<<

ASRXXX10.1177/0003122415598534American Sociological ReviewVasi et al. 5985342015

American Sociological Review 1 –26 “No Fracking Way!” © American Sociological Association 2015 DOI: 10.1177/0003122415598534 Documentary Film, Discursive http://asr.sagepub.com Opportunity, and Local Opposition against in the United States, 2010 to 2013

Ion Bogdan Vasi,a Edward T. Walker,b John S. Johnson,c and Hui Fen Tand

Abstract Recent scholarship highlights the importance of public discourse for the mobilization and impact of social movements, but it neglects how cultural products may shift discourse and thereby influence mobilization and political outcomes. This study investigates how activism against hydraulic fracturing (“fracking”) utilized cultural artifacts to influence public perceptions and effect change. A systematic analysis of Internet search data, social media postings, and newspaper articles allows us to identify how the documentary reshaped public discourse. We find that Gasland contributed not only to greater online searching about fracking, but also to increased social media chatter and heightened mass media coverage. Local screenings of Gasland contributed to anti-fracking mobilizations, which, in turn, affected the passage of local fracking moratoria in the Marcellus Shale states. These results have implications not only for understanding movement outcomes, but also for theory and research on media, the environment, and energy.

Keywords social movements, environment, hydraulic fracturing, social media, mass media

Social movements attempt to change public 2012). When social movements “do” cul- perceptions of their grievances not only by ture—not just consume culture—they possess organizing collective actions, but also by “an extraordinarily powerful mode for both using cultural products and artifacts. Move- ments may introduce new ideas through cul- a tural products such as music (Danaher 2010; University of Iowa bUniversity of California, Los-Angeles Eyerman and Jamison 1998; Roy 2010), cHarmony Institute books (Meyer and Rohlinger 2012), and films dCornell University (Andits 2013; Whiteman 2003). These cul- tural products do not mobilize masses and Corresponding Author: Ion Bogdan Vasi, Department of Sociology and generate government response by themselves: College of Business, University of Iowa, 120 they require intensive and organized efforts to Seashore Hall, Iowa City, IA, 52241 generate activism (Meyer and Rohlinger E-mail: [email protected] 2 American Sociological Review solidifying commitment to collective action documentary film. Documentary filmmaking and for helping collectivities achieve their is increasingly a central component of organ- goals” (Roy 2010:258). But although a num- ized, multimedia social action campaigns. We ber of studies of movement mobilization and argue that documentary film may represent a outcomes analyze the role of cultural prod- discursive opportunity available to social ucts, they primarily examine how issues that movement actors. Documentary films can be social movements address relate to cultural seen as discursive opportunities that generate content (Berezin 1994; Eyerman and Jamison public interest, discussion, problematization, 1995; Hanson 2008; Lipsitz 2000; McAdam and new political preferences around critical 1994; Steinberg 2004; Zolberg 1997). social issues, as we will explain in further Empirical research on social movements’ depth. Documentary films create new topics use of cultural products to alter public percep- of popular discourse, inspire political activ- tions has two important shortcomings. First, ism, and even influence policymaking. Some these studies do not examine the influence of pieces of legislation, in fact, are informally movement-related artifacts on the broader known by the name of the documentary that public sphere, especially public spheres that inspired them, such as the recent “Blackfish extend beyond mainstream media. Recent bill” in California that would ban Sea World studies point out that features of the broader from keeping orcas in captivity.1 There are political culture moderate movements’ likeli- now entire film production companies, such hood of generating social and political as entrepreneur Jeff Skoll’s Participant Media, change, as contentious collective actors must that focus primarily on creating documenta- also develop persuasive accounts that align ries to inspire social change. However, social with themes in public discourse (Bail 2012; scientists have yet to systematically explore Ferree 2003; Giugni et al. 2005; Koopmans how social movements use documentary 2004; Koopmans and Olzak 2004; McCam- films to leverage social change campaigns mon et al. 2008; McCammon et al. 2007). In (Karlin and Johnson 2011). In particular, it is this body of research, scholars draw attention not known if documentaries are effective to whether a movement’s claims-making tools for local mobilizations, and the extent to aligns with broader societal discursive oppor- which they may influence policymaking tunities, that is, ideas about what is sensible, either directly or indirectly. realistic, or legitimate in political culture Beyond social movement theories, under- (Koopmans and Statham 1999). These oppor- standing how activists use cultural products tunities may be stable or volatile and they to alter discourses and shape policymaking is may be narrowly selective or broadly inclu- important for theories of media and environ- sive and accepting of movement ideas; fur- mental sociology. Research on media focuses thermore, these opportunities may shape on the processes of gathering and disseminat- movement actors’ decisions to tailor their ing news in mass media (Gans 1979; Gitlin frames to resonant themes (McCammon et al. 1980; Tuchman 1978) or through social media 2007). These studies have widened the lens of (Bennett and Segerberg 2012; Hussain and research on social movement outcomes, but Howard 2013; Tufekci and Wilson 2012), but they have not examined how activists exploit not in both simultaneously. Moreover, this cultural products as discursive opportunities, research primarily examines the role of dis- particularly outside the realm selectively cov- ruptive events and political contexts on news- ered by the mass media. paper coverage (see, e.g., Amenta et al. 2009), A second shortcoming in research on but not the role of artifacts such as documen- movements’ use of cultural products is that it taries.2 At the same time, research on environ- does not explain systematically how artifacts mental sociology shows that, even though are used to mobilize sympathizers and affect many communities are plagued by environ- change. A particularly important, yet insuffi- mental problems, the perception or public ciently researched, type of artifact is the expression of grievances is frequently lacking Vasi et al. 3

