Abstract
To successfully manage a crisis in sports, it is necessary to examine the public’s emotions and other emotion-related responses. Without this knowledge, an organization will likely be disconnected from the needs of the public. Inspired by the key constructs in the integrated crisis mapping model, this study examined emotion, coping, and responsibility attribution through a content analysis of tweets during the Larry Nassar scandal involving Michigan State University and USA Gymnastics. This study analyzed a sample of 3,088 tweets generated by the public several days following the sentencing of Larry Nassar. The findings had both theoretical and practical implications. Theoretically, the findings provided an extension of emotions examined and suggested ways in which the model could be expanded to accommodate varied crisis situations. Furthermore, the study revealed important findings regarding levels of attributions and the relationship of the public’s emotions and attribution of responsibilities. On a practical level, the findings offered tangible suggestions for sports communication managers when developing appropriate strategies and tactics considering public sentiments such as emotions, coping strategies, and attributions on social media.
“I just signed your death warrant,” stated Judge Rosemarie Aquilina as she told Larry Nassar when she sentenced him from 40 to 175 years in prison on January 24, 2018. About 1 year prior to the sentence, on September 12, 2016, Tim Evans of the Indianapolis Star broke the first story of the sexual abuse scandal of Larry Nassar, a former Michigan State University (MSU) employee and USA Gymnastics team physician, who was charged with three counts of felony criminal sexual conduct. This crisis in sports history has resulted in more than 400 victims of sexual assault and one of the most publicized mishandlings of a sexual abuse case by a state university (Thompson & Wolcott, 2018). The publicity surrounding this story has been extensive. Bloomberg notes that the Larry Nassar case may pose an even greater challenge for MSU than the Pennsylvania State University’s Jerry Sandusky scandal (Smith, 2018). Sandusky, former Penn State football defensive coordinator, sexually abused 10 young males over a period of at least 15 years. In 2012, he was sentenced to decades in prison (Hobson, 2017). The Washington Post called the Nassar crisis one of the most tragic sexual abuse scandals in not only the sport’s history but perhaps in the history of the United States (Jenkins, 2018). The same article also attributed responsibilities to USA Gymnastics, the U.S. Olympic Committee (USOC), and MSU, claiming the organizations allowed Larry Nassar unsupervised access to young women athletes.
Beyond the overwhelming attention in traditional media outlets, several remarkable events of the case sparked heated discussions on social media. For example, days after the public hearing, MSU trustee, Joel Ferguson, expressed support for former MSU president, Lou Anna Simon, in an interview and reminded the public that the university had more things going on than “just this Nassar thing” (Manzullo, 2018). The sentencing day was also emotional and dramatic. In sentencing Larry Nassar, Judge Aquilina said, “As much as it was my honor and privilege to hear the sister survivors, it is my honor and privilege to sentence you. Because, sir, you do not deserve to walk outside of a prison ever again.” From these quotes and theatrical moments, social media was flooded with various emotions, quotations, memes, GIFs, and opinions concerning the decision of the courts, the victims, and the perpetrator.
Although both traditional and social media targeted the court case, the parties responsible for the crisis were also negatively affected. The scandal dramatically affected MSU’s reputation and subsequently damaged the athletic department and overall administration (Connor, 2018; Hobson & Svrluga, 2018; Mencarini, 2018; Tracy, 2018). USA Gymnastics has also suffered negative repercussions. In 1996, Nassar was appointed national medical coordinator and the following year team physician (Kamp, 2018). The Nassar scandal ultimately resulted in the resignation of Steve Penny as the president of USA Gymnastics in 2017 (Chavez & Sutton, 2018). The organization has been accused by many of negligence, has been included as defendants on multiple civil lawsuits, and has experienced reputation problems (Branch, 2017). When considering the involvement of both MSU and USA Gymnastics, the Nassar crisis potentially brings to light a number of issues regarding administrative oversight, sexual abuse, and questionable coaching practices within the sport.
Hardin (2014) has called for more theoretically grounded Twitter research in sports communication. Given the high priority of this athletic team doctor scandal and the substantial public reactions, it provides a unique opportunity to examine the emotions and coping strategies expressed by the public in light of the integrated crisis mapping (ICM) model (Jin, Pang, & Cameron, 2012). This study contributes to the crisis communication scholarship in sports and the ICM model in three ways. First, the ICM model proposes that four emotions (anger, fright, anxiety, and sadness) are commonly expressed by the pubic during times of crises, and previous studies have limited the examination of discrete emotions on social media to those four (e.g., Brummette & Sisco, 2015). This study not only applies the ICM to a crisis in sports but also attempts to expand the model by identifying new emotions expressed in tweets that have not been previously included in the model.
Second, coping strategies are a key component of the ICM model. Specifically, the model highlights the importance of coping strategies employed by the public as a psychological device to handle the emotions and cognitive discord resulting from a crisis. Once again, ICM has been employed in only a few studies examining social media platforms (Brummette & Sisco, 2015; Guo, 2017). This narrow application of the model to social media platforms provides several opportunities to examine a new crisis situation and at the same time evaluate how the coping strategies featured in the model align with those expressed by the public. We hope to assess the applicability of these coping strategies to this sport crisis.
Finally, a crisis always involves discussion of the actors or parties responsible for the situation. The ICM model briefly mentions the role of the attribution of responsibility in the examination of a crisis situation (Jin et al., 2012). There is a strong relationship between emotion and attribution (Coombs & Holladay, 2004). This study examines how to more thoroughly integrate attribution into the ICM model. In summary, it is the hope of the authors that the current study contributes to the extant literature on crisis emotions by expanding the discrete emotions based on a sports crisis situation and exploring coping strategies and responsibility attribution.
