Influencer Endorsement: what if it goes wrong? – a Study on the Effectiveness of Crisis Response Strategy

Influencer Endorsement: What if it goes wrong? – A Study on the

Effectiveness of Crisis Response Strategy

Student: Mileau Holman 11953578 Supervisor: Dr. James Slevin

University of Amsterdam

Master Track Corporate Communication Date: January 31th, 2020

Table of Content

Abstract ... 3

Introduction ... 4

Theoretical framework ... 8

Influencer Endorsement Crisis ... 8

The Effect of Crisis Responsibility on Organizational Reputation and Purchase Intention .. 8

The Moderating Effect of Crisis Response Strategy on Organizational Reputation and Purchase Intention ... 11

The Mediating Role of Brand Image ... 13

The Mediating Role of Influencer Identification ... 15

Methodology ... 17

Experimental Design and Sample ... 17

Procedure and Stimulus Material ... 18

Measures ... 19

Manipulation check ... 21

Results ... 22

Manipulation and Randomization Check ... 22

Crisis Responsibility Effects ... 22

The Mediating Effect of Brand Image and Influencer Identification of the Moderation of

Crisis Response Strategy ... 23

The Moderation Effect of Brand Image and Influencer Identification ... 24

The Effect of Crisis Response Strategy on Brand Image and Influencer Identification ... 26

Discussion ... 26

Crisis Responsibility ... 26

Moderation effect of Crisis Response Strategy ... 28

Mediation effect of Brand Image ... 29

Mediation effect of Influencer Identification ... 30

Conclusion ... 31 References ... 34 Appendix A ... 43 Appendix B ... 47 Appendix C ... 49 Appendix D ... 51

Abstract

Over the past decade, organizations are increasingly using influencers in their marketing strategies. However, influencer endorsements carry high risks and possible negative consequences for the organization as their behaviour and occurrence cannot be controlled. Especially, when the endorsers get themselves involved in some kind of scandal, which can be associated with the organization or not. These scandals are often irreversible and can damage organizational outcomes and can therefore be seen as an influencer endorsement crisis. This study focusses on finding the best possible crisis response strategy in case of an influencer endorsement crisis in order to reduce the negative effect on the post-crisis evaluation, being organizational reputation and purchase intention. By using the crisis responsibility of the organization in combination with two crisis response strategies, managerial implications for the organization could be made. This produced a 2 (Crisis

Responsibility: external locus of control vs. internal locus of control) x 2 (Crisis Response Strategy: ending the endorsement vs. continuing the endorsement) between-subject design. As

it is important for both practitioners and researchers to clearly identify factors that can

influence the effectiveness of crisis responses, the emotions and perceptions of the consumers among the organization and the influencer were taken into account. Brand image and

influencer identification were added to see if there was a mediating effect between crisis response strategy and the relation between crisis responsibility on the post-crisis evaluation. Results of this study (N = 148) showed that there is no significant effect of crisis

responsibility on organizational reputation or purchase intention. Neither does the crisis response strategy moderate this relation and no significant mediation effect of brand image and influencer identification was found. However, a direct effect of brand image on

organizational reputation and purchase intention was found, and respondents successfully identified the level of crisis responsibility and the used crisis response strategy. This suggests

that, consumers make perceptions after a crisis and are aware of the response of the organization, however the best suitable crisis response strategy still needs to be found. Therefore, organizations should spread risks by not only focussing on influencer marketing but keep investing in the traditional media. Also, organizations could choose for endorsing with micro-influencers as it is less likely for them to get involved in a scandal. This study provides a valuable starting point to design an effective crisis response strategy by taking the crisis responsibility of the organization and the emotions of the consumer towards the

organization and the influencer into account.

Keywords: Influencer endorsement, crisis communication, crisis responsibility, crisis

response strategy, brand image, influencer identification

Introduction

Nowadays, influencer marketing plays an important role in the daily marketing efforts of small, medium and large organizations. Influencers upload brand-related content on social media, both on their own initiative or partnering with brands in exchange for monetary benefits (Lu, Chang, & Chang, 2014). When partnering with a brand, the organization is endorsed with an influencer as part of their marketing strategy in order to promote the organization’s products or services (Tripp, Jenson & Carlson, 1994). One of the goals is creating word of mouth, as influencer marketing has the ability to influence consumer’s opinions, and therefore affect their purchase behaviour (Djafarova & Rushworth, 2017). Besides word of mouth, organizations want to create brand awareness and favourability, as influencers can transfer meaning and significance to the brand’s products (Erdogan, 1999; Lear, Runyan, & Whitaker, 2008). Also, as brands want to establish perceived credibility of

the endorsement, they choose their influencers precisely based on their attractiveness, trustworthiness, and expertise (Erdogan, 1999).

Besides the benefits influencer endorsement brings to the brand, it has its pitfalls, as the influencer’s future behaviour and occurrence of events cannot be controlled by the brand (Upadhyay & Singh, 2010). If particular influencer behaviour leads to negative publicity, it can damage the influencer’s reputation (Koo, Ruihley & Dittmore, 2012). Due to the close collaboration shared by the influencer and the organization that partners with them (Walker, Langmeyer & Langmeyer, 1992), this negative publicity about the endorsed influencer can negatively influence their attractiveness, and consequently the appeal and the credibility of the brand and therefore the organization (Till & Shimp, 1998).

Although, organizations tend to ignore the risks and negative consequences influencers can cause when involved in some kind of scandal. These scandals are often irreversible and can damage organizational outcomes (Jiang, Huang, Wu, Choy & Lin, 2015). Therefore this can be seen as an organizational scandal, as a crisis is described as ‘the

perception of an unpredictable event that threatens important expectancies of stakeholders related to health, safety, environmental, and economic issues and can seriously impact an organization’s performance and generate negative outcomes’ (Coombs & Holladay, 2015, p.146).

In this case, the organizational crisis is caused by an influencer’s shortcoming or involvement in an event (Upadhyay & Singh, 2010), which leads to organizational risks such as negative publicity, embarrassment to an organization or its products, which leads to damage to the organizational reputation (Jiang et al., 2015) and decrease of the purchase intention (Um & Kim, 2016).

Previous research showed that besides the crisis itself, the crisis responsibility plays an important part during a crisis (Coombs, 2007; Jiang et al., 2015; Weiner, 2006). Crisis

responsibility is occupied with the question who can be held responsible for the crisis (Coombs, 1995). As after a crisis, questions will pop up in the heads of the consumer, for example, is the scandal of the influencer related to the brand of the organization? The

Situational Crisis Communication Theory (SCCT) states that crisis responsibility is one of the factors in the crisis situation that shape the reputational threat (Coombs, 2007). However, when the crisis is effectively managed, the negative effects on the organizational reputation can be reduced (Coombs, 2007). In order to do so, the organization needs to react in a proper way, choosing the right response strategy (Bartz, Molchanov, & Stork, 2013). The SCCT provides the organizations with a guideline about which response strategy is most appropriate to use for each crisis type (Nekmat & Kong, 2019). Coombs (2007) introduced well-known types of crisis response strategies, like the rebuild and the response strategies. As it is an influencer crisis, these strategies need to be adjusted. The organization can react by

continuing the endorsement or ending the endorsement with the influencer (Jiang et al., 2015). When ending the endorsement with the influencer, they trie to improve the organization’s reputation by actively changing the situation. On the other hand, when choosing to continue the endorsement, the organization asserts that there is no crisis and continues the endorsement with the influencer.

