Amsterdam Business School, University of Amsterdam
Master Thesis in Business Administration -- Digital Business track
Celebrity eWoM or traditional endorsement? The impact
of advertisement types on short-term and long-term brand
recall: an experiment with social distance as mediator
(2016-2017)
Supervisor: Dr. Ruben de Bliek
Pujun Li
Student number: 11360801
Statement of originality
This document is written by Student Pujun Li who declares to take full responsibility for the contents of this document.
I declare that the text and the work presented in this document is original and that no sources other than those mentioned in the text and its references have been used in creating it. The Faculty of Economics and Business is responsible solely for the supervision of completion of the work, not for the contents.
Table of contents
Statement of originality ... 2
Table of contents ... 3
List of Figures and Tables ... 4
Abstract ... 5
Introduction ... 6
Literature review ... 9
Brand recall ... 9
Celebrity endorsement... 10
eWoM (Electronic word-of-mouth) ... 12
Social Distance ... 14
Conceptual Modal & Research Hypotheses ... 15
Data and Methodology ... 18
Stimulus ... 18
Operationalization ... 20
Procedure ... 22
Sample ... 24
Analysis and Results ... 26
Examine whether the celebrities are well-known ... 26
Examine whether the brands are not well-known ... 26
Reliability Analysis ... 27
Hypothesis Testing ... 27
Discussion and Implications ... 37
Limitations and Future Research ... 40
References ... 42
List of Figures and Tables
Table 1 Advertisement Types ... 20
Table 2 Paired Samples Statistics of Short-term Brand Recall ... 29
Table 3 Paired Samples Statistics of Long-term Brand Recall ... 31
Table 4 Paired Samples Statistics-Affinity ... 33
Table 5 Paired Samples Statistics-Connection ... 33
Table 6 Paired Samples Statistics of Social Distance ... 34
Table 7 The Regression Tables of the Mediation Analysis ... 35
Table 8 The Regression Tables of the Mediation Analysis ... 37
Figure 1 Concept Model ... 18
Figure 2 Four Traditional Celebrity Endorsement Advertisements ... 21
Figure 3 Four eWoM Advertisements ... 21
Figure 4 Age Distribution of Participants- stage 1 ... 25
Figure 5 Age Distribution of Participants- stage 2 ... 26
Figure 6 Mean scores of short-term brand recall ... 28
Figure 7 Mean Scores of Long-term Brand Recall ... 30
Figure 8 Mean scores of affinity and connection... 32
Abstract
The purpose of this research was to examine the effect of two different advertisement types on short-term and long-term brand recall which are the foundations of consumer loyalty and whether there is mediation effect of social distance in these relationships. The two advertisement types are celebrity eWoM on social media and traditional celebrity endorsement respectively. To test the hypotheses, a 2 (ad types) x 4 (brands-celebrities) between-subject and within-subject mixed lab experiment was conducted among undergraduate students with the help of research software Qualtrics. Results showed that celebrity eWoM on Twitter generates both higher short-term brand recall and long-term brand recall than traditional celebrity endorsement advertisement (short-term and long-term brand recall of eWoM are significantly higher than short-term and long-term brand recall of traditional celebrity endorsement). Although perceived social distance is significantly lower towards celebrity followed on Twitter than a celebrity in the traditional advertisement, there is no significant mediation effect of social distance in the relationships. Social distance is not a mediator during this process. This paper has filled up the research gap of the effectiveness of these two advertisement types on brand recall and mediation effect of social distance. What’s more, this paper also provides practical implications for businesses and implications for future research.
Introduction
It is quite common for companies to have celebrities in their advertisements, in almost all the industries including NGOs, political campaign and health propaganda (Knoll and Matthes 2017). It has been a popular marketing strategy (Erdogan 1999). One of the most successful celebrity endorsement examples could be Nike and Michael Jordan. Since 1984, basketball shoes endorsed by Jordan have occupied 75% of the basketball shoes market and nearly 10% of shoe market in America (Robert 2013). Consumers link the brand image and this famous basketball player together which are congruent perfectly with each other. Jordan brand, producing the best basketball shoes with the highest quality involved the basketball player’s design, represents Michael Jordan himself-- the greatest basketball player of all time with numerous accomplishments. Consumers are purchasing not only the shoes, but also the passion towards basketball, love and respect to Michael Jordan. Celebrity endorsements are so effective that even when companies just release the news of hiring famous athletes as endorsers, the average stock value of 341 sample companies rises 0.23% and the weekly sales of endorsed product increase by 4% in average (Elberse and Verleun 2012).
Except for celebrity endorsement, another form of celebrity involvement in marketing is emerging-- celebrity eWoM. That is celebrities share their opinions about a product or brand on social media intentionally or unconsciously. This can have huge impact on consumer’s attitude towards the brand (Lee and Youn 2009), brand awareness and brand recall (Friedman and Friedman 1979), and sales (Amblee and Bui 2011) in a very short time, because of the fast speed of the Internet to reach a large amount of consumers (Hung and Li 2007). For example, an unplanned selfie photo taken by Ellen Degeneres in 2014’s Oscar ceremony was estimated to bring $800 million to $1 billion value to Samsung (Vranica 2014). During the ceremony, she took out a Samsung Galaxy Note to do a selfie with a group of splashy superstars and that photo went viral on Twitter immediately with 3.2 million
retweets and 2.3 million likes, exposing the white Galaxy Note 3 in front of millions of viewers. Similarly, a photo posted by Taylor Swift on Instagram increased the exposure of a ring brand-- Dana Rebecca dramatically with 1.4 million likes on this photo (Mindy 2016). An unintentional mention of a brand by a celebrity on social media can have great impact on brand and make this brand noticeable and memorable to their followers.
One of the differences between the two kinds of celebrity participation above is the social distance between consumers and celebrities. Social distance is the perceived closeness and familiarity between one person and another person (Magee and Smith 2013). The more familiar between two individuals, the more affinity they feel to each other, the lower the social distance between them. In other words, the more people know each other, and the more people interact with each other, the lower the social distance. And lower social distance can generate higher trust (Binzel and Fehr 2013). Therefore, people are more willing to accept messages from people towards whom they perceive lower social distance.
Usually, celebrity is unreachable and untouchable (Warner 2011). Celebrity is well-known to be famous (Elliot 1998) in any field including the entertainment industry, sports, politics, business, religion, medicine, and science (McCutcheon et al. 2002). Before the advent of the Internet, people rarely have the opportunity to interact with a celebrity or know a celebrity’s personal life. And people usually feel psychological distant towards celebrity. So social distance between ordinary people and celebrities is relatively high.
However, the development of the Internet has changed the relationship between people and celebrities. Most people in consumers’ social media are acquaintances, since they will frequently interact with each other (Zajonc 1968), thus be more familiar (Lawler and Yoon 1996), the social distance between them is low. Once a user follows or likes a celebrity
Then the follower is able to track the celebrity, from his/her commercial activities, recent works to personal lives. Therefore, I expect that the more the follower knows about the celebrity, the more familiar the celebrity will become to the follower, the higher affinity the follower will feel to the celebrity, resulting in a lower perceived social distance. Thus, the social distance between consumer and celebrity in social media is lower than the social distance between consumer and celebrity in a traditional advertisement.