Literature Review
Crisis Communication in Sports
Crisis communication is very relevant to sports communication as athletes, sports teams, and sport organizations face crises on a day-to-day basis. When facing crises, sports organizations or individual athletes need to manage their reputation and image through effective strategies and tactics to maintain a mutually beneficial relationship with their stakeholders. Researchers attempt to understand the role that both sports organizations and the public play in the escalation or resolution of a crisis in sports. A growing number of studies have examined crisis response strategies in sports regarding organizations, athletic teams, individual players/coaches, and others from an organizational perspective (e.g., Brown, Murphy, & Maxwell, 2018; De Haan, Osborne, & Sherry, 2015; Hambrick, Frederick, & Sanderson, 2015; Onwumechili & Bedeau, 2017). For example, using Benoit’s image repair theory, Hambrick, Frederick, and Sanderson (2015) analyzed Lance Armstrong’s response strategies when he faced a doping investigation and admitted to using performance-enhancing drugs. Adopting the same theory, Onwumechili and Bedeau (2017) examined the International Federation of Association Football’s response strategies after its key officials were arrested for corruption. However, the public’s reaction to sports scandals from an emotional perspective has been understudied in the literature. A few studies explored how fans employed crisis communication strategies and reacted to the sports organizations’ crisis response (Brown & Billings, 2013; Brown, Brown, & Billings, 2015). Still, more research is needed to examine the emotional component of crisis response as sports scandals are often emotionally charged.
Given the two-way symmetric communication approach adopted by most communication practitioners, organizations base their crisis communication plans, strategies, and tactics on the knowledge and reactions from their key public(s) and stakeholders (Coombs, 2010). As the public communicates through traditional and new media channels (i.e., social media), it is essential to understand how they are receiving communication regarding a crisis and their subsequent reactions. The ability to understand the public(s) response including emotions allow organizations to adapt crisis strategies and tactics at each stage of a crisis to better meet the needs of the public and reach a state of resolution. The importance of psychometrics, including emotions, has been recognized as an important element in the past, but recent theoretical models, such as the ICM, have given researchers the ability to examine the role of emotions in crisis communications. Therefore, the current study attempts to apply this model to a sports communication crisis.
Mapping Discrete Emotions in Crises: The ICM Model
In recent years, crisis communication research has focused on a public-driven approach, which puts a great deal of effort in finding out how the public’s reaction affects an organization’s response strategies (e.g., Choi & Lin, 2009; Coombs, 1998; Jin et al., 2012; Kim & Niederdeppe, 2013). One particular line of research has integrated the public’s emotional responses into the audience-centered approach and has gained increasing popularity among researchers (Coombs & Holladay, 2005; Jin, 2010; Jin, Liu, Anagondahalli, & Austin, 2014). This line of research is important in crisis communication as emotions may “facilitate or negate the effectiveness of the various crisis response strategies” (Coombs & Holladay, 2005, p. 274). Researchers have focused on exploring discrete emotions rather than general positive or negative emotions and how these emotions interact with other key variables such as predictability, controllability, and coping in crisis situations (Jin, 2010).
Several studies have made significant contributions to the development of crisis emotion research. To understand what emotions are experienced and expressed by the public so that the organizations involved in the crisis can develop appropriate strategies to address needs, Jin, Pang, and Cameron (2012) proposed an emotion-based model of crisis communication, the ICM model (Jin et al., 2012). The ICM model argued that in a crisis situation, “emotions are one of the anchors of the publics’ interpretation of the unfolding and evolving events” (Jin et al., 2012, p. 268). That means, the public’s perception, evaluation, and interpretation of events are related to emotions. Building upon Lazarus’s (1991) cognitive appraisal theory in emotion research, the ICM model was posited on four quadrants based on two axes, with the X-axis indicating the public’s coping strategy and the Y-axis indicating the organizations’ level of engagement. Scholars in this field (e.g., Jin, 2010; Jin et al., 2012) explicated and tested the four core negative emotions most likely to be triggered by various crisis types from a cognitive appraisal perspective: anger, sadness, fright, and anxiety. For example, anger was experienced when individuals were offended by an organization or they perceived something threatening their well-being (Jin, 2010). Anger was typically evoked when a crisis situation was controllable and predictable. When an organization should have controlled or prevented the harmful situation but failed to do so, the public were likely to experience anger. Sadness was likely to be experienced when individuals felt a sense of loss and when a crisis was predictable but beyond someone’s control (Jin, 2010). Fright was experienced when one was facing uncertainty or a threat (Jin, 2010). Typically, fright was evoked when an organization or an individual was not able to predict and control what was happening. Anxiety was also experienced when one was facing danger or a threat, and this emotion was evoked from perceived uncertainty. However, it was reported that anxiety was not a consequence of the public’s cognitive appraisal (Jin, 2010) because according to Jin (2010), anxiety was the default emotion the public experienced in a crisis situation. The four core negative emotions and their relationship with other variables shed light on the understanding of public sentiment during various crisis situations.
In a similar vein, a wide range of distinct and discrete emotions have been examined by scholars of crisis communication. For example, Coombs and Holladay (2005) examined three emotions, including sympathy, anger, and schadenfreude (i.e., feeling happy about the pain of the organization), and their relationship with crisis responsibility. Choi and Lin (2009) explored anger, surprise, contempt, and relief based on attribution. More recently, Brummette and Sisco (2015) applied the ICM model to an analysis of tweets immediately after a college campus crisis and examined emotions, predictability, controllability, and coping methods used by the public. The study found that the most frequently displayed emotions on Twitter were anger and fright.
In addition to the aforementioned negative emotions, researchers have examined positive emotions in the public’s crisis responses such as hope (Jin, Park, & Len-Rios, 2010), relief (Choi & Lin, 2009), and sympathy (Coombs & Holladay, 2005; Kim & Niederdeppe, 2013). For example, Coombs and Holladay (2005) found that sympathy was associated with crisis responsibility attribution. Kim and Niederdeppe (2013) reported that in response to an H1N1 outbreak, college students most frequently experienced interest/alert/curiosity (41.6%), sympathy/compassion (38.2%), anxiety/worry/concern (28.6%), and gratefulness/appreciativeness/thankfulness (22.4%). Three of the four dominant emotions experienced by college students in response to the outbreak were positive, according to the authors’ typology. Furthermore, students also reported that they experienced some of the positive emotions such as interest and sympathy more frequently than negative ones (Kim & Niederdeppe, 2013). Guo’s (2017) qualitative analysis of Facebook posts after the Boston Marathon bombing found that positive emotions such as pride, hope, respect, compassion, and joy co-occurred with negative emotions such as sadness, anxiety, anger, shock, and worry.
