate discourse. Although this study does not
provide causal evidence of message diffusion, it provides a complementary angle
of examining message diffusion different from the type of diffusion tracked by
message sharing.
In doing so, we offer new and important theoretical and empirical insights for future
work on nonprofit advocacy. First, it expands the nonprofit advocacy literature by bring-
ing the notion of networked publics into the focus of nonprofit advocacy. This theoretical
framework provides an entry point to theorize the potential influence of nonprofits to the
broader climate discourse on Twitter. Specifically, we conceptualize network influence
by capturing topical similarity in tweet messages across the climate discourse, beyond
repeating words verbatim. Our focus also complements the networked publics literature
that tends to focus on media, activists, or anonymous influencers (Jackson et al., 2020),
with a pivot to nonprofit actors. Second, our approach toward the measure of “who
leads, who echoes” is empirically innovative. It extends prior work by analyzing a dif-
ferent type of diffusion network based on the flow of similar (rather than same) mes-
sages. It allows us to ascertain in ways not previously possible through a temporal
network analysis of massive Twitter messages. This article thus expands the method-
ological possibility of social media research by combining computational text analysis,
network science, and content analysis techniques for analyzing online discourse. Third,
examining the various appeals of leading climate action and contrarian nonprofits and
the audiences who possibly echoed the nonprofits’ social media messages will enable the
organizations to identify likeminded stakeholders for a more strategic engagement.
Implications for Practice
Our findings provide important implications for practice. In particular, nonprofits
should strategically tailor different talking points when targeting stakeholders and
those from opposing camps. Identifying those who echo their social media messages
will enable the organizations to identify like-minded stakeholders for more in-depth
engagement. Advocacy organizations can intentionally reach out and target stakehold-
ers of opposing camps’ stakeholders to counteract echo chambers or filter bubbles
prevalent in an algorithmic public sphere. Furthermore, it is crucial for both scholars
and practitioners to probe bot-like activities and their connections to advocacy non-
profits in disseminating misinformation and disinformation. Because gaining public
attention is vital to social media-based advocacy, the methodological framework pre-
sented here can be adapted for practitioners to track their effectiveness in engaging
audiences and audit their possible discursive influence over time in a complex and
multistakeholder social media environment.
The authors would like to thank Shang Yat Lam, who provided excellent research assistance on
content analysis and verification of tweets. The authors thank the NVSQ Editors Jaclyn Piatak,
Wu and Xu 25
Joanne Carman, Susan Phillips, and three anonymous reviewers for their editorial support and
insightful feedback. Sincere appreciation also goes to Chao Guo, Beth Gazley, Aseem Prakash,
Juniper Katz, John McNutt, and participants of the C limate C hange and Voluntary Sector
Conference and Regional Nonprofit Scholar Meeting, who contributed and provided helpful
guidance to the initial draft.
Declaration of Conflicting Interests
The author(s) declared no potential conflicts of interest with respect to the research, authorship,
and/or publication of this article.
The author(s) received no financial support for the research, authorship, and/or publication of
this article.
Viviana Chiu Sik Wu
Supplemental Material
Supplemental material for this article is available online.
1. After the infamous Cambridge Analytica scandal, major platforms including Facebook
restricted and tightened up public access to platform data at the time of study (Freelon,
2. NRDC sent out 24 messages between December 15, 2017, and December 31, 2017; how-
ever, only seven of them passed the similarity threshold in RNewsflow. CCL sent out 24
messages between December 15, 2017, and December 31, 2017; however, only 10 passed
the similarity threshold in RNewsflow.Daily Caller sent out three messages between
December 15, 2017, and December 31, 2017, that passed the similarity threshold in
RNewsflow. CFACT sent out three messages between December 15, 2017, and December
31, 2017; however only two passed the similarity threshold in RNewsflow.
3. More information can be found on the websites and publications of the DCNF and the
Daily Caller, Inc., retrieved on September 28, 2022: https://dailycallernewsfoundation.
org/about-us/, and
uploads/2016%20Deck.pdf (p. 14).
4. We have considered analyzing all the groups on the list. However, the climate Twitter data
set by Littman and Wrubel (2019), likely due to the keyword terms used for curating the
data set, did not capture too many nonprofit groups (obstruction nonprofits are rare and
often in the shadow in reality).
