‘ wizymztiza . . u , V v
a S. .. r5352 figmm. :.
i ‘ .rznsnvmu...‘

! 313500.51

‘ ..
ifizgt n44“.
. .‘ll.a\.“..vr .r
1 1:33.!
2 o... ‘

is
m.

 

 

$.23 «ii-row.
‘t. x!
1.0.12!ng-

A‘I}. . vi 3 .5...
I41|I~n¢| 6'2:

 

z LIBRARY
q? 5‘} Michigan §tate
’ Unnversnty

This is to certify that the
dissertation entitled

A CONTINGENCY FRAMEWORK FOR GLOBAL BRANDING:

A MULTl-LEVEL INTERACTION MODEL

presented by

BURCU TASOLUK

has been accepted towards fulfillment
of the requirements for the

PhD. degree in Business Administration

 

 

 

WIajoH’rofessor’s Signature

0'7/ll/2D06

Date

 

MSU is an Affirmative Action/Equal Opportunity Institution

.,-.-.-.-.-.-._.—.-.Q.-V—.-.-.—.-.-...

PLACE IN RETURN BOX to remove this checkout from your record.
TO AVOID FINES return on or before date due.
MAY BE RECALLED with earlier due date if requested.

 

DATE DUE DATE DUE DATE DUE

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

2/05 p2/CIRC/Date0ue,indd-p.1

 

A CONTINGENCY FRAMEWORK FOR GLOBAL BRANDING:
A MULTI-LEVEL INTERACTION MODEL

By

Burcu Tasoluk

A DISSERTATION

Submitted to
Michigan State University
in partial fulfillment of the requirements
for the degree of

DOCTOR OF PHILOSOPHY

Department of Marketing and Supply Chain Management

2006

ABSTRACT
A CONTINGENCY FRAMEWORK FOR GLOBAL BRANDING:
A MULTI-LEVEL INTERACTION MODEL

By

Burcu Tasoluk

This dissertation builds and tests a contingency model for global branding and explores
the factors which influence the likelihood of global brand ownership. Drawing on the
contingency approach from strategic management, this study expands our understanding
of both specific dimensions of global brands which appeal to consumers across countries
and specific characteristics of consumers across countries who are most receptive to
global brands.

Global brands are among the most important intangible assets a company can
have. According to the Interbrand/Business Week Study, the financial value of the top
100 global brands exceeds $1 trillion in 2004. In line with its importance, many
researchers have investigated different aspects of global branding. Still, the literature
lacks a framework that analyzes global brand perceptions of consumers within the
broader development context of these perceptions.

This thesis develops a contingency model where the relationship between the
global brand perceptions of consumers and the likelihood of global brand ownership is
hypothesized to be dependent on consumer characteristics. Consumer characteristics
include both demographic (age and income) and psychographic (personal values)

variables. Schwartz’s value system is used to classify consumers along two bipolar

dimensions: self-enhancement versus self-transcendence and openness to change versus
conservation. The national economic development of the countries (GDP per capita) is
included as a macro-level variable to explain the variation in the characteristics of global
brand-prone consumers across countries. By incorporating both consumer-level and
country-level variables into a single model, the interactions between micro and macro
units are analyzed. Hierarchical Linear Modeling is employed as the research method to
deal with the existence of multiple units of analysis in the same model and to show how
variables at one level influence relationships occurring at another. The hypotheses are
empirically tested by using data from more than 31,000 consumers across 31 countries.
This study is an in-depth consumer research study with important managerial
implications. It will not only reveal how various dimensions of global brands relate to the
likelihood of global brand ownership contingent on consumer characteristics, but also
will help to identify the consumer characteristics that define consumers most receptive to

global brands.

Copyright by
BURCU TASOLUK
2006

This dissertation is dedicated to my Mom and Dad, who have always supported me for
going after my dreams; and to my brother, who has always brought joy and happiness to

my life: I love you all very much.

ACKNOWLEDGEMENTS

I would like to acknowledge the guidance and support of many people who have
been with me throughout this long Ph.D. process. My greatest thanks goes to my advisor,
Roger Calantone. He has been a patient mentor to me and he was with me in every step
of the way. Whenever I got nervous, he knew how to calm me down and make things
seem a lot less complicated under a different light. He always pushed me harder though,
so that I would exceed my limits. As a researcher, he is always after excellence; as a
mentor, he knows how to motivate and encourage; and as a person, he is very
understanding and considerate. I learned a lot from him. Tamer Cavusgil also deserves
special acknowledgment; he was the reason why I applied to MSU in the first place and
his support gave me confidence at the hardest times of this journey. Cornelia Droge is a
very smart and capable researcher; I would like to thank her for the care and attention she
showed toward my work, as well as for her sincere efforts to teach. I am grateful to David
Griffith for his recommendations of the positioning of my research, as well as for
preparing me for my job search and job interviews. I’d also like to thank Regina McNally
and Chris White for spending their time providing valuable insight into the best way of
presenting my work at job interviews.

I am so grateful to our department secretaries: Kathy Mullins, Kathy Waldie,
Sherry Danner, Laurie Pitch, and Cheryl Lundeen. They were so helpful during the most

stressful times. They always smile and make you feel good.

vi

My fellow classmates know very well what it takes to spend such a long time with
such challenges in a Ph.D. program. It was fun to be with them and knowing that you
would be understood when you complain. Thou shant be alone in tough times ©

I’d like to especially acknowledge and thank my dear friends N ilufer Agirdir,
from Istanbul, and Nazan Ugur and Halide Sayar, from Brussels. Their love and support
was very important to get me going. Whenever I needed a friend, they were there ready to
help me. They taught me that friendships can stay alive even across continents and
physical distance is not important if you want to be near your friend. The friends I made
here in the US. were also very supportive and made these years pass a lot more
pleasantly and with more fun than what would have been without them. I am very glad
that I met them. They are certainly among the most important rewards that I gained in
these four years.

My parents Ayse Tasoluk and Cevdet Tasoluk, and my brother, Berker Tasoluk,
to whom this dissertation is dedicated, made me feel their love and support always
nearby. I know that they would prefer me to be not so far away, but they respected the
path I chose, and once I was on the way, they supported me always. Words cannot
express my gratitude towards them.

I am very glad that l undertook this mission. I am very glad that I completed it. It
was an experience that taught me so many things, not only academically, but also about

life. Thank you to everyone who has helped me achieve this.

vii

TABLE OF CONTENTS

LIST OF TABLES .............................................................................................................. x
LIST OF FIGURES ........................................................................................................... xi
Chapter 1 ............................................................................................................................. 1
INTRODUCTION .............................................................................................................. 1
1.1 Background ......................................................................................................... 1
1.2 Gaps in Previous Research .................................................................................. 2
1.3 Purpose of the Study ........................................................................................... 3
1.4 Methodological Basis for Answering Research Objectives ................................ 4
1.5 Research Questions Addressed ........................................................................... 5
1.6 Summary and Overview of Remaining Chapters ............................................... 5
Chapter 2 ............................................................................................................................. 7
REVIEW OF LITERATURE AND HYPOTHESIS DEVELOPMENT ............................ 7
2.1 Overview of Global Branding Research ................................................................... 7
2.2 Theoretical Foundation and Hypothesis Development ........................................... 20
2.2.1 Associative Network Memory Model .............................................................. 20
2.2.2 Contingency View ........................................................................................... 22
2.2.3 Consumer Demographics ................................................................................. 22
2.2.4 Consumer Psychographics: Personal Values Research and Schwartz’s Values
Theory ....................................................................................................................... 30
Chapter 3 ........................................................................................................................... 39
RESEARCH METHOD .................................................................................................... 39
3.1 Unit of Analysis ...................................................................................................... 39
3.2 Hierarchical Linear Modeling (HLM) .................................................................... 41
3.3 Hierarchical Generalized Linear Models (HGLM) ................................................. 43
3.4 Data ......................................................................................................................... 47
3.5 Model Variables and Their Descriptions ................................................................ 48
3.6 Model Specification ................................................................................................ 53
3.6.1 Fully Unconditional Model .............................................................................. 53
3.6.2 Level-1 Conditional Model .............................................................................. 57
3.6.3 Level-2 Unconditional Model .......................................................................... 57
3.6.4 Level-2 Conditional Model .............................................................................. 58
3.6.5 Mixed Model .................................................................................................... 59

viii

Chapter 4 .......................................... 6O

RESEARCH RESULTS ................................................................................................... 60
4.1 Descriptive Statistics ............................................................................................... 60
4.2 Estimated Results .................................................................................................... 63

4.2.1 Direct Effects ................................................................................................... 63
4.2.2 Moderating Effects ........................................................................................... 65
4.2.3 Random Effects ................................................................................................ 71
4.3 Interpretation of the Estimated Coefficients ........................................................... 77

Chapter 5 ........................................................................................................................... 86

DISCUSSION OF CONTRIBUTIONS, FUTURE DIRECTIONS AND LIMITATIONS

........................................................................................................................................... 93
5.1 Discussion of Contributions and Managerial Implications ..................................... 93
5.2 Limitations and Future Research Directions ......................................................... 101

References .......................................................................................... 103

ix

LIST OF TABLES

Table 2.1 — Schwartz’s 10 Value Types ........................................................................... 32
Table 3.1 — Description of Variables ................................................................................ 49
Table 3.2 — Value Domains .............................................................................................. 51
Table 3.3 — Value Poles .................................................................................................... 52
Table 3.4 — List of Interaction Terms ............................................................................... 52
Table 3.5 — Estimation of Variance Components ............................................................. 58
Table 4.1 — Descriptive Statistics ...................................................................................... 60
Table 4.2 — Direct Effects ................................................................................................. 63
Table 4.3 - Moderating Effects of Consumer Characteristics .......................................... 66
Table 4.4 — Moderating Effects of Perceived Brand Globalness (PBG) .......................... 70
Table 4.5 — Random Effects .............................................................................................. 71
Table 4.6 — Cross-level Interactions ................................................................................. 73
Table 4.7 — Final Estimation of Variance Components .................................................... 73
Table 4.8 — Significant Model Coefficients ...................................................................... 8O

LIST OF FIGURES

Figure 2.1 — The Model .................................................................................................... 19
Figure 2.2 — Schwartz’s Circumplex Model of Values ..................................................... 33
Figure 4.1 — Relationship between RC and the likelihood of GBO across Countries ...... 72
Figure 4.2 — Relationship between Age and the likelihood of GBO across Countries ..... 74
Figure 4.3 — Relationship between RSE and the likelihood of GBO across Countries 75
Figure 4.4 - Relationship between Income and the likelihood of CEO across Countries 76

Figure 4.5 - Model Graphs for Significant Moderation Relationships ............................ 86

xi

Chapter 1

INTRODUCTION

1.1 Background

Global brands can be defined as brands that consumers can find under the same
name in multiple countries; these brands have generally similar and centrally coordinated
marketing strategies (Steenkamp et al. 2003; Yip 1992).

The key supply-side trends that are responsible for the rise of global branding are
identified by researchers as economies of scale effects, exploitation of experience curves
and knowledge transfers, capitalizing on geographic image spillover, seeking new growth
opportunities, consolidation and globalization of the retail trade, and decreased time to
market in the absence of local modifications (Gregory and Wiechmann 2002; Neff 1999;
Yip 1992). The demand-side factors suggested by researchers are the emergence of a
global consumer culture and global consumer segments (Dawar and Parker 1994; Hassan
and Katsanis 1994; Kapferer 1997; Levitt 1983; Walker 1996), and consumer preference
for brands with a global image (Aaker 1991; Kapferer 1997; Shocker et al. 1994;
Steenkamp et al. 2003; Yip 1992).

Several scholars suggest that the perception of a brand as global enhances brand
equity (Gregory and Wiechmann 2002; Kapferer 1997; Shocker et al. 1994). Indeed,
Alden, Steenkamp, and Batra (1999) found that corporations take advantage of image-

enhancing effects by positioning brands as “global” in their communications.

1.2 Gaps in Previous Research

The global branding literature is fragmented. Divergent topics have been
investigated by researchers in this literature. Main research streams can be summarized as
standardization/adaptation of global brands (usually in the broader context of marketing
strategy), country of origin and brand origin studies, global brand management strategy
and organizational coordination issues, global brand personality studies, cultural
influences of global brands, global brand associations, global brand image and
positioning strategies, global brand equity measurement issues, corporate social
responsibility of global brands, and many case studies investigating individual global
branding cases.

A niche, but very important area that has been recently investigated, is to what
extent globalness of a brand actually matters and what specific global brand associations
consumers value worldwide. Global branding practices undertaken by companies rest on
the widely held belief that globalness of a brand signals certain characteristics to
consumers. The first empirical research on this topic was presented at MSI’s Global
Branding Conference in Milan (Batra et al. 2000b). Later, the same researchers published
a paper which investigated consumers’ perceptions about global brands (Steenkamp et al.
2003), where they tested the effect of perceived brand globalness (PBG) on brand
purchase likelihood. Their findings show that PBG is positively associated with both
perceived brand quality and brand prestige, and higher perceived quality and prestige are
key sources of enhanced consumer value for global brands. The direct effect of perceived

brand globalness on brand purchase likelihood was not found significant.

The other empirical study on dimensions of global brands was conducted by Holt,
Quelch, and Taylor (2004a; 2004b). Their empirical analyses show that most people
prefer one global brand over another because of differences in the brands' global
dimensions. They identified the four dimensions of global brands as quality, global status
symbol, best-in-class, and social responsibility.

Lastly, J ohansson and Ronkainen (2005) tested the effect of globalness of a brand
on brand esteem and found that a brand’s globality is associated with greater esteem after
controlling for perceived quality and familiarity. They concluded that the globality of a
brand had a definite role in building sustainable competitive advantage.

These pioneer studies identify a number of global brand associations and imply
that the globalness of a brand matters; however, they do not state for what type of
consumers. The literature lacks a framework that analyzes global brand perceptions of
consumers within the broader development context of these perceptions. Furthermore,
although researchers have suggested that consumers may choose global brands because
they want to be part of a special culture, global consumer culture (Alden et al. 1999), the
argument that there are consumers who buy global brands consistently has not been

empirically tested, either.

1.3 Purpose of the Study

This study uses recent global branding studies as a starting point and introduces
consumer-level and country-level contingencies. The main premise is that the effect of

different global brand associations on the likelihood of global brand ownership will be

different for different types of consumers. Furthermore, the assertion that there is a
specific segment of consumers that is most prone to global brands will be empirically
tested. The robustness of this “global brand-prone” consumer segment across countries

will also be investigated.

1.4 Methodological Basis for Answering Research Objectives

The model proposed in this study includes variables that belong to multiple units
of analysis: global brand perceptions of consumers and consumer characteristics are the
level-l units; these are nested within countries, the level-2 units.

Two basic statistical assumptions, independence of observations and
homoscedasticity, are violated when different units of analysis are combined in the same
model. Hierarchical Linear Modeling (HLM), also known as multilevel modeling, was
developed to deal with such hierarchical (nested) data. Each of the levels in the data
structure has its own sub-model, which indicates the relationships among variables within
a given level, and specifies how variables at one level influence relations occurring at
another (Raudenbush and Bryk 2002).

Although the nested data structure is quite common in marketing and international
business, HLM is not a familiar method of analysis in these disciplines. Scholars from
sociology, biometrics, econometrics, statistics, and education research historically have
employed and contributed to the development of this technique. A more specific version

of HLM, Hierarchical Generalized Linear Modeling (HGLM), will be utilized in this

study and both the theoretical foundations and the practical application of this method

will be presented.

1.5 Research Questions Addressed

Based on the objectives set forth, the following three research questions will be
investigated:

1. Which dimensions of global brands are more effective for what type of
consumers?

2. Can we identify a “global brand-prone” consumer segment? If we can, what are
the demographic and psychographic characteristics of these consumers?

3. Is this consumer segment robust across countries?

1.6 Summary and Overview of Remaining Chapters

Chapter 1 provides an overview of this study. Background information about
global branding and a review of gaps in the literature were presented. The research
objectives of the study, a brief overview of the method used, and specific research
questions addressed were also provided. Chapter 2 presents a review of the literature and
provides the theoretical rationale for the development of hypotheses relevant to the
research objectives. This chapter also summarizes previous research limitations,
discusses gaps in the literature, and presents the research hypotheses to be tested. Chapter

3 gives information on the theoretical background of the method used in this study.

Chapter 4 presents empirical results. Lastly, Chapter 5 presents conclusions and
contributions of the study. It also discusses the implications of the study results for

managers. Future research directions and limitations of the study are also presented.

Chapter 2

REVIEW OF LITERATURE AND HYPOTHESIS DEVELOPMENT

2.1 Overview of Global Branding Research

Global brands are among the most important intangible assets a company can
have. According to the Interbrand/BusinessWeek Study (2004), the financial value of the
top 100 global brands exceeds $1 trillion. In line with its importance, many researchers
have investigated different aspects of global branding.

Of these, the standardization/adaptation research is probably the oldest. The
majority of the studies in this stream investigated promotion strategies of global brands,
e. g., international advertising. A grawal (1995) traced the beginning of this debate to the
19205, when Goodyear’s manager for advertising, David L. Brown, and Bausch &
Lomb’s manager of sales promotion, Carl F. Propson, took two opposite sides. Brown
(1923) argued for standardization because be viewed humans as possessing common
attributes, whereas Propson (1923) insisted on the need for adaptation to appeal to
divergent local markets. The debate that concerned practitioners at first then appealed to
scholars. The late 19505 saw the first academic engagement on the issue by Pratt (1956).
Since then, international marketing strategy standardization has been the central focus of
both academics and practitioners (Agrawal 1995). As acknowledged also by a recent
literature review on the standardization/adaptation literature by Ryans, Griffith, and

White (2003), a vast amount of studies has been conducted in the field (e. g., Boddewyn

et al. 1986; Calantone et al. 2004; Calantone et al. 2006; Cavusgil and Zou 1994; Jain
1989; Kotler 1986; Levitt 1983; Samiee and Roth 1992; Wills et al. 1991; Wind 1986;
Zou and Cavusgil 2002).

The central research question in this line of research has been whether a
standardized or an adapted international marketing strategy is more effective. Agrawal
(1995) identified three schools of thought: standardization, adaptation, and the
contingency perspective. Proponents of the standardization school of thought (Fatt 1967;
Levitt 1983) stressed the benefits of standardization, mainly stemming from cost
reductions. Their assumption was that consumers around the world were not very
different and universal attributes appealing to all consumers could be found. This school
of thought argued that cost reductions would outweigh revenue loss, if any. Proponents of
the adaptation school of thought (Britt 1974; Nielsen 1964; Ricks and Arpan 1974)
argued that inter-country differences were too big to ignore and applying the same
marketing strategy worldwide would result in revenue loss, which would far outweigh
cost reductions. As the debate went on unresolved, a third group took a moderate
approach and acknowledged that the evaluation of possible factors that might affect the
success of a strategy in a particular country would be necessary when deciding on the
degree of standardization/adaptation (Kotler 1986; Walters 1986). Today, even after a
vast amount of studies has been conducted on the same issue, the debate has still not been
resolved (Ryans et al. 2003).

The second oldest main stream of research addresses the effects of country of
origin (the country with which a firm is associated) and country of manufacture (the

location of manufacture or assembly of a product) on consumer attitude and behavior

(Ashill and Sinha 2004; Batra et al. 2000a; Bilkey and Nes 1982; Han 1989; Johansson et
al. 1985; Kaynak and Cavusgil 1983; Samiee 1994; Samiee et al. 2005; Thakor 1996;
Thakor and Lavack 2003; Tse and Gorn 1993; Tse and Lee 1993). The earlier studies
rarely make a distinction between country of origin and country of manufacture. The
general name used in the literature is “country of origin (C0)” and CO effects are defined
as “intangible barriers-to-entry in the form of negative consumer bias toward products of
foreign origin” (Wang and Lamb 1983, p. 71). Although these studies have divergent
findings (Al-Sulaiti and Baker 1998), the great majority of country of origin studies
found that in general, consumers display a preference for products made in some
countries more than others (Samiee 1994).

J ohansson et al. (1985) concluded that there was little evidence of stereotyping
based on country of origin and suggested that the impact of country of origin on product
evaluations might be substantially more complex than had been previously assumed.
Later studies evolved in complexity. Han (1989; 1990) showed that familiarity of
consumers with a country’s products was an important moderator. When consumers are
not familiar with a country's products, country image serves as a halo from which
consumers infer a brand’s product attributes, which in turn impact their attitude toward
the brand. On the contrary, when consumers are familiar with a country's products,
country image summarizes consumers' beliefs about product attributes and directly
affects their attitude toward the brand. This presents an evolutionary perspective;
consumers form the image of the country through experience with their products.

As global companies started to use off-shore outsourcing and shift their factories

overseas to take advantage of lower costs, the distinction between country of origin and

country of manufacture is stated more clearly in the literature and another construct
appeared: brand origin. The country a brand is associated with, or the perceived
headquarters of the company of the brand, determines brand origin. This
conceptualization is independent of country of manufacture (Samiee et al. 2005; Thakor
and Lavack 2003).