However, what is still missing when implementing these crisis response strategies, are the perceptions and emotions of the consumers towards the products of the brand and the influencers. In this social media age, people can be emotionally attracted to influencers via the internet (Um, 2013). Also, people can feel a certain connection with a brand (Abbratt & Kleyn, 2012). These can be factors which can influence the effectiveness of the crisis

response (Klein & Dawar, 2004). Therefore, it is important to tailor crisis response strategies in order to make them successful. In order to test the emotions towards the brand, brand image will be taken into account. As brand image can play a predictive role in the

persuasiveness of a company’s crisis response strategy (Lee & Atkinson, 2019). In order to test if the emotions towards the influencer will affect the effectiveness of the crisis response strategy, influencer identification will be included. Burke (1950) stated that communication effectiveness depends on an audience’s identification with a fictional character, in this case the influencer.

For the reason explained above, the following research questions should be answered:

RQ1: What is the effect of an influencer-endorsement crisis based on crisis responsibility on

the post-crisis evaluation, being organizational reputation and purchase intention?

RQ2: How does the effect of crisis responsibility on the post-crisis evaluation differ when

ending or continuing the endorsement?

RQ2a: To what extent does brand image explain the effect of crisis response strategy on the

effect of crisis responsibility on the post-crisis evaluation?

RQ2b: To what extent does influencer identification explain the effect of crisis response

strategy on the effect of crisis responsibility on the post-crisis evaluation?

Bringing the answers to the research questions under the attention, will provide new insights about how to best respond to endorsed influencer negative publicity. Firstly, this could help marketeers and communication professionals to choose the best suitable crisis response strategy during an influencer endorsement crisis. Secondly, it will allow them to tailor the crisis response strategy to the level of crisis responsibility. Finally, it will help understand the effects of the emotions of the consumer on the effectiveness of the crisis

response strategies. All this in order to reduce the damage on the post crisis evaluation of the organization.

Theoretical framework Influencer Endorsement Crisis

Influencers refer to the phenomenon of micro-celebrities, who have built an extensive social network of followers by communicating narration of their everyday lives online (Abidin, 2016). When an influencer is partnering with a brand in exchange for monetary benefits, it can be seen as an influencer endorsement (Lu et al., 2014). Increasingly, brands prefer using influencers as marketing strategy over traditional media, because of their trustworthiness and attractiveness (Domingues & Reijmersdaal, 2018; Evans, 2017).

Including influencers in the marketing strategies has benefits, such as word of mouth, brand awareness and favourability (Lear et al., 2008) and will therefore positively affect the purchase behaviour (Djafarova & Rushworth, 2017).

Despite the benefits influencer endorsement brings to the brand, is also has it pitfalls. Brands cannot control the behaviour of the influencer (Upadhyay & Singh, 2010), which means that particular influencer behaviour can lead to negative publicity and damages the influencer’s own reputation (Koo et al., 2012). Although as these scandals are irreversible it can damage the organizational outcomes of the endorsed brand such as the organizational reputation and purchase intention (Jiang et al., 2015). As these scandals can permanently damage organizational outcomes, it can be seen as an influencer endorsement crisis.

The Effect of Crisis Responsibility on Organizational Reputation and Purchase Intention

An influencer endorsement contains of mainly two parties who are connected, namely the brand and the influencer. During an influencer endorsement crisis those two parties are

connected as well, therefore it is important to be aware of who is responsible for the crisis. Crisis responsibility can be defined as the stakeholder attributions of personal control for the crisis by the organization, in other words, how much stakeholders believe organizational actions caused the crisis (Coombs, 1995).

As an influencer endorsement crisis can harm the organizational reputation (Jiang et al., 2015), this paper looked at the well-known Situational Crisis Communication Theory which provides a framework to better understand how to anticipate on stakeholders in order to maximize the organizational reputation (Coombs, 2007). It states that the crisis responsibility is one of the factors in the crisis situation that shape the reputational threat (Coombs, 2007).

Although crisis responsibility consists of two components, the first component determines if the accused is held responsible for the action, and second, if that act is considered offensive (Benoit & Dorries, 1996). In both components, perceptions are more important than reality.

Perceptions of the crisis can be seen as attributions, as the Attribution Theory states that people search for the causes of events (make attributions), especially those that are negative and unexpected (Weiner, 1985, 2006). In this Social Media Age, where nothing happens unnoticed, it is important to take the attributions made by the public into account. The public will react more favourably to companies where the crisis is out of their control (Lu et al., 2014). Therefore the Attribution Theory will be used to determine the level of crisis responsibility of the organization. Locus of control will be applied to determine how the consumer makes attribution judgments about the causes of crisis. In the context of influencer endorsement crisis, locus of control refers to whether a crisis is caused by the internal parties of an organization or by external agents (Jiang et al., 2015). In this context, internal locus stands for a situation where the undesirable behaviours of the endorsed influencer is perceived to be associated with the organization, in contrary to external locus of control, where the

undesirable behaviours of the endorsed influencer is not perceived to be associated with the organization.

Previous research stated that internal locus of control results in the attribution of more responsibility to organizations, which in turn generates more unfavourable feelings of

stakeholders (Jiang et al., 2015). Negative evaluations, like unfavourable feelings, can harm the organizational reputation (Jeong, 2009).

Besides the Attribution Theory, recent research sheds light on the Image Repair Theory, as an additional explanation for how stakeholders respond to crisis situations (Page, 2019). It argues that both responsibility and offensiveness are necessary for a negative reputation threat as a result of a crisis. The study developed new measures which combined attributed responsibility and offensiveness and concluded that attribution and offensiveness predicts reputation more than attribution alone (Page, 2019).

Although as a result of this different method of manipulating responsibility, it did not manipulated attribution appropriately. Therefore, in this study the focus will be on the attributed responsibility only (Page, 2019).

In sum, the SCCT suggests that the reputation threat, derived from a crisis situation, comes from the crisis responsibility. However, within this Social Media Age, the made attributions by the public should be considered wisely as internal locus of control results in more responsibility to the organization, which in turn can harm the organizational reputation (Jeong, 2009). This leads to formulating the following hypothesis:

H1a: Influencer endorsement crisis perceived with external locus of control will lead

to more positive organizational reputation than influencer endorsement perceived with internal locus of control.

Also, when locus of control is external, stakeholders tend to attribute responsibility to external factors. Contrary, when locus of control is internal, responsibility is attributed to the organization. In this case, consumer behaviour such as blame and anger is directed towards the organization. Moreover, the consumer behaviour blame, has a strong negative effect on brand evaluations, and therefore on purchase intention (Klein & Dawar, 2004). Therefore the following hypotheses is formulated:

H1b: Influencer endorsement crisis perceived with external locus of control will lead

to more positive purchase intention than influencer endorsement crisis perceived with internal locus of control.