(Aronoff and Gunter 1994; Crenson 1971; socially constructed and politically contested Gaventa 1982; Roscigno 2011). Yet, research (Freudenburg and Jones 1991; Leiserowitz has not explored how activists use cultural 2004; Pidgeon, Kasperson, and Slovic 2003; products to overcome quiescence. Further- Vasi and King 2012). more, while research shows that activism may We begin by providing background on con- change perceptions of environmental risks tention surrounding hydraulic fracturing and (Vasi and King 2012), it does not examine the value of this research site for investigating how activism can shape perceptions of risks, our theoretical questions. We then discuss the mobilize the public, and affect policy change. importance of activist documentaries and dis- We argue that by scrutinizing activism sur- cursive opportunities. After laying out our rounding documentaries, we can identify expectations for the study, we offer background changes in perceptions of environmental risks on the rich online and social media data that we in mass and social media spheres, and we use to supplement our data sources on media explore how these changes lead to public discourse, fracking activity, and other meas- claims-making and local political change. ures. We then describe results from our investi- We undertake an analysis of the strategy gations at both national and local levels. At the and effects of activist documentary through a national level, we examine the role of activism detailed investigation of social movement surrounding Gasland for the emergence of dis- efforts to halt the controversial natural gas cursive opportunities in social and mass media. extraction method known as hydraulic frac- At the local level, we examine the use of Gas- turing, or “fracking.” We use a detailed con- land as an organizing tool and assess its contri- ceptualization and measurement of the shifts bution to mobilizations that led to municipal in public discourse that follow from such fracking bans in the Marcellus Shale region of opportunities, which is made possible by the the eastern United States. relatively recent availability of social media data. This allows us to understand these pro- cesses starting with public attention to a topic HYDRAULIC FRACTURING (Internet searches), moving to the overall AND GASLAND volume of chatter about it (on social media Hydraulic fracturing—also known as “frack- sites), and, in turn, the generation of diagnos- ing”—is a method of extracting natural gas tic (problem identification and blame attribu- that goes back to the 1940s, although it was in tion) and prognostic (suggesting remedies only limited use until recent years. Tradition- through online chatter) frames (see Benford ally, the method involved pumping large and Snow 2000). In the same fashion, we also amounts of pressurized water—mixed with track discourses found in conventional media various chemicals—into porous sandstone or reports, as done by previous analysts. How- limestone to release natural gas. Starting in ever, we believe our richer investigations into the early 2000s, drilling companies devel- the dynamics of the contemporary public oped unconventional methods to harvest nat- sphere are valuable for understanding the true ural gas trapped in shale rock. These methods importance of movements’ alignment with involve drilling horizontal wells that extend the rhythms of mass public (rather than only from their vertical well shafts along relatively mass media) discourse. As we will explain, thin, horizontal shale layers. Applying the this strategy helps us address key debates in hydraulic fracturing process to a horizontal the sociologies of media and the environ- well requires much larger volumes of water ment; the former in terms of how the new mixed with sand and chemicals under greater social media landscape shapes public dis- pressure than conventional methods, and this course and activism around critical issues poses engineering challenges and increases (Bennett and Segerberg 2012; Chadwick the potential for environmental pollution. 2013; Kreiss, Finn, and Turner 2011), and the Opposition to hydraulic fracturing is moti- latter in terms of how environmental risks are vated primarily by local environmental and 4 American Sociological Review health concerns. The environmental risks of “This [raising awareness] happened all over fracking, previously off the public agenda as the place. . . . It would take quite a bit of time an unknown practice, are now being vigor- to document all of the impacts, but I would ously debated (Osborn et al. 2011; Wilber say there were dozens of examples of this in 2012). To date, no conclusive research exists specific cases and the cumulative impact was to prove that hydraulic fracturing is danger- a big factor.” ous. On one side, natural gas industry groups This documentary’s impact on the move- and supporters argue that the risk of water ment can be assessed at national and local contamination is overblown. For example, the levels. First, if Gasland had a nationwide Independent Oil and Gas Association of New effect on the public debate about fracking, York argues that the chemicals used in frack- then this effect should be strongest when the ing are safe and are made of “a small amount film was released nationally on HBO and of dilute, benign additives found in common when it was nominated for an Academy household products” (Wilber 2012:117). On Award.6 Second, if Gasland had an effect on the other side, environmental activists claim local areas, this effect should be strongest in the industry is misleading the public by communities where the movie was screened, downplaying the risk of water contamination and soon after it was screened.7 Given that and other environmental problems, and civil multiple issues compete for public attention society groups are monitoring water quality at any given moment, it is extremely difficult for evidence of harms (Jalbert, Kinchy, and to maintain high levels of awareness about Perry 2014). Some grassroots groups, such as new issues. Any discernible effects of films Marcellus Protest, claim “there are almost as and other artifacts on activism are thus likely many violations as there are wells. Numerous to last for a relatively short period. incidents of spills, contamination and blow outs have been documented.”3 Awareness about fracking risks increased DISCURSIVE OPPORTUNITIES: rapidly in 2010, when the documentary Gas- TOWARD A RICHER land—directed by filmmaker —was UNDERSTANDING OF PUBLIC released. Gasland’s extraordinary potential DISCOURSE for raising awareness about the environmen- tal risks associated with hydraulic fracturing Most research on framing processes examines comes primarily from a few scenes in which how collective action frames contribute to Fox visits residents living near natural gas recruitment and collective identities (for a wells and witnesses them light their (appar- review, see Benford and Snow 2000), but a ently methane-contaminated) tap water on number of recent studies have generated criti- fire. Similar to how pictures of the burning cal insights into how frames can be linked Cuyahoga River energized the 1960s environ- directly to social movement outcomes (Bail mental movement, powerful images of burn- 2012; Cress and Snow 2000; McCammon ing water from Gasland energized activists. et al. 2008; McCammon et al. 2007). In so Film critics immediately recognized that Gas- doing, these studies draw attention to the land could be used as a mobilizing tool: one ways contentious collective actors are subject critic wrote in early 2010 that “Gasland may to features of the broader public discourse become to the dangers of natural gas drilling (Steinberg 1999). These interests have con- what Silent Spring was to DDT.”4 Activists gealed into a focus on how social movement also recognized that Gasland had a major frames, to have policy effectiveness, must not impact. As one organizer put it, “Josh Fox’s only resonate with potential activists but must Gasland blew the doors off what was happen- also connect with broader themes in public ing, raising awareness a tremendous amount.”5 discourse. When a movement’s ideas are con- In a personal communication, Fox told us, gruent with these broader public discourses, Vasi et al. 5 actors benefit from “discursive opportuni- participants (Chadwick 2013; Kreiss et al. ties,” or moments when movement ideas 2011). Furthermore, activist groups often align with what the culture at large deems to communicate directly through these services, be sensible, realistic, or legitimate (Koop- such that SMOs become less important as mans and Statham 1999; McCammon et al. spokespeople (Bennett 2003); this leads to 2007). what some call a “personalization” of collec- A limitation of these studies, however, is tive action (Bennett and Segerberg 2013). that their conceptualization and measurement Twitter, for instance, is now widely recog- of public discourse often privileges discourses nized as an effective mechanism on its own represented in the mass media without suffi- for generating and amplifying claims-making ciently investigating mass discourse more among collective actors (Bennett and broadly. Understanding the differences Segerberg 2012; Hussain and Howard 2013; between coverage of critical social issues in Tufekci and Wilson 2012). Because social the mass media versus social media offers media provides the capacity to broadcast ideas insight into a number of key questions in the and to organize directly through online means sociology of the media (for reviews, see (Shirky 2008), coverage of the risks posed by Schudson 1989, 2002). Most directly, meth- technological practices such as hydraulic frac- ods of news-gathering and dissemination are turing may be more substantial and more seen by media sociologists as critical to the negative in social media than in mass media. content of coverage on a topic (e.g., Gans Another important omission of media and 1979; Gitlin 1980; Tuchman 1978). The social movement studies is an analysis of bureaucratic routines of traditional news- changing discursive opportunities. We explore gathering, for instance, encourage a reliance the complex nature of movements’ discursive on official sources and a general deference to opportunities ranging from general to specific: official statements (Fishman 1980). When (1) initial online public attention to an issue; covering contentious politics or protest (2) evolution of social media chatter about, events, this should mean that events involving and mass media coverage of, a topic; (3) iden- social movement organizations (SMOs) will tification of problems associated with a topic gain greater coverage than those that do not in social and mass media (analogous to diag- (Earl and Kimport 2011); also, the disruptive- nostic framing); and (4) discussion in social ness, resource mobilization, and policy envi- and mass media of what can be done to remedy ronment surrounding protest groups shape the issue (analogous to prognostic framing).8 prominent newspaper coverage (Amenta et al. Related to the argument above, existing 2009). When covering events associated with research on mass media and discursive oppor- controversial or environmentally risky pro- tunities does not consider in sufficient depth jects, mass media act as either social amplifi- how movement-related artifacts serve as key cation or social attenuation stations (Flynn, sources of changing discursive opportuni- Slovic, and Kunreuther 2001; Pidgeon et al. ties.9 This is an important omission because, 2003). It is also well known that the predomi- as research on political opportunity structures nant norm of conventional news reporting is shows, social movements can sometimes cre- to seek out a balanced perspective in report- ate opportunities for themselves and for other ing (Benson 2004; Gamson and Modigliani movements at later points in time. Indeed, as 1989), even if this means including a certain Kriesi (2004:79) notes, “episodes of conten- amount of false equivalence. tious interaction are likely to modify the rel- Contemporary social media sources like evant configuration of actors and, thus, to Twitter and do not share these fea- change the specific opportunities for future tures, partly because the work of reporting options for collective action.” on events of the day is distributed, non- We argue that social movements use cul- bureaucratic, and mainly done by lay tural artifacts to influence public debates and 6 American Sociological Review capitalize on discursive opportunities in mul- people together through poetry reading or tiple public spheres. Specifically, we theorize singing can solidify commitment and mobi- that discursive opportunities are made possi- lize actors toward achieving goals (Roy ble by the national release of documentaries 2010). and their nomination for major awards. In this Looking at films in particular, social scien- case, Gasland is thought to have had a large tists suggest that activist cinema can have impact on building the movement because it substantial effects, but there is only limited identified water pollution as a serious prob- empirical evidence for these arguments to lem and called for organizing, helping to date. For example, Whiteman (2003) argues counter dominant frames about the safety of that the 1992 documentary From the Ground natural gas drilling. Up, which examined the environmental impacts of mining in Wisconsin, fostered Hypothesis 1: Gasland’s release and award local opposition against mining in Wisconsin nominations increased (1) online public through video screenings in diverse venues: attention toward fracking; (2) social me- bars, county fairs, churches, and sporting dia chatter and mass media coverage; (3) group meetings. Leiserowitz (2004) uses sur- discussion of water problems in social and vey data to show that individuals who watched mass media; and (4) discussion of bans and The Day After Tomorrow, which depicts fic- moratoria in social and mass media. tional disasters spawned by climate change, were more likely to perceive global warming CULTURAL PRODUCTS, as a threat and more willing to take action. MOBILIZATION, AND Jacobsen (2011) conducts what is arguably POLITICAL OUTCOMES the most rigorous analysis of activist cinema, exploiting spatial variation in the release of Analysts of contentious politics argue that former vice-president Al Gore’s climate movements use cultural artifacts to advance change documentary An Inconvenient Truth, their cause and contribute to positive social and shows that the documentary caused indi- change. Movement participants introduce viduals to purchase carbon offsets. A variety new ideas through such products as films of other accounts show that films and docu- (Andits 2013; Whiteman 2003), music (Dana- mentary programming may influence social her 2010; Eyerman and Jamison 1998; processes.10 Roscigno and Danaher 2001; Roy 2010), and Public screenings of documentaries are a books (Meyer and Rohlinger 2012). Existing more moderate protest form, consistent with research, however, focuses primarily on other information-distribution strategies such social movements’ impact on the broader cul- as teach-ins, tabling, leafleting, and holding tural context (Berezin 1994; Eyerman and press conferences to educate the public.11 But Jamison 1995; Hanson 2008; Lipsitz 2000; the moderate nature of screenings need not McAdam 1994; Steinberg 2004; Zolberg imply they will have little impact. Documen- 1997). For example, studies show that aboli- taries can be used to overcome quiescence, or tionists introduced African American spiritu- the absence of grievance perception in the als into white mainstream culture (Cruz face of significant inequality. The fact that 1999); leftist artists had a major influence on individual grievances about environmental American painting, sculpture, literature, and problems may exist in a community where theater (Hemingway 2002); and the women’s polluting industries are located does not mean movement of the 1960s and 1970s shaped grievances will be expressed collectively American poetry (Reed 2005). What is fre- (Crenson 1971; Gaventa 1982; Roscigno quently underappreciated in these accounts is 2011). Numerous environmental justice stud- an analysis of how social movements “do ies show that residents “endure years of culture” and its consequences. This is conse- uncertainty in the face of contradictory evi- quential, because movements that bring dence on the health and environmental risks Vasi et al. 7 posed by contaminants” (Aronoff and Gunter harming your community.” By the end of 1994:243). Because scientists and officials November 2013, more than 183,900 actions often cannot provide definitive answers about were reportedly taken through this website. risks, ambiguity and conflicting interpreta- Like the proverbial tree that falls in the for- tions frequently emerge (Auyero and Swistun est, Gasland screenings may have no effect 2008; Brown, Kroll-Smith, and Gunter 2000; because activists are not present. We argue that Brown and Mikkelsen 1990; Elliott and Fric- this is unlikely: activists are probably involved kel 2013; Fowlkes and Miller 1982; Kroll- in organizing local screenings, because the Smith and Couch 1993; Kroll-Smith, Couch, documentary was not shown in large commer- and Levine 2002). Even when information is cial cinemas. Moreover, even if anti-fracking made available, it may be presented using activists did not organize the screenings, they scientific jargon that is not accessible to many likely knew about the screenings and promoted citizens (Freudenburg and Jones 1991). them in their local communities. Therefore, we Moreover, in an unequal political system, citi- argue that Gasland screenings influenced local zens may distrust information presented by mobilizations against fracking, but we expect experts (Cable 2012). Documentaries created the effect of these screenings was relatively by non-experts may present facts in a more short term, that is, it lasted only a few weeks or accessible manner and overcome such prob- months. At any given moment, multiple issues lems of mistrust. compete for public attention (Baumgartner and In the case of Gasland, filmmaker Josh Jones 1993), which makes it extremely diffi- Fox attempted to alter public perceptions of cult to maintain high levels of public interest in social reality and victimhood by showing new issues and to organize long-term collec- evidence that fracking can contaminate drink- tive actions. Indeed, studies of local opposition ing water, proximity to hydraulic fracturing against energy projects find that not all at-risk wells puts public health at risk, and the natu- communities experience mobilizations and, ral gas industry takes advantage of landown- when they do occur, local mobilizations are ers who sign leases—which generally come usually short lived (McAdam and Boudet with non-disclosure agreements (NDAs), a 2012). topic we will return to—without fully under- standing the consequences. Gasland also Hypothesis 2: Local screenings of Gasland will legitimizes anti-fracking activism, prescribes have a short-term effect on local mobiliza- courses of action, and calls for mobilizations: tions against fracking. the filmmaker argues that local opposition is not only legitimate but also necessary, given We also expect that screenings influenced that natural gas drilling is exempt from the the adoption of anti-fracking policies in a num- Safe Act and government ber of ways. First, screenings likely affect the regulators appear unwilling to protect affected adoption of local bans because they increase communities. Additionally, Gasland has sus- the number of mobilizations. Communities tained the movement against fracking through with more mobilizations will likely have a screenings in numerous communities across higher level of commitment to the cause, even the country. Indeed, the documentary was if the number of participants in these events is instrumental in connecting individuals inter- not very large. Second, screenings of Gasland ested in joining the opposition against frack- are likely to attract new activists and sympa- ing. The film’s website has links that allow thizers. According to Josh Fox, screenings people to share information through Twitter, frequently involved hundreds of people, and Facebook, and YouTube, as well as links to many viewers likely became sympathetic to contact elected officials and participate in the cause after seeing the documentary. Indeed, local events. The website tracks participation after screenings many people expressed out- in actions to “help prevent gas drilling from rage and their determination to “spread the 8 American Sociological Review word,” as evidenced in statements such as: “I exist when relevant cultural products become had no idea,” “I can’t believe they’re getting important topics of social media discussion. away with this,” and “I’m going to tell every- Of particular importance is the Twittersphere, one I know to see this film.”12 Third, commu- or the postings made on the social media nities with screenings likely have a local website Twitter, which provides a space political context that is favorable to activism where people often share opinions and dis- due to the presence of influential allies. For seminate political information (Gerbaudo example, councilmembers from cities such as 2012; Howard and Hussain 2011; Papacha- New York and Pittsburgh organized screen- rissi and Oliveira 2012; Vasi and Suh 2012). ings.13 The effect of screenings on local poli- Moreover, unlike Facebook and other social cies can be direct or indirect. media platforms, Twitter facilitates the down- loading of large volumes of messages on a Hypothesis 3: Local screenings of Gasland will certain topic. have a direct effect on the adoption of frack- Next, we examine how the documentary ing bans. may have influenced local mobilizations and policymaking. We focus on cities within the Hypothesis 4: Local screenings of Gasland will Marcellus Shale states (Pennsylvania, Ohio, have an indirect effect on the adoption of New York, and West Virginia). The Marcellus fracking bans, by contributing to local mo- bilizations against fracking. Shale is the largest shale formation in the United States (Kargbo, Wilhelm, and Camp- bell 2010), it currently generates more than RESEARCH DESIGN twice as much natural gas as the next largest formation (Eagle Ford in Texas), and the area We use two separate sets of analyses. First, was the primary focus of Gasland. Despite its we explore how a documentary represents the incredible potential for natural gas production, opening of discursive opportunities nation- fracking in the Marcellus is a relatively new wide. Several theoretical and methodological phenomenon in the past decade. This forma- concerns motivate our choice of examining tion has also had a higher growth rate than all discourses in what we conceptualize as dis- other formations, increasing nearly 13-fold tinct public spheres. Like previous scholars, between January 2007 and March 2015.14 we are attentive to the public discourses Drilling in the Marcellus Shale has grown reported in the media and their particular rapidly, and this activity has apparently served influences on the political impacts of move- as a “suddenly imposed grievance” (Walsh ment action. Indeed, a primary way that 1981) for residents. Given the vast gas reserves movements “enter the public sphere” is in this shale play, the prospect of long-term through appearing in media reports (Oliver societal changes due to fracking is a signifi- and Myers 1999). Yet, movements are also cant component of public discourse surround- enmeshed in broader public discourses that ing drilling in the Marcellus. extend well beyond traditional media. Move- Also, in models reported in the online sup- ments may find that discursive opportunities plement for this article, we took additional also exist when cultural products on themes steps to account for the endogenous selection related to the movement become popular of Gasland screenings, examining differences online. Indeed, recent research shows that between matched cities with and without online public attention to a social move- screenings. ment—as measured by Internet searches for In each of our models of local mobiliza- terms associated with the movement—is a tions and fracking bans we focus on the city useful predictor of movement emergence level because the processes we wish to explain (Vasi and Suh 2012). Additionally, move- occurred primarily within cities; it is reason- ments may find that discursive opportunities able to assume that activists’ city-level Vasi et al. 9 organizational activities were more closely We used a parsing algorithm to remove spam associated with a city’s politics than with a and to include Twitter messages that mention state’s politics. To date, only New York and “fracking” and are related to the topic of Vermont have adopted a state-level morato- interest: hydraulic fracturing.20 We measure rium or freeze on hydraulic fracturing; more chatter as the monthly number of Twitter than 240 cities have adopted local bans in the messages that mention “fracking” between Marcellus Shale region. Our dataset includes January 2010 and May 2013, relative to the all cities in the Marcellus Shale states—a total number of Twitter messages in a month.21 total of 3,322 cities—observed between July We also measure discussion of potential water 29, 2010, and May 31, 2013; a total of 1,038 pollution, because this was the most impor- city-day observations.15 The most fine- tant problem discussed in Gasland and indi- grained measures were available at the day cates the formation of a diagnostic frame. We level; city-days are thus the appropriate units conducted an initial analysis of the context in of observation. The rich history of local activ- which water is mentioned, and we found that ism and the decentralized nature of politics in the word “water” is most frequently men- the United States present us with another tioned in association with words such as motivation for choosing the city-level. “contamination,” “pollution,” and “chemi- cals”; therefore, mentions of water indicate the discussion is focused on a range of water- Dependent Variables related problems. Using all Twitter messages Our analyses use four types of dependent that mention fracking, we conducted an auto- variables: social media discourse, mass media mated content analysis for mentions of the discourse, local mobilizations, and local bans. word “water” using WordStat. We measure Because we are interested in the complex discussion of potential solutions (prognostic nature of the opportunities that followed from frames) for water contamination problems by this documentary film—ranging from public conducting an automated content analysis in attention to an issue, to social media chatter WordStat for mentions of the words “ban” (or about the topic, and contentious discussions bans) or “moratorium” (or moratoria). Again, of remedies—we investigate both social the search was conducted for all Twitter mes- media interest (search) and social media chat- sages that mention fracking. ter. We choose the slang term “fracking” We operationalize mass media discourse rather than the industry preferred “hydraulic on fracking through newspaper coverage and fracturing” because it is a more popular term discussion. We measure mass media coverage in the public discourse.16 using a Lexis-Nexis search of all U.S. news- We measure public attention using Internet papers that mentioned “fracking” between and YouTube searches through the Google January 2009 and May 2013.22 Using Word search engine.17 In Google Trends, we con- Stat, we conducted an automated content ducted simultaneous Web and YouTube analysis of all newspaper articles that men- searches for “fracking” and “Gasland ” tion fracking for mentions of the words between January 2009 and May 2013.18 We “water,” “ban,” and “moratorium.” included simultaneous but separate searches We measure local mobilizations against for “fracking” and “Gasland ”—not “fracking hydraulic fracturing using data from the web- and Gasland ”—to obtain the percent of site MarcellusProtest.org, a grassroots organi- searches for “fracking” relative to the percent zation that acts as an information clearinghouse of searches for “Gasland ” by month. This about Marcellus Shale gas drilling and activ- allows us to compare trends in searches for ism. This website contains a calendar of all fracking with trends in searching for Gasland. anti-fracking events planned after July 29, We measure chatter and discussion using 2010, in the Marcellus Shale region. We cre- data from the social media platform Twitter.19 ated a web scraping program in Python, an 10 American Sociological Review open source programing language, to obtain it is unlikely that the effect of local screenings information about all mobilizations. This var- can be detected in the Google or Twitter data iable takes the value 1 for the day when a (which is aggregated at the national level) mobilization took place in a city, and 0 for the because, although the HBO release has other days.23 We measure the adoption of reached millions of viewers, each screening anti-fracking municipal ordinances in the has reached at most a few hundred viewers.27 Marcellus Shale region using information We also include a measure of the overall from Food & Water Watch, an organization volume of fracking-related Twitter chatter in that “works to ensure the food, water and fish a given city-day. We coded all Twitter users we consume is safe, accessible and sustaina- according to the location they entered in the bly produced.”24 We found the dates for most “location” field associated with a user’s Twit- municipal bans in documents posted on Food ter handle at the time each respective tweet & Water Watch, but some dates were missing; was posted. We first excluded all tweets not in these cases, we obtained dates by searching issued from the eastern time zone, because all municipalities’ websites and contacting four Marcellus states are in that location; offices of city clerks.25 The ban variable was among this subset, 87 percent of tweets were coded 1 for the day when a city adopted a ban associated with a user with an entry in the and 0 for the other days, with observations location field. We then coded these entries by after the ban was passed right-censored. city and state, finding that approximately 40 percent of tweets with an entry in the location field were posted by a user account located in Independent Variables New York, Pennsylvania, Ohio, or West Vir- Regarding the documentary, we utilize data ginia. We then converted these tweets into a about local Gasland screenings. We obtained city-day sum total of fracking-related tweets data from Josh Fox about all screenings of the per 10,000 people, logged the measure, and documentary in the United States. These lagged it by one day.28 screenings were organized by individuals and We control for the adoption of municipal nonprofits in local libraries, cinemas, schools, ordinances and population size, because and other public spaces. In our models, we larger cities are more likely to have a critical examine the effects of Gasland screenings mass of citizens who mobilize against frack- using measures of the effects of screenings in ing. We measure population size using data the short to medium term (i.e., from a few from the 2010 U.S. Census, and we apply the weeks to a few months post-screening) and natural logarithm to correct for skewness. We within a close radius (five miles between the control for the dominant political ideology, city where a screening took place and other because environmental campaigns are mainly cities).26 Some screenings were initiated by associated with left-of-center politics (Brulle Fox and others affiliated with the film; other 2000). We measure left-of-center political screenings were initiated by local community ideology using the percentage of votes for members who sought a copy to screen at a Barack Obama, at the county level, during the local theater or other public venue (Josh Fox, 2008 presidential election. We control for personal communication, March 17, 2014). In unemployment, because cities with high lev- a personal communication, Fox said he aimed els of unemployment are less likely to oppose to inspire local activists to initiate local frack- drilling for natural gas. Indeed, the natural ing bans and to give communities the tools gas industry frequently touts hydraulic frac- they would need to go forward. These local turing as beneficial for local economies. We screenings were carried out on a variety of also control for income, because cities with dates starting in August 2010, thus making it high-income residents are more likely to difficult to see the effect of these screenings organize Not In My Back Yard (NIMBY) on Google searches or Twitter chatter. Indeed, campaigns (Bullard 2000). We measure Vasi et al. 11 unemployment rate and income using data include dummy variables for New York, from the U.S. Census Bureau about the unem- Ohio, and Pennsylvania (West Virginia is the ployment rate and median income in 2010. reference category). We include a measure of We control for the presence of the oil and environmental organizations, because com- gas industry using data from the Bureau of munities with a strong environmental move- Labor Statistics about the total number of ment infrastructure are likely aware of people employed in the oil and gas extraction hydraulic fracturing’s environmental impacts industry in a year, by county. We standardized and possess resources for mobilizing against this measure per 10,000 and lagged it by one this practice.34 We also control for the density year to avoid simultaneity bias. We include as of overall nonprofit organizations as a meas- a control whether a city is located directly ure of general civic and political capacity of above the Marcellus. We expect cities located communities, consistent with the finding in directly above the Marcellus Shale will per- many studies that high general civic capacity ceive a greater risk from drilling than cities represents a powerful force to counterbalance located elsewhere.29 Data about Marcellus industry interests in the siting of environmen- Shale locations come from maps developed tal bads (Aldrich 2008; McAdam and Boudet by state agencies.30 2012; McAdam et al. 2010). All data come We include measures of proximity to natu- from the National Center for Charitable Sta- ral gas wells because this might (1) increase tistics (NCCS): the environmental groups citizens’ grievances and lead to mobilizations include all nonprofits categorized in NTEE and the adoption of bans, or (2) depress activ- section C (environmental), and the general ism because individuals who live very close civic measure includes all nonprofits in the to wells have likely leased their land and NCCS overall.35 signed the associated non-disclosure agree- Finally, we generated a diffusion variable ments (NDAs) that prevent their activism, or to examine whether previous adoptions of they benefit financially from the industry. We bans in geographically proximate cities influ- obtained data on permits from each state’s ence the passage of local bans. We experi- department of natural resources/environmen- mented with different radii for proximity (e.g., tal protection.31 Using these data, we calcu- 30, 60, or 100 miles) as well as different func- lated the sum of active hydraulic fracturing tions (cumulative, an inverse square root of wells within a specified radius from each city distance, and various decay functions) that in the previous year. We experimented with model how the influence of proximate cities various radii, but for simplicity we include decreases with geographic distance.36 Alterna- only a five-mile radius.32 To calculate the sum tive specifications produced similar results, so of all wells we used GeoNear, a Stata module we describe results from the inverse square that allows the calculation of distance-based root of distance measure with a 100-mile variables; these are log-transformed. radius, because this functional form was used We also include a measure of water con- in previous research on diffusion (Andrews tamination events due to fracking, and here and Biggs 2006; Hedström, Sandell, and Stern we use the same five-mile radius as we do for 2000). Table 1 shows the means, standard proximity to wells. To create this measure, we deviations, and correlations for all variables. relied on state disclosures on fracking regula- tory violations from Pennsylvania and Ohio, Estimation Techniques combined with an MIT research team’s pri- mary systematic report on all fracking viola- Because the dependent variables are the day a tions to date (Moniz et al. 2011) and national mobilization occurred in a city and the day a reports on fracking-related water contamina- ban against fracking was adopted, hazard tion from Earthjustice.33 We control for the models are appropriate (Bennett 1999; Box- state in question because each state has a Steffensmeier and Jones 1997, 2004). We use specific regulatory and economic context. We a single-events Cox proportional hazards 1 17

