Speaking and Being Heard: How Nonprofit Advocacy Organizations Gain Attention on Social Media

Abstract

The social media era ushers in an increasingly “noisy” information environment that renders it more difficult for nonprofit advocacy organizations to make their voices heard. How then can an organization gain attention on social media? We address this question by building and testing a model of the effectiveness of the Twitter use of advocacy organizations. Using number of retweets and number of favorites as proxies of attention, we test our hypotheses with a 12-month panel dataset that collapses by month and organization the 219,915 tweets sent by 145 organizations in 2013. We find that attention is strongly associated with the size of an organization’s network, its frequency of speech, and the number of conversations it joins. We also find a seemingly contradictory relationship between different measures of attention and an organization’s targeting and connecting strategy.

Keywords

advocacy attention nonprofit organizations social media Twitter

In just over 10 years, social media have arguably become both the best ally and worst enemy of nonprofit organizations in their effort to advocate for the people and communities they serve. Regarded by some as a “microphone for the masses” (Murthy, 2011), a “voice for the people” (Wattam, 2009), or “cyberactivism” (McNutt & Menon, 2008), social media offer an alternative broadcast and communication medium for nonprofit organizations, a low-cost, interactive tool to speak out and to educate, engage, mobilize, and build rapport with large audiences of supporters. Yet in the meantime, and ironically, social media have ushered in a “noisy” information environment that renders it more difficult for nonprofits to make their voices heard. Operating in an increasingly information-saturated world populated by people with limited information-processing capacities and short attention spans (Simon, 1971), organizations are struggling to effectively grab and hold the attention of their supporters and the general public.

Public attention is thus a key prerequisite for social change and an immediate measure of the effectiveness of an organization’s social media usage. This is especially true for nonprofit advocacy organizations: When using social media to speak out on a cause, their voices must be heard by audiences of current and potential supporters before any more tangible outcomes might occur. Yet prior research, with a heavy focus on analyzing whether and how nonprofit advocacy organizations use social media (e.g., Bortree & Seltzer, 2009; Edwards & Hoefer, 2010; Greenberg & MacAulay, 2009; Guo & Saxton, 2014), has barely touched on the effectiveness of that social media usage.

This article represents a focused effort to address this gap. Our key research question is the following: How does an organization gain supporters’ attention with its social media messages? To answer this question, we investigate the Twitter use of 145 nonprofit advocacy organizations. Given the dearth of existing theory, we develop an original four-factor explanatory model for understanding why some advocacy organizations get attention on social media while others do not. Specifically, the amount of attention an organization receives is modeled as a function of network characteristics and communication strategy. In colloquial terms, we argue that the extent to which an organization is “being heard” (i.e., the attention it receives) depends on the size of the audience, how much and how it speaks, and what it says.

We test our hypotheses with a 12-month panel dataset that collapses by month and organization the 219,915 tweets sent by the 145 organizations over the entire 12 months of 2013. Using number of retweets and number of favorites as proxies of attention, we find that attention is positively associated with the size of an organization’s network (i.e., number of followers), its volume of speech (i.e., number of tweets sent), and how many “conversations” it joins (i.e., number of hashtags employed). We also find interesting relationships between attention and various measures of an organization’s targeting and connecting strategy. For instance, we find that the frequency with which an organization retweets or shares other users’ tweets is positively associated with the number of retweets of the organization’s own messages (one measure of attention) and yet is negatively associated with the number of favorites of the organization’s tweets (the other measure of attention). We find the opposite relationship with the use of targeted public reply messages, which are negatively related to the number of retweets received but positively associated with the number of favorites.

This article makes several contributions to the literature. To start, it adds to the literature on the effectiveness of computer-mediated organizational communication. To our knowledge, this is the first organizational-level study of social media message effectiveness. As nonprofit organizations increasingly turn to social media to engage their publics, it is critical they develop effective strategies that make better usage of their limited organizational resources and capacities. This study also contributes to the nonprofit advocacy literature by providing a framework for understanding the antecedents and consequences of effective social media advocacy. Along this line of thought, we establish “attention” as a key intermediate goal and important resource that a nonprofit advocacy organization must acquire to effectively promote its causes. We find that the extent to which an organization is “being heard” depends not only on how much it speaks, the size of its network, its targeting and connecting strategy, and visual content, but also on the specific measure of audience attention examined. Our study thus provides some meaningful practical guidance to organizations seeking to garner attention from the public in an increasingly noisy yet exciting information environment.

Theory and Hypothesis Development

The Role of Attention in Social Media Advocacy

As seen in the policy window (Howlett, 1998; Kingdon, 1984) and punctuated equilibrium (Baumgartner & Jones, 2010; Jones & Baumgartner, 2005) arguments, the policy agenda-setting literature has long recognized that the government, the public, and the media have limited attentional capacities and that they only pay attention to a restricted number of issues at any given time. Most of the agenda-setting models operate at the macro level, with the nation or public as their main unit of analysis (Wood &Vedlitz, 2007). A general implication of these macro-level theories is that social movements and advocacy organizations influence policy agenda setting by reacting to broad shifts in public attention. We are interested here, however, in the micro-level. Namely, within the current policy agenda context, what drives the public’s attention to an organization and its cause? In short, while advocacy organizations certainly adapt to “windows of opportunity” and macro-level changes in societal attention when promoting their causes, we are interested in how a particular organization garners attention to the messages it sends on social media by connecting with its target audiences at the micro-level.

At the micro-level, advocacy organizations are presented with a different challenge. With so much information available through numerous venues, donors and supporters today are better equipped to assess a wider variety of causes and organizations than at any time in the past. Yet this abundance of information comes with a price. As Herbert Simon (1971) pointed out, “a wealth of information creates a poverty of attention” (p. 40). In effect, we find ourselves drowning in information but short of the attention to make sense of it (Lanham, 2006). For this very reason, attention, defined as the allocation of “information-processing capacity . . . to environmental stimuli over time” (Simon, 1971; Sproull, 1984, p. 10), is regarded by some as the scarcest resource in today’s organizations (Davenport & Beck, 2001; Ocasio, 2011; Ren & Guo, 2011).

