Description
Celebrity endorsements have proven very successful around the world where, due to increasing consumerism, an individual is considered a status symbol when they purchase a celebrity-endorsed product.
I NT E R NAT I ONE L L A HANDE L S HÖGS KOL AN
HÖGSKOLAN I JÖNKÖPING
Cel ebri t y Endorsement -
Hidden factors to success
Master`s thesis within Business Administration
Author: Chabo Dimed
Saouma Joulyana
1utor: Gusta·sson Veronica
Co-examinator: Da·idsson Per
Jonkoping: May 2005
J ÖNKÖP I NG I NT E R NAT I ONAL BUS I NE S S S CHOOL
Jonkoping Uni·ersity
Cel ebri t y Endorsement -
Hidden factors to success
Master`s thesis within Business Administration
Author: Chabo Dimed
Saouma Joulyana
1utor: Gusta·sson Veronica
Co-examinator: Da·idsson Per
Jonkoping May 2005
i
Master Thesis in Business administration
Title: Celebrity Endorsement
Authors: Chabo Dimed, Saouma Joulyana
Tutor: Gustavsson Veronica
Co-examinator: Davidson Per
Date: May 27
th
200S
Subject terms: Celebrity endorsement, Source credibility, Source
attractiveness, Negative publicity, Conjoint Analysis
Abstract
1he use oí celebrity endorsement strategy is nowadays more írequently used by
marketers in order to increase their sales and thereby extend their market shares. Many
celebrities are used in ·arious marketing campaigns and in most cases, the use oí
celebrities as endorsers is seen írom mainly positi·e aspects. 1his made the authors
curious whether the negati·e aspects, that also exists when using celebrities as endorsers,
aííects consumers in their purchasing decisions when a celebrity gets associated with
negati·e publicity. Another cause oí interest is which íactors oí a certain celebrity are
most important and crucial in consumers` perceptions, in the case oí negati·e publicity.
Purpose: 1he purpose oí this thesis is to study which íactors consumers íind important
íor a company to consider when a celebrity gets negati·e publicity, to maintain
successíul brand recognition.
Literature review: 1he use oí pre·ious studies within the íield oí celebrity
endorsement clariíies many important aspects when it comes to celebrity endorsement
and this chapter is elaborated írom 4 diííerent perspecti·es, Company, Celebrity, Brand
and Consumer. Based on pre·ious studies, the authors identiíied 6 crucial attributes
when using celebrities as endorsers and this can also be seen as a pre-study that the
research process has been based upon. lurthermore, the 6 attributes are chosen írom
the three íirst mentioned perspecti·es in order to be able to íulíil the purpose. lence,
this thesis is conducted írom a consumer`s point oí ·iew.
Method: A quantitati·e method is used in this thesis since the authors want to base the
results on collected data that is expressed in numbers and also to generate a general
apprehension in this phenomenon. Moreo·er, the combinations containing the 6
attributes are used in the conjoint experiment.
Conclusions: It was pro·en in this study that consumers do get aííected by celebrities
as endorser, when the attributes írom the literature re·iew are in a combination. But, the
consumers` perception oí the attributes diííers in diííerent cases. lowe·er, the main
íinding was that there are two crucial attributes, trustworthiness and expertise that
companies should take into account when using celebrities in their ad·ertising
campaign.
ii
Table of Contents
1 Introduction............................................................................ 1
1.1 The emergence of celebrity endorsement ..................................... 1
1.2 The two-sided effect of celebrity endorsement .............................. 2
1.3 Limitations ..................................................................................... 4
1.4 Purpose......................................................................................... 4
1.5 Research questions....................................................................... 4
1.6 Outline of this study....................................................................... 4
2 Literature review.................................................................... 6
2.1 Introduction ................................................................................... 6
2.2 Celebrity endorsement strategy .................................................... 7
2.3 Company....................................................................................... 7
2.3.1 Choose and use celebrities as endorsers........................... 8
2.4 Celebrity........................................................................................ 9
2.4.1 Source credibility ................................................................ 9
2.4.2 Source attractiveness ....................................................... 10
2.4.3 Source Power ................................................................... 11
2.5 Brand........................................................................................... 11
2.5.1 Brand equity ..................................................................... 11
2.5.2 Meaning transfer............................................................... 12
2.5.3 Multiple brand endorsement ............................................. 13
2.6 Consumer.................................................................................... 14
2.6.1 Consumer behaviour and negative publicity ..................... 14
2.7 Chapter summary........................................................................ 15
2.7.1 Hypotheses....................................................................... 16
3 Methodology ........................................................................ 18
3.1 Introduction ................................................................................. 18
3.2 Sampling ..................................................................................... 18
3.3 Conjoint Analysis......................................................................... 20
3.3.1 Attributes and attribute level ............................................. 21
3.3.2 Different approaches of conjoint analysis ......................... 22
3.3.3 Experimental design ......................................................... 22
3.3.4 Pilot study ......................................................................... 23
3.3.5 Data collection .................................................................. 23
3.3.6 The survey........................................................................ 24
3.4 Statistics...................................................................................... 24
4 Analysis and results............................................................ 26
4.1 Introduction ................................................................................. 26
4.2 Research question 1 ................................................................... 26
4.2.1 Testing the hypothesis...................................................... 27
4.2.2 Results research question 1 ............................................. 27
4.3 Research question 2 ................................................................... 28
4.3.1 Case 1 .............................................................................. 28
4.3.2 Case 2 .............................................................................. 28
4.3.3 Testing the hypothesis...................................................... 29
4.3.4 Results research question 2 ............................................. 29
iii
4.4 Research question 3 ................................................................... 29
4.4.1 Perceived importance of attributes ................................... 30
4.4.2 Results research question 3 ............................................. 30
4.5 Chapter summary........................................................................ 31
5 Conclusions and discussion.............................................. 33
5.1 Introduction ................................................................................. 33
5.2 Research questions..................................................................... 33
5.3 Discussion................................................................................... 34
5.4 Final words from the authors....................................................... 35
5.5 Prospects for future researches .................................................. 35
6 Critique of this research process....................................... 37
References................................................................................. 38
Figures
Figure 1-1: The Optimal use of celebrity endorsement (Chabo & Saouma,
2005) ................................................................................................. 3
Figure 1-2: The effect of negative information (Chabo & Saouma, 2005) ...... 3
Figure 2-1: The structure of the literature review............................................ 6
Figure 2-2: A triangular relationship (Pringle, 2004)....................................... 8
Figure 2-3: Meaning transfer in the endorsement process (Schlecht, 2003) 13
Figure 1-1: The network of optimal use of a celebrity (Chabo & Saouma,
2005) ............................................................................................... 35
Tabels
Table 3-1: Numbers of Students at Jönköping University ............................ 20
Table 3-2: Attributes and attribute level of the conjoint experiment.............. 21
Table 4-1: Outline of research process ........................................................ 26
Table 4-2: Ranking the attributes ................................................................. 30
Appendices
Appendix 1- 8 profiles for the six attributes at two levels.............................. 42
Appendix 2 - Research Survey.................................................................... 43
Appendix 3- Extract from pilot study............................................................. 63
Appendix 4 - Appendix concerning reserach question 1 ............................. 64
Appendix 5- Appendix concerning reserach question 2 case 1................... 66
Appendix 6- Appendix concerning reserach question 2 case 2................... 67
Appendix 7- Anova tables ........................................................................... 68
Appendix 8- Perception of attributes ........................................................... 70
Introduction
1
1 Introduction
1be avtbor. ritt ¡re.evt a geverat vvaer.tavaivg of tbe to¡ic, cetebrit, evaor.evevt, iv tbe fir.t cba¡ter.
ívrtbervore, tbe avtbor. ritt at.o ¡re.evt tbe ¡robtev area, rbicb tbi. .tva, ritt be ba.ea ov, fottorea b,
tbe ¡vr¡o.e ava tbe ovttive of tbe .tva,.
1.1 The emergence of celebrity endorsement
1he use oí ad·ertising has changed o·er the past 150 years, írom the classical to the
modern school. In the modern ad·ertising strategies ·arious appeals are included, such as
sexual, chock, emotional, íear, and humour ,Se·ern, Belch & Belch, 1990, Belch and Belch,
2001, Soderlund, 2003,. 1he main purpose oí these appeals is to deli·er the iníormation
that the company seeks to send to gain high brand awareness and brand recognition among
a large audience. lowe·er, when using any oí these appeals there is always a person
included, sometimes someone unknown or in most cases a well known person in the public
eye. According to McCracken ,1989,, a well-known person tends to ha·e a greater eííect
on the consumer buying beha·iour. 1his is especially ií people more easily can identiíy
themsel·es with this particular person ,Lrdogan, Baker, & 1agg, 2001,. McCracken ,1989,
also states that celebrity endorsement ad·ertising has been recognized as a ubiquitous
íeature oí modern marketing.
1he concept itselí, celebrity endorsement, is recognized by marketers because it has an
eííect that iníluences the message ,brand, the company is trying to send írom someone
that the consumers íeel a sense oí similarity with. Consumers tend to e·aluate iníormation
írom a communicator ,celebrity,, which they ha·e similar a goal, interest or liíestyle with
greater than someone they do not ,Lrdogan et al, 2001,. lurthermore, celebrities ha·e an
ability to transíer their image to a speciíic product that is being ad·ertised ,\heeler, 2003,.
Companies ha·e taken this opportunity into account and tries, through using it as an
ad·ertising tool, to gain a high brand exposure, attention, interest, desire and action ,Belch
& Belch, 2001,.
Celebrities are people who enjoy public recognition by a large share oí a certain group oí
people and they ha·e distincti·e characteristics, such as attracti·eness and trustworthiness
,Sil·era & Austad, 2004,. Many companies ha·e realized the importance oí celebrity
endorsement as a marketing communication tool ,Belch & Belch, 2001, Soderlund, 2003,.
In using this ad·ertising tool, a company will in the short run generate a high awareness
among a larger audience as well as an increased market share. Since this ad·ertising tool has
gained such attention írom ·arious companies, the price le·el is extremely high
,Danesh·ary & Schwer, 2000, Kambitsis, larahousou, 1heodorakis, & Chatzibeis, 2002,.
As billions oí dollars are spent per year on celebrity endorsement contracts, which indicates
that celebrities play an enormous and important role íor the ad·ertising industry. A recent
example oí this is the Super Bowl in the United States oí America, where thirty seconds oí
ad·ertising cost approximately se·enteen million SLK. Pepsi used many diííerent
celebrities, such as Cindy Crawíord, P Diddy and Carson Kressley to promote diet Pepsi.
leineken, on the other hand, used Brad Pitt in their endorsement strategy, where he
earned around three to íi·e million USD íor his eííort ,Lundell, 2005,.
1his shows that celebrity endorsement is a ·ery popular ad·ertising tool, due to the prices
companies are ready to pay and the eííect it has on consumers buying beha·iour ,Belch &
Belch, 2001, McCracken, 1989, Soderlund, 2003,. 1hereíore, a disputable issue rises
Introduction
2
upon which attributes and personal characteristics oí a celebrity a company base their
choice. lurthermore, Sil·era & Austad ,2004, state that celebrities are eííecti·e endorsers
because they are ·iewed as highly belie·able, likeable, pursuable, and trustworthy.
lowe·er, celebrity endorsement might ·ary when it comes to the íit between the celebrity
and the ad·ertised product, as well as the le·el oí eííecti·eness on the purchasing
beha·iour ,1ill & Shimp, 1998,. 1his entails that the use oí celebrity endorsement might
ha·e a risk when there is a lack oí íit or when the celebrity gets associated with negati·e
publicity ,Soderlund, 2003,. Negati·e publicity occurs when negati·e iníormation is spread
and associated with the celebrity. lence, negati·e iníormation can be direct, indirect
and,or percei·ed subjecti·ely.
1.2 The two-sided effect of celebrity endorsement
1here are not many researchers who ha·e embarked on the concept oí negati·e
iníormation, which celebrities` causes íor consumers` e·aluations oí endorsed brands ,1ill
& Shimp, 1998,, but still there are many diííerent aspects on how negati·e iníormation can
be ·iewed. lurthermore, Soderlund ,2003, argues that celebrities can be percei·ed as less
trustworthy when they are ad·ertising too many brands ,multiple brand endorsement, since
the consumers get di·erse iníormation. On one hand, a celebrity who is linked with
positi·e iníormation has a greater opportunity to reach out to the targeted audience. On
the other hand, negati·e iníormation brings an association that decreases the
trustworthiness oí a celebrity and the íit with the brand, which is also the authors`
deíinition oí negati·e iníormation.
One example where the íit between the celebrity and the brand has been ·ery successíul is
Michael Jordan and Nike. 1he reason that the authors assume the períect íit with the brand
is based on the íact that Nike captured their endorser to a higher le·el where they
de·eloped a new brand within Nike called Air Jordan. One can assume that Michael Jordan
has not shown any indications oí being re·olutionary in the media. Basically, one can say
that Michael Jordan has been a ·ery good endorser íor this brand, in the sense oí his high
expertise and him being trustworthy, where a large audience íeels a sense oí similarity with.
Now twenty years later, they still use him in their ad·ertising, which strengthens the
assumption oí the good íit between Michael and Nike ,Belch & Belch, 2001,.
la·ing this iníormation in mind the authors decided to create a model to íacilitate the
understanding oí how the use oí a celebrity aííects the company. ligure 1-1 shows, when
the iníormation the celebrity is sending out is optimal, and there are no obstacles between
the celebrity, brand, consumers, and company. As one can see, when looking at the model,
the company, celebrity, brand and consumer are interconnected as a continuous network.
In this optimal model, there are no particular obstacles that harm the celebrity
endorsement process. 1his implies that the íit between the celebrity and brand are
integrated.
Introduction
3
Soderlund ,2003, argues that celebrities are not proíitable to be used in a long-run
perspecti·e, ií they in any matters are not maintaining rele·ance with the consumers.
1his is an issue íor companies,
because consumers are easily
adapting to another celebrity,
sometimes because one cannot
íully predict the liíe-span oí a
celebrity ,1ill & Busler, 1998,.
1his can aííect the brand image
and also the le·el oí loyalty a
consumer has towards a brand.
lurthermore, Soderlund ,2003,
discusses that demographical
changes oí a celebrity cannot be
seen as anything negati·e
towards the brand.
lowe·er, e·eryone cannot be as
períect as Michael Jordan. 1here
are always some celebrities that get negati·e publicity, sometimes what media creates and
sometimes their own personal actions. 1his indicates that companies ha·e to be more
careíul when selecting a celebrity who will endorse one oí their brands. 1he authors ha·e in
íigure 1-2 chosen to demonstrate how negati·e publicity oí a celebrity aííects the company,
brand, and consumer. 1his íigure tells us, that when a celebrity is associated with negati·e
publicity, the celebrity is only aííecting each part separately and that there is no connection
between the company, brand and consumer. lurthermore, there is no ílow in the network
oí connections which also aííects the lack oí íit between the celebrity and brand. 1he
questions that arise are, how are these íour parts ,company, celebrity, consumer and brand,
connected, when a
celebrity is associated
with negati·e
publicity· low can a
company o·ercome
these obstacles in the
long run· \ould the
use oí a celebrity who
gets negati·e publicity
destroy the brand
image and the le·el oí
loyalty consumers íeel towards a brand· Does the íit between a celebrity and company
aííect the consumers when a celebrity gets associated with negati·e iníormation· Does the
meaning, a celebrity is transíerring, aííect the consumer`s perception oí the brand, ií a
celebrity is associated with negati·e iníormation· 1he authors` belie·e that írom the
perspecti·e oí this thesis, the eííect oí negati·e iníormation on a celebrity is the most
interesting aspect to study, as it is not limited to certain exceptional elements but can make
use oí a ·ariety oí elements like source attracti·eness, source credibility, and source power.
Brand Company Consumer
Celebrity
ligure 1-2: 1he eííect oí negati·e iníormation ,Chabo &
Saouma, 2005,.
Company
Brand
Consumer
Celebrity
ligure 1-1: 1he Optimal use oí celebrity
endorsement ,Chabo & Saouma, 2005,.
Introduction
4
1.3 Limitations
According to íigure 1-2, the authors discuss the eííect oí negati·e iníormation and how the
celebrity aííects the company, consumers and brand. lowe·er, the authors belie·e that the
consumers are the end group oí the network because they are the ones that the company is
trying to reach through their brand and the ad·ertising tool, celebrity endorsement.
1hereíore, the authors will base their study on a consumer perspecti·e.
1.4 Purpose
1he purpose oí this thesis is to study which íactors consumers íind important íor a
company to consider when a celebrity gets negati·e publicity, to maintain successíul brand
recognition.
1.5 Research questions
1his thesis is intended to in·estigate whether the attributes that ha·e been conducted
through the literature re·iew ,chapter 2, really aííects consumer purchase decisions.
lurthermore, the authors are ·ery interested to study ií the consumers e·aluate the
attributes diííerently, when diííerent endorsers, who get negati·e publicity, are ad·ertising a
product or ser·ice. Lastly, the authors want to see how the consumers percei·ed the
attributes separately. 1he reason in doing this is to compare the results írom the conjoint
and the sur·ey to show which attributes are percei·ed as the most important ones. 1hus,
this thesis addresses more precisely the íollowing research questions.
1. \hich íactors írom the literature re·iew aííect the consumers` willingness to
purchase a product or ser·ice that is endorsed by a celebrity who is associated with
negati·e iníormation, when the attributes are in a combination·
2. Do the attributes írom the literature re·iew ha·e diííerent impact on consumers`
willingness to purchase a product or ser·ice, when diííerent celebrities are used as
endorsers·
3. \hich attributes do consumers íind to be oí most importance when e·aluating
them separately·
1.6 Outline of this study
Chapter J: 1he introduction chapter consists oí an explanation concerning the importance
oí celebrity endorsement in the ad·ertising industry. 1he authors will íurther on explain the
two-sided eííect oí celebrity endorsement, positi·e and negati·e. Also, the purpose oí this
thesis will be addressed íollowed by the research questions.
Chapter 2: 1he literature re·iew is di·ided into íour diííerent phases, Company, Celebrity,
Brand and Consumer. 1hese parts are all related to celebrity endorsement strategy with its
main purpose to íind important attributes that will be studied íurther on in this thesis.
Lastly, this will lead to a conclusion oí this chapter and the hypotheses.
Chapter 3: 1he methodology chapter explains how the authors will approach this research
process in order to íulíil the purpose. 1his will be done by using a quantitati·e method and
more precisely a conjoint experiment.
Introduction
5
Chapter 4: In the analysis and results chapter, the main íindings will be presented íollowed
by an analysis oí the most crucial aspects within this subject. Lastly, the authors will present
a chapter summary and conclude the most important attributes.
Chapter S: In this chapter, the authors will answer the research questions and draw the
main conclusions íollowed by the íinal words írom the authors. linally, the authors will
present suggestions íor íuture researches.
Chapter 6: 1he íinal chapter will pro·ide the reader with critique oí this study.
Literature re·iew
6
2 Literature review
1bi. cba¡ter .ee/. to fiva tbe vo.t crvciat attribvte. tbat i. vo.tt, a..ociatea ritb cetebrit, evaor.evevt
.trateg,. 1bi. iv tvrv to a¡¡t, tbe fottorivg attribvte. iv tbe vetboaotog, ¡art of tbi. .tva,.
2.1 Introduction
\hen a celebrity gets associated with negati·e iníormation a company has to consider
·arious actions to be able to maintain a good position on the market and a similar le·el oí
brand recognition as well. 1his indicates that a company needs to de·elop a strategy that
pre·ents upcoming obstacles ,1ill & Shimp, 1998,, ií a celebrity that they are using as an
endorser gets associated with negati·e iníormation. \ithin this írame, which is also the
main body oí this thesis, the authors will di·ide it into íour diííerent phases that are directly
related to each other. lence, these are the main pillars. 1he purpose oí this chapter is to
identiíy crucial attributes that will íurther on be used in the research process. 1o íacilitate
íor the reader, the authors ha·e chosen to explain the structure oí this chapter in the íigure
below.
ligure 2-1: 1he structure oí the literature re·iew
As one can see írom the íigure 2-1, this chapter is outlined with the basis oí celebrity
endorsement strategy which is also the main íocus when elaborating later, concerning the
íour diííerent phases. 1hese parts will be intergrated and brought together as a summary oí
the most important issues, which will also enhance the process oí choosing rele·ant
attributes íor the experimental design ,see section 3.3.3,.
Company
Phase 1
Celebrity
Phase 2
Brand
Phase 3
Consumer
Phase 4
Celebrity endorsement strategy
Conclusions - identiíication oí attributes
Literature re·iew
¯
2.2 Celebrity endorsement strategy
Marketers usually use indi·iduals who ha·e achie·ed some íorm oí celebrity status to ser·e
as spokespersons íor their companies. Most oí the celebrities that are hired by a company
to pitch their products or ser·ices are popular people, mo·ie stars, entertainers, athletics, or
pop-stars, although occasionally a politician or some other well-known public íigure may
be used ,Belch & Belch, 2001,. lurthermore, when a company decides upon using an
endorsement strategy as their marketing communication tool, the main íocus lies in
exposing their brand ,Kotler, Armstrong, Saunders, & \ong, 2001,. In an endorsement
strategy a new sort oí product is gi·en a new brand name that is unique íor that product
,Riezebos, 2003,. Besides the unique brand name, they also get pro·ided with the name oí
an endorser. In this case, the endorser is a celebrity and íunctions as endorsement which
means an appro·al or support that can be seen as a guarantee oí recommendation íor
consumers. According to Riezebos ,2003, it is only ad·isable to use endorsers íor brands ií
there is a high le·el oí brand-added ·alue. 1his means that the name oí the endorser should
be clearly ·isible next to the name oí the branded article.
Companies ha·e jointly been using their brands and themsel·es, through the use oí
celebrity endorsers, in hope that celebrities may boost eííecti·eness oí their marketing
attempts in the long-term ,Belch & Belch, 2001,. Basically, a company is trying to send
·arious types oí iníormation to their target audience. 1o be able to de·elop an eííecti·e
ad·ertising and promotional campaign, a company has to select their endorser appropriate
to diííerent channels and media ,1ill & Shimp, 1998,, such as source, message, and recei·er
,Belch & Belch, 2001,. 1hus, the brand can be seen as the message the company is trying
to send to their audience. Moreo·er, the source which is intended to send this message in
an endorsement strategy is in this case, the celebrity. lurthermore, the recei·er in the
communication process is the consumer ,Belch & Belch, 2001,.
1he authors ha·e chosen to present each phase oí this thesis separately. 1he reason in
doing this is to bring the reader an understanding oí the eííect celebrity endorsement
strategy has on each phase.
2.3 Company
According to Lrdogan & Baker ,2004,, there are se·eral reasons why companies choose to
use celebrity endorsement to a larger extent. Managers seek to reíresh the brand image,
awareness and attention getting and also to add new dimensions to the brand image.
According to Pringle ,2004,, the best identiíied celebrity that is seen as the best íor
promoting a brand, is one oí the more important decisions considering how consumers
will percei·e the brand. 1here are less important decisions íor a brand compared to the
choice oí celebrity such as what it is named, places where it is seen and sold and what kind
oí ad·ertising campaign it runs. 1hese considerations are taken into account and this is also
the reason why many companies are ready to in·est a huge amount oí money in choosing
and using a certain celebrity, whose identity íits well with their brand. lowe·er, it is not
only the in·ol·ement between the celebrity and the brand, the consumers are also
integrated. Pringle ,2004, has outlined this relationship and it is shown in íigure 2-2.
Literature re·iew
8
ligure 2-2: A triangular relationship ,Pringle, 2004,
1hese three íactors are important íor a company to take into account in order to a·oid any
obstacles when it comes to the triangular relationship. lence, another important area is
that the company only can regulate these three to a certain extent, íor the process oí
celebrity endorsement to be successíul.
Uníortunately, when looking at pre·ious experiences írom the use oí celebrity
endorsement, there are many examples that celebrities might change their beha·iour, ·iews
and their percei·ed personality rather drastically ,Soderlund, 2003,. Certainly, consumers
might change their perception oí the celebrities since the latter mentioned ha·e changed.
lowe·er, the worst outcome is that a celebrity is certainly no longer the indi·idual that the
company chose to promote their brand in the íirst place. 1his leads to a decrease in the
connection between the celebrity and brand ,Belch & Belch, 2001,.
2.3.1 Choose and use celebrities as endorsers
According to Lrdogan et al. ,2001, managers choose celebrity endorsers depending on the
product type and how that íits with the characteristics oí a celebrity. 1his will, according to
Louie & Obermiller ,2002,, lead to a more eííicient ad·ertising.
Celebrities can be chosen by companies to increase their competiti·e ad·antage íor three
main reasons, launch, reiníorcement and repositioning ,Pringle, 2004,.
• 1he íirst opportunity is when launching the brand íor the íirst time and the use oí
celebrity can be ·ery poweríul in this situation. 1his can be particularly strong when
establishing a new category and the consumers need reassurance and an appropriate
star will pro·ide them this.
• Celebrity endorsement can be ·ery eííecti·e when maintaining and reiníorcing a
brand`s competiti·e position in the market. 1he use oí celebrities in this situation
can be especially eííecti·e ií other brands ha·e entered the marketplace and thereby
changed its dynamics.
• \hen the consumers` needs create greater potential in a diííerent sector in the
market than the one which the brand is currently positioned, repositioning with the
help oí a celebrity will be in order. 1he appropriate celebrity can be used as the
íocus oí the brand communication and this shows consumers that it`s positioning is
changing in order to suit an emerging target audience.
In a research done by Bielli ,2003, it is shown that 18° oí all the tested commercials
íeature íamous celebrities. In o·erall, this study shows that celebrities are used to grab
attention, generate interest and in·ol·ement in the brand. Also, the right` celebrity can
add ·alues by associations. A suggestion made by this research is that the celebrity should
not o·ershadow the brand and instead be beneíicial ií the celebrity is used íor what the
My personality
Celebrity personality Brand personality
Literature re·iew
9
brand is already íamous íor. Instead, the brand should borrow and build on the celebrity. A
conclusion made by Bielli ,2003, is that a celebrity with a generally likeable personality is
more likely to make íor success. Moreo·er, ad·ertisers and marketers need to ask
themsel·es íour questions beíore engaging too deep in a celebrity. low íamous is the
celebrity· low well does the celebrity íit with the brand· \hich íacets oí this celebrity can
best work íor the brand proíile· low much oí this can the brand íinance· Many studies
show that the deeper íit between the celebrity and the brand it is the more likely to be
eííecti·e in the marketplace. 1hereíore, the company should stri·e to create a close
connection as possible between the celebrity and the brand ,Pringle, 2004,.
2.4 Celebrity
A celebrity can be considered as the source oí the message a company seeks to send to
their target audience. According to Belch & Belch ,2001,, the term source, when talking
about the in·ol·ement in communicating a marketing message, can occur either directly or
indirectly. Directly can be the celebrity who íunctions as a spokesperson and who sends
out the iníormation that the company wants to deli·er to their target audience. Indirectly is
when a celebrity does not send the message, but instead draw attention to and,or enhances
the appearance oí the ad ,Belch & Belch, 2001,. lowe·er, as mentioned in phase one, a
company ha·e to careíully select a celebrity that has a good íit with the brand, which is
intended to be exposed ,Pringle, 2004,.
\hen a company decides on the use oí a celebrity in their endorsement strategy, there are
three ·ery important source íactors, source credibility, source attracti·eness, and source
power ,Belch & Belch, 2001, 1ill & Shimp, 1998, Ohanian, 1990,. 1hese will now be
presented.
2.4.1 Source credibility
Credibility is the extent to which the recei·er sees the source as ha·ing rele·ant knowledge,
skills, experience and trust to gi·e unbiased and objecti·e iníormation. Source credibility is
used to imply a communicator`s positi·e characteristics that will aííect the recei·er`s
acceptance oí a message ,Ohanian, 1990,. Basically, one can say that a communicator,
celebrity, can be seen as knowledgeable and a person with expertise. lurthermore, the
source needs to be trustworthy, in the sense oí honesty, ethics, and belie·ability ,Belch &
Belch, 2001,. 1hese two attributes, that a celebrity needs to ha·e to be a successíul
endorser in an ad·ertising campaign, are presented more in-depth below.
• í·¡erti.e: Belch & Belch ,2001, discusses that a spokesperson are oíten chosen due
to their knowledge, experience, and expertise in a particular product or ser·ice.
lurthermore, Ohanian ,1990, states that the percei·ed expertise oí celebrity
endorsers is more important in explaining purchase intentions rather than their
attracti·eness and trustworthiness. She also argued that celebrity endorsers are
more eííecti·e when they are knowledgeable, experienced, and qualiíied to talk
about the product they are endorsing. On the whole, .ovrce e·¡erti.e in persuasi·e
communication, indicates generally that the source`s percei·ed expertise has a
positi·e impact on attitude change.
• 1rv.trortbive..: In comparison to expertise, a celebrity needs to be trustworthy when
endorsing a product or a ser·ice ,Schiííman & Kanuk, 2004,. 1his is logically based
on how honest the celebrity is about what he,she says concerning the brand.
lurthermore, Belch & Belch ,2001, discusses that the target audience must íind the
Literature re·iew
10
source ,celebrity, belie·able. Moreo·er, Ohanian ,1990, states that when a celebrity
is percei·ed more trustworthy, the message will be more eííecti·e and the recie·er
will be more integrated. lence, trustworthiness is the degree oí coníidence in the
communicator`s intention to communicate the assertions he,she considers being
the most ·alid ,Ohanian, 1990,.
Belch & Belch ,2001, argues that when the iníormation írom a credible source iníluences
the belieís, opinions, and attitudes oí the recei·er, the latter mentioned adopts the opinion
oí the credible communicator. 1his is based on the assumption that the iníormation írom
the source is accurate. Ií the celebrity achie·es to integrate the recei·er with the
iníormation that he,she meant to send, the company will in the long run gain a loyal
consumer, in the sense that the consumer is more integrated with the brand and not to a
high scope with the celebrity.
1he authors belie·e that using these attributes, expertise and trustworthiness, will be
accurate in the research process. 1he authors belie·e it will pro·ide the thesis with ·aluable
results. lurthermore, it will also be considered as the underlying íactor in the process oí
íinding suitable attributes which consumers íind important íor a company to consider
when a celebrity gets associated with negati·e iníormation.
2.4.2 Source attractiveness
Source attracti·eness is more related to physical attributes, such as similarity, íamiliarity,
and likeability. 1hese are important in the indi·idual`s initial judgment oí another person
,Ohanian, 1990,. Similarity is a supposed resemblance between the source and the recei·er
oí the message, while íamiliarity reíers to the knowledge oí the source through exposure.
Liability is aííection íor the source as a result oí physical appearance, beha·iour or other
personal traits ,Belch & Belch, 2001,. Source attracti·eness aííects the recei·er in the sense
oí that they are identiíying themsel·es with the celebrity. 1his does also moti·ate the
recei·er to seek some type oí relationship with the source and thus adopts similar belieís,
attitudes, preíerences, or beha·iour. lurthermore, ií a celebrity changes position, the
recei·er might íollow and adapt to these changes ,Belch & Belch, 2001, Ohanian, 1990,. Ií
this happens, the celebrity has íulíilled its íunction. lowe·er, not many celebrities achie·e
this, but as mentioned beíore, the main purpose with source attracti·eness is to make a
consumer íeel a part oí the celebrity as well as the company and brand.
1he three attributes similarity, íamiliarity, and likeability are more elaborated below.
• ´ivitarit,: \hen talking about similarity, Belch & Belch ,2001, mentions that the
consumers are more easily iníluenced by a message coming írom someone with
whom they íeel a sense oí similarity with. lurthermore, Ohanian ,1990, elaborated
on the íact that similarity can be measured ií the communicator and recei·er ha·e
similar needs, goals, interest and liíestyle.
• íavitiarit,: According to Belch & Belch ,2001,, íamiliarity can be considered as the
le·el oí knowledge a celebrity possesses oí a brand. \hen a company considers
choosing a celebrity íor their ad·ertising campaign, they need to analyze the
pre·ious knowledge a celebrity has or how he,she will utilize their knowledge in
the exposure phase. 1he authors belie·e that this will be an attribute that the
respondents will obser·e as too diííicult to e·aluate and thereíore the authors will
exclude it írom the research process. lurthermore, the respondent might íind it to
be to diííusi·e.
Literature re·iew
11
• íi/eabitit,: Marketers ha·e an important íunction when choosing which celebrity to
use in their ad·ertising campaign. Basically, these celebrities need to be admired or
at least well known in the public eye ,Belch & Belch, 2001,. lurthermore,
celebrities ha·e to be popular on the market and ha·e certain characteristics that
are extra·agant. Moreo·er, the company has to íind a balance to make sure that the
celebrity does not o·ershadow the brand itselí ,Ohanian, 1990,. 1his can be a
problem íor the company when a celebrity might be associated with negati·e
iníormation.
Summarizing the eííect oí source attracti·eness, one can say that it is used to create
eííecti·e messages, where the attribute attracti·eness reíers to the endorser`s physical
appearance, personality, likeability, and similarity ,Salomon, 2002, Ohanian, 1990,. 1he
authors ha·e chosen to íurther elaborate on the attributes similarity and likeability. 1his,
because they are key attributes when creating eííecti·e messages, thus they will aííect the
research process and pro·ide it with crucial iníormation that will íacilitate the approach to
íulíil the purpose.
2.4.3 Source Power
\hen mentioning the power a celebrity has in ad·ertising, one mainly reíers to how well
they can persuade the consumer to a purchase. 1his is ·ery beneíicial in personal selling,
where personal communication can be an eííicient way to con·ince or lead a consumer
into a purchase ,Salomon, 2002, Ohanian, 1990,. lowe·er, the power as source
characteristics is ·ery diííicult to apply in a non-personal iníluence situation such as
ad·ertising. 1he reason is that a celebrity in an ad generally cannot apply any sanctions to
the recei·er or determine any compliance that will actually occur ,Belch & Belch, 2001,.
lowe·er, the source power can be beneíicial in an endorsement strategy when using an
indi·idual with an authoritati·e personality as a spokesperson.