The Moderating Effect of Crisis Response Strategy on Organizational Reputation and Purchase Intention

During a crisis, caused by an influencer endorsement, the public’s attitude concerning perceived credibility of the organization and its products can be damaged if the crisis is not managed effectively (Bartz, Molchanov, & Stork, 2013; Biswas, Hussain, & Donnell, 2009; Koo et al., 2012; Money, Shimp, & Sakano, 2006; Till & Shimp, 1998; Um, 2013; Upadhyay & Singh, 2010). The stakeholders will evaluate the appropriateness and the quality of the actions of the organization during and after the crisis (Stieglitz, Mirbabaie, Kroll & Marx, 2019). In case that the actions do not comply with the expectations of the stakeholders, it can result in illegitimacy and reputation damage (Stieglitz et al., 2019). As both are types of organizational evaluation, both concepts can be used interchangeably (Deephouse &

Suchman, 2008). Legitimacy can be defined as ‘a generalized perception or assumption that the actions of an entity are desirable, proper, or appropriate within some socially constructed system of norms, values, beliefs, and definitions’ (Suchman, 1995, p.574). In order to stay

legitimate, it is important what the organization says after the crisis, otherwise known as the ‘Crisis Response Strategy’, as it can affect important crisis communication outcomes, including the company’s reputation (Coombs & Holladay, 2009).

The way an organization’s reputation is evaluated is thus influenced by the crisis response strategy. Therefore, the right response strategy should be identified, in order to reduce the negative effects on the organizational reputation (Coombs, 2007). Coombs (1995) introduced the rebuild response strategy which tries to ‘improve the organization’s reputation by offering material and/or symbolic forms of aid to victims’ (Coombs, 2007, p.172), and the denial response strategy, where the organization asserts that there is no crisis (Coombs, 2007). In case of an influencer crisis, one of the strategies the organization can choose, is ending the endorsement with the influencer, which can be slightly compared with the rebuild strategy. The organization can also choose to continue the endorsement, where the organization asserts there is no crisis (Jiang et al., 2015).

When the influencer crisis is not perceived to be associated with the brand, and perceived to be outside the control of the organization, the organization can be seen as the victim of the crisis (Coombs, 2007). This means that the organization holds a low degree of responsibility for the crisis, therefore a defensive denial strategy should be adopted in order to reduce damage to organizational outcomes (Chao, 2019). When continuing the influencer endorsement, any connection between the crisis and the organization will be denied (Chao, 2019). When the influencer crisis is perceived to be associated with the brand, the crisis can be perceived to be inside the control of the organization (Coombs, 2007). When the

consumers think that the crisis is associates with the organization, the level of responsibility for the crisis is high. This means that the organization should implement an accommodating strategy in order to reduce negative organizational outcomes (Chao, 2019). When ending the influencer endorsement, the organization actively anticipates to the situation.

This leads to the following research questions:

H2a: The positive effect of influencer endorsement crisis perceived with external locus

of control on organizational reputation compared to influencer endorsement perceived with internal locus of control will be more prominent when the organization continues the influencer endorsement than when it ends the influencer endorsement.

H2b: The positive effect of influencer endorsement crisis perceived with external locus

of control on purchase intention compared to influencer endorsement crisis perceived with internal locus of control will be more prominent when the organization continues the influencer endorsement than when it ends the influencer endorsement.

The Mediating Role of Brand Image

The difference between the type of crisis response strategy according to the level of crisis responsibility is clear from the distinction outlined above. This leads us to think that when the organization is perceived to be associated with the influencer scandal, the

organization should end the endorsement, but when the organization is not associated with the scandal, the organization should continue with the collaboration. Although the emotions and feeling of the consumers towards the brand should also be taken into account. Jeong, Paek and Lee (2013) have argued that brand image is critical to the understanding of consumer responses.

The meaning and definitions of brand image developed over time. James states that brand image consists of the attributes and associations that consumers connect to a brand (James, 2005). Moreover, it is the immediate reflection stakeholders have towards a brand (Bick, Jacobson & Abratt, 2003). According to Keller, brand image is defined as:

‘Perceptions about a brand as reflected by the brand associations held in consumer memory’ (Keller, 1993, p.3). These perceptions generate a set of associations that contribute to a total impression of an organization or brand (Andreassen, 2001). The perceptions of a brand are of great importance as they are considered to be the reflection of the total corporate

communication of the brand (Balmer, 2017).

Brand image plays a predictive role in the persuasiveness of a company’s crisis response strategy (Lee & Atkinson, 2019). The consumer is emotional attachment to the brand and will perceive the crisis more seriously (Lee & Atkinson, 2019). This means that a responding action from the brand is expected. Also, the study by Raithel, Wilczynski, Schloderer and Schwaiger (2010) states that the overall negative impact of a crisis on the organization cannot be avoided, but the magnitude of influence is depending on the individual firm dynamics. In particular, these value dynamics are significantly associated with a

reputation’s affective component as perceived by the general public (Raithel et al., 2010). Therefore, customers who are attached to the brand, will be more in favour of ending the endorsement. This said, the following hypotheses are formulated:

H3a: The positive effect of influencer endorsement crisis perceived with external locus

of control on organizational reputation compared to influencer endorsement perceived with internal locus of control will be more prominent when the organization continues the influencer endorsement than when it ends the influencer endorsement, this effect can be explained by participants who have a less positive brand image than by those who have a more positive brand image.

H3b: The positive effect of influencer endorsement crisis perceived with external locus

internal locus of control will be more prominent when the organization continues the influencer endorsement than when it ends the influencer endorsement, this effect can be explained by participants who have a less positive brand image than by those who have a more positive brand image.

The Mediating Role of Influencer Identification

Besides brand image, there is another factor that might influence the effectiveness of the crisis response strategy, namely the identification with the endorsed influencer. The article by Burke (1950) stated that communication effectiveness depends on an audience’s

identification with a fictional character, in this case the influencer. His dramatism theory explains that identification is largely formed through the connections established between a character and audience members (Burke, 1950). Kelman (1961) proposed that compliance, identification and internalization are the three processes of social influence. In this article, identification is described as a process which occurs when an individual adopts an attitude or behaviour from another person because of its association with a satisfying self‐defining relationship with that other person (Kelman, 1961). In other words, when an individual enjoys being or acting like someone else, it can be seen as identification. The article by Bandura (1986) proposed that identification occurred when individuals viewed a model as being similar to themselves. The common point these three articles stated, is that identification is an important factor which explains attitude and behaviour change (Um, 2013).

The identification process with an endorsed influencer happens through social media. This is in line with the identification process, called ‘parasocial relationship’. Parasocial relationship occurs when an individual develops a sense of intimacy and identification with the celebrity or media personality via the media (Hartmann & Goldhoorn, 2011). This means that consumers with a high level of identification with the influencer will be more likely to

adopt their thoughts, feelings and behaviours than those who identify little with the influencer. The article by Johnson (2005) proposed that consumers with a high level of identification are less likely to react negatively on scandals, than those who are lightly

identified with the influencer. Also, consumers who are strongly identified with the influencer are more convinced of the innocence of the influencer and will continue buying the endorsed products (Johnson, 2005).

