Analysis of Newspaper Coverage of Offshore Wind Energy in the United States

The Block Island Wind Farm (BIWF) is the first operational offshore wind farm in the United States. Two different theoretical frameworks are applied in an attempt to understand the nature of media coverage about offshore wind energy in the United States. Chapter 1 utilizes gatekeeping theory (Shoemaker & Vos, 2009) in comparing the local and regional newspaper coverage of the BIWF. Specifically, chapter 1 is focused on understanding the use of sources. The data from chapter 1 suggests that the gatekeeping processes at regional and local outlets are more similar than different, but business sources are more dominant than has been proud in previous research. Chapter 2 uses Stephens et al. (2009) and Smith et al.’s (2016) adaptation of Luhmann’s theory of ecological communication (1989) comparing the first five years of local newspaper coverage of the Cape Wind offshore wind farm and the BIWF. The data from chapter 2 suggests coverage of the Cape Wind offshore farm utilized more risk narratives than the local coverage of the BIWF, yet coverage of both wind farms centered around the political dimensions related to the projects.


Introduction
A little before 6:00 AM on May 1, 2017, Block Island, a small island community off the coast of Rhode Island, began receiving energy powered by offshore wind. The five-turbine offshore wind farm, built by Deepwater Wind, is the first in the United States. The wind farm's completion is being seen by those in the offshore wind industry as a pivotal moment and many are hopeful that this will be a catalyst event that spurs additional development in the United States . While the Block Island Wind Farm (BIWF) was successful, the permitting process was complicated, intensive, and took almost a decade . The nine-year process however, is short when compared to the first proposed offshore wind farm by Cape Wind Associates in Nantucket Sound, Massachusetts. Cape Wind Associates recently ended its 16-year battle when it terminated its lease with the Bureau of Ocean Energy Management (BOEM) in 2017 .
Despite the complexities in permitting and building offshore wind farms in the United States, there remains strong interest in the future of the offshore wind industry, as evidenced by the numerous ocean area leases granted to companies by BOEM in the past decade (BOEM, 2017). Deepwater Wind, the company responsible for the BIWF is proposing offshore wind farms off the coast of Maryland, New Jersey, and Massachusetts . In addition, states like Massachusetts have announced aggressive renewable energy goals calling for 1,600 megawatts of energy to be produced by offshore wind by 2027 .
The BIWF remains the only realized offshore wind farm in the United States at the time of this thesis, and for this reason it has become a site of academic research in numerous fields .
No research to date has analyzed the media coverage of the BIWF to understand the nature of the coverage and identify any discursive themes or narrative structures that were used to describe the wind farm and its impacts. Given the infancy of the offshore wind industry in the United States and the numerous projects currently on the horizon, this is an ideal opportunity to study the media discourse of the BIWF. Of particular interest in this research study are the sources present in newspaper coverage of the BIWF; in other words, who were the frequently quoted individuals and groups in framing the BIWF? To answer this question, we turn to gatekeeping theory.

Theoretical Framework: Gatekeeping Theory
There is a strong history in the communication discipline of studying media representations of complex issues. Framing theory, agenda-setting theory, priming theory, and a myriad of other nuanced theoretical frameworks are continuously being updated to help explore the relationship between the media and the public . Research across theoretical approaches, has shown that media representations or coverage of complex stories has some link to public perceptions . Offshore wind energy, given the ever-improving technical nature of the turbines, and the public perceptions about these "green energy machines," can be a complicated issue for reporters to cover (Wüstenhagen, Wolsink, & Buerer, 2007).
The premise of gatekeeping theory is that journalistic choices, either at the level of an individual journalist or an entire media organization, are key factors in the construction of media narratives. One key tenet of the theory that serves as the foundation for this study is that source selection and the presence or lack of source diversity is a result of the gatekeeping process Smith & Norton, 2013). This theoretical framework suggests that information that makes it "past" the proverbial gatekeepers in the media is reflective of underlying assumptions about who is a credible source and whose voices should be heard ). Recent gatekeeping theory research has also suggested that the individual and routine forces that influence media coverage are applicable across print and online journalists and organizations .
In a study on source diversity,  found, "the use of sources in news narratives is an extremely important part of not only the story's construction but also its orientation and, ultimately, the point of view being supported in a given story" (p. 449). Assuming that gatekeeping theory's proposition that the information and sources included in newspaper stories are reflective of what journalists deem as important, evaluating the newspaper coverage of the BIWF could provide insight into the broader narrative of offshore wind energy in the United States; particularly focusing on identifying which sources are telling the story. As  notes, "If journalists serve as legitimizers, reflecting events in society to readers and viewers, the sources and subjects of those stories are going to present and reinforce certain ideals" (p. 140).