Many companies have been using celebrity to prompt their products because this strategy can enhance brand recall, brand awareness (Friedman and Friedman 1979) and purchase intention thus sales (Hakimi et al. 2011). Many researchers have proved that brand recall is not only the prerequisite of final purchase behavior (Jacoby et al. 1977; Park and Lessig 1981; Nedungadi 1990; Chakravarti et al. 2003) but also the foundation of achieving brand resonance, the top of customer-based brand equity pyramid (Keller 1993). Before consumers make the purchase decision, they usually already have several brands in mind. And this memory retrieval could be the decisive factor of buying decision when there is no other stimulus such as promotions (Bettman and Park 1980). Therefore, it is vital that whether a consumer can think of a brand during the purchase process. In other words, it is essential that whether a brand can pop up in a consumer’s mind. So it is interesting to see which one of these two different forms of advertisements can create higher brand recall which is an important premise for business success.
Overall, celebrity eWoM through social media and celebrity endorsement in an advertisement are both effective sales generators. Usually, consumers will feel closer to celebrity in social media than those in an advertisement. Companies can choose either of them to prompt their products or service. However, it is still uncertain which one can create higher brand recall, thus lead to more purchase behavior and bring benefits to a company. And inviting a celebrity to prompt brands is always expensive. So it is necessary to choose
the most effective one. However, there is no existing research focusing on the different effects of these two kinds of celebrity involvements. There is no literature studying the mediator impact of social distance in this procedure. To fill this gap, based on the theory of social distance, this paper is going to examine which one will generate higher brand recall, celebrity eWoM on social media or celebrity endorsement in traditional advertisement and the mediation effect of social distance.
This study tries to answer the following research questions by conducting a 2 (ad types)x4 (brand-celebrity combinations) lab experiment:
1. Are there differences in the impact on short-term and long-term brand recall between traditional celebrity endorsement and celebrity eWoM in social media?
2. Is perceived social distance the mediator of the effect of advertisement types on consumer’s short-term and long-term brand recall?
3. Is social distance lower towards celebrity on Twitter than celebrity in a traditional advertisement?
To answer the preceding questions, the rest of this paper will present literature review, raise conceptual models and hypotheses, develop the methodology, show results, advise implications, report limitations and directions of future research.
Literature review Brand recall
Brand recall is very important to a company, since it is an effective predictor of marketing success. According to Keller (1993), brand recall is to what extent can a consumer
recall is the prerequisite of brand equity which will drive the consumer to purchase, to achieve brand attachment and brand loyalty. That is, better brand recall enhances consumer’s buying intention (Nedungadi, 1990). Higher brand recall increases the possibility of a brand to be in a consumer’s purchase consideration, thus enhance the chance to be purchased (Chakravarti et al., 2003). It has been shown that consumers are more likely to buy something they are familiar with or to choose famous brands (Jacoby et al., 1977). Hence, brand recall will contribute to purchase decision when consumer need to make a choice among different products or brands (Jacoby et al., 1977; Park & Lessig, 1981). It can even be the determining factor of a purchase decision, especially when consumers are in low involvement environment where consumers can make quick purchase decisions that are relatively routine, e.g. bread, shampoo, socks, etc. (Bettman and Park, 1980).
Brand recall also contributes to brand image construction. By rooting the existence of brand in consumer’s memory, it helps to create psychological nodes which associate various information with the brand, such as brand characteristics, images, and name. (Keller, 1993).
Therefore, brand recall is an important indicator for companies to judge marketing effectiveness and success. With the increasingly fierce business competition, it is essential for brands to be memorized. And both celebrity endorsement and celebrity eWoM are proven to be effective but expensive facilitators of brand recall. Thus, it is necessary to examine which one is more effective, which one can generate higher brand recall, celebrity endorsement or celebrity eWoM on social media.
Celebrity endorsement
According to McCracken (1989), celebrity endorsement is a form of marketing strategy where a famous person makes use of his/her public recognition on behalf of a
consumer good by showing up within an advertisement. Using famous people to prompt products is not a recent phenomenon. Back to 1760s, the owner of the company Wedgwood which produced pottery and chinaware—Josiah Wedgwood used royal endorsements to increase his brand’s awareness (Erdogan BZ 1999).
Many studies have examined the benefits of celebrity endorsement in advertising. Having celebrity in an advertisement will influence consumers’ feeling towards endorsed product (Miller and Allen 2012), such as higher perceived quality of the product (Ohanian 1991), perceived information quality and lower perceived risk of purchasing (Biswas et al. 2006). Celebrity endorsement generates more positive attitude to the brand (Till and Busler 2000), and higher buying intentions and intention-to-use (Ohanian 1991). It can also facilitate higher brand and message recall (Friedman and Friedman 1979).
Famous people themselves can have an impact on the effectiveness of the endorsement. Consumers expect an endorser’s figure, image and characteristics would be in line with the product or brand he/she endorsed (Knoll and Matthes 2017). So consumers tend to believe that an endorsed product has similar attributes as the celebrity has (Miller and Allen 2012). In other words, endorser’s perceived traits such as healthy can be transferred to the perceived traits of brand or product (Langmeyer and Walker 1991). Thus celebrities with attractive and charming appearance can make the product more appealing and then increase brand image and consumer’s purchase intention (Hakimi et al. 2011). Inner characteristics of celebrity also contribute to advertising effectiveness (Tzoumaka et al. 2016). Trustworthy celebrities with relevant knowledge, experience or capabilities are able to convince the consumer that they are unbiased (Ohanian 1990). What’s more, if an endorser is perceived experienced and knowledgeable in the endorsed product area, the endorsement will enhance consumer’s purchase intention (Ohanian 1991). Thus companies tend to select celebrities
who are reliable, professional in certain filed or attractive, while their image is congruent with product characteristic or brand image (Hakimi et al. 2011).
There are many other researchers focus on other issues such as celebrity himself as a mediating factor, demographic differences, and product categories. Spry et al. (2011) found that brand endorser himself can be a mediating factor of the positive linkage between celebrity’s trustfulness and consumer’s perceived brand value. Klaus and Bailey (2008) examined endorsement effectiveness based on gender differences and indicated that celebrity endorsement is more effective on female customers, especially female endorsers. Tzoumaka et al. (2016) focused on perceived celebrity’s quality and consumer’s characteristics on endorsement effectiveness. It is supported by Friedman and Friedman (1979) that when the endorsed product is related to taste and flavor, celebrity endorsement can create higher purchase intention.
eWoM (Electronic word-of-mouth)
People are exposed to thousands of messages and advertisements every day, and it is impossible to absorb all the information. Chances are, consumers do see the advertisements which company has spent millions of money on, but they remember nothing about the target brand, product or service. The investment company made has little effect on their goal. However, WoM (word of mouth) can help a company to stand out from millions of information. It has shown that when the content comes from people’s network, message acceptance is higher which means people can absorb the information better (Roggeveen 2012). In other words, better brand recall. And the content from the network is known for Word of Mouth. In the era of the Internet, eWoM (electronic word of mouth) has been the mainstream.