In regards to discrete emotions experienced and expressed directly by the public on social media, there is a question as to whether there are any other emotions in addition to the four emotions plotted in the ICM model. Brummette and Sisco’s (2015) study was one of the few that examined emotions on Twitter, but it only examined the four emotions outlined in the ICM model. Jin et al. (2012) called for more emotions to be examined other than the four emotions in the ICM model. Guo (2017) also pointed out the importance of studying positive emotions in crisis situations. In the emotionally charged Nassar case, it is highly possible that people expressed a wider range of mixed emotions toward Nassar, the organizations involved, and the victims, including positive ones. Thus, the first research question attempts to directly examine what emotions Twitter users expressed in response to this high-profile sports crisis:
Coping as an Important Concept in Crisis Emotion Research
Because the line of crisis emotion research is built upon the cognitive appraisal framework, scholars in this research area also examined the relationship of emotions and some key appraisal-related variables such as crisis responsibility, crisis controllability, crisis predictability, and coping (e.g., Choi & Lin, 2009; Coombs & Holladay, 2005; Guo, 2017; Jin, 2009, 2010; Jin & Hong, 2010). Coping is a key concept in the appraisal theory of emotions (Jin, 2010; Jin & Hong, 2010). In a crisis situation, the public are not just passive receivers of an organization’s responses. Instead, they engage in a variety of coping methods to make sense of the situation, seek emotional support, plan concrete steps to resolve the negative situation, or simply vent their emotions (Jin, 2010; Jin & Hong, 2010). As noted by Lazarus (1991), the research of emotion and coping was closely related because coping emerged as a consequence of changing the environment where emotional state originated.
Lazarus and Folkman (1984) first introduced two types of coping: problem focused (i.e., to change the stressful environment through action) and emotion focused (i.e., to regulate an individual’s emotional status). Duhachek (2005) argued that the problem- and emotional-focused dichotomy of coping oversimplified reality. Grounded in consumer research, Duhachek developed an eight-dimensional coping construct in three higher hierarchical categories: active coping (action, rational thinking, and positive thinking), expressive support seeking (emotional support seeking, instrumental support seeking, and emotional venting), and avoidance (avoidance and denial). The eight dimensions were distinct but interrelated (Duhachek, 2005). The researcher also suggested that both emotions and appraisal affected the coping strategies used by consumers in a stressful situation. For example, consumers’ sense of self-efficacy and negative emotions conjunctively affected coping outcomes.
Building upon Lazarus’s (1991) cognitive appraisal theory, crisis communication researchers (Jin et al., 2012) posited that the public engaged in two types of coping during a crisis, namely, problem-focused coping and cognitive-focused coping. The former involved actual steps likely to be taken by the public to change the current situation or relationship between the organization and the public, while the latter involved a cognitive level of interpretation of the crisis situation.
Duhachek (2005) proposed that coping should meet two properties. First, coping emerges as a result of emotions. Second, the coping process should include cognitive, emotional, and behavioral dimensions. To address the complex reality of coping, Jin (2010) added one more coping category and further developed a three-dimensional coping construct, which included cognitive coping, conative coping, and emotional coping. Jin tested the relationships among various coping strategies and appraisal factors, such as perceived crisis predictability and controllability. For example, she found that when the public perceived a crisis as predictable, but not controllable, they were most likely to engage in cognitive coping (i.e., rational thinking and positive thinking). Conversely, when the perceived predictability and the controllability of a crisis were both low, individuals were most likely to engage in affective coping (i.e., emotional support and emotional venting). Similarly, building upon Duhachek’s (2005) conceptualization and operationalization of coping, Jin and Hong (2010) identified four coping strategies most relevant to crisis communication research, which were rational thinking (cognitive coping), emotional venting (emotional coping), instrumental support (conative coping), as well as action (conative coping), and examined the interrelationship of the four coping strategies at a hierarchical level.
More recently, a few studies explored the public’s coping strategies on social media in response to crises in general, using both quantitative and qualitative methods (Brummette & Sisco, 2015; Guo, 2017). For example, Brummette and Sisco’s (2015) content analysis of tweets found that the three most commonly used coping strategies during a campus safety crisis were instrumental support (31%), emotional venting (25%), and emotional support (16%). Guo (2017) examined public sentiment on the Boston Marathon Facebook page in the month following the Boston Marathon bombing crisis. The study found that Facebook users not only engaged in cognitive, affective, instrumental, and action coping identified in the ICM model and coping literature, they also contemplated lessons learned and established correcting mechanisms (community-wide attitudes and behaviors), which were not identified in the extant crisis coping literature. Furthermore, the Boston Athletic Association (the main organization involved in the crisis) facilitated public coping through active listening, digital community nurturing, facilitating action, and renewing the organization’s mission. These coping facilitation strategies used by the public and organizations further advanced emotion and coping literature in a real-time crisis setting.
Understanding the public’s coping preference can help sports crisis communication practitioners and managers better facilitate public coping (Jin et al., 2012). Jin et al. (2012) reported the public mostly engaged in conative coping by examining the public’s emotions and coping strategies as was evident in the news coverage of five major U.S. newspapers. According to the authors, one of the limitations of this study was that analyzing news stories was an indirect way to understand public emotions and coping strategies. Although prestigious newspapers were often at the forefront of reporting a crisis, the stories “are filtered through the eyes of journalists who may frame issues according to their perceptions of what had happened” (Jin et al., 2012, p. 290). This begs the question as to whether other coping types such as emotional and cognitive coping are also frequently used on a platform where the public can act as “self-publishers.” Social media such as Twitter provide an ideal platform for the public to express their own feelings and coping methods directly (as opposed to the gatekeeping role of news outlets) and timely (as opposed to the publishing cycle of news media). In addition, the relationship of emotions and coping strategies used on social media in a sports communication setting has been understudied. Therefore, the following questions (Research Question 2 and Research Question 3) focus on examining public coping on social media and the relationship of coping and emotions.