Local versus Regional Newspaper Coverage
Previous media research using gatekeeping theory to analyze source use has found justification for analyzing newspaper coverage based on geographical location . Research has shown that source diversity "differs between wire services and local, state or national newspaper coverage" (Maddison & Watts, 2012, p. 50).
Additionally, research into the concept of NIMBYism ("not in my backyard") suggests that there may be meaningful differences between local newspaper coverage, national newspaper coverage and corresponding public perceptions when it comes to environmental issues . Despite this research, there has been limited research into local newspaper coverage of renewable energy or biotechnology stories (Wakefield & Elliot, 2003;. Local and regional in the context of this study are directly related to geographical setting. The Block Island Times (BIT), which provides a large portion of articles to be analyzed in this study, is a newspaper located on Block Island that is intended for visitors and year-round residents: it will be the only newspaper in this study considered "local." The initial intent of this study was to compare three levels of BIWF newspaper coverage, but due to the limited number of national newspaper articles (n=24), all newspaper articles that are not from the BIT will be considered regional. This differentiation is not based on readership or circulation numbers; it is rather based on the physical proximity of the newspaper outlet to the BIWF.

Hypotheses and research questions
Given the newness of the BIWF, there are more questions than answers when considering the nature of media coverage about the BIWF. It seems reasonable to expect that there would be more local newspaper articles on the BIWF than national newspaper articles, but in what ways would the sources relied on differ, if at all?
Would local and regional stories about the BIWF rely more heavily on government or elected officials because "they (government/elected officials") can provide a steady stream of newsworthy information" and smaller newspapers do not have as much capital as national outlets (Maddison & Watts, 2012, p. 41)? Or are the processes of source gathering so routine that they bridge geographies )?
Therefore, based on gatekeeping theory, the research question was:

RQ1:
Are there statistically significant differences in the people/organizations used as sources between the local and regional coverage of the BIWF?
Ultimately, we hope to evaluate the source coverage as a way to understand the broader newspaper coverage of offshore wind in the United States. The role of the BIWF as a first presents a unique research opportunity to assess if previous research on source diversity and geographical source differences are consistent in the coverage of this new media topic. With that in mind, the research question is purposefully broad in order to present the findings in their entirety.

Data Collection
All newspaper articles identified for analysis were found utilizing online databases including LexisNexis Academic (now Nexis Uni), National Newspapers Core (ProQuest), Google News, the Boston Globe archives, Infotrac Newsstand, and the BIT archives. Search terms and dates were consistent in using the terms "Block Island Wind" OR "Deepwater Wind" AND "Rhode Island" with a date range of January 1, 2008 to July 1, 2017. January 1, 2008 as the beginning date is reflective of Rhode Island announcing interest in offshore wind development late in 2007 (Voskamp, 2008). July 1, 2017 was selected as the end date in order to create a bounded data set. July 1, 2017 allowed for articles to be included that were published during the first summer season in which the BIWF was operational, but ultimately allowed the research to move from the data collection phase into the analysis phase.
The only exception to search terms occured in using the Infotrac Newstand database to find Providence Journal articles; the search terms were narrowed down to "Block Island Wind" due to the heavy prevalence of articles about the company Deepwater Wind with no mention of the BIWF. Articles were then skimmed by the primary researcher to determine if the newspaper reporting was appropriate and if so, the article was catalogued into NVivo software.
All articles from the BIT were labeled as "local newspaper articles" and all other newspaper articles were catalogued as "regional newspaper articles." Again, due to the small number of national articles found between 2008 and 2017, all national newspaper articles are being included in the "regional" category. All editorial or opinion articles were omitted and no articles from global newspapers were included (Smith & Norton, 2013;.
A total of 483 newspaper articles were identified and included in the data set for analysis.

