CELEBRITY ENDORSEMENT
SCANDAL
THE CASE OF SPORTS CELEBRITIES
Charles Gilbert
student number S2693755
M.Sc. IB&M (2018/19) – Master Thesis
Faculty of Economic and Business
University of Groningen
1
Abstract
Celebrity endorsement is one of the most important forms of advertising in today’s marketing arena. However, celebrity endorsement also has several risks, one of which arises from the possibility that the celebrity endorser undergoes a public scandal. Although research has investigated the effect of a celebrity endorser scandal on purchasing intent, it has resulted in contradicting evidence. Hence this research aims to improve and validate the theories of celebrity endorsements by researching two variables that ought to influence purchasing intent after a celebrity scandal has occurred. These two variables are celebrity product-congruence and the response strategy of the firm.
Research indicates that purchasing intent will be affected more severely when the celebrity-product congruence is high, as it aids the transfer of adverse spillover effects from the celebrity to the product. Additionally, it is believed that purchasing intent will react more favourable when the organization utilizes a separation strategy: the organization disassociates itself from the celebrity and illustrates that the public scandal is not tolerated.
2
Table of Content
1. Introduction ... 4
2. Literature review ... 6
2.1 Celebrity endorsement effect on purchasing intent ... 6
2.2 Theoretical analysis celebrity endorsements ... 6
2.3 Congruence ... 8
2.4 Risks of using celebrity endorsers ... 9
2.4.1 Celebrity endorser undergoing a scandal ...10
2.4.2 Factors influencing purchasing intent after a celebrity scandal ...11
2.5 Response strategies ...14
2.5.1 Why should the organization respond? ...14
2.5.2 Variables organization should consider when responding to a celebrity endorsement scandal ...15
2.5.3 Image Repair Typologies ...16
3. Formulating Research Hypotheses ...19
3.1: Celebrity endorsement effect on purchasing intent ...19
3.2 Match-Up ...20
3.3 Scandalous effect and congruency ...21
3.4 Response strategies ...22
3.5 Level of congruence and response strategies ...22
3.6 Control variables ...23
4. Methods ...25
4.1 Overview ...25
4.2 Pilot study ...26
4.3 Experiment study ...28
4.4 Participants, design and Procedure ...30
4.4 Questions and Measures. ...32
4.4.1 Independent variable: celebrity congruence ...32
4.4.2 Moderator: measuring response strategies ...33
4.4.3 Moderator: measuring scandalous effect ...33
4.4.4 Dependent variable: measuring purchasing intent: ...33
4.5 Data Collection and Sampling strategy ...33
3 5.1: Manipulation check ...34 5.2: Hypotheses testing ...34 5.2.1 Experiment 1 ...34 5.2.2 Experiment 2 ...38 6. Discussion ...41 7. Managerial implications ...48 8. Theoretical implications ...49
9. Future research and limitations ...50
10. Conclusion ...52
11. Bibliography ...53
12. Appendix A. ...59
13. Appendix B. ...64
14.1: Pilot study survey. ...64
4
1. Introduction
Celebrity endorsement has evolved into one of the most popular forms of advertising in today’s world (Choi and Rifon, 2007). It has developed into a significant marketing tool worldwide and is utilized in one in four advertisements (MarketWatch, 2006).
The popularity of celebrity endorsement can be attributed to the abundant advantages of utilizing this type of marketing. For example, research suggests that the personality, fame, and characteristics of the celebrity induce greater attention to the products and services they deliver (Atkin and Block, 1983; O’Mahony and Meenaghan, 1997). Additionally, celebrity endorsement enables higher marketing message recall, which is of particular relevance in today’s cluttered advertising environment. In turn, celebrity endorsement enhances consumers’ acknowledgement of the product and increases the purchasing intent of the consumers (Erdogan, 1999). Furthermore, research suggests that the level of congruency plays an important role in influencing consumers. Congruency, composed of the match-up hypotheses, infers that a high level of physical attractiveness and expertise endorser regarding the functional intention of the product will enable more favourable responses from consumers (Erdogan, 1999). Hence, brands often seek high congruent celebrity endorsers for their brands (Till, Stanley, and Priluck, 2008).
5
the match-up hypotheses has also not received much attention when a scandal occurs. Similarly, how the endorsing brand reacts after the celebrity scandal has also revealed scant literature attention. Goldsmith, Lafferty and Newel (2000) illustrate that a company’s response influences consumers perception of the celebrity scandal. Thus, in order to understand the effect of celebrity endorser scandal, we should understand how consumers react to the behaviour of firms towards the scandalous celebrity endorser (Louie and Obermiller, 2013). Do consumers approve of a brand dropping an endorser for the scandal? Do consumers respect a company for sackings their endorser, or do they want them to stick by him?
Therefore, this paper will focus primarily on how a celebrity endorser scandal influences purchase intention, and how the congruency among celebrity and brand/product affect this relationship. This paper further seeks to add value by assessing how an organization should respond in the aftermath of a celebrity endorser scandal, to leverage more favourable purchasing intent from the consumer (White, Goddard and Willbur, 2000). Resulting in the following research question:
To what degree does the congruence between a celebrity and a product influence the purchasing intent of consumers after a celebrity endorser scandal, and how is this relationship moderated by the organizational response strategy?
6
2. Literature review
2.1 Celebrity endorsement effect on purchasing intent
Utilizing celebrity endorsement is a standard advertising practice, and has received extensive practical and academic attention (Erdogan, 1990).
A celebrity endorsement occurs when a well-known person is used to promote a company or a product brand (Broderick and Pinkton, 2005). Celebrity endorsement is applicable to a variety of functions but is most commonly used in the advertising context.Research suggests that the personality, fame, and characteristics of the celebrity induce more attention to the products and services they deliver (Atkin and Block, 1983; O’Mahony and Meenaghan, 1997). In fact, celebrity endorsement does not only increase the attention consumers pay but also enables higher recall rates of marketing messages in today's cluttered environment (Erdogan, 1999). Celebrity endorsement further influences perceived product quality and uniqueness (Dean, 1999), improved stock market value of the company (Agrawal and Kamakura, 1995), more positive brand evaluations, (Silvera and Austad, 2004), and more favourable word of mouth (Bush, Martin, and Bush, 2004). Thus, celebrity endorsements are widely utilized in advertising to positively influence purchase intention of consumers (Shimp 2003; Bowman 2002).
2.2 Theoretical analysis celebrity endorsements
In order to analyse specifically what effect celebrity endorsement has on consumers, celebrity endorsement can be explained using different perspectives.
Firstly, Kamins (1989) suggests that when three processes of social influence are addressed, a consumer is more likely to allow himself to be influenced by another individual or group of individuals.These processes consist of identification, compliance and internalization.
7
those of the other person or group (Kelman, 1961). When a company utilises a celebrity endorsement marketing strategy, the celebrity utilises these three processes to influence consumers.