1he authors ha·e decided to exclude the whole concept oí source power írom the research
process, since it does not belong within the írame oí this thesis. 1he main reason
underlying this statement is because the purpose oí this thesis does not in·ol·e any
personal selling, more than the ad·ertising perspecti·e oí the use oí celebrity.
2.5 Brand
A brand, according to Keller ,2002, is a name, term, sign, symbol or e·en a combination oí
them in order to identiíy the goods and ser·ices that are being ad·ertised. 1he consumers
will be able to more easily diííerentiate one brand írom their competitors due to the
diííerent attributes associated with the brand. Ad·ertising has a great impact on how
consumers will percei·e a brand. Branding is about creating diííerences and pro·iding
products,ser·ices with the power oí brand equity ,Aaker, 1991,.
Brand is one oí the main pillars oí this thesis and the authors explained earlier its
importance on the celebrity endorsement process. Since branding is an extensi·e area, there
will be a special íocus within branding such as brand equity, the meaning transíer process,
and multiple brand endorsement. 1he concepts oí brand linked together with the
ad·ertising tool celebrity endorsement will be discussed in the diííerent sections below.
2.5.1 Brand equity
L·en though brand equity diííers among companies it should in general be deíined within
the area oí marketing eííects that is distincti·e to a brand. 1he concept oí brand equity
Literature re·iew
12
arises when consumers react to more preíerable products and the way it is marketed when
the brand gets identiíied. One has to be aware, according to \algren, Ruble & Donthu
,1995, that brand equity will be less ·aluable íor the manuíacturers and retailers ií it does
not ha·e any meaning to the consumers. Keller ,2002, continues explaining that brand
knowledge is composed oí awareness ,recall and recognition, whereas brand image is more
about ía·ourability, strength and uniqueness oí a brand association.
Ad·ertisers spend great sums oí money to ha·e celebrities promoting their
products,ser·ices with the expectation that consumers will react positi·ely to the
celebrity`s association with a certain brand. Consumers might say to themsel·es Ií she
uses it, it must be good` and Ií I use it, I will be like her` ,Belch & Belch, 2001,. In the
long term, this way oí thinking may lead to an increase in the sales and thereaíter the brand
equity. Brand equity ,Riezebos, 2003, indicates the intrinsic ·alue in a well-known brand
name. 1he amount oí ·alue that is reíerred to in a brand name depends on consumer`s
perception oí the brand domination and through social esteem that is pro·ided when using
it as well as the consumers trust and identiíication with the brand. 1he most ·aluable assets
in many companies are their brand names such as Coca-Cola and Nike which are also
reíerred to as mega brands ,Riezebos, 2003,.
Brand equity can be iníluenced by ad·ertising in se·eral ways ,\algren et. al, 1995,.
Awareness oí the brand can be created and increase the possibility that the brand is
included in the consumer`s mind. Moreo·er, when these brand associations get stored in
the consumers accessible memory, it can later lead to beha·ioural actions. 1he usage
experience can be iníluenced through the use oí ad·ertising and it can also aííect the
percei·ed quality oí a brand ,\algren et. al, 1995,. A celebrity spokesperson is used in
order to promote a company`s product,brand and íor these in turn to be associated with
the celebrity ,Belch & Belch, 2001,. 1his relates to the image,meaning transíer model
described in the next section.
2.5.2 Meaning transfer
McCracken`s transíer model is based on meanings and he suggests that the eííecti·eness oí
the endorser depends on the meaning the celebrity is bringing into the endorsement
process and the brand ,Schlecht, 2003, McCracken, 1989,. lence, he created the meaning
transíer model in order to explain the celebrity endorsement process. Celebrities are íull oí
diííerent meanings e.g. demography ,age, gender, personality and liíestyle types. 1his
makes it ob·ious that a celebrity represents not only one meaning but rather a ·ariety oí
se·eral ones. 1hese celebrity spokespersons are ·ery useíul in marketing brands since they
pro·ide the consumers with quite a íew characteristics when e·aluating the brands in
question. Celebrities add ·alue to the image transíer process because they are oííering
meanings oí deepness and power írom their personality and liíe styles, in comparison to
non-íamous endorsers ,Schlecht, 2003,.
Literature re·iew
13
1he model illustrates a three-stage process oí meaning transíer. 1his in·ol·es the creation
oí the celebrity image, transíer oí meaning írom the celebrity to the brand and the third
stage is how the brand transíers image onto the consumers ,Schlecht, 2003,.
Meaning Acquisition Lndorsement Consumption
ligure 2-3: Meaning transíer in the endorsement process ,Schlecht, 2003,
\hen the brand`s representati·e íeatures should be determined, the consumers` needs
should be considered. 1hen the ad·ertising company has to decide on what celebrity to
choose and who possesses the most appropriate characteristics in relation to the brand.
Simply, there has to be a congruence or íit between the celebrity and the product,brand
,Schlecht, 2003,.
2.5.3 Multiple brand endorsement
Nowadays, it is not unlikely that celebrity spokespersons can and are endorsing se·eral or a
speciíic brand. 1his situation is called multiple brand endorsement or e·en multiple
celebrity endorsement. Ad·ertising íirms might share certain spokespersons and thus the
celebrity will end up promoting more than one brand. An example oí this is the actress
Catherine Zeta-Jones who promoted both 1-Mobile and Llizabeth Arden. Also, the golí
champion 1iger \oods endorsed as much as three brands, American Lxpress, Rolex and
Nike. 1he marketers ha·e to question ií this kind oí celebrity endorsement does aííect
consumer brand attitudes ,Schlecht, 2003,. lowe·er, consumers are becoming more
knowledgeable within the íield oí marketing and they might think that the celebrity rather
preíers to get paid instead oí transíerring any meaning to the product, which in turn will
aííect the consumer buying beha·iour ,Belch & Belch, 2001,.
Pre·ious experience explained by Redenbach ,2005, shows that an endorsement oí íour
diííerent brands or products does indeed iníluence the celebrity`s credibility,
trustworthiness, expertise and likeability. 1he reason íor this is that a íamous person, who
endorses se·eral products instead oí only íocusing and representing one speciíic brand, will
e·entually attain a lack oí distincti·eness. lowe·er, one has to be aware that the use oí
multiple brand endorsement does not ha·e to imply that it is useless. Researches ha·e
actually showed some potential positi·e eííects like transíer oí positi·e brand images. Also,
the shape oí consumers response gets aííected positi·ely when more than íour products
are being endorsed ,Redenbach, 2005,.
It can also be beneíicial to endorse a product with multiple celebrities. Celebrity
spokespersons represent a diííerent mix oí types, like gender and age. Using multiple
celebrities in collaboration they can more eííecti·ely endorse a speciíic brand. 1he cosmetic
manuíacturer L`Oreal matches its wide range oí product lines depending on the celebrities
and their meanings ,Redenbach, 2005,.
Stage 1
Stage 2
Stage 3
Role 1
Role 2
Role 3
Celebrity Product Consumer
Literature re·iew
14
2.6 Consumer
1he continuously increasing competiti·eness has triggered many ad·ertisers to realize that
they need to detect the consumers` actual needs in order to satisíy them. By identiíying the
consumer buying beha·iour, it is more likely that the marketers will target products and
ser·ices directly towards the consumers` needs. Marketing is about satisíying needs and
thereíore it is crucial íor marketers to understand the rele·ance oí human needs to buyer
beha·iour. Consumers tend to search íor, purchase, use, e·aluate, and dispose
products,ser·ices that they expect will satisíy their needs ,Schiííman & Kanuk, 2004,.
1he uses oí endorsers or spokespersons as credible sources are nowadays being írequently
used by ad·ertisers in order to iníluence consumer`s attitudes and purchase intentions
,Goldsmith, Laííerty & Newell, 2000,. Credibility according to Belch & Belch ,2001,
means in this context the extent which the source is percei·ed as possessing expertise
rele·ant to the communication topic and can be trusted to gi·e an objecti·e opinion related
to the subject. 1rustworthiness reíers to the honesty and belie·ability oí the source whereas
expertise is originated írom the knowledge oí the subject. 1rustworthiness along with
expertise are dimensions that are important to theorize credibility and it has been shown
that they are ·ery iníluential when iníluencing attitudes and persuading consumers.
2.6.1 Consumer behaviour and negative publicity
Publicity tends to be more credible and ha·e more power than general marketing
communications exerted by a company ,Dean, 2004,. Lspecially, negati·e publicity seems
to ha·e the tendency to damage the company`s image. 1his is mainly due to the íact that
high credibility as well as the negati·ity eííect has a tendency to be more reílected upon
than positi·e iníormation in the consumer`s e·aluations. It is more likely that companies
will recei·e bad exposure since the media preíers to present bad news. 1ill & Shimp ,1998,
support this statement and continues, explaining that companies ha·e to be aware oí the
possibility oí attaining negati·e publicity when using celebrities as endorsers, since this may
eííect the consumers` perception oí the brand.
1he consumers` support oí a brand is oí major importance íor the prospect existence oí a
brand and the organization ,Riezebos, 2003,. Ií an incident occurs in relation to the brand,
the consumer`s trust in the brand will íade and the consequence will aííect the consumers
purchasing beha·iour. \hen consumers get questioned by marketers about their
purchasing beha·iour, they gi·e the impression that incidents with brands do not iníluence
their choices. Usually they also claim that neither ad·ertising nor negati·e publicity aííects
them. Riezebos ,2003, also states that the most objecti·e way to determine this eííect on
consumers is to list changes in the market shares.
Moreo·er, ií negati·e publicity leads to brand damage, the media can be seen both as the
source,spreader oí publicity simultaneously as challengers. In the case oí source,spreader
the media íunctions as a gate-keeper whereby it concentrates on those incidents that bring
about newsworthiness. Other incidents with a high newsworthiness are rele·ant with a
particular brand in crisis. 1his is true mainly because the other e·ents can act as deílections
and push away the negati·e publicity oí the brand. An example oí this push away e·ent`
can be a natural disaster. 1he media can thereíore also play an essential role in the
challenge oí negati·e publicity. In this case, one can think oí the incident as possible but on
the other hand the attention during this news should be spent on the rebuilding oí a brand
in order to retain the consumers and maintain the brand recognition ,Riezebos, 2003,.
Louie & Obermiller ,2002, explains the case oí negati·e e·ents` as problems that can take
Literature re·iew
15
place when a íamous person gets in·ol·ed in incidents that change or damage his,her
reputation. 1hese kinds oí circumstances can diííer widely írom misused exposure to an
accident that holds back a celebrity`s ability to períorm. In a study made by 1ill & Shimp
,1998, it was íound that negati·e iníormation about a celebrity can harm how consumers
percei·e the product,brand through the connected link between the brand and celebrity.
\hen a company has a potential relationship with a celebrity, the consumers will not íocus
on the company but rather on the circumstances surrounding the celebrity. Consumer`s
reaction may not be unpleasant when íirms reject high blame potential endorsers since
these could be harmíul to the company`s image and the consumer buying beha·iour.
lowe·er, due to the reason oí consumers being deíensi·e, they are supposed to not react
hea·ily when the potential endorsers ha·e low blame. Also, endorser candidates can be
more complex than the existing endorsers ,Louie & Obermiller, 2002,.
2.7 Chapter summary
1he main purpose oí this section is to introduce the íeatures that are rele·ant when
in·estigating which attributes consumers consider to be most appropriate íor a company to
keep in mind, beíore de·eloping a celebrity endorsement strategy. 1he attributes ha·e been
elaborated írom pre·ious empirical íindings within celebrity endorsement, where the main
íocus ha·e been addressed on celebrities and the eííect they ha·e had on company, brand
and consumer. lurthermore, the authors ha·e narrowed this additionally and put the íocal
point on how negati·e iníormation will aííect the consumers and how this is linked to the
three other phases in this thesis. 1hese attributes will play the most important role when
conducting the conjoint analysis, and they will be presented in section 3.3.1.
By analyzing many earlier studies, the authors came to the conclusion that most oí these
ha·e been structured in a similar way. Basically, the starting point was to elicit the
enormous role a celebrity ha·e had on the modern ad·ertising industry. lurthermore, many
oí these researchers ha·e embarked on how the consumers get aííected by celebrities ,1ill
& Shimp, 1998, Soderlund, 2003, Sil·era & Austad, 2004, Belch & Belch, 2001,
McCracken, 1989,. Only a íew oí them ha·e concentrated on the negati·e eííect a celebrity
can ha·e on consumers. 1his is the reason why, the authors chose to conduct this thesis
írom that nature.
1hese attributes are based on studies írom both qualitati·e and quantitati·e research
methods. 1he íact that similar or exact attributes are mentioned and discussed in the both
research methods, only strengthen the thought that the attributes within this thesis are
comprehensi·e. lowe·er, it can also be seen as a weakness in only using the already
analyzed attributes since there might be other more important attributes that consumers
might e·aluate in this dynamic market. lence, the time span limits the author`s ability to
search íor other inno·ati·e attributes. Despite this, earlier attributes will hopeíully bring
another understanding related to consumer`s perspecti·e on celebrities in ad·ertising.
1he íollowing attributes are elaborated on pre·ious studies. 1hese attributes are related to
the company, celebrity and brand. 1he reason why no attributes are directly related to the
consumer is because the research is conducted írom a consumer perspecti·e, which
indirectly will lead to íindings on what attributes consumers íeel are most important íor
companies to take into account when celebrities get associated with negati·e publicity.
Literature re·iew
16
Company
1he most important aspect that has been brought up is how well the celebrity fit.;vatcbe.)
with the company`s en·ironment and the brand as well. Consumers might percei·e a brand
diííerently depending on what celebrity a company chooses. 1he greater fit between the
celebrity and the brand, it is more likely íor a consumer to accept this ad·ertising tool. A
good congruence between the belieís oí a celebrity and a company will enhance the brand
recognition. 1hereíore, the authors belie·e that the fit ;vatcb) between the celebrity and
company is rather important to in·estigate íurther on when a celebrity gets associated with
negati·e iníormation and ií a change oí the fit will in turn change consumer`s perception
oí a brand. 1his assumption is strengthen by Lrdogen & Baker ,2004,, Pringle ,2004,,
Brown ,2003,, Danesh·ary & Schwer ,2000,, Louie & Obermiller, ,2002,, Kamins ,1990,,
& 1ill & Shimp ,1998, who claim that íit` ,match, can be considered as an attribute.
Celebrity
\hen talking about source credibility, both the attributes e·¡erti.e ava trv.trortbive.. are
included. lurthermore, the eííecti·eness oí a message depends on the celebrities` e·¡erti.e
ava trv.trortbive... 1he belieís, opinions, attitudes and beha·iour can be iníluenced by the
iníormation oí a credible source. 1his occurs when the consumer ,recei·er, accepts a
source iníluence when it comes to their personal attitude. Source attracti·eness includes
.ivitarit, ava ti/eabitit, which are also two attributes that will be used in the conjoint analysis.
1he underlying reason íor this is because the authors íeel they can play an important role
when analyzing the celebrity`s e·¡erti.e, trv.trortbive.., .ivitarit, ava ti/eabitit, and how that
iníluences the consumers` purchase beha·iour when the celebrity gets associated with
negati·e iníormation ,Lrdogan et al, 2001, Ohanian, 1990, 1ill & Shimp, 1998, Sil·era &
Austad, 2004, Kamins, 1990, Danesh·ary & Schwer, 2000, Kamins, Brand, loeke &
Moe,1989,.
Brand
1he most important aspect to consider when talking about brand and celebrity is to what
extent a celebrity can trav.fer av, veavivg to the brand itselí. 1hereíore, the more a celebrity
has the ability to trav.fer veavivg. to the brand, the more a consumer will associate with the
celebrity. lence, when a celebrity gets associated with negati·e iníormation, it might occur
that the celebrity looses the ability to transíer the right meaning to the brand as the
company would like to. 1his indicates also that the consumer will get aííected, sometimes
in a positi·e way and sometimes negati·ely. 1hereíore, the authors assume this is an
important attribute to study, in the sense on how consumers react to the meaning a
celebrity transíers, when negati·e iníormation is associated with a certain celebrity
,Schlecht, 2003, Redenbach, 2005, 1ill & Shimp, 1998, \algren, Ruble & Donthu, 1995,
Gwinner, 199¯,.
2.7.1 Hypotheses
1o summarize this, one can conclude that the presented attributes are considered as the
personal characteristics oí a celebrity. lowe·er, the main purpose oí this thesis is to study
which attributes that a company has to consider when a celebrity gets associated with
negati·e iníormation. 1o be able to íulíil the purpose oí this thesis and also answer the
research questions ,the íirst and second, that are posed in section 1.5, the authors íind it
·ery appropriate to in·estigate ií there is a relationship between the consumer willingness
to purchase a product or ser·ice, with the presented attributes in a conjunction. 1hereíore,
Literature re·iew
1¯
the íirst hypothesis deals with ií the consumers really get aííected by the chosen attributes
írom the literature re·iew. 1his leads us in to the íollowing hypothesis:
H
A
: 1he celebrity-product combination attributes expertise, trustworthiness,
similarity, likeability, meaning transfer, and ´fit¨ (match) affect the consumers'
willingness to purchase a product or service also under condition of negative
information.
As in research question two, the authors are aiming to understand ií the consumers are
aííected diííerently by these attributes when diííerent celebrities are used as endorsers.
1hereíore, the authors are ·ery keen on in·estigating ií consumers` willingness to purchase
a product or ser·ice diííers when using diííerent celebrities. 1his leads us into the second
hypothesis:
H
B
: 1he relative importance of specific celebrity-product combination attributes is
different depending on which specific celebrity is used as endorser.
Methodology
18
3 Methodology
This chapter will describe the choice of methods and how the process of gathering
empirical data will occur to fulfil the purpose of this thesis. The approach that will be
used is a quantitative one, and the survey will be analysed through a conjoint
experiment.
3.1 Introduction
Many researchers ha·e supported the idea oí using a multi-method approach in a
marketing research study. llick ,2002, & Carson, Gilmore, Perry & Gronhaug ,2001,, are
just a íew to mention and they íurther claimed that a combination oí qualitati·e and
quantitati·e approaches can lead to interesting and exciting explorations within a research
study. 1he integration oí these two methods should also be seen as complementary rather
than ri·alry. Moreo·er, llick ,2002, suggested that a thorough qualitati·e pre-study should
be carried out in order íor a quantitati·e study to be successíul. lowe·er, due to the time
span the authors ha·e with the thesis they decided that a multi-method-approach cannot be
undertaken. 1hereíore, the authors ha·e decided upon conducting a quantitati·e research.
In a quantitati·e method according to Lsaiasson, Gilljam, Oscarsson & \ägnerud ,2002,
the study is based on collected data that can be expressed in numbers to be estimated later.
1he reason íor this choice oí method is based on the íact that many companies claim
themsel·es to be consumer-dri·en, but still íail when conducting a marketing research, this
because they are only running íocus groups, sending out questionnaires, and analysing sales
data ,Gustaísson, lerrmann, & luber, 2001,. lowe·er, the main reason íor this íailure is
basically that the company does not understand the consumer needs and has a lack oí
iníormation about the market ,Iggland, 1989,. lrom the perspecti·e oí the purpose oí this
thesis, the authors are aiming to identiíy the most important íactors a company should
consider írom consumers point oí ·iew, when a celebrity gets associated with negati·e
iníormation. 1hereíore, a quantitati·e approach will pro·ide this thesis with better results
that will lead to the process oí drawing general conclusions and to get an o·erall picture
among the larger sample ,Gustaísson et al., 2001,. As stated in the multi-method approach,
conducting íocus groups as a pre-study will ía·our the study oí this thesis. But, at the same
time many researchers ha·e embarked on the concept or the use oí celebrity endorsement,
and thereíore they ha·e come to the conclusion that some íactors are more beneíicial íor a
company to consider when using a celebrity in their endorsement strategies ,McCracken,
1989, Se·ern, Belch & Belch, 1990, Lrdogen et al., 2001, \heeler, 2003, Sil·era & Austad,
2004, 1ill & Shimp, 1998, 1ill & Busler, 1998,. 1he authors oí this thesis ha·e thereíore
chosen to use the recommended attributes when conducting a quantitati·e approach, and
exclude the use oí íocus group or any other pre-study.
1he attributes ha·e been di·ided within the írame oí this thesis, company, celebrity, brand,
and consumer. Basically, one can say that this thesis has íour diííerent phases. lence,
rele·ant attributes ha·e been directly addressed to each phase and research questions ha·e
been de·eloped. 1he authors will aim to íind proper results íor each research question in
an eííicient way in order to identiíy the di·erse consumer needs.
3.2 Sampling
1he most important aspect a researcher has to keep in mind is how to determine an
appropriate research population and a proper sampling procedure. According to Churchill
,1995, the sampling procedure can be di·ided into probability and non-probability
Methodology
19
sampling. lurthermore, Saunders et al. ,2003, explains probability sampling, as the chance
oí each case being selected írom the population which is known. In non-probability
sampling there is an assumption that there is an e·en distribution oí characteristics within
the population. \ithin business research it is oíten not possible to speciíy the probability
that any case will be included in the sample and thus, the sample must be selected some
other way ,Churchill, 1995,.
Probability sampling in·ol·es the selection oí a sample írom a population, based on the
principle oí randomization or chance ,Aczel & Sounderpandian, 2002,. lence, probability
sampling is more complex in the sense that it sometimes in·ol·es two diííerent stages oí
sampling. 1hereíore, it can be considered to be more time consuming and more costly than
a non-probability sampling ,Saunders et. al, 2003,. Non-probability sampling is cheaper and
used when a sampling írame is not a·ailable. Also, this method is used in a research where
there is an interest oí obtaining an idea oí responses on ideas that people ha·e ,Churchill,
1995,. Based on this discussion, the sampling technique that will be used in this thesis is
non-probability sampling. 1he main argument íor this is the limited time and the expenses.
Additionally, the authors ha·e no sampling írame a·ailable and want to keep the research
to a low cost while putting the main interest on getting hold oí responses that people ha·e.
\hen conducting non-probability sampling, the main assumption a researcher makes is
that there is an e·en distribution oí characteristics within the population. In doing this, the
sample would be representati·e which will also lead to that the results will be truthíul
,Aczel & Sounderpandian, 2002,. lurthermore, non-probability sampling pro·ides a range
oí alternati·e techniques based on the researchers` subjecti·e judgement and examples oí
these are: quota sampling, purposi·e sampling, snowball sampling and con·enience
sampling ,Saunders et al, 2003,.
1o be able to íulíil the purpose oí this thesis, the main íocus lies on the consumers`
assumptions, and also the most important íactors that are related to celebrities that get
associated with negati·e iníormation. 1hereíore, the authors belie·e quota sampling is
most appropriate. 1his sampling technique means that a researcher has the ability to get
iníormation írom a respondent in the easiest way ,Lsaiasson et al., 2002,, which in this case
is on the campus area in Uni·ersity oí Jonkoping. Quota sampling was chosen because
there will be a sampling íocus on students at Uni·ersity oí Jonkoping. 1he underlying
íactor íor this statement is according to 1ang, Kim & 1ang ,2002,, 1ill & Shimp ,1998,,
that students are more conscious about celebrities and the authors also belie·e that they
will gain more ·aluable iníormation írom students. lurthermore, to be able to get a wider
understanding and strengthen the assumption that the students that are selected are similar
to those not selected, the authors will not íocus on one íaculty and thereíore take the íour
íaculties into consideration. Basically, Saunders et al. ,2003, argue that the sampling is
done, when a speciíic number oí units ,quotas, íor ·arious sub-populations ha·e been
selected. 1his means that the main population in this case are the students at Uni·ersity oí
Jonkoping, and the sub-population is the students at each íaculty.
Shepherd and Zacharakis ,199¯, state that the sample size used in a conjoint analysis is
smaller then 250 respondents, which is the normal amount in a required standard sur·ey.
lurthermore, they state that 50 respondents is a ·ery suííicient amount oí respondent,
when conducting a conjoint analysis. As mentioned beíore, the total population is the
students at Jonkoping Uni·ersity. lurthermore, Saunders et al. ,2003, state that the quotas
may be based on population proportions. 1o íacilitate the understanding íor the reader, the
authors chose to present the sample population and how the quotas are selected in table 3-
1.
Methodology
20
1able 3-1: Numbers oí Students at Jonkoping Uni·ersity
School Number oí students Percentage oí
students
Quota oí students in
sample oí ¯5
JIBS
1
1 ¯50 20 15
ING
2
1 800 20 15
lLK
3
3 000 36 2¯
llJ
4
2 116 24 18
1otal 8 666 100 ¯5
3.3 Conjoint Analysis
Conjoint analysis is a measurement technique that is concerned with an understanding oí
how people make choices between products or ser·ices, or a combination oí them that
better meet the consumers underlying needs ,Gustaísson, lerrmann, & luber, 2001,.
Rather then asking respondents íor their selí-perceptions oí what attributes determine their
choices, conjoint analysis in·estigations ask consumers to rank or rate complete` or íull
proíile` descriptions objects. 1he importance oí each speciíic attributes is tested out in the
statistical analysis. 1hereíore, conjoint analysis has been recognized as an extremely
poweríul way oí capturing what really dri·es consumers to purchase a product o·er
another and what consumers really ·alue ,Green & Srini·asan, 1990, Green, 2001,.
Moreo·er, conjoint analysis can according to Malthotra ,1996, and Green & Srini·asan
,1990, be used when de·eloping a model that enables companies to understand consumers`
needs. 1his is also important knowledge to be able to impro·e their market shares and as
well as understand how competitors` beha·iour will aííect their consumers.
Conjoint analysis has been used within many diííerent research areas, such as consumer
beha·ior ,Bruns, 2004,, and in diííerent marketing areas ,segmentation, positioning,
preíerence models, ,Green & Srini·asan, 1990, Malhotra, 1996, Gustaísson, lerrmann &
luber, 2001,. As one can tell, conjoint analysis is an eííicient research method when the
aim is to in·estigate and understand how consumers e·aluate diííerent combinations.
\ithin the írame oí the thesis, the authors aim to understand what the underlying íactors
consumers consider to be most important considering the aim oí this thesis.
1he main reason to the decision to conduct a conjoint analysis is de·eloped írom the aim
to íully understand how consumers are dri·en in e·aluating a celebrity that gets associated
with negati·e iníormation. Also, a standard sur·ey will not pro·ide the authors with
enough data to be able to draw conclusions and present a model íor companies that seek to
use a celebrity in their endorsement strategy. lence, the respondents in a standard sur·ey
1
Jonkoping International Business School
2
Lngineering School
3
School oí Lducation and Communication
4
School oí lealth and Science
Methodology
21
will only e·aluate each attribute ,íactor, separately, and thereíore not pro·ide a general
impression oí the use oí celebrity in an endorsement strategy. But, in a conjoint analysis the
respondents are asked to rate diííerent combinations oí attributes at diííerent le·els and in
doing this the authors belie·e they will gain more ·aluable iníormation that will íacilitate
the creating oí a model that embarks the concept, celebrity endorsement.
1here are no speciíic and reliable approaches to e·aluate the ·alidity and reliability in a
conjoint analysis. lence, according to Antilla, leu·el & Moller ,1980, the choice oí
attributes and their le·els are ·ery determined when it comes to ·alidity and reliability. 1o
be able to reach a high ·alidity, Antilla et al ,1980,, argues that a respondent needs to
e·aluate realistic attributes, such as expertise and trustworthiness.
3.3.1 Attributes and attribute level
\hen choosing the attributes and the le·els oí attributes, it is according to Lkdahl ,199¯,,
·ery important that they are realistic and related to the problem. lurthermore, he states
that the number should be as small as possible to minimize estimation eííorts. Bruns
,2004,, also states that the íirst steps in designing a conjoint study is to de·elop a set oí
attributes and their le·els.
lurthermore, to be able to conduct a conjoint analysis one has to identiíy the most
important attributes that a consumer assumes to be oí most importance in their purchase
beha·iour ,Antilla et al, 1980, Green & Srini·asan, 1990,. lrom the perspecti·e oí this
thesis, the attributes can be chosen írom pre·ious studies, both qualitati·e and quantitati·e
research methods, concerning the use oí celebrity endorsement in marketing.
\hen deciding upon diííerent le·els oí an attribute, one has to consider whether the
attributes are quantitati·e or qualitati·e. 1he attributes that are presented in table 3-2 are
írom a quantitati·e base. 1his indicates that the le·el oí the attributes will be easier to
estimate.
1able 3-2: Attributes and attribute le·el oí the conjoint experiment
1o be able to create attribute le·els both the authors and the respondents ha·e to
understand the clear diííerence oí le·el which an attribute is attached to ,Van Der Pol &
Ryan, 1996,. lrom the literature re·iew, it has been pro·en se·eral times, that the attributes
and their le·els presented in table 3-2 are ·ery applicable when studying celebrity
endorsement ,Lrdogan, Baker & 1agg, 2001, Ohanian, 1990, 1ill & Shimp, 1998, Sil·era &
Attribute Level J Level 2
Lxpertise ligh Low
1rustworthiness ligh Low
Similarity ligh Low
Likeability ligh Low
Meaning transíer ligh Low
lit` Match ligh Low
Methodology
22
Austad, 2004, Kamins, 1990, Danesh·ary & Schwer, 2000, Kamins, Brand, loeke & Moe,
1989,. 1he reason íor the choice oí attributes were elaborated and explained in depth in
section 2.¯.
3.3.2 Different approaches of conjoint analysis
\hen one considers conducting a conjoint analysis, there are many diííerent approaches to
reílect upon. lowe·er, the most popular one is a íull proíile approach ,Green &
Srini·asan, 1990, when the amounts oí attributes are not less then six and at the most
eight. 1hey recommended using trade-oíí matrices, when a larger number oí attributes are
used. lurthermore, they also argued that when the number reaches ten or more attributes,
a researcher should use selí-explicated data and methods in·ol·ing a combination oí selí-
explicated and conjoint data. 1he latest recommendation can be compared to an .aa¡tire
Cov;oivt .vat,.i. ,ACD,. 1he ACD has gained more acceptances in recent years, due to that
it can accommodate a larger number oí attributes, where it combines selí-explicated data
with pairs in comparison to intensity ratings ,Johnson, 198¯,.
1he approach, íull proíile, is chosen by the authors to gather rele·ant data íor íurther
analysis. 1he main reason is based on the amount oí attributes that the authors decided to
support their research on ,see table 3-2,. 1he amounts oí attributes are six, which indicates
that a íull proíile is the most suitable approach íor this study. lowe·er, this can be seen as
a negati·e aspect due to that the respondents should not be o·erloaded. 1hereíore, the
authors decided upon attributes that are rele·ant to the purpose oí this thesis and also not
o·erwhelming íor the respondent.
3.3.3 Experimental design
Beíore designing the sur·ey one has to distinguish and make clariíications concerning
dependent and independent ·ariables. According to Antilla et al. ,1980, the dependent
·ariable can be seen as the preíerence which can be explained by a number oí independent
·ariables. 1he dependent ·ariable used in this study is: tbe rittivgve.. to ¡vrcba.e. In other
words, the analysis aims at íinding out what attributes aííect the most respondents in their
purchasing decisions, in the context oí a celebrity being associated with negati·e publicity.
1he independent ·ariables can also be considered as attributes, which in this case are six
diííerent ones at two le·els. lurthermore, the basic assumption íor a conjoint experiment is
that the attributes are known in ad·ance ,Bruns, 2004,. 1his means that the researcher can
select the attributes that are most appropriate íor the study írom literature and pre·ious
empirical íindings. 1he authors support this assumption and ha·e presented the choice oí
attributes more in-depth in chapter two. Aíter clariíying and deciding upon the dependent
and independent ·ariables one can start designing the experiment.
\hen designing the experiment one has to keep in mind how the respondents will
interpret it and as well understand it. In the case oí conjoint analysis, the experiment will be
based on diííerent generated combinations or conjoint proíiles, in this case 64, which also
is a íull íactorial design. 1he sur·eys are usually not períormed as íull íactorial design, but
rather as íractional íactorial designs, which basically are íractions oí the íull design. 1he
reason that researchers use íractional íactorial designs, according to Lkdahl ,199¯,, is that
they usually are utilized in order to add more attributes into the combinations and at the
same time to not increase the strain on the respondents. One can íor sure use a íull
íactorial design, when collecting data, but according to Bruns ,2004,, there will be too
many combinations íor a respondent to rate and still pro·ide ·alid results. 1hereíore, the
Methodology
23
most important reason íor limiting the experiment according to Bruns ,2004, is to keep the
number oí cases manageable while still creating ·alid results.
In this case the authors came to the decision that two le·els and six attributes will pro·ide
this thesis with ·alid results. lurthermore, the assessment oí six attributes with two le·els
íor each ·ariable would lead to 64 ,2
6
, hypothetical combinations which is a íull íactorial
design. 1his according to Green & Srini·asan ,19¯8, would be an o·erwhelming task íor
each respondent to e·aluate and still gi·e consistent and meaningíul answers. 1hereíore,
Gustaísson et al. ,2001, argues that it is important to keep the number oí cases to be
e·aluated to as íew as appropriate and possible, but still ha·e enough cases to generate
signiíicant and ·alid results. 1his is the reason why an orthogonal íractional íactorial design
will be applied ,Green & Srini·asan, 19¯8, Bruns, 2004,. In doing this, the number oí
combinations will be reduced írom 64 to 8 which is according to Lkdahl ,199¯, a suííicient
number oí combinations to conduct a conjoint analysis. But, Lkdahl ,199¯, also argued
that a suitable íractional íactorial design íor six attributes at two le·els could be 2
6-2
~2
4
~16
combinations. lurthermore, he stated that the íractional íactorial designs are generally
more resource eííecti·e, and in the case oí conjoint analysis more íeasible.
