Tweeting in the midst of disaster: A comparative case study of journalists’ practices following four crises

Abstract

This comparative case study examines how local journalists used Twitter as a crisis communication tool during four emergency situations in the United States. The public’s retweeting and liking patterns also identified messages that resonated with them. A content analysis found that although local journalists used objective reporting most frequently across all crises, there were variances in Twitter practices of journalists covering the two human-made crises. The two natural disasters showed more similarities. These findings can help develop best-practices strategies for journalists as they cover different types of crises.

Keywords

community journalism reporting techniques Twitter social media crisis communication

As journalists continue to refine the ways in which they use social media to perform their work, a handful of activities consistently have been identified as effective: to convey objective information, engage in conversations, share opinions, and issue calls to action (Hermida et al., 2014; Lasorsa et al., 2011; Poell & Rajagopalan, 2015; Revers, 2014). This project explores the execution of those professional norms on a particular platform during four specific time periods. This comparative case study examines how local journalists used Twitter as a crisis communication tool in the first week after different emergency situations in various U.S. regions. Two crises were natural disasters: thousand-year flooding in West Virginia, and one of state history’s worst wildland-urban fires in Northern California. The other two crises were human-made: protests following the death of Michael Brown in Ferguson, MO, and a nightclub shooting in Orlando, FL, that killed 49.

Through these comparisons, it is possible to begin parsing how journalists cover different types of natural disasters and human-made crises on Twitter and which messages are most resonant with the public. By identifying these patterns, we can begin to develop best-practices strategies for journalists that are specific to different types of crises.

Almost three fourths of Twitter users get news on the platform (Shearer & Matsa, 2018) and nearly half (45%) go there specifically seeking news (Gottfried & Shearer, 2016). That desire for information is even more acute during times of turmoil, when social media use surges (Brown et al., 2012; Heverin & Zach, 2012; Wendling et al., 2013) and journalists are among the primary sources for reliable news (Blair, 2015; Fung, 2014). Journalists depend on professional routines such as objective reporting when covering crises on their traditional platforms (Mourão et al., 2018; Riegert & Olsson, 2007), although research has found that journalists also are more likely to voice their opinions on Twitter (Lawrence et al., 2013) as well as to have conversations with sources and the public (Revers, 2014) and issue calls to action that raise awareness of public issues (Poell & Rajagopalan, 2015). Tweeting during a crisis can challenge local journalists’ reliance on professional norms as they balance the demand for real-time reporting and sharing information about the crisis unfolding in their community.

By examining journalists’ tweets related to four types of crises that elicited varied community responses, it is possible to identify how the news workers used Twitter differently in each case, including the extent to which they were guided by certain professional tenets, as well as which types of messages were most resonant with the public through their retweets and favorites/likes. (Twitter switched from favorites to likes in late 2015, so tweets related to Ferguson received favorites and the tweets from other crises received likes.)

The value of using a comparative case study is to emphasize comparison within and across contexts (Goodrick, 2014). This study sought to understand and explain how certain contextual qualities might influence journalists’ use of Twitter in a crisis. With this information, we can begin to better understand the role journalists play in a crisis in the modern media age, and how citizens interpret and interact with the information journalists disseminate.

Background on the Crises in Ferguson and Orlando

Michael Brown’s death at age 18 became an international news story, but it initially garnered little news media attention after he was shot on August 9, 2014, by then-Ferguson police officer Darren Wilson. Twitter users broke the news of Brown’s death, which came to represent fractured race relations, police discrimination and brutality, and uneven processes in St. Louis’s multiple municipalities. That August, #Ferguson was the most popular hashtag in the world (Halimi, 2014). A few months later, in November, a grand jury voted not to indict Wilson. Federal investigators also examined Brown’s death and related issues: They did not find fault in Wilson’s actions, but they did identify discriminatory patterns by Ferguson police (Deere et al., 2015). Since Brown’s death, and deaths of several others in Georgia, Minnesota, New Mexico, New York, South Carolina, Texas, and elsewhere at the hands of law enforcement, police behavior first received greater scrutiny and then protests in hundreds of U.S. cities by June 2020.

Dozens of international and national news crews were in Ferguson in the week following Brown’s death, but follow-up coverage has been sparse at that level. Local journalists and their organizations received praise from the Columbia Journalism Review for their continuous coverage, aided by their “prior understanding of the many players involved and long-standing issues in play . . . [They] offered the deepest dive into the complexities of the Ferguson story” (D. Lee, 2014, para. 13).

Almost 2 years later, Omar Mateen killed 49 people on June 12, 2016, at the Pulse nightclub, a popular spot for the Orlando LGBTQ (lesbian, gay, bisexual, transgender, and queer) community. At the time, it was the worst mass shooting in U.S. history (Shapiro & Chan, 2016). Mateen’s specific motives were unclear, but various reports suggested mental illness, homophobia, and a connection to radical Islamic terrorism (Saraiya, 2016; Schneider, 2016). Journalists received conflicting details in the early hours of the crisis, especially during the standoff between police and Mateen from 2:00 a.m. to 5:00 a.m. As the timeline and victims’ names were released, reporters struggled because “no one was an expert when it came to covering a mass shooting and ISIS-inspired attack” (Hayes, 2016, para. 3).

Although the human-made crises studied here were both shootings, they were quite different and thus present case study comparisons that highlight the need for journalists to use distinct strategies when covering various crises via Twitter. Whereas the death of Brown—an African American teen—at the hands of a White police officer divided the community and gave rise to movements such as Black Lives Matter, the Orlando deaths seemed to unite their community in shared grief over a terrorist’s acts.

Background on the Crises in West Virginia and Northern California

When rain began falling in western West Virginia on June 23, 2016, it was the start of a series of storms that led to flash flooding that killed 23 residents and damaged thousands of homes and businesses. It was, as scientists described, a “one-in-a-thousand-year event” with some places receiving more than 1 ft of rain in a few hours—“the water drained down the mountains into the valleys where roads, homes and entire neighborhoods flooded” (Rice, 2016, para. 9). Journalists scrambled to cover the story in areas with poor wireless connectivity as roads washed out, homes were swept from their foundations, and county after county was declared a disaster area. For one local television reporter, the despair was so overwhelming that she sobbed in a live broadcast. Her response went viral, as did her subsequent letter to the community, explaining that she “had to be a human first . . . This flood—by no asking of our own—has ruined memories and towns and homes. It’s ruined belongings but it hasn’t ruined our spirit” (Spears, 2016, para. 15).

