Emoticons in Crisis Communication and their effect on Corporate Reputation

Faculty of Social and Behavioural Science Dr. Joost Verhoeven

Emoticons in Crisis Communication and their effect on Corporate Reputation.

Master Thesis Selina Stolzenbach

June 28, 2019

UVA-ID: 11767340

selina.stolzenbach@student.uva.nl Master Communication Science Track: Corporate Communication

Abstract

Our daily communication becomes more and more computer mediated, in an interpersonal environment as well as in a corporate communication context. As computer mediated

communication via social media platforms enables organizations to inform their stakeholders quickly and directly, it is seen as highly useful for crisis communication. However, it also bears challenges as the nonverbal cues, like mimic and gestures are missing. In CMC

emoticons assist as nonverbal cues (Hof, 2016), but literature has also shown, that receivers of computer-mediated messages, had extreme difficulties to perceive the correct emotions, attentions, intentions and attitude (Lo, 2008). The aim of this study is to examine the effect of emoticons in organizational crisis communication on the corporate reputation. The research question was answered with help of a survey-embedded experiment and a mixed-factorial design. The effects of the emoticons were predicted to vary depending on the crisis type and the participants’ persuasion knowledge. However, this study did not find a significant

relationship between emoticons in a crisis response and the corporate reputation. Neither was the relationship influenced by the crisis type or the persuasion knowledge of the participants. Nevertheless, the study gives interesting insights, on how the effects of emoticons in

organizational crisis responses could be further researched.

Keywords: computer-mediated communication, crisis communication, crisis response, emoticons, emotional corporate communication

Emoticons in Crisis Communication and their effect on Corporate Reputation. In times of organizational crises, it is seen as very important to inform stakeholders quickly and directly (Utz, Schultz & Glocka). Social media are mostly seen as opportunity to achieve this. And while they do enable a more direct and faster way to communicate with stakeholders (Veil, Buehner & Palenchar, 2011), it’s also challenging as the nonverbal cues, e.g. facial expressions, are missing in computer mediated communication (CMC) (Dresner & Herring, 2010; Krohn, 2004; Walther & D’Addario, 2001).

Emotions that can be conveyed through mimic and gestures and can influence the interpretation of a message (Homan & Cheshin, 2012; Van Kleef, 2012) are very important in organizational crisis communication, as crises are not only sudden and unexpected events, but also emotional events (Coombs, 2007). While stakeholders attribute responsibility for an event to an organization, they will react emotional and develop feelings of anger (Utz et al., 2012). However, emotions are not only evoked on the side of stakeholders during a crisis, but also on the side of the organization. An experiment revealed that communicated emotions like shame and regret have a positive effect on corporate reputation. (Van der Meer & Verhoeven, 2014).

However, research has focused on media, where the nonverbal cues like gestures and mimic was visible for the public eye, e.g. audiovisual material of a spokesperson. But how can emotions be conveyed in text messages that are distributed over the internet?

In CMC emoticons are used to assist to carry emotions and attitudes. About 92% of online users use emoticons and people sent about six billion emoticons every day (Hof, 2016). Emoticons are not only used by stakeholders, but they have also found their way into

corporate communication. Customers of Domino’s Pizza could order a Pizza by tweeting a pizza slice emoticon and Pepsi sent out customized emoticons to people that tweeted the hashtag #PepsiHalftime during the Super Bowl (Hof, 2016).

(Utz et al., 2012) and emotions displayed by the organization affect the corporate reputation as well, emoticons in the crisis responses may enhance this relationship. On the other hand, literature has shown, that receivers of computer-mediated messages, had extreme difficulties to perceive the correct emotion, attentions, intentions and attitude (Lo, 2008) and therefore research is needed to better understand how people perceive and react to emoticons in

corporate communication. With the help of a survey-embedded experiment, this study aims to assess the research question: Does the use of emoticons in crisis communication affect the corporate reputation?

Theoretical Framework

In this chapter the theoretical background for this study is presented, to explain how the communication of emotions is connected to the use of emoticons. Furthermore, previous research about the perception of emoticons is examined and it’s connected to crisis

communication. The chapter goes on about the role of persuasion knowledge and crisis type in the context of emoticons in crisis communication.

Emotional Communication

Communication does not only include information for the receiver but also convey emotions. Research on effects of emotions suggest that emotions do not only affect the person who expresses and experiences emotions, but also the person who observes and perceives them (Hareli & Rafaeli, 2008; van Kleef, 2014; van Kleef et al., 2012). The emotions as social information theory (EASI) is about “interpersonal consequences of emotions in social and organizational life” (van Kleef, 2014, p.1147) and can be used for several organizational phenomena, e.g. group decision making, leadership, persuasion and negotiation (van Kleef, 2014). EASI theory claims that there are two processes that influence receivers of emotional expressions: (1) inferential processes and (2) affective reactions. In the latter, the person that perceives the emotion of a message, behaves in a certain way because of the affective reaction

to the emotional message. Additionally, by receiving the emotional message, the receiver also learns about the sender’s interpretation of the situation (inferential process) (van Kleef, 2014). Nonverbal cues

To react affective to emotional messages, the observer first needs to grasp the senders’ emotions, e.g. with the help of nonverbal cues (van Kleef, 2012). Nonverbal behaviours include actions that differ from speech, such as postures, facial expressions, positions or hand and arm gestures (Mehrabian, 1972). In a broader sense vocal phenomenon like speech pauses, frequency range or speech duration can also be understood as nonverbal behaviour (Mehrabian, 1972). One of the five functions of nonverbal behaviour as proposed by Ekman & Friesen (1969) is affect display, which is the most important function in this context. The other categories are emblem (nonverbal cues that can be translated into words, e.g. a smile), illustrator (act of emphasis, e.g. nodding), regulator (signals for the speaker to clarify or hurry up) and adaptor (movements that refer to bodily needs, e.g. scratching) (Mehrabian, 1972).

In an organizational context, research has shown, that nonverbal cues had an impact on the public and that it negatively affected a company’s stock market (ten Brinke & Adams, 2015). The adequate demonstration of the emotion sadness during a public apology increased the organization’s performance (ten Brinke & Adams, 2015). In contrast, when happiness was displayed by the spokesperson in the same situation, the organizational performance

decreased. This indicates that not only communicating emotions has an impact on the stakeholders, but also which emotion is displayed (ten Brinke & Adams, 2015).

What does it mean to show off emotions adequately? Research has highlighted that the affective reactions and the inferential processes depend on the appropriateness of the

displayed emotion (van Kleef, et al. 2012). Emotions can be perceived as inappropriate, when they seem inauthentic, which can lead to negative reactions through a manipulative, dishonest and unethical perception of the sender (Coté, Hideg & van Kleef, 2013; van Kleef, 2014). A

study by Cheshin, Amit & van Kleef (2018) found that the intensity of the emotional display affected the perceived appropriateness of messages by a service provider. This also affected the trust in the organization as well the satisfaction with the service. As the intensity of emotional display can be inferred from the nonverbal cues (Mehrabian, 1972), it shows that the nonverbal cues affect the perception of the sender.