Although it has yet to be studied explicitly, the attention-deficit problem would appear to be especially critical for advocacy organizations, for whom the capture of public attention—whether offline or on social media—is often a prerequisite for achieving any more tangible strategic outcomes (see Tufekci, 2013). For the purposes of this study, we build on prior research (Webster, 2011) to define public attention to an organization on social media as the extent to which multiple audience members (individuals and organizations) react to the messages sent by an organization on its social media platform(s). We argue public attention is a necessary step for achieving social outcomes, for it is through attention that advocacy organizations are able to convince, connect, counteract, recruit, and mobilize.

It is within this context that social media have been offered as a valuable tool for mobilizing and garnering attention for nonprofit advocacy organizations, and a growing number of studies have begun to explore whether, why, and how these organizations are using various Web 2.0 and social media tools (Bortree & Seltzer, 2009; Edwards & Hoefer, 2010; Greenberg & MacAulay, 2009; McNutt & Menon, 2008; Petray, 2011). Among other things, these studies have highlighted the decentralized nature of social media where, with its highly participatory structure and low barriers to entry, controlling the message it conveys or the direction of discussion presents a formidable organizational challenge (Edwards & Hoefer, 2010; Farrar-Myers & Vaughn, 2015). At the same time, the sheer amount of (often redundant) information made available by social media has resulted in a substantial signal-to-noise problem (see Hermida, 2010; Klein, 2012) whereby it is increasingly difficult for the recipients of information to distinguish useful, desirable information (“signals”) from irrelevant information (“noise”).

On social media, the “signals” come in the form of the series of brief dynamic updates, or more simply messages, the organizations send to their followers. It is through their messages—their tweets, pins, videos, and status updates—that advocacy organizations seek to bring awareness to their struggle (Petray, 2011), tighten ties with their community of followers (Guo & Saxton, 2014), and mobilize constituents to collective action (Obar, Zube, & Lampe, 2012). Social media are, in effect, used for a variety of strategic purposes, yet the achievement of each purpose is dependent on the public paying attention to what the organization is saying. Therefore, a key immediate resource for the organization—and measure of the effectiveness of its work on social media—is the level of attention the public pays to its messages.

Differently put, if we want to understand whether organizations’ actions on social media are “working,” we need to zero in on the audience’s reaction to the key tools the organizations use—the messages. The audience’s immediate reactions to organizations’ messages—their sharing, liking, and commenting behaviors—reflects the level of attention they are paying to these messages. Because these reactions are publicly visible and in a standardized, accessible format (Saxton & Waters, 2014), we are able to employ quantifiable and comparable measures of attention different organizations are acquiring from their social media efforts.

Despite this, existing research focuses almost exclusively on either the adoption or organizational uses of social media (e.g., Bortree & Seltzer, 2009; Edwards & Hoefer, 2010; Greenberg & MacAulay, 2009; Guo & Saxton, 2014). Scholars have yet to examine the extent to which advocacy organizations—or, indeed, nonprofit organizations in general—are effective in their social media use. We maintain that, in the context of information overload, a critical (and immediate) measure of the effectiveness of any given message—whether online or offline—is the level and type of attention it receives. While offline such attention has been difficult to capture, social media (as a form of Big Data) make visible—and testable—certain phenomena that were previously invisible or not amenable to testing (Clark & Golder, 2015). It is thus with audience attention to social media messages we are able to examine attention in a context where individual pieces of organizational communication are related to specific, visible, comparable indicators of public attention.

In the next section, we present a framework for understanding the factors that facilitate or hinder the effort of advocacy organizations to gain attention on social media.

Hypotheses

This is the first study of which we are aware to statistically model why some organizations get attention on social media while others do not. We cannot therefore build directly off an existing model; we can, however, build a model that is informed from diverse prior literature in marketing, communication, public relations, computer science, and nonprofit studies that deal with, on one hand, the effects of different types of social media messages and, on the other hand, the determinants of audience reactions to social media messages. Based on our evaluation of this literature and our own understanding of this phenomenon, we present and test a four-factor explanatory model of the determinants of audience attention to social media messages as shown in Figure 1.

Figure 1.

Factors determining the level of attention an organization receives on social media.

The four factors can be arranged into two broad categories. The first factor deals with audience- or network-level characteristics, while the remaining three—timing and pacing, targeting and connecting, and content—cover message-level characteristics that reflect the organization’s communication strategy. Although it is not the focus of our current study, the final element in the model, “Dialogue and mobilization for social change,” highlights our argument that attention is a near-term and/or immediate goal for advocacy organizations on social media—The ultimate goal is often to mobilize its online network to support the cause.

Network characteristics

Our first argument is that the extent to which an organization is “heard” depends on characteristics of the organization’s social media network, particularly its size. There is much intuitive sense in this assertion, and a growing line of research has found evidence of the resources that can accrue to organizations from their social networks. For instance, Eng, Liu, and Sekhon (2012) found that informal and personal relationships, so-called “relationally embedded” network ties, are strongly linked to resource acquisition such as donor or volunteer support. Reddick and Ponomariov (2013), meanwhile, found that people who are members of voluntary associations are more likely to donate online. Saxton and Wang (2014), in turn, found the social network effect to be one of the most important determinants of the amount of donations received by organizations on the crowdfunding site Facebook Causes.

The positive relationship between network ties and resource acquisition is not limited to tangible financial or human resources. Entrepreneurship research provides evidence that social networks can also generate intangible resources such as access to information, advice, and new ideas (e.g., Hills, Lumpkin, & Singh, 1997). We consider attention a vital intangible resource—one whose potential is contingent on the size of an organization’s social media network. More specifically, we posit the informal belonging to an online network implied by following an organization makes it more likely for a given user to afford attention to that organization in the form of retweeting or favoriting the organization’s messages. In fact, individual-level research on Twitter has found that network size has a strong relationship with retweeting and message popularity (e.g., Bakshy, Hofman, Mason, & Watts, 2011). We thus hypothesize as follows:

Hypothesis 1 (H1): The level of attention an organization receives will be positively associated with the number of followers the organization has.

Communication strategy

Just as critical as network characteristics are features of the messages that reflect the organization’s overall communication strategy. We identify three salient dimensions of communication strategy: timing and pacing, targeting and connecting, and content.

Timing and pacing

Timing and pacing concerns when and how often an organization communicates. In this initial test, we argue that the extent to which an organization is “heard” depends on how loud it “speaks,” as reflected in the volume and frequency of messages sent. Supporting evidence can be found in the literature on, among other places, medical research advocacy. For example, findings from a study of the relationship between funding by the National Institutes of Health and the “burden” of various conditions or diseases show that some (AIDS, breast cancer, diabetes mellitus, and dementia) receive disproportionate funding and that these diseases happen to be the same ones that have the most vocal advocates (Gross, Anderson, & Powe, 1999; see also Elman, Ogar, & Elman, 2000).

