EFFECTS OF CELEBRITY ENDORSEMENT ON CONSUMER

PURCHASING INTENTION OF APPAREL PRODUCTS

AU-YEUNG PUI YI, PRISCILLA

BA (Hons) Scheme in Fashion and Textiles
(Fashion Marketing and Merchandising Specialism)

INSTITUTE OF TEXTILES & CLOTHING

THE HONG KONG POLYTECHNIC UNVERSITY
2012

EFFECTS OF CELEBRITY ENDORSEMENT

ON CONSUMER PURCHASING INTENTION
OF APPAREL PRODUCTS

A Thesis Submitted
In Partial Fulfillment of the Requirements
For the Degree of
Bachelor of Arts (Honours)
in
Fashion & Textiles
(Fashion Marketing and Merchandising Specialism)
Under the Supervision of
Dr. ZHANG Z.M.
by
Au-Yeung Pui Yi, Priscilla

Institute of Textiles & Clothing

The Hong Kong Polytechnic University
2012

ACKNOWLEDGEMENTS

First and foremost I am heartily thankful to my supervisor, Dr. Zhang Zhi-Ming

who has supported me, throughout my thesis with his patience and knowledge
whilst allowing me the room to work in my own way. His encouragement,
constant guidance and invaluable advice from the preliminary to the concluding
level enabled me to develop an understanding of the subject.

Secondly I would like to thank my family and dearest friends Jennifer Chu, Peter
Lee, who were always willing to help, support and give their best suggestions. My
research would not have been possible without their helps.

Finally, I offer my regards and blessings to all of those who supported me in any
respect during the completion of the research.

CERTIFICATE OF ORIGINALITY

I hereby declare that this thesis is my own work and that, to the best of my knowledge and belief, it

reproduces no material previously published or written, nor material that has been accepted for the award

of any other degree or diploma, except where due acknowledgement had been made in the text.

_____________________________________________________________(Signed)

______________________________________________________(Name of student)

ABSTRACT
Hong Kong is a famous international city with an influential role in Asia. Because of a business
environment of cut-throat competition, companies must find ways to strike for survival and sustain a
competitive position in the market. Celebrity endorsement is one of the strategies which companies
frequently used to promote the brand thus enhance sales and ultimately increase market share. This is
especially common in the highly competitive fashion industry. There are a number of researches
which investigated the effects of celerity endorsement on consumers purchase intention. However,
they are not general to all and gender differences have not been thoroughly discussed. Moreover,
relationships between good-looking, trustworthiness, expertise and attractiveness of celebrity
endorsers have not been investigated. Therefore, this research is to investigate effects on celebrity
endorsement on consumers purchase intention of apparel product by using Source credibility model
(Ohanian, 1990).

In order to obtain a more reliable result, a quantitative research is carried out to investigate the topic.
In general it was found that whether a celebrity is expert and trustworthy had the most significant
effect on purchase intention, regardless of gender. For females, they perceive celebrity endorsement
in a rather different way than that of males for instance the attractiveness of the endorser was found
to be the most important source while it was trustworthiness that male chose to be the most
important.

Moreover it seems the two genders have a different view on good-looking celebrity. A good-looking
celebrity meant attractive, trustworthy but not expert to female, whereas male would associate a
good-look celebrity with trustworthy, expertise and attractive.
As for the perception towards a plain-looking celebrity, females would only think it was expertise,
while male would consider it as both trustworthy and expertise.
Generally the effect of celebrity endorsement is found to have little direct impact on consumers
purchase decision, yet it is still an effective strategy because it is suggested as a good way to grab
consumers attention although they may not willingly spend a higher price if the products are
endorsed by a celebrity.

This research also suggests some criteria for selecting the celebrity endorsers with the right attributes
for their products. Besides, it is proved that genders differences are existed in the perception of
celebrity endorsement. Therefore companies are recommended to conduct sufficient research in order
to generate the best effect of a celebrity endorsement.

CONTENTS
Page
ACKNOWLEDGEMENT
CERTIFICATE OF ORIGINALITY
ABSTRACT
CHAPTER 1 INTRODUCTION
1.1 Background of study

1.1.1 Situation of celebrity endorsement in Hong Kong

1.1.2 Significance of the research

1.2 Objectives

1.3 Scope of study

1.4 Outline of the study

1.5 Chapter summary

CHAPTER2 LITERATURE REVIEW

2.1 Introduction

2.2 Consumer behaviour

2.2.1 Definition of consumer behaviour

2.2.2 Consumer decision process

2.2.2.1 Consumer decision process model

10

2.2.2.2 Need recognition

11

2.2.2.3 Information search

12

2.2.2.4 Pre-purchase evaluation of alternatives

12

2.2.2.5 Purchase

13

2.2.2.6 Consumption

14

2.2.2.7 Post-consumption evaluation

15

2.2.2.8 Divestment

15

2.3 Gender

16

2.3.1 Definitions of gender

16

2.3.2 Characteristics of male

17

2.3.3 Characteristics of female

17

2.4 Purchase intention

2.4.1 Definition of purchase intention
2.5 Celebrity endorsement
2.5.1 Definition of celebrity endorsement

18
18
20
20

2.5.2 Forms of celebrity endorsement

20

2.5.3 Benefits of celebrity endorsement

21

2.5.4 Celebrity endorsement as reference group

23

2.6 Source credibility model

24

2.6.1 Definition of source credibility

24

2.6.2 Source credibility model

24

2.6.3 Trustworthiness

26

2.6.3.1 Definition of trustworthiness

26

2.6.3.2 Effects of trustworthiness

27

2.6.4 Expertise

28

2.6.4.1 Definition of expertise

28

2.6.4.2 Effects of expertise

28

2.6.5 Attractiveness

29

2.6.5.1 Definition of attractiveness

29

2.6.5.2 Effects of attractiveness

30

2.6.6 Good-looking
2.6.6.1 Definition of good-looking
2.6.7 Modification of Source credibility model

31
32
32

2.7 Celebrity endorsement in Hong Kong

33

2.8 Chapter summary

35

CHAPTER 3 METHODOLOGY
3.1 Introduction

37

3.2 Research design

37

3.2.1 Nature of research

37

3.2.2 Exploratory research

38

3.2.3 Quantitative research

38

3.3Conceptual framework

39

3.4 Research hypothesis

40

3.4.1 Impact of attractiveness of celebrity endorsers and purchase intention.

40

3.4.2 Impacts of trustworthiness of celebrity endorsers and purchase intention.

41

3.4.3 Impacts of expertise of celebrity endorsers and purchase intention.

42

3.5 Research approach

3.5.1 Data gathering process

43
44

3.5.2 Preliminary test

44

3.5.2.1 Selection of celebrities

44

3.5.2.2 Criteria for selecting celebrities

46

3.6 Questionnaire survey

47

3.6.1 Questionnaire design

47

3.6.2Questionnaire details

48

3.6.3 Measurement and scaling

50

3.6.4 Pilot test

51

3.6.5 Sampling method and distribution of questionnaires

52

3.7 Data entry and analysis techniques

53

3.8Chapter summary

55

CHAPTER 4 FINDINGS AND RESULTS

4.1 Introduction

57

4.2 Sample size and cleaning invalid data

57

4.3 Demographic data analysis

58

4.3.1 Gender

58

4.3.2 Age

59

4.3.3 Education Level

60

4.3.4 Occupations

61

4.3.5 Average monthly income

64

4.4 Consumers purchase intention under the perception of celebrity endorsement.

4.4.1 Reliability test
4.4.1.1Reliability of the data set
4.4.2 Factors analysis
4.4.2.1 Attributes of celebrity endorsers to purchase intention
4.4.3Impacts of celebrity endorsements

65
65
65
66
66
70

4.4.3.1 Impacts of attributes of celebrity endorsement to

consumers purchase intention

70

4.4.3.2 Impacts of source credibility of celebrity endorsers to

consumers purchase intention

73

4.4.3.3 Important dimension of source credibility

for celebrity endorsers

77

4.5 Source credibility

81

4.6 Perceived image of celebrity endorsers

83

4.6.1 Reliability of data set

83

4.6.2 Good-Looking, Trustworthiness, Expertise and Attractiveness

84

4.6.2.1 Ratings

84

4.6.3 Relationships between Good-looking, Trustworthiness, Expertise

and Attractiveness of celebrity endorsers.

92

4.6.3.1 Good-looking and Trustworthiness

92

4.6.3.2 Good-looking and Attractiveness

95

4.6.3.3 Good-looking and Expertise

99

4.6.4 Favorite celebrity endorsers

4.7 Impacts of celebrity endorsement

102
104

4.7.1 Reliability of data set

104

4.7.2 Consumer buying behavior processes

105

4.7.3Effectiveness of celebrity endorsements on fashion apparel products

108

4.7.4 Perceived value of the endorsed fashion products and brands

110

4.7.5 Influence of celebrity endorsements on fashion apparel products

111

4.8 Chapter Summary

113

CHAPTER 5 CONCLUSION AND RECOMMENDATION

5.1 Introduction

114

5.2Conclusionson the findings

114

5.2.1 Consumers purchase intention under the perception of

celebrity endorsement

114

5.2.2 Source credibility

115

5.2.3 Perceived image of celebrity endorsers

116

5.2.4 Impacts of celebrity endorsement

118

5.3 Recommendation

120

5.4 Limitations of study

121

5.5 Suggestions for future study

122

5.6 Chapter summary

122

APPENDIX

124

REFERENCES

168

APPENDIX

Appendix I

Questionnaire

Appendix II

Demographic data analysis

Appendix III

Reliability test of the attributes of celebrity endorsers

Appendix IV

Factors analysis of the attributes of celebrity endorsers

Appendix V

Independent t-test (Attributes of celebrity endorsement)

Appendix VI

Paired sample t-test (Source credibility of celebrity endorsement)

Appendix VII

Independent t-test (Source credibility of celebrity endorsement, by genders)

Appendix VIII

Paired sample t-test (Important dimension of source credibility)

Appendix IX

Independent t-test (Important dimension of source credibility by genders)

Appendix X

Descriptive Statistics (Credibility source of celebrity endorsement)

Appendix XI

Reliability test of four attributes of celebrity endorsers

Appendix XI

Compare means of ten celebrity endorsers

Appendix XII

Relationships between Good-looking, Trustworthiness, Expertise and

Attractiveness of celebrity endorsers.

Appendix XIII

Reliability test of the impacts of celebrity endorsement

Appendix IX

Independent t- test for impacts of celebrity endorsement

LIST OF TABLES
Table 3.1

Names of celebrity generated from preliminary test

Table 3.2

Names of celebrity after elimination

Table 3.3

Semantic differential scale in Source credibility model

Table 3.4

5-point Likert scale

Table 4.1

Reliability statistics (i)

Table 4.2

KMO and Bartlett's Test table

Table 4.3

Total Variance Explained table

Table 4.4

Rotated Component Matrix Table

Table4.5

Mean table of attributes in Source credibility model

Table4.6

Independent Samples Test- Equality of means table

Tale 4.7

Paired Samples Statistics table, with means

Table 4.8

Paired Samples Correlations table (i)

Table 4.9

Paired Differences table (ii)

Table 4.10

Report table, by genders

Table 4.11

Independent Samples Test table

Table 4.12

Paired Samples Statistics table (ii)

Table 4.13

Paired Differences table (iii)

Table 4.14

Report table of Independent Samples Test

Table 4.15

Reliability statistics (ii)

Table 4.16

Average mean table, by genders

Table 4.17

Celebrities rating table

Table 4.18

Regression Model Summary table (i)

Table 4.19

Regression Model ANOVA table (i)

Table 4.20

Coefficients table (i)

Table 4.21

Regression Model Summary tablem(ii)

Table 4.22

Regression Model ANOVA table (ii)

Table 4.23

Coefficients table (ii)

Table 4.24

Regression Model Summary table (iii)

Table 4.25

Regression Model Summary ANOVA table (iii)

Table 4.26

Coefficients table (iii)

Table 4.27

Regression Model Summary table (iv)

Table 4.28

Regression Model Summary ANOVA table (iv)

Table 4.29

Coefficients table (iv)

Table 4.30

Regression Model Summary table (v)

Table 4.31

Regression Model Summary ANOVA table (v)

Table 4.32

Coefficients table (v)

Table 4.33

Regression Model Summary table (vi)

Table 4.34

Regression Model Summary ANOVA table (vi)

Table 4.35

Coefficients table (vi)

Table 4.36

Top3 favorite celebrities table

Table 4.37

Descriptive Statistics table of 3 favorite celebrities

Table 4.38

Reliability statistics (ii)

Table4.39

Report table of Independent Samples Test

Table4.40

Report table of Independent Samples Test (ii)

Table4.41

Report table of Independent Samples Test (iii)

Table4.42

Report table of Independent Samples Test (iv)

LIST OF FIGURES

Figure 2.1

Consumer decision process model (Blackwell, et al. 2006)

Figure 2.2 1

5-point Source-credibility scale (Ohanian, 1990)

Figure 3.1

Conceptual Framework

Figure 3.2

Example of error

Figure 3.3

Example of modification of error

Figure 4.1

Gender of respondents

Figure 4.2

Age of respondents

Figure 4.2b

Age of respondents, by genders

Figure 4.3

Education levels of respondents

Figure 4.3b

Education level of respondents, by genders

Figure 4.4

Occupations of respondents

Figure 4.4b

Occupations of respondents, by genders

Figure 4.5

Average monthly income of respondents

Figure 4.5b

Average monthly income of respondents, by genders

Figure 4.6b

Most credible source of celebrity endorsement, by genders

Figure 4.6

Most credible source of celebrity endorsement

LIST OF FIGURES

Figure 2.1

Consumer decision process model (Blackwell, et al. 2006)

Figure 2.2 1

5-point Source-credibility scale (Ohanian, 1990)

Figure 3.1

Conceptual Framework

Figure 3.2

Example of error

Figure 3.3

Example of modification of error

Figure 4.1

Gender of respondents

Figure 4.2

Age of respondents

Figure 4.2b

Age of respondents, by genders

Figure 4.3

Education levels of respondents

Figure 4.3b

Education level of respondents, by genders

Figure 4.4

Occupations of respondents

Figure 4.4b

Occupations of respondents, by genders

Figure 4.5

Average monthly income of respondents

Figure 4.5b

Average monthly income of respondents, by genders

Figure 4.6b

Most credible source of celebrity endorsement, by genders

Figure 4.6

Most credible source of celebrity endorsement

CHAPTER 1
INTRODUCTION

CHAPTER 1 INTRODUCTION
This introductory chapter aims to present an insight into the research area by briefly
discussing the background about celebrity endorsement, significance of the research,
objectives, scope and outline of study.

1.1 Background of study

Nowadays celebrity endorsements are commonly used strategies for promoting various
types of products, services or brands, especially in fashion and cosmetic industry. In Hong
Kong, most of the fashion brands are endorsed by artists, singers or young models. It is
because celebrity endorsement can arouse public awareness and peoples interest both
effectively and efficiently. Due to the expansion of China market there is a trend for
international fashion companies to use Chinese and Hong Kong celebrities to be their
endorsers. It helps corporations to build up brand awareness and to gain acceptance while
adopting local celebrities.

In the past only one-eighth of the advertisements showed the celebrities but the situation
has changed nowadays. In the United States and United Kingdom, 20%-25% of the
advertisements showed a celebrity (Shimp, 2000; Erdogan, Baker, and Tagg, 2001; Mistry,
2006); 57% in Korea (Choi, Lee and Kim, 2005); and 70% in Japan (Money, Shimp and
1

Sakano, 2006). On the other hand, around 40% of the youth-products advertisement
featured at least one celebrity in China (Chan, 2008). These figures support celebrity
endorsements can bring about a more favorable rating to the advertisement and product
evaluation (Dean and Biswas, 2001), as well as the financial returns for companies which
use celebrity endorsement for promotion (Erdogan, et. al. 2001).

However it found that 90% of the celebrity endorsement in Asia is done by 1% of the
celebrities (Marketing-interactive, 2012). So it is very important to grab consumers
attention and interest among those similar advertisements by using a celebrity who has
unique characteristics that can truly arouse consumers purchase intention.

1.1.1 Situation of celebrity endorsement in Hong Kong

As both international city and gateway to China market, Hong Kong attracts many
overseas and international companies to enter into her market. In Hong Kongs fashion
industry, celebrity endorsement is not a new phenomenon whereas its trend is increasing.
This is because celebrities can generate brand awareness and it can recall brand attitudes,
brand equity, emotional attachments to celebrities, as well as purchase intentions.
Although Hong Kongs consumer market is mature and almost saturated, the fashion
marketers are in severe competitions for grabbing consumers attention. As a result, most
2

of the fashion brands at least have one celebrity endorsers, either is an artist, singer or
young model. According to The Hong Kong Advertisers Association (HK2A, 2006), in
such a celebrity culture and a growing advertising market, there is an increasing number
of advertisers who consider to employ celebrities as spokespersons of their
advertisements.

Moreover, a celebrity-chasing phenomenon has been emerging in China since her growth.
The Chinese consumers, especially the young consumers are keen to buy products which
are endorsed by celebrities, as long as they are famous and attractive (Liu, Huang and
Jiang, 2007). Celebrity endorsement has been portrayed as a status symbol to China
consumers, therefore many fashion marketers would like to use Hong Kongs top
celebrities in their advertisements in order to attract individual visits consumers from
China. In 2011, Mainland China continued to be the largest visitor source market of Hong
Kong with 28.1 million arrivals, which accounts for 67% of total arrivals (Tourism
Commission, 2012). As a result, it provides a favorable condition for the fashion
marketers to target the Chinese tourists by using celebrity endorsement.

1.1.2 Significance of the research

Consumers are over-exposed to the media nowadays. A person living in an over populated
country will be expose to one thousand advertisements on average per day. But the
3

viewers tend to forget 80% of the information within 24 hours. In order to deal with this,
marketers use celebrity endorsement to reinforce the marketing message.

As a celebrity endorser, they must certainly have some unique factors which allow people
to remember them, for instance they are good looking. On the other hand, consumers
usually interpret the source of marketing messages by judging whether the celebrity
endorsers are trustworthy, expert and attractive. However the relationship between level
of good-looking, trustworthiness, expertise, attractiveness, especially the quality of
selecting the celebrity endorsers have not been investigated in the literature. Therefore, it
is important to conduct a research which aims to find out the above effects and
relationships. Finally, we can summarize and recommend an effective way for marketers
to use the right celebrities which can arouse consumers purchase intention in fashion
marketing communication.

1.2 Objectives
By gaining the insights from previous research and news about the celebrity endorsement
in Hong Kong, this study would focus on consumers purchase intention under the
impacts of celebrity endorsement, especially the impacts of source credibility of celebrity
4

endorsement. Moreover, the relationships between good-looking, trustworthiness,

expertise and attractiveness of the celebrities would be examined. Those impacts to
genders would also be investigated. As a result, we are able to understand male and
female consumers preference and perceived difference on celebrity endorsement. Finally,
we will recommend some effective strategies to the fashion companies who are using
celebrity endorsement. So they are able to choose the right celebrity and bring positive
impact to consumers purchase behaviour. In particular, this study seeks to answer the
following questions:
To study the impact of source credibility of celebrity to consumers purchase
intention.
To compare the effectiveness and impacts of celebrity endorsement on male and
female.
To understand the perceived image difference of celebrity endorsers between male
and female.
To indentify the quality of a popular celebrity endorser.
To understand the preference of celebrity endorsers between male and female.
To understand the relationships between good-looking, trustworthiness, expertise and
attractiveness of celebrity endorsers between male and female.

1.3 Scope of study

The research aims to identify the effects of celebrity endorsement in fashion apparel
products to consumers purchase intention in Hong Kong. In order to retain the integrity
of research and investigate the effects of celebrity endorsement to different age groups,
the questionnaires will be delivered to all ages of male and female consumers who are
Hong Kong residents. Moreover the research focuses on the possible difference between
male and female consumers in terms of perceived image of celebrity endorsement, source
credibility, impacts of celebrity endorsement and quality and preference of a celebrity
endorser.

1.4 Outline of the study

Chapter1 aims to provide insights for the research area by briefly discussing the
background, objectives, the scope of study, and the outline of the research. Chapter 2 aims
to provide relevant concepts and detailed literature reviews about celebrity endorsement,
source credibility, purchase intention and consumer behaviour and other related subjects.
Chapter 3 aims to explain in detail the conceptual framework and the methods which will
be employed for data collection and analysis. It aims to develop research hypothesis
based on the research topic too. Chapter 4 aims to illustrate the result of data analysis and
findings. In Chapter 5, it is to discuss findings and limitations, draw conclusions and
6

make suggestions for future research.

1.5 Chapter summary

This chapter gives a clear background and overview to the subject matters, which also
provide an initial understanding and brief insight of the problem area which study in the
research. By setting the objectives for the research, it also helps to clarify the purposes,
values and contributions about the research.

CHAPTER 2
LITERATURE REVIEW

CHAPTER 2 LITERATURE REVIEW

2.1 Introduction
This chapter aims to review the literatures which relate to the area of communication in
marketing, advertising, celebrity endorsement, psychology, as well as consumer
behaviour. Moreover, some relevant research about celebrity endorsements in Hong Kong
would also be examined. The relevant theories and models were introduced in order to
provide insights and deeper understanding about the research topic.

2.2 Consumer behaviour

2.2.1 Definition of consumer behaviour
Consumer behavior referred to the study of the processes involved when individuals or
groups select, purchase, use, or dispose of products, services, ideas, or experiences to
satisfy needs and desires (Solomon, 2006). If the company can understand consumer
behaviour well, they will able to develop good business. For instance, they must
understand the way to satisfy the consumers needs, gain more knowledge and data about
customers. So they could define the market and identify threats or opportunities to a
brand easier (Blackwell, Miniard, and Engel, 2006).

2.2.2 Consumer decision process

2.2.2.1 Consumer decision process model
The consumer decision process model illustrated a roadmap of consumers minds that the
marketers and managers could use to help and guide product mix, communication as well
as sales strategy. The model captured the activities which would occur when decisions are
made in a schematic format. It also showed how different internal and external forces
would interact and affect consumers thinking, evaluation, and acts. Furthermore, it could
help people to solve the problems which lead them to make a purchase and consume the
products (Blackwell, et al. 2006). Thus, culture, social class, personal factors, group
association, celebrities and celebrity endorsement advertising were some of the examples
of external forces. For the internal forces, it included own attitudes, personality,
perceptions, self-concepts, emotions, knowledge, value and lifestyle.
Meanwhile, the steps of consumer decision process included, problem recognition,
information search, evaluation of alternatives, and product choice. After that were
consumption, post purchase evaluation and divestment. As shown in Figure 2.1.

Figure 2.1 Consumer decision process model (Blackwell, et al. 2006)

2.2.2.1 Need recognition

First of all, significant differences were existed between the current state of affairs and
the ideal state. So consumers recognized there was a need to find the product or service
which could deliver some benefits (Blackwell, et al. 2006). For instance, the product or
service had the ability to fill consumers needs or to solve the problems which was worth
more than the cost of buying it. On the other hand, opportunity recognition could be
occurred when consumers were exposed to different or better quality products (Solomon,
2006).
In fact, need recognition could occur naturally or induce by marketers efforts. The
10

marketers created primary and secondary demand for the consumers. Primary demand
referred to the consumer was encouraged to use a product or service regardless of the
brand which they chose. Secondary demand referred to the marketer persuaded the
consumer to use specific brand of product (Solomon, 2006). For example, when
consumers were exposed to the advertisement which the celebrity endorser wear the latest
fashion and look beautifully, it might enlarge the difference between consumers current
state and ideal state. Then a need was created. As a result, they would like to make a
purchase on the apparel products which featured in the advertisement which endorsed by
the celebrity.

2.2.2.2 Information search

After the consumers recognized a need was existed, they needed to acquire sufficient
information to resolve it. So the process of information search was necessary for them to
survey on their environment to collect appropriate data to make a reasonable decision.
Information could be searched internally or externally. Internal search of information
referred to scan and retrieve the decision-relevant knowledge which stored in our memory,
such as experience. External search referred to the information was collected from the
external environment. Consumers were able to gather information from the source which
was non-marketer-dominated and marketer-dominated. Non-marketer-dominated source
11

was simply come from word-of moth, which was considered as an objective source. For
instance, it included family, friends, opinion leaders, reference group, consumer reports
and government and industry reports. On the other hand, marketer-dominated source was
anything that the marketers did for the purpose of persuasion, which included
advertisement in fashion magazines, catalogs, books and newspaper, salespeople, fashion
show, web-site and window and in-store displays (Blackwell, et al. 2006). Besides, the
intensity of information research depended on the risk which consumer perceived, such as
the cost of wrong choice.
Furthermore, celebrity was one of the common information sources for consumers. It is
because celebrity acted as a reference group for the consumers, which could influence
individuals evaluations, aspirations, or behaviour (Park and V. Parker, 1977). Consumers
might imitate the behaviors of their admired celebrities in order to enhance self-esteem
through identification with certain desirable images (Alsmadi, 2006). So celebrity
endorsement was one of the persuasive sources of information which might affect
consumers purchase decision.

