Product-, Corporate-, and Country-Image Dimensions and Purchase Behavior: A Multicountry Analysis

You might also like

Download as pdf or txt
Download as pdf or txt
You are on page 1of 20

JOURNAL OF THE ACADEMY

Hsieh
10.1177/0092070304264262
et al.
OF/ BRAND
MARKETING
IMAGESCIENCE ARTICLE SUMMER 2004

Product-, Corporate-, and Country-Image


Dimensions and Purchase Behavior:
A Multicountry Analysis
Ming-Huei Hsieh
Yuan Ze University

Shan-Ling Pan
Rudy Setiono
National University of Singapore

This research focuses on consumer perceptions that are istics that could affect the success of its brand-image strat-
developed on the basis of a firm’s advertising appeals as egies. In so doing, a firm can thus determine whether it
well as other factors. In conceptualizing brand-image per- should apply the same brand-image appeal across markets
ceptions, the authors extend the frequent use of product- (Motameni and Shahrokhi 1998). To explore this issue,
related images to include corporate and country images Roth (1995) conducted a survey from the firm’s perspec-
attached to brands. The authors report findings based on tive, which supported his hypotheses that factors at the
secondary economic and cultural data at the macro level national level, such as socioeconomic development and
and the results of a global brand-image survey conducted cultural characteristics, affect the identified brand
in the top 20 international automobile markets at the indi- image–brand performance linkage. In his work, brand-
vidual level. The findings suggest that while consumers’ image measures represent managers’ assessments based
attitudes toward corporate image and country image exert on factual information.
main effects on their brand purchase behavior, the effects However, in line with Keller (2000), we believe that
of certain product-image appeals are moderated by “ultimately, the power of a brand lies in the minds of con-
sociodemographics and national cultural characteristics. sumers or customers” (p. 157) and that the meaning that
The empirical results are broadly supportive of the pro- consumers attach to a brand may be different from that
posed hypotheses and provide a consumer-based exten- which the firm intends. To address this issue of validity, we
sion of Roth’s work on global brand image. examine the effects of contextual factors from the con-
sumer’s perspective, using consumers’ responses from an
Keywords: brand image; cross-national study; automo- extensive cross-national empirical study of automobile
tive; multilevel analysis purchasers. In addition to verifying Roth’s (1995) work,
we also seek to complement his framework by extending
our study in two ways. First, we conceptualize a rather
The emergence of global brands gives rise to the issue comprehensive brand-image perception that extends the
of whether brand-image appeals affect consumer re- traditional definition of product image to include corpo-
sponses differently in different countries. A firm involved rate and country image, and we perceive these images as
in multiple markets should identify the national character- branding at both the corporate and the country level. Sec-
ond, we propose a framework that distinguishes individ-
ual- and national-level moderating variables and brand-
Journal of the Academy of Marketing Science.
Volume 32, No. 3, pages 251-270. level covariates in the relationship between brand-image
DOI: 10.1177/0092070304264262 perceptions and brand-purchase behavior.
Copyright © 2004 by Academy of Marketing Science.
252 JOURNAL OF THE ACADEMY OF MARKETING SCIENCE SUMMER 2004

Furthermore, we outline the procedures necessary to image research and development, Gardner and Levy
obtain an aggregate value from multiple indicators of (1955) were credited with the concept that products have a
brand associations underlying consumers’ perceptions. social and psychological nature as well as a physical one.
Specifically, these procedures include the following: (1) In the late 1960s and early 1970s, brand image was defined
identification of the structural nature of brand image, by as “an attitude about a given brand” (Bird, Channon, and
which we represent each image dimension as a member of Ehrenberg 1970). Drawing on the cognitive and psycho-
the higher-order category, such that brand associations are logical perspective of brand image, Gensch (1978) linked
building blocks; (2) development and modification of the brand image with brand preference by asserting that brand
measurement models of brand image and fitting each preference is a function of the perception space associated
model to the national sample; and (3) generation of the with other brands. Since the 1980s, attitude-based image
aggregate value for each image dimension on the basis of research has come to dominate image research, and there
the measurement models specified in a confirmatory fac- has emerged an interest in linking psychological concepts
tor analysis. The aggregate values of the multiple brand- to strategic research in brand management (Zinkhan and
image dimensions are then used as the predictors of pur- Hirschheim 1992). Researchers have continued on this
chase behavior in a multilevel regression model. Whereas path into the 1990s, and brand image is considered the key
the aggregate values of the dimensions of a brand’s prod- component of brand equity (Keller 1993).
uct, corporate, and country images are specified as in- The definitions and operationalizations of brand image
dependent variables, the dependent variable is opera- have been fairly irregular, although they are not without
tionalized as brand-purchase behavior (i.e., whether some patterns and commonalities. In their summary of the
consumers purchased the brand under investigation). The definitions and conceptualizations of brand image from 28
moderators include national-level factors, such as a coun- studies on the history of marketing literature, Dobni and
try’s level of economic development and its cultural char- Zinkhan (1990) provide a centralized collection of defini-
acteristics, and individual-level factors, such as socio- tions based on authors’ principle emphasis. They further
demographic variables including annual household group the definitions on the basis of their focus into five
income, sex, age, and marital status. Finally, the level of categories: (1) blanket, (2) symbolism, (3) meanings or
brand awareness, market share, and the product origin messages, (4) personification, and (5) cognition or psy-
(i.e., local or foreign) are incorporated into the model as chology. Across groups, image is defined in terms of gen-
covariates. eral characteristics, evoked feelings or impressions,
perceptions of products, and beliefs and/or attitudes.
More recently, following cognitive and psychological
CONCEPTUAL FRAMEWORK principles, brand equity researchers (Keller 1993) have
defined brand image as a set of perceptions about a brand
Brand-Image Perceptions as reflected by the brand associations held in consumers’
memory. Brand associations are classified into three major
A successful brand image enables consumers to iden- categories with respect to their level of abstraction (i.e.,
tify the needs that the brand satisfies and to differentiate attribute, benefit, and overall brand attitude). Here, attrib-
the brand from its competitors, and it consequently ute refers to descriptive features that characterize a prod-
increases the likelihood that consumers will purchase the uct or service, benefit is the personal value that consumers
brand. Brand-image research has long been recognized as attach to the product or service, and brand attitude is
central to marketing and advertising research, not only consumers’ overall evaluation of the brand (Wilkie 1986).
because it serves as a foundation for tactical marketing- Ideally, in consumers’ memory, brand-image perception
mix issues but also because it plays an integral role in should encompass all three types of brand associations.
building long-term brand equity (Aaker 1996; Keller However, given the entailed complexity, it is difficult to
1993; Park and Srinivasan 1994; Srivastava and Shocker design a specific study to incorporate all types of associa-
1991). However, a consensus and immutable definition of tions when a parsimonious model is required. As such, we
brand image has not been established (Stern, Zinkhan, and incorporate only benefit association as the key element.
Holbrook 2002). It is thus necessary for us to review the Benefit is a kind of “perceived” information that is self-
brand-image literature in some detail. relevant and corresponds to the notion that brand-image
Rooted in psychology, brand image can be considered a perception is a largely subjective and perceptual phenom-
new variety of “old” psychological variables that aim to enon formed through consumer interpretation (Dobni and
explain consumer behavior. Since the early 1950s, image Zinkhan 1990).
has been an important concept in consumer behavior In addition, unlike attributes, benefit associations are
research (Gardner and Levy 1955), and the term brand closely related to consumers’ needs. To reflect the diverse
image quickly gained currency. In the early stage of brand- needs of consumers, Park, Jaworski, and MacInnis (1986)
Hsieh et al. / BRAND IMAGE 253