According to Henning-Thurau et al. (2004), electronic word-of-mouth (eWOM) is defined as any unaided positive or negative brand/product/company related message or information made by consumers (former, actual, potential), which can be seen by other people via the Internet. Different from traditional WoM (informal, person to person information exchange about a brand, product, organization or service), eWoM is faster, more convenient, more reachable and eWoM reduces the pressure of face-to-face communication (Cui et al. 2012).
Considerable researchers have studied the impact of eWoM in business domain, including perceived popularity of brand or product (Park and Lee 2009), perceived product quality (Lee and Lee 2009), expected cost, thought and interest of product (Bickart and Schindler 2001), buying intention (Park et al. 2007; Lee and Youn 2009), trust (Awad and Ragowsky 2008; Liu 2006), brand choice and purchase decision (Senecal and Nantel 2004; Gupta and Harris 2010). Many researchers focus on the differences between positive and negative eWoM. Chevalier and Mayzlin (2006) indicated that positive eWoM increases sales, while the negative one is more powerful and influential on consumer purchase behavior (East et al. 2008), twice the effect of positive WoM (Arndt 1967).
It has been examined that when the information of eWoM is perceived trustworthy and credible, consumers are more likely to make a decision based on that eWoM (Fan and Miao 2012). Since people in a same social network know each other to some extent, perceived content reliability is higher in social media platform than other forms of Internet platform. Several researchers have studied the effect of eWoM in this special channel. Coulter and Roggeveen (2012) examined the effectiveness of eWoM on the online social network, regarding the origin of content, information, and network by tracking number of
user’s network. What’s more, the more influential the source of message is, the more effective the message have on other’s behavior (Trusov et al. 2010). Since celebrities are always more influential than ordinary people, at the same time social media facilitates spread of eWoM (McCarthy et al. 2014), their recommendations on social media platform have a positive effect on brand recall.
Social Distance
One of the differences between the two kinds of celebrity participation above is the social distance between consumers and celebrities. Social distance can be defined as the psychological feeling of how far from another person or other people that is perceived closeness between one person and another person (Magee and Smith 2013). The more familiar between two people, the higher degree of perceived affinity to each other, the lower the social distance between them. In other words, the more people know each other, and the more people interact with each other, the lower the social distance.
Over the past years, various influencers of social distance have been examined by previous psychological research, such as similarity between two individuals (Stephan et al. 2011) or differences between one and another (Meirick 2005), congruence of an individual’s identity and group’s (Kramer and Brewer 1984), interaction quality and frequency (Zajonc 1968) and the familiarity and interdependence generated from repeated interaction (Lawler and Yoon 1996).
Magee and Smith (2013) summarized the measurements of social distance and divided them into three categories: affiliation motivation, feeling close, and behaving close. Researchers developed many scales, such as interpersonal orientation scale (Hill 1987), perceived interpersonal closeness scale (Popovic et al. 2003), relationship closeness
inventory- strength subscale and relationship closeness inventory- frequency and diversity subscales (Berscheid et al. 1989).
Physical distance can have an impact on social distance. Schiffenbauer and Schiavo (1976) have shown that residents who live close to each other are more likely to develop friendship than people who live further, because neighbors have more opportunities to see, interact and exchange information with each other, thus are more likely to be familiar and interdependent. Before the advent of the Internet, it is rare for people to get in touch with celebrities. So it is impossible for ordinary people to know celebrity’s personal lives, communicate with celebrities. Celebrities are reachless. And people know little about them. Thus the social distance between people and celebrities is quite high. However, with the development of technology, physical distance won’t be a problem anymore. People can interact with each other through the Internet, check status and lives of someone through social media. Sometimes, people can interact with celebrities on the Internet. Nowadays, many celebrities share their personal lives, communicate with fans on social media. Celebrities have become part of people’s “network”. Thus perceived social distance is lower.
Conceptual Modal & Research Hypotheses
Previous research has shown the positive impact of celebrity involvement in marketing on brand recall (Friedman and Friedman 1979), because consumers think the celebrity is trustworthy and the information provided is unbiased (Ohanian 1990). Although effective, celebrity endorsement is also an expensive marketing tool. Therefore, it is necessary to know whether and to what extent can this tool achieve its goal of positively influencing brand recall.
People usually follow or like celebrities to trace their recent development or activities on social media platforms such as Facebook or Twitter and celebrities sometimes interact with people who have followed or liked him/her by sharing their feelings, personal lives and photos or videos. For example, Taylor Swift has 84.1M followers on Twitter (data viewed in April 2017). She shares funny videos of dogs, her commercial events, links of songs, and retweets, likes and comments people’s shares. These communications among social media help to improve information exchange between followers and celebrities. Through these accumulating interactions, celebrities and followers are becoming closer and more familiar with each other (Newcomb 1978), thus lower social distance, at least from the followers’ perspective. A follower will feel like the celebrity he following is kind of “his friend”, because they do exactly what friends will do—share each others’ lives. What’s more, because of the convenience of the Internet, “physical distance” between consumer and celebrity have been shorted. Namely consumers don’t have to travel a long way to see celebrity in person, they can watch them on TV, PC or mobile phone easily. Lower physical distance can enhance the probability of communicating, in other words, develop a friendship, to be closer.
Thus, this paper suggests that:
H1: Perceived social distance is lower towards celebrity followed on social media than celebrity in the traditional advertisement.
Most previous researchers only focus on short time brand recall, for example, Nedungadi (1990) measured brand recall immediately after the experiment without any time difference. However, if the sample of brands is small enough, for example only testing one brand, people can usually recall the brand name correctly right after seeing it. According to Ebbinghaus Forgetting Curve, people’s memory retention decrease over time when put no
effort to retain it. After 24 hours, people can only recall 33.7% of the information, and after 48 hours, only 27.8% remains. Usually, there will be a time lag between when people see the advertisement or brand endorsement and purchase behavior. So it is essential for companies that consumers are able to recall the advertised brand when they are going to buy something in that product categories. Since brand recall is so important to companies, it will be interesting also to measure long-term recall, to see whether there is a different impact on both short-term recall and long-term recall.
One of the differences between the two kinds of celebrity participation above is the social distance between consumers and celebrities. Social distance is the perceived closeness and familiarity between one person and another person (Magee and Smith 2013). The more familiar between two individuals, the more affinity they feel to each other, the lower the social distance between them. In other words, the more people know each other, and the more people interact with each other, the lower the social distance. And lower social distance can generate higher trust (Binzel and Fehr 2013). Therefore, people are more willing to accept messages from their network. Roggeveen (2012) has shown that message acceptance is higher when the message comes from closer affinity. In other words, people are more willing to acquiring information from their network or people they know. And they will pay more attention to the information from “friends”. This kind of information is known as WoM, and eWoM in the digital environment. Because of absorbing information better, people have a longer memory of the content, thus better long-term recall.
So this paper raises the following hypotheses:
H2: eWoM advertisement on Twitter generates higher short-term brand recall than traditional celebrity endorsement advertisement.