Shepherd & Zacharakis ,199¯, ha·e another ·iew oí the amount oí combinations that is
suííicient when conducting a conjoint analysis. 1hey claimed that duplexing the amount oí
combinations will lead to more ·alid results as well as reduced bias. 1his means that the
number oí the total amount oí combinations will be duplicated into 16, when the íractional
íactorial design is 2
3
,8 combinations, and into 32 when the íractional íactorial design is 2
4
,16 combination,, and still be manageable. Con·ersely, Smith, Schullen & Barr ,2002,
argues that it is e·en applicable and manageable to use less than 100 combinations where
the respondents are able to judge satisíactorily. 1hey also state that the amount oí
combinations is related to the nature oí the purpose and what the authors want to achie·e.
By taking both oí these approaches into consideration, the amount oí combinations ,8 or
16,, which will be suííicient to íulíil the purpose oí this thesis, ha·e made the authors
question it. 1hereíore, to not decrease the reliability oí this thesis and the ability to
maintain accurate data to reach a sati íactional conclusion, the authors chose to test both 8
and 16 combination in duplex íorm, beíore conducting the collection oí data which leads
us into the next section.
3.3.4 Pilot study
\hen constructing the experiment, crucial íeedback írom the respondents were gi·en to
the authors, who realized that 32 combinations would be too extensi·e íor the respondents
and thereíore the authors chose to keep it to 16 combinations. Other crucial íeedbacks
were gi·en, both positi·e and negati·e. 1he positi·e ones were that the instructions and
description oí the attributes and le·els were straightíorward. 1he main concern was how to
deal with the combinations and the question related to them. 1he respondents interpreted
the question diííerently, which also aííected their ratings. \hen asking them how they were
thinking, the authors could clearly see a pattern oí diííerent interpretations. 1hereíore, the
authors íound it highly important to re-phrase the question, which later on showed that the
interpretation oí the question did not diííer in a crucial matter. In appendix 3, an extract
írom the pilot study will be presented.
3.3.5 Data collection
1here are usually two types oí scales used in conjoint analysis sur·eys, rating and rank
order ,Lkdahl, 199¯,. lurthermore, Gustaísson ,1996, explains íi·e diííerent response
Methodology
24
scales, where a conjoint analysis can be conducted. 1hese are, rank order, rating, ·erbal
rating, íirst rank order and then distribute the concepts on a scale, and íirst rate and then
rank order among combinations with the same rating. \hen the respondents are asked to
rank the combinations, which is rather easy íor them, they will only pro·ide iníormation
that is in order oí preíerences and not stating the degree oí preíerences ,Green &
Srini·asen, 19¯8,. Ranking is also not suitable íor postal questionnaires, since the
respondents require help to íacilitate their task. Verbal rating ,would buy and would not
buy, is also not useíul íor postal questionnaires as it is most applicable íor low in·ol·ement
products. 1he rating scale ,1-¯ or 1-9,, makes the respondents e·aluate each combination
separately in regard to other combinations. 1he íirst mentioned scale will be used in this
thesis íor data collection. 1he reason íor this is that the authors want each combination to
be e·aluated separately, while the respondents assume that a celebrity gets associated with
negati·e publicity. Also, the results will be more integrated and understandable in
comparison to a scale oí 1-9, which can pro·ide too a·eraged results. lurthermore, aíter
the pilot study was conducted, the authors came to the conclusion that a rating scale írom
1 to ¯ will make it easier íor the respondents to e·aluate the diííerent combinations. 1his is
also stated by Smith et al. ,2002, who claimed that this rating scale is normally expected to
show íulíilling reliability and ·ariability.
3.3.6 The survey
1he main sur·ey is di·ided into íi·e diííerent parts: ,1, presentation and task instructions
íor the respondent, ,2, description oí the attributes and their le·els, ,3, the combinations
,conjoint proíiles,, ,4, a post-experiment questionnaire, ,5, a short questionnaire
concerning the respondents` characteristics. 1he complete sur·ey is presented in appendix
2.
In the íirst part the respondents were instructed to assume that the celebrity was associated
with negati·e iníormation, and írom that rate each combination in order to show which
íactors a company should consider in such particular case. 1he second part the authors
belie·e it is highly important íor the respondents to understand the meaning oí each
attribute and their le·els, to íacilitate íor the respondent and to pro·ide the authors with
more accurate data. 1hirdly, the combinations will be presented in a ·erbal description
where the attributes will be introduced sequentially. 1he order in which the attributes are
presented is crucial since the results depend highly on this. 1wo diííerent cases will be
pro·ided íor the respondents to íacilitate their rating oí the attributes. In the íourth part
the respondents will be asked to e·aluate the diííerent attributes and state how important
they are íor a company to consider when a celebrity gets associated with negati·e publicity.
Lastly, the authors are not seeking to segment the total population based on the answers
that are gi·en. 1his leads to no particular interest in demographical iníormation, such as
age, sex, income etc. 1his is why, in the short questionnaire, the authors only asked about
their personal characteristics and how well íashion-oriented they are. L·en though the
authors randomly chose respondents within each school, they a·oided asking respondents
that are not conscious about íashion in order to enhance the ·alidity. By using non-
probability sampling the authors` choice oí data collection was supported.
3.4 Statistics
\hen conducting the regression analysis ,multiple regressions, in SPSS, an understanding
is needed to see how well the experiment has been explained by the regression. \hen
looking at the signiíicance oí l ,P-·alue, it shows how much oí the results that can be
Methodology
25
attributed to chance. 1his means the percentage oí data, which is not explained by the right
choice oí attributes and attribute le·els. 1he signiíicance ,P-·alue, has to be higher than 5
percent in order to support the results ,Gustaísson et al. 2001,.
Moreo·er, the signiíicance is di·ided into diííerent le·els depending on how signiíicant the
results are, and the íollowing le·els are presented below: ,Korner & \ahlgren, 1998,
1hree star signiíicant ,,: P-·alue·0.1 °
1wo star signiíicant ,,: 0.1 ° ·P-·alue·1.0 °
One star signiíicant ,,: 1.0 ° ·P-·alue·5.0 °
Analysis and results
26
4 Analysis and results
ív tbi. cba¡ter, tbe avtbor. ¡re.evt tbe avat,.i. ava re.vtt.. 1be vo.t reteravt fivaivg. i. brovgbt v¡, rbicb
tbe avtbor. betiere are iv¡ortavt to fvtfit tbe ¡vr¡o.e. ívrtbervore, av avat,.i. i. covavctea rbicb teaa. to
tbe aeci.iov ov rbicb attribvte. are vo.t iv¡ortavt.
4.1 Introduction
1he research sur·ey íor this thesis has been di·ided into a conjoint experiment and two
short questionnaires to be able to íulíil the purpose oí this thesis. lirstly, the consumers`
willingness to purchase a product or ser·ice when the attributes are in a conjunction is
presented. Secondly, an in·estigation will be brought up to see ií the consumers`
willingness to purchase a product or ser·ice diííers when diííerent endorsers are used. 1he
results íor both these parts are recei·ed írom the conjoint experiment. Data which is
deri·ed írom the conjoint experiment, according to Shepherd & Zacharakis ,199¯, can be
used through two statistical techniques, regression analysis and analysis oí ·ariance
,ANOVA,. linally, an analysis will be presented oí the standard de·iation and mean ·alue
when it comes to how consumers percei·e each attribute separately.
1he authors will introduce the results írom the research process, where each part is related
to a research question and there are also hypotheses addressed to the two íirst research
questions. 1he last research question deals with the two last parts in the research process.
1o íacilitate the structure oí this chapter the authors want to make it clear íor the reader
how the diííerent research questions are linked to the research process. 1his will be
explained in the table 4-1.
1able 4-1 Outline oí research process
Research question 1 Conjoint experiment, combination 1-16
Research question 2 Conjoint experiment, combination 1-8 case 1 ,Michael Jackson,.
Conjoint experiment, combination 9-16 case 2 ,Bill Clinton,.
Research question 3 Questionnaires regarding post-experiment and respondents
characteristics.
4.2 Research question 1
!bicb factor. frov tbe titeratvre rerier affect tbe cov.vver.` rittivgve.. to ¡vrcba.e a ¡roavct or .errice
tbat i. evaor.ea b, a cetebrit, tbat i. a..ociatea ritb vegatire ivforvatiov, rbev tbe attribvte. are iv a
covbivatiov.
lirstly, the le·el oí importance oí each attribute is presented when the respondents
e·aluated the combinations. 1here are only small diííerences between the attributes in
general ,see appendix 4A,. lowe·er, when comparing each bar some attributes are on a
higher le·el than the other ones. In all oí the combinations, expertise was seen as the most
important attribute íollowed by trustworthiness, similarity and likeability. All oí these
attributes, as was mentioned in the literature re·iew, are directly connected to the celebrity
and company. 1his implies that the choice and use oí a celebrity is crucial since it appears
Analysis and results
2¯
clearly in the íindings that these attributes iníluence the consumer mostly. Moreo·er, e·en
though the le·el oí importance between these íour attributes diííers slightly, they seem to
aííect the consumers combined with each other. 1he other two attributes, meaning transíer
and íit` match, showed a lower le·el oí importance than the íirst mentioned íour
attributes. 1he results indicate that the meaning transíer a celebrity shiíts to a certain brand
is more important than how well the íit between the celebrity and brand is. Both meaning
transíer and íit` are directly connected to celebrity and brand.
Looking íurther into how each indi·idual rated each combination, one can clearly see ,see
appendix 4B, that íit` and meaning transíer do not show any indications oí high
di·ergence. 1he respondents seem to ha·e similar opinions regarding these attributes
meaning that they are not considered to be the most crucial attributes when purchasing a
product,ser·ice. At the same time, the authors can strengthen this statement that the other
íour attributes, expertise, trustworthiness, similarity and likeability ha·e the greatest impact
on the consumers` willingness to purchase when a celebrity is used as an endorser. One can
íurther on see that combination 12 has the highest mean ·alue with a le·el oí 4.9, which
makes it the most important combination and basically the winning concept. \hen looking
at the concept ,see appendix 4C, the most important attributes are expertise,
trustworthiness, similarity and likeability, which also strengthen what is gi·en írom the
importance summary.
4.2.1 Testing the hypothesis
As mentioned beíore this part oí the research deals with the 16 combinations. \hen
conducting a multiple regression analysis, one has to take into account the diííerent
statistical terms, such as R-square and the signiíicance le·el. 1hese terms are presented in
the model summary ,see appendix ¯A,. \hen looking at the table, R-square is 0.316 which
means that 31.6 ° oí the obser·ed ·ariability oí the willingness to purchase is explained by
the attributes. lurthermore, R
2
is the squared correlation coeííicient between the obser·ed
·alues oí the dependent ·ariable and the predicted ·alue based on the regression model. A
·alue close to 1 tells you that the dependent ·ariable can be períectly predicted írom the
independent ·ariable, whereas a ·alue close to 0 pro·es the opposite. 1he R
2
·alue in this
case is 31.6 °, which is a moderately good model íit íor this type oí data.
Due to loss oí degrees oí íreedom, howe·er, the adjusted R
2
is only 12.8°. 1his loss oí
degrees oí íreedom also explains why the signiíicant ·alue íor the model is as high as 0.0¯8
that is higher than the 5° con·entionally used in signiíicant testing. lowe·er, this is still a
signiíicant result on the 10° le·el and gi·en the theoretical support íor these attributes our
conclusion is that the results are more in ía·our oí the than against hypothesis A.
H
A
: 1he celebrity-product combination attributes expertise, trustworthiness,
similarity, likeability, meaning transfer, and ´fit¨ (match) affect the consumers'
willingness to purchase a product or service also under condition of negative
information.
4.2.2 Results research question 1
As consumer willingness to purchase a product or ser·ice gets aííected by the attributes,
taken írom the literature re·iews, one can argue that these attributes are rele·ant when it
comes to celebrity endorsement strategy. lurthermore, these attributes showed diííerent
le·els oí importance íor consumers` willingness to purchase a product or ser·ice that is
endorsed by a celebrity that is associated with negati·e iníormation. It was also pro·en in
Analysis and results
28
the íindings that expertise, trustworthiness, similarity and likeability oí a celebrity ,source
attracti·eness and credibility, ha·e a strong impact on consumers. 1he percei·ed expertise
oí an endorser is more important when it comes to the purchasing intentions, rather than
trustworthiness. lence, the celebrity ,source, needs to íirst and íoremost be an expert in
the endorsement strategy and secondly be trustworthy in order to reach to the consumer
,recei·er,. 1he other attributes linked to a celebrity are also oí importance and a crucial
aspect to bear in mind is that a celebrity, who is seen as ha·ing all these six attributes linked
to him,her, is more likely to reach a certain target compared to a celebrity who only is
percei·ed as linked with one main attribute. As the íit` and meaning transíer do not show
any indications oí importance, one can say that consumers do not íind a celebrity that is
associated with negati·e iníormation to aííect the relationship it has with the brand or
company. linally, when looking at the importance summary one can see that the attributes
are relati·ely important, but mostly when it comes to expertise, trustworthiness, similarity
and likeability. Studying the 16 combinations in the multiple regressions, the authors íind
the regression to be signiíicant since there is no high skewness and at the same time it
strengthens the acceptance oí the hypothesis.
4.3 Research question 2
Do tbe attribvte. frov tbe titeratvre rerier bare aifferevt iv¡act. ov cov.vver.` rittivgve.. to ¡vrcba.e a
¡roavct or .errice, rbev aifferevt cetebritie. are v.ea a. evaor.er..
In this section, an in·estigation on how the respondents rated the diííerent combinations
related to each case is presented. 1he reason in doing this is to see how consumers percei·e
the use oí diííerent celebrities as endorser. 1his is also done to íind ií consumers react
equally no matter what celebrity is used at the same time as the particular celebrity is
associated with negati·e iníormation.
4.3.1 Case 1
Looking at the importance summary ,see appendix 5A, oí how the respondents rated the
attributes in this case, the authors can conclude that the most important attribute was
likeability íollowed by expertise. 1rustworthiness was nearly as important íollowed by
meaning transíer and similarity. Lastly, íit` was not considered to aííect the respondents`
willingness to purchase, as much as the pre·iously mentioned attributes. 1he indi·idual
subject importance ,see appendix 5B, shows that likeability is clearly the most important
attribute since the respondents agree upon its importance, while the other attributes are
percei·ed ·ery diííerent indi·idually. 1his indicates that the respondents ha·e diííerent
perceptions to the characteristics oí the celebrity.
4.3.2 Case 2
1he importance summary ,see appendix 6A, tells us that the most important attributes íor
the respondents were trustworthiness and meaning transíer. 1he third attribute that the
respondents considered to be important was íit`. 1he three other attributes, expertise,
similarity and likeability were considered to be oí equal importance. 1he authors can
íurther on claim, when looking at the indi·idual subject importance ,see appendix 6B,, that
trustworthiness was more important than meaning transíer based on that it was rated
higher. Meaning transíer and íit` were rated equally, but one can see that meaning transíer
was in most cases more important. Lastly, when looking at appendix 6B, one can see that
expertise, similarity and likeability are, as mentioned beíore, like wisely important. Similarity
was slightly more important in comparison with the other two, due to the di·erse rating.
Analysis and results
29
4.3.3 Testing the hypothesis
\hen testing this hypothesis the authors intend to compare the importance ·alues across
the two regressions in case one and two. lurthermore, a comparison between the R
2
·alues
in both cases is made. Looking at the importance summary oí the attributes ,see appendix
5A and 6A, in both cases, one can clearly see a signiíicant diííerence on how the
respondents percei·e the attributes when diííerent celebrities are used.
In case one the R
2
is 18.5° and in case two it is 25.6° ,see appendix ¯C and ¯L,. As
shown there is a diííerence between the R
2
·alues, which also supports the hypothesis B
that there is a diííerence in the perception oí the attributes, depending on which speciíic
celebrity is used as endorser.
H
B
: 1he relative importance of specific celebrity-product combination attributes is
different depending on which specific celebrity is used as endorser.
4.3.4 Results research question 2
1he hypothesis shows that consumers willingness to purchase or not is aííected by a
celebrity`s characteristics ,the attributes írom the literature re·iew,. 1his means that
consumers think and act diííerently depending on a particular celebrity who is used in the
endorsement strategy. 1his also indicates that companies need to take this into account
when choosing a celebrity, since the mechanism oí negati·e publicity associated with a
celebrity does not necessarily has to mean that the consumers choose not to purchase the
endorsed brand. Instead, a company need to ha·e in mind that the willingness to purchase
a product,ser·ice depends highly on to what extent consumers percei·e a celebrity as
possessing the six attributes. 1he higher perception oí a celebrity, the more likely and
willing they will be to purchase. Since two diííerent celebrities were used in each case, one
can clearly see that consumers tend to e·aluate a celebrity`s personal characteristics in
·arious ways.
In case one, the attributes likeability and expertise were rated as the most important ones
while íit` was not seen as playing an important role. In case two howe·er, trustworthiness
and meaning transíer were rated as most important whereas likeability surprisingly was seen
as the least important attribute. 1his clearly shows that there are clear diííerences between
how consumers percei·e diííerent celebrities depending on what characteristics they
possess and in what certain situation they are in. Also, as was mentioned in chapter two, it
was indicated that when consumers get asked about their purchasing beha·iour they usually
gi·e the impression that negati·e publicity in relation with brands do not usually aííect their
choices. lowe·er, the authors íeel that consumers purchasing beha·iours might occur
unconsciously and e·en though they claim that negati·e publicity does not ha·e any
iníluence on them, they most likely can. lurthermore, the authors can argue that the
attributes írom the literature re·iew are diííerently percei·ed in relationships with diííerent
celebrities.
4.4 Research question 3
!bicb attribvte. ao cov.vver. fiva of vo.t iv¡ortavce rbev eratvativg tbev .e¡aratet,.
In this section oí the empirical study, the respondents were asked to e·aluate the six
diííerent attributes separately. 1he reason is to compare the results írom the conjoint with
the sur·ey, which in can either strengthen the research process or indicate that the
respondents are not íully aware oí their own decision criteria.
Analysis and results
30
4.4.1 Perceived importance of attributes
In this section, the respondents were asked to rate the selí-percei·ed importance oí each
attribute. Lach oí the six attributes was e·aluated on a scale ranging írom 1, vviv¡ortavt, to
5, iv¡ortavt. 1he results oí the respondents` percei·ed importance oí the attributes are
a·eraged and illustrated in table 4-1.
1able 4-2 Ranging oí attributes
Mean Std. De· Ranking
Lxpertise 4.0 1.25 2
1rustworthiness 4.3 0.93 1
Similarity 3.4 1.1 4.5
Likeability 3.0 1.06 6
Meaning transíer 3.4 1.24 4.5
lit` Match 3.6 1.1 3
1he mean oí the selí-percei·ed weights ranged írom 3.0 íor ti/eabitit, to 4.3 íor
trv.trortbive... 1his indicates that the respondents percei·ed trv.trortbive.. as the most
important íollowed by expertise. 1he fit vatcb was ranked as the third most important
attribute íollowed by .ivitarit, and veavivg trav.fer which both were rated oí equal
importance. Lastly, the least important attribute is ti/eabitit, ,see appendix 8,.
4.4.2 Results research question 3
1he respondents rated the importance oí the attributes in general and this is why one can
see clear diííerences between which attributes are important and less important. Since the
consumers belie·e that the celebrity`s credibility is the most important, where both
trustworthiness and expertise are included, one can say that the consumers are more
aííected by a celebrity`s positi·e characteristics. lurthermore, how well the celebrity
communicates the message that the company is trying to send, aííects the consumers
purchasing decision. As it was stated that expertise and trustworthiness are the most
importance attributes, one can clearly state that the consumers ·alue the knowledge,
understanding and ethical aspects oí celebrity more than the other personal characteristics.
1his also implies that companies should take these attributes more into account when
choosing a celebrity as an endorser. 1his is because ií a particular celebrity gets associated
with negati·e iníormation it will not aííect the brand equity or lead to a decrease in the
market shares as much. At the same time, the celebrity will still transíer necessary
iníormation that a consumer can íind crucial and that potentially can lead to a purchase. It
was also íound in this research that a celebrity`s attracti·eness does not aííect the
consumers in any way, which can be chocking íor many consumers since they may assume
that the use oí celebrity endorsement is mainly beneíicial íor a company due to a celebrity`s
attracti·eness.
Analysis and results
31
4.5 Chapter summary
As the journey to íind the hidden íactors that companies need to consider when using
celebrities in ad·ertising is heading towards its end, the authors need to clariíy some crucial
aspects. Since, the íocus throughout this thesis has been conducted írom consumer`s point
oí ·iew, the authors will írom now on consider the respondents in this research process as
consumers.
1he authors can clearly state that the strategy oí using celebrity endorsement is an eííicient
way to reach out to a di·erse audience. 1he íocal point throughout this study was the
in·estigation on how willing consumers are to purchase a brand based on the attributes
elaborated írom the literature re·iew, in relation to a celebrity that gets associated with
negati·e iníormation. At this point the authors can clearly present íacts that consumers are
aííected by celebrities and that negati·e iníormation that a celebrity is associated with can
play an important role occasionally. In the promotional campaign where celebrities are
included, the purpose is to send out optimal iníormation to the consumers that will lead to
a purchase. 1aken the hypothesis A into account one can say that consumers are willing to
purchase a product or ser·ice based on the attributes taken írom the literature re·iew,
when they are in a combination. On the other hand, when in·estigating hypothesis B the
authors came to the understanding that the consumers` willingness to purchase a product
or ser·ice is aííected by a particular celebrity, when the celebrity has the mentioned
attributes as personal characteristics. 1hereíore, marketers in ad·ertising agencies need to
be aware that a celebrity`s personality together with brand- and the consumers indi·idual
personalities ,triangular relationship, are integrated, and the higher the linkage is between
these three parts the less likely consumers will get aííected negati·ely by negati·e
iníormation.
Since it was shown írom the three diííerent sections in this chapter ,research question 1-3,
that the consumers rated the attributes diííerently based on the diííerent cases, the authors
can at this point claim that consumers tend to percei·e attributes in ·arious ways. But, to
keep the íocus within the írame oí this thesis, the authors chose to analyze the relationship
between the ratings oí the diííerent concepts with how the consumers rated each attribute
separately. lirstly, when rating the 16 combinations, the most important attributes
,expertise, trustworthiness, similarity, likeability, are linked with the celebrity`s
characteristics ,see section 2.4,. On the other hand, the attribute meaning transíer which is
linked with the brand and íit` which in turn is linked with the company are not
considered to ha·e a particular importance. In the case oí Michael Jackson, the three most
important íactors ,likeability, expertise, trustworthiness, are also linked to the celebrity. At
the same time, meaning transíer and íit` are once again not percei·ed as crucial íactors.
lowe·er, in the case oí Bill Clinton both meaning transíer and íit` were considered to be
important íollowed by trustworthiness, and the other three attributes as less important.
\hat can be said írom this, írom the diííerent le·els oí importance in the diííerent cases·
It is as simple as the celebrities themsel·es and the brand they are endorsing aííects how
consumers react to the diííerent attributes in a combination. \hen comparing the two
cases with the total rating oí the 16 combinations, one can see that trustworthiness in all
cases is ·ital whereas the other attributes are rated diííerently. One can also clearly
understand that in the 16 combinations, expertise was seen as the absolute important
attribute which is also indicated to be important in the case oí Michael Jackson but less
important in the case oí Bill Clinton. 1his can be due to the product Clinton was
endorsing. 1hereíore, the authors belie·e that expertise also has a crucial aííect when a
celebrity gets associated with negati·e iníormation. 1his due to that expertise was percei·ed
Analysis and results
32
and ranked as the second most important when the consumers rated the attributes
separately. lurthermore, as the authors aim to reach a general conclusion, expertise would
be an accurate attribute to include when choosing a celebrity in ad·ertising purposes.
linally, based on e·erything mentioned so íar in this thesis, the authors can summarize the
most important íactors that companies need to consider when a celebrity gets associated
with negati·e iníormation. Basically, all attributes are important when using a celebrity as
an endorser but in the case when a celebrity gets associated with negati·e iníormation, the
authors e·aluated all the six attributes and compared them and came to decide upon the
íollowing attributes to be the most crucial attribute to be taken into account when
choosing and using a certain celebrity.
1rv.trortbive..: the reason why this attribute is important is basically that it was ranked
highly both when the consumers rated it in combination in all cases and when they rated it
separately. 1hereíore, there are se·eral reasons in concluding it as the most crucial
attribute.
í·¡erti.e: has been rated diííerently in each case but it was only rated as ha·ing a low
importance in the case oí Bill Clinton. Despite this, the authors belie·e that it is an
important attribute to consider when a celebrity gets associated with negati·e iníormation.
1he reason íor the low rating with the Clinton study could be explained as the product
which he promoted and also that consumers did not consider him being ·ery
knowledgeable in that particular íield.
1he remaining íour attributes, meaning transíer, likeability, similarity, and íit` did not in
most cases show a high le·el oí importance but rather an indication oí a·erage importance.
1hus, the authors do not íind them to be crucial íor a company when they use a celebrity
in the endorsement strategy. lowe·er, these can still be oí importance in conjunction with
the two íirst mentioned attributes. 1his is based upon that likeability in the case oí Michael
Jackson showed to be the most important attribute whereas when the attributes were rated
separately, this attribute was seen as the least important one. 1he remaining three showed a
relati·ely a·eraged importance and are thereíore not crucially unimportant. But, at the same
time they are not the dri·ing mechanism to successíul brand recognition when using a
celebrity in ad·ertising.
Conclusions and discussion
33
5 Conclusions and discussion
ív tbi. cba¡ter tbe avtbor. ritt fivatt, ¡roriae tbe reaaer ritb airect av.rer. to tbe ¡o.ea re.earcb qve.tiov..
ívrtbervore, fivat rora. frov tbe avtbor. ritt be girev. ía.tt,, a ai.cv..iov ritbiv tbe re.earcb area ritt
be etaboratea ava iv¡ticatiov. to ¡ro.¡ect .tvaie. iv tbe area of cetebrit, evaor.evevt ritt be girev.
5.1 Introduction
1his thesis in·estigated consumers` willingness to purchase a product or ser·ice based on
the six attributes that were retrie·ed írom the literature re·iew. In the introduction chapter,
three research questions were identiíied. 1he direct answers to the research questions will
be more speciíied in section 5.2. 1o be able to answer these questions, the quota sampling
oí ¯5 uni·ersity students in the campus oí Uni·ersity oí Jonkoping was studied. 1hese
students were asked to e·aluate 16 hypothetical cases, which are based on the attributes
retrie·ed írom the literature re·iew and they were asked to indicate the willingness to
purchase a product or ser·ice.
lurthermore, the literature re·iew pro·ided the authors with rele·ant iníormation oí the
crucial attributes that are associated with celebrity endorsement. 1hese attributes ha·e
pre·iously been used when gaining an optimal use oí a celebrity as an endorser ,see íigure
1-1,. 1hus, it has led to the íocal point oí this thesis which is to in·estigate whether the
attributes írom the literature re·iew really aííect the consumers` willingness to purchase a
product or ser·ice, when a celebrity is associated with negati·e iníormation.
1he authors came to the conclusion aíter conducting the research process that two oí the
six attributes are the most important attributes a company should consider beíore choosing
and using a celebrity, due to the possibility that a celebrity can get associated with negati·e
iníormation. 1his will íurther on be elaborated on in section 5.3. Beíore tackling and
starting this discussion, the authors will answer the research questions. 1he last section oí
this chapter will, aíter all the diííerent elements within this thesis, present suggestions that
can be undertaken in íuture studies.
5.2 Research questions
!bicb factor. frov tbe titeratvre rerier affect tbe cov.vver.` rittivgve.. to ¡vrcba.e a ¡roavct or .errice
tbat i. evaor.ea b, a cetebrit, tbat i. a..ociatea ritb vegatire ivforvatiov, rbev tbe attribvte. are iv a
covbivatiov.
As it has been pro·en in this thesis, the authors can now conclude that the consumers
purchasing beha·iour in general do get aííected by the mentioned attributes írom the
literature re·iew. 1his is also strengthened in the support oí the hypothesis, which implies
that the attributes in combination really aííect the consumers` willingness to purchase
under conditions oí negati·e iníormation. 1hereíore, the authors íound it ·ery important
to identiíy these crucial attributes. As mentioned in section 4.2.2 expertise, trustworthiness,
likeability and similarity were the main ones.
Marketers need to be aware that consumers percei·e celebrities in diííerent ways based on
what personal characteristics they possess, thus leading to a purchase. \hen a celebrity is in
a particular situation and promoting diííerent products or ser·ices, the consumers tend to
preíer a celebrity who is an expert and trustworthy within that situation. Moreo·er, the
celebrity should preíerably be likeable among the targeted consumers and ha·e similar
goals, interest etc. 1hese attributes in a combination could be the winning concept in a
celebrity endorsement strategy and increased market shares. 1he more knowledge and the
Conclusions and discussion
34
higher acceptance the consumers ha·e regarding the celebrity, when e·aluating the
problematic` endorser, the more likely they still are to make a purchase oí a
product,ser·ice.
Do tbe attribvte. frov tbe titeratvre rerier bare aifferevt iv¡act. ov cov.vver.` rittivgve.. to ¡vrcba.e a
¡roavct or .errice, rbev aifferevt cetebritie. are v.ea a. evaor.er..
1he two diííerent cases with two diííerent celebrities show clearly that consumers react in
·arious ways to celebrities who are associated with negati·e iníormation. It is also crucial to
point out that consumers may percei·e a separate attribute diííerently in comparison to the
six attributes in a combination and in relation to diííerent endorsers. 1he diííerent
e·aluations oí the attributes with ·arious endorsers can be related to the answer oí research
question one, regarding what situation the celebrity is in and what the endorser is
promoting. lence, the attributes írom the literature re·iew ha·e diííerent impacts on the
consumer beha·iour depending on who the endorser is. linally, the authors cannot see a
direct association between the used attributes írom the literature re·iew and the le·el oí
negati·e iníormation.
!bicb attribvte. ao cov.vver. fiva of vo.t iv¡ortavce rbev eratvativg tbev .e¡aratet,.
\hen the consumers e·aluated the attributes separately, the attribute trustworthiness was
the most important one íollowed by expertise and íit`. Similarity and meaning transíer
were oí equal importance and likeability was seen as the attribute which played the least
important role. 1hese íindings together with the chapter summary in section 4.5, the
authors ha·e in a conjunction íound the two most important attributes that consumers
íound to be oí most importance when e·aluating them separately and e·en in
combinations. 1rustworthiness was throughout this thesis indicated as a ·ery important
attribute and expertise was almost oí equal importance. 1hereíore, the authors can
conclude that the consumers preíer a celebrity who has the characteristics oí being an
expert and trustworthy within the íield oí ad·ertising.
5.3 Discussion
Bringing the íindings írom the research questions together, the authors ha·e decided to
pro·ide the readers with a model ,íigure 5-1, in order to gi·e a general apprehension oí
their conclusions. 1his model is a deeper elaboration on íigure 1-2 since the purpose oí this
thesis was to íind the most important attributes that marketers need to consider when
using celebrities in their endorsement strategies. ligure 1-2 did not indicate what íactors oí
a celebrity that is oí main importance but aíter conducting this research, the authors can
íinally conclude that the attributes trustworthiness and expertise are the hidden íactors that
need to be considered when ad·ertisers use the strategy oí celebrity endorsement.
Conclusions and discussion
35
ligure 5-1 1he network oí optimal use oí a celebrity
1he attributes trustworthiness and expertise are combined in a network within the celebrity
endorsement strategy and since the consumers e·aluated them as the most important ones,
the authors assume this model to be as the optimal use oí a celebrity. A company needs to
be aware oí íigure 5-1 and the interaction in order to gain success or e·en pre·ent íailure
when using celebrity endorsers. 1hese attributes are directly linked to the celebrity and
these are what marketers should consider when choosing a particular celebrity. 1hese will
e·entually aííect the company, consumer and brand when a celebrity gets associated with
negati·e publicity. It is important to ha·e in mind that a celebrity who is linked with
negati·e iníormation and that possess these attributes to a ·ery high degree, is less likely to
aííect the consumer beha·iour negati·ely in comparison to the opposite case.
5.4 Final words from the authors
1he main objecti·e oí this thesis was to come to an understanding oí which íactors
consumers íind to be important íor a company to consider when a celebrity gets associated
with negati·e iníormation. 1hese íactors were as mentioned earlier trustworthiness and
expertise. But, the authors claim that only considering these íactors does not eliminate the
other ones that has been used in the research process but rather that the identiíied most
important íactors are the ·ital ones. Now that the purpose is íulíilled, the authors ha·e
gained more insight within the area oí celebrity endorsement and hopeíully pro·ide
marketers with crucial iníormation and knowledge that will beneíit their company and
strengthen their brand equity. It is also crucial to bare in mind that the selection oí a
celebrity should not only include the two most important attributes but also weigh the
remaining íour as a priority, since these may strengthen the perception oí a celebrity in
collaboration with the top listed attributes. Moreo·er, there is no strong and determinant
indication that the íound attributes are the only ones to consider since there may be indeed
other hidden íactors that consumers consider. 1he authors can now claim that this íield is a
continuously ongoing process and one should be clear that there is no correct way oí doing
things, but rather pro·ide helpíul insights on what íactors marketers should consider when
a celebrity gets associated with negati·e publicity.
5.5 Prospects for future researches
Some oí the results in this research process suggest interesting prospects íor íuture
research. 1he íindings írom the post-experiment showed diííerent results compared to the
general understanding oí each attribute. Consumers tend to think and act diííerently and
Celebrity
Company Brand Consumer
1rustworthiness Lxpertise
Conclusions and discussion
36
this is the main reason why the authors ha·e not claimed that the results are the accurate
ones. 1his research pro·ided the readers with a general understanding oí this subject and it
is crucial to íurthermore state that this subject can be narrowed down e·en more, mostly
when it comes to psychological aspects. Potentially, this can lead to new and exiting
íindings that can pro·ide marketers with more understanding and a deep íocus on certain
psychological issues to take into account.