To capture the volume of an organization’s speech, we concentrate on the number of tweets an organization sends in a given month. Although it is likely there is a point of diminishing returns on tweet frequency, we expect that, all other things being equal, organizations that tweet more frequently will be more likely to receive attention.

Hypothesis 2 (H2): The level of attention an organization receives will be positively associated with the number of tweets sent by the organization.

Targeting and connecting strategy

The third factor in the model posits that the extent to which an organization is “heard” depends on its targeting and connecting strategy. Bonk (2010) discussed the crucial role strategic communications can play in developing nonprofits—with a particular emphasis on targeted audiences reached through targeted messages. Such ideas have their roots in the “relational turn” that has occurred in public relations since the late 1990s, where theorizing has shifted from an emphasis on managing communication toward building and maintaining relationships (Broom, Casey, & Ritchey, 1997; Ledingham & Bruning, 1998). Dovetailing with this idea, Guo and Saxton (2010) found evidence in support of a positive relationship between stakeholder communication and the scope and intensity of nonprofit advocacy activities, at least within the offline context. And in the online context, evidence is mounting that social media have greatly expanded organizations’ potential for interacting and building relationships with stakeholders through direct and unmediated targeting and connecting efforts (e.g., Waters, Burnett, Lamm, & Lucas, 2009). We develop separate hypotheses to examine these two interrelated efforts.

First, with respect to targeting, communication directed at specific audience members is more likely to receive attention from those audience members inasmuch as targeted communication both fosters a sense of reciprocity and conveys a perception of personalness: The former, as highlighted by social exchange theory (Homans, 2013), triggers an obligation to reciprocate in a social interaction; whereas the latter signals trust and intimacy, which motivates those audience members to respond to the communication (see Jang & Stefanone, 2011). On Twitter, targeted messages are indicated by the use of the “@USERNAME” convention at the start of the message; these public reply messages are a form of “public email” that represent a departure from “broadcast” messages in that they are directed at a specific audience member. Moreover, they reflect an organization’s explicit attempts at dialogic conversation with a specific user or set of users (Lovejoy & Saxton, 2012). This leads to our third hypothesis:

Hypothesis 3 (H3): The level of attention an organization receives will be positively associated with the number of public reply messages sent by the organization.

In a related manner, we posit a relationship between an organization’s connecting behavior and levels of audience attention. This conforms with marketing research that finds audience engagement in social media linked to increased advertising effectiveness (Calder, Malthouse, & Schaedel, 2009). On Twitter, there are four types of tweeting behavior that reflect organizations’ efforts to engage and connect with specific Twitter entities. First, organizations can connect with messages (and message senders) through sharing or retweeting an existing message. Retweeting a message sent by another user shows the organization is paying attention to that user; retweeting amounts to a form of connection in that sharing a message is an implicit endorsement of the user who originally posted the message, forging a “message tie” between the originator and disseminator (Saxton & Waters, 2014). Second, organizations can connect to existing topics by including a specific hashtag (#) in its tweets. Third, an organization can connect to existing content through the inclusion of URLs (hyperlinks). And fourth, an organization can connect to specific people through the use of user mentions, which occur when the organization includes a user’s “@USERNAME” anywhere except the start of a tweet (tweets that begin with “@USERNAME” are, as described in H3, a form of public email coded separately as public reply messages). Each of these four tools represents a different facet of the connecting quality of a social media message. We thus posit the following four hypotheses:

Hypothesis 4a (H4a): The level of attention an organization receives will be positively associated with the number of the organization’s tweets that are retweets of other users’ tweets.

Hypothesis 4b (H4b): The level of attention an organization receives will be positively associated with the number of hashtags included in an organization’s tweets.

Hypothesis 4c (H4c): The level of attention an organization receives will be positively associated with the number of URLs included in an organization’s tweets.

Hypothesis 4d (H4d): The level of attention an organization receives will be positively associated with the number of user mentions included in an organization’s tweets.

Visual content

The fourth and final factor in the model posits that the extent to which an organization is “heard” depends on what it communicates; in other words, content matters. There are numerous ways of approaching and measuring the content of a message. For this initial test, we concentrate on visual content—on the inclusion of videos and images—which have been shown to be related to the shareability of messages on Facebook (Saxton & Waters, 2014). In marketing and advertising research, pictorial illustrations are considered the most important element in print advertisements for capturing consumers’ attention (see Pieters & Wedel, 2004). We draw a parallel argument and expect social media messages with visual content to receive greater audience attention.

Given the nature of the Twitter application programming interface (API), videos and images can be included in a tweet in different ways. Some pictures are “included” in a tweet, such that when a tweet is viewed, the image will appear automatically. Other pictures, in contrast, such as those posted via Instagram, do not appear in the tweet itself but are instead included via a link. All videos are likewise included solely in the form of links. We thus formulate three hypotheses to capture these different ways of including visual content:

Hypothesis 5a (H5a): The level of attention an organization receives will be positively associated with the number of an organization’s messages that include a photo.

Hypothesis 5b (H5b): The level of attention an organization receives will be positively associated with the number of an organization’s messages that include one or more photo links.

Hypothesis 5c (H5c): The level of attention an organization receives will be positively associated with the number of an organization’s messages that include one or more video links.

Method

Sample

Our sample comprises the 188 “Civil Rights and Advocacy” organizations rated in 2011 by Charity Navigator, an independent nonprofit organization that evaluates the financial health of U.S. charities. To be evaluated, an organization must be a 501(c)(3) charitable organization, have available at least four consecutive years of IRS Form 990, and receive public support greater than US$500,000 and total revenue greater than US$1,000,000. The average organization in our sample had US$8.7 million in total revenues and US$8.8 million in total expenses in the most recently completed fiscal year. They cover a range of sizes and advocacy issue areas, including health, education, civil rights, the environment, and others.

Data Collection

In this test, we focus on organizations’ use of Twitter. Twitter is arguably the world’s premier message network (Lovejoy & Saxton, 2012). It is well suited to advocacy work, has an open API, and broadly serves as a proxy for organizations’ overall social media use. We first determined each organization’s Twitter adoption profile through a review of its website supplemented by queries on the Google and Twitter search interfaces. We then gathered detailed Twitter data on each of the 145 organizations with an active Twitter profile. Python code (available upon request) was then written to access the Twitter API and download all Twitter activity for organizations with an active profile over the entire 12 months of 2013.