The first studies done on this topic found sourcing country stimuli to have greater
effects than brand name on consumer attitudes toward bi-national products (Han and
Terpstra 1988). In one study, Tse and Gorn (1993) concluded a well-known global
brand—Sony, in their experiment—could not override the country of origin effect. Subjects
perceived electronic products made in Japan to be significantly higher in quality than
those made in Indonesia. The authors’ suggestion to managers was to develop new
products made in a favorable country of origin. However, in another study, Tse and Lee
(1993) decomposed the negative country image into component and assembly origins,
and considered effects of both global branding and product experience. They obtained a
contradictory result this time; a strong positive brand was found to override negative
assembly origin effects. After the product experience, the negative component origin
effect was also gone.

A recent study also showed that the brand effect might outweigh the country of
origin effect. Ashill and Sinha (2004) showed that the effect of brand equity was three
times more important than country of origin effects on consumers’ purchase intentions.
Their suggestion to marketing managers was to choose a country of manufacture for cost

efficiencies regardless of any associations consumers may have with that country.

10

Yet in another recent study, Samiee et al. (2005) questioned the basic assumption
of all these studies, i.e., that consumers actually know brand origins. They claimed that
previous research might have overstated the effect of brand origin on consumer attitude
and behavior since the studies were mostly experimental designs in which the brand
origins of the hypothetical brands were manipulated. They used actual brands in a
national survey. Their argument was that if brand origin played a salient role in
consumers’ everyday decision making, then consumers would know the brand origins
reasonably accurately. The opposite result they obtained indicated that this was not the
case.

Looking at other widely researched topics on global branding, one notices quite a
variety: global brand management strategy and organizational coordination issues (Aaker
and Joachimsthaler 1999; Das 1993; Farquhar 2005; Gregory and Wiechmann 2002;
Hamel and Prahalad 1985; Quelch 1999; Quelch and Hoff 1986; van Gelder 2004; van
Gelder 2003; van Gelder 2005), global brand personality studies (Aaker 1997; Sung and
Tinkham 2005), cultural influences of global brands (Klein 1999: Smart 1999; Thompson
and Arsel 2004), global brand associations (Batra et al. 2000b; Holt et al. 2004a; Holt et
al. 2004b; Steenkamp et al. 2003), global brand image and positioning strategies (Alden
et al. 1999; Hsieh 2002; Hsieh et al. 2004; Roth 1992; Roth 1995a; Roth 1995b), global
brand equity measurement issues (Aaker 1996; Hsieh 2004), corporate social
responsibility of global brands (Macrae et al. 2003; Werther and Chandler 2005), global
marketing universals and global consumer segments (Dawar and Parker 1994; Hassan
and Katsanis 1994; Holt et al. 2004a; Holt et al. 2004b; Levitt 1983), and many case

studies investigating individual global branding cases (Abratt and Motlana 2002; Buri

ll

and Findlay 1996; Crippen et al. 1995; Kanter and Dretler 1998; Kapferer 2002; Kern et
al. 1990; Kish et al. 2001; Raman 2003; Rozin and Magnusson 2003; Strong 1987;
Wright and Nancarrow 1999).

Although all these studies address a variety of global branding issues, the
common effort is to understand what works best for global brands under what
circumstances. In order to be able to give an answer to this question, it is very important
that world consumers’ attitudes and behaviors toward global brands are known. However,
the main limitation of conducting such studies is the difficulty in obtaining data from
consumers. As a result, with a few exceptions (e.g., Holt et al. 2004a; Holt et al. 2004b;
Steenkamp et al. 2003), either experimental designs with student subjects were used, or
global brand managers were surveyed and asked information about their consumers. Even
when actual consumers were used, the number of countries included in the studies was
naturally limited.

A niche, but very important area, is the value in the “globalness” and the global
brand associations consumers value worldwide. The idea that “globalness” of a brand
may be a value-added for consumers was suggested by researchers before (Gregory and
Wiechmann 2002; Kapferer 1997; Shocker et al. 1994). However, this was empirically
tested only recently, in the 2000s (B atra et al. 2000b; Holt et al. 2004a; Holt et al. 2004b;
Johansson and Ronkainen 2005; Steenkamp et a1. 2003).

Steenkamp et al. (2003) tested the effect of perceived brand globalness (PBG) on
brand purchase likelihood in two countries (United States and South Korea) across four
product categories. Their findings show that PBG is positively associated with both

perceived brand quality and brand prestige; higher perceived quality and prestige are key

12

sources of enhanced consumer value for global brands. The strongest effect on PBG,
however, is through perceived quality. The direct effect of perceived brand globalness on
brand purchase likelihood (what they called “belongingness pathway”) was not found
significant.

Johansson and Ronkainen (2005) tested the effect of globalness of a brand on
brand esteem. They used data from eight countries: France, Germany, Italy, Spain,
Sweden, UK, USA, and Japan. Their results show that the reach of a brand is associated
with higher esteem when brand familiarity is controlled. They concluded that the
globality of a brand has a definite role in building sustainable competitive advantage.
“Esteem” is a major construct coming from Landor Associates’ global surveys of brand
strength.l The exact items were not provided in the study; however, it was reported to be
based on multiple items, including the question, “In what personal regard do you hold
this brand? By ‘personal regard’ we mean how highly you think and feel about the
brand.”2

However, one caveat concerning their research is that the data used for esteem
and familiarity variables were only the rankings, not the actual scores. This casts some
doubt on the validity of their results. Furthermore, their operationalization of the
globalness of a brand was questionable, as will be discussed next.

There are different conceptualizations of what “global brands” means. Although

there is a shortage of formal definitions of “global brands” in the literature (Steenkamp et

 

I Landor Associates is a brand consultancy affiliated with the Young & Rubicam advertising agency. The
survey data is used to build their proprietary Brand Asset Valuator (BAV®) model.

The final score for brand esteem was derived from a factor analysis and a weighing together of the
multiple indicators. The level of familiarity was, likewise, derived from several items, one of the items
being, “What is your overall awareness and understanding of what kind of product this brand represents?”
The perceived quality score was derived from a rating of the degree to which the respondent associates
‘high quality’ with the brand.

13

al. 2003), global brands can be defined as brands that consumers can find under the same
name in multiple countries, with generally similar and centrally coordinated marketing
strategies (Steenkamp et al. 2003; Yip 1992). Consultancy companies’ definitions are
more specific. ACNielsen (2001) classifies brands as global if they are present in the four
major regions of the world—North America, Latin America, Europe/Middle East/Africa,
and Asia—with at least 5 percent of sales coming from outside the home region, and total
revenues of at least $1 billion. Interbrand (2004) has slightly different criteria; a brand
should derive about a third of its earnings from outside its home country and have a value
greater than $1 billion to qualify for The 100 Top Global Brands list published by
Interbrand and NewsWeek.

Global branding research has borrowed from different definitions when
operationalizing global brands. For example, Steenkamp et a]. (2003, p. 58) stated that
“consumers can form the perception that a brand is ‘global’ if they come to believe that
the brand is also available in other countries through media coverage, word of mouth, or
travel and/or if they see brand communications stressing ‘globalness’ through
associations with global consumer culture (GCC) symbols.” They operationalized
perceived brand globalness (PBG) by measuring the degree to which consumers thought
the same brand was marketed in countries other than their own, and used the following
items: “To me, this is a global brand,” “I do think consumers overseas buy this brand,”
and “This brand is sold all over the world” (Steenkamp et al. 2003, p. 64). Johansson and
Ronkainen (2005, p. 340) defined global as “the multi-market reach of products that are
perceived as the same brand worldwide both by consumers and internal constituents”;

however, their operationalization included only 8 countries, not “the world.” They

l4

analyzed data from 8 countries, with approximately 150 global and local brands coming
from each. They calculated “reach” as a measure of globality in these 8 countries;
therefore, if a brand was listed in one country (i.e., included in the Top 150 list of
Landor), it was assigned the value “1.” If it was listed in all eight countries, it was
assigned the value “.8” The other pioneer study in the field done by Holt et al. (2004a;
2004b) did not use a separate measure of globalness, but analyzed only global brands, all
but one of which came from lnterbrand’s Top 100 list.

The most recent study on this subject (Holt et al. 2004a; Holt et al. 2004b)
conducted an exploratory qualitative research in 41 countries, followed by a survey with
1800 people in 12 countries. Seventeen global brands were evaluated across 5
dimensions: quality signal (international success), global myth (global status symbol),
social responsibility, best-in-class, and American values.3 Their empirical analyses show
that most people prefer one global brand over another because of differences in the
brands' global dimensions. They identified the four dimensions of global brands as
quality, global status symbol, best-in-class, and social responsibility (Holt et al. 2004b).
This study is especially interesting because the quotations they report from consumers
worldwide are very rich. Some examples include (Holt et al. 2004b, p.186-189):

“Global brands must be very good in order to sustain a global presence. Some
local brands are good too, but they don’t have the ambition to venture into other
countries. They are not as successful. ” (Hong Kong)

“The more people who buy a brand, the better quality it is. ” (Russia)

 

3 The authors use different construct names in their two papers; the alternated names are provided in
parentheses.

15

“When you think of an international brand, you already assume it has better
quality. ” (United Kingdom)

“They (global brands) are very dynamic, always upgrading themselves.” (India)

“They (global brands) are more exciting because they come up with new products
all the time, whereas you know what you will get with the local ones. ” (Australia)
As is evident from the examples above, global brands symbolize “quality” and
“innovativeness.”4

They also convey “belongingness,” e.g., consumers may choose global brands
because they feel themselves being part of a special group connected via the brand:

“Global brands make you feel part of something bigger and give you a sense of

belonging. ” (New Zealand)

Global brands also give the consumers the feeling of “prestige”; consumers may
feel themselves in a higher class when they use global brands:

“If you buy something because Americans use it, you feel at their same level. ”
(Mexico)

“You appear to be of the same class when you use global brands. ” (Nigeria)

Another key factor consumers consider when buying global brands is
“genuineness,” e.g., consumers may think that the brand is from a country that has a
particular expertise in the product category, and that’s why they may buy them:

“If it is electronics, I will go for Japanese products because they are the best, but

for clothing I prefer American labels and British made. ” (Nigeria)

 

4 Both Holt et al. (2004b) and Steenkamp et al. (2003) treated quality and innovativeness as if they were
the same association. This will be the first study where the two are acknowledged to be distinct.

16

“German cars are very reliable. They are produced by employing the highest
technology. ” (Turkey)

Consumers expect global brands to act in a socially responsible manner, thus
“social responsibility” is another key dimension of global brands on which consumers
base their purchasing decisions:

“McDonald’s pays back locally, but it is their duty. They are making so much

money, they should be giving back. ” (Australia)

The results of all these studies identify the global brand associations. A list
includes: quality, innovativeness, belongingness, prestige, genuineness, and social
responsibility (Batra et al. 2000b; Holt et al. 2004a; Holt et al. 2004b; Steenkamp et al.
2003). Even though the names of the associations differ between studies, the distinction
is semantic more than conceptual. For example, brand prestige in one study becomes
global status symbol in another; what is called perceived brand genuineness in this study
is called best-in-class in another, etc.

In sum, research so far has shown that consumers attribute to global brands
certain associations. However, no international study has been conducted yet on which
attributes are more effective for which types of consumers.

The increasing interconnectedness of local cultures and the emergence of a world
culture (Hannerz 1990), theories about global cultural flow (Appadurai 1990), the wide-
spread influence of media, and certain consumers (e. g., teenagers) displaying similar
consumption signs like jeans or sneakers led the researchers to the idea that a global

consumer culture is emerging (Alden et al. 1999). Alden, Steenkamp, and Batra (1999, p.

17

76) state that “From the semiotics perspective, it can be argued that certain consumers
(e.g., elite, post-World War 11 [WW II] consumers, teens) will desire consumption
experiences and objects that they consider ‘signs’ of these scripts (of imagined lives) in
order to act out imagined or real participation in the more cosmopolitan global consumer
culture communicated by the media.” Global brands are argued to symbolize this global
consumer culture (B atra et al. 2000b).

However, it has not been empirically investigated whether there are, in fact,
specific consumers who are global brand-prone, that is, consumers who systematically
buy global brands, or whether such a segment of consumers is robust across countries.
These questions will be empirically tested in this study by using data from more than

31,000 consumers across 31 countries. The model is shown in Figure 2.1.

18

 

65 Seacoaou Essex -

 

_"_— l

 

 

 

 

 

 

 

 

Ammmv EoEoocancm.:om Hana—swam -
2:85 - 283682 65.5 Ego—O
SEwO com ADO - ow< -
mumummaoaoauasu mug—=50 wozmtoaowuasv 35.5.80 Ezfiacoamom
. . . Eoom

38m 338qu

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

mace—320 $255280
e55 nozoocom team cozoocom
¢
ownmocn—
E:m..o:.$O 985 328qu
6:95 A 4
Eco—G wwocwfiwcofim

 

 

 

28m uoioocom

 

 

32.35305:
285 ©3680;

 

 

 

 

3:95
985 338qu

 

 

 

 

 

 

.252 2:. t 3 955

19

2.2 Theoretical Foundation and Hypothesis Development

2.2.1 Associative Network Memory Model

A brand can be defined as “a name, term, sign, symbol, or design, or combination
of them which is intended to identify the goods and services of one seller or group of
sellers and to differentiate them from those of competitors” (Kotler 1991, p. 442). A
broad literature has investigated the effects of brands on consumer behavior and
marketing program effectiveness (for a review, see Hoeffler and Keller 2003; Yoo et al.
2000). Research so far reveals that there are many advantages to creating a strong brand,
where brand strength may reflect both macro-level brand targets such as market
leadership or market share position, and micro-level targets such as consumer familiarity,
consumer knowledge, preferences, or loyalty (Keller 2002).

Brand equity can be defined as the differential effect that brand knowledge has on
consumer or consumer response to the marketing of a brand (Keller 2002). The
associative network memory model can be used to explain how brand associations
influence consumer responses toward a brand (Keller 1993). The associative network
memory model (see Anderson 1983; Srull and Wyer 1989) views knowledge as a set of
nodes and links. Nodes are stored information connected by links, which vary in strength.
While a spreading activation process from node to node determines the extent of retrieval
in memory (Collins and Loftus 1975; Raaijmakers and Shiffrin 1981; Ratcliff and
McKoon 1988), it is the strength of the association between the activated node and all

linked nodes that determines the extent of the spreading activation. In this context, brand

20

associations are viewed as the informational nodes that are linked to the brand node in
memory and contain the meaning of the brand for consumers (Keller 1993).

The global brand associations identified by previous research, namely quality,
innovativeness, belongingness, prestige, genuineness, and social responsibility are all
hypothesized to be positively related to the likelihood of global brand ownership (GBO).
Perceived brand globalness is also included in the model as a separate variable to test
both its direct effect on the likelihood of CEO, and its interaction with other global brand
associations. More formally, the hypotheses are:

H]: Perceived brand quality is positively related to the likelihood of CEO.

H 2: Perceived brand innovativeness is positively related to the likelihood of GBO.

H3: Perceived brand belongingness is positively related to the likelihood of GBO.

H4: Perceived brand prestige is positively related to the likelihood of GBO.

H5: Perceived brand genuineness is positively related to the likelihood of GBO.

H6: Perceived brand social responsibility is positively related to the likelihood of
CEO.

H7: Perceived brand globalness is positively related to the likelihood of CEO.

H8: Perceived brand globalness strengthens the efi‘ects of all other perceived

brand associations on the likelihood of G80.

However, not all the associations can be expected to have the same effect on the
likelihood of CEO for all types of consumers. The contingency approach will thus be
introduced in the next section and it will be hypothesized that the relationship between
global brand perceptions of consumers and the likelihood of GBO is dependent on

consumer characteristics.

21

2.2.2 Contingency View

Contingency approaches are positioned within management as mid-range theories
between the two extreme views which state either that management is a science where
universal principles and laws can be found, or that it is an art where each case is unique
and each situation must be analyzed separately. Contingency theories stand in between,
and state that the appropriateness of different strategies depends on various factors
(Zeithaml et al. 1988). Such theories stress: “It depends.”

This thesis rests on the contingency view, borrowed from the strategic
management literature. Consumer-level contingencies included in the model consist of
both demographic (age and income) and psychographic (personal values) variables.
Schwartz’s values inventory is used to classify consumers along two bipolar dimensions:

self-enhancement versus self-transcendence and openness to change versus conservation.

2.2.3 Consumer Demographics

Several researchers emphasized the importance of using demographics variables
to classify consumers (Gatignon and Robertson 1985; Gupta and Chintagunta I994;
Zeithaml 1985). Past research suggests that age and income levels of consumers may
influence consumer behaviour (Green and Cunningham 1975; Holbrook and Schindler
1994; Zeithaml 1985).

For example, Zeithaml (1985) f inds that older shoppers plan more for shopping
than younger consumers, tend to use information more, and economize more. Shopping
time, number of supermarkets visited weekly, and number of weekly shopping trips are

all higher for older consumers. Older consumers see shopping as more important than do

22

younger consumers. These findings suggest that older consumers may be more selective
and thus more sensitive to the existence or lack of existence of brand attributes than
younger consumers. Respectively, the effects of quality, prestige, and genuineness
associations of global brands on purchasing decisions can be expected to be stronger for
older consumers than younger consumers. Therefore, it is hypothesized that:

H9: The effect of perceived brand quality on the likelihood of GBO is stronger for
older consumers than younger consumers.

H10: The eflect of perceived brand prestige on the likelihood of GBO is stronger
for older consumers than younger consumers.

H“: The effect of perceived brand genuineness on the likelihood of GBO is

stronger for older consumers than younger consumers.

According to Kohlberg's (1981) cognitive moral development theory, our
cognition, emotion, and judgment change as we move through stages of moral
development. Terpstra et al. (1993) have also argued that people tend to become more
ethical as they grow older. It is suggested that as people age, they tend to become less
concerned with wealth and more interested in personal growth (Singhapakdi et al. 2001).
Thus, it can be hypothesized that older consumers are more sensitive to the existence of
social responsibility than younger consumers and the existence or lack of existence of
this attribute will have a higher impact on older consumers’ purchasing decisions than it
will have on younger consumers’ purchasing decisions. Thus, it is hypothesized that:

H12: The effect of percei ved brand social responsibility on the likelihood of GBO

is stronger for older consumers than younger consumers.

23

Studies on new product adoption and consumer innovativeness suggest that
younger consumers are more open to innovative products. Steenkamp et al. (1999) show
a strong negative association between age and consumer innovativeness. In an empirical
study, lm et al. (2003) found that consumers who are younger are more likely to adopt
new products than consumers who are older. Previously identified relationships between
younger age and innovativeness-friendly consumer behavior suggest that the effect of
perceived brand innovativeness on the likelihood of GBO may be higher for younger
consumers than for older consumers. Therefore, it is hypothesized that:

H13: The effect of percei ved brand innovativeness on the likelihood of GBO is

stron ger for younger consumers than older consumers.

In a cross-cultural empirical study of consumer tendencies, Keillor et al. (2001)
found that older consumers are more likely to have a higher sense of national identity and
to be more ethnocentric than younger consumers. They concluded that, regardless of
culture and national background, younger consumers might indeed be more similar and
have less marketplace prejudice than older consumers. They argued as these consumers
age and older consumers drop out of the market, the existence of a globalized consumer
might become a reality.

Global media is often held responsible for this trend. Studies on consumerism and
young people suggest that global media is influential in promoting particular brand
images, and through encouraging brand identity, it provides young people with a
common language among their peers (Anderson and Miles 1999). Global advertising is
one way through which this influence occurs. Lifestyle advertising, e.g., associating a

pair of sneakers with an adventurous lifestyle to make the consumer identify himself with

24

the product’s personality, and emotional branding are among the best practices in
branding. These practices are used by global brands, which can be seen as successful
actors of branding. Thus, one can expect this influence to become even stronger for
global brands. Anderson and Miles (1999) suggest that the television campaigns
emphasizing the values of individuality and freedom, like Reebok and Nike’s advertising
campaigns “Let UBU” and “Just Do It,” in fact appeal to young people by satisfying their
needs for conformity and their desire to feel part of a group. Therefore, it is hypothesized
that:

H14: The effect of perceived brand belongingness on the likelihood of GBO is

stronger for younger consumers than older consumers.

Indeed, global brands were reported to provide an added-value to young people
such as confidence. Anderson and Miles (1999, p. 109) report quotations from a
qualitative research study funded by the Economic and Social Research Council (ESRC)

conducted in Yorkshire and Strathclyde, UK:

“Danny: ‘Yeah, brands likeLevi ’s give you confidence and that. If you like what
you’re wearing and if you ’ve spent a lot of money on it like, it gives you a lot of
confidence... I don ’t know. Like I ’d buy stufi’ like that bottle of Jack Daniels. I wouldn ’t
drink it, I ’d just have it in my bedroom. Just because I like it. I like how it looks and that.

And it makes me happy.”