H3: eWoM advertisement on Twitter generates higher long-term brand recall than traditional celebrity endorsement advertisement.
H4: There is mediation effect of social distance on short-term brand recall. H5: There is mediation effect of social distance on long-term brand recall.
In conclusion, this paper suggests the following concept model.
Figure 1 Concept Model
Data and Methodology
The purpose of this study is to research the effect of advertisement types on brand recall and whether there is mediation effect of social distance as mediator. A 2 (ad types)x 4 (brands-celebrities) between-subject and within-subject mixed lab experiment is conducted.
Stimulus
To eliminate the effect of the well-known brand that everyone knows it so well that people have remembered the name before and can easily recall the name (Nedungadi 1990), this study picks up four less well-known brands. And Biswas et al. (2006) have shown that it is difficult to affect knowledgeable consumers via advertisements with celebrity involvement. So if this experiment uses famous brand such as Lipton which already has had good brand recall among consumers, the impact of advertisement type will be quite small, thus cannot
effectively reflect the difference. What’s more, Petty et al. (1983) found that celebrity endorsers do increase brand recall of low-involvement products (consumers can make quick decisions that are relatively routine, e.g. bread, shampoo), but cannot significantly affect recall under high-involvement conditions (purchase decision required extensive information and thought, e.g. buying a house). According to these principles, four low-involvement less well-known product brands which are also frequently used in daily life, are chosen: shampoo: REDKEN, beverage: CURIO, tooth care: VITIS, face cleanser: TULA. For each of these four brands, there are two kinds of advertisements will be shown to the participants. One is traditional celebrity endorsement, a poster with the celebrity, the brand name, logo and the product. The other is celebrity eWoM on social media, a screenshot of Twitter from a celebrity with the product along with a piece of positive description and recommendation. Because there are 4 brands and each brand has 2 advertisement types, so there will be 8 different kinds of advertisements. Four of them are eWoM, four of them are traditional advertisements.
In terms of the celebrity, to make sure that every participant can recognize him/her, this celebrity should be as famous as possible. And since most of the participants are millennials generation, so this study will pick the celebrity who is the most popular among this group of people. According to the data from Twitter Counter (data viewed in April 2017), this paper has selected four celebrities with most followers. They are Katy Perry, Justin Bieber, Taylor Swift, Rihanna and they have 97 million, 93million, 84 million and 71 million followers respectively. (Barack Obama is ranking number 3 with 86 million followers, but as the former president of the United States, he is not appropriate to be an endorser in this experiment. So this paper picks up celebrities ranking 1, 2, 4 and 5).
Operationalization
This paper chooses advertisement type as the independent variable which has two types: traditional endorsement advertisement and eWoM on Twitter. With the help of Photoshop, the author makes eight advertisements. There are one eWoM and one traditional endorsement advertisement for each fixed celebrity combination. There are four brand-celebrity combinations. So altogether, there are eight types of advertisements. See all the advertisement types combinations in Table 1, traditional celebrity endorsement advertisements in Figure 3 and eWoM advertisements in Figure 4.
Table 1 Advertisement Types
TULA & Taylor Swift CURIO & Katy Perry VITIS & Justin Bieber REDKEN & Rihanna eWoM Endorsement
In this study, brand recall is the dependent variable. To measure it, this paper uses the same method as Chakravarti et al., (2003) did in their study-- unaided brand recall, simply asking participants to write down the brand name they saw in the advertisements with product categories as cues. Later the answer of brand recall will be coded as 0=incorrect, 0.5=somewhat correct, 1=almost/totally correct.
Social distance is the mediator in this research. According to Emory Bogardus, social distance is the affinity between people, and it is primarily as subjective psychological distance. Some sociologists such as Georg Simmel think social distance is a judgment of a person as an “insider” or “outsider”. Other researchers measure social distance base on
the frequency of interaction and familiarity. In conclusion, this paper measures social distance with perceived affinity and perceived degree of connection.
Procedure
There are 4 different brand-celebrity combinations with 2 kinds of advertisement types, so there are 8 types in total. Participants will be randomly shown 2 eWoM advertisements and 2 traditional ones. This can be achieved by the randomizer function of Qualtrics. There are 2 phases in this experiment, using phase 1 to test the short-term brand recall and phase 2 to test the long-term brand recall. The online experiment was programmed in Qualtrics.
Phase 1:
Participants will be tested whether they know the celebrities by asking them to select the name of the celebrity among five distraction names according to the picture of the celebrity. Then the familiarity between the participants and celebrities will be tested. Participants will indicate the familiarity by answering “Could you indicate how well you know this person?” with a 5-point scale, ranging from 1 nothing; 3 know the name; 5 fan of this celebrity. These questions are used to test whether the selected celebrities are well-known among participants.
Then participants will see the advertisements along with following questions:
1. Please imagine that you and Rihanna follow each other on Twitter, select the degree of affinity you feel between you and her. (when participants view celebrity eWoM)/ Please select the degree of affinity you feel between you and Rihanna in this advertisement. (when participants view celebrity endorsement)
2. Please imagine that you and Rihanna follow each other on Twitter, how connected do you feel to her? (when participants view celebrity eWoM)/ How connected do you feel to Rihanna in this advertisement? (when participants view celebrity endorsement)
Participants will answer these two questions with seven-point Likert scale ranging from “very low affinity” to “very high affinity” and from “not connected at all” to “connected like a personal friend” respectively.
Followed by these two questions, there is another question to test the familiarity of the brands to participants, “Could you indicate how familiar you are with this shampoo brand: REDKEN” with 1-3 scale, 1 never heard of it before; 2 know the brand, but not a user; 3 user of the brand. This question is used to make sure the selected brands are not well-known among participants.
Before testing short term brand recall, a 3minutes 30 seconds video will be shown to participants, which can distract their attention to eliminate the advantage of last shown advertisement. After viewing the video, they are required to write down the brand names while giving product categories as cues.
Last, participants are required to leave their email to receive the information about the second phase of experiment and dismissed.
Phase 2:
24 hours later, all the participants will receive an email which is automatically sent by Qualtrics with one question:
Please write down the brand names you saw in the advertisements yesterday.
The first phase of the experiment will take participant about 12 minutes to finish. And the whole experiment will take 24 hours to finish.
Sample
Undergraduate students are the targeting participants of this experiment. Because they are active on social media and at the same time they can get in touch with many traditional advertisements. Therefore, the writer distributed the online experiment survey among students at University of Amsterdam. The author sent emails with experiment link to fellow students, posted experiment link in Facebook groups. In order to encourage the students to participate the experiment, the author offered them an opportunity to win one of the three 15 euros voucher prizes. About 450 students received the link, and 115 students participated the experiment in the end. And they took 11.7 minutes (704 seconds) on average to finish the first stage of the experiment. They finished the experiment through Qualtrics.