Second, McCracken’s (1989) transfer model demonstrates how celebrity endorsement influences consumers. This model elaborates on how the popularity of the celebrity is transferred to the product or brand, and finally to the consumer in three steps. McCracken visions a celebrity as an individual who is symbolized by an abundance of characteristics such as lifestyle, personality, age, and gender. In step one, the characteristics of the celebrity are mingled together to form the portrait of the celebrity. In the second step, the characteristics are conveyed to the brand or commodity and in the third and final step, the meanings of the celebrity and the brand/product are conveyed to the consumer through consumption. Moreover, McCracken (1989) infers that it is important that companies choose celebrities based on their characteristics which are beneficial to the company, and with which they want to be associated. The three steps of McCracken transfer model are illustrated in figure 1.
Figure 1: McCracken Transfer Model regarding Celebrity endorsement
8
Third, another perspective regarding the effectiveness of celebrities’ endorsement is illustrated by McGuire’s (1968) source attractiveness model. The model is based around four dimensions, namely: likability, similarity, familiarity, and attractiveness. The higher the celebrity scores on these four dimensions, the more likely he or she is to persuade the consumer to buy the product endorsed (Ohanian 1990).
Lastly, another factor relating to the effectiveness of the celebrity endorsement is the source credibility model, which stipulates that an endorser’s effectiveness has three essential segments: physical attractiveness, trustworthiness, and expertise regarding the product they advertise (McGuire,1968).
2.3 Congruence
The previously mentioned perspectives are influenced by one vital variable: the congruence between the celebrity and the brand/product. Previous research illustrates that the perception of congruency is theoretically composed of the Match-Up hypothesis (Till and Busler, 1998,2000; Till, Stanley, and Priluck, 2008) and has two dimensions.
Early studies highlighted the Match-Up effectiveness with regards to the physical attractiveness. These studies illustrated that an attractive celebrity endorser is more effective than an unattractive celebrity endorser when the product is related to attractiveness (e.g., cosmetics, perfume, hair products) (Kahle and Homer, 1985). However, functional or technological related products might not need the same appeal due to the irrelevance of the attractiveness of the product. Therefore, Till and Busler (2000) introduced a second dimension of the Match-Up hypothesis, relating to the expertise of the celebrity endorser and the functional intention of the product.
9
Misra and Beatty (1990) underscore that the general consumer perception of congruence between the celebrity and the brand/product being endorsedis the most salient variable as In a similar vein, Marshall and Deuskar (2008) illustrate that if the picture of the celebrity endorsing a product and the self-image of the consumer is too contrasting, the consumer can create post-purchase cognitive dissonance, which will negatively affect purchasing intention.
Thus, the Match-Up hypothesis advocates that when there is a high level of congruence between the celebrity and the brand – along with the messages the celebrity conveys regarding the product – the consumer response will be more favourable (Erdogan, 1999).
2.4 Risks of using celebrity endorsers
Although celebrity endorsement has several advantages, it also entails certain risks marketers should be wary about. Bailey (2007) highlights the importance of considering consumer scepticism. Consumer scepticism refers to consumers realising that marketers have particular purposes, such as convincing the consumers to buy certain products, and that advertisers may be untruthful. In turn, consumers will be more sceptical in accepting the message of the advertisers, and hence reducing the intended effect of the advertisement. Consumers scepticism can also be reinforced by the celebrity endorser endorsing too many different brands and products. The influence and credibility that the celebrity endorser has on the product will diminish as the association between the celebrity, and the particular product are no longer unique.
10
2.4.1 Celebrity endorser undergoing a scandal
Past literature illustrates that celebrity endorsement is a beneficial marketing strategy for an organization. However, using endorsers can be risky, as the behaviour of an endorser can be unpredictable (Money, Shimp, and Sakano, 2006). Such unpredictable behaviour could lead to the celebrity muddling in a scandal.
Thwaites, Lowe, Monkhouse and Barnes (2012) support this notion by illustrating that celebrity endorsement is a double-edged sword, with an endorser scandal potentially having the opposing intended effect. In order words, when a scandalous affair becomes the focus of media attention, it can have a detrimental effect on the consumers’ opinions of the celebrity and potentially spill over to the product they endorse (Lear, Runyan, Whitaker, 2009). Also, the Match-Up hypothesis, would no longer prove effective but damage the celebrity endorser’s reputation and alters consumers’ perceptions of the product being endorsed.
Furthermore, people tend to be more perceptive to negative information than positive information when composing overall judgement of the news (Klein, 1996; Mizerski, 1982; Skowronski and Carlston, 1989). The Merriam formula, which measures the media impact, illustrates that negative news impacts people four times higher than positive news (Krolof, 1988). However, previous literature has been somewhat scant and contractionary with regards to the resulting effect on purchasing intent (Edwards and Ferle, 2009; Thwaites et al, 2012).
11
In contrast, several other authors illustrate that celebrity endorser scandals have led to unfavourable assessments of the purchasing intentions of the brands they endorse (Bailey 2007; White et al 2009). Till and Shimp (2008), illustrate that when a celebrity receives negative press it will reflect badly on the commodity he or she endorses. Moreover, Edwards and Ferle (2009) infer that public scandals caused by the celebrity negatively impacts the celebrities with regards to their credibility. This impact reduces their trustworthiness and attractiveness along with the appropriateness of being the spokesperson of the brand. Similarly, Klebba and Unger (1983) found that negative information regarding the celebrity endorser tarnished the perception of the celebrity as well as the product. As the connection between the celebrity and the product is frequently repeated in advertisements, consumers will start associating the product with the celebrity, in turn also allowing the possibility of negative information spillovers to the product. When the celebrity endorser undergoes a public scandal, the customers acquire additional information about the celebrity. Consequently, the meanings a customer associated with the celebrity are affected. In turn, McCracken's (1989) transfer model is adjusted, as the bundles of meanings which create the image of the celebrity have been changed: the characteristics of the celebrity added to the product or brand have also changed, which eventually will influence whether people will decide to consume the product. Hence negatively affecting purchasing intent of the consumer.
2.4.2 Factors influencing purchasing intent after a celebrity scandal
12
cognitive dissonance can vary in magnitude. As a result, variables influencing the magnitude of cognitive dissonance were identified and can be viewed in figure 1 below.
Firstly, the nature of a controversy that the celebrity endorser scandal was involved in plays a key factor Baileys (2007). Naturally, a more severe scandal (i.e. celebrity endorser being accused of rape or murder) would impact the purchasing intent negatively. The severity of a scandal is influenced by the culpability and degree of a celebrity scandal.