With the raise of social media and therefore computer-mediated communication it is questionable how nonverbal cues can be perceived in text-based CMC. In the 1980s

emoticons were invented to help with that exact situation. Emoticons

The term emoticon is a merger of the two words “emotion” and “icon”. It refers to a graphic sign which is often used in combination with text in computer-mediated

communication. It was first used in 1982 by a computer scientist to indicate if something that was posted to a discussion forum was meant as a joke or not. Back then an emoticon was a rotated smiley (e.g. “:-)” and “:-( “) (Dresner & Herring, 2010; Krohn, 2004). Since ongoing development, the image and number of emoticons has developed to a large degree. A lot of them still use only ASCII (American Standard Code for Information Interchange) symbols (as the original emoticons) and others are now graphically edited (e.g. “😊”). While the graphic design of emoticons began to become more sophisticated, numerous emoticons have been developed that represent objects (a heart or a flower). However, the most used ones mimic facial expressions (Dresner & Herring, 2010).

The two original emoticons were created to indicate affective states, which are implied through nonverbal cues in Face-to-Face communication. Because these nonverbal cues are missing in the computer-mediated communication, emoticons were invented to surrogate them (Dresner & Herring, 2010; Walther & D’Addario, 2001). However, they do not only function as indicators of emotions, which are translated to facial expressions, but also to

express the way certain messages should be interpreted, e.g. an anxious smile or indicating joking intention with a wink (Dresner & Herring, 2010).

In general, it seems that emoticons can be used and interpreted in two ways: an emotion in terms of a facial expression or to express underlying intention and implications of the speech act in terms of facial expressions but also body gestures and illocutionary forces (Dresner & Herring, 2010). Additionally, emoticons often indicate a certain playfulness, e.g. when emoticons are used to downplay messages for example with a happy face after a rather serious message (Dresner & Herring, 2010). Research has shown, that happiness ratings were improved, when a smiley face was added to a positive message, while the positivity was reduced when a frown was added (Walther & D’Addario, 2001). However, literature has also shown, that receivers of computer-mediated messages, had extreme difficulties to perceive the correct emotion, attentions, intentions and attitude (Lo, 2008). The presence of emoticons significantly changed the internet user’s perception of the received message. Therefore, emoticons can assist with the correct interpretation of attitude and emotions in CMC (Lo, 2008). It was also found that with emoticons present, the level of arousal was higher. It increased smiling and decreased frowning activity (Thompson et al., 2016). Following, emoticons increased enjoyment in CMC (Thompson et al., 2016). When emoticons were used by an organization’s employee while providing information online (web care) they were perceived as warmer, but less competent (Li et al., 2019).

In conclusion, research has shown that emoticons have the functions to downplay a computer-mediated text message (Dresner & Herring, 2010), help the receiver of CMC with the correct interpretation of emotions and attitudes (Lo, 2008), enhance the level of arousal and smiling activity (Thompson et al., 2016) and improve the happiness rating (Walther & D’Addario, 2001). Furthermore, emotions communicated by the organization can improve the corporate reputation (van der Meer & Verhoeven, 2014) and thus, following assumption was made:

H1: A crisis response with emoticons present will lead to a more positive corporate reputation than a crisis response without emoticons.

Corporate Reputation

Corporate reputation “is an aggregate evaluation stakeholder make about how well an organization is meeting stakeholder expectations based on its past beaviors” (Coombs, 2007, p.164). Stakeholders are defined as any group that can have an impact on the organization’s behaviour or can be affected by it (Coombs, 2007). A positive corporate reputation is crucial to gain customers, increase the financial performance and build an advantage over

competitors (Coombs, 2007). A stakeholder builds the reputation based on experiences with the brand, reports about the organization from media as well as from other customers (word of mouth) (Coombs, 2007). An organizational crisis, e.g. a product recall or misbehaviour of an employee, can give stakeholders reason to think and feel about the organization negatively, which in turn threatens the corporate reputation. This doesn’t only change how customers think about the organization, but it also influences customers’ behavioural intentions as well. It means that stakeholders change the way they interact with the company which may results in not supporting the organization any longer or stop buying their products (Coombs, 2007).

Therefore, corporate reputation is an important asset of an organizations, especially because the pre-crisis relationship with an organization and the corporate reputation based on it, work as a buffer in times of a crisis and can mitigate the negative effect of a crisis. Hence it is important for an organization not to lose too much of their reputation, in case future crises. The role of emotions during a crisis

Crises are not only sudden and unexpected events, threatening the corporate reputation., but they provoke affective reactions as well, e.g. anger (Coombs, 2007).

According to the situational crisis communication theory, emotions are generated depending on the attribution of crisis responsibility: higher attribution of crisis responsibility towards the

organization lead to stronger feelings of anger on side of the stakeholders, which in turn influenced the behavioural intentions of the organization’s stakeholder (Coombs, 2007). This is supported by studies, where the public used emotions, like anger or fright to cope with crises (Jin, 2010; Kim & Jin, 2016). If a corporate preventable crisis happened, the public described more sadness and used proactive coping strategies, e.g. positive thinking (Jin, 2010). While this is focused on emotions that stakeholders experience in times of a crisis, it is also possible to think of emotions that are experienced and communicated by the organization during a crisis. Communicating the emotions regret and shame lead to a better corporate reputation, compared to no communicated emotions. They increased the acceptance of the message by the organization and reduced feelings of anger (van der Meer & Verhoeven, 2014). Sadness can be understood as reaction to something that happened but was not

supposed to occur and therefore doesn’t meet the expectations (Tam, 2015). It is a response to incongruency between the expectations and the reality (Tam, 2015). According to the theory of emotional disparity (Tam, 2015), sadness marks the stage of acceptance of the situation, not finding an adequate solution and agreeing that the reality can’t be changed. It comes after the initial stage, which is shock or surprise and marks the detection of incongruency. When the shock has passed, a feeling of anger spreads, as the observation is made that nothing can be changed about the incident (Tam, 2015). The emotional state of sadness seems highly applicable for a crisis response of an organization, as a crisis is a sudden and unexpected event. The moment an organization issues a response, it should have overcome the state of shock and anger and conclude that there is nothing that they can change about the situation now, except for accommodating the affected stakeholders.

Hence, the nonverbal cues can affect the corporate reputation in times of a crisis. With the rise of new media and the missing nonverbal cues in text messages, it seems necessary to examine how emoticons can support with conveying emotions and attitudes in CMC,

Crisis type

The publics’ evaluation of the corporate reputation after a crisis, depends on the fact to what extent the stakeholders hold an organization responsible for the incident. The SCCT distinguishes three crisis types, which reflect the public’s attribution of crisis responsibility to the organization: victim cluster, accidental cluster and preventable cluster (Coombs, 2007). Crises with weak attribution of the organization’s responsibility fall within the victim cluster (e.g. product-tampering or a natural disaster), while crises with a minimal crisis responsibility attribution, which are regarded uncontrollable or unintentional are defined in the accidental cluster (e.g. product harms, technical-error accidents). When stakeholder attribution of crisis responsibility is very strong and the incident is perceived intentional and purposeful, the crisis falls within the preventable cluster (e.g. organizational misdeeds, human-error accidents) (Coombs, 2007).