1 16 –.00 1 15 .01 –.01 1 14 .14 .09 –.01 1 13 .35 .07 .05 –.02 1 12 .01 .31 .04 –.01 –.01 1 11 .07 –.46 –.08 –.05 –.14 –.02 1 10 .09 .05 .03 –.40 –.46 –.14 –.10 9 1 .04 .00 .30 .00 –.10 –.06 –.01 –.00 8 1 .11 .49 .03 .03 .24 .06 .02 –.15 –.37 7 1 .39 .00 –.28 –.18 –.15 –.01 –.01 –.00 –.12 –.10 6 1 .66 .04 –.47 –.12 –.04 –.31 –.27 –.05 –.03 –.07 –.01 5 1 .26 .30 .01 .01 –.26 –.05 –.14 –.11 –.06 –.04 –.10 –.01 4 1 .36 .30 .17 .00 –.19 –.16 –.08 –.01 –.09 –.11 –.06 –.10 –.03 3 1 .10 .09 .05 .03 .01 .00 –.04 –.00 –.01 –.03 –.02 –.01 –.02 –.01 2 1 .72 .08 .07 .04 .02 .01 .00 –.03 –.00 –.00 –.02 –.01 –.01 –.01 –.01 1 1 .02 .02 .01 .01 .01 .00 .00 .00 –.01 –.00 –.00 –.00 –.00 –.00 –.00 –.00 SD .01 .09 .14 .08 .49 .45 .45 .47 .34 .61 1.43 1.97 3.08 74.14 11.14 83.34 454.2 .00 .01 .01 .48 .47 .29 .28 .34 .05 .10 Mean 7.54 8.86 2.16 1.78 4.51 168.0 106.6 same day (5m) 2 months (5m) 4 months (10m) . Means, Standard Deviations, and Correlations Gasland Gasland Gasland 1. 2. 3. 4. Population (ln) 5. Votes for Obama 5. Votes 6. Unemployment 7. Income (ln) 8. Oil and gas industry 9. Marcellus location Table 1 Table 10. Location: NY 11. Location: OH 12. Location: PA 13. NGOs (all) 14. Environmental NGOs 15. Proximity to wells (5m) 16. Proximity to water c. (5m) 17. Twitter messages 17. Twitter