\hat can be interesting to elaborate deeper on is how the negati·e iníormation oí a
celebrity aííects low- and high in·ol·ement products,ser·ices or luxury- and low cost
products,ser·ices. Moreo·er, what happens ií a celebrity in the case oí negati·e
iníormation promotes a brand that do not íit well with the celebrity is also oí peculiar
interest.
Another major area except íor the psychological aspect is the demographical area. 1his
implies that studies regarding perceptions in diííerent countries, ages, sex etc. can be
conducted to clariíy more speciíic issues that can be helpíul and beneíicial íor marketers.
lurthermore, as it has been stated in pre·ious studies, that one cannot predict the liíe span
oí a celebrity and combined with the íindings in this thesis, there are no clear assumptions
ií the consumers` willingness to purchase a product or ser·ice that is promoted by a
celebrity and associated with negati·e iníormation. 1he authors belie·e that a new research
area is de·eloped. 1his one deals with how the e·er changing identity oí a celebrity might
aííect the consumers` willingness to purchase, while some consumers might íind this
change oí the celebrity as something negati·e. 1hus, the authors assume that a potential
research area can be to in·estigate ií and in that case how consumers react to a celebrity`s
identity change during their time in the spotlight and also ií the celebrity might through a
business perspecti·e beneíit more ií they start their own business, which will lead us in to a
new phenomenon oí celebrity entrepreneurship.
Lastly, it would also be interesting to conduct a study within this subject, mainly írom
marketers point oí ·iew or e·en an intertwined study írom both consumers and marketers
perspecti·es. Concluding this, the authors want to make it clear and agree with the íact that
the ad·ertising industry is really eííicient when using celebrities as endorsers due to its
ability to reach a wider audience where the consumers can identiíy themsel·es with. 1he
use oí ad·ertising is and will change in the íuture, but at this time celebrities are the dri·ing
mechanism to successíul ad·ertising within this industry.
Critique oí this research process
3¯
6 Critique of this research process
´ivce tbi. tbe.i. ba. beev covavctea ritb a tive cov.traivt ava tivitea re.ovrce., tbe avtbor. cav covre,
.ererat .etf·critici.v. ava ;vage tbe re.earcb ¡roce.. .o tbat tbe reaaer ritt vot get vi.teaa or ivter¡ret bia.ea
ivforvatiov.
1he authors íeel that the limited time and resources ha·e aííected the methodology. A
qualitati·e study ,íocus groups, could ha·e beneíited this thesis e·en more in the sense that
it could ha·e underpinned the main method used ,quantitati·e,. lurthermore, it was
theoretically supported that at least 50 respondents can be used in a conjoint experiment
and e·en though ¯5 respondents were used in the research process, the authors íeel that a
higher number oí respondents could ha·e beneíited this study to a greater extent and
pro·ided more accurate results. Moreo·er, the R
2
le·el in all combinations ,1-16, 1-8, 9-16,
were on a low le·el which indicates that the obser·ed ·ariability also was low. 1he R
2
le·el
was on an a·erage le·el throughout the results and ií the R
2
le·els were closer to 1 it would
increase the reliability oí the thesis. 1his can be due to the low amount oí respondents.
Another reason íor this can be the number oí combinations oí the conjoint experiment. A
larger amount oí combinations ,cards, and íewer attributes would gi·e less statistical
uncertainty and more signiíicant results.
1he results show that the celebrity-product attribute that ha·e been íound important in
pre·ious research hold up in a context oí a celebrity that is associated with negati·e
iníormation. lowe·er, a more ideal design would ha·e included negati·e iníormation
instead oí as a consequence oí the attributes, allowing íor a more direct comparison.
lurthermore, the authors came to the understanding that as much as conjoint came across
as time consuming, they were going more íor well answered questionnaires then halí done
ones. lurthermore, the authors belie·e that the conjoint experiment is such a well co·ered
research process that many diííerent results could ha·e been elaborated írom it. But, the
authors decided to keep it as accurate to the purpose oí this thesis, which was to íind the
hidden íactors and not in·estigating the diííerent relationships between the diííerent
attributes. 1he concern was more on which attributes the respondents percei·ed to be
most important linked with the purpose oí this thesis. 1his is why the authors only kept the
number oí hypotheses to two instead oí a larger number. 1here was a direct interest to
in·estigate other aspects, mainly ií the consumers were willing to purchase a product or
ser·ice ad·ertised by a celebrity that is associated with negati·e iníormation, and also ií
consumers respond diííerently in reaction to negati·e iníormation or the celebrities
negati·e iníormation.
1he last point oí criticism that is important to point out is the order oí combinations.
\hen looking at appendix 4 A, one can see that the le·el oí importance has a certain
symmetry where it starts oíí with a high le·el and decreases slightly aíter each combination
used. lence, ií the combinations where placed in a diííerent order the outcome might ha·e
diííered and lead to other results.
Reíerences
38
References
Aaker, D. A. ,2000,. ßrava íeaaer.bi¡. A Di·ision oí Simon & Schuster, Inc: New \ork.
Aaker, D. A. ,1991,. Mavagivg ßrava íqvit,. 1he íress press, New \ork.
Aczel, A. D., & Sounderpandian, J. ,2002,. ßv.ive.. ´tati.tic. ,5
th
ed.,. McGraw-lill ligher
Lducation: New \ork.
Antilla, M., leu·el, R. R., & Moller, K. ,1980,. Conjoint Measurement íor Marketing
Management. ívro¡eav ]ovrvat of Mar/etivg. 11,¯,, 39¯-408.
Belch, G.E., & Belch, M.A. (2001). Advertising and Promotion: An integrated
Marketing Communications Perspective (5
th
ed.). Boston: Irwin/MaGraw-
Hill.
Bielli, A. (2003).The Research Power Behind Great Brands. www.millwardbrown.com.
Retrieved 2005-02-15.
Carson, D., Gilmore A,. Perry C & Gronhaug K. ,2001,. Qvatitatire Mar/etivg Re.earcb.
SAGL Publications Ltd. Guildíord: Great Britain.
Chabo, D., & Saouma, J. ,2005,. 1be O¡tivat v.e of Cetebrit, ívaor.evevt. Jonkoping
International Business School: Jonkoping.
Chabo, D., & Saouma, J. ,2005,. 1be íffect of ^egatire ívforvatiov. Jonkoping International
Business School: Jonkoping.
Chabo, D., & Saouma, J. ,2005,. 1be ^etror/ of O¡tivat v.e of a Cetebrit,. Jonkoping
International Business School: Jonkoping.
Churchill, G. A. ,1995,. Mar/etivg Re.earcb· Metboaotogicat íovvaatiov. ,6
th
Ld., New \ork:
1he Dryden Press.
Danesh·ary, R., & Schwer, K.R. ,2000,. 1he Association and Consumers Intention to
Purchase: ]ovrvat of Cov.vver Mar/etivg, 1¨,3,, 203-213.
Dean, l. D. ,1999,. Brand Lndorsement, Popularity, and L·ent Sponsorship as
Ad·ertising Cues Aííecting Consumer Pre-Purchase Attitudes. ]ovrvat of
.arerti.ivg. 2º ,3,.
Bruns, V. ,2004,. !bo Receire. ßav/ íoav..· . .tva, of tevaivg officer.` a..e..vevt. of toav. to
grorivg .vatt ava veaivv·.i¸ea evter¡ri.e.. Jonkoping International Business
School, Jonkoping Uni·eristy.
Lkdahl, l. ,199¯,. ívcrea.ea Cv.tover ´ati.factiov |.ivg De.igv of í·¡erivevt., Cov;oivt .vat,.i.
ava QíD. Linkoping: Di·ision oí Quality 1echnology and Management
Department oí Mechanical Lngineering, Linkoping Uni·ersity.
Lrdogan, Z.B., Baker, M.J., & 1agg, S. ,2001,. Selecting Celebrity Lndorsers: 1he
Practitioner`s Perspecti·e: ]ovrvat of .arerti.ivg Re.earcb, 11, 1-26.
Reíerences
39
Lsaiasson, P., Gilljam, M., Oscarsson, l., & \ängnerud, L. ,2002,. Metoa¡ra/ti/av: Kov.tev
att .tvaera .avbatte, ivairia ocb var/vaa. Stockholm: Nordstedts Juridik.
llick, U. ,2002,. .v ívtroavctiov to Qvatitatire Re.earcb. ,2
nd
ed., SAGL Publications Ltd.
\iltshire: Great Britain.
Goldsmith, L. R, Laííerty, A. B, Newell, J.S. ,2000,. 1he Impact oí Corporate Credibility
and Celebrity Credibility on Consumer Reaction to Ad·ertisements and
Brands. ]ovrvat of .arerti.ivg, 2· ,3,.
Green, P.L., & Srini·asan, V. ,1990,. Conjoint Analysis in Marketing: New de·elopments
with implications íor research and practise. ]ovrvat of Mar/etivg. :1,4,, 3-19.
Green, P.L., & Srini·asan, V. ,19¯8,. Conjoint Analysis in Consumer Research: Issues and
Outlook. ]ovrvat of cov.vver Re.earcb, :: 103-123.
Green, P. L. ,2001,. loreword. In A. Gustaísson, A. lerrmann, & l. luber ,Lds.,,
Cov;oivt vea.vrevevt·vetboa. ava a¡¡ticatiov., ,2
nd
ed., Berlin: Springer Verlag.
Gustaísson, A., lerrman, A., & luber, l. ,2001,. Conjoint Analysis as an Instrument oí
Market Reserach Practice. In A. Gustaísson, A. lerrman, & l. luber ,Lds.,,
Cov;oivt vea.vrvevt: vetboa. ava a¡¡ticatiov., 2
nd
ed. Berlin:Springer Verlag.
Gwinner, K. ,199¯,. . Moaet of ívage Creatiov ava ívage 1rav.fer iv írevt ´¡ov.or.bi¡.
International Marketing Re·iew, 11 ,3,, 145-158. North Carolina: USA.
Iggland, B. ,1989,. .vravaarbeaövivg för ¡roav/tvtrec/tivg· ígev./a¡.raraerivgar vea Cov;oivt
.vat,.i.. Linkoping: Department oí Management and Lconomics: Linkoping
Uni·ersity.
Johnson, R. M. ,198¯,. Adapti·e Conjoint Analysis, in: ´artootb .vat,.i. Covferevce
Proceeaivg., Ketchum, ID: Sawtooth Soítware, July, 253-65.
Kambitsis, C., larahousou, \., 1heodorakis, N., & Chatzibeis, G. ,2002,. Sports
Ad·ertising in Print Media: 1he case oí 2000 Olympic Games: Cor¡orate
covvvvicatiov: .v ivtervatiovat ]ovrvat, ¨,3,, 155-161.
Kamins, A.M. ,1990,. An in·estigation into the match-up` hypothesis in celebrity
ad·ertising: when beauty may be only skin deap. ]ovrvat of .arerti.ivg, 1·:
American Academy oí Ad·ertising.
Kamins, A.M., Brand, J.M., loeke. A.S & Moe, C.J. ,1989,. 1wo-Sided ·ersus One-Sided
Celebrity Lndorsements: 1he Impact on Ad·ertising Lííecti·eness and
Credibility. ]ovrvat of .arerti.ivg, 1º: American Academy oí Ad·ertising.
Keller, L. K. ,2002,. Branding and Brand Lquity. Mar/etivg ´cievce ív.titvte. Cambridge:
USA.
Kotler, P., Armstrong, G., Saunders, J., & \ong, V. ,2001,. Privci¡te. of Mar/etivg. Prentice
lall: larlow.
Korner, S., & \ahlgren, L. ,1998,. ´tati.ti./a vetoaer. Lund: Studentlitteratur.
Reíerences
40
Louie, 1.A., & Obermiller, C. ,2002,. Consumer Response to a lirm`s Lndorser ,Dis,
association Decisions: ]ovrvat of .arerti.ivg, ²1.
Lundell, M. ,2005,. 1¨ Mit;over för ²0 .e/vvaer. Aítonbladet.
lttp:,,www.aítonbladet.se,·ss,ekonomi,story,utskriít,0,3258,600119,00.ht
m. Retrie·ed 2005-02-08.
Malhotra, K. N. ,1996,. Mar/etivg re.earcb: .v a¡¡tiea orievtatiov ,2th ed.,. New
Jersey
rentice lall.
McCracken, G. ,1989,. \ho is celebrity endorser· Cultural íoundations oí the celebrity
endorsement process`. ]ovrvat of Cov.vver Re.earcb, 1ó,3,. 310-21.
O´Mahony, S. & Meenaghan, 1. ,199¯,98,. 1he impact oí Celebrity Lndorsements on
Consumers: íri.b Mar/etivg Rerier, 10 ,2,, 15-24.
Ohanian, R. ,1990,. Construction and Validation oí a scale to measure celebrity endorsers`
percei·ed expertise, trustworthiness and attracti·eness. ]ovrvat of .arerti.ivg, 1·:
American Academy oí Ad·ertising.
Pringle, l. ,2004,. Celebrity Sells. ]obv !ite, c ´ov.. \iltshire: UK.
Redenbach, A. ,2005,. A Multiple Product Lndorser Can Be A Credible Source. C,ber·
]ovrvat of ´¡ort Mar/etivg. ISSN. 132¯-6816. Griííith Uni·ersity.
Riezebos, R., Kist, B., Koostra. G. (2003), Brand Management. A theoretical and
practical approach. Prentice Hall.
Saunders, M., Lewis P., & 1hornhill, A. ,2003,. Re.earcb vetboa. for bv.ive.. .tvaevt., Person
education limited: larlow.
Schiííman, G. Leon & Kanuk, L. Lazar. ,2004,. Cov.vver ßebariovr. Pearson education inc:
New Jersey.
Schlecht, C. ,2003,. Celebrities` Impact on Branding. Cevter ov Ctobat ßrava íeaaer.bi¡.
Columbia Business School: New \ork.
Se·ern, J., Belch G.L. & Belch M.A. ,1990,. 1he eííects oí sexual and non-sexual
ad·ertising appeals and iníormation le·el on cogniti·e processing and
communication eííecti·eness. ]ovrvat of aarerti.ivg, 1·1, 14-22.
Sil·era, D.l., & Austad. B. ,2004,. lactors predicting the eííecti·eness oí celebrity
endorsement ad·ertisements: ívro¡eav ]ovrvat of Mar/etivg, ²º,11,12,, 1509-
1526.
Shepherd, D. A., & Zacharakis, A. ,199¯,. Conjoint analysis: A window oí opportunity íor
entrepreneurship research. .aravce. iv ívtre¡revevr.bi¡, íirv ívergevce, ava
Crortb, ², 203-248.
Smith, A.L., Schullen, L.S., & Barr, l.S. ,2002,.Orgavi¸atiovat Re.earcb Metboa., : ,4,,
October 2002, 388-414.
Reíerences
41
Solomon, M.R. ,2002,. Cov.vver ßebarior: ßv,ivg, íarivg,ava ßeivg, 5
th
ed.Prentice lall: New
Jersey.
Soderlund, M. ,2003,. ívotiov.taaaaa Mar/vaa.förivg. Liber: Malmo.
1ang, L.P., Kim, K.J & 1ang, L.1. ,2001,. Lndorsement oí the Money Lthic, Income, and
Liíe Satisíaction - A comparison oí íull-time employees, part-time employees,
and non-employed uni·ersity students. ]ovrvat of Mavageriat P.,cbotog,, 1¨ ,6,,
1ennessee: USA.
1ill, B.D., & Schimp, 1.A. ,1998,. Lndorsers in Ad·ertising: 1he case oí Negati·e Celebrity
Iníormation: ]ovrvat of .arerti.ivg, 2¨.
1ill, B.D., & Busler, M. ,1998,. Matching products with Lndorsers: Attracti·eness ·ersus
Lxpertise: ]ovrvat of Cov.vver Mar/etivg, 1: ,6,, 5¯6-586.
1hurrot, P. ,2004,. \ith iPod success, Apple takes network eííect` írom Microsoít.
lttp:,,www.windowsitpro.com,Articles,Print.cím·ArticleID~443¯6.
Retrie·ed 2005-03-11.
Van de Pol, M., & Ryan, M. ,1996,. Using Conjoint analysis to establish consumer
preíerences íor íruit and ·egetables. ßritti.b íooa ]ovrvat, ·º,8,, 5-12.
\algren, J. C, Ruble, A.C & Donthu, N. ,1995,. Brand Lquity, Brand Preíerence &
Purchase Intentions. ]ovrvat of .arerti.ivg, 21 ,3,.
\heeler, R. ,2003,. Choosing Celebrity endorsers: 1ips and traps: ^ov¡rofit !orta., 21,4,,
1¯-20.
42
Appendix 1- 8 profiles for the six attributes at two levels
Lxpertise 1rust Similarity Likeability Meaning
transíer
Match
íit`
Concept
Low Low Low Low Low Low Design 1
ligh ligh Low Low ligh Low Design 2
Low ligh ligh Low ligh ligh Design 3
ligh Low ligh Low Low ligh Design 4
Low ligh Low ligh Low ligh Design 5
ligh Low Low ligh ligh ligh Design 6
ligh ligh ligh ligh Low Low Design ¯
Low Low ligh ligh ligh Low Design ¯
43
Appendix 2 - Research Survey
Dear re.¡ovaevt!
\ou ha·e been chosen to take part oí our sur·ey based on our Master`s thesis in marketing.
1his thesis has its main íocus on celebrity endorsement and as you might ha·e noticed, this
concept is being more írequently used by marketers in the ad·ertising industry.
lurthermore, most oí the celebrities that ha·e been used ha·e generated a high brand
exposure and ha·e been associated with a certain brand, which in terms ha·e aííected the
consumer buying beha·iour.
1he use oí celebrity endorsement has mostly a positi·e outcome íor companies, since the
celebrity`s characteristics íit well with the characteristics oí a brand. Most oí the celebrities
used by companies ha·e been chosen since they are considered to ha·e a strong power on
consumers. 1his means that consumers may buy a certain brand that is associated with a
celebrity to identiíy themsel·es with this particular brand and also íeel a sense oí similarity
with the celebrity. lowe·er, one can wonder how consumers get aííected by celebrities
that ha·e been associated with negati·e publicity. 1his is what we want you to consider
when rating the different combinations of attributes and how the negative publicity
of a celebrity affects YOU.
\e will pro·ide you with two diííerent cases in order íor you to get a better understanding
oí how to rate when imagining that particular situation, and how that would eííect you in
your buying decision. \e want to emphasize that the generated cases are made up. Simply,
what we want from you is to rate each combination of attributes from J to 7. Please
e·aluate each combination as a separate situation independent oí all others. Once you ha·e
completed one combination, you are not supposed to go back and take a pre·ious look.
\ou should also read through the descriptions oí the attributes and their le·el one time
only. Moreo·er, ií you consider one combination not to ha·e a great impact on you, you
can rate it as number 1 and írom that point rate the combinations a higher number the
more impact they ha·e on you with ¯ being the highest.
\e would like to make it clear that there is no right or wrong answer, we are more
interested in what you consider to be most important when a celebrity gets associated with
negati·e publicity. With negative publicity we mean any information that has
negative association that decreases the trustworthiness of the celebrity and the fit
with the brand.
1he whole experiment might be considered as time consuming but it will probably not take
you more than 15 minutes to complete this task. linally, we emphasize that your response
will remain anonymous and thank you íor your participation.
44
Attribute Level J Level 2
í·¡erti.e: 1he le·el oí
knowledge, experience
and expertise a celebrity
has in a particular brand.
íigb: 1he celebrity
has a high le·el oí
knowledge,
experience and
expertise.
íor: 1he celebrity has a
low le·el oí knowledge,
experience and expertise.
1rv.trortbive..: low
honest and belie·able a
celebrity is when
ad·ertising a brand.
íigb: 1he celebrity
is considered to be
·ery trustworthy.
íor: 1he celebrity is not
percei·ed to be honest
and belie·able enough.
´ivitarit,: 1he sense oí
similarity a consumer íeel
with a celebrity endorser.
íigb: 1he consumer
íeels highly linked
with the celebrity
when it comes to
the characteristics.
íor: 1he consumer does
not íeel any connection
with the celebrity.
íi/eabitit,: 1he le·el oí a
celebrity`s íame in the
public eye and how
popular this person is.
íigb: 1he celebrity
is highly íamous
and popular.
íor: 1he celebrity is not
likeable or popular.
Meavivg trav.fer: 1he
meaning a celebrity brings
or transíers to a certain
brand with the help oí
their personality and liíe-
styles.
íigb: 1he celebrity
is successíully
transíerring
meaning to a brand.
íor: 1he celebrity does
not succeed in
transíerring any meaning
to the brand.
íit Matcb: low well the
íit between the celebrity
and the brand is.
íigb: A ·ery good
íit.
íor: No íit at all.
45
Case J. Plastic Iantastic AB
1he last two years, Plastic lantastic AB has been sponsoring the most amazing and
extreme make-o·er 1V-show ,1he Swan, íor regular people in the USA. 1his has been
·ery successíul in the domestic market as well as internationally. 1hereíore, Plastic
lantastic has now realized that this can be a beneíicial way to reach out to a larger audience
in a diííerent way than the original 1V-show promotion. 1alking about 1V-shows, Michael
Jackson is currently not only notorious íor attending his own court on international 1V,
where he is accused oí child molesting, but he is also promoting the ser·ices Plastic
lantastic AB oííers.
Combination J
Lxpertise Low
1rustworthiness Low
Similarity Low
Likeability Low
Meaning transfer Low
´Iit¨ Match Low
Based on the attributes mentioned abo·e, how willing are you to buy this ser·ice·
,Please rate this combination,
Unwilling 1 2 3 4 5 6 ¯ \illing
46
Combination 2
Lxpertise High
1rustworthiness High
Similarity Low
Likeability Low
Meaning transfer High
´Iit¨ Match Low
Based on the attributes mentioned abo·e, how willing are you to buy this ser·ice·
,Please rate this combination,
Unwilling 1 2 3 4 5 6 ¯ \illing
4¯
Combination 3
Lxpertise Low
1rustworthiness High
Similarity High
Likeability Low
Meaning transfer High
´Iit¨ Match High
Based on the attributes mentioned abo·e, how willing are you to buy this ser·ice·
,Please rate this combination,
Unwilling 1 2 3 4 5 6 ¯ \illing
48
Combination 4
Lxpertise High
1rustworthiness Low
Similarity High
Likeability Low
Meaning transfer Low
´Iit¨ Match High
Based on the attributes mentioned abo·e, how willing are you to buy this ser·ice·
,Please rate this combination,
Unwilling 1 2 3 4 5 6 ¯ \illing
49
Combination S
Lxpertise Low
1rustworthiness High
Similarity Low
Likeability High
Meaning transfer Low
´Iit¨ Match High
Based on the attributes mentioned abo·e, how willing are you to buy this ser·ice·
,Please rate this combination,
Unwilling 1 2 3 4 5 6 ¯ \illing
50
Combination 6
Lxpertise High
1rustworthiness Low
Similarity Low
Likeability High
Meaning transfer High
´Iit¨ Match High
Based on the attributes mentioned abo·e, how willing are you to buy this ser·ice·
,Please rate this combination,
Unwilling 1 2 3 4 5 6 ¯ \illing
51
Combination 7
Lxpertise High
1rustworthiness High
Similarity High
Likeability High
Meaning transfer Low
´Iit¨ Match Low
Based on the attributes mentioned abo·e, how willing are you to buy this ser·ice·
,Please rate this combination,
Unwilling 1 2 3 4 5 6 ¯ \illing
52
Combination 8
Lxpertise Low
1rustworthiness Low
Similarity High
Likeability High
Meaning transfer High
´Iit¨ Match Low
Based on the attributes mentioned abo·e, how willing are you to buy this ser·ice·
,Please rate this combination,
Unwilling 1 2 3 4 5 6 ¯ \illing
53
Case 2. Dell and Bill
Dell computers are launching their new product, Monica 69 \l, which includes a certain
system íor managers` at large companies to enhance the process oí keeping track oí
numbers. 1his indicates that Dell`s main target is business-oriented people. 1o be able to
succeed with reaching this particular group, Dell ha·e been using Bill Clinton as a
spokesperson in promoting Monica 69 \l.
Combination 9
Lxpertise Low
1rustworthiness High
Similarity High
Likeability Low
Meaning transfer High
´Iit¨ Match High
Based on the attributes mentioned abo·e, how willing are you to buy this product·
,Please rate this combination,
Unwilling 1 2 3 4 5 6 ¯ \illing
54
Combination J0
Lxpertise Low
1rustworthiness Low
Similarity Low
Likeability Low
Meaning transfer Low
´Iit¨ Match Low
Based on the attributes mentioned abo·e, how willing are you to buy this product·
,Please rate this combination,
Unwilling 1 2 3 4 5 6 ¯ \illing
55
Combination JJ
Lxpertise Low
1rustworthiness High
Similarity Low
Likeability High
Meaning transfer Low
´Iit¨ Match High
Based on the attributes mentioned abo·e, how willing are you to buy this product·
,Please rate this combination,
Unwilling 1 2 3 4 5 6 ¯ \illing
56
Combination J2
Lxpertise High
1rustworthiness High
Similarity High
Likeability High
Meaning transfer Low
´Iit¨ Match Low
Based on the attributes mentioned abo·e, how willing are you to buy this product·
,Please rate this combination,
Unwilling 1 2 3 4 5 6 ¯ \illing
5¯
Combination J3
Lxpertise High
1rustworthiness Low
Similarity Low
Likeability High
Meaning transfer High
´Iit¨ Match High
Based on the attributes mentioned abo·e, how willing are you to buy this product·
,Please rate this combination,
Unwilling 1 2 3 4 5 6 ¯ \illing
58
Combination J4
Lxpertise High
1rustworthiness High
Similarity Low
Likeability Low
Meaning transfer High
´Iit¨ Match Low
Based on the attributes mentioned abo·e, how willing are you to buy this product·
,Please rate this combination,
Unwilling 1 2 3 4 5 6 ¯ \illing
59
Combination JS
Lxpertise Low
1rustworthiness Low
Similarity High
Likeability High
Meaning transfer High
´Iit¨ Match Low
Based on the attributes mentioned abo·e, how willing are you to buy this product·
,Please rate this combination,
Unwilling 1 2 3 4 5 6 ¯ \illing
60
Combination J6
Lxpertise High
1rustworthiness Low
Similarity High
Likeability Low
Meaning transfer Low
´Iit¨ Match High
Based on the attributes mentioned abo·e, how willing are you to buy this product·
,Please rate this combination,
Unwilling 1 2 3 4 5 6 ¯ \illing
61
How important do you consider each of the attributes mentioned to be of
importance when considering a celebrity that gets negative publicity.
,Please e·aluate each attribute on the scale oí 1 to 5,
Lxpertise
Unimportant 1 2 3 4 5 Important
1rustworthiness
Unimportant 1 2 3 4 5 Important
Similarity
Unimportant 1 2 3 4 5 Important
Likeability
Unimportant 1 2 3 4 5 Important
Meaning transfer
Unimportant 1 2 3 4 5 Important
´Iit¨ Match
Unimportant 1 2 3 4 5 Important
62
,Please put a cross on the situation that is addressed to each question,
J. Which faculity do you belong to?
2. Do you consider yourself to be fashion conscious?
3. Do you consider yourself to be affected by celebrities in advertising?
4. Do you purchase brands that are advertised by celebrities?
S. Does your purchasing decisions change if a celebrity get assocaiated with
negative publicity?
JIBS ING lLK llJ
\es No Occassionally
\es No Occassionally
\es No Occassionally
\es No
63
Appendix 3- Extract from pilot study
Combination J
Lxpertise Low
1rustworthiness Low
Similarity Low
Likeability Low
Meaning transfer Low
´Iit¨ Match Low
Based on the attributes mentioned above, how important do you find the celebrity's
negative publicity affecting your purchasing decision.
,Please rate this combination,
Low importance 1 2 3 4 5 6 ¯ ligh
importance
64
Appendix 4 - Appendix concerning reserach question 1
Appendix 4A
Importance summary
Factor
Match
meaning transf er
likeability
similarity
trsutworthiness
expertise
A
v
e
r
a
g
e
d
I
m
p
o
r
t
a
n
c
e
30
20
10
0
Appendix 4B
Individual Subject Importance
Factor
Match
meaning transf er
likeability
similarity
trsutworthiness
expertise
A
v
e
r
a
g
e
d
I
m
p
o
r
t
a
n
c
e
60
50
40
30
20
10
0
65
66
67
68
69
70
71
72
73
74
75
65
Appendix 4C
CARD12
7,0 6,0 5,0 4,0 3,0 2,0 1,0
CARD12
F
r
e
q
u
e
n
c
y
30
20
10
0
Std. Dev = 1,45
Mean = 4,9
N = 75,00
66
Appendix 5- Appendix concerning reserach question 2
case 1
Appendix SA
Importance summary
Factor
Match
meaning transf er
likeability
similarity
trsutworthiness
expertise
A
v
e
r
a
g
e
d
I
m
p
o
r
t
a
n
c
e
30
20
10
0
Appendix SB
Individual Subject Importance
Factor
Match
meaning transf er
likeability
similarity
trsutworthiness
expertise
A
v
e
r
a
g
e
d
I
m
p
o
r
t
a
n
c
e
80
60
40
20
0
65
66
67
68
69
70
71
72
73
74
75
6¯
Appendix 6- Appendix concerning reserach question 2
case 2
Appendix 6A
Importance summary
Factor
Match
meaning transf er
likeability
similarity
trsutworthiness
expertise
A
v
e
r
a
g
e
d
I
m
p
o
r
t
a
n
c
e
30
20
10
0
Appendix 6B
Individual Subject Importance
Factor
Match
meaning transf er
likeability
similarity
trsutworthiness
expertise
A
v
e
r
a
g
e
d
I
m
p
o
r
t
a
n
c
e
80
70
60
50
40
30
20
10
0
65
66
67
68
69
70
71
72
73
74
75
68
Appendix 7- Anova tables
Reserach question J
Model Summary(b) Appendix 7A
Model R R-square
Adjusted R-
square
Std. Error of
the Estimate
1
,562(a) ,316 ,128 20,357
a Predictors: (Constant), CARD16, CARD10, CARD7, CARD3, CARD12, CARD11, CARD1, CARD14,
CARD13, CARD5, CARD8, CARD2, CARD6, CARD4, CARD15, CARD9
b Dependent Variable: RESP
ANOVA(b) Appendix 7B
Model
Sum of
Squares df Mean Square F Sig.
Regressio
n
11114,608 16 694,663 1,676 ,078(a)
Residual
24035,392 58 414,403
1
Total
35150,000 74
a Predictors: (Constant), CARD16, CARD10, CARD7, CARD3, CARD12, CARD11, CARD1, CARD14,
CARD13, CARD5, CARD8, CARD2, CARD6, CARD4, CARD15, CARD9
b Dependent Variable: RESP
Research question 2 case J
Model Summary Appendix 7C
Model R R-square
Adjusted R-
square
Std. Error of
the Estimate
1
,430(a) ,185 ,086 20,832
a Predictors: (Constant), CARD8, CARD7, CARD5, CARD4, CARD2, CARD1, CARD3, CARD6
ANOVA(b) Appendix 7D
Model
Sum of
Squares df Mean Square F Sig.
Regressio
n
6507,286 8 813,411 1,874 ,079(a)
Residual
28642,714 66 433,981
1
Total
35150,000 74
a Predictors: (Constant), CARD8, CARD7, CARD5, CARD4, CARD2, CARD1, CARD3, CARD6
b Dependent Variable: RESP
Research question 2 case 2
69
Model Summary Appendix 7E
Model R R-square
Adjusted R-
square
Std. Error of
the Estimate
1
,506(a) ,256 ,166 19,909
a Predictors: (Constant), CARD16, CARD10, CARD12, CARD14, CARD15, CARD11, CARD13, CARD9
ANOVA(b) Appendix 7F
Model
Sum of
Squares df Mean Square F Sig.
Regressio
n
8988,430 8 1123,554 2,834 ,009(a)
Residual
26161,570 66 396,387
1
Total
35150,000 74
a Predictors: (Constant), CARD16, CARD10, CARD12, CARD14, CARD15, CARD11, CARD13, CARD9
b Dependent Variable: RESP
¯0
Appendix 8- Perception of attributes
Rating
5,0 4,0 3,0 2,0 1,0
Expertise
F
r
e
q
u
e
n
c
y
40
30
20
10
0
Std. Dev = 1,25
Mean = 4,0
N = 75,00
Rating
5,0 4,0 3,0 2,0 1,0
Trustworthiness
F
r
e
q
u
e
n
c
y
50
40
30
20
10
0
Std. Dev = ,93
Mean = 4,3
N = 75,00
¯1
Rating
5,0 4,0 3,0 2,0 1,0
Similarity
F
r
e
q
u
e
n
c
y
30
20
10
0
Std. Dev = 1,10
Mean = 3,4
N = 75,00
Rating
5,0 4,0 3,0 2,0 1,0
Likeability
F
r
e
q
u
e
n
c
y
30
20
10
0
Std. Dev = 1,06
Mean = 3,0
N = 75,00
¯2
Rating
5,0 4,0 3,0 2,0 1,0
Meaning transfer
F
r
e
q
u
e
n
c
y
30
20
10
0
Std. Dev = 1,24
Mean = 3,4
N = 75,00
Rating
5,0 4,0 3,0 2,0 1,0
"Fit" Match
F
r
e
q
u
e
n
c
y
30
20
10
0
Std. Dev = 1,10
Mean = 3,6
N = 75,00
doc_800796532.pdf
Celebrity endorsements have proven very successful around the world where, due to increasing consumerism, an individual is considered a status symbol when they purchase a celebrity-endorsed product.