Further Examining Crisis Responsibility in the ICM Model
Weiner’s (1986) attribution theory posited that people needed to assign responsibility to an actor(s) or organization(s) for a negative event. In the context of crisis communication, Coombs (1998) defined crisis responsibility as “the degree to which stakeholders blame the organization for a crisis event” (p. 180). One of the primary outcome variables of the Situational Crisis Communication Theory (SCCT) was crisis responsibility (Coombs & Holladay, 2002). The link between attribution and crisis communication strategies developed by crisis managers has been examined within the theoretical framework of SCCT (e.g., Coombs & Holladay, 2004). This stream of research also explored how attributions can shape public perception, attitudes, and behaviors toward an organization as well as how an organization’s reputation will be affected (e.g., Coombs & Holladay, 2004; Schwarz, 2012). Based on the premise of attribution theory, which stated that the public are motivated to seek causal explanations of unexpected events, one of the central notions of SCCT argued that the more responsibility the public attributed to an organization or an individual for a crisis, the more pronounced their emotions (Bundy, Pfarrer, Short, & Coombs, 2016).
According to the cognitive appraisal theory of emotions (Lazarus, 1991), people engage in the attribution process as part of the appraisal of the environment. Lazarus (1991) suggested that people engage in either the blame or credit process to cope with a crisis situation and plan future steps to protect one’s well-being. Jin et al. (2012) advanced Lazarus’s position, discovering that blame takes precedence over credit in a crisis situation. Relying on the cognitive appraisal theory of emotions (Lazarus, 1991), the ICM model incorporated attribution in the plotting of emotions.
In the extant theoretical frameworks of crisis communication, there is a clear connection between attribution and emotion. Both attribution of responsibility and emotion have been studied as independent and outcome variables. Some studies explored how emotions can shape attribution after a crisis event (Small, Lerner, & Fischhoff, 2006). Small, Lerner, and Fischhoff (2006) investigated the causal relationship of emotions (i.e., anger and sadness) and attribution and found that anger triggered more attribution thoughts than sadness regarding a terrorist attack. Coombs and Holladay (2005) explored the relationship of crisis responsibility and three emotions evoked by a crisis: anger, schadenfreude, and sympathy. They found that crisis responsibility was positively related to both anger and schadenfreude but was negatively correlated with sympathy. The authors recommended integrating emotions into SCCT, and that research should examine the role of emotions. Kim and Niederdeppe (2013) found that regardless of the valence (positive or negative), discrete emotions were positively associated with crisis responsibility, meaning the more intense the emotions (both positive and negative), the more responsibility the public attributed to the organizations for playing a role in a crisis. Crisis responsibility was most strongly associated with anxiety or concern (Kim & Niederdeppe, 2013). Similarly, Jin, Liu, Anagondahalli, and Austin (2014) argued that both positive and negative emotions are closely linked to crisis responsibility attribution.
A few studies attempted to further explore the relationship between emotion and attribution and to categorize emotions based on crisis responsibility (Choi & Lin, 2009; Jin et al., 2014). Based on attribution theory and the SCCT model, Choi and Lin (2009) empirically explored two types of emotions, attribution-dependent and attribution-independent emotions. They found that anger, fear, surprise, worry, contempt, and relief were associated with crisis responsibility attribution, which were identified as attribution-dependent emotions. Anger was most associated with crisis responsibility. On the contrary, alertness and confusion were identified as attribution-independent emotions. Negative public emotions are found to be associated with crisis responsibility attribution (Choi & Lin, 2009). Jin et al. (2014) further extended Choi and Lin’s (2009) types of emotions experienced during crises, adding a third layer, internal attribution-dependent emotions. They developed and empirically tested three clusters of crisis emotions: attribution-independent, external attribution-dependent, and internal attribution-dependent emotions. The first cluster consisted of anxiety, apprehension, sympathy, and fear; the second cluster, external attribution-dependent emotions, included disgust, contempt, sadness, and anger; the third cluster consisted of embarrassment, guilt, and shame. Jin et al. (2014) developed an emotional response scale that measures the public’s emotions evoked by a crisis situation. This study advanced crisis emotion theories at the measurement level. Despite the methodology difference between the two studies (Choi & Lin, 2009, vs. Jin et al., 2014), they reached similar findings in terms of the relationship of emotions and responsibility attribution. For example, both studies suggested anger and contempt were associated with crisis responsibility attribution, whereas conflictive findings were revealed about fear. In Choi and Lin’s (2009) study, fear was associated with attribution, while in Jin et al.’s (2014) study, fear was identified as an attribution-independent emotion. The mixed findings in the literature of emotions and attribution further echo the notion that attribution-dependent and attribution-independent emotions coexist with each other (Jin et al., 2014; Weiner, 1986). Some emotions, such as fear, can be both attribution independent or dependent.
The ICM model advanced crisis communication research by putting public emotions at the center of crisis communication research. However, the connection between emotions and attribution is only briefly mentioned. More work is needed to find out how to incorporate attribution into the current ICM framework. Choi and Lin (2009) and Jin et al. (2014) not only examined discrete emotions likely to be experienced by the public during crises, more importantly, they attempted to establish a relationship between emotions and crisis responsibility attribution. As Jin et al. (2014) suggested, “attribution…seems to be the fundamental factor for grouping different crisis emotions” (p. 315). In reality, attribution and emotion can be spontaneously and simultaneously triggered by a crisis event and can be further expressed on social media platforms. The next research question will explore how different emotions are expressed based on responsibility attribution.