Altogether 115 participants joined the online experiments which took them about 12 minutes to complete. 14 participants dropped out during the process. Among the remained completed participants, 76 results are valid. The criteria of invalid responses are: participants spent significant shorter time than average; participants who wrote “no idea”, “don’t know”, “don’t remember” etc. for all the four brand recall questions (if they followed the instructions and read the advertisements carefully, they should at least remember something instead of nothing. If they failed to recall anything, this means that they didn’t pay enough attention to the experiment. Thus their responses could be invalid). Because of the “force response” function of Qualtrics, there were no missing values. Therefore, there were 76 valid respondents in this research.
The average age of the participants is 23.63 (N=76, M=23.63, SD=2.19) while the youngest participant is 18, the oldest is 29. In terms of gender, 51 (67.1%) of the participants are female, and 25 (32.9%) are male.
Figure 4 Age Distribution of Participants- stage 1
However, in terms of the second stage of the experiments, only 39 participants replied to the email which was sent to them automatically by Qualtrics in 24 hours after they finished the first stage of the experiment. So there were 39 valid respondents for the second stage (measure long-term brand recall) in this experiment.
The average age of the second stage participants is 24.03 (N=39, M=24.03, SD=2.39) while the youngest participant is 20, the oldest is 29. In terms of gender, 29 (74.4%) of them are female, and 10 (25.6%) are male.
Figure 5 Age Distribution of Participants- stage 2
Analysis and Results
Examine whether the celebrities are well-known
Top four celebrities on Twitter with most followers are chosen for this experiment. They are Katy Perry, Justin Bieber, Taylor Swift, Rihanna ranking 1, 2, 4 and 5 respectively. One question in the experiment is used to make sure the celebrity is “celebrity” from participant’s perspective. The results show that the selected celebrities are well-known in the participants. Taylor Swift: 74 out of 76 participants recognized her name among five disturbing options. Katy Perry: 74 out of 76 participants recognized her name among five disturbing options. Rihanna: 74 out of 76 participants recognized her name among five disturbing options. Justin Bieber: 75 out of 76 participants recognized his name among five disturbing options.
Examine whether the brands are not well-known
To eliminate the effect of the well-known brand that everyone knows it so well that people have remembered the name before and can easily recall the name (Nedungadi 1990),
this study picks up four less well-known brands. The results show that the selected brands are not well-known among participants. CURIO: 71 out of 76 participants never heard of it before. VITIS: 70 out of 76 participants never heard of it before. TULA: 73 out of 76 never heard of it before. REDKEN: 52 out of 76 never heard of it before, 22 out of 76 know the brand but not a user.
Reliability Analysis
Based on theory from sociologists Emory Bogardus and Georg Simmel, social distance can be measured by affinity and connection. So these two variables are used as scales of social distance. Participants are required to indicate the degree of perceived affinity and degree of feeling connected towards the celebrity in scale of 7 points. From the reliability analysis results, the social distance scale has high reliability with Cronbach’s Alpha = 0.922. The corrected item-total correlations indicate that all the items have a good correlation with the total score of the scale (Affinity 0.880, Connection 0.880).
Hypothesis Testing H1
eWoM advertisement on Twitter generates higher short-term brand recall than traditional celebrity endorsement advertisement.
A paired-samples T-Test with Bootstrapping was conducted to examine whether there was a significant difference of short-term brand recall between eWoM advertisement and traditional celebrity endorsement advertisement. There were no outliers in the data, as
distributed. So paired-samples T-Test with Bootstrapping should be conducted which can be successfully used to estimate the difference of two variables under nonnormality (Cribari-Neto 1999). The bootstrap paired-samples T-Test results showed that there was a statistically significant difference (p=0.035<0.05). As the result shown, there is a significant difference of short-term brand recall between these two ad types. From the mean comparison of Paired Samples Statistics, eWoM (M=1.1053, SD=0.74, N=76) has higher short-term brand recall score than celebrity endorsement (M=0.875, SD=0.66, N=76). Therefore, the writer does not reject H1.
Please see the results of the mean comparison in Figure 6 and results of paired-samples T-Test with Bootstrapping in Table 2.
Table 2 Paired Samples Statistics of Short-term Brand Recall
Paired Samples Statistics of Short-term Brand Recall
Statistic Bootstrapa Bias Std. Error 95% Confidence Interval Lower Upper Short-term brand recall eWoM Mean 1.11 -.0006 .0813 .9342 1.2630 N 76 Std. Deviation .74 -.00602 .03785 .65228 .80008 Std. Error Mean .084 Short-term brand recall celebrity endorsement Mean .88 .0001 .0771 .7239 1.0197 N 76 Std. Deviation .66 -.00694 .04027 .58205 .72920 Std. Error Mean .076
a. Unless otherwise noted, bootstrap results are based on 1000 bootstrap samples
H2
eWoM advertisement on Twitter generates higher long-term brand recall than traditional celebrity endorsement advertisement.
A paired-samples T-Test with Bootstrapping was conducted to examine whether there was a significant difference of long-term brand recall between eWoM advertisement and traditional celebrity endorsement advertisement. There were no outliers in the data, as examined by boxplot. However, through the normality test results, the data was not normally
successfully used to estimate the difference of two variables under nonnormality (Cribari-Neto, 1999). Because the sample is small (N=39), so alpha-level of 0.10 is acceptable. The bootstrap paired-samples T-Test results showed that there was statistically significant difference (p=0.089<0.10). As the result shown, there is significant difference of long-term brand recall between these two ad types. Therefore, long-term brand recall of eWoM (M=0.96, SD=0.77) has a higher mean than celebrity endorsement (M=0.72, SD=0.63). Therefore, the writer does not reject H2.
Please see the results of mean comparison in Figure 7 and results of paired-samples T-Test with Bootstrapping in Table 3.
Table 3 Paired Samples Statistics of Long-term Brand Recall
Paired Samples Statistics of Long-term Brand Recall
Statistic Bootstrapa Bias Std. Error 95% Confidence Interval Lower Upper Long-term brand recall eWoM Mean .72 -.0079 .0974 .5256 .9099 N 39 Std. Deviation .63 -.01053 .05784 .49939 .73303 Std. Error Mean .10 Long-term brand recall celebrity endorsement Mean .96 -.0049 .1201 .7308 1.1923 N 39 Std. Deviation .77 -.00890 .04751 .66355 .85208 Std. Error Mean .12
a. Unless otherwise noted, bootstrap results are based on 1000 bootstrap samples H3
Perceived social distance is lower towards celebrity followed on social media than the celebrity in a traditional advertisement.
Based on theory from sociologists Emory Bogardus and Georg Simmel, social distance can be measured by affinity and connection. In other words, the higher the affinity and the higher the connection, the lower the social distance. From the mean comparison of Paired Samples Statistics, affinity and connection scores are significantly higher towards celebrity
Because there are two items (affinity & connection) used to measure social distance, so two paired-samples T-Test were conducted to examine whether there was significant difference of affinity and connection between celebrity on Twitter and celebrity in a traditional advertisement. There were no outliers in the data, as examined by boxplot. And both the data of affinity and connection were normally distributed, as examined by Shapiro-Wilk test of normality. And as examined before, the social distance scale has high reliability with Cronbach’s Alpha = 0.922.