Research suggests that when a celebrity endorsement is responsible for the occurrence of a negative scandal, consumers will reciprocate a more unfavourable attitude towards the celebrity as opposed to when the celebrity is viewed as not responsible. Subsequently, consumers purchasing intent will be more negatively influenced by a celebrity endorsement scandal in which the celebrity endorser has a high degree of culpability (Louie and Obermiller,2002). Lastly, the severity of a scandal increases if the scandal results in innocent people being hurt rather than the scandal mainly impacting the celebrity endorser himself/herself, or whether (Money et al, 2006). Second, the association between product and celebrity can also have an impact on consumers’ purchasing intent (Baileys, 2007). In Bailey’s (2007) study, he infers that consumers might not be as influenced by the celebrity endorsement scandal, if the celebrity does not have a strong association with the brand. In a similar vein, Sato, Arai, Tsuji and Kay (2019) illustrate that when a celebrity endorser undergoes a public scandal relevant to their expertise domain, consumers yield a higher negative emotion towards the celebrity, as opposed to when the celebrity endorser undergoes a public scandal unrelated to their field of expertise.
13
celebrity endorser will be more likely to believe in the celebrity endorser innocence’s, and hence the credibility of the endorser does not diminish. In a similar vein, Lee and Kwak and Braunstein-Minkove (2016) infer that high identification between a celebrity and consumer will enable the consumer to withstand negative information. Additionally, when a consumer has a strong liking of the celebrity endorser prior to the celebrity scandal, that they will show more sympathy towards to celebrity endorser and the scandal will have a less detrimental effect on purchasing intent (F and Wyler, 2012).
Furthermore, a celebrity endorser’s immoral action can trigger cognitive dissonance (Bhattacharjee, Berman and Reed, 2012; Festinger and Carlsmith, 1959). Cognitive dissonance induces consumers to utilize certain moral reasoning methods to handle the negative information created by the celebrity scandal. Consumers will tend to have a higher cognitive dissonance when there is a high level of scandal severity, low level of celebrity expertise and a high association between the product and the celebrity. In turn, this will decrease the purchasing intent of the consumer. This can be summarized in the graph below.
Graph 1: The factors influencing purchasing intent.
Culpability Nature of scandal Self-inflected
The severity of the scandal Prior scandals Liking Prior scandals Identification Strong association Relevance of domain Credibility of celebrity Association of brand and
14
2.5 Response strategies
To counteract negative spill-overs which a celebrity scandal creates, past research illustrates that it is of interest to examine how the organization handles a negative event for their celebrity endorser (Louie and Obermiller, 2002). The organization has two possibilities. Either the celebrity endorser builds prior positive associations with the consumer before the scandal occurs or a post-scandal response of the organization should be elicited (Sato, et al, 2019). As the former option is not feasible after the celebrity scandal occurs, this paper will focus on the post-scandal response of the organization, also known as the crisis communication of the organization. Here the intention is to mitigate the damage, by communicating effectively and appropriately to the stakeholders (Croombs, 2007).
2.5.1 Why should the organization respond?
It is of importance that the organization responds after a crisis occurs, in order to deal with the organizational perception management.Organizational perception management includes actions that are constructed and achieved by the organizational representative in order to affect the audience’s impressions of the organization (Croombs, 2014). This implies that an organizational image is a relatively short-lived perception by the outside audience, to the fit with particular categories, such as to the organizational legitimacy, correctives, consistency and trustworthiness.
Organizational legitimacy refers to “a generalized perception 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. 5740). Organizations that are deemed legitimate are recognized as more trustworthy, meaningful and predictable. As a direct result, an organization is more likely to receive loyal support and resources from its constituents when it is perceived as legitimate. In particular, in times of crisis, the value of legitimacy is of increasing importance, as the legitimacy of the organization is challenged.
15
Organizational trustworthiness is defined as “perceptions that an organization displays competence, benevolence, and integrity in its behaviours and beliefs” (Elsbach, 2003, P. 303). In this definition, integrity refers to company’s obedience to ideals or principles that accommodate social norms. Benevolence implies the firm’s willingness to do good and competence is referred to as the abilities and skill that allow the organization its intended goals.
An organization should also respond to its celebrity endorser undergoing a public scandal, in order to protect its reputation. Organizational reputation differs with the image of the organization; with reputation being more-long lasting and it being more general. Organizational reputation is defined as “the overall estimation in which a company is held by its constituents.” (Fombrun, 1996, p.37). The reputation of an organization also illustrates how the company’s strategies, products, jobs and prospects compare to competing organizations, thus reputations are characterized by status categorizations. It is importance to keep the reputation intact when an organization endures a celebrity endorser’s public crisis. Research illustrates that when the organization has a positive reputation, it increases the chances of constituents to give their support and resources (Elsbach, 2003).
2.5.2 Variables organization should consider when responding to a celebrity endorsement scandal
16
Murnighan, 2002). Therefore, this paper will analyse two different types of response strategies, two find the preferred response strategy of key stakeholder, the consumer.
By doing so, this paper will look at Benoit’s (1995) Image Repair Typologies {I.R.T.}, which illustrates the repair strategies an organization can employ after a crisis.
2.5.3 Image Repair Typologies
Throughout the past decades a range of different image repair strategies have been developed focussing on potential actions an entity can undertake in order to address an image in crisis context (Fink, 1986; Benoit, 1995; Hearit, 2001; Seeger, 2002; Seeger and Ulmer, 2002; Hearit, 2006; Fearn-Banks, 2007). This research addresses Benoit’s Image Repair Typologies (I.R.T) as well as the advocacy-accommodation strategy.
Benoit (1995) has constructed a classification of strategic action, which allows for traceable categorization of actions as well as interpreting the motives and action utilised in one of the I.R.T. strategies (Benoit and Nill,1998; Pfahl and Bates, 2008). The I.R.T. model of Benoit focuses on scrutinizing the rhetor’s role in formulating a statement with the focus on protecting and defining the reputation of the entity (Seeger and Ulmer, 2008).
17
reducing offensiveness, corrective action and mortification (Benoit, 1997).These five response strategies are summarized in the table below.
Table 2: Image Repair Typologies.
Image Repair Dimension Tactics Organization Employs
Denial Deny or shift blame to different cause than celebrity endorser.
Evasion of responsibilities No comment, or claim provocation
Reduce Offensiveness Organization minimizing the importance of the act by:
• making the act look less offensive • diminish offensiveness by comparing
it to more severe issues
• attacking accuser to shift attention • compensating the victim.
Corrective Action Organization communicates a plan that solves the problem. Cuts ties with celebrity endorser in order to prevent future problems arising.
Mortification Organization asks for forgiveness of endorsing the celebrity and cancels all ties with him.
18
endorser scandal, this would imply that the organization decides to either stand with the celebrity endorser during the public scandal or give in, make concessions and take action. The advocacy and accommodation are polar opposites; therefore, most companies utilize a response strategy that is somewhere in between the two extremes (Cancel et al, 1997).