2.2.2.3 Pre-purchase evaluation of alternatives

After the search process, there would be possible amounts of alternative options which
generated by the consumers. They believed the selected products or brands were able to
12

resolve their needs. Because of limited resources, consumers had to narrow down their
choices by assessing some standards. They might use pre-existing evaluations, prior
purchase or consumption experience which stored in memory to assess the selected
options that could provide the greatest satisfaction with the purchase and consumption.
They could also rely on the experiences or impressions gained second-hand. Moreover,
they could construct new evaluation standards base on the information they search in the
previous process. Although different consumers had different standards in the evaluation
process, the decision rule was usually based on appropriate to personal style, economy,
aesthetics, quality, other-people-directed, country of origin and so on. For example, the
consumer was totally enthralled by the celebrity which endorsed in the advertisement.
Then the consumer would evaluate the products based on the celebrity endorser.

2.2.2.4 Purchase
After narrowed down the product choices, consumers could purchase the products by
choosing a specific retailer and in-store choice. Meanwhile, consumers must decide
whether, where, what, when and how they buy. Usually, they would like to choose a store
which the image matched with their personality and purchase characteristics. Their past
experience was able to influence their store choice too. Even in the stage of purchasing,
consumers purchase intention still could be influenced by several factors. For example,
13

in-store promotions, discounts, persuasion of salespeople, failure to find the product in a

store or lack of financial resources. As a result, if consumers saw an in-store
advertisement with their admired celebrity, it could reinforce their purchase intention.

2.2.2.5 Consumption
After the consumer purchased and owned the product, consumption would be occurred.
Consumption was the process of utilizing the purchased product or service to satisfy the
needs of consumer. However, it did not guarantee consumers needs could be satisfied. It
is because positive and negative feeling could be resulted from consumption. Positive
feeling resulted in positive reinforcement, which occurred when consumer received
positive outcome from product usage. Besides, negative reinforcement could occur when
consumption helps to avoid negative outcomes. On the contrary, negative feeling resulted
in punishment, which referred to consumption lead to negative outcomes. As a result, it
was a key for the companies to enhance positive feeling while eliminate negative feeling.
So a pleasant and positive experience could be created.

2.2.2.6 Post-consumption evaluation

Consumption was an important determinant of the level of satisfaction. Consumer was
satisfied by the product, if their expectation was matched by the perceive performance.
On the contrary, consumers felt dissatisfaction if their experiences and performance fell
14

short of expectations. In other words, customer satisfaction was very important because it
might affect whether consumers would remain loyal and repeat buying the product or not.
So the company must ensure the product performance by creating positive consumption
feeling to the consumers. Furthermore, satisfaction was an important determinate factor
for shaping word-of-mouth and word-of-mouse communication. As a result, consumers
could communicate with others about their consumption experiences, no matter positive
or negative. Finally, it might affect the purchase decisions of the others. In sum,
companies must try their hardest to meet and exceed consumers expectations to avoid
losing good views.

2.2.2.7 Divestment
At the end, consumers would dispose the product after its useful life. Normally, there
were three types of divestment, which included disposal, remarketing and recycling.
Since the fashion trend was always changing, so disposal of apparel items were
commonly found. Besides, there were some celebrities in Hong Kong who engaged in the
business of selling second-hand clothes. Many consumers, especially the young people
were adored to these clothes. It is because the clothes were from their favorite celebrities,
the price was attractive and to support environmental protection.

15

2.3 Gender
2.3.1 Definitions of gender
Gender referred to the physical attributes that differentiate males and females (Phang and
De Run, 2007).Every society had their own set of expectations about the appropriate
behaviors for males and females (Solomon, 2006). People were well-trained by the social
norm since they were very young. They tended to behave according to their genders,
which included the way they acted, spoke and dressed. However it did not determine male
and female will follow the characteristics which are stereotypically associated with their
gender. In the society, there were gender differences which the marketers had to further
investigate.
In addition, it was a common demographic variable, which existed in almost all the
researches but only casual mentioned in the respondents personal profile. In other words,
it seldom fully utilized in the research which related to celebrity endorsement. As a result,
it provided a room for the research to focus on the possible gender differences among the
perception and selection of celebrity endorsers.

2.3.2 Characteristics of male

Males were more likely to make emotional buying base on partially digested information
(Elliott, 2006). They did not need to acquire a lot of information or details when making a
purchase decision. Besides, males viewed shopping as purchase-driven activities. They
16

would go shopping only when there was a need. According to the research done by a
famous global research agency, Millward Brown (2011), males were easily to be appealed
by humor, distinctive creative styles, and sexual imagery in the advertisement. Besides,
they preferred the marketing message which was straight to the point rather than
complicated. It was also found that colors and pictures in the advertisement could grab
males attentions. Meanwhile, in Germany, males liked the women which featured in the
advertisement than female did (Millward Brown, 2011).

2.3.3 Characteristics of female

Females tended to be more sensitive and careful to details of the information. So they
preferred detailed advertisements in order to get a clear message. They also favored
objectives over subjective claims but it would be better if the marketer were able to
combine the emotional and rational factors (Elliott, 2006). Furthermore, they enjoyed the
advertisements which featured with children or a slice of life. Besides, female hated the
advertisements for being sexist or erotic (Millward Brown, 2011).

Some researchers investigated females were far more influenced by the experts and expert
advertising than males (Aronson, 1972). Besides women were more fashion conscious
and bigger spenders in fashion than males (Goldsmith, Freiden, and Kilsheimer, 1993).
17

Furthermore, women had been shown to score higher on the opinion leadership and
fashion innovativeness than males (Stith and Goldsmith, 1989). On the other hand,
women were more likely to go shopping and usually exhibited positive attitude towards
shopping. But male hated shopping. (Alreck and Settle, 2001; Falk and Campbell, 1997)

In addition, women viewed shopping as enjoyment which could satisfy their needs at the
same time, especially went for fashion shopping (Falk and Campbell, 1997). As the
purchase behaviour of females was totally different from males, so the possible
differences among genders were a critical factor to marketers as well this research.

2.4 Purchase intention

2.4.1 Definition of purchase intention
Purchase intention was the probability of consumer to buy a product (Burton,
Lichtenstein, Netemeyer, and Garretson,1998) or ones conscious plan to make an effort
to purchase a brand in the future (Spears and Singh, 2004). It was also a component of
consumer cognitive behavior on how consumer intended to buy a specific product,
service or brand (Kwek, 2010).
On the other hand, customers' consideration and expectation in buying a brand were the
variables which could be used to measure purchase intention. For example, customers
18

interest, attending, information and evaluation were the consideration factors which
determined purchase intention (Laroche Zhou, 1996).
Furthermore, consumers purchase intention was easily affected by many external factors
during their decision processes. For instance in the processes of information search, they
were easily affected by word-of-mouth and word-of-mouse communication. During the
purchase process, their purchase intention would be affected by the product attributes,
price, and endorser performance which perceived by consumers (Sheu, 2010). Since
consumers purchase intentions and arousal were easily triggered by the promotion
strategies, so their attitude towards the promoted product, service or brand was also easily
affected by the popularity and image of the celebrity. Especially when the consumers
treated the celebrities as reference group, they wanted to imitate them in an effort to feel
successful too (Forbes, 2011). As a result, consumers purchase intention would positively
influenced by their admired celebrities.
In sum, the study of the effects on purchase intention was complicated. It is because there
were many variables which might have effects on it. Besides, there would be some
possible hidden variables which would affect consumers purchase intention too.

19

2.5 Celebrity endorsement

2.5.1 Definition of celebrity endorsement
Celebrity endorsements could be traced back to a print advertisement for Waltham
watches which endorsed by Reverend Henry Ward Beecher in Harpers Weekly at 1870s
(Subhadip, 2006).
Celebrity was a person whose name could grab public attention, arouse public interest and
generate profit from the public (Gupta, 2009). Hence, the celebrity endorsement was the
use of famous people to endorse a particular product, service or brand. Celebrity
endorsement was very useful, which could help to gain sales, increase brand awareness,
create positive feelings towards the brand, entertain the customers, as well as recall the
brand value. In other words, by getting famous people to represent a brand could gain a
higher degree of attractiveness, believability, and recall when comparing with those
unknown models (Kambitsis, Harahousou, Theodorakis, and Chatzibeis, 2002).

2.5.2 Forms of celebrity endorsement

There were certain forms of celebrity endorsements, which included print advertising in
magazines, television advertising, products used in movies and television programs,
photographs of paid celebrities using the products bearing the endorsed brand,
photographs of unpaid celebrities using the products bearing the endorsed brand, mention
20

of luxury brand in music, inviting celebrities to be co-creators in designing products and

naming products after celebrities (Eshaghpour, 2010). The companies should select the
form of celebrity endorsement which suit their promotion strategy, budget and brand
image the best. So the company was able to reach the target customers and delivered the
marketing message in an efficient way.

2.5.3 Benefits of celebrity endorsement

Though celebrity endorsement was expensive, there were many benefits which celebrity
endorsement could bring about. So it was still a commonly used strategy. By using the
celebrities public recognition, the companies were able to gain many benefits. (Mcaleer,
2010)
Firstly, it could add instant credibility. Some of the consumers believed that if a celebrity
was willing to use a particular product, service, or brand, the company who produced the
product or service must have certain degree of high quality.
Secondly, it could grab instant attention. With celebrities to endorse particular kind of
product, service or brand, the consumers attentions were most likely to be grabbed by the
celebrities and willing to pay more attention to watch.
Thirdly, it could enhance media exposure. With celebrity endorsement, the media would
report the news about the celebrity who participated in the events. So the companies were
21

able to get free publicity in the newspaper, magazine or TV. Moreover consumers liked to
participate in the events which their favorite celebrities would be involved. It helped to
increase brand awareness too
Thirdly, it could receive superior market presence. Products which were endorsed by
celebrities were relatively easier to get in stores and being displayed in a good location.
Because of consumers acceptances of the endorsed product, service or brand increased,
they were willing to pay for the goods.
Fourthly, celebrity endorsement might materially improve the financial returns for the
companies who employed celebrities as an investment in the advertising campaigns
(Farrell, Karels, Monfort and McClatchey, 2000; Erdogan et al., 2001).
Fifthly, celebrity endorsement strategy was an effective way to differentiate among the
similar products in the market. It is because consumers were better able to identify the
products by associating with the endorsers (Burroughs and Feinberg, 1987). But it did not
work if the celebrity endorsed had already endorsed several brands of product. Besides, it
was a useful strategy when consumers did not perceive many differences among
competitors. So the company was able to create differentiation by using celebrity
endorsement, especially during the mature stage of the product life cycle (Solomon,
2006).
In conclusion, celebrity endorsement could increase the brand awareness, reach the target
22

market effectively and efficiently, create a positive feeling towards the brand, deliver
marketing message, generate immediate attention within short period of time and gain a
profit. Last but not least, celebrity endorsements had a positive influence on purchase
intentions (Karina, P. R., 2008). Thats why celebrity endorsement was one of the most
effective ways of establishing a long-running brand building campaign (Pringle, 2004).

2.5.4 Celebrity endorsement as reference group

Reference group was one of the important influences during consumers pre-purchase
evaluation process. Reference group was any person or group of people which could
influence ones behaviour significantly (Bearden and Etzel, 1982). Consumers often
compared themselves with the others or a group of people which they did not interact but
treat as a reference. Celebrities, athletes, successful businessmen or performers were one
of the reference groups whose consumers did not know but admire, which was
aspirational reference groups (Solomon, 2006). Value-expressive influence occurred when
the consumers used the groups position to bolster his/her ego through identification or
association with the group.
Celebrity endorsers were interpreted as reference group because consumers would like to
be associated with them. They would like to behave in the same manner as the celebrity
endorsers because of symbolic meanings. In other words, celebrity endorsements served
23

as a role model and were able to trigger consumers feeling of it will be nice to look like
the celebrity if I use their endorsed fashion brand or I can enhance my image and show
the others I look like a star. As a result celebrity endorsement played an important role in
influencing consumers purchase intention and buying behaviour.

2.6 Source credibility

2.6.1 Definition of source credibility
Source credibility was the degree to which the receiver would believe the source has
certain degree of relevant knowledge and/or expertise and they choose to believe the
information offered by the source (Ohanian, 1990). At first, source credibility simply
meant endorsers credibility in an advertisement (Aronson, Turner and Carlsmith, 1963).
Eventually it was considered as an important factor which might affect consumers
purchase intentions and attitudes toward advertising (Lutz, MacKenzie, and Belch, 1983).
The literature identified three dimensions as components of source credibility:
Trustworthiness, Expertise and Attractiveness (Baker and Churchill, 1977; Giffin,
1967; Hovland, Janis, and Kelley, 1953; Joseph, 1982; Kahle and Homer, 1985; Maddux
and Rogers, 1980; Mills and Harvey, 1972).

2.6.2 Source credibility model

Ohanian (1990) proposed that the effectiveness of celebrity endorsement very much
24

depends on the endorsers perceived attractiveness and perceived credibility, which are
trustworthiness and expertise. So the source credibility was an important construct of this
research. This model was widely used by researchers when they were studying the
effectiveness of celebrity endorsement. As shown in Figure 2.2.

Figure 2.2 15-point Source-credibility scale (Ohanian, 1990)

In the model, there were three main dimensions which were,

Trustworthiness, which included not dependable/dependable, dishonest/honest,
unreliable/reliable, insincere/sincere, and untrustworthy/trustworthy.
Expertise,

which

included

not

expert/expert,

inexperienced/experienced,

unknowledgeable/knowledgeable, unqualified/qualified, and unskilled/skilled.

Attractiveness,

which

included,

unattractive/attractive,

not

classy/classy,

ugly/beautiful (or handsome), plain/elegant, and not sexy/sexy.

Thus, the above three dimensions were the credibility source of celebrity endorsement. In
the model, each dimensions had five adjectives along with 5 sub-scale scores. It was
25

using the semantic differential scale for measuring individuals concepts to a particular
celebrity endorser. According to Ohanian (1990), name of a specific celebrity endorser
would be provided. Next, respondents were required to rate the celebrity endorsers by
using the set of provided adjectives. By combining different perspectives, it helped
researchers to examine the credibility level of celebrity endorser and effectiveness of
celebrity endorsement. Furthermore, it helped to determine the most persuasive
dimension which could induce and influence consumers purchase intention. So this
model was very important in this research.

2.6.3 Trustworthiness
2.6.3.1 Definition of trustworthiness
Trustworthiness was the degree of confidence in the communicator's intent to
communicate the assertions which he or she considered the most valid (Hovland, 1953). It
also meant consumer's confidence in the source for providing information in an objective
and honest manner (Ohanian, 1991) In other words, if the communicator was perceived as
highly trustworthy, the opinionated message would be considered as more persuasive and
effective in producing attitude change (Miller and Baseheart, 1969). Moreover, the
perceived trustworthiness had been shown to produce a greater change than the perceived
expertise (McGinnies & Ward 1980). Trustworthiness was also an important predictor of
26

celebrity endorsement effectiveness.

2.6.3.2 Effects of trustworthiness

In early year, Miller and Baseheart (1969) investigated the way source trustworthiness
affected the persuasibility of the communication. They found that when the
communicator was perceived as highly trustworthy, the opinionated message was
considered as more effective in producing attitude change. As a result, when a person
wanted to buy a product, they would consider a good friend trustworthy on most matters.
On the other hand, endorsers or sales people often give professional knowledge and
opinion on particular kinds of products to consumer than the inexpert friends do.
Harrison-Walker (2001) believed that the consumers preferred to trust and to be guided by
information from friends contacts rather than a companys formal promotion mix.
Although celebrities were usually viewed as highly credible and trustworthy, Tripp,
Jensen and Carlson (1994) thought that if the celebrities who had already endorsed
several kinds of products would be viewed as less credible and trustworthy than those
who had endorsed only one product. Furthermore, the celebrities who were blamed for
some negative events could have harmful effects on the products they endorse too. (Louie
and Obermiller, 2002)

27

2.6.4Expertise
2.6.4.1 Definition of expertise
Expertise was the extent to which a communicator was perceived to be a source of valid
assertion. They usually supported by professional knowledge and skills thus their claims
about the product became a factor that increased the persuasiveness to consumers. Some
researchers believed that the endorser with expertise would be more potent than those
only have physical attractiveness (Till and Busler 1998).

2.6.4.2 Effects of expertise

Expertise was the person who gave useful and creditable advice or opinion to the
consumer. They usually supported by professionals who possessed relevant knowledge
and skills. The claims about the product could become a factor that increase the
persuasiveness and beyond the effects of trustworthiness, if the celebrity endorsers had
professional product knowledge (David H. Silvera, Benedikte Austad, 2004). According
to Miciak and Shanklin (1994), if a company wants the celebrity endorsers to be seen as
credible then the consumers must perceive them to be trustworthy and have the expertise
to speak about a product or a service.As a result the consumers could be better induced
by the endorsers and buy the products. Some researchers even believed that the endorser
with expertise might be more potent than those only have physical attractiveness (Till and
28

Busler 1998).
Furthermore, consumers reactions in response to the recommendations were varied
directly with the sources perceived level of expertise and the target persons level of
agreement with those recommendations. When consumers exposed to a source perceived
as high expert, they would exhibit a higher level of agreement with the sources
recommendation than did those exposed to a source with lower level of expertise
(Ohanian 1990). As a result, the level of perceived celebrity expertise could be use to
predict the effectiveness of celebrity endorser.

2.6.5 Attractiveness
2.6.5.1 Definition of attractiveness
Attractiveness was something that could arouse the interest of most people. It was not
limited to the size of the body, but it encompassed all of ones physical attributes. For
example, hair color, facial features and physical appearance. Moreover, endorsers
personality, likeability, similarity to the receiver and receivers perceived social value
would also affect the interpretation of attractiveness. Attractiveness of the celebrity
endorsers could be affected by ones thoughts, beliefs, attitudes and the society (Temple,
2009). Many researches in advertising and communication suggested that physical
attractiveness was an important cue in how an individual judge another person at first
29

sight. It is because beauty usually made a better first impression. Because of increasing
use of celebrity endorsement, attractiveness became an important dimension of source
credibility (Ohanian, 1991).

2.6.5.2 Effects of attractiveness

Attractive endorsers tended to be more easily and effectively in product promotion and
grabbing consumers attention (Till and Busler, 1998). Quoting Aristotle, Beauty is
greater recommendation than any letter of introduction. So the highly attractive models
and celebrity endorsers were easily linked the source credibility. As a result, consumers
believed that physical attractiveness was the most important criteria of source credibility
nowadays. Therefore they were commonly featured in mass media and print media, in
order to attract consumers attentions. In other words, physically attractive celebrities
were an important predictor of advertising effectiveness (Till & Busler 2000).
On the other hand, attractive people had greater influence on the consumers compared to
unattractive people. As a result, attractive endorsers were more easily and effectively in
product promotion and attracting consumers attention. Apart from fashion industry,
cosmetics industry was also demanding the celebrity endorsers who were highly attractive
and looked beautifully. Researchers found that female consumers who idealized the
women in traditional roles were likely to use cosmetics in the pursuit of beauty (Forbes,
30

Jung and Haas, 2006).

Within the broader context of celebrity endorsement, attractiveness of the celebrity
endorsers was an important construct. However, the nature and scope of the attractiveness
construct was still uncertain. Therefore additional attention should be pay to it.

2.6.6 Good-looking
2.6.6.1 Definition of good-looking
Good-looking was a subjective term. A person was considered as good-looking if he or
she had a pleasing appearance. The standard of beauty changed over time and across
cultures. Some African tribes would perforate their lips, ears, or noses in order to insert
shells, colored stones, or gems. In QING dynasty of China, females, especially the
nobility would bind their feet in order to fit in the tiny bow shoes. It is because it was an
important standard of beauty and symbol of status. Classical cultures of Greece and Rome
developed some standards and rules for measuring beauty, which based on proportion and
composition (Goldstein, 1998).
British scientists discovered that human infants were born with an innate concept of what
makes an attractive face (Discover Magazine, 2004). Physical symmetry and golden ratio
were good example of innate beauty. Apart from the innate beauties, ordinary people who
were healthy and clean, with confident and dress well, with a smile and reflect unique
31

personality could also acquire good-looking nowadays. As a result, there was an

increasing use of ordinary endorsers in advertisements.
Nonetheless, the situation in fashion industry did not change much. Innate good-looking
endorsers still dominated the market because beauty made a better first impression.
Therefore, they were considered as highly persuasive. Due to good-looking was very
subjective, it was hard to measure and consumers were hard to tell.

2.6.7 Modification of Source credibility model

In order to achieve a better result of evaluation, adjustments would be made when using
the source credibility model. Firstly, semantic differential scale would not be employed
but only positive adjectives would be used. Secondly, statements and situations would be
used when the credibility source of celebrity endorsement was being tested. Thirdly, any
specific celebrity endorsers or brand name would not be provided in order to test the
general reaction of celebrity endorsement to consumers purchase intention. By
modifying the model, it could examine and understand consumers purchase intention
under the perception of celebrity endorsement in general, not limit to a particular celebrity
endorser.
Meanwhile, good-looking would be added to test the model. It is because it could
examine if good-looking means the celebrity endorsers are trustworthy, expertise or
32

attractive. When good-looking was being tested, name of predetermined celebrity would
be provided. Therefore the kind of characteristics which the celebrity possessed could
help to understand consumers preference. It also helped to understand the criteria of
selecting celebrity endorsers.

2.7 Celebrity endorsement in Hong Kong

Celebrity endorsement is a commonly used strategy for many companies in the world. A
successful endorser can somehow increase the purchase intentions and buying behaviour
of a consumer. According to Shimp (2000), about 25 percent of the commercials in the
USA would use celebrity endorsers for promotion. The situation in Hong Kong is that,
there were an increasing number of advertisers who would like to employ celebrities as
spokespersons for their ads (HK2A, 2006). The Hong Kong Advertisers Association
believed that the increasing use of celebrities in advertising was a trend in the world.
Furthermore, there were some researches which supported celebrity endorsement could
result in facilitating product evaluations and receive better advertisement ratings (Dean
and Biswas, 2001) and had substantial positive impacts on financial returns (Erdogan,
2001).
On the other hands, there were several researches which studied the celebrity
phenomenon in Hong Kong It provided an insight to further study the effects of celebrity
33

endorsement in Hong Kong. Fam and Waller (2006) conducted a qualitative research to
investigate the likeability of advertisement in Hong Kong, Shanghai and other South-East
Asian countries. They found that the likeable advertisements in Hong Kong had the
elements of escapism and the celebrities are fun, features, humors, which were highly
entertaining.
Another research which conducted by (Ling, Gee and Elaine, 2008) focused on the
impact of celebrities on brand decisions to non-advertising students. The study aimed at
identifying celebrities attractiveness, meaning transfer, credibility, celebrities impact and
whether they could match up with products. The study found that actors and singers in
Hong Kong have a greater influence on university students brand choice due to transfer
images of fun, entertainment, and escapism.
In fact, the study in 2006 did not mention what were the criteria for selecting celebrity
endorsers. Still, the study in 2008 did not focus on the perceptual difference of celebrities
image between genders and the possible different of purchase intention among genders.
Moreover the impacts of credibility of celebrity endorsements to consumers in Hong
Kong are needed to be investigated too.
In summary, the previous research about the celebrity endorsements in Hong Kong
provided a good insight for this research. It also shown that, there was possible research
area for further investigating the celebrity endorsement phenomenon in Hong Kong.
34

Research questions would be based on the area which indentified.

2.8 Chapter summary

This chapter provided useful theoretical models, principles and information about
consumer decision process, genders, purchase intention, celebrity endorsement, reference
group, as well as the source credibility model. Moreover, some relevant researches about
the phenomenon of celebrity endorsement in Hong Kong were also discussed. As a result,
a research which was more applicable to the situation of celebrity endorsements in Hong
Kong could be conducted. Meanwhile the literature review served as guidelines for
construction of the questionnaire survey.

35

CHAPTER 3
METHODOLOGY

CHAPTER 3 METHODOLOGY
3.1 Introduction
The chapter aimed to explain the processes involved in conducting the research, from
research design, to report preparation, as well as setting up the hypothesis for the research.
Firstly, the nature of research would be identified. Secondly, the conceptual framework
would be introduced. Thirdly, the research hypothesis would be set up. Fourthly, the
research approach and data collection processes would be clarified. Fifthly, all the details
about the questionnaire would be discussed. Finally, some important techniques for data
entry and analysis would be introduced. By defining the research methods in a clear and
systematic manner, the research would be more organized and the mistakes could be
minimized, which was critical to the entire study.

3.2 Research design

3.2.1 Nature of research
This research was based on the exploratory and quantitative research in order to examine
the effects of celebrity endorsements in fashion apparel products to consumers' purchase
intention. To begin with the study, exploratory research aimed to gain insights, identify
and define the topic better. In the second stage, quantitative research would be applied in
order to further discover consumers responses towards the celebrity endorsement and
37

find out any hidden variables.