The paired-samples T-Test results of Affinity showed that there was statistically significant difference (t(75)=-5.673, p=0.000<0.05) between celebrity on Twitter (M=3.4803, SD=1.55) and celebrity in a traditional advertisement (M=2.6711, SD=1.48). The paired-samples T-Test results of Connection showed that there was a statistically significant difference (t(75)=-3.773, p=0.000<0.05) between celebrity on Twitter (M=3.4868, SD=1.46) and celebrity in a traditional advertisement (M=2.8882, SD=1.50).
Table 4 Paired Samples Statistics-Affinity
Paired Samples Statistics-Affinity
Mean N Std. Deviation Std. Error Mean
Affinity Celebrity in Traditional ad 2.6711 76 1.47773 .16951 Celebrity on Twitter 3.4803 76 1.54799 .17757
Table 5 Paired Samples Statistics-Connection
Paired Samples Statistics-Connection
Mean N Std. Deviation Std. Error Mean
Connection Celebrity in Traditional ad 2.8882 76 1.50022 .17209 Celebrity on Twitter 3.4868 76 1.45825 .16727
Before running the paired-samples T-Test for Social Distance, a scale mean computation was conducted to compute the mean of the two measure items (affinity and connection). There were no outliers in the data, as examined by boxplot. And the data was normally distributed, as examined by Shapiro-Wilk test of normality. Then a paired-samples T-Test was conducted to examine whether there was a significant difference of social distance between celebrity on Twitter (M=3.4836, SD=1.43) and celebrity in a traditional advertisement (M=2.7796, SD=1.44). The results showed that there was a statistically significant difference (t(75)=-5.068, p=0.000<0.05). Therefore, the writer does not reject H3.
Please see the results of the mean comparison in Figure 9 and results of paired-samples T Test in Table 6.
Figure 9 Mean scores of social distance
Table 6 Paired Samples Statistics of Social Distance
Paired Samples Statistics of Social Distance
Mean N Std. Deviation Std. Error Mean
Social distance Celebrity in Traditional ad 3.4836 76 1.43256 .16433 Celebrity on Twitter 2.7796 76 1.44047 .16523 H4
There is mediation effect of social distance on short-term brand recall.
There are two advertisement types, and the writer has coded eWoM ad as 1 and traditional celebrity endorsement as 0. IV is ad type, DV is the short-term brand recall, and social distance is the mediator. There were no outliers in the data, as examined by boxplot.
However, through the normality test results of residual, the residual was not normally distributed. So regression with Bootstrapping should be conducted which can be successfully used to estimate the regression effect of variables under nonnormality (Cribari-Neto 1999). Two regressions with bootstrapping were conducted to check the direct effect of ad type and indirect effect of social distance on short-term brand recall. The P-value of the indirect effect was 0.094 > 0.05, and it wasn’t significant. So the mediation effect does not take place. Therefore, the writer rejects H4.
Table 7 The Regression Tables of the Mediation Analysis
Consequent
Social Distance(M) Short-term Brand Recall(Y)
Antecedent Coeff. SE P Coeff. SE P
Ad type(X) a1 0.704 0.233 0.003 c1' 0.254 0.117 0.032
Social Distance(M) — — — b1 -0.034 0.039 0.398
constant i1 2.779 0.165 0.000 i2 0.969 0.137 0.000
R2=0.057 R2=0.313
Effect SE P LLCI ULCI
Direct effect c1' 0.254 0.117 0.032 -1.197 0.202
Total effect c1 0.2303 0.114 0.045 -0.615 0.214
Boot SE Boot LLCI Boot ULCI
Indirect effect a1b1 -0.034 0.032 -0.104 0.027
H5
There is mediation effect of social distance on long-term brand recall.
There are two advertisement types, the writer has coded eWoM ad as 1 and traditional celebrity endorsement as 0. IV is ad type, DV is the long-term brand recall, and social distance is the mediator. There were no outliers in the data, as examined by boxplot. However, through the normality test results of residual, the residual data was not normally distributed. So regression with Bootstrapping should be conducted which can be successfully used to estimate the regression effect of variables under nonnormality (Cribari-Neto 1999). Two regressions with bootstrapping were conducted to check the direct effect of ad type and indirect effect of social distance on long-term brand recall. The P-value of the indirect effect was 0.283 > 0.05, and it wasn’t significant. So the mediation effect does not take place. Therefore, the writer rejects H5.
Table 8 The Regression Tables of the Mediation Analysis
Consequent
Social Distance(M) Long-term Brand Recall(Y)
Antecedent Coeff. SE P Coeff. SE P
Ad type(X) a1 0.724 0.319 0.026 c1' 0.265 0.165 0.114
Social Distance(M) — — — b1 -0.029 0.057 0.616
constant i1 2.692 0.226 0.000 i2 0.796 0.192 0.000
R2=0.063 R2=0.033
F(1, 76)=, p=0.026 F(2,75)=1.285, p=0.283
Effect SE P LLCI ULCI
Direct effect c1' 0.265 0.165 0.114 -4.043 0.346
Total effect c1 0.244 0.159 0.130 -1.489 0.358
Boot SE Boot LLCI Boot ULCI
Indirect effect a1b1 -0.021 0.043 -0.138 0.044
Discussion and Implications
This study examined the effect of two different commonly used advertisement types on short-term brand recall and long-term brand recall. One of the two advertisements is
tested whether there is a mediation effect of social distance in this relationship. An online experiment in which ad types were manipulated was conducted to examine the effects. It was expected that eWoM generates higher short-term brand recall and long-term brand recall than celebrity endorsement. And this paper supposed that perceived social distance towards celebrity on Twitter is lower towards celebrity in an advertisement. And perceived social distance towards the celebrity will mediate this effect.
Currently, there is no existing research focusing on the different effects of these two kinds of celebrity involvements. There is no literature studying the mediator impact of social distance in this process. Therefore, this study has filled up this research gap.
First, as expected, the results show that perceived social distance is lower towards celebrity on Twitter than celebrity in a traditional advertisement. This result is congruence with previous studies that people are becoming familiar and closer to each other by cumulated interactions on social media (Newcomb 1978). Since people follow the celebrity on Twitter, they have more opportunities to get to know and interact with the celebrity. So the celebrity is closer and familiar to the follower. What’s more, people consider the celebrity they followed on Twitter as part of their network and people always feel closer to their friends (Emory Bogardus 1933). However, people rarely interact with celebrity in advertisement and they won’t consider the celebrity as part of their network. Therefore, participants have lower perceived social distance towards celebrity on social media –Twitter than celebrity in a traditional advertisement.
Second, as predicted, celebrity eWoM generates both higher short-term and long-term brand recall than traditional celebrity endorsement. Since people consider celebrity followed on Twitter is part of their “network” and they are more willing to acquiring information from their network, they will pay more attention to the information from their network – the
celebrity they followed on social media (Roggeveen 2012). This relationship is effective for both short-term and long-term brand recall. People have higher short-term and long-term brand recall of brand/product recommended by a celebrity on Twitter than brand/product endorsed by a celebrity in a traditional advertisement.