However, past literature is contradictory with regards to which type of I.R.T. to utilise in order to minimise the damage (Cancel et al, 1997). Brown, Anderson and Dickhaus (2016) infer that there are no I.R. T’s that will create a significant more positive result over another type of strategy. Yet other literature suggests that different types of I.R.T. strategies do make a difference with regards to the consumer perception of the brand and celebrity. For instance, Sanderson (2008) illustrates that celebrity endorser Paul Clement did more harm than good when he utilised a mixture of denial and attacking the accuser I.R.T. strategies. The public must have a genuine believe that the athlete is sincerely trying to rebuild its image (Meng and Pan, 2013) unless the entity has credible evidence that the scandal did not occur, literature advises to steer clear from these two options. Indeed, Walsch and McAllister-Spooner (2011) illustrate that the I.R.T. strategies reduction of offensiveness, corrective action and mortification brings forth a more favourable perception from the public. Additionally, Arendt, Lafeche and Limperopulus (2017), show that corrective action is the most successful I.R.T. strategy, closely followed by reducing offensiveness. Therefore, this paper will analyse the I.R.T. typologies of reducing offensiveness and corrective action, as both dimensions are in between the two extremes of Benoit’s (1995) scenarios. Strategies to reduce the offensiveness of the celebrity endorser include bolstering the celebrity image, minimizing the seriousness of the act, differentiating the offense to similar yet more offensive scandals, displaying that the act committed by the celebrity served a greater purpose, attacking and blaming the accused party, and finally compensating the attacker (Brazeal, 2008).
19
involves putting the blame on a specific individual or group that is part of the organization. For example, by terminating the celebrity’s endorsement, the organization ensures it will not be damaged further by the celebrity (endorsement). The strategy which is the most suitable will be analyzed in the formulation research hypotheses below.
3. Formulating Research Hypotheses
3.1: Celebrity endorsement effect on purchasing intent
Previous literature illustrates mixed findings with regard to the effect of a celebrity endorser scandal on consumers purchasing intent (Money, et al, 2006; Edwards and Ferle, 2009; Thwaites, et al 2012)
A stream of literature suggests that when a celebrity endorser undergoes a public scandal, there is barely any negative effect on purchasing intent. In fact, Donaton (2002) proves that after the celebrity endured a public scandal, the sales of the product endorsed by the celebrity actually went up. Moreover, Pringle (2004) also illustrates that once the celebrity endorser has undergone a public scandal, the products he or she endorsed inadvertently increased in awareness and lead to higher purchasing intent of the consumers.
Yet, several other authors demonstrate that celebrity endorser scandals have led to unfavourable assessments of the purchasing intentions of the brands endorsed (Bailey 2007; White, et al, 2009). Till and Shimp (2008), show that when a celebrity receives negative press it will reflect badly on the commodity he or she endorses.
20
customer associates with the celebrity are affected. In turn, McCracken’s (1989) transfer model is adjusted, as the bundles of meanings which create the image of the celebrity have been changed: the characteristics of the celebrity added to the product or brand have also changed, which will eventually influence whether people decide to consume the product or not.
Therefore, based on the preceding literature review, the following hypotheses is created: H1: A celebrity endorser scandal will have a negative effect on consumer purchasing intent.
3.2 Match-Up
Previous literature has extensively analysed the influence of the Match-Up hypothesis on purchasing intention. Consistent with the Match-Up hypotheses, the validity of the celebrity endorser is the result of congruence between the product/brand and the endorser (Kahle and Holmer, 1985)
The advantages of congruent association between the celebrity endorser and the brand are extensively established in the literature. A high congruence association will allow the endorser to be more convincing (Kamins and Gupa, 1994) and generate a more positive opinion towards the advertisement (Misra and Beaty, 1990) as well as a better demeanour with brand attitude and finally purchasing intention.
Hence past literature suggests that the attractiveness and the expertise of the celebrity endorser and the functional intention of the product lead to a more favourable perception in consumer purchasing intention.
Therefore, based on the preceding discussion the following hypotheses is created:
21
3.3 Scandalous effect and congruency
Literature has been limited in analysing the influence of the Match-Up hypothesis when the celebrity endorser undergoes a public scandal and its resulting effect on purchasing intention (Money, et al 2006; Carrilat, d’Astou and Lazaure, 2013).
Nevertheless, Ohanian (1990) illustrates that when the congruence between the brand and the celebrity endorsement is low, the celebrity has mainly been selected because of his/her reputation. In turn, when the celebrity undergoes a public scandal, the central feature that ties the product and the endorser, the reputation, is damaged. Hence the resulting negative magnification is greater than it if the congruence is higher (Carrillat and Lazure, 2014).
In contrast, Thwaites et al (2012) infer that when congruence between the celebrity and the product is high, the negative information regarding the celebrity strongly influences the endorsed product and hence reduces purchasing intent. White et al (2009) also explain that the purchasing intention of consumers will be effected at a greater rate when the congruence between the product and the celebrity is high: because consumers associate the product with the celebrity, the scandal that the celebrity undergoes is associated with the product they endorse. Additionally, Carrillat and Lazure (2014) argue that negative spillover effects of a celebrity endorser undergoing a scandal are more easily transferred to the product they endorse when a high congruence exists.
Therefore, based on past research, the purchasing intentions will be affected more negatively when the congruence between the celebrity and the product is high, resulting in the following hypotheses:
22
3.4 Response strategies
The fourth hypothesis will measure the moderating influence of the organization response strategy on the relationship between the level of congruence between product and celebrity endorser and its effect on purchasing intent. It is believed that a corrective action strategy will yield more favourable purchasing intentions as opposed to reducing offensiveness strategy (Benoit and Drew, 1997). A potential explanation could be that a consumer prefers to avoid uncertainty when it comes to making a purchasing decision. Thus, corrective action signifies a clear stance of the organization: it disassociates itself from the celebrity. Hence, in this scenario, uncertainty decreases for the consumer, since the decision whether to trust the endorser is not necessary. Furthermore, when the organization parts company with the celebrity endorser, the organization illustrates that the act is not acceptable and cannot be tolerated. By separating the company from the endorser, the consumer can draw a clear line between what the organization represents and the activity the celebrity endorser undertook in the public scandal. In turn, when the company reduces the offensiveness of the act, it illustrates that what the celebrity endorser did was acceptable. Therefore, the following hypothesis is created.
H4a: Consumers purchasing intent will be affected more favourably when the organization utilizes a corrective action response strategy as opposed to a reducing offensiveness response strategy.
H4b: Consumers purchasing intent will be affected less favourably when the organization utilises a reducing offensiveness response strategy, as opposed to a correcting action response strategy.
3.5 Level of congruence and response strategies
The fifth hypotheses will measure the interaction effect of the level of congruence between the product and the celebrity and the response strategy, and how this relationship influences the purchasing intent of the consumer.
23
When the congruence between the celebrity endorser and the product is high, consumers will strongly associate the product with the celebrity. Hence if the celebrity endorser undergoes a public scandal, the negativity surrounding the celebrity will transfer to the product. This is mostly due to the associations which consumers draw more easily if high congruence exists (i.e. a sports celebrity endorsing products for a sports company). If there is no congruence between the celebrity and the organization (i.e. sports person endorsing a financial institution), this association may not be drawn, resulting in less pressure for the organization to separate itself from the celebrity endorser and rather opting for a strategy that reduces offensiveness and enables the company to keep the sponsorship going ((Erdogan, 1999; Till, et al, 2008).