Agostino, D., & Arnaboldi, M. (2016). A measurement framework for assessing the contribu-
tion of social media to public engagement: An empirical analysis on Facebook. Public
Management Review, 18(9), 1289–1307.
26 Nonprofit and Voluntary Sector Quarterly 00(0)
Almiron, N., Boykoff, M., Narberhaus, M., & Heras, F. (2020). Dominant counter-frames in
influential climate contrarian European think tanks. Climatic Change, 162(4), 2003–2020.
Anderson, A. A., & Huntington, H. E. (2017). How Twitter discussions of climate change
use sarcasm and incivility. Science Communication, 39(5), 598–620. https://doi.
Ausserhofer, J., & Maireder, A. (2013). National politics on Twitter: Structures and topics of a
networked public sphere. Information, Communication & Society, 16(3), 291–314. https://
Bruns, A., & Highfield, T. (2015). Is Habermas on Twitter?: Social media and the public sphere.
In A. Bruns, G. Enli, E. Skogerbø, A. O. Larsson, & C. Christensen (Eds.) The Routledge
companion to social media and politics (pp. 56–73). Routledge.
Bail, C. A. (2016). Combining natural language processing and network analysis to exam-
ine how advocacy organizations stimulate conversation on social media. Proceedings
of the National Academy of Sciences, 113(42), 11823–11828.
Barberá, P., Casas, A., Nagler, J., Egan, P. J., Bonneau, R., Jost, J. T., & Tucker, J. A. (2019).
Who leads? Who follows? Measuring issue attention and agenda setting by legislators and
the mass public using social media data. American Political Science Review, 113, 883–901.
Benkler, Y., Roberts, H., & Faris, R. (2018). Network propaganda: Manipulation, disinfor-
mation, and radicalization in American politics. Media and Journalism, 19(34), 329–331.
Bennett, W. L., & Segerberg, A. (2012). The logic of connective action: Digital media and
the personalization of contentious politics. Information, Communication & Society, 15(5),
Borgatti, S. P., Mehra, A., Brass, D. J., & Labianca, G. (2009). Network analysis in the social
sciences. Science, 323(5916), 892–895.
Bourdieu, P. (1989). Social space and symbolic power. Source: Sociological Theory, 7(1),
Brulle, R. J. (2014). Institutionalizing delay: Foundation funding and the creation of U.S. cli-
mate change counter-movement organizations. Climatic Change, 122(4), 681–694. https://
Brulle, R. J. (2018). The climate lobby: A sectoral analysis of lobbying spending on cli-
mate change in the USA, 2000 to 2016. Climatic Change, 149(3), 289–303. https://doi.
Brulle, R. J., Carmichael, J., Jenkins, J. C., Brulle, R. J., Carmichael, J., & Jenkins, J. C. (2012).
Shifting public opinion on climate change: An empirical assessment of factors influenc-
ing concern over climate change in the U.S., 2002–2010. Climatic Change, 114, 169–188.
Brulle, R. J., Hall, G., Loy, L., & Schell-Smith, K. (2021). Obstructing action: Foundation fund-
ing and U.S. climate change counter-movement organizations. Climatic Change, 166(1),
Cann, H. W., & Raymond, L. (2018). Does climate denialism still matter? The prevalence of
alternative frames in opposition to climate policy. Environmental Politics, 27(3), 433–454.
Chinn, S., Hart, P. S., & Soroka, S. (2020). Politicization and polarization in climate
change news content, 1985-2017. Science Communication, 42(1), 112–129. https://doi.
Wu and Xu 27
Citizens Climate Lobby. (2021).
Climate Social Science Network. (2021). CSSN research report 2021:2: The Mises Institute
network and climate policy. 9 Findings.
Committee for a Constructive Tomorrow. (2021).
Daily Caller News Foundation. (2022).
Elgin, D. J., & Weible, C. M. (2013). A stakeholder analysis of Colorado climate and energy
issues using policy analytical capacity and the advocacy coalition framework. Review of
Policy Research, 30(1), 114–133.