Coding Procedure
A preliminary codebook was developed by the primary researcher based on previous source analysis research Smith & Norton, 2013;. The codebook emphasized that a source will be "(a) named or anonymous individual(s) that provided information or opinion in a direct quote, partial quote, or paraphrase" (Sumpter & Garner, 2007, p. 460). When an article is specifically attributing a statement, whether fact or opinion, to an individual or a larger organization, this inclusion of ownership is important to identify ).
The codebook, identical for identifying sources in both the local newspaper articles and the regional articles, had six dominant or parent source codes: no sources identified, business or for-profit representative, government or elected official, scientist or expert, individual, or nonprofit representative (see Appendix A). These six sources are reflective of previous research into source diversity:  work on the inclusion of government officials, at all levels, in newspapers;  inclusion of business representatives, consumers or individual source categories; and the work by  on the citing of experts and scientists. The nonprofit category was included based on preliminary readings of the articles and the presence of individuals quoted who were members or leaders, often times, of environmental organizations. Smith & Norton (2013) note that in regards to environmental topics, nonprofit environmental groups are given a voice where they might not have previously had one: "But this is an age of organizations and in no sector is this more true than non-profit environmental social movement organizations" (p. 53). A critical piece of gatekeeping theory is that journalists and media organizations are dynamic; the most dominant source is limited to a specific time frame and media environment ).
Within each of these parent codes, the primary researcher identified more specific sub or child codes during preliminary readings of the articles . An example of a child code within "business" was "developers of offshore wind." In the case of an individual from Deepwater Wind, the company behind the BIWF, being quoted, the entirety of that statement would be included in both the child code of "developers of offshore wind" and the parent code of "business." By being more specific within each of the six dominant codes, researchers hoped for a more nuanced understanding of the most common voices in newspaper coverage . Instances of no subcode being identified occurred, but it was necessitated that one of the six dominant codes always be referenced in coding a citation.
Given the narrative structure of the articles coded, researchers emphasized coding based on identifiers in a specific article. Even if a source had been mentioned in a previous article, coders analyzed a source based on what was presented in each piece of data. In other words, while "Dr. Patel" may have been identified as a marine biologist in a previous article, if he was noted only as being a Block Island resident in a subsequent article, Dr. Patel should be coded as an "individual" despite the previous information.
The six dominant codes were also mutually exclusive. Citations where individuals were introduced by multiple attributes, researchers identified the dominant code. In other words, it seems safe to generally assume that an elected member of local government is also a local resident. But to code any local government official as also a local resident would be misleading if a quote did not identify a source in this way. While previous research has allowed non-mutually exclusive source coding, given the heavy presence of local newspaper coverage in the data, researchers decided it would more appropriately reflect the coverage to identify a dominant code for each source and remain consistent throughout a specific article . This is consistent with media research that identifies a primary or dominant frame given the presence of multiple narratives .  suggested coding as source using the first qualifier provided in maintaining mutually exclusive source categories. Hermida et al (2014) used a consensus method to come to an agreement among researchers in cases of ambiguity or disagreement about a source's primary identifier.
The frequency of specific sources referenced was also of interest in this study.
If throughout an article, for example, the same state government official was quoted in three separate, different paragraphs, this seems to suggest not only the importance of this individual, but also the multiple areas in which this organization has been given credibility (Dimitrova & Strömbäck, 2009;Smith & Norton, 2013). The codebook specified that if an individual was cited in one part of the article and again in another part of the article, if those two quotes were separated by at least another line of text, sentence, etc., those two quotes would be counted as two citations from the same coded category. Additionally, if multiple people from the same organization or dominant code were quoted, they were quoted as two different codes, as the inclusion of two voices from the same group is important to consider.
To ensure intercoder reliability, a random sample of 10% of the dataset (n=49) was coded by both researchers Dimitrova & Strömbäck, 2009

Results
Researchers coded a total of 483 articles; local articles, n=352; regional articles, n=131. The two tables below show the total number of articles that included a quote from one of the six parent codes (code present) and the total number of quotes included from that parent code (total code count); table 1 details the local newspaper coverage data and table 2 the regional newspaper coverage. The most commonly cited source for both local and regional newspapers were business and for-profit representatives. Of local articles that included at least one source (n=328), 82% cited at least one business representative. For regional articles with at least one source quoted (n=121), 83% had at least one business or for-profit individual cited. A twotailed z-test did not support a statistically significant difference between how local and regional newspapers used business citations (p=0.77182). The raw data is being provided here as suggested by Simmons, Nelson, and Simonsohn (2011).

For-Profit Businesses
Within this business or for-profit representative parent code, the majority of quotes came from "developers of wind farms." 81% of local articles that included a quote from a business or for-profit representative had a quote from a representative of a wind farm company: company examples included Deepwater Wind, Allco, and Cape Wind. For regional articles, 84% of articles that included a quote from a for-profit representative included a quote from a wind farm developer. The code "developers of wind farms" is inclusive of onshore wind farm developers, but the data came overwhelmingly from offshore wind developers, particularly Deepwater Wind, the company that developed the BIWF. A two-tailed z-test did not support a statistically significant difference between the way local and regional newspapers used wind farm representatives as sources (p=0.44726).

Scientists
Local newspaper stories that included at least one source (n=328) included a citation from an expert or scientist 7% of the time. Regional newspapers, using the same parameters, included a quote from an expert in 13% of articles. While the majority of newspaper articles did not include a quotation from a researcher, those articles that did, seemed to select who to include based on geographical proximity. In local newspaper stories that included a quotation from an expert, 88% of those quoted were identified as being scientists working in the New England area (i.e. local experts). However, in regional coverage, 38% of expert citations belonged to local scientists. A two-tailed z-test showed that this difference was significant at the 0.01 level (p=0.00094). Regional newspapers relied more heavily on national experts, using them 56% of the time, versus local newspapers using national scientists in 21% of articles that quoted an expert. A two-tailed z-test showed that this difference was not statistically significant at the 0.01 level (p=0.02144). International experts were almost never used and were quoted in a total of four articles across both local and regional newspapers (n=483).