12 Vasi et al. 13

120 Web searches "fracking" Web searches "Gasland" YouTube searches "fracking" YouTube searches "Gasland" 100

80

60

40

20

0 Jul-10 Jul-11 Jul-12 Jan-10 Jan-11 Jan-12 Jan-13 Jun-10 Jun-11 Jun-12 Oct-10 Oct-11 Oct-12 Apr-10 Apr-11 Apr-12 Apr-13 Feb-10 Sep-10 Feb-11 Sep-11 Feb-12 Sep-12 Feb-13 Dec-10 Dec-11 Dec-12 Aug-10 Aug-11 Aug-12 Nov-10 Nov-11 Nov-12 Mar-10 Mar-11 Mar-12 Mar-13 May-10 May-11 May-12

Figure 1. Gasland’s Influence on Online Public Attention Note: The figure shows the percent of Google and YouTube searches for “fracking” and for “Gasland” relative to the highest number of searches between January 2010 and May 2013 (HBO release on June 2010; Oscar award nomination on February 2011). model to estimate the adoption of bans against measured via Google web searches only for fracking because these events occur only “fracking,” and public attention toward frack- once; thus, once a city experiences a ban it is ing as measured via Google web searches dropped from the analysis for the remaining only for “Gasland.” As predicted, we find days.37 We use stratified Cox models with that online public attention toward fracking— repeated events for the analysis of mobiliza- as measured by Internet searches—increases tions because these events occur multiple immediately after Gasland’s release and times. Following Box-Steffensmeier and award nominations. Gasland contributes to Jones (2004), we use conditional gap time an increase in public attention toward frack- models because they offer the best solution ing in two ways: first, Internet searches for for dealing with events that have sequential “Gasland” increase when the documentary is risks. Therefore, the dependent variable in the released and nominated—we call this the case of anti-fracking events is reset to zero Gasland-specific public attention to hydraulic after each event. fracturing; second, Internet searches for “fracking” increase after the documentary is released and nominated for an Academy RESULTS Award—we refer to this as fracking-specific Changing Discursive Opportunities public attention. These two contributions are apparent in We begin by exploring trends in public atten- Figure 1. First, searches for the documentary tion toward fracking. Figure 1 displays trends peak in June 2010, when it is released on in overall public attention toward fracking as HBO, and in February 2011, when it is 14 American Sociological Review nominated for an Oscar award. Remarkably, ‘fracking,’ which never leaves the media after Internet searches for the documentary surpass that point.” searches for fracking in June and July 2010. Next, we explore trends in social media For most months, searches for Gasland chatter about fracking. We find that chatter in account for less than 5 percent of the highest the Twittersphere about fracking increases value of overall public attention to fracking, over time, particularly after Gasland is but in June 2010 they account for over 40 released on HBO and is nominated for an percent of the searches, in July 2010 they Oscar award. Figure 2 displays trends in social account for approximately 24 percent, and in media chatter about fracking as measured by February 2011 they account for about 20 per- Twitter messages about fracking or about cent. Second, Internet searches for fracking Gasland. The figure shows that the percentage increase immediately after the HBO release of Twitter messages about Gasland is larger and the Oscar nomination. When examining than the percentage of Twitter messages about searches for both fracking and Gasland, we fracking when Gasland was released, in June see a clear increase in searches for these terms 2010. While the percentage of Twitter mes- after the release and nomination. sages about Gasland is not larger than that of Figure 1 also displays trends in overall Twitter messages about fracking in February online public attention toward fracking as 2011, when Gasland is nominated for an measured by YouTube searches for “frack- Oscar, it is relatively close. The release and ing” and for “Gasland.” Again, we find that nomination events are accompanied by an overall public attention toward fracking—as increase in chatter not only about Gasland but measured by YouTube searches—increases also about fracking: chatter increases by over time, particularly after Gasland’s release approximately 6 percent after the release and and award nominations. Gasland contributes by approximately 9 percent after the nomina- to an increase in public attention toward tion (compared to the previous month). fracking in two ways. First, YouTube searches We now examine trends in mass media for “Gasland ” increase when the documen- coverage of fracking. Figure 3 shows the num- tary is released and nominated—the Gasland- ber of newspaper articles that mention “frack- specific online public attention to hydraulic ing” or “Gasland ” between 2010 and early fracturing. Figure 1 shows that YouTube 2013. As predicted, we find that mass media searches for the documentary peak in June coverage of fracking—as measured by news- 2010, when it is released on HBO, and in paper articles—increases over time, particu- February 2011, when it is nominated for an larly after Gasland’s release and award Oscar award. Interestingly, YouTube searches nominations. Gasland contributes to an for the documentary surpass overall searches increase in coverage of fracking in two ways: for fracking not only during these two critical first, coverage of Gasland increases when the months but throughout the entire period from documentary is released and nominated—we January 2010 until May 2011. This is most call this the Gasland-specific coverage of likely because YouTube is a video-sharing hydraulic fracturing. Remarkably, more than website; therefore, searches for a documen- half of the articles that mention fracking in tary are likely to be higher on YouTube than June 2010 and January 2011 also mention on the web. Second, YouTube searches for Gasland. Second, coverage of fracking “fracking” increase after the documentary is increases after the documentary is released released and nominated—the fracking- and nominated for an Academy Award—we specific public attention to hydraulic fractur- refer to this as the fracking-specific coverage. ing. Indeed, the findings presented in Figure Next, Figure 4 examines changes in social 1 are consistent with experiences described in and mass media discussion about fracking— the field. As Gasland creator Josh Fox told us specifically, how diagnostic and prognostic in a personal communication, “When Gas- frames emerged through a focus on water land hits, it changes the game on the word problems and on bans and moratoria Vasi et al. 15

0.35

Twier messages that menon "fracking"

0.3 Twier messages that menon "Gasland"

0.25

0.2

0.15

0.1

0.05

0 0 0 1 1 2 2 3 3 Jul-10 Jul-11 Jul-12 Jan-10 Jan-11 Jan-12 Jan-13 Jun-10 Jun-11 Jun-12 Oct-10 Oct-11 Oct-12 Apr-10 Apr-11 Apr-12 Apr-13 Feb-10 Sep-10 Feb-11 Sep-11 Feb-12 Sep-12 Feb-13 Dec-10 Dec-11 Dec-12 Aug-10 Aug-11 Aug-12 Nov-10 Nov-11 Nov-12 Mar-1 Mar-1 Mar-1 Mar-1 May-1 May-1 May-1 May-1

Figure 2. Gasland’s Influence on Twitter Chatter about Fracking Note: The figure shows the percent of messages that mention “fracking” and “Gasland” between January 2010 and May 2013 relative to the total number of messages (HBO release on June 2010; Oscar award nomination on February 2011).