The SCCT proposes that organizations need to choose specific response strategies depending on the crisis type to minimize the reputational damage. Stakeholders will expect more from an organization if the attribution of crisis responsibility is strong and it calls for a crisis response strategy that approaches all the stakeholders which are affected (Coombs & Holladay, 2002). When organizations experience a preventable crisis, the organizations are held accountable for it and the corporate reputation is the less favorable than for organizations that suffer crises within the victim or accidental cluster. Coombs & Holladay (2002) showed that in all three clusters crisis responsibility had a significant impact on corporate reputation.

Depending on how the organization’s role is perceived during an incident, the emoticon in the organization’s crisis response might be perceived differently. Dresner & Herring (2010) observed that emoticons indicate a certain playfulness. The public expects an organization that experienced a crisis within the preventable cluster to be accommodating the affected stakeholders (Coombs & Holladay, 2002). An emoticon in the crisis response may

have a negative effect on the corporate reputation, because the public evaluates the emoticon as inappropriate for the situation.

H2: A crisis response with emoticons will lead to a more negative corporate reputation than a crisis response without emoticons, if the crisis is within the preventable cluster.

This is expected to be different for crises within the victim cluster, as the playfulness may be forgiven by the public due to the weak attribution of crisis responsibility. When the crisis responsibility was higher, emotions like sadness, anger and more frightened feelings were stronger than for crises with crisis that seemed uncontrollable for the organization

(victim cluster) (Jin, 2010). The public is also in higher need of emotional support when crises are unpredictable and uncontrollable (Jin, 2010). Therefore, the effect of emoticons to help with the interpretation of emotions in CMC (Lo, 2008) and the ability to lift the feeling of happiness (Walther & D’Addario, 2001) is predicted to outweigh the negative effect of playfulness in the victim cluster as the emoticons seem more appropriate.

H2a: A crisis response with emoticons will lead to a more positive corporate reputation than a crisis response without any emoticons, if the crisis falls within the victim cluster.

Persuasion Knowledge

The difference between emoticons in CMC and nonverbal cues in Face-to-Face communication is the intentional use of them (Dresner & Herring, 2010; Walther &

D’Addario, 2001). In the latter, facial and body expressions are used rather unconsciously and unintentionally (Dresner & Herring, 2010; Krohn, 2004), but when using an emoticon for a computer mediated message, it happens consciously and intentionally. Emotive states that are given unintentionally are usually more reliable than intentionally given emotional expressions (Dresner & Herring, 2010; Walther & D’Addario, 2001). Depending on the awareness of the receiver about the intentional characteristic of emoticons, it may influence the perception of the message and the emoticon. This relates closely to the concept of persuasion knowledge.

Persuasion knowledge refers to the phenomenon, that consumers over time develop knowledge, which helps them to identify, how, why and when organizations try to influence them (Friestad & Wright, 1994). This knowledge can be gained in many ways, such as social interaction with friends, family and co-workers, from observing organization’s behavior and commentary on advertising tactics in the media. The persuasion knowledge can play an important role in affecting behavioral intentions of persuasion targets (Friestad & Wright, 1994). The persuasion knowledge model differs between three different sub-constructs. First, persuasion knowledge, which refers to the knowledge of persuasion practices. For crisis communication this means that stakeholders with a higher persuasion knowledge have a greater ability to understand underlying motives of crisis response strategies. Second, agent knowledge (knowledge about the organization) and it helps the public to understand why a company uses a specific crisis response strategy (Ham & Kim, 2017). The third sub-concept is topic knowledge (information about the situation in general, e.g. the organizations stock market performance) (Ham & Kim, 2017). A person’s persuasion knowledge is not always activated, but it is predicted to be activated in situations where they notice something

unpredicted (Friestad & Wright, 1994). This relates to crises, as they are unexpected events by definition and it furthermore relates to emoticons in the crisis response of an organization, as the stakeholders are predicted not to expect them.

Hence, it is highly interesting to look further into the connection between emoticons and persuasion knowledge, as most people not only perceive emoticons but also use them in their own CMC. Therefore, it is possible that some people are aware of the intentional character of emoticon use, which could be perceived as strategic use of it. Hence, persuasion knowledge could be very high in terms of emoticon use in a strategic way. When stakeholders notice a strategic meaning behind a message, it is possible that they disengage from the conversation, get sidetracked from the message or disregard the message (Friestad & Wright,

1994). In conclusion, persuasion knowledge is predicted to moderate the relationship between emoticon use and corporate reputation.

H3: A crisis response with emoticons present will lead to a more negative corporate reputation than a crisis response without any emoticons, particularly for participants with higher persuasion knowledge.

If a persuasion attempt is identified, stakeholders also evaluate how effective, fair and appropriate they think of the strategy (Friestad & Wright, 1994). As predicted in H2 and H2a, the appropriateness of emoticons is predicted to differ depending on the crisis type. Higher persuasion knowledge is forecasted to enhance the relationship, as a strategic attempt to make stakeholders believe that the organization feels emotional is evaluated even more negative than just the felt inappropriateness of the emoticon itself.

H3a: A crisis response with emoticons present and to a crisis within the victim cluster will lead to a more positive corporate reputation for participants with higher persuasion knowledge than with lower persuasion knowledge.

H3b: A crisis response with emoticons present and to a crisis within the preventable cluster will lead to a more negative corporate reputation for participants with higher persuasion knowledge than with lower persuasion knowledge.

Method

Research Design

The research question was answered in two steps. Firstly, a pretest in form of a survey was conducted to find crisis scenarios according to the criteria and to test the emotional perception of several emoticons. Secondly, to test the hypotheses, the main study was carried out in form of a survey-embedded experiment with a 2x2 mixed-factorial design. Emoticons was the within-subject factor (2 levels, namely: absent and present) and Crisis Type was the

between-subject factor (2 levels, namely: victim and preventable cluster). There were two experimental groups, who received two tweets each, representing the organisational crisis responses. Both tweets referred to the same crisis type, but one tweet contained emoticons and the other didn’t.

Development stimulus materials

To ensure that the stimuli were according to the criteria (relevant, realistic, emotional, realistic to tweet about), a pretest with four crisis scenarios was conducted (see Appendix A). The pretest was also used to assess the emotional perception of different combinations of emoticons.

Design Pretest

The pretest was designed as a survey and has guided the decisions which crisis

scenarios and emoticons were used for the main study. Four crisis scenarios were considered. The survey had two versions: Version 1 included two different crisis scenarios in the victim cluster and version 2 included the other two crisis scenarios in the intentional cluster.

Procedure Pretest

The participants from the personal network of the researcher received a link to the survey experiment via social media posts on Facebook and LinkedIn. As a first step, the participants received general information about the experiment, where they gave their consent to participate. Afterwards, they were randomly assigned to see two stimuli, either in the victim or the preventable condition. They read texts, which described fictitious product

recalls, including the company name, the product and the reason for recalling it (see Appendix A). After the exposure to each stimulus, participants answered questions which assessed the selection criteria for the crisis scenarios.

The second part of the pretest assessed the perceptions of the expression of the emotion sadness, by emoticons. They were presented with multiple combinations of

emoticons, all of them were ought to be sad. To make sure that the manipulation was not to subtle, a string of emoticons was presented (see Table 1).