3.2.2 Exploratory research

In the exploratory research, it aimed to examine the effects of celebrity endorsement to
consumers purchase intention, both in Hong Kong and other countries. It helped to
identify and define the problem better. Moreover, it could develop an approach to the
research topic and interpret primary data more insightfully. In order to have a more
in-depth understanding about the existing topic, related materials such as articles, theories,
academic journals, and internet based source were used as well. So fundamental
knowledge and secondary data could be provided to research.

3.2.3 Quantitative research

Quantitative research approach was employed to the research. Questionnaires were
adopted in order to collect the quantitative data to examine the effects of celebrity
endorsement to consumers purchase intention in Hong Kong. It provided facts and
estimate which used to make accurate prediction about the relationship between celebrity
endorsement and consumers purchase intention. Therefore, it was able to gain insights
and understand the relationships between variables and possible differences between
groups. Details about the questionnaires design and setting would be discussed in the
38

following parts.

3.3 Conceptual framework

In this research, the main objective was to examine if source credibility of celebrity
endorsers had any effects on consumers purchase intention. A conceptual framework was
developed, which was presented in Figure 3.1. On the left hand side of the framework, it
illustrated the three dimensions of source credibility model, which were Trustworthiness,
Expertise and Attractiveness. On the bottom was Good-looking, which represented it
might have an impact on the source credibility model. Furthermore, source credibility
model and Good-looking were the factors which might affect the credibility of celebrity
endorsement and finally affect consumers purchase intention. By developing this
conceptual framework, the relationship between different components were illustrated
and summarized. It also facilitated the hypothesis setting.

Figure 3.1 Conceptual Framework

39

3.4 Research hypothesis

Ohanian (1990) proposed that effectiveness of celebrity endorsement very much
depended on the endorsers perceived attractiveness and perceived credibility, which
were expertise and trustworthiness. So the source credibility was the main focus in
hypotheses setting.

3.4.1 Impact of attractiveness of celebrity endorsers and purchase

intention.
Attractiveness was a unique quality which could arouse the interest of other people.
Consumers would more likely to demonstrate positive buying behaviors when they saw
the physically attractive endorsers in the advertisements. Many researchers proved that
the opinions of an attractive person are more likely to be agreed and accepted by the
others when compared to the less attractive one (Phau and Lum, 2000). Kahle and Homer
(1985) also proved that a positive relationship is found between physical attractive source
and persuasiveness in marketing communications. So it was common to find the
physically attractive celebrity endorsers or models in the fashion advertisement. It is
because the marketers believed that the attractive endorsers with attractive body image
and unique characteristics were more convincing when promoting fashion apparel
products. In short, an attractive celebrity endorser should have positive impact on
consumers purchase intention. Based on the above discussion, the following hypotheses
40

were made.

H1a. Attractiveness of celebrity endorsers will have less impact on consumers

purchase intention of fashion apparel products.

H1b. Different gender had different purchase intention under the impact of attractive
celebrity endorsement.

3.4.2 Impacts of trustworthiness of celebrity endorsers and purchase

intention.
Trustworthiness was related to the degree of actual believability and credibility of the
celebrity endorsers. When the endorser was viewed as highly trustworthy, their opinions
would be more convincing to the consumers. James Houran, a clinical psychologist
whose research targeted on trustworthiness of celebrities in Hollywood claimed that
people trust the celebrities because they felt that they knew the celebrities very well.
Owing to the fact that the celebrities were familiar and successful, therefore people
wanted to copy them in an effort to feel successful too (Forbes, 2011). In modern society
we were frequently exposed to the celebrity endorsement, thus we tended to like, trust and
accept it. As a result, consumers tended to trust the celebrity endorsers and perceived
highly trustworthiness towards the celebrities after repeatedly receiving the marketing
41

message from celebrity endorsers. Finally, consumers purchase intention and decision
were affected. Based on the above discussion, the following hypotheses were made.

H2a. Trustworthiness of celebrity endorsers will have less impact on the consumers
purchase intention of fashion apparel products.

H2b. Different gender had different purchase intention under the impact of
trustworthy celebrity endorsement.

3.4.3 Impacts of expertise of celebrity endorsers and purchase intention.

Expertise is degree which a communicator is perceived to be a source of valid assertion.
In fashion industry, the rules to fashion expertise are to know everything about fashion,
starting from fashion history to the latest trends. The fashion experts should know how to
mix and match the fashion apparel items and accessories. Moreover they should know
how to represent themselves and have already earned reputations from the public.
Celebrities are often perceived as highly fashionable and the leaders of the fashion trend.
As a result, the celebrity endorsers are able to increase persuasiveness. Ohanian (1991)
proved that celebrity expertise can make influence on purchase intentions. Thus it is
42

necessary to prove Ohanian (1991)s research can also apply to the situation in Hong
Kong. Based on the above discussion, the following hypotheses were made.

H3a. Expertise of celebrity endorsers will have less impact on the consumers
purchase intention of fashion apparel products.

H3b. Different gender had different purchase intention under the impact of expertise
of celebrity endorsement.

3.5 Research approach

After the hypotheses were set up, the research approach could be determined. In the study,
two stages were involved. In the first phase, fundamental theories and principles would be
studied in order to obtain basic understandings and know ledges for constructing the
research and conducting the questionnaire. Literatures and previous researches about
celebrity endorsements and consumer behavior were able to provide insights and idea
about the ways to construct the questionnaire. Especially the questions about celebrity
endorsements, source credibility, consumers purchase intention and related behaviors
would be developed based on the theories and conceptual frameworks.
43

The second stage was about the construction of questionnaire. The questionnaire would
be created after the preliminary test. In order to enhance the accuracy and reliability of the
questionnaire, pre-test and pilot test would be held. Self-administered online surveys and
intercept surveys were conducted as well. All the respondents would be volunteers
without any benefit transfer. Furthermore, the legal and ethical aspects of research with
the respondents would be adhered to. Through the survey, the objectives and hypotheses
about the research could be verified.

3.5.1 Data gathering process

Before set up the questionnaire, preliminary test should be conducted in order to select
ten suitable male and female celebrities in Hong Kong for examining the relationships
between good-looking, trustworthiness, expertise and attractiveness. Pre-test and pilot test
would be conducted afterwards to eliminate mistakes and refine the wordings which
employed in the questions.

3.5.2 Preliminary test

3.5.2.1 Selection of celebrities
The celebrities who possessed the characteristics of good-looking, attractive, trustworthy
and expertise were selected through several phases of preliminary test by using the
44

convenience samples of The Hong Kong Polytechnic University undergraduate students.

20 respondents were given five minutes to list as many celebrities name as possible, as
least had 5 male and 5 female. They were required to select the celebrity according to the
characteristics of good-looking, attractiveness, trustworthiness, expertise and who are
related to fashion. However they were not required to specify the characteristics.
Celebrity who had been chosen more than 10 times was presented in Table 3.1.

MALE CELEBRITY

FEMALE CELEBRITY

Daniel Wu (18)

Angela Baby (16)

Takeshi Kaneshiro (17)

Gaile Lai (15)

Nicholas Tse (15)

Jolin Tsai (15)

Wyman Wong (13)

Cecilia Cheung (14)

PakHo Chau (13)

Rainie Yang (14)

Lee MinHo(13)

Lin Hsi-Lei(13)

Chang Chen (13)

Hilary Tsui(12)

Edison Chen (12)

Charlene Choi (12)

Sam Lee (12)

Lin Chi-ling(12)

Louis Koo (11)

Lena Fujii (11)

Jackie Chan (11)

Josie Ho (11)

Table 3.1 Names of celebrity generated from preliminary test

The number in the bracket represented the number of times which the celebrity was being
chosen by the respondents. After the preliminary test, some of the celebrities would be
eliminated out of the pool based on some criteria, which will be discussed below.

45

3.5.2.2 Criteria for selecting celebrities

In the research, it was very important to identify ways to examine the perceived image of
celebrity endorsers and how they affected consumers purchase intention. As a result,
numerous celebrity names were generated from the preliminary test. Due to the necessity
of limiting the amount of celebrities which would be tested in the survey, some criteria
were made.
The celebrity is well-known in Hong Kong.
The celebrity is Hong Kong based.
The celebrity must be fashionable.
The celebrity who had negative publicity or scandal previously must necessarily be
eliminated.
After elimination, the celebrities who have top five votes would be chosen for this
research, as shown in Table 3.2.

MALE CELEBRITY

FEMALE CELEBRITY

Daniel Wu

Angela Baby

Takeshi Kaneshiro

Gaile Lai

Wyman Wong

Hilary Tsui

PakHo Chau

Charlene Choi

Sam Lee

Josie Ho

Table 3.2 Names of celebrity after elimination

The celebrities who were not able to fulfill the criteria were eliminated. The majority of
46

the remaining celebrities were fashionable, well known and Hong Kong based celebrities.
Moreover they did not have significant negative publicity and scandals. Therefore they
were able to use in the survey for examining the relationships between good-looking,
trustworthiness, expertise and attractiveness.

3.6 Questionnaire survey

Questionnaire was a formalized set of questions which could obtain the require
information from the respondents. It must translate the information need into a set of
specific questions that the respondents can understand and willing to answer (Malhotra,
2006).

3.6.1 Questionnaire design

In this research, structured questions such as multiple-choice and scale would be used.
Therefore it could specify the set of response alternatives and the response format. Before
the questionnaires were distributed to the public, pilot test must be held for improving the
reliability and accuracy of the questionnaire. In order to obtain sufficient amounts of
response for the research within a short period, self-administered online surveys and
intercept surveys would be conducted.

47

3.6.2 Questionnaire details

The questionnaire consisted of five parts in order to obtain sufficient information from the
respondents. These include their attitudes towards celebrity endorsements, the effects of
source credibility to celebrity endorsement, impacts to consumers purchase intention and
other related behavior and so on.

Part one
This part aimed to understand consumers purchase intention under the attributes of
celebrity endorsement (Q1-18). The Source credibility model (Ohanian, 1990) was
applied along with some modifications. At first, the model was presented in a semantic
differential scale, as illustrated in Table 3.3. Later on the model was being modified and
presented in statements with positive adjectives. For instance, Beautiful celebrity
endorsers can trigger my purchase to their endorsed fashion apparel products. Moreover,
no specific celebrity endorsers or brand name will be shown in the questions.

Strongly

Disagree

Neutral

Agree

Disagree

Ugly

Strongly
Agree

Beautiful

Table 3.3 Semantic differential scale in Source credibility model

Part two
This part allowed the respondents to select the most credible source of celebrity
48

endorsement from attractiveness, trustworthiness and expertise (Q19). The question was
made straightforward and easy to understand in order to minimize errors made from
misinterpretation.

Part three
The fourth part was the perceived image of celebrities in Hong Kong. Ten celebrities
name which generated from the preliminary test would be applied in this part. In terms of
good-looking, attractiveness, trustworthiness and expertise, respondents were asked to
rate the celebrity by using the semantic differential scale (Q20-23). Finally the
respondents were asked to select three favorite celebrities (Q24).

Part four
This part allowed the respondents to outline the impacts of celebrity endorsement. The
questions were designed base on several principles. Firstly, there were six questions
(Q25-30) which related to the consumer buying decision model except divestment.
Secondly, there were four questions (Q31-34) which used for measuring the effectiveness
of celebrity endorsements on fashion apparel products. Thirdly, four questions (Q35-38)
were designed to investigate the perceived value of the products and brands which
endorsed by celebrities. Finally, there were three questions (Q39-41) which used to
49

explore whether the attractiveness of the fashion apparel products would increase, if the
products were endorsed by celebrities.

Part five
The last part of the questionnaire was to collect general personal data from the
respondents. For instance, gender, age, educational level, occupation and monthly income
level, which would be useful to examine the relationships between variables and
demographics.

3.6.3 Measurement and scaling

The 5-point Likert scale was used in the questionnaire, as shown in Table 3.4. It required
the respondent to indicate the degree of agreement or disagreement with each of the
statement about the stimulus objects (Malhotra, 2006). It was easy to construct,
administer and understand. In the research, negative statements would not be used in
order to avoid mistakes make by the respondents and increase the accuracy of the survey.

1=Strongly

2=Disagree

Disagree

3=Neutral

4=Agree

5=Strongly
Agree

Table 3.4 5-point Likert scale

50

3.6.4 Pilot test

Before the distribution of the questionnaires, a pilot test must be held. The purposes were
to ensure the questions were able to administer in a real world environment by the
respondents. Moreover, pilot test helped to ensure the respondents would understood the
meanings of the questions, appropriateness of the response categories and clarified and
refined the questions, which could helped to increase the reliability and validity of the
survey.

In the pilot test, five printed questionnaires were distributed by using the convenience
samples in The Hong Kong Polytechnic University. Online questionnaires were
disseminated to friends and the feedbacks were immediately collected through telephone
conversations. Finally, twelve questionnaires were collected and some comments and
feedback were made by the respondents.
Respondents could hardly understand the adjectives used in part one (Q1-18)
because they were too similar to each other.
Respondents were confused about the English statements used in the questionnaires.
In the online questionnaires, several respondents missed out the question of
occupations in part five. Because the options were hidden by the list since the
question type was choose from a list, illustrated in Figure 3.3.
51

Figure 3.3 Example of error

Modifications were made in both printed and online questionnaire, according to the
comments and feedback provided by the respondents.
Although the adjectives used in part one were similar, they were the model
adjectives proposed by Ohanian (1990) for standard reference. In order to minimize
the chance to misinterpret, Chinese meanings were added to each statement.
To avoid the respondents missing out any question, Question type was change to
multiple choices, thus all the options could be presented as illustrated in Figure 3.4.

Figure 3.4 Example of modification of error

3.6.5 Sampling method and distribution of questionnaires

Due to limited resources, all the respondents participated in the survey were volunteers
and without any benefits transfer. Before the questionnaires were distributed, some
targets were set,
The sample size should be at least 100.
The proportions of male and female respondents must not be less than 6:4 or 4:6.
52

 The range of occupation collected should be as wide as possible.

The data must be collected within a week, from 6th February, 2012 16:00p.m to 13th
February, 2012 16:00p.m.

Convenience sampling, online sampling and snowball sampling were used in order to
achieve the above targets. Parts of questionnaires were distributed to the undergraduate
students in the Hong Kong Polytechnic University. For the online questionnaires, it was
constructed by using the spreadsheet in GOOGLE. The questionnaires were disseminated
throughout the internet social network and snowball sampling was triggered at the same
time. Finally, 45 copies of printed questionnaires were collected and 43 copies were valid.
132 copies of online questionnaire were collected and 129 of them were valid. The total
copies of valid questionnaires were 172. All other targets set were also satisfied, detailed
results would be discussed in the later chapter.

3.7 Data entry and analysis techniques

There were several ways to assist data entry and data analysis. Firstly, Microsoft Excel
2007 was used for the work of data entry. Then the data were further edited, processed
and analyzed by using the IBM Statistical Package of Social Science (SPSS) Statistics
19.0.0 for windows. Other useful analytical methods would also be applied for analyzing
53

the data.

I. Descriptive Statistics
Firstly, descriptive such as mean, standard deviation would be used for simple
comparison between variables. Secondly, frequencies and crosstabs would be able to
present the data into tables of counts, percentages, in terms of groups and univariate
statistics, including the mean, median, and mode (SPSS Inc, 2007).

II. Compare Means

Firstly, means were used to present the group means and related univariate statistics for
dependent variables within categories of one or more independent variables. Secondly,
several types of T-test such as one sample, independent samples, and paired samples
T-tests would be used for hypotheses testing (SPSS Inc, 2007).

III. Scale
Reliability test aimed to estimate the reliability statistics for the components of
multiple-item additive scales (SPSS Inc, 2007).

54

IV. Multiple-response
Frequency tables and cross tabulations for multiple-response data would help to analyze
multiple response questions (SPSS Inc, 2007).

V. Multiple-regression
It was a technique which functioned to explore the relationship between one continuous
dependent variable and a number of independent variables or predictors (usually
continuous). Multiple-regression was based on correlation, but it allowed a more
sophisticated exploration of the interrelationship among a set of variables.

VI. Dimension Reduction

Factors analysis would be used in order to identify the underlying variables, or factors,
that explained the pattern of correlations within a set of observed variables (SPSS Inc,
2007).

3.8 Chapter summary

In summary, conceptual framework, hypotheses, research types, questionnaires and
method of analysis were determined in this chapter. By carrying out the pilot test before
the distribution of questionnaire, it could enhance the quality of the questionnaires.
55

Moreover, in twenty-first century, there were great advances in the techniques and
technologies employed in survey research, from systematic sampling method, online
survey, in which they could also enhance the quality of questionnaire design and
computerize the data analysis process. For instance, the SPSS provided convenient and
powerful functions to users for analyzing data and generating graphics. Detailed data
analysis would be discussed in the next chapter.

56

CHAPTER 4
FINDINGS AND RESULTS

CHAPTER 4 FINDINGS AND RESULTS

4.1 Introduction
In this chapter, the collected data such as respondents perspectives on celebrity
endorsements and their effects to consumers purchase intention, source credibility of
celebrity endorsers and other important data were analyzed statistically by SPSS. The
detailed findings and analyses were made and illustrated by using the tables, charts and
graphs from SPSS.

4.2 Sample size and cleaning invalid data

Both self-administered online surveys and intercept surveys were conducted for the
research. After a week of the collection period, 176 copies of questionnaire were received.
45 copies were done by intercept surveys, while 132 copies were completed through
internet. Before the data analysis, it was necessary to check the data set for entry error and
any invalid and missing data. With the assist of SPSS, the entry errors were detected and
then corrected. Since missing data were found in 4 copies of questionnaires, so the valid
questionnaires were accounted for 172 copies, with 97.7% of validity. In order to ensure
all the errors were corrected, a double-checked was made. Finally, all of the invalid
questionnaires were deleted. Since this research was a relative small scaled-research, a
sample size of 172 was an appropriately-sized pool, which enabled the research to draw a
57

reliable conclusion.

4.3 Demographic data analysis

In the survey, the study of respondents demographic data was vital. It is because these
data could further help to understand and study their perceptions and behaviors towards
celebrity and purchase intention. They also served as a supportive data for drawing
conclusions. Apart from that, age, education level, occupations and average monthly
income would be analyzed both in terms of sample populations and genders. Thus we
could obtain a more comprehensive comparison and result.

4.3.1 Gender

Before the questionnaires were distributed,

some targets were set for the data collection.
One of the targets was that the proportions of
male and female respondents must not less
than 6:4 or 4:6. In Figure 4.1, there were 97

Figure 4.1 Gender of respondents

female (56.4%) and 75 male (43.6%) who participated in the survey, which met with the
target. Detailed table was illustrated in Appendix II.
58

4.3.2 Age

Figure 4.2 Age of respondents

Figure 4.2b Age of respondents, by genders

In Figure 4.2, it showed that majority of the respondents were young adult who aged
18-24, which accounted 84.9% (146 respondents). 10.5% of the respondents were adult
who aged 25-39 (18 respondents). Minority of the respondents were teenagers and middle
aged, which accounted for both 2.3% (4 respondents) respectively. The uneven
distribution of respondents age could` be explained by the method of collecting data.
Convenience sampling and online sampling and snowball sampling were applied. After
the respondents did the questionnaires, they might referral to their friends. As a result,
most of the respondents were in the similar age group. Detailed table was illustrated in
Appendix II.

In Figure 4.2b, the respondents age were separated by gender in terms of percentages in
59

order to overcome the uneven gender proportion. It shown that, 90% of female
respondents and 79% of male respondents were 18-24 years old. There were 13% of male
respondents aged 25-39, which accounted for a higher percentage than female, who had
8% of respondents in this age group. Male had the youngest and eldest respondents,
where 4% of the respondents were younger than 17 and between 40-54 years old. Only
had 1% of female respondent was younger than 17 and between 40-54 years old. Detailed
table was illustrated in Appendix II.

4.3.3 Education Level

Figure 4.3 Education levels of respondents

Figure 4.3b Education level of respondents, by genders

In Figure 4.3, it is observed that majority of the respondents were educated at university
level, which accounted for 60.5% (104 respondents). 27.9% (48 respondents) were
educated at high diploma or associate degree level. 11% (19 respondents) were educated
60

at secondary school level. 0.6% (1 respondent) was educated at other level. Detailed table
was illustrated in Appendix II.

In Figure 4.3b, the respondents educational levels were separated by gender in terms of
percentages due to uneven gender proportion. It shown that, 63% out of the total male
respondents had attained university level, which was higher than female respondents 59%.
There were 30% of female respondents who had the high diploma or associate degree
level, which was higher than the male respondents 25%. However, there was 12% of the
male respondents attained secondary school level, which was slightly higher than female
respondents of 10%. Besides there was 1% of female respondent educated at other level.
Detailed table was illustrated in Appendix II.

4.3.4 Occupations

Figure 4.4 Occupations of respondents

61

Figure 4.4b Occupations of respondents, by genders

A target for occupations was set before the distribution of questionnaires, which was to
collect as many respondents who were currently employed as possible. In Figure 4.3, it
shown that 61% of them (105 respondents) were students and 38.4% of them (66
respondents) were on a job. The pool collected was acceptable due to time constraint.
Among the respondents who were on a job, 15.7% (27 respondents) were working in
Wholesale, Retail and Import/Export industry. Some respondents were working in the
Community,

Social

and

Personal

Services

industry,

Transport,

Storage

and

Communications industry as well as Financing, Insurance, Real Estate and Business

Services industry, which accounted for 5.2%s (9 respondent) respectively. Minority of the
62

respondents were working in the Restaurants and Hotels industry, Manufacturing industry
and Electricity, Gas and Water industry, which accounted for 3.5% (6 respondents), 2.3%
(4 respondents) and 0.6% (1 respondent) respectively. 1.2% (2 respondents) claimed that
they were unemployed. Detailed table was illustrated in Appendix II.
In Figure 4.3b, it shown that majority of male and female respondents were students. 21%
of female respondents were working in the Wholesale, Retail and Import/Export industry,
which were much higher than the male respondents (9%). Male respondents were mostly
working in the Financing, Insurance, Real Estate and Business Services industry,
Transport, Storage and Communications industry, Community, Social and Personal
Services industry and Restaurant and Hotel industry than female respondents. They
accounted for 11%, 9%, 8% and 4% respectively. 1% of the female respondent was
working in the Electricity, Gas and Water industry and 2% of them were unemployed.
Detailed table was illustrated in Appendix II.

63

4.3.5. Average monthly income

Figure 4.5Average monthly income of respondents

Figure 4.5b Average monthly income of respondents, by

genders

In Figure 4.5, it revealed that 60% (104 respondents) had average monthly income of
$0-$4999. It could be explained by the fact that majority of the respondents were students.
14.53% (25 respondents) earned $5000-$9999 per month. Besides 16.38% (28
respondents) earned $10000-$14999 and 6.4% (11 respondents) earned$15000-$39999
per month, which could be classified as middle income group. 2.33% (4 respondents)
earned the highest monthly income. Detailed table was illustrated in Appendix II.
In Figure 4.5b, it showed that majority of the respondents earned $0-$4999 per month.
15% of the female respondents and 13% of male respondents earned $5000-$9999 per
month. It also showed that, male respondents earned a higher income than female
respondents in terms of other three income groups. Detailed table was illustrated in
Appendix II.

64

4.4 Consumers purchase intention under the perception of celebrity

endorsement.
In the questionnaire, fifteen attributes of the Source credibility model about the celebrity
endorsement were applied to Q1- 18. It aimed to investigate consumers purchase
intention under the attributes of celebrity endorsement.

4.4.1 Reliability test

Reliability test was an important analytical method which could estimate the reliability
and quality of components of multiple-item. If the Alpha value was >0.7, it considered as
acceptable and preferable when the values > 0.8.

4.4.1.1Reliability of the data set

As shown in Table 4.1, fifteen items were subjected to the reliability test. Those items
were the adjectives which used to describe the three dimensions of Trustworthiness,
Expertise and Attractiveness in the Source credibility model (Ohanian 1990).
According to Pavot, Diener, Colvin and Sandvik (1991), the fifteen attributes of source
credibility model had good internal consistency, with a Cronbach alpha coefficient
reported of 0.85. In the current study the Cronbach alpha coefficient was 0.935. As a
result, they were considered as highly reliable.
65

Reliability Statistics
Cronbach's
Alpha
.935

N of
Items
15

Table 4.1 Reliability statistics (i)

The Corrected Item-Total Correlation helped us to indicate the degree to which each item
correlates with the total score (Pallant, J., 2005). Since all of the values were >0.5, so that
none of the items had low item-total correlations and they could continue the analysis.
Detailed tables about the results of reliability test were illustrated in Appendix III.

4.4.2 Factors analysis

Factors analysis could reduce the amount of data due to possible existence of similar and
redundant variables. So the problems could be simplified with little loss of information.

4.4.2.1 Attributes of celebrity endorsers to purchase intention

According to the Ohanian (1990), the source credibility model included three dimensions
with fifteen adjectives. In the previous chapter, the adjustment made on the model had
been discussed. To avoid any negative impacts made on the variables and to simplify the
model, it was necessary to carry out the factors analysis.

66

I.