Operationalization

We operationalize two dependent variables that serve as proxies for attention: (a) Number of Retweets, defined as the total number of retweets of an organization’s tweets by other users in a given month, and (b) Number of Favorites, defined as the total number of favorites of an organization’s tweets by other users in a given month. A retweet is a reposting of someone else’s tweet; it is a popular feature for spreading news or sharing information on Twitter. A favorite is represented by a small star icon below a tweet; this feature is an archiving function that is used when a user likes a particular tweet and/or would like to save it for later. Both features indicate that a user has paid some attention to a particular tweet.

We also operationalize the following 10 independent variables, all measured monthly for each organization. To start, H1 is operationalized with Followers, which indicates the organization’s number of Twitter followers. H2, which relates to the timing and pacing of the organization’s communication strategy, is operationalized with Tweets, a count of the total number of tweets sent by the organization. The targeting element of the communication strategy, reflected in H3, is operationalized through Public Reply, a measure of the total number of public reply messages (tweets starting with @USERNAME) sent by the organization. The connecting element of the organization’s strategy (H4a-H4d), meanwhile, is operationalized through a series of four variables: Retweets, measured as the total number of an organization’s messages that are retweets of other users’ tweets; Hashtags, the total number of hashtags included in an organization’s tweets; Hyperlinks, the total number of hyperlinks (URLs) included in an organization’s tweets; and User Mentions, the total number of user mentions included in an organization’s tweets. Finally, the content aspect of communication strategy (H5a-H5c) is operationalized through three variables that tap the inclusion of visual content: Photos, which measures the total number of an organization’s messages that include one or more photos; Photo Links, which measures the total number of an organization’s messages that include one or more photo links; and Video Links, which measures the total number of an organization’s messages that include one or more video links.

Finally, we operationalize the following two control variables. Organization Size: The size of the organization in terms of annual revenue. It is defined as a continuous variable, and measured as an organization’s reported annual revenue for the year 2011. Organization Age: The age of the organization in years. It is defined as a continuous variable, and measured as the difference between the year of 2013 and the year when the Internal Revenue Service granted an organization 501(c)(3) status.

Results

Before turning to multivariate tests of our model, we report findings from descriptive statistics (shown in Table 1) on all of the monthly level Twitter variables in our model.

Table 1.

Monthly Descriptive Statistics, January to December 2013 for 145 Advocacy Nonprofits.

Variable	Observations	M	SD	Minimum	Maximum
Message Attention Proxies
# Retweets of organization’s tweets	1,679	2,183.42	9,206.90	0	206,624
# Favorites of organization’s tweets	1,679	209.59	892.78	0	14,895
Network Size
# Followers of the organization	1,679	15,056.54	35,126.05	22	334,949
Volume
Total # of tweets	1,679	130.97	126.30	1	1,341
Targeting and Connecting Strategy
# of public reply messages	1,679	11.64	22.37	0	252
# of retweets of others’ tweets	1,679	27.38	44.45	0	826
# Tweets with ≥1 URL	1,679	86.01	84.08	0	850
# Tweets with ≥1 hashtag	1,679	77.56	94.54	0	1,010
# Tweets with ≥1 mention	1,679	74.87	88.76	0	947
Content
# Tweets with ≥1 photo	1,679	5.90	10.80	0	102
# Tweets with ≥1 photo link	1,679	1.12	4.28	0	60
# Tweets with ≥1 video link	1,679	1.24	2.70	0	32
Control Variables
Organization size	1,679	9,771,249	17,700,000	522,960	118,854,085
Organization age	1,679	30.44	13.68	6	89

Note. Data shown are based on each organization’s monthly values for the variables indicated. Number of observations is less than 1,740 (145 organizations × 12 months) due to the fact that some organizations did not tweet in a given month.

Descriptive Statistics

Proxies for attention: Retweeting and favoriting

To start, in terms of the amount of attention garnered, the data reveal interesting variation in the amount of retweets and favorites afforded by the Twitter community to the messages sent by the 145 organizations. As shown in Table 1, in a typical month, an organization received on average 2,183 retweets (median = 350) and 210 favorites (median = 34) from their audience, though the range could be dramatic, ranging from zero to 206,624 retweets and from zero to 14,895 favorites. Basically, as suggested by the large difference between the mean and median scores and the wide range, the distribution of these data approximate not a normal “bell curve” distribution but rather—as is common with many social network-based phenomena—a power law distribution (Barabási & Albert, 1999). We thus see a large number of organizations garnering low levels of attention and a smaller subset generating very high levels of attention.

Network size

The average organization in our sample had 15,057 followers (median = 5,349). The range was vast, from a low of 22 followers for the National Child Safety Council in January 2013, to a high for the Human Rights Campaign Foundation of 334,949 in December.

Volume of tweets

The most important technological feature of Twitter is the ability to send short messages, or tweets, or 140 characters or fewer. The 145 organizations in our sample with Twitter accounts sent a total of 219,915 tweets during the 12 months of 2013. As shown in Table 1, on average, an organization sent about 131 tweets per month during the 12-month period, roughly 4.4 tweets per day. Compared with 2.3 tweets/day sent out by NPTimes 100 organizations (Lovejoy, Waters, & Saxton, 2012) in 2009, these advocacy organizations in 2013 are heavier tweeters. Yet there is much variation: Some organizations sent as many as 1,000 tweets, while several sent only one, and five of them (not included in final sample) did not send a single one.

Targeting and connecting strategy

There are a number of technological tools available to organizations in their tweets that allow organizations to connect with and target specific constituencies: public reply messages (also known as direct messages), retweets, hyperlinks, hashtags, and user mentions. We found 8.89% (n = 19,544) of all tweets were direct messages, well below the 16% by nonprofits on the NPTimes 100 list (Lovejoy et al., 2012) or the 22% sent by the individuals studied by Hughes and Palen (2009), respectively. As shown in Table 1, the average organization sent about 12 public reply messages in a typical month. We found 20.91% of the tweets (n = 45,978) were retweets, with the average organization sending 27.38 retweets in a typical month. This is more than the 16.2% found by Lovejoy et al. (2012) for NPTimes 100 organizations, and substantially less than the 28% found by Hughes and Palen (2009) for individuals during natural emergencies.