I NT E R NAT I ONE L L A HANDE L S HÖGS KOL AN
HÖGSKOLAN I JÖNKÖPING
Cel ebri t y Endorsement -
Hidden factors to success
Master`s thesis within Business Administration
Author: Chabo Dimed
Saouma Joulyana
1utor: Gusta·sson Veronica
Co-examinator: Da·idsson Per
Jonkoping: May 2005
J ÖNKÖP I NG I NT E R NAT I ONAL BUS I NE S S S CHOOL
Jonkoping Uni·ersity
Cel ebri t y Endorsement -
Hidden factors to success
Master`s thesis within Business Administration
Author: Chabo Dimed
Saouma Joulyana
1utor: Gusta·sson Veronica
Co-examinator: Da·idsson Per
Jonkoping May 2005
i
Master Thesis in Business administration
Title: Celebrity Endorsement
Authors: Chabo Dimed, Saouma Joulyana
Tutor: Gustavsson Veronica
Co-examinator: Davidson Per
Date: May 27
th
200S
Subject terms: Celebrity endorsement, Source credibility, Source
attractiveness, Negative publicity, Conjoint Analysis
Abstract
1he use oí celebrity endorsement strategy is nowadays more írequently used by
marketers in order to increase their sales and thereby extend their market shares. Many
celebrities are used in ·arious marketing campaigns and in most cases, the use oí
celebrities as endorsers is seen írom mainly positi·e aspects. 1his made the authors
curious whether the negati·e aspects, that also exists when using celebrities as endorsers,
aííects consumers in their purchasing decisions when a celebrity gets associated with
negati·e publicity. Another cause oí interest is which íactors oí a certain celebrity are
most important and crucial in consumers` perceptions, in the case oí negati·e publicity.
Purpose: 1he purpose oí this thesis is to study which íactors consumers íind important
íor a company to consider when a celebrity gets negati·e publicity, to maintain
successíul brand recognition.
Literature review: 1he use oí pre·ious studies within the íield oí celebrity
endorsement clariíies many important aspects when it comes to celebrity endorsement
and this chapter is elaborated írom 4 diííerent perspecti·es, Company, Celebrity, Brand
and Consumer. Based on pre·ious studies, the authors identiíied 6 crucial attributes
when using celebrities as endorsers and this can also be seen as a pre-study that the
research process has been based upon. lurthermore, the 6 attributes are chosen írom
the three íirst mentioned perspecti·es in order to be able to íulíil the purpose. lence,
this thesis is conducted írom a consumer`s point oí ·iew.
Method: A quantitati·e method is used in this thesis since the authors want to base the
results on collected data that is expressed in numbers and also to generate a general
apprehension in this phenomenon. Moreo·er, the combinations containing the 6
attributes are used in the conjoint experiment.
Conclusions: It was pro·en in this study that consumers do get aííected by celebrities
as endorser, when the attributes írom the literature re·iew are in a combination. But, the
consumers` perception oí the attributes diííers in diííerent cases. lowe·er, the main
íinding was that there are two crucial attributes, trustworthiness and expertise that
companies should take into account when using celebrities in their ad·ertising
campaign.
ii
Table of Contents
1 Introduction............................................................................ 1
1.1 The emergence of celebrity endorsement ..................................... 1
1.2 The two-sided effect of celebrity endorsement .............................. 2
1.3 Limitations ..................................................................................... 4
1.4 Purpose......................................................................................... 4
1.5 Research questions....................................................................... 4
1.6 Outline of this study....................................................................... 4
2 Literature review.................................................................... 6
2.1 Introduction ................................................................................... 6
2.2 Celebrity endorsement strategy .................................................... 7
2.3 Company....................................................................................... 7
2.3.1 Choose and use celebrities as endorsers........................... 8
2.4 Celebrity........................................................................................ 9
2.4.1 Source credibility ................................................................ 9
2.4.2 Source attractiveness ....................................................... 10
2.4.3 Source Power ................................................................... 11
2.5 Brand........................................................................................... 11
2.5.1 Brand equity ..................................................................... 11
2.5.2 Meaning transfer............................................................... 12
2.5.3 Multiple brand endorsement ............................................. 13
2.6 Consumer.................................................................................... 14
2.6.1 Consumer behaviour and negative publicity ..................... 14
2.7 Chapter summary........................................................................ 15
2.7.1 Hypotheses....................................................................... 16
3 Methodology ........................................................................ 18
3.1 Introduction ................................................................................. 18
3.2 Sampling ..................................................................................... 18
3.3 Conjoint Analysis......................................................................... 20
3.3.1 Attributes and attribute level ............................................. 21
3.3.2 Different approaches of conjoint analysis ......................... 22
3.3.3 Experimental design ......................................................... 22
3.3.4 Pilot study ......................................................................... 23
3.3.5 Data collection .................................................................. 23
3.3.6 The survey........................................................................ 24
3.4 Statistics...................................................................................... 24
4 Analysis and results............................................................ 26
4.1 Introduction ................................................................................. 26
4.2 Research question 1 ................................................................... 26
4.2.1 Testing the hypothesis...................................................... 27
4.2.2 Results research question 1 ............................................. 27
4.3 Research question 2 ................................................................... 28
4.3.1 Case 1 .............................................................................. 28
4.3.2 Case 2 .............................................................................. 28
4.3.3 Testing the hypothesis...................................................... 29
4.3.4 Results research question 2 ............................................. 29
iii
4.4 Research question 3 ................................................................... 29
4.4.1 Perceived importance of attributes ................................... 30
4.4.2 Results research question 3 ............................................. 30
4.5 Chapter summary........................................................................ 31
5 Conclusions and discussion.............................................. 33
5.1 Introduction ................................................................................. 33
5.2 Research questions..................................................................... 33
5.3 Discussion................................................................................... 34
5.4 Final words from the authors....................................................... 35
5.5 Prospects for future researches .................................................. 35
6 Critique of this research process....................................... 37
References................................................................................. 38
Figures
Figure 1-1: The Optimal use of celebrity endorsement (Chabo & Saouma,
2005) ................................................................................................. 3
Figure 1-2: The effect of negative information (Chabo & Saouma, 2005) ...... 3
Figure 2-1: The structure of the literature review............................................ 6
Figure 2-2: A triangular relationship (Pringle, 2004)....................................... 8
Figure 2-3: Meaning transfer in the endorsement process (Schlecht, 2003) 13
Figure 1-1: The network of optimal use of a celebrity (Chabo & Saouma,
2005) ............................................................................................... 35
Tabels
Table 3-1: Numbers of Students at Jönköping University ............................ 20
Table 3-2: Attributes and attribute level of the conjoint experiment.............. 21
Table 4-1: Outline of research process ........................................................ 26
Table 4-2: Ranking the attributes ................................................................. 30
Appendices
Appendix 1- 8 profiles for the six attributes at two levels.............................. 42
Appendix 2 - Research Survey.................................................................... 43
Appendix 3- Extract from pilot study............................................................. 63
Appendix 4 - Appendix concerning reserach question 1 ............................. 64
Appendix 5- Appendix concerning reserach question 2 case 1................... 66
Appendix 6- Appendix concerning reserach question 2 case 2................... 67
Appendix 7- Anova tables ........................................................................... 68
Appendix 8- Perception of attributes ........................................................... 70
Introduction
1
1 Introduction
1be avtbor. ritt ¡re.evt a geverat vvaer.tavaivg of tbe to¡ic, cetebrit, evaor.evevt, iv tbe fir.t cba¡ter.
ívrtbervore, tbe avtbor. ritt at.o ¡re.evt tbe ¡robtev area, rbicb tbi. .tva, ritt be ba.ea ov, fottorea b,
tbe ¡vr¡o.e ava tbe ovttive of tbe .tva,.
1.1 The emergence of celebrity endorsement
1he use oí ad·ertising has changed o·er the past 150 years, írom the classical to the
modern school. In the modern ad·ertising strategies ·arious appeals are included, such as
sexual, chock, emotional, íear, and humour ,Se·ern, Belch & Belch, 1990, Belch and Belch,
2001, Soderlund, 2003,. 1he main purpose oí these appeals is to deli·er the iníormation
that the company seeks to send to gain high brand awareness and brand recognition among
a large audience. lowe·er, when using any oí these appeals there is always a person
included, sometimes someone unknown or in most cases a well known person in the public
eye. According to McCracken ,1989,, a well-known person tends to ha·e a greater eííect
on the consumer buying beha·iour. 1his is especially ií people more easily can identiíy
themsel·es with this particular person ,Lrdogan, Baker, & 1agg, 2001,. McCracken ,1989,
also states that celebrity endorsement ad·ertising has been recognized as a ubiquitous
íeature oí modern marketing.
1he concept itselí, celebrity endorsement, is recognized by marketers because it has an
eííect that iníluences the message ,brand, the company is trying to send írom someone
that the consumers íeel a sense oí similarity with. Consumers tend to e·aluate iníormation
írom a communicator ,celebrity,, which they ha·e similar a goal, interest or liíestyle with
greater than someone they do not ,Lrdogan et al, 2001,. lurthermore, celebrities ha·e an
ability to transíer their image to a speciíic product that is being ad·ertised ,\heeler, 2003,.
Companies ha·e taken this opportunity into account and tries, through using it as an
ad·ertising tool, to gain a high brand exposure, attention, interest, desire and action ,Belch
& Belch, 2001,.
Celebrities are people who enjoy public recognition by a large share oí a certain group oí
people and they ha·e distincti·e characteristics, such as attracti·eness and trustworthiness
,Sil·era & Austad, 2004,. Many companies ha·e realized the importance oí celebrity
endorsement as a marketing communication tool ,Belch & Belch, 2001, Soderlund, 2003,.
In using this ad·ertising tool, a company will in the short run generate a high awareness
among a larger audience as well as an increased market share. Since this ad·ertising tool has
gained such attention írom ·arious companies, the price le·el is extremely high
,Danesh·ary & Schwer, 2000, Kambitsis, larahousou, 1heodorakis, & Chatzibeis, 2002,.
As billions oí dollars are spent per year on celebrity endorsement contracts, which indicates
that celebrities play an enormous and important role íor the ad·ertising industry. A recent
example oí this is the Super Bowl in the United States oí America, where thirty seconds oí
ad·ertising cost approximately se·enteen million SLK. Pepsi used many diííerent
celebrities, such as Cindy Crawíord, P Diddy and Carson Kressley to promote diet Pepsi.
leineken, on the other hand, used Brad Pitt in their endorsement strategy, where he
earned around three to íi·e million USD íor his eííort ,Lundell, 2005,.
1his shows that celebrity endorsement is a ·ery popular ad·ertising tool, due to the prices
companies are ready to pay and the eííect it has on consumers buying beha·iour ,Belch &
Belch, 2001, McCracken, 1989, Soderlund, 2003,. 1hereíore, a disputable issue rises
Introduction
2
upon which attributes and personal characteristics oí a celebrity a company base their
choice. lurthermore, Sil·era & Austad ,2004, state that celebrities are eííecti·e endorsers
because they are ·iewed as highly belie·able, likeable, pursuable, and trustworthy.
lowe·er, celebrity endorsement might ·ary when it comes to the íit between the celebrity
and the ad·ertised product, as well as the le·el oí eííecti·eness on the purchasing
beha·iour ,1ill & Shimp, 1998,. 1his entails that the use oí celebrity endorsement might
ha·e a risk when there is a lack oí íit or when the celebrity gets associated with negati·e
publicity ,Soderlund, 2003,. Negati·e publicity occurs when negati·e iníormation is spread
and associated with the celebrity. lence, negati·e iníormation can be direct, indirect
and,or percei·ed subjecti·ely.
1.2 The two-sided effect of celebrity endorsement
1here are not many researchers who ha·e embarked on the concept oí negati·e
iníormation, which celebrities` causes íor consumers` e·aluations oí endorsed brands ,1ill
& Shimp, 1998,, but still there are many diííerent aspects on how negati·e iníormation can
be ·iewed. lurthermore, Soderlund ,2003, argues that celebrities can be percei·ed as less
trustworthy when they are ad·ertising too many brands ,multiple brand endorsement, since
the consumers get di·erse iníormation. On one hand, a celebrity who is linked with
positi·e iníormation has a greater opportunity to reach out to the targeted audience. On
the other hand, negati·e iníormation brings an association that decreases the
trustworthiness oí a celebrity and the íit with the brand, which is also the authors`
deíinition oí negati·e iníormation.
One example where the íit between the celebrity and the brand has been ·ery successíul is
Michael Jordan and Nike. 1he reason that the authors assume the períect íit with the brand
is based on the íact that Nike captured their endorser to a higher le·el where they
de·eloped a new brand within Nike called Air Jordan. One can assume that Michael Jordan
has not shown any indications oí being re·olutionary in the media. Basically, one can say
that Michael Jordan has been a ·ery good endorser íor this brand, in the sense oí his high
expertise and him being trustworthy, where a large audience íeels a sense oí similarity with.
Now twenty years later, they still use him in their ad·ertising, which strengthens the
assumption oí the good íit between Michael and Nike ,Belch & Belch, 2001,.
la·ing this iníormation in mind the authors decided to create a model to íacilitate the
understanding oí how the use oí a celebrity aííects the company. ligure 1-1 shows, when
the iníormation the celebrity is sending out is optimal, and there are no obstacles between
the celebrity, brand, consumers, and company. As one can see, when looking at the model,
the company, celebrity, brand and consumer are interconnected as a continuous network.
In this optimal model, there are no particular obstacles that harm the celebrity
endorsement process. 1his implies that the íit between the celebrity and brand are
integrated.
Introduction
3
Soderlund ,2003, argues that celebrities are not proíitable to be used in a long-run
perspecti·e, ií they in any matters are not maintaining rele·ance with the consumers.
1his is an issue íor companies,
because consumers are easily
adapting to another celebrity,
sometimes because one cannot
íully predict the liíe-span oí a
celebrity ,1ill & Busler, 1998,.
1his can aííect the brand image
and also the le·el oí loyalty a
consumer has towards a brand.
lurthermore, Soderlund ,2003,
discusses that demographical
changes oí a celebrity cannot be
seen as anything negati·e
towards the brand.
lowe·er, e·eryone cannot be as
períect as Michael Jordan. 1here
are always some celebrities that get negati·e publicity, sometimes what media creates and
sometimes their own personal actions. 1his indicates that companies ha·e to be more
careíul when selecting a celebrity who will endorse one oí their brands. 1he authors ha·e in
íigure 1-2 chosen to demonstrate how negati·e publicity oí a celebrity aííects the company,
brand, and consumer. 1his íigure tells us, that when a celebrity is associated with negati·e
publicity, the celebrity is only aííecting each part separately and that there is no connection
between the company, brand and consumer. lurthermore, there is no ílow in the network
oí connections which also aííects the lack oí íit between the celebrity and brand. 1he
questions that arise are, how are these íour parts ,company, celebrity, consumer and brand,
connected, when a
celebrity is associated
with negati·e
publicity· low can a
company o·ercome
these obstacles in the
long run· \ould the
use oí a celebrity who
gets negati·e publicity
destroy the brand
image and the le·el oí
loyalty consumers íeel towards a brand· Does the íit between a celebrity and company
aííect the consumers when a celebrity gets associated with negati·e iníormation· Does the
meaning, a celebrity is transíerring, aííect the consumer`s perception oí the brand, ií a
celebrity is associated with negati·e iníormation· 1he authors` belie·e that írom the
perspecti·e oí this thesis, the eííect oí negati·e iníormation on a celebrity is the most
interesting aspect to study, as it is not limited to certain exceptional elements but can make
use oí a ·ariety oí elements like source attracti·eness, source credibility, and source power.
Brand Company Consumer
Celebrity
ligure 1-2: 1he eííect oí negati·e iníormation ,Chabo &
Saouma, 2005,.
Company
Brand
Consumer
Celebrity
ligure 1-1: 1he Optimal use oí celebrity
endorsement ,Chabo & Saouma, 2005,.
Introduction
4
1.3 Limitations
According to íigure 1-2, the authors discuss the eííect oí negati·e iníormation and how the
celebrity aííects the company, consumers and brand. lowe·er, the authors belie·e that the
consumers are the end group oí the network because they are the ones that the company is
trying to reach through their brand and the ad·ertising tool, celebrity endorsement.
1hereíore, the authors will base their study on a consumer perspecti·e.
1.4 Purpose
1he purpose oí this thesis is to study which íactors consumers íind important íor a
company to consider when a celebrity gets negati·e publicity, to maintain successíul brand
recognition.
1.5 Research questions
1his thesis is intended to in·estigate whether the attributes that ha·e been conducted
through the literature re·iew ,chapter 2, really aííects consumer purchase decisions.
lurthermore, the authors are ·ery interested to study ií the consumers e·aluate the
attributes diííerently, when diííerent endorsers, who get negati·e publicity, are ad·ertising a
product or ser·ice. Lastly, the authors want to see how the consumers percei·ed the
attributes separately. 1he reason in doing this is to compare the results írom the conjoint
and the sur·ey to show which attributes are percei·ed as the most important ones. 1hus,
this thesis addresses more precisely the íollowing research questions.
1. \hich íactors írom the literature re·iew aííect the consumers` willingness to
purchase a product or ser·ice that is endorsed by a celebrity who is associated with
negati·e iníormation, when the attributes are in a combination·
2. Do the attributes írom the literature re·iew ha·e diííerent impact on consumers`
willingness to purchase a product or ser·ice, when diííerent celebrities are used as
endorsers·
3. \hich attributes do consumers íind to be oí most importance when e·aluating
them separately·
1.6 Outline of this study
Chapter J: 1he introduction chapter consists oí an explanation concerning the importance
oí celebrity endorsement in the ad·ertising industry. 1he authors will íurther on explain the
two-sided eííect oí celebrity endorsement, positi·e and negati·e. Also, the purpose oí this
thesis will be addressed íollowed by the research questions.
Chapter 2: 1he literature re·iew is di·ided into íour diííerent phases, Company, Celebrity,
Brand and Consumer. 1hese parts are all related to celebrity endorsement strategy with its
main purpose to íind important attributes that will be studied íurther on in this thesis.
Lastly, this will lead to a conclusion oí this chapter and the hypotheses.
Chapter 3: 1he methodology chapter explains how the authors will approach this research
process in order to íulíil the purpose. 1his will be done by using a quantitati·e method and
more precisely a conjoint experiment.
Introduction
5
Chapter 4: In the analysis and results chapter, the main íindings will be presented íollowed
by an analysis oí the most crucial aspects within this subject. Lastly, the authors will present
a chapter summary and conclude the most important attributes.
Chapter S: In this chapter, the authors will answer the research questions and draw the
main conclusions íollowed by the íinal words írom the authors. linally, the authors will
present suggestions íor íuture researches.
Chapter 6: 1he íinal chapter will pro·ide the reader with critique oí this study.
Literature re·iew
6
2 Literature review
1bi. cba¡ter .ee/. to fiva tbe vo.t crvciat attribvte. tbat i. vo.tt, a..ociatea ritb cetebrit, evaor.evevt
.trateg,. 1bi. iv tvrv to a¡¡t, tbe fottorivg attribvte. iv tbe vetboaotog, ¡art of tbi. .tva,.
2.1 Introduction
\hen a celebrity gets associated with negati·e iníormation a company has to consider
·arious actions to be able to maintain a good position on the market and a similar le·el oí
brand recognition as well. 1his indicates that a company needs to de·elop a strategy that
pre·ents upcoming obstacles ,1ill & Shimp, 1998,, ií a celebrity that they are using as an
endorser gets associated with negati·e iníormation. \ithin this írame, which is also the
main body oí this thesis, the authors will di·ide it into íour diííerent phases that are directly
related to each other. lence, these are the main pillars. 1he purpose oí this chapter is to
identiíy crucial attributes that will íurther on be used in the research process. 1o íacilitate
íor the reader, the authors ha·e chosen to explain the structure oí this chapter in the íigure
below.
ligure 2-1: 1he structure oí the literature re·iew
As one can see írom the íigure 2-1, this chapter is outlined with the basis oí celebrity
endorsement strategy which is also the main íocus when elaborating later, concerning the
íour diííerent phases. 1hese parts will be intergrated and brought together as a summary oí
the most important issues, which will also enhance the process oí choosing rele·ant
attributes íor the experimental design ,see section 3.3.3,.
Company
Phase 1
Celebrity
Phase 2
Brand
Phase 3
Consumer
Phase 4
Celebrity endorsement strategy
Conclusions - identiíication oí attributes
Literature re·iew
¯
2.2 Celebrity endorsement strategy
Marketers usually use indi·iduals who ha·e achie·ed some íorm oí celebrity status to ser·e
as spokespersons íor their companies. Most oí the celebrities that are hired by a company
to pitch their products or ser·ices are popular people, mo·ie stars, entertainers, athletics, or
pop-stars, although occasionally a politician or some other well-known public íigure may
be used ,Belch & Belch, 2001,. lurthermore, when a company decides upon using an
endorsement strategy as their marketing communication tool, the main íocus lies in
exposing their brand ,Kotler, Armstrong, Saunders, & \ong, 2001,. In an endorsement
strategy a new sort oí product is gi·en a new brand name that is unique íor that product
,Riezebos, 2003,. Besides the unique brand name, they also get pro·ided with the name oí
an endorser. In this case, the endorser is a celebrity and íunctions as endorsement which
means an appro·al or support that can be seen as a guarantee oí recommendation íor
consumers. According to Riezebos ,2003, it is only ad·isable to use endorsers íor brands ií
there is a high le·el oí brand-added ·alue. 1his means that the name oí the endorser should
be clearly ·isible next to the name oí the branded article.
Companies ha·e jointly been using their brands and themsel·es, through the use oí
celebrity endorsers, in hope that celebrities may boost eííecti·eness oí their marketing
attempts in the long-term ,Belch & Belch, 2001,. Basically, a company is trying to send
·arious types oí iníormation to their target audience. 1o be able to de·elop an eííecti·e
ad·ertising and promotional campaign, a company has to select their endorser appropriate
to diííerent channels and media ,1ill & Shimp, 1998,, such as source, message, and recei·er
,Belch & Belch, 2001,. 1hus, the brand can be seen as the message the company is trying
to send to their audience. Moreo·er, the source which is intended to send this message in
an endorsement strategy is in this case, the celebrity. lurthermore, the recei·er in the
communication process is the consumer ,Belch & Belch, 2001,.
1he authors ha·e chosen to present each phase oí this thesis separately. 1he reason in
doing this is to bring the reader an understanding oí the eííect celebrity endorsement
strategy has on each phase.
2.3 Company
According to Lrdogan & Baker ,2004,, there are se·eral reasons why companies choose to
use celebrity endorsement to a larger extent. Managers seek to reíresh the brand image,
awareness and attention getting and also to add new dimensions to the brand image.
According to Pringle ,2004,, the best identiíied celebrity that is seen as the best íor
promoting a brand, is one oí the more important decisions considering how consumers
will percei·e the brand. 1here are less important decisions íor a brand compared to the
choice oí celebrity such as what it is named, places where it is seen and sold and what kind
oí ad·ertising campaign it runs. 1hese considerations are taken into account and this is also
the reason why many companies are ready to in·est a huge amount oí money in choosing
and using a certain celebrity, whose identity íits well with their brand. lowe·er, it is not
only the in·ol·ement between the celebrity and the brand, the consumers are also
integrated. Pringle ,2004, has outlined this relationship and it is shown in íigure 2-2.
Literature re·iew
8
ligure 2-2: A triangular relationship ,Pringle, 2004,
1hese three íactors are important íor a company to take into account in order to a·oid any
obstacles when it comes to the triangular relationship. lence, another important area is
that the company only can regulate these three to a certain extent, íor the process oí
celebrity endorsement to be successíul.
Uníortunately, when looking at pre·ious experiences írom the use oí celebrity
endorsement, there are many examples that celebrities might change their beha·iour, ·iews
and their percei·ed personality rather drastically ,Soderlund, 2003,. Certainly, consumers
might change their perception oí the celebrities since the latter mentioned ha·e changed.
lowe·er, the worst outcome is that a celebrity is certainly no longer the indi·idual that the
company chose to promote their brand in the íirst place. 1his leads to a decrease in the
connection between the celebrity and brand ,Belch & Belch, 2001,.
2.3.1 Choose and use celebrities as endorsers
According to Lrdogan et al. ,2001, managers choose celebrity endorsers depending on the
product type and how that íits with the characteristics oí a celebrity. 1his will, according to
Louie & Obermiller ,2002,, lead to a more eííicient ad·ertising.
Celebrities can be chosen by companies to increase their competiti·e ad·antage íor three
main reasons, launch, reiníorcement and repositioning ,Pringle, 2004,.
• 1he íirst opportunity is when launching the brand íor the íirst time and the use oí
celebrity can be ·ery poweríul in this situation. 1his can be particularly strong when
establishing a new category and the consumers need reassurance and an appropriate
star will pro·ide them this.
• Celebrity endorsement can be ·ery eííecti·e when maintaining and reiníorcing a
brand`s competiti·e position in the market. 1he use oí celebrities in this situation
can be especially eííecti·e ií other brands ha·e entered the marketplace and thereby
changed its dynamics.
• \hen the consumers` needs create greater potential in a diííerent sector in the
market than the one which the brand is currently positioned, repositioning with the
help oí a celebrity will be in order. 1he appropriate celebrity can be used as the
íocus oí the brand communication and this shows consumers that it`s positioning is
changing in order to suit an emerging target audience.
In a research done by Bielli ,2003, it is shown that 18° oí all the tested commercials
íeature íamous celebrities. In o·erall, this study shows that celebrities are used to grab
attention, generate interest and in·ol·ement in the brand. Also, the right` celebrity can
add ·alues by associations. A suggestion made by this research is that the celebrity should
not o·ershadow the brand and instead be beneíicial ií the celebrity is used íor what the
My personality
Celebrity personality Brand personality
Literature re·iew
9
brand is already íamous íor. Instead, the brand should borrow and build on the celebrity. A
conclusion made by Bielli ,2003, is that a celebrity with a generally likeable personality is
more likely to make íor success. Moreo·er, ad·ertisers and marketers need to ask
themsel·es íour questions beíore engaging too deep in a celebrity. low íamous is the
celebrity· low well does the celebrity íit with the brand· \hich íacets oí this celebrity can
best work íor the brand proíile· low much oí this can the brand íinance· Many studies
show that the deeper íit between the celebrity and the brand it is the more likely to be
eííecti·e in the marketplace. 1hereíore, the company should stri·e to create a close
connection as possible between the celebrity and the brand ,Pringle, 2004,.
2.4 Celebrity
A celebrity can be considered as the source oí the message a company seeks to send to
their target audience. According to Belch & Belch ,2001,, the term source, when talking
about the in·ol·ement in communicating a marketing message, can occur either directly or
indirectly. Directly can be the celebrity who íunctions as a spokesperson and who sends
out the iníormation that the company wants to deli·er to their target audience. Indirectly is
when a celebrity does not send the message, but instead draw attention to and,or enhances
the appearance oí the ad ,Belch & Belch, 2001,. lowe·er, as mentioned in phase one, a
company ha·e to careíully select a celebrity that has a good íit with the brand, which is
intended to be exposed ,Pringle, 2004,.
\hen a company decides on the use oí a celebrity in their endorsement strategy, there are
three ·ery important source íactors, source credibility, source attracti·eness, and source
power ,Belch & Belch, 2001, 1ill & Shimp, 1998, Ohanian, 1990,. 1hese will now be
presented.
2.4.1 Source credibility
Credibility is the extent to which the recei·er sees the source as ha·ing rele·ant knowledge,
skills, experience and trust to gi·e unbiased and objecti·e iníormation. Source credibility is
used to imply a communicator`s positi·e characteristics that will aííect the recei·er`s
acceptance oí a message ,Ohanian, 1990,. Basically, one can say that a communicator,
celebrity, can be seen as knowledgeable and a person with expertise. lurthermore, the
source needs to be trustworthy, in the sense oí honesty, ethics, and belie·ability ,Belch &
Belch, 2001,. 1hese two attributes, that a celebrity needs to ha·e to be a successíul
endorser in an ad·ertising campaign, are presented more in-depth below.
• í·¡erti.e: Belch & Belch ,2001, discusses that a spokesperson are oíten chosen due
to their knowledge, experience, and expertise in a particular product or ser·ice.
lurthermore, Ohanian ,1990, states that the percei·ed expertise oí celebrity
endorsers is more important in explaining purchase intentions rather than their
attracti·eness and trustworthiness. She also argued that celebrity endorsers are
more eííecti·e when they are knowledgeable, experienced, and qualiíied to talk
about the product they are endorsing. On the whole, .ovrce e·¡erti.e in persuasi·e
communication, indicates generally that the source`s percei·ed expertise has a
positi·e impact on attitude change.
• 1rv.trortbive..: In comparison to expertise, a celebrity needs to be trustworthy when
endorsing a product or a ser·ice ,Schiííman & Kanuk, 2004,. 1his is logically based
on how honest the celebrity is about what he,she says concerning the brand.
lurthermore, Belch & Belch ,2001, discusses that the target audience must íind the
Literature re·iew
10
source ,celebrity, belie·able. Moreo·er, Ohanian ,1990, states that when a celebrity
is percei·ed more trustworthy, the message will be more eííecti·e and the recie·er
will be more integrated. lence, trustworthiness is the degree oí coníidence in the
communicator`s intention to communicate the assertions he,she considers being
the most ·alid ,Ohanian, 1990,.
Belch & Belch ,2001, argues that when the iníormation írom a credible source iníluences
the belieís, opinions, and attitudes oí the recei·er, the latter mentioned adopts the opinion
oí the credible communicator. 1his is based on the assumption that the iníormation írom
the source is accurate. Ií the celebrity achie·es to integrate the recei·er with the
iníormation that he,she meant to send, the company will in the long run gain a loyal
consumer, in the sense that the consumer is more integrated with the brand and not to a
high scope with the celebrity.
1he authors belie·e that using these attributes, expertise and trustworthiness, will be
accurate in the research process. 1he authors belie·e it will pro·ide the thesis with ·aluable
results. lurthermore, it will also be considered as the underlying íactor in the process oí
íinding suitable attributes which consumers íind important íor a company to consider
when a celebrity gets associated with negati·e iníormation.
2.4.2 Source attractiveness
Source attracti·eness is more related to physical attributes, such as similarity, íamiliarity,
and likeability. 1hese are important in the indi·idual`s initial judgment oí another person
,Ohanian, 1990,. Similarity is a supposed resemblance between the source and the recei·er
oí the message, while íamiliarity reíers to the knowledge oí the source through exposure.
Liability is aííection íor the source as a result oí physical appearance, beha·iour or other
personal traits ,Belch & Belch, 2001,. Source attracti·eness aííects the recei·er in the sense
oí that they are identiíying themsel·es with the celebrity. 1his does also moti·ate the
recei·er to seek some type oí relationship with the source and thus adopts similar belieís,
attitudes, preíerences, or beha·iour. lurthermore, ií a celebrity changes position, the
recei·er might íollow and adapt to these changes ,Belch & Belch, 2001, Ohanian, 1990,. Ií
this happens, the celebrity has íulíilled its íunction. lowe·er, not many celebrities achie·e
this, but as mentioned beíore, the main purpose with source attracti·eness is to make a
consumer íeel a part oí the celebrity as well as the company and brand.
1he three attributes similarity, íamiliarity, and likeability are more elaborated below.
• ´ivitarit,: \hen talking about similarity, Belch & Belch ,2001, mentions that the
consumers are more easily iníluenced by a message coming írom someone with
whom they íeel a sense oí similarity with. lurthermore, Ohanian ,1990, elaborated
on the íact that similarity can be measured ií the communicator and recei·er ha·e
similar needs, goals, interest and liíestyle.
• íavitiarit,: According to Belch & Belch ,2001,, íamiliarity can be considered as the
le·el oí knowledge a celebrity possesses oí a brand. \hen a company considers
choosing a celebrity íor their ad·ertising campaign, they need to analyze the
pre·ious knowledge a celebrity has or how he,she will utilize their knowledge in
the exposure phase. 1he authors belie·e that this will be an attribute that the
respondents will obser·e as too diííicult to e·aluate and thereíore the authors will
exclude it írom the research process. lurthermore, the respondent might íind it to
be to diííusi·e.
Literature re·iew
11
• íi/eabitit,: Marketers ha·e an important íunction when choosing which celebrity to
use in their ad·ertising campaign. Basically, these celebrities need to be admired or
at least well known in the public eye ,Belch & Belch, 2001,. lurthermore,
celebrities ha·e to be popular on the market and ha·e certain characteristics that
are extra·agant. Moreo·er, the company has to íind a balance to make sure that the
celebrity does not o·ershadow the brand itselí ,Ohanian, 1990,. 1his can be a
problem íor the company when a celebrity might be associated with negati·e
iníormation.
Summarizing the eííect oí source attracti·eness, one can say that it is used to create
eííecti·e messages, where the attribute attracti·eness reíers to the endorser`s physical
appearance, personality, likeability, and similarity ,Salomon, 2002, Ohanian, 1990,. 1he
authors ha·e chosen to íurther elaborate on the attributes similarity and likeability. 1his,
because they are key attributes when creating eííecti·e messages, thus they will aííect the
research process and pro·ide it with crucial iníormation that will íacilitate the approach to
íulíil the purpose.
2.4.3 Source Power
\hen mentioning the power a celebrity has in ad·ertising, one mainly reíers to how well
they can persuade the consumer to a purchase. 1his is ·ery beneíicial in personal selling,
where personal communication can be an eííicient way to con·ince or lead a consumer
into a purchase ,Salomon, 2002, Ohanian, 1990,. lowe·er, the power as source
characteristics is ·ery diííicult to apply in a non-personal iníluence situation such as
ad·ertising. 1he reason is that a celebrity in an ad generally cannot apply any sanctions to
the recei·er or determine any compliance that will actually occur ,Belch & Belch, 2001,.
lowe·er, the source power can be beneíicial in an endorsement strategy when using an
indi·idual with an authoritati·e personality as a spokesperson.