Cote and Allahar (1996) criticize global media in selling young people an element
of identity they taught them to crave. The argument is that standardizing young people’s
consuming habits is seen by the global media as an allowance to getting more advertising

income or increasing the appeal of their youth-oriented programs. Standardization of the

25

media images young people consume is thought to be instrumental for that reason. Their
argument is open to debate; however, the importance of the media, and by implication the
idealized images of which global brands are part, is straightforward. “The media play a
key role as an arbiter of consumption, in a young person’s life during which they are
struggling to find a focus” (Anderson and Miles 1999, p. 110). So do global brands, it is
suggested. Therefore, it is hypothesized that not only the globalness association of global
brands has a stronger effect on the likelihood of GBO for younger consumers than for
older consumers (i.e., a moderator role of age on the relationship between perceived
brand globalness and the likelihood of GBO), but also younger consumers are more likely
to buy global brands in general (i.e., a direct effect of age on the likelihood of GBO).

H15: The effect of percei ved brand globalness on the likelihood of GBO is

stronger for younger consumers than older consumers.

H16: Age is negatively related to the likelihood of GBO.

Cote and Allahar (1996) argue that consumption ethos is poured out upon young
people by the mass media. “Since they are bombarded with tantalizing images of the
‘good life,’ it is not surprising that the young are dispirited by the reality of their poor
prospects” (Cote and Allahar 1996, p. 148). The gap between the idealized life and the
actual life is expected to be even wider in less—developed countries than in more-
developed ones. Thus, the differences between younger and older people in terms of their
behavior towards global brands may be argued to be wider as well. This suggests that the
marginal effect of age on the likelihood of global brand ownership is higher in less-

developed countries than in more-developed countries. Thus, ceteris paribus, the negative

26

effect of age on the likelihood of GBO can be expected to be stronger in less-developed
countries than in more-developed countries. Therefore, it is hypothesized that:

H17: The negative effect of age on the likelihood of GBO is stronger in less-

developed countries than in more-developed countries.

Classification of consumers based on their income categories rests on the
assumption that more income results in more disposable income and this leads
consumption away from subsistence products toward material goods. This also suggests
that as consumer income increases, consumers may expect more from their purchases and
be more sensitive to additional product attributes. Zeithaml (1985) indeed shows that
income affects the amount of time spent shopping, the number of supermarkets visited
weekly, the extent of planning, the amount of purchase, and the importance of shopping.
Higher income consumers plan less and they spend more time shopping.

Gupta and Chintagunta (1994) find that income and household size significantly
affect the segment membership probabilities and lower income consumers tend to be
price and promotion sensitive. Granzin (1981), for example, suggests that stores which
serve younger shoppers with larger households and lower incomes should stock a wide
selection and large quantities of generic products. Im et al. (2003) found that higher
income consumers are more likely to adopt new products than lower income consumers.

Related with the more disposable income argument above, higher income
consumers can also be expected to want more from their products, be more willing to pay
premiums for products with many attributes, and be more sensitive to the existence of
these attributes. Thus, it can be argued that higher income consumers are more sensitive

to the existence of global brand associations when making their purchasing decisions than

27

lower income consumers (i.e., income positively moderates the relationships between
global brand associations and the likelihood of GBO).

H18: The effect of perceived brand quality on the likelihood of GBO is stronger for
higher income consumers than lower income consumers.

H19: The effect ofperceived brand innovativeness on the likelihood of GBO is
stronger for higher income consumers than lower income consumers.

H 20: The efiect of perceived brand belongingness on the likelihood of GBO is
stronger for higher income consumers than lower income consumers.

H2]: The effect of perceived brand prestige on the likelihood of GBO is stronger
for higher income consumers than lower income consumers.

H22: The efiect of perceived brand genuineness on the likelihood of GBO is
stronger for higher income consumers than lower income consumers.

H 2 3: The effect of percei ved brand social responsibility on the likelihood of GBO

is stronger for higher income consumers than lower income consumers.

Keillor et al. (2001) argued that as consumers’ income increase, they tend to
consume a wide range of products, thus their participation in the global marketplace is
more likely than that of lower income consumers. They found that higher income
consumers have a weaker sense of national identity than lower income consumers and
they suggested that higher levels of income might be associated with a global consumer,
a consumer who is similar across countries in terms of their psychological tendencies in
the global marketplace.

Other studies investigating the relationship between income and consumer
behavior found that higher income consumers hold relatively more favorable attitudes

toward and are more likely to accept non-local products than do lower income consumers

28

(Bilkey and Nes 1982; Hyllegard et al. 2005; Schooler 1971; Wall et al. 1991),
suggesting a more positive attitude toward global brands.

Another factor that may be argued as important is ability; higher income
consumers may be more able to afford the price premiums global brands command.

Therefore, it is hypothesized that not only the globalness association of global
brands has a stronger effect on the likelihood of GBO for higher income consumers than
for lower income consumers (i.e., a moderating effect of income on the relationship
between perceived brand globalness and the likelihood of GBO), but also higher income
consumers are more likely to buy global brands in general (i.e., a positive direct effect of
income on the likelihood of GBO). Therefore, it is hypothesized that:

H 24: The effect of percei ved brand globalness on the likelihood of GBO is
stronger for higher income consumers than lower income consumers.

H 2 5: Income is positively related to the likelihood of GBO.

In developing countries, imported brands are preferred to national goods (Usunier
2000). Batra et al. (2000a) showed that developing country consumers who had a greater
admiration for lifestyles in economically developed countries had a stronger preference
towards non-local, especially western, brands than other consumers who did not have
such an admiration. Therefore, we can expect that in less-developed countries, the
marginal effect of income on the likelihood of GBO will be higher than in more-
developed countries. Thus, the positive effect of income on the likelihood of GBO can be
expected to be stronger in less-developed countries than in more-developed countries.

Therefore, it is hypothesized that:

29

H 26: The positive efifect of income on the likelihood of GBO is stronger in less-

developed countries than in more-developed countries.

2.2.4 Consumer Psychographics: Personal Values Research and Schwartz’s
Values Theory

Psychographics examine consumer motivations when purchasing products.
Rokeach (1973, p. 5) defines personal value as an "enduring belief that a specific mode of
conduct or end-state of existence is personally or socially preferable to an opposite or
converse mode of conduct or end-state of existence." Values serve as guiding principles
in everyday life (Rokeach 1973). They are abstract social cognitions and they provide a
foundation for both attitudes and behaviors (Homer and Kahle 1988). While
demographics provide a basic means of categorizing consumers, psychographic variables
such as values can also be used to assess the similarities between consumers (Kahle et al.
2000). Indeed, research shows that values can be illuminating to explain why consumers
purchase certain products (Rose and Shoham 2000) and can be used as a basis for
consumer segmentation.

The centrality of the value concept is probably proclaimed best by Rokeach
(1973, p. 3): “The value concept, more than any other, should occupy a central position
able to unify the apparently diverse interests of all the sciences concerned with human
behavior.” Personal values have been shown to be the underlying determinant of
consumer attitudes and consumption behavior (Homer and Kahle 1988; Scott and Lamont
1977; Vinson et al. 1977). Allen (2001) investigates the impact of personal values on
product preferences in a recent study and finds that values influence product preference

both directly and indirectly via prioritizing the importance of product attributes.

30

Schwartz (1992) has focused on the search for a fundamental set of basic human
values. Through an extensive series of studies across more than 40 countries, he
constructed a values inventory representing the values that are universally important and
identified 10 general value types (Schwartz and Sagiv 1995). These values are power,
achievement, hedonism, stimulation, self-direction, universalism, benevolence, tradition,
conformity, and security (see Table 2.1). He developed a Circumplex model by arranging
these 10 values in a two-dimensional circular structure oriented by two axes: (1) self-
enhancement versus self-transcendence and (2) openness to change versus conservation
(see Figure 2.2). The first dimension indicates the degree to which values are self-
oriented (hedonism, power, achievement) versus other-oriented (benevolence,
universalism). The second dimension shows values that emphasize change (stimulation,
self-direction) versus preservation of the status quo (conformity, tradition, security)
(Burroughs and Rindfleisch 2002). Complementary values stand next to each other and
competing values stand opposite (e. g., 180 degrees) from each other. Value conflicts
happen when individuals’ value systems include competing values, which in turn causes
psychological tension. "The pursuit of each type of value [has] psychological, practical,
and social consequences that may conflict or may be compatible with the pursuit of other

value types" (Schwartz 1994, p. 23).

31

Table 2.1 - Schwartz’s 10 Value Types

1. Power: Social status and prestige, control or dominance over people and resources.

2. Achievement: Personal success through demonstrating competence according to
social standards.

3. Hedonism: Pleasure and sensuous gratification for oneself.

4. Stimulation: Excitement, novelty and challenge in life.

5. Self-Direction: Independent thought and action-choosing, creating, exploring.

6. Universalism: Understanding, appreciation, tolerance and protection for the
welfare of all people and for nature.

7. Benevolence: Preservation and enhancement of the welfare of people with whom
one is in frequent personal contact.

8. Tradition: Respect, commitment and acceptance of the customs and ideas that
traditional culture or religion provide the self.

9. Conformity: Restraint of actions, inclinations, and impulses likely to upset or
harm others and violate social expectations or norms.

10. Security: Safety, harmony, and stability of society, of relationships, and of self.

Source: Schwartz and Sagiv (1995)

 

32

Figure 2.2 — Schwartz’s Circumplex Model of Values

A

 

 

Self-Transcendence
A
Universalism Benevolenc
Tradition
Self-Direction .
Conformrty Conser-
Openness‘ ,. Nation
to Change _ , ,,,,,,,,, j, ”””
Strmulatron .............. f, .
................ Security
............. Hedonism ,
" Power
......,.-"':,A,chievemenl
Self-Enhancement

Source: Schwartz (1992)

33

Schwartz’s value dimensions express fundamental motivational goals which
reflect existential needs and which, in turn, relate theoretically to potentially any attitude
and behavior. In fact, Clark (1987) suggests that values provide an important link
between consumers and marketers; successful marketers are able to develop brands that
work effectively across borders because they understand both consumer values and brand
values.

Consumers with different values can be expected to prioritize different
associations of global brands. For example, people whose dominant values lie closer to
the self-enhancement pole rather than the self-transcendence pole tend to be motivated by
their own personal interests, even at the expense of others. Their main value types are
hedonism, power, and achievement. On the other hand, people whose dominant values
are closer to the self—transcendence pole rather than the self-enhancement pole tend to be
motivated by promoting the welfare of others. Benevolence and universalism are their
main values (Schwartz 1992). Resultant self-enhancement (RSE) construct, which is the
importance attached by the consumer to self-enhancement minus the importance attached
to self-transcendence (Feather 1995; Steenkamp et al. 1999), will be used in this research
to indicate the dominant value types of the consumers.

It was suggested that new and functionally better products and brands have the
potential to improve consumption (Steenkamp et al. 1999) and to help gain social status
(Rogers 2003). It can be hypothesized that global brands are also a way of enhancing
one’s status and improving consumption experience. Thus, it can be expected that the

effects of quality, innovativeness, prestige, genuineness, and globalness associations of

34

global brands on the likelihood of GBO are stronger for higher RSE consumers than
lower RSE consumers (i.e., a positive moderator role of RSE). It can also be
hypothesized that higher RSE consumers are more likely to buy global brands in general
than lower RSE consumers (i.e., a direct effect of RSE on the likelihood of GBO). On the
other hand, belongingness and social responsibility associations are aligned more with the
dominant values of lower RSE consumers and these associations can be expected to have
a stronger effect on the likelihood of GBO for lower RSE consumers than for higher RSE

consumers (i.e., a negative moderator role of RSE).

Therefore, it is hypothesized that:

H 27: The effect of perceived brand quality on the likelihood of GBO is stronger for
higher RSE consumers than lower RSE consumers.

H 28: The eflect of percei ved brand innovativeness on the likelihood of GBO is
stronger for higher RSE consumers than lower RSE consumers.

H 29: The eflect of percei ved brand belongingness on the likelihood of GBO is
stronger for lower RSE consumers than higher RSE consumers.

H 30: The effect of percei ved brand prestige on the likelihood of GBO is stronger
for higher RSE consumers than lower RSE consumers.

H3]: The effect of percei ved brand genuineness on the likelihood of GBO is
stronger for higher RSE consumers than lower RSE consumers.

H 32: The eflect of percei ved brand social responsibility on the likelihood of GBO
is stronger for lower RSE consumers than higher RSE consumers.

H 3 3: The effect of percei ved brand globalness 0n the likelihood of GBO is
stronger for higher RSE consumers than lower RSE consumers.

H34: RSE is positively related to the likelihood of GBO.

35

Just as was the case with higher income consumers, again the marginal effect of
RSE on the likelihood of GBO can be expected to be higher for consumers in less-
developed countries than consumers in more-developed countries. In less-developed
countries, consumers may see global brands as a means for self-enhancement (Friedman
1990). Such an effect of global brands may not be present for consumers from more-
developed countries, thus the purchasing decisions of global brands may be more
sensitive to consumer RSE in less-developed countries than in more-developed countries.
Therefore, it is hypothesized that:

H35: The positive effect of RSE on the likelihood of GBO is stronger in less-

developed countries than in more—developed countries.

The second bipolar dimension is openness to change versus conservation. People
whose dominant values lie closer to the openness to change pole rather than the
conservation pole tend to be motivated by following their own intellectual and emotional
interests in unpredictable and uncertain directions. Their main value types are stimulation
and self-direction. On the other hand, people whose dominant values are closer to the
conservation pole rather than the openness to change pole tend to be motivated by
preserving the status-quo and the certainty it provides in relationships with close others,
institutions, and traditions. Conformity, tradition, and security are their main values
(Schwartz 1992). Resultant conservation (RC) construct, which is the importance
attached by the consumer to conservation minus the importance attached to openness to
change (Feather 1995; Steenkamp et al. 1999), will be used in this research to indicate the

dominant value types of the consumers.

36

Innovativeness association of global brands is aligned with the values of lower
RC consumers. Past research indeed has identified a negative association between
conservation and consumer innovativeness (Steenkamp et al. 1999). Thus, it can be
expected that the effect of innovativeness association on the likelihood of GBO is higher
for lower RC consumers than higher RC consumers. Furthermore, global brands are the
symbols of modernity and change, which again align with the values of lower RC
consumers. Thus, we can expect globalness association to have a stronger effect on the
likelihood of global brand ownership for lower RC consumers than higher RC consumers
(i.e., a moderator role of RC on the relationship between perceived brand globalness and
the likelihood of GBO) and lower RC consumers to be more likely to buy global brands
in general (i.e., a direct effect of RC on the likelihood of GBO). For the other

associations, no differences between higher RC and lower RC consumers are expected.

The hypotheses for the effect of RC are:

H 36: The efiect of perceived brand quality on the likelihood of GBO is not
difi‘erent for higher RC consumers vs. lower RC consumers.

H 37: The effect of percei ved brand innovativeness on the likelihood of GBO is
stronger for lower RC consumers than higher RC consumers.

H 38: The eflect of percei ved brand belongingness on the likelihood of GBO is not
difi‘erent for higher RC consumers vs. lower RC consumers.

H 39: The effect of percei ved brand prestige on the likelihood of GBO is not
different for higher RC consumers vs. lower RC consumers.

H40: The effect of percei ved brand genuineness on the likelihood of GBO is not
diflerent for higher RC consumers vs. lower RC consumers.

H41: The effect of percei ved brand social responsibility on the likelihood of GBO

is not diflerent for higher RC consumers vs. lower RC consumers.

37

H42: The effect of perceived brand globalness on the likelihood of GBO is
stronger for lower RC consumers than higher RC consumers.

H43: RC is negatively related to the likelihood of GBO.

Consistent with the arguments discussed for higher RSE consumers, the
perceptional symbolic value, e. g., modernity, of global brands can be expected to be
higher for consumers in less-developed countries than for consumers in more-developed
countries. Hence, the marginal effect of resultant conservation on the likelihood of GBO
may be higher in less-developed countries than in more-developed countries. Thus, the
negative effect of RC on the likelihood of GBO can be expected to be stronger for
consumers in less-developed countries than consumers in more—developed countries.
Therefore, it is hypothesized that:

H44: The negative effect of RC on the likelihood of GBO is stronger in less-

developed countries than in more-developed countries.

Lastly, it can be hypothesized that average global brand ownership is higher in
more-developed countries than less-developed countries, since it is the more-developed
countries that are homes of global brands in general and consumers in developed
countries have on average greater purchasing power. Thus, it is hypothesized:

H45: GDP per capita is positively related to the likelihood of GBO.

In the next section, the theoretical background for the research method is

presented.

38

Chapter 3

RESEARCH lVIETHOD

3.1 Unit of Analysis

Model variables describe both consumers and countries. These variables belong to
different units of analysis that form a hierarchical structure; namely consumers are nested
within countries. Two common traditional methods to deal with such a data structure
have been:

1. To disaggregate all higher order (group level) variables to the individual level,
and do the analyses on the individual level,

2. To aggregate the individual level variables to the group level and do the
analyses on the group level.

The problem with the first approach is that the consumers in the same country will
have the same value on each of the country variables. Therefore, the “independence of
observations” assumption, which is basic for classical statistical techniques, does not hold
(Raudenbush and Bryk 2002).

The second approach aggregates consumer characteristics over countries, and
does a higher level analysis. However, all the within-group information is lost, the
relations between aggregated variables often seem to be much stronger than what they
actually are, and the relations between the aggregate variables can be very different from
the relations between the nonaggregate variables. Waste of information and distortion of

interpretation are the downsides of this second approach (Raudenbush and Bryk 2002).

39

Individuals in the same group are often closer or more similar than individuals in
different groups. Consumers in different countries can be independent, but consumers in
the same country share the same value on certain variables. If unaccounted for, these
unobserved variables vanish into the error term of the linear model and cause correlation
between disturbances. The disturbances have a group and an individual component.
Group components are correlated within groups and independent between groups,
whereas individual components are all independent. In addition, some groups may be
more homogeneous than other groups, thus the variance of the group components can
differ (Raudenbush and Bryk 2002).

Therefore, combining all variables that belong to different levels of analysis into
one regression equation undermines two basic assumptions of traditional linear model
analysis: homoscedasticity and independence.

One approach to solve these problems is to include an effect in the model that
corresponds to the grouping of the lower-level units, thus employing ANOVA or
ANCOVA. However, there are a number of problems with this approach (Luke 2004). As
the number of groups increase, there are more parameters to estimate, the model has less
power, and greater complexity. The treatment of group parameters as fixed effects
ignores the random variability of the group characteristics. Furthermore, ANOVA
methods are not very flexible in handling missing data or unbalanced designs (Luke
2004).

Hierarchical Linear Modeling (Bryk and Raudenbush 1992; Raudenbush and
Bryk 2002) was developed to deal specifically with hierarchical (nested) data in

education research. This method avoids the weaknesses outlined above. Each of the

40

levels in the data structure has its own sub-model, which indicates the relationships
among variables within a given level and specifies how variables at one level influence
relations occurring at another (Raudenbush and Bryk 2002). Disciplines such as
sociology, biometrics, econometrics, and statistics have all contributed to the
development of such models for nested data structure. Although not very common,
organizational behavior (see Hofmann 1997; Hofmann and Gavin 1998; Klein et al.
1994) and strategic management research (see Song et al. 2002) also utilize this
technique. Other names used in different literatures include multi—level linear models,
mixed-effects models, random-effects models, random-coefficient regression models,
covariance components models, etc. A more generalized version of HLM is called
“Hierarchical Generalized Linear Modeling (HGLM)” and is employed when the
dependent variable is discrete (see Goldstein 1991; Wong and Mason 1985). Both HLM,

and HGLM are quite new to the marketing discipline.

3.2 Hierarchical Linear Modeling (HLM)

A simple example of HLM follows for a 2-level model, where there are
individual- (level-1) and group-level (level-2) variables.

To first isolate and then account for the effects of group-level variables, the
individual-level variables are modeled as having a separate regression equation for each
group. The parameters of these regression equations are then regressed on the group-level

variables. This procedure lets group-level variables be used to explain variation in the

41

individual-level parameters and allows testing for main effects and interactions within
and between levels.

Assume there are 2 level—l variables (x. and x2) and 1 level-2 variable (0).). The
value of the dependent variable (Yij) can be predicted from the values of level-1

independent variables. The regression equation is:

Yij=l301+l3u * Xlij + 321 * m + Fir [3-11
“i” refers to the person number and “j” refers to the group number.