proposed three brand concepts in their normative frame- (Narayana 1981; Shimp, Samiee, and Madden 1993). For
work: (1) functional benefits, (2) symbolic benefits, and example, in general, research shows that products origi-
(3) experiential benefits. Brand concepts are defined as nating in developed countries receive favorable attitudes
brand-unique abstract meanings that typically originate from consumers in developing countries. Specific product
from a particular configuration of product features and country image is the overall perception consumers form of
from a firm’s effort to create meaning from the arrange- specific product categories from a particular country
ments (Park, Milberg, and Lawson 1991). A brand with (Roth and Romeo 1992). According to research in this
functional benefits is one designed to satisfy consumers’ stream, the effect of country image varies considerably,
needs to solve consumption-related problems. A brand depending on the product group under consideration
with symbolic (or social) benefits is one designed to fulfill (Hugstad and Durr 1986). In general, durable goods such
consumers’ desires for self-enhancement, role position, as automobiles are more sensitive to country image than
group membership, or ego identification. Finally, a brand are nondurable goods. We use this definition of specific
with experiential benefits is one designed to fulfill con- product country image because most researchers have
sumers’ desires for sensory pleasure, variety, or cognitive used it since the 1990s (e.g., Cordell 1991; Niss 1996;
stimulation. As they have for the relationship between Roth and Romeo 1992; Tse and Gorn 1992; Zhang 1996).
functional needs and consumption, researchers have Han (1989) attempted to explain country-image effect
proved that the relationship between symbolic needs and on product evaluation through the halo and summary mod-
consumption and that between experiential needs and con- els, in which country-image perception influences overall
sumption are significant (Holbrook and Hirschman 1982). evaluations indirectly through beliefs (halo effect) and/or
The current study follows Keller’s (1993) definition of directly (summary effect). On one hand, consumers tend
brand image and adopts the three previously mentioned to use halo images of a country to judge quality and to gen-
brand concepts as product-related brand-image dimen- erate inferential brand beliefs when other brand-specific
sions. In addition, this study considers umbrella-brand information is not available. On the other hand, as con-
images (e.g., country image and corporate image) that can sumers become more familiar with a country’s products,
be attached to the product to form an overall brand-image they summarize product information and form product
perception. To some extent, umbrella-brand image has evaluations directly. Consequently, we anticipate that con-
been described as part of branding strategy at the country sumers’ positive impression of a brand’s country of origin
and corporate levels (Meenaghan 1995). This is because, results in a positive attitude toward the brand itself and
as with product-related brand image, corporate image and toward subsequent behavior. In contrast, consumers who
country image can provide consumers with meaningful have a negative image or attitude toward a particular coun-
associations to create a desirable, appealing image; thus, try may not consider purchasing a brand that originates
they can affect consumer behavior (Al-Sulaiti and Baker from that country, regardless of its specific associations
1998; Saunders and Fu 1997). In addition, because of the (Hong and Wyer 1989).
strong link between product brand and umbrella brand, Corporate image, like country image, works as a sum-
the presence of product brand might evoke corporate and mary or inference information in the consumer decision-
country information (e.g., Johansson 1993; Leclerc, making process (Erickson, Johansson, and Chao 1984).
Schmitt, and Dube 1994). For example, consumers asso- Corporate image is often interchangeable with corporate
ciate the Mercedes-Benz brand with German origin even identity and may be defined as “the impressions of a par-
though the location does not appear explicitly in the brand ticular company or [corporation] held by some segment of
name. Given this, consumers’ perceptions may not be the public” (Johnson and Zinkhan 1990:347). Perceived
purely product specific; they may combine both corporate corporate information can influence brand-image percep-
and country images (Erickson, Johansson, and Jacobson tion by transferring consumers’ attitude toward the corpo-
1992). ration to their attitude toward the product (Homer 1990)
Similar to the problem of many definitions of brand or by influencing them to create new beliefs (Simmons
image, there are many definitions of country image, which and Lynch 1991), which subsequently affects their prefer-
can be classified into three groups: (1) overall country ences and product choice. For instance, when General
image, (2) aggregate product country image, and (3) spe- Motor positions itself as concerned with quality, consum-
cific product country image. Overall country image is the ers’ associations of General Motors quality might transfer
total of all descriptive, inferential, and informational to their knowledge of a particular General Motors models.
beliefs that a consumer has about a particular country (e.g., As such, we anticipate that a positive impression of the
Martin and Eroglu 1993). Aggregate product country corporation that is affiliated with the brand results in a pos-
image is the entire cognitive “feel” associated with a par- itive attitude toward the brand itself and a positive change
ticular country’s products or with the perceived overall in consumers’ subsequent behavior. Conversely, a nega-
quality of the products from that particular country tive corporate image might result in spillover effects on a
254 JOURNAL OF THE ACADEMY OF MARKETING SCIENCE SUMMER 2004

FIGURE 1
The Relationship Between Brand Image Perception and Purchase Behavior

Nation-Level Moderator
• Cultural dimensions
Brand-Image Perception • Level of economic development
and media penetration

Product-image dimensions
Individual Level Moderator
• Age Brand Purchase
• Income
• Sex Behavior
Corporate-image dimensions • Marital Status

Country-image dimensions
Brand/Nation Interaction
• Brand awareness
• Market share
• Local versus foreign origin
Covariates

particular brand. In summary, like country image, cor- tation (e.g., Helsen, Jedidi, and DeSarbo 1993; Jain 1993)
porate image not only adds value but also creates a halo over consumer segmentation.
effect for all of a company’s products. Advocates of country segmentation attempt to explore
The preceding literature review clarifies the definitions the underlying national factors that are attributable to the
of the key concepts in our study. In addition, it rationalizes differences between segments. In contrast, advocates of
our attempt to conceptualize brand image with the multi- consumer segmentation focus on identifying segments
ple dimensions of product, corporate, and country image. with similar consumer characteristics. Consequently, as
units of analysis, the former emphasize macro variables,
and the latter focus on micro variables. This dichotomy
Moderating Effects of suggests that there is a need to establish a linkage between
National- and Individual-Level Variables the two research streams. The use of a multiple-tier struc-
ture in our research design enables us to evaluate the feasi-
We now examine the linkages between brand-image bility of a standardization strategy from both the country
perceptions and purchase behavior in a cross-national and the consumer perspective.
context (Figure 1). In the early 1990s, some scholars and
practitioners advocated that firms exploit the global mar- Level of economic development. From the perspective
ket using a standardized strategy (e.g., Jain 1989; Levitt of international consumer research, Clark (1990) pro-
1983). Arguments for this strategy included efficient posed that consumer behavior in the marketplace reflects
worldwide brand positioning and the enhancement of cus- consumers’ varying cultural and personality traits, which
tomer preferences as a result of global availability, ser- can be described by macro-level national characteristics
viceability, and recognition (Yip 1995), as well as global such as level of economic development. A country’s level
credibility (Alden, Steenkamp, and Batra 1999; Kapferer of economic development and its effect on brand-image
1992). In line with this, proponents of the strategy (Hassan perception can be understood from the process of a soci-
and Katsanis 1994) were interested in attaining global ety’s “consumerization.” Developing countries are ex-
consumer segments that associate similar meanings with a pected to fall mostly within the early stages, whereas
certain brand across national boundaries. However, some developed countries are more likely to be in the later stages
scholars have expressed skepticism about the value of a of the consumerization process. In low-consumerized
standardization strategy because important national dif- markets, a brand’s role is centered on product difference
ferences (e.g., economic, political, ethnic, or geographic (i.e., utilitarian). The brand’s role changes when the mar-
variation) are often overlooked (Whitelock and Chung ket moves toward high consumerization, at which point
1989). Accordingly, skeptics advocated country segmen- the brand not only is used for reference but also is imbued
Hsieh et al. / BRAND IMAGE 255

with personality and/or symbolic meanings as the market power distance implies a higher tendency toward privi-
becomes more competitive and as consumers become leged positions, suggesting that buyers from high-power-
more discriminating (Goodyear 1996). As such, whereas distance cultural groups (e.g., the “Chinese Cluster”) in
consumers in developing countries might value utilitarian which social and economic class differences are more
appeals, nonutilitarian appeals are more likely to be effec- emphasized are more likely to prefer symbolic appeals
tive for consumers in developed countries. (Lowe and Corindale 1998).
A country’s level of economic development is also It is empirically evident in Roth’s (1995) work that cul-
closely related to product affordability and media penetra- tural dimensions, similar to economic factors, exert a
tion, which in turn shape consumers’ attitudes toward con- moderating effect on the relationship between brand-
sumption. Here, “affordability” indicates the amount of image strategy and brand performance. In summary,
resources consumers are able to spend on goods and ser- the previous discussion leads us to form the following
vices. When resources are limited, only the most basic and hypotheses:
utilitarian needs will be satisfied. Therefore, utilitarian
appeals, which attempt to solve basic consumption prob- Hypothesis 1: The effect of utilitarian image on brand-
lems, should have a stronger effect on consumer behavior purchase behavior is greater in countries where the
in countries where the level of economic development is economic development level is low.
low. Media penetration dictates the amount of information Hypothesis 2: The effect of symbolic image on brand-
that influences consumers’ perceptions of specific prod- purchase behavior is greater in countries where the
ucts. A lack of media exposure might confine consumers’ economic development level is high.
perceptions of a product to fundamental utilitarian capa- Hypothesis 3: The effects of sensory image on brand-
bilities and prevent them from learning about other non- purchase behavior are greater for consumers in coun-
utilitarian values of the product. tries that are characterized by high individualism.
Hypothesis 4: The effects of symbolic image on brand-
Cultural dimensions. National culture has long been purchase behavior are greater for consumers in coun-
identified as an important characteristic that determines tries that are characterized by high power distance.
consumer behavior in different countries. Hofstede’s Hypothesis 5: The effects of utilitarian image on brand-
(1983, 1991) influential work on cross-cultural value sys- purchase behavior are greater for consumers in coun-
tries that are characterized by high uncertainty
tems identifies five aspects of culture dimensions: (1)
avoidance.
power distance, (2) uncertainty avoidance, (3) individual-
ism, (4) masculinity, and (5) long-term orientation. Al- Sociodemographics. Key segmentation variables in-
though Hofstede’s cultural scores were based on work- clude sociodemographic variables, psychographic vari-
related values, the results were validated with numerous ables (e.g., value system, lifestyle), and behavioral vari-
cross-cultural studies from various disciplines, including ables (e.g., brand choice or usage data collected at point of
the consumer research areas of tipping behavior (Lynn, sales). Much segmentation involves combinations of these
Zinkhan, and Harris 1993), product usage (de Mooij bases, yet little research exists that examines how con-
2000), consumer innovativeness (Steenkamp, Hofstede, sumer characteristics moderate the effects of certain
and Wedel 1999), perceived good-life image (Zinkhan and image perceptions on brand-purchase behavior. In other
Prenshaw 1994), and brand image (e.g., Kale 1995; Roth words, what kind of image appeal might be more effective
1995). Specifically, Kale and Roth associate three of in a certain consumer segment? In the limited literature,
Hofstede’s cultural dimensions (i.e., power distance, un- there are findings that older people are less likely to ask for
certainty avoidance, and individualism) with the three “relaxation” benefits and more likely to care about practi-
brand concepts discussed previously. cal benefits; that women tend to value goods for symbolic
According to Kale (1995), high individualism is related and emotional reasons, while men value goods for func-
to a strong hedonistic motive in consumption; this sug- tional and activity-related reasons (Dittmar, Beattie, and
gests that products positioned to emphasize distinct senses Friese 1995; Gitelson and Kerstetter 1990); and that high-
can enhance the purchase behavior of consumers from income groups are more sensitive to perceived differences
countries with a higher degree of individualism (e.g., the in quality (Wheatley and Chiu 1977). We examine these
“Anglo Cluster”), but not necessarily of consumers from possible relationships, which we formulate as follows:
other countries. A high degree of uncertainty avoidance
implies a higher reluctance to accept risk, which suggests Hypothesis 6: The effect of utilitarian image appeal on
that, to attract consumers, marketers should reduce the brand-purchase behavior is greater among male re-
perceived risk of buyers from countries that show high spondents.
uncertainty avoidance (e.g., the “Roman/Latin Cluster”) Hypothesis 7: The effect of symbolic image on brand-
by ensuring utilitarian satisfaction (i.e., focusing on prob- purchase behavior is greater among female respon-
lem solving and reducing perceived risk). Finally, high dents.
256 JOURNAL OF THE ACADEMY OF MARKETING SCIENCE SUMMER 2004