1he authors ha·e decided to exclude the whole concept oí source power írom the research
process, since it does not belong within the írame oí this thesis. 1he main reason
underlying this statement is because the purpose oí this thesis does not in·ol·e any
personal selling, more than the ad·ertising perspecti·e oí the use oí celebrity.
2.5 Brand
A brand, according to Keller ,2002, is a name, term, sign, symbol or e·en a combination oí
them in order to identiíy the goods and ser·ices that are being ad·ertised. 1he consumers
will be able to more easily diííerentiate one brand írom their competitors due to the
diííerent attributes associated with the brand. Ad·ertising has a great impact on how
consumers will percei·e a brand. Branding is about creating diííerences and pro·iding
products,ser·ices with the power oí brand equity ,Aaker, 1991,.
Brand is one oí the main pillars oí this thesis and the authors explained earlier its
importance on the celebrity endorsement process. Since branding is an extensi·e area, there
will be a special íocus within branding such as brand equity, the meaning transíer process,
and multiple brand endorsement. 1he concepts oí brand linked together with the
ad·ertising tool celebrity endorsement will be discussed in the diííerent sections below.
2.5.1 Brand equity
L·en though brand equity diííers among companies it should in general be deíined within
the area oí marketing eííects that is distincti·e to a brand. 1he concept oí brand equity
Literature re·iew
12
arises when consumers react to more preíerable products and the way it is marketed when
the brand gets identiíied. One has to be aware, according to \algren, Ruble & Donthu
,1995, that brand equity will be less ·aluable íor the manuíacturers and retailers ií it does
not ha·e any meaning to the consumers. Keller ,2002, continues explaining that brand
knowledge is composed oí awareness ,recall and recognition, whereas brand image is more
about ía·ourability, strength and uniqueness oí a brand association.
Ad·ertisers spend great sums oí money to ha·e celebrities promoting their
products,ser·ices with the expectation that consumers will react positi·ely to the
celebrity`s association with a certain brand. Consumers might say to themsel·es Ií she
uses it, it must be good` and Ií I use it, I will be like her` ,Belch & Belch, 2001,. In the
long term, this way oí thinking may lead to an increase in the sales and thereaíter the brand
equity. Brand equity ,Riezebos, 2003, indicates the intrinsic ·alue in a well-known brand
name. 1he amount oí ·alue that is reíerred to in a brand name depends on consumer`s
perception oí the brand domination and through social esteem that is pro·ided when using
it as well as the consumers trust and identiíication with the brand. 1he most ·aluable assets
in many companies are their brand names such as Coca-Cola and Nike which are also
reíerred to as mega brands ,Riezebos, 2003,.
Brand equity can be iníluenced by ad·ertising in se·eral ways ,\algren et. al, 1995,.
Awareness oí the brand can be created and increase the possibility that the brand is
included in the consumer`s mind. Moreo·er, when these brand associations get stored in
the consumers accessible memory, it can later lead to beha·ioural actions. 1he usage
experience can be iníluenced through the use oí ad·ertising and it can also aííect the
percei·ed quality oí a brand ,\algren et. al, 1995,. A celebrity spokesperson is used in
order to promote a company`s product,brand and íor these in turn to be associated with
the celebrity ,Belch & Belch, 2001,. 1his relates to the image,meaning transíer model
described in the next section.
2.5.2 Meaning transfer
McCracken`s transíer model is based on meanings and he suggests that the eííecti·eness oí
the endorser depends on the meaning the celebrity is bringing into the endorsement
process and the brand ,Schlecht, 2003, McCracken, 1989,. lence, he created the meaning
transíer model in order to explain the celebrity endorsement process. Celebrities are íull oí
diííerent meanings e.g. demography ,age, gender, personality and liíestyle types. 1his
makes it ob·ious that a celebrity represents not only one meaning but rather a ·ariety oí
se·eral ones. 1hese celebrity spokespersons are ·ery useíul in marketing brands since they
pro·ide the consumers with quite a íew characteristics when e·aluating the brands in
question. Celebrities add ·alue to the image transíer process because they are oííering
meanings oí deepness and power írom their personality and liíe styles, in comparison to
non-íamous endorsers ,Schlecht, 2003,.
Literature re·iew
13
1he model illustrates a three-stage process oí meaning transíer. 1his in·ol·es the creation
oí the celebrity image, transíer oí meaning írom the celebrity to the brand and the third
stage is how the brand transíers image onto the consumers ,Schlecht, 2003,.
Meaning Acquisition Lndorsement Consumption
ligure 2-3: Meaning transíer in the endorsement process ,Schlecht, 2003,
\hen the brand`s representati·e íeatures should be determined, the consumers` needs
should be considered. 1hen the ad·ertising company has to decide on what celebrity to
choose and who possesses the most appropriate characteristics in relation to the brand.
Simply, there has to be a congruence or íit between the celebrity and the product,brand
,Schlecht, 2003,.
2.5.3 Multiple brand endorsement
Nowadays, it is not unlikely that celebrity spokespersons can and are endorsing se·eral or a
speciíic brand. 1his situation is called multiple brand endorsement or e·en multiple
celebrity endorsement. Ad·ertising íirms might share certain spokespersons and thus the
celebrity will end up promoting more than one brand. An example oí this is the actress
Catherine Zeta-Jones who promoted both 1-Mobile and Llizabeth Arden. Also, the golí
champion 1iger \oods endorsed as much as three brands, American Lxpress, Rolex and
Nike. 1he marketers ha·e to question ií this kind oí celebrity endorsement does aííect
consumer brand attitudes ,Schlecht, 2003,. lowe·er, consumers are becoming more
knowledgeable within the íield oí marketing and they might think that the celebrity rather
preíers to get paid instead oí transíerring any meaning to the product, which in turn will
aííect the consumer buying beha·iour ,Belch & Belch, 2001,.
Pre·ious experience explained by Redenbach ,2005, shows that an endorsement oí íour
diííerent brands or products does indeed iníluence the celebrity`s credibility,
trustworthiness, expertise and likeability. 1he reason íor this is that a íamous person, who
endorses se·eral products instead oí only íocusing and representing one speciíic brand, will
e·entually attain a lack oí distincti·eness. lowe·er, one has to be aware that the use oí
multiple brand endorsement does not ha·e to imply that it is useless. Researches ha·e
actually showed some potential positi·e eííects like transíer oí positi·e brand images. Also,
the shape oí consumers response gets aííected positi·ely when more than íour products
are being endorsed ,Redenbach, 2005,.
It can also be beneíicial to endorse a product with multiple celebrities. Celebrity
spokespersons represent a diííerent mix oí types, like gender and age. Using multiple
celebrities in collaboration they can more eííecti·ely endorse a speciíic brand. 1he cosmetic
manuíacturer L`Oreal matches its wide range oí product lines depending on the celebrities
and their meanings ,Redenbach, 2005,.
Stage 1
Stage 2
Stage 3
Role 1
Role 2
Role 3
Celebrity Product Consumer
Literature re·iew
14
2.6 Consumer
1he continuously increasing competiti·eness has triggered many ad·ertisers to realize that
they need to detect the consumers` actual needs in order to satisíy them. By identiíying the
consumer buying beha·iour, it is more likely that the marketers will target products and
ser·ices directly towards the consumers` needs. Marketing is about satisíying needs and
thereíore it is crucial íor marketers to understand the rele·ance oí human needs to buyer
beha·iour. Consumers tend to search íor, purchase, use, e·aluate, and dispose
products,ser·ices that they expect will satisíy their needs ,Schiííman & Kanuk, 2004,.
1he uses oí endorsers or spokespersons as credible sources are nowadays being írequently
used by ad·ertisers in order to iníluence consumer`s attitudes and purchase intentions
,Goldsmith, Laííerty & Newell, 2000,. Credibility according to Belch & Belch ,2001,
means in this context the extent which the source is percei·ed as possessing expertise
rele·ant to the communication topic and can be trusted to gi·e an objecti·e opinion related
to the subject. 1rustworthiness reíers to the honesty and belie·ability oí the source whereas
expertise is originated írom the knowledge oí the subject. 1rustworthiness along with
expertise are dimensions that are important to theorize credibility and it has been shown
that they are ·ery iníluential when iníluencing attitudes and persuading consumers.
2.6.1 Consumer behaviour and negative publicity
Publicity tends to be more credible and ha·e more power than general marketing
communications exerted by a company ,Dean, 2004,. Lspecially, negati·e publicity seems
to ha·e the tendency to damage the company`s image. 1his is mainly due to the íact that
high credibility as well as the negati·ity eííect has a tendency to be more reílected upon
than positi·e iníormation in the consumer`s e·aluations. It is more likely that companies
will recei·e bad exposure since the media preíers to present bad news. 1ill & Shimp ,1998,
support this statement and continues, explaining that companies ha·e to be aware oí the
possibility oí attaining negati·e publicity when using celebrities as endorsers, since this may
eííect the consumers` perception oí the brand.
1he consumers` support oí a brand is oí major importance íor the prospect existence oí a
brand and the organization ,Riezebos, 2003,. Ií an incident occurs in relation to the brand,
the consumer`s trust in the brand will íade and the consequence will aííect the consumers
purchasing beha·iour. \hen consumers get questioned by marketers about their
purchasing beha·iour, they gi·e the impression that incidents with brands do not iníluence
their choices. Usually they also claim that neither ad·ertising nor negati·e publicity aííects
them. Riezebos ,2003, also states that the most objecti·e way to determine this eííect on
consumers is to list changes in the market shares.
Moreo·er, ií negati·e publicity leads to brand damage, the media can be seen both as the
source,spreader oí publicity simultaneously as challengers. In the case oí source,spreader
the media íunctions as a gate-keeper whereby it concentrates on those incidents that bring
about newsworthiness. Other incidents with a high newsworthiness are rele·ant with a
particular brand in crisis. 1his is true mainly because the other e·ents can act as deílections
and push away the negati·e publicity oí the brand. An example oí this push away e·ent`
can be a natural disaster. 1he media can thereíore also play an essential role in the
challenge oí negati·e publicity. In this case, one can think oí the incident as possible but on
the other hand the attention during this news should be spent on the rebuilding oí a brand
in order to retain the consumers and maintain the brand recognition ,Riezebos, 2003,.
Louie & Obermiller ,2002, explains the case oí negati·e e·ents` as problems that can take
Literature re·iew
15
place when a íamous person gets in·ol·ed in incidents that change or damage his,her
reputation. 1hese kinds oí circumstances can diííer widely írom misused exposure to an
accident that holds back a celebrity`s ability to períorm. In a study made by 1ill & Shimp
,1998, it was íound that negati·e iníormation about a celebrity can harm how consumers
percei·e the product,brand through the connected link between the brand and celebrity.
\hen a company has a potential relationship with a celebrity, the consumers will not íocus
on the company but rather on the circumstances surrounding the celebrity. Consumer`s
reaction may not be unpleasant when íirms reject high blame potential endorsers since
these could be harmíul to the company`s image and the consumer buying beha·iour.
lowe·er, due to the reason oí consumers being deíensi·e, they are supposed to not react
hea·ily when the potential endorsers ha·e low blame. Also, endorser candidates can be
more complex than the existing endorsers ,Louie & Obermiller, 2002,.
2.7 Chapter summary
1he main purpose oí this section is to introduce the íeatures that are rele·ant when
in·estigating which attributes consumers consider to be most appropriate íor a company to
keep in mind, beíore de·eloping a celebrity endorsement strategy. 1he attributes ha·e been
elaborated írom pre·ious empirical íindings within celebrity endorsement, where the main
íocus ha·e been addressed on celebrities and the eííect they ha·e had on company, brand
and consumer. lurthermore, the authors ha·e narrowed this additionally and put the íocal
point on how negati·e iníormation will aííect the consumers and how this is linked to the
three other phases in this thesis. 1hese attributes will play the most important role when
conducting the conjoint analysis, and they will be presented in section 3.3.1.
By analyzing many earlier studies, the authors came to the conclusion that most oí these
ha·e been structured in a similar way. Basically, the starting point was to elicit the
enormous role a celebrity ha·e had on the modern ad·ertising industry. lurthermore, many
oí these researchers ha·e embarked on how the consumers get aííected by celebrities ,1ill
& Shimp, 1998, Soderlund, 2003, Sil·era & Austad, 2004, Belch & Belch, 2001,
McCracken, 1989,. Only a íew oí them ha·e concentrated on the negati·e eííect a celebrity
can ha·e on consumers. 1his is the reason why, the authors chose to conduct this thesis
írom that nature.
1hese attributes are based on studies írom both qualitati·e and quantitati·e research
methods. 1he íact that similar or exact attributes are mentioned and discussed in the both
research methods, only strengthen the thought that the attributes within this thesis are
comprehensi·e. lowe·er, it can also be seen as a weakness in only using the already
analyzed attributes since there might be other more important attributes that consumers
might e·aluate in this dynamic market. lence, the time span limits the author`s ability to
search íor other inno·ati·e attributes. Despite this, earlier attributes will hopeíully bring
another understanding related to consumer`s perspecti·e on celebrities in ad·ertising.
1he íollowing attributes are elaborated on pre·ious studies. 1hese attributes are related to
the company, celebrity and brand. 1he reason why no attributes are directly related to the
consumer is because the research is conducted írom a consumer perspecti·e, which
indirectly will lead to íindings on what attributes consumers íeel are most important íor
companies to take into account when celebrities get associated with negati·e publicity.
Literature re·iew
16
Company
1he most important aspect that has been brought up is how well the celebrity fit.;vatcbe.)
with the company`s en·ironment and the brand as well. Consumers might percei·e a brand
diííerently depending on what celebrity a company chooses. 1he greater fit between the
celebrity and the brand, it is more likely íor a consumer to accept this ad·ertising tool. A
good congruence between the belieís oí a celebrity and a company will enhance the brand
recognition. 1hereíore, the authors belie·e that the fit ;vatcb) between the celebrity and
company is rather important to in·estigate íurther on when a celebrity gets associated with
negati·e iníormation and ií a change oí the fit will in turn change consumer`s perception
oí a brand. 1his assumption is strengthen by Lrdogen & Baker ,2004,, Pringle ,2004,,
Brown ,2003,, Danesh·ary & Schwer ,2000,, Louie & Obermiller, ,2002,, Kamins ,1990,,
& 1ill & Shimp ,1998, who claim that íit` ,match, can be considered as an attribute.
Celebrity
\hen talking about source credibility, both the attributes e·¡erti.e ava trv.trortbive.. are
included. lurthermore, the eííecti·eness oí a message depends on the celebrities` e·¡erti.e
ava trv.trortbive... 1he belieís, opinions, attitudes and beha·iour can be iníluenced by the
iníormation oí a credible source. 1his occurs when the consumer ,recei·er, accepts a
source iníluence when it comes to their personal attitude. Source attracti·eness includes
.ivitarit, ava ti/eabitit, which are also two attributes that will be used in the conjoint analysis.
1he underlying reason íor this is because the authors íeel they can play an important role
when analyzing the celebrity`s e·¡erti.e, trv.trortbive.., .ivitarit, ava ti/eabitit, and how that
iníluences the consumers` purchase beha·iour when the celebrity gets associated with
negati·e iníormation ,Lrdogan et al, 2001, Ohanian, 1990, 1ill & Shimp, 1998, Sil·era &
Austad, 2004, Kamins, 1990, Danesh·ary & Schwer, 2000, Kamins, Brand, loeke &
Moe,1989,.
Brand
1he most important aspect to consider when talking about brand and celebrity is to what
extent a celebrity can trav.fer av, veavivg to the brand itselí. 1hereíore, the more a celebrity
has the ability to trav.fer veavivg. to the brand, the more a consumer will associate with the
celebrity. lence, when a celebrity gets associated with negati·e iníormation, it might occur
that the celebrity looses the ability to transíer the right meaning to the brand as the
company would like to. 1his indicates also that the consumer will get aííected, sometimes
in a positi·e way and sometimes negati·ely. 1hereíore, the authors assume this is an
important attribute to study, in the sense on how consumers react to the meaning a
celebrity transíers, when negati·e iníormation is associated with a certain celebrity
,Schlecht, 2003, Redenbach, 2005, 1ill & Shimp, 1998, \algren, Ruble & Donthu, 1995,
Gwinner, 199¯,.
2.7.1 Hypotheses
1o summarize this, one can conclude that the presented attributes are considered as the
personal characteristics oí a celebrity. lowe·er, the main purpose oí this thesis is to study
which attributes that a company has to consider when a celebrity gets associated with
negati·e iníormation. 1o be able to íulíil the purpose oí this thesis and also answer the
research questions ,the íirst and second, that are posed in section 1.5, the authors íind it
·ery appropriate to in·estigate ií there is a relationship between the consumer willingness
to purchase a product or ser·ice, with the presented attributes in a conjunction. 1hereíore,
Literature re·iew
1¯
the íirst hypothesis deals with ií the consumers really get aííected by the chosen attributes
írom the literature re·iew. 1his leads us in to the íollowing hypothesis:
H
A
: 1he celebrity-product combination attributes expertise, trustworthiness,
similarity, likeability, meaning transfer, and ´fit¨ (match) affect the consumers'
willingness to purchase a product or service also under condition of negative
information.
As in research question two, the authors are aiming to understand ií the consumers are
aííected diííerently by these attributes when diííerent celebrities are used as endorsers.
1hereíore, the authors are ·ery keen on in·estigating ií consumers` willingness to purchase
a product or ser·ice diííers when using diííerent celebrities. 1his leads us into the second
hypothesis:
H
B
: 1he relative importance of specific celebrity-product combination attributes is
different depending on which specific celebrity is used as endorser.
Methodology
18
3 Methodology
This chapter will describe the choice of methods and how the process of gathering
empirical data will occur to fulfil the purpose of this thesis. The approach that will be
used is a quantitative one, and the survey will be analysed through a conjoint
experiment.
3.1 Introduction
Many researchers ha·e supported the idea oí using a multi-method approach in a
marketing research study. llick ,2002, & Carson, Gilmore, Perry & Gronhaug ,2001,, are
just a íew to mention and they íurther claimed that a combination oí qualitati·e and
quantitati·e approaches can lead to interesting and exciting explorations within a research
study. 1he integration oí these two methods should also be seen as complementary rather
than ri·alry. Moreo·er, llick ,2002, suggested that a thorough qualitati·e pre-study should
be carried out in order íor a quantitati·e study to be successíul. lowe·er, due to the time
span the authors ha·e with the thesis they decided that a multi-method-approach cannot be
undertaken. 1hereíore, the authors ha·e decided upon conducting a quantitati·e research.
In a quantitati·e method according to Lsaiasson, Gilljam, Oscarsson & \ägnerud ,2002,
the study is based on collected data that can be expressed in numbers to be estimated later.
1he reason íor this choice oí method is based on the íact that many companies claim
themsel·es to be consumer-dri·en, but still íail when conducting a marketing research, this
because they are only running íocus groups, sending out questionnaires, and analysing sales
data ,Gustaísson, lerrmann, & luber, 2001,. lowe·er, the main reason íor this íailure is
basically that the company does not understand the consumer needs and has a lack oí
iníormation about the market ,Iggland, 1989,. lrom the perspecti·e oí the purpose oí this
thesis, the authors are aiming to identiíy the most important íactors a company should
consider írom consumers point oí ·iew, when a celebrity gets associated with negati·e
iníormation. 1hereíore, a quantitati·e approach will pro·ide this thesis with better results
that will lead to the process oí drawing general conclusions and to get an o·erall picture
among the larger sample ,Gustaísson et al., 2001,. As stated in the multi-method approach,
conducting íocus groups as a pre-study will ía·our the study oí this thesis. But, at the same
time many researchers ha·e embarked on the concept or the use oí celebrity endorsement,
and thereíore they ha·e come to the conclusion that some íactors are more beneíicial íor a
company to consider when using a celebrity in their endorsement strategies ,McCracken,
1989, Se·ern, Belch & Belch, 1990, Lrdogen et al., 2001, \heeler, 2003, Sil·era & Austad,
2004, 1ill & Shimp, 1998, 1ill & Busler, 1998,. 1he authors oí this thesis ha·e thereíore
chosen to use the recommended attributes when conducting a quantitati·e approach, and
exclude the use oí íocus group or any other pre-study.
1he attributes ha·e been di·ided within the írame oí this thesis, company, celebrity, brand,
and consumer. Basically, one can say that this thesis has íour diííerent phases. lence,
rele·ant attributes ha·e been directly addressed to each phase and research questions ha·e
been de·eloped. 1he authors will aim to íind proper results íor each research question in
an eííicient way in order to identiíy the di·erse consumer needs.
3.2 Sampling
1he most important aspect a researcher has to keep in mind is how to determine an
appropriate research population and a proper sampling procedure. According to Churchill
,1995, the sampling procedure can be di·ided into probability and non-probability
Methodology
19
sampling. lurthermore, Saunders et al. ,2003, explains probability sampling, as the chance
oí each case being selected írom the population which is known. In non-probability
sampling there is an assumption that there is an e·en distribution oí characteristics within
the population. \ithin business research it is oíten not possible to speciíy the probability
that any case will be included in the sample and thus, the sample must be selected some
other way ,Churchill, 1995,.
Probability sampling in·ol·es the selection oí a sample írom a population, based on the
principle oí randomization or chance ,Aczel & Sounderpandian, 2002,. lence, probability
sampling is more complex in the sense that it sometimes in·ol·es two diííerent stages oí
sampling. 1hereíore, it can be considered to be more time consuming and more costly than
a non-probability sampling ,Saunders et. al, 2003,. Non-probability sampling is cheaper and
used when a sampling írame is not a·ailable. Also, this method is used in a research where
there is an interest oí obtaining an idea oí responses on ideas that people ha·e ,Churchill,
1995,. Based on this discussion, the sampling technique that will be used in this thesis is
non-probability sampling. 1he main argument íor this is the limited time and the expenses.
Additionally, the authors ha·e no sampling írame a·ailable and want to keep the research
to a low cost while putting the main interest on getting hold oí responses that people ha·e.
\hen conducting non-probability sampling, the main assumption a researcher makes is
that there is an e·en distribution oí characteristics within the population. In doing this, the
sample would be representati·e which will also lead to that the results will be truthíul
,Aczel & Sounderpandian, 2002,. lurthermore, non-probability sampling pro·ides a range
oí alternati·e techniques based on the researchers` subjecti·e judgement and examples oí
these are: quota sampling, purposi·e sampling, snowball sampling and con·enience
sampling ,Saunders et al, 2003,.
1o be able to íulíil the purpose oí this thesis, the main íocus lies on the consumers`
assumptions, and also the most important íactors that are related to celebrities that get
associated with negati·e iníormation. 1hereíore, the authors belie·e quota sampling is
most appropriate. 1his sampling technique means that a researcher has the ability to get
iníormation írom a respondent in the easiest way ,Lsaiasson et al., 2002,, which in this case
is on the campus area in Uni·ersity oí Jonkoping. Quota sampling was chosen because
there will be a sampling íocus on students at Uni·ersity oí Jonkoping. 1he underlying
íactor íor this statement is according to 1ang, Kim & 1ang ,2002,, 1ill & Shimp ,1998,,
that students are more conscious about celebrities and the authors also belie·e that they
will gain more ·aluable iníormation írom students. lurthermore, to be able to get a wider
understanding and strengthen the assumption that the students that are selected are similar
to those not selected, the authors will not íocus on one íaculty and thereíore take the íour
íaculties into consideration. Basically, Saunders et al. ,2003, argue that the sampling is
done, when a speciíic number oí units ,quotas, íor ·arious sub-populations ha·e been
selected. 1his means that the main population in this case are the students at Uni·ersity oí
Jonkoping, and the sub-population is the students at each íaculty.
Shepherd and Zacharakis ,199¯, state that the sample size used in a conjoint analysis is
smaller then 250 respondents, which is the normal amount in a required standard sur·ey.
lurthermore, they state that 50 respondents is a ·ery suííicient amount oí respondent,
when conducting a conjoint analysis. As mentioned beíore, the total population is the
students at Jonkoping Uni·ersity. lurthermore, Saunders et al. ,2003, state that the quotas
may be based on population proportions. 1o íacilitate the understanding íor the reader, the
authors chose to present the sample population and how the quotas are selected in table 3-
1.
Methodology
20
1able 3-1: Numbers oí Students at Jonkoping Uni·ersity
School Number oí students Percentage oí
students
Quota oí students in
sample oí ¯5
JIBS
1
1 ¯50 20 15
ING
2
1 800 20 15
lLK
3
3 000 36 2¯
llJ
4
2 116 24 18
1otal 8 666 100 ¯5
3.3 Conjoint Analysis
Conjoint analysis is a measurement technique that is concerned with an understanding oí
how people make choices between products or ser·ices, or a combination oí them that
better meet the consumers underlying needs ,Gustaísson, lerrmann, & luber, 2001,.
Rather then asking respondents íor their selí-perceptions oí what attributes determine their
choices, conjoint analysis in·estigations ask consumers to rank or rate complete` or íull
proíile` descriptions objects. 1he importance oí each speciíic attributes is tested out in the
statistical analysis. 1hereíore, conjoint analysis has been recognized as an extremely
poweríul way oí capturing what really dri·es consumers to purchase a product o·er
another and what consumers really ·alue ,Green & Srini·asan, 1990, Green, 2001,.
Moreo·er, conjoint analysis can according to Malthotra ,1996, and Green & Srini·asan
,1990, be used when de·eloping a model that enables companies to understand consumers`
needs. 1his is also important knowledge to be able to impro·e their market shares and as
well as understand how competitors` beha·iour will aííect their consumers.
Conjoint analysis has been used within many diííerent research areas, such as consumer
beha·ior ,Bruns, 2004,, and in diííerent marketing areas ,segmentation, positioning,
preíerence models, ,Green & Srini·asan, 1990, Malhotra, 1996, Gustaísson, lerrmann &
luber, 2001,. As one can tell, conjoint analysis is an eííicient research method when the
aim is to in·estigate and understand how consumers e·aluate diííerent combinations.
\ithin the írame oí the thesis, the authors aim to understand what the underlying íactors
consumers consider to be most important considering the aim oí this thesis.
1he main reason to the decision to conduct a conjoint analysis is de·eloped írom the aim
to íully understand how consumers are dri·en in e·aluating a celebrity that gets associated
with negati·e iníormation. Also, a standard sur·ey will not pro·ide the authors with
enough data to be able to draw conclusions and present a model íor companies that seek to
use a celebrity in their endorsement strategy. lence, the respondents in a standard sur·ey
1
Jonkoping International Business School
2
Lngineering School
3
School oí Lducation and Communication
4
School oí lealth and Science
Methodology
21
will only e·aluate each attribute ,íactor, separately, and thereíore not pro·ide a general
impression oí the use oí celebrity in an endorsement strategy. But, in a conjoint analysis the
respondents are asked to rate diííerent combinations oí attributes at diííerent le·els and in
doing this the authors belie·e they will gain more ·aluable iníormation that will íacilitate
the creating oí a model that embarks the concept, celebrity endorsement.
1here are no speciíic and reliable approaches to e·aluate the ·alidity and reliability in a
conjoint analysis. lence, according to Antilla, leu·el & Moller ,1980, the choice oí
attributes and their le·els are ·ery determined when it comes to ·alidity and reliability. 1o
be able to reach a high ·alidity, Antilla et al ,1980,, argues that a respondent needs to
e·aluate realistic attributes, such as expertise and trustworthiness.
3.3.1 Attributes and attribute level
\hen choosing the attributes and the le·els oí attributes, it is according to Lkdahl ,199¯,,
·ery important that they are realistic and related to the problem. lurthermore, he states
that the number should be as small as possible to minimize estimation eííorts. Bruns
,2004,, also states that the íirst steps in designing a conjoint study is to de·elop a set oí
attributes and their le·els.
lurthermore, to be able to conduct a conjoint analysis one has to identiíy the most
important attributes that a consumer assumes to be oí most importance in their purchase
beha·iour ,Antilla et al, 1980, Green & Srini·asan, 1990,. lrom the perspecti·e oí this
thesis, the attributes can be chosen írom pre·ious studies, both qualitati·e and quantitati·e
research methods, concerning the use oí celebrity endorsement in marketing.
\hen deciding upon diííerent le·els oí an attribute, one has to consider whether the
attributes are quantitati·e or qualitati·e. 1he attributes that are presented in table 3-2 are
írom a quantitati·e base. 1his indicates that the le·el oí the attributes will be easier to
estimate.
1able 3-2: Attributes and attribute le·el oí the conjoint experiment
1o be able to create attribute le·els both the authors and the respondents ha·e to
understand the clear diííerence oí le·el which an attribute is attached to ,Van Der Pol &
Ryan, 1996,. lrom the literature re·iew, it has been pro·en se·eral times, that the attributes
and their le·els presented in table 3-2 are ·ery applicable when studying celebrity
endorsement ,Lrdogan, Baker & 1agg, 2001, Ohanian, 1990, 1ill & Shimp, 1998, Sil·era &
Attribute Level J Level 2
Lxpertise ligh Low
1rustworthiness ligh Low
Similarity ligh Low
Likeability ligh Low
Meaning transíer ligh Low
lit` Match ligh Low
Methodology
22
Austad, 2004, Kamins, 1990, Danesh·ary & Schwer, 2000, Kamins, Brand, loeke & Moe,
1989,. 1he reason íor the choice oí attributes were elaborated and explained in depth in
section 2.¯.
3.3.2 Different approaches of conjoint analysis
\hen one considers conducting a conjoint analysis, there are many diííerent approaches to
reílect upon. lowe·er, the most popular one is a íull proíile approach ,Green &
Srini·asan, 1990, when the amounts oí attributes are not less then six and at the most
eight. 1hey recommended using trade-oíí matrices, when a larger number oí attributes are
used. lurthermore, they also argued that when the number reaches ten or more attributes,
a researcher should use selí-explicated data and methods in·ol·ing a combination oí selí-
explicated and conjoint data. 1he latest recommendation can be compared to an .aa¡tire
Cov;oivt .vat,.i. ,ACD,. 1he ACD has gained more acceptances in recent years, due to that
it can accommodate a larger number oí attributes, where it combines selí-explicated data
with pairs in comparison to intensity ratings ,Johnson, 198¯,.
1he approach, íull proíile, is chosen by the authors to gather rele·ant data íor íurther
analysis. 1he main reason is based on the amount oí attributes that the authors decided to
support their research on ,see table 3-2,. 1he amounts oí attributes are six, which indicates
that a íull proíile is the most suitable approach íor this study. lowe·er, this can be seen as
a negati·e aspect due to that the respondents should not be o·erloaded. 1hereíore, the
authors decided upon attributes that are rele·ant to the purpose oí this thesis and also not
o·erwhelming íor the respondent.
3.3.3 Experimental design
Beíore designing the sur·ey one has to distinguish and make clariíications concerning
dependent and independent ·ariables. According to Antilla et al. ,1980, the dependent
·ariable can be seen as the preíerence which can be explained by a number oí independent
·ariables. 1he dependent ·ariable used in this study is: tbe rittivgve.. to ¡vrcba.e. In other
words, the analysis aims at íinding out what attributes aííect the most respondents in their
purchasing decisions, in the context oí a celebrity being associated with negati·e publicity.
1he independent ·ariables can also be considered as attributes, which in this case are six
diííerent ones at two le·els. lurthermore, the basic assumption íor a conjoint experiment is
that the attributes are known in ad·ance ,Bruns, 2004,. 1his means that the researcher can
select the attributes that are most appropriate íor the study írom literature and pre·ious
empirical íindings. 1he authors support this assumption and ha·e presented the choice oí
attributes more in-depth in chapter two. Aíter clariíying and deciding upon the dependent
and independent ·ariables one can start designing the experiment.
\hen designing the experiment one has to keep in mind how the respondents will
interpret it and as well understand it. In the case oí conjoint analysis, the experiment will be
based on diííerent generated combinations or conjoint proíiles, in this case 64, which also
is a íull íactorial design. 1he sur·eys are usually not períormed as íull íactorial design, but
rather as íractional íactorial designs, which basically are íractions oí the íull design. 1he
reason that researchers use íractional íactorial designs, according to Lkdahl ,199¯,, is that
they usually are utilized in order to add more attributes into the combinations and at the
same time to not increase the strain on the respondents. One can íor sure use a íull
íactorial design, when collecting data, but according to Bruns ,2004,, there will be too
many combinations íor a respondent to rate and still pro·ide ·alid results. 1hereíore, the
Methodology
23
most important reason íor limiting the experiment according to Bruns ,2004, is to keep the
number oí cases manageable while still creating ·alid results.
In this case the authors came to the decision that two le·els and six attributes will pro·ide
this thesis with ·alid results. lurthermore, the assessment oí six attributes with two le·els
íor each ·ariable would lead to 64 ,2
6
, hypothetical combinations which is a íull íactorial
design. 1his according to Green & Srini·asan ,19¯8, would be an o·erwhelming task íor
each respondent to e·aluate and still gi·e consistent and meaningíul answers. 1hereíore,
Gustaísson et al. ,2001, argues that it is important to keep the number oí cases to be
e·aluated to as íew as appropriate and possible, but still ha·e enough cases to generate
signiíicant and ·alid results. 1his is the reason why an orthogonal íractional íactorial design
will be applied ,Green & Srini·asan, 19¯8, Bruns, 2004,. In doing this, the number oí
combinations will be reduced írom 64 to 8 which is according to Lkdahl ,199¯, a suííicient
number oí combinations to conduct a conjoint analysis. But, Lkdahl ,199¯, also argued
that a suitable íractional íactorial design íor six attributes at two le·els could be 2
6-2
~2
4
~16
combinations. lurthermore, he stated that the íractional íactorial designs are generally
more resource eííecti·e, and in the case oí conjoint analysis more íeasible.