Suitability evaluation

The Correlation Matrix table, illustrated in Appendix IV, showed that there were many
variables which are correlated (r> 0.5). From the KMO and Bartlett's Test table (Table
4.2), the Kaiser-Meyer-Oklin value was 0.903, which exceeded the recommended value
of 0.6 (Kaiser, 1970, 1974). So there was a satisfactory amount of variance in original
variables that might be caused by underlying factors. Barletts Test of Sphericity (Bartlett,
1954) was 0.000 < 0.05 reached statistical significance, supporting the factorability of the
correlation matrix. Therefore the original variables were not unrelated.

KMO and Bartlett's Test

Kaiser-Meyer-Olkin Measure of Sampling
Adequacy.
Bartlett's Test of
Approx.
Sphericity
Chi-Square

.903
2225.044

df

105

Sig.

.000

Table 4.2 KMO and Bartlett's Test table

II.

Factor Extraction Evaluation

From the Total Variance Explained table (Table 4.3), it revealed that three underlying
factors were extracted with their eigenvalues of 7.929 (>1), 2.1(>1) and 1.448(>1), they
totally accounted for 76.516% (>60%) of the total variance of all variables. Therefore by
using just three factors instead of fifteen variables, the complexity of the data set could be
considerably reduced
67

Total Variance Explained

Component

Initial Eigenvalues

Extraction Sums of Squared

Loadings

Rotation Sums of Squared

Loadings

% of Cumulative
% of Cumulative
% of
Cumulative
Total Variance
%
Total Variance
%
Total Variance
%
1

7.929

52.858

52.858 7.929

52.858

52.858 4.214

28.093

28.093

2.100

14.003

66.860 2.100

14.003

66.860 3.941

26.270

54.363

1.448

9.656

76.516 1.448

9.656

76.516 3.323

22.153

76.516

.596

3.975

80.491

Table 4.3 Total Variance Explained table

III. Factor loading finding and factor naming

The rotated component matrix table (Table 4.4) illustrated that reliable, honest, sincere,
dependable and trustworthy had the highest factor loading (>0.5) on factor 1. Experience,
experts, qualified, skilled and knowledgeable had the highest factor loading (>0.5) on
factor 2. Beautiful, attractive, classy, sexy and elegant had the highest factor loading
(>0.5) on factor 3.
According to the source credibility model (Ohanian 1990), reliable, honest, sincere,
dependable and trustworthy were exactly the variables of Trustworthiness, which
was factor 1. Experience, experts, qualified, skilled and knowledgeable were the
variables of Expertise, which was the factor 2. Finally, beautiful, attractive, classy,
sexy and elegant were the variables of Attractiveness. It proved that the three
dimensions did not interrupted by the changes which made to the model. As a result, the

68

results of factors analysis were satisfied. Detailed tables and Figures about the results of
factors analysis were illustrated in Appendix IV.
Rotated Component Matrixa
Component
1

Q10Reliable
Q9Honest
Q11Sincere
Q8Dependable

.892
.876
.876
.846

.191
.199
.157
.241

.193
.185
.202
.298

Q7Trustworthy

.779

.281

.293

Q14Experienced
Q13Experts

.152
.110

.871
.825

.200
.333

Q16Qualified
Q17Skilled
Q15Knowledgeable

.297
.223
.241

.823
.805
.795

.077
.306
.265

Q2Beautiful
Q1Attractive
Q3Classy
Q4Sexy

.156
.223
.217
.192

.269
.187
.363
.245

.853
.793
.696
.680

Q5Elegant

.414

.075

.642

Extraction Method: Principal Component Analysis.

Rotation Method: Varimax with Kaiser
Normalization.
a. Rotation converged in 6 iterations.
Table 4.4 Rotated Component Matrix Table

69

4.4.3Impacts of celebrity endorsements

4.4.3.1 Impacts of attributes of celebrity endorsement to consumers
purchase intention
In the Independent Samples Test, if the Sig. value of the Levene's Test for Equality of
Variances is >0.05, the Sig. (2-tailed) under equal variances assumed and would be used.
If not, equal variances not assumed would be used.

As shown in Table 4.5, in general, Sexy had the lowest mean (2.81) among, the
respondents, which revealed sexy celebrities were not able to trigger consumers purchase
on the endorsed fashion and apparel products. Experience had the highest mean (3.6)
among the populations, and then was Knowledgeable (3.59) and Experts (3.58). It
depicted that experienced, knowledgeable and celebrity endorsers who were experts in
fashion were able to trigger consumers purchase intention on fashion apparel items than
the others.

By genders
Sexy had the lowest means, males was2.73, females was 2.87. As shown in Table 4.6,
due to the Sig. (2-tailed) of Sexy between genders was 0.405 > 0.05, it proved that
statistically significant difference was not existed. So that males and females through that
70

their purchase intentions on the endorsed fashion apparel products were not able to be
triggered by sexy celebrity endorsers.
Males thought that beautiful celebrity endorsers were not able to trigger their purchase
intention, with the mean of 2.92 which was lower than the total mean (3.09). The mean of
Beautiful for females was 3.23. Due to the Sig. (2-tailed) of Beautiful between genders
was 0.047 < 0.05, it proved that statistically significant difference was existed. Therefore
the females purchase intentions on fashion apparel products were able to be trigger by
beautiful celebrity endorsers.

Although female respondents gave a higher mean to Classy, Elegant, Trustworthy,

Dependable, Honest, Reliable, Experts, Experience, Knowledgeable, Qualified
and Skill of celebrity endorsers than males, the Sig. (2-tailed) of the above mentioned
attributes were >0.05, which means statistically significant difference was not existed. As
a result, males purchase intentions were also able to be triggered by celebrity endorsers
who had the above attributes. But females purchase intentions were easier to be triggered
by Attractive, Beautiful, Attractive and Sincere of the celebrity endorsers than
males. Due to the Sig. (2-tailed) of <0.05, so the statistically significant difference was
existed. Detailed tables about the mean scores of fifteen attributes were illustrated in
Appendix V.
71

Q1

Mean

Q7

Q8

Q10

Q11

Q14

Q5
Trustw Depend
Q9 Reliabl Sincer
Elegant orthy
able
Honest
e
e

Q13
Experts

Q15

Experi Knowle
enced dgeable

Q16
Qualified

Q17
Skilled

3.12

2.92

3.21

2.73

3.20

3.37

3.35

3.41

3.44

3.35

3.56

3.53

3.45

3.40

3.49

75

75

75

75

75

75

75

75

75

75

75

75

75

75

75

3.43

3.23

3.49

2.87

3.31

3.54

3.52

3.62

3.64

3.65

3.60

3.66

3.70

3.67

3.60

97

97

97

97

97

97

97

97

97

97

97

97

97

97

97

Mean

3.30

3.09

3.37

2.81

3.26

3.47

3.44

3.53

3.55

3.52

3.58

3.60

3.59

3.55

3.55

172

172

172

172

172

172

172

172

172

172

172

172

172

172

172

N
Female Mean
N
Total

Q3

Attract Beauti Class Q4

ive
ful
y
Sexy

Gender
Male

Q2

Table4.5 Mean table of attributes in Source credibility model

Levene's

Levene's

T-test for
Equality of
Means

Test for
Equality of
Variances

T-test for
Equality of
Means

Test for
Equality of
Variances

Sig.

Sig. (2-tailed)

Sig.

Sig.
(2-tailed)

Q1Attractive

Equal variances assumed

.019

.038 Q10Reliable

.724

Equal variances assumed

.201

Q2Beautiful

Equal variances not assumed

.722

.042 Q11Sincere

.125

Equal variances assumed

.042

Q3Classy

Equal variances assumed

.067

.065 Q13Experts

.170

Equal variances assumed

.792

Q4Sexy

Equal variances assumed

.234

.405 Q14Experienced

.087

Equal variances assumed

.363

Q5Elegant

Equal variances assumed

.857

.454 Q15Knowledgeable

.225

Equal variances assumed

.077

Q7Trustworthy Equal variances assumed

.957

.276 Q16Qualified

.005

Equal variances not assumed

.071

Q8Dependable Equal variances assumed

.800

.264 Q17Skilled

.023

Equal variances not assumed

.466

Q9Honest

.469

.169

Equal variances assumed

Table4.6 Independent Samples Test- Equality of means table

72

4.4.3.2 Impacts of source credibility of celebrity endorsers to consumers

purchase intention
After the factors analysis, fifteen similar variables were reduced to three factors. By
calculating the average means of sub-variables of the dimensions and using the paired
sample T-test, comparison between Trustworthiness, Expertise and Attractiveness
were able to be made.
Mean
Pair 1
Pair 2
Pair 3

Mean of Attractiveness

3.1663

172

Mean of Trustworthiness
Mean of Attractiveness
Mean of Expertise
Mean of Trustworthiness

3.5012
3.1663
3.5767
3.5012

172
172
172
172

Mean of Expertise

3.5767

172

Tale 4.7 Paired Samples Statistics table, with means

In Table 4.7, it shown that in general, the average means of Expertise was the highest
(3.58) secondly was Trustworthiness (3.5) thirdly was Attractive (3.17).

N
Pair 1
Pair 2
Pair 3

Correlation

Sig.

Mean of Attractiveness & Mean of

Trustworthiness
Mean of Attractiveness & Mean of
Expertise

172

.574

.000

172

.578

.000

Mean of Trustworthiness & Mean of

Expertise

172

.502

.000

Table 4.8 Paired Samples Correlations table (i)

In Table 4.8, three dimensions of the Source credibility model were grouped into three
73

pairs. Each pair had a value of correlation >0.5 and a significant level <0.05, which
suggested they were correlated.

Paired Differences
Mean

Std.
Deviation

Sig. (2-tailed)

Pair 1

Mean of Attractiveness Mean of Trustworthiness

-.33488

.78670

.000

Pair 2

Mean of Attractiveness Mean of Expertise

Mean of Trustworthiness
- Mean of Expertise

-.41047

.73993

.000

-.07558

.85381

.247

Pair 3

Table 4.9 Paired Differences table (ii)

In Table 4.9, it showed that the average means of pair 1 (Attractiveness &
Trustworthiness) had statistically significant difference, with the Sig. (2-tailed) of 0.000<
0.05. Average means of pair 2 (Attractiveness & Expertise) also had statistically
significant difference, with the Sig. (2-tailed) of 0.000< 0.05. However the average means
of pair 3 (Trustworthiness & Expertise) had no statistically significant difference, with
Sig. (2-tailed) of 0.247> 0.05. As a result, the Trustworthiness and Expertise of
celebrity endorsers were able to trigger consumers purchase intention. Detailed tables
about paired sample test were illustrated in Appendix VI.

Therefore the hypothesis,

74

 H1a: Attractiveness of celebrity endorsers have less positive impact on consumers

purchase intention of fashion apparel products was accepted.
H2a: Trustworthiness of celebrity endorsers have less positive impact on the
consumers purchase intention of fashion apparel products was rejected.
H3a: Expertise of celebrity endorsers have less positive impact on the consumers
purchase intention of fashion apparel products was rejected.

In conclusion, although Expertise had the highest mean value than Trustworthiness,
the results of paired-samples t-test shown that there was no statistically significant
difference. So, Expertise and Trustworthiness had greater impact on consumers
purchase intentions on fashion apparel products.

By genders
In the Independent Samples Test, if the Sig. value of the Levene's Test for Equality of
Variances is >0.05, the Sig. (2-tailed) under equal variances assumed and would be used.
If not, equal variances not assumed would be used.

As shown in Table 10, by genders, females had a higher average means of Expertise
(3.65) than males (3.49). Moreover females had a higher average means of

75

Trustworthiness (3.6) and Attractiveness (3.3) than males, whose average means were
just 3.39 and 3.04.

Gender

Mean of Attractiveness Mean of Trustworthiness Mean of Expertise

Male

Mean

3.0373

3.3840

3.4880

Female

Mean

3.2660

3.5918

3.6454

Total

Mean

3.1663

3.5012

3.5767

Table 4.10 Report table, by genders

Levene's Test for

t-test for

Equality of
Variances

Equality of
Means

Sig.

Sig.
(2-tailed)

Mean of Attractiveness

Equal variances assumed

.330

.063

Mean of Trustworthiness

Equal variances assumed

.558

.131

Mean of Expertise

Equal variances assumed

.212

.207

Table 4.11 Independent Samples Test table

In Table 4.11, it showed that the Sig. (2-tailed) of Expertise was 0.207 > 0.05, which
means statistically significant difference was not existed. Secondly, the Sig. (2-tailed) of
Trustworthiness was 0.131 > 0.05, which means statistically significant difference was
not existed. Thirdly, the Sig. (2-tailed) of Attractiveness was 0.063 > 0.05, which means
statistically significant difference was not existed. As a result, in terms of Expertise,
Trustworthiness and Attractiveness, there were no statistically significant difference
between gender.
76

Therefore the hypothesis,

'H1b: Different gender had different purchase intention under the impact of
attractive celebrity endorsement was rejected.
H2b: Different gender had different purchase intention under the impact of
trustworthy celebrity endorsement was rejected.
H3b: Different gender had different purchase intention under the impact of
expertise of celebrity endorsement was rejected.

In conclusion, different genders tend to have similar purchase intention under the three
dimensions. Detailed tables about independent-samples t-test were illustrated in Appendix
VII.

4.4.3.3 Important dimension of source credibility for celebrity endorsers

In Table 4.12, it showed that in general, the means of Importance of Trustworthiness
was the highest (3.9), followed by Importance of Expertise (3.63) and Importance of
Attractive (3.43).

Mean
Pair 1

Pair 2

Q6 Importance of Attractiveness

3.43

172

Q12 Importance of Trustworthiness

3.90

172

Q6 Importance of Attractiveness

3.43

172
77

Pair 3

Q18 Importance of Expertise

3.63

172

Q12 Importance of Trustworthiness

3.90

172

Q18 Importance of Expertise

3.63

172

Table 4.12 Paired Samples Statistics table (ii)

Paired Differences
Std.
Mean
Pair 1

Q6Importance of Attractiveness -

Deviation

Sig.
(2-tailed)

-.465

1.078

.000

-.198

1.231

.037

.267

1.048

.001

Q12 Importance of Trustworthiness

Pair 2

Q6Importance of Attractiveness Q18 Importance of Expertise

Pair 3

Q12 Importance of Trustworthiness Q18 Importance of Expertise

Table 4.13 Paired Differences table (iii)

In Table 4.13, it showed that average means of pair 1 had statistically significant
difference, with the Sig. (2-tailed) of 0.000< 0.05. Average means of pair 2 also had
statistically significant difference, with the Sig. (2-tailed) of 0.037< 0.05. Lastly, average
means of pair 3 had statistically significant difference, with the Sig. (2-tailed) of 0.001<
0.05 .As a result, consumers thought that Trustworthiness of celebrity endorsers was the
most important and had a significant difference with Expertise and Attractiveness.
Detailed tables about the test were illustrated in Appendix VIII.

By genders
78

In the Independent Samples Test, if the Sig. value of the Levene's Test for Equality of
Variances is >0.05, the Sig. (2-tailed) under equal variances assumed and would be used.
If not, equal variances not assumed would be used.

As shown in Table 4.14, females thought that Trustworthiness was very important to the
celebrity endorsers, with a mean of 3.99. Males also thought that Trustworthiness was
important but with a lower mean than female (3.77). Secondly, females and males agreed
that Expertise was important, with a mean of 3.65 and 3.6 respectively. Lastly, females
thought that Attractiveness of the celebrity endorsers important, with a mean of 3.48,
which was a higher mean than males (3.36).

Gender
Q6Importance of Attractiveness

Mean

Male

3.36

Female

3.48

(Equal variances assumed)

Q12 Importance of Trustworthiness

Male

3.77

Female

3.99

(Equal variances assumed)

Q18 Importance of Expertise

Male

3.60

Female

3.65

(Equal variances assumed)

Levene's Test

t-test for

for Equality of
Variances

Equality of
Means

Sig.

Sig. (2-tailed)

.757

.432

.262

.123

.059

.734

Table 4.14 Report table of Independent Samples Test

Moreover it showed that Trustworthiness had a Sig. (2-tailed) of 0.123> 0.05. So males
79

and females agreed that it was the most important dimensions to the celebrity endorsers.
Secondly, Expertise had a Sig. (2-tailed) of 0.734> 0.05. It meant no statistically
significant difference was existed between genders, in terms of the importance of
Expertise of the celebrity endorsers. Lastly, Attractiveness had a Sig. (2-tailed) of
0.432> 0.05. It meant statistically significant difference was not existed among genders.
Detailed tables about the average means, by genders were illustrated in Appendix IX.

80

4.5 Source credibility

In Figure 4.6, it showed that about 36.6 % of the respondents (63) thought that the most
credible source of celebrity endorsement was came from Attractiveness of the celebrity.
At the same time, 36% of the respondents (62) thought that Trustworthiness was the
most important credibility source. Around 27.3% of the respondents (47) thought that
Expertise was the most credible source of celebrity endorsement. However these results
present difference from the results discovered in previous part. Detailed tables were
illustrated in Appendix X.

By gender
As shown in Figure 4.6b, 46% of female thought that the most credible source was came
from Attractiveness of the celebrities. Secondly was Trustworthiness (29%). The last
was Expertise (25%). It showed that females tend to believe the fashion apparel
products which were endorsed by attractive celebrities. For males, they thought that
Trustworthiness was the most credibility source of celebrity endorsements (45%).
Secondly was Expertise (31%). The last was Attractiveness (24%). It showed that
males tend to believe the fashion apparel products which were endorsed by trustworthy
celebrities. Detailed tables were illustrated in Appendix X.

81

In summary, consumers tend to believe the fashion apparel products which were endorsed
by attractive celebrities. But when the data was analyzed by genders, it showed that males
thought that Trustworthiness was the most credibility source of celebrity endorsements,
not Attractiveness.

Figure 4.6 Most credible source of celebrity endorsement

Figure 4.6b Most credible source of celebrity endorsement, by

genders

82

4.6 Perceived image of celebrity endorsers

To examine the relationship between good-looking, trustworthiness, expertise and
attractiveness, ten celebrities were chosen and matched with the above attributes which
consumer perceived.

4.6.1 Reliability of data set

As shown in Table 4.15, all the celebrities with particular attribute of forty items were
subjected to the reliability test. According to Pavot, Diener, Colvin and Sandvik (1991),
the celebrities with particular attribute had good internal consistency, with a Cronbach
alpha coefficient reported of 0.85. In the current study the Cronbach alpha coefficient was
0.912. As a result, they were considered as highly reliable.

Reliability Statistics
Cronbach's
Alpha
.912

N of
Items
40

Table 4.15 Reliability statistics (ii)

The Corrected Item-Total Correlation can help to indicate the degree to which each item
correlates with the total score (Pallant, J., 2005). Some of the values were lower than 0.3,
which meant the items were measuring something different from the scale as a whole.
Since the overall Cronbach alpha was greater than 0.7, thus those with low item-total
83

correlations needed not to be removed. Detailed tables about the results of reliability test
were illustrated in Appendix XI.

4.6.2 Good-Looking, Trustworthiness, Expertise and Attractiveness

4.6.2.1 Ratings
Table 4.16 was a simplified table which showed the ratings of the celebrities. In order to
find out how Good-looking would affect consumers perceptions on celebrity
endorsement, only the ratings related to best-looking and plain-looking celebrities would
be discussed. Moreover average means of the four attributes for the ten celebrities, by
genders, were calculated, as shown in Table 4.17. So it could serve as a guide for the
ratings. All the tables related to the ratings of celebrities were illustrated in Appendix XI.

Gender

Average

Average

Average

Average mean of

mean of

mean of

mean of

celebrities

celebrities

celebrities

celebrities

Good-Looking

Trustworthy

Attractive

Expertise

Male

Mean

3.2253

3.3320

3.2080

3.3373

Female

Mean

3.1856

3.1845

3.1454

3.2010

Table 4.16 Average mean table, by genders

Male Respondents

Mean

Female Respondents

Mean

Daniel Wu
Q20 Good-Looking

4.24 Q20Good-Looking

4.51

Q21Trustworthy

3.91 Q21Trustworthy

3.82

Q22Attractive

4.01 Q22Attractive

4.40
84

Q23Expertise

3.35 Q23Expertise

3.39

Takeshi Kaneshiro
Q20 Good-Looking

4.24 Q20 Good-Looking

4.19

Q21 Trustworthy

3.91 Q21 Trustworthy

3.76

Q22 Attractive

4.00 Q22 Attractive

4.04

Q23 Expertise

3.35 Q23 Expertise

3.27

Wyman Wong
Q20 Good-Looking

2.56 Q20 Good-Looking

2.26

Q21 Trustworthy

3.91 Q21 Trustworthy

3.46

Q22 Attractive

2.88 Q22 Attractive

2.61

Q23 Expertise

4.24 Q23 Expertise

4.06

Sam Lee
Q20 Good-Looking

2.64 Q20 Good-Looking

2.39

Q21 Trustworthy

3.25 Q21Trustworthy

3.01

Q22 Attractive

2.76 Q22Attractive

2.67

Q23 Expertise

3.41 Q23Expertise

2.96

PakHo Chau
Q20 Good-Looking

3.60 Q20 Good-Looking

3.18

Q21 Trustworthy

3.36 Q21 Trustworthy

2.89

Q22 Attractive

3.45 Q22 Attractive

3.01

Q23 Expertise

3.09 Q23 Expertise

2.79

Angelababy
Q20 Good-Looking

3.68 Q20 Good-Looking

3.53

Q21 Trustworthy

2.72 Q21 Trustworthy

2.67

Q22 Attractive

3.47 Q22 Attractive

3.13

Q23 Expertise

2.96 Q23 Expertise

2.59

Gaile Lai
Q20 Good-Looking

3.05 Q20 Good-Looking

3.31

Q21 Trustworthy

3.19 Q21Trustworthy

3.32

Q22 Attractive

3.17 Q22 Attractive

3.15

Q23 Expertise

3.47 Q23 Expertise

3.55
85

Hilary Tsui
Q20 Good-Looking

2.48 Q20 Good-Looking

2.60

Q21 Trustworthy

3.07 Q21 Trustworthy

3.06

Q22 Attractive

2.61 Q22 Attractive

2.69

Q23 Expertise

3.61 Q23 Expertise

3.48

Josie Ho
Q20 Good-Looking

2.53 Q20 Good-Looking

2.65

Q21 Trustworthy

3.11 Q21 Trustworthy

3.04

Q22 Attractive

2.61 Q22 Attractive

2.59

Q23 Expertise

3.19 Q23 Expertise

3.07

Charlene Choi
Q20 Good-Looking

3.23 Q20 Good-Looking

3.26

Q21 Trustworthy

2.91 Q21 Trustworthy

2.80

Q22 Attractive

3.11 Q22 Attractive

3.15

Q23 Expertise

2.71 Q23 Expertise

2.85

Table 4.17 Celebrities rating table

86

I.

Good-looking

For male respondents:

Best-looking (mean >4): Daniel Wu and

Kaneshiro

Ordinary-looking (3.23< mean <4): PakHo Chau, Angelababy and Charlene Choi.
Plain-looking (mean<3.23): Wyman Wong, Sam Lee, Gaile Lai, Hilary Tsui, and Josie
Ho

For female respondents:

Best-looking (mean >4): Daniel Wu and

Kaneshiro

Ordinary-looking (3.19<mean, 4): Angelababy, Gaile Lai and Charlene Choi.

Plain-looking (mean<3.19): Wyman Wong, Sam Lee, PakHo Chau, Hilary Tsui, and
Josie Ho

As a result, both male and female respondents thought that Daniel Wu and

Kaneshiro

were the best-looking celebrities among the listed celebrities. Wyman Wong and Hilary
Tsui got the lowest mean values, which mean respondents thought that they had a really
plain-looking.

87

II

Trustworthiness

For male respondents:

Highly trustworthy (mean >3.5): Daniel Wu,

Kaneshiro and Wyman Wong

Quite trustworthy (3.33< mean <3.5): PakHo Chau

Untrustworthy (mean<3.33): Sam Lee, Angelababy, Gaile Lai, Hilary Tsui, Josie Ho and
Charlene Choi

For female respondents:

Highly trustworthy (mean >3.5): Daniel Wu,

Kaneshiro and Wyman Wong Quite

trustworthy (3.18< mean <3.5): Gaile La

Untrustworthy (mean<3.18): Sam Lee, PakHo Chau, Angelababy, Hilary Tsui, Josie Ho
and Charlene Choi

Male and female respondents thought that the best- looking celebrity, Daniel Wu and
Kaneshiro were relatively more trustworthy. It is because the mean values of
trustworthiness were very much closed to 4. However one of the plain-looking celebrity,
Wyman Wong, he got the same mean value from the male respondents as that of Daniel
Wu and

Kaneshiro received. He also got the third highest mean value of

trustworthiness from female respondents. Whereas, Angelababy, one of the ordinary

88

looking celebrity who actually got the third highest mean value in terms of good-looking.
She got the lowest mean value in trustworthiness from both male and female respondents.