Local journalists were in much the same situation in October 2017, when several fires swept through Napa and Sonoma counties in Northern California. “How do we cover this when our community is burning?” asked photographer Kent Porter in a video posted on the Santa Rosa Press-Democrat website (Smith, 2018). The fires, fueled by high winds and dry vegetation, burned for days. At the time, it was the “worst wildland-urban cluster of fires in state history” (Fagan et al., 2017, para. 10) and the costliest in U.S. history (Rossman, 2017). Forty-three people were killed and more than 8,000 buildings were destroyed.

In West Virginia and Northern California, journalists faced the “risk and responsibility of community journalism” as they endeavored to cover the events (Smith, 2018). Although these were different types of natural disasters, homes and businesses in each area were decimated. Neighbors were killed. The entire community, including local news professionals, felt the suffering. Would journalists in each area cover their respective crises similarly on Twitter?

Local journalists in the four regions are the focus of this study because their coverage required significant shifts in their routines and additional resources to provide live reports and extended coverage. The crises in Ferguson, Orlando, West Virginia, and Northern California offer a unique opportunity to study journalists as they tweeted about times of turmoil in their communities. Furthermore, their ties as local residents raises questions about how they would cover a crisis happening in their community.

Twitter as a Journalistic Tool

Through Twitter, journalists can engage with the public on a one-to-one channel that is open in both directions at all times. Such direct, open access is unprecedented in the industry, and reporters are still navigating their relationships and discerning best practices for using Twitter and other social media platforms as journalistic tools.

Nearly 80% of U.S. journalists use Twitter and find it to be the most useful social media tool for their work (Oriella PR Network, 2013; Santana & Hopp, 2016). Relatedly, the more time that they spend on Twitter professionally, the more they value the social media platform (N. Y. Lee et al., 2016). In a survey of U.S. journalists, Willnat et al. (2017) found more than one third of news professionals (34.6%) spend between 30 and 60 min on social media every day gathering information and reporting stories.

Journalists rely on Twitter for four primary functions: disseminating objective information, engaging in conversations, posting opinions, and issuing calls to action (Hermida et al., 2014; Lasorsa et al., 2011; Poell & Rajagopalan, 2015; Revers, 2014; Willnat et al., 2017). They often use the platform to share objective information, such as live-tweeting events as well as posting updates and links to stories (Hermida et al., 2014; Willnat et al., 2017). Although journalists promote objective reporting in their “traditional” formats (Weaver et al., 2007) and on social media, they are more likely to post their opinions on Twitter than elsewhere in their reporting (Lasorsa et al., 2011). Journalists also interact with the public via Twitter, asking and answering questions (Revers, 2014). Occasionally, they use Twitter to issue specific calls to action, such as journalists in India who raised awareness about gender violence following a high-profile gang rape (Poell & Rajagopalan, 2015).

Journalists’ Use of Twitter as a Crisis Communication Tool

These emerging professional norms on social media provide journalists with a framework for covering crises, and they depend on those practices to guide their news-gathering and other activities. Journalists’ main goal during a crisis is to provide information to help people (Skinner, 2013). The public relies on journalists to help them make sense of events during times of unrest and as a result, the public also is more likely to view news professionals as “truth-tellers” if they occasionally depart from objective reporting to offer their own experiences and opinions during a crisis (Riegert & Olsson, 2007).

Research examining how news workers cover crises has found it centers on journalistic practices of sharing frequent factual updates on the situation that come from officials as well as the journalists’ own observations from the scene (Nee & Fusco, 2015; Riegert & Olsson, 2007; Skinner, 2013). Their news coverage follows classic crisis communication patterns—it is primarily one-way, pushing out updates to the public with details on how widespread the crisis is and what people can expect as it continues, as well as what is being done for crisis recovery. Journalists do occasionally engage in interactive communication on social media, responding to comments and sending questions to other users (Nee & Fusco, 2015).

Journalists must utilize social media platforms to ensure their objective voices remain part of the crisis narrative (Veil et al., 2011). The power of social media is that it contributes to the worldview of those who consume it; the social medium may not always build cohesion, but it can cultivate the spread of information and incite change (Breuer & Groshek, 2014).

Although little research examines journalists’ use of Twitter as a crisis communication tool, a handful of case studies provide some insight. In one research project that focused exclusively on Michael Brown’s death in Ferguson, the authors noted that professional routines like relying on police and government officials for information were standard operating procedure for journalists as they tweeted about events there (Blackstone et al., 2017). Another study examining a human-made crisis—a mass shooting in Australia—found that media professionals’ tweets focused on “conflict-oriented narratives . . . because conflict has an entrenched news value” (Kwon et al., 2019, p. 2668). An analysis of journalists’ social media accounts in the hours following a San Diego power outage concluded that journalists’ main roles were as aggregators and verifiers of information and as “facilitators of temporary communities online through the use of hashtags, questions and retweets . . . as well as adding value by crowd sourcing and keeping the conversation going” (Nee & Fusco, 2015, p. 207). Similarly, Takahashi et al. (2015) determined that journalists performed their professional roles to varying degrees during and after a typhoon in the Philippines, with the majority of their tweets providing objective reports, issuing calls to action in aid of relief efforts, and expressing their personal feelings and experiences. Other research suggests that the public seeks out information on social media to varying degrees, depending on the crisis type and size of the community (Graham et al., 2015).

The present study sought to better understand the differences in how journalists used Twitter following four crises. Thus, the first four research questions are as follows:

RQ1:How did the general Twitter practices differ between local journalists in the week following Michael Brown’s death and the Pulse Nightclub deaths?

RQ2:How did the general Twitter practices differ between local journalists in the week following flooding in West Virginia and fires in Northern California?

RQ3:How did professional norms of local journalists differ following the human-made crises?

RQ4:How did professional norms of local journalists differ following the natural disasters?

Sharing on Twitter: Retweets and Likes/Favorites

One in five U.S. adults use Twitter, and the information they receive is in part determined by the popularity of the tweet (Mitchell & Matsa, 2014; Hughes & Wojcik, 2019). Journalists make up about one quarter (24.6%) of all verified Twitter accounts, and they along with their media organizations had particularly powerful voices in the days following the crises studied here.