Therefore, it would be better for the company to utilize a separation strategy {Corrective action} when the congruence between brand and celebrity is high, as purchasing intent of the consumers will react more favourably (Arendt, et al,2017).
H5: When the congruence between celebrity and product is high, a corrective response strategy will mitigate the negative effect on purchasing intent; the negative effect will be exacerbated, if the organization adopts a reducing offensiveness strategy.
3.6 Control variables:
The control variables utilized in this study are age, gender and education. Alsmadi (2006) illustrates that males are more influenced by celebrity endorsers advertisement. Additionally, Larson, Woloshin and Schwarzt, (2005) infer that higher educational attainment consumers are less impacted by a celebrity endorser advertising. Thus, both variables are fruitful to control for, to analyse how an endorser scandal, as well the brands response, influences gender and educational differences.
24
different generational effects are on purchasing intent. The four generations that are taken into account are, Generation X, Z, millennials and baby-boomers.
25
The hypothesis and the control variables lead to the following conceptual model: Figure 2: Conceptual model.
4. Methods
4.1 Overview
26
to low, medium and high congruency associations between brand/product and celebrity endorser was tested in the pilot study.
In order to test the moderator, the different response strategies of organizations are tested by analysing different organizational response strategies after the occurrence of the celebrity scandal, in the second experiment.
4.2 Pilot study
The main goal of the pilot study was to select an appropriate celebrity endorser and brand/product to endorse in the main experiment. Both the perceived fit between celebrity endorser and brand/product as weIl as additional variables related to familiarity and likability of celebrity and product/brand were taken into consideration. The pilot study utilized a mix of snowballing and convenience sampling.
During the study 59 people (49.2% male, 50.8% female, mean age 31) were randomly assigned to one of three groups (low, medium and high congruence). Based on the assigned group, respondents were presented with different matches of celebrities and brands and asked to evaluate the perceived fit between the respective celebrity and product/brand. The matches of different level of congruency grouping consisted of the same celebrities’ endorsers {Roger Federer, Lewis Hamilton, Usain Bolt, Max Verstappen, Zlatan Ibrahimovic}, however, the brands/products differed per congruency category. Lastly, questions related to familiarity and likability of the celebrity and product/brand were asked.
27
The high category was chosen instead of the medium since it resulted in a higher significance level and a higher average mean
Table 1: Pilot study’s congruency categories.
Congruency Level Brands Mean
Perceived fit score
Significance level
Low congruency Colgate, B.M.W, Ray-Ban,
Eastpak, Calvin Klein.
3.63 With Medium: 0.007** With High: 0.001*
Medium congruency Sony, I-phone, Red-bull, The North-Face, Rolex
4.49 With low: 0.007** With high: 0.786
High congruency Gym-Shark, Reebok, New-Balance, Nike, Adidas 4.67 With low: 0.001*** With medium: 0.786 *=p<0.01 **=P<0.05 ***=P<0.001
Among all the celebrities tested, Federer was the most likable (m=5.14, P<0.001), and familiar (m=5.46, p=0.01), hence as Roger Federer is the most popular, he is deemed the best hypothetical celebrity endorser (Carrilat, et all 2013). For comparison across the different congruency associations, please see table 1, in appendix A.
28
P<0.001), and was thus chosen as the high congruency manipulation. Additionally, Colgate scored the lowest perceived fit with Lewis Hamilton, and was thus chosen as the low congruency manipulation.1 In addition, familiarity with the brand {Colgate: M= 6.4; Adidas: M=6.05, and
attitude towards the brand (Colgate: M=4.95, Adidas: M= 5.84, were statistically different (P=0.00). As likeability and familiarity lead to a high purchasing intent (Kent and Kellaris, 2001), both variables will be controlled for in the main study. Based on these results, the combinations Rodger Federer with Adidas and Colgate were selected for the main study.
Lastly, the pilot study also asked questions pertaining to the best form of celebrity advertising for consumers. The results indicate that consumers are most influenced by online social media advertisements and are most familiar with Facebook. Hence, in the main experiment, the announcement of the celebrity endorsement between brand and celebrity will be shown as a Facebook advertisement, whilst the announcement on the scandal will be announced via a newspaper outlet, as demonstrated by Carrillat, et al (2013), as is illustrated in figure 1 on appendix A.
4.3 Experiment study
The variables were measured by performing two experimental designs. The first with a 2 (high congruence vs low congruence) x 2 (Scandal vs No scandal) experimental factorial design ANOVA, where the celebrity-brand congruence and the occurrence of the celebrity endorser scandal are manipulated (see table 2). The no scandal scenarios are analysed in order to measure the scandalous effect of the celebrity endorser scandal, whilst simultaneously testing different levels of congruency.
29
Table 2 - 2x2 Matrix of celebrity-firm congruence and the scandalous effect.
The second experimental design analyses a 2 (high congruence vs low congruence) X 2 (reducing offensiveness vs corrective action) experimental factorial design ANOVA, where the response strategy and the celebrity-firm congruence are manipulated. As the scandal has already occurred in this experimental design, there is no value in analysing the scandalous effect. Hence this study needs two experimental designs.
Table 3 - 2x2 Matrix of celebrity-firm congruence and response strategy.
30
Resulting in the following 4 main and 4 sub-scenarios:
(1) Scenario A: No scandal occurs. The respondents are confronted with a high congruency between brand and celebrity endorsement.
(2) Scenario B: The celebrity endorser undergoes a public scandal. The respondents are confronted with a high congruency between the brand and the celebrity endorser.
(a) Scenario B1: In case of high congruence between the celebrity and the firm, the company stands by the celebrity and reduces the offensiveness of the celebrity scandal in their official statement.
(b) Scenario B2: In case of high congruence between the celebrity and the firm, the company drops the celebrity, thus taking corrective action.
(3) Scenario C: No scandal occurs. The respondents are confronted with a low congruency between brand and celebrity endorsement.
(4) Scenario D: The celebrity endorser undergoes a public scandal. The responses are confronted with a low congruency among brand and the celebrity endorser.
(a) Scenario D1: In case of low congruence between the celebrity and the firm, the company stands by the celebrity and reduces the offensiveness of the celebrity scandal in their official statement.
(b) Scenario D2: In case of low congruence between the celebrity and the firm, the company drops the celebrity, thus taking corrective action.
The scenarios will be divided equally and randomly among the participants to ensure comparison among the scenarios.
4.4 Participants, design and Procedure
31
the survey. The remaining 298 participants consisted of 164 males (55.6%), 130 females (44.1%) and 1 person who did not identify with either of the genders (0.03%). When comparing age generations, 19.9% of the respondents represent generation Z (younger than 23), 34.7% of the respondents are millennials (23-38), 17.2% are from generation X (39-54), and lastly, 27.6% are baby boomers (55 years or older).