The more consumers identify with the influencer, the less likely negative press will influence them negatively (Um, 2013), and might therefore be more in favour of continuing the endorsement. In contrast with consumers less identified with the influencer, who will feel ashamed and guilty of being connected with the influencer (Um, 2013), and might therefore be advocate ending the endorsement. Therefore, the following hypotheses are drawn:

H4a: The positive effect of influencer endorsement crisis perceived with external locus

of control on organizational reputation compared to influencer endorsement perceived with internal locus of control will be more prominent when the organization continues the influencer endorsement than when it ends the influencer endorsement but this effect can be explained by participants who have a higher level of influencer identification than by those who have a lower level of influencer identification.

H4b: The positive effect of influencer endorsement crisis perceived with external locus

of control on purchase intention compared to influencer endorsement perceived with internal locus of control will be more prominent when the organization continues the influencer endorsement than when it ends the influencer endorsement but this effect can be explained by participants who have a higher level of influencer identification than by those who have a lower level of influencer identification.

Figure 1. Conceptual framework

Methodology Experimental Design and Sample

This experimental study was conducted using the Qualtrics platform and used a 2x2 between subject full-factorial design, with Crisis Responsibility as between-subjects variable (includes 2 levels, namely: internal locus of control and external locus of control) and Crisis

Response Strategy as between-subjects variable (includes 2 levels, namely: continuing the

endorsement and ending the endorsement).

The appropriate sample mainly consisted of Dutch females who follow influencers on Instagram and are familiar with the social media platform Twitter. Although the age of Twitter users varies in age, the participants should have their own purchasing ability and therefore this study required an age minimum of eighteen years old. Also, the respondents had to have obtained at least a high school degree.

An earlier study about the effects of the brand response strategy (revoke vs. continuing the endorsement) after a celebrity immersed in a scandal on brand attitude and purchase intention used a sample of 252 respondents (Carrillat, Astous & Lazure, 2013). As this study focuses on a similar research aim and taking a minimum count of 30 participants per

condition and the four conditions of this study, a minimum of 120 participants were required. In this study, a participant pool of N =169 completed the questionnaire, of which 21 were deleted because of missing values. In total, 148 respondents remained for further analysis, and 116 of them were female (78.4%) and 32 male (21.6%). Most of the respondents have

obtained a Master degree (31,8%), and the majority is still a student (70,2%). The average age of the participants was 24 years old (SD = 5.54).

Procedure and Stimulus Material

Most of the respondents were recruited through WhatsApp and in order to ease to process of snowball sampling (Biernacki & Waldorf, 1981) the respondents were asked to invite their friends to join this study. Although it will lower external validity, the respondents were recruited trough convenience sampling, not random sampling, as the University of Amsterdam’s network of students will be contacted and due to high expenses of carrying out a research with a random sample, this is accompanied by limited budget available for this research.

After clicking the link the participants entered the introduction and instructions of the questionnaire. After giving consent, all participants were randomly assigned to one of the four conditions, in order to keep a high internal validity. Each condition started with a brief

introduction about the influencer, the brand and the endorsement. After that, they were shown a fictional article in a gossip magazine (Story) and photograph, telling a story about a fictional scandal related to the influencer (Anna Nooshin), who is endorsed with a non-fictional brand (H&M). The article fell within one of two conditions of the independent variable, Crisis

Responsibility, a nominal variable that consists of two levels, namely internal locus of control

Nooshin. The text, which is placed beneath the photograph, explained for each condition (internal vs. external locus of control) a different story, and was therefore the manipulation.

Then, the respondents were exposed to a Twitter message of the brand, the Crisis

Response Strategy. The stimulus consisted of a Tweet sent from the H&M’s Twitter account

on the social media platform Twitter. The participants saw the feed of H&M, which contained the brand logo, brand description, the number of tweets (13,3 thousand), the number of

following (330) and the number of followers (8,406,429).The tweet had this exact same layout for each condition. The manipulation was the content of the Tweet consisting of the answer to the scandal, given by H&M. For the first condition, continuing the endorsement, the Tweet consisted of a statement of the brand stating that they are aware of the negative

publicities concerning Anna Nooshin, but that they will continue the endorsement with her. For the second condition, ending the endorsement, the Tweet consisted of a statement of H&M, stating that they are aware of the negative publicities concerning Anna Nooshin and that they end the endorsement with the influencer (See Appendix A for stimulus material).

After being exposed to one of the four conditions, the respondents were asked to give their extent of agreement with Organizational Reputation and Purchase Intention

measurement items, the dependent variables. Then the questions about the two mediators,

Brand Image and Influencer Identification, were asked in order to not influence the

participants when questioning about the organizational reputation and purchase intention. In the end of the questionnaire the manipulation check was included, in order to avoid bias as well.

Measures

Organizational Reputation was measured by adopting the organizational reputation

scale with 5 items as the Cronbach’s Alpha (α = .91) showed it is reliable (Coombs & Holladay, 2002). ‘Organization’ was replaced by ‘H&M’ and for the third item on the scale ‘crisis’ was replaced by ‘scandal’, in order to make it more clear for the respondents. The scale consisted of questions like: ‘H&M is concerned with the well-being of its public’, ‘H&M is basically dishonest’ or ‘I do not trust H&M to tell the truth about the scandal’ and was measured on a 7-point Likert-scale (1 = strongly disagree, 7 = strongly agree). 3 Items of the scale were recoded so that the higher the participants scored the more positive they were about the organizational reputation. The 5-item scale proved to be reliable as indicated by a Cronbach’s Alpha of α = .74. Then, the average of the scale was computed into the new variable, Reputation_total (M = 4.33, SD = 1.06).

Purchase Intention was measured using a 4-item scale of Bian & Forsythe (2012) as

this scale demonstrates good internal consistency and reliability (Cronbach’s Alpha of α = .90) (Bian & Forsythe, 2012). The scale is slightly adapted to the purpose of this study, as ‘brand’ was replaced by ‘H&M’ in order to make it more clear. Questions like: ‘If I were going to purchase clothes, I would consider buying H&M’ and ‘If I were shopping for a similar brand, the likelihood I would purchase H&M is high’ were measured on a 7-point Likert-scale (1 = strongly disagree, 7 = strongly agree) and no recoding was necessary. The 4-item scale also proved to be reliable as indicated by a Cronbach’s Alpha of α = .93. Then, the average of the scale was computed into the new variable, PurchaseIntention_total (M = 4.56, SD = 1.47).

Brand Image was measured adopting the Martínez, Montaner, & Pina (2009) scale as

it is the most frequently used scale to measure brand image (Plumeyer, Kottemann, Böger, & Decker, 2017), only ‘brand’ was changed into ‘H&M’. The scale consisted of 10 statements, for example: ‘H&M is nice’, ‘H&M has a personality that distinguish itself from competitors’

brands’ and ‘It’s H&M that doesn’t disappoint its customers’, measured with a 7-point Likert-scale (1 = strongly disagree, 7 = strongly agree). The 9th item was removed from the scale in order to optimize the reliability, with Cronbach’s Alpha of α = .90. Then, the average of the scale was computed into the new variable, BrandImage_total (M = 3.98, SD = 1.08).

