ABBREVIATIONS

ADT, affective disposition theory; CSR, corporate social responsibility, SCCT, situational crisis communication theory

INTRODUCTION

With the development of technology, people are exposed to global crises more often and they expect companies to respond well during their major crises [1]. The average number of headlines in Forbes signaling crisis of the top 100 companies reported increased two times between 2010 and 2016 compared to the period between 2000 and 2009 [2]. It is no longer a question of whether a corporation will face a crisis. How to handle the crisis afterward is now as important as how to prevent it. Crises are often followed by severe and sometimes fatal damage, including decreased revenue, tarnished corporate reputation, increased media scrutiny, loss of consumer loyalty, critical social media discussion, and more [3]. Crisis managers need to know the effective ways to communicate after crises to strategically handle crises in order to prevent further damage. To communicate effectively with people around corporations that are experiencing crises, it is crucial to understand how those people feel about the corporations.

Emotion is one of many attributions that would decide how people react to crises [4–6]. Even though people often evaluate a crisis cognitively, their emotions also lead them to certain decisions [7]. For example, when British Petroleum (BP) faced the oil spill crisis in the Gulf Coast, angry publics did not hesitate to post negative comments on social media channels even on BP’s company Facebook page. BP mostly used one of the corrective action strategies, such as compensations and cleaning the shores; however, the strategy did not change the attitude of angry publics [8]. Similarly, Choi and Lin [6] found that people who felt stronger responsibility of the company in Mattel’s product recall crisis in 2007, were more likely to generate emotions toward Mattel including anger, surprise, worry, fear, contempt, and relief. Among those emotions, anger led people to have the intention to boycott Mattel’s products.

In a crisis, people’s moral judgment about the crisis response influences their level of anger [9]. Affective disposition theory (ADT) explains that people wish for protagonists’ success and antagonists’ failure. Once people establish a liking for a character, they experience empathetic reactions to the character which ultimately create enjoyment or distress. Furthermore, people tend to optimize their mood; thus, they choose media contents that can strengthen their mood states [10]. Applying these premises, disposition toward a company may work for or against the company when a crisis occurs. People would develop a certain emotion toward the company or the crisis itself, and it can influence their reaction to the company.

ADT has mostly been applied to people’s exposure to entertainment media. However, in this study, we wanted to examine if a Corporate Social Responsibility (CSR) reputation is a dispositional factor in a crisis. The prior reputation of a corporation has been discussed as one of the possible factors that make it easier to repair the corporation’s reputational assets during a crisis [11]. For example, Eisingerich, Rubera [12] found that people tried to avoid negative information about a company with an excellent CSR reputation in a crisis because it would make them feel uncomfortable. In addition, we watned to explore whether the type of crisis (low responsibility vs high responsibility) would moderate the affective disposition process in line with situational crisis communication theory (SCCT) [13]. Lastly, we investigated which crisis responses are more acceptable under a specific emotional state in a crisis.

EMOTION IN CRISIS

How people react to a crisis is closely related to how they process crisis information. A crisis is “an event perceived by managers and stakeholders as highly salient, unexpected, and potentially disruptive can threaten an organization’s goals and have profound implications for its relationships with stakeholders” [14]. In the event of crisis, people naturally seek the cause of it, which often raises negative emotions [15]. People would respond more negatively to an organization that has a crisis if theye perceive that the organization is responsible for the crisis [13]. In other words, the publics’ negative response can be tied to a high level of perceived organizational responsibility in a crisis. The developed negative responses then lead negative public emotions [6,16,17]. People would cognitively try to process the information about how the crisis happened and who might be responsible for it, but their cognitive efforts of processing the information would affect their emotions as well.

When people are exposed to crisis news about a company they feel different emotions based on the types of crises. More specifically, Jin found that people felt anger in a controllable and predictable crisis, while they felt more sadness when the crisis is not controllable; when crisis is not predictable and not controllable, people felt frightened [18]. Choi and Lin said that anger is highly related to the attribution of crises called attribution-dependent emotion; some emotions like fear are not necessarily dependent on attributions called attribution-independent emotion [6]. Jin et al. found that crises that are attributed to internal causes led to more attribution-dependent emotion, including anger, contempt, and disgust. Attribution-independent emotions are more likely to be generated when an organization’s responsibility is perceived to be low in a crisis and when the crisis situation is uncertain [18,19]. These findings all show that people process crisis information cognitively based on various crises, which also entails certain emotions.