Farrell, J. (2015). Network structure and influence of the climate change counter-movement.
Nature Climate Change, 6(4), 370–374.
Farrell, J. (2016). Corporate funding and ideological polarization about climate change.
Proceedings of the National Academy of Sciences, 113(1), 92–97.
Feezell, J. T. (2018). Agenda setting through social media: The importance of incidental news
exposure and social filtering in the digital era. Political Research Quarterly, 71(2), 482–
Freelon, D. (2018). Computational research in the post-API age. Political Communication,
35(4), 665–668.
Freelon, D., Marwick, A., & Kreiss, D. (2020). False equivalencies: Online activism from left
to right. Science, 369(6508), 1197–1201.
Fu, J. S., & Shumate, M. (2016). Hyperlinks as institutionalized connective public goods for
collective action online. Journal of Computer-mediated Communication, 21(1), 298–311.
Finn, S., Maher, J. K., & Forster, J. (2006). Indicators of information and communication
technology adoption in the nonprofit sector: Changes between 2000 and 2004. Nonprofit
Management and Leadership, 16(3), 277–295.
Gibbs, M., Meese, J., Arnold, M., & Carter, M. (2015). #Funerals and Instagram: Death, social
media and platform vernacular. Information, Communication & Society, 18(3), 255–268.
Giglietto, F., Righetti, N., Rossi, L., & Marino, G. (2020). It takes a village to manipulate
the media: Coordinated link sharing behavior during 2018 and 2019 Italian elections.
Information, Communication & Society, 23(6), 867–891.
Grimmer, J., & Stewart, B. M. (2013). Text as data: The promise and pitfalls of automatic
content analysis methods for political texts. Political Analysis, 21(3), 267–297. https://doi.
Guo, C., & Saxton, G. D. (2014). Tweeting social change: How social media are changing non-
profit advocacy. Nonprofit and Voluntary Sector Quarterly, 43(1), 57–79.
Guo, C., & Saxton, G. D. (2018). Speaking and being heard: How nonprofit advocacy organi-
zations gain attention on social media. Nonprofit and Voluntary Sector Quarterly, 47(1),
Guo, C., & Saxton, G. D. (2020). The quest for attention: Nonprofit advocacy in a social media
age. Stanford University Press.
Gupta, K., Ripberger, J., & Wehde, W. (2018). Advocacy group messaging on social media:
Using the narrative policy framework to study Twitter messages about nuclear energy pol-
icy in the United States. Policy Studies Journal, 46(1), 119–136.
28 Nonprofit and Voluntary Sector Quarterly 00(0)
Halpin, D. R., Fraussen, B., & Ackland, R. (2021). Which audiences engage with advocacy
groups on twitter? Explaining the online engagement of elite, peer, and mass audiences
with advocacy groups. Nonprofit and Voluntary Sector Quarterly, 50, 842–865. https://doi.
Hoffman, A. J. (2015). How culture shapes the climate change debate. Stanford University
Jackson, S. J., Bailey, M., & Welles, B. F. (2020). HashtagActivism: Networks of race and
gender justice. MIT Press.
Jackson, S. J., & Welles, B. F. (2015). #Ferguson is everywhere: Initiators in emerging counter-
public networks. Information, Communication & Society, 19(3), 397–418.
Jasny, L., Waggle, J., & Fisher, D. R. (2015). An empirical examination of echo chambers in
U.S. climate policy networks. Nature Climate Change, 5, 782–786.
Jung, K., & Valero, J. N. (2016). Assessing the evolutionary structure of homeless network:
Social media use, keywords, and influential stakeholders. Technological Forecasting and
Social Change, 110, 51–60.
Kirilenko, A. P., & Stepchenkova, S. O. (2014). Public microblogging on climate change: One
year of Twitter worldwide. Global Environmental Change Jo, 26(1), 171–182. https://doi.
Kim, Y. M., Hsu, J., Neiman, D., Kou, C., Bankston, L., Kim, S. Y., Heinrich, R., Baragwanath,
R., & Raskutti, G. (2018). The stealth media? Groups and targets behind divisive issue
campaigns on Facebook. Political Communication, 35(4), 515–541.
Littman, J., & Wrubel, L. (2019). Climate change tweets ids.