Respondents began the online questionnaire by being welcomed and receiving a quick introduction to the survey. Then the participants were asked about their demographics, including their education, gender and age. All respondents were distributed equally among the six different scenarios: scandal vs no scandal, high vs low congruence and corrective action vs reducing offensiveness response strategy scenarios, as is illustrated in table 4. By arbitrarily distributing the respondents to one of the six scenarios’, the survey allows for comparing and contrasting the resulting changes in purchasing behaviour (Hox, 2008; Webster, 2007). The table illustrates how many respondents were assigned to each scenario.
Table 4 – Overview of experimental design and respondents.
Scandalous effect Response strategy
32
Following each scenario, the respondents were asked additional control questions pertaining to their familiarity and attitude of the celebrity, familiarity with the brand, the brand and athlete congruence, and credibility of Roger Federer endorsing the product. Then participants were sked questions related to the purchasing intention of the product endorsed Roger Federer (Adidas or Colgate), using a 7 item Likert scales. Finally, all were respondents are thanked for their participation and reminded that all information would stay confidential.
4.4 Questions and Measures.
4.4.1 Independent variable: celebrity congruence
Celebrity-firm congruence is manipulated by presenting the participants with one of two scenarios. In one scenario, participants are told that Roger Federer endorses Adidas (see Appendix A, Figure 1). This scenario represents a high congruency between the celebrity and the product. On the other hand, half of the participants are presented with a scenario with Roger Federer and Colgate toothpaste, representing a low congruence association. In order to test the perceived congruence between the celebrity and the product, questions are taken from Gwinner and Eaton (1999), and adapted to suit the needs of this study (i.e.: “my image of Roger Federer is very dissimilar from the image I have of Colgate products”); with answers ranging from totally disagree/ totally agree on a 7-item Likert scale2.
A factor analysis is performed in order to test the validity of the congruency questions and can be seen in Table 2 in appendix A. The factor analysis illustrates that all variables can be retained, hence improving its validity.
2 All Likert scale questions were recoded to have following answer scale: 1: extremely unlikely → 7: extremely likely. For example, 1: extremely unlikely → 7: extremely likely. Please see appendix B for clear overview.
33
4.4.2 Moderator: measuring response strategies
The response strategy of the organization is manipulated by presenting the participants with one of two scenarios. In one scenario, participants are told that the organization has used the reducing offensiveness I.R.T. strategy, whilst in the other scenario, the participants are told that the organization has utilised the corrective action strategy. The participants are then asked how the different response strategy influences their purchasing intent. By analysing the different scenario’s, the researcher can compare how this variable influences the level of congruence between product/celebrity and its effect on purchasing intent.
4.4.3 Moderator: measuring scandalous effect
The scandalous effect of the celebrity endorser undergoing a public scandal is measured by splitting the participants into one of two groups. Either the respondents are shown the occurrence of the scandal or not. Hence, 50% of the respondents are put into the no-scandal scenario and the other 50% are put into the no scandal effect. Ultimately, the purchasing intent means of both groups are compared, in order to determine the scandalous effect.
4.4.4 Dependent variable: measuring purchasing intent:
Purchasing behaviour can be defined as the decision process and includes the act of being involved with buying and consuming products (Sharma, 2014). Purchasing intent is measured by asking respondents various questions regarding the likelihood of them purchasing the product endorsed by Roger Federer. The questions are asked using a 7-item Likert scale, taken from Spears and Singh (2004).
4.5 Data Collection and Sampling strategy
34
5: Analysis and results
5.1: Manipulation check
In order to check whether the congruence manipulation was effective, 298 respondents answered several questions pertaining to the congruency between Roger Federer and the product/brand. The five congruency questions share a Cronbach Alpha of 0.885 (α> 0.7) and are therefore able to be computed into a congruency sum variable. A one-way ANOVA test of group congruence and the group congruencesum variable (indicating which congruency scenario the respondents are put in), resulting in the table 4 in appendix A. The manipulation check questions illustrate that the respondents find the perceived fit among Adidas-Federer more favourable (M=5.16, SE=0.087), relative to the congruency among Colgate-Federer(M=3.56, SE= 0.08). Hence the priming of the scenarios is deemed a success (P=0.00).
5.2: Hypotheses testing
5.2.1 Experiment 1
35
0.77. However, as the variance explained by variable purchasing behaviour question 3 is more than 10% (O’Rouke and Hatcher, (2013) {PB3 explains 13.5% of total variance), this variable will be analysed separately in the analysis section. Next, an univariate analysis is utilised in order to demonstrate the results.
Table 5: Univariate analysis celebrity-firm congruence and the scandalous effect.
*=p<0.01 **=P<0.05 ***=P<0.001
Independent Variables F-value Standard
Deviation Mean Scandal 6.34 No Scandal 1.25 4.67** Scandal 1.2 4.97** Congruency 4.844 Adidas-Federer 1.27 4.68** Colgate-Federer 1.19 4.96** Scandal x Congruence 4.833
No Scandal & Adidas 1.29 4.40 **
No Scandal & Colgate 1.20 4.94**
Scandal & Adidas 1.20 4.72**
Scandal & Colgate
Control Variables:
1.18 4.98**
Age 4.79 1.1 2.53 **
Familiar with brand in adv. 0.02 1.44 5.49**
36
In H1, the scandalous effect is significant in determining to what extent purchasing behaviour is affected (F=6.34, P=0.012). Moreover, the level of congruency among Roger Federer and the brand he endorses (H2) also significantly impacts to what extent purchasing behaviour is affected (F=4.83, P=0.029). Additionally, the interaction (H3) effect between the level of congruency and the scandalous effect, also significantly impacts purchasing behaviour (F=4.91, P=0.029).
However, the results indicate that hypothesis 1 is significant, but not in the way hypotheses 1 predicts. Hypotheses 1 states that when Roger Federer undergoes a public scandal, the purchasing behaviour of consumers will be more negative, as opposed to when Roger Federer does not undergo a scandal. However, the results indicate that a scandalous effect has a positive effect on purchasing intent. In specific, the respondents who received the scenario in which Roger Federer underwent the doping scandal have a higher purchasing intention (M= 4.97; SE= 0.084), compared to the respondents who received the no scandal scenario (M=4.67 SE= 0.083). These results indicate that respondents are more likely to buy the product Roger Federer endorsers if Federer undergoes a doping scandal. Therefore, the results find statistical evidence against hypotheses 1.
37
Another surprising effect is that respondents illustrate a more favourable purchasing intention when they are supplied with a low congruence celebrity endorsement, versus a high congruent association. Statistical evidence is found a low congruence association scoring a more favourable purchasing intent as opposed to a high congruence association, thus this study finds evidence against hypotheses 2. This is demonstrated by respondents having a more favourable purchasing intent when they are shown the Colgate-Federer celebrity endorsement, with an average 4.96 purchasing intent (SE= 0.086). In turn, the celebrity endorsement with Adidas-Federer results in a statistically significant lower purchasing intention (m=4.68, SE= 0.087). Hence, the association between Colgate-Roger Federer yields a more positive purchasing behaviour intent, as opposed to the purchasing behaviour intent that is stimulated by the association of Federer-Adidas.