TABLE 1
Countries/Metropolitan Areas Surveyed
Country Metropolitan Areas Passenger Car Sample
Australia Sydney, Melbourne, Brisbane 143
Belgium Brussels, Liege, Charleroi, Antwerp, Gent 181
Brazil Sao Paulo, Rio de Janeiro, Belo Horizonte 124
United Kingdom London, Birmingham, Manchester 178
Canada Toronto, Montreal, Vancouver 94
China Beijing, Shanghai 44
France Paris, Lille, Lyon, Marseilles 171
Germany Nurnberg, Leipzig, Bochum, Hamburg 189
India Mumbai, Delhi, Calcutta, Madras, Bangalore 125
Italy Rome, Milan, Bari 158
Japan Fukuoka, Tokyo, Osaka, Sapporo 156
Mexico Mexico City, Guadalajara, Monterey 136
the Netherlands Amsterdam, Rotterdam, The Hague 172
Russia Moscow, St. Petersburg, Nizhni-Novgorod 124
South Korea Seoul, Kyunggi 142
Spain Madrid, Barcelona, Seville, Bilbao 165
Taiwan Taipei, Kaohsiung 143
Thailand Bangkok, Chiangmai, Haad Yai 57
Turkey Istanbul, Ankara, Izmir 150
United States Boston, Los Angeles, Minneapolis, Philadelphia, Atlanta, St. Louis 176
Total 2,828

Hypothesis 8: The effect of utilitarian image on brand- and consists of 4,320 new car buyers. We limit our scope
purchase behavior is greater among older respon- to passenger car owners only (leaving out truck owners,
dents. which the data set also covers), for a total sample size of
Hypothesis 9: The effect of symbolic image on brand- 2,828. Respondents were asked to evaluate a set of brands1
purchase behavior is greater among high-income in a survey conducted from September 1997 to October
respondents.
1997. In view of heterogeneity within a country, the data
Covariates. Previous research suggests that character- set covers only metropolitan areas, of which 70 are identi-
istics such as brand awareness, market share, and local fied. A multistage sampling procedure was adopted. The
versus foreign origin affect consumer choices. Therefore, sample size of each country was determined arbitrarily by
we include related variables in our framework as co- MORPACE at the initial stage. Initial samples of 200 were
variates. Controlling for these variables allows for stron- chosen for each country, with the exception of Japan (300)
ger tests of our hypotheses. Various authors have sug- and the United States (370) because of the large sales vol-
gested that awareness reflects the salience of a brand in umes in these two countries. Subsequently, the sample in
consumers’ minds (Motameni and Shahrokhi 1998), each country was divided into mutually exclusive and col-
which in turn affects brand perceptions and attitudes and lectively exhaustive subgroups on the basis of the propor-
thus drives brand choice and loyalty (Aaker 1996). Other tion of total population in each metropolitan area. In each
than brand awareness, market share has been used as a sig- area, local agents then determined the quota of each brand
nal for brand quality. All else being equal, consumers tend according to the proportion of automobile sales volumes
to reduce their levels of uncertainty by choosing better- in terms of makes and segments. Finally, trained inter-
selling brands. Finally, because ethnocentric tendency viewers were recruited. They interviewed respondents at
affects consumers’ product evaluation, we include brand major commercial centers using the intercept method.
origin (i.e., local or foreign) as a covariate. Before the main survey was conducted, exploratory
research employing focus group discussions was con-
ducted in each market to offset the limitations of the free-
METHOD elicitation method, which usually generates low quantities
of elicited thoughts. Participants indicated the desirability
Description of the Existing Data Set and importance of several associations, in which benefit-
oriented associations that most participants mentioned
The data set used, which is owned by MORPACE Inter- (i.e., ones that had the same attitudinal meaning for most
national, covers the top 20 automobile markets (Table 1) respondents at this stage) were selected as salient beliefs
Hsieh et al. / BRAND IMAGE 257

about automobiles. In total, 17 benefit associations that associations to presence/absence decreases analytic sensi-
were commonly elicited across the surveyed nations were tivity and limits analytic options.
included for analysis. The associations reflect product Sociodemographic measures include marital status,
image (i.e., durability, reliability, safety in accidents, lux- age, annual household income, and sex. Marital status,
ury features, prestige, latest technology, fun to drive, ex- sex, and education were measured dichotomously. Age
citing, stylish, good acceleration, and sporty), corporate range was measured using an ordinal scale. Household
image (i.e., “does a lot of advertising,” “in the news a lot,” income was also measured using an ordinal scale, and
“good for this country,” “good for the world economy,” pretax total household income was measured in the local
“shows concern for the environment,” “the company that currency and subsequently converted into U.S. dollars.
makes the best vehicle,” and “the most successful automo-
bile manufacturer”), and country image (i.e., “the country Collection of secondary data. National economic data
that makes the best vehicle” and “the country that makes were drawn from publicly available statistical data
vehicles I accept”). A questionnaire was then developed sources. Because no agreement has been reached as to
on the basis of the findings from the focus group discus- what constitutes “economic development,” multiple indi-
sions. During the interviews, respondents were asked to cators pertaining to consumer purchase behaviors (gross
connect each recognized brand to the applicable benefit national product per capita), product usage (penetration of
associations that are contained in the show cards. vehicles), and information diffusion (penetration of mass
An extensive cross-national pretest in each surveyed media, including television, newspapers, telephones, per-
metropolitan area was conducted before fieldwork was sonal computers, and the Internet) were compiled for the
carried out. The questionnaires were then fine-tuned 20 surveyed countries. In line with other analyses of multi-
according to the feedback from each country after the pilot item socioeconomic data, confirmatory factor analysis
study and assigned scale items. To establish calibration was used to develop a single-dimension scale and the cor-
equivalence, a few measures, such as income, were con- responding aggregate value. With respect to culture
verted before analysis so that they were comparable across dimensions, given the quantitative rigor and the intuitive
countries. Moreover, to establish translation equivalence, appeal of Hofstede’s study (1983, 1991), three dimensions
questionnaires were initially constructed in English. Par- of his five-dimensional cultural framework (power dis-
ticipants in the United States, Canada, Australia, and Brit- tance, individualism, and uncertainty avoidance) were
ain, where English is the official language, received an used to develop the cultural profile of each country in our
English version of the questionnaire. Bilingual personnel research.
who were fluent in English translated the questionnaires
for the remaining countries into the local languages. The
questionnaires were then back-translated into English to DATA ANALYSIS
ensure accuracy and to enhance translation equivalence
(Douglas and Craig 1983). Prior to applying multilevel regression modeling to
examine how consumers from different countries weigh
Measures. A dichotomous measure was applied to image appeal in relation to brand-purchase behavior, three
associations in the data set; that is, a person believes either questions related to brand image need to be addressed: (1)
that the brand under investigation is associated with a ben- What are the image dimensions that underlie related, sub-
efit (1) or that the brand is not associated with that benefit ordinate brand associations, which comprise brand-image
(0). Similarly, in terms of the measure of brand-purchase perception? (2) Is the factor structure of each image
behavior, a person either has purchased (1) or has not (0) dimension identified in each individual country similar
purchased the specific brand. The use of a nominal scale, across the globe? (3) How may aggregate scales be gener-
to some extent, can facilitate scalar equivalence by pre- ated for measuring brand-image dimensions based on the
venting the midpoint or extreme responses associated with well-fitted measurement models across nations and for
a multiple-point scale (Mullen 1995). The use of a dichot- use as predictors in the subsequent hypotheses testing?
omous measure also has practical implications. Often,
researchers find that their respondents are not knowledge- Exploring Brand-Image
able enough to make detailed judgments or to respond in Dimensions Using Global Data
terms of a continuous scale for a long list of stimuli. In
view of this difficulty, the use of binary or dichotomous We first explore the dimensionality of brand image per-
scales has been proposed to measure consumer percep- ceptions by using principal components analysis (PCA) of
tions and preferences in developing countries (Malhotra the global data. We extracted the first four components
1988). While the use of a dichotomous measure could ease that had eigenvalues greater than one. (The findings given
the difficult or time-consuming task of judging that is here are brief, but the detailed results of the PCA are avail-
imposed on respondents, we note that reducing the scale of able from the authors on request.) The first three benefit-
258 JOURNAL OF THE ACADEMY OF MARKETING SCIENCE SUMMER 2004