DVN/5QCCUU, Harvard Dataverse, V1
Lam, W. F., & Nie, L. (2020). Online or offline? Nonprofits’ choice and use of social media in
Hong Kong. VOLUNTAS: International Journal of Voluntary and Nonprofit Organizations,
31(1), 111–128.
Manovich, L. (2012). Trending: The promises and the challenges of big social data. In M.
K. Gold (Ed.), Debates in the digital humanities (pp. 460–475). University of Minnesota
Marlow, T., Miller, S., & Roberts, J. T. (2021). Bots and online climate discourses: Twitter dis-
course on President Trump’s announcement of US withdrawal from the Paris Agreement.
Climate Policy, 21, 765–777.
Meraz, S., & Papacharissi, Z. (2013). Networked gatekeeping and networked framing
on #Egypt. The International Journal of Press/Politics, 18(2), 138–166. https://doi.
Muchnik, L., Aral, S., & Taylor, S. J. (2013). Social influence bias: A randomized experiment.
Science, 341(6146), 647–651.
Natural Resources Defense Council. (2021).
Pearce, W., Holmberg, K., Hellsten, I., & Nerlich, B. (2014). Climate change on twitter: Topics,
communities and conversations about the 2013 IPCC Working Group 1 report. PLOS ONE,
9(4), Article 94785.
Pearce, W., Niederer, S., Özkula, S. M., & Sánchez Querubín, N. (2019). The social media
life of climate change: Platforms, publics, and future imaginaries. Wiley Interdisciplinary
Reviews: Climate Change, 10(2), e569.
Wu and Xu 29
Pond, P., & Lewis, J. (2019). Riots and Twitter: Connective politics, social media and fram-
ing discourses in the digital public sphere. Information, Communication & Society, 22(2),
Saffer, A. J., Yang, A., Morehouse, J., & Qu, Y. (2019). It takes a village: A social network
approach to NGOs’ international public engagement. American Behavioral Scientist,
63(12), 1708–1727.
Saxton, G. D., & Guo, C. (2020). Social media capital: Conceptualizing the nature, acquisi-
tion, and expenditure of social media-based organizational resources. International
Journal of Accounting Information Systems, 36, Article 100443.
Schmidt, J.-H. (2014). Twitter and the rise of personal publics. In K. Weller, A. Bruns, J.
Burgess, C. Puschmann, & M. Mahrt (Eds.), Twitter and society (pp. 3–14). Peter Lang.
Shin, J., Jian, L., Driscoll, K., & Bar, F. (2018). The diffusion of misinformation on social media:
Temporal pattern, message, and source. Computers in Human Behavior, 83, 278–287.
Trzyna, T. C., & Didion, J. (2001). World directory of environmental organizations: A hand-
book of national and international organizations and programs, governmental and non-
governmental, concerned with protecting the Earth’s resources. Routledge.
Tucker, J. A., Guess, A., Barberá, P., Vaccari, C., Siegel, A., Sanovich, S., Stukal, D., & Nyhan,
B. (2018). Social media, political polarization, and political disinformation: A review of
the scientific literature.
Turunen, J., & Weinryb, N. (2020). Organizing service delivery on social media platforms?
Loosely organized networks, co-optation, and the welfare state. Public Management
Review, 22(6), 857–876.
Tyson, A., & Kennedy, B. (2020, June 23). Two-thirds of Americans think government should do
more on climate. Pew Research Center.
Te’eni, D., & Young, D. R. (2003). The Changing Role of Nonprofits in the Network
Economy. Nonprofit and Voluntary Sector Quarterly, 32(3), 397–414. https://doi.
Vu, H. T., Do, H. V., Seo, H., & Liu, Y. (2019). Who leads the conversation on climate change?
A study of a global network of NGOs on Twitter. Environmental Communication, 14(4),
Wang, L., Yang, A., & Thorson, K. (2019). Serial participants of social media climate discussion
as a community of practice: A longitudinal network analysis. Information, Communication
& Society, 24(7), 941–959.
Welbers, K., van Atteveldt, W., Kleinnijenhuis, J., & Ruigrok, N. (2018). A gatekeeper among
gatekeepers: News agency influence in print and online newspapers in the Netherlands.