400 Newspaper menons of "fracking" Newspaper menons of "Gasland" 350

300

250

200

150

100

50

0 0 0 1 1 2 2 3 Jul-10 Jul-11 Jul-12 Jan-10 Jan-11 Jan-12 Jan-13 Jun-10 Jun-11 Jun-12 Oct-10 Oct-11 Oct-12 Apr-10 Apr-11 Apr-12 Apr-13 Feb-10 Sep-10 Feb-11 Sep-11 Feb-12 Sep-12 Feb-13 Dec-10 Dec-11 Dec-12 Aug-10 Aug-11 Aug-12 Nov-10 Nov-11 Nov-12 Mar-1 Mar-1 Mar-1 Mar-1 May-1 May-1 May-1

Figure 3. Gasland’s Influence on Mass Media Coverage of Fracking Note: The figure shows the number of newspaper articles that mention “fracking” and “Gasland” between January 2010 and May 2013 (HBO release on June 2010; Oscar award nomination on February 2011). 16 American Sociological Review

Menons of "ban" & "moratorium" - Twier Menons of "ban" & "moratorium" - Newspapers

Menons of "water" - Twier Menons of "water" - Newspapers 3

2.5

2

1.5

1

0.5

0 3 0 0 1 1 2 2 Jul-10 Jul-11 Jul-12 Jan-13 Jan-10 Jan-11 Jan-12 Jun-10 Jun-11 Jun-12 Oct-12 Oct-10 Oct-11 Apr-13 Apr-10 Apr-11 Apr-12 Feb-13 Feb-10 Sep-10 Feb-11 Sep-11 Feb-12 Sep-12 Dec-12 Dec-10 Dec-11 Aug-10 Aug-11 Aug-12 Nov-12 Nov-10 Nov-11 Mar-1 Mar-1 Mar-1 Mar-1 May-1 May-1 May-1

Figure 4. Gasland’s Influence on Social and Mass Media Discussion about Fracking over Time Note: The figure shows mentions of the words “water” and “ban” and “moratorium” as a percent of the total number of words in Twitter messages and newspaper articles about fracking (HBO release on June 2010; Oscar award nomination on February 2011). immediately after Gasland’s release and that, overall, these words are used less fre- award nominations. Our automated content quently in mass media than in social media. analysis reveals that Twitter mentions of the Twitter mentions of bans and moratoria are word “water” increase after Gasland is highly variable over time, reflecting concen- released on HBO (from 1.2 percent in May trated periods of activity when particular 2010 to 1.6 percent in June 2010) and after it fracking bans were being proposed, debated, is nominated for an Oscar (from 1.5 percent and in some cases enacted. in January 2011 to 1.8 percent in February 2011). At the same time, newspaper mentions Local Mobilizations and of the word “water” decrease after Gasland is Policymaking publicly released (from 2 percent in May 2010 to 1.1 percent in June 2010) and after it Table 2 displays results of models that esti- is nominated for an Academy Award (from mate the effects of Gasland and covariates on 1.3 percent in January 2011 to 1.1 percent in mobilizations. Table 3 displays results of a set February 2011). Twitter mentions of the of models that estimate effects of the anti- words “ban” and “moratorium” increase after fracking events (and covariates) on the pas- Gasland is nominated for an Oscar (from .3 sage of local moratoria against fracking. The percent in January 2011 to .6 percent in Feb- models in Table 2 are from conditional gap ruary 2011) but do not change after Gasland time event history analysis models (clustered is released on HBO (.3 percent in May and on city and stratified by event number); in June 2010). Newspaper mentions of these Table 3, results are from single-event event words increase very little after the documen- history models. In both tables, models include tary is released (from 0 to .1 percent) but measures of proximity to permitted fracking decrease after it is nominated for an academy wells and proximity to water contamination award (from .2 to .1 percent). Figure 4 shows accidents (within a five-mile radius).38 Vasi et al. 17

Table 2. Effects of Gasland Screenings on Anti-fracking Mobilizations in the Marcellus Region States

Model 1 Model 2 Model 3

Gasland Screenings (five miles) Four months after, not including day of .171* screening (.075) Two months after, not including day of .375*** screening (.099) Same day of screening 2.669*** (.806) Controls Twitter chatter, total number of messages .052*** .053*** .053*** (.010) (.010) (.010) Population (ln) .940*** .935*** .938*** (.167) (.163) (.167) Votes for Obama –2.230* –2.304* –2.293* (.974) (.983) (.963) Unemployment –.228 –.236 –.234 (.123) (.126) (.123) Income (ln) –.001 –.001 –.001 (.001) (.001) (.001) Marcellus location 2.083*** 2.119*** 2.111*** (.379) (.392) (.378) Oil and gas industry .011* .011* .011* (.005) (.004) (.005) New York .569 .563 .570 (.511) (.514) (.513) Ohio 1.302* 1.354* 1.343* (.628) (.641) (.624) Pennsylvania 1.717*** 1.727*** 1.728*** (.529) (.531) (.528) NGOs (all) .154*** .153*** .154*** (.020) (.019) (.020) Environmental NGOs –.111 –.111 –.112 (.134) (.134) (.134) Proximity to wells (five miles) –.003** –.003** –.003** (.001) (.001) (.001) Proximity to water contamination (five miles) .018 .017 .017 (.043) (.043) (.043) N Observations 3,448,236 3,448,236 3,448,236 Chi Square 514*** 160*** 113***

Note: Conditional gap time models clustered on city and stratified by event number. *p < .05; **p < .01; ***p < .001 (two-tailed tests).

Results in Table 2 show that screenings of mobilizations in the same day. Results in Gasland in a given community lead to a sig- Models 2 and 3 show that the effect of Gas- nificant increase in anti-fracking events in the land does not last very long if we are not relatively short term, and this is true regard- including the day of the screening: for exam- less of how the model specifies proximity to ple, the effect is significant ( p < .001) when active wells or to wells that have contami- it is assumed to last two months, and it is nated water. Results in Model 1 show that weaker but still significant ( p < .05) when it local screenings are associated with local is assumed to last four months after the 18 American Sociological Review screening. Thus, the effect of Gasland screen- added. Therefore, we conclude that Hypothe- ings does not last much beyond four months.39 sis 3 is not supported but Hypothesis 4 is Results in Table 2 show that cities with supported. growing Twitter chatter about fracking have a As mentioned earlier, we also estimated greater likelihood of organizing events, an additional models (reported in the online sup- important finding that suggests local activists plement) investigating the effects of Gasland used Twitter as an organizing tool.40 Results screenings on local mobilizations using a in this table also show that population size matched set of cities with and without screen- and location in any region directly above the ings, given concerns of endogenous selection; Marcellus Shale (with or without existing the results add further support to the argument wells) increase the likelihood of organizing that screenings had a significant effect on events in a city. We also find a very strong mobilizations within the following two effect of the overall scale of the nonprofit sec- months.42 tor, suggesting that a city’s overall civic Another important finding is that geo- capacity represents a substantial influence. graphic proximity to previous adopters Indeed, our review of the field of organiza- increases the likelihood of adopting a ban, tions active in making anti-fracking claims which suggests activists are influenced by illustrates that the majority of these groups neighboring communities. These effects are are ad hoc coalitions, built from preexisting stronger if we assume that closer cities exert blocs of citizens, which have sprung up in a stronger influence than do distant cities.43 recent years to fight against this contested Unlike mobilizations, for moratoria we do not activity. The presence of environmental find significant effects of local Twitter chat- organizations and proximity to water con- ter. However, as with our findings on mobili- tamination sites are not significant influences zation patterns, general civic capacity was a on anti-fracking mobilizations.41 Proximity to factor in the passage of local moratoria; the wells and a greater share of Democratic vot- overall number of nonprofit organizations in ers decrease the likelihood of organizing a city was a significant factor in the passage events; both are surprising findings. How- of moratoria, whereas we find no effect of the ever, it is also clear that individuals who live presence of specific environmental nonprof- in close proximity to fracking often benefit its. As in the previous case, we expect that the materially from the activity or sign NDAs. population of nonprofit organizations repre- The finding on partisanship likely reflects the sented a significant resource on which anti- property-rights orientation of many conserva- fracking activists could build and represent tive landowners. Finally, local employment in themselves against the threat of a potentially the oil and gas industry increases the likeli- risky new practice. We also find that areas hood of mobilization—such workers are often with higher rates of unemployment were less a source of resentment in communities—as likely to pass fracking moratoria, confirming does location in Pennsylvania or Ohio. the expectation that economically vulnerable Moving to our findings on local moratoria, areas would not want to cut off a potential our primary finding in Table 3 is that the anti- source of investment. In contrast, areas with fracking mobilizations—which were, in part, greater population size and income are more motivated by screenings of Gasland—had a likely to pass fracking moratoria. significant influence on the passage of mora- toria against fracking. Model 2 shows that local Gasland screenings had a significant DISCUSSION AND ( p < .05) direct effect on the adoption of CONCLUSIONS bans; yet, this effect becomes nonsignificant This study makes a number of important con- at p < .05 (although it is significant at p < .10) tributions to the literatures on social move- in Model 3, when the mobilization variable is ment outcomes, media, the environment, and Vasi et al. 19

Table 3. Effect of Gasland Screenings and Anti-fracking Mobilizations on Anti-fracking Municipal Bans in the Marcellus Region

Model 1 Model 2 Model 3

Mobilizations and Screenings Anti-fracking events (ln) .765*** (.242) Gasland screening five miles (ln) .634* .511 (.266) (.262) Controls Proximity to previous adopters (100m; inverse sqr. .266*** .257*** .266*** dist.) (.048) (.048) (.049) Twitter chatter, total number of messages (ln) .035 .036 .024 (.034) (.034) (.036) Population (ln) .577*** .538*** .449*** (.077) (.078) (.083) Votes for Obama –.533 –1.533 –1.332 (1.428) (1.482) (1.480) Unemployment –.246* –.251* –.240* (.115) (.113) (.112) Income (ln) 1.448*** 1.457*** 1.398*** (.262) (.265) (.266) Marcellus location –.001 –.001 –.001 (.002) (.002) (.002) Oil and gas industry –.042 –.044 –.044 (.041) (.043) (.042) New York 1.408 1.436 1.256 (.841) (.842) (.844) Ohio 1.620 1.720 1.617 (.885) (.882) (.881) Pennsylvania –.445 –.487 –.944 (.862) (.862) (.887) NGOs (all) .101*** .101*** .093*** (.020) (.020) (.022) Environmental NGOs .005 .001 .008 (.229) (.224) (.238) Proximity to wells (five miles) –.002 –.002 –.002 (.002) (.002) (.002) Proximity to water contamination (five miles) .350 .351 .358 (.259) (.251) (.264) N Observations 3,401,471 3,401,471 3,401,471 Chi Square 220*** 225*** 233***