Government and elected officials
The second most frequently cited source for both local and regional newspapers were government or elected officials. 58% of local articles that included at least one citation included a quote from a government official. 61% of regional articles using the same parameters cited at least one government official. A two-tailed z-test did not support a statistically significant difference between the use of local and regional governmental citations (p=0.50286). Table 3 shows the presence of the four sub-codes for government or elected officials for local articles and table 4 shows the same data for regional articles. In local articles that quoted at least one elected official (n=189), 61% of these articles specifically cited a local government official. In regional coverage, the percentage of local officials cited in an article that quoted at least one elected official (n=74) was 18%. This difference in local government citation was statistically significant at the 0.01 level (p=0.00000). The most commonly cited government official for regional coverage were Rhode Island state-level government officials, such as the governor or state house representatives (54%). Less than half of local articles that cited at least one governmental source (n=189) quoted a state government official (47%). However, this difference was not statistically significant at the 0.01 level (p=0.30772).

Individuals
In local articles that included at least one source (n=328), 27% included a quote from an individual. Almost all (92%) of these individuals were noted as local residents; local residents include both Block Island residents and residents of neighboring towns, such as Narragansett and Kingston. Regional articles with at least one citation (n=121) included an individual 12% of the time; 93% of these individuals were local residents as defined by the parameters above. The difference in the use of individuals between local and regional newspapers was shown to be statistically significant at the 0.01 level (p=0.00104). Tourists, coded within the dominant code of "individuals," appeared in almost no articles (n=3) across local and regional coverage (n=483).

Non-profit organizations
Local newspapers cited nonprofit representatives in less than 10% of articles that included at least one citation and regional newspapers quoted nonprofit representatives in 16% of the articles with at least one quote. The difference in usage between local and regional newspapers of nonprofit representatives was not statistically significant at the 0.01 level (p=0.01046). 84% of all nonprofit quotations, from both local and regional newspapers, came from environmental groups, such as the Conservation Law Foundation.

For-Profit Business Sources Most Common
What is immediately clear is that the prevalence of for-profit company quotations in local and regional newspaper coverage of the BIWF suggests a similar gatekeeping process in source selection. This strong reliance on business sources is a surprising finding. Previous research has empirically supported the hypothesis that journalists commonly defer to "elite" sources of information, but much of this work has found that government officials fulfill this role Smith & Norton, 2013).
Researchers have also found that this reliance on politicians or government officials is commonplace on social media sites, such as Twitter . Additionally, the focus of the news stories that this research has been conducted on has spanned numerous topics: national disasters , business crises , political elections , technological disasters , and environmental issues including carbon emissions and federal land policies Smith & Norton, 2013). This highlights the strong reliance of journalists on government sources, regardless of "beat." It should be noted that , in their work on newspaper coverage of the Clean Air Act, did find that governmental sources, while most common, were closely followed by business sources: 43% of total sources came from government sources and 39% from business sources (Lacy & Coulson, 2000, p. 18).
But, it appears that this study is the first instance where business citations have outpaced governmental sources.
Given the similarity between the local and regional business source usage in this study, gatekeeping theory might suggest that there is something distinct about the BIWF that is impacting the historical channels between journalists and sources . Considering that previous research into environmental topics and the use of sources did not find that business sources were most common, it may be the nature of the BIWF as the "first in the nation" that is influencing the newspaper source coverage.
There was a prevailing feeling of uncertainty present in the data, particularly in regards to the permitting process from government officials. Local government officials, the most prominent type of governmental source in local newspaper coverage, were repeatedly quoted commenting on unknowns: "Town Manager Nancy Dodge said too many questions remained unanswered, including whether the island could purchase electricity from the mainland grid through the wind farm's cable and sell excess energy back" ; and: "'I'm sure there are downsides we don't know about,' said John Warfel (town councilman). 'There are a lot of unknowns. We can't truly predict what's going to happen'" .
It stands to reason that in this circumstance, while an elected official remains an "elite" source, the overwhelming unknowns related to permitting an offshore wind project might make these traditional relationships less attractive ). It may be that in previous studies, the government officials literally knew more than any other source, leading to their source domination Smith & Norton, 2013). In this circumstance however, the data suggest that business representatives, particularly those from developers of wind farms, served the function of the "elite" source more often than governmental sources.
Additionally, while previous research that has suggested that the use of diverse sources increases credibility to readers, there continues to be a heavy reliance on "elite" sources Slovic, 1987;Cozma, 2006).
Again, the finding that business sources are the most common type of citation is present in both the local coverage, as noted above, and the regional coverage; this finding suggests that at least in relation to the BIWF, the developer of the wind farm, Deepwater Wind, held a powerful position in building the narrative about the project.
Whether this was an intentional tactic is a topic for future research.