Which Emotion do the Emoticons evoke for you?

Open question

How sad would you rate the combinations of Emoticons?

7-point Likert Scale

Table 1: Pretest questions about emoticons

On the last page, the participants were thanked for taking part in the experiment and debriefed. The debriefing emphasized that the crisis scenarios and the companies were fictitious.

Sample Pretest

The sample originally included 36 respondents. Of these 36 participants, six were excluded caused by a dropout. Thus, analyses were conducted over a final sample of 30 participants in total (N=30).

Results Pretest

Crisis Scenarios

Four crisis scenarios were tested for the criteria relevance, realism, emotional and tweet expectations. For each crisis scenario a new variable was computed that showed the mean of the four questions (see Table 2).

Crisis scenario Relevance Realistic Emotional Tweet expectation

VRIBA (baby food) 6.13 6.19 4.56 5.75

DÄCK (car tire) 6.13 5.00 3.94 5.63

RENTVA (laundry detergent)

6.57 5.93 4.00 5.86

PIVA (beer bottles) 6.57 6.00 4.93 6.14

Table 3: Emoticons and Sadness

The beer bottle scenario scored highest in the categories relevance, emotional and tweet expectation. Only in the category realistic the incident scored second highest, while the baby food scenario had the highest mean. In the categories relevance and tweet expectation the baby food scenario scored lower than the laundry detergent scenario. However, it had a higher mean in the emotional category, which was considered very important for the current study. Following, the baby food incident was chosen over the laundry detergent scenario. As second product recall for the main study the beer bottle incident was chosen. As the car tire recall had the lowest scores in all categories it wasn’t further considered.

Manipulation check crisis type

The pretest included questions to determine whether the manipulation check of crisis type was valid. Three questions were asked for this manipulation check, which were

computed to one new variable for the analysis. Respondents that saw the crisis scenarios within the victim cluster (N = 16) scored lower (M = 4.25, SD = 0.71) than participants that saw the incidents within the preventable cluster (N = 14; M = 4.38, SD = 0.51). An

independent t-test showed that there was no significant difference between the two means (t (28) = -0.57, p = .575). Therefore, the manipulation of the crisis type did not work. This did not affect the results of the pretest but had to be considered when creating the stimuli for the main study.

Emoticons

When asking what emotion was provoked by the

emoticon, the desired emotion was sad. Sadness was mentioned for each of the four emoticon that were displayed. However, they were differences on how often the emotion sadness was

spontaneously mentioned. The crying emoticon evoked sadness by 80% of the participants. Emoticon “Sadness” 8 4 24 6

Table 4: Means Emoticon Combinations

For the other three emoticons sadness was mentioned by 13.34% up to 26.67% of the respondents.

When asking how sad different combinations of three emoticons were rated, combination 1 and 5 were rated the saddest. Combination

1 included the three emoticons that also provoked sadness in the open question the most. Thus, it was used in the stimuli for the main study.

Stimuli

The aim of the pretest study was to ensure that the crisis scenarios were realistic, relevant, emotional and realistic to tweet about and which meanings different emoticons have. Especially, in combination with the emotion sadness and to make sure that the manipulation of the crisis type worked. The tested crisis scenarios can be found in Appendix A. Following the pretest results two crisis scenarios were selected: One regarding baby food and the other regarding beer bottles. Depending on the crisis type, the reason for the product recall differed for both crisis scenarios. As the manipulation of the crisis type did not work in the pretest, they were rewritten. In the victim cluster the responsible party was changed from a former employee to a competitor who tampered with the product. In the preventable cluster the rather vague reason of “non-compliant behavior of an employee” was changed to a missed

maintenance due to the organizations intent to save money. The Stimuli consisted of two parts. The first part was a description of the situation (Appendix B) and the second part was an organizational tweet (Appendix C), including an information about the product recall, an apology by the organization, and in the emoticon present condition the tweet will be

completed with a combination of three emoticons. This string of emoticons was selected based on the pretest results. It evoked the saddest reaction and consisted of three emoticons that also induced the emotion sadness spontaneously.

Number Emoticons Mean

1 4.20

2 3.97

3 4.13

4 3.83

Procedure Main Study

The participants received a link to the survey-embedded experiment via social media posts on Facebook and LinkedIn. As a first step, the participants read general information about the experiment and gave their consent to participate in the experiment.

Thereafter, the participants were presented with the description of a product recall (see Appendix B) and were asked to imagine the product recall in real life. Participants were randomly assigned to either see the crisis scenario within the victim or the preventable cluster. This was manipulated by blaming a competitor or the organization. After being introduced to the crisis scenario, they saw the organizational tweet about the respective incident, without any emoticons. Following, they were introduced to a second crisis scenario. The participant was confronted with the same crisis type but this time emoticons were used in the

organizational tweet. After exposure to each organizational tweet the corporate reputation was measured. In the end, the persuasion knowledge the manipulation check and the

demographics were assessed.

On the last page, the participants were thanked for taking part in the experiment and debriefed. It was emphasized that the crisis scenarios and the companies were fictitious and that the tweets were constructed by the researcher.

Sample Main Study

The sample originally included 169 respondents. Of these 169 participants, 16 of them were excluded because of a dropout. Another four were deleted because the participants took less than three minutes to complete the survey and it is assumed that they did not pay enough attention to the survey. Analyses were thus conducted over a final sample of 149 participants in total (N=149). Most of them (59.1%) were between 18 and 25 and the average age was 26.3 years. Of the 149 participants 27.5% were male and 72.5% were female. Women scored lower (M = 2.61, SD = 1.08) than men (M = 2.95, SD = 0.97) on the question ‘How often do you use

emoticons in your online communication?’. A lower score suggests a higher use of emoticons, because of the answer scale (1 = Always; 5 = Never). The persuasion knowledge is quite high throughout the sample, 48% of the participants stated that they use emoticons always or most of the time, while 23.5% indicated that they use them about half the time and 28.2% said they would use them rarely or never (M = 2.70, SD = 1.06).

Independent variables

The independent variable in this design were the emoticons in the organization’s crisis response. The levels of this variable are absent and present. Therefore, they saw a crisis statement by an organization, one with three sad-face-emoticons and the other without. The crisis and the organization was fictitious, to avoid that any previous knowledge of the crisis or relationship with the organization influenced the evaluation of the corporate reputation. Accordingly, this will strengthen the internal validity.

Moderator Variables

One moderator variable is the crisis type, which is the between-subjects factor in the experiment. Even though Coombs (2007) distinguishes between the crisis types, this paper looked only at the two extremes: victim cluster (weakest attribution of crisis responsibility) and preventable cluster (strongest attribution of crisis responsibility). The decision was made to assure the participants will recognize the difference between the crisis types. Therefore, this variable has two levels: victim and preventable cluster, and it was manipulated by giving different causes for the crises. Product recalls were chosen as scenarios, because it was easy to manipulate the cause of the incidents. (see Appendix B).