Third, it was predicted that there is mediation effect of social distance on the relationships. However, they are not supported by the results of this study. There is no mediation effect of social distance. So social distance is not the reason why there are differences between the two advertisement types on brand recall. Social distance is one of the differences between the two advertisement types. It has been proved that perceived social distance is significant lower towards celebrity on Twitter than celebrity in a traditional advertisement. But the differences of social distance do not influence the effectiveness of two advertisement types on brand recall. Since no previous studies are researching the mediation effect of social distance on brand recall, they may not have enough theory supported.
This study helps to understand the differences between the two advertisement types— celebrity eWoM on Twitter and traditional celebrity endorsement. The results have shown that there are significant differences of social distance towards celebrity in these two advertisement types. Social distance means psychological feeling of how far from another person or other people, that is perceived closeness between one person and another person (Magee and Smith 2013). Consumers feel more connected towards celebrity on Twitter. However, the difference of social distance won’t affect the brand recall effects. There are other reasons affecting the effectiveness of the two ad types on brand recall.
The results of this study contribute to the application of advertising for companies, advertisers, and marketers. Involving celebrities in marketing has been proven an effective
consumer’s brand recall which is the foundation of consumer loyalty. Brand recall is not only the prerequisite of final purchase behavior (Jacoby et al., 1977), but also the foundation of achieving brand resonance (Keller, 1993). This study has proven that celebrity eWoM on social media is significantly more effective on brand recall both short-term and long-term than traditional celebrity endorsement. Therefore, companies should invest more in celebrity eWoM than traditional advertisement. When conduct marketing plan, marketers and advertisers should take the effect of different advertisement types into consideration. Since celebrity involvement in advertising is usually expensive, they should choose the most effective advertisement to achieve the best effect of advertisement with the lowest cost.
Limitations and Future Research
This experiment used four specific daily used product categories (shampoo, toothpaste, soft drink, face cleanser) to examine the effect of advertisement type on brand recall. The results may only apply to these four product categories. Whether the findings are also applicable to other product categories still need to be studied further. What’s more, four particular celebrities (Taylor Swift, Katy Perry, Justin Bieber, Rihanna) were chosen in this experiment. Whether the results remain the same with other celebrities still need to be tested in the future.
Second, the valid number of samples is 76 in this research, which is relatively a small sample. This may lead to insignificant statistical mediation effect analysis results. Because of time and budget limitation, the author only managed to get 115 participants (76 valid samples) for the first period of experiment and 39 participants for the final stage of the experiment. So future research can try to enlarge the number of participants and repeat this experiment to retest the mediation effect.
Third, the number of female participants (67.1% for the first experiment stage, 74.4% for the second experiment stage) in the experiment is much more than male participants (32.9% for the first experiment stage, 25.6% for the second stage experiment stage). It will be better to have nearly 50% female participant and 50% male participants. Besides, since this research focuses on undergraduate students who are familiar with both social media advertisements and traditional advertisements, all the participants’ ages are between 18 and 29 years old. So whether the results are still applicable to other age groups remains unknown. Future studies could enlarge this experiment to other age groups to check the results.
Fourth, in terms of long-term brand recall, this research took a period of 24 hours because of feasibility. However, in real life, it always takes longer than 24 hours for a consumer to buy the product after viewing the advertisement. So future research could extend the time difference to more than 24 hours to test the effectiveness on long-term brand recall.
Last, this research conducted a lab experiment which simplifies the situations. But in the real business world, the situations are more complicated. There are many other unexpected factors, such as the profile of the celebrity followers, time and location of the advertisements. So future research could repeat the experiment in field experiment form. In terms of social media, this research used Twitter as an example. It will be interesting to see whether the same effect happens on other social media platforms such as Facebook, LinkedIn, Google +, Pinterest etc. Therefore, future research could apply this experiment to other social media platforms and check the effectiveness.
References
1. Amblee N, Bui T. Harnessing the influence of social proof in online shopping: The effect of electronic word of mouth on sales of digital microproducts. International Journal of Electronic Commerce. 2011 Dec 1;16(2):91-114.
2. Arndt J. Role of product-related conversations in the diffusion of a new product. Journal of marketing Research. 1967 Aug 1:291-5.
3. Awad NF, Ragowsky A. Establishing trust in electronic commerce through online word of mouth: An examination across genders. Journal of Management Information Systems. 2008 Apr 1;24(4):101-21.
4. Bettman JR, Park CW. Effects of prior knowledge and experience and phase of the choice process on consumer decision processes: A protocol analysis. Journal of consumer research. 1980 Dec 1;7(3):234-48.
5. Bickart B, Schindler RM. Internet forums as influential sources of consumer information. Journal of interactive marketing. 2001 Jan 1;15(3):31-40.
6. Binzel C, Fehr D. Social distance and trust: Experimental evidence from a slum in Cairo. Journal of Development Economics. 2013 Jul 31;103:99-106.
7. Biswas, D., Biswas, A., & Das, N. (2006). The differential effect of celebrity and expert endorsements on consumer risk perceptions. The role of consumer knowledge, perceived congruency, and product technology orientation. Journal of Advertising, 35(2), 17–31.
8. Bohnet, I., & Frey, B. S. (1999). Social distance and other-regarding behavior in dictator games: Comment. American Economic Review, 89, 335-339.
9. Chakravarti A, Janiszewski C. The influence of macro-level motives on consideration set composition in novel purchase situations. Journal of Consumer Research. 2003 Sep 1;30(2):244-58.
10. Cheung MY, Luo C, Sia CL, Chen H. Credibility of electronic word-of-mouth: Informational and normative determinants of on-line consumer recommendations. International Journal of Electronic Commerce. 2009 Jul 1;13(4):9-38.
11. Chevalier JA, Mayzlin D. The effect of word of mouth on sales: Online book reviews. Journal of marketing research. 2006 Aug 1;43(3):345-54.
12. Cowley E, Mitchell AA. The moderating effect of product knowledge on the learning and organization of product information. Journal of Consumer Research. 2003 Dec 1;30(3):443-54.
13. Cribari-Neto, Francisco, and Spyros G. Zarkos. "Bootstrap methods for heteroskedastic regression models: evidence on estimation and testing." Econometric Reviews 18.2 (1999): 211-228.
14. Cui G, Lui HK, Guo X. The effect of online consumer reviews on new product sales. International Journal of Electronic Commerce. 2012 Oct 1;17(1):39-58.
15. East R, Hammond K, Lomax W. Measuring the impact of positive and negative word of mouth on brand purchase probability. International journal of research in marketing. 2008 Sep 30;25(3):215-24.
16. Elberse A, Verleun J. The economic value of celebrity endorsements. Journal of Advertising Research. 2012 Jun 1;52(2):149-65.
17. Elliot, A. (1998). Celebrity and political psychology: Remembering Lennon. Political Psychology, 19, 833–852.
18. Erdogan BZ. Celebrity endorsement: A literature review. Journal of marketing management. 1999 May 1;15(4):291-314.
19. Fan YW, Miao YF. Effect of electronic word-of-mouth on consumer purchase intention: The perspective of gender differences. International Journal of Electronic
20. Friedman, H. H., and L. Friedman. 1979. “Endorser Effectiveness by Product Type.” Journal of Advertising Research 9 (5): 63 – 71.