III

Attractiveness

For male respondents:

Highly attractive (mean >4): Daniel Wu and

Kaneshiro

Quite attractive (3.21< mean <4): PakHo Chau, Angelababy

Unattractive (mean<3.21): Wyman Wong, Sam Lee, Gaile Lai, Hilary Tsui, Josie Ho and
Charlene Choi

For female respondents:

Highly attractive (mean >4): Daniel Wu and

Kaneshiro

Quite attractive (3.15< mean <4): Gaile Lai and Charlene Choi
Unattractive (mean<3.15): Wyman Wong, Sam Lee, PakHo Chau, Angelababy; Hilary
Tsui, Josie Ho

Both male and female respondents thought that the best looking and trustworthy
celebrities, Daniel Wu and

Kaneshiro were very attractive. Although Wyman Wong

was trustworthy, male and female respondents thought that he was not attractive at all
89

because of his plain-looking. In summary, most of the celebrities who had plain-looking
usually rated as not attractive at the same time and vise versa, such as Hilary Tsui and
Josie Ho.

IV Expertise
For male respondents:
Expert in fashion (mean >4): Wyman Wong.
Know a little about fashion (3.34< mean <4): Daniel Wu,

Kaneshiro, Sam Lee, Gaile

Lai and Hilary Tsui

Not expert in fashion (mean<3.34): PakHo Chau, Angelababy; Josie Ho and Charlene Choi

For female respondents:

Expert (mean >4): Wyman Wong.
Know a little (3.2< mean <4): Daniel Wu,

Kaneshiro, Gaile Lai and Hilary Tsui

Not expert (mean<3.2): Sam Lee, PakHo Chau, Angelababy, Charlene Choi and Josie Ho

In terms of Expertise, both male and female respondents thought that Wyman Wong, who
was trustworthy however plain-looking and not attractive, was the expert in fashion. On

90

the other hand, Hilary Tsui was plain looking, neither trustworthy nor attractive but she
was still able to receive an above average mean.

From the above discussion, it is observed that the good-looking celebrities such as Daniel
Wu and

Kaneshiro, were able to get a high rating in all the attributes than other

ordinary-looking or plain-looking celebrities. But some of the plain-looking celebrities

such as Wyman Wong and Hilary Tsui were able to get a high rating in some attributes. It
suggested that there might be some possible underlying relationships between
good-looking, trustworthiness, attractiveness and expertise. In order to investigate this,
two celebrities were selected.

Daniel Wu , who had the highest mean value for good-looking

Wyman Wong, who the lowest mean value for good-looking.

91

4.6.3 Relationships between Good-looking, Trustworthiness, Expertise

and Attractiveness of celebrity endorsers.

4.6.3.1 Good-looking and Trustworthiness

Good-looking celebrity & Trustworthiness
From Table 4.18, the Coefficient of Determination R2of male was 18.9% and 11.6% for
female. It indicated that 18.9% and 11.6% of the variations in Good-looking of the
celebrity could be explained by the Trustworthiness.
From Table 4.19, it showed that the p-value of F-test in Table 4.22and p-value of t-test in
Table 4.20 for male was 0.000 <0.05 and 0.000 <0.05. It explained that there was a
significant linear relationship between Good-looking and Trustworthiness. Moreover
the p-value of F-test and p-value of t-test for female was 0.001< 0.05 and 0.001 >0.05,
which meant there was also a significant linear relationship between Good-looking and
Trustworthiness. Detailed tables were shown in Appendix XII.

Model Summaryb
Gender

Model

R Square

Male

.434a

.189

Female

.341a

.116

a. Predictors: (Constant), Q20Daniel_ Good-Looking

b. Dependent Variable: Q21Daniel_Trustworthiness
Table 4.18 Regression Model Summary table (i)

92

ANOVAb
Gender Model

Sig.

Regression

.000a

Female 1

Regression

.001a

Male

a. Predictors: (Constant), Q20Daniel_ Good-Looking

b. Dependent Variable: Q21Daniel_ Trustworthiness
Table 4.19 Regression Model ANOVA table (i)

Coefficientsa
Gender

Model

Male

Female

Sig.
(Constant)

.000

Q20Daniel_Good-Looking

.000

(Constant)

.001

Q20Daniel_G Good-Looking

.001

a. Dependent Variable: Q21Daniel_ Trustworthiness

Table 4.20 Coefficients table (i)

Plain-looking celebrity & Trustworthiness

From Table 4.21, the Coefficient of Determination R2 of male was 1% and 5% for female.
It indicated that 1% and 5% of the variations in Good-looking of the celebrity could be
explained by the Trustworthiness.
It showed that the p-value of F-test in Table 4.22 and p-value of t-test in Table 4.23 for
males was 0.382 > 0.05 and 0.382 > 0.05. Both of them were rejected. It meant that there
was no significant linear relationship between Good-looking and Trustworthiness.
Besides the p-value of F-test and p-value of t-test for female was 0.028< 0.05 and 0.028
93

<0.05, which meant there was significant linear relationship between Good-looking and
Trustworthiness. Detailed tables were shown in Appendix XII.

Model Summaryb
Gender

Model

R Square

Male

.010

Female

.050

a. Predictors: (Constant), Q20Wyman_

Good-Looking
b. Dependent Variable: Q21Wyman_T
Table 4.21 Regression Model Summary tablem(ii)

ANOVAb
Gender

Model

Sig.

Male

Regression

.382a

Female

Regression

.028a

a. Predictors: (Constant), Q20Wyman_G_ Good-Looking

b. Dependent Variable: Q21Wyman_ Trustworthiness
Table 4.22 Regression Model ANOVA table (ii)

Coefficientsa
Gender Model
Male

Female 1

Sig.
(Constant)

.000

Q20Wyman_ Good-Looking

.382

(Constant)

.000

Q20Wyman_ Good-Looking

.028

a. Dependent Variable: Q21Wyman_ Trustworthiness

Table 4.23 Coefficients table (ii)

94

In conclusion, if good-looking celebrity endorsed the fashion apparel products, significant

linear relationship between Good-looking and Trustworthiness was existed for both
genders. It is because Trustworthiness would be positively affected by Good-looking
of the celebrity.
On the other hand, males thought that the Trustworthiness of the celebrity would not be
affected by Good-looking, if the celebrity was plain-looking. It is because there was no
significant linear relationship between Good-looking and Trustworthiness for
plain-looking celebrities. However, females thought that there was a significant linear
relationship between Good-looking and Trustworthiness for plain-looking celebrities.
Therefore they do not believe plain-looking celebrities have trustworthiness.

4.6.3.2 Good-looking and Attractiveness

Good-looking celebrity & Attractiveness
From Table 4.24, the Coefficient of Determination R2of male was 40.6% and 60.3% for
female. It indicated that 40.6% and 60.3% of the variations in Good-looking of the
celebrity could be explained by the Attractiveness.
It showed that the p-value of F-test in Table 4.25 and p-value of t-test in Table 4.26 for
male and female were 0.000<0.05 and 0.000 <0.05. Both of them were accepted, so that
there was a significant linear relationship between Good-looking and Attractiveness.
Detailed tables were shown in Appendix XII.
95

Model Summaryb
Gender Model R Square
Male

.460

Female 1

.603

a. Predictors: (Constant), Q20Daniel_ Good-looking

b. Dependent Variable: Q22Daniel_ Attractiveness
Table 4.24 Regression Model Summary table (iii)

ANOVAb
Gender Model

Sig.

Regression

.000a

Female 1

Regression

.000a

Male

a. Predictors: (Constant), Q20Daniel_ Good-Looking

b. Dependent Variable: Q22Daniel_ Attractiveness
Table 4.25 Regression Model Summary ANOVA table (iii)

Coefficientsa
Gender Model
Male

Female 1

Sig.
(Constant)

.003

Q20Daniel_ Good-Looking

.000

(Constant)

.178

Q20Daniel_ Good-Looking

.000

a. Dependent Variable: Q22Daniel_ Attractiveness

Table 4.26 Coefficients table (iii)

Plain-looking celebrity & Attractiveness

96

From Table 4.27, the Coefficient of Determination R2 of male was 27.3% and 37.9% for
female. It indicated that 27.3% and 37.9% of the variations in Good-looking of the
celebrity could be explained by the Attractiveness.
It showed that the p-value of F-test in Table 4.28 and p-value of t-test in Table 4.29 for
males and females were 0.000 < 0.05 and 0.000 < 0.05. Both of them were accepted and
could be explained by that there was a significant linear relationship between
Good-looking and Attractiveness. Detailed tables were shown in Appendix XII.

Model Summaryb
Gender Model R Square
Male

.273

Female 1

.379

a. Predictors: (Constant), Q20Wyman_ Good-Looking

b. Dependent Variable: Q22Wyman_ Attractiveness
Table 4.27 Regression Model Summary table (iv)

ANOVAb
Gender Model

Sig.

Regression

.000a

Female 1

Regression

.000a

Male

a. Predictors: (Constant), Q20Wyman_ Good-Looking

b. Dependent Variable: Q22Wyman_ Attractiveness
Table 4.28 Regression Model Summary ANOVA table (iv)

Coefficientsa
Gender Model
Male

Sig.
(Constant)

.000
97

Female 1

Q20Wyman_ Good-Looking

.000

(Constant)

.001

Q20Wyman_ Good-Looking

.000

a. Dependent Variable: Q22Wyman_ Attractiveness

Table 4.29 Coefficients table (iv)

In conclusion, no matter the fashion apparel products were endorsed by good-looking or

plain looking celebrities, there was a significant linear relationship between
Good-looking and Attractiveness. It is because Attractiveness was positively
affected by Good-looking of the celebrity. As a result, good-looking celebrity would
have a certain high degree of attractiveness. But plain-looking celebrity would not
possess this due to significant linear relationship were only found between Good-looking
and Attractiveness.

98

4.6.3.3 Good-looking and Expertise

Good-looking celebrity & Expertise
From Table 4.30, the Coefficient of Determination R2of male was 6.5% and 2.3% for
female. It indicated that 6.5% and 2.3% of the variations in Good-looking of the
celebrity could be explained by the Expertise.
It showed that the p-value of F-test in Table 4.31and p-value of t-test in Table 4.32 for
male was 0.028 <0.05 and 0.028 <0.05. It explained that there was a significant linear
relationship between Good-looking and Expertise. Moreover the p-value of F-test and
p-value of t-test for female was 0.137> 0.05 and 0.137 >0.05, which meant there was no
significant linear relationship between Good-looking and Expertise. Detailed tables
were shown in Appendix XII.

Model Summaryb
Gender Model R Square
Male

.065

Female 1

.023

a. Predictors: (Constant), Q20Daniel_Good-Looking

b. Dependent Variable: Q23Daniel_Expertise
Table 4.30 Regression Model Summary table (v)

ANOVAb
Gender Model

Sig.

Regression

.028a

Female 1

Regression

.137a

Male

a. Predictors: (Constant), Q20 Daniel_Good-Looking

99

b. Dependent Variable: Q23 Daniel_Expertise

Table 4.31 Regression Model Summary ANOVA table (v)

Coefficientsa
Gender Model
Male

Female 1

Sig.
(Constant)

.000

Q20Daniel_ Good-Looking

.028

(Constant)

.000

Q20Daniel_ Good-Looking

.137

a. Dependent Variable: Q23 Daniel_Expertise

Table 4.32 Coefficients table (v)

Plain-looking celebrity & Expertise

From Table 4.33, the Coefficient of Determination R2 of male was 4% and 3% for female.
It indicated that 4% and 3% of the variations in Good-looking of the celebrity could be
explained by the Expertise.
It showed that the p-value of F-test in Table 4.34and p-value of t-test in Table 4.35 for
male was 0.607 > 0.05 and 0.607 > 0.05. Both of them were rejected. It meant that there
was no significant linear relationship between Good-looking and Expertise. Moreover
the p-value of F-test and p-value of t-test for female was 0.137> 0.05 and 0.137 >0.05,
which meant there was also no significant linear relationship between Good-looking and
Expertise. Detailed tables were shown in Appendix XII.

100

Model Summaryb
Gender Model R Square
Male

.004

Female

.030

a. Predictors: (Constant), Q20Wyman_ Good-Looking

b. Dependent Variable: Q23Wyman_ Expertise
Table 4.33 Regression Model Summary table (vi)

ANOVAb
Gender

Model

Sig.

Male

Regression

.607a

Female

Regression

.088a

a. Predictors: (Constant), Q20Wyman_ Good-Looking

b. Dependent Variable: Q23Wyman_ Expertise
Table 4.34 Regression Model Summary ANOVA table (vi)

Coefficientsa
Gender Model
Male

Female 1

Sig.
(Constant)

.000

Q20Wyman_ Good-Looking

.607

(Constant)

.000

Q20Wyman_ Good-Looking

.088

a. Dependent Variable: Q23Wyman_ Expertise

Table 4.35 Coefficients table (vi)

In conclusion, for males, if good-looking celebrity endorsed the fashion apparel products,
significant linear relationship between Good-looking and Expertise was existed. It is
because Expertise would be positively affected by Good-looking of the celebrity. For
females, there was no such kind of significant linear relationship between Good-looking
and Expertise
101

On the other hand, both males and female thought that the Expertise of the celebrity
would not be affected by Good-looking, if the celebrity was plain-looking. It is because
there was no significant linear relationship between Good-looking and Expertise for
plain-looking celebrities.

4.6.4 Favorite celebrity endorsers

Respondents were allowed to select three favorite celebrities from the listed ten
celebrities. As shown in Table 4.36, Daniel Wu was the most favorite celebrity among
males and female who obtained a very high percentage of votes.

Kaneshiro was the

second favorite celebrity for both males and female. For males, their third favorite
celebrity was Pakho Chau. For females, Charlene Choi was their third favorite celebrity.

Male

Female
Daniel Wu

Pakho Chau

56 (74.7%)
40 (53.3%)
30 (40%)

Charlene Choi

84(86.6%)
40(50.5%)
31 (32%)

Angela Baby

25 (33.3%)

Angela Baby

30 (30.9%)

Charlene Choi

25 (33.3%)

Gaile Lai

30 (30.9%)

Wyman Wong

16 (21.3%)

Pakho Chau

21 (21.6% )

Sam Lee

14 (18.7%)

Hilary Tsui

14 (14.4%)

Gaile Lai

11 (14.7%)

Josie Ho

11 (11.3%)

Hilary Tsui

7 (9.3%)

Wyman Wong

11 (11.3%)

Josie Ho

1 (1.3%)

Sam Lee

10 (10.3%)

Daniel Wu
Kaneshiro

Kaneshiro

Count

75

Count

97

% of Total

100%

% of Total

100%

Table 4.36 Top3 favorite celebrities table

102

Gender
Male

Female

Sum

Mean

No. of male celebrity chose

75

154

2.05

No. of female celebrity chose

75

69

.92

Valid N (listwise)

75

No. of male celebrity chose

97

178

1.84

No. of female celebrity chose

97

113

1.16

Valid N (listwise)

97

Table 4.37 Descriptive Statistics table of 3 favorite celebrities

In Table 4.37, it showed among the three votes, male had selected 2 male celebrities and
one female celebrity. In average, female also had selected 2 male celebrities and one
female celebrity.

In conclusion, base on the discussions in previous part, some attributes about the
celebrities were sum up by first that Daniel Wu and

Kaneshiro had the attributes of

Best-looking, Highly trustworthy, Highly attractive and Know a little about fashion.
So that males and females could be attracted by these attributes.

Secondly, Pakho Chau got the comments of Ordinary-looking, Quite trustworthy,

Quite attractive and Not expert in fashion from males. Males were attracted by Pakho
Chau s other three attributes except for the Expertise.

Thirdly, Charlene Choi got the comments ofOrdinary-looking, Untrustworthy, Quite

103

attractive and Not expert in fashion from females. Females were attracted by Charlene
Chois other two attributes except for the Trustworthiness and Expertise.

Fourthly, the common attributes of those celebrities who got the lowest percentage of
votes were Plain-looking and Unattractive.

4.7 Impacts of celebrity endorsement

4.7.1 Reliability of data set
As shown in Table 4.38, all possible impacts of celebrity endorsements were subjected to
the reliability test. According to what suggested by Pavot, Diener, Colvin and Sandvik
(1991), those possible impacts had good internal consistency, with a Cronbach alpha
coefficient reported of 0.85. In the current study the Cronbach alpha coefficient was 0.914.
As a result, they were considered as highly reliable.

Reliability Statistics
Cronbach's

N of

Alpha

Items

.914

17

Table 4.38 Reliability statistics (ii)

The Corrected Item-Total Correlation helped us to indicate the degree to which each item
correlates with the total score (Pallant, J., 2005). Some of the values were lower than 0.5,
104

which meant the items were measuring something different from the scale as a whole.
Due to the overall Cronbach alpha greater than 0.7, thus those with low item-total
correlations needed not to be removed. Detailed tables about the results of reliability test
were illustrated in Appendix XIII.

4.7.2 Consumer buying behavior processes

In the Independent Samples Test, if the Sig. value of the Levene's Test for Equality of
Variances is >0.05, the Sig. (2-tailed) under equal variances assumed and would be used.
If not, equal variances not assumed would be used.

Levene's Test t-test for

for Equality Equality of

Q25 Product awareness

Gender

Mean

Male
Female
Total

3.61
3.62

Male
Female
Total
Male
Female
Total

(Equal variances assumed)

Sig.

Sig.
(2-tailed)

.675

.968

.084

.049

.968

.664

3.57
3.31
3.42

(Equal variances assumed)

Q27 Brand recall before shopping

Means

3.62

(Equal variances assumed)

Q26 Brand Acceptance

of Variances

3.52
3.58
3.55

105

Q28Remember the brand during shopping

Male

3.67

Female
Total

3.66
3.66

(Equal variances assumed)

Q29 Brand alternative evaluation process

Male
Female
Total
Male
Female
Total

.958

.475

.096

.368

.367

3.39
3.15
3.26

(Equal variances assumed)

Q30 Help buying decision

.236

3.15
3.02
3.08

(Equal variances assumed)

Table4.39 Report table of Independent Samples Test

As shown in Tabe4.39, due to the Sig. (2-tailed) of Q25, 27-30 were > 0.05, so
statistically significant difference was not existed between males and females. While the
Sig. (2-tailed) of Q26 was <0.05, statistically significant difference was existed between
genders. Firstly, both genders thought that their awareness on fashion apparel products
was highly increased by the celebrity endorsements, which mean of 3.62.
Secondly, males thought that their acceptance of the fashion brands were able to be
increased by the celebrity endorsements, with mean of 3.57. However, fewer females
thought that their brand acceptance could be increased simply by celebrity endorsements,
with a mean of 3.31. Thirdly, both genders agreed that celebrity endorsements could help
them to recall the endorsed fashion brand before shopping, which had a mean of 3.55.
Fourthly, both gender agreed that celebrity endorsements were useful and able to help
them remember the endorsed fashion brand during shopping, which had the highest mean
106

of 3.66. Fifthly, both genders agreed that celebrity endorsements were able to help them
to evaluate alternative brands during the brand alternatives evaluation process, with a
mean of 3.26. Lastly, both genders thought that celebrity endorsements could help them to
make purchase decisions but the impact might be not as large as other processes, which
had a mean of 3.08.
In conclusion, consumers thought that celebrity endorsement had many positive impacts
on their buying behavior processes. It helped to increase product awareness and brand
acceptance of the consumer, especially males. Moreover consumers could have a higher
chance to recall the fashion brands when shopping if it was endorsed by celebrities.
Furthermore, celebrities played an important role in consumers brand alternative
evaluation process. However, celebrity endorsements might have fewer impacts on
consumers purchase decision. It is because celebrity endorsements could arouse their
purchase intention but the final buying decisions were determined by other factors, such
as price, quality and style of the products. Detailed tables about the test were illustrated in
Appendix IX.

107

4.7.3Effectiveness of celebrity endorsements on fashion apparel

products
Levene's Test t-test for
for Equality Equality of
of Variances
Means
Gender
Q31Feel confused on products, will
believe celebrity endorsement

Mean

Male
Female

2.99
2.75

Total

2.85

(Equal variances assumed)

Q32The brand become more

Male

3.45

favorable

Female
Total

3.13
3.27

(Equal variances assumed)

Q33Grab Attention

Male
Female
Total
Male
Female
Total

(Equal variances assumed)

Sig.

.374

.096

.932

.020

.006

.008

.121

.095

3.80
3.46
3.69

(Equal variances not assumed)

Q34Arouse interested

Sig.
(2-tailed)

3.43
3.21
3.30

Table4.40 Report table of Independent Samples Test (ii)

In Table 4.40, it showed that when consumers felt confused about products, they would
not believe in celebrity endorsements. As Sig. (2-tailed) >0.05, statistically there was no
significant difference among males and females, with a mean value of 2.85.
Secondly, males thought that the fashion brands became more favorable if they were
endorsed by celebrities. They got a mean value of 3.45, which was higher than females
108

mean value of 3.13. Statistically significant difference was existed because of Sig.
(2-tailed) <0.05.
Thirdly, males agreed that the fashion brands with celebrity endorsements could grab
their attentions. They got the highest mean values of 3.8 than females mean of 3.46.
Moreover, significant difference was existed between genders because of Sig. (2-tailed)
<0.05.
Lastly, males and females thought that their interested were aroused by the fashion brands
which had celebrity endorsement. Significant difference was not existed, Sig. (2-tailed)
>0.05, the mean was 3.3.

In conclusion, celebrity endorsements were very effective. It was able to increase the
favorability of the fashion brand, to grab consumers attention more easily and arouse
their interests. Besides, males found that celebrity endorsements were more effective.
Detailed tables about the test were illustrated in Appendix IX.

109

4.7.4 Perceived value of the endorsed fashion products and brands

Gender
Q35Increase product value

Male
Female
Total

Mean

Male
Female

3.67
3.65

Total

3.66

Male
Female
Total

.088

.372

.577

.891

.229

.963

.497

.793

3.71
3.70
3.70

(Equal variances assumed)

Q38 Increase the worth of less

Male

3.65

valued fashion products

Female
Total

3.69

(Equal variances assumed)

Sig.

Sig.
(2-tailed)

3.51

(Equal variances assumed)

Q37 Increase Image and value

t-test for
Equality of
Means

3.57
3.45

(Equal variances assumed)

Q36 Increase brand value

Levene's
Test for
Equality of
Variances

3.67

Table4.41 Report table of Independent Samples Test (iii)

In Table4.41, it showed that no significant different were found between genders, due to
the Sig. (2-tailed) >0.05. Firstly, both genders thought that celebrity endorsements could
increase the value of the endorsed fashion apparel products, with means around 3.5.
Secondly, males and females thought that the value of the fashion brands was largely
increased by celebrity endorsements, with means both over 3.65. Thirdly, both males and
females agreed that the celebrity endorsers' image and value could positively increase the
110

image and value of the endorsed fashion apparel products. The mean values for males and
females were both over 3.7, which were the highest mean scores.

Lastly, both genders

thought that celebrity endorsements could increase the worth of less valued fashion
apparel products, with mean values over 3.65.
In conclusion, celebrity endorsements could increase the perceived value of the endorsed
fashion products and brands. Males and Females agreed that the product value and brand
value were enhanced due to celebrity endorsement. Furthermore, they thought that
celebrity endorsers were able to use their own image and value to positively influence the
image and value of the endorsed products. On the other hand, they agreed that the less
valued products could increase their value if endorsed by the celebrities. Detailed tables
about the test were illustrated in Appendix IX.

4.7.5 Influence of celebrity endorsements on fashion apparel products

Gender
Q39Intend to buy

Mean

Male
Female

3.05
2.86

Total

2.94

(Equal variances assumed)

Q40Know more about the
product
(Equal variances assumed)

Male
Female
Total

Levene's Test
for Equality of
Variances

t-test for
Equality of
Means

Sig.

Sig.
(2-tailed)

.116

.152

.524

.419

3.27
3.15
3.20
111

Q41Willing to pay higher

Male

2.67

price

Female
Total

2.71
2.69

(Equal variances assumed)

.404

.784

Table4.42 Report table of Independent Samples Test (iv)

In Table4.42, it shown that no statistically significant different were found between

genders due to the Sig. (2-tailed) >0.05. Firstly, both genders thought that they were not
intended to buy the fashion apparel products which endorsed by celebrities, with a
relatively low mean of 2.94. Secondly, both genders thought that they would like to know
more about the fashion apparel products which endorsed by celebrities, with mean of 3.2.
Lastly, both genders disagreed that they would pay a higher price for fashion apparel
products which endorsed by celebrities. It got the mean of 2.69, which was the lowest.
In conclusion, both genders thought that they were not intend to purchase the products
which endorsed by the celebrities. In addition, they were not willing to pay higher price
for the fashion apparel products which endorsed by celebrities. However, they would like
to know more about the products which were endorsed by the celebrities. As a result, the
influence of celebrity endorsements on fashion apparel products was neural. Detailed
tables about the test were illustrated in Appendix IX.

112

4.8 Chapter Summary

This chapter has had an in-depth analysis on data about source credibility, celebrity
endorsements and purchase intentions. Further discussion and conclusion of the findings,
limitations and recommendations would be included in the following chapter.