Journalism Studies, 19, 315–333.
Wetts, R. (2020). Models and morals: Elite-oriented and value-neutral discourse dominates
American organizations’ framings of climate change. Social Forces, 98(3), 1339–1369.
Williams, H. T. P., McMurray, J. R., Kurz, T., & Hugo Lambert, F. (2015). Network
analysis reveals open forums and echo chambers in social media discussions of cli-
mate change. Global Environmental Change, 32, 126–138.
30 Nonprofit and Voluntary Sector Quarterly 00(0)
Wojcik, S., & Hughes, A. (2019). Sizing up Twitter users. Pew Research Center: Internet,
Science & Tech.
Xu, W., & Saxton, G. D. (2019). Does stakeholder engagement pay off on social media? A
social capital perspective. Nonprofit and Voluntary Sector Quarterly, 48(1), 28–49. https://
Yang, A., & Saffer, A. (2018). NGOs’ advocacy in the 2015 refugee crisis: A study of agenda
building in the digital age. American Behavioral Scientist, 62(4), 421–439. https://doi.
Yang, A., & Saffer, A. (2019). Embracing a network perspective in the network society: The
dawn of a new paradigm in strategic public relations. Public Relations Review, 45(4),
Article 101843.
Yang, S. J., & González-Bailón, S. (2017). Semantic networks and applications in public opin-
ion research. In J. Victor, A. Montgomery, & M. Lubell (Eds.), The Oxford handbook of
political networks (pp. 327–353). Oxford University Press.
Author Biographies
Viviana Chiu Sik Wu, PhD, is an assistant professor at the University of Massachusetts
Amherst, School of Public Policy. Her research combines. Using mixed methods and computa-
tional approaches, to examine how nonprofits and individuals organize to advance social justice
and policy change through community philanthropy, public engagement, and social media
advocacy, as well as how nonprofits shape and are shaped by place-based disadvantage and
Weiai Wayne Xu, PhD, is an associate professor in the Department of Communication at the
University of Massachusetts Amherst. His research focuses on social media and online com-
munities, particularly how networked diffusion and relationship building afford public institu-
tions a new way to engage the public. He uses computational techniques to analyze content and
connections in social media content.
This article is a comprehensive empirical overview of environmental protection and conservation nonprofits’ discourse on social media. To what extent have these nonprofits framed climate change in their public discourse and how has it evolved over time? How do organizational characteristics and resources affect their social media behavior? To address our research questions, we use machine learning with texts—specifically topic modeling—to track the activity of 120 environmental nonprofits during a 14-year time span on X, formerly known as Twitter. Our analysis of more than 1.3 million tweets shows that climate change, although not closely aligned with the missions for more than half of the top tweeting organizations included in our sample, has consistently been a prevalent priority issue on their social media agendas for more than a decade. This heightened attention to climate change discourse by the environmental nonprofit sector denotes their uniform efforts to inspire government for climate action.
Full-text available
Numerous studies to date have interrogated United States (US) think tanks—and their networks—involved in climate change countermovement (CCM). Comparatively in Europe (EU), research has been lacking. This investigation therefore attends to that gap. We conducted a frame analysis on eight most prominent contrarian think tanks in six countries and four languages in Europe over 24 years (1994–2018). We found that there has been consistent contrarian framing through think tanks in the EU regarding climate change. Yet, we found a proliferation of contrarian outputs particularly in recent years. This uptick in quantity correlates with increases in CCM activities in the US. Our content analyses showed that well-worn climate change counter-frames spread by US CCM organizations were consistently circulated by European organizations as well. Moreover, we found that, as in the US, neoliberal ideological stances stood out as the most frequently taken up by contrarian think tanks in Europe. As such, we documented that CCM tropes and activities have flowed strongly between US and EU countries.