Choi and Lin (2009) and Jin et al. (2014) further advanced the traditional negative or positive emotion analysis in crisis communication research and shed light on the relationship of discrete emotions and attribution, regardless of the valence of the emotion (Jin et al., 2014). However, both studies did not distinguish between levels of responsibility or actors responsible for the crises. Jin et al. (2014) called for a multidimensional construct of crisis responsibility. In a complex case such as Nassar’s, multiple sports entities were held responsible by the public (i.e., Nassar, MSU, USA Gymnastics, USOC). It is important to examine whether the public attribute more responsibilities for the crisis to individuals or organizations. In some crisis cases, the public might mainly blame the individuals who did the wrongdoing while ignoring any organizational-level responsibilities. In other cases, the public might hold the organizations that enabled the individual responsible for the crisis. If that was the case, the organizations would be forced to use crisis response strategies to protect their reputation. Therefore, understanding the public’s different levels of responsibility attributions would assist crisis communication decision makers in developing appropriate strategies based on the public’s attribution decision. An, Gower, and Cho (2011) examined how news media used individual- and organization-level responsibility frames when reporting crises. More recently, Starke and Flemming (2017) explored how German newspapers used individual- or organization-level responsibility frames when discussing doping in sports. However, little is known about how the public attribute different levels of responsibilities directly on social media. In the Nassar case, many individuals and several sports organizations were affected. Therefore, the next research question intends to fill-in this gap in the literature.
Method
A content analysis was appropriate to address the research questions for several primary reasons. First, this method has been noted as being “the systematic, objective, quantitative analysis of message characteristics” (Neuendorf, 2002, p. 1). Our research questions focused on message characteristics including examining emotions, coping strategies, and responsibility attributions found in social media posts. Second, content analysis has been used by researchers as a viable methodology to extract data from social media (Wong, Gao, Xu, & Li, 2017). Finally, previous studies have employed content analysis for examining social media content in sports crises (Brown & Billings, 2013; Brown et al., 2015).
Sampling
Tweets were collected from January 24 to January 28, 2018. The time frame was selected based on the real-world events and literature. Larry Nassar was sentenced by Judge Rosemarie Aquilina on January 24, 2018. Previous studies had shown that a minimum of 1 day of data collection on Twitter was acceptable concerning the large amount of tweets produced (e.g., Brummette & Sisco, 2015). The key word “Nassar” was selected to capture any mention of the crisis situation. This key word was chosen to ensure that a comprehensive sampling frame was produced. The Twitter archive service, Sifter, was used to acquire the initial sampling frame (9,599 tweets in total). To ensure the comprehensiveness of user profiles, all tweets from various users were included in the sample. Such users might include individuals, organizations, business, media, and so on. Then a systematic random sample of 3,400 tweets was extracted to code; 312 tweets were discarded from this sample due to the tweet being off topic. As a result, 3,088 tweets were coded.
Coding Scheme
The coding scheme included the following categories: emotions, coping strategies, and attribution of responsibility. Each category was binary coded as 1 = present or 0 = not present.
Emotions
Subcategories in this category included six discrete emotions: anger, anxiety, disgust, fright, joy, sadness, and other emotions. These subcategories were inspired by the ICM model, previous work (e.g., Choi & Lin, 2009; Guo, 2017; Jin et al., 2014), and the particular crisis situation in this study. Because emotions could be interrelated (Jin et al., 2014), each tweet was coded for as many emotions as was expressed. For example, a tweet such as “I’m so mad, sad, and heartbroken. I’ve never hated anybody more in my life than Larry Nassar. I’m mad, for what he has done to these women.” was classified as both demonstrating anger and sadness. When a tweet displayed no emotion, we also examined whether the tweet contained the user’s original content/sentiments or it was simply a retweet or a link to a news article with a news headline. Operational definitions for each emotion and tweet examples are shown in Table 1.
Operational Definitions and Examples of Emotions Expressed on Twitter About the Nassar Case.
a Definitions were adapted form Lazarus (1991) and Nabi (2002).
Coping strategies
Coping strategies was adopted from prior literature (e.g., Brummette & Sisco, 2015; Duhachek, 2005; Guo, 2017; Jin, 2010; Jin & Hong, 2010; Jin, Pang, & Cameron, 2010). This category comprised of the following subcategories based on three types of coping strategies: emotional, cognitive, and conative coping. The specific coding items were rational thinking, positive thinking, denial, avoidance, emotional support, emotional venting, and action. Rational thinking was coded when a tweet expressed ways of making sense of the crisis. Positive thinking was coded when tweets featured content that downplayed stressors and highlighted good things. Tweets expressing denial were clearly denying that the event occurred or that the actors in the crisis conducted specific behaviors. Avoidance was coded when a tweet attempted to impart distance (psychological or physical) between themselves and the crisis situation. Emotional support was coded when a tweet expressed how an individual felt or attempted to improve their emotional and/or mental state by connecting with other users. Emotional venting was coded when the tweet attempted to express any emotions. The final coping strategy, action, was coded when a tweet provided tangible recommendations or actual steps to a solution for the crisis. Avoidance was originally developed in the codebook but was dropped later in the final data analysis due to extremely low frequencies. Tweet examples for coping strategies are shown in Table 2.
Examples of Coping Strategies Used on Twitter About the Nassar Case.
Note. MSU = Michigan State University.
Attribution of responsibility
We examined whether the tweet mentioned one or more of the actors/organizations involved in the crisis. The list of actors for this category were derived from past work in this field (e.g., An & Gower, 2009; An, Gower, & Cho, 2011; Schwarz, 2012) and the specific events. Based on An and Gower’s (2009) work, we divided the entities into two levels, individual and organization levels. In terms of the individual level of entities, we identified the following actors: Larry Nassar, the former MSU President, and other individuals. Although many individuals from multiple organizations, such as Steve Penny, former USA Gymnastics president, were involved in the scandal, an initial name search in the data set showed that these individuals were not frequently mentioned at the time of data collection. As a result, these individuals were coded in the “others” category. In terms of organizations, we identified MSU, the USOC, the National Collegiate Athletic Association (NCAA), USA Gymnastics, and other organizations.