Note: Single-event Cox models. *p < .05; **p < .01; ***p < .001 (two-tailed tests). political change. Our findings make clear that growing body of work highlights the benefits cultural artifacts created by actors allied with that accrue to activists when they find “dis- social movements have both national and cursive opportunities” to make their claims local effects; in our case, they helped displace politically resonant, but earlier work does not expectations that gas drilling is safe with an examine in sufficient depth how discursive alternative frame of fracking risk. At the opportunities emerge and change. We demon- national level, we show that the documentary strate that Gasland’s nationwide release on Gasland created a discursive opportunity, HBO and nomination for an Oscar award particularly in social media. An important and contributed to the emergence of a more 20 American Sociological Review favorable set of discursive opportunities for documentary films are released each year, the anti-fracking movement. Specifically, and most will not generate very much discus- Gasland contributed not only to an increase sion or activism. Given such considerations, in online public attention to fracking, but also we believe our study is generalizable for to increased chatter about fracking on social understanding the impact of documentary media and to increased coverage of the topic films that gain widespread popular attention, in mass media. Moreover, Gasland influ- and future researchers could investigate simi- enced the social media discussion about lar effects for analogous films like Blackfish, fracking by supporting diagnostic frames An Inconvenient Truth, and The Cove. related to water pollution and, to a lesser At the local level, we show that screenings degree, supporting prognostic frames regard- of Gasland in different locations had an effect ing local bans and moratoria. Yet, Gasland’s on the mobilization of local campaigns influence on mass media’s discussion about against the controversial practice of hydraulic fracking was weaker: mass media discussion fracturing; in turn, these local mobilizations did not become more focused on water prob- made local policymakers significantly more lems or on bans and moratoria. likely to take action to ban the practice of Our study shows that the release of the fracking in cities across the Marcellus Shale Gasland documentary influenced public dis- states. Social movements try to use cultural course on fracking by highlighting environ- artifacts, such as documentaries, to build sup- mental risks and offering possible solutions. port for their cause and win policy changes. This influence was larger in the Twittersphere These cultural artifacts are not merely epiphe- than in the mass-mediasphere, in part because nomenal traces of movements’ organization- mass media coverage was more “balanced,” building efforts but are themselves key giving equal weight to critics and proponents mobilizing tools that help movements address of fracking.44 We argue that using only mass the problem of quiescence (Crenson 1971; media sources to understand discursive Gaventa 1982; Roscigno 2011) and counter opportunities may lead scholars to wrongly opponents. Other studies of social movement conclude in the direction of a “false nega- outcomes show the importance of cultural tive,” in which opportunities are overlooked. artifacts, but this study is among the first to Social media discourses may, in many cases, provide systematic evidence of their impact be much more aligned with movement claims on both mobilization and political outcomes. and preferences than what is represented in Taken together, these findings suggest that the newspaper or on television. Some of this Gasland opened up a new discursive opportu- divergence may be due to mass media’s bal- nity for the movement, increasing sympathetic ance norms (Boykoff and Boykoff 2004; social media chatter. That opportunity, in turn, Gans 1979; Tuchman 1978), or to elite public supported anti-fracking mobilization in local relations campaigns and other efforts by communities. We find that both Twitter chatter counter-movements to challenge ideas favora- and Gasland screenings had direct effects on ble to a social movement (Vasi 2011; Vasi and organized events. But when it came to actually King 2012; Walker 2009, 2014). passing local bans, neither Twitter chatter nor Readers might express concern that we the Gasland screenings were directly influen- have investigated the impact of a documen- tial. What mattered more were anti-fracking tary film that was perhaps unusual in its abil- mobilizations and diffusion processes. Discur- ity to generate social change. Indeed, we find sive opportunities were critically influential in it remarkable that Figures 1 and 2 both show setting the stage for influencing social move- periods when discussion and Internet search- ment outcomes. In this respect, our finding is ing of “Gasland ” surpassed overall discus- consistent with bodies of work on both politi- sion/searching for “fracking” (on Google, cal and discursive opportunities, which find YouTube, and Twitter). Hundreds of political that these effects are indirect. Vasi et al. 21

This study also makes clear the importance of Given the issue’s shift from a place of almost civic capacity for a community’s ability to complete novelty to such an established and respond to threatened environmental changes. contentious position over less than a decade, Other studies have found similar evidence fracking serves as an excellent example of regarding the siting of contentious facilities how movements grapple with contentious (Aldrich 2008; McAdam and Boudet 2012). energy issues and the environment. Here we find that communities’ capacities for mobilizing and changing policy were very much Acknowledgments influenced by the density of preexisting nonprof- The authors would like to thank Burton DeWilde, Doris its. In addition, the presence of local activists who Domoszlai, and Joanna Raczkiewicz at the Harmony Insti- were active on social media increased communi- tute (for data collection and research assistance), the Uni- ties’ capacity for mobilization. These factors mat- versity of Iowa Computing Cluster, the Urban Institute (for tered more than the density of environmental access to the NCCS data), and colloquium audiences at nonprofits, suggesting that anti-fracking activity Brown University, Northwestern University, Stanford University, University of Arizona, University of California- needed to build from the overall civic capacity of Irvine, UNC-Chapel Hill, University of Colorado-Boulder, an area. Of course, consistent with previous University of Colorado-Denver, University of Michigan, research, much of this NIMBY activism reflects University of Southern California, the West Coast Non- a more civically advantaged community’s ability profits Data Conference, the European Group for Organi- to resist unwanted land uses. zational Studies, and the 2014 ASA meetings. Additionally, we thank Chris Bail, Steve Barley, Neal Caren, Joe We encourage future researchers to inves- Galaskiewicz, Don Grant, Tanya Heikkila, Lori Hunter, tigate the effects of other documentaries, Colin Jerolmack, Michael Kennedy, Francesca Polletta, films, and consciousness-raising tactics. Woody Powell, Timmons Roberts, Sarah Soule, Patricia Given that most of these events do not take Strach, and Lori Yue for helpful feedback. place outdoors, on streets or public squares, but indoors, in cinemas, public libraries, or Notes college campuses, these events are less visi- 1. See http://www.usnews.com/news/articles/2014/ ble and, therefore, harder to study. However, 03/07/california-lawmaker-introduces-blackfish- further research is necessary to understand inspired-orca-captivity-bill. why only some documentaries lead to large- 2. It is important to distinguish between mass media as scale mobilizations or how these less visible a technological platform (newspapers, magazines, tactics contribute to social change. We also radio, television, and the Internet) and the artifacts (essays, plays, songs, films, and documentaries) encourage future researchers to investigate that are sometimes disseminated through these plat- the role of industry counter-mobilization forms in addition to news and other information. An through grassroots efforts and public relations artifact may have its own technological platform (a campaigns on behalf of natural gas interests, CD or MP3 format for a song; a DVD or BluRay as this may also affect community mobiliza- format for a film) but, to reach the majority of the public, it often needs a mass media platform as a tion processes, media discourse, and ultimate means of communication. However, artifacts can political outcomes. There is some initial evi- also be distributed through either formal or infor- dence of these influences. First, it is clear that mal activism: for example, samizdat literature was natural gas interests have been heavily mobi- distributed in the former communist countries not lized in favor of fracking, through campaigns through mass media but through informal networks of dissidents. organized by trade associations such as 3. See MarcellusProtest.org, accessed November 2013 America’s Natural Gas Alliance and third- (http://www.marcellusprotest.org/myths). party groups like Energy in Depth. Further- 4. See “Sundance 2010: Josh Fox’ GASLAND.” more, the natural gas industry helped promote January 29, 2010, accessed November 2013 (http:// the documentary Truthland, which raised edendale.typepad.com/weblog/2010/01/sundance- 2010-josh-fox-Gasland.html). questions about the accuracy of Gasland, and 5. Mark Schlosberg, EcoWatch, accessed November the counter-documentary FrackNation, which 2013 (http://ecowatch.com/2013/04/05/Gasland- attacked Fox’s evidence and credibility. explosive-growth-anti-fracking-movement/). 22 American Sociological Review

6. HBO had 28 million subscribers in 2011 (http:// March 2015) and Bakken (8 times the growth, only www.economist.com/node/21526314). Even if the 9 percent of Marcellus gas production; Bakken’s documentary was initially viewed by only a modest high volume of oil extraction is not included here). 5 percent of subscribers, we would estimate view- Nonetheless, given that fracking has been taking ership of 1.4 million. The film’s producers claim a place for longer in other formations, the total num- premiere to 3 million (see http://www.Gaslandthe ber of fracked wells is higher in the non-Marcellus movie.com/pdf/GaslandScreeningGuide.pdf). Shale plays. 7. According to Gasland’s producers, the movie was 15. July 29, 2010, was the earliest date for which mobi- seen by more than 150,000 people in 250 cities lization data were available; May 31, 2013 was the (http://www.Gaslandthemovie.com/pdf/Gasland last day for which we collected data and the day ScreeningGuide.pdf). before Gasland 2 (the sequel) was released. There 8. A second dimension could be described as an insti- are more observations in the conditional gap time tutional opportunity structure, which opens with the models used to predict mobilizations than in the growth of new social media. single-event models used to predict the adoption of 9. This reflects two distinct ways that cultural forces bans (3,448,236 versus 3,401,471) because in the matter for social movements: as cultural products former, the variable counting the passage of time is (documentaries enable the emergence of new ideas reset after each event. and practices) and as structures of discourse and 16. For example, we found that in 2013, “fracking” meaning-making (which offer more or less fertile was used approximately 7 times more frequently in environments for the reception of social movement Google searches, and 10 times more frequently in claims-making). A third option would be the tool- YouTube searches, than “hydraulic fracturing.” kit or cultural repertoires approach, which remains 17. We acknowledge the limitation of using search data critical to understanding social movement pro- from Google. However, this is not a major limita- cesses (Clemens 1997; Walker, Martin, and McCar- tion because Google is the most popular search thy 2008) but is beyond the scope of this article. We engine, with a 75 percent market share (http://www. note, however, that anti-fracking groups have cre- comscore.com/Insights/Press_Releases/2010/6/ atively borrowed and adapted rhetorics and tactical comScore_Releases_May_2010_U.S._Online_ approaches from other movements, most notably Video_Rankings). environmentalism. 18. Google Trends data are normalized by dividing a 10. These include, for instance, Kearney and Levine’s term’s frequency by the total number of searches (2014) study on how the MTV program 16 and for a topic in a given period. Google fixes a certain Pregnant reduced teen childbearing and Pautz’s “low-interest threshold”; only searches involving a (2015) quasi-experiment on how Argo and Zero level of interest above this threshold are presented, Dark Thirty increased trust in government. and the rest are considered missing data. Results 11. In a supplementary analysis, we examined the are presented on a scale from 0 to 100. To arrive Dynamics of Collective Action data on all protests at this final number, Google takes the normalized reported in (1960 to 1995), look- data and assigns 100 to the case that presents the

ing for the co-occurrence of film screenings (actn = highest overall search interest (for time series, this 12) with other tactics. We found that film screenings is the date on which the term received the greatest co-occurred with such practices as dramaturgical attention) and scales other cases proportionally. presentations, public discussions, press conferences, 19. One of this study’s authors purchased these data ceremonies, and activist photo exhibitions. through The Harmony Institute from DataSift. We 12. See “Gasland tour draws huge crowds,” accessed acknowledge that our study may be limited by using November 2014 (http://rooftopfilms.com/blog/ one social media platform, albeit a widely influen- 2010/09/Gasland-tour-draws-huge-crowds-lively- tial one. debate.html). 20. We used a topic modeling model, Latent Dirichlet 13. See http://www.riverkeeper.org/news-events/events Allocation, to remove spam; we implemented this /rvk-events/Gasland-screening-with-council-mem using the lda package in the R statistical software. ber-stephen-levin/ and http://newsgatheringblog. We experimented with the number of topics param- dentonrc.com/2011/08/denton-councilman-to-host- eter to obtain an accurate spam filter; the final filter gasl.html. allowed removal of span from most messages. We 14. Other high-producing formations grew at a much treat the remaining messages unrelated to hydraulic slower rate over this period, such as Eagle Ford fracturing as randomly distributed. (4.6 times), Haynesville (1.8 times), Niobrara (1.2 21. We were not able to obtain data from 2009 because times), and Permian (1.4 times). Drilling in other our Twitter data provider, DataSift, only provides formations grew at a high rate but accounts for data from 2010 onward. only a fraction of the gas production of the Mar- 22. We used Lexis-Nexis because it is the most com- cellus: Utica (12.1 times the growth, but only 12 prehensive database of U.S. newspapers. We exper- percent of the gas extraction of the Marcellus as of imented with different search queries to eliminate Vasi et al. 23