Retweets give Twitter users equal power to share information easily and broadly (Kwak et al., 2010). Twitter users’ primary incentive for accessing social media is to get information but they also want to share what they find with others, especially if it has the potential to help others (Berger & Milkman, 2012). Users find this social media action rewarding and valuable (Tan et al., 2014). boyd et al. (2010) posit that retweets represent more than a desire to share information—they reflect the sharer’s use of those posts to speak to their followers. Sharing volatile information during a crisis increases the likelihood of receiving exponential retweets, which amplifies the desire to engage in that behavior (Heverin & Zach, 2012). Alternately, liking or favoriting a tweet can be compared with an agreement or endorsement of the original post (Alhabash & McAlister, 2015; Coombs, 2014). Favoriting requires greater cognitive processing than retweeting because the former communicates support of the message (Alhabash & McAlister, 2015). Research also has found that having a greater number of favorites does not increase the likelihood of a message being retweeted, which suggests users see these features as separate functions (Alhabash & McAlister, 2015).

Conversations are specific to the individuals involved and often are not relevant for a majority of people. Retweeting and liking conversations takes them out of context for users’ followers, so it is expected that they are not retweeted or liked at high rates. It does not, however, mean Twitter conversations are not valuable to the people involved, and they can make up a significant portion of journalists’ jobs (Wasike, 2013). Being able to ask questions of journalists and get answers also helps assuage distress during a crisis, and at any time it can build credibility and brand loyalty to the journalist and their organization (Parmelee et al., 2019; Xu & Feng, 2014).

This study sought to better understand which messages were more salient with Twitter users in each crisis. Since retweets and favorites/likes by others indicate resonance of the message with the public, this study also asked the following:

RQ5:How did the retweeting and liking patterns by others of the journalists’ tweets differ following each human-made crisis?

RQ6:How did the retweeting and liking patterns by others of the journalists’ tweets differ following each natural disaster?

RQ7:Which professional norm messages were more likely to be retweeted and liked by others following each human-made crisis?

RQ8:Which professional norm messages were more likely to be retweeted and liked by others following each natural disaster?

Method

This study compared how journalists used Twitter in the week after Michael Brown’s death in August 2014 and the Orlando Pulse nightclub shooting in June 2016, and devastating flooding in West Virginia in June 2016 and deadly fires in Northern California in October 2017. Local news professionals were selected because of their preexisting ties to each community.

Initial lists of up to 45 local print and broadcast journalists were compiled based on those who tweeted about each crisis and related events in the first week after the onset of the emergency. From this list, the top 10 local journalists were selected based on the number of crisis-related tweets they produced during that week and their number of followers. Journalists’ tweets were accessed with ExportTweet.com, which provides a user’s most recent 3,200 tweets. ExportTweet also specifies whether each message was a retweet and the number of times it was liked and retweeted.

For Ferguson tweets, the top 10 local (five broadcast; five print) journalists generated 4,646 tweets during the first week of the crisis. A random sample based on a 99% confidence level was selected; 582 tweets were analyzed. In Orlando, 2,497 tweets were generated by the top 10 local journalists (seven broadcast; three print) during the first week. A random sample based on a 99% confidence level yielded 526 tweets. For West Virginia tweets, the top 10 local journalists (seven broadcast; three print) produced 685 tweets related to the first week of the flooding; the random sample based on a 99% confidence level yielded 338 tweets. Finally, the top 10 local journalists (six broadcast; four print) for the Northern California fires created 1,292 tweets during the first week. A random sample based on a 99% confidence level yielded 440 tweets.

One researcher and a research assistant participated in two 1-hr trainings, and each person coded 10% of the sample. Scott’s Pi was used to calculate intercoder reliability. All variables had sufficient intercoder reliability coefficients. (Scott’s Pi for hashtags: Ferguson 1.0; Orlando 1.0; West Virginia 0.88; Northern California 1.0. Scott’s Pi for retweets: Ferguson 1.0; Orlando 1.0; West Virginia 1.0; Northern California 1.0. Scott’s Pi for likes: Ferguson 1.0; Orlando 1.0; West Virginia 0.98; Northern California 1.0. Scott’s Pi for media: Scott’s Pi: Ferguson 0.97; Orlando 0.98; West Virginia 0.94; Northern California 1.0. Scott’s Pi for professional norms: Ferguson 0.77; Orlando 0.89; West Virginia 0.88; Northern California 0.86. Scott’s Pi for crisis location: Scott’s Pi: Ferguson 1.0; Orlando 1.0; West Virginia 1.0; Northern California 1.0.)

Tweet Variables to Answer RQs

Operational definitions were as follows: Number of Hashtags: The number of hashtags in each tweet; Retweets: Number of times a tweet was retweeted (further recoded as whether it was retweeted [1] or not [0]); Likes: Number of times a tweet was liked (further recoded as whether it was liked [1] or not [0]); Media: Whether the tweet had an external link or attached media (1) or not (0); Professional norms: Primary professional norm codes were guided by existing literature on journalists’ Twitter use (Hermida et al., 2014; Lasorsa et al., 2011) and developed from preliminary sample coding.

Coders selected one dominant norm for each tweet, based on the first sentence of the tweet: Objective/neutral reporting: Objective statement of what was occurring. It quoted opinion statements from others, such as from an interview or press conference. It also included debunking rumors and correcting errors. Opinion: Expressed explicit criticism of law enforcement, terrorists, city leaders, media or protestors, as well as racial divides. Positive tweets, such as support of the police, community leaders, and Muslims, also were included. Conversations: Response to inquiries or conversing with others. Tweet may start with @username of others or include a question/response clearly directed to another user. Call for action: Sought to mobilize engagement, directing people to follow particular journalists or read certain stories, as well as encouraging others to help/support/respond in a particular way. Others: Primary message did not fall into another category. Messages included tweets that only contained hashtags, mentions, or URLs. Crisis location: Whether the crisis was in Ferguson, MO (1), Orlando, FL (2), West Virginia, (3) or Northern California (4).

Results

RQ1 asked how general Twitter practices differed between local journalists in the week following the human-made crises of Michael Brown’s death and the Pulse Nightclub deaths. Table 1 provides a comprehensive illustration of general Twitter practices between the crises.

Table 1

General Twitter Practices Between Ferguson and Orlando

Twitter Practice	Ferguson	Orlando
Original tweets	64.4% (n = 375)	65.8% (n = 346)
Retweets	35.6% (n = 207)	34.2% (n = 180)
Media included	41.6%*** (n = 242)	54.2%*** (n = 285)
Mean number of retweets	173.99 (SD = 1,096.63)	154.78 (SD = 1,089.37)
Mean number of likes	104.36** (SD = 865.49)	6.03** (SD = 27.52)
Mean number of hashtags	0.77* (SD = 0.76)	0.66* (SD = 0.77)

*p < .05. **p < .01. ***p < .001.