In the early days of social media, social scientists speculated that it could support democracy because large media conglomerates could not dominate it. Regarding climate change, research documents the influence of the fossil fuel industry in climate denial discourses in the traditional media. Therefore, a similar hope emerged that a more democratized platform would see less polarization in belief about climate change's scientific basis. However, online public opinion about climate change continues to be extremely polarized, and the social forces behind it are not well understood. Here, we examine the role of one proposed mechanism for the systemic polarization and spread of disinformation in online discourses - the automated social media bot. This article pioneers the use of bot detection software on climate change discussion on Twitter, using automated monitoring of 6.8 million tweets. We examine the period around the time of U.S. President Donald Trump's announcement of U.S. withdrawal from the Paris Agreement on June 1, 2017. We find that the announcement generated an immediate online social movement response with very low suspected bot presence. Prior to and afterwards, however, suspected bots were responsible for approximately 25% of original tweets. Additionally, we find that suspected bots were more frequent in some topic areas than others, including denialist discourses focused on research questioning the reality or importance of climate change. Key policy insights • On an average day during our study period, automated bots produced an estimated one-quarter of all original tweets referencing climate change and global warming. • Bots were more active in some discussion areas than others - including climate denialist messages. • President Trump's announcement of the U.S.'s planned withdrawal from the Paris Agreement generated an immediate online social movement response with very low suspected bot presence. • For social media to deliver on its promise of a decentralized and democratic forum, automated accounts need to be identified, marginalized, and removed.
How are nonprofit organizations utilizing social media to engage in advocacy work? We address this question by investigating the social media use of 188 501(c)(3) advocacy organizations. After briefly examining the types of social media technologies employed, we turn to an in-depth examination of the organizations’ use of Twitter. This in-depth message-level analysis is twofold: A content analysis that examines the prevalence of previously identified communicative and advocacy constructs in nonprofits’ social media messages; and an inductive analysis that explores the unique features and dynamics of social media-based advocacy and identifies new organizational practices and forms of communication heretofore unseen in the literature.
Gaining an audience on social media is an important goal of contemporary policy advocacy. While previous studies demonstrate that advocacy-dedicated nonprofit organizations—what we refer to as advocacy groups—use different social media tools, we still know little about what specific audiences advocacy groups set out to target on social media, and whether those audiences actually engage with these groups. This study fills this gap, deploying survey and digital trace data from Twitter over a 12-month period for the Australian case. We show that while groups target a variety of audiences online, there are differences between group types in their strategic objectives and the extent to which particular audiences engage with them. Business groups appear to target elite audiences more often compared with citizen and professional groups, whereas citizen groups receive more online engagement from mass and peer audiences.
Digital media are critical for contemporary activism-even low-effort "clicktivism" is politically consequential and contributes to offline participation. We argue that in the United States and throughout the industrialized West, left- and right-wing activists use digital and legacy media differently to achieve political goals. Although left-wing actors operate primarily through "hashtag activism" and offline protest, right-wing activists manipulate legacy media, migrate to alternative platforms, and work strategically with partisan media to spread their messages. Although scholarship suggests that the right has embraced strategic disinformation and conspiracy theories more than the left, more research is needed to reveal the magnitude and character of left-wing disinformation. Such ideological asymmetries between left- and right-wing activism hold critical implications for democratic practice, social media governance, and the interdisciplinary study of digital politics.
Over the last few years, a proliferation of attempts to define, understand and fight the spread of problematic information in contemporary media ecosystems emerged. Most of these attempts focus on false content and/or bad actors detection. In this paper, we argue for a wider ecological focus. Using the frame of media manipulation and a revised version of the ‘coordinated inauthentic behavior’ original definition, the paper presents a study based on an unprecedented combination of Facebook data, accessed through the CrowdTangle API, and two datasets of Italian political news stories published in the run-up to the 2018 Italian general election and 2019 European election. By focusing on actors’ collective behavior, we identified several networks of pages, groups, and verified public profiles (‘entities’), that shared the same political news articles on Facebook within a very short period of time. Some entities in our networks were openly political, while others, despite sharing political content too, deceptively presented themselves as entertainment venues. The proportion of inauthentic entities in a network affects the wideness of the range of news media sources they shared, thus pointing to different strategies and possible motivations. The paper has both theoretical and empirical implications: it frames the concept of ‘coordinated inauthentic behavior’ in existing literature, introduces a method to detect coordinated link sharing behavior and points out different strategies and methods employed by networks of actors willing to manipulate the media and public opinion.