based image dimensions that correspond to the three brand matrix as input and weighted least squares as the fitting
concepts (Park et al. 1986) are (1) the symbolic-economic function using LISREL VIII. Of the different types of cor-
dimension, which includes positive prestigious and luxury relation matrix, the performance of the tetrachoric correla-
features and negative fuel economy and dealer service; (2) tions coefficient is superior to that of other correlation
the sensory dimension, which includes “exciting,” “good coefficients when the observed variables in the model are
acceleration and speed,” “fun to drive,” and “sporty; and all dichotomous. This is because the tetrachoric correla-
(3) the utilitarian dimension, which includes “reliable,” tion coefficient produces essentially unbiased parameter
“made to last,” and “safe in accidents.” The findings imply estimates.
that the brand associations in each dimension are interde-
pendent as related measures of the converged image
Model Comparison and
dimension at a higher level of abstraction. The fourth
Modification Using National Data
dimension consists mainly of “stylish” and “latest technol-
ogy,” which are the two important automobile associations
that are not related to general brand concepts. The results of the chi-square test for the alternative
In the same vein, we employ PCA to explore the dimen- models are summarized in Table 2. As Table 2 shows, the
sionality of associations related to a brand’s corporate three-factor model with a smaller chi-square value is a sig-
image and country image. The first dimension highlights nificantly better fit than the four-factor model, indicating
corporate associations: “does a lot of advertising,” “in the no evidence of discriminant validity of the fourth factor
news a lot,” “good for this country,” “good for the world (i.e., economic dimension). However, when we use the
economy,” and “shows concern for the environment.” The three-factor model on some national samples (e.g., Brazil,
first two associations involve corporate publicity, while Canada, Mexico, China), we face the problem of non-
the other three associations involve corporate responsibil- convergence, indicating an identification problem proba-
ity. The other dimension (which comprises “the company bly caused by the data characteristics. We thus respecified
that makes the best vehicle,” “the most successful automo- the model in an exploratory manner according to model
bile manufacturer,” “the country that makes the best vehi- development strategy (Byrne 1998). Although heavily
cle” and “the country that makes vehicles I accept”) repre- used in practice, model development strategy has often
sents overall attitudes toward a brand’s corporate image been criticized for being data driven. Nonetheless, in many
and country image. applications, theory can only provide a starting point for
the development of a model that is both substantively
meaningful and statistically a good fit.
Developing Measurement Models That Therefore, we institute a series of appropriate modifi-
Reflect Multiple Dimensional Brand Image cations based on the results derived from the confirmatory
factor analysis and PCA of each country, so as to attain a
We examine the extracted dimensionality using struc- closer fit between the model and market data in each
tural equation modeling on a country-by-country basis. national sample. As a result, the chi-square statistics for
Although the PCA identifies the symbolic-economic fac- the revised three-factor model for each nation become in-
tor as a single dimension, in practice, marketing appeals significant. The resulting goodness-of-fit indexes, includ-
are used independently to target different audiences. We ing the Comparative Fit Index (CFI), the Goodness-of-Fit
therefore split the symbolic-economic dimension into two Index (GFI), and the Adjusted Goodness-of-Fit Index
individual image dimensions: economic and symbolic. (AGFI), also indicate well-fitting models in general. In
The idea of treating the economic dimension as distinct is summary, the overall fit statistics for the three-factor mea-
also supported by Doyle (1989), who suggested that one of surement models support measurement quality, suggest-
consumers’ needs is for products to be economic or func- ing that the indictors specified in the models acceptably
tional (rather than emotional and psychological). We fur- converge to the corresponding construct (Steenkamp and
ther propose an alternative model in which the economic van Trijp 1991) in each country.
dimension is omitted so that the dimensions correspond Item-factor loadings, which are estimates of the validity
with these three image dimensions used in Roth’s (1995) of the observed variables (i.e., brand associations), are also
study. Model comparison strategy is then adopted, in that summarized in Table 2. The larger the factor loadings or
the proposed four-factor model is compared with the alter- coefficients, the stronger is the evidence that the measured
native three-factor model in terms of chi-square statistics items represent the underlying constructs (Bollen 1989).
and goodness-of-fit indexes (Byrne 1998; Hair, Anderson, In most countries surveyed, each loading (λx) is signifi-
Tatham, and Black 1995). cantly related to its underlying factor (on the basis of t-val-
Because of the nominal nature of the variables, we esti- ues). However, the validity of the scales is less than desir-
mate these two measurement models with the correlation able for India, Russia, Mexico, Spain, South Korea, and
(text continues on p. 262)
TABLE 2
Properties of Measurement Models
Australia Belgium Brazil Britain Canada Germany India Italy Japan Russia
(n = 3,395) (n = 3,894) (n = 2,548) (n = 3,698) (n = 1,879) (n = 4,174) (n = 1,792) (n = 2,961) (n = 2,983) (n = 2,246)
Product image
Chi-square four-factor model (df = 38) 220.21 127.7 NC 86.73 NC 124.08 NC 184.99 137.59 126.77
Chi-square three-factor model (df = 24) 180.65 116.74 NC 92.17 NC 187.03 65.77 NC NC 74.31
Chi-square revised three-factor model 6.5 22.24 19.33 7.04 16.11 17.71 5.24 21.83 16.36 13.5
df for the revised three-factor model 6 11 11 6 11 11 6 11 6 6
Fit indexes for the revised three-factor model
GFI 1.00 1.00 1.00 1.00 1.00 1.00 1.00 1.00 1.00 1.00
AGFI 1.00 1.00 1.00 1.00 1.00 1.00 1.00 1.00 1.00 1.00
CFI 1.00 0.96 0.97 1.00 0.98 0.99 1.00 0.95 0.92 0.88
RMR 0.03 0.05 0.05 0.03 0.07 0.06 0.04 0.05 0.05 0.05
Factor loading
Utilitarian
Durable 0.39 0.48 0.51 0.42 0.56 0.38 0.59 0.35
Reliable 0.76 0.60 0.73 0.38 0.65 0.55 0.39
Safe in accidents 0.39 0.50 0.68 0.09 0.52 0.47 0.27
Symbolic
Luxury features 0.46 0.77 0.98 0.54 0.70 0.55 0.13 0.57 0.38 0.50
Prestigious 0.96 0.56 0.23 0.67 0.80 0.93 0.15 0.59 0.84 0.55
a
Safe in accidents
a
Latest technology
a
Stylish
Sensory
Exciting 0.53 0.67 0.69 0.58 0.55 0.66 0.38 0.44
Fun to drive 0.33 0.70 0.70 0.67 0.34 0.34 0.22
Good acceleration and speed 0.61 0.46 0.80 0.25
Sporty 0.70 0.72 0.54 0.53
Corporate image
Chi-square two-factor model (df = 3) 5.34 10.47 14.37 0.96 1.15 7.79 57.86 11.25 7.45 10.09
Fit indexes for the two-factor model
GFI 1.00 1.00 1.00 1.00 1.00 1.00 0.99 1.00 1.00 1.00
AGFI 1.00 1.00 1.00 1.00 1.00 1.00 0.97 1.00 1.00 1.00
CFI 0.97 0.93 0.90 1.00 1.00 0.95 0.61 0.63 0.87 0.89
Factor loading
Publicity
Does a lot of advertising 0.59 0.89 0.20 0.30 0.52 0.37 0.12 0.22 0.11 0.13
Is in the news a lot 0.50 0.31 0.99 0.45 0.74 0.55 0.74 0.07 0.31 0.78
Social responsibility
Good for the world economy 0.50 0.56 0.98 0.08 0.34 0.45 0.86 0.30
Shows concern for the environment 0.45 0.55 0.39 0.50 0.71 0.55 0.37 0.38 0.17 0.71
a

259
Good for this country 0.37 0.22

(continued)
260
TABLE 2 (continued)
Australia Belgium Brazil Britain Canada Germany India Italy Japan Russia
(n = 3,395) (n = 3,894) (n = 2,548) (n = 3,698) (n = 1,879) (n = 4,174) (n = 1,792) (n = 2,961) (n = 2,983) (n = 2,246)

Overall attitude toward corporation and country


Chi-square two-factor model (df = 3) 7.48 8.65 3.28 8.77 5.09 4.84 1.84 5.91 9.86 11.1
Fit indexes for the two-factor model
GFI 1.00 1.00 1.00 1.00 1.00 1.00 1.00 1.00 1.00 1.00
AGFI 1.00 1.00 1.00 1.00 1.00 1.00 1.00 1.00 1.00 1.00
CFI 0.99 1.00 1.00 1.00 1.00 1.00 1.00 1.00 0.99 1.00
Factor loading
Overall attitude toward brand’s corporate
This corporation makes the best vehicle 0.50 0.78 0.70 0.65 0.74 0.76 0.74 0.56 0.76 0.77
This is the most successful auto manufacturer 0.97 0.89 0.89 0.87 0.77 0.78 0.77 0.84 0.84 0.89
Overall attitude toward brand’s country
This country makes the best vehicle 0.91 0.98 0.85 0.97 0.85 0.96 0.97 0.95 0.93 0.98
This country makes the vehicle I accept 0.55 0.83 0.71 0.74 0.84 0.89 0.96 0.78 0.67 0.89

United The South


Taiwan Turkey States China France Mexico Netherlands Korea Spain Thailand
(n = 2,970) (n = 3,156) (n = 3,872) (n = 851) (n = 3,225) (n = 2,216) (n = 3,963) (n = 2,820) (n = 3,017) (n = 1,323)
Product image
Chi-square four-factor model (df = 38) NC 70.29 NC NC 183.67 NC 141.73 132.5 NC 321.77
Chi-square three-factor model (df = 24) 158.76 58.94 230.04 NC 189.05 NC NC 73.29 NC 650.72
Chi-square revised three-factor model 9.18 22.34 9.79 1.38 23.29 14.75 14.15 18.59 14.17 10.29
df for the revised three-factor model 6 11 6 6 11 6 6 11 6 6
Fit indexes for the revised three-factor model
GFI 1.00 1.00 1.00 1.00 1.00 1.00 1.00 1.00 1.00 1.00
AGFI 1.00 1.00 1.00 1.00 1.00 1.00 1.00 1.00 1.00 1.00
CFI 0.99 0.92 0.99 1.00 0.97 0.84 0.97 0.95 0.92 1.00
RMR 0.04 0.05 0.03 0.04 0.06 0.05 0.05 0.05 0.05 0.07
Factor loading
Utilitarian
Durable 0.75 0.50 0.45 0.81 0.77 0.30 0.79 0.62 0.65 0.08
Reliable 0.43 0.34 0.16 0.55 0.42 0.46 0.38 0.22
Safe in accidents 0.51 0.45 0.48 0.22
Symbolic
Luxury features 0.84 0.33 0.97 0.88 0.32 0.27 0.42
Prestigious 0.74 0.54 0.61 0.17 0.39 0.63
a
Safe in accidents 0.26 0.69
a
Latest technology 0.27 0.83
a
Stylish 0.72 0.40 0.26
Sensory
Exciting 0.57 0.66 0.69 0.69 0.57 0.85
Fun to drive 0.65 0.68 0.75 0.50 0.18
Good acceleration and speed 0.34 0.59 0.33 0.48
Sporty 0.38 0.40 0.58 0.61 0.32 0.13
Corporate image
Chi-square two-factor model (df = 3) 9.71 9.87 5.2 5.75 7.68 3.84 9.82 9.31 6.32 4.08
Fit indexes for the two-factor model
GFI 1.00 1.00 1.00 1.00 1.00 1.00 1.00 1.00 1.00 1.00
AGFI 1.00 1.00 1.00 0.99 1.00 1.00 1.00 1.00 1.00 1.00
CFI 0.64 0.91 0.99 0.87 0.87 1.00 0.92 0.96 0.96 0.82
Factor loading
Publicity
Does a lot of advertising 0.51 0.34 0.73 0.35 0.48 0.73 0.56 0.73 0.54 0.20
Is in the news a lot 0.26 0.15 0.55 0.19 0.50 0.71 0.55 0.55 0.45 0.31
Social responsibility
Good for the world economy 0.56 0.23 0.50 0.35 0.36 0.20 0.13 0.59 0.31 0.47
Shows concern for the environment 0.41 0.30 0.45 0.43 0.22 0.71 0.67 0.44 0.55 0.38
a
Good for this country
Overall attitude toward corporation and country
Chi-square two-factor model (df = 3) 3.89 0.22 5.24 2.12 5.56 5.62 3.99 27.88 9.38 3.25
Fit indexes for the two-factor model
GFI 1.00 1.00 1.00 1.00 1.00 1.00 1.00 1.00 1.00 1.00
AGFI 1.00 1.00 1.00 1.00 1.00 1.00 1.00 0.99 1.00 1.00
CFI 1.00 1.00 1.00 1.00 1.00 1.00 1.00 0.98 1.00 1.00
Factor loading
Overall attitude toward brand’s corporation
This corporation makes best vehicle 0.72 0.72 0.82 0.71 0.67 0.58 0.53 0.78 0.59 0.32
This is the most successful auto manufacturer 0.71 0.77 0.89 0.77 0.78 0.84 0.84 1.00 0.96 0.95
Overall attitude toward brand’s country
This country makes the best vehicle 0.90 0.83 0.97 0.84 0.95 0.99 0.98 0.90 1.00 0.93
This country makes the vehicle I accept 0.71 0.71 0.71 0.84 0.84 0.71 0.71 0.53 0.80 0.71