Local vs. Regional Differences
Beyond the similar routines utilized by both local and regional newspapers in relying on for-profit sources, much of the data support previous research findings related to gatekeeping theory. Namely, that geographic difference in newspaper location appears to impact the coverage of a particular story . This research study does not propose to understand the broader mechanism underlying these differences, but in relation to the BIWF, there were statistically significant differences between the local and regional coverage related to the use of local experts, local government officials, and individuals. This suggests that continuing to differentiate between newspapers based on geographic proximity, particularly in relation to renewable energy projects, is meaningful. Salwen (2009) suggests, "the local media organization might turn out to have more influence in the outcome of an issue or event" (p. 105). Future research may attempt to determine if the differences in local and regional coverage is reflective of the geographical proximity to the sited offshore wind farm project or is more representative of the resources available to local reporters.
While the original research question of this study was supported, RQ1: Are there statistically significant differences in the people/organizations used as sources between the local and regional coverage of the BIWF, it is interesting to note that the data appear to emphasize the similarities in coverage rather than the differences when it comes to the BIWF.

Conclusion and implications
The development of the BIWF is an important research area where we can apply previously supported theoretical frameworks. Offshore wind energy, while not a new technology, continues to be novel in the United States: at the time of publication, the BIWF remains the only offshore wind farm in the United States.
As the data above suggest, there are interesting differences between the coverage of the BIWF and previous topics studied through gatekeeping theory. Most notably, the domination of the business sources across both local and regional coverage has surpassed governmental officials who had previously been most common.
In addition, it is always important to consider the sources who were not present in the coverage. Neither experts (i.e. scientists or researchers) nor non-profit organizations were major sources in the coverage of the BIWF. Additionally, perhaps surprisingly given the dependence of Rhode Island on tourism and recreation for the state economy, almost no tourists were cited in either the local or regional coverage.
Gatekeeping theory suggests this is because of the heavy dependence on traditional channels of information by journalists, and this also suggests that a particular story is less important than routine source relationships ).
The particular nuances of siting large renewable energy projects will continue to be a topic of importance assuming the United States continues to lean less heavily on energy produced by fossil fuels. The offshore wind industry, if one believes the industry publications and projections, is poised to begin historic growth in the United States. It is the intention of this study that the analysis of the newspaper coverage presented here will serve as a window into understanding the groups and individuals shaping the social discourse about offshore wind energy production.

Limitations and Future Research
A limitation of this study is that the regional coverage is inclusive of national newspaper coverage due to the low number of articles returned (n=24). A potentially meaningful follow-up study should investigate whether the reliance on business sources is as strongly reflected in the national coverage as in the local and regional (specifically New England geographically-situated newspapers) coverage.
It is also important to point out that gatekeeping theory, while originally theorized in the pre-electronic media age, has empirical support for continuing to be an appropriate theoretical frame through which to analyze newspaper coverage Hermida et al., 2014). Future studies may suggest adaptations, but there is strong experimental support for its continued use. "Only named or anonymous individuals that provided information or opinion in a direct quote, partial quote, or paraphrase were coded. Coders were instructed to ignore anonymous sources when they were used casually in collective attribution like "Engineers predicted" or "Several senior accident investigators say"" (Sumpter & Garner, 2007, p. 460) General Rules • Typically, anytime a sentence includes "said," "says," "noted," "stated," etc. it should be coded. • Anything in quotes will be coded, including references to a written document.
• A general sentence such as "Deepwater Wind said" will not be coded. Nor will sentences such as "according to a state press release" be coded. o A more specific individual must be responsible for a sentence in order to be coded. • If the same person is quoted multiple times and these quotes are separated by other sentences, quotes etc., that same person should be coded multiple times. o This provides frequency of singular source use. • In the case where an individual is introduced by numerous identities, the coder will need to determine the primary or dominant category; there will be no double coding. Parent codes are to remain mutually exclusive. o Also, each quote should be identified only by the information provided in that article; no knowledge from previous articles should be used in the coding process. • Quotes from different individuals from the same group will be coded as unique codes. • A quote from a lawyer will be coded as being from the company that they work for. • An individual who is introduced as a seasonal resident will be coded as a "local resident" not a "tourist."