38
When only considering purchasing behaviour in question 3, the interaction effect infers results that were initially predicted. Namely, in a no scandal scenario, respondents’ purchasing behaviour is higher in the Adidas-Federer association (m=4.6 SE which Federer undergoes a doping scandal, the association between Colgate-Federer has a more favourable purchasing intent (M=3.8 SE=0.106), as opposed to the Adidas-Federer association (M=3.74 SE=0.105), as can be seen in table 5 in appendix A. Nonetheless, the significance value of this variable is P=0.123, and hence as p>0.1, no statistical evidence is found to support hypotheses 3.
5.2.2 Experiment 2
Additionally, another 2 (low congruency vs high congruency) X 2 (Corrective action response strategy vs Reducing offensiveness strategy) ANOVA between subject’s factorial design is performed, followed by an univariate analysis in order to demonstrate the results.
4 4,2 4,4 4,6 4,8 5 5,2 Colgate Adidas
Graph 2 : Level of Congruency vs Scandalous
Effect
39
Table 6: Univariate analysis celebrity-firm congruence and response strategy.
*=p<0.01 **=P<0.05 ***=P<0.001
In experiment 2, the respondents are all faced with a scenario in which Roger Federer undergoes a public scandal. This experiment analyses whether corrective action or reducing offensiveness I.R.T. strategy, adopted by the brand’s organization, can alter respondents purchasing intent regarding the product. Specifically, H4a predicts that a corrective action strategy of the organization results in a more favourable purchasing intention of respondents, as opposed to when the organization utilises a reducing offensiveness strategy. Additionally, H5 claims that in a high congruency association between the celebrity endorser and the brand, when the
Independent Variables F Value Standard
Deviation Mean Congruence 0.07 Adidas 1.20 4.89 Colgate 1.18 4.94 Response Strategy 0.84 Corrective Action 1.21 5.0 Reducing Offensiveness 1.18 4.83
Response Strategy x Congruence 0.39
Corrective Action & Adidas 1.26 5
Corrective Action & Colgate 1.09 4.9
Reducing Offensiveness & Adidas 1.1 4.8
Reducing Offensiveness & Colgate
Control Variables:
1.27 4.86
Age 4.5 1.1 2.53 **
Familiar with brand in advertisement 5.1 1.44 5.49**
40
organization adopts a corrective action strategy, the negative effect on purchasing intent will be mitigated. In turn, the negative effect will be amplified if the organization adopts a reducing offensiveness strategy.
Based on the results there is no statistical evidence that supports that consumers purchasing intent is significantly better when respondents receive a corrective action strategy, relative to a reducing offensiveness strategy. Respondents’ purchasing intent is not significantly influenced by the different response strategies the organization utilises (F=0.846, P=0.359), hence the findings offer no evidence in favour of H4a and h4b. Nonetheless, if the organization adopts a corrective action strategy, the purchasing intent of respondents does result in a more favourable outcome. This is inferred by the average purchasing intent scoring 5 (SE= 0.123) if the respondents are shown the corrective action strategy, whereas the average purchasing intent of the reducing offensiveness strategy scored an average of 4.83 (SE=0.12). This illustrates that a corrective action strategy does result in a more favourable purchasing intent of consumers, however, the difference is not enough to be statistically significant.
41
6. Discussion
Overall, the findings did not support the initial predictions, yet the survey results do indicate potential significant practical and theoretical implications.
First of all, the results of the survey go against the intuitive notion that a celebrity endorsement scandal results in a less favourable purchasing intent (Till and Shimp, 2008; Edwards and Ferle, 2009; Klebba and Unger, 1982). Contrary to expectation, the purchasing intent of respondents who faced with a celebrity doping scandal is significantly more positive relative to the no scandal scenario respondents. Numerous other researchers have also found that a celebrity endorsement scandal increases the purchasing behaviour of consumers. In specific, Money, et al, (2006) find that their participants who received no negative information about the celebrity endorser resulted into weaker purchasing intention, relative to the participants who did receive negative information regarding the celebrity endorser. According to Money, et al, (2006), this result is due to how people make their causal attributions. Specifically, it is suggested that participants attribute the celebrity endorser drug use to a situational cause, and not due to dispositional deficiency of the celebrity. The former refers to the cause of behaviour coming from an internal characteristic of an individual, whilst the latter refers to the blame being assigned to a cause that is outside the individuals' control (Fiske, 1991).
Hence it is argued that consumers believe that drug use among celebrities is not due to the personal shortcoming of the celebrity, but rather due to environmental pressures, such as group norms, the pressure to perform and other situational pressures, which in ordinary circumstances, would be viewed as inappropriate (Money, Et all 2006). According to this logic, such situational pressure could effectively shield the celebrity from negative, dispositional attribution, that most ordinary people would receive.
42
that any publicity is good publicity, hence the negative news of a celebrity doping scandal, also leads to positive publicity.
Second, the findings of the survey also show an adverse effect of what the Match-Up predicts in a no scandal scenario (H2). Surprisingly, the respondents illustrated a more favourable purchasing intention towards the low congruent association between Colgate-Federer, as opposed to the high congruent association between Adidas-Federer. Thus, the paper’s findings are in disagreement with Kamins and Gupa (1994) who illustrate that a high congruency among brand and fit should enable the celebrity endorser to be more convincing, create more positive opinions towards the advertisement (Misra and Beaty, 1990), and ultimately create a superior attitude towards the brand and greater purchasing intent. However, these results are in line with Lynch and Schuler (1994) who theorise that an incongruent association between product/brand and the celebrity endorser may induce consumers to pay more attention to the message. Similar to this study, Lee and Thorson (2008) illustrated that a moderately incongruent celebrity endorser association produced higher purchasing behaviour than a high congruent association. Both Lee and Thorson (2008) and Lynch and Schuler (1994) utilise the schema theory framework in order to explain why the Match-Up hypotheses works in the opposite way it predicts. (Torn, 2012). Social Schema theory stems from cognitive psychology and demonstrates how concepts or ideas are characterized in an individual’s brain. Accordingly, we create schemas by bringing information together which is related to the original concept of the association (Torn, 2012).
43
1983). As a result, the Colgate-Federer association has created more conflict into the consumer’s mind, and the consumer is more driven to process the celebrity endorser communication, in order to diminish the conflict with the pre-existing brand schemas.