Three remaining tools—hyperlinks, hashtags, and user mentions—are available within tweets and are not mutually exclusive. Hyperlinks were included in 65.67% of the tweets (n = 144,417), with the average organization sending 86 tweets with a hyperlink over the month. Hashtags were included in 59.22% of all tweets (n = 130,236), with the average organization sending around 78 tweets with at least one hashtag over the course of a typical month. We found 57.17% of the tweets (n = 125,728) contained at least one user mention, and the average nonprofit sent 75 such messages over the course of a given month.

Visual content

Organizations are also able to include a variety of different media in their tweets. We found 4.51% (n = 9,913) of all tweets sent contained a photo, with a typical organization sending out six such tweets in an average month; 0.95% (n = 2,079) of all tweets contained a link to a video, and 0.86% (n = 1,882) contained a link to a photo, with the average organization sending out 1.24 and 1.12 such tweets, respectively, in a typical month.

Multivariate Tests

Table 2 contains zero-order correlations for all model variables. Our analysis of the table and variance inflation factors shows no problems with multicollinearity.

Table 2.

Zero-Order Correlation Matrix for All Model Variables.

	1	2	3	4	5	6	7	8	9	10	11	12	13	14	15
1. Retweets	1
2. Favorites	0.65*	1
3. Organization age	0.00	−0.03	1
4. Total revenue	0.04	0.04	0.39*	1
5. Followers	0.49*	0.78*	−0.02	0.17*	1
6. Tweets	0.26*	0.26*	0.16*	0.17*	0.25*	1
7. Public reply	0.18*	0.19*	0.08*	0.11*	0.17*	0.50*	1
8. Retweets of others	0.20*	0.11*	0.08*	0.03	0.11*	0.70*	0.27*	1
9. Photos	0.30*	0.40*	0.11*	0.13*	0.29*	0.51*	0.29*	0.42*	1
10. Video links	0.25*	0.28*	0.01	0.19*	0.31*	0.31*	0.20*	0.22*	0.41*	1
11. Photo links	0.14*	0.18*	0.08*	0.11*	0.20*	0.26*	0.18*	0.19*	0.18*	0.15*	1
12. Tweets with ≥1 URL	0.24*	0.27*	0.09*	0.18*	0.29*	0.91*	0.41*	0.50*	0.37*	0.28*	0.18*	1
13. Tweets with ≥1 hashtag	0.26*	0.25*	0.21*	0.21*	0.23*	0.90*	0.36*	0.63*	0.55*	0.30*	0.28*	0.77*	1
14. Tweets with ≥1 mention	0.27*	0.25*	0.16*	0.15*	0.23*	0.93*	0.57*	0.79*	0.50*	0.29*	0.25*	0.78*	0.84*	1
15. Month	−0.02	0.07*	0.01	0.00	0.07*	0.05*	0.07*	0.04	0.13*	0.02	0.01	0.06*	0.02	0.06*	1

p < .05.

All continuous variables in our models are transformed to natural log values to account for their skewed distribution. For each of the two dependent variables, we run two separate tests: ordinary least squares (OLS) regression and fixed effects regression. The fixed effects model is a useful additional model when (as in cases like ours) there may be some unknown or untested organizational characteristic that is influencing the dependent variable; in essence, the fixed effects model helps control for a potential omitted variable bias by removing the effects of time-invariant organizational characteristics to better evaluate the effect of the time-varying independent variables on our dependent variables.¹ For comparison purposes, we present the results of both OLS and fixed effects models in Table 3. The coefficients in Models 1 and 2 indicate the effects of each independent variable on the number of retweets, whereas the coefficients in Models 3 and 4 indicate the effects of each independent variable on the number of favorites.

Table 3.

Determinants of Audience Attention to Social Media Messages—Multivariate Analyses.

Model	Hypothesis	Predicted sign	(1)	(2)	(3)	(4)
			Dependent Variable = number of retweets		Dependent Variable = number of favorites
			OLS	Fixed effects	OLS	Fixed effects
# Followers	H1	+	0.596*** (0.021)	0.094 (0.129)	0.691*** (0.016)	1.085*** (0.075)
# Tweets	H2	+	0.914*** (0.105)	0.646*** (0.112)	0.378*** (0.080)	0.695*** (0.065)
Public Reply	H3	+	−0.071** (0.029)	−0.077** (0.033)	0.078*** (0.022)	0.042** (0.019)
Retweets of others	H4a	+	0.351*** (0.023)	0.383*** (0.042)	−0.047** (0.022)	−0.072*** (0.024)
Hashtags	H4b	+	0.053* (0.030)	0.137*** (0.045)	0.054** (0.023)	0.063** (0.026)
URLs	H4c	+	−0.089 (0.079)	−0.036 (0.085)	0.313*** (0.060)	0.011 (0.049)
Mentions	H4d	+	−0.064 (0.042)	0.073 (0.058)	−0.131*** (0.032)	−0.103*** (0.033)
Photos	H5a	+	0.046 (0.030)	0.035 (0.033)	0.172*** (0.023)	0.140*** (0.019)
Photo links	H5b	+	−0.043 (0.038)	−0.087** (0.039)	0.009 (0.029)	−0.011 (0.023)
Video links	H5c	+	0.080** (0.040)	0.046 (0.042)	0.113*** (0.030)	0.031 (0.024)
Organization size	Control		−0.087*** (0.024)	—	−0.041** (0.018)	—
Organization age	Control		0.150** (0.059)	—	0.033 (0.044)	—
Month	Control		−0.021*** (0.007)	—	0.023*** (0.005)	—
Constant			−2.795*** (0.367)	0.570 (1.116)	−4.865*** (0.278)	−8.722*** (0.645)
# of observations			1,679	1,679	1,679	1,679
F statistic (df)			520.37 (13)	126.39 (10)	667.49 (13)	156.19 (10)
R²/adjusted R²			.80/.80	.72	.84/.84	.82

Note. All continuous variables are transformed to natural log values to account for their skewed distribution. OLS = ordinary least squares.

p < .10. **p < .05. ***p < .01; standard errors in parentheses.

To recap, each of the independent variables is associated with a specific hypothesis related to one of the four primary factors in our explanatory model. In line with our hypothesis testing, we present our results here briefly factor by factor, before discussing the most important implications of these findings in the “Discussion” section.