AFFECTIVE DISPOSITION THEORY IN CRISIS

Emotions explained in ADT, however, are not generated cognitively. It begins with an assumption of selective exposure theory: People always instinctively seek pleasurable stimuli [20]. Zillmann and Bryant [21] claimed, “Selective exposure, instead, designates behavior that is deliberately performed to attain and sustain perceptual control of particular stimulus events” (p. 2). Furthermore, people tend to avoid uncomfortable situations [22]. It is also true that people instinctively seek positive experiences rather than negative experiences according to their survival instincts [23]. This explanation draws on the limbic system’s coordination of the basic survival function by rewarding pleasure. Activities that are rewarded with pleasure are sustained, and activities that are punished with discomfort are discouraged. As a result, people use all five senses to minimize discomfort and maximize pleasure. This is especially true when people are enjoying media entertainment; for example, when they watch a crime drama, viewers’ emotional states are decided more affectively than cognitively. Based on the assumption of selective exposure theory, in the context of a corporation’s crisis situation, people would instinctively feel a certain emotion when they are exposed to the information about the crisis.

Yet, when people generate emotions after being exposed to certain information, there should be predisposition factors that stimulate certain emotions. ADT explains how one’s predisposition toward the characters of entertainment, e.g., movies, crime drama, sports, etc., influences his or her enjoyment (positive hedonic valence) or distress (negative hedonic valence) [20]. For example, Raney [23] described the effect of moral judgment in television crime dramas and insisted that moral judgment is a key factor to increasing enjoyment because it creates likes and dislikes for certain characters, which, in turn, enhances enjoyment. He explained the mechanism of disposition in a drama: “Once characters are liked, we are able to empathize with their plights and hope for their triumph over them. In fact, the stronger the positive feelings, the stronger our empathic reaction” (p. 351). Vorderer, Knobloch, and Schramm [24] also found that when people watched movies, empathy toward protagonists was a powerful predictor of increased enjoyment. Regarding these findings, the original concept of ADT is that the emotions obtained from media entertainment come from moral considerations: a morally desirable character is liked, and the opposites are disliked [25]. After all, according to ADT, emotions are not fallouts after information processing.

This mechanism of ADT has been examined in different contexts. For example, when watching a political talk show, viewers’ assessment of the political issue of the show is usually predetermined based on their predisposition toward the guest [26]. Robinson and Knobloch-Westerwick [25] found that predisposition toward a protagonist in a novel strengthened the self-efficacy of readers and influenced their outcome expectations. ADT has not been examined in a crisis, however. Therefore, it will be interesting to see if any factor of a company works as a predisposition in a crisis. More specifically, following ADT, a factor working as a predisposition in a crisis should lead to a negative or positive emotional state. More specifically, as Figure 1 shows, people usually have a certain predisposition toward a company before they receive crisis information about the company. If the company is a protagonist in people’s minds, they will develop instinctively negative emotional states when they hear or watch the crisis news about the company. If a company is a villain, on the other hand, people may develop positive emotional states when they are exposed to crisis information about the company.

CSR REPUTATION AS A PREDISPOSITON FACTOR IN CRISIS

To see if the model illustrated in Figure 1 works, it is necessary to examine possible predisposition factors for companies. In a crisis, like in a crime drama, moral judgment about a company can be a predisposition factor. Reputation can be one of the predisposition factors as well. Reputation is an intangible asset of organizations that influences various publics of the organizations, including customers, investors, employees, media, and many more [3]. Publics shape the reputation of organizations based on the experiences or information they receive about the organizations.

Gilbert explained that CSR activities work as a dispositional motive before consumers take other factors such as competitive pressure and financial motivations into consideration using the two-stage model of attributions [27]. This is because CSR activities provide specific experiences or information about a company to its stakeholders, which forms the reputation of the company. In a 2012 study by the Reputation Institute, 42% of a company’s reputation was based upon perceptions of their CSR efforts. Godfrey also insisted that CSR activities protect the reputational capital of a company [28]. This protected reputational capital because of CSR activities often positively violate stakeholders’ expectations toward the corporation and thus engender buffering effects in a crisis [29]. It also strengthens a positive attitude toward the organization [30,31]. Therefore, companies are motivated to perform socially responsible activities as insurance for their risk management [32–34] as well as for improving financial performance [35]. After all, practicing CSR activities would establish a positive predisposition to a company’s stakeholders. Like in a crime drama, stakeholders would hope for the success of the company if CSR activities create a positive predisposition toward the company. Following the assumption of ADT, this established predisposition due to CSR activities can be shown as a negative or positive emotional state when stakeholders are exposed to crisis information of the company.