The schema congruity theory is also an explanation for hypotheses 3. In fact, when a scandal occurs, the association between Colgate-Federer scores a moderately significant higher purchasing intent score, relative to the association between Adidas-Federer. In turn, this is in agreement with what H3 claims, as H3 claims that when the doping scandal is announced, the Adidas-Federer celebrity endorsement will stimulate a more adverse effect on purchasing intent, relative to the Colgate-Federer stimulated purchasing intent (Thwaites et al,2012; Carrillat and d’Astous,2014). As the schema congruity theory specifies, an incongruent celebrity-brand association will indirectly force the consumers into deeper cognition (Torn, 2012). More thought and time are spent in analysing the endorsement, leading to more favourable purchasing intent. Therefore, in the no scandal scenario, the purchasing intent is significantly higher for the association of Colgate-Federer as opposed to the purchasing intent for the association Adidas-Federer. In the scandal scenario’s, consumers could be accustomed to a high congruent celebrity-brand association engaging in a scandal, thus a pre-existing schema is already in consumer’s mind. In contrast, the consumer could be unfamiliar with or not be as familiar with a scandal occurrence in an incongruent celebrity-brand association, hence again deeper cognition is required by the consumer, resulting in more favourable purchasing intent.
Hence, a low congruent association will force people to pay more attention and indirectly encourage them to think about the information provided, ensuing deeper cognition (Fiske, Kinder and Larler, 1983). Additionally, previous literature indicates that negative spill-over effects of a celebrity endorser undergoing a scandal, is more easily transferred to the product they endorse when a high congruency exists (White, et all, 2009). Therefore, the greater negative spill-over effect may also be an explanation of the lower purchasing intent score in the high congruence association.
44
However, if purchasing behaviour question 3 {PB3} is analysed, the conclusion of the hypothesis change. Namely, the data infers that purchasing behaviour is affected, more positively when the celebrity endorser does not undergo a scandal, conforming H1. Specifically, when Federer undergoes a doping scandal, the purchasing behaviour for the product, he endorses is significantly lower, relative to the no scandal scenario. This finding is in line with Thwaites, et all (2012), who finds that any negative information about the celebrity endorser will reflect badly on the brand being endorsed, hence purchasing intent will decrease. Similarly, Klebba and Unger (1982) found that negative celebrity endorser information tarnishes the perception of the celebrity as well as the product. PB3 asks: “Imagine you just purchased a product from this brand, how would such information affect your purchasing behaviour?" This question indicates that respondents usually buy products from this brand. The concept of loyalty may play a part. Past empirical studies illustrate that brand loyalty directly influences purchasing intention (Mittal, Ross, and Baldasare, 1998; Hennig-Thurau, Gwinner, and Gremler, 2002). The average mean of pb3 is also significantly lower than the Purchasingsum mean, whilst the no scandal scenario is significantly higher. This may indicate that loyal customers may be affected significantly more severely by the scandalous news of the celebrity endorser.
Additionally, the interaction effect between the level of congruence and scandalous effect also finds evidence in support of hypotheses 3. Namely the scandal scenario respondents purchasing behaviour is higher in the Colgate-Federer association, relative to the Adidas-Federer association. This finding agrees with Thwaites et al (2012) and White et a (2009) who illustrate that purchasing intention will be affected at a greater rate when the congruence between the product and the celebrity is high. However, as the P>0.1, the results of the variable are not significant, thus pb3 also finds not statistical evidence in support of H3.
45
celebrity endorser scandal has on the purchasing intent of the public (Benoit and Drew, 1997; Arendt, Lafeche and Limperopulus,2017). This study also finds that when the organization adopts a corrective action strategy, the purchasing intent of respondents does result in a more favourable outcome. In turn, this demonstrates that a corrective action strategy results in a more favourable purchasing intent. However, the results indicate that there is not a significant difference regarding which I.R.T. the brand utilises in order to minimise the damage done on the consumer's purchasing behaviour. Hence, the findings of this study suggest that it does not matter if the brand utilises a corrective action or reducing offensiveness strategy as the resulting purchasing behaviour is very similar. The findings of this study are in line with Anderson, Brown and Dickhaus (2016), who also illustrate that all I.R.T strategies yield a similar result. Anderson, et all (2016) give three possible explanations why different I.R.T. strategies do not yield a statistically different result in purchasing intent of the consumers.
46
Lastly, H5 infers that in a high congruent celebrity endorser association, the corrective action response strategy will mitigate the negative effect on purchasing intent, whilst the negative effect will be exacerbated, if the brand adopts a reducing offensiveness strategy. This is predicted as the literature illustrates that a corrective action strategy will mitigate the negative effect of purchasing intent after the celebrity endorser undergoes a public scandal (Walsch and McAllister-Spooner, 2011). However, the study's result demonstrates that the different repair strategies also do not make a significant difference when there is a high or low congruent situation.
Although the corrective action strategy scores a more favourable purchasing intent in both Colgate-Federer and Adidas-Federer associations, it is not significantly different. A potential explanation is inferred from Dardis and Haigh (2006), who empirically tested Benoit’s 5 image restoration strategies against each other and demonstratedthat the reducing offensiveness strategy is the only I.R.T. which created a positive statistically significant difference. This is a potential explanation of why the I.R.T are not statistically different, in the results of this study, as the reducing offensiveness strategy scores more favourably than expected. Literature assumes that a more advocacy position creates a more favourable reaction towards the company (Croombs and Schmidt, 2000), this study, however, indicates differently. The reducing offensiveness strategy, which is positioned in the middle of the continuum {of advocacy vs accommodative}, may be more adequate in enhancing stakeholder’s perspective of the brand. Simultaneously, the reducing offensiveness strategy enables the organization to avoid taking responsibility for the scandal, thereby avoiding potential lawsuits, and hence deflect any threats that harm the brand's credibility (Croombs and Schmidt,2000).
Both experiments illustrate that age, familiarity with the brand in advertising and likability of the brand, are all significant control variables in this study.
47
that younger generations are less purchasing behaviour is less severely affected, relative to older generations.
Table 7: Generational different influences P.I.
Generation categories Purchasing intent
Generations Z 4.97
Millennials 4.85
Generation X 4.89
48
7. Managerial implications
The implications of this study have resulted in intriguing managerial implications. Firstly, the results signify that it is more favourable for a brand to choose an incongruent celebrity endorser. An incongruent association indirectly forces the consumer to think more about the association, ensuring that the brand/product stays in the mind of the consumer, enabling higher PI. Secondly, it is also more beneficial for the brand when the celebrity endorser undergoes a doping scandal. In turn, this could imply that if a brand is looking for a celebrity endorser, the organization may be better off to choose a celebrity endorser who has undergone previous public scandals. Additionally, the study also indicates that both the I.R.T. strategies lead to a favourable assessment of the consumers purchasing intent. Yet, although corrective action does result into a slightly more favourable outcome, I would advise a brand to utilize the reducing offensiveness strategy. This study highlights that in a crisis situation in which the organizational responsibly is not obvious, a reducing offensiveness strategy could yield into a more statisfacotry long run strategy. As this strategy will enable the brand to not take any responsibilities in the occurrence of the scandal, killing any potential lawsuits before they could arise. Additionally, the brand avoids the trouble of having to terminate the contract with the celebrity endorser and find a new one. In turn, saving the organization a time and money.
Additionally, PB3 also illustrates that loyal consumers might be affected significantly more severely, by the occurrence of a scandal, resulting in a sharp decline in purchasing intent for the brand. Hence, specific response strategies may be applicable to loyal consumers.