Intercoder Reliability
The authors, who were knowledgeable of the research topic, coded the tweets. The two coders were trained and test coded the variables during multiple sessions. During the initial training session, each coding item was explained with example tweets. Then the coders conducted test coding with a set of tweets separate from the intercoder reliability sample. The test coding allowed the coders to further learn the coding items. Any conflicts were resolved. Then another test coding and training session was conducted. After that, the coders randomly selected 10% of the sample that did not overlap with the actual data and coded separately for intercoder reliability calculations. Cohen’s (1960) κ values were between 0.53 and 1 for all items. The percentage agreements were between 82.83% and 100%. Some items had low Cohen’s κ values but high percentage agreement due to low frequency. For example, the coding item “joy” had the lowest intercoder reliability coefficient, 0.53, but the percentage agreement for this item was 91.78%, which is considered relatively high. Although there is no discipline-specific consensus about the minimum levels for the reliability coefficients (Neuendorf, 2008), in this specific study, the authors accepted the intercoder reliability coefficients after examining both Cohen’s κ values and percentage agreement. The coders then split the sample into half and coded separately.
Data Analysis
Descriptive statistics were conducted for Research Questions 1, 2, and 5. Chi-square tests were conducted for Research Questions 3 and 4. For Research Question 3, we ran a series of χ2 analyses with each specific emotion (binary coded) as the independent variables (IVs) and coping strategy (binary coded) as the dependent variables (DVs). To answer Research Question 4, we first recoded responsibly to a binary variable, with “1” indicating responsibility mentioned and “0” indicating responsibility not mentioned. Then we ran χ2 analyses with responsibility attribution as the IV and each specific emotion as the DV. Given our data variables were nonparametric and categorical, χ2 is an appropriate test for determining differences between observed and expected values.
Results
Emotions Expressed About the Nassar Case (Research Question 1)
The most frequently exhibited emotion through the tweets was anger (n = 974, 31.6%), followed by disgust (n = 898, 29.1%) and joy (n = 860, 27.9%). In this specific case, people expressed joy toward different individuals. Of the 860 tweets exhibiting joy, 441 (51.28%) tweets expressed happiness directed towards Judge Aquilina. An example was “the judge is incredible. Such powerful women!” A number of users, 269 (31.28%), felt happy about the sentencing of Nassar. An example was “I’m glad to see he gets what he deserves. Justice is served.” The remaining tweets expressed sadness (n = 90, 2.9%), anxiety (n = 11, 0.4%), fright (n = 5, 0.2%), and other emotions (n = 492, 15.9%). A total of 591 tweets expressed no emotions (19.1%). Further analysis of those tweets expressing no emotion showed that 249 (8.1%) tweets were simply retweets of news stories with news headlines and 342 (11.1%) tweets contained original content by the user. A total of 834 (27%) tweets contained multiple emotions. The results are shown in Table 3.
Frequencies of Emotions, Coping Strategies, and Attribution of Responsibilities.
Note. MSU = Michigan State University; USOC = U.S. Olympic Committee.
Coping Strategies Used by the Public (Research Question 2)
The predominate coping strategy was emotional venting (n = 2,396, 77.6%), followed by positive thinking (n = 762, 24.7%) and rational thinking (n = 527, 17.1%). The remaining tweets used emotional support (n = 220, 7.1%), action (n = 19, 0.6%), and denial (n = 6, 0.2%). A total of 918 (29.7%) tweets contained multiple coping strategies (see Table 3).
Coping Strategies and Emotions (Research Question 3)
A series of χ2 analyses showed that different coping strategies were used when different emotions were expressed. When users expressed anger, disgust, and sadness in the tweets, they were less likely to use rational thinking, positive thinking, emotional support, and emotional venting coping strategies. When people expressed joy in their tweets, they were less likely to use rational thinking, emotional support, emotional venting, and action strategies but were more likely to use positive thinking strategies (Table 4).
Chi-Square Tests Results on Comparisons of Emotions and Coping Strategies.
Emotions Expressed Based on the Attribution of Responsibility (Research Question 4)
This study found a significant difference of emotions expressed on Twitter when attribution of responsibilities was present/absent. When attribution of responsibilities was present, anger (n = 897, 92.1%) and disgust (n = 842, 93.8%) were most frequently expressed. However, when attribution of responsibilities was absent, joy (n = 574, 63.3%) was likely to be expressed (Table 5).
Chi-Square Tests Results on Comparisons of Emotions and Attribution of Responsibilities.
Individual- and Organizational-Level Responsibility Attribution (Research Question 5)
The majority of the tweets contained at least one responsibility attribution (n = 1,842, 59.7%). Only 13 tweets (0.42%) contained multiple responsibility attributions. Over half of the tweets expressed individual-level attribution (n = 1,657, 53.7%). Among those tweets, 1,415 tweets (45.8%) blamed Nassar and 116 tweets (3.8%) blamed the former MSU president. There were a small number of tweets (n = 235, 7.6%) blaming other individuals. Only 222 tweets expressed organizational-level attribution (7.2%). Among those tweets, 139 (4.5%) blamed MSU. Only 54 tweets (1.7%) blamed USA Gymnastics, and the NCAA was held responsible in 42 tweets (1.4%). The USOC received the smallest amount of responsibility attributions (n = 19, 0.6%; see Table 3).
Discussion
Responding to Jin’s (2010) call for exploring positive emotions in the crisis emotion research, the current study adds joy to the analysis. In the Nassar case, there appears to be nuances regarding the expression of joy. Our findings reveal a further dimension of joy depending on the actors involved. Some tweets were “happy” about Nassar’s sentence. For example, many wrote they were glad to see Nassar was going to spend the rest of his life in jail. Others were happy for Judge Aquilina for her role in the sentencing. Immediately after the sentencing of Nassar, many in the public called the judge a hero. People’s joy about Nassar going to jail seems to still have some negative sentiment undertones and can be related to other emotions such as anger. Whereas the positive emotions expressed towards Judge Aquilina are more of a pure happy feeling and can be associated with other positive feelings such as gratefulness, thankfulness, and admiration.
The current analysis pointed out the necessity to incorporate positive emotions in the crisis emotions research. Other scholars also advocated adding positive emotions to the ICM model. Drawing from research in social psychology, Kim and Niederdeppe (2013) advocated the importance of examining positive emotions that often co-occur alongside negative ones and play an important role in crisis coping. Most recently, Guo (2017) qualitatively examined how positive emotions, such as gratitude, fearlessness, determination, hope, pride, and joy, emerged after the Boston bombing. The research recommended adding the positive emotions to Quadrant 3 of the ICM model. Our findings concur with the previous studies’ suggestion (e.g., Guo, 2017; Kim & Niederdeppe, 2013) and, specifically in this case, reveal that joy is one of the most predominant emotions. This study also demonstrates the importance for researchers to methodologically employ grounded theory in the creation of coding categories identifying emotions during the coding process.