NOTE: NC = not convergent; GFI = Goodness-of-Fit Index; AGFI = Adjusted Goodness-of-Fit Index; CFI = Comparative Fit Index; RMR = root mean square residual.
a. Additional measures.

261
262 JOURNAL OF THE ACADEMY OF MARKETING SCIENCE SUMMER 2004

Thailand, which are considered less-developed countries. uous variables have been normalized (i.e., put in
In summary, the results of item-factor loadings in most deviation-score form with means of zero) to help ensure
countries indicate a modestly satisfactory validity of con- unbiased parameter estimates in regression analysis
vergence, suggesting that the revised three-factor model is (Aiken and West 1991).
a possible and plausible way of relating brand associations
to second-order image dimensions (i.e., sensory, symbolic,
and utilitarian) across most countries. Nevertheless, RESULTS OF HYPOTHESES TESTING
unlike product-related brand-image dimensionality, the
umbrella brand consists of the same pattern of factor struc- In this section, we specify a three-level multilevel re-
tures and has higher loadings (λx > .50) across nations, gression model: the country is Level 3 (between coun-
suggesting a satisfactory reliability and validity level. tries), the respondent is Level 2 (between individuals), and
However, it is notable that national differences are the brand clustered within individuals is Level 1. In the
found in terms of the combination of brand associations fixed part of the model, brand-image perceptions, includ-
corresponding to each image dimension. For example, the ing the seven latent scores derived from product, corpo-
sensory dimension is formed with the inclusion of three of rate, and country-image dimensions, are specified as pre-
the four measures extracted from the global sample (i.e., dictors. National characteristics (e.g., level of economic
“exciting,” “fun to drive,” and “good acceleration and development, cultural dimension) and sociodemographic
speed”) for Belgium and Brazil, but only two measures factors (e.g., age, household income, sex, marital status)
(i.e., “exciting” and “fun to drive”) for Canada and India. are moderators, and brand characteristics are covariates.
The various combinations of measures applied to each
national sample suggest that factor constructs of brand- Moderating Effects of National Factors
image perceptions are similar but not statistically equiva-
lent across nations (Hsieh and Lindridge forthcoming). Although the hypotheses did not address main effects,
Nonetheless, development of psychometrically rigorous the main effects of all predictors, moderators, covariates,
cross-cultural scales is not the main purpose of our analy- and dummy variables are included in the analysis to esti-
sis. Rather, it is our intention to identify image dimen- mate moderating effects more precisely. Consistent with
sions in each country and to aggregate multiple indicators Roth’s (1995) results, the main effects of brand image are
underlying each image dimension in order to represent not significant (Table 4).3 Nonetheless, the corporation’s
unidimensionality. social responsibility image (β = .022), overall attitude
toward a brand’s corporate image (β = .153), and overall
Calculating Image Scores attitude toward a brand’s country image (β = .023) show
their significant main effects on purchase behavior in the
Instead of incorporating a large number of brand asso- positive direction, as we proposed, indicating that individ-
ciations, we aggregate multiple indicators into a single ual brands benefit from consumers’ positive attitude
composite of latent variables for multilevel analysis. The toward umbrella-brand tiers.
estimated latent variable scores can be analyzed as if they Our framework posits cross-level interactions between
were observed, given that the correlation matrix of the indvidual differences and national economic and cultural
constructs was used as the input matrix in generating latent dimensions. Table 4 shows that four interactive terms (i.e.,
variable scores. As such, any problem with scale reliability Symbolic Image × Economic Development, Symbolic
merely renders conservative the strength of the direct Image × Power Distance, Sensory Image × Power Dis-
and moderating effects in the subsequent analysis. The tance, and Sensory Image × Individualism) are positive
summed latent variable scores reduce model complexity and statistically significant (β = .026, β = .038, β = .026,
and avoid the limitations of multilevel application with β = .030), which provides evidence to validate Roth’s
categorical data. (1995) findings. Symbolic image has a greater effect on
In summary, seven latent scores, including three prod- purchase behavior in countries with higher levels of
uct-image dimensions (i.e., utilitarian, sensory, and sym- economic development (Hypothesis 1) and countries that
bolic appeals) and four corporate and country image are characterized by higher power distance (Hypothesis
dimensions (i.e., corporate publicity, corporate social re- 4). The effect of sensory image on purchase behavior is
sponsibility, overall attitude toward a brand’s corporate greater in countries that are characterized by higher
image, and overall attitude toward a brand’s country individualism (Hypothesis 3). However, contrary to our
image) have been generated from related multiple indica- expectations, Hypotheses 1 and 5, which state that the
tors.2 Table 3 reports Pearson correlations and descriptive effect of utilitarian image on purchase behavior is greater
statistics for the variables used to examine the research in countries characterized by higher uncertainty avoidance
hypotheses. Because of scale differences across measures and lower levels of economic development, respectively,
and to reduce any possible multicollinearity, some contin- are not supported. The formal hypothesis (Hypothesis 5)
TABLE 3
Correlations and Descriptive Statistics for Dependent, Independent, Moderator, and Covariate Variables
Dependent Independent Moderators Covariates

Normalized Normalized
Brand Normalized Normalized Normalized Normalized Local Brand Market
Variable Purchased Utilitarian Symbolic Sensory Publicity NSR NOAM NOAC NPDI NUA NINDI ED Marital Age Gender Income Origin Awareness Share

Brand purchased 1.000


Normalized utilitarian .066 1.000
Normalized symbolic (.003) (.012) 1.000
Normalized sensory (.004) (.395) .017 1.000
Normalized publicity .093 .073 .037 .001 1.000
NSR .049 .067 .034 .017 (.049) 1.000
NOAM .202 .077 .116 .040 .075 .057 1.000
NOAC .088 .045 .118 .106 .048 .048 .403 1.000
NPDI .010 (.001) (.000) (.001) .000 .002 .000 .000 1.000
NUA .008 (.001) .000 (.001) .001 (.004) .000 .000 .613 1.000
NINDI (.007) .001 (.000) .000 (.001) .000 .001 (.000) (.697) (.480) 1.000
ED (.002) .001 (.000) .001 .000 (.008) .001 (.000) (.749) (.217) .716 1.000
Marital status .000 .011 (.005) (.019) .021 .020 (.009) (.017) .027 (.033) (.002) (.005) 1.000
Age .003 .019 (.012) (.031) .006 .020 (.003) (.018) (.160) (.148) .239 .181 .443 1.000
Sex (.014) .006 (.011) .001 .002 .013 (.012) (.010) .118 .053 (.170) (.186) .071 .021 1.000
Income (.005) .011 .008 (.009) (.015) (.011) (.003) (.005) (.136) (.088) (.086) .070 .172 .036 (.003) 1.000
Local origin .141 .006 (.001) .029 .109 .030 .112 .147 (.071) .124 (.029) .185 .030 .032 (.027) .038 1.000
Normalized brand awareness .109 .101 .106 .036 .164 .072 .161 .187 (.194) (.063) .139 .184 .004 .061 (.104) .030 .168 1.000
Normalized market share .367 .102 (.045) (.059) .229 .053 .153 .082 .017 .036 (.010) .011 .002 .006 (.038) (.023) .401 .301 1.000

Normalized Normalized
Brand Normalized Normalized Normalized Normalized Local Brand Market
Purchased Utilitarian Symbolic Sensory Publicity NSR NOAM NOAC NPDI NUA NINDI ED Marital Age Gender Income Origin Awareness Share

High 0.00 –3.81 –1.92 –3.13 –2.16 –1.62 –2.21 –0.97 –1.30 –1.78 –1.64 –1.74 0.00 1.00 0.00 1.00 0.00 -5.00 –0.55
Low 1.00 4.72 5.33 5.98 4.31 5.58 5.91 2.56 1.82 1.32 1.27 1.57 1.00 5.00 1.00 6.00 1.00 0.65 7.29
M 0.04 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.22 0.63 2.42 0.68 3.57 0.07 0.00 0.00
SD 0.20 1.00 1.00 1.00 1.00 1.00 1.00 1.00 1.00 1.00 1.00 0.89 0.48 1.21 0.47 1.57 0.25 1.00 1.00
N 56,983 56,983 56,983 56,983 56,982 56,982 56,983 56,983 56,983 56,983 56,983 56,983 56,871 56,933 56,983 44,281 56,983 56,983 56,983

NOTE: NSR = normalized social responsibility; NOAM = normalized overall attitude toward brand’s corporate; NOAC = normalized overall attitude toward brand’s country image; NPDI = Normalized Power Dis-
tance Index; NUA = normalized uncertainty avoidance; NINDI = normalized individualism; ED: economic development.