Each group will have a separate regression equation, and the coefficients BO, [31,
and [32 will be allowed to change from group to group. Further analyses can explain their
variability. Thus, level-2 regression equations are formed to predict the value of the level-
1 parameters using values of the level-2 independent variable:

[301' =Y00+Y01 * (Du +110j 13-2]
Bij=Yio+Yrr *wrj+urj 13-3]
32;“ =720+Y21 * (Du +U2j [3-4l
Note that there is a separate equation for each parameter in Equation [3.1].
If the level-2 equations [3.2], [3.3], and [3.4] are substituted into the level-1
equation [3.1], the combined model is:
Yij = 700 + 'Yor * (Du + qu
+ (Ylo + 711* (On + ulj) * Xttj
+(Y20+'th * (Du +l12j) * X2ij +fij 13-5]
Coefficients that are allowed to vary from group to group, e.g., BOj, [3”, and sz are

referred to as random coefficients. Coefficients Y- are not assumed to vary across

groups (and hence they lack the subscript j); therefore, they are referred to as “fixed”

42

 

 

coefficients (Hox 1995). Standard OLS cannot be used to estimate this equation. As
reviewed above, one necessary condition to conduct OLS is that the random errors are
independent, normally distributed, and have constant variance. Note that the random error
in Equation [3.5], which is [qu + (Ulj * Xlij) + (Uzj * X2ij) + rij ], is not independent across
groups since the components uoj, u”, and u2J- are common to every individual within group
j. The errors do not have equal variances either, since u0j, Ulj, and UZj vary across groups
and Xlij and x30- vary across individuals. Although standard regression analysis is
inappropriate, iterative maximum likelihood procedures can be used to estimate such
models (Raudenbush and Bryk 2002).

Note that if UOj, ulj, and uzj were null for every j, Equation [3.5] would be

equivalent to an OLS regression model.

3.3 Hierarchical Generalized Linear Models (HGLM)

In the focal model of this study, the outcome variable, global brand ownership, is
dichotomous. When the dependent variable is dichotomous, linear regression methods
should not be employed for two reasons (Snijders and Bosker 1999). First, although the
range of dichotomous variables is restricted, the ordinary linear regression model might
give a fitted value which is outside this allowed range. For example, global brand
ownership can be represented as 0 (failure; the consumer does not own the brand) and 1
(success; the consumer owns the brand). A fitted value of “0.80” can be interpreted as a

probability of 80% for success (owning the brand) and a probability of 20% for failure

43

(not owning the brand). Getting a fitted value of “1.1,” however, would not be
meaningful.

Second, the mean and the variance of a Bernoulli distribution are related; the
variance is determined by the mean (Snijders and Bosker 1999). For a dichotomous
variable Y that has probability p for outcome 1 (success), the probability for outcome 0
(failure) is l-p. The mean and variance are:

E (Y) = P

Var (Y) = p 0-1)) [3.6]

Thus, the parameters in the fixed part and the parameters of the random part of the
model are related. The variance is not a free parameter, but it is determined by the mean.
The variance of the level-1 random effects is a function of the predicted value; higher
predicted values are associated with larger residual variance (Raudenbush and Bryk
2002).

Two assumptions of HLM are linear relationship between the predictors and the
dependent variable, and normality of the random effects (Raudenbush and Bryk 2002).
These assumptions are violated when the dependent variable is dichotomous. The level-1
random effect cannot be normally distributed since it can take only two values contingent
on the predicted value of the outcome. This can be shown as follows:

In a multi-level setting, the dichotomous outcome, Yij for level-1 unit i in group j,
can be represented as the sum of the probability (average proportion of successes) in
group j plus a residual for the individual i (assuming the probability of success is constant
in each group).

Yij = Pj + Rij [3.7]

 

 

The residual Rij has a mean zero; however, it can have only the values -Pj and
l—Pj, since Yij must be 0 or 1. Given the value of the probability Pj, the variance of the
residual is:

Var (Rii) = Pj (l-Pj) [3.8]

When the outcome variable is dichotomous, the linearity assumption does not
hold either. That the dependent variable must lie between the (0, 1) interval requires a
nonlinear transformation of the predicted value. Thus, the outcome is modeled as a
nonlinear function of explanatory variables. The analysis of nonlinear structural models
and nonnormally distributed errors is accomplished by hierarchical generalized linear
models (HGLMs). The level-1 model in HGLM consists of three parts: a sampling
model, a link function, and a structural model.

1. Level-1 Sampling Model

It is assumed that the level-1 outcome variable arises from a specific probability
distribution, holding constant the level-1 expected value. This level-1 probability

distribution is called the level—1 sampling model.

Ila-[(0,3- ~ B(mU9 Q1) [3.9]

In the Equation 3.9, Yij is defined as the number of successes in mij trials and (pij
as the probability of success in each trial. Yij has a binomial distribution with mij trials
and probability of success per trial as (pij. The expected value and variance of Yij are

shown in Equation 3.10.

E(Yij (pa) = mug)” , var(Yij|¢ij) = mrjgoij(1_ (pry) [310}

45

When mi]- = 1, Yij can only take a value of zero or one. This distribution is also
known as the Bernoulli distribution and it is the focal case here.

2. Level-l Link Function

Instead of the probability of some event, e. g., global brand ownership, one can
consider the odds (Snijders and Bosker 1999). The odds of success are defined as the

ratio of the probability of success ((pij) to the probability of failure (1- (pg)

¢ij
1 — (pg.

Odds =

 

[3.11]

Unlike probabilities, odds can take any value from zero to infinity.

“Link function” is the general term for a transformation function. It links the
expected values of the dependent variable to the explanatory variables. “Log-odds” is one
of the most frequently used link function for probabilities. The logit function is an
increasing function defined for numbers between 0 and 1, and its range goes from -00 to
+00. The level—1 predicted value, (pij, is transformed via this link function to ensure that

the predicted values lie in the given interval. Equation 3.12 shows the logit link function.

(pit
1_ l, [3.12]

ij

 

77;] = ln(

where llij is the log-odds (natural logarithm) of the odds of success.

When the probability of success, (pg, is .5, the odds of success are ‘Pij / (1- (pg) =
.5/.5 = 1 and the log-odds is 0. When the probability of success is less than .5, the odds
are less than 1 and the logit is negative, whereas when the probability of success is more
than .5, the odds are more than 1 and the logit is positive. By forming the link function,

(pg is constrained to stay in the interval (0, 1), whereas nij can take any real value.

46

3. Level-1 Structural Model

The level-1 link function, nu, can be equated to a linear model having level-1
coefficients. Level-l structural model refers to this linear model. A predicted log-odds is
converted to an odds by taking the exp (nu), whereas it is converted to a predicted
probability by the logistic function stated in Equation 3.13. The logistic and logit
functions are the inverses of each other.

2 exp(n,-,) = 1
1+eXP(n.,-) 1+exp(—n,,)

 

 

ij [3.13]

Please note that the value of (pi,- will be in the interval (0, 1).
HLM can be considered as a special case of HGLM where the sampling model is
normal, the link function is the identity link, and the structural model is linear

(Raudenbush and Bryk 2002).

3.4 Data

For level-1 variables, proprietary data were obtained from a large, global
marketing research consultancy. The data were collected in late 2003 via one-hour
closed-ended interviews with consumers. The samples were designed to represent the
national population aged 13-65. A multistage, stratified area probability sample was used.
For the selection of households, interviewers followed randomly selected, predetermined
interviewing routes with skip intervals. All interviews were conducted face-to-face at
respondents’ homes. The sample size was 31,397 respondents from 31 countries. The

countries included in the research are: Argentina, Australia, Brazil, Canada, China, Czech

47

 

 

Republic, Egypt, France, Germany, Hong Kong, Hungary, India, Indonesia, Italy, Japan,
South Korea, Mexico, Philippines, Poland, Russia, Saudi Arabia, Singapore, South
Africa, Spain, Sweden, Taiwan, Thailand, Turkey, U.K., USA, and Venezuela.

The research includes thirty-six global brands, which span 6 categories:
automotive, technology, media services, consumer goods, personal care, and financial
services.

Public secondary data were used for the level—2 variable. Gross domestic product
(GDP) per capita values were taken from The World Factbook (2004) of the Central

Intelligence Agency (CIA).

3.5 Model Variables and Their Descriptions

The variables used in the model were global brand associations (Quality,
Innovativeness, Belongingness, Prestige, Globalness, Genuineness, and Social
Responsibility), consumer characteristics (Age, Income, RSE, and RC) and country
characteristics (GDP per Capita). All the brand variables are dichotomous; consumer and
country variables are continuous. The names and descriptions of the variables are

provided in Table 3.1.

48

 

Table 3.1 - Description of Variables

 

Variable

Description

 

Level-l Variables

 

Perceived Brand Quality

QUALITY is a dichotomous, uncentered variable, coded

1 if the brand is perceived as high quality and 0 if it is not.

 

Perceived Brand

Innovativeness

IN NOVATI is a dichotomous, uncentered variable, coded

1 if the brand is perceived as innovative and 0 if it is not.

 

Perceived Brand

Belongingness

BELONGIN is a dichotomous, uncentered variable, coded
1 if the consumer feels s/he is part of a group connected

by the brand and 0 if s/he is does not.

 

Perceived Brand Prestige

PRESTIGE is a dichotomous, uncentered variable, coded

1 if the brand is perceived as prestigious and 0 if it is not.

 

Perceived Brand

Globalness

GLOBAL is a dichotomous, uncentered variable, coded 1

if the brand is perceived as global and 0 if it is not.

 

Perceived Brand

Genuineness

GENUINE is a dichotomous, uncentered variable, coded

1 if the brand is perceived as genuine and 0 if it is not.

 

Perceived Brand Social

SOCRESP is a dichotomous, uncentered variable, coded 1

 

 

Responsibility if the brand is perceived as being socially responsible and
0 if it is not.

Age AGE is a continuous variable. It is grand-mean centered
and expressed in 103.

Income INCOME is a continuous variable calculated by using

Purchasing Power Parity (PPP) for consumers from
different countries. It is grand-mean centered and

expressed in 1,0005.

 

Resultant Self-
Enhancement (RSE)

RSE is a continuous variable calculated from Schwartz’s

value dimensions. It is grand-mean centered.

 

 

Resultant Conservatism

(RC)

 

RC is a continuous variable calculated from Schwartz’s

value dimensions. It is grand-mean centered.

 

49

 

Table 3.1, cont.

 

 

 

Variable Description
Level-2 Variable
GDP per Capita GDPCAPIT is a continuous variable; it is GDP on a

purchasing power parity basis divided by population. It is

grand-mean centered and expressed in 1,0005.

 

 

 

While dichotomous variables were left uncentered, all of the continuous variables
were grand-mean centered. Grand-mean centering of a predictor was done by subtracting
the grand mean of the predictor from the original predictor for each level-1 case. Grand-
mean centering makes the interpretation of the intercept term meaningful (Raudenbush
and Bryk 2002). The intercept term is interpreted as the expected value of the outcome
variable when the values of explanatory variables are zero. For example, if AGE were left
uncentered, the intercept would show the log—odds of brand ownership for a consumer
who is zero years old, which is not meaningful. Grand-mean centering makes the
intercept the expected value of the outcome variable for an individual with an “average”
level of explanatory variable, e.g., average age. Previous research found that grand-mean
centered models provide a computational advantage by reducing the correlation between
the intercept and slope estimates across groups (Kreft et al. 1995). The reduction of the
correlation between the random intercepts and slopes can help to mitigate potential level-
2 estimation problems due to multicollinearity (Kreft et al. 1995).

GDPCAPITA variable used purchasing power parity (PPP) basis instead of
official exchange rate. GDP at PPP is the total value of all goods and services produced

in the country, valued at market prices in the United States. According to the CIA (2004),

50

 

“This is the measure most economists prefer when looking at per-capita welfare and
when comparing living conditions or use of resources across countries.”

The level-1 INCOME variable that was originally coded in local currency was
also converted to its PPP equivalent. The World Bank’s 2004 World Development
Indicators were used. PPP Conversion Factor is “the number of units of a country’s
currency required to buy the same amount of goods and services in the domestic market
as a US. dollar would buy in the United States” (WorldBank 2004). By making this
conversion, comparison of the income levels between consumers from different countries
is made meaningful.

Transformation of Resultant Self-Enhancement (RSE) and Resultant
Conservation (RC) were calculated as follows. First, mean values for each of the 10 value
types were calculated by averaging the importance scores of individual values. Then, the
mean of the value types for each domain was calculated (Table 3.2). This method ensures
equal weighting of the value types in the formation of the value domains (Schwartz
1992). As proposed by Feather (1995), the score for resultant self-
enhancement/conservation was computed by subtracting the mean importance score for
self-transcendence/openness to change from the mean importance score for self-
enhancement/conservation (Table 3.3).

Table 3.2 - Value Domains

 

 

 

 

 

Value Domain Value Types
Self-transcendence Universalism, benevolence
Self-enhancement Power, achievement, hedonism
Openness to change Stimulation, self-direction
Conservation Tradition, conformity, security

 

 

 

 

51

 

Table 3.3 - Value Poles

 

Value Poles

Calculation

 

Resultant self-enhancement (RSE)

Self-enhancement — Self-transcendence

 

Resultant conservation (RC)

 

 

Conservation — Openness to Change

 

Finally, a list of interaction terms is provided in Table 3.4.

Table 3.4 — List of Interaction Terms

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Level-1 Interaction Term Notation Description
INCQUAL INCOME*QUALITY
INCINNOV INCOME*INNOVATI
INCBELON INCOME*BELONGIN
INCPREST INCOME*PRESTIGE
INCGLOBA INCOME*GLOBAL
INCGENUI INCOME*GENUINE
INCSOCRE INCOME*SOCRESP
AGEQUAL AGE*QUALITY
AGEINNOV AGE*INNOVATI
AGEBELON AGE*BELONGIN
AGEPREST AGE*PRESTIGE
AGEGLOBA AGE*GLOBAL
AGEGENUI AGE*GENUINE
AGESOCRE AGE*SOCRESP
RSEQUAL RSE*QUALITY
RSEINNOV RSE*INNOVATI
RSEBELON RSE*BELONGIN
RSEPREST RSE*PRESTIGE
RSEGLOBA RSE*GLOBAL

 

52

 

 

Table 3.4, cont.

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Level-l Interaction Term Notation Description
RSEGENUI RSE*GENUINE
RSESOCRE RSE*SOCRESP
RCQUAL RC*QUALITY
RCINNOV RC*INNOVATI
RCBELONG RC*BELONGIN
RCPREST I RC*PRESTIGE
RCGLOBAL RC*GLOBAL
RCGENUIN RC*GENUINE
RCSOCRES RC*SOCRESP
GLOBQUAL GLOBAL*QUALITY
GLOBINNO GLOBAL*INNOVATI
GLOBBELO GLOBAL*BELONGIN
GLOBPRES GLOBAL*PRESTIGE
GLOBGENU GLOBAL*GENUINE
GLOBSOCR GLOBAL*SOCRESP
3.6 Model Specification

3.6.1 Fully Unconditional Model

 

Formulation of a fully unconditional model, where no predictors are specified at
either level, is suggested as the first step by Raudenbush and Bryk (2002). The purpose of
this stage is to see how much variation in the outcome lies within and between countries.
The level-1 model is given in Equation 3.14, and level-2 model is given in Equation 3.15.

ml = [301 [3'14]

Bpj = too+ no; qu ~ N (0. Too) [3.15]

53

where yoo is the average log-odds of global brand ownership across countries, and too is
the variance between countries in country-average log-odds of repetition. The estimated
results are 700 = -l.084824 (se = 0.051468), too = 0.07687 (sd = 0.27726). The
interpretation is that for a country with a typical global brand ownership rate (with uoj =
0), the expected log-odds of global brand ownership is -1.084824. This corresponds to an
odds of exp (-1.084824) = .34 and a probability of 1 / (1 + exp (1084824)) = .25.
Equation 3.14 does not include a level-1 residual because it is an equation for ‘lij’ not for
Yij.

Here, the explanation of an important index used in multi—level models, intra-class
correlation, is necessary and the explanation requires a detour.

Intra-Class Correlation (ICC)

In a traditional HLM, a fully unconditional model would be of the form:

vi,- = so, + n, r0,- ~ N (0, 02) [3.16]

B0] = Too + qu qu ~ N (0. Too) [3.17]

This model is equivalent to a one-way ANOVA with random effects. ij is
modeled as a function of an individual-level mean and a random error. More specifically:

Yij is the outcome for the individual i in group j

l30j is the mean outcome for the jth group

rij is a random person (level-1) effect. Each level—l error, rij, is assumed to be
normally distributed with a mean of zero and a constant level-1 variance, 02

700 is the grand-mean outcome in the population

54

uiJ- is the random group (level-2) effect associated with group j and it represents
the deviation of group j’s mean from the grand mean. Each level-2 effect is assumed to
be normally distributed with a mean of O and variance too.

The variance of the outcome is:

Var (Yij) = Var (rij + uij) = 02 + 1'00 [3.18]

The 02 parameter gives within-group variability, whereas the parameter too gives
the between- group variability. The intra-class correlation (ICC) is a widely used statistic
by multilevel researchers. It measures the proportion of the variance in the outcome that
is between groups. Equation 3.19 provides the formula:

p =IOO/(t00+oz) [3.19]

Notice that the model shown in Equation 3,14 and 3.15 does not have a parameter
for the level-1 variance. This is because the level-1 residual variance of the dichotomous
outcome variable directly comes from the success probability (Snijders and Bosker
1999), as shown in the Equation 3.8.

Accordingly, the partitioning of variance into different levels is not very simple in
HGLM. Decomposing variance of an outcome variable into different levels is most
appropriate when the outcome variable is continuous and the error terms are additive and
independent. Traditional HLM analyses usually model the realized values of the outcome
variable as a deterministic function of independent variable(s) plus error terms, which are
realized values of random variables. The models are considered "perfect" when they can
explain the actual values of the outcome variable without error terms, i.e., when there is

no residual variance. When the models are not “perfect,” the left-over variance can be

55

attributed to the error term(s) and apportioned among them according to each one's
variance (Feiveson 2001).

When the outcome variable is binary, however, multi-level researchers usually do
not model the “values” of it, but they rather model the probability, p, that the outcome
variable is l as a nonlinear function of the independent variables and error terms at Level
2 or higher. Even if the model is "perfect," (i.e., the higher level error terms have zero
variance) there would still be level-1 variation, which is p*(l-p), since p is modeled, not
the value of the outcome variable (Feiveson 2001). This would still be the case even
though p is “perfectly” modeled. If p is not “perfectly” modeled, decomposition of the
sources of variation at level-2 or higher may still be desirable; however, these error terms
contribute to the overall higher-level variance in a non-additive way. Thus, each error
term's contribution to the hi gher-level variance cannot be claimed to be proportional to its
own variance (Feiveson 2001). Therefore, the intraclass correlation, which is a very
useful index for hierarchical linear models, is not very informative for hierarchical
generalized linear models (Raudenbush and Bryk 2002).

It is still theoretically possible to get this index. The multilevel logistic regression
can be formulated as a so-called threshold model. The assumption is that the dichotomous
outcome variable is the manifestation of an underlying latent continuous variable. The
level-1 residual of the underlying latent variable has a logistic distribution with a mean of
O and a variance of n2/3 = 3.29 (Snijders and Bosker 1999). Thus, the ICC formula
becomes:

p = “[00 / (:00 + n2/3) [3.20]

56

The estimate is provided in Equation 3.15 as too = 0.07687. Using this value in
Equation 3.20 gives p = 0.07687 / (0.07687 + 3.29) = 0.0228. This means that 2.28% of

the variability in the global brand ownership variable is between the countries.

3.6.2 Level-l Conditional Model

The level-1 conditional model is formulated by using the log-odds of global brand
ownership as the outcome. The model is indicated below with i = 1, 2, ..., nij consumers
nested within each of j = 1, 2, ..., j countries.
m = BOj + B]j*(QUALITY)ij + B2j*(INNOVATI)ij + B3j*(BELONGIN)ij +
B4j*(PRESTIGE)U + B5j*(GLOBAL)ij + B6j*(GENUINE)ij + B7j*(SOCRESP)ij +
B8j*(INCOME)ij + [39j*(AGE)ij + BIOj*(RSE)ij + Br tj*(RC)ij + Brzj*(INCQUAL)ij +
B.3j*(INcrNNOV).,- + B14j*(INCBELON)ij + B.5j*(INCPREST)ij + B.6,-*(1NCGLOBA).,- +
Bl7j*(INCGENUI)ij + B18j*(INCSOCRE)ij + B|9j*(AGEQUAL)ij + B20j*(AGEINNOV)ij +
B21j*(AGEBELON)ij + B22j*(AGEPREST)ij + D23j*(AGEGLOBA)ij + B24j*(AGEGENUI)ij
+ B25j*(AGESOCRE)U + B26J'*(RSEQUAL)U + B27j*(RSEINNOV)ij + B23j*(RSEBELON)ij
+ B29j*(RSEPREST)ij + B30j*(RSEGLOBA)ij + B31j*(RSEGENUI)ij + B32j*(RSESOCRE)ij
+ B33j*(RCQUAL)ij + D34j*(RCINNOV)ij + B35j*(RCBELONG)ij + B36j*(RCPRESTI)ij +
B37j*(RCGLOBAL)ij + B38j*(RCGENUIN)ij + B39j*(RCSOCRES)ij + B40j*(GLOBQUAL)ij
+ B4|j*(GLOBINNO)ij + B42j*(GLOBBELO)ij + B43j*(GLOBPRES)ij +
B44j*(GLOBGENU)ij + B45j*(GLOBSOCR)ij [3.21]

In Equation 3.14, flu is the log-odds of global brand ownership for consumer i in

country j. The parameter Boj is the intercept term. Bpj are the slopes for the level-1

variables, associated with country j.