Although original tweets were more frequent than retweets for both Orlando (65.8%, n = 346) and Ferguson (64.4%, n = 375), Ferguson tweets had a higher number of likes (M = 104.36, SD = 865.49) compared with Orlando tweets (M = 6.03, SD = 27.52), t(1,106) = 2.60, p < .01. Ferguson tweets (M = 0.77, SD = 0.76) also had a greater number of hashtags compared with Orlando (M = 0.66, SD = 0.77), t(1,106) = 2.36, p < .05, while media was more likely to be included for Orlando tweets (54.2%, n = 285) compared with Ferguson (41.6%, n = 242), χ²(1) = 17.59, p < .001.

RQ2 asked how general Twitter practices differed between local journalists in the week following flooding in the natural disasters in West Virginia and fires in Northern California. Original tweets were more frequent than retweets for both West Virginia (52.4%, n = 177) and California (63%, n = 277), and the differences were statistically meaningful, χ²(1) = 8.82, p < .01. California tweets had a significantly greater number of likes (M = 2.63, SD = 5.74) compared with West Virginia tweets (M = 1.71, SD = 6.04), t(706.13) = −2.15, p < .05; however, the number of retweets between West Virginia (M = 36.74, SD = 198.61) and California tweets (M = 28.30, SD = 86.72) was not different in a statistically meaningful way, p = .43.

West Virginia tweets (M = 0.43, SD = 0.78) had fewer hashtags compared with California tweets (M = 1.06, SD = 1.18), t(760.57) = 2.36, p < .001. Both crises more often than not included media in a statistically meaningful way, χ²(1) = 43.90. Z-score comparisons showed that West Virginia tweets were more likely to include media (73.7%, n = 249) than California tweets (50.2%, n = 221), p < .001. Table 2 provides a comprehensive illustration of general Twitter practices between the crises in West Virginia and Northern California.

Table 2

General Twitter Practices Between West Virginia and Northern California

Twitter Practice	West Virginia	Northern California
Original tweets	52.4% (n = 177)	63% (n = 277)
Retweets	47.6%** (n = 161)	37%** (n = 163)
Media included	73.7%*** (n = 249)	50.2%*** (n = 221)
Mean number of retweets	36.74 (SD = 198.61)	28.30 (SD = 86.72)
Mean number of likes	1.71* (SD = 6.04)	2.63* (SD = 5.74)
Mean number of hashtags	.43*** (SD = .78)	1.06*** (SD = 1.18)

*p < .05. **p < .01. ***p < .001.

RQ3 asked how the professional norms of local journalists differed following each human-made crisis. For Ferguson, objective reporting (73.9%, n = 430) appeared most frequently, followed by conversation (21%, n = 122), opinion (2.2%, n = 13), other (1.5%, n = 9), and calls for action (1.4%, n = 8). Objective reporting (84.4%, n = 444) appeared most frequently for Orlando tweets as well, followed by opinion (6.3%, n = 33), conversation (5.7%, n = 30), other (2.1%, n = 11), and calls for action (1.5%, n = 8).

Chi-square analysis with z-score comparisons found statistically significant differences between Ferguson and Orlando tweets, χ²(4) = 62.12, p < .001. Orlando tweets were more likely to include objective reporting (84.4%, n = 444) and opinion (6.3%, n = 33) compared with Ferguson (objective reporting: 73.9%, n = 430, opinion: 2.2%, n = 13). Meanwhile, Ferguson tweets were more likely to have conversations (21%, n = 122) compared with Orlando (5.7%, n = 30). Table 3 illustrates the cross-tabulation of professional norms between the two crises.

Table 3

Cross-Tabulation of Professional Norms Between the Two Man-Made Crises

Professional Norms	Ferguson	Orlando
Objective reporting	73.9% (n = 430)	84.4%^a (n = 444)
Opinion	2.2% (n = 13)	6.3% (n = 33)
Conversations	21% (n = 122)	5.7%^a (n = 30)
Call for action	1.4% (n = 8)	1.5% (n = 8)
Other	1.5% (n = 9)	2.1% (n = 11)
Total	100% (n = 582)	100% (n = 526)

Chi-square analysis with z-score comparisons found statistically significant differences between Ferguson and Orlando tweets for objective reporting and conversations, χ²(4) = 62.12, p < .001.

RQ4 asked how professional norms of local journalists differed following the natural disasters. For West Virginia, objective reporting (84.6%, n = 286) appeared most frequently, followed by conversation (5.9%, n = 20), opinion (4.4%, n = 15), calls for action (3.8%, n = 13), and other (1.2%, n = 4). Identical patterns appeared for California tweets, with objective reporting (81.4%, n = 358) appearing most frequently, followed by conversation (8.9%, n = 39), opinion (5.7%, n = 25), calls for action (3.9%, n = 17), and other (0.2%, n = 2).

Chi-square analysis with z-score comparisons found no statistically significant differences in the ways in which professional norms were practiced between West Virginia and California tweets, χ²(4) = 5.73, p = .22. Table 4 illustrates the cross-tabulation of professional norms between the two crises.

Table 4

Cross-Tabulation of Professional Norms Between the Two Natural Crises

Professional Norms	West Virginia	Northern California
Objective reporting	84.6% (n = 286)	81.4% (n = 358)
Opinion	4.4% (n = 15)	5.7% (n = 25)
Conversations	5.9% (n = 20)	8.9% (n = 39)
Call for action	3.8% (n =13)	3.9% (n = 17)
Other	0.2% (n = 2)	0.2% (n = 2)
Total	100% (n = 338)	100% (n = 440)

Note. Chi-square analysis with z-score comparisons found no statistically significant differences between professional norms in the West Virginia and California tweets, χ²(4) = 5.73, p = .22.

RQ5 asked how the retweeting and liking patterns by others differed following each shooting. The two crises showed similarities as well as differences. Both Orlando tweets (81.6%, n = 429) and Ferguson tweets (83.8%, n = 488) had high “retweetability” and high likability (Orlando: 50.6%, n = 266; Ferguson tweets: 79.6%, n = 463). However, only Ferguson tweets were more likely to get liked compared with Orlando tweets, χ²(1) = 103.13, p < .001, and had a higher number of likes (M = 104.36, SD = 865.49) compared with Orlando tweets (M = 6.03, SD = 27.52), t(1,106) = 2.60, p < .01. There were no statistically significant differences for the number of retweets between Orlando (M = 154.78, SD = 1,089.37) and Ferguson tweets (M = 173.99, SD = 1,096.63), p = .77.