The impact of the reputation established by CSR activities has been studied in crisis research. Coombs and Holladay said that reputation is the factor that links CSR and crisis and that CSR is one of the variables that influence the relationship between crisis and reputation [29]. Here, reputation is also well connected with the concept of expectation. If the expectation does not match the company’s performance, it may bring about a reputational threat [36]. Evaluation of the issue would be negative when performance does not meet expectations and more positive if performance exceeds expectations [37]. However, the confirmatory bias effect could sometimes be bigger than the expectancy-violation effect [38]. Namely, people would withhold their evaluation in a crisis and continue to support a company with a high reputation.

Therefore, CSR-focused messages can be an effective bolstering strategy in a crisis [39,40]. The messages are useful, especially when the organization has already established a good reputation [41]. If a good reputation has not been established, the stakeholders will experience skepticism instead. Naturally, crisis damages an organization’s reputation [42], so stakeholders may be confused if the company does not have an established reputation when they receive the message about a crisis of the company. Bae and Cameron also found that focusing on CSR messages in a crisis was not valid if the company did not have a good reputation [43]. In addition, Sohn and Lariscy [44] found that people punished a company more with stronger disappointment when they had a higher CSR reputation toward the company. Klein and Dawar [45] found that CSR reputation influences the attribution of crisis directly when people consider CSR to be an important business practice. Various evidence supports that established CSR reputation is one of the important factors that decide people’s reaction to crisis information. This suggests that CSR reputation can function as a predisposition factor. Therefore, we posit our first research question:

RQ1: Does CSR reputation of a company work as a predisposition factor in a crisis?

Furthermore, based on the ADT model in Figure 1, we can predict that a crisis for a protagonist, in other words, positive CSR of a company, would bring about more negative emotional states in the audience. The opposite would happen for an antagonist, in other words, negative CSR of a company. However, it is still premature to compose hypotheses about ADT in a crisis as CSR reputation has also been known to be the source of the buffering effect. Therefore, here we posit another research question:

RQ2: If CSR reputation of a company works as a predisposition factor in a crisis, which emotional states would it elicit?

SITUATIONAL CRISIS COMMUNICATION THEORY

Once a crisis occurs, the image repair process needs to begin right away. In a crisis, an organization will take action somewhere along the continuum of acting in the organization’s interest or acting in the interests of various stakeholders [46]. These actions affect the public’s perception of an organization. SCCT indicates that the higher the publics’ perception of organizational responsibility, the higher the accommodation the organization must make to repair damage to its relationship with the stakeholders [3]. SCCT can be explained based on attribution theory; when a crisis occurs, people usually try to find out the causes of it [47]. They will have a more negative attitude toward an organization if they feel that it has more substantial responsibility for the crisis. Even when a company in crisis provides assistance that could help people who are affected, people are supportive only when they attribute less responsibility for the crisis to the organization [47]. Therefore, it is important for a company with a crisis to accurately gauge the magnitude of the crisis and monitor how media report the crisis and how people respond to it.

According to the SCCT, crises consist of three major categories based on external control and locus/personal control dimensions: (a) a crisis that is external and unintentional is a victim crisis cluster, (b) a crisis with internal and unintentional control is an accidental crisis cluster, and (c) a crisis caused internally and intentionally is an intentional crisis cluster [42]. Coombs and Holladay suggested proper crisis response strategies based on the level of concern for victims and responsibility. For a crisis with low or moderate concern for victims and responsibility, they suggested defensive response strategies. More specifically, for a low responsibility crisis, attack the accuser, denial, and scapegoat strategies are necessary. Crises that cause moderate concern for victims and responsibility acceptance need excuse and justification strategies. Lastly, crises with great concern for victims and responsibility acceptance need a more accommodative strategy, such as compensation and apology.