113

CHAPTER 5
CONCLUSION AND
RECOMMENDATION

CHAPTER 5 CONCLUSION AND RECOMMENDATION

5.1 Introduction
This chapter aims to summarize the key findings in the questionnaire survey which have
been reported in chapter 4. All the test results and findings will be integrated and being
discussed together. As a result, a more comprehensive conclusion can be drawn. On the
other hand, the limitations existed in this research and its according recommendations for
further study would also be discussed in detail manner.

5.2Conclusionson the findings

5.2.1 Consumers purchase intention under the perception of celebrity
endorsement
Attributes in source credibility model
Among 15 attributes in the source credibility model, males purchase intentions tended to
be positively aroused by the celebrity endorsers who are experts and experienced in
fashion. But sexy and beautiful celebrities are not able to trigger their purchase on fashion
apparel products.
For females, their purchases are relatively easier to be triggered by knowledgeable
celebrities. On the whole, females purchase intentions are easier to trigger by most of the
attributes than males, especially attractive, beautiful and sincere celebrities. However,
114

sexy celebrities are not able to prompt their purchase intention.

Dimensions of source credibility model

Trustworthiness and expertise of the celebrity endorsement are the most influential
dimension that can positively trigger consumers purchase intention.
However, different genders tend to have similar purchase intention under the three
dimensions of source credibility model. Before the attributes grouped in to three
dimensions, some differences in purchase intentions are existed between genders. As a
result, we can focus on the effects of attributes rather than the dimensions, if we want to
study which particular attribute of celebrities can influence the purchase intentions of
male and female consumers.

Important dimension of source credibility for celebrity endorsers

Trustworthiness is the most important dimensions that the celebrity endorsers must
possess. Second most important factor is found to be expertise while the third is
attractiveness. From the statistics, there is no significant difference among genders within
the dimensions mentioned.

5.2.2 Source credibility

Generally speaking, attractiveness is the most credible source of the celebrity endorsers.
115

Trustworthiness comes the second while expertise ranks the third. When considering the
result according to the genders, attractiveness is only credible to females but not males.
Males think that the most credible source of the celebrity endorsers is trustworthiness
followed by expertise and attractiveness respectively. Meanwhile, the results for females
are same as the population.

5.2.3 Perceived image of celebrity endorsers

By using ten celebrities, the relationships between good-looking, trustworthiness,
attractiveness and expertise are discovered.
Good-looking, Plain looking and Trustworthiness
Trustworthiness is positively affected by Good-looking. As a result, the good-looking
celebrity usually perceive as trustworthy by both genders.
If the fashion apparel products are endorsed by plain-looking celebrity, males think that
Trustworthiness of the celebrity is not affected by Good-looking while females
perceived that the plain celebrities are not trustworthy.

Good-looking, Plain looking and Attractiveness

Attractiveness is positively affected by Good-looking of the celebrity. Thus the
good-looking celebrities usually have a high degree of attractiveness while plain-looking
116

celebrities do not have. Both genders tend to have the same opinion towards this
relationship.

Good-looking, Plain looking and Expertise

If the fashion apparel products are endorsed by good-looking celebrity, males think that
Expertise is positively affected by Good-Looking. In other words they perceive
good-looking celebrities are expertise. For females, they will not perceive the
good-looking celebrities as expertise.
If the fashion apparel products are endorsed by plain-looking celebrity, both males and
females think that the Expertise of the celebrity will not be affected by Good-looking.

Three Favorite celebrity endorsers

Both genders tend to select the celebrities who possess the attributes of Best-looking,
Highly trustworthy, Highly attractive and Know a little about fashion. But the
celebrities who are Plain-looking and Unattractive are the least popular. Furthermore,
both genders tend to select two male celebrities and one female celebrity.

117

5.2.4 Impacts of celebrity endorsement

Consumer buying behavior processes
Celebrity endorsements have positive impacts on consumers buying behavior processes.
Firstly, product awareness and brand acceptance of the consumer are enhanced, especially
for males. Secondly consumers can recall the fashion brands before and during shopping
better with celebrity endorsements. Thirdly celebrities can affect consumers brand
alternative evaluation process. However, celebrity endorsements have fewer impacts on
consumers purchase decision. It is because the purchase intention on fashion apparel
products is affected by other determinant factors, for instance customers can control the
decision themselves.

Effectiveness of celebrity endorsements on fashion apparel products

Celebrity endorsements are very effective to promote the fashion apparel products.
Meanwhile the favorability of the fashion brands is enhanced and consumers attention is
easily grabbed by the celebrity endorsements. Moreover, celebrity endorsements are more
effective on males than females.

Perceived value of the endorsed fashion products and brands

The perceived value of the endorsed fashion products and brands can be enhanced by
118

celebrity endorsements. On the other hand, celebrity endorsers are able to use their own
image and value to positively influence the image and value of the products. In other
words, less valuable products can enhance their values accordingly by celebrity
endorsements.

Influence of celebrity endorsements on fashion apparel products

Consumers do not intended to purchase and are not willing to pay higher price even for
the fashion apparel products are endorsed by the celebrities. However, they want to know
more about the products endorsed by the celebrities. Although consumers are not
intended to make a purchase on the fashion apparel products endorsed by celebrities,
celebrity endorsement have the ability to arouse consumers interest and grab their
attentions. If a need is created, celebrity endorsements can influence consumers
preference and decision. At last it can consequently trigger their purchase.

119

5.3 Recommendation
In Hong Kong, most of the celebrity endorsers are singers, actors or models. They possess
different attributes which are suitable to endorse different kind of products. In fashion
apparel sector, consumers will have different preference on the celebrity endorsers and
their purchase intentions can only be triggered by a particular kind of celebrity endorsers.
In order to make the effect a significant one, there are some suggestions to marketers.

First, the selection of celebrity endorsers with the right attributes is crucial to the success
of celebrity endorsement. It is because celebrity endorsements can help to create and
reinforce the value and image of the fashion product and brand. Thus the markets must
carefully evaluate the celebrities and make sure the image which perceived by the target
customers are positive. For instance males and females think that trustworthiness is the
most important dimension that the celebrity endorsers must possess, so the prerequisite of
the fashion companies is to select the celebrity endorsers who are highly trustworthy.

Second, the markets must understand the relationships between good-looking,

trustworthiness, attractiveness and expertise. Good-looking can give a positive impression
of attractiveness, trustworthiness and expertise to the celebrities and the products.
However, celebrity with plain-looking or ordinary looking can also promote the fashion
120

product effectively if the companies would like to focus on the trustworthiness and
expertise of the celebrities as well as the products.

Finally, the company must have thorough consideration and sufficient research before the
endorsement takes place. They must understand the preference and purchase behaviour of
their target customers. So that the marketers can react to customer and market changes
rapidly

In summary, celebrity endorsement is an effective promotion tool. The companies must

select the right celebrity endorsers in order to grab the attentions and arouse the interest of
the target customers.

5.4 Limitations of study

In this research, there are some limitations. Firstly, the target respondents of the research
should be from males and females from different age groups, however due to the time
constraints and the snow balling effects, most of the respondents are students. Secondly,
the sample of the present research is not derived from probability sampling and therefore
the data obtained should be treated with caution. Moreover, the sample population s is
relatively small to be a formal marketing research. Lastly, the questions which related to
the attributes of the Hong Kong celebrities may have some perceptual difference among
121

the respondents. It is because the perception of good-looking, trustworthiness,

attractiveness and expertise are rather subjective thus it is uneasy to make judgments on
those words.

5.5 Suggestions for future study

In this research, the analysis usually focuses on investigating the significant difference
between genders. However, the results showed that there is not much significant
difference between genders in the purchase intention. As a result, there must be other
factors rather than genders which can affect consumers purchase intention under the
effects of celebrity endorsement. For example, age, occupation or income level.
Besides, focus group, in-depth interview and other exploratory research are highly
recommended for the company who want to investigate the effect of a particular celebrity
to a specific kind of products.

5.6 Chapter summary

By summarizing the relevant data, a comprehensive summary, recommendation and some
suggestions for future studies are presented clearly on the above paragraphs. The research
is generalized without over-focus on specific celebrities, thus accurate information and
insight could be given to the marketers and retailers. Finally, despite some limitations, the
122

research findings could demonstrate in a clear manner that the role of celebrity endorsers
perceived image to consumers purchase intention and the relationships between
good-looking and the dimensions of source credibility model.

123

APPENDIX

124

Appendix I: Questionnaire

I am an undergraduate of the Hong Kong Polytechnic University. I am conducting a research on the Effects of
celebrity endorsement in fashion apparel products to consumers' purchase intention. All the data collected will
be used for academic purpose only and kept strictly confidential. Thank you!

!
Part I: Consumers purchase intention under the perception of celebrity endorsement
.
The following statements describe your purchase intention under the perception of celebrity endorsement in
fashion apparel products.
*Please put a in the appropriate box to indicate how much you agree with each statement.
1.
*
1=Strongly Disagree

2=Disagree

3=Neutral

4=Agree

5=Strongly Agree

1
1

Highly attractive celebrity endorsers can trigger my purchase to their

endorsed fashion apparel products.

Beautiful celebrity endorsers can trigger my purchase to their endorsed

fashion apparel products.

Classy celebrity endorsers can trigger my purchase to their endorsed fashion

apparel products.

Sexy celebrity endorsers can trigger my purchase to their endorsed fashion

apparel products.

Elegant celebrity endorsers can trigger my purchase to their endorsed

fashion apparel products.

Attractiveness of celebrity endorsers is very important.

Highly trustworthy celebrity endorsers can trigger my purchase to their

endorsed fashion apparel products.
125

Dependable celebrity endorsers can trigger my purchase to their endorsed

fashion apparel products.

Honest celebrity endorsers can trigger my purchase to their endorsed fashion

apparel products.

10 Reliable celebrity endorsers can trigger my purchase to their endorsed

fashion apparel products.

11 Sincere celebrity can trigger my purchase to their endorsed fashion apparel

products.
12 Trustworthiness of a celebrity endorser is very important.

13 If celebrity endorsers are experts in fashion, they can trigger my purchase to

their endorsed fashion apparel products./

14 If celebrity endorsers are experienced in fashion, my purchase to their

endorsed fashion apparel products. /

15 If celebrity endorsers are knowledgeable in fashion, they can trigger my

purchase to their endorsed fashion apparel products.
/

16 If celebrity endorsers are qualified in fashion, they can trigger my purchase

to their endorsed fashion apparel products.
/

17 If celebrity endorsers are skilled in fashion, they can trigger my purchase to

their endorsed fashion apparel products.
/

18 Expertise of a celebrity endorser is very important.

Part II. Source credibility

.
*Please put a in the appropriate box to indicate the appropriate answer
*
19

The most credible source of celebrity endorsement is the celebritys____________?

____________

Trustworthiness

Expertise

Attractiveness
126

Part III. Perceived image of celebrity endorsers

.
20-23. In terms of Good-looking, Trustworthiness, Expertise and Attractiveness, please give a score to the
following celebrity endorsers (1= the lowest score, 5= the highest score)
(1=, 5=)
20.Good-looking 21.Trustworthiness

22.Expertise

23.Attractiveness

Daniel Wu
Kaneshiro
Wyman Wong

Sam Lee
PakHo Chau
Angela Baby
Gaile Lai
Hilary Tsui
Josie Ho
Charlene Choi

24. Please select THREE favorite celebrity endorsers. (Remember you can ONLY choose three)
*Please put a in the appropriate box to indicate 3 favorite celebrity endorsers
***()
*

Wyman Wong

Sam Lee

Kaneshiro

PakHo Chau

Angela Baby

Gaile Lai

Hilary Tsui

Josie Ho

Charlene Choi

Daniel Wu

127

Part IV. Impacts of celebrity endorsement

.
The following sentences describe the impacts of celebrity endorsement to consumers.
*Please put a in the appropriate box to indicate the appropriate answer

*
1=Strongly Disagree

2=Disagree

3=Neutral

4=Agree

5=Strongly Agree

1
25

26

27

28

29

30

31

32

33

34

35

36

37

Celebrity endorsements increase my awareness of their endorsed fashion

brands.
Celebrity endorsements increase my acceptance of their endorsed fashion
brands.
Celebrity endorsements help me to recall their endorsed fashion brand before
shopping.
Celebrity endorsements help me to remember their endorsed fashion brand
during shopping.
Celebrity endorsements affect my selection of alternative brands during
evaluation process.
Celebrity endorsements can help me to make the buying decision.

When I am confused about product, I believe celebrity endorsements.

Fashion brand with celebrity endorsement is more favorable.

Fashion brand with celebrity endorsement can grab my attention.

I am interested in fashion brands which use celebrity endorsement.

Celebrity endorsements increase the value of their endorsed fashion apparel

products.
Celebrity endorsements increase the value of their endorsed fashion brand.

Celebrity endorsers' image and value increase their endorsed fashion apparel
products' image and value.
128

38

39

40

41

Celebrity endorsements increase the worth of less valued fashion apparel

products.
I intend to buy the fashion apparel products which endorsed by celebrities.

I want to know more about the fashion apparel products which endorsed by
celebrities.
I am willing to pay higher prices for fashion apparel products which endorsed
by celebrities.

Part V: Personal Data

:
Gender :
Male

Female

Age :
17

25-39

40-54

55

()/

University

Other

18-24

Education level :
Primary school
or below

Secondary
School

(High) Diploma/
Associate degree

Occupations :
Agriculture and Fishing

Mining and Quarrying

Electricity, Gas and Water

Community, Social and Personal Services

Manufacturing

Restaurants and Hotels

Wholesale, Retail and Import/Export

Transport, Storage and Communications

Financing, Insurance, Real Estate and Business

Services

Housewives

Student

Retired

Self-employed

Unemployed

Average monthly income :

$0 - $4999

$5000 - $9999

$10000 - $14999 $15000 - $39999

$40000

-END--

129

Appendix II Demographic data analysis

Gender
Frequency Percent
Valid Male
Female
Total

Valid
Percent

Cumulative Percent

75

43.6

43.6

43.6

97

56.4

56.4

100.0

172

100.0

100.0

Age
Frequency Percent
Valid 17

Valid
Percent

Cumulative
Percent

2.3

2.3

2.3

18-24

146

84.9

84.9

87.2

25-39

18

10.5

10.5

97.7

40-54

2.3

2.3

100.0

Total

172

100.0

100.0

Age * Gender Crosstabulation

Gender
Male
Age

17

Count

Total

4.0%

1.0%

2.3%

59

87

146

78.7%

89.7%

84.9%

10

18

13.3%

8.2%

10.5%

4.0%

1.0%

2.3%

75

97

172

% within Gender 100.0%

100.0%

100.0%

% within Gender
18-24 Count
% within Gender
25-39 Count
% within Gender
40-54 Count
% within Gender
Total

Female

Count

130

Education
Frequency Percent
Valid Secondary
School
HD/AD
University

Cumulative
Percent

19

11.0

11.0

11.0

48

27.9

27.9

39.0

104

60.5

60.5

99.4

.6

.6

100.0

172

100.0

100.0

Others
Total

Valid
Percent

Education * Gender Crosstabulation

Gender
Male
Education Secondary
School
HD/AD

Count
% within Gender

University
Others

10

19

12.0%

10.3%

11.0%

19

29

48

25.3%

29.9%

27.9%

47

57

104

62.7%

58.8%

60.5%

.0%

1.0%

.6%

75

97

172

100.0%

100.0%

100.0%

Count
% within Gender
Count
% within Gender

Total

Count
% within Gender

Total

Count
% within Gender

Female

Occupations

Valid

Frequency

Percent

Valid
Percent

Cumulative
Percent

Electricity, Gas and Water

.6

.6

.6

Community, Social and

5.2

5.2

5.8

Manufacturing

2.3

2.3

8.1

Restaurants and Hotels

3.5

3.5

11.6

Wholesale, Retail and

Import/Export

27

15.7

15.7

27.3

Personal Services

131

Transport, Storage and

5.2

5.2

32.6

5.2

5.2

37.8

105

61.0

61.0

98.8

1.2

1.2

100.0

172

100.0

100.0

Communications
Financing, Insurance, Real
Estate and Business
Services
Student
Unemployed
Total

Occupations * Gender Crosstabulation

Gender
Male
Occupations

Electricity, Gas and

Water

% within Gender

.0%

1.0%

.6%

Community, Social and Count

Personal Services
% within Gender

8.0%

3.1%

5.2%

1.3%

3.1%

2.3%

4.0%

3.1%

3.5%

20

27

% within Gender

9.3%

20.6%

15.7%

Transport, Storage and Count

Communications
% within Gender

6.7%

4.1%

5.2%

10.7%

1.0%

5.2%

45

60

105

60.0%

61.9%

61.0%

.0%

2.1%

1.2%

75

97

172

100.0%

100.0%

100.0%

Count
% within Gender

Restaurants and Hotels Count

% within Gender
Wholesale, Retail and
Import/Export

Count

Financing, Insurance,
Real Estate and
Business Services

Count

Student

Count

% within Gender

% within Gender
Unemployed

Count
% within Gender

Total

Total

Manufacturing

Count

Female

Count
% within Gender

132

Average monthly Income

Frequency

Percent

Valid
Percent

104

60.5

60.5

60.5

$5000 $9999

25

14.5

14.5

75.0

$10000 $14999

28

16.3

16.3

91.3

$15000 $39999

11

6.4

6.4

97.7

$40000

2.3

2.3

100.0

172

100.0

100.0

Valid $0 - $4999

Total

Cumulative
Percent

Average monthly Inocme * Gender Crosstabulation

Gender
Male
Inocme $0 - $4999

Count
% within
Gender

$5000 - $9999

Count
% within
Gender

$10000 $14999

Count

$15000 $39999

Count

$40000

Count

% within
Gender
% within
Gender
% within
Gender

Total

Count
% within
Gender

Female

Total

41

63

104

54.7%

64.9%

60.5%

10

15

25

13.3%

15.5%

14.5%

14

14

28

18.7%

14.4%

16.3%

11

9.3%

4.1%

6.4%

4.0%

1.0%

2.3%

75

97

172

100.0%

100.0%

100.0%

133

Appendix III: Reliability test of the attributes of celebrity endorsers

Case Processing Summary

N
Cases

Valid
Excluded

Total

172

100.0

.0

172

100.0

a. Listwise deletion based on all

variables in the procedure.

Reliability Statistics
Cronbach's

N of

Alpha

Items

.935

15

Item-Total Statistics
Scale Mean
Scale
if Item
Variance if
Deleted
Item Deleted

Corrected
Item-Total
Correlation

Cronbach's
Alpha if Item
Deleted

Q1Attractive
Q2Beautiful
Q3Classy
Q4Sexy
Q5Elegant
Q7Trustworthy
Q8Dependable
Q9Honest
Q10Reliable
Q11Sincere

47.92
48.13
47.85
48.41
47.96
47.76
47.78
47.69
47.67
47.70

96.725
95.633
95.766
97.016
97.817
94.735
94.138
95.641
94.901
95.941

.628
.673
.675
.578
.595
.751
.776
.700
.708
.683

.932
.931
.931
.934
.933
.929
.928
.930
.930
.931

Q13Experts
Q14Experienced
Q15Knowledgeable
Q16Qualified
Q17Skilled

47.64
47.62
47.63
47.67
47.67

96.641
97.443
96.481
97.042
96.176

.676
.651
.702
.641
.723

.931
.932
.930
.932
.930

134

Appendix VI:
Appendix IV: Factors analysis of the attributes of celebrity endorsers

Correlation

Correlation Matrix
Q1
Attractive

Q2
Beautiful

Q3
Classy Q4Sexy

Q5
Elegant

Q1Attractive

1.000

.758

.487

.506

.473

Q2Beautiful

.758

1.000

.637

.540

.442

Q3Classy

.581

.640

.492

.544

.476

Q4Sexy

.487

.637

1.000

.434

.410

Q5Elegant

.506

.540

.434

1.000

.491

Q7Trustworthy

.473

.442

.410

.491

1.000

Q8Dependable

.463

.457

.442

.524

.883

Q9Honest

.380

.363

.376

.447

.719

Q10Reliable

.399

.370

.354

.472

.721

Q11Sincere

.389

.358

.321

.514

.694

Q13Experts

.430

.499

.421

.318

.392

Q14Experienced

.357

.441

.376

.286

.405

Q15Knowledgeable

.416

.451

.377

.376

.494

Q16Qualified

.313

.365

.343

.250

.510

Q17Skilled

.438

.508

.457

.362

.458

Correlation Matrix

Correlation

Q7
Trustwort
Q8
Q9
Q10
hy
Dependable Honest Reliable

Q11
Sincere

Q1Attractive

.581

.463

.380

.399

.389

Q2Beautiful

.640

.457

.363

.370

.358

Q3Classy

1.000

.467

.402

.401

.394

Q4Sexy

.492

.442

.376

.354

.321

Q5Elegant

.544

.524

.447

.472

.514

Q7Trustworthy

.476

.883

.719

.721

.694

Q8Dependable

.467

1.000

.792

.785

.791

Q9Honest

.402

.792

1.000

.882

.789
135

Q10Reliable

.401

.785

.882

1.000

.852

Q11Sincere

.394

.791

.789

.852

1.000

Q13Experts

.595

.364

.344

.366

.306

Q14Experienced

.440

.404

.368

.344

.309

Q15Knowledgeable

.551

.485

.391

.405

.379

Q16Qualified

.366

.467

.413

.402

.395

Q17Skilled

.525

.467

.404

.412

.418

Correlation Matrix
Q13

Q14

Q15

Correlation

Expert Experienc Knowledgea

Q16
s
ed
ble
Qualified

Q17
Skilled

Q1Attractive

.430

.357

.416

.313

.438

Q2Beautiful

.499

.441

.451

.365

.508

Q3Classy

.595

.440

.551

.366

.525

Q4Sexy

.421

.376

.377

.343

.457

Q5Elegant

.318

.286

.376

.250

.362

Q7Trustworthy

.392

.405

.494

.510

.458

Q8Dependable

.364

.404

.485

.467

.467

Q9Honest

.344

.368

.391

.413

.404

Q10Reliable

.366

.344

.405

.402

.412

Q11Sincere

.306

.309

.379

.395

.418

Q13Experts

1.000

.792

.703

.671

.734

Q14Experienced

.792

1.000

.722

.690

.776

Q15Knowledgeable

.703

.722

1.000

.719

.714

Q16Qualified

.671

.690

.719

1.000

.707

Q17Skilled

.734

.776

.714

.707

1.000

KMO and Bartlett's Test

Kaiser-Meyer-Olkin Measure of Sampling
Adequacy.
Bartlett's Test of
Approx.
Sphericity
Chi-Square

.903
2225.044

df

105

Sig.

.000
136

Total Variance Explained

Initial Eigenvalues

Extraction Sums of Squared

Loadings

Rotation Sums of Squared

Loadings

Component

% of Cumulative
% of Cumulative
% of Cumulative
Total Variance
%
Total Variance
%
Total Variance
%
1

7.929

52.858

52.858

7.929

52.858

52.858

4.214

28.093

28.093

2.100

14.003

66.860

2.100

14.003

66.860

3.941

26.270

54.363

1.448

9.656

76.516

1.448

9.656

76.516

3.323

22.153

76.516

.596

3.975

80.491

.505

3.367

83.859

.452

3.016

86.875

.417

2.781

89.656

.297

1.978

91.634

.264

1.759

93.393

10 .254

1.691

95.085

11 .209

1.395

96.480

12 .189

1.261

97.741

13 .152

1.014

98.755

14 .104

.695

99.450

15 .083

.550

100.000

Extraction Method: Principal Component Analysis.

137

Appendix V: Independent t-test

4.4.3.1 Impacts of attributes of celebrity endorsement to consumers purchase intention
Report
Q1

Q3

Q7

Attract Beauti Class Q4

ive
ful
y
Sexy

Gender
Male

Q2

Mean

Q8

Q10

Q11

Q5
Trustw Depend
Q9 Reliabl Sincer
Elegant orthy
able
Honest
e
e

Q14
Q13
Experts

Q15

Experi Knowle
enced dgeable

Q16
Qualified

Q17
Skilled

3.12

2.92

3.21

2.73

3.20

3.37

3.35

3.41

3.44

3.35

3.56

3.53

3.45

3.40

3.49

75

75

75

75

75

75

75

75

75

75

75

75

75

75

75

.997 1.056 1.095

.959

.969

.979

.974

1.017

1.007

1.003

.977

.934

1.053

1.005

3.43

3.23

3.49

2.87

3.31

3.54

3.52

3.62

3.64

3.65

3.60

3.66

3.70

3.67

3.60

97

97

97

97

97

97

97

97

97

97

97

97

97

97

97

Std.
Deviation

.912

.995

.926

.986

.939

.969

.980

.962

1.002

.925

.874

.840

.880

.838

.825

Mean

3.30

3.09

3.37

2.81

3.26

3.47

3.44

3.53

3.55

3.52

3.58

3.60

3.59

3.55

3.55

172

172

172

172

172

172

172

172

172

172

172

172

172

172

172

Std.
Deviation

.985

1.004

.992 1.034

.947

.970

.981

.970

1.010

.970

.930

.902

.910

.944

.907

N
Std.