RQ6 asked how the retweeting and liking patterns by others differed following each natural disaster. Both West Virginia tweets (82.5%, n = 279) and California tweets (76.4%, n = 336) had high “retweetability,” χ²(1) = 4.41, p < .05. Interestingly, both crises had more non-likes than likes for their tweets (West Virginia: 63.9%, n = 216; California: 56.9%, n = 230). However, the difference between whether a tweet was liked or not was not statistically meaningful, χ²(1) = 3.73, p = .053.

Regarding the number of retweets and likes for each crisis, California tweets had more likes (M = 2.63, SD = 5.74) compared with West Virginia tweets (M = 1.71, SD = 6.04), t(706.13) = −2.15, p < .05. There were no statistically significant differences for the number of retweets between West Virginia (M = 36.74, SD = 198.61) and California tweets (M = 28.30, SD = 86.72), p = .43.

RQ7 asked which professional norm messages were more likely to get retweeted and liked by others following each shooting. Only the findings with statistical differences are reported here. For Ferguson, objective reporting (97.2%, n = 488) was more likely to get retweeted than not (2.8%, n = 12), whereas conversation tweets were more likely to not get retweeted (63.1%, n = 77) than to be retweeted (36.9%, n = 45), Fisher’s exact (30% of cells have expected count less than 5) = 248.13, p < .001.

In terms of likes, objective reporting (88.8%, n = 420) was more likely to get liked than not (11.2%, n = 48), whereas conversation tweets (absent: 51.6%, n = 63; present: 48.4%, n = 59) and call to action tweets (absent: 62.5%, n = 5; present: 37.5%, n = 3) were less likely to get liked, Fisher’s exact = 95.81, p < .001.

For Orlando, objective reporting (present: 86.3%, n = 383; absent: 13.79%, n = 61) and opinion (present: 60.6%, n = 20; absent: 39.4%, n = 13) were more likely to get retweeted than not. However, conversation tweets were more likely to not get retweeted (63.3%, n = 19) than to get tweeted (36.7%, n = 11), Fisher’s exact = 48.55, p < .001.

Also for Orlando, objective reporting was less likely to get liked (53.2%, n = 236) than to get liked (46.8%, n = 208), whereas opinions were more likely to get liked (90.9%, n = 30) than not (9.1%, n = 3), Fisher’s exact = 28.80, p < .001. Table 5 provides an overview for RQ7.

Table 5

Retweets and Likes for Professional Norms Between Cities

	Crisis city
	Ferguson				Orlando
	Retweeted		Liked		Retweeted		Liked
Professional norms	n	%	n	%	n	%	n	%
Objective	488	97.2^a	420	88.6^b	383	86.3^c	208	46.8^d
Opinion	12	92.3	13	100	20	60.6^c	30	90.9^d
Conversations	45	36.9^a	59	48.4^b	11	36.7^c	15	50
Call to action	6	75	3	37.5^b	8	100	6	75
Other	7	77.8	6	66.7	7	63.6	7	63.6

Note. Table only displays cases when retweeting and liking were present.

Ferguson retweets: Objective reporting was more likely to get retweeted than not. Conversations were more likely to not get retweeted than to be retweeted (Fisher’s exact = 248.13, p < .001). ^b Ferguson likes: Objective reporting was more likely to get liked than not. Conversation and call to action tweets were less likely to get liked (Fisher’s exact = 95.81, p < .001). ^c Orlando retweets: Objective reporting and opinion were more likely to get retweeted than not. Conversations were more likely to not get retweeted than to get tweeted (Fisher’s exact = 48.55, p < .001). ^d Orlando likes: Opinions were more likely to get liked than not. Objective reporting was less likely to get liked (Fisher’s exact = 28.80, p < .001).

RQ8 asked which professional norm messages were more likely to be retweeted and liked by others following each natural disaster. For West Virginia, objective reporting (89.2%, n = 255) was more likely to get retweeted than not (10.8%, n = 31), whereas conversation tweets were more likely to not get retweeted (90%, n = 18) than to get retweeted (109%, n = 2), Fisher’s exact = 65.68, p < .001. In terms of the likes, objective reporting was less likely to get liked (69.6%, n = 199) than to get liked (30.4%, n = 87), whereas opinions and calls for action were more likely to get liked (80%, n = 12; 76.9%, n = 10) than not (20%, n = 3; 23.1%, n = 3), Fisher’s exact = 31.81, p < .001.

Identical patterns appeared for California tweets: Objective reporting (83.8%, n = 300) was more likely to get retweeted than not (16.2%, n = 58), whereas conversation tweets were more likely to not get retweeted (present: 10.3%, n = 2; absent: 89.7%, n = 35), Fisher’s exact = 90.10, p < .001. In terms of the likes, objective reporting was less likely to get liked (39.8%, n = 135) than to get liked (60.2%, n = 204), whereas opinions and calls for action were more likely to get liked (90%, n = 18; 90.9%, n = 10) than not (10%, n = 2; 9.1%, n = 1), Fisher’s exact = 33.85, p < .001. Table 6 provides an overview for RQ8.

Table 6

Retweets and Likes for Professional Norms Between Cities

	Crisis area
	West Virginia				Northern California
	Retweeted		Liked		Retweeted		Liked
Professional norms	n	%	n	%	n	%	n	%
Objective	255	89.2^a	87	30.4^b	300	83.8^c	135	39.8^d
Opinion	10	66.7	12	80^b	17	68	18	90^d
Conversations	2	10^a	9	45	4	10.3^c	10	30.3
Call to action	10	76.9	10	76.9^b	14	82.4	10	90.9^d
Other	2	50	4	100	1	100	1	100

Note. Table only displays cases when retweeting and liking were present.

West Virginia retweets: Objective reporting was more likely to get retweeted than not. Conversations were more likely to not get retweeted than to get retweeted (Fisher’s exact = 65.68, p < .001). ^b West Virginia likes: Opinion and calls for action were more likely to get liked than not. Objective reporting was less likely to get liked than to get liked (Fisher’s exact = 31.81, p < .001). ^c Northern California retweets: Objective reporting was more likely to get retweeted than not. Conversations were more likely to not get retweeted than to get retweeted (Fisher’s exact = 90.10, p < .001). ^d Northern California likes: Opinion and calls for action were more likely to get liked than not. Objective reporting was less likely to get liked than to get liked (Fisher’s exact = 33.85, p < .001.).