263
264 JOURNAL OF THE ACADEMY OF MARKETING SCIENCE SUMMER 2004

TABLE 4 TABLE 4
Standardized Multilevel Coefficients (Continued)
Fixed Part of Model β p > |Z| Fixed Part of Model β p > |Z|

Constant –.065 .019** Sensory × Individualism .030 .000***


Sensory image –.005 .768 Symbolic × Individualism .003 .701
Symbolic image –.011 .470 Utilitarian × Individualism –.009 .288
Utilitarian image –.006 .726 Market share .346 .000***
Corporate publicity .003 .498 Brand awareness –.003 .656
Corporate social responsibility .022 .000*** Local origin –.106 .000***
Overall attitude toward brand’s corporate .153 .000***
Overall attitude toward brand’s country .023 .000*** BMW –.053 .066*
Marital status .002 .842 Chrysler .075 .015**
Age .001 .733 Daewoo .161 .000***
Gender .000 .992 Daihatsu .091 .011***
Income .002 .507 Fiat .110 .000***
Sensory × Marital –.017 .130 Ford .019 .545
Symbolic × Marital .001 .957 GM .057 .075*
Utilitarian × Marital –.018 .127 Honda .124 .000***
Sensory × Age .001 .885 Hyundai .162 .000***
Symbolic × Age .004 .410 Isuzu .038 .289
Kia .112 .004***
Utilitarian × Age .011 .020**
Mazda .101 .001***
Sensory × Sex .009 .421
Mercedes Benz –.168 .000***
Symbolic × Sex .007 .495
Mitsubishi .044 .150
Utilitarian × Sex .005 .615 Nissan .072 .018**
Sensory × Income .010 .003*** Peugeot .106 .001***
Symbolic × Income .001 .807 Renault .095 .002***
Utilitarian × Income .005 .097* Saab .077 .017**
Level of economic development .017 .238 Samsung .132 .003***
Power distance .018 .198 Subaru .091 .008***
Uncertainty avoidance –.012 .205 Suzuki .120 .000***
Individualism –.009 .382 Toyota .025 .415
Sensory × Economic Development .007 .564 Volkswagen –.054 .074*
Symbolic × Economic Development .026 .010*** Volvo .038 .221
Utilitarian × Economic Development .010 .389
Sensory × Power Distance .026 .012** Random Part of Model tauhat p > |Z|
Symbolic × Power Distance .038 .000*** Nation level .000 .093***
Utilitarian × Power Distance –.014 .167 Individual level –.031 .000***
Sensory × Uncertainty Avoidance .004 .543 Brand level .828 .000***
Symbolic × Uncertainty Avoidance –.024 .000***
NOTE: –2 log-likelihood: decreased from 122815 in 1 iteration to 115457
Utilitarian × Uncertainty Avoidance –.003 .622
in 4 iterations.
(continued) *p < .10. **p < .05. ***p < .01.

was also rejected in Roth’s study. The rejection of Hypoth- (β = .010), which suggests that high-income groups value
esis 1 might be due to the relatively low scale reliability of a sensory image. However, the hypotheses regarding the
the measurement model, which might moderate the strength difference between female and male respondents, in
of the effects reported for less-developed countries. which women put more emphasis on symbolic appeals
(Hypothesis 7), while men focus more on utilitarian image
Moderating Effects of appeals (Hypothesis 6), are not supported. Also, Hypothe-
Sociodemographic Factors sis 8, which states that people with higher incomes are
more likely to value the symbolic image appeal, is not
Findings were mixed for the moderating effects of supported.
sociodemographics. In line with our expectation (Hypoth- To demonstrate the generalizability of the previously
esis 8), the findings show that older respondents are more discussed findings, we conduct further diagnosis on statis-
attracted to the utilitarian image (β = .011). Although the tically significant effects using the individual country data
interaction of sensory image and income is not addressed (see Table 5). Logit regressions are performed using the
by a formal hypothesis, note that it has a significant effect same set of variables, but excluding macro-level inde-
Hsieh et al. / BRAND IMAGE 265

pendent and moderating variables (i.e., Macro-Level Vari- and national-level factors in a single study to enhance the
ables × Image Dimensions). Chi-square tests are used to validity of international market segmentation research. It
assess whether the effect (i.e., coefficient) differs from is this gap that we have sought to bridge.
zero. Overall, the results from the individual countries We have tested our hypotheses across 20 countries and
support the aggregate results. In general, the effects of have used large samples of consumers, which allows for a
overall attitude toward corporate and country image on strong test of the empirical generalizability of our find-
brand-purchase behavior are confirmed for most of the ings. As the globalization process quickens and several
countries with an alpha of .10.4 less-developed countries join the global market, there is a
As far as the moderating effect is concerned, some indi- need to investigate emerging markets. In general, the find-
cations were found to support the results of aggregate anal- ings suggest the following: (1) benefit-based multidimen-
ysis. For instance, symbolic image is more effective in sional brand-image perception corroborates brand con-
France, which has a relatively higher degree of power dis- cepts (i.e., functional, sensory, and social concepts) in
tance, and in the Netherlands, which has a relatively lower corresponding to consumers’ diverse needs (Park et al.
degree of uncertainty avoidance. In contrast, symbolic 1986) across economically and culturally diverse nations,
image is less effective in Belgium, which has the highest (2) the perceptions of and attitude toward umbrella brands
degree of uncertainty avoidance, and Germany, which has have main effects on brand-purchase behavior, and (3)
the lowest degree of power distance among surveyed national characteristics as well as sociodemographics play
countries. On the contrary, sensory image enhances pur- significant moderating roles in the relationship between
chase behavior in Belgium and Canada, both of which brand-image perceptions and brand-purchase behavior.
have higher degrees of individualism. Sensory image is With respect to the effects of umbrella-brand image,
less effective in Taiwan, which has the lowest degree of our findings indicate that brand-purchase behavior is posi-
individualism among surveyed countries, and Germany, tively affected by consumers’ feelings toward the corpo-
which has the lowest degree of power distance. In line with rate image and country image of an individual brand. An
the aggregate analysis, the moderating effect of utilitarian interesting point emerging from this study is that the
image on purchase behavior was not found. With respect effects of umbrella-brand images in consumers’ purchase
to the moderating effect of sociodemographics, the inter- behavior is evident not only in developed countries but
active term of sensory image with income was found in also in emerging markets, including former Socialist
Germany, Taiwan, and the United States. However, other countries where related research is scarce. Accordingly,
interactions of sociodemographics and product-image we suggest that firms continue to engage in activities that
dimensions that are significant in aggregate analysis are build country and corporate equity, given that both equi-
found to be insignificant across countries. The insignifi- ties are likely to benefit brand performance. Also, sup-
cance of the interactions might be due to the small sample ported by categorization theory, basic information (e.g.,
sizes used in the individual countries. information about umbrella-brand tiers) is more predomi-
nant and primary than is more complex information (e.g.,
information about specific brand associations) and there-
THEORETICAL IMPLICATIONS fore is easier to retrieve, which also supports the impor-
tance of consumers having a positive image of umbrella
In this study, we have examined how macro-level and brands.
micro-level factors moderate the effect of brand-image In terms of moderating effects, our findings extend
perception on consumer brand-purchase behavior. Our Roth’s (1995) work, which suggests that brand-image
framework incorporates product-image dimensions and strategies and their effects on firm performance should
umbrella-brand-image dimensions (i.e., corporate image generalize across national markets that are similar eco-
and country image) and considers individual- and nomically and culturally to consumer-based brand-image
national-level factors that moderate the relationship perceptions and their effects. The findings highlight the
between brand-image perception and brand-purchase important role of national characteristics and consumers’
behavior relationship. Incorporation of product, corpo- sociodemographic variables in moderating the effects of
rate, and country image into brand-image perception product-related brand-image dimensions on brand-pur-
enhances the validity and practical utility of perceptions. It chase behavior. As the findings illustrate, an emphasis on
is evident that, like product image, corporate and country symbolic increases the brand-purchase behavior of con-
images can provide consumers with meaningful associa- sumers from countries with higher levels of economic
tions to create an appealing image and thus can affect con- development and higher degrees of power distance, and
sumer behavior. However, not much research has consid- sensory appeals enhance the purchase behavior of con-
ered various image sources in a single framework. Also, sumers from countries characterized by higher degrees of
there is a need to bring both consumer and country seg- individualism and power distance. With respect to socio-
mentation strategies together to consider both individual- demographics, we find that older respondents tend to be
266
TABLE 5
Significant Effects on a Country-by-Country Basis
The South United
Australia Belgium Brazil Britain Canada China France Germany India Italy Japan Mexico Netherlands Russia Korea Spain Taiwan Thailand Turkey States

Symbolic image – + – +
Sensory image + + – –
Utilitarian image
Publicity + +
Social responsibility + + + +
Overall attitude toward brand’s corporation + + + + + + + + + + + + + + + +
Overall attitude toward brand’s country + + + + + + + + + +
Utilitarian Image × Age – +
Utilitarian × Income + –
Sensory × Income – + + +

2
R .302 .333 .536 .325 .416 .305 .432 .348 .718 .350 .383 .421 .323 .390 .638 .347 .381 .381 .503 .300
2
χ 242 221 467 236 165 61 368 240 412 191 350 316 211 42 619 238 305 151 358 285
df = 43, significance = .000