First, in H1, we proposed that the level of attention an organization received would be positively associated with the organization’s number of followers. The regression analyses reveal a positive and significant relationship between the number of followers and both measures of attention (number of retweets and number of favorites) in three of the four models. More specifically, Model 1 (OLS) indicates a positive and significant association between number of followers and number of retweets, but this association disappears in Model 2 (fixed effects); Models 3 to 4 reveal a positive and significant association between number of followers and number of favorites. Thus our hypothesis is supported at least with regard to one measure of attention (number of favorites).

H2 predicted that the level of attention an organization received would be positively associated with the number of tweets sent by the organization. The regression analyses reveal a positive and significant relationship between the number of tweets sent by the organization and both measures of attention (number of retweets and number of favorites) across all four models. Thus, H2 is supported.

H3 examined the effect of targeting strategy. It predicted that the level of attention an organization received would be positively associated with the number of public reply messages (i.e., “@USER” messages) sent by the organization. Interestingly, the regression analyses reveal a significant relationship between the number of public reply messages and the dependent measures across the four models, but the direction of the relationship differs between the two dependent measures. More specifically, there is a negative relationship between the number of public reply messages and the number of retweets of the organization’s tweets; by contrast, there is positive relationship between the number of public reply messages and the number of favorites of the organization’s tweets. Therefore, H3 is partially supported.

Next, we created four hypotheses to examine the effect of connecting strategy: H4a through H4d described positive relationships with attention for the number of the organization’s tweets that are retweets of other users’ tweets and the number of hashtags, hyperlinks, and user mentions included in an organization’s tweets. Again, the regression analyses reveal a significant relationship between the number of an organization’s tweets that are retweets of others’ tweets and the dependent measures across the four models, but the direction of the relationship differs between the two dependent measures. More specifically, there is a positive relationship between an organization’s tweets that are retweets of others’ tweets and the number of retweets of the organization’s tweets, while there is a negative relationship between an organization’s tweets that are retweets of others’ tweets and the number of favorites of the organization’s tweets. Therefore, H4a is partially supported. The analyses reveal a positive and significant relationship between the number of hashtags included in an organization’s messages and both measures of attention (number of retweets and number of favorites) across all four models. Thus, H4b is supported. H4c, which posited a positive relationship between attention and hashtags, is partially supported, with a positive, significant coefficient on Hashtags in Model 3. H4d is not supported.

Finally, H5a, H5b, and H5c proposed that three types of visual content—tweets with photos, tweets with links to photos, and tweets with links to videos—would obtain a positive relationship with both dependent measures. The results are mixed. The regression analyses revealed a positive and significant relationship between photos and the number of favorites of the organization’s tweets, thus providing some support to H5a. However, H5b (links to photos) and H5c (links to videos) are largely rejected.

Discussion and Conclusions

This study represents a much-needed investigation into the social media use of nonprofit advocacy organizations. Our study is the first to examine the extent to which nonprofit advocacy organizations are effective in grabbing their audiences’ attention through social media channels. It is also to our knowledge the first organizational-level analysis of the impact of social media message strategies. Altogether, the study sheds new light on how social media help organizations engage in advocacy work, and it amounts to an important step in elucidating the challenges and opportunities of managing in the altered information landscape of the social media age.

The explanatory model we have developed points to interesting relationships for each of its four main factors: Aspects of network size, tweeting frequency, targeting and connecting strategy, and visual content all had a relationship to the amount of attention paid to an organization’s social media messages. Some of the findings deserve further discussion here. For example, the number of retweets of others—a measure of connecting strategy—is found to be positively related to the number of retweets and negatively related to the number of favorites. These seemingly contradictory results suggest that, while both the number of retweets and the number of favorites are indicators of the aggregated amount of attention that an organization’s tweets have received from the Twitterverse, there are some qualitative differences between the two. Retweeting as a function is often used as a reciprocal act of “giving and receiving attention”; that is, user A retweeting user B’s tweets increases the likelihood that user B will retweet user A’s tweets. Our finding corroborates the existence of such reciprocal attention. Favoriting, on the contrary, is often used as a bookmarking tool where a user keeps useful tweets for future reference. Our results suggest a user is more likely to favorite a tweet that has been targeted at them (a public reply message) or is original and/or visually stimulating, rather than a tweet that is simply a retweet of others (i.e., nonoriginal, noncreative).

These findings also reinforce the idea that there are different types of tweets, that serve different purposes, and that generate different outcomes. This is something of which nonprofit advocacy organizations should be aware. In terms of practical implications, our findings suggest that, to garner attention, nonprofits should seek to join conversations (by employing existing hashtags), speak often, and grow their follower base. They should also be aware that there are different levels and forms of attention gained from various tweeting activities. For instance, while targeted public reply messages may not be effective in garnering mass attention in the form of retweets, they may serve to strengthen ties with the users targeted in those messages, as reflected in favoriting behavior. Of course, our two chosen measures are not intended to provide a complete picture of attention; rather, we are interested in forms of attention that are more “intense” or “interactive” than simply reading the messages. Still, future research should look for ways to further operationalize the notion of audience attention.

The nonfindings of this study also raise some interesting issues. With the exception of tweets with photos, the visual content measures are largely insignificant, suggesting that providing photo or video links in an organization’s tweets does not increase the amount of attention those tweets receive from other users, at least in terms of the numbers of retweets and favorites. However, it is important to note that our data do not have information on whether other users have clicked on the photo or video links. Nevertheless, these findings suggest it would be a mistake for organizations to focus narrowly on visually pleasing content. Substantive textual content likely matters more than visual content, as it is mainly through the former that an advocacy organization disseminates policy-relevant ideas and position itself to be an expert or thought leader on a specific subject issue.

We should also stress that the goal of our study was not to deliver evidence at the message level or, that is, to bring evidence to bear on which specific message characteristics yield an audience response on a message-by-message basis. There is a growing body of message-level research in this area (e.g., Saxton & Waters, 2014). Our aim here, however, was to do something new: to shed light on a broader organizational communication strategy as seen in the aggregate monthly use of various communication tactics and tools. At the same time, our findings (and nonfindings) imply the need for further theoretical development on what leads to both a successful message and a successful communication strategy.

For instance, prior message-level research has found the number of followers to be the most powerful variable in determining levels of audience reaction (e.g., Bakshy et al., 2011; Saxton & Waters, 2014). While we find it to be an important factor, it is not the most important in all cases and, in one of four models, is insignificant. These findings suggest the need for additional audience characteristics measures such as the size or level of activity of the followers’ follower networks. Clearly, some followers are more influential or valuable than others, and future research might get good mileage from exploring such features.