Discussion

The primary goals of this comparative case study were to determine whether local journalists covering these four crises engaged in unique tweeting patterns, which types of tweets were most resonant with the public, and what these findings suggest for how journalists should approach using Twitter as a crisis communication tool.

Journalists in all four places (Ferguson, Orlando, West Virginia, Northern California) produced more original tweets rather than retweeting others, which aligns with normative practices in the profession to create their own reports based on what they witness and the sources they interview.

One possible explanation for why journalists covering Ferguson used more hashtags than Orlando journalists is that #Ferguson quickly evolved to represent much more than the shooting death of an African American teen by a White police officer. #Ferguson was the world’s most popular hashtag in August 2014 (Halimi, 2014), and Ferguson journalists may have needed additional hashtags to signal the specific focus of their tweets.

In the Orlando crisis, journalists included more media in their tweets than their Ferguson counterparts, usually either with links to their stories or attached photos. In interviews 1 year after Brown’s death, local journalists, activists, and government officials all stressed the crucial function of including visual evidence in their tweets—“Quite simply: photos and videos were the currency of credibility in this crisis” (Hinsley et al., 2016, p. 18). The Orlando shooting was nearly 2 years later, and scattered evidence suggests journalists were increasing their use of photos and videos in tweets during that time (Molyneux & Mourão, 2019; Mortensen et al., 2017).

Journalists covering the Northern California fires more often produced their own tweets rather than retweeting others, compared with the West Virginia journalists, who more often included photos or videos in their tweets. Differences in staffing, geography, and the events themselves might explain these differences, but no direct evidence was gathered on those points. Again, this study examined differences in the professional norms practiced by the various journalists. Orlando news workers were more likely than their Ferguson counterparts to include objective reports and opinions in their tweets because what happened was quickly settled. But Ferguson journalists were more likely to engage in Twitter conversations than Orlando reporters because, simply, Ferguson journalists had more conversations to ask and answer more questions.

Identical patterns of demonstrating professional norms being found in the tweets from West Virginia and Northern California means both groups relied on objective reporting, followed by engaging in conversations, sharing opinions, and issuing calls to action—a shared understanding of how journalists are expected to cover natural disasters. Journalists recognized their primary function was informing the public, but also understood that responding to questions helped assuage public uncertainty and fear.

Objective tweets from Ferguson being more likely than not to be retweeted and liked is connected to the public’s desire for information during turmoil, which fuels their motivation to share it (Heverin & Zach, 2012). But retweets and likes mean different things to users. The public retweets to share information they find helpful and as a way to “speak” to their followers (boyd et al., 2010; Heverin & Zach, 2012). Users tend to view likes (favorites in 2014) as endorsements of the tweet (Alhabash & McAlister, 2015; Coombs, 2014).

The Ferguson journalists’ call-to-action tweets were less likely to receive likes, and their messages perhaps reveals why: Most of the eight tweets merely directed people to follow other journalists or check out a linked story. By comparison, journalists’ call-to-action posts during and after a typhoon were composed the majority of tweets and helped coordinate relief efforts (Takahashi et al., 2015).

Tweet conversations related to Ferguson also were less likely to receive retweets or likes, and conversations between Orlando journalists and other Twitter users also were less likely to get retweeted. These results are consistent with Wasike (2013), Parmelee et al. (2019), and Xu and Feng (2014).

Orlando opinion tweets were more likely to be retweeted and liked than not, signaling that the public in situation seemed amenable to opinions. After initial confusion about the shooting, Orlando crisis tweets quickly coalesced around messages of unity and healing for the community and perhaps journalists occasionally crossing over from objective reporting to sharing their opinions in this way were seen as truth-tellers aligned with the audience, as in the Riegert and Olsson (2007) study.

In Orlando, opinion tweets had strong resonance, but objective ones were split: again, objective tweets were more likely to be retweeted than not, but were not more likely to be liked. Followers may have been unwilling to “like” such awful news, especially about named persons who had been killed.

In West Virginia and Northern California, again, objective tweets were more likely to get retweeted than not, but that was not the case for likes, possibly again because followers sought to share information they found helpful (boyd et al., 2010; Heverin & Zach, 2012) but they did not want to be seen as “liking” it. As in Orlando and Ferguson, conversation tweets were not likely to get retweeted or liked since doing so would have taken them out of context.

Opinion tweets from the West Virginia and Northern California journalists, as well as tweets with calls to action, were more likely than not to receive retweets and likes from Twitter users, probably because these disasters did not divide the communities or raise issues for journalists.

Conclusion

Findings here are specific to local journalists in these crises, but highlight the need for journalists to consider the crisis type when developing their strategy for using Twitter as a crisis communication tool. Again, tweets related to the Pulse shooting in Orlando had more in common with the natural disaster tweets than with the other human-made crisis. Clearly, an act of terrorism tends to elicit feelings of devastation and unity that are more present with natural disasters, and this research suggests that a “one-size-fits-all” approach cannot be applied when using Twitter as a crisis communication tool.

Objective information is a priority focus in the tweets studied here, and previous research shows such information dominates journalists’ feeds even during noncrisis periods (Lasorsa et al., 2011). Objective information had high “retweetability” during these crises, which reinforces the public’s desire to share helpful information. But the public also is wary of implying support for a crime like terrorism or for disasters that devastated their communities by clicking “like.” In the absence of Twitter revising its choices for user reactions, news organizations should consider the crisis type before emphasizing to reporters a preference for receiving retweets over likes, or vice versa.

Another caution for news organizations and journalists is the use of opinion tweets related to a crisis: Opinion tweets help the audience identify with journalists, but news workers must understand public sentiment regarding the crisis and consider how their credibility would be affected if they are seen as violating that sentiment. The public generally feels greater kinship with its local journalists—broadcast journalists in particular because they are more recognizable—and residents expect them to know the community and its standards.

Finally, news organizations should not underestimate the importance of encouraging their journalists to use Twitter as a conversation tool during a crisis. Although the value of that activity is not reflected in retweets and likes, a higher level of community engagement is demonstrated when journalists ask questions of their followers during a crisis and when the news professionals respond to inquiries. By helping to lessen the anxiety and uncertainty that residents experience during times of turmoil, journalists foster goodwill that can manifest as users’ increased loyalty to the news organization—a long-term dividend (Xu & Feng, 2014).