All sub-codes must also be coded as the dominant code. There will be instances of quotes just being coded as the dominant code with no sub-code.
Example: Block Island resident Bryce Ye noted, "I haven't really seen sharks lately, but I know they are out there" would be coded as "LOCAL RESIDENT (sub-code) To be sure, these proposed projects have been met with opposition from certain groups who have a relationship to the areas being leased . In an Associated Press (AP) article published in late 2017, the headline read "Winds of worry: US fishermen fear forests of power turbines" (Marcelo). In the recreational and commercial fishing industry along the east coast of the United States, there is fear that expanding offshore wind farms will result in limited fishing grounds . Regulatory agencies and even the developers of offshore wind have attempted to address these fears by excluding specific habitats and fishing grounds, but uncertainty remains .
While the opposition to the Cape Wind farm was not orchestrated primarily by fishermen or union groups, the broader lesson learned was that unrelenting, wellfunded resistance can be effective at stalling development . Jim Gordon, the business leader behind the proposed Cape Wind farm, noted this about the opposition: "In a football game, if you have a tie, there's an overtime period, a sudden death period...We were kept in a repeated sudden death period, and the goal posts kept moving" . The Alliance to Protect Nantucket Sound, the opposition group supported by Senator Edward M. Kennedy and funded by William I. Koch, was determined in its argument that Nantucket Sound was not the place for an offshore wind farm . It was during these repeated negotiations that the BIWF marched, albeit slowly, through the permitting process eventually obtaining the first in the nation title. Both the failure of Cape Wind and the success of the BIWF will remain important historical markers for the offshore wind industry in the United States for decades to come.
Of interest in this study is the utilization of risk frames by local newspapers covering the proposal and subsequent processes of permitting these offshore wind Residents of Block Island, on the other hand, frequently paid some of the highest electricity costs in the nation due to their reliance on aging diesel generators . Additionally, installation of the underwater electrical cable, a necessity for the offshore wind farm, would connect Block Island to mainland Rhode Island for the first time. The cable held the promise of a fiber optic cable, which would increase internet speeds, allow for more stable connection for business owners, and lessen the occurrences of brownouts and electricity surges . For many, the cost of building such a connection cable without a company like Deepwater Wind fronting the cost would be impossible . In the same year that the BIWF was proposed (2008), local government official Barbara MacMullan was quoted as saying, "I do think that given the possibility that we may get a cable from this project should weigh heavily on any decision to invest ratepayer funds" .
The circumstances surrounding Cape Wind and the BIWF are unique; it is not the intention of this research to suggest otherwise. However, while acknowledging the disparate situations into which these two projects were proposed, comparing the local newspaper coverage in regards to the use of risk framing is meaningful. Given the geographical presence of offshore wind farms, analyzing local newspapers may present a more direct understanding of the risks as understood by the local citizens .

Local newspaper coverage
Neither Cape Cod nor Block Island are isolated communities; their local energy decisions impact neighboring communities. Previous research has suggested that in siting renewable energy projects, social acceptance can be a powerful barrier (Wüstenhagen, Wolsink, & Bürer, 2007;.
In the case of Cape Wind, the wind farm was sited for Nantucket Sound, and both Martha's Vineyard and Nantucket populations would have had their viewsheds impacted . And while Block Island was promised lower electricity rates, Narragansett, Rhode Island, the town across the ocean, was quite concerned about the potential increase in their electricity prices as a direct result of the offshore wind farm; this concern extended statewide .
The theoretical concept of NIMBY (not in my backyard), while cognitively attractive, has struggled to be supported empirically as an explanation for geographical differences in opinions about wind energy . It seems rational to expect that those closest to a proposed offshore wind farm, the turbines of which can reach over five hundred feet in the air, would be more opposed than those geographically farther away . In other words, people may have positive opinions of offshore wind energy, until confronted with a very real, potential project that affects them directly ). Yet, the "NIMBY theory has not been substantiated with empirical evidence as a major source of opposition" which suggests that there is need for additional research into local contexts and the reasons for opposition or support for renewable energy projects , p. 1372).
Outside of the NIMBY framework, research into newspaper representations of emerging renewable energy technologies has found substantial support for evaluating coverage in regards to a newspaper organization's proximity to the proposed site  Researchers emphasized that media coverage differed "partially related to regional attributes, needs, and the perceived appropriateness of CCS (carbon capture storage)" (Feldpausch-Parker et al., 2013, p. 347). That is to say, the specific local or state context impacts the newspaper coverage: "Perhaps there really is a better understanding of local readers' needs amongst local journalists because, unlike the nationals, local journalists are also local readers: crucially, are local" (Ross, 2007, p. 452).
The lack of a comprehensive framework to understand these geographical nuances makes clear the need for additional studies into local newspaper representation of renewable energy topics. Media research suggests that how a complex story is covered is reflected in the public understanding and conceptualization of that issue . Regardless of the circulation numbers of a newspaper organization or the clout provided to the journalists, the construction of a news article is intentional in what sources are used, what information is provided, and importantly, what information is omitted .
Local newspapers, occasionally referred to as "community" newspapers, are media organizations that provide services for a specific geographical area . While this definition has changed given the geographic boundlessness of the internet, for the purposes of this research, local and community will be used interchangeably. A local newspaper will be understood as a media organization that writes for a small, well-defined group of individuals . Given the distinct audience for local newspapers, there is an increased likelihood for daily coverage of stories directly impacting the community that might not be covered extensively in larger newspaper outlets. Research has also suggested that while there exists a more robust understanding of elite newspaper coverage, such as stories from The New York Times and The Washington Post, there is a dearth of research that has focused specifically on local, geographically based newspaper coverage .
We argue that given the uncertainties related to building the first offshore wind farm, local newspaper coverage, whether on Cape Cod or Block Island, would include discussion of risks. And while the United States has had power generated from onshore wind farms since the 1980s, offshore wind should be considered as a distinct and new (to the United States) technology . Priest and Ten Eyeck (2004) have suggested that stories focused on emerging technology are first framed as stories of risk at the local level which then inform how these topics are subsequently covered in elite, national newspapers . To explore the differences in coverage of Cape Wind and the BIWF, we turn to  adaptation of  theory of ecological communication.