SCCT can further imply that when people follow crisis information, they may cognitively determine the attribution of the crisis; at the same time, the level of responsibility itself can also generate negative or positive emotion. Weiner [48], who developed attribution theory, insisted that the perception of controllability is related with affective responses such as anger, annoyance, or pity. Specifically, a controllable mistake is linked with anger and an uncontrollable mistake is linked with pity [49]. Weiner and Handel [50] found that more controllable mistakes with internal cause brought about more anger. Weiner, Perry, and Magnusson [51] added the dimension of stability to attribution theory and found that a more stable and uncontrollable situation led to pity, not anger. Cho [9] confirmed that emotion is influenced by crisis attribution; specifically, internal and controllable crises made people angrier, as did repeated crisis (stability). Kühne, Weber, and Sommer [49] also said if the crisis is framed as being the organization’s responsibility, stakeholders feel anger and eventually a demand for punishment. All in all, crisis type, more specifically, people’s perception of the level of responsibility, would also influence people’s emotional states when they are exposed to crisis information about a company. Therefore, we propose a 2 (CSR reputation: Good vs Bad) × 2 (Crisis responsibility: High vs Low) model here (See Figure 2). Using the model, our first hypothesis is:

H1: The level of crisis responsibility shows an interaction effect with CSR reputation toward emotional states.

FIGURE 2

Figure 2. Proposed interaction model between CSR reputation and level of crisis responsibility.

ACCEPTANCE OF CRISIS RESPONSE STRATEGIES

Lastly, regardless of how people perceive its level of responsibility or established reputation, a company that faces a crisis needs to communicate with its stakeholders. It is still uncertain, though, which response strategies work the best for companies. For example, because how people determine the level of responsibility is more of a perception than a fact, taking defensive strategies such as denials or shifting blame is riskier than taking accommodative strategies in any situation [52]. In addition, many times, companies experience negative social evaluations, including stigmatization, regardless of the response strategies they use [14]. As Bundy et al. [14] said, there must be more studies regarding crisis response strategies and how people react to each strategy.

Emotional states following crisis response strategies are one of the puzzles that have been investigated as well. In an accident cluster crisis which has a medium level of responsibility, Coombs and Holladay [53] found that expressing sympathy and compensation were as effective as apology. Jin [4] also found that people who felt more anger than sympathy were more likely to accept denial and scapegoat strategies, while people who felt more sympathy than anger were more likely to accept full apology. In addition, as people felt that a crisis is more severe, they were more likely to accept defensive strategies. Even though there are not many studies available regarding emotional states regarding crisis strategies, based on what Jin [4] found, two hypotheses are posited here:

H2: People who have positive emotion about a company after receiving news about a crisis are more likely to accept accommodative strategies (compensation and apology) after receiving news about a crisis.

H3: People who have negative emotion after receiving news about a crisis are more likely to accept defensive crisis response strategies (attack the accuser, denial, scapegoat, excuse, and justification) after receiving news about a crisis.

MATERIALS AND METHODS

We employed a randomized 2 (CSR reputation: Good vs Bad) × 2 (Product-harm crisis type: Tampering (low crisis responsibility) vs Preventable (High crisis responsibility)) full factorial design to examine the suggested hypotheses and answer the research questions. Two different industries (food and retail) were examined to strengthen the external validity of the study findings.

The scope of this study pertains to general consumers. Participants were recruited through two different means of constructing consumer samples. First, we used Qualtrics’ research participant pool. Then, we recruited participants from diverse courses at a large university in the United States. College students were included because young adults aged 18–25 years old are one of the major target publics for the food industry [54]. They make decisions about food consumption daily and eat in restaurants more frequently than any other age group [55].

A total of 227 participants were recruited through Qualtrics’ research participant pool and from a large southeastern university. The average age of participants was 30.28 (SD = 15.97). There were more female participants (n = 169, 74.4%) than male participants (n = 58, 25.6%). In terms of ethnicity and race, most participants were of non-Hispanic origin (n = 219, 96.5%), with the majority identifying as White (n = 183, 80.6%), followed by African American or Black (n = 21, 9.3%), and American Indian or Alaska Native (n = 1, 0.4%).

Experimental groups were randomly assigned by Qualtrics and distribution of participants was fairly even. A total of 59 participants (26%) were in the good CSR reputation and victim crisis case, 56 participants (24.7%) for good CSR reputation and preventable crisis, 50 participants (22%) for bad CSR reputation and victim crisis, and 62 participants (27.3%) for bad CSR reputation and preventable crisis.