The second moderator variable was persuasion knowledge. To measure the

participant’s persuasion knowledge, a scale consistent of six items was used. The scale was developed by Bearden, Hardesty & Rose (2001) and is considered as highly reliable (Holmes, Beitelspacher, Hochstein & Bolander, 2017). The items were measured on a 7-point Likert

scale (items in Appendix D). An exploratory factor analysis indicated two components. The first one had an Eigenvalue of 3.08 explaining 51.33% of the variance and included all items. The second factor included three items that also loaded on the first component. It had an Eigenvalue of 1.05 and explained 17.49% of the variance. After a Varimax Rotation four items loaded on the first component and the items 1 and 2 on the second. Accordingly, these items were excluded from further analysis. The remaining 4-items proved to be reliable, Cronbach’s alpha = .79 (M = 21.40, SD = 3.47) and the reliability of the scale could not be further improved by deleting any items. Based on these results a new variable including the four remaining items was computed (PersuasionKnowledge).

Dependent variable

The dependent variable in this study was Corporate Reputation. The Organizational Reputation Scale by Coombs & Holladay (2002) was used to measure it. The original scale was developed by Coombs (1998) and was a 10-items scale. To reduce the length of this scale, only five items, were selected. Coombs & Holladay (2002) reviewed previous research that implemented the Organizational Reputation Scale and selected five items that resulted in a high reliability. The scale was measured on a 7-point Likert scale (items in Appendix E). In the present study, Corporate Reputation was measured twice. First, it was measured after seeing an organizational tweet with no emoticons. An exploratory factor analysis showed that the 5-items scale was unidimensional (Appendix G). The scale also proved to be reliable, Cronbach’s alpha = .81 (M = 24.09, SD = 5.50) and the reliability of the scale could not be further improved by deleting any items. Based on these results a new variable including all items was computed (CorpRep_NE). Second, the corporate reputation was assessed after the participants were exposed to the organizational tweet with emoticons. The scale was

unidimensional as well (Appendix H) and proved to be reliable, Cronbach’s alpha = .84 (M = 24.17, SD = 5.66). The reliability of the scale could not be further improved by deleting any

items and based on these results a new variable including all items was computed (CorpRep_E).

Manipulation check

To verify the manipulation of the crisis type, three items were measured, which were adopted from Lee (2005), who used them to measure “’participants’ perception of internal-controllable and external-internal-controllable causes”. The items were measured on a 7-point Likert scale (items in Appendix F). An exploratory factor analysis indicated that the scale was unidimensional (Appendix I). The 3-item scale also proved to be reliable, Cronbach’s alpha = .78 (M = 7.60, SD = 2.82) and the reliability of the scale could not be further improved by deleting any items. Based on these results a new variable including all three items was computed (CrisisResponsibility). Respondents that saw the crisis scenarios which fell within the victim cluster (N = 76) scored higher (M = 3.00, SD = 0.93) on the scale for crisis

responsibility than participants that saw the incidents within the preventable cluster (N = 73; M = 2.04, SD = 0.66). Therefore, participants perceived the crisis responsibility as higher in the preventable cluster than in the victim cluster. An independent t-test showed that there was a significant difference between the two means (t (136.07) = 7.36, p < .000). Therefore, the manipulation of the crisis type was successful.

To check, if the manipulation for the independent variable worked the question “Did one of the organizational tweets include emoticons?” was asked with the answer options “Yes”, “No” and “I don’t know”. Of the 149 participants 141 indicated that one of the

companies used emoticons, three stated that they did not use emoticons and five didn’t know. Therefore, the latter eight cases were excluded from further analysis.

Randomization check

Randomization checks were conducted for the control variables age and gender. A Chi-squared test with the variables condition (crisis type) and age showed that there was no

association between the two variables, Chi-squared (24, N = 141) = 23.34, p = .500. Another Chi-squared test was conducted with the variables condition (crisis type) and gender. The test showed also no association between the two variables, Chi-squared (1, N = 141) = 0.21, p = .645. As both Chi-squared tests were not significant, the distribution of age and gender was similar in both experimental conditions. Therefore, the randomization was successful.

Results

To assess if the crisis response of the organization that included emoticons lead to a more positive corporate reputation than the crisis response without emoticons, a paired t-test was conducted. Results showed that the corporate reputation was perceived as almost the same in both conditions. The paired t-test was not significant, t (148) = -0.18, p = .857 d = 0.95, 95% CI [-0.18, 0.15]. Therefore, there was no significant difference between the means of

corporate reputation after seeing an emoticon in the crisis response or not. In both conditions the participants were generally positive about the organizations. In the “no emoticons” condition 71.8% and in the “emoticons” condition 75.8% rated the corporate reputation 4 or higher (on a 7-point Likert scale). Hypothesis 1 cannot be supported.

To test if this relationship was moderated by crisis type or persuasion knowledge, a repeated measures ANOVA was conducted. To include persuasion knowledge into the test, the continuous variable was split into two groups, with help of the median split procedure. The results of the repeated measures ANOVA showed that there was no interaction effect of emoticons and crisis type, F (1, 145) = 0.47, p = .493. Thus, a crisis response with emoticons didn’t lead to a more negative or positive corporate regardless of the crisis type. Therefore, Hypotheses 2 and 2a can’t be supported.

There was also no main effect of persuasion knowledge, F (1, 145) = 0.18, p = .672 and no interaction effect of emoticons and persuasion knowledge, F (1, 145) = 2.65, p = .106. Hence, persuasion knowledge didn’t affect the relationship between emoticons present and

the corporate reputation. As the persuasion knowledge was rather high throughout the sample, the test was repeated with splitting the variable persuasion knowledge in three equal groups. However, there was still neither a main effect of persuasion knowledge, F (2, 143) = 1.15, p = .320, nor an interaction effect of emoticons and persuasion knowledge, F (2, 143) = 1.33, p = .268. To look further into this, the test was repeated one more time with splitting persuasion knowledge in four equal groups. There was still no main effect of persuasion knowledge, F (3, 141) = 1.20, p = .314 and no interaction effect of emoticons and persuasion knowledge, F (3, 141) = 1.09, p = .357. Thus, Hypothesis 3 can’t be supported.

Furthermore, there was no interaction effect of emoticons, persuasion knowledge and crisis type, F (1, 145) = 0.01, p = .939, which means that high or low persuasion knowledge did not affect the relationship between emoticons and corporate reputation depending on the crisis type. Again, the same test was repeated with persuasion knowledge split in 3 and 4 groups. However, these tests did not show significant results, F (2, 143) = 0.29, p = .748; F (3, 141) = 1.67, p = .176. Hence, Hypotheses 3a and 3b can’t be supported.

Corporate Reputation was used as dependent variable, as it is an important asset for organizations, especially during a crisis. However, as the analyses did not result in any significant results, it was considered if the variable was too broad as a dependent variable. To check if more specific parts of the variable, such as if the organization is perceived as warm, would have resulted in a significant result, only the two items of the corporate reputation scale that dealt with the organizations concern were computed as a new variable, as they were predicted to be associated with warmth. However, it didn’t reveal any new results (Appendix J).

As the gender groups were not equal in the sample and gender can have an impact on how someone uses and interprets emoticons, the repeated measures ANOVA was conducted excluding all male participants (n = 41) However, this did not result in any new findings (Appendix K).