Journalists must utilize social media to ensure their relevance in a crisis (Veil et al., 2011) because social media can spread information and incite change, if not build cohesion (Breuer & Groshek, 2014). Therefore, this project’s findings point to how news organizations can be more effective when developing their social media strategies for responding to crises in their communities: Their first priority should be to relay frequent, factual updates, and even repeating information to ensure it does not get drowned out in the volumes of tweets related to the crisis. Second, it is crucial to have staff dedicated to answering questions on social media to help alleviate concerns about the disaster, and finally, it is acceptable to use personal commentary as a way to connect with shared suffering in the community.

Limitations and Further Research

This study utilized broad-based categories for coding professional norms. This approach allowed a macro-level examination of the professional norm differences between the crises. But in doing so, it compromises nuanced differences within the subcategories of the coding scheme. Future research may expand categorization of messages within the professional norms, such as the types of objective information being shared (e.g., focus on the victims, response from officials, correcting false reports, and rumors) as well as whether calls to action centered on bolstering community resilience or merely promoted the journalists’ work.

While this study included multiple, significant crisis cases, they were nevertheless situated in the context of human-made crises related to shooting deaths, as well as two instances of natural disasters. Therefore, the type of the crisis and geographic location may have affected the journalistic practices on Twitter. Future research should include more crises, other types of crises, and in other areas, to further identify best practices for using Twitter as a crisis communication tool.

Finally, in-depth interviews with the local journalists from the crisis areas can offer rich explanations of their Twitter practices during crises (including but not limited to whether they are indeed mindful of the specificities of the crisis when covering breaking news), which may or may not be consistent with explanations found in the various relevant studies.

Footnotes

ORCID iD

Amber Hinsley

References

Alhabash

McAlister

A. R.

(2015). Redefining virality in less broad strokes: Predicting viral behavioral intentions from motivations and uses of Facebook and Twitter. New Media & Society, 17(8), 1317–1339. https://doi.org/10.1177/1461444814523726

Berger

Milkman

K. L.

(2012). What makes online content go viral? Journal of Marketing Research, 49(2), 192–205.

Blackstone

G. E.

Cowart

H. S.

Saunders

L. M.

(2017). TweetStorm in #ferguson: How news organizations framed dominant authority, anti-authority, and political figures in a restive community. Journal of Broadcasting & Electronic Media, 61(3), 597–614.

Blair

(2015, August 22). All in on covering Ferguson: Huffington Post Q&A. American Journalism Review. http://ajr.org/2014/08/22/covering-ferguson-huffington-post-q/

boyd

Golder

Lotan

(2010, January). Tweet, tweet, retweet: Conversational aspects of retweeting on Twitter [Conference session]. 43rd Hawaii International Conference on System Sciences, Kauai, HI, United States.

Breuer

Groshek

(2014). Online media and offline empowerment in post-rebellion Tunisia: An analysis of internet use during democratic transition. Journal of Information Technology & Politics, 11(1), 25–44.

Brown

Guskin

Mitchell

(2012). The role of social media in the Arab uprisings. https://www.journalism.org/2012/11/28/role-social-media-arab-uprisings/

Coombs

W. T.

(2014). Ongoing crisis communication: Planning, managing, and responding. Sage.

Deere

Raasch

Kohler

(2015, March 5). DOJ finds Ferguson targeted African-Americans, used courts mainly to increase revenue. St. Louis Post-Dispatch. https://www.stltoday.com/news/local/crime-and-courts/doj-finds-ferguson-targeted-african-americans-used-courts-mainly-to-increase-revenue/article_d561d303-1fe5-56b7-b4ca-3a5cc9a75c82.html

10.

Fagan

Tucker

Johnson

Fimrite

Lang

Alexander

Mobley

Cabanatuan

(2017, October 15) A fire’s first, fatal hours. San Francisco Chronicle. https://www.sfchronicle.com/bayarea/article/Wine-Country-fires-first-fatal-hours-12278092.php

11.

Fung

(2014, August 14). How social media freed reporters Wesley Lowery and Ryan Reilly from Ferguson Police. Washington Post. https://www.washingtonpost.com/news/the-switch/wp/2014/08/14/how-social-media-freed-reporters-wesley-lowery-and-ryan-reilly-from-ferguson-police/

12.

Goodrick

(2014). Comparative case studies, methodological briefs: Impact evaluation 9. UNICEF Office of Research.

13.

Gottfried

Shearer

(2016). News use across social media platforms 2016. Pew Research Center. http://www.journalism.org/2016/05/26/news-use-across-social-media-platforms-2016/

14.

Graham

M. W.

Avery

E. J.

Park

(2015). The role of social media in local government crisis communications. Public Relations Review, 41, 386–394. https://doi.org/10.1016/j.pubrev.2015.02.001

15.

Halimi

(2014). Top Twitter hashtags of the last 28 days. http://blog.similarweb.com/top-twitter-hashtags-of-the-last-28-days/

16.

Hayes

(2016, December 20). The Pulse nightclub shooting taught me to question everything, even what I heard and saw. Poynter. http://www.poynter.org/2016/the-pulse-nightclub-shooting-taught-me-to-question-everything-even-what-i-heard-and-saw/443040/

17.

Hermida

Lewis

S. C.

Zamith

(2014). Sourcing the Arab Spring: A case study of Andy Carvin’s use of sources on Twitter during the Tunisian and Egyptian revolution. Journal of Computer-Mediated Communication, 3(19), 479–499.

18.

Heverin

Zach

(2012). Use of microblogging for collective sense-making during violent crises: A study of three campus shootings. Journal of the American Society for Information and Science Technology, 1(63), 34–47.

19.

Hinsley

Lee

Blank

Wray

Sandoval

J. S. O.

Jupka

Cioni

(2016, August). Asserting credibility in a crisis: How journalists, activists and police/government officials used Twitter during Ferguson [Conference session]. Association for Education in Journalism and Mass Communication National Conference, Minneapolis, MN, United States.

20.

Hughes

Wojcik

(2019, Aug. 2). 10 facts about Americans and Twitter. Pew Research Center. https://www.pewresearch.org/fact-tank/2019/08/02/10-facts-about-americans-and-twitter/

21.

Kwak

Lee

Park

Moon

(2010, April). What is twitter, a social network or a news media? [Conference session]. Proceedings of the 19th International Conference on World Wide Web, WWW’ 10, Raleigh North, CA, United States.

22.