Regarding practical implications, the findings indicate that for sports communication managers, they need to know how to prioritize their limited resources in relation to the public’s emotional reactions. The majority of the tweets are emotionally charged immediately after the sentencing. That means sports communication managers need to figure out what emotions are dominating public conversation. Then they need to craft messages that address these emotions to reduce the public’s stress level, before any concrete steps are proposed. Furthermore, these findings point out the importance of social media monitoring of public emotions, before and after the crisis event occurs.
Responding to Jin and Hong’s (2010) call for more research on public coping in crises via social media, the current analysis reveals that the public mainly engaged in two types of coping strategies regarding the Nassar scandal, cognitive (rational and positive thinking) and affective (emotional venting). Some tweets showed positive thinking, conveying strength and hope for the victims of sexual abuses. Also, almost one fifth of the public engaged in rational thinking. Some tweets questioned the responsibility of major organizations such as MSU, the USOC, and USA Gymnastics. Such voices go beyond complaining or emotional venting. As Jin and Hong (2010) note, rational thinking allows the public to cope with a crisis by making plans that may reduce stress and negativity. For example, questioning organizational responsibilities is a logical cognitive process that would decrease the public’s stress in the crisis.
In the Nassar case, emotional and cognitive copings are more evident than conative coping. Immediately after the sentencing, Twitter users did not engage in approach- and action-oriented coping, such as considering how to solve the problem and taking direct action. This finding departs from Jin and colleagues’ (2012) application of the ICM model and Guo’s (2017) qualitative analysis, which find conative coping is dominant. The conflictive findings can be explained in several ways. First, Jin and colleagues’ (2012) study content analyzed news stories in major newspapers. Significant differences exist between news coverage and tweets. For example, news stories allow an in-depth investigation of a crisis, whereas Twitter limits user expression to 280 characters. Second, news stories may be action oriented. As Jin and colleagues’ (2012) argued, “conative-based strategies…are often preferred, and recorded” (p. 287) by the public. Finally, the news stories may contain agenda-setting/framing elements, while tweets are more individualistic and are less likely to advance a specific agenda. These differences, although unavoidable, point out an important direction for crisis communication managers and practitioners. It is insufficient if just one type of media content is examined and monitored during a crisis. This activity may lead to misrepresentation and misinterpretation of the reality of public coping. Therefore, it is recommended that both traditional and social media be monitored for public sentiment to facilitate organizational-level strategies and tactics to assist in public coping.
Another explanation lies in the features of Twitter. As a microblogging site, Twitter allows compositions of 280 characters or less. Offering tangible solutions or steps in 280 characters is challenging at best and frustrating at worst. Instead, we found that the public employed Twitter as a platform to express emotions, as discussed in Research Question 1.
Crisis type may also play a role in the mixed findings. Guo (2017) examined a terrorist crisis in a major metropolitan area. People affected by the threat were more likely to take actions to protect their well-being. In addition, Guo limited the analysis to the organization’s official Facebook page, while the current study examines the tweets of all users. It is possible that, as Guo suggested, people who post on the official Facebook page already belonged to a runner community prior to the bombing. Therefore, the public in that case may be more proactive in planning out concrete steps to improve the crisis situation in the community.
Coping is a highly complicated mechanism related to crisis type, threat level, and individual characteristics such as efficacy (Duhachek, 2005; Folkman & Lazarus, 1985). Jin (2009) suggested that individuals’ coping preference was related to one’s emotions, the predictability, and controllability of the crisis. For example, when members of the public predominantly feel sad and when the situation is predictable but out of control, they are likely to use action and instrumental support coping. When members of the public predominantly feel scared and believe the situation is out of the control of the organization, they are likely to use an avoidance strategy. In the current case, the public did not express sadness or fear predominately. Therefore, action-oriented strategies and avoidance were not used. Our finding, in some way, supports Jin’s relationship about discrete emotions and coping. Furthermore, Folkman and Lazarus (1985) suggested that when the threat level was high, the individual might use an avoidance coping strategy. Duhachek (2005) noted that when consumers experience threat in conjunction with high self-efficacy, they would be more likely to engage in expressive support coping such as emotional venting, emotional support, or instrumental support. These theoretical propositions may explain why the coping strategies vary significantly from study to study. Although the current study provides a basic picture of coping strategies used by Twitter users, future studies investigating this topic should also consider other key factors such as an individual’s self-reported efficacy to further advance the coping component of the ICM model.
It is possible that Twitter users are not sharing actions on social media as they perceive Twitter predominantly as a self-expression platform. However, it does not mean actions are not executed off-line. For example, an alum may stop donating to MSU because of disappointment, but this may not be expressed on social media. An implication for crisis managers is that they should closely survey social media, but this activity should be used conjunctionally with other issues of management strategies, such as surveying public opinion about action intendancy.
The current findings support Choi and Lin’s (2009) and Jin et al.’s (2014) findings that in general, there is a clear distinction among emotions expressed depending on whether the public engage in responsibility attribution. Specifically, the findings reveal that when the public think about who is responsible for the crisis situation, they are likely to express anger and disgust. When the public do not engage in responsibility processing, they are likely to express joy. The current finding indirectly supports that anger is often associated with blaming (Jin et al., 2012) and is one of the most common attribution-dependent emotions (Choi & Lin, 2009).