Shepherd & Zacharakis ,199¯, ha·e another ·iew oí the amount oí combinations that is
suííicient when conducting a conjoint analysis. 1hey claimed that duplexing the amount oí
combinations will lead to more ·alid results as well as reduced bias. 1his means that the
number oí the total amount oí combinations will be duplicated into 16, when the íractional
íactorial design is 2
3
,8 combinations, and into 32 when the íractional íactorial design is 2
4
,16 combination,, and still be manageable. Con·ersely, Smith, Schullen & Barr ,2002,
argues that it is e·en applicable and manageable to use less than 100 combinations where
the respondents are able to judge satisíactorily. 1hey also state that the amount oí
combinations is related to the nature oí the purpose and what the authors want to achie·e.
By taking both oí these approaches into consideration, the amount oí combinations ,8 or
16,, which will be suííicient to íulíil the purpose oí this thesis, ha·e made the authors
question it. 1hereíore, to not decrease the reliability oí this thesis and the ability to
maintain accurate data to reach a sati íactional conclusion, the authors chose to test both 8
and 16 combination in duplex íorm, beíore conducting the collection oí data which leads
us into the next section.
3.3.4 Pilot study
\hen constructing the experiment, crucial íeedback írom the respondents were gi·en to
the authors, who realized that 32 combinations would be too extensi·e íor the respondents
and thereíore the authors chose to keep it to 16 combinations. Other crucial íeedbacks
were gi·en, both positi·e and negati·e. 1he positi·e ones were that the instructions and
description oí the attributes and le·els were straightíorward. 1he main concern was how to
deal with the combinations and the question related to them. 1he respondents interpreted
the question diííerently, which also aííected their ratings. \hen asking them how they were
thinking, the authors could clearly see a pattern oí diííerent interpretations. 1hereíore, the
authors íound it highly important to re-phrase the question, which later on showed that the
interpretation oí the question did not diííer in a crucial matter. In appendix 3, an extract
írom the pilot study will be presented.
3.3.5 Data collection
1here are usually two types oí scales used in conjoint analysis sur·eys, rating and rank
order ,Lkdahl, 199¯,. lurthermore, Gustaísson ,1996, explains íi·e diííerent response
Methodology
24
scales, where a conjoint analysis can be conducted. 1hese are, rank order, rating, ·erbal
rating, íirst rank order and then distribute the concepts on a scale, and íirst rate and then
rank order among combinations with the same rating. \hen the respondents are asked to
rank the combinations, which is rather easy íor them, they will only pro·ide iníormation
that is in order oí preíerences and not stating the degree oí preíerences ,Green &
Srini·asen, 19¯8,. Ranking is also not suitable íor postal questionnaires, since the
respondents require help to íacilitate their task. Verbal rating ,would buy and would not
buy, is also not useíul íor postal questionnaires as it is most applicable íor low in·ol·ement
products. 1he rating scale ,1-¯ or 1-9,, makes the respondents e·aluate each combination
separately in regard to other combinations. 1he íirst mentioned scale will be used in this
thesis íor data collection. 1he reason íor this is that the authors want each combination to
be e·aluated separately, while the respondents assume that a celebrity gets associated with
negati·e publicity. Also, the results will be more integrated and understandable in
comparison to a scale oí 1-9, which can pro·ide too a·eraged results. lurthermore, aíter
the pilot study was conducted, the authors came to the conclusion that a rating scale írom
1 to ¯ will make it easier íor the respondents to e·aluate the diííerent combinations. 1his is
also stated by Smith et al. ,2002, who claimed that this rating scale is normally expected to
show íulíilling reliability and ·ariability.
3.3.6 The survey
1he main sur·ey is di·ided into íi·e diííerent parts: ,1, presentation and task instructions
íor the respondent, ,2, description oí the attributes and their le·els, ,3, the combinations
,conjoint proíiles,, ,4, a post-experiment questionnaire, ,5, a short questionnaire
concerning the respondents` characteristics. 1he complete sur·ey is presented in appendix
2.
In the íirst part the respondents were instructed to assume that the celebrity was associated
with negati·e iníormation, and írom that rate each combination in order to show which
íactors a company should consider in such particular case. 1he second part the authors
belie·e it is highly important íor the respondents to understand the meaning oí each
attribute and their le·els, to íacilitate íor the respondent and to pro·ide the authors with
more accurate data. 1hirdly, the combinations will be presented in a ·erbal description
where the attributes will be introduced sequentially. 1he order in which the attributes are
presented is crucial since the results depend highly on this. 1wo diííerent cases will be
pro·ided íor the respondents to íacilitate their rating oí the attributes. In the íourth part
the respondents will be asked to e·aluate the diííerent attributes and state how important
they are íor a company to consider when a celebrity gets associated with negati·e publicity.
Lastly, the authors are not seeking to segment the total population based on the answers
that are gi·en. 1his leads to no particular interest in demographical iníormation, such as
age, sex, income etc. 1his is why, in the short questionnaire, the authors only asked about
their personal characteristics and how well íashion-oriented they are. L·en though the
authors randomly chose respondents within each school, they a·oided asking respondents
that are not conscious about íashion in order to enhance the ·alidity. By using non-
probability sampling the authors` choice oí data collection was supported.
3.4 Statistics
\hen conducting the regression analysis ,multiple regressions, in SPSS, an understanding
is needed to see how well the experiment has been explained by the regression. \hen
looking at the signiíicance oí l ,P-·alue, it shows how much oí the results that can be
Methodology
25
attributed to chance. 1his means the percentage oí data, which is not explained by the right
choice oí attributes and attribute le·els. 1he signiíicance ,P-·alue, has to be higher than 5
percent in order to support the results ,Gustaísson et al. 2001,.
Moreo·er, the signiíicance is di·ided into diííerent le·els depending on how signiíicant the
results are, and the íollowing le·els are presented below: ,Korner & \ahlgren, 1998,
1hree star signiíicant ,,: P-·alue·0.1 °
1wo star signiíicant ,,: 0.1 ° ·P-·alue·1.0 °
One star signiíicant ,,: 1.0 ° ·P-·alue·5.0 °
Analysis and results
26
4 Analysis and results
ív tbi. cba¡ter, tbe avtbor. ¡re.evt tbe avat,.i. ava re.vtt.. 1be vo.t reteravt fivaivg. i. brovgbt v¡, rbicb
tbe avtbor. betiere are iv¡ortavt to fvtfit tbe ¡vr¡o.e. ívrtbervore, av avat,.i. i. covavctea rbicb teaa. to
tbe aeci.iov ov rbicb attribvte. are vo.t iv¡ortavt.
4.1 Introduction
1he research sur·ey íor this thesis has been di·ided into a conjoint experiment and two
short questionnaires to be able to íulíil the purpose oí this thesis. lirstly, the consumers`
willingness to purchase a product or ser·ice when the attributes are in a conjunction is
presented. Secondly, an in·estigation will be brought up to see ií the consumers`
willingness to purchase a product or ser·ice diííers when diííerent endorsers are used. 1he
results íor both these parts are recei·ed írom the conjoint experiment. Data which is
deri·ed írom the conjoint experiment, according to Shepherd & Zacharakis ,199¯, can be
used through two statistical techniques, regression analysis and analysis oí ·ariance
,ANOVA,. linally, an analysis will be presented oí the standard de·iation and mean ·alue
when it comes to how consumers percei·e each attribute separately.
1he authors will introduce the results írom the research process, where each part is related
to a research question and there are also hypotheses addressed to the two íirst research
questions. 1he last research question deals with the two last parts in the research process.
1o íacilitate the structure oí this chapter the authors want to make it clear íor the reader
how the diííerent research questions are linked to the research process. 1his will be
explained in the table 4-1.
1able 4-1 Outline oí research process
Research question 1 Conjoint experiment, combination 1-16
Research question 2 Conjoint experiment, combination 1-8 case 1 ,Michael Jackson,.
Conjoint experiment, combination 9-16 case 2 ,Bill Clinton,.
Research question 3 Questionnaires regarding post-experiment and respondents
characteristics.
4.2 Research question 1
!bicb factor. frov tbe titeratvre rerier affect tbe cov.vver.` rittivgve.. to ¡vrcba.e a ¡roavct or .errice
tbat i. evaor.ea b, a cetebrit, tbat i. a..ociatea ritb vegatire ivforvatiov, rbev tbe attribvte. are iv a
covbivatiov.
lirstly, the le·el oí importance oí each attribute is presented when the respondents
e·aluated the combinations. 1here are only small diííerences between the attributes in
general ,see appendix 4A,. lowe·er, when comparing each bar some attributes are on a
higher le·el than the other ones. In all oí the combinations, expertise was seen as the most
important attribute íollowed by trustworthiness, similarity and likeability. All oí these
attributes, as was mentioned in the literature re·iew, are directly connected to the celebrity
and company. 1his implies that the choice and use oí a celebrity is crucial since it appears
Analysis and results
2¯
clearly in the íindings that these attributes iníluence the consumer mostly. Moreo·er, e·en
though the le·el oí importance between these íour attributes diííers slightly, they seem to
aííect the consumers combined with each other. 1he other two attributes, meaning transíer
and íit` match, showed a lower le·el oí importance than the íirst mentioned íour
attributes. 1he results indicate that the meaning transíer a celebrity shiíts to a certain brand
is more important than how well the íit between the celebrity and brand is. Both meaning
transíer and íit` are directly connected to celebrity and brand.
Looking íurther into how each indi·idual rated each combination, one can clearly see ,see
appendix 4B, that íit` and meaning transíer do not show any indications oí high
di·ergence. 1he respondents seem to ha·e similar opinions regarding these attributes
meaning that they are not considered to be the most crucial attributes when purchasing a
product,ser·ice. At the same time, the authors can strengthen this statement that the other
íour attributes, expertise, trustworthiness, similarity and likeability ha·e the greatest impact
on the consumers` willingness to purchase when a celebrity is used as an endorser. One can
íurther on see that combination 12 has the highest mean ·alue with a le·el oí 4.9, which
makes it the most important combination and basically the winning concept. \hen looking
at the concept ,see appendix 4C, the most important attributes are expertise,
trustworthiness, similarity and likeability, which also strengthen what is gi·en írom the
importance summary.
4.2.1 Testing the hypothesis
As mentioned beíore this part oí the research deals with the 16 combinations. \hen
conducting a multiple regression analysis, one has to take into account the diííerent
statistical terms, such as R-square and the signiíicance le·el. 1hese terms are presented in
the model summary ,see appendix ¯A,. \hen looking at the table, R-square is 0.316 which
means that 31.6 ° oí the obser·ed ·ariability oí the willingness to purchase is explained by
the attributes. lurthermore, R
2
is the squared correlation coeííicient between the obser·ed
·alues oí the dependent ·ariable and the predicted ·alue based on the regression model. A
·alue close to 1 tells you that the dependent ·ariable can be períectly predicted írom the
independent ·ariable, whereas a ·alue close to 0 pro·es the opposite. 1he R
2
·alue in this
case is 31.6 °, which is a moderately good model íit íor this type oí data.
Due to loss oí degrees oí íreedom, howe·er, the adjusted R
2
is only 12.8°. 1his loss oí
degrees oí íreedom also explains why the signiíicant ·alue íor the model is as high as 0.0¯8
that is higher than the 5° con·entionally used in signiíicant testing. lowe·er, this is still a
signiíicant result on the 10° le·el and gi·en the theoretical support íor these attributes our
conclusion is that the results are more in ía·our oí the than against hypothesis A.
H
A
: 1he celebrity-product combination attributes expertise, trustworthiness,
similarity, likeability, meaning transfer, and ´fit¨ (match) affect the consumers'
willingness to purchase a product or service also under condition of negative
information.
4.2.2 Results research question 1
As consumer willingness to purchase a product or ser·ice gets aííected by the attributes,
taken írom the literature re·iews, one can argue that these attributes are rele·ant when it
comes to celebrity endorsement strategy. lurthermore, these attributes showed diííerent
le·els oí importance íor consumers` willingness to purchase a product or ser·ice that is
endorsed by a celebrity that is associated with negati·e iníormation. It was also pro·en in
Analysis and results
28
the íindings that expertise, trustworthiness, similarity and likeability oí a celebrity ,source
attracti·eness and credibility, ha·e a strong impact on consumers. 1he percei·ed expertise
oí an endorser is more important when it comes to the purchasing intentions, rather than
trustworthiness. lence, the celebrity ,source, needs to íirst and íoremost be an expert in
the endorsement strategy and secondly be trustworthy in order to reach to the consumer
,recei·er,. 1he other attributes linked to a celebrity are also oí importance and a crucial
aspect to bear in mind is that a celebrity, who is seen as ha·ing all these six attributes linked
to him,her, is more likely to reach a certain target compared to a celebrity who only is
percei·ed as linked with one main attribute. As the íit` and meaning transíer do not show
any indications oí importance, one can say that consumers do not íind a celebrity that is
associated with negati·e iníormation to aííect the relationship it has with the brand or
company. linally, when looking at the importance summary one can see that the attributes
are relati·ely important, but mostly when it comes to expertise, trustworthiness, similarity
and likeability. Studying the 16 combinations in the multiple regressions, the authors íind
the regression to be signiíicant since there is no high skewness and at the same time it
strengthens the acceptance oí the hypothesis.
4.3 Research question 2
Do tbe attribvte. frov tbe titeratvre rerier bare aifferevt iv¡act. ov cov.vver.` rittivgve.. to ¡vrcba.e a
¡roavct or .errice, rbev aifferevt cetebritie. are v.ea a. evaor.er..
In this section, an in·estigation on how the respondents rated the diííerent combinations
related to each case is presented. 1he reason in doing this is to see how consumers percei·e
the use oí diííerent celebrities as endorser. 1his is also done to íind ií consumers react
equally no matter what celebrity is used at the same time as the particular celebrity is
associated with negati·e iníormation.
4.3.1 Case 1
Looking at the importance summary ,see appendix 5A, oí how the respondents rated the
attributes in this case, the authors can conclude that the most important attribute was
likeability íollowed by expertise. 1rustworthiness was nearly as important íollowed by
meaning transíer and similarity. Lastly, íit` was not considered to aííect the respondents`
willingness to purchase, as much as the pre·iously mentioned attributes. 1he indi·idual
subject importance ,see appendix 5B, shows that likeability is clearly the most important
attribute since the respondents agree upon its importance, while the other attributes are
percei·ed ·ery diííerent indi·idually. 1his indicates that the respondents ha·e diííerent
perceptions to the characteristics oí the celebrity.
4.3.2 Case 2
1he importance summary ,see appendix 6A, tells us that the most important attributes íor
the respondents were trustworthiness and meaning transíer. 1he third attribute that the
respondents considered to be important was íit`. 1he three other attributes, expertise,
similarity and likeability were considered to be oí equal importance. 1he authors can
íurther on claim, when looking at the indi·idual subject importance ,see appendix 6B,, that
trustworthiness was more important than meaning transíer based on that it was rated
higher. Meaning transíer and íit` were rated equally, but one can see that meaning transíer
was in most cases more important. Lastly, when looking at appendix 6B, one can see that
expertise, similarity and likeability are, as mentioned beíore, like wisely important. Similarity
was slightly more important in comparison with the other two, due to the di·erse rating.
Analysis and results
29
4.3.3 Testing the hypothesis
\hen testing this hypothesis the authors intend to compare the importance ·alues across
the two regressions in case one and two. lurthermore, a comparison between the R
2
·alues
in both cases is made. Looking at the importance summary oí the attributes ,see appendix
5A and 6A, in both cases, one can clearly see a signiíicant diííerence on how the
respondents percei·e the attributes when diííerent celebrities are used.
In case one the R
2
is 18.5° and in case two it is 25.6° ,see appendix ¯C and ¯L,. As
shown there is a diííerence between the R
2
·alues, which also supports the hypothesis B
that there is a diííerence in the perception oí the attributes, depending on which speciíic
celebrity is used as endorser.
H
B
: 1he relative importance of specific celebrity-product combination attributes is
different depending on which specific celebrity is used as endorser.
4.3.4 Results research question 2
1he hypothesis shows that consumers willingness to purchase or not is aííected by a
celebrity`s characteristics ,the attributes írom the literature re·iew,. 1his means that
consumers think and act diííerently depending on a particular celebrity who is used in the
endorsement strategy. 1his also indicates that companies need to take this into account
when choosing a celebrity, since the mechanism oí negati·e publicity associated with a
celebrity does not necessarily has to mean that the consumers choose not to purchase the
endorsed brand. Instead, a company need to ha·e in mind that the willingness to purchase
a product,ser·ice depends highly on to what extent consumers percei·e a celebrity as
possessing the six attributes. 1he higher perception oí a celebrity, the more likely and
willing they will be to purchase. Since two diííerent celebrities were used in each case, one
can clearly see that consumers tend to e·aluate a celebrity`s personal characteristics in
·arious ways.
In case one, the attributes likeability and expertise were rated as the most important ones
while íit` was not seen as playing an important role. In case two howe·er, trustworthiness
and meaning transíer were rated as most important whereas likeability surprisingly was seen
as the least important attribute. 1his clearly shows that there are clear diííerences between
how consumers percei·e diííerent celebrities depending on what characteristics they
possess and in what certain situation they are in. Also, as was mentioned in chapter two, it
was indicated that when consumers get asked about their purchasing beha·iour they usually
gi·e the impression that negati·e publicity in relation with brands do not usually aííect their
choices. lowe·er, the authors íeel that consumers purchasing beha·iours might occur
unconsciously and e·en though they claim that negati·e publicity does not ha·e any
iníluence on them, they most likely can. lurthermore, the authors can argue that the
attributes írom the literature re·iew are diííerently percei·ed in relationships with diííerent
celebrities.
4.4 Research question 3
!bicb attribvte. ao cov.vver. fiva of vo.t iv¡ortavce rbev eratvativg tbev .e¡aratet,.
In this section oí the empirical study, the respondents were asked to e·aluate the six
diííerent attributes separately. 1he reason is to compare the results írom the conjoint with
the sur·ey, which in can either strengthen the research process or indicate that the
respondents are not íully aware oí their own decision criteria.
Analysis and results
30
4.4.1 Perceived importance of attributes
In this section, the respondents were asked to rate the selí-percei·ed importance oí each
attribute. Lach oí the six attributes was e·aluated on a scale ranging írom 1, vviv¡ortavt, to
5, iv¡ortavt. 1he results oí the respondents` percei·ed importance oí the attributes are
a·eraged and illustrated in table 4-1.
1able 4-2 Ranging oí attributes
Mean Std. De· Ranking
Lxpertise 4.0 1.25 2
1rustworthiness 4.3 0.93 1
Similarity 3.4 1.1 4.5
Likeability 3.0 1.06 6
Meaning transíer 3.4 1.24 4.5
lit` Match 3.6 1.1 3
1he mean oí the selí-percei·ed weights ranged írom 3.0 íor ti/eabitit, to 4.3 íor
trv.trortbive... 1his indicates that the respondents percei·ed trv.trortbive.. as the most
important íollowed by expertise. 1he fit vatcb was ranked as the third most important
attribute íollowed by .ivitarit, and veavivg trav.fer which both were rated oí equal
importance. Lastly, the least important attribute is ti/eabitit, ,see appendix 8,.
4.4.2 Results research question 3
1he respondents rated the importance oí the attributes in general and this is why one can
see clear diííerences between which attributes are important and less important. Since the
consumers belie·e that the celebrity`s credibility is the most important, where both
trustworthiness and expertise are included, one can say that the consumers are more
aííected by a celebrity`s positi·e characteristics. lurthermore, how well the celebrity
communicates the message that the company is trying to send, aííects the consumers
purchasing decision. As it was stated that expertise and trustworthiness are the most
importance attributes, one can clearly state that the consumers ·alue the knowledge,
understanding and ethical aspects oí celebrity more than the other personal characteristics.
1his also implies that companies should take these attributes more into account when
choosing a celebrity as an endorser. 1his is because ií a particular celebrity gets associated
with negati·e iníormation it will not aííect the brand equity or lead to a decrease in the
market shares as much. At the same time, the celebrity will still transíer necessary
iníormation that a consumer can íind crucial and that potentially can lead to a purchase. It
was also íound in this research that a celebrity`s attracti·eness does not aííect the
consumers in any way, which can be chocking íor many consumers since they may assume
that the use oí celebrity endorsement is mainly beneíicial íor a company due to a celebrity`s
attracti·eness.
Analysis and results
31
4.5 Chapter summary
As the journey to íind the hidden íactors that companies need to consider when using
celebrities in ad·ertising is heading towards its end, the authors need to clariíy some crucial
aspects. Since, the íocus throughout this thesis has been conducted írom consumer`s point
oí ·iew, the authors will írom now on consider the respondents in this research process as
consumers.
1he authors can clearly state that the strategy oí using celebrity endorsement is an eííicient
way to reach out to a di·erse audience. 1he íocal point throughout this study was the
in·estigation on how willing consumers are to purchase a brand based on the attributes
elaborated írom the literature re·iew, in relation to a celebrity that gets associated with
negati·e iníormation. At this point the authors can clearly present íacts that consumers are
aííected by celebrities and that negati·e iníormation that a celebrity is associated with can
play an important role occasionally. In the promotional campaign where celebrities are
included, the purpose is to send out optimal iníormation to the consumers that will lead to
a purchase. 1aken the hypothesis A into account one can say that consumers are willing to
purchase a product or ser·ice based on the attributes taken írom the literature re·iew,
when they are in a combination. On the other hand, when in·estigating hypothesis B the
authors came to the understanding that the consumers` willingness to purchase a product
or ser·ice is aííected by a particular celebrity, when the celebrity has the mentioned
attributes as personal characteristics. 1hereíore, marketers in ad·ertising agencies need to
be aware that a celebrity`s personality together with brand- and the consumers indi·idual
personalities ,triangular relationship, are integrated, and the higher the linkage is between
these three parts the less likely consumers will get aííected negati·ely by negati·e
iníormation.
Since it was shown írom the three diííerent sections in this chapter ,research question 1-3,
that the consumers rated the attributes diííerently based on the diííerent cases, the authors
can at this point claim that consumers tend to percei·e attributes in ·arious ways. But, to
keep the íocus within the írame oí this thesis, the authors chose to analyze the relationship
between the ratings oí the diííerent concepts with how the consumers rated each attribute
separately. lirstly, when rating the 16 combinations, the most important attributes
,expertise, trustworthiness, similarity, likeability, are linked with the celebrity`s
characteristics ,see section 2.4,. On the other hand, the attribute meaning transíer which is
linked with the brand and íit` which in turn is linked with the company are not
considered to ha·e a particular importance. In the case oí Michael Jackson, the three most
important íactors ,likeability, expertise, trustworthiness, are also linked to the celebrity. At
the same time, meaning transíer and íit` are once again not percei·ed as crucial íactors.
lowe·er, in the case oí Bill Clinton both meaning transíer and íit` were considered to be
important íollowed by trustworthiness, and the other three attributes as less important.
\hat can be said írom this, írom the diííerent le·els oí importance in the diííerent cases·
It is as simple as the celebrities themsel·es and the brand they are endorsing aííects how
consumers react to the diííerent attributes in a combination. \hen comparing the two
cases with the total rating oí the 16 combinations, one can see that trustworthiness in all
cases is ·ital whereas the other attributes are rated diííerently. One can also clearly
understand that in the 16 combinations, expertise was seen as the absolute important
attribute which is also indicated to be important in the case oí Michael Jackson but less
important in the case oí Bill Clinton. 1his can be due to the product Clinton was
endorsing. 1hereíore, the authors belie·e that expertise also has a crucial aííect when a
celebrity gets associated with negati·e iníormation. 1his due to that expertise was percei·ed
Analysis and results
32
and ranked as the second most important when the consumers rated the attributes
separately. lurthermore, as the authors aim to reach a general conclusion, expertise would
be an accurate attribute to include when choosing a celebrity in ad·ertising purposes.
linally, based on e·erything mentioned so íar in this thesis, the authors can summarize the
most important íactors that companies need to consider when a celebrity gets associated
with negati·e iníormation. Basically, all attributes are important when using a celebrity as
an endorser but in the case when a celebrity gets associated with negati·e iníormation, the
authors e·aluated all the six attributes and compared them and came to decide upon the
íollowing attributes to be the most crucial attribute to be taken into account when
choosing and using a certain celebrity.
1rv.trortbive..: the reason why this attribute is important is basically that it was ranked
highly both when the consumers rated it in combination in all cases and when they rated it
separately. 1hereíore, there are se·eral reasons in concluding it as the most crucial
attribute.
í·¡erti.e: has been rated diííerently in each case but it was only rated as ha·ing a low
importance in the case oí Bill Clinton. Despite this, the authors belie·e that it is an
important attribute to consider when a celebrity gets associated with negati·e iníormation.
1he reason íor the low rating with the Clinton study could be explained as the product
which he promoted and also that consumers did not consider him being ·ery
knowledgeable in that particular íield.
1he remaining íour attributes, meaning transíer, likeability, similarity, and íit` did not in
most cases show a high le·el oí importance but rather an indication oí a·erage importance.
1hus, the authors do not íind them to be crucial íor a company when they use a celebrity
in the endorsement strategy. lowe·er, these can still be oí importance in conjunction with
the two íirst mentioned attributes. 1his is based upon that likeability in the case oí Michael
Jackson showed to be the most important attribute whereas when the attributes were rated
separately, this attribute was seen as the least important one. 1he remaining three showed a
relati·ely a·eraged importance and are thereíore not crucially unimportant. But, at the same
time they are not the dri·ing mechanism to successíul brand recognition when using a
celebrity in ad·ertising.
Conclusions and discussion
33
5 Conclusions and discussion
ív tbi. cba¡ter tbe avtbor. ritt fivatt, ¡roriae tbe reaaer ritb airect av.rer. to tbe ¡o.ea re.earcb qve.tiov..
ívrtbervore, fivat rora. frov tbe avtbor. ritt be girev. ía.tt,, a ai.cv..iov ritbiv tbe re.earcb area ritt
be etaboratea ava iv¡ticatiov. to ¡ro.¡ect .tvaie. iv tbe area of cetebrit, evaor.evevt ritt be girev.
5.1 Introduction
1his thesis in·estigated consumers` willingness to purchase a product or ser·ice based on
the six attributes that were retrie·ed írom the literature re·iew. In the introduction chapter,
three research questions were identiíied. 1he direct answers to the research questions will
be more speciíied in section 5.2. 1o be able to answer these questions, the quota sampling
oí ¯5 uni·ersity students in the campus oí Uni·ersity oí Jonkoping was studied. 1hese
students were asked to e·aluate 16 hypothetical cases, which are based on the attributes
retrie·ed írom the literature re·iew and they were asked to indicate the willingness to
purchase a product or ser·ice.
lurthermore, the literature re·iew pro·ided the authors with rele·ant iníormation oí the
crucial attributes that are associated with celebrity endorsement. 1hese attributes ha·e
pre·iously been used when gaining an optimal use oí a celebrity as an endorser ,see íigure
1-1,. 1hus, it has led to the íocal point oí this thesis which is to in·estigate whether the
attributes írom the literature re·iew really aííect the consumers` willingness to purchase a
product or ser·ice, when a celebrity is associated with negati·e iníormation.
1he authors came to the conclusion aíter conducting the research process that two oí the
six attributes are the most important attributes a company should consider beíore choosing
and using a celebrity, due to the possibility that a celebrity can get associated with negati·e
iníormation. 1his will íurther on be elaborated on in section 5.3. Beíore tackling and
starting this discussion, the authors will answer the research questions. 1he last section oí
this chapter will, aíter all the diííerent elements within this thesis, present suggestions that
can be undertaken in íuture studies.
5.2 Research questions
!bicb factor. frov tbe titeratvre rerier affect tbe cov.vver.` rittivgve.. to ¡vrcba.e a ¡roavct or .errice
tbat i. evaor.ea b, a cetebrit, tbat i. a..ociatea ritb vegatire ivforvatiov, rbev tbe attribvte. are iv a
covbivatiov.
As it has been pro·en in this thesis, the authors can now conclude that the consumers
purchasing beha·iour in general do get aííected by the mentioned attributes írom the
literature re·iew. 1his is also strengthened in the support oí the hypothesis, which implies
that the attributes in combination really aííect the consumers` willingness to purchase
under conditions oí negati·e iníormation. 1hereíore, the authors íound it ·ery important
to identiíy these crucial attributes. As mentioned in section 4.2.2 expertise, trustworthiness,
likeability and similarity were the main ones.
Marketers need to be aware that consumers percei·e celebrities in diííerent ways based on
what personal characteristics they possess, thus leading to a purchase. \hen a celebrity is in
a particular situation and promoting diííerent products or ser·ices, the consumers tend to
preíer a celebrity who is an expert and trustworthy within that situation. Moreo·er, the
celebrity should preíerably be likeable among the targeted consumers and ha·e similar
goals, interest etc. 1hese attributes in a combination could be the winning concept in a
celebrity endorsement strategy and increased market shares. 1he more knowledge and the
Conclusions and discussion
34
higher acceptance the consumers ha·e regarding the celebrity, when e·aluating the
problematic` endorser, the more likely they still are to make a purchase oí a
product,ser·ice.
Do tbe attribvte. frov tbe titeratvre rerier bare aifferevt iv¡act. ov cov.vver.` rittivgve.. to ¡vrcba.e a
¡roavct or .errice, rbev aifferevt cetebritie. are v.ea a. evaor.er..
1he two diííerent cases with two diííerent celebrities show clearly that consumers react in
·arious ways to celebrities who are associated with negati·e iníormation. It is also crucial to
point out that consumers may percei·e a separate attribute diííerently in comparison to the
six attributes in a combination and in relation to diííerent endorsers. 1he diííerent
e·aluations oí the attributes with ·arious endorsers can be related to the answer oí research
question one, regarding what situation the celebrity is in and what the endorser is
promoting. lence, the attributes írom the literature re·iew ha·e diííerent impacts on the
consumer beha·iour depending on who the endorser is. linally, the authors cannot see a
direct association between the used attributes írom the literature re·iew and the le·el oí
negati·e iníormation.
!bicb attribvte. ao cov.vver. fiva of vo.t iv¡ortavce rbev eratvativg tbev .e¡aratet,.
\hen the consumers e·aluated the attributes separately, the attribute trustworthiness was
the most important one íollowed by expertise and íit`. Similarity and meaning transíer
were oí equal importance and likeability was seen as the attribute which played the least
important role. 1hese íindings together with the chapter summary in section 4.5, the
authors ha·e in a conjunction íound the two most important attributes that consumers
íound to be oí most importance when e·aluating them separately and e·en in
combinations. 1rustworthiness was throughout this thesis indicated as a ·ery important
attribute and expertise was almost oí equal importance. 1hereíore, the authors can
conclude that the consumers preíer a celebrity who has the characteristics oí being an
expert and trustworthy within the íield oí ad·ertising.
5.3 Discussion
Bringing the íindings írom the research questions together, the authors ha·e decided to
pro·ide the readers with a model ,íigure 5-1, in order to gi·e a general apprehension oí
their conclusions. 1his model is a deeper elaboration on íigure 1-2 since the purpose oí this
thesis was to íind the most important attributes that marketers need to consider when
using celebrities in their endorsement strategies. ligure 1-2 did not indicate what íactors oí
a celebrity that is oí main importance but aíter conducting this research, the authors can
íinally conclude that the attributes trustworthiness and expertise are the hidden íactors that
need to be considered when ad·ertisers use the strategy oí celebrity endorsement.
Conclusions and discussion
35
ligure 5-1 1he network oí optimal use oí a celebrity
1he attributes trustworthiness and expertise are combined in a network within the celebrity
endorsement strategy and since the consumers e·aluated them as the most important ones,
the authors assume this model to be as the optimal use oí a celebrity. A company needs to
be aware oí íigure 5-1 and the interaction in order to gain success or e·en pre·ent íailure
when using celebrity endorsers. 1hese attributes are directly linked to the celebrity and
these are what marketers should consider when choosing a particular celebrity. 1hese will
e·entually aííect the company, consumer and brand when a celebrity gets associated with
negati·e publicity. It is important to ha·e in mind that a celebrity who is linked with
negati·e iníormation and that possess these attributes to a ·ery high degree, is less likely to
aííect the consumer beha·iour negati·ely in comparison to the opposite case.
5.4 Final words from the authors
1he main objecti·e oí this thesis was to come to an understanding oí which íactors
consumers íind to be important íor a company to consider when a celebrity gets associated
with negati·e iníormation. 1hese íactors were as mentioned earlier trustworthiness and
expertise. But, the authors claim that only considering these íactors does not eliminate the
other ones that has been used in the research process but rather that the identiíied most
important íactors are the ·ital ones. Now that the purpose is íulíilled, the authors ha·e
gained more insight within the area oí celebrity endorsement and hopeíully pro·ide
marketers with crucial iníormation and knowledge that will beneíit their company and
strengthen their brand equity. It is also crucial to bare in mind that the selection oí a
celebrity should not only include the two most important attributes but also weigh the
remaining íour as a priority, since these may strengthen the perception oí a celebrity in
collaboration with the top listed attributes. Moreo·er, there is no strong and determinant
indication that the íound attributes are the only ones to consider since there may be indeed
other hidden íactors that consumers consider. 1he authors can now claim that this íield is a
continuously ongoing process and one should be clear that there is no correct way oí doing
things, but rather pro·ide helpíul insights on what íactors marketers should consider when
a celebrity gets associated with negati·e publicity.
5.5 Prospects for future researches
Some oí the results in this research process suggest interesting prospects íor íuture
research. 1he íindings írom the post-experiment showed diííerent results compared to the
general understanding oí each attribute. Consumers tend to think and act diííerently and
Celebrity
Company Brand Consumer
1rustworthiness Lxpertise
Conclusions and discussion
36
this is the main reason why the authors ha·e not claimed that the results are the accurate
ones. 1his research pro·ided the readers with a general understanding oí this subject and it
is crucial to íurthermore state that this subject can be narrowed down e·en more, mostly
when it comes to psychological aspects. Potentially, this can lead to new and exiting
íindings that can pro·ide marketers with more understanding and a deep íocus on certain
psychological issues to take into account.