The results did show a main effect of condition (crisis type), F (1, 145) = 19.06, p < .000, suggesting that based on the crisis type there was a difference in how the participants rated the corporate reputation. The corporate reputation was assessed higher in the victim cluster, after seeing the organizational tweet with no emoticons, M = 5.18, SD = 0.99, and after seeing the one with emoticons, M = 5.14, SD = 1.01, than in the preventable cluster after seeing no emoticons, M = 4.45, SD = 1.09 and after seeing the emoticons, M = 4.52, SD = 1.17.

Discussion

The aim of this study was to test whether emoticons can affect the corporate reputation in times of a crisis. However, the hypotheses couldn’t be supported. This could have multiple reasons. Emoticons did influence the perception of messages in previous research, e.g. when participants indicated higher arousal and enhanced smiling activity when a message included an emoticon (Thompson et al., 2016). However, these responses are more direct, while corporate reputation builds up overtime. The emoticons in the crisis response may had a spontaneous positive response, like a smile, but it did not translate into the evaluation of the corporate reputation. This adds to the discussion, if corporate reputation was the appropriate dependent variable in this study. It seemed to be appropriate, as it is the most important outcome for the organization in times of a crisis. The study of Li et al. (2019) found that employees that used emoticons online (web care), were perceived as warmer but less

competent. Therefore, warmth and competence could be interesting variables in this context and should be considered in further research designs.

Furthermore, in the study by Thompson et al., 2016 messages that were complemented with an emoticon were ironic, praising, critical or literal, but none of them tried to convey a sad message. Therefore, the impact of emoticons could differ depending on the original attitude of the message they are added to. Moreover, the study of Walther & D’Addario

(2001) found that happiness ratings were improved, when a smiley face was added to a positive message, while the positivity was reduced when a frown was added to a positive message. Walther & D’Addario (2001) only focused on a positive message, while the message in this study was sad by definition and the emoticons were ought to make the emotion clearer. These findings suggest, that the manipulation of text messages with

emoticons works rather for messages with a positive tone than a negative tone. Additionally, Mehrabian (1972 suggested that the expression of negative emotions was more subtle, as the openly expression negative feelings, such as sadness, could be perceived as inappropriate. As the emoticons emphasized the sadness of the message, the emotion could have been expressed too explicit. Again, this could imply that the effect on the perception of emoticons that convey positive emotions is stronger than for negative ones. However, future research would have to look further into this difference.

Nevertheless, other studies came to the conclusion that emoticons helped people to interpret the intention and attitude in the right way, e.g. Lo (2008), who showed that most people had extreme difficulty to perceive the right emotion and attitude from a computer mediated message and that added emoticons significantly changed the perception of it. As the stimuli in the present study were straighter forward, it could suggest that the manipulation with emoticons works better, if the attitude, intention and emotion of the message itself is not very clear. This could be further researched by adding more experimental groups to the present research design.

Ten Brinke & Adams (2015) showed, the appropriate display of sadness during an apology did improve the organization’s performance and if happiness was displayed by the spokesperson, the organizational performance decreased. This could have been included in this study, by adding another experimental group, that exposed participants to a tweet with inappropriate emoticons, e.g. a laughing smiley.

design. Even though the study of ten Brinke & Adam (2015) suggested so, other emotions like disappointment, anger or shame, may had revoked stronger feelings about the corporate reputation.

Persuasion Knowledge

The persuasion knowledge was rather high throughout the sample. The concept was measured on a 7-point Likert scale and about half of the participants had a persuasion knowledge higher than 5.25. This could explain why there was no influence of persuasion knowledge, as the groups of participants with low and high persuasion knowledge was not evenly distributed.

Gender

The sample comprised of more women (72.5%) than men (27.5%), which may has influenced the results. Even though this distribution was similar in both experimental groups, it could have affected the results, as research has shown that there is a difference in the use of emoticons depending on gender. The study of Wolf (2000) suggested that women use

emoticons in computer mediated messages more often than men. This goes along with the finding of Tossel, Kortum, Shepard, Barg-Walkow, Rahmati & Zhong (2012), who found out that their female participants were using emoticons almost twice as often as the male

participants in their study. This skewness between males and females was also found in this study. Women in this sample used emoticons more often than men. However, the group of male participants is too small to make any statistically significant claims about this. Hence, future research about the perception of emoticons in corporate communication should pay attention to a sample with equally distributed gender groups.

The high number of female participants could explain the high use of emoticons throughout the sample, almost half of the participants stated that they use emoticons always or half of the time. An analysis of the effect of the use of emoticons in their own online

communication on the perception of emoticons in corporate communication, may be interesting for further research. If so, it could explain why there were now effects found in this study, as the use of emoticons of the participants was high in general.

Conclusion

This study is a contribution to a field in the corporate communication area, which has a lot of potential for further research. Even though, this study didn’t find any effects of the use of emoticons in organizational crisis responses on the corporate reputation, it does not mean that emoticons could not be of any use regarding crisis communication. Firstly, it must be noted that the emoticons in the crisis response did not affect the corporate reputation

negatively. Secondly, further research that includes more attention to gender, personal use of emoticons, appropriateness of the displayed emotion and other theoretical approaches is needed in this area, to get a better understanding of the chances and challenges emoticons can pose in the context of corporate communication.

Especially, the question if the expression of positive feelings vs negative feelings affects the perception of emoticons needs further research. Because many emoticons are used in the daily communication, it’s possible, that they tend to sympathize with organization’s that use them in their communication. The emoticons could distract from the actual message, which could lead to a more surficial evaluation of the message. In terms of the elaboration likelihood model, this could mean that receivers of a message with emoticons could tend to use the peripheral route and tend to accept messages with emoticons easier.

In this context, another and a newer way to communication nonverbal cues should be considered as well: GIFs. They are animated images, and due to the moving picture, they could convey messages even better, as they don’t just show a facial expression, but also body gestures or hand and arm movements. On the other hand, GIF’s often portray common TV shows or celebrities. Hence, they could mean different things to different receivers, depending

on their relationship to the celebrity or TV Show. However, further research into the meaning of GIFs in the context of corporate communication could bear interesting opportunities about the communication of emotions.

References

Bearden, W. O., Hardesty, D. M., Rose, R. L. (2001). Consumer Self-Confidence:

Refinements in Conceptualization and Measurement. Journal of Consumer Research, 28, 121-134.

Cheshin, A. Amit, A & van Kleef, G. A. (2018). The interpersonal effects of emotion

intensity in customer service: Perceived appropriateness and authenticity of attendants’ emotional display shape customer trust and satisfaction. Organizational Behavior and Human Decision Processes, 144, 97-111.

Chang, C. Y. (2016). EFL Reviewers’ Emoticon Use in Asynchronous Computer-Mediated Peer Response. Computer and Composition, 40, 1-18.

Claeys, A.-S., Cauberghe, V. & Vyncke, P. (2010). Restoring reputation in times of crisis: An experimental study of the Situational Crisis Communication Theory and the

moderating effects of locus of control. Public Relation Review, 36, 265-262.

Coombs, W. T. (1998). An analytic framework for crisis situations. Better responses from a better understanding of the situation. Journal of Public Relations Research, 10, 177-191.