A pretest using college students (N = 83) was conducted to choose companies for stimuli. Ten companies (Panera, Subway, McDonald’s, Starbucks, Chick-Fil-A, Whole Foods, Costco, Target, Sam’s Club and Wal-Mart) were initially chosen from the two industries. After the pre-test, Target and Wal-Mart were chosen to represent the retail industry and Panera and McDonald’s to represent the food industry. The chosen pairs are in the same sector (e.g., retail industry) and have relatively distinctive CSR reputations according to the reputation institution’s RepTrak Pulse Ranking [56]. The pretest tested participants’ familiarity with the companies. We matched consumers’ familiarity with the chosen companies in an attempt to control for the subjects’ familiarity by using real companies. When using real companies, it is impossible to control for potential contamination from pre-existing knowledge about the companies. In order to manipulate crisis-type, real cases of product-harm crises in the food and retail industries were researched. We used victim and preventable product-harm crises, as these crises tend to bring more extreme differences in consumers’ reactions [41]. An E. coli outbreak was chosen to manipulate a preventable product-harm case, while cyanide poisoning was selected to manipulate a product tampering (victim crisis) case.

We created 16 news articles as stimuli: each condition has two stimuli (i.e., CSR reputation stimulus and product-harm crisis stimulus). Although we carefully selected companies to represent good or bad CSR reputation based on secondary data, we believe that using a stimulus would secure clearly distinctive perceptions of CSR reputation between the chosen companies. The stimuli of CSR reputation describe the selected company as the most (or least) responsible company in 2014. In addition, a table containing letter grades for CSR attributes (transparent corporate governance, environmental sustainability, community involvement, corporate giving, and excellence in the workplace) was presented. All other elements aside from CSR reputation were the same across stimuli. The stimuli for the product-harm crises contained information on either an E. coli outbreak (preventable) or cyanide poisoning (tampering/victim). For the preventable product-harm crisis, the stimuli described a multistate outbreak of E. coli infections as being linked to the company’s unsanitary conditions and improper refrigeration system during distribution. The infected food items were described as distributed nationwide while being stored at an improper temperature, causing the cardboard packaging of meats to harbor bacteria, and the company’s facilities were described as showing clear health code violations. For a product-harm crisis, cyanide poisoning was reported to have occurred at one of the company’s largest meat suppliers. The primary suspect was thought to have mixed potassium cyanide into the meat product in an attempt to get his supervisor in trouble. Police are still investigating the incident and are interviewing supply facility managers for further information. The extent of the crisis was the same across stimuli: the incident killed three and sickened 418. The crisis stimuli were made based on real crisis news articles regarding E. coli infections and food tampering.

The second pre-test checked the manipulation of the stimuli. A total of 71 college students were recruited for this pre-test. When respondents accessed the experiment site, the site randomly assigned respondents to one of the 16 conditions. They first read a stimulus about CSR reputation and then read another stimulus regarding crisis type. They were asked to answer manipulation check questions including believability of the stimuli and recall of crisis type. Believability was checked using a three-item 7-point semantic differential scale (Cronbach’s α = 0.93) with three items: unconvincing/convincing, unbelievable/believable, and not credible/credible [57]. Participants were then asked to recall which crisis type the news article addressed. The pretest results revealed no significant differences in terms of believability across conditions, and participants perceived the stimuli as generally believable (M = 5.71, SD = 1.5). Perceived CSR reputation varied as expected and most recalled the type of crisis correctly. Because the manipulation was successful, the study used the developed stimuli for the actual experiment.

We used a CSR reputation measure developed by Kim and Woo [58]. They developed the measure by adopting and modifying the Reputation Institute’s measure, RepTrak Pulse. RepTrak Pulse uses seven key pillars to measure general reputation: products/services, innovation, workplace, governance, citizenship, leadership, financial performance; three of these (i.e., governance, citizenship, workplace) are designed to assess CSR reputation [56]. Five 7-point Likert scale items (1 = strongly disagree, 7 = strongly agree) were used for corporate governance (Cronbach α = 0.95), four items were used for citizenship (Cronbach α = 0.97), and six items were utilized for workplace dimension (Cronbach α = 0.98).

The most frequently expressed feelings in a crisis are reported to be anger and sympathy [4]. Jin used three different words to ask how much people were feeling angry toward a company: “angry,” “irritated,” and “annoyed.” We asked respondents to rank how much they agree that they felt these emotions toward the company using a 7-point Likert-type scale (1 = strongly disagree; 7 = strongly agree). Cronbach’s alpha was 0.95. Jin [3] also used three different words to measure sympathy: sympathy, concern, and compassion. However, after getting a low Cronbach’s alpha, we tried different words chosen from the thesaurus, specifically “sorry” and “pity” instead of “concern.” Again, a 7-point Likert-type scale (1 = strongly disagree, 7 = strongly agree) was used. Cronbach’s alpha was 0.92.