21. Gupta P, Harris J. How e-WOM recommendations influence product consideration and quality of choice: A motivation to process information perspective. Journal of Business Research. 2010 Oct 31;63(9):1041-9.
22. Hakimi, B.Y., Abedniya, A. and Zaeim, M.N. (2011), “Investigate the impact of celebrity endorsement on brand images”, European Journal of Scientific Research, Vol. 58 No. 1, pp. 116‐132.
23. Hennig-Thurau, T., Gwinner, K.P., Walsh, G., Gremler, D.D., 2004. Electronic word-of-mouth via consumer-opinion platforms: what motivates consumers to articulate themselves on the Internet? Journal of Interactive Marketing 18 (1), 38–52
24. Hung, K.H., Li, S.Y., 2007. The influence of eWOM on virtual consumer communities: social capital, consumer learning, and behavioral outcomes. Journal of Advertising Research 47 (4), 485–495.
25. Jacoby, J., Syzabillo, G. J., & Busato-Schach, J. (1977). Information acquisition behavior in brand choice situations. Journal of Consumer Research, 3, 209–216. 26. Keller KL. Conceptualizing, measuring, and managing customer-based brand equity.
the Journal of Marketing. 1993 Jan 1:1-22.
27. Klaus, N. and Bailey, A.A. (2008), “Celebrity endorsements: an examination of gender and consumers' attitudes”, American Journal of Business, Vol. 23 No. 2, pp. 53‐61.
28. Knoll J, Matthes J. The effectiveness of celebrity endorsements: a meta-analysis. Journal of the Academy of Marketing Science. 2017 Jan 1;45(1):55-75.
29. Kramer, R. M., & Brewer, M. B. (1984). Effects of group identity on resource use in a simulated commons dilemma. Journal of Personality and Social Psychology, 46, 1044-1057.
30. Langmeyer, Lynn, and Mary Walker. 1991. A first step to identify the meaning in celebrity endorsers. Advances Consumer Research 18:364–371.
31. Lawler, E. J., & Yoon, J. (1996). Commitment in exchange relations: Test of a theory of relational cohesion. American Sociological Review, 61, 89-108.
32. Lee J, Lee JN. Understanding the product information inference process in electronic word-of-mouth: An objectivity–subjectivity dichotomy perspective. Information & Management. 2009 Jun 30;46(5):302-11.
33. Lee M, Youn S. Electronic word of mouth (eWOM) How eWOM platforms influence consumer product judgement. International Journal of Advertising. 2009 Jan 1;28(3):473-99.
34. Liu Y. Word of mouth for movies: Its dynamics and impact on box office revenue. Journal of marketing. 2006 Jul 1;70(3):74-89.
35. Magee JC, Smith PK. The social distance theory of power. Personality and Social Psychology Review. 2013 May;17(2):158-86.
36. McCracken G. Who is the celebrity endorser? Cultural foundations of the endorsement process. Journal of consumer research. 1989 Dec 1;16(3):310-21.
37. McCutcheon LE, Lange R, Houran J. Conceptualization and measurement of celebrity worship. British journal of psychology. 2002 Feb 1;93(1):67-87.
38. Meirick, P. C. (2005). Rethinking the target corollary: The effects of social distance, perceived exposure, and perceived predispositions on first-person and third-person perceptions. Communication Research, 32, 822-843.
39. Miller, F. M., & Allen, C. T. (2012). How does celebrity meaning transfer? Investigating the process of meaning transfer with celebrity affiliates and mature brands. Journal of Consumer Psychology, 22(3), 443–452.
40. Mindy Weinstein, 5 Examples of Social Media Celebrity Endorsements That Paid off Big Time, Search Engine Journal, 2016 Feb, viewed 2017 Jan 30,
https://www.searchenginejournal.com/5-examples-social-media-celebrity-endorsements-paid-off-big-time/156342/
41. Mullainathan S. A memory-based model of bounded rationality. The Quarterly Journal of Economics. 2002 Aug 1;117(3):735-74.
42. Nedungadi, P. (1990). Recall and consumer consideration sets: Influencing choice without altering brand evaluations. Journal of Consumer Research, 17, 263–276. 43. Newcomb TM. The acquaintance process: Looking mainly backward. Journal of
Personality and Social Psychology. 1978 Oct;36(10):1075.
44. Ohanian, R. (1990), “Construction and validation of a scale to measure celebrity endorser's perceived expertise, trustworthiness and attractiveness”, Journal of Advertising, Vol. 19 No. 3, pp. 39‐52.
45. Ohanian, R. (1991), “The impact of celebrity spokesperson's perceived image on consumers' intention to purchase”, Journal of Advertising Research, Vol. 31 No. 1, pp. 46‐52.
46. Park DH, Lee J, Han I. The effect of on-line consumer reviews on consumer purchasing intention: The moderating role of involvement. International journal of electronic commerce. 2007 Jul 1;11(4):125-48.
47. Park DH, Lee J. eWOM overload and its effect on consumer behavioral intention depending on consumer involvement. Electronic Commerce Research and Applications. 2009 Feb 28;7(4):386-98.
48. Park, C. W., & Lessig, V. P. (1981). Familiarity and its impact on consumer biases and heuristics. Journal of Consumer Research, 8, 223–230.
49. Petty, R. E., Cacioppo, J. T., & Schumann, D. (1983). Central and peripheral routes to advertising effectiveness: the moderating role of involvement. Journal of Consumer Research, 10(2), 135–146.
50. Robert Passikoff , When It Comes To Nike Celebrity Endorsements, They Have To Make Sure The Shoe Fits (LeBron), 2013 Dec, views 2017 Jan 30, http://www.forbes.com/sites/robertpassikoff/2013/12/12/if-the-nike-brand-dont-fit-lebron-cannot-commit-2/#4c84a1ce67fa
51. Schiffenbauer A, Schiavo RS. Physical distance and attraction: An intensification effect. Journal of Experimental Social Psychology. 1976 May 31;12(3):274-82.
52. Senecal S, Nantel J. The influence of online product recommendations on consumers’ online choices. Journal of retailing. 2004 Dec 31;80(2):159-69.
53. Spry, A., Pappu, R. and Cornwell, T.B. (2011), “Celebrity endorsement, brand credibility and brand equity”, European Journal of Marketing, Vol. 45 No. 6, pp. 882‐ 909
54. Stephan, E., Liberman, N., & Trope, Y. (2011). The effects of time perspective and level of construal on social distance. Journal of Experimental Social Psychology, 47, 397-402.
55. Suzanne Vranica, Behind the Preplanned Oscar Selfie: Samsung's Ad Strategy. The Wall Street Journal. 2014 Mar, viewed 2017 Jan 30, http://www.wsj.com/articles/SB10001424052702304585004579417533278962674 56. Till, B. D., and M. Busler. 2000. “The Match-Up Hypothesis: Physical Attractiveness,