Theoretical Framework
Stephens et al (2009)   . Essentially, rather than trying to understand the relationship between humans and the environment by evaluating physical interactions between these two systems, the communication within the human sphere is reflective of the underlying relationship .  titled these communicative activities "internal resonance" and noted that the process is ongoing and dynamic.
For the sake of this research, the adaptation first operationalized by  is the framework for this study.  adapted theoretical framework of Luhmann's theory posits that a newspaper story focused on an environmental topic, such as wind energy as a climate change mitigation tool, can be narrated in six distinct frames: technical, economic, environmental, health and safety, political, and aesthetic/cultural. Within each of these frames, either the risks or the benefits can be emphasized . Following previous media research, a frame here is defined as the "central organizing idea" used to make sense of a news event (Gamson & Modigliani, 1989, p Subsequent research  has applied this framework in studying state-level newspaper coverage of carbon capture technology  and the relationship between regional newspaper coverage about renewable energy stories and the introduction of bills in the Maine state legislature .  updated the codebook by including a "none" category for articles in which no frame was identifiable.

Stephens et al. (2009) adaptation of Luhmann's theory of ecological
communication is an appropriate lens through which to evaluate newspaper coverage of offshore wind because offshore wind as a technology is a deliberate mechanism through attempting to offset fossil fuel emissions . Renewable energy, at least in the United State's context, is understood as intimately connected to the environment ). Utilizing Stephen's theoretical adaptation will also help to test and apply this adapted framework to more cases, assessing its validity across contexts.
This article will focus on the following research questions:

Data Collection
Both the Cape Cod Times and The Block Island Times were selected for analysis based on their geographical proximity to a proposed offshore wind farm site (Massachusetts and Rhode Island, respectively) and their roles as community focused newspapers   (Smith & Norton, 2013;. In order to remain as consistent as possible in comparing the newspaper coverage of the two projects, the first five years after the initial proposal were selected to be analyzed. This follows previous research which has used important dates or meaningful events to inform the time period selected for analysis .  . Previous media research has utilized constructed week sampling as a more intentional sampling method that is reflective of the nature of newspapers to vary coverage based on the day of the week (Sunday vs. Monday coverage) (Riffe, Aust, & Lacy, 1993). Constructed week samples have varied in length with previous research having been conducted on two randomly constructed weeks including Saturday and Sunday coverage (Hester & Dougall, 2007;. Recent work has advised that in considering the impact of online newspaper content, constructed week sample sizes need to be larger (Hester & Dougall, 2007).

Coding Procedure
A four-week constructed sample, 28 days, was utilized to analyze the data set.
For each newspaper, four randomly selected dates were selected for each day of the week (inclusive of Saturday and Sunday). All articles published on that random day were included in the sample set. However, there were no instances of days on which multiple articles had been published, resulting in samples from both newspaper consisting of 28 articles.
Each article, the unit of analysis, was coded as having a dominant narrative structure reflective of one of the six frames provided by Luhmann's theory of ecological communication  or no frame . Within each of these frames, coders identified the article as portraying risks or benefits .
Coders were instructed to emphasize the title and leading paragraphs of an article in identifying the focus of the narrative . In the case of lengthy investigate pieces where multiple frames were utilized, coders were instructed to count paragraphs belonging to each frame to determine the dominant narrative structure .
To ensure the validity of the findings, intercoder reliability metrics were calculated following researchers' initial training and before the entirety of the sample set was coded . Due to the small sample size (n=56), 20% of the articles were selected for intercoder reliability testing ). The codebook achieved a Krippendorff's alpha of 1.0 between the two researchers.

Results
As previously noted in the methods section, a total of fifty-six articles, twentyeight articles per newspaper, were analyzed as a representative sample of the total data set (n=485). Only one article out of the four-week constructed sample was identified as having no frame. This article, published in The Block Island Times, was less than one hundred words and was accompanied by simulation photos of the proposed wind farm. Figure 1 shows a breakdown of the identified themes between the two newspapers as separated by risk and benefit frames. Overall, the Cape Cod Times utilized a risk frame in 68% of its articles. The Block Island Times, on the other hand, emphasized a risk frame in 43% of articles. Figure 2 shows the percentage of articles from each newspaper that utilized a particular theme (regardless of risk or benefit).

Figure 2. Overall frame results from Cape Cod Times and The Block Island Times
Again, this analysis focused on identifying the dominant frame of an article rather than the presence or lack thereof in any one newspaper article.