One important question is whether the attention gained by the organization leads to any tangible or intangible organizational outcome. This is an area strongly in need of future research. Above all, for advocacy and social moment organizations, the ultimate long-term goal is not attention but instead spreading awareness, building coalitions, or mobilizing supporters to achieve some broader societal or policy goal. We argue that attention is a crucial first step toward such meaningful advocacy actions and outcomes, but this idea needs to be tested further.

As a preliminary check of the connection between attention and broader or longer term outcomes, we analyzed the relationship between the total amount of attention an organization received over the course of 2013 (as indicated by the total number of retweets as well as favorites) and the total number of new followers gained by the organization over the same time period. Figure 2 shows scatter plots (with regression line and 95% confidence intervals) for the increase in followers for retweets and favorites, respectively. The number of followers is important insofar as it is a rough indicator of social media capital, or social media–based social resources that an organization acquires via its social media efforts and which can be leveraged to produce longer term organizational outcomes (Saxton & Guo, 2014). What Figure 2 shows is a clear positive association between attention and growth in the follower networks. An increase in attention, it appears, leads to an increase in social media capital.

Figure 2.

Total audience reactions (retweets and favorites) in 2013 versus total gain in # of followers over 2013.

Another important question to consider is whether and to what extent this online attention generated by the organization can create a “real” impact. We briefly discussed the relationship between attention and more tangible outcomes but did not test it in our empirical analyses. Indeed, many people have raised doubts about the effectiveness of advocacy campaigns carried out in online forums or organized through social media, calling them “clicktivism” or “slacktivism.” The rationale behind such skepticism is that social media might turn activism into a button-pressing game analogous to channel flipping with the TV remote or point-and-click adventures on video-game consoles; in that case, the audience’s actions would only ultimately amount to information consumerism and likely fail to achieve any real impact. While such pessimistic possibilities certainly exist, recent social movement research has actually demonstrated a more optimistic view of the issue. For example, in a study of the consequences of Internet technology for social movements, Earl and Kimport (2011) have documented numerous instances where Web usage has transformed activism through E-mobilizations, E-tactics, and E-movements. In this study, we contribute to this discussion by highlighting “attention” as a key intermediate goal and important resource for advocacy organizations. Yet we also acknowledge that attention is not the end game: To effectively promote their causes, not only must these organizations obtain and sustain public attention on social media, but they also must be able to turn that attention into action.²

In any event, helping nonprofit advocacy organizations understand how to succeed in this new information era should be a critical goal for future research. Nonprofit organizations are increasingly mobilizing, engaging, and communicating with their stakeholders on social media. The stakes are getting higher, as is the amount of time nonprofits are expending on their social media efforts. Understanding what makes for successful communication in this arena is thus important. We hope our findings here help start an ongoing dialogue that ultimately helps nonprofit leaders make better decisions regarding how to allocate their time and effort in this fast-paced, sometimes frustrating, always fascinating environment.

Footnotes

Acknowledgements

The authors thank Editor Lucas Meijs and the three anonymous reviewers of this journal for their thoughtful comments. The authors also acknowledge participants in the George Washington University Trachtenberg School Research Colloquium for their useful input. Seongho An and Viviana Wu provided excellent research assistance. All of the usual caveats apply.

Authors’ Note

This article was accepted by the previous Nonprofit and Voluntary Sector Quarterly (NVSQ) editor, Dr. Lucas Meijs. An earlier version of this article was presented at the Association for Research on Nonprofit Organizations and Voluntary Action’s 43rd Annual Conference in Denver, Colorado, in 2014, and won the Best Conference Paper that year.

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.

Funding

The author(s) received no financial support for the research, authorship, and/or publication of this article.

Notes

Author Biographies

Chao Guo, PhD, is an associate professor in the School of Social Policy & Practice at the University of Pennsylvania. Guo is also founding codirector of the China Institute for Philanthropy and Social Innovation at Renmin University of China. His interests are in collaboration within and across sectors, new media and nonprofit advocacy, and social entrepreneurship.

Gregory D. Saxton, PhD, is assistant professor in the Schulich School of Business at York University in Toronto, Canada. Previously, he was associate professor in the Department of Communication at the University at Buffalo, State University of New York (SUNY). His interests are in the role and effects of technology—especially Big Data and social media—on the flow of information to and from organizations.

References

Bakshy

Hofman

J. M.

Mason

W. A.

Watts

D. J.

(2011). Everyone’s an influencer: Quantifying influence on Twitter. In Proceedings of the Fourth ACM International Conference on Web Search and Data Mining (pp. 65-74). New York, NY.

Barabási

A. L.

Albert

(1999). Emergence of scaling in random networks. Science, 286, 509-512.

Baumgartner

F. R.

Jones

B. D.

(2010). Agendas and instability in American politics. Chicago, IL: The University of Chicago Press.

Bonk

(2010). Strategic communications. In Renz

(Ed.), The Jossey-Bass handbook of nonprofit leadership and management (3rd ed., pp. 329-346). San Francisco, CA: Wiley.

Bortree

D. S.

Seltzer

(2009). Dialogic strategies and outcomes: An analysis of environmental advocacy groups’ Facebook profiles. Public Relations Review, 35, 317-319.

Broom

G. M.

Casey

Ritchey

(1997). Toward a concept and theory of organization-public relationships. Journal of Public Relations Research, 9, 83-98.

Calder

B. J.

Malthouse

E. C.

Schaedel

(2009). An experimental study of the relationship between online engagement and advertising effectiveness. Journal of Interactive Marketing, 23, 321-331.

Clark

W. R.

Golder

(2015). Big Data, causal inference, and formal theory: Contradictory trends in political science? PS: Political Science & Politics, 48, 65-70.

Davenport

T. H.

Beck

J. C.

(2001). The attention economy: Understanding the new currency of business. Cambridge, MA: Harvard Business Press.

10.

Earl

Kimport

(2011). Digitally enabled social change: Activism in the Internet age. Cambridge, MA: MIT Press.

11.

Edwards

H. R.

Hoefer

(2010). Are social work advocacy groups using Web 2.0 effectively? Journal of Policy Practice, 9, 220-239.

12.

Elman

R. J.

Ogar

Elman

S. H.

(2000). Aphasia: Awareness, advocacy, and activism. Aphasiology, 14, 455-459.

13.