3.6.3 Level-2 Unconditional Model

As in linear HLM, the level-1 intercept and slope terms become outcome
variables in the level-2 model in HGLM. The hypotheses require a test of whether the
relationships between AGE, INCOME, RSE, and RC and Global Brand Ownership,

together with the average global brand ownership, vary across countries or not. Thus,

57

respective coefficients, together with the intercept term, are modeled as random. This is

expressed as follows:

BPI=YPO+UPI ifP=098,9,10,11
Bpj = ypo otherwise [3,22]

Estimation of variance components are provided in Table 3.5. The results suggest
that there is no significant variation across countries to model the relationship between

RC and the dependent variable (p > .500).

Table 3.5 - Estimation of Variance Components

 

 

 

 

 

 

Random Effect Standard Variance Chi-square P-value
Deviation Component df

INTRCPTl, U0 0.25994 0.06757 30 210.02038 0.000

INCOME slope, U8 0.04048 0.00164 30 54.67346 0.004

AGE slope, U9 0.04362 0.00190 30 42.94462 0.059

RSE slope, U10 0.04115 0.00169 30 40.87888 0.089

RC slope, U11 0.01526 0.00023 30 27.61276 >.500

 

 

 

 

 

 

 

 

3.6.4 Level-2 Conditional Model

When building the conditional model for level-2, nonsignificant variation in the
RC slope is taken into account, and this variable is treated as fixed in the subsequent
analyses. The relationships between age and the likelihood of GBO, income and the
likelihood of GBO, and RSE and the likelihood of GBO are identified as group-variant.

Therefore, in order to better understand to what degree this variation is a function of

58

economic development levels of the countries, GDPCAPIT variable is added to the

model. The resulting level-2 conditional model is as follows:

Bpj = YpO + Ypl*(GDPCAPIT)j + Upj If p = 0, 8, 9, IO
Bpj = Ypo otherwise [3.23]
3.6.5 Mixed Model

Substituting Equation 3.23 in Equation 3.21 gives the mixed model:

11,,- = 700 + yo. *(GDPCAPITA)j + y.0*(QUALITY)U + y20*(INNOVATI)ij +
730*(BELONGIN)U + Y40*(PRESTIGE)§j + Y50*(GLOBAL)U + 760*(GENUINE)U +
770*(SOCRESPM + Y80*(INCOME)U + Y8l *(GDPCAPITA)j*(INCOME)ij + 790*(AGE)ij +
Y9I*(GDPCAPITA)j *(AGEM + Y100*(RSE)ij + 'YIOI*(GDPCAPITA)J*(RSE)ij +
7110*(RCIij + 7120*(1NCQUAth + Y130*(1NCINNOV)ij + 7140*(INCBELONhj +
Y150*(INCPREST)ij + 7160*(INCGLOBA)U + Y]70*(INCGENUI)U + 7180*(INCSOCRE)ij +
7190*(AGEQUAL)U + 7300*(AGEINNOVM + 7210*(AGEBELONXJ' + 7220*(AGEPREST)U
+ 7230*(AGEGLOBA)U + Y340*(AGEGENUI)U + Y250*(AGESOCRE)ij +
7260*(RSEQUALM + Yz70*(RSEINNOV)ij + 7280*(RSEBELON)ij + 7290*(RSEPRESTM +
7300*(RSEGLOBA)ij + Y3]0*(RSEGENUI)U + Y320*(RSESOCRE)ij + Y330*(RCQUAL)ij +
7340*(RCINNOV)U + 7350*(RCBELONGM + 7360*(RCPRESTI)ij + Y370*(RCGLOBAL)U +
7380*(RCGENUINM + 7390*(RCSOCRESMJ' + Y4oo*(GLOBQUAL)ij +
Y410*(GLOBINNO)ij + Y420*(GLOBBELO)ij + 7430*(GLOBPRESM +
Y440*(GLOBGENU)U + Y450*(GLOBSOCR)U + Lloj + u8j*(INCOME)ij + U9j*(AGE)ij +
u10j*(RSE)ij [3.24]

59

4.1 Descriptive Statistics

Chapter 4

RESEARCH RESULTS

The summary statistics for consumer-level variables and the county-level variable

is presented in Table 4.1.

Table 4.1 - Descriptive Statistics

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

LEVEL-l DESCRIPT IVE STATISTICS
VARIABLE N MEAN STANDARD MINIMUM MAXIMUM
NAME DEVIATION
QUALITY 31337 0.35 0.48 0.00 1.00
INNOVATI 31337 0.21 0.41 0.00 1.00
BELONGIN 31337 0.19 0.39 0.00 1.00
PRESTIGE 31337 0.24 0.43 0.00 1.00
GLOBAL 31337 0.23 0.42 0.00 1.00
GENUINE 31337 0.14 0.34 0.00 1.00
SOCRESP 31337 0.10 0.30 0.00 1.00
BOWNERSH 31337 0.25 0.43 0.00 1.00
INCOME 26863 -0.00 2.06 -2.34 17.00
AGE 31329 0.00 1.45 -2.11 2.74
RSE 31397 -0.00 1.77 -5.75 7.67
RC 31397 0.00 1.77 -7.16 6.30
INCQUAL 29772 0.02 1.20 -2.29 14.79
INCINNOV 30417 0.00 0.88 -2.34 14.79
INCBELON 30668 -0.03 0.82 -2.29 14.79

 

60

 

Table 4.1, cont.

 

LEVEL-1 DESCRIPT IVE STATISTICS

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

VARIABLE N MEAN STANDARD MINIMUM MAXIMUM
NAME DEVIATION

INCPREST 30294 0.01 1.00 —2.29 14.79
INCGLOBA 30388 0.04 0.98 -2.34 14.79
INCGENU I 30746 -0.00 0.73 -2.34 14.79
INCSOCRE 30909 0.00 0.67 -2.29 14.79
AGEQUAL 31319 -0.01 0.85 -2.11 2.74
AGEIN NOV 31326 -0.03 0.66 -2.1 1 2.74
AGEBELON 31329 -0.01 0.63 -2.11 2.74
AGEPREST 31314 -0.01 0.70 -2.11 2.74
AGEGLOBA 31319 -0.00 0.67 -2.11 2.74
AGEGENUI 31321 -0.00 0.53 -2.11 2.74
AGESOCRE 31329 -0.01 0.45 -2.1 l 2.74
RSEQUAL 31337 0.00 1.05 -5.56 7.67
RSEINNOV 31337 0.02 0.82 -5.56 6.78
RSEBELON 31337 0.03 0.76 -5.67 7.67
RSEPREST 31337 0.02 0.89 -5.58 6.64
RSEGLOBA 31337 0.01 0.85 -5.67 7.42
RSEGENUI 31337 0.01 0.65 -5.67 7.67
RSESOCRE 31337 0.01 0.56 -5.56 6.78
RCQUAL 31337 -0.01 1.06 -6.92 6.24
RCINNOV 31337 -0.00 0.82 -6.92 6.24
RCBELONG 31337 0.02 0.77 -6.49 6.30
RCPRESTI 31337 -0.01 0.88 -6.92 6.03
RCGLOBAL 31337 -0.01 0.85 -7.16 6.24
RCGENUIN 31337 -0.00 0.65 -6.59 6.30
RCSOCRES 31337 0.00 0.56 -6.80 6.24

 

 

 

 

 

 

61

 

Table 4.1, cont.

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

LEVEL-l DESCRIPT IVE STATISTICS

VARIABLE N MEAN STANDARD MINIMUM MAXIMUM
NANIE DEVIATION

GLOBQUAL 31337 0.10 0.30 0.00 1.00

GLOBINNO 31337 0.07 0.25 0.00 1.00

GLOBBELO 31337 0.06 0.23 0.00 1.00

GLOBPRES 31337 0.08 0.28 0.00 1.00

GLOBGEN U 31337 0.05 0.21 0.00 1.00

GLOBSOCR 31337 0.04 0.20 0.00 1.00

LEVEL-2 DESCRIPT IVE STATISTICS

GDPCAPIT 31 16.63 10.34 2.90 37.80

 

 

 

 

 

 

 

 

For the purposes of the analysis, the HLM 6.02 (Raudenbush et al. 2000) software
was used. The two most frequently used estimation methods for hierarchical generalized
linear models are MQL and PQL (Snijders and Bosker 1999). MQL (marginal quasi-
likelihood) does the approximation around the estimated fixed part, whereas PQL
(penalized quasi-likelihood) does it around an estimate for the fixed part and the random
part (Snijders and Bosker 1999). Guo and Zhao (2000) recommend the use of PQL over
MQL for binary data. PQL, which is also the default estimation method used by the
program, is chosen as the estimation procedure here.

The program provides two sets of estimated coefficients: one for unit-specific
models, the other for population average models. When the research questions are unit-
specific, e. g., when one investigates how differences in level-2 explanatory variables

relate to the level-1 processes in each level-1 unit, unit-specific model is appropriate.

62

This study tries to find an answer to a population-average question. The topic of
interest is not country-specific; not specific countries but rather the differences between
countries that are more-developed versus less-developed are investigated. Thus,
population-average model estimates are appropriate and they are used. It should be noted
that population-average estimates have the additional benefits of being quite robust to
erroneous assumptions about the distribution of random effects in the model

(Raudenbush and Bryk 2002).

4.2 Estimated Results

4.2.1 Direct Effects
Table 4.2 shows the hypothesized direct effects of the consumer- and country-
level variables on the likelihood of global brand ownership, together with the analysis

results.

Table 4.2 - Direct Effects

 

 

 

 

Direct Effects Hypothesized Coefficient t-ratio Resultant
Directional Relationship
Relationship
w/GBO
Perceived Brand Quality Positive 0.181040 5.234*** Positive
Perceived Brand Positive 0.261474 6.414*** Positive
Innovativeness
Perceived Brand Positive -0.058745 -1.326 NS
Belongingness

 

 

 

 

 

 

 

63

Table 4.2, cont.

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Direct Effects Hypothesized Coefficient t-ratio Resultant
Directional Relationship
Relationship
w/GBO
Perceived Brand Prestige Positive -0.446787 -10.326*** Negative
Perceived Brand Positive 0.228780 4.667*** Positive
Genuineness
Perceived Brand Social Positive 0.103335 1.653** Positive
Responsibility
Perceived Brand Positive 0.001675 0.031 NS
Globalness
Age Negative -0.076071 -4.350*** Negative
Income Positive 0.053120 3.689*** Positive
RSE Positive 0.009460 0.663 NS
RC Negative 0.041719 3.188*** Positive
GDP per Capita Positive 0.004895 0.987 NS
***p<.01 **p<.05

All of the brand associations were hypothesized to be positively related to the

 

likelihood of global brand ownership. All but belongingness and globalness associations
do appear to have significant direct effects on the likelihood of global brand ownership.
Perceived brand prestige association was negatively related to the likelihood of GBO (p <
.01). [H], H2, H5 and H6 are supported; H3, H4, and H7 are not.]

The direct effects of consumer demographics are as expected. Age was negatively
related (p < .01) and income was positively related (p < .01) to the likelihood of GBO.

[H16 and H25 are supported] The direct effect of RSE was found to be nonsignificant. The

64

direct effect of RC was found to be significant (p<.01) but positive, in contrast to what
was hypothesized. [H34 and H43 are not supported]

The direct effect of GDP per Capita on the likelihood of GBO was nonsignificant.
[H45 is not supported] The level of economic development does not seem to explain the

differences in the average odds of global brand ownership between countries.

4.2.2 Moderating Effects
Table 4.3 shows the hypothesized moderation effects of Age, Income, RSE, and
RC on the relationships between the perceived brand associations and the likelihood of

global brand ownership, together with the analysis results.

65

 

mz memo 8285 058m 305320 95m 8:85

 

 

 

 

 

 

 

 

 

 

bzfimmcomwom

mZ cmfid mnemood o>Emom Boom venom coZoouom

“Scion .55va Nwmfivod 328m $255:ch 98cm 3&8qu—

o>Emom *mmmg 35486 o>Emom owcmoi Baum welcocom

m2 omvd wmmwood 328m mmocwfiwaflom 29m @0385.»

ozfiwoz *maNA- 338.0- “Scion 3053355: 985 uoioouom
mZ oomd SoNood o>Ewom bzmsd 985m @3683 2:85

o>Emom .3an .N 3.936 o>uawoz amen—EEO 22m coZoocom

mZ vowd moowmod o>cawoz mmocwEwco—om Baum quoBoL

mZ wmmd mwvwood oifimoz wmoco>§>ocfi ccfim @038qu

 

bEnicommom

 

 

 

 

 

 

 

 

 

 

 

mz ME: N323 258a seem 2.5 8385

m2 53 388.0 228a 322550 BE eazoeaa

mz Name- 228.0- 22.8 amass 82m @0385;
m2 $3 335.0 828.. £20 23m 8285 aw<
3:25:20“ fingers—um 12.2825 mosmtogoanasv
2.3—3.3— canué 320580 weaver—Hon»: m=c=£oomm< 35.5 Ego—G BEE—EU

 

macaw—80330 BEE—50 mo $8.5.— wnaauoeez I m4. 933.

66

5.
~q rwvsv‘

=I

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

2.x: avg: 395:
03332 *mvmg- mi; 8.0. 9:832 305320 Enum— @0385;
bzfiucoawom
m2 02.6 8886 m2 Eoow 9.8m cozuouom
m2 Svd- M3556- m2 wmocucEcoO ucfim BZQBQL
m2 026 mwmmood mZ owsmoi Exam 8&8th
mZ 3mg- 588.0- m2 wwocwcmwcoflom wcfim cuZoBum
03332 ***N\.m.v- mwoowod- o>cawuz mwoco>ca>oc£ 235 wozoobm
mZ oowd- nmmiod- mZ 9130 28m cutoflum 0%
m2 mmod- wofi 50.0. 328% 805320 Baum 838“?“
bzémcoamom
o>uawoz ***vmo.~- wmofiwod- ozawoz 300m 985 cozoobm
m2 m2 .0 mommood o>Emom $255550 22m @3680.»
328m ***Nm\l.m mofimod 328m owcfii 28m cozoobm
928m **mow._ wwwwmod Diamoz mmocwszoBm 38m 8385;
03332 "353.7 558.0- u>Emom 8052855: 255 @3688
m2 3nd- $35.0- 3:55 5:20 ESE Bzoohom mmm
35:23.0: aim—8:23— 3:2325 85:23.3:0
2.5.33— 83?» 339580 conic—=25: m:c=n_u°mm< 6.5.5 .326 E

4:60 ans? v—A—NH.

67

Results show that age acts as a moderator only for the relationship between
perceived brand globalness and the likelihood of global brand ownership (p < .02).
Contrary to the hypothesis, this relationship was positive; i.e., the effect of perceived
brand globalness on the likelihood of global brand ownership was stronger for older
consumers than it was for younger consumers. [H5 is not supported] Older and younger
consumers did not seem to differ in terms of the effects of other brand associations on
their purchasing decisions. [H9 through H14 are not supported] One interesting point to
note is that although the direct effect of globalness association on the likelihood of GBO
was not significant, this effect was significant for older consumers.

When the moderation effect of income was analyzed, significant differences were
seen in the impact of perceived brand innovativeness, prestige, and genuineness on the
likelihood of GBO for higher vs. lower income consumers. Ceteris paribus, genuineness
(p < .02) and prestige (p<.07) associations of global brands had more impact on the
likelihood of GBO for consumers with higher income than for consumers with lower
income. [H21 and H32 are supported] Note that while prestige association had a
significantly negative direct effect on the likelihood of GBO, this effect was significantly
positive for higher income consumers. Interestingly, innovativeness association had less
impact on the likelihood of GBO for higher income consumers than for lower income
consumers (p < .10). [H19 is not supported] Other brand associations tested did not have
significantly different effects for higher income vs. lower income consumers on their
purchasing behaviors. [H13, H20, H23 and H24 are not supported]

Belongingness (p < .04) and prestige (p < .01) associations of global brands had

more effect on the likelihood of GBO for higher RSE consumers than lower RSE

68

consumers. [H29 is not supported; H30 is supported] Higher effect of prestige on the
likelihood of GBO was consistent with the hypothesis; however, the higher effect of
belongingness on the likelihood of GBO was surprising. Belongingness association is
related with identification with a group, which aligns more with the dominant values of
self-transcendence. One explanation can be that higher RSE consumers may see
belongingness as a way of enhancing their own self, rather than replacing it. It is worth
noting that the direct effect of belongingness association on the likelihood of GBO was
not significant. This association seems to be only effective for higher RSE consumers.

Social responsibility (p < .01) and innovativeness (p < .03) associations had more
effect on the likelihood of GBO for lower RSE consumers than for higher RSE
consumers. [H28 is not supported; H32 is supported] This was expected for social
responsibility association, but the higher effect of innovativeness on the likelihood of
GBO for lower RSE consumers was contrary to the hypothesis. One explanation may be
that lower RSE consumers may see innovative brands as a means to share with others and
enhance the welfare of others, and thus value the effect of innovativeness association in
their purchasing decisions more than do higher RSE consumers. The effects of other
brand associations on the likelihood of global brand ownership were not found
significantly different for higher RSE vs. lower RSE consumers. [H27 H31, and H33 are not
supported]

Finally, innovativeness (p < .01) and globalness (p < .07) associations of global
brands had less effect on the likelihood of GBO for higher RC consumers than lower RC
consumers. [H37 and H42 are supported] These are in line with the hypothesized effects.

As expected, the effects of other brand associations on the likelihood of global brand

69

 

 

‘I
l
J
F-
an! IU I ' '

 

 

 

 

ownership were not found significantly different for higher RC vs. lower RC consumers.
[H36, H33, H39, H40 and H41 are supported]

Table 4.4 shows the hypothesized moderation effects of Perceived Brand
Globalness (PBG) on the relationship between global brand associations and the

likelihood of global brand ownership, together with the analysis results.

Table 4.4 — Moderating Effects of Perceived Brand Globalness (PBG)

 

 

 

 

 

 

 

Global Brand Hy pothesized Coefficient t-ratio Resultant
Associations Directional Relationship
Relationship
w/PBG
PBG *Perceived Brand Positive -0.071317 -1.020 NS
Quality
PBG *Perceived Brand Positive 0.015328 0.202 NS
Innovativeness
PBG *Perceived Brand Positive -0.051300 -0.630 N S
Belongingness
PBG*Perceived Brand Positive 0.133255 l.728** Positive
Prestige
PBG*Perceived Brand Positive 0.191394 2.158* * Positive
Genuineness
PBG*Perceived Brand Positive 0.127033 1308* Positive
Social Responsibility

 

 

 

 

 

 

 

***p<.01 **p<.05 *p<.10

It was hypothesized that globalness association of brands would intensify the

strength of the relationships between each of the other brand associations and the

70

likelihood of global brand ownership. The results show that the interaction effect was
significant only for prestige (p < .05), genuineness (p < .02), and social responsibility (p
< .10) associations. [H8 is partially supported] One interesting finding to note is that,
although the direct effect of prestige on the likelihood of global brand ownership was

negative, the interaction effect of prestige and globalness was positive.

4.2.3 Random Effects
The details on the result of variance components analysis were already presented
in Table 3.5. Table 4.5 presents the results once more.

Table 4.5 - Random Effects

 

 

 

 

 

 

Random Effect Standard Variance Chi-square P-value
Deviation Component df

INTRCPT 1, U0 0.25994 0.06757 30 210.02038 0.000

INCOME slope, U8 0.04048 0.00164 30 54.67346 0.004

AGE slope, U9 0.04362 0.00190 30 42.94462 0.059

RSE slope, U10 0.04115 0.00169 30 40.87888 0.089

RC slope, U11 0.01526 0.00023 30 27.61276 >.500

 

 

 

 

 

 

 

 

The results show that there is significant variability between countries in terms of
average log-odds of GBO, as well as the direct effect of age, income, and RSE on the
likelihood of GBO. The relationship between RC and the likelihood of GBO did not vary
across countries, and it is treated as fixed (non-varying) in the subsequent analyses. Thus
H44 is rejected right away. Note that previous analyses showed a significant positive
effect of RC on the likelihood of GBO. Lack of variation of this relationship across

countries hints that there is a relatively robust segment of conservative consumers who

71

buy global brands more than do consumers who are open to change. The relationship

between RC and the likelihood of GBO across countries is shown in Figure 4.1.5

Figure 4.1 - Relationship between RC and the likelihood of GBO across Countries

 

Relationship between RC and the Likelihood of GBO

 

 

1.001
_ GDPCAPIT=-11.982

, — GDPCAPIT=13.729
o J
m 0.75‘
(5
h
o 1
'0 a
O 0.50
O 1
:5
E 9—H
3 0.25:///

G i I Y 1 1 v 1 v v v u l t v r r l v 1

-5.71 -2.34 1.02 4.38
RC

 

 

 

Once the group—variant relationships were identified, it was tested whether their
variability could be explained by the country-level variable: GDP per capita. Table 4.6
shows the results of this test of cross-level interactions; the interactions of the consumer-

level variables Age, Income, and RSE with the country-level variable GDP per capita.