Research question 1 stated: Are there thematic differences in the use of risk/benefit narratives between coverage of the Cape Wind farm and the BIWF?
Due to the exploratory nature of this analysis, no statistical tests of significance were used to assess the data, rather emphasis was placed on trying to understand the context of the articles. The inclusion of percentages and frequency counts is in an attempt to provide as much of the data as possible. Interestingly, at first glance, the data seem to suggest that the coverage from the Cape Cod Times and The Block Island

Discussion
Given the different outcomes of the two proposed wind farms, it is perhaps unsurprising that the newspaper coverage also differed. The constructed week samples from both newspapers employed the same majority theme --politics --to cover their respective wind farms. What this suggests is that rather than emphasizing or focusing on the differences between the wind farm coverage, the similarities should also be explored. The data suggest that the process of trying to become the "first" offshore wind farm shaped the newspaper coverage. The following discussion will focus on two themes: political uncertainty, which was dominant in both newspapers; and the unique coverage from The Block Island Times about the underwater cable.

Political uncertainty
Perhaps the most apparent feeling within the entire constructed sample set was that of uncertainty. Even in articles that included the potential benefits of a future offshore wind farm, there were sections dedicated to the unknowns. For example, in an article published about the preemptive environmental monitoring Deepwater Wind was conducting prior to siting the wind farm, the article noted, "Spar buoys have been used for years as aids to navigation; however, this is the first buoy of its kind that will be used to collect wind data. It is still experimental" . In the same sentence, the journalist noted Deepwater Wind's acceptance of the due diligence related to building an offshore wind farm but explicated the potential possibility that the technology utilized might not be appropriate.
This uncertainty was presented in connection with the political frame, which was the most common theme, for risks and benefits across both newspapers. The domination of the political theme is distinct from previous work about onshore wind.
The uncertainty explicated above is reflective not only of Kerry's unknown position, but also of a potential relationship between the wind developer and state politicians.
Given that offshore wind development is a new industry in the United States, there are no established or historical rules. For the readers of the Cape Cod Times, this article suggests a level of ambiguity among the key players in the state.
It should be noted that previous research has found that media coverage of energy projects is likely to detail potential political risks for politicians, but this research vein has been confined to fossil fuel energy projects . This comparable finding in offshore wind coverage suggests that political backlash may be an inherent theme in covering energy development projects, regardless of the type of power production. In fact, a 2006 Cape Cod Times article summarizing the governor's race in Massachusetts was titled "Wind farm stance didn't hurt Patrick"  However, the definition of "local" in the case of the BIWF is complicated.
Nowhere is this clearer than in the split between the articles coded as technical risks versus the articles coded as technical benefits. All of the technical risk articles were focused on the proposed mainland landing sites of the underwater cable: first at the Narragansett town beach and then at the Scarborough State Beach. Opposition from Narragansett and Scarborough residents, who could be considered local residents in relation to the BIWF, was strong. Narragansett Town Council president James Callaghan is quoted as saying: You have to look at it from our perspective. You have to look at it as offensive that you want to come to the town beach. When it was first proposed, I was surprised. When you think about it, this is not the best for the town when it goes through our most precious resource .
There was an overall feeling that while Block Island residents may be benefitting from an underwater cable, the mainland residents of Rhode Island would be placed at a disadvantage; particularly because of the potential that the underwater cable could impact beach tourism. This dynamic between Block Island residents and local mainland Rhode Islanders remains contentious, now due however, to the increased electricity rates being paid by Rhode Island, in general due to costs associated with building the wind farm. Block Island electricity rates are expected to decline, but this is more reflective of the previously high power costs on the island rather than direct savings related to the BIWF . Island would truly need to consider, or it may be reflective of Cape Wind's true role as the first proposed wind farm in the nation. Cape Wind Associates was the first company to try to court a community to allow an offshore wind farm off its shores; there was no precedent on how to accomplish this goal. The coverage suggests that rather than offering something in conjunction with the wind farm, Cape Wind Associates tried to simply sell the project on its own merits. Deepwater Wind, potentially having learned from the Cape Wind project, combined the idea of a wind farm with a connection cable that Block Island desperately needed.

Conclusion
Ultimately, this article hopes to serve as an introductory analysis into the media coverage of offshore wind. Given the newness of the industry in the United States and the current momentum in the field, studying offshore wind is poised to become an interesting line of research. By applying previous theories to this new renewable energy project, the literature can become stronger and more useful.
What the findings of this research study suggest is that the process of permitting a new technology is not only complicated, but also news-making. For both the Cape Cod Times and The Block Island Times, the numerous steps and permitting victories and defeats overwhelmingly framed the media conversation during the first five years after each project was proposed. It also supports previous findings that there are meaningful differences in the newspaper coverage of different types of renewable energy projects and that these are influenced by location (i.e. state) .
This study also adds to the body of literature utilizing Luhmann's theory of ecological communication as adapted by  and suggests that multiple types of renewable energy projects are appropriate areas to apply this framework.

Limitations and Future Research
There are a number of limitations to this study. The sample size, while supported by previous literature, should be expanded in future studies to strengthen the findings. And as previously noted, the inclusion of multiple local newspapers may help to give additional insight into the broader local and regional media coverage.
Subsequent analysis should also allow for the coding of multiple frames in a single article. While this is a more complicated research method, this would allow for a more nuanced comparison between the coverage of different newspapers.