Finally, Influencer identification, was measured with a 7-point Likert-scale (1 =

strongly disagree, 7 = strongly agree) consisting of five items, for example: ‘I like influencer

Anna Nooshin’ or ‘I can easily relate to influencer Anna Nooshin’ and was adopted from the Um (2013) scale as it is proved to be reliable (Cronbach’s Alpha of α = .83) (Um, 2013). Consumers’ level of identification with an influencer was proved to be reliable, indicated by a Cronbach’s Alpha of α = .86. Then, the average of the scale was computed into the new variable, Influencer_total (M = 3.26, SD = 1.22).

Manipulation check

At the end of the survey, in order to avoid bias, three questions as manipulation check were used in order to check if respondents perceived the experimental manipulation as intended. The first question was for the variable Crisis Responsibility (internal vs. external locus of control) and consisted of the following question: ‘Is the influencer scandal related to the organization?’. The answer consisted of two answer options: yes or no. The manipulation check for the variable Crisis Response Strategy consisted of the two following questions: ‘Did the brand continue the endorsement with the influencer?’ and ‘Did the brand ended the

endorsement with the influencer?’, the answer consisted of two answer options: yes or no. These three questions were the same for all the conditions.

Results Manipulation and Randomization Check

Almost 87% correctly identified if the influencer scandal was related to the

organization. The manipulation of the Crisis Responsibility can be seen as successful, using a chi-square test (χ2 = 60.53, p < .001). Lastly, the manipulation for the Crisis Response

Strategy is successful as well, because 92% of the participants identified the used crisis

response strategy correctly (χ2 = 91.02, p < .001).

The respondents were almost equally divided over the four conditions: external locus of control/ ending endorsement (N = 36), external locus of control/ continuing endorsement (N = 41), internal locus of control/ ending endorsement (N = 33), internal locus of control/ continuing endorsement (N = 38).

Crisis Responsibility Effects

H1a and H1b were tested with an Independent T-Test, with Crisis responsibility (internal locus of control vs. external locus of control) as the independent variable and

Organizational reputation and Purchase intention as the dependent variables. Organizational

reputation did not significantly differ between the internal locus of control condition (M = 4.43, SD = 1.14) and the external locus of control condition (M = 4.23, SD = .98), t(146) = .56, p = .455, IC [-.54,.14], which rejects H1a. Purchase intention did also not significantly differ between the internal locus of control condition (M = 4.37, SD = 1.39) and the external locus of control condition (M = 4.74, SD = 1.53), t(146) = .21, p = .648, IC [-.11,.84], which rejects H1b.

Moderation Effect of Crisis Response Strategy

H2a and H2b were tested with a two-way ANOVA, with Crisis responsibility (internal locus of control vs. external locus of control) as the independent variable, Crisis response

strategy (ending vs. continuing) as the moderator and Organizational reputation and

Purchase intention as the dependent variables. No significant direct effect of crisis response

strategy on organizational reputation F(1, 144) = 2.47, p = .136, or on purchase intention F(1, 144) = 3.08, p = .081 was found.

The interaction between crisis responsibility and crisis response strategy did not significantly predict organizational reputation F(1,144) = 2.22, p = .138, which rejects H2a. Purchase intention is also not significantly predicted by the interaction between crisis responsibility and crisis response strategy F(1,144) = .25, p = .616, which rejects H2b. The means are shown in Table 1.

Table 1: Means of interaction effect of crisis responsibility and crisis response strategy External locus of control Internal locus of control

Continue End Continue End

Dependent variables M (SD) M (SD) M (SD) M (SD)

Organizational reputation 4.47 (.93) 3.96 (.98) 4.43 (1.18) 4.18 (1.06) Brand image 4.99 (1.44) 4.45 (1.60) 4.51 (1.41) 4.21 (1.38)

Note: N = 148, No sig. difference between conditions for all measured variables

The Mediating Effect of Brand Image and Influencer Identification of the Moderation of Crisis Response Strategy

To test the moderated mediation model proposed in H3 and H4, model 2 was used in the PROCESS 3.4 Macro extension in SPSS (Hayes, 2018) A bootstrap with 5,000 samples

was included, for both the dependent variables, to estimate the confidence intervals that account for 95% of bias-correction. Then, an Independent T-Test was conducted.

Model 2 was conducted for the independent variable Crisis Responsibility (internal locus of control vs. external locus of control), the moderators Brand Image and Influencer

Identification and the dependent variables Organizational Reputation and Purchase Intention

(See Appendix C for models). The independent T-Test tested if the Crisis Response Strategy (ending vs. continuing) as independent variable influenced Brand Image and Influencer

Identification as dependent variables.

The Moderation Effect of Brand Image and Influencer Identification

Brand image. A significant direct effect of brand image on organizational reputation

(b2a = .26, p < .05) and on purchase intention (b2b = .90, p < .001), was found. The interaction

between crisis responsibility and brand image did not significantly predict organizational reputation (b4a = .18, p = .254), which rejects H3a. Purchase intention is also not significantly

predicted by the interaction between crisis responsibility and brand image (b4b = -.15, p =

.429), which partly rejects H3b.

Influencer identification. The direct effect of influencer identification on

organizational reputation was insignificant (b3a = .12, p = .224). The direct effect of

influencer identification on purchase intention was also not significant (b3b = -.07, p = .557),

Organizational reputation is not significantly predicted by the interaction between crisis responsibility and influencer identification (b5a = -.07, p = .641), which rejects H4a. Also, the

interaction between crisis responsibility and influencer identification did not significantly predict purchase intention (b5b = -.11, p = .551), which partly rejects H4b. Figure 2 and 3

Figure 2. Effects measured in PROCESS Model 2 with dependent variable Organizational

reputation

The Effect of Crisis Response Strategy on Brand Image and Influencer Identification The results of the Independent T-Test show that brand image did not significantly differ between the continuing condition (M = 4.08, SD = 1.05) and the ending condition (M = 3.84, SD = 1.11), t(146) = .34, p = .561, IC [-.11,.59], which rejects H3a and H3b.

Lastly, influencer identification did not significantly differ between the continuing condition (M = 3.36, SD = 1.28) and the ending condition (M = 3.15, SD = 1.15), t(146) = .94,

p = .333, IC [-.19,.61], all of H4a and H4b is rejected. This means all hypotheses are rejected

(See Appendix D, table 3).

Discussion

With a major increase in influencer marketing as influencer crises, it is more important than ever for both researchers and practitioners to know what kind of impact crisis

responsibility has and to clearly identify which factors can influence the effectiveness of the crisis response and to tailor successful crisis response strategies. Therefore, this study wanted to find out how organizations need to react to endorsed influencer negative publicity.

Accordingly, this study wanted to provide new insights about how crisis responsibility influenced organizational reputation and purchase intention, and to find out which crisis response strategy was best suitable. Finally, what differentiate this study from others, was to find out how the perceptions and emotions of the consumer might affect the effectiveness of the crisis response strategy, being brand image and influencer identification.

Crisis Responsibility

Firstly, this study failed to reproduce significant crisis responsibility effects on the post-crisis evaluation of an organization. Previous studies suggested that negative

organizational outcome threats could be derived from the crisis responsibility within a crisis situation (Coombs, 1995; Coombs, 2007; Jeong, 2009; Page, 2019). Internal locus of control,

where the responsibility is attributed to the organization, leads to more unfavourable feelings of the consumer towards the organization (Jiang et al., 2015) which in turn harms the

organizational reputation (Jeong, 2009) and negatively effects the purchase intention (Klein & Dawar, 2004). This current literature on crisis responsibility contradicts the findings of this study.