 

5 In the model graphs, averaged lower-upper quartiles are used for the range of GDP per capita.

72

Table 4.6 - Cross-level Interactions

 

 

 

 

 

Interaction w/GDP per Capita Coefficient t-ratio Results
GDPCapita*Intercept 0.004895 0.987 NS
GDPCapita*Age -0.000614 -0.476 NS
GDPCapita*lncome -0.001764 -1.539* Negative
GDPCapita*RSE -0.001994 -1.863** Negative

 

 

 

 

 

 

***p<.01 **p<.05 *p<.10

Note that there was significant variability between countries on the average odds
of GBO and the effects of age, income, and RSE on the likelihood of GBO (see Table
4.5). The differences between countries in the relationship between age and the likelihood
of GBO did not seem to be a function of the level of economic development. [H17 is not
supported] GDP per capita is significant in explaining the variability of the relationships
between income and the likelihood of GBO (p < .07), and RSE and the likelihood of
GBO (p < .04). The final estimation of variance components—after GDP per capita is

included in the model—is provided in Table 4.7.

Table 4.7 — Final Estimation of Variance Components

 

Random Effect Standard Variance Chi-square P-value

Deviation Component df

 

 

 

 

 

 

 

 

 

 

 

INTRCPT 1, U0 0.26328 0.06931 29 190.18740 0.000
INCOME slope, U8 0.03926 0.00154 29 50.04304 0.009
AGE slope, U9 0.04716 0.00222 29 46.83109 0.019
RSE slope, U10 0.03789 0.00144 29 38.32455 0.115

 

73

The variation in the relationship between RSE and the likelihood of GBO was not
significant after GDP per capita is included, meaning the variable added explained all the
significant variation between countries. However, there was still unexplained variability
left in the relationships between age and the likelihood of GBO and between income and
the likelihood of GBO. This hints at the existence of other variables that are not included
in the model and that can explain the variation in the relationships. The relationship

between age and the likelihood of GBO across countries is shown in Figure 4.2.

Figure 4.2 - Relationship between Age and the likelihood of GBO across Countries

 

 

 

 

 

 

Relationship between Age and the Likelihood of (580
100*
. — GDPCAPlT=-11.982
j — GDPCAPIT=13.729
J
O .
m 0.75‘
0
'5 .
g 0.50«
5 1
- l
o
x _
5 0.25:
04.1............4r1.
-1.13 0.08 1.29 2.51
Age

 

 

 

Note that the direct effect of RSE was nonsignificant. The significant negative
interaction effect of RSE and GDP per capita implies that the effect of RSE on the
likelihood of GBO is higher in less-developed countries than it is in more-developed

countries. [H35 is supported] In fact, as can be seen from Figure 4.3, as RSE increases,

74

 

 

‘H- 'n

consumers in less-developed countries appear to be more likely to buy global brands than
consumers in more-developed countries; and as RSE increases in more-developed
countries, the likelihood of GBO appears to decrease. Indeed, computation reveals that
the odds of global brand ownership decrease by 0.864 when RSE score and GDP per
capita are two standard deviations higher than the average, whereas they increase by
1.158 when RSE score is two standard deviations higher and GDP per capita is two

standard deviations lower than the average (see Table 4.8 for computation method).

Figure 4.3 -— Relationship between RSE and the likelihood of GBO across Countries

 

Relationship between RSE and the Likelihood of GBO

1.001
, — GDPCAPIT: 41.982

+ — GDPCAPIT: 13.729

 

 

Likelihood of GBO

 

 

/
0.25~F______'_/———f

G V T j Y I T V 1 V I V T U I I V V Y

-5.70 -2.34 1.01 4.36

RSE

 

 

 

The significant negative interaction effect between income and the likelihood of GBO
shows that although there is a positive relationship between income and the likelihood of
GBO in a country with an average level of development, this relationship is stronger for

less-developed countries. [H36 is supported] In fact, as can be seen from Figure 4.4, as

75

income increases, consumers in less-developed countries appear to be more likely to buy
global brands than consumers in more—developed countries. Indeed, computation reveals
that the odds of global brand ownership decrease by 0.86 when income and GDP per
capita are two standard deviations higher than the average, whereas they increase by 1.19
when income is two standard deviations higher and GDP per capita is two standard

deviations lower than the average (see Table 4.8 for computation method).

Figure 4.4 — Relationship between Income and the likelihood of GBO across
Countries

 

 

 

Relationship between Income and the Likelihood of GBO
1.00-
. —- GDPCAPlT=-11.982
— GDPCAPIT:13.729
O .
m 0.75?
0
fi
0 4
1:: .
O 0.50
o .
5 .
g :
3 0.25:
o ....1T........,....1
-2.34 2.49 7.33 12.16 17.00
Income

 

 

 

76

4.3 Interpretation of the Estimated Coefficients

So far, the presentation of the results has been done in terms of their directionality
(positive association vs. negative association) and significance. This section discusses the
interpretation of the magnitude of the coefficients.

In HGLM, the effects of independent variables on the outcome is usually
interpreted using the odds ratio. The coefficients provided in the program output give the
log-odds (nij). Predicted log-odds is converted to odds ratio by taking the exp (nij). It is
converted to a predicted probability by the formula in Equation 3.13.

It should be noted that the logit model is linear and additive for the log-odds, but
multiplicative for the odds. For example, 1 unit increase in AGE changes the logit by [39
and multiplies the odds by egg. The interpretation for AGE, a continuous variable, is as
follows: When AGE is increased by c units, the odds of global brand ownership increase
by a factor of ecl’g, controlling for other predictors. If [39 is 2, increasing AGE by 1 unit
increases the odds by a factor of e2. The interpretation of the effect of a dummy variable
is different. For example, the effect of QUALITY can be interpreted as the odds of global

Bl times greater than the odds for a non-quality

brand ownership for a quality brand are e
brand, controlling for other predictors, where [31 is the estimated effect for QUALITY.

The interpretation of coefficients is done following the ANOVA-like procedure
suggested by Subedi (2005).

The main effects are calculated by taking the exponential of the estimated

parameter coefficients after multiplying it with a constant, c, where c is two standard

77

deviations of the respective continuous variable, e.g., AGE in the example below. The

66 99

constant c is selected as 1 for dichotomous variables.

Main CffCCtAGE = exp (CAGE * ’Ygo) [4.1]

Cross-level interaction effects are calculated by taking the exponential of the
estimated parameter coefficients after multiplying with two constants, where c’s are the
two standard deviations associated with the respective variables, e.g., AGE and GDP in

the example below. A gain. the constant “c” is selected as l for dichotomous variables.

Interaction CffCCIAGE*GDp = exp (CAGE * CGDP * 79]) [4.2]

The interpretation of the significant model coefficients is provided in Table 4.8.

It must be kept in mind that the reference brand is an attribute-free global brand, a
brand that has a value of zero on all the brand-level variables. A reference consumer is a
consumer that has average age, income, RSE and RC scores. A reference country is a
country that has an average GDP per capita.

Thus, the coefficients for the direct effects of brand associations show the change
in odds of brand ownership when the brand is perceived to have the associated attribute
versus when it is perceived not to have the associated attribute, all else being equal (e.g.,
for the quality association, the log-odds of 0.181 translates into an odds ratio of exp

(0.181) = 1.20, meaning the expected odds of ownership of a high quality global brand

78

are 1.20 times the odds of ownership of an otherwise-similar global brand which is not
perceived as high quality).

The interaction effects are calculated in accordance with Equation 4.2. For
example, income is associated with higher likelihood of global brand ownership, for a
reference brand in a reference country, ceteris paribus. Two standard deviations
difference in income is associated with a difference in the log-odds of ownership of
4.12*0.06=0.219 or a relative odds of exp(0.219)=1.245. However, the interaction effect
of income and GDP per capita is negative. For a reference global brand, there is a
decrease in the odds ratio by a factor of exp(4.12*20.68*-0.002)=0.86 due to two
standard deviations increase in income that is associated with two standard deviations
increase in GDP per capita. This means the odds for global brand ownership of a
reference brand are 0.86 times lower for consumers with two standard deviations above-
average income in countries with two standard deviations above-average GDP per capita
than average income consumers in average GDP per capita countries.

Likewise, the effect of e.g., prestige association of a global brand for a consumer
that has two standard deviations higher income is a factor increase of
exp(4.12*1*0.025)=1.107 in the odds ratio. This means that the odds of global brand
ownership are 1.107 times higher for consumers with two standard deviations above-
average income when the brand is prestigious than the base-line case where an average
income consumer and a reference brand is considered.

The graphical representations of significant interaction effects are provided in

Figure 4.5.

79

 

656305: 8 858.6:

:0: mm 553 :55 152w Ezemméfizofio :a 5 952236
.5 55 5.. 88: omAuCcNdExo 2a :55 152w 26965:
:a .5 9:20:30 do ago 8:898 2F .35.: 2: E 28585
.550 2: 2:858 wEEo: 2:20:30 90 movomfl .5sz

5:5 @8383: m_ ..mwo:o>:m>o::r. 8388a .8551520

806

whiomd

ONO

$083355:

 

5:96

:wB mm 8385: 8: mm 5053 :55 850% RESTomSCofio
5» do £52830 do move 2: 8:5 CNAHA—wfiovmxo 2a :85
152w 5:8: :wE a do E52830 5 55 @2898 2C. .358
05 :2 29052: 850 05 5858 mEEo: dEfioEso :0 5:0

-mB 25mm: 5:? 8:588 mm .5295: 8388: .555 1520

oood

0:236

SU

596

 

.mm.ou:2.:-§o
2: >558 8:222 a “E: 5:558 8:222 a 5:3 @2282:

5:5 8:222 a :8 9:20:30 ::a5 250% do 55 :28me

086

3:24-

80

58:25

 

55889.2:—

 

2:5»...—

 

2.8-2:
:

 

 

 

 

3:33:80 .352 2.8.255

 

 

356580 .252 28555 n a... 2.3.

80

 

05:50:02

3:058 00 003020: 5: 2 :053 0:05 _0:0_w 20:50
0232050 :0 5 9:20:30 5 0000 05 0080 24180898
20 0:05 _0:0_w 05:50:02 3:058 0 5 9:20:30 5 0000
00:00:08 0:9 9:20:30 0:05 :0:0_w 5 m00o-w0_ 20:9: :93

0205020 2 0:05 :0:0_w 0 5 ..b:_:_m:0:m2 300%. 003020:

awed

mmmmC—d

2‘0

0030808: 380

 

0:5:0w 00

003020: 5: 2 :053 0:05 20:56-02320fio :0 5 9:20:30
5 0000 05 0080 0N._namm.ov:x0 20 0:05 :0:0_w 09::0w

0 5 9:20:30 5 0000 00:00:08 0:9 :0008 0:: 9 28902:
20:5 05 58500 w:_0_0: 9:20:30 5 0000-w0_ 20:9:

:93 0000:0020 2 :20:0:5:0w: 003020: .00:05 5:06

Good

owhwmmd

000

20:85:00

 

.5005800 09:02:

05 0>0: 5: 0000 :053 0:05 _0:0_w 2020920232050 :0 5
9:20:30 5 0000 0:: 00:5 00.0H§00.0¢:x0 20 0:05 _0:0_w
0:99:02: 0 5 9:20:30 5 0000 00:00:08 0:9 :0005 0:0

9 22902: 20:50 20 0:05:00 w:_0_0: 9:20:30 5 0000:0—

530: :93 0205020 2 :0wsw29. 003020: .00:05 _0:0_0

89.0

53.036.

000

09:02:

 

 

:035029805

 

030?:

 

008.00:
:

 

 

4:00 £0 030,—.

 

 

81

 

.mowdHCNNd

-Exo mo £60 03:28 a .8 ~NN.o-quo.o-v*a.~ mo 9:30:30

.«0 mvwo-wo_ 05 E oocouotwu a 5:.» 38683 $ own 5
858:6 32838 @8283 03H .2593 €88 $5553

285 130% mo ago-wo_ $32 53> @8300me mm ow<

Good

:ochcd-

cow

om<

 

Same you mQO E 8355 Sonar/ow 23:82
95 £3, 88683 a 85 0885 E 8385 mcouagow

Bang; 22 2 26 cw.on§o.o-*wo.o~§ $93 mo 883 a E

03E $60 05 E 338% a mm 0.85 .285 Eco—w 8:832 a 8m

hood

vetoed-

E0

samumnouwoaoos

 

.Qfiuaadaxo

mo was 2622 a s Edonooduwfle ho 955:3

mo $60-?— ofi S 85:56 a ~23 38683 mm 2:85 E
mucoBtE mcosmgon 2%:me 03H .mzntam $580 $55.56

285 Boo—w no $60-9: 85?: 2:3 38688 mm 2:85

Sod

o2 mmod

owO

2:85

 

 

.5333385

 

2:5-.—

 

38.3:
r

 

 

4:8 £6 039—.

 

 

82

 

.28 was as 5 2:883? _ 8&8 2o 8865 :98

a 2 8888 8:2: 80:28: 28:83 03: mm: :2: 883:8

a .8: :82: 8.0% m :0 :38888w: 8288: :0 828 2:.

~56

$33.0

050

38888808885

 

.28 8o 05 5 8888.8 *2 $98 :0 8885 .56:
a 2 2:008 8:2: 80:28: 2883 02: mm: :2: 883:8

a :0: :82: 150% a :0 $2329.. 8388: :0 828 23.

mood

E Sumod

030

8232:8888

 

.28 8o 05 a 05.828.918.38 2o 8882 as:

a m: 8888 8:2: 80:28: 2883 03: mm: :2: 883:8

a :0: :82: 330% a .8 $888908? 8288: :0 828 of.

hood

Sum _ N06.

020

3883888888:—

 

52833.88

.20 3::0 2:32 a :0 widnvoduwvflm :0 8:38:30

.20 3::0-w0_ 2: 8 80:22:: a 6:5 8:838 2 UM 8
80:22:: 80:28: 2883 03,—. 38:8 3:28 8:38:30

:82: 330% :0 8:082 88:: 8:5 8:803: 2 um

Sod

3:36

0:0

UM

 

.888 8: mQO 8 3888 80:88: 2883
03: 5:5 88803: 2 :2: mmm 8 3888 80:28: 2883

92 9 26 Emonaocdywcdwén25 .8 88 a 3

0:8 8:0 2: 8 3:28: a 2 22: :82: 68% 8:222 a :0”:

:86

voaocd-

2:0

9:88:83:

 

 

8:58.888—

 

2:92.

 

2.8.2:
=

 

 

.28 .3... ~38.

 

 

83

 

6:8 m::o 2: E 03.82999? ~ 30398 no 8888: 88£
a m_ mmm 8:3: 30:88: :8::Sm oz: 8: 85 883:8 a :8

:88 Eco—m a mo :bafimfiamo: :88: 8388: Ho 888 2;

wood

wmofiwod-

CNmO

bananas:
.mmoomfim:

 

.28 $8 2: 5 :wfifiq? _ 83:6 :o 3355

888 a 2 Mmm 8:3: 80:88: :8::Sm 02: mm: :2: 883:8

a :8 :85 Ego—w a mo :owsmoa: 8388: .8 Soto 2F

wood

wcvaod

8N0

88:82

 

.28 88 2: a 8.788.? _ $0993 :0 8855

883 g £ mmm 8&2 80:28: :8::8m 03: mm: :3: 863:8

a :8 :85 392m a mo ..mmgwfiwg—on: 8388: mo 888 2:.

:86

vawmod

owNO

mmocwfiwg—omufl‘am

 

65:6quon
-* fl “30398 m0 883 a 3 0:8 m::o 2: E 8888:
: mmm 8&2 80:28: :8::Sm 03: mm: 85 883:8

a :8 :85 Eco—w : .8 $883888? 8388: Mo 8&8 2F

cmod

gamed-

CBNU

m8:o>:a>o::~*m_mm

 

.28 88 2: a o:._u€mo.o:3.~§u :o 38.55

88E : fl own 8sz 828:8: :8::fim 95 mm: :3: 883:8

a :8 :85 :BBw a mo ammofiano—w: 8388: «0 888 SF

wad

mbmvmcd

OMNO

m8:_mno_0*ow<

 

 

83:888.:—

 

82.3-:

 

88-3:

 

 

4:8 £6 035,—.

 

 

84

 

.028 :25 05 5 mm g . WARN“ 6* m *3me mo 8885 863
: fl £5.00 9 985 Eco—w a no 82388:: :bmzflmcnamo:

E68: 328qu can :mmuEano—w: 338:2. :0 Soto 2:

08.9

among .o

8:10

3338?:

Efiowuwmmofimno—O

 

6:8 2:8 2: a 5.13: 6:393 :0 3885 .533

a W: “25-8 2 985 Eco—w a mo 283E083 :mmocufiscow:

338:3 was :mmoEmnc—w: 338:3 :0 “onto 05.

Bod

van—2.0

03:0

mmocofizcouuwmmofiano—O

 

5:2 :25 2: 5 m3._u§_.:*::&o

Ho 8385 563 a mm 368-8 9 985 Ego—w a Mo 82868::

:owtmoa: 338qu can ..mmoEBEm: 338:3 mo Soto 2:

N36

mnmmid

Ova

owcmofiuwmmofiano—G

 

5:5 was 2: E 3:.on5mo.o-§£0393 :o 3850:.

883 a m_ Um SnwE £5925“: BEES: 025 was 35 583:8

a :8 985 33% a no ..wmoEBo—w: @3583 :0 Babe 2;

wood

xv: mod-

OED

$052206:

 

5:8 :25 2: s :m:.on5wo.:-v* _ 3.998 :o 885%

888 a fl 0% SawE mcocmgon 235:: oz: mm: 85 883:8

a :8 955 Ego—w : mo :wmoco>ca>o::_: 338th .8 Soto 2:.

oood

mwccwod-

OVMO

mmoco>§>o=£uwum

 

 

53829.8:—

 

2:5»-..

 

223?:
r

 

 

.28 .3 :35.

 

 

85

Figure 4.5 — Model Graphs for Significant Moderation Relationships6

 

Relationship between Innovativeness and Likelihood of GBO

I: INCOME = -1 .848
m INCOME = 3.306

Likelihood Of GBO

 

O 1
Innovafiveness

 

 

 

Relationship between Prestige and the Likelihood of GBO

0.500
i:J INCOME = -1 .848

"'23 3 INCOME = 3.306

 

0.375

0.250

Likelihood of GBO

0.125

 

1

 

Prestige

 

 

 

6 In the model graphs, averaged lower-upper quartiles are used for the range of z-axis.

86

 

Figure 4.5, cont.

 

Relationship between Genuineness and the Likelihood Of GBO

1
[:1 INCOME = -1.848

m INCOME = 3.306

Likelihood of GBO

 

O 1
Genuineness

 

 

 

 

Relationship between Globalness and the Likelihood Of GBO

1 .00
[:J AGE = .1 .735

m AGE = 2.049

Likelihood of GBO

 

O 1
Globalness

 

 

 

87

 

Figure 4.5, cont.

 

Relationship between Innovativeness and Likelihood of GBO

1
E3 RSE = -2.330

m RSE = 2.485

Likelihood of GBO

0.25

 

0 1
lnnovafiveness

 

 

 

 

Relationship between Belongingness and Likelihood Of GBO

0
[j RSE = -2.330

m RSE = 2.485

0.375

0.250

Likelihood of GBO

0.125

 

0 1
Belongingness

 

88

 

Figure 4.5, cont.

 

Relationship between Prestige and the Likelihood Of GBO

0
[:3 RSE = 2.330

$5232? RSE = 2.485

 

0.375

0.250

Likelihood Of GBO

0.125

 

1
Prestige

 

 

 

 

Relationship between Soc. Resp. and the Likelihood of GBO

1
[:1 RSE = 2.330

m RSE = 2.485

0.75

Likelihood of GBO

0.25

 

0 1
Social Responsibility

 

 

 

89

 

Figure 4.5, cont.

 

Relationship between Innovativeness and Likelihood of GBO

1
CI] RC = -2.302
m RC = 2.438

0.75

0.25

Likelihood Of GBO

 

0 1
Innovafiveness

 

 

 

 

Relationship between Globalness and the Likelihood of GBO

1
[:1 RC = 2.302

m RC = 2.438

Likelihood of GBO

 

0 1
Globalness

 

 

9O

 

 

Figure 4.5, cont.