To see if the manipulation of crisis types worked correctly, we asked participants three questions about whom they would blame after they read the crisis news: “circumstances, not the company, are responsible for the crisis” (reverse), “the blame for the crisis lies with the company”, and “the blame for the crisis lies in the circumstances, not the company” (reverse). A 7-point Likert-type scale (1 = strongly disagree, 7 = strongly agree) was used. After adjusting the reverse questions, cronbach’s alpha for this measurement was 0.89.

Following Jin [3], we asked how much participants accept the organization’s crisis responses. Seven response strategies were provided and participants indicated how much they would accept each strategy (1 = not acceptable at all; 7 = totally acceptable).

RESULTS

To answer RQ1 and RQ2, independent sample t-tests were run. Overall, people in the high CSR reputation group felt more sympathy (M = 4.04, SD = 1.39) toward the company than those in the low CSR reputation group did (M = 2.74, SD = 1.30). People in the low CSR reputation group were angrier (M = 4.90, SD = 1.40) than people in the high CSR reputation group (M = 3.74, SD = 1.64), t (225) = −5.71, p < 0.001. In addition, when we ran paired sample t-tests on anger and sympathy, people in the low CSR reputation group felt more anger (M = 4.90, SD = 1.40) than sympathy (M = 2.74, SD = 1.30), t (111) = 9.73, p < 0.001. In the high reputation group, people felt more sympathy (M = 4.04, SD = 1.39) than anger (M = 3.74, SD = 1.64); however, there was no significantly statistical difference. The results show that the CSR reputation worked as a predispositional factor; however, the direction of the results turned out to be opposite of what ADT predicted. Positive CSR reputation entailed more positive emotional states and negative CSR reputation led to more negative emotional states; this represents the buffering effect of positive CSR reputation.

To examine H1, two-way ANOVA tests were run. For anger, both CSR reputation, F(1, 223) = 34.420, p < 0.001, ƞ² = 0.13 and crisis type, F(1, 223) = 51.833, p < 0.001, ƞ² = 0.19, had main effects (see Table 1). Table 1 also shows that CSR reputation, F(1, 223) = 53.60, p < 0.001, ƞ² = 0.19 and crisis type, F(1, 223) = 26.74, p < 0.001, ƞ² = 0.11, also had main effects on sympathy. As both CSR reputation and crisis types showed main effects, we ran independent sample t-tests for crisis type on people’s emotional states as well. Those who read the preventable crisis scenario were angrier (m = 4.98, sd = 1.39) than those who read the product tampering or victim crisis scenario (m = 3.59, sd = 1.57), t(225) = −7.12, p < 0.001. For sympathy, participants who read the victim crisis scenario were more sympathetic (m = 3.90, sd = 1.38) than those who read the preventable crisis scenario (m = 2.94, sd = 1.45), t(225) = 5.12, p < 0.001. However, an interaction effect was not found. Therefore, H1 was not supported.

TABLE 1

Table 1. Univariate results for Anger and Sympathy.

Because both CSR reputation and crisis type showed main effects, we wanted to determine which variable was a bigger predictor on emotional states using stepwise regression analyses. To do that, we wanted to use CSR reputation and blame measures as those were continuous variables. First of all, we ran an independent sample t-test to see if these measures represent experimental groups used for ANOVA tests. For CSR reputation, participants perceived the companies that were used for the good reputation group to be better CSR companies (m = 5.05, sd = 0.85) than the companies that were used for the bad reputation group (m = 4.13, sd = 1.19), t (225) = 6.743, p < 0.001. For blame on companies, participants put more blame on companies in the preventable crisis scenario (m = 4.55, sd = 1.79) than in the victim crisis scenario (m = 3.97, sd = 1.51), t (225) = −2.61, p < 0.05. Therefore, we decided to use these measurement for stepwise regression analyses.

As Table 2 illustrates, when running a stepwise regression analysis on anger, blame was the first predictor, β = 0.26, t (225) = 4.07, p < 0.001, and the model was also significant, R² = 0.07, F (1, 226) = 16.60, p < 0.001. CSR reputation was added in the second block. In the second block, the standardized coefficient beta for blame was increased and statistically significant, β = 0.27, t (224) = 4.26, p < 0.001. CSR reputation was also a statistically significant but negative predictor, β = −0.23, t (224) = −3.66, p < 0.001. The statistical power of the second model was increased and the model was statistically significant, R² = 0.12, F (2, 226) = 15.50, p < 0.001.