Kwon

K. H.

Chadha

Wang

(2019). Proximity and Networked News Public: Structural topic modeling of global twitter conversations about the 2017 Quebec mosque shooting. International Journal of Communication, 13, 2652–2675.

23.

Lasorsa

Lewis

S. C.

Holton

(2011). “Normalizing” Twitter: Journalism practice in an emerging communication space. Journalism Studies, 1(13), 19–36.

24.

Lawrence

Molyneux

Coddington

Holton

(2013). Tweeting conventions. Journalism Studies, 6(15), 789–806.

25.

Lee

(2014, August 20). In Ferguson, local news coverage shines (para. 13). Columbia Journalism Review. http://www.cjr.org/united_states_project/local_coverage_ferguson_michae.php?page=all

26.

Lee

N. Y.

Kim

Sang

(2016). How do journalists leverage Twitter? Expressive and consumptive use of Twitter. The Social Science Journal, 54(2), 139–147.

27.

Mitchell

Matsa

(2014, March 26). 8 Key takeaways about social media and news. Pew Research Center. http://www.journalism.org/2014/03/26/8-key-takeaways-about-social-media-and-news/

28.

Molyneux

Mourão

R. R.

(2019). Political journalists’ normalization of Twitter. Journalism Studies, 20(2), 248–266.

29.

Mortensen

T. M.

Hull

Boling

K. S.

(2017). Really social disaster: An examination of photo sharing on Twitter during the #SCFlood. Visual Communication Quarterly, 24(4), 219–229.

30.

Mourão

R. R.

Kilgo

D. K.

Sylvie

(2018). Framing Ferguson: The interplay of advocacy and journalistic frames in local and national newspaper coverage of Michael Brown. Journalism. Advance online publication. https://doi.org/10.1177/1464884918778722

31.

Nee

Fusco

(2015). Tweets during crisis follow one-way communication. Newspaper Research Journal, 36(2), 197–211.

32.

Oriella

Network. (2013). “The new normal for news,” Oriella Digital Journalism Study. http://www.oriellaprnetwork.com/research

33.

Parmelee

J. H.

Roman

Beasley

Perkins

S. C.

(2019). Gender and generational differences in political reporters’ interactivity on Twitter. Journalism Studies, 20(2), 232–247.

34.

Poell

Rajagopalan

(2015). Connecting activists and journalists: Twitter communication in the aftermath of the 2012 Delhi rape. Journalism Studies, 16(5), 719–733. https://doi.org/10.1080/1461670X.2015.1054182

35.

Revers

(2014). The Twitterization of news making: Transparency and journalistic professionalism. Journal of Communication, 5(64), 806–826.

36.

Rice

(2016, June 27). Why the W.Va. floods were so deadly and destructive. USA Today. https://amp.usatoday.com/story/weather/2016/06/27/west-virginia-floods-storm-train/86429020/

37.

Riegert

Olsson

(2007). The importance of ritual in crisis journalism. Journalism Practice, 2(1), 143–158.

38.

Rossman

(2017, October 31) Tubbs, Nuns, Pocket fires fully contained in Sonoma and Napa counties. Santa Rosa Press Democrat. https://www.pressdemocrat.com/news/7584410-181/nuns-fire-contained-burned-more

39.

Santana

A. D.

Hopp

(2016). Tapping into a new stream of (personal) data: Assessing journalists’ different use of social media. Journalism & Mass Communication Quarterly, 93(2), 383–408.

40.

Saraiya

(2016, June 12). Media coverage of Orlando tragedy follows familiar scripts. Variety. http://variety.com/2016/tv/opinion/media-coverage-orlando-shooting-networks-news-1201793981/

41.

Schneider

(2016, June 15). Correction: Nightclub shooting-Florida story. Associated Press. http://bigstory.ap.org/article/435dd4b47ec7403888f74e9c5c3480b5/orlando-police-multiple-injuries-after-nightclub-shooting

42.

Shapiro

Chan

(2016, June 12). What to know about the Pulse nightclub shooting in Orlando. Time. http://time.com/4365260/orlando-shooting-pulse-nightclub-what-know/

43.

Shearer

Matsa

(2018). News use across social media platforms 2018. Pew Research Center. https://www.journalism.org/2018/09/10/news-use-across-social-media-platforms-2018/

44.

Skinner

(2013). Natural disasters and Twitter: Thinking from both sides of the tweet. First Monday, 18(9), 1–7.

45.

Smith

(2018, January 6) Video: Press Democrat’s Kent Porter about covering the Northern California fires. Santa Rosa Press Democrat. https://www.pressdemocrat.com/news/7840010-181/chris-smith-kent-porters-fire?sba=AAS

46.

Spears

(2016, June 30). Broadcast journalist: We are people 1st, reporters 2nd. WOWK-TV. https://www.wowktv.com/archives/broadcast-journalist-we-are-people-1st-reporters-2nd/865489662

47.

Takahashi

Tandoc

E. C.

Jr. Carmichael

(2015). Communicating on Twitter during a disaster: An analysis of tweets during Typhoon Haiyan in the Philippines. Computers in Human Behavior, 50, 392–398.

48.

Tan

Lee

Pang

(2014). The effect of wording on message propagation: Topic- and author-controlled natural experiments on Twitter. Paper presented at the 52nd Annual Meeting of the Association for Computational Linguistics, at Baltimore, MD, June.

49.

Veil

S. R.

Buehner

Palenchar

M. J.

(2011). A work-in-process literature review: Incorporating social media in risk and crisis communication. Journal of Contingencies & Crisis Management, 19(2), 110–122. https://doi.org/10.1111/j.1468-5973.2011.00639.x

50.

Wasike

(2013). Framing news in 140 characters: How social media editors frame the news and interact with audiences via Twitter. Global Media Journal: Canadian Edition, 6(1), 5–23.

51.

Weaver

D. H.

Beam

R. A.

Brownlee

B. J.

Voakes

P. S.

Wilhoit

G. C.

(2007). The American journalist in the 21st century: U.S. news people at the dawn of a new millennium. Lawrence Erlbaum.

52.

Wendling

Radisch

Jacobzone

(2013). The use of social media in risk and crisis communication. Organisation for Economic Co-operation and Development.

53.

Willnat

Weaver

Wilhoit

G. C.

(2017). The American Journalist in the digital age: A half-century perspective. Peter Lang.

54.

W. W.

Feng

(2014). Talking to the broadcasters on Twitter: Networked gatekeeping in Twitter conversations with journalists. Journal of Broadcasting & Electronic Media, 58(3), 420–437.