However, a highly significant finding in this research is that the respondents recognized which level of responsibility was intended per crisis, meaning that consumers make attributions judgements about the causes of a crisis (Jiang et al., 2015) when negative and unexpected events occur (Weiner, 2006). Organizations should know that consumers are aware of the difference between higher involvement and lower involvement of the

organization in an influencer scandal. Although, they should not worry about the level of responsibility attribution as this will not influence the organizational outcomes.

For both theory as well for practitioners, it is useful to further investigate if crisis responsibility should be approached in a different way than locus of control. The research by Um (2013), suggest that the level of responsibility should be divided into dispositional and situational attributions when a celebrity crisis occurs. In case of situational attributions, people attribute the causal structure of the crisis to the environment whereas the qualities of the person involved in the event will be held responsible in case of dispositional attribution (Um, 2013). The study found that consumers who make dispositional attributions judge the endorsed brand more than consumers who make situational attributions (Um, 2013).

A practical implication is that brands should persuade consumers to focus on situational factors. This means that, when bringing out the bad news, communication

professionals should emphasize situational factors, like social and physical circumstances of the influencer crisis.

Moderation effect of Crisis Response Strategy

Secondly, it was expected that the crisis response strategy moderated the relation between crisis responsibility and the post-crisis evaluation. Although results revealed that ending or continuing the endorsement did not significantly change the effects on the relation between crisis responsibility and the post-crisis evaluation of the organization, as well as no direct effect on organizational reputation nor on purchase intention. This contradicts the literature, stating that if organizations use the right response strategy after a crisis it will reduce damage to organizational outcomes (Coombs, 2007; Coombs & Holladay, 2009; Chao, 2019; Stieglitz et al., 2019) and that in case of a higher degree of responsibility, a more

defensive response strategy should be adopted (Chao, 2019).

Although not significant, the results revealed that in this sample, both for external and internal locus of control, continuing the endorsement leads to higher organizational reputation and purchase intention. If the organization is not responsible for the influencer crisis, they should continue the endorsement. However not significant, it can be assumed that the consumer expects the organization to take their responsibility independently of the level of responsibility.

Ending or continuing the endorsement did not significantly change the post-crisis evaluation, therefore communication professionals should not focus on ‘what to say’ but on ‘how to say’ it. Future research should more focus on the content style and presentation format of the message (Lee & Atkinson, 2019) to see how this influences the efficacy of the response.

However, the respondents were significantly capable of recognizing the used response strategy by the organization. A practical implication of this result is that the organization needs to be attentive of the implementation of the response strategy in their after-crisis communication as the consumer is aware and recognizes the used response strategy.

Mediation effect of Brand Image

Thirdly, this study did not find a mediating effect of brand image on the crisis response strategy on the relation of crisis responsibility and the post-crisis evaluation of the organization. As the results showed that brand image did not moderate the relation between crisis responsibility and the post-crisis evaluation and crisis response strategy did not

influence brand image. This is not in line with previous literature, stating that brand image is critical in understanding the responses of the consumer (Jeong et al., 2013). The study by Lee and Atkinson (2019) discussed that the persuasiveness of the chosen crisis response strategy depends on brand image, as the consumer will take the crisis more seriously when emotional attached to the brand.

Communication managers should understand and be aware that brand image will not save or reduce negative effects if the organization chooses the wrong crisis response strategy. Therefore, it can be concluded that brand image is not a factor that changes the effectiveness of a response strategy.

Perhaps for future research, brand commitment should be considered, as the research by Um (2013) suggests that consumer’s brand commitment plays an important role and can be defined as an emotional or psychological attachment to a brand within a product class. The research argues that consumers with a high-commitment to the brand are more likely to counterargue negative publicity and therefore maintain their original attitude, which leads to a more positive brand evaluation as well as higher purchase intention (Ahluwalia, Burnkrant & Unnava, 2000; Um, 2013).

Another study suggests that brand image should be categorized as being functional or symbolic (Lee & Atkinson, 2019). Functional brand image can be seen as the brand’s

(Lee & Atkinson, 2019). Symbolic brand image is about the brand’s product’s self-expressive value (Jeong et al., 2013), related to uniqueness, self-enhancement and uniqueness (Lee & Atkinson, 2019). The study found that when individuals are more emotionally connected to the brand, they will show more favourable attitudes towards the brand and the message. Meaning that, a symbolic brand image will lead to more favourable attitude towards the brand and the message in combination with higher purchase intention after a crisis than a functional brand image as they have a weaker association with the crisis (Lee & Atkinson, 2019).

However, this study found a direct effect of brand image on the organizational

reputation and on the purchase intention. The result offers additive insights to those of Raithel and colleagues (2010), by showing that the magnitude of influence on the organizational reputation and purchase intention is depending on the brand image of the consumer. The effect of brand image on purchase intention was larger than the effect of brand image on organizational reputation.

Communication professionals should understand that a more positive consumer’s perception about the organization after the crisis will more positively affect the purchase intention than the organizational reputation. For future research, it could be beneficial to examine if purchase intention and organizational reputation are related, perhaps a level of purchase intention leads to a higher organizational reputation.

Mediation effect of Influencer Identification

Lastly, this study failed to find evidence for the mediating effect of influencer

identification on the crisis response strategy and the relation between crisis responsibility on the post-crisis evaluation. The results showed that influencer identification did not moderate the relation between crisis responsibility and the post-crisis evaluation, also, the crisis response strategy did not influence influencer identification.

This contradicts the study by Burke (1950) which stated that the effectiveness of communication depends on the identification with a fictional character. Consumers who share thoughts, feelings and behaviours with the influencer (Um, 2013) via the media (Horton & Wohl, 1956) are less likely to react negatively on scandals (Johnson, 2005; Um, 2013). Therefore it was suggested that consumers have a high level of identification with the influencer, would be more in favour of continuing the endorsement.

A practical implication is that however choosing an influencer with a high level of attractiveness, trustworthiness and expertise (Erdogan, 1999; Lear et al., 2010), it will not reduce the damage of organizational outcomes after an influencer crisis. Furthermore, both consumers with high or low level of influencer identification will respond in the same way after implementing a response strategy. This means that influencer identification is not a factor that influences the effectiveness of a response strategy.

Also, only one influencer was used in this study, namely Anna Nooshin. For future research it is suggested to include more influencer to see if this would change the results and in order to be able to better generalize the results.

Conclusion

All by all, this research shed light on the importance of being aware of the risks influencer marketing can bring. More and more organizations tend to choose influencer marketing over the traditional media channels as it often seems less expensive and more efficient. Taking the results of this study into account, it appears that the risks of only focussing on influencer marketing are not taken away, as the perfect crisis response strategy after an influencer crisis still has to be found. Therefore, it is important, for start-ups in

particular, to not only focus on influencer marketing but to be keep investing in the traditional media. However, to reduce the risks, organizations could also choose for endorsing with

micro-influencers (Cruz, 2018) as it is less likely for them to get involved in a scandal. Micro-influencers tend to build long-term relationships with consumers which are built on trust, and create consumer engagement, while macro-influencers are more likely to enhance short-term brand awareness (Cruz, 2018).