TABLE 2

Table 2. Stepwise regression analyses predicting anger.

Table 3 shows the results of a stepwise regression analysis on sympathy. At this time, CSR reputation was the first predictor, β= 0.40, t (225) = 6.63, p < 0.001, and the model was also significant, R² =0.16, F (1, 226) = 43.92, p < 0.001. Blame was added in the second block. In the second block, the standardized coefficient beta for CSR reputation was increased and statistically significant, β = 0.41, t (224) = 7.02, p < 0.001. Blame was also a statistically significant, but negative predictor, β = −0.27, t (224) = −4.59, p < 0.001. The statistical power of the second model was increased and the model was statistically significant, R² = 0.24, F (2, 226) = 34.48, p < 0.001. In summary, blame was a bigger predictor for anger, while CSR reputation was a bigger predictor for sympathy.

TABLE 3

Table 3. Stepwise regression analyses predicting sympathy.

Emotional states and acceptance of crisis response

To examine H2 and H3, we ran regression analyses on all crisis response strategies by anger and sympathy. Table 4 shows that the attack the accuser, compensation, and apology strategies, which had generally higher acceptance by participants, did not vary based on emotion. It was also found that anger did not predict acceptance of any response. Sympathy was a strong predictor for strategies of denial, β = 0.28,F(2, 224) = 11.52, p < 0.001, scapegoat, β = 0.37, F(2, 224) = 28.57, p < 0.001, excuse, β = 0.36, F(2, 224) = 20.56, p < 0.001, and justification, β = 0.32, F(2, 224) = 10.91, p < 0.001. Among these, emotion explained 20% of variances for the scapegoat strategy, which represented the strongest model. Therefore, both hypotheses were not supported and H3 was turned out to be opposite of the expectation.

TABLE 4

Table 4. Multiple regression results on acceptance of crisis response strategies by Anger and Sympathy.

DISCUSSION

First, both CSR reputation and crisis type influenced people’s emotional states. Speaking of CSR reputation, Kim [41] and Shon and Lariscy [44] both said that CSR reputation would not be tarnished much in a product-harm crisis, but would be more damaged in CSR-related crises, such as those related to sustainability and organizational ethics. Our results also showed that CSR reputation acted as a buffer in a product-harm crisis regardless of whether it was a victim crisis or a preventable crisis. CSR reputation played a role as a predispositional factor as people had more sympathy, which is a positive feeling, toward a company with a better CSR reputation. Participants were also angrier toward a company with a bad CSR reputation. When we looked more closely into CSR reputation and the level of blame assigned to companies in a crisis, we found that CSR reputation was a bigger predictor for sympathy, which would illustrate ADT that was explained by Zillmann and Bryant [20] at least partially. In more detail, ADT explains that observing protagonists’ failure would entail distress, or negative hedonic valence. Sympathy is not a negative feeling; however, it does not have a positive hedonic valence, such as pleasurable feelings, either [59]. Schindler, Körner, Bauer, Hadji, and Rudolph [60] also said that people feel sympathy when a protagonist is not responsible for his or her misfortune, which they explained as a negative hedonic quality.

However, we also found same results as previous studies; the type of crisis was a main effect. Like Coombs and Holladay [61], we also found more anger in the case of preventable crisis and more sympathy in a victim crisis. Primarily, how much people blame a company in a crisis was a bigger predictor of anger than CSR reputation. Of course, CSR reputation also predicted anger; however, positive CSR reputation brought about less anger, which is opposite of what ADT predicts. Therefore, our findings show that anger is better explained by SCCT than ADT.

Second, sympathy was the emotion that drove people to accept defensive strategies (denial, excuse, scapegoat, and justification). In other words, participants who felt sympathy toward a company were more likely to accept denial, excuse, scapegoat, and justification strategies of the company. In addition, as they had a more positive CSR reputation toward a company, they were more likely to accept scapegoat, justification, and compensation strategies.

Lastly, participants who received a preventable crisis scenario were more likely to accept apology and less likely to accept scapegoat and excuse strategies. These are different results from Jin’s [3]; however, it is reasonable considering she used four different cases including dog fighting, the Virginia Tech University shooting, bridge collapse, and product tampering, which were more emotionally volatile than the cases we used. However, our results may be explained by Jin’s finding that sympathy was connected with proactive coping strategy and more rational thinking. Anger was not connected well with a cognitive coping strategy. Accepting a crisis response strategy may require more rational and cognitive thinking than an emotional reaction. Furthermore, Ducassy [62] said that the buffering effect of CSR is temporary. He looked at financial performance after crisis and found that this buffering effect did not last longer than six months. Our asking about the participant’s acceptance of each crisis response would have stimulated more long-term oriented thinking as opposed to asking about their emotion.