 

Relationship between Prestige and the Likelihood of GBO

0
:3 GLOBAL -.- 0

m GLOBAL = 1

0.375

0.250

0.125

Likelihood of GBO

 

1
Prestige

 

 

 

 

Relationship between Genuineness and the Likelihood of GBO

1
E2] GLOBAL -.- o

m GLOBAL = 1

0.75

Likelihood of GBO

0.25

 

0 1
Genuineness

 

 

91

Figure 4.5, cont. ,

 

Relationship between SOC. Resp. and the Likelihood of GBO

1.
[:1 GLOBAL = 0

m GLOBAL = 1

0.75

Likelihood of GBO

 

0 1
Social Responsibility

 

 

 

92

Chapter 5

DISCUSSION OF CONTRIBUTIONS, FUTURE DIRECTIONS AND
LIMITATIONS

5.1 Discussion of Contributions and Managerial Implications

Although marketing scholars have been interested in different aspects Of global
branding and have studied a wide variety Of topics, only a handful Of studies empirically
investigated what values consumers perceive tO be getting from global brands. This study
presents a framework which analyzes global brand perceptions Of consumers within the
broader development context Of these perceptions. First, it introduces a contingency
perspective tO the issue and models the effects Of specific associations as a function Of
consumer characteristics. Second, it empirically investigates tO what extent different
brand associations vary in their effects on the likelihood Of global brand ownership for
consumers that have different characteristics. Third, it tests the notion that there is a
segment of consumers who find global brands appealing and identifies their demographic
and psychographic characteristics. Fourth, it examines the robustness Of this segment
across countries by looking at the variations in the dominant characteristics Of these
consumers across countries. Fifth, it illustrates use Of multi-level modeling techniques
which are especially important when the data structure is nested. Both theoretical
foundations and practical applications Of the method are presented. Sixth, by using data
coming from more than 31,000 consumers from 31 countries, it contributes tO the

empirical generalizations in global branding. In summary this research finds an answer tO

93

one question which lies underneath the debate between standardization vs. adaptation
proponents since the 19205: what works best for global brands under what
circumstances?

The results Of the analyses are interesting. Six global brand associations identified
by previous research are empirically tested in this research; quality, innovativeness,
belongingness, prestige, genuineness, and social responsibility. Furthermore, globalness
is included as a separate association and the interaction of globalness with other
associations were tested. The results show that quality, innovativeness, genuineness, and
social responsibility are indeed positively related tO the likelihood of global brand
ownership. However, neither belongingness nor globalness had a significant direct effect
on the likelihood Of global brand ownership. Many researchers have suggested that
consumers choose global brands because they want to be part Of a special global
consumer culture (Alden et al. 1999), or because Of the psychological benefits Of
globalness per se (Steenkamp et a1. 2003). Steenkamp et al. (2003) called this
“belongingness pathway,” and indeed their quantitative research results showed that this
path was nonsignificant. However, the anecdotal evidence from qualitative research
studies (e. g., Anderson and Miles 1999; Holt et al. 2004b) showed that consumers aspire
tO belong to and tO be part Of something bigger by using global brands. In fact, Alden,
Steenkamp, and Batra (1999) found that corporations take advantage of image-enhancing
effects by positioning brands as “global” in their communications, thus using “global
consumer culture positioning.” As far as consumer behavior toward global brands is
concerned, the results Of this dissertation support Steenkamp et al. (2003) and the premise

that global consumer culture is still in its infancy (Samli 1995).

94

One counter-intuitive significant result is the negative relationship between the
prestige association and the likelihood of GBO for a reference consumer (a consumer
with an average income, age, RSE, and RC scores from a country that has an average
GDP per capita), when all other brand associations are controlled. This is contrary to
previous studies which found a positive association between prestige and purchase
likelihood (Holt et al. 2004b; Steenkamp et al. 2003). The study results show that the
effect of prestige on the likelihood of GBO is stronger for higher income consumers than
lower income consumers and it is stronger for higher RSE consumers than lower RSE
consumers. The interaction effect Of globalness association and prestige is also positive,
i.e., globalness association has a positive effect on the relationship between prestige and
the likelihood of global brand ownership. Higher income or higher RSE consumers
appear to be good target segments for the global brands to emphasize and communicate
prestige association. More research on this topic (e. g., examining the relationship
between prestige and brand equity) may provide more insight into the issue Of why
prestige has a negative effect on the likelihood of global brand ownership for other
consumers.

The effect of quality association of global brands on the likelihood of global
brand ownership was not found to be significantly different for consumers from different
ages, from different income levels, and having different value structures. High quality is
an association that global brand marketers should emphasize regardless of the profile of
their target consumers.

When the direct effects of age, income, RSE, and RC are tested, the dominant

characteristics of global brand-prone consumers are identified. Younger consumers,

95

higher income consumers, and consumers with higher RC scores are more likely to buy
global brands systematically. Creating global brands for higher income or younger
consumers may be a profitable strategy for global marketers.

In fact, the negative effect of age and positive effect Of income on the likelihood
of global brand ownership are in line with the expectations, but the positive effect Of RC
is not consistent with the hypothesis. Higher RC consumers were thought to be more
likely to resist change and global brands, which are symbols of modernity and change.
Indeed, when these consumers perceive a global brand as global, they are less likely to
buy it compared to the brands that are not perceived as global. This implies that higher
RC consumers buy global brands without considering—or ignoring—their globalness.
Indeed, Usunier (2000) talks about “self-contradicting individuals”_ who drink Coke while
complaining about capitalization, or the Americanization of their society. These
consumers may not be thinking about their brand as “global” or symbols Of globalism,
since admitting otherwise would create dissonance. It is suggested that individuals who
hold oppositional values experience values conflict in the form of psychological tension,
which would decrease their welfare (Burroughs and Rindfleisch 2002). This notion can
be linked to classic cognitive consistency theories which state individuals are engaged in
a constant psychological struggle to align cognitions in a consistent manner in order to
decrease tension (Burroughs and Rindfleisch 2002). Thus, it can be argued that as a way
of coping with the dissonance, consumers may be ignoring the globalness of these brands
as a way to let themselves consume these brands. Therefore, when marketing global
brands to conservative consumers, downplaying the globalness association may be a good

strategy.

96

When the robustness of the global brand-prone consumer characteristics across
countries was tested, it was seen that there was no significant variability across countries
in terms of the relationship between RC and the likelihood Of GBO. This is very
interesting, since it shows that there is a robust segment Of higher RC consumers across
countries who consistently buy global brands in general. This segment presents an
Opportunity for marketing managers to target. Another association Of global brands that is
negatively related to the likelihood of GBO for higher RC consumers is innovativeness.
The negative effect of innovativeness association is in line with the findings Of past
research (e.g., Steenkamp et al. 1999) and the negative effect of globalness association, as
explained above, is expected by the values theory. Quality can be a common denominator
and may be important to emphasize for these consumers.

Although there is significant variability across countries in the relationship
between age and the likelihood Of GBO, the economic development level Of countries
(GDP per capita) does not explain this variability. Past research indicated that in most
countries, the correlation between per capita income and modernity is quite high
(Johansson and Moinpour 1977; Roth 1995a; Sethi 1971). Note that convergence theory
proposes that along with modernization and industrialization, nations are becoming
increasingly alike (Williamson and Fleming 1996). The study results do not seem to
support convergence theory, as applied to consumer behavior toward global brands, and it
is in line with the arguments of convergence theory opponents (e. g., de MOOij 2004; de
MOOij 2003).

The relationship between income and the likelihood Of GBO shows significant

variation across countries as well. GDP per capita can explain this variation partially.

97

Note that there was a positive relationship between income and the likelihood Of GBO in
a country with an average level of development. The significant negative interaction
effect between consumer income and GDP per capita shows that consumers in less-
developed countries are more sensitive to changes in income than consumers in more-
developed countries in their purchasing decisions Of global brands. Thus, the effect Of
income on the likelihood of GBO is stronger in less—developed countries than in more-
developed countries, i.e., the marginal effect of income is higher in less-developed
countries than in more-developed countries. Consumers in less-developed countries seem
to be spending more on global brands as their income levels increase than consumers in
more-developed countries. Likewise, the drop in spending is relatively higher than more-
developed country consumers when the income levels decrease. Global brand managers
should be cautious in income fluctuations in the less—developed countries since this may
cause relatively higher fluctuations in consumer demand toward their products titan what
they used to Observe in more-developed countries.

The inspection Of Figure 4.4 also suggests that higher income consumers in less-
developed countries may be more prone to global brands than higher income consumers
in more-developed countries, i.e., the average likelihood Of GBO appears to be higher for
higher income consumers in less-developed countries than higher income consumers in
more-developed countries. Global marketing managers may have an Opportunity in that
segment; targeting higher income individuals in less-developed countries may be a
rewarding strategy.

Although the direct effect of RSE was nonsignificant, the significant negative

interaction effect of RSE and GDP per capita shows that the effect Of RSE is stronger in

98

less-developed countries than it is in more-developed countries, i.e., the marginal effect
of RSE is higher in less-developed countries than in more-developed countries.

The inspection of Figure 4.3 also suggests that higher RSE consumers in less-
developed countries may be more prone to global brands than higher RSE consumers in
more-developed countries, i.e., the average likelihood Of GBO appears to be higher for
higher RSE consumers in less-developed countries than higher RSE consumers in more-
developed countries. This may be explained by the demonstration effect, a phenomenon
where poorer consumers buy symbolic and sensory products to identify with
consumption societies (Keyfitz 1982; Nurske 1953). This effect was used to also explain
why consumers in poorer countries buy hedonic goods (see Roth 1995a). In fact, the
relationship between RSE and the likelihood of GBO appears to be almost flat for more-
developed countries, whereas it is positive for less-developed countries. Consumers in
less-developed countries may be using global brands as a way Of “enhancing” the self,
whereas more-developed country consumers may not see such an enhancement value in
using them. Therefore, emphasizing self—enhancing values (e.g., power, achievement, and
hedonism) in the communication strategy of global brands may be effective for higher
RSE consumers in less-developed countries.

Indeed, on average, prestige and belongingness associations have stronger effects
on the likelihood of GBO for higher RSE consumers than for lower RSE consumers.
These two associations seem to be the means for higher RSE consumers to promote their
personal interests. Prestige is an association that is expected; belongingness association
shows that higher RSE consumers see being part of a group connected via the global

brand as a way to enhance themselves. Therefore, these two global brand associations

99

should especially be emphasized when targeting higher RSE consumers. Innovativeness
and social responsibility have more effects on the likelihood Of GBO for lower RSE
consumers than for higher RSE consumers, and should be emphasized for these types of
consumers.

The effect of different brand associations on purchase decisions does not seem to
vary for consumers that belong to different age groups. The only association that has a
significantly different effect on the likelihood of GBO for Older vs. younger consumers is
globalness; for older consumers globalness seems to be a “summary” attribute more than
it is for younger consumers. When the differences in the income groups are considered,
prestige and genuineness associations appear to have more effect on the likelihood Of
GBO for higher income consumers than for lower income consumers. If the brand origin
is a country famous for the product category Of the brand, the emphasis on this
association may pay Off. As mentioned before, prestige has a positive impact on the
likelihood of GBO for higher income consumers as well, and should be emphasized for
these consumers.

A counter-intuitive result is the negative interaction effect of income and
innovativeness on the likelihood of GBO; perceived brand innovativeness has more effect
on the likelihood of GBO for lower income consumers than higher income consumers, all
other associations being controlled for. Steenkamp et al. (1999) found no significant
relationship between income and consumer innovativeness, although the sign Of the
estimated coefficient of income in their model was also negative. The significance level
of this coefficient in this study is not great either (p = .097), but it is still significant and

interesting. More research is needed to understand the reason of this anomaly.

100

Lastly, three associations which have positive interaction effects on the likelihood
of GBO when they interact with globalness are: prestige, genuineness, and social
responsibility.

Brown and Dacin (1997, p. 68) quoted the concerns of one manager at a major
American retailer:

“We do all these good things we build buildings, give money away but we

don’t know if we get anything out of it. ”

The results of this research suggest that as far as consumer behavior toward global
brands are concerned, they may be getting something out of it. Social responsibility and
globalness, prestige and globalness, and genuineness and globalness all have positive
impact on the likelihood of GBO. An investigation of the target consumer characteristics
may shed some light on the final managerial decision as to whether emphasizing

“globalness” would be desirable or not.

5.2 Limitations and Future Research Directions

Limitations Of this study should be addressed. First, the variables used to measure
brand associations are dichotomous, as well as the model outcome variable, global brand
ownership. This limits the variance in the relationships (e.g., creates a restrictive range
problem). Second, because of the increased complexity associated with introduction of
random variables to the model, the variation of the relationships between global brand

associations and consumer characteristics across countries were not investigated in this

101

study. Future researchers may want to study to what extent the effects Of brand
associations change for different consumers across countries, by employing perhaps a
simpler model with fewer associations. Third, some of the findings of this research are
counter-intuitive; e. g., the direct negative effect Of prestige on the likelihood Of global
brand ownership and the positive interaction between age and globalness association.
More research should be carried out to understand the reasons. Fourth, the outcome
variable employed in this model is a behavioral variable, global brand ownership. It
would be very interesting to employ an attitudinal variable (e.g., global brand equity) and
compare and contrast the differences. This may shed some light on some of the counter-
intuitive results found. Fifth, this study investigates the consumer-side of global branding
and examines consumer perceptions. Other research studies that take the company-side
and link how global brands’ actual efforts (e.g., branding investments) transmute into
consumer perceptions would be extremely interesting. Sixth, related with the last point
made, this study shows the relationship between consumer perceptions and consumer
purchase behavior for different consumers. It does not show the effect Of proposed
changes (e. g., emphasizing one association more for certain consumers) on the cost
structure of the company. A cost and benefit analysis should be carried out to understand
to what degree the benefits from creating new communication programs would outweigh

the costs associated with it.

102

References
Aaker, David A. (1991), Managing Brand Equity. New York, NY: The Free Press.

---- (1996), "Measuring Brand Equity across Products and Markets," California
Management Review, 38 (3), 102-20.

Aaker, David A. and Erich J oachimsthaler (1999), "The Lure of Global Branding,"
Harvard Business Review, 77 (6), 137-46.

Aaker, Jennifer L. (1997), "Dimensions of Brand Personality," Journal of Marketing
Research, 34 (3), 347-56.

Abratt, Russell and Patience Motlana (2002), "Managing CO-Branding Strategies:
Global Brands into Local Markets," Business Horizons, 45 (5), 43-50.

ACNielsen (2001 ), Reaching the Billion-Dollar Mark: A Review of Today's Global
Brands. Chicago, IL: ACNielsen Inc.

Agrawal, Madhu (1995), "Review Of a 40-Year Debate in lntemational Advertising,"
lntemational Marketing Review, 12 (1), 26-48.

Al-Sulaiti, Khalid I. and Michael J. Baker (1998), "Country Of Origin Effects: A
Literature Review," Marketing Intelligence and Planning, 16 (3), 150-99.

Alden, Dana L., Jan-Benedict E.M. Steenkamp, and Rajeev Batra (1999), "Brand
Positioning through Advertising in Asia, North America, and Europe: The Role Of Global

Consumer Culture," Journal Of Marketing, 63 (1), 75-87.

Allen, Michael W. (2001), "A Practical Method for Uncovering the Direct and Indirect
Relationships between Human Values and Consumer Purchases," Journal Of Consumer
Marketing, 18(2), 102-20.

Anderson, Alison and Steven Miles (1999), ""Just Do It?" Young People, the Global
Media and the Construction of Consumer Meanings," in Youth and the Global Media,
Sue Ralph and J o Langham Brown and Tim Lees, Eds. Luton, U.K.z University of Luton
Press.

Anderson, John R. (1983), The Architecture Of Cognition. Cambridge, MA: Harvard
University Press.

Appadurai, Arjun (1990), "Disjuncture and Difference in the Global Cultural Economy,"
Theory, Culture and Society, 7 (2-3), 295-310.

103

Ashill, Nicholas J. and Ashish Sinha (2004), "An Exploratory Study into the Impact Of
Components of Brand Equity and Country of Origin Effects on Purchase Intention,"
Journal of Asia-Pacific Business, 5 (3), 27-44.

Batra, Rajeev, Venkatram Ramaswamy, Dana L. Alden, J an-Benedict E. M. Steenkamp,
and S. Ramachander (2000a), "Effects of Brand Local and Nonlocal Origin on Consumer
Attitudes in Developing Countries," Journal of Consumer Psychology, 9 (2), 83—95.

Batra, Rajeev, J an—Benedict E. M. Steenkamp, and Dana L. Alden (2000b), "Global
Brands: Consumer Motivations and Mechanisms," in Global Branding, Pedro Sousa

(Ed.). Milan, Italy: Marketing Science Institute.

Bilkey, Warren J. and Erik Nes (1982), "Country-Of-Origin Effects on Product
Evaluations," Journal Of lntemational Business Studies, 13 (1), 89-99.

Boddewyn, Jean J., Robin Soehl, and Jacques Picard (1986), "Standardization in
International Marketing: Is Ted Levitt in Fact Right?" Business Horizons, 29 (6), 69-75.

Britt, Steuart Henderson (1974), "Standardizing Marketing for the lntemational Market,"
Columbia Journal Of World Business, 9 (4), 39-45.

Brown, David L. (1923). Export Advertising. New York, NY: Ronald Press.

Brown, T om J. and Peter A. Dacin (1997), "The Company and the Product: Corporate
Associations and Consumer Product Responses," Journal Of Marketing, 61 (1), 68-84.

Bryk, Anthony S. and Stephen W. Raudenbush (1992), Hierarchical Linear Models.
Newbury Park, CA: Sage Publications.

Buri, Christof and Andrew Findlay (1996), "Here Today, Here Tomorrow, The
Contribution of Marketing Research to the Making of UBS' Global Brand Campaign,"
Marketing and Research Today, 24 (4), 208-15.

Burroughs, James E. and Aric Rindfleisch (2002), "Materialism and Well-being: A
Conflicting Values Perspective," Journal of Consumer Research, 29 (3), 348-70.

Calantone, Roger J ., S. Tamer Cavusgil, Jeffrey B. Schmidt, and Geon-Cheol Shin
(2004), "Intemalization and the Dynamics of Product Adaptation - An Empirical
Investigation," The Journal of Product Innovation Management, 21 (3), 185-98.

Calantone, Roger J ., Daekwan Kim, Jeffrey B. Schmidt, and S. Tamer Cavusgil (2006),
"The Influence of Intemal and External Firm Factors on lntemational Product Adaptation
Strategy and Export Performance: A Three-Country Comparison," Journal of Business

Research, 59 (2), 176-85.

104

Cavusgil, S. Tamer and Shaoming Zou (1994), "Marketing Strategy-Performance
Relationship: An Investigation of the Empirical Link in Export Market Ventures,"
Journal Of Marketing, 58 (1), 1-21.

CIA (2004), "The World Factbook."

Clark, Harold F ., Jr. (1987), "Consumer and Corporate Values: Yet Another View on
Global Marketing," lntemational Journal Of Advertising, 6 (1), 29-42.

Collins, Allan M. and Elizabeth F. Loftus (1975), "A Spreading Activation Theory Of
Semantic Processing," Psychological Review, 82 (6), 407-28.

Cote, James E. and Anton L. ‘Allahar (1996), Generation on Hold: Coming Of Age in the
Late Twentieth Century. London, U.K.: New York University Press.

Crippen, Kaye, Pauline Tng, and Patricia Mulready (1995), "DuPont Lycra Shifts
Emphasis to Global Brand Management," The Journal Of Product and Brand
Management, 4 (3), 27-37.

Das, Gurcharan (1993), "Local Memoirs of a Global Manager," Harvard Business
Review, 71 (2), 38-46.

Dawar, Niraj and Philip Parker (1994), "Marketing Universals: Consumers' Use Of Brand
Name, Price, Physical Appearance, and Retailer Reputation as Signals Of Product
Quality," Journal of Marketing, 58 (2), 81-95.

de MOOij, Marieke (2004), Consumer Behavior and Culture: Consequences for Global
Marketing and Advertising. Thousand Oaks, CA: Sage Publications.

(2003), "Convergence and Divergence in Consumer Behavior: Implications for
Global Advertising," International Journal of Advertising, 22 (2), 183-202.

Farquhar, Peter H. (2005), "Brand Alignment Across Organisational Boundaries,"
Journal of Brand Management, 13 (2), 96-100.

Fatt, Arthur C. (1967), "The Danger Of "Local" lntemational Advertising," Journal of
Marketing, 31 (1), 60-62.

Feather, N. T. (1995), "Values, Valances, and Choice: The Influence Of Values on the
Perceived Attractiveness and Choice of Alternatives," Journal of Personality and Social

Psychology, 68 (6), 1135-51.

Feiveson, Alan H. (2001), "Variance Decomposed in Bernoulli Models," in The
Multilevel Modeling Discussion List.

105

Friedman, Jonathan (1990), "Being in the World: Globalization and Localization,"
Theory, Culture and Society, 7 (2-3), 311-28.

Gatignon, Hubert and Thomas S. Robertson (1985), "A Propositional Inventory for New
Diffusion Research," Journal Of Consumer Research, 11 (4), 849-67.

Goldstein, Harvey (1991), "Nonlinear Multilevel Models, with an Application to Discrete
Response Data," Biometrika, 78 (1), 45-51.

Granzin, Kent L. (1981), "An Investigation of the Market for Generic Products," Journal
of Retailing, 57 (4), 39-55.

Green, Robert T. and Isabella C. M. Cunningham (1975), "Feminine Role Perception
and Family Purchasing Decisions," Journal Of Marketing Research, 12 (3), 325-32.