However, some limitations to this study need to be acknowledged. First, the organization used in this experiment is H&M, which is a well-known and existing

organization. This could have introduced some bias to the results, as the respondents could have earlier experiences with the brand and its products. Second, Anna Nooshin is a well-known influencer in the Netherlands which can possibly bias the results, as the respondents already have an opinion and feeling about her before joining the experiment. However, during a real influencer endorsement crisis, the consumers also have past experiences with both the organization and the influencer. Therefore, it is relevant for the results to use an existing, well-known organization and influencer. Third, only one influencer and one brand is used in this experiment, therefore it is hard to generalize the results for all influencers and brands active on Instagram. Fourth, all the materials, namely the articles in the gossip magazine and the Twitter messages, used in this experiment were fictional. A limitation here, is that the respondents were aware of the materials being fictional, meaning that this could have influenced there answers as they possibly took the crisis not as serious as they would in real life. On the other hand, using fictional materials leads to less bias, as this is the first time the respondents are exposed to the crisis.

Finally, this study shows the importance of tailoring crisis responses strategies attentively. Given the influence of brand image on the post-crisis evaluation and the fact that the influence of locus of control and influencer identification need to be further researched, it can be advised that a one-size-fits-all approach is not to be used when undertaking influencer crisis communication. Alternately, guidelines should be developed for more targeted

approaches in case of an influencer crisis. This would not only optimize the creation of the best possible apology message, both content and design, but also result in maintaining the effectiveness of crisis communication. Lastly, since finding and applying the best suitable response strategy is often as difficult as the crisis itself, this study provides a valuable starting point to design an effective crisis response strategy by taking the crisis responsibility of the organization and the emotions of the consumer towards the organization and the influencer into account.

As it occurs that consumers make perceptions about the causes of a crisis and

recognize the crisis response, future research should look for other emotional indicators that could adjust the response strategy. Also, the possible changing character of a crisis in this social media age could be researched. Another opportunity would be expanding the research field, for example to find out if the chosen spokesman who brings out the news about the crisis will affect the crisis response strategy, or perhaps a different type of media channel should be taken into account.

References

Abidin, C. (2016). Visibility labour: Engaging with influencers’ fashion brands and# OOTD advertorial campaigns on Instagram. Media International Australia, 161(1), 86-100. doi: 10.1177/1329878X16665177

Abratt, R., & Kleyn, N. (2012). Corporate identity, corporate branding and corporate reputations. European Journal of Marketing, 46(7-8), 1048‐1063. doi:

10.1108/03090561211230197

Abratt, R., & Mofokeng, T. N. (2001). Development and management of corporate image in South Africa. European Journal of Marketing, 35(3-4), 368-386. doi:

10.1108/03090560110382075

Ahluwalia, R., Burnkrant, R. E., & Unnava, H. R. (2000). Consumer response to negative publicity: The moderating role of commitment. Journal of Marketing Research, 37(2), 203-214. doi: 10.1509/jmkr.37.2.203.18734

Andreassen, T. W. (2001). From disgust to delight: Do customers hold a grudge? Journal of

Service Research, 4(1), 39-49. doi: 10.1177/109467050141004

Balmer, J. M. (2017). The corporate identity, total corporate communications, stakeholders’ attributed identities, identifications and behaviours continuum. European Journal of

Bandura, A. (1986). The explanatory and predictive scope of self-efficacy theory. Journal of

Social and Clinical Psychology, 4(3), 359-373. doi: 10.1521/jscp.1986.4.3.359

Bartz, S., Molchanov, A., & Stork, P. A. (2013). When a celebrity endorser is disgraced: A twenty-five-year event study. Marketing Letters, 24(2), 131-141. doi: 10.1007/s11002-013-9229-2

Benoit, W. L., & Dorries, B. (1996). Dateline NBC's persuasive attack on Wal-Mart.

Communication Quarterly, 44(4), 463-477. doi: 10.1080/01463379609370032

Bian, Q., & Forsythe, S. (2012). Purchase intention for luxury brands: A cross cultural comparison. Journal of Business Research, 65(10), 1443-1451. doi:

10.1016/j.jbusres.2011.10.010

Bick, G., Jacobson, M. C., & Abratt, R. (2003). The corporate identity management process revisited. Journal of Marketing Management, 19(7-8), 835-55. doi:

10.1080/0267257X.2003.9728239

Biernacki, P., & Waldorf, D. (1981). Snowball sampling: Problems and techniques of chain referral sampling. Sociological Methods and Research, 10(2), 141-163. doi:

10.1177/004912418101000205

Biswas, S., Hussain, M., & O'Donnell, K. (2009). Celebrity endorsements in advertisements and consumer perceptions: A cross-cultural study. Journal of Global Marketing, 22(2), 121-137. doi: 10.1080/08911760902765940

Burke, K. (1950). A rhetoric of motives. Berkeley: University of California Press.

Carrillat, F. A., Astous, A., & Lazure, L. (2013). For better, for worse? What to do when celebrity endorsements go bad. Journal of Advertising Research, 51(1), 15-30. doi: 10.2501/JAR-53-1-015-030

Chao, Q. (2019). The effectiveness of corporate response to brand crises: The crisis response strategies, and customer-based brand equity. International Conference on

Economics, Management and Humanities Science, 5, 363-370. doi:

10.25236/ecomhs.2019.077

Coombs, W.T. (1995). Choosing the right words: The development of guidelines for the selection of the “appropriate” crisis response strategies. Management Communication

Quarterly, 8, 447-476. doi: 10.1177/0893318995008004003

Coombs, T., Holladay, S. (2015). CSR as crisis risk: Expanding how we conceptualize the relationship, corporate communications. An International Journal, 2, 144-162. Doi: 10.1108/CCIJ-10-2013-0078

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

Coombs, W. T. (2007). Protecting organization reputations during a crisis: The development and application of situational crisis communication theory. Corporate Reputation

Review, 10(3), 163-176. doi: 10.1057/palgrave.crr.1550049

Cruz, F. (2018, May 16). Influencers: Micro & macro. Tapinfluence. https://www.tapinfluence.com/influencers-micro-macro/

Deephouse, D. L., & Suchman, M. (2008). Legitimacy in organizational institutionalism. The

Sage Handbook of Organizational Institutionalism, 49, 77.

Djafarova, E., & Rushworth, C. (2016). Exploring the credibility of online celebrities’ Instagram profiles in influencing the purchase decisions of young female users.

Computers in Human Behavior, 68, 1-7. doi: 10.1016/j.chb.2016.11.009

Domingues Aguiar & Van Reijmersdal (2018). Influencer Marketing. Chapter 2: What is influencer marketing. SWOCC: Amsterdam

Erdogan, B. Z. (1999). Celebrity endorsement: A literature review. Journal of Marketing

Management, 15(4), 291-314. doi: 10.1362/026725799784870379

Evans, N. J., Phua, J., Lim, J., & Jun, H. (2017). Disclosing Instagram influencer advertising: The effects of disclosure language on advertising recognition, attitudes, and behavioral intent. Journal of Interactive Advertising, 17(2), 138-149. doi:

Hayes, A. F. (2018). Introduction to mediation, moderation, and conditional process

analysis: A regression-based approach (2nd ed.). Guilford Publications.