PRACTICAL AND THEORETICAL IMPLICATIONS

In a product-harm crisis, a high CSR reputation makes people feel less angry and more sympathetic toward the company. Considering sympathy, assuming CSR reputation works as a predisposition factor, people would view companies with a high CSR reputation as a protagonist. According to affective disposition theory, people hope to observe the success of a protagonist and the failure of an antagonist [23]. Our results indeed showed that people were more sympathetic toward the protagonist (a company with high CSR reputation) and angrier with the antagonist (a company with low CSR reputation). It is important to invest in CSR activities because they are an important part of corporate reputation [63], and can help to make a company a protagonist, not an antagonist, in people’s mind.

The findings of CSR reputation as a predisposition in a crisis provide insights not only to crisis communication experts. ADT also explains how fandom works for people who follow their beloved characters and sports teams. A protagonist of a crime drama can bring joy to the viewers [64]. Wann [65] found that sports fandom even improves the well-being of sports fans. Similarly, CSR activities also improve job satisfaction, the well-being of employees [66], and customer engagement [67]. Considering CSR’s function as a predisposition, the findings of our study suggest to management-level officials of companies that CSR can also generate enjoyment for its employees as well as customers.

However, crisis managers also need to be aware that crisis communication is a long-term effort even though it begins immediately after a crisis occurs. CSR reputation can work positively for a company, but its effect does not last long. Crisis managers should pay careful attention to their crisis responses. Even though people would have certain feelings about a company in a crisis, they also tend to accept crisis responses rationally rather than emotionally and generally accept accommodative strategies including compensation and apology. They will also accept a strategy of attacking the accuser. Therefore, crisis managers should collect correct information about the crisis and decide on the appropriate responses. As Kim [39], Coombs [3], and Ducassy [62] all warned, the mechanism between CSR reputation and crisis is very complex. Therefore, it is important for crisis managers to keep researching different cases and updating their crisis manuals.

This study also has a theoretical implication. Even though emotion and expectations have been studied in crisis communication research, media effect theories have not been investigated much. We introduced affective disposition theory and found a possible connection and contribution of these theory for crisis communication research. Especially in a crisis, people get information through various media channels [68], which would initiate affective disposition process. Future researchers need to include more and different situations and measurements of emotions utilizing theories related to media effects.

LIMITATIONS

As this is an experimental study, we have some notable limitations. First, even though we used the names of real companies and tried to provide realistic news articles, participants might notice that the crisis was hypothetical. In addition, because we used real companies, participants might already have established strong predispositions toward them. Our results show that people do not easily change their opinion about organizations in a crisis, and this might influence the level of manipulation in our experiment. We tried to follow the tradition of experiment studies applying ADT, which use real names often [64,69,70]. However, ADT studies mostly use existing situations or entertainment contents, while crisis studies cannot truly realistically simulate a situation with real companies. Therefore, future experiments may need to use hypothetical companies. Second, as Jin [3] did, we only measured anger and sympathy even though many different emotions can be generated in a crisis [71]. Schadenfreude would be a good emotion to examine as well, as Coombs and Holladay [61] did. Third, it is possible that participants might not feel anger toward the company, but they might feel anger toward the issue. Majid [72] noted that it is difficult to reflect emotional diversity using words because when people see a word, they may conceptualize the word instead of thinking of how they experience the feeling. Therefore, future researchers need to try different ways of measuring emotion. Especially, to examine ADT more effectively, a spectrum of hedonic valence (e.g., pleasing to displeasing, appealing to unappealing, etc.) should be measured instead of discrete emotions, such as anger and sympathy. Fourth, there are more crisis types than product-harm. As Kim [39] said, expectancy-violation may be observed if a more CSR-related crisis is examined. More and various types of crises need to be investigated. Lastly, we only looked at external publics in our study. However, internal publics may be influenced by CSR activities as well.

AUTHOR CONTRIBUTIONS

CWW and YK worked together on the literature review, experimental design, including the stimuli development, and data collection. YK developed the CSR reputation scale. CWW analyzed the data and wrote results, discussion, and implications sections with the consultation of YK.

CONFLICTS OF INTEREST

The authors declare that there is no conflict of interest.