Gregory, James R. and Jack G. W iechmann (2002), Branding across Borders: A Guide
to Global Brand Marketing. New York, NY: McGraw-Hill.

Guo, Guang and Hongxin Zhao (2000), "Multilevel Modeling for Binary Data," Annual
Review Of Sociology, 26, 441-62.

Gupta, Sachin and Pradeep K. Chintagunta (1994), "On Using Demographic Variables to
Determine Segment Membership in Logit Mixture Models," Journal Of Marketing
Research, 31 (1), 128-36.

Hamel, Gary and C. K. Prahalad (1985), "DO You Really Have a Global Strategy?"
Harvard Business Review, 63 (4), 139-48.

Han, C. Min (1989), "Country Image: Halo Or Summary Construct?" Journal Of
Marketing Research, 26 (2), 222-29.

---- (1990), "Testing the Role of Country Image in Consumer Choice Behaviour,"
European Journal Of Marketing, 24 (6), 24-40.

Han, C. Min and Vern Terpstra (1988), "Country-Of-Origin Effects For Uni-National
And Bi-National Products," Journal Of lntemational Business Studies, 19 (2), 235-55.

Hannerz, Ulf (1990), "Cosmopolitans and Locals in World Culture," Theory, Culture and
Society, 7 (2-3), 237-51.

Hassan, Salah S. and Lea Prevel Katsanis (1994), "Global Market Segmentation
Strategies and Trends," in Globalization Of Consumer Markets: Structures and Strategies,

Salah S. Hassan and Erdener Kaynak, Eds. New York: lntemational Business Press.

Hoeffler, Steve and Kevin Lane Keller (2003), "The Marketing Advantages Of Strong
Brands," Journal of Brand Management, 10 (6), 421-45.

106

Hofmann, David A. (1997), "An Overview of the Logic and Rationale Of Hierarchical
Linear Models," Journal of Management, 23 (6), 723-44.

Hofmann, David A. and Mark B. Gavin (1998), "Centering Decisions in Hierarchical
Linear Models: Implications for Research in Organizations," Journal Of Management, 24
(5), 623-41.

Holbrook, Morris B. and Robert M. Schindler (1994), "Age, Sex, and Attitude toward the
Past as Predictors Of Consumers' Aesthetic Tastes for Cultural Products," Journal Of
Marketing Research, 31 (3), 412-22.

Holt, Douglas B., John A. Quelch, and Earl L. Taylor (2004a), "How Global Brands
Compete," Harvard Business Review, 82 (9), 68-75.

Holt, Douglas B., John A. Quelch, and Earl L. Taylor (2004b), "Managing the Global
Brand: A Typology of Consumer Perceptions," in The Global Market: Developing a
Strategy to Manage across Borders, John Quelch and Rohit Deshpande, Eds. San
Francisco, CA: Jossey-Bass.

Homer, Pamela M. and Lynn R. Kahle (1988), "A Structural Equation Test Of the Value-
Attitude-Behavior Hierarchy," Journal of Personality and Social Psychology, 54 (4), 638-
46.

Hox, Joop J. (1995), Applied Multilevel Analysis. Amsterdam: TT-Publikaties.

Hsieh, Ming-Huei (2002), "Identifying Brand Image Dimensionality and Measuring the
Degree of Brand Globalization: A Cross-National Study," Journal Of lntemational
Marketing, 10 (2), 46-67.

---- (2004), "Measuring Global Brand Equity Using Cross-National Survey Data,"
Journal of lntemational Marketing, 12 (2), 28-57.

Hsieh, Ming-Huei, Shan-Ling Pan, and Rudy Setiono (2004), "Product-, Corporate-, and
Country-Image Dimensions and Purchase Behavior: A Multicountry Analysis," Journal
of the Academy Of Marketing Science, 32 (3), 251-70.

Hyllegard, Karen, Molly Eckman, Alejandro Molla Descals, and Miguel Angel Gomez
Borja (2005), "Spanish Consumers' Perceptions of US Apparel Speciality Retailers'
Products and Services," Journal of Consumer Behaviour, 4 (5), 345-62.

Im, Subin, Barry L. Bayus, and Charlotte H. Mason (2003), "An Empirical Study Of

Innate Consumer Innovativeness, Personal Characteristics, and New-Product Adoption
Behavior," Journal of the Academy Of Marketing Science, 31 (1), 61-73.

107

Interbrand/BusinessWeek (2004), "The Global Brand Scoreboard: The 100 Top Brands,"
in BusinessWeek Vol. August 2.

Jain. Subhash C. (1989), "Standardization of lntemational Marketing Strategy: Some
Research Hypotheses," Journal of Marketing, 53 (1), 70-79.

J Ohansson, Johny K. and Ilkka A. Ronkainen (2005), "The Esteem Of Global Brands,"
Journal of Brand Management, 12 (5), 339-55.

Johansson, Johny K., Susan P. Douglas, and IkujirO Nonaka (1985), "Assessing the
Impact Of Country of Origin on Product Evaluations: A New Methodological
Perspective," Journal of Marketing Research, 22 (4), 388-96.

J ohansson, Johny K. and Reza Moinpour (1977), "Objective and Perceived Similarity of
Pacific Rim Countries," Columbia Journal Of World Business, 12 (4), 65—76.

Kahle, Lynn R., Gregory Rose, and Aviv Shoham (2000), "Findings of LOV Throughout
the World, and Other Evidence of Cross-National Consumer Psychographics:
Introduction," in Cross-National Consumer Psychographics, Lynn R. Kahle, Ed. New
York, NY: lntemational Business Press.

Kanter, Rosabeth Moss and Thomas D. Dretler (1998), ""Global Strategy" and Its
Impact on Local Operations: Lessons from Gillette Singapore," The Academy of
Management Executive, 12 (4), 60-68.

Kapferer, J can-Noel (2002), "Is There Really NO Hope for Local Brands?" Journal Of
Brand Management, 9 (3), 163-70.

(1997), Strategic Brand Management: Creating and Sustaining Brand Equity Long
Term (2nd ed.). London, UK: Kogan Page.

Kaynak, Erdener and S. Tamer Cavusgil ( 1983), "Consumer Attitudes Towards Products
of Foreign Origin: DO They Vary Across Product Classes?" lntemational Journal Of
Advertising, 2 (2), 147-57.

Keillor, Bruce D., Michael D. AmicO, and Veronica Horton (2001), "Global Consumer
Tendencies," Psychology and Marketing, 18 (1), 1-19.

Keller, Kevin Lane (2002), Branding and Brand Equity. Cambridge, MA: Marketing
Science Institute.

---- (1993), "Conceptualizing, Measuring, and Managing Customer-Based Brand Equity,"
Journal of Marketing, 57 (1), 1-22.

Kern, Horst, Hans-Christian Wagner, and Roswitha Hassis (1990), "European Aspects of
a Global Brand: The BMW Case," Marketing and Research Today, 18 (1), 47-57.

108

Keyfitz, Nathan (1982), "Development and the Elimination of Poverty," Economic
Development and Cultural Change, 30(3), 649-70.

Kish, Paulette, Dwight R. Riskey, and Roger A. Kerin (2001), "Measurement and
Tracking of Brand Equity in the Global Marketplace - The PepsiCo Experience,"
International Marketing Review, 18 (1), 91-96.

Klein, Katherine J., Fred Dansereau, and Rosalie J. Hall (1994), "Levels Issues in
Theory Development, Data Collection, and Analysis," The Academy Of Management
Review, 19(2), 195-229.

Klein, Naomi (1999), NO Logo: Taking Aim at the Brand Bullies. New York, NY:
Picador.

Kohlberg, Lawrence (1981), The Meaning and Measurement Of Moral Development.
Worcester, MA: Clark University Press.

Kotler, Philip (1986), "Global Standardization-Courting Danger," Journal of Consumer
Marketing, 3 (2), 13-15.

Kotler, Philip H. (1991), Marketing Management: Analysis, Planning, and Control (8th
ed.). Englewood Cliffs, NJ: Prentice-Hall, Inc.

Kreft, Ita G. G., Jan de Leeuw, and Leona S. Aaken (1995), "The Effect Of Different
Forms Of Centering in Hierarchical Linear Models," Multivariate Behavioral Research,
30(1), 1-21.

Levitt, Theodore (1983), "The Globalization Of Markets," Harvard Business Review, 61
(3), 92-102.

Luke, Douglas A. (2004), Multilevel Modeling. Thousand Oaks, CA: Sage Publications.
Macrae, Chris, Ian Ryder, Jack Yan, John Caswell, Tim Kitchin, Thomas Power,
Malcolm McQuarrie, and Simon Anholt (2003), "Editorial: Can Brand Leadership
Recover Local Trust and Global Responsibility?" Journal of Brand Management, 10

(4/5), 268-78.

Neff, Jack (1999), "P&G and Unilever's Giant Headaches," Advertising Age, 70 (22), 22-
28.

Nielsen, Arthur C. Jr. (1964), "DO's and Don'ts in Selling Abroad," in lntemational
Handbook Of Advertising, S. W. Dunn, Ed. New York, NY: McGraw-Hill.

Nurske, Ragnar (1953), Problems of Capital Formation in Undeveloped Countries. New
York, NY: Oxford University Press.

109

Pratt, E. E. (1956), "Building Export Sales-Advertising," The International Advertiser, 8
(I), 19-22.

Propson, Carl F. (1923), "Illustrating the Foreign Campaign," in Export Advertising
Practice. New York, NY: Prentice-Hall.

Quelch, John (1999), "Global Brands: Taking Stock," Business Strategy Review, 10 (1),
1-14.

Quelch, John A. and Edward J. Hoff (1986), "Customizing Global Marketing," Harvard
Business Review, 64 (3), 59-69.

Raaijmakers, J eroen G. W. and Richard M. Shiffrin (1981), "Search Of Associative
Memory," Psychological Review, 88 (2), 93-134.

Raman, Anand P. (2003), "HBR Case Study: The Global Brand Face-Off," Harvard
Business Review, 81 (6), 35-50.

Ratclif f, Roger and Gail McKoon (1988), "A Retrieval Theory Of Priming in Memory,"
Psychological Review, 95 (3), 385-408.

Raudenbush, Stephen W. and Anthony S. Bryk (2002), Hierarchical Linear Models:
Applications and Data Analysis Methods (2nd ed.). Thousand Oaks, CA: Sage
Publications.

Raudenbush, Stephen W., Anthony S. Bryk, and Richard Congdon (2000), "HLM6:
Hierarchical Linear and Nonlinear Modeling." 6.02 ed. Lincolnwood, IL: Scientific
Software lntemational.

Ricks, David A. and Jeffrey S. Arpan (1974), "Pitfalls in Advertising Overseas," Journal
Of Advertising Research, 14 (6), 47-51.

Rogers, Everett M. (2003), Diffusion of Innovations (5th ed.). New York, NY: The Free
Press.

Rokeach, Milton J. (1973), The Nature of Human Values. New York: The Free Press.
Rose, Gregory M. and Aviv Shoham (2000), "The Values Of American and Japanese
Mothers: An Application of LOV in the US. and Japan," in Cross-National Consumer
Psychographics, Lynn R. Kahle, Ed. New York, NY: lntemational Business Press.

Roth, Martin S. (1992), "Depth versus Breadth Strategies for Global Brand Image
Management," Journal of Advertising, 21 (2), 25-36.

110

(1995a), "The Effects of Culture and Socioeconomics on the Performance of Global
Brand Image Strategies," Journal of Marketing Research, 32 (2), 163-75.

---- (1995b), "Effects of Global Market Conditions on Brand Image Customization and
Brand Performance," Journal Of Advertising, 24 (4), 55-75.

Rozin, Randall S. and Liz Magnusson (2003), "Processes and Methodologies for
Creating a Global Business-tO-Business Brand," Journal Of Brand Management, 10(3),
185-207.

Ryans, John K., Jr., David A. Griffith, and D. Steven White (2003), "Standardization/
Adaptation of International Marketing Strategy: Necessary Conditions for the
Advancement of Knowledge," lntemational Marketing Review, 20 (6), 588-603.

Samiee, Saeed (1994), "Customer Evaluation Of Products in a Global Market," Journal Of
International Business Studies, 25 (3), 579-604.

Samiee, Saeed, Terence A. Shimp, and Subhash Shanna (2005), "Brand Origin
Recognition Accuracy: Its Antecedents and Consumers' Cognitive Limitations," Journal
of lntemational Business Studies, 36 (4), 379-97.

Samiee, Saeed and Kendall Roth (1992), "The Influence Of Global Marketing
Standardization on Performance," Journal Of Marketing, 56 (2), 1-17.

Samli, Coskun A. (1995), International Consumer Behavior: Its Impact on Marketing
Strategy Development. Westport, CT: Quorum Books.

Schooler, Robert (1971), "Bias Phenomena Attendant to the Marketing of Foreign Goods
in the U.S.," Journal Of International Business Studies, 2 (1), 71-80.

Schwartz, Shalom H. (1994), "Are There Universal Aspects in the Structure and Content
of Human Values?" Journal of Social Issues, 50 (4), 19-45.

(1992), "Universals in the Content and Structure Of Values: Theoretical Advances
and Empirical Tests in 20 Countries," in Advances in Experimental Social Psychology,
Mark P. Zanna, Ed. Vol. 25. Orlando, FL: Academic Press, Inc.

Schwartz, Shalom H. and Lilach Sagiv (1995), "Identifying Culture-Specifics in the
Content and Structure Of Values," Journal Of Cross-Cultural Psychology, 26 (1), 92-116.

Scott, Jerome E. and Lawrence M. Lamont (1977), "Relating Consumer Values to
Consumer Behavior: A Model and Method for Investigation," in Increasing Marketing

Productivity, T. W. Greer, Ed. Chicago, IL: American Marketing Association.

Sethi, S. Prakash (1971), "Comparative Cluster Analysis for World Markets," Journal Of
Marketing Research, 8 (3), 348-54.

111

Shocker, Allan D., Rajendra K. Srivastava, and Robert W. Ruekert (1994), "Challenges
and Opportunities Facing Brand Management: An Introduction to the Special Issue,"
Journal Of Marketing Research, 31 (2), 149-58.

Singhapakdi, Anusom, Kiran Karande, C.P. Rao, and Scott J. Vitell (2001), "How
Important are Ethics and Social Responsibility? - A Multinational Study of Marketing
Professionals," European Journal Of Marketing, 35 (1/2), 133-52.

Smart, Barry (1999), "Resisting McDonaldization: Theory, Process and Critique," in
Resisting McDonaldization, Barry Smart, Ed. London, UK: Sage Publications.

Snijders, Tom AB. and Roel J. Bosker (1999), Multilevel Analysis: An Introduction to
Basic and Advanced Multilevel Modeling. London: Sage Publications.

Song, Michael, Roger J. Calantone, and C. Anthony Di Benedetto (2002), "Competitive
Forces and Strategic Choice Decisions: An Experimental Investigation in the United
States and J apan," Strategic Management Journal, 23 (10), 969-78.

Srull, Thomas K. and Robert S. Jr. Wyer (1989), "Person Memory and Judgment,"
Psychological Review, 96 (1), 58-83.

Steenkamp, Jan-Benedict E.M., Frenkel ter Hofstede, and Michel Wedel (1999), "A
Cross-National Investigation into the Individual and National Cultural Antecedents Of
Consumer Innovativeness," Journal Of Marketing, 63 (2), 55-69.

Steenkamp, Jan-Benedict E.M., Rajeev Batra, and Dana L. Alden (2003), "How
Perceived Brand Globalness Creates Brand Value," Journal of lntemational Business
Studies, 34(1), 53-65.

Strong, Frank P. (1987), "A Company Study Kodak: Beyond 1990," The Journal Of
Business & Industrial Marketing, 2 (4), 29-36.

Subedi, Bidya Raj (2005), "A Demonstration Of the Three-Level Hierarchical
Generalized Linear Model Applied to Educational Research," The Florida State
University.

Sung, Yongiun and Spencer F. Tinkham (2005), "Brand Personality Structures in the
United States and Korea: Common and Culture-Specific Factors," J oumal Of Consumer
Psychology, 15 (4), 334-50.

Terpstra, David E., Elizabeth J. Rozell, and Robert K. Robinson (1993), "The Influence

Of Personality and Demographic Variables on Ethical Decisions Related to Insider
Trading," The Journal of Psychology, 127 (4), 375-89.

112

Thakor, Mrugank V. (1996), "Brand Origin: Conceptualization and Review," The Journal
Of Consumer Marketing, 13 (3), 27-42.

Thakor, Mrugank V. and Anne M. Lavack (2003), "Effect Of Perceived Brand Origin
Associations on Consumer Perceptions of Quality," The Journal of Product and Brand
Management, 12 (6/7), 394-407.

Thompson, Craig J. and Zeynep Arsel (2004), "The Starbucks Brandscape and
Consumers' (Anticorporate) Experiences Of Glocalization," Journal Of Consumer
Research, 31 (3), 631-42.

Tse, David K. and Gerald J. Gorn (1993), "An Experiment on the Salience Of Country-of-
Origin in the Era of Global Brands," Journal of lntemational Marketing, 1 (1), 57-76.

Tse, David K. and Wei-na Lee (1993), "Removing Negative Country Images: Effects Of
Decomposition, Branding, and Product Experience," Journal of lntemational Marketing,
1 (4), 25-48.

Usunier, Jean-Claude (2000), Marketing across Cultures (3rd ed.). Harlow, UK: Prentice
Hall.

van Gelder, Sicco (2004), "Global Brand Strategy," Journal Of Brand Management, 12
(1), 39-48.

---- (2003), Global Brand Strategy: Unlocking Brand Potential across Countries, Cultures
& Markets. London and Sterling, VA: Kogan Page.

---- (2005), "The New Imperatives for Global Branding: Strategy, Creativity and
Leadership," Journal Of Brand Management, 12 (5), 395—404.

Vinson, Donald E., Jerome E. Scott, and Lawrence M. Lamont (1977), "The Role Of
Personal Values in Marketing and Consumer Behavior," Journal of Marketing, 41 (2), 44-
50.

Walker, Chip (1996), "Can TV Save the Planet?" American Demographics, 18 (5), 42-48.
Wall, Marjorie, John Liefeld, and Louise A Heslop (1991), "Impact of Country-Of-Origin
Cues on Consumer Judgments in Multi-Cue Situations: A Covariance Analysis," Journal
Of the Academy of Marketing Science, 19(2), 105-13.

Walters, Peter G. (1986), "International Marketing Policy: A Discussion of the

Standardization Construct and Its Relevance for Corporate Policy," Journal of
lntemational Business Studies, 17 (2), 55-69.

113

Wang, Chih-Kang and Charles W. Lamb, Jr. (1983), "The Impact Of Selected
Environmental Forces upon Consumers' Willingness to Buy Foreign Products," Journal
of the Academy of Marketing Science, 11 (1), 71-84.

Werther, William B., J r. and David Chandler (2005), "Strategic Corporate Social
Responsibility as Global Brand Insurance," Business Horizons, 48 (4), 317-24.

Williamson, John B. and Jeanne J. Fleming (1996), "Convergence Theory and the Social
Welfare Sector: A Cross-National Analysis," in Comparing Nations and Cultures:
Readings in a Cross-Disciplinary Perspective, Alex Inkeles and Masamichi Sasaki, Eds.
Englewood Cliffs, NJ: Prentice Hall.

Wills, James, A. Coskun Samli, and Laurence Jacobs (1991), "Developing Global
Products and Marketing Strategies: A Construct and a Research Agenda," Journal Of the
Academy Of Marketing Science, 19 (1), 1-10.

Wind, Yoram (1986), "The Myth of Globalization," The Journal Of Consumer Marketing,
3 (2), 23-26.

Wong, George Y. and William M. Mason (1985), "The Hierarchical Logistic Regression
Model for Multilevel Analysis," Journal of the American Statistical Association, 80
(391), 513-24.

WorldBank (2004), "World Development Indicators."

Wright, Len Tiu and Clive Nancarrow (1999), "Researching lntemational "Brand
Equity": A Case Study," lntemational Marketing Review, 16 (4/5), 417-31.

Yip, George S. (1992), Total Global Strategy: Managing for Worldwide Competitive
Advantage. Englewood Cliffs, NJ: Prentice Hall.

YOO, Boonghee, Naveen Donthu, and Sungho Lee (2000), "An Examination of Selected

Marketing Mix Elements and Brand Equity," Journal Of the Academy of Marketing
Science, 28 (2), 195-211.

Zeithaml, Valarie A. (1985), "The New Demographics and Market Fragmentation,"
Journal of Marketing, 49 (3), 64-75.

Zeithaml, Valarie A., Rajan P. Varadarajan, and Carl P. Zeithaml (1988), "The
Contingency Approach: Its Foundations and Relevance to Theory Building and Research
in Marketing," European Journal Of Marketing, 22 (7), 37-64.

Zou, Shaoming and S. Tamer Cavusgil (2002), "The GMS: A Broad Conceptualization Of

Global Marketing Strategy and Its Effect on Firm Performance," Journal Of Marketing,
66 (4), 40-56.

114

   

IIIIIIIIIIIIIIIIIIIIIIIIIIIIII