Hartmann, T., & Goldhoorn, C. (2011). Horton and Wohl revisited: Exploring viewers' experience of parasocial interaction. Journal of Communication, 61(6), 1104-1121. doi: 10.1111/j.1460-2466.2011.01595.x

James, D. (2005). Guilty through association: brand association transfer to brand alliances.

Journal of Consumer Marketing, 22(1), 14-24. doi:

doi.org/10.1108/07363760510576518

Jeong, S. H. (2009). Public’s responses to an oil spill accident: A test of the attribution theory and situational crisis communication theory. Public Relations Review, 35(3), 307-309. doi: 10.1016/j.pubrev.2009.03.010

Jeong, H. J., Paek, H. J., & Lee, M. (2013). Corporate social responsibility effects on social network sites. Journal of Business Research, 66(10), 1889-1895. doi:

10.1016/j.jbusres.2013.02.010

Jiang, J., Huang, Y. H., Wu, F., Choy, H. Y., & Lin, D. (2015). At the crossroads of inclusion and distance: Organizational crisis communication during celebrity-endorsement crises in China. Public Relations Review, 41(1), 50-63. doi:

Johnson, A. R. (2005). When a celebrity is tied to immoral behavior: Consumer reactions to Michael Jackson and Kobe Bryant. Advances in Consumer Research, 32, 100-101.

Keller, K. L. (1993). Conceptualizing, measuring, and managing customer-based brand equity. Journal of marketing, 57(1), 1-22. doi: 10.1177/002224299305700101

Kelman, H. C. (1961). Processes of attitude change. Public Opinion Quarterly, 25(1), 57-78.

Klein, J., & Dawar, N. (2004). Corporate social responsibility and consumers' attributions and brand evaluations in a product–harm crisis. International Journal of Research in

Marketing, 21(3), 203-217. doi: 10.1016/j.ijresmar.2003.12.003

Koo, G. Y., Ruihley, B. J., & Dittmore, S. W. (2012). Impact of perceived on-field

performance on sports celebrity source credibility. Sport Marketing Quarterly, 21, 147-158.

Lear, K. E., Runyan, R. C., & Whitaker, W. H. (2008). Sports celebrity endorsement in retail products advertising. International Journal of Retail and Distribution Management,

37(4), 308–321. doi: 10.1108/09590550910948547

Lee, S. Y., & Atkinson, L. (2019). Never easy to say “sorry”: Exploring the interplay of crisis involvement, brand image, and message appeal in developing effective corporate apologies. Public Relations Review, 45(1), 178-188. doi:

Lu, L. C., Chang, W. P., & Chang, H. H. (2014). Consumer attitudes toward blogger's sponsored recommendations and purchase intention: The effect of sponsorship type, product type, and brand awareness. Computers in Human Behavior, 34, 258-266. doi: 10.1016/j.chb.2014.02.007

Martínez, E., Montaner, T., & Pina, J. M. (2009). Brand extension feedback: The role of advertising. Journal of Business Research, 62(3), 305-313. doi:

10.1016/j.jbusres.2008.05.009

Money, R. B., Shimp, T. A., & Sakano, T. (2006). Celebrity endorsements in Japan and the United States: Is negative information all that harmful? Journal of Advertising

Research, 46(1), 113-123. doi: 10.2501/S0021849906060120

Nekmat, E., & Kong, D. (2019). Effects of online rumors on attribution of crisis responsibility and attitude toward organization during crisis uncertainty. Journal of Public Relations

Research, 1-19. doi: 10.1080/1062726X.2019.1644644

Page, T. G. (2019). Beyond attribution: Building new measures to explain the reputation threat posed by crisis. Public Relations Review, 45(1), 138-152. doi:

10.1016/j.pubrev.2018.10.002

Plumeyer, A., Kottemann, P., Böger, D., & Decker, R. (2019). Measuring brand image: A systematic review, practical guidance, and future research directions. Review of

Raithel, S., Wilczynski, P., Schloderer, M. P., & Schwaiger, M. (2010). The value-relevance of corporate reputation during the financial crisis. Journal of Product and Brand

Management, 19(6), 389-400. doi: 10.1108/10610421011085703

Stieglitz, S., Mirbabaie, M., Kroll, T., & Marx, J. (2019). Silence” as a strategy during a corporate crisis–the case of Volkswagen’s “Dieselgate. Internet Research, 29(4), 921-939. doi: 10.1108/INTR-05-2018-0197

Suchman, M. C. (1995). Managing legitimacy: Strategic and institutional approaches. Academy of Management Review, 20(3), 571-610. doi: 10.5465/amr.1995.9508080331

Till, B. D., & Shimp, T. A. (1998). Endorsers in advertising: The case of negative celebrity information. Journal of Advertising, 27(1), 67-82. doi:

10.1080/00913367.1998.10673543

Tripp, C., Jensen, T. D., & Carlson, L. (1994). The effect of multiple product endorsements by celebrities on consumer attitudes and intentions. Journal of Consumer Research,

20(4), 535-547. doi: 10.1086/209368

Um, N. H. (2013). Celebrity scandal fallout: How attribution style can protect the sponsor.

Um, N. H., & Kim, S. (2016). Determinants for effects of celebrity negative information: When to terminate a relationship with a celebrity endorser in trouble? Psychology and

Marketing, 33(10), 864-874. doi: 10.1002/mar.20923

Upadhyay, Y., & Singh, S. K. (2010). When sports Celebrity doesn't perform: How consumers react to celebrity endorsement? Vision, 14(1-2), 67-78. doi: 10.1177/097226291001400107

Walker, M., Langmeyer, L., & Langmeyer, D. (1992). Celebrity endorsers: Do you get what you pay for? Journal of Consumer Marketing, 9(2), 69-76. doi:

10.1108/07363769210037033

Weiner, B. (1985). An attributional theory of achievement motivation and emotion. Psychological Review, 92(4), 548.

Weiner, B. (2006). Social motivation, justice, and the moral emotions: An attributional approach. Psychology Press.

Appendix A

Visual stimuli

Appendix B

Measurement scales

Table 1: Organizational reputation

Item Statement Item scales

1 The organization is concerned with the well-being of its public 7-point Likert scale (1 = strongly disagree, 7 = strongly agree) 2 The organization is basically dishonest (R)

3 I do not trust the organization to tell the truth about the scandal (R) 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 public (R)

Table 2: Purchase intention

Item Statement Item scales

1 If I were going to purchase clothes, I would consider buying H&M 7-point Likert scale (1 = strongly disagree, 7 = strongly agree) 2 If I were shopping for a similar brand, the likelihood I would

purchase H&M is high

3 My willingness to buy H&M would be high if I were shopping for a similar products

4 The probability I would consider buying H&M is high

Table 3: Brand image

Item Statement Item scales

1 H&M is nice 7-point Likert scale

(1 = strongly 2 H&M has a personality that distinguish itself from competitors’