NOTE: + denotes a significant positive effect at the .10 level; – denotes a significant negative effect at the .10 level.
Hsieh et al. / BRAND IMAGE 267

attracted by utilitarian image appeal, and the effect of sen- LIMITATIONS AND FUTURE RESEARCH
sory image on purchase behavior is greater among high-
income respondents. These findings are important from As does any empirical study, this research has several
a theoretical perspective, because they aid in the under- limitations that present opportunities for further research,
standing of why certain image appeals are not effective in including (1) the examination of the moderating effect on
some countries and among some consumer segments; they the relationship between umbrella-brand images and pur-
show that the macro environment or micro characteristics chase behavior, (2) extensions of the present framework to
of consumers might not be conducive. other product categories, (3) extensions of contextual fac-
In terms of methodological implications, the current tors and an examination of those factors’ antecedents and
study illustrates necessary steps to measure association- indicators, (4) extensions of the image construct, and (5)
based brand-image perception by developing measure- measurement improvement. First of all, given that this
ment models and by aggregating multiple indicators of study is a consumer-based extension of Roth’s (1995)
brand associations that underlie each image dimension. work, only the moderating effects on product-image
The proposed approach has implications for data mining, dimensions have been examined. However, potential
particularly when multiple indicators are measured by moderating effects on the corporate- and country-image
nominal scales. dimensions might exist. For instance, more educated con-
sumers might put more weight on social responsibility, or
younger consumers might pay more attention to publicity.
Future research can explore such possibilities. Second,
MANAGERIAL IMPLICATIONS although to some extent the findings can be generalized to
consumer goods that have a variety of appeals, additional
Our findings also generate insights directly applicable work is needed to explore the generalizability of the
for marketing managers. First, brand image building helps results in other product categories. For example, brand-
consumers understand what a brand can do for them in image research pertaining to industrial rather than con-
terms of either product-, corporate- or country-related sumer goods and services would be an interesting area for
benefits. This finding contributes to the understanding of investigation.
the underlying image dimensions of benefit-based brand Third, additional research could investigate other
associations and provides a conceptual basis for linking typologies of national cultural dimensions, such as the one
brand-purchase behavior and consumer needs (Gutman developed by Schwartz (1994), or it could extend the cur-
1992). The different image dimensions can be related to rent framework by including the Confucian dynamic
different actions by brand managers. In addition, because dimension that Hofstede identified in his later work
much more information (higher level of abstraction) is (1991). This later work distinguishes a dynamic, future-
summarized in benefit-oriented brand association than in oriented mentality from a static mentality that is focused
product attributes, the benefit-oriented image dimensions on the past and the present. However, this cultural dimen-
provide guidelines at the strategic marketing level. Thus, sion has attracted relatively little attention, probably
the findings facilitate effective communication strategies. because of its small coverage of country groups. Nonethe-
For example, an advertising message containing an appeal less, the potential of this cultural dimension in marketing
of reliability can support a utilitarian brand image. Simi- research applications is anticipated. Apart from cultural
larly, corporate image can be enhanced by delivering a dimensions, there are other variables from the national,
social responsibility message to the target audience. brand, and individual levels that could be investigated. For
Second, our study evaluates the feasibility of region- example, national-level factors include automobile import
alization strategies based on country segments and that of regulations and the tendency of consumer ethnocentrism
globalization strategies based on consumer segments (Shimp and Sharma 1987). Brand-level factors include the
(Hassan and Katsanis 1994). Because the findings suggest length of time the brand has been in the market, brand dis-
that both national and sociodemographic variables moder- tribution aspects, and advertising and marketing expen-
ate the effects of image dimension on purchase behavior, diture variables. Individual-level variables other than
we assume that regionalization based only on macro-level sociodemographic variables could also be considered. A
variables might not be sufficient to maximize the effect of potentially interesting construct is consumer innovation,
image appeals; supplementary segmentation based on which could be operationalized through well-developed
micro-level variables (i.e., sociodemographics) might pro- scales, such as the exploratory acquisition of a product
vide additional value. The empirical findings enable mar- scale (Baumgartner and Steenkamp 1996). Consumer
keting managers to work on a strategy premise that may be innovation may moderate the effect of brand image on
applied across borders while it is still being tailored, where purchase behavior for a product introduced in a new for-
necessary, to the characteristics of the local market and eign market. Broadening the scope of information used to
specific sociodemographic segments. assess national and individual differences would afford
268 JOURNAL OF THE ACADEMY OF MARKETING SCIENCE SUMMER 2004

better insights into key cross-market differences, therefore Mitsubishi, Nissan, Peugeot, Porsche, Renault, Saab, Samsung, Subaru,
Suzuki, Toyota, Volkswagen, Volvo, Acura, Alfa Romeo, Buick, Cadil-
ensuring the selection of the most appropriate market-
lac, Chevrolet, Citroen, Dodge, Eagle, GMC, Holden, Infiniti, Jaguar,
entry strategy. Jeep, Lancia, Land Rover, Lexus, Lincoln, Mercury, Oldsmobile, Opel,
Fourth, the brand associations underlying image per- Plymouth, Pontiac, Rover, Seat, Saturn, Talbot, and Vauxhall.
ceptions could be expanded. User imagery that also has
distinct meanings for consumers could be included in the 2. It has been suggested that single-indicator constructs are unattrac-
association set. This might be a more effective strategy tive because they ignore the unreliability of measurement (Bollen 1989).
than the attempt that associates brand with benefit-based 3. The standard output of multilevel analysis provided by LISREL
2
application does not include statistics such as the R , F test, and z-scores.
images only. Also, the interdimensional correlation of We therefore perform the model using binary logistic regression provided
these distinctive image dimensions could shed some light 2
by the SPSS application. The results indicate a moderate R of .342 and a
on the complexity of the brand-image perception struc- significant F test with 62 degrees of freedom. Although the coefficients
ture. For example, according to means-end chain theory, obtained from SPSS are not identical for the results derived from
the representation of cognitive structures in memory is LISREL, the same significant main and moderator effects are identified.
However, it is noted that logistic regression does not take the multilevel
based on the notion that the product and the consumer may nature of the data into consideration.
be hierarchically linked through an interconnected set of 4. When setting the alpha level, we were concerned with the power of
cognitive elements with different levels of abstraction the statistical tests. In addition to avoiding Type I error by setting a con-
(Olson and Reynolds 1983). The hierarchical linkages servative value for alpha, we also need to be able to identify the significant
among these image dimensions provide a possible contribution of the variables if they do indeed contribute to the prediction
model. If alpha was set too small when we used the individual country
structure of brand-image preconceptions. data—where the sample size is smaller—the power could be too low to
Finally, measurement improvements are possible and identify the significant variables in the model. As such, we increase alpha
may be important for further research. Given that we have to the .10 level.
used an existing data set in this study, the potential mea-
surement problem arising from questionnaire design and
international survey implementation is beyond our con- REFERENCES
trol. Although the measurement model has proved accept-
able at a pool level, the result of our national-level analy- Aaker, David A. 1996. “Measuring Brand Equity Across Products and
sis, which seeks invariant factor patterns across countries Markets.” California Management Review 38 (Spring): 102-120.
Aiken, Leona and Stephen West. 1991. Multiple Regression: Testing and
in a more rigorous way, is unsatisfactory. Also, although Interpreting Interaction. Newbury Park, CA: Sage.
the use of integrated latent variable scores to represent Alden, Dana L., Jan-Benedict E. M. Steenkamp, and Rajeev Batra. 1999.
multi-item measures is practically justifiable in the estima- “Brand Positioning Through Advertising in Asia, North America,
and Europe: The Role of Global Consumer Culture.” Journal of Mar-
tion of coefficients in the multilevel model, this approach keting 63 (1): 75-87.
might reduce the full structural equation model to the level Al-Sulaiti, Khalid I. and Michael J. Baker. 1998. “Country of Origin
of a path model. Regarding the use of self-reporting to Effects: A Literature Review.” Marketing Intelligence & Planning 16
measure key concepts, the variance explained may be (3): 150-199.
Baumgartner, Hans and Jan-Benedict E. M. Steenkamp. 1996. “Explor-
biased upwardly because of the self-reported nature of the atory Consumer Buying Behaviour: Conceptualization and Measure-
individual behavior response. To address this validity ment.” International Journal of Research in Marketing 13:121-137.
issue, future replications might use multiple methods to Bird, M., C. Channon, and A. S. C. Ehrenberg. 1970. “Brand Image and
Brand Usage.” Journal of Marketing Research 7 (August): 307-314.
eliminate the rival explanation of shared method dissent. Bollen, Kenneth. 1989. Structural Equations With Latent Variables. New
York: John Wiley.
Byrne, Barbara M. 1998. Structural Equation Modeling With LISREL,
PRELIS and SIMPLIS: Basic Concepts, Applications, and Program-
ACKNOWLEDGMENTS ming. Mahwah, NJ: Lawrence Erlbaum.
Clark, Terry. 1990. “International Marketing and National Character: A
Review and Proposal for an Integrative Theory.” Journal of Market-
The authors acknowledge the contribution of ing 54 (October): 66-79.
MORPACE International Inc. in providing the data set for Cordell, V. 1991. “Competitive Context and Price as Moderators of
analysis. The authors also are grateful for the reviewers’ Country of Origin Preference.” Journal of the Academy of Marketing
Science 19 (Spring): 123-128.
and the editor’s comments. This study was partially de Mooij, Marieke. 2000. “The Future Is Predictable for International
funded by the National Science Council under the research Marketers Converging Income Lead to Diverging Consumer Behav-
project NSC 92-2416-H-155-035. ior.” International Marketing Review17 (2): 103-113.
Dittmar, Helga, Jane Beattie, and Susanne Friese. 1995. “Gender Identity
and Material Symbols: Objects and Decision Considerations in
Impulse Purchase.” Journal of Economic Psychology 16:491-511.
NOTES Dobni, Dawn and George M. Zinkhan. 1990. “In Search of Brand Image:
A Foundation Analysis.” In Advances in Consumer Research, Vol. 17.
Eds. Marvin Goldberg, Gerald Gorn, and Richard Pollay. Provo, UT:
1. Only the evaluations for the top 25 brands were included in this Association for Consumer Research, 110-119.
study. These brands were BMW, Chrysler, Daewoo, Daihatsu, Fiat, Ford, Douglas, Susan P. and Samuel Craig. 1983. International Marketing
General Motors, Honda, Hyundai, Izuzu, Kia, Mazda, Mercedes Benz, Research. Englewood Cliffs, NJ: Prentice Hall.
Hsieh et al. / BRAND IMAGE 269