false positives and off-topic reports. We found 33. We included contamination documented by Earthjus- that the most accurate query was [“fracking” and tice only if we could verify it using media sources. (“hydraulic fracturing” or “natural gas”)]. 34. A full census of anti-fracking SMOs is beyond our 23. This measure of anti-fracking mobilizations scope. However, many events were sponsored by includes any type of collective gathering to organize ad hoc community groups, such as the Gas Drilling opposition against fracking, not just street demon- Awareness Coalition (in Dallas, Pennsylvania). strations. We are unable to distinguish between 35. We acknowledge, as in other studies that rely on events that included large numbers of people and nonprofit tax data to estimate associational popula- were highly visible, such as street demonstrations, tions, that these figures underestimate informal and and events that included a relatively small number short-lived associations. We included other versions of activists and were not highly visible, such as a of the nonprofit measures, such as diversity of non- debate at a local library or college. profits and total revenues, as controls in alternative 24. See http://www.foodandwaterwatch.org/. specifications. None of these measures had signifi- 25. We excluded three cases where municipal ordi- cant effects. nances were later repealed. 36. Results available upon request. 26. We experimented with larger radii (10, 15, and 20 37. Because the estimation of proportional hazards miles) but the main effects did not change. We used models when hazards are non-proportional can information about the geographic coordinates of the result in biased estimates, incorrect standard errors, cities where the film was screened. and faulty inferences (Box-Steffensmeier and Jones 27. See http://www.Gaslandthemovie.com/pdf/Gasland 2004), we use the Grambsch and Therneau tests for ScreeningGuide.pdf. non-proportional hazards. The main effects do not 28. See the online supplement (http://asr.sagepub.com/ change when we include time interaction for the supplemental) for additional information about how covariates with non-proportional hazards. Twitter data were obtained. 38. We also estimated measures of proximity to wells 29. Areas not located directly above shale formations and water contamination accidents beyond a five- often protest fracking, too, due to concerns about mile radius (10, 15, and 20 miles); the main effects related activities (e.g., trucking, compressor sta- remain the same. tions, and downstream pollution). 39. More detailed illustrations of the decay are avail- 30. We used the following sources for each: New able upon request. York (http://www.nysm.nysed.gov/nysgs/research/ 40. We also found that local screenings increase the oil-gas/marcellus.html), Pennsylvania (http:// number of fracking-related Twitter messages posted www.dep.state.pa.us/dep/deputate/minres/oilgas/ from the city where the film was screened, but the BOGM%20Website%20Pictures/2010/Marcel- effect lasts only for a few days post-screening. lus%20Shale%20Formation.jpg), Ohio (http:// 41. However, we find a significant interaction effect geosurvey.ohiodnr.gov/portals/geosurvey/Energy/ between Gasland screenings and total number of Utica/Utica_Marcellus_Ohio_8x11.pdf), and West NGOs or number of environmental NGOs; these Virginia (http://www.wvgs.wvnet.edu/www/data- are omitted due to space limitations. stat/devshales.htm). 42. See the online supplement for additional robustness 31. We used the following disclosure websites for checks. each: Pennsylvania (https://www.paoilandgasreport 43. For example, we find that the cumulative number of ing.state.pa.us/publicreports/Modules/Welcome/ previous adopters in a certain radius has a weaker Agreement.aspx), New York (http://www.dec (but significant) effect than does weighted number .ny.gov/cfmx/extapps/GasOil/search/wells/index of previous adopters. We report results only from f .cfm), Ohio (http://oilandgas.ohiodnr.gov/shale), = 1 / sqrt(d), but we experimented with other decay and West Virginia (http://www.dep.wv.gov/oil-and- functions (e.g., f = 1/2^[x/10]; f = 1/2^[x/20]); gas/databaseinfo/Pages/default.aspx). Local per- the main results remain unchanged. We also tried mits issued in New York State are preempted by its alternative radii of influence (e.g., 30 miles and 60 state-level moratorium. miles) with similar results. 32. We included the number of wells within a given 44. In supplemental analyses (available upon request), radius in the previous year because we wanted to we found that mass media coverage of fracking was include only wells that were already active and, significantly more positive in sentiment than was therefore, acted as “suddenly imposed grievances” the Twitter discourse. (Walsh 1981). Given that both dependent variables are measured in days but the well data is available only by year, we could not be sure that a well was References already drilled by the time people mobilized against Aldrich, Daniel P. 2008. Site Fights. Ithaca, NY: Cornell hydraulic fracturing. However, we also included the University Press. sum of active fracking wells within a certain radius Amenta, Edwin, Neal Caren, Sheera J. Olasky, and James from each city in the same year; results were similar. E. Stobaugh. 2009. “All the Movements Fit to Print: 24 American Sociological Review

Who, What, When, Where, and Why SMO Families Brown, Phil, Steve Kroll-Smith, and Valerie J. Gunter. Appeared in the New York Times in the Twentieth 2000. “Knowledge, Citizens, and Organizations: An Century.” American Sociological Review 74(4):636– Overview of Environments, Diseases, and Social 56. Conflict.” Pp. 9–28 in Illness and the Environment, Andits, Petra. 2013. “Movies and Movements.” Pp. 1–2 edited by S. Kroll-Smith, P. Brown, and V. J. Gunter. in The Wiley-Blackwell Encyclopedia of Social and New York: NYU Press. Political Movements, (published online) edited by Brown, Phil, and Edwin Mikkelsen. 1990. No Safe Place. D. A. Snow, D. della Porta, B. Klandermans, and D. Berkeley: University of California Press. McAdam. West Sussex, UK: Blackwell Publishing Ltd. Brulle, Robert J. 2000. Agency, Democracy, and Nature. Andrews, Kenneth T., and Michael Biggs. 2006. “The Cambridge, MA: MIT Press. Dynamics of Protest Diffusion: Movement Organiza- Bullard, Robert D. 2000. Dumping in Dixie. Boulder, CO: tions, Social Networks, and News Media in the 1960s Westview. Sit-Ins.” American Sociological Review 71(5):752–77. Cable, Sherry. 2012. Sustainable Failures. Philadelphia: Aronoff, Marilyn, and Valerie Gunter. 1994. “A Pound Temple University Press. of Cure: Facilitating Participatory Processes in Tech- Chadwick, Andrew. 2013. The Hybrid Media System. nological Hazard Disputes.” Society and Natural Oxford, UK: Oxford University Press. Resources 7(3):235–52. Clemens, Elisabeth S. 1997. The People’s Lobby. Chi- Auyero, Javier, and Debora Swistun. 2008. “The Social cago: University of Chicago Press. Production of Toxic Uncertainty.” American Socio- Crenson, Matthew. 1971. The Un-politics of Air Pollu- logical Review 73(3):357–79. tion. Baltimore, MD: John Hopkins Press. Bail, Christopher A. 2012. “The Fringe Effect: Civil Soci- Cress, Daniel M., and David A. Snow. 2000. “The ety Organizations and the Evolution of Media Dis- Outcomes of Homeless Mobilization: The Influ- course about Islam since the September 11th Attacks.” ence of Organization, Disruption, Political Media- American Sociological Review 77(6):855–79. tion, and Framing.” American Journal of Sociology Baumgartner Frank R., and Bryan D. Jones. 1993. Agen- 105(4):1063–1104. das and Instability in American Politics. Chicago: Cruz, Jon. 1999. Culture on the Margins. Princeton, NJ: University of Chicago Press. Princeton University Press. Benford, Robert D., and David A. Snow. 2000. “Fram- Danaher, William F. 2010. “Music and Social Move- ing Processes and Social Movements: An Over- ments.” Sociology Compass 4(9):811–23. view and Assessment.” Annual Review of Sociology Earl, Jennifer, and Katrina Kimport. 2011. Digitally 26(1):611–39. Enabled Social Change. Cambridge, MA: MIT Press. Bennett, D Scott. 1999. “Parametric Models, Duration Elliot, James R., and Scott Frickel. 2013. “The Historical Dependence, and Time-Varying Data Revisited.” Nature of Cities: A Study of Urbanization and Haz- American Journal of Political Science 43(1):256–70. ardous Waste Accumulation.” American Sociological Bennett, W. Lance. 2003. “Communicating Global Review 78(4):521–43. Activism.” Information, Communication & Society Eyerman, Ron, and Andrew Jamison. 1995. “Social Move- 6(2):143–68. ments and Cultural Transformation: Popular Music in Bennett, W. Lance, and Alexandra Segerberg. 2012. “The the 1960s.” Media, Culture, and Society 17(3):449–68. Logic of Connective Action.” Information, Commu- Eyerman, Ron, and Andrew Jamison. 1998. Music and nication & Society 15(5):739–68. Social Movements. Cambridge, UK: Cambridge Uni- Bennett, W. Lance, and Alexandra Segerberg. 2013. The versity Press. Logic of Connective Action. Cambridge, UK: Cam- Fishman, Mark. 1980. Manufacturing the News. Austin: bridge University Press. University of Texas Press. Benson, Rodney. 2004. “Bringing the Sociology of Media Ferree, Myra Marx. 2003. “Resonance and Radicalism: Back In.” Political Communication 21(3):275–92. Feminist Framing in the Abortion Debates of the Berezin, Mabel. 1994. “Cultural Form and Political United States and Germany.” American Journal of Meaning: State-Subsidized Theater, Ideology, and the Sociology 109(2):304–344. Language of Style in Fascist Italy.” American Journal Flynn, James, Paul Slovic, and Howard Kunreuther. of Sociology 99(5):1237–86. 2001. Risk, Media and Stigma: Understanding Public Box-Steffensmeier, Janet M., and Bradford S. Jones. Challenges to Modern Science and Technology. Ster- 1997. “Time Is of the Essence: Event History Models ling, VA: Earthscan. in Political Science.” American Journal of Political Fowlkes, Martha, and Patricia Miller. 1982. Love Canal. Science 41(4):1414–61. Washington, DC: FEMA. Box-Steffensmeier, Janet M., and Bradford S. Jones. Freudenburg, William R., and Timothy R. Jones. 1991. 2004. Event History Modeling: A Guide for Social “Attitudes and Stress in the Presence of Technologi- Scientists. New York: Cambridge University Press. cal Risk: A Test of the Supreme Court Hypothesis.” Boykoff, Maxwell T., and Jules M. Boykoff. 2004. “Bal- Social Forces 69(4):1143–68. ance as Bias: Global Warming and the US Prestige Gamson, William A., and Andre Modigliani. 1989. Press.” Global Environmental Change 14(2):125–36. “Media Discourse and Public Opinion on Nuclear Vasi et al. 25