Coombs, W. T. (2004). Impact of past crises on current crisis communication. Insights from Situational Communication Crisis Theory. Journal of Business Communication, 41, 3, 265-289.

Coombs, W. T. (2007). Protecting Organization Reputations During a Crisis: The

Development and Application of Situational Crisis Communication Theory. Corporate Reputation Review, 10, 163-176.

Coombs, W.T. & Holladay, S. J. (2002). Helping crisis managers protect reputational assets: Initial tests of the situational crisis communication theory. Management

Communication Quarterly, 16, 165-186.

Coté, S., Hideg, I. & van Kleef, G. A. (2013). The consequences of faking anger in negotiations. Journal of Experimental Social Psychology, 49, 453-463.

Dresner, E. & Herring, S. C. (2010). Functions of the Nonverbal in CMC: Emoticons and Illocutionary Force. Communication Theory, 20, 249-268.

Ekman, P. & Friesen, W. V. (1969). The Repertoire of Nonverbal Behavior: Categories, Origins, Usage, and Coding. Semiotica, 1, 49-98.

Friestadt & Wright (1994). The Persuasion Knowledge Model: How People Cope with Persuasion Attempts. Journal of Consumer Research, 21 (1), 1-31.

Ham, C.-D. & Kim, J. (2017). The Role of CSR in Crises: Integration of Situational Crisis Communication Theory and the Persuasion Knowledge Model. Journal of Business Ethics, Oct 2017, 1-20.

Hof, R. D. (06.03.2016). Picture this: Marketers Let Emojis Do the Talking. Retrieved from https://www.nytimes.com/2016/03/07/business/media/picture-this-marketers-let-emojis-do-the-talking.html?_r=1

Holmes, Y. M., Beitelspacher, L. S., Hochstein, B. & Bolander, W. (2017). “Let’s make a deal:” Price outcomes and the interaction of customer persuasion knowledge and salesperson negotiation strategies. Journal of Business Research, 78, 81-92. Jin, Y. (2010). Making Sense Sensibly in Crisis Communication. How Publics’ Crisis

Appraisals Influence Their Negative Emotions, Coping Strategy Preferences, and Crisis Response Acceptance. Communication Research, 37, 4 522-552.

Kim, J. & Yan, J. (2016). Understanding emotionally involved publics. The effects of crisis type and felt involvement on publics’ emotional responses to different consumer product crises. Corporate Communications: An International Journal, 21, 4, 465-482. Krohn, F. (2004). A generational approach to using emoticons as non-verbal communication.

Journal of Technical Writing in Communication, 34, 4, 321-328.

Lee, B. K. (2005). Hong Kong Consumers’ Evaluation in an Airline Crash: A Path Model Analysis. Journal of Public Relations Research, 17, 4, 363-391.

Li, X., Chan, K. W., Kim, S. (2019). Service with Emoticons: How Customers Interpret Employee use of Emoticons in Online Service Encounters. Journal of Consumer Research, 45, 973-987.

Lo, S.-K. (2008). The Nonverbal Communication Functions of Emoticons in Computer-Mediated Communication. Cyber Psychology & Behavior, 11, 5, 595-597. Matsa, K. E., Shearer, E. (10.09.2018). News Use Across Social Media Platforms 2018.

Retrieved from https://www.journalism.org/2018/09/10/news-use-across-social-media-platforms-2018/.

Mehrabian, A. (1972). Nonverbal Communication. New York: Routledge

Mehrabian, A. & Williams, A. (1969). Nonverbal Concomitants of Perceived and Intended Persuasiveness. Journal of Personality and Social Psychology, 13, 1, 37-58.

Tam, N. D. (2015). Quantitative assessment of sad emotion. Psychology and Behavioral Sciences, 4, 2, 36-43.

Ten Brinke, L. & Adams, G. S. (2015). Saving face? When emotion displays during public apologies mitigate damage to organizational performance. Organizational Behavior and Human Decision Process, 130, 1-12.

Thompson, D., Mackenzie, I. G., Leuthold, H. & Filik, R. (2016). Emotional responses to irony and emoticons in written language: Evidence from EDA and facial EMG. Psychophysiology, 53, 1054-1062.

Tossel, C. C., Kortum, P., Shepard, C., Barg-Walkow, L. H., Rahmati, A. & Zhong, L. (2012). A longitudinal study of emoticon use in text messaging from smartphones. Computers in Human Behavior, 28, 659-663.

Utz, S., Schultz, F. & Glocka, S. (2012): Crisis communication online: How medium, crisis type, and emotions affected public reactions in the Fukushima Daiichi nuclear disaster. Public Relations Review, 39, 40-46.

Van der Meer, T. G.L.A. & Verhoeven, J. W.M. (2014). Emotional crisis communication. Public Relations Review, 4, 526-536.

Van Kleef, G. A. (2014). Understanding the positive and negative effects of emotional expressions in organizations: EASI does it. Human relations, 67, 9, 1145-1164. Van Kleef, G. A., Homan, A. C. & Cheshin, A. (2012). Emotional influence at work: Take it

Veil, S. R., Buehner, T., Palenchar, M. J. (2011). A Work-In-Process Literature Review: Incorporating Social Media in Risk and Crisis Communication. Journal of Contingencies and Crisis Management, 19, 2, 110-122.

Wolf, Alecia (2000). Emotional Expression Online: Gender Differences in Emoticon Use. CyberPsychology & Behavior, 3, 5, 827-833.

Appendix Appendix A.: Crisis Scenarios Pretest

Appendix A.1: VRIBA Baby Food

The company VRIBA is well-known for their high-quality baby food. They have announced that they are recalling three of their products worldwide, because a bacterium that causes botulism was found in the flavours Apple & Pear, Banana & Peach and Banana, Pear & Mango. Botulism is a dangerous illness, particularly for infants and young children. It can suddenly paralyze muscles, leading to weakness, trouble seeing, speaking and breathing. If untreated the illness can cause paralysis of the arms, legs and muscles in general, which can end in death due to the inability to breathe. If any of these symptoms occur, immediate medical attention is required.

After investigations, VRIBA found that a former employee tampered with the production of VRIBA’s baby food, which lead to the bacterium in these 3 products. The employee was fired from VRIBA 3 months ago and started tampering with the production line because he was upset about the termination.

Appendix A.2: DÄCK car tire

The car tire manufacturer DÄCK is recalling two of their tire models worldwide. The models AT tires and AT+ Tires. The tread on the AT models tends to peel of, resulting in very high failure rates. The National Highway Traffic Safety Administration linked hundreds of accidents to the problems with the tread on the tires. The tires mostly fail at high-speed, risking roll-overs which could lead to fatal accidents.

After investigations it was found that a former employee tampered with the production of DÄCK’s tires, which lead to the problems with the tread. The employee was fired from DÄCK 3 months ago and started tampering with the production line because he was upset about the termination.

Appendix A.3: RENTVA laundry detergent

The company RENTVA is well-known for their high-quality laundry detergent. They are recalling two of their products worldwide: 'Pure White' and 'Sensitive'. RENTVA announced that the concentration of toxic chemicals in the preservatives of the products is too high. This can lead to allergic reactions, e.g. red skin, blisters that ooze, burning, itching and swelling of the eyes and face. The allergic reaction can occur after the first exposure to the landry

detergent or slowly after repeated exposure. If any of these symptoms occur a doctor should be seen for professional treatment.