1.052

Deviation
Female Mean
N

Total

138

Independent Samples Test

Levene's Test for
Equality of Variances

t-test for Equality of Means

95% Confidence

F
Q1Attractive

Equal variances assumed

Sig.

5.625

.019

Equal variances not assumed

Q2Beautiful

Equal variances assumed

.127

.722

Equal variances not assumed

Q3Classy

Equal variances assumed

3.400

.067

Equal variances not assumed

Q4Sexy

Equal variances assumed

1.428

.234

Equal variances not assumed

Q5Elegant

Equal variances assumed

.033

.857

Equal variances not assumed

Q7Trustworthy

Equal variances assumed

.003

.957

Equal variances not assumed

Q8Dependable

Equal variances assumed

.064

.800

Equal variances not assumed

Q9Honest

Equal variances assumed

.527

.469

t
-2.087

df

Sig.
(2-tailed)

Mean
Std. Error
Difference Difference

Interval of the
Difference
Lower

Upper

170

.038

-.313

.150

-.609

-.017

-2.049 146.750

.042

-.313

.153

-.615

-.011

-2.004

170

.047

-.307

.153

-.609

-.005

-2.004 159.126

.047

-.307

.153

-.609

-.004

-1.859

170

.065

-.282

.151

-.580

.017

-1.828 147.783

.070

-.282

.154

-.586

.023

170

.405

-.133

.159

-.447

.181

-.823 150.322

.412

-.133

.161

-.451

.186

-.750

170

.454

-.109

.146

-.397

.178

-.748 157.682

.456

-.109

.146

-.398

.179

170

.276

-.163

.149

-.457

.131

-1.092 159.232

.276

-.163

.149

-.457

.132

-1.120

170

.264

-.169

.151

-.466

.129

-1.120 159.331

.264

-.169

.151

-.466

.129

-1.380

.169

-.205

.149

-.499

.088

-.834

-1.092

170

139

Equal variances not assumed

Q10Reliable

Equal variances assumed

.125

.724

Equal variances not assumed

Q11Sincere

Equal variances assumed

2.373

.125

Equal variances not assumed

Q13Experts

Equal variances assumed

1.902

.170

Equal variances not assumed

Q14Experienced

Equal variances assumed

2.958

.087

Equal variances not assumed

Q15Knowledgeable Equal variances assumed

1.482

.225

Equal variances not assumed

Q16Qualified

Equal variances assumed

8.680

.004

Equal variances not assumed

Q17Skilled

Equal variances assumed

Equal variances not assumed

5.238

.023

-1.378 158.335

.170

-.205

.149

-.499

.089

-1.285

170

.201

-.199

.155

-.505

.107

-1.282 158.136

.202

-.199

.155

-.506

.108

-2.049

170

.042

-.303

.148

-.595

-.011

-2.027 152.138

.044

-.303

.149

-.598

-.008

170

.792

-.038

.143

-.321

.245

-.260 147.238

.795

-.038

.146

-.326

.250

-.911

170

.363

-.126

.139

-.400

.147

-.894 146.003

.373

-.126

.141

-.406

.153

170

.077

-.248

.139

-.522

.027

-1.768 154.353

.079

-.248

.140

-.524

.029

-1.874

170

.063

-.270

.144

-.555

.014

-1.820 138.650

.071

-.270

.148

-.563

.023

170

.455

-.105

.140

-.380

.171

-.731 141.558

.466

-.105

.143

-.388

.178

-.265

-1.782

-.749

140

Appendix VI: Paired sample t-test

4.4.3.2 Impacts of source credibility of celebrity endorsers to consumers purchase intention
Paired Samples Statistics
Mean
Pair 1
Pair 2
Pair 3

Std.

Std. Error

Deviation

Mean

Mean of Attractiveness

3.1663

172

.80002

.06100

Mean of Trustworthiness
Mean of Attractiveness
Mean of Expertise
Mean of Trustworthiness

3.5012
3.1663
3.5767
3.5012

172
172
172
172

.89456
.80002
.81012
.89456

.06821
.06100
.06177
.06821

Mean of Expertise

3.5767

172

.81012

.06177

Paired Samples Correlations

Correlatio
N
Pair 1
Pair 2
Pair 3

Mean of Attractiveness & Mean of Trustworthiness

Mean of Attractiveness & Mean of Expertise
Mean of Trustworthiness & Mean of Expertise

172
172
172

Sig.
.574
.578
.502

.000
.000
.000

141

4.4.3.2 Impacts of source credibility of celebrity endorsers to consumers purchase intention

Paired Samples Test

Paired Differences

Mean
Pair 1 Mean of Attractiveness
- Mean of
Trustworthiness
Pair 2 Mean of Attractiveness
- Mean of Expertise
Pair 3 Mean of
Trustworthiness -

Std.
Deviation

Std. Error
Mean

95% Confidence Interval of

the Difference
Lower

Upper

df

Sig.
(2-tailed)

-.33488

.78670

.05999

-.45329

-.21648

-5.583

171

.000

-.41047

.73993

.05642

-.52183

-.29910

-7.275

171

.000

-.07558

.85381

.06510

-.20409

.05293

-1.161

171

.247

Mean of Expertise

142

Appendix VII: Independent t-test

4.4.3.2 Impacts of source credibility of celebrity endorsers to consumers purchase intention, by genders

Report
Gender
Male

Mean of Attractiveness
Mean

3.3840

3.4880

75

75

75

Std. Deviation

.80987

.90151

.87379

Mean

3.2660

3.5918

3.6454

97

97

97

Std. Deviation

.78194

.88314

.75471

Mean

3.1663

3.5012

3.5767

172

172

172

.80002

.89456

.81012

N
Total

Mean of Expertise

3.0373

N
Female

Mean of Trustworthiness

N
Std. Deviation

143

Independent Samples Test

Levene's Test for
Equality of
Variances

F
Mean of Attractiveness Equal variances
assumed

.956

Equal variances not

assumed
Mean of
Trustworthiness

Equal variances
assumed

.345

Equal variances not

assumed
Mean of Expertise

Equal variances
assumed
Equal variances not
assumed

1.571

Sig.

t-test for Equality of Means

95% Confidence
Interval of the
Difference

Sig.
(2-tailed)

Mean
Difference

Std. Error
Difference

170

.063

-.22865

.12212

-.46971

.01242

-1.864 156.463

.064

-.22865

.12267

-.47095

.01366

170

.131

-.20775

.13703

-.47825

.06275

-1.512 157.652

.133

-.20775

.13739

-.47912

.06361

170

.207

-.15736

.12435

-.40282

.08810

-1.242 146.434

.216

-.15736

.12670

-.40775

.09303

.330 -1.872

.558 -1.516

.212 -1.265

df

Lower

Upper

144

Appendix VIII: Paired sample t-test

4.4.3.3 Important dimension of source credibility for celebrity endorsers
Paired Samples Statistics
Mean
Pair 1
Pair 2
Pair 3

Std. Deviation

Std. Error Mean

Importance of Attractiveness

3.43

172

1.026

.078

Importance of Trustworthiness

3.90

172

.912

.070

Importance of Attractiveness
Importance of Expertise
Importance of Trustworthiness

3.43
3.63
3.90

172
172
172

1.026
.943
.912

.078
.072
.070

Importance of Expertise

3.63

172

.943

.072

Paired Samples Test

Paired Differences

Mean
Pair 1
Pair 2
Pair 3

Importance of Attractiveness Importance of Trustworthiness

Importance of Attractiveness Importance of Expertise
Importance of Trustworthiness Importance of Expertise

Std.
Deviation

Std. Error
Mean

95% Confidence Interval

of the Difference
Lower

Upper

df

Sig.
(2-tailed)

-.465

1.078

.082

-.627

-.303

-5.658

171

.000

-.198

1.231

.094

-.383

-.012

-2.105

171

.037

.267

1.048

.080

.110

.425

3.348

171

.001

145

Appendix IX: Independent t-test

4.4.3.3 Important dimension of source credibility for celebrity endorsers, by genders
Group Statistics
Gender

Mean

Std.
Deviation

Std. Error
Mean

Importance of
Attractiveness

Male

75

3.36

1.022

.118

Female

97

3.48

1.032

.105

Importance of
Trustworthiness

Male

75

3.77

.924

.107

Female

97

3.99

.896

.091

Importance of
Expertise

Male

75

3.60

1.027

.119

Female

97

3.65

.878

.089

Independent Samples Test

Levene's Test for
Equality of Variances

F
Importance of
Attractiveness

Equal variances assumed

Importance of
Trustworthiness

Equal variances assumed

Importance of

Equal variances assumed

Expertise

Sig.
.096

.757

Equal variances not assumed

1.265

.262

Equal variances not assumed

Equal variances not assumed

3.608

.059

t-test for Equality of Means

t
-.788

df

Sig.

Mean

Std. Error

95% Confidence
Interval of the
Difference

(2-tailed)

Difference

Difference

Lower

Upper

170

.432

-.125

.158

-.436

.187

-.789 160.013

.431

-.125

.158

-.436

.187

170

.123

-.216

.140

-.492

.059

-1.544 156.799

.125

-.216

.140

-.493

.060

170

.734

-.049

.145

-.337

.238

-.334 145.542

.739

-.049

.148

-.343

.244

-1.550
-.340

146

Appendix X: Descriptive Statistics

The most credibility source of celebrity endorsement
Q19SourceCredibility
Frequency Percent
Valid

Valid
Percent

Cumulative
Percent

Trustworthiness

62

36.0

36.0

36.0

Expertise

47

27.3

27.3

63.4

Attractiveness

63

36.6

36.6

100.0

172

100.0

100.0

Total

Q19SourceCredibility * Gender Crosstabulation

Gender
Male
Q19SourceCredibility Trustworthiness

Count
% within
Gender

Expertise

Count
% within

Female

Total

34

28

62

45.3%

28.9%

36.0%

23

24

47

30.7%

24.7%

27.3%

18

45

63

24.0%

46.4%

36.6%

75

97

172

100.0%

100.0%

100.0%

Gender
Attractiveness

Count
% within
Gender

Total

Count
% within
Gender

147

Appendix XI: Reliability test

Four attributes of celebrity endorsers
Case Processing Summary
N
Cases

Valid
Excluded

Total

172

100.0

.0

172

100.0

a. Listwise deletion based on all

variables in the procedure.

Reliability Statistics
Cronbach's

N of

Alpha

Items

.912

40

Item-Total Statistics
Scale
Mean if
Item
Deleted

Scale
Variance if

Corrected
Item-Total

Item Deleted Correlation

Cronbach's
Alpha if
Item
Deleted

Q20Daniel_Good-Looking
Q20 _Good-Looking
Q20Wyman_Good-Looking
Q20Sam_Good-Looking
Q20PakHo_Good-Looking
Q20Angelababy_Good-Looking
Q20Gaile_Good-Looking
Q20Hilary_Good-Looking
Q20Josie_Good-Looking

124.46
124.64
126.46
126.35
125.49
125.26
125.65
126.30
126.25

332.203
330.150
324.694
323.269
319.783
321.595
327.363
320.095
319.007

.232
.259
.435
.420
.498
.425
.318
.483
.552

.912
.912
.910
.911
.910
.911
.912
.910
.909

Q20Charlene_Good-Looking
Q21Daniel_Trustworthiness
Q21 _Trustworthiness
Q21Wyman_Trustworthiness
Q21Sam_Trustworthiness
Q21PakHo_Trustworthiness

125.60
124.99
125.02
125.19
125.73
125.76

321.656
331.064
330.362
323.630
322.291
323.110

.473
.285
.300
.492
.513
.547

.910
.912
.912
.910
.909
.909
148

Q21Angelababy_Trustworthiness

126.16

325.478

.385

.911

Q21Gaile_Trustworthiness
Q21Hilary_Trustworthiness
Q21Josie_Trustworthiness
Q21Charlene_Trustworthiness
Q22Daniel_Attractiveness
Q22 _Attractiveness
Q22Wyman_Attractiveness
Q22Sam_Attractiveness
Q22PakHo_Attractiveness
Q22Angelababy_Attractiveness

125.59
125.78
125.78
126.00
124.62
124.83
126.12
126.14
125.65
125.57

324.653
320.532
319.670
323.427
330.203
326.788
320.283
321.185
319.388
323.720

.506
.589
.596
.407
.280
.335
.455
.469
.508
.368

.910
.909
.908
.911
.912
.912
.910
.910
.909
.911

Q22Gaile_Attractiveness
Q22Hilary_Attractiveness
Q22Josie_Attractiveness

125.69
126.19
126.25

325.655
319.910
318.902

.363
.490
.585

.911
.910
.909

Q22Charlene_Attractiveness
Q23Daniel_Expertise
Q23 _Expertise
Q23Wyman_Expertise
Q23Sam_Expertise
Q23PakHo_Expertise
Q23Angelababy_Expertise
Q23Gaile_Expertise

125.72
125.48
125.55
124.71
125.69
125.92
126.10
125.34

321.304
332.859
330.179
325.728
322.425
322.222
322.967
326.190

.439
.215
.301
.424
.506
.502
.427
.358

.910
.913
.912
.910
.910
.910
.910
.911

Q23Hilary_Expertise
Q23Josie_Expertise
Q23Charlene_Expertise

125.31
125.73
126.06

319.828
316.691
322.189

.466
.641
.530

.910
.908
.909

149

Appendix XI: Compare means

Rating of four attributes for celebrities, by genders
Report
Average mean Average mean
of celebrities of celebrities
Good-Looking Trustworthy

Gender
Male

Mean
N
Std.
Deviation

Female Mean
N
Std.
Deviation

Average mean Average mean

of celebrities of celebrities
Attractive
Expertise

3.2253

3.3320

3.2080

3.3373

75

75

75

75

.59980

.52406

.54469

.49831

3.1856

3.1845

3.1454

3.2010

97

97

97

97

.54467

.54397

.59844

.64202

Ratings of celebrities
Means of four attributes for celebrities, by genders
Daniel Wu

Report
Q22Daniel_
Q20Daniel_G Q21Daniel_Trust Attractivene Q23Daniel_E
ood-Looking
worthiness
ss
xpertise

Gender
Male

Mean

4.24

3.91

4.01

3.35

75

75

75

75

.998

.738

.993

.626

4.51

3.82

4.40

3.39

97

97

97

97

.647

.829

.731

.930

Mean

4.39

3.86

4.23

3.37

172

172

172

172

Std.
Deviation

.827

.790

.874

.810

N
Std.
Deviation
Female Mean
N
Std.
Deviation
Total

150

Takeshi Kaneshiro

Report
Q20
Q21
Q22
_Good-Look _Trustworthi _Attractivene
ing
ness
ss

Gender
Male

Mean

4.24

3.91

4.00

3.35

75

75

75

75

.998

.791

1.040

.668

4.19

3.76

4.04

3.27

97

97

97

97

Std.
Deviation

.894

.826

.967

.930

Mean

4.21

3.83

4.02

3.30

172

172

172

172

Std.
Deviation

.938

.812

.997

.825

N
Std.
Deviation
Female Mean
N

Total

Q23
_Expertise

Wyman Wong

Report
Q20Wyman_
Q22Wyman
Good-Looki Q21Wyman_T _Attractiven Q23Wyman
ng
rustworthiness
ess
_Expertise

Gender
Male

Mean

2.56

3.91

2.88

4.24

75

75

75

75

.858

.756

1.026

.803

2.26

3.46

2.61

4.06

97

97

97

97

Std.
Deviation

.939

.914

1.186

.922

Mean

2.39

3.66

2.73

4.14

172

172

172

172

Std.
Deviation

.914

.874

1.124

.874

N
Std.
Deviation
Female Mean
N

Total

151

Sam Lee
Report
Q21Sam_ Q22Sam_
Q20Sam_Go Trustworth Attractiven Q23Sam_E
od-Looking
iness
ess
xpertise

Gender
Male

Mean

2.64

3.25

2.76

3.41

75

75

75

75

1.048

.887

.883

.718

2.39

3.01

2.67

2.96

97

97

97

97

1.006

.919

1.161

.999

Mean

2.50

3.12

2.71

3.16

172

172

172

172

1.029

.910

1.047

.914

N
Std.
Deviation
Female Mean
N
Std.
Deviation
Total

Std.
Deviation

PakHo Chau
Report
Q20PakHo_ Q21PakHo_ Q22PakHo_
Good-Looki Trustworthin Attractivene Q23PakHo_
ng
ess
ss
Expertise

Gender
Male

Mean

3.60

3.36

3.45

3.09

75

75

75

75

1.040

.765

.949

.918

3.18

2.89

3.01

2.79

97

97

97

97

1.051

.802

1.113

.924

Mean

3.36

3.09

3.20

2.92

172

172

172

172

1.064

.818

1.065

.930

N
Std.
Deviation
Female Mean
N
Std.
Deviation
Total

Std.
Deviation

152

Angelababy

Report
Q20Angelab Q21Angelab Q22Angelab Q23Angelab
aby_Good-L aby_Trustwo aby_Attracti aby_Expertis
ooking
rthiness
veness
e

Gender
Male

Mean

3.68

2.72

3.47

2.96

75

75

75

75

1.080

.980

.859

.877

3.53

2.67

3.13

2.59

97

97

97

97

1.146

.965

1.280

1.116

Mean

3.59

2.69

3.28

2.75

172

172

172

172

1.117

.969

1.125

1.032

N
Std.
Deviation
Female Mean
N
Std.
Deviation
Total

Std.
Deviation

Gaile Lai

Report
Q21Gaile_ Q22Gaile_
Q20Gaile_G Trustworthi Attractiven Q23Gaile_
ood-Looking
ness
ess
Expertise

Gender
Male

Mean

3.05

3.19

3.17

3.47

75

75

75

75

.943

.748

.891

.905

3.31

3.32

3.15

3.55

97

97

97

97

1.034

.836

1.093

1.041

Mean

3.20

3.26

3.16

3.51

172

172

172

172

1.001

.799

1.007

.982

N
Std.
Deviation
Female Mean
N
Std.
Deviation
Total

Std.
Deviation

153

Hilary Tsui

Report
Q20Hilary_ Q21Hilary_ Q22Hilary_
Good-Looki Trustworthi Attractiven Q23Hilary_
ng
ness
ess
Expertise

Gender
Male

Mean

2.48

3.07

2.61

3.61

75

75

75

75

1.031

.875

.957

1.089

2.60

3.06

2.69

3.48

97

97

97

97

1.115

.888

1.158

1.156

Mean

2.55

3.06

2.66

3.54

172

172

172

172

1.078

.880

1.073

1.126

N
Std.
Deviation
Female Mean
N
Std.
Deviation
Total

Std.
Deviation

Josie Ho

Report
Q21Josie_T Q22Josie_
Q20Josie_G rustworthin Attractiven Q23Josie_E
ood-Looking
ess
ess
xpertise

Gender
Male

Mean

2.53

3.11

2.61

3.19

75

75

75

75

1.031

.894

.928

.940

2.65

3.04

2.59

3.07

97

97

97

97

Std.
Deviation

.990

.923

.987

1.003

Mean

2.60

3.07

2.60

3.12

172

172

172

172

1.007

.909

.959

.975

N
Std.
Deviation
Female Mean
N

Total

Std.
Deviation

154

Charlene Choi

Report
Q20Charlene Q21Charlene Q22Charlene
_Good-Look _Trustworthi _Attractivene Q23Charlene
ing
ness
ss
_Expertise

Gender
Male

Mean

3.23

2.91

3.11

2.71

75

75

75

75

1.098

1.153

1.098

.802

3.26

2.80

3.15

2.85

97

97

97

97

Std.
Deviation

.950

.964

1.112

.950

Mean

3.24

2.85

3.13

2.78

172

172

172

172

1.014

1.049

1.103

.889

N
Std.
Deviation
Female Mean
N

Total

Std.
Deviation

155

Appendix XII: Relationships between Good-looking, Trustworthiness, Expertise

and Attractiveness of celebrity endorsers.
4.6.3.1 Good-looking and Trustworthiness
Good-looking celebrity & Trustworthiness

Model Summaryb
Gender Model
Male

Female 1

Adjusted R
Square

R Square

Std. Error of
the Estimate

.189

.178

.670

.116

.107

.784

.434
.341

a. Predictors: (Constant), Q20Daniel_ Good-Looking

b. Dependent Variable: Q21Daniel_Trustworthiness

ANOVAb
Sum of
Squares

Gender Model
Male

Female 1

Regression

Mean
Square

df

7.611

7.611

Residual

32.736

73

.448

Total

40.347

74

7.686

7.686

Residual

58.335

95

.614

Total

66.021

96

Regression

Sig.

16.971

.000a

12.516

.001a

a. Predictors: (Constant), Q20Daniel_ Good-Looking

b. Dependent Variable: Q21Daniel_ Trustworthiness

Coefficientsa
Unstandardized
Coefficients
Gender Model
Male

B
(Constant)
Q20Daniel_
Good-Looking

Female 1

(Constant)
Q20Daniel_
Good-Looking

Standardized
Coefficients

Std. Error

2.544

.340

.321

.078

1.856

.562

.437

.124

Beta

t
.434

.341

Sig.

7.489

.000

4.120

.000

3.302

.001

3.538

.001

a. Dependent Variable: Q21Daniel_ Trustworthiness

156

Plain-looking celebrity & Trustworthiness

Model Summaryb
Gender Model
Male

Female 1

Adjusted R
Square

R Square

Std. Error of
the Estimate

.010

-.003

.758

.050

.040

.895

.102
.224

a. Predictors: (Constant), Q20Wyman_ Good-Looking

b. Dependent Variable: Q21Wyman_ Trustworthiness

ANOVAb
Sum of
Gender Model
Male

Female 1

Squares
Regression

Mean
df

Square

.444

.444

Residual

41.902

73

.574

Total

42.347

74

4.005

4.005

Residual

76.119

95

.801

Total

80.124

96

Regression

Sig.

.774

.382a

4.998

.028a

a. Predictors: (Constant), Q20Wyman_ Good-Looking

b. Dependent Variable: Q21Wyman_ Trustworthiness

Coefficientsa
Unstandardized
Coefficients
Gender Model
Male

B
(Constant)
Q20Wyman_
Good-Looking

Female 1

(Constant)
Q20Wyman_
Good-Looking

Standardized
Coefficients

Std. Error

3.675

.277

.090

.103

2.973

.238

.218

.097

Beta

t
.102

.224

Sig.

13.271

.000

.880

.382

12.499

.000

2.236

.028

a. Dependent Variable: Q21Wyman_ Trustworthiness

157

4.6.3.2 Good-looking and Attractiveness

Good-looking celebrity & Attractiveness

Model Summaryb
Gender Model
Male

Female 1

Adjusted R
Square

R Square

Std. Error of
the Estimate

.460

.453

.734

.603

.599

.463

.679
.777

a. Predictors: (Constant), Q20Daniel_ Good-looking

b. Dependent Variable: Q22Daniel_ Attractiveness

ANOVAb
Sum of
Squares

Gender Model
Male

Female 1

Mean
Square

df

Sig.

62.294

.000a

30.959 144.450

.000a

Regression

33.606

33.606

Residual

39.381

73

.539

Total

72.987

74

Regression

30.959

Residual

20.361

95

Total

51.320

96

.214

a. Predictors: (Constant), Q20Daniel_ Good-Looking

b. Dependent Variable: Q22Daniel_ Attractiveness

Coefficientsa
Unstandardized
Coefficients
Gender Model
Male

B
(Constant)

Standardized
Coefficients

Std. Error

1.150

.373

.675

.086

(Constant)

.451

.332

Q20Daniel_
Good-Looking

.877

.073

Q20Daniel_

Beta

t
.679

Sig.

3.086

.003

7.893

.000

1.357

.178

12.019

.000

Good-Looking
Female 1

.777

a. Dependent Variable: Q22Daniel_ Attractiveness

158

Plain-looking celebrity & Attractiveness

Model Summaryb
Adjusted R
Square

R Square

Std. Error of
the Estimate

Gender Model

Male

.522a

.273

.263

.881

Female 1

.616a

.379

.372

.940

a. Predictors: (Constant), Q20Wyman_ Good-Looking

b. Dependent Variable: Q22Wyman_ Attractiveness

ANOVAb
Sum of
Squares

Gender Model
Male

Female 1

Mean
Square

df

Regression

21.269

21.269

Residual

56.651

73

.776

Total

77.920

74

Regression

51.194

51.194

Residual

83.919

95

.883

135.113

96

Total

Sig.

27.407

.000a

57.954

.000a

a. Predictors: (Constant), Q20Wyman_ Good-Looking

b. Dependent Variable: Q22Wyman_ Attractiveness

Coefficientsa
Unstandardized
Coefficients
Gender Model
Male

Female 1

B
(Constant)

Standardized
Coefficients

Std. Error

1.280

.322

Q20Wyman_
Good-Looking

.625

.119

(Constant)

.851

.250

Q20Wyman_
Good-Looking

.778

.102

Beta

t
.522

.616

Sig.