\hat can be interesting to elaborate deeper on is how the negati·e iníormation oí a
celebrity aííects low- and high in·ol·ement products,ser·ices or luxury- and low cost
products,ser·ices. Moreo·er, what happens ií a celebrity in the case oí negati·e
iníormation promotes a brand that do not íit well with the celebrity is also oí peculiar
interest.
Another major area except íor the psychological aspect is the demographical area. 1his
implies that studies regarding perceptions in diííerent countries, ages, sex etc. can be
conducted to clariíy more speciíic issues that can be helpíul and beneíicial íor marketers.
lurthermore, as it has been stated in pre·ious studies, that one cannot predict the liíe span
oí a celebrity and combined with the íindings in this thesis, there are no clear assumptions
ií the consumers` willingness to purchase a product or ser·ice that is promoted by a
celebrity and associated with negati·e iníormation. 1he authors belie·e that a new research
area is de·eloped. 1his one deals with how the e·er changing identity oí a celebrity might
aííect the consumers` willingness to purchase, while some consumers might íind this
change oí the celebrity as something negati·e. 1hus, the authors assume that a potential
research area can be to in·estigate ií and in that case how consumers react to a celebrity`s
identity change during their time in the spotlight and also ií the celebrity might through a
business perspecti·e beneíit more ií they start their own business, which will lead us in to a
new phenomenon oí celebrity entrepreneurship.
Lastly, it would also be interesting to conduct a study within this subject, mainly írom
marketers point oí ·iew or e·en an intertwined study írom both consumers and marketers
perspecti·es. Concluding this, the authors want to make it clear and agree with the íact that
the ad·ertising industry is really eííicient when using celebrities as endorsers due to its
ability to reach a wider audience where the consumers can identiíy themsel·es with. 1he
use oí ad·ertising is and will change in the íuture, but at this time celebrities are the dri·ing
mechanism to successíul ad·ertising within this industry.
Critique oí this research process
3¯
6 Critique of this research process
´ivce tbi. tbe.i. ba. beev covavctea ritb a tive cov.traivt ava tivitea re.ovrce., tbe avtbor. cav covre,
.ererat .etf·critici.v. ava ;vage tbe re.earcb ¡roce.. .o tbat tbe reaaer ritt vot get vi.teaa or ivter¡ret bia.ea
ivforvatiov.
1he authors íeel that the limited time and resources ha·e aííected the methodology. A
qualitati·e study ,íocus groups, could ha·e beneíited this thesis e·en more in the sense that
it could ha·e underpinned the main method used ,quantitati·e,. lurthermore, it was
theoretically supported that at least 50 respondents can be used in a conjoint experiment
and e·en though ¯5 respondents were used in the research process, the authors íeel that a
higher number oí respondents could ha·e beneíited this study to a greater extent and
pro·ided more accurate results. Moreo·er, the R
2
le·el in all combinations ,1-16, 1-8, 9-16,
were on a low le·el which indicates that the obser·ed ·ariability also was low. 1he R
2
le·el
was on an a·erage le·el throughout the results and ií the R
2
le·els were closer to 1 it would
increase the reliability oí the thesis. 1his can be due to the low amount oí respondents.
Another reason íor this can be the number oí combinations oí the conjoint experiment. A
larger amount oí combinations ,cards, and íewer attributes would gi·e less statistical
uncertainty and more signiíicant results.
1he results show that the celebrity-product attribute that ha·e been íound important in
pre·ious research hold up in a context oí a celebrity that is associated with negati·e
iníormation. lowe·er, a more ideal design would ha·e included negati·e iníormation
instead oí as a consequence oí the attributes, allowing íor a more direct comparison.
lurthermore, the authors came to the understanding that as much as conjoint came across
as time consuming, they were going more íor well answered questionnaires then halí done
ones. lurthermore, the authors belie·e that the conjoint experiment is such a well co·ered
research process that many diííerent results could ha·e been elaborated írom it. But, the
authors decided to keep it as accurate to the purpose oí this thesis, which was to íind the
hidden íactors and not in·estigating the diííerent relationships between the diííerent
attributes. 1he concern was more on which attributes the respondents percei·ed to be
most important linked with the purpose oí this thesis. 1his is why the authors only kept the
number oí hypotheses to two instead oí a larger number. 1here was a direct interest to
in·estigate other aspects, mainly ií the consumers were willing to purchase a product or
ser·ice ad·ertised by a celebrity that is associated with negati·e iníormation, and also ií
consumers respond diííerently in reaction to negati·e iníormation or the celebrities
negati·e iníormation.
1he last point oí criticism that is important to point out is the order oí combinations.
\hen looking at appendix 4 A, one can see that the le·el oí importance has a certain
symmetry where it starts oíí with a high le·el and decreases slightly aíter each combination
used. lence, ií the combinations where placed in a diííerent order the outcome might ha·e
diííered and lead to other results.
Reíerences
38
References
Aaker, D. A. ,2000,. ßrava íeaaer.bi¡. A Di·ision oí Simon & Schuster, Inc: New \ork.
Aaker, D. A. ,1991,. Mavagivg ßrava íqvit,. 1he íress press, New \ork.
Aczel, A. D., & Sounderpandian, J. ,2002,. ßv.ive.. ´tati.tic. ,5
th
ed.,. McGraw-lill ligher
Lducation: New \ork.
Antilla, M., leu·el, R. R., & Moller, K. ,1980,. Conjoint Measurement íor Marketing
Management. ívro¡eav ]ovrvat of Mar/etivg. 11,¯,, 39¯-408.
Belch, G.E., & Belch, M.A. (2001). Advertising and Promotion: An integrated
Marketing Communications Perspective (5
th
ed.). Boston: Irwin/MaGraw-
Hill.
Bielli, A. (2003).The Research Power Behind Great Brands. www.millwardbrown.com.
Retrieved 2005-02-15.
Carson, D., Gilmore A,. Perry C & Gronhaug K. ,2001,. Qvatitatire Mar/etivg Re.earcb.
SAGL Publications Ltd. Guildíord: Great Britain.
Chabo, D., & Saouma, J. ,2005,. 1be O¡tivat v.e of Cetebrit, ívaor.evevt. Jonkoping
International Business School: Jonkoping.
Chabo, D., & Saouma, J. ,2005,. 1be íffect of ^egatire ívforvatiov. Jonkoping International
Business School: Jonkoping.
Chabo, D., & Saouma, J. ,2005,. 1be ^etror/ of O¡tivat v.e of a Cetebrit,. Jonkoping
International Business School: Jonkoping.
Churchill, G. A. ,1995,. Mar/etivg Re.earcb· Metboaotogicat íovvaatiov. ,6
th
Ld., New \ork:
1he Dryden Press.
Danesh·ary, R., & Schwer, K.R. ,2000,. 1he Association and Consumers Intention to
Purchase: ]ovrvat of Cov.vver Mar/etivg, 1¨,3,, 203-213.
Dean, l. D. ,1999,. Brand Lndorsement, Popularity, and L·ent Sponsorship as
Ad·ertising Cues Aííecting Consumer Pre-Purchase Attitudes. ]ovrvat of
.arerti.ivg. 2º ,3,.
Bruns, V. ,2004,. !bo Receire. ßav/ íoav..· . .tva, of tevaivg officer.` a..e..vevt. of toav. to
grorivg .vatt ava veaivv·.i¸ea evter¡ri.e.. Jonkoping International Business
School, Jonkoping Uni·eristy.
Lkdahl, l. ,199¯,. ívcrea.ea Cv.tover ´ati.factiov |.ivg De.igv of í·¡erivevt., Cov;oivt .vat,.i.
ava QíD. Linkoping: Di·ision oí Quality 1echnology and Management
Department oí Mechanical Lngineering, Linkoping Uni·ersity.
Lrdogan, Z.B., Baker, M.J., & 1agg, S. ,2001,. Selecting Celebrity Lndorsers: 1he
Practitioner`s Perspecti·e: ]ovrvat of .arerti.ivg Re.earcb, 11, 1-26.
Reíerences
39
Lsaiasson, P., Gilljam, M., Oscarsson, l., & \ängnerud, L. ,2002,. Metoa¡ra/ti/av: Kov.tev
att .tvaera .avbatte, ivairia ocb var/vaa. Stockholm: Nordstedts Juridik.
llick, U. ,2002,. .v ívtroavctiov to Qvatitatire Re.earcb. ,2
nd
ed., SAGL Publications Ltd.
\iltshire: Great Britain.
Goldsmith, L. R, Laííerty, A. B, Newell, J.S. ,2000,. 1he Impact oí Corporate Credibility
and Celebrity Credibility on Consumer Reaction to Ad·ertisements and
Brands. ]ovrvat of .arerti.ivg, 2· ,3,.
Green, P.L., & Srini·asan, V. ,1990,. Conjoint Analysis in Marketing: New de·elopments
with implications íor research and practise. ]ovrvat of Mar/etivg. :1,4,, 3-19.
Green, P.L., & Srini·asan, V. ,19¯8,. Conjoint Analysis in Consumer Research: Issues and
Outlook. ]ovrvat of cov.vver Re.earcb, :: 103-123.
Green, P. L. ,2001,. loreword. In A. Gustaísson, A. lerrmann, & l. luber ,Lds.,,
Cov;oivt vea.vrevevt·vetboa. ava a¡¡ticatiov., ,2
nd
ed., Berlin: Springer Verlag.
Gustaísson, A., lerrman, A., & luber, l. ,2001,. Conjoint Analysis as an Instrument oí
Market Reserach Practice. In A. Gustaísson, A. lerrman, & l. luber ,Lds.,,
Cov;oivt vea.vrvevt: vetboa. ava a¡¡ticatiov., 2
nd
ed. Berlin:Springer Verlag.
Gwinner, K. ,199¯,. . Moaet of ívage Creatiov ava ívage 1rav.fer iv írevt ´¡ov.or.bi¡.
International Marketing Re·iew, 11 ,3,, 145-158. North Carolina: USA.
Iggland, B. ,1989,. .vravaarbeaövivg för ¡roav/tvtrec/tivg· ígev./a¡.raraerivgar vea Cov;oivt
.vat,.i.. Linkoping: Department oí Management and Lconomics: Linkoping
Uni·ersity.
Johnson, R. M. ,198¯,. Adapti·e Conjoint Analysis, in: ´artootb .vat,.i. Covferevce
Proceeaivg., Ketchum, ID: Sawtooth Soítware, July, 253-65.
Kambitsis, C., larahousou, \., 1heodorakis, N., & Chatzibeis, G. ,2002,. Sports
Ad·ertising in Print Media: 1he case oí 2000 Olympic Games: Cor¡orate
covvvvicatiov: .v ivtervatiovat ]ovrvat, ¨,3,, 155-161.
Kamins, A.M. ,1990,. An in·estigation into the match-up` hypothesis in celebrity
ad·ertising: when beauty may be only skin deap. ]ovrvat of .arerti.ivg, 1·:
American Academy oí Ad·ertising.
Kamins, A.M., Brand, J.M., loeke. A.S & Moe, C.J. ,1989,. 1wo-Sided ·ersus One-Sided
Celebrity Lndorsements: 1he Impact on Ad·ertising Lííecti·eness and
Credibility. ]ovrvat of .arerti.ivg, 1º: American Academy oí Ad·ertising.
Keller, L. K. ,2002,. Branding and Brand Lquity. Mar/etivg ´cievce ív.titvte. Cambridge:
USA.
Kotler, P., Armstrong, G., Saunders, J., & \ong, V. ,2001,. Privci¡te. of Mar/etivg. Prentice
lall: larlow.
Korner, S., & \ahlgren, L. ,1998,. ´tati.ti./a vetoaer. Lund: Studentlitteratur.
Reíerences
40
Louie, 1.A., & Obermiller, C. ,2002,. Consumer Response to a lirm`s Lndorser ,Dis,
association Decisions: ]ovrvat of .arerti.ivg, ²1.
Lundell, M. ,2005,. 1¨ Mit;over för ²0 .e/vvaer. Aítonbladet.
lttp:,,www.aítonbladet.se,·ss,ekonomi,story,utskriít,0,3258,600119,00.ht
m. Retrie·ed 2005-02-08.
Malhotra, K. N. ,1996,. Mar/etivg re.earcb: .v a¡¡tiea orievtatiov ,2th ed.,. New
Jersey
rentice lall. McCracken, G. ,1989,. \ho is celebrity endorser· Cultural íoundations oí the celebrity
endorsement process`. ]ovrvat of Cov.vver Re.earcb, 1ó,3,. 310-21.
O´Mahony, S. & Meenaghan, 1. ,199¯,98,. 1he impact oí Celebrity Lndorsements on
Consumers: íri.b Mar/etivg Rerier, 10 ,2,, 15-24.
Ohanian, R. ,1990,. Construction and Validation oí a scale to measure celebrity endorsers`
percei·ed expertise, trustworthiness and attracti·eness. ]ovrvat of .arerti.ivg, 1·:
American Academy oí Ad·ertising.
Pringle, l. ,2004,. Celebrity Sells. ]obv !ite, c ´ov.. \iltshire: UK.
Redenbach, A. ,2005,. A Multiple Product Lndorser Can Be A Credible Source. C,ber·
]ovrvat of ´¡ort Mar/etivg. ISSN. 132¯-6816. Griííith Uni·ersity.
Riezebos, R., Kist, B., Koostra. G. (2003), Brand Management. A theoretical and
practical approach. Prentice Hall.
Saunders, M., Lewis P., & 1hornhill, A. ,2003,. Re.earcb vetboa. for bv.ive.. .tvaevt., Person
education limited: larlow.
Schiííman, G. Leon & Kanuk, L. Lazar. ,2004,. Cov.vver ßebariovr. Pearson education inc:
New Jersey.
Schlecht, C. ,2003,. Celebrities` Impact on Branding. Cevter ov Ctobat ßrava íeaaer.bi¡.
Columbia Business School: New \ork.
Se·ern, J., Belch G.L. & Belch M.A. ,1990,. 1he eííects oí sexual and non-sexual
ad·ertising appeals and iníormation le·el on cogniti·e processing and
communication eííecti·eness. ]ovrvat of aarerti.ivg, 1·1, 14-22.
Sil·era, D.l., & Austad. B. ,2004,. lactors predicting the eííecti·eness oí celebrity
endorsement ad·ertisements: ívro¡eav ]ovrvat of Mar/etivg, ²º,11,12,, 1509-
1526.
Shepherd, D. A., & Zacharakis, A. ,199¯,. Conjoint analysis: A window oí opportunity íor
entrepreneurship research. .aravce. iv ívtre¡revevr.bi¡, íirv ívergevce, ava
Crortb, ², 203-248.
Smith, A.L., Schullen, L.S., & Barr, l.S. ,2002,.Orgavi¸atiovat Re.earcb Metboa., : ,4,,
October 2002, 388-414.
Reíerences
41
Solomon, M.R. ,2002,. Cov.vver ßebarior: ßv,ivg, íarivg,ava ßeivg, 5
th
ed.Prentice lall: New
Jersey.
Soderlund, M. ,2003,. ívotiov.taaaaa Mar/vaa.förivg. Liber: Malmo.
1ang, L.P., Kim, K.J & 1ang, L.1. ,2001,. Lndorsement oí the Money Lthic, Income, and
Liíe Satisíaction - A comparison oí íull-time employees, part-time employees,
and non-employed uni·ersity students. ]ovrvat of Mavageriat P.,cbotog,, 1¨ ,6,,
1ennessee: USA.
1ill, B.D., & Schimp, 1.A. ,1998,. Lndorsers in Ad·ertising: 1he case oí Negati·e Celebrity
Iníormation: ]ovrvat of .arerti.ivg, 2¨.
1ill, B.D., & Busler, M. ,1998,. Matching products with Lndorsers: Attracti·eness ·ersus
Lxpertise: ]ovrvat of Cov.vver Mar/etivg, 1: ,6,, 5¯6-586.
1hurrot, P. ,2004,. \ith iPod success, Apple takes network eííect` írom Microsoít.
lttp:,,www.windowsitpro.com,Articles,Print.cím·ArticleID~443¯6.
Retrie·ed 2005-03-11.
Van de Pol, M., & Ryan, M. ,1996,. Using Conjoint analysis to establish consumer
preíerences íor íruit and ·egetables. ßritti.b íooa ]ovrvat, ·º,8,, 5-12.
\algren, J. C, Ruble, A.C & Donthu, N. ,1995,. Brand Lquity, Brand Preíerence &
Purchase Intentions. ]ovrvat of .arerti.ivg, 21 ,3,.
\heeler, R. ,2003,. Choosing Celebrity endorsers: 1ips and traps: ^ov¡rofit !orta., 21,4,,
1¯-20.
42
Appendix 1- 8 profiles for the six attributes at two levels
Lxpertise 1rust Similarity Likeability Meaning
transíer
Match
íit`
Concept
Low Low Low Low Low Low Design 1
ligh ligh Low Low ligh Low Design 2
Low ligh ligh Low ligh ligh Design 3
ligh Low ligh Low Low ligh Design 4
Low ligh Low ligh Low ligh Design 5
ligh Low Low ligh ligh ligh Design 6
ligh ligh ligh ligh Low Low Design ¯
Low Low ligh ligh ligh Low Design ¯
43
Appendix 2 - Research Survey
Dear re.¡ovaevt!
\ou ha·e been chosen to take part oí our sur·ey based on our Master`s thesis in marketing.
1his thesis has its main íocus on celebrity endorsement and as you might ha·e noticed, this
concept is being more írequently used by marketers in the ad·ertising industry.
lurthermore, most oí the celebrities that ha·e been used ha·e generated a high brand
exposure and ha·e been associated with a certain brand, which in terms ha·e aííected the
consumer buying beha·iour.
1he use oí celebrity endorsement has mostly a positi·e outcome íor companies, since the
celebrity`s characteristics íit well with the characteristics oí a brand. Most oí the celebrities
used by companies ha·e been chosen since they are considered to ha·e a strong power on
consumers. 1his means that consumers may buy a certain brand that is associated with a
celebrity to identiíy themsel·es with this particular brand and also íeel a sense oí similarity
with the celebrity. lowe·er, one can wonder how consumers get aííected by celebrities
that ha·e been associated with negati·e publicity. 1his is what we want you to consider
when rating the different combinations of attributes and how the negative publicity
of a celebrity affects YOU.
\e will pro·ide you with two diííerent cases in order íor you to get a better understanding
oí how to rate when imagining that particular situation, and how that would eííect you in
your buying decision. \e want to emphasize that the generated cases are made up. Simply,
what we want from you is to rate each combination of attributes from J to 7. Please
e·aluate each combination as a separate situation independent oí all others. Once you ha·e
completed one combination, you are not supposed to go back and take a pre·ious look.
\ou should also read through the descriptions oí the attributes and their le·el one time
only. Moreo·er, ií you consider one combination not to ha·e a great impact on you, you
can rate it as number 1 and írom that point rate the combinations a higher number the
more impact they ha·e on you with ¯ being the highest.
\e would like to make it clear that there is no right or wrong answer, we are more
interested in what you consider to be most important when a celebrity gets associated with
negati·e publicity. With negative publicity we mean any information that has
negative association that decreases the trustworthiness of the celebrity and the fit
with the brand.
1he whole experiment might be considered as time consuming but it will probably not take
you more than 15 minutes to complete this task. linally, we emphasize that your response
will remain anonymous and thank you íor your participation.
44
Attribute Level J Level 2
í·¡erti.e: 1he le·el oí
knowledge, experience
and expertise a celebrity
has in a particular brand.
íigb: 1he celebrity
has a high le·el oí
knowledge,
experience and
expertise.
íor: 1he celebrity has a
low le·el oí knowledge,
experience and expertise.
1rv.trortbive..: low
honest and belie·able a
celebrity is when
ad·ertising a brand.
íigb: 1he celebrity
is considered to be
·ery trustworthy.
íor: 1he celebrity is not
percei·ed to be honest
and belie·able enough.
´ivitarit,: 1he sense oí
similarity a consumer íeel
with a celebrity endorser.
íigb: 1he consumer
íeels highly linked
with the celebrity
when it comes to
the characteristics.
íor: 1he consumer does
not íeel any connection
with the celebrity.
íi/eabitit,: 1he le·el oí a
celebrity`s íame in the
public eye and how
popular this person is.
íigb: 1he celebrity
is highly íamous
and popular.
íor: 1he celebrity is not
likeable or popular.
Meavivg trav.fer: 1he
meaning a celebrity brings
or transíers to a certain
brand with the help oí
their personality and liíe-
styles.
íigb: 1he celebrity
is successíully
transíerring
meaning to a brand.
íor: 1he celebrity does
not succeed in
transíerring any meaning
to the brand.
íit Matcb: low well the
íit between the celebrity
and the brand is.
íigb: A ·ery good
íit.
íor: No íit at all.
45
Case J. Plastic Iantastic AB
1he last two years, Plastic lantastic AB has been sponsoring the most amazing and
extreme make-o·er 1V-show ,1he Swan, íor regular people in the USA. 1his has been
·ery successíul in the domestic market as well as internationally. 1hereíore, Plastic
lantastic has now realized that this can be a beneíicial way to reach out to a larger audience
in a diííerent way than the original 1V-show promotion. 1alking about 1V-shows, Michael
Jackson is currently not only notorious íor attending his own court on international 1V,
where he is accused oí child molesting, but he is also promoting the ser·ices Plastic
lantastic AB oííers.
Combination J
Lxpertise Low
1rustworthiness Low
Similarity Low
Likeability Low
Meaning transfer Low
´Iit¨ Match Low
Based on the attributes mentioned abo·e, how willing are you to buy this ser·ice·
,Please rate this combination,
Unwilling 1 2 3 4 5 6 ¯ \illing
46
Combination 2
Lxpertise High
1rustworthiness High
Similarity Low
Likeability Low
Meaning transfer High
´Iit¨ Match Low
Based on the attributes mentioned abo·e, how willing are you to buy this ser·ice·
,Please rate this combination,
Unwilling 1 2 3 4 5 6 ¯ \illing
4¯
Combination 3
Lxpertise Low
1rustworthiness High
Similarity High
Likeability Low
Meaning transfer High
´Iit¨ Match High
Based on the attributes mentioned abo·e, how willing are you to buy this ser·ice·
,Please rate this combination,
Unwilling 1 2 3 4 5 6 ¯ \illing
48
Combination 4
Lxpertise High
1rustworthiness Low
Similarity High
Likeability Low
Meaning transfer Low
´Iit¨ Match High
Based on the attributes mentioned abo·e, how willing are you to buy this ser·ice·
,Please rate this combination,
Unwilling 1 2 3 4 5 6 ¯ \illing
49
Combination S
Lxpertise Low
1rustworthiness High
Similarity Low
Likeability High
Meaning transfer Low
´Iit¨ Match High
Based on the attributes mentioned abo·e, how willing are you to buy this ser·ice·
,Please rate this combination,
Unwilling 1 2 3 4 5 6 ¯ \illing
50
Combination 6
Lxpertise High
1rustworthiness Low
Similarity Low
Likeability High
Meaning transfer High
´Iit¨ Match High
Based on the attributes mentioned abo·e, how willing are you to buy this ser·ice·
,Please rate this combination,
Unwilling 1 2 3 4 5 6 ¯ \illing
51
Combination 7
Lxpertise High
1rustworthiness High
Similarity High
Likeability High
Meaning transfer Low
´Iit¨ Match Low
Based on the attributes mentioned abo·e, how willing are you to buy this ser·ice·
,Please rate this combination,
Unwilling 1 2 3 4 5 6 ¯ \illing
52
Combination 8
Lxpertise Low
1rustworthiness Low
Similarity High
Likeability High
Meaning transfer High
´Iit¨ Match Low
Based on the attributes mentioned abo·e, how willing are you to buy this ser·ice·
,Please rate this combination,
Unwilling 1 2 3 4 5 6 ¯ \illing
53
Case 2. Dell and Bill
Dell computers are launching their new product, Monica 69 \l, which includes a certain
system íor managers` at large companies to enhance the process oí keeping track oí
numbers. 1his indicates that Dell`s main target is business-oriented people. 1o be able to
succeed with reaching this particular group, Dell ha·e been using Bill Clinton as a
spokesperson in promoting Monica 69 \l.
Combination 9
Lxpertise Low
1rustworthiness High
Similarity High
Likeability Low
Meaning transfer High
´Iit¨ Match High
Based on the attributes mentioned abo·e, how willing are you to buy this product·
,Please rate this combination,
Unwilling 1 2 3 4 5 6 ¯ \illing
54
Combination J0
Lxpertise Low
1rustworthiness Low
Similarity Low
Likeability Low
Meaning transfer Low
´Iit¨ Match Low
Based on the attributes mentioned abo·e, how willing are you to buy this product·
,Please rate this combination,
Unwilling 1 2 3 4 5 6 ¯ \illing
55
Combination JJ
Lxpertise Low
1rustworthiness High
Similarity Low
Likeability High
Meaning transfer Low
´Iit¨ Match High
Based on the attributes mentioned abo·e, how willing are you to buy this product·
,Please rate this combination,
Unwilling 1 2 3 4 5 6 ¯ \illing
56
Combination J2
Lxpertise High
1rustworthiness High
Similarity High
Likeability High
Meaning transfer Low
´Iit¨ Match Low
Based on the attributes mentioned abo·e, how willing are you to buy this product·
,Please rate this combination,
Unwilling 1 2 3 4 5 6 ¯ \illing
5¯
Combination J3
Lxpertise High
1rustworthiness Low
Similarity Low
Likeability High
Meaning transfer High
´Iit¨ Match High
Based on the attributes mentioned abo·e, how willing are you to buy this product·
,Please rate this combination,
Unwilling 1 2 3 4 5 6 ¯ \illing
58
Combination J4
Lxpertise High
1rustworthiness High
Similarity Low
Likeability Low
Meaning transfer High
´Iit¨ Match Low
Based on the attributes mentioned abo·e, how willing are you to buy this product·
,Please rate this combination,
Unwilling 1 2 3 4 5 6 ¯ \illing
59
Combination JS
Lxpertise Low
1rustworthiness Low
Similarity High
Likeability High
Meaning transfer High
´Iit¨ Match Low
Based on the attributes mentioned abo·e, how willing are you to buy this product·
,Please rate this combination,
Unwilling 1 2 3 4 5 6 ¯ \illing
60
Combination J6
Lxpertise High
1rustworthiness Low
Similarity High
Likeability Low
Meaning transfer Low
´Iit¨ Match High
Based on the attributes mentioned abo·e, how willing are you to buy this product·
,Please rate this combination,
Unwilling 1 2 3 4 5 6 ¯ \illing
61
How important do you consider each of the attributes mentioned to be of
importance when considering a celebrity that gets negative publicity.
,Please e·aluate each attribute on the scale oí 1 to 5,
Lxpertise
Unimportant 1 2 3 4 5 Important
1rustworthiness
Unimportant 1 2 3 4 5 Important
Similarity
Unimportant 1 2 3 4 5 Important
Likeability
Unimportant 1 2 3 4 5 Important
Meaning transfer
Unimportant 1 2 3 4 5 Important
´Iit¨ Match
Unimportant 1 2 3 4 5 Important
62
,Please put a cross on the situation that is addressed to each question,
J. Which faculity do you belong to?
2. Do you consider yourself to be fashion conscious?
3. Do you consider yourself to be affected by celebrities in advertising?
4. Do you purchase brands that are advertised by celebrities?
S. Does your purchasing decisions change if a celebrity get assocaiated with
negative publicity?
JIBS ING lLK llJ
\es No Occassionally
\es No Occassionally
\es No Occassionally
\es No
63
Appendix 3- Extract from pilot study
Combination J
Lxpertise Low
1rustworthiness Low
Similarity Low
Likeability Low
Meaning transfer Low
´Iit¨ Match Low
Based on the attributes mentioned above, how important do you find the celebrity's
negative publicity affecting your purchasing decision.
,Please rate this combination,
Low importance 1 2 3 4 5 6 ¯ ligh
importance
64
Appendix 4 - Appendix concerning reserach question 1
Appendix 4A
Importance summary
Factor
Match
meaning transf er
likeability
similarity
trsutworthiness
expertise
A
v
e
r
a
g
e
d
I
m
p
o
r
t
a
n
c
e
30
20
10
0
Appendix 4B
Individual Subject Importance
Factor
Match
meaning transf er
likeability
similarity
trsutworthiness
expertise
A
v
e
r
a
g
e
d
I
m
p
o
r
t
a
n
c
e
60
50
40
30
20
10
0
65
66
67
68
69
70
71
72
73
74
75
65
Appendix 4C
CARD12
7,0 6,0 5,0 4,0 3,0 2,0 1,0
CARD12
F
r
e
q
u
e
n
c
y
30
20
10
0
Std. Dev = 1,45
Mean = 4,9
N = 75,00
66
Appendix 5- Appendix concerning reserach question 2
case 1
Appendix SA
Importance summary
Factor
Match
meaning transf er
likeability
similarity
trsutworthiness
expertise
A
v
e
r
a
g
e
d
I
m
p
o
r
t
a
n
c
e
30
20
10
0
Appendix SB
Individual Subject Importance
Factor
Match
meaning transf er
likeability
similarity
trsutworthiness
expertise
A
v
e
r
a
g
e
d
I
m
p
o
r
t
a
n
c
e
80
60
40
20
0
65
66
67
68
69
70
71
72
73
74
75
6¯
Appendix 6- Appendix concerning reserach question 2
case 2
Appendix 6A
Importance summary
Factor
Match
meaning transf er
likeability
similarity
trsutworthiness
expertise
A
v
e
r
a
g
e
d
I
m
p
o
r
t
a
n
c
e
30
20
10
0
Appendix 6B
Individual Subject Importance
Factor
Match
meaning transf er
likeability
similarity
trsutworthiness
expertise
A
v
e
r
a
g
e
d
I
m
p
o
r
t
a
n
c
e
80
70
60
50
40
30
20
10
0
65
66
67
68
69
70
71
72
73
74
75
68
Appendix 7- Anova tables
Reserach question J
Model Summary(b) Appendix 7A
Model R R-square
Adjusted R-
square
Std. Error of
the Estimate
1
,562(a) ,316 ,128 20,357
a Predictors: (Constant), CARD16, CARD10, CARD7, CARD3, CARD12, CARD11, CARD1, CARD14,
CARD13, CARD5, CARD8, CARD2, CARD6, CARD4, CARD15, CARD9
b Dependent Variable: RESP
ANOVA(b) Appendix 7B
Model
Sum of
Squares df Mean Square F Sig.
Regressio
n
11114,608 16 694,663 1,676 ,078(a)
Residual
24035,392 58 414,403
1
Total
35150,000 74
a Predictors: (Constant), CARD16, CARD10, CARD7, CARD3, CARD12, CARD11, CARD1, CARD14,
CARD13, CARD5, CARD8, CARD2, CARD6, CARD4, CARD15, CARD9
b Dependent Variable: RESP
Research question 2 case J
Model Summary Appendix 7C
Model R R-square
Adjusted R-
square
Std. Error of
the Estimate
1
,430(a) ,185 ,086 20,832
a Predictors: (Constant), CARD8, CARD7, CARD5, CARD4, CARD2, CARD1, CARD3, CARD6
ANOVA(b) Appendix 7D
Model
Sum of
Squares df Mean Square F Sig.
Regressio
n
6507,286 8 813,411 1,874 ,079(a)
Residual
28642,714 66 433,981
1
Total
35150,000 74
a Predictors: (Constant), CARD8, CARD7, CARD5, CARD4, CARD2, CARD1, CARD3, CARD6
b Dependent Variable: RESP
Research question 2 case 2
69
Model Summary Appendix 7E
Model R R-square
Adjusted R-
square
Std. Error of
the Estimate
1
,506(a) ,256 ,166 19,909
a Predictors: (Constant), CARD16, CARD10, CARD12, CARD14, CARD15, CARD11, CARD13, CARD9
ANOVA(b) Appendix 7F
Model
Sum of
Squares df Mean Square F Sig.
Regressio
n
8988,430 8 1123,554 2,834 ,009(a)
Residual
26161,570 66 396,387
1
Total
35150,000 74
a Predictors: (Constant), CARD16, CARD10, CARD12, CARD14, CARD15, CARD11, CARD13, CARD9
b Dependent Variable: RESP
¯0
Appendix 8- Perception of attributes
Rating
5,0 4,0 3,0 2,0 1,0
Expertise
F
r
e
q
u
e
n
c
y
40
30
20
10
0
Std. Dev = 1,25
Mean = 4,0
N = 75,00
Rating
5,0 4,0 3,0 2,0 1,0
Trustworthiness
F
r
e
q
u
e
n
c
y
50
40
30
20
10
0
Std. Dev = ,93
Mean = 4,3
N = 75,00
¯1
Rating
5,0 4,0 3,0 2,0 1,0
Similarity
F
r
e
q
u
e
n
c
y
30
20
10
0
Std. Dev = 1,10
Mean = 3,4
N = 75,00
Rating
5,0 4,0 3,0 2,0 1,0
Likeability
F
r
e
q
u
e
n
c
y
30
20
10
0
Std. Dev = 1,06
Mean = 3,0
N = 75,00
¯2
Rating
5,0 4,0 3,0 2,0 1,0
Meaning transfer
F
r
e
q
u
e
n
c
y
30
20
10
0
Std. Dev = 1,24
Mean = 3,4
N = 75,00
Rating
5,0 4,0 3,0 2,0 1,0
"Fit" Match
F
r
e
q
u
e
n
c
y
30
20
10
0
Std. Dev = 1,10
Mean = 3,6
N = 75,00
doc_800796532.pdf