After investigations it was found that non-compliant behaviour of an employee regarding workplace procedures in one of RENTVA’s factories lead to the chemical within the product. Appendix A.4: PIVA Beer Bottles

The company PIVA is recalling two of their products worldwide. The recall applies to PIVA IPA 6-packs and 12-packs. They have discovered that a glass packaging flaw could cause small pieces of glass to break off and fall into the bottle. If this is first noticed when already

swallowed some of the glass, it can have fatal consequences and immediate medical attention is required. PIVA is recommending all customers to not open any of the bottles that could be affected.

After investigations it was found that non-compliant behaviour of an employee regarding workplace procedure of PIVA’s beer bottle factory lead to the glass packaging flaw.

Appendix B.: Crisis Scenarios Main Study Appendix B.1: VRIBA Baby Food

The company VRIBA is well-known for their high-quality baby food. They have announced that they are recalling two of their products worldwide, because a bacterium that causes botulism was found in the flavours Apple & Pear and Banana & Peach. Botulism is a dangerous illness, particularly for infants and young children. It can suddenly paralyze muscles, leading to weakness, trouble seeing, speaking and breathing. If untreated the illness can cause paralysis of the arms, legs and muscles in general, which can end in death due to the inability to breathe. If any of these symptoms occur, immediate medical attention is required.

Reason for victim cluster

After investigations, it was found that a competitor tampered with the production of VRIBA’s baby food, which lead to the bacterium in these two products. This happened after the

products left VRIBA’s factory and the products were no longer under control of the company. Reason for preventable cluster

After investigation, VRIBA found that a missed maintenance of the machines in the factory lead to the bacterium in these 3 products. The company released information that they had changed the monthly maintenance of their machines to quarterly in order to save money. During this longer period withoutservicing the machines the bacterium spread.

Appendix B.2: PIVA Beer Bottles

The company PIVA is recalling two of their products worldwide. The recall applies to PIVA IPA 6-packs and 12-packs. It was discovered that a glass packaging flaw could cause small pieces of glass to break off and fall into the bottle. If this is first noticed when already

swallowed some of the glass, it can have fatal consequences and immediate medical attention is required. PIVA is recommending all customers to keep the bottles closed and return them to the store for a refund.

Reason for victim cluster

After investigations, it was found that a competitor tampered with the production of PIVA’s bottle production, which laed to the packaging flaw. This happened after the products left VRIBA’s factory and the products were no longer under control of the company.

Reason for preventable cluster

After investigation, it was found that a missed maintenance of the machines in the factory lead to the packaging flaw of the beer bottles. The company released information that they had changed the monthly maintenance of their machines to quarterly in order to save money. During this longer period without servicing one of the machines broke.

Appendix C.: Organizational Tweets Appendix C.1: VRIBA

Appendix C.2: PIVA

Appendix D.: Items Persuasion Knowledge Scale 1) “I know when an offer is “too good to be true” 2) “I can tell when an offer has strings attached”

3) “I have no trouble understanding the bargaining tactics used by salespeople” 4) “I know when a marketer is pressuring me to buy”

5) “ I can see through scales gimmicks used to get consumers to buy” 6) “I can separate fact from fantasy in advertising”.

Appendix E.: Items Organizational Reputation Scale

1) “The organization is concerned with the well-being of its publics” 2) “The organization is basically dishonest”

3) “I do not trust the organization to tell the truth about the incident”

4) “Under most circumstances, I would be likely to believe what the organization says” 5) “The organization is not concerned with the well-being of its publics”.

Appendix F.: Items Manipulation Check Crisis Type

1) “To what degree to you think the incident was caused by the organization?”

2) “To what degree do you think the organizations could have prevented the incident?” 3) “To what degree do you think the organization had control over the incident?” Appendix G.: Factor analysis corporate reputation no emoticons

Factor Matrix

Factor 1 The organization is concerned with the

well-being of its public.

.821 The organization is basically dishonest. .696 I trust the organization to tell the truth about

the incident.

.673 Under most circumstances, I would be likely to believe what the organization says.

.617 The organization is concerned with the

well-being of its public.

.615

Note: Factor analysis of the corporate reputation scale after seeing the crisis response with no emoticons. For this purpose, the items 2, 3 and 5 were reversed, as they were formulated negatively.

Appendix H.: Factor analysis corporate reputation emoticons

Factor Matrix

Factor 1 The organization is

concerned with the well-being of its public.

.824

The organization is .814 Total Variance Explained

Factor Total % of Variance Cumulative %

1 2.876 57.518 57.518

2 .860 17.203 74.721

3 .543 10.850 85.571

4 .465 9.308 94.879

5 .256 5.121 100.000

Total Variance Explained

Factor Total % of Variance Cumulative %

1 3.056 61.125 61.125

2 .714 14.280 75.404

3 .576 11.526 86.931

4 .402 8.035 94.966

basically dishonest. I trust the organization to tell the truth about the incident.

.707

Under most

circumstances, I would be likely to believe what the organization says.

.617

The organization is concerned with the well-being of its public.

.615

Note: Factor Analysis of the Corporate Reputation Scale after seeing the crisis response with emoticons. For this analysis the items two, three and five were reversed, as they were

formulated negatively.

Appendix I.: Factor analysis crisis responsibility

Factor Matrix

Factor 1 The organization could have prevented the

incidents.

.833 The incidents were caused by the

organization.

.702 The organizations had control over the

incidents.

.694

Note: Factor Analysis of the Crisis Responsibility Scale. Appendix J.: Repeated Measures Corporate Warmth

Tests of Within-Subjects Contrasts

Source df F Sig. Factor 1 1 .019 .891 Factor1*ConditionCrisisType 1 .085 .771 Factor1*PK_4Groups 3 .878 .454 Factor1*ConditionCrisisType* PK_4Groups 3 .163 .921 Error(factor1) 141

Total Variance Explained

Factor Total % of Variance Cumulative %

1 2.101 70.025 70.025

2 .514 17.127 87.153

Tests of Between-Subjects Effects Source df F Sig. Intercept 1 2793.770 .000 ConditionCrisisType 1 15.999 .000 PK_4Groups 3 1.289 .281 ConditionCrisisType* PK_4Groups 3 .669 .573 Error 141

Note: Results of repeated measure test of the whole model, with Corporate Warmth as dependent variable, instead of corporate communication.

Appendix K.: Repeated Measures only women Tests of Within-Subjects Contrasts

Source df F Sig. Factor 1 1 .048 .872 Factor1*ConditionCrisisType 1 .206 .651 Factor1*PK_4Groups 3 .446 .721 Factor1*ConditionCrisisType* PK_4Groups 3 .891 .449 Error(factor1) 100

Tests of Between-Subjects Effects

Source df F Sig. Intercept 1 3076.393 .000 ConditionCrisisType 1 22.894 .000 PK_4Groups 3 2.611 .056 ConditionCrisisType* PK_4Groups 3 1.690 .174 Error 100

Note: Results of the repeated measure test of the whole model, excluding the male participants from the sample.