The differences of emotions expressed by attribution revealed in this study provided important practical implications for sports communication managers. If the public hold the organization or individual responsible, the major emotions expressed by the public on social media are negative emotions such as anger and disgust. From an SCCT standpoint, when the individual or organization is at fault, more accommodative responses are recommended to calm the public and fix the relationship. Coombs and Holladay (2004) suggested that more accommodative crisis response strategies were recommended when the public assign greater attributions of crisis responsibility. Practitioners and crisis managers should pay particular attention when the public predominantly express anger and disgust and when they are clear about who should be held responsible for the crisis. Future corrective action should be incorporated into message development to protect organizational reputation. In the Nassar case, for example, it is possible that involved organizations such as MSU and USA Gymnastics could first express sympathy to the victims and support the decision of sentencing Nassar to mitigate any negative impact to the organization. Feelings of sympathy can “make it easier for an organization to engender potential supportive behavior from stakeholders” (Coombs & Holladay, 2005, p. 274). Then they should develop key points such as future steps and corrective actions to prevent the similar tragedy from happening within the organization. This strategy may help the organizations gain public support and protect the organizations’ reputation.
Segmenting and categorizing emotions based on crisis responsibility attribution can be a challenging task. The challenge lies in two aspects. First, emotions are often interrelated and can hardly be separated into clear-cut clusters (Jin et al., 2014). This can be understood from the mixed findings of previous studies (e.g., Choi & Lin, 2009; Jin et al., 2014). Second, new emotions may be added to the model depending on the crisis situation (such as joy in the current study). However, this study, along with previous research, provides meaningful first steps for sports communication managers and practitioners to better understand public emotions and thus to develop effective response strategies in accordance with the public’s emotional status and crisis attributions.
Analyzing the level of responsibility attribution can help us paint a picture of who the public believe should be responsible for the crisis situation, especially when multiple entities are involved. Although MSU and other organizations were mentioned frequently in the news media, our finding reveals that only a small portion of responsibility attribution is targeted at the organization level such as MSU, USOC, and USA Gymnastics. The majority of the tweets are focused on the individual level, mainly blaming Nassar. This finding implies that in the first several days after the sentencing, few members of the public are thinking about the role of sports organizations or individuals in enabling Nassar’s abuses. This is a time window within which the involved sports organizations should act proactively to prepare for the possible forthcoming attribution and develop messages that might change public’s attitude. Coombs (2010) notes that often times, crisis managers use dissociations in hopes to reduce the overall threat to an organization’s reputation. If the public accept an individual-group dissociation, then the public infers that the organization is not bad, just some people inside the organization (Coombs, 2010). It was beyond the scope of this article to analyze the crisis strategies employed by MSU, USOC, and USA Gymnastics, whereas this finding begs to question whether these organizations successfully used an individual-group dissociation strategy. More research would be necessary to examine whether the media sparked this dissociation. On another note, An and colleagues (2011) suggest that the traditional news media can potentially direct people to blame the organization, and if the public consume this information first, it could anchor opinions regarding attribution of responsibility. Additionally, An and colleagues concluded mainly that in preventable crises, the media typically focuses on an individual level of attribution. Practically, our finding suggests that crisis managers need to incorporate environmental scanning methods during a crisis that carefully analyze the focus of social media posts by media outlets and individuals. This process may reveal whether an individual-group dissociation is successful and/or whether it is naturally occurring.
It should be noted that the Nassar case has a unique set of circumstances due to the length of the abuses, the number of victims affected, and the number of organizations/characters involved. Our data show that Twitter users experience quite complicated emotions such as anger, disgust, sadness, and joy. Any communication strategies developed by MSU or other organizations can be interpreted in many different ways, mostly negatively, by sports fans, athletes, students, alumni, and community members. However, the strategies put forward by MSU can provide some takeaways for sports communication managers. For example, MSU developed a website (https://msu.edu/ourcommitment/news/nassar-information.html) to address various concerns regarding the Nassar scandal, including news, FAQs, media contacts, and future steps. This website had the potential of assisting the organization in identifying concerns expressed by the public. Tactics such as this are useful for organizations to connect to the public and should promote two-way systematic communication channels. When the emotions are overwhelming like in this case, transparent communication with the public is the key.
Limitations and Future Research
When a crisis in sports happens, a variety of publics including fans, news media, and organizations will tweet about the crisis. As such, a number of additional variables, including Twitter user types, would be useful to examine in future research as different user types may generate distinct emotions depending on their viewpoints and group affiliations (Brown & Billings, 2013; Brown et al., 2015). Benoit and Pang (2008) note that identification of the key public is important as each public group has its very own interests, concerns, and goals. It would be vital for future research to examine how different types of users express their emotions and how they cope with a crisis.
Second, emotions are crisis-specific. This study examined one scandal in sports. Future studies are encouraged to analyze multiple cases in one study, ideally, including different crisis types in sports. It is likely that a more complex picture of commonly expressed emotions will emerge with different emotions being more common or rarer depending on the nature of the crisis. Some emotions (i.e., anger) may be viewed as positive by some external stakeholders regarding a crisis. For example, the media or various individuals in the public could be reassured of a society’s ethics and values in the public outrage of a crisis. This would require a possible reclassification of the valence of emotions and a future extension of the ICM model. Additionally, a longitudinal tracking of emotions will be helpful. As a crisis progresses, more emotions and coping strategies may emerge depending on the stage of the crisis. It will be interesting to compare themes during multiple stages of a crisis. Furthermore, emotions are often interrelated (Jin et al., 2014). One tweet expressing anger can also express disgust at the same time. Future studies are encouraged to explore how emotions are related with each other.
Emotional contagion may play a role in the type and quantity of emotions expressed by the public during a sports crisis. When one browses social media messages filled with emotions, it is possible that the displayed emotions will influence the individual’s affective status and behavior. This phenomenon can be explained from an emotional contagion perspective (McColl-Kennedy & Anderson, 2005). Since emotional contagion occurs frequently in interpersonal relationships (McColl-Kennedy & Anderson, 2005), it would be interesting for future studies to examine how emotional contagion applies in social media contexts. Although our current analysis does not allow us to comment on whether emotional contagion resulted in the frequency of specific emotions on Twitter, future studies can apply network analysis to examine whether the specificity and frequency of affections is indeed driven by emotional contagion.
Footnotes
Declaration of Conflicting Interests
The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.
Funding
The author(s) received no financial support for the research, authorship, and/or publication of this article.