Power: A Constructionist Approach.” American Jour- Koopmans, Ruud, and Paul Statham. 1999. “Challeng- nal of Sociology 95(1):1–37. ing the Liberal Nation-State? Postnationalism, Mul- Gans, Herbert J. 1979. Deciding What’s News. Evanston, ticulturalism, and the Collective Claims Making of IL: Northwestern University Press. Migrants and Ethnic Minorities in Britain and Ger- Gaventa, John. 1982. Power and Powerlessness. Chi- many.” American Journal of Sociology 105(3):652–96. cago: University of Chicago Press. Kreiss, Daniel, Megan Finn, and Fred Turner. 2011. “The Gerbaudo, Paolo. 2012. Tweets and the Streets. New Limits of Peer Production: Some Reminders from York: Pluto Press. Max Weber for the Network Society.” New Media & Gitlin, Todd. 1980. The Whole World Is Watching. Berke- Society 13(2):243–59. ley: University of California Press. Kriesi, Hanspeter. 2004. “Political Context and Oppor- Giugni, Marco, Ruud Koopmans, Florence Passy, and tunity.” Pp. 67–90 in Blackwell Companion to Social Paul Statham. 2005. “Institutional and Discursive Movements, edited by D. A. Snow, S. A. Soule, and Opportunities for Extreme-Right Mobilization in H. Kriesi. West Sussex, UK: Blackwell Publishing. Five Countries.” Mobilization 10(1):145–62. Kroll-Smith, Stephen J., and Stephen R. Couch. 1993. Hanson, Michael. 2008. “Suppose James Brown Read “Symbols, Ecology, and Contamination: Case Stud- Fanon: The Black Arts Movement, Cultural Nation- ies in the Ecological-Symbolic Approach to Disas- alism and the Failure of Popular Musical Praxis.” ter.” Research in Social Problems and Public Policy Popular Music 27(3):341–65. 5:47–73. Hedström, Peter, Rickard Sandell, and Carlotta Stern. Kroll-Smith, Stephen J., Stephen R. Couch, and Adeline 2000. “Mesolevel Networks and the Diffusion of Levine. 2002. “Technological Hazards and Disas- Social Movements.” American Journal of Sociology ters.” Pp. 295–328 in Handbook of Environmental 106(1):145–72. Sociology, edited by R. Dunlap and W. Michelson. Hemingway, Andrew. 2002. Artists on the Left. New Westport, CT: Greenwood Press. Haven, CT: Yale University Press. Leiserowitz, Anthony A. 2004. “Before and After The Howard, Philip N., and Muzammil M. Hussain. 2011. Day After Tomorrow: A U.S. Study of Climate “The Upheavals in Egypt and Tunisia: The Role of Change Risk Perception” Environment 46(9):23–37. Digital Media.” Journal of Democracy 22(3):35–48. Lipsitz, George. 2000. Time Passages. Minneapolis: Uni- Hussain, Muzammil M., and Philip N. Howard. 2013. versity of Minnesota Press. “What Best Explains Successful Protest Cascades? McAdam, Doug. 1994. “Culture and Social Movements.” ICTs and the Fuzzy Causes of the Arab Spring.” Pp. 36–56 in New Social Movements: From Ideology International Studies Review 15(1):48–66. to Identity, edited by E. Larana, H. Johnston, and Jacobsen, Grant. 2011. “The Al Gore Effect: An Inconvenient J. R. Gusfield. Philadelphia: Temple University Press. Truth and Voluntary Carbon Offsets.” Journal of Envi- McAdam, Doug, and Hilary Schaffer Boudet. 2012. Put- ronmental Economics and Management 61(1):67–78. ting Social Movements in Their Place: Explaining Jalbert, Kirk, Abby J. Kinchy, and Simona L. Perry. 2014. Opposition to Energy Projects in the United States, “Civil Society Research and Marcellus Shale Natural 2000–2005. Cambridge, UK: Cambridge University Gas Development: Results of a Survey of Volunteer Press. Water Monitoring Organizations.” Journal of Envi- McAdam, Doug, Hilary Schaffer Boudet, Jennifer Davis, ronmental Studies & Sciences 4(1):78–86. Ryan J. Orr, W. Richard Scott, and Raymond E. Lev- Kargbo, David M., Ron G. Wilhelm, and David J. Camp- itt. 2010. “‘Site Fights’: Explaining Opposition to bell. 2010. “Natural Gas Plays in the Marcellus Shale: Pipeline Projects in the Developing World.” Socio- Challenges and Potential Opportunities.” Environ- logical Forum 25(3):401–427. mental Science & Technology 44(15):5679–84. McCammon, Holly J., Soma Chaudhuri, Lyndi N. Karlin, Beth, and John Johnson. 2011. “Measuring Hewitt, Courtney Sanders Muse, Harmony D. New- Impact: The Importance of Evaluation for Documen- man, Carrie Lee Smith, and Teresa M. Terrell. 2008. tary Film Campaigns.” M/C Journal 14(6) (http:// “Becoming Full Citizens: The U.S. Women’s Jury journal.media-culture.org.au/index.php/mcjournal/ Rights Campaigns, the Pace of Reform, and Stra- article/viewArticle/444). tegic Adaptation.” American Journal of Sociology Kearney, Melissa, and Phillip Levine. 2014. “Media Influ- 113(4):1104–47. ences on Social Outcomes: The Impact of MTV’s 16 McCammon, Holly J., Courtney Sanders Muse, Harmony and Pregnant on Teen Childbearing.” NBER Work- D. Newman, and Teresa M. Terrell. 2007. “Movement ing Paper #19795. Framing and Discursive Opportunity Structures: The Koopmans, Ruud. 2004. “Movements and Media: Selec- Political Successes of the U.S. Women’s Jury Move- tion Processes and Evolutionary Dynamics in the ments.” American Sociological Review 72(5):725–49. Public Sphere.” Theory and Society 33(3–4):367–91. Meyer, David S., and Deana A. Rohlinger. 2012. “Big Koopmans, Ruud, and Susan Olzak. 2004. “Discursive Books and Social Movements: A Myth of Ideas and Opportunities and the Evolution of Right-Wing Vio- Social Change.” Social Problems 59(1):136–53. lence in Germany.” American Journal of Sociology Moniz, Ernest, Henry D. Jacoby, and Anthony J. M. 110(1):198–230. Meggs. 2011. “The Future of Natural Gas: An 26 American Sociological Review

Interdisciplinary MIT Study.” MIT Energy Initiative, mental Risk and Financial Performance.” American Massachusetts Institute of Technology, Cambridge, MA. Sociological Review 77(4):573–97. Oliver, Pamela E., and Daniel J. Myers. 1999. “How Vasi, Ion Bogdan, and Chan Suh. 2012. “Public Atten- Events Enter the Public Sphere: Conflict, Location, tion, Social Media, and the Spread of the ‘Occupy’ and Sponsorship in Local Newspaper Coverage Movement in the United States.” Presented at the of Public Events.” American Journal of Sociology annual meeting of the American Sociological Asso- 105(1):38–87. ciation, Denver, CO. Osborn, Stephen G., Avner Vengosh, Nathaniel R. Walker, Edward T. 2009. “Privatizing Participation: Warner, and Robert B. Jackson. 2011. “Methane Civic Change and the Organizational Dynamics of Contamination of Drinking Water Accompanying Grassroots Lobbying Firms.” American Sociological Gas-Well Drilling and Hydraulic Fracturing.” PNAS Review 74(1):83–105. 108(20):8172–76. Walker, Edward T. 2014. Grassroots for Hire: Public Papacharissi, Zizi, and Maria de Fatima Oliveira. 2012. Affairs Consultants in American Democracy. Cam- “Affective News and Networked Publics: The bridge, UK: Cambridge University Press. Rhythms of News Storytelling on #Egypt.” Journal Walker, Edward T., Andrew W. Martin, and John D. of Communication 62(2):266–82. McCarthy. 2008. “Confronting the State, the Cor- Pautz, Michelle C. 2015. “Argo and Zero Dark Thirty: poration, and the Academy: The Influence of Insti- Film, Government, and Audiences.” PS 48:120–28. tutional Targets on Social Movement Repertoires.” Pidgeon, Nick, Roger E. Kasperson, and Paul Slovic. American Journal of Sociology 114(1):35–76. 2003. The Social Amplification of Risk. New York: Walsh, Edward J. 1981. “Resource Mobilization and Cambridge University Press. Citizen Protest in Communities around Three Mile Reed, T. V. 2005. The Art of Protest: Culture and Activ- Island.” Social Problems 29(1):1–21. ism from the Civil Rights Movement to the Streets of Whiteman, David. 2003. “Reel Impact.” Stanford Social Seattle. Minneapolis: University of Minnesota Press. Innovation Review 1(1):60–63. Roscigno, Vincent J. 2011. “Power, Revisited.” Social Wilber, Tom. 2012. Under the Surface: Fracking, For- Forces 90(2):349–73. tunes, and the Fate of the Marcellus Shale. Ithaca, Roscigno, Vincent J., and William F. Danaher. 2001. NY: Cornell University Press. “Media and Mobilization: The Case of Radio and Zolberg, Vera. 1997. Outsider Art. Cambridge, UK: Cam- Southern Textile Worker Insurgency, 1929 to 1934.” bridge University Press. American Sociological Review 66(1):21–48. Roy, William. 2010. Reds, Whites, and Blues. Princeton, Ion Bogdan Vasi is Associate Professor in the Depart- NJ: Princeton University Press. ment of Sociology and the College of Business at Univer- Schudson, Michael. 1989. “The Sociology of News Pro- sity of Iowa. His research examines how social duction.” Media, Culture and Society 11(3):263–82. movements contribute to organizational change, industry Schudson, Michael. 2002. “The News Media as Politi- creation, and policymaking. He also studies the social cal Institutions.” Annual Review of Political Science processes that shape the diffusion of technological inno- 5(1):249–69. vations. He is the author of Winds of Change: The Envi- Shirky, Clay. 2008. Here Comes Everybody. New York: ronmental Movement and the Global Development of the Penguin. Wind Energy Industry (Oxford University Press 2011). Steinberg, Marc W. 1999. “The Talk and Back Talk of Col- lective Action: A Dialogic Analysis of Repertoires of Edward T. Walker is Associate Professor and Vice Discourse among Nineteenth-Century English Cotton Chair in the Department of Sociology at the University of Spinners.” American Journal of Sociology 105(3):736–80. California-. His research investigates the Steinberg, Marc W. 2004. “When Politics Goes Pop: On mobilization and outcomes of popular participation both the Intersections of Popular and Political Culture and by social movement organizations and by business firms the Case of Serbian Student Protests.” Social Move- and trade associations. He is author of Grassroots for ment Studies 3(1):3–29. Hire: Public Affairs Consultants in American Democracy Tuchman, Gaye. 1978. Making News. New York: Free Press. (Cambridge University Press 2014). Tufekci, Zeynep, and Christopher Wilson. 2012. “Social Media and the Decision to Participate in Political John S. Johnson is co-founder of the social news and Protest: Observations from Tahrir Square.” Journal entertainment site Buzzfeed and is founder and executive of Communication 62(2):363–79. director of the Harmony Institute, an independent, Vasi, Ion Bogdan. 2011. Winds of Change: The Environmen- applied quantitative research institute studying the rela- tal Movement and the Global Development of the Wind tionship between media and social change. Energy Industry. Oxford, UK: Oxford University Press. Vasi, Ion Bogdan, and Brayden King. 2012. “Social Hui Fen Tan is a PhD student in statistics at Cornell Movements, Risk Perceptions, and Economic University. She works on inference and interpretability of Outcomes: The Effect of Primary and Secondary machine learning methods. She is also interested in Stakeholder Activism on Firms’ Perceived Environ- causal inference and computational social science.