Eng

Liu

Sekhon

(2012). The role of relationally embedded network ties in resource acquisition of British nonprofit organizations. Nonprofit and Voluntary Sector Quarterly, 41, 1092-1115.

14.

Farrar-Myers

V. A.

Vaughn

J. S.

(Eds.). (2015). Controlling the message: New media in American political campaigns. New York: New York University Press.

15.

Greenberg

MacAulay

(2009). NPO 2.0? Exploring the web presence of environmental nonprofit organizations in Canada. Global Media Journal—Canadian Edition, 2, 63-88.

16.

Gross

C. P.

Anderson

G. F.

Powe

N. R.

(1999). The relation between funding by the National Institutes of Health and the burden of disease. New England Journal of Medicine, 340, 1881-1887.

17.

Guo

Saxton

G. D.

(2010). Voice-in, voice-out: Constituent participation and nonprofit advocacy. Nonprofit Policy Forum, 1(1), Article 5. doi:10.2202/2154-3348.1000

18.

Guo

Saxton

G. D.

(2014). Tweeting social change: How social media are changing nonprofit advocacy. Nonprofit and Voluntary Sector Quarterly, 43, 57-79.

19.

Hermida

(2010). From TV to Twitter: How ambient news became ambient journalism. Media/Culture Journal, 13(2). Available at SSRN: https://ssrn.com/abstract=1732603.

20.

Hills

Lumpkin

G. T.

Singh

R. P.

(1997). Opportunity recognition: Perceptions and behaviors of entrepreneurs. Frontiers of Entrepreneurship Research, 17(4), 203-218.

21.

Homans

G. C.

(2013). The human group (Vol. 7). New York, NY: Routledge.

22.

Howlett

(1998). Predictable and unpredictable policy windows: Institutional and exogenous correlates of Canadian federal agenda-setting. Canadian Journal of Political Science, 31, 495-524.

23.

Hughes

A. L.

Palen

(2009). Twitter adoption and use in mass convergence and emergency events. International Journal of Emergency Management, 6(3-4), 248-260.

24.

Jang

C. Y.

Stefanone

M. A.

(2011). Non-directed self-disclosure in the blogosphere: Exploring the persistence of interpersonal communication norms. Information, Communication & Society, 14, 1039-1059.

25.

Jones

B. D.

Baumgartner

F. R.

(2005). The politics of attention: How government prioritizes problems. Chicago, IL: The University of Chicago Press.

26.

Kingdon

(1984). Agendas, alternatives and public choices. Boston, MA: Little, Brown.

27.

Klein

(2012). How to harvest collective wisdom on complex problems: An introduction to the MIT Deliberatorium. (Center for Collective Intelligence Working Paper).

28.

Lanham

(2006). The economics of attention. Chicago, IL: The University of Chicago Press.

29.

Ledingham

J. A.

Bruning

S. D.

(1998). Relationship management in public relations: Dimensions of an organization-public relationship. Public Relations Review, 24, 55-65.

30.

Lovejoy

Saxton

G. D.

(2012). Information, community, and action: How nonprofit organizations use social media. Journal of Computer-Mediated Communication, 17, 337-353.

31.

Lovejoy

Waters

R. D.

Saxton

G. D.

(2012). Engaging stakeholders through Twitter: How nonprofit organizations are getting more out of 140 characters or less. Public Relations Review, 38, 313-318.

32.

McNutt

J. G.

Menon

G. M.

(2008). The rise of cyberactivism: Implications for the future of advocacy in the human services. Families in Society: The Journal of Contemporary Social Services, 89, 33-38.

33.

Murthy

(2011). Twitter: Microphone for the masses? Media, Culture & Society, 33, 779-789.

34.

Obar

J. A.

Zube

Lampe

(2012). Advocacy 2.0: An analysis of how advocacy groups in the United States perceive and use social media as tools for facilitating civic engagement and collective action. Journal of Information Policy, 2, 1-15.

35.

Ocasio

(2011). Attention to attention. Organization Science, 22, 1286-1296.

36.

Petray

T. L.

(2011). Protest 2.0: Online interactions and Aboriginal activists. Media, Culture & Society, 33, 923-940.

37.

Pieters

Wedel

(2004). Attention capture and transfer in advertising: Brand, pictorial, and text-size effects. Journal of Marketing, 68, 36-50.

38.

Reddick

Ponomariov

(2013). The effect of individuals’ organization affiliation on their Internet donations. Nonprofit and Voluntary Sector Quarterly, 42, 1197-1223.

39.

Ren

C. R.

Guo

(2011). Middle managers’ strategic role in the corporate entrepreneurial process: Attention-based effects. Journal of Management, 37, 1586-1610.

40.

Saxton

G. D.

Guo

(2014, November 20-22). Conceptualizing social media capital. Paper presented at the annual meeting of the Association for Research on Nonprofit Organizations and Voluntary Action, Denver, CO.

41.

Saxton

G. D.

Wang

(2014). The social network effect: The determinants of giving through social media. Nonprofit and Voluntary Sector Quarterly, 43, 850-868.

42.

Saxton

G. D.

Waters

R. D.

(2014). What do stakeholders like on Facebook? Examining public reactions to nonprofit organizations’ informational, promotional, and community-building messages. Journal of Public Relations Research, 26, 280-299.

43.

Simon

H. A.

(1971). Designing organizations for an information-rich world. In Greenberger

(Ed.), Computers, communications and the public interest (pp. 37-72). Baltimore, MD: Johns Hopkins University Press.

44.

Sproull

L. S.

(1984). The nature of managerial attention. Advances in Information Processing in Organizations, 1, 9-27.

45.

Tufekci

(2013). “Not this one”: Social movements, the attention economy, and microcelebrity networked activism. American Behavioral Scientist, 57, 848-870.

46.

Waters

R. D.

Burnett

Lamm

Lucas

(2009). Engaging stakeholders through social networking: How nonprofit organizations are using Facebook. Public Relations Review, 35, 102-106.

47.

Wattam

(2009). A voice for the people: The role of social media technologies in community empowerment and regeneration. In Proceedings from Salford Postgraduate Annual Research Conference (pp.417-427). Salford, UK: University of Salford.

48.

Webster

J. G.

(2011). The duality of media: A structurational theory of public attention. Communication Theory, 21, 43-66.

49.

Wood

B. D.

Vedlitz

(2007). Issue definition, information processing, and the politics of global warming. American Journal of Political Science, 51, 552-568.