3.976

.000

5.235

.000

3.410

.001

7.613

.000

a. Dependent Variable: Q22Wyman_ Attractiveness

159

4.6.3.3 Good-looking and Expertise

Good-looking celebrity & Expertise

Model Summaryb
Gender Model
Male

Female 1

Adjusted R
Square

R Square

Std. Error of
the Estimate

.065

.052

.609

.023

.013

.924

.254
.152

a. Predictors: (Constant), Q20Daniel_Good-Looking

b. Dependent Variable: Q23Daniel_Expertise

ANOVAb
Sum of
Squares

Gender Model
Male

Female 1

Regression

Mean
Square

df

1.877

1.877

Residual

27.110

73

.371

Total

28.987

74

1.926

1.926

Residual

81.188

95

.855

Total

83.113

96

Regression

Sig.

5.054

.028a

2.254

.137a

a. Predictors: (Constant), Q20 Daniel_Good-Looking

b. Dependent Variable: Q23 Daniel_Expertise

Coefficientsa
Unstandardized
Coefficients
Gender Model
Male

B
(Constant)
Q20Daniel_
Good-Looking

Female 1

(Constant)
Q20Daniel_
Good-Looking

Standardized
Coefficients

Std. Error

2.670

.309

.160

.071

2.406

.663

.219

.146

Beta

t
.254

.152

Sig.

8.637

.000

2.248

.028

3.628

.000

1.501

.137

a. Dependent Variable: Q23 Daniel_Expertise

160

Plain-looking celebrity & Expertise

Model Summaryb
Gender Model

R Square

Adjusted R
Square

Std. Error of
the Estimate

Male

.060a

.004

-.010

.807

Female

.174a

.030

.020

.913

a. Predictors: (Constant), Q20Wyman_ Good-Looking

b. Dependent Variable: Q23Wyman_ Expertise

ANOVAb
Gender

Sum of
Squares

df

Mean
Square

Sig.

Regression

.174

.174

.268

.607a

Residual

47.506

73

.651

Total

47.680

74

Regression

2.471

2.471

2.965

.088a

Residual

79.158

95

.833

Total

81.629

96

Model

Male

Female

a. Predictors: (Constant), Q20Wyman_ Good-Looking

b. Dependent Variable: Q23Wyman_ Expertise

Coefficientsa
Unstandardized
Coefficients
Gender
Male

Female

Model
1

Std. Error

(Constant)

4.385

.295

Q20Wyman_
Good-Looking

-.057

.109

(Constant)

3.676

.243

Q20Wyman_
Good-Looking

.171

.099

Standardized
Coefficients
Beta
-.060

.174

Sig.

14.869

.000

-.517

.607

15.157

.000

1.722

.088

a. Dependent Variable: Q23Wyman_ Expertise

161

Appendix XIII: Reliability test of the impacts of celebrity endorsement

Case Processing Summary

N
Cases

Valid
a

Excluded
Total

Reliability Statistics

172

100.0

.0

172

100.0

Cronbach's
Alpha

N of
Items

.914

17

a. Listwise deletion based on all

variables in the procedure.

Item-Total Statistics

Q25 Product awareness

Q26 Brand Acceptance
Q27 Brand recall before
shopping
Q28 Remember the brand
during shopping
Q29 Brand alternative

Scale Mean
if Item

Scale
Variance if

Corrected
Item-Total

Cronbach's
Alpha if

Deleted

Item Deleted

Correlation

Item Deleted

53.39
53.58
53.45

86.333
84.373
85.255

.522
.638
.594

.911
.908
.910

53.34

85.069

.613

.909

53.75

85.230

.555

.911

evaluation process
Q30 Buying decision

53.93

84.826

.582

.910

Q31Confused_Believe
Q32More_Favorable
Q33Grab_Attention
Q34Interested

54.15
53.73
53.40
53.70

85.860
83.905
83.854
83.800

.510
.646
.694
.687

.912
.908
.907
.907

Q35Product_Value
Q36Brand_Value
Q37Image_Value
Q38Increase_worthness

53.50
53.35
53.30
53.33

84.953
86.170
86.294
86.153

.602
.565
.577
.487

.909
.910
.910
.913

Q39Intend_To_Buy
Q40Know_More
Q41Pay_Higher_Price

54.06
53.80
54.31

84.774
82.861
83.842

.592
.715
.537

.910
.906
.912

162

Appendix IX: Independent t- test

4.7.2 Consumer buying behavior processes

Group Statistics
Gender
Q25 Product awareness Male
Female
Q26 Brand Acceptance Male
Female
Q27 Brand recall before Male
shopping
Female
Q28Remember the brand Male
during shopping
Female
Q29 Brand alternative
Male
evaluation process
Female
Q30 Buying decision
Male
Female

N
75
97
75
97
75
97
75
97
75
97
75
97

Mean
3.61
3.62
3.57
3.31
3.52
3.58
3.67
3.66
3.39
3.15
3.15
3.02

Std.
Deviation
.868
.847
.791
.917
.844
.864
.920
.789
.914
.894
.926
.890

Std. Error
Mean
.100
.086
.091
.093
.097
.088
.106
.080
.106
.091
.107
.090

163

Independent Samples Test

Levene's Test
for Equality
of Variances

Q25Product Awareness
Q26 Brand Acceptance
Q27 Brand recall before shopping
Q28Remember the brand during
shopping
Q29 Brand alternative evaluation
process
Q30 Help buying decision

Equal variances assumed

Equal variances not assumed
Equal variances assumed
Equal variances not assumed
Equal variances assumed
Equal variances not assumed
Equal variances assumed
Equal variances not assumed
Equal variances assumed
Equal variances not assumed
Equal variances assumed
Equal variances not assumed

F
.177

Sig.
.675

3.014

.084

.002

.968

1.415

.236

.513

.475

.816

.368

t-test for Equality of Means

t
-.040
-.040
1.986
2.024
-.436
-.437
.053
.052
1.672
1.667
.905
.901

df
170
157.320
170
167.906
170
161.015
170
145.744
170
157.525
170
156.080

Sig.
Mean
(2-tailed Differe
)
nce
.968 -.005
.969 -.005
.049
.264
.045
.264
.664 -.057
.663 -.057
.958
.007
.959
.007
.096
.232
.097
.232
.367
.126
.369
.126

95% Confidence
Std.
Interval of the
Error
Difference
Differe
nce
Lower Upper
.132
-.265
.255
.132
-.266
.256
.133
.002
.526
.130
.007
.522
.132
-.317
.202
.131
-.316
.202
.130
-.251
.264
.133
-.256
.270
.139
-.042
.506
.139
-.043
.507
.139
-.149
.401
.140
-.150
.402

164

4.7.3Effectiveness of celebrity endorsements on fashion apparel products

Group Statistics
Gender
Q31Feel confused on products, will
Male
believe celebrity endorsement
Female
Q32 The brand become more favorable Male
Female
Q33Grab Attention
Male
Female
Q34Arouse interested
Male
Female

N
75
97
75
97
75
97
75
97

Mean
2.99
2.75
3.45
3.13
3.80
3.46
3.43
3.21

Std.
Std. Error
Deviation
Mean
.862
.100
.947
.096
.827
.095
.931
.095
.753
.087
.890
.090
.756
.087
.924
.094

Independent Samples Test

Levene's Test
for Equality of
Variances

Q31Feel confused on
products, will believe
celebrity endorsement
Q32 The brand become
more favorable
Q33Grab Attention
Q34Arouse interested

Equal variances assumed

F
.794

Sig.
.374

Equal variances not assumed

Equal variances assumed
Equal variances not assumed
Equal variances assumed
Equal variances not assumed
Equal variances assumed
Equal variances not assumed

.007

.932

7.902

.006

2.425

.121

t-test for Equality of Means

t
1.671

95% Confidence
Interval of the
Sig.
Mean
Difference
(2-taile Differen Std. Error
df
d)
ce
Difference Lower Upper
170
.096
.234
.140
-.042
.511

1.692 165.465

.093

.234

.138

-.039

.507

2.340
170
2.376 166.722
2.622
170
2.679 168.585
1.677
170

.020
.019
.010
.008
.095

.319
.319
.336
.336
.220

.136
.134
.128
.125
.131

.050
.054
.083
.088
-.039

.589
.585
.589
.584
.480

1.720 169.408

.087

.220

.128

-.032

.473
165

4.74 Perceived value of the endorsed fashion products and brands

Group Statistics
Gender
Male
Female
Q36 Increase brand value
Male
Female
Q37 Increase Image and value Male
Female
Q38 increase the worth of less Male
valued fashion products
Female

Q35Increase product value

Std.
Deviation
.774
.936
.759
.854
.731
.831
.993
.870

Mean
3.57
3.45
3.67
3.65
3.71
3.70
3.65
3.69

75
97
75
97
75
97
75
97

Std. Error
Mean
.089
.095
.088
.087
.084
.084
.115
.088

Independent Samples Test

Levene's Test for
Equality of
Variances

F
Q35Increase product value
Q36 Increase brand value
Q37 Increase Image and
value
Q38 increase the worth of
less valued fashion products

Equal variances assumed

Equal variances not assumed
Equal variances assumed
Equal variances not assumed
Equal variances assumed
Equal variances not assumed
Equal variances assumed
Equal variances not assumed

Sig.

2.937

.088

.312

.577

1.460

.229

.462

.497

t-test for Equality of Means

95%
Confidence
Interval of the
Difference
Std. Error
Difference Lower Upper

Sig.
(2-tailed)

Mean
Difference

.896
170
.918 169.190
.137
170
.139 166.676
.046
170
.047 167.145
-.263
170

.372
.360
.891
.889
.963
.962
.793

.120
.120
.017
.017
.006
.006
-.037

.134
.130
.125
.123
.121
.119
.142

-.144
-.138
-.230
-.226
-.234
-.230
-.318

.384
.377
.264
.261
.245
.241
.244

-.258 147.806

.797

-.037

.145

-.323

.249

df

166

4.7.5 Influence of celebrity endorsements on fashion apparel products

Group Statistics

Q39Intend to buy
Q40 Know more
about the product
Q41 Willing to pay
higher price

Gender
Male
Female
Male
Female
Male
Female

N
75
97
75
97
75
97

Mean
3.05
2.86
3.27
3.15
2.67
2.71

Std.
Deviation
.820
.946
.859
.928
1.031
1.080

Std. Error
Mean
.095
.096
.099
.094
.119
.110

Independent Samples Test

Levene's Test
for Equality of
Variances

t-test for Equality of Means

95%
Confidence
Interval of the
Difference

Q39 Intend to buy

Q40 Know more about
the product
Q41 Willing to pay
higher price

Equal variances assumed

Equal variances not assumed
Equal variances assumed
Equal variances not assumed
Equal variances assumed

F
Sig.
t
df
2.493 .116 1.438
170
1.465 167.744
.409 .524 .811
170
.819 164.504
.701 .404 -.274
170

Equal variances not assumed

-.276 162.589

Sig.
Mean
Std. Error
(2-tailed) Difference Difference Lower Upper
.152
.198
.137 -.074
.469
.145
.198
.135 -.069
.464
.419
.112
.138 -.161
.385
.414
.112
.137 -.158
.382
.784
-.045
.163 -.366
.277
.783

-.045

.162

-.364

.275
167

REFERENCES
Alsmadi, S. (2006). The Power of Celebrity Endorsement in Brand Choice Behavior:
An Empirical Study of Consumer Attitudes. Journal of Accounting, Business &
Management, 13, 69-84.
Aronson, E., Turner, J. A., & Carlsmith, J. M. (1963). Communicator credibility and
communications discrepancy as determinants of attitude change. Journal of Abnormal
and Social Psychology, 67(July), 3136.
Aronson, E. (1972). The Social Animal. San Francisco, CA: W.H. Freeman and
Company.
Baker, M. & Churchhill, G.A. (1977) The impact of physically attractive models on
advertising evaluations. Journal of Marketing Research, 14(4), pp. 538555.
Bartlett, M. S. (1954). A note on the multiplying factors for various chi square
approximations. Journal of the Royal Statistical Society, 16 (Series B), 296298.
Bearden, W.O. and Etzel, M.J. (1982), Reference group influence on product and
brand purchase decisions, Journal of Consumer Research, Vol. 9, pp. 183-94.
Burroughs, W. Jeffrey and Richard A. Feinberg (1987), "Using Response Latency to
Assess Spokesperson Effectiveness," Journal of Consumer-Research, 14, 295-299.
Burton, S, Lichtenstein, D.R., Netemeyer, R.G., and Garretson, J.A., (1998). A scale
for measuring attitude toward private label products and an examination of its
psychological and behavioral correlates. Journal of the Academy of Marketing
Science, 26 (4), 293-306.
Caballero, M.J., Lumpkin, J. & Madden, C.D. (1989) Using physical attractiveness as
an advertising tool: an empirical test of attraction phenomenon. Journal of Advertising
Research, 29(4), pp. 1623.
Chan, Kara. (2008) Use of Celebrity in Television Commercials of Youth Products.
Working paper. Hong Kong Baptist University
Choi, S. M., Lee, W. N., Kim, H. J., (2005), Lessons from the rich and famous: A
168

cross-cultural comparison of celebrity endorsement in advertising. Journal of

Advertising, 34 (2), 85-98.
Cohen, J. (1988). Statistical power analysis for the behavioral sciences. Hillsdale, NJ:
Erlbaum.
David H. Silvera, Benedikte Austad, (2004) "Factors predicting the effectiveness of
celebrity endorsement advertisements", European Journal of Marketing, Vol. 38 Iss:
11/12, pp.1509 1526
Dean, D.H. and Biswas, A. (2001), Third-party organization endorsement of
products: an advertising cue affecting consumer prepurchase evaluation of goods and
services, Journal of Advertising, Vol. 30 No. 4, pp. 41-57.
Discover Magazine (2004). Gene Expression: Innate sense of beauty. Retrieved
February 18, 2012 from http://www.gnxp.com/MT2/archives/002687.html
Dean, D.H. and Biswas, A. (2001), Third-party organization endorsement of
products: an advertising cue affecting consumer prepurchase evaluation of goods and
services, Journal of Advertising, Vol. 30 No. 4, pp. 41-57.
Elliott, R. (2006). Strategic Brand Management. Oxford: Oxford University Press.
Erdogan, B. Z., Baker, M. J., Tagg, S (2001), Selecting celebrity endorsers: the
practitioners perspective, Journal of Advertising Research, Vol. 41 No. 3, pp. 39-48
Falk, Pasi & Campbell, Colin (1997). The shopping experience. London; Thousand
Oaks, Calif.: Sage Publications, pp. 166-175.
Fam, K. S., Waller, D. S., 2006. Identifying likeable attributes: A qualitative study of
television advertisements. Qualitative Market Research: An International Journal, 9
(1), 38- 50.
Farrell, K.A., Karels, G.V., Monfort, K.W. & McClatchey, C.A. (2000) Celebrity
performance and endorsement value: the case of Tiger Woods. Managerial Finance,
26(7), pp. 115.
Forbes, G., Jung, J. and Haas, K. (2006.), Benevolent sexism and cosmetic use: a
169

replication with three college and one adult sample, The Journal of Social
Psychology,146(5), 635-640
Forbes (2011). Hollywood's Most Trustworthy Celebrities - Forbes.com. Retrieved
February 17, 2012 from
http://www.forbes.com/2011/02/07/most-trustworthy-celebrities-business-entertainme
nt.html
Giffin, K. (1967),The Contribution of Studies of Source Credibility to a Theory of
Interpersonal Trust in the Communication Process." Psychological Bulletin 68,
104-19.
Goldsmith, R.E., Freiden, J.B. and Kilsheimer, J.C. (1993). Social Values and Female
Fashion Leadership: A Cross-Cultural Study. Psychology and Marketing, 10(5),
399-413.
Goldstein, R. E. (1993) : Esthetic Dentistry-a health service ? Journal of Dental
Research. 72 : 641-642.
Harrison-Walker, L. J.(2001), The Measurement of Word-of-Mouth Communication
and an Investigation of Service Quality and Customer Commitment as Potential
Antecedents , Journal of Service Research, Vol.4, No.1, pp.60-75
Harrison-Walker, L. J.(2001), The Measurement of Word-of-Mouth Communication
and an Investigation of Service Quality and Customer Commitment as Potential
Antecedents , Journal of Service Research, Vol.4, No.1, pp.60-75
Hovland, C. 1.; I. L. Janis; and H. H. Kelley. Communication and Persuasion. New
Haven, CT: Yale University Press, 1953.
Hovland, I., & Weiss, W. (1951). The Influence of Source Credibility on
Communication Effectiveness. Public Opinion Quarterly 15(4), 635-650.
HK2A.Com, 2006. Adspend projections for 2006. Retrieved 7 February, 2012, from
http://www.hk2a.com/pr_20060221.pdf
Ian Phau, Linda Lum, (2000) "Effects of physical attractiveness in the evaluation of
print advertisements", Asia Pacific Journal of Marketing and Logistics, Vol. 12 Iss: 4,
170

pp.41 59
Joseph, W. B (1982). "The Credibility of Physically Attractive Communicators: A
Review." Journal of Advertising 11, 3 (1982): 15-24.
Kahle, L. and Homer, P. (1985), Physical Attractiveness of the Celebrity
Endorser: A Social Adaptation Perspective, Journal of Consumer Research, Vol. 11,
pp.954-961.
Kaiser, H. (1970). A second generation Little Jiffy. Psychometrika, 35, 401415.
Kaiser, H. (1974). An index of factorial simplicity. Psychometrika, 39, 3136.
Kambitsis, C., Harahousou, Y., Theodorakis, N., & Chatzibeis, G. (2002). Sports
advertising in print media: The case of 2000 Olympic games. Corporate
Communications, 7(3), 155-161.
Karina, P. R. (2008). Apparel Brand endorsers and their effects on purchase intentions:
A study of Philippine consumers. Philippine Management Review, 15, 83-99
Kwek C.L (2010). The Effects of Shopping Orientations, Online Trust and Prior
Online Purchase Experience toward Customers Online Purchase Intention.
International Business Research Vol. 3, No. 3
Laroche, M., Kim, C. and Zhou, L. (1996) 'Brand familiarity and confidence as
determinants of purchase intention: an empirical test in a multiple brand context,
Journal of Business Research, 37(2), 115-120.
Ling. P, Gee.C, Elaine.W, (2008). Impact of celebrities on brand decisions of Hong
Kong university students. Retrieved February 11, 2012 from Edith Cowan University,
Advertising/Marketing
Communication
Web
site:
http://www.anzmac-2008.org/_Proceedings/PDF/S01/Ling%20&%20Wong_S6%20S
1%20P2.pdf
Louie, T.A. and Obermiller, C. (2002), Consumer response to a firms endorser
(dis)association decisions, Journal of Advertising, Vol. 30 No. 4, pp. 41-52.
Lutz, R.J., MacKenzie, S.B., and Belch, G.E. (1983), "Attitude Toward the Ad as a
171

Mediator of Advertising Effectiveness:

Determinants and Consequences," 532-9 in

Bagozzi, R.P.and Tybout, A.M. (eds.), Advances in Consumer Research, Vol. X,

(Ann Arbor, MI: Association for Consumer Research)
Maddux, J. E., and R. W. Rogers. "Effects of Source Expertness, Physical
Attractiveness and Supporting Arguments on Persuasion: Persuasion: A Case of
Brains over Beauty." Journal of Personality and Social Psychology 39, 2 (1980):
235-44.
Malhotra, N. K. (2006).Marketing Research: An Applied Orientation and SPSS 14.0
Student CD, 5/E.New Jersey: Prentice Hall.
Marketing-interactive (2012). Marketing TV: Star Power In Social Media. Retrieved
March 15, 2012 from http://www.marketing-interactive.com/news/30953
Martin, J., Jamal, A., & Foxall, G. (2009). Consumer behavior: Chichester, England:
Hoboken. NJ : Wiley
Matthew Tingchi Liu, Yu-Ying Huang, Jiang Minghua, (2007) "Relations among
attractiveness of endorsers, match-up, and purchase intention in sport marketing in
China", Journal of Consumer Marketing, Vol. 24 Iss: 6, pp.358 - 365
McGinnies, E. & Ward, C. (1980) Better liked than right: trustworthiness and
expertise as factors in credibility. Personality and Social Psychology Bulletin, 6(3), pp.
467472.
Miciak, A.R. and Shanklin, W.L. (1994). Choosing celebrity endorsers. Marketing
Management, 3 (3). 50-59.
Michael R. Solomon, Gary Bamossy, Soren Askegaard, Margaret K. Hogg (2006)
Consumer Behaviour: A European Perspective. 3rdEdn. England: Pearson Education
Ltd.
Miller, Gerald P and John Basehart (1969), "Source Trustworthiness, Opinionated
Statements, and Response to Persuasive Communication," Speech Monogropfu, 36
(1), (March), 1-7.
Mills, J., and J. Harvey. (1972)"Opinion Change as a Function of When Information
172

About the Communicator Is Received and Whether He Is Attractive or Expert."

Journal of Personality and Social Psychology 21, 1: 52-55.
Millward Brown (2011). Do men and women respond differently to ads?. Retrieved
March
17,
2012
from
http://www.wpp.com/NR/rdonlyres/1DD86CF1-C93A-49C6-94FC-653889D60652/0/
millward_brown_men_women.pdf
Mistry, B., (2006). Star spotting. Marketing (Jun), 33-34.
Money, R. B., Shimp, T. A., Sakano, T., (2006), Celebrity endorsements in Japan and
the United States: Is negative information all that harmful? Journal of Advertising
Research, 46 (1), 113-123
Muijs, D. (2004). Doing quantitative research in education with SPSS. London: Sage.
Ohanian, R. (1990), Construction and validation of a scale to measure celebrity,
Journal of Advertising, Vol. 19, pp. 39-53.
Ohanian, R. (1991), The impact of celebrity spokespersons perceived image on
consumersintention to purchase, Journal of Advertising Research, Vol. 31 No. 1, pp.
46-54.
Pallant, J. (2005).SPSS SURVIVAL MANUAL, A step by step guide to data analysis
using SPSS for Windows (Version 12),Australia: Allen & Unwin Pty Ltd.
Pamela Alreck, & Robert B Settle. (2002). Gender effects on Internet, catalogue and
store shopping. Journal of Database Marketing, 9(2), 150-162.
Harrison-Walker, L. J.(2001), The Measurement of Word-of-Mouth Communication
and an Investigation of Service Quality and Customer Commitment as Potential
Antecedents , Journal of Service Research, Vol.4, No.1, pp.60-75
Park,C.Whan, andV.Parker Lessig (1977). Students and housewives: Differences in
susceptibility to reference group influence, Journal of Consumer Research 4:
10210.
Pavot, W., Diener, E., Colvin, C. R., & Sandvik, E. (1991). Further validation of the
173

Satisfaction with Life scale: Evidence for the cross method convergence of well being
measures. Journal of Personality Assessment, 57, 149161.
Phang. Grace , De Run Ernest Cyril (2007). Celebrity endorser and respondents
gender: Its impact on company, behavioral and attitudinal variables, Journal of
Humanities, Vol. 10, 38-55
Pringle, H. (2004). Celebrity Sells. John Wiley & Sons, Wiltshire: UK.
Roger D. Blackwell, Paul W. Miniard, James F. Engel (2006). Consumer Behavior.
Ohio: South-Western College.
Sheu, J. B. (2010), A hybrid dynamic forecast model for analyzing celebrity
endorsement effects on consumer attitudes, Mathematical and Computer Modelling,
52 (9), 1554-156
Shimp, T.A. (2000), Advertising Promotion: Supplemental Aspects of Integrated
Marketing Communications, 5th ed., Dryden Press, Fort Worth, TX.
Spears, N., & Singh, S.N. (2004). Measuring attitude towards the brand and purchase
intentions. Journal of Current Issues and Research in Advertising, 26(2), 53-556.
SPSS Inc. (2007).SPSS Statistics 17.0 Command Syntax Reference.USA: Microsoft
Corporation.
Stith, M.T. and Goldsmith, R.E. (1989). Race, Sex and Fashion Innovativeness: A
Replication. Psychology and Marketing, 6(4), 249-262.
Subhadip R (2006). An exploratory study in Celebrity Endorsements. J. Creat.
Commun., 1 (2).
Temple Northup (2009), The Self-Body Image 1 The Self-Body Image: An Integrated
Model of Body Image and Beauty Ideals, Journalism & Mass Communication, 2009
Till, B.D. and Busler, M. (1998). Matching products with attractiveness versus
expertise. Journal of Consumer Marketing, 15(6), 576-586.
Till, B.D. & Busler, M. (2000) The match-up hypothesis: physical attractiveness,
174

expertise, and the role of fit on brand attitude, purchase intentions, and brand beliefs.
Journal of Advertising, 29(3), pp. 113.
Tourism Commission (2012). Tourism Commission - Tourism Performance. Retrieved
March
16,
2012
from
http://www.tourism.gov.hk/english/statistics/statistics_perform.html
Tripp, C., Jensen, T.D. and Carlson, L. (1994), The effects of multiple product
endorsements by celebrities on consumers attitudes and intentions, Journal of
Consumer Research, Vol. 20 No. 4, pp. 535-47.

175