Doyle, P. 1989. “Building Successful Brands: The Strategic Options.” Kapferer, Jean-Noël. 1992. “How Global Are Global Brands?” In The
Journal of Marketing Management 5 (Summer): 77-95. Challenge of Branding Today and in the Future, proceedings of the
Erickson, Gary M., Johnny K. Johansson, and Paul Chao. 1984. “Image conference of the European Society for Opinion and Marketing
Variables in Multi-Attribute Product Evaluations: Country of Origin Research (ESOMAR), October 28-30, Brussels, Belgium. Brussels,
Effects.” Journal of Consumer Research 11 (September): 694-699. Belgium: ESOMAR, 199-215.
, , and R. Jacobson. 1992. “Competition for Market Share Keller, Kevin Lane. 1993. “Conceptualizing, Measuring, and Managing
in the Presence of Strategic Invisible Assets: The US Automobile Customer-Based Brand Equity.” Journal of Marketing 57 (January):
Market, 1971-1981.” International Journal of Research in Marketing 1-22.
9:23-37. . 2000. “The Brand Report Card.” Harvard Business Review 78
Gardner, Burleigh and Sidney J. Levy. 1955. “The Product and the (January-February): 147-157.
Brand.” Harvard Business Review 33 (March-April): 33-39. Leclerc, France, Bernd H. Schmitt, and Laurette Dube. 1994. “Foreign
Gensch, Dennis H. 1978. “Image-Measurement Segmentation.” Journal Branding and Its Effects on Product Perceptions and Attitudes.” Jour-
of Marketing Research 15 (August): 384-394. nal of Marketing Research 31 (May): 263-270.
Gitelson, Richard J. and Deborah L. Kerstetter. 1990. “The Relationship Levitt, Theodore. 1983. “The Globalization of Markets.” Harvard Busi-
Between Sociodemographic Variables, Benefits Sought and Subse- ness Review 61 (May-June): 92-102.
quent Vacation Behaviour: A Case Study.” Journal of Travel Lowe, Anthony Chun-Tung and David R. Corindale. 1998. “Differences
Research 28 (Winter): 24-29. in ‘Cultural Values’ and Their Effects on Responses to Marketing
Goodyear, Mary. 1996. “Divided by a Common Language: Diversity and Stimuli: A Cross-Cultural Study Between Australians and Chinese
Deception in the World of Global Marketing.” Journal of the Market From the People’s Republic of China.” European Journal of Market-
Research Society 38 (2): 105-122. ing 32 (9/10): 843-867.
Gutman, Jonathan. 1992. “Exploring the Nature of Linkages Between Lynn, Michael, George M. Zinkhan, and Judy Harris. 1993. “Consumer
Consequences and Values.” Journal of Business Research 22 (2): 143- Tipping: A Cross-Country Study.” Journal of Consumer Research 19
148. (December): 478-888.
Hair, Joseph F., R. Anderson, R. Tatham, and W. Black. 1995. Malhotra, Naresh K. 1988. “A Methodology for Measuring Consumer
Multivariate Data Analysis With Readings. 4th ed. Englewood Cliffs, Preference in Developing Countries.” International Marketing
NJ: Prentice Hall. Review 5 (3): 52-66.
Han, C. Min. 1989. “Country Image: Halo or Summary Construct?” Jour- Martin, Ingrid M. and Sevgin Eroglu. 1993. “Measuring a Multi-Dimen-
nal of Marketing Research 26 (May): 222-229. sional Construct: Country Image.” Journal of Business Research 28
Hassan, Salah S. and Lea Prevel Katsanis. 1994. “Global Market Seg- (November): 191-210.
mentation Strategies and Trends.” In Globalization of Consumer Meenaghan, Tony. 1995. “The Role of Advertising in Brand Image De-
Markets: Structure and Strategies. Eds. Salah S. Hassan and Erdener velopment.” Journal of Product and Brand Management 4 (4): 23-34.
Kaynak. New York: International Business Press, 47-62.
Motameni, Reza and Manuchehr Shahrokhi. 1998. “Brand Equity Valua-
Helsen, Kristiaan, Kamel Jedidi, and Wayne S. DeSarbo. 1993. “A New
tion: A Global Perspective.” Journal of Product and Brand Manage-
Approach to Country Segmentation Utilizing Multinational Diffu-
ment 7 (4): 275-290.
sion Patterns.” Journal of Marketing 57 (October): 60-71.
Mullen, Michael R. 1995. “Diagnosing Measurement Equivalence in
Hofstede, Geert H. 1983. “National Cultural in Four Dimensions: A
Cross-National Research.” Journal of International Business Studies
Research-Based Theory of Cultural Differences Among Nations.”
26 (3): 573-596.
International Studies of Management and Organization 13 (1-2): 46-74.
Narayana, Chem L. 1981. “Aggregate Images of American and Japanese
 (1991), Cultures and Organizations: Software of the Mind. Lon-
Product: Implications on International Marketing.” Columbia Jour-
don: McGraw-Hill.
nal of World Business 16 (Summer): 31-35.
Holbrook, Morris B. and Elizabeth C. Hirschman. 1982. “The Experi-
Niss, H. 1996. “Country of Origin Marketing Over the Products Life
ential Aspects of Consumption: Consumer Fantasies, Feelings, and
Cycle: A Danish Case Study.” European Journal of Marketing 30 (3):
Fun.” Journal of Consumer Research 9 (2): 132-140.
6-22.
Homer, Pamela M. 1990. “The Mediating Role of Attitude Toward the
Ad: Some Additional Evidence.” Journal of Marketing Research 27 Olson, J. and R. Reynolds. 1983. “Understanding Consumers’ Cognitive
(February): 78-86. Structures: Implications for Advertising Strategy.” In Advertising
and Consumer Psychology, Vol. 1. Eds. L. Percey and A. Woodside.
Hong, Sung-Tai and Robert S. Wyer Jr. 1989. “Effects of Country-of-
Lexington, MA: Lexington Books, 77-91.
Origin and Product Attribute Information on Product Evaluation: An
Information Processing Perspective.” Journal of Consumer Research Park, C. Whan, Bernard J. Jaworski, and Deborah J. MacInnis. 1986.
16:175-187. “Strategic Brand Concept-Image Management.” Journal of Market-
Hsieh, M. H. and A. Lindridge. Forthcoming. “Universal Appeals With ing 50 (October): 135-145.
Local Specifications.” Journal of Product & Brand Management 13. , Sandra Milberg, and Robert Lawson. 1991. “Evaluation of
Hugstad, P. and M. Durr. 1986. “A Study of Country of Manufacturer Brand Extension: The Role of Product Level Similarity and Brand
Impact on Consumer Perceptions.” In Development in Marketing Sci- Concept Consistency.” Journal of Consumer Research 18 (Septem-
ence, Vol. 9. Eds. J. Hawes and N. K. Malhotra. Coral Gables, FL: ber): 185-193.
Academy of Marketing Science, 115-119. Park, Chan Su and V. Srinivasan. 1994. “A Survey-Based Method for
Jain, Subhash. 1989. “Standardization of International Marketing Strat- Measuring and Understanding Brand Equity and Its Extendibility.”
egy: Some Research Hypotheses.” Journal of Marketing 53 (Janu- Journal of Marketing Research 58 (May): 271-288.
ary): 70-79. Roth, Martin S. 1995. “The Effects of Culture and Socioeconomics on the
. 1993. International Marketing Management. 4th ed. Boston: Performance of Global Brand Image Strategies.” Journal of Market-
PWS-KENT. ing Research 32 (May): 163-175.
Johansson, Johny K. 1993. “Missing a Strategic Opportunity: Managers’  and Jean B. Romeo. 1992. “Matching Product Category and
Denial of Country-of-Origin Effect.” In Product-Country Image: Country Image Perceptions: A Framework for Managing Country-of-
Impact and Role in International Marketing. Eds. N. Heslop and L. A. Origin Effects.” Journal of International Business Studies l (3): 477-
Papadopoulos. New York: International Business Press, 77-86. 497.
Johnson, Madeline and George M. Zinkhan. 1990. “Defining and Mea- Saunders, John and Guoqun Fu. 1997. “Dual Branding: How Corporate
suring Company Image.” In Developments in Marketing Science, Vol. Names Add Value.” Journal of Product and Brand Management 6 (1):
13. Ed. B. J. Dunlap. Provo, UT: Academy of Marketing Science, 40-48.
346-350. Schwartz, Shalom H. 1994. “Beyond Individualism/Collectivism: New
Kale, Sudhir H. 1995. “Grouping Euroconsumers: A Culture-Based Cultural Dimensions of Values.” In Individualism and Collectivism:
Clustering Approach.” Journal of International Marketing 3 (3): 35- Theory, Method, and Application. Eds. U. Kim et al. Thousand Oaks,
48. CA: Sage, 85-119.
270 JOURNAL OF THE ACADEMY OF MARKETING SCIENCE SUMMER 2004

Shimp, Terence A., Saeed Samiee, and Thomas J. Madden. 1993. “Coun- and Prudential in Taiwan. She is currently studying a variety of
tries and Their Products: A Cognitive Structure Perspective.” Journal
topics in cross-national consumer research. Specifically, she is
of the Academy of Marketing Science 21 (4): 323-330.
 and Subhash Sharma. 1987. “Consumer Ethnocentrism: Con- exploring topics in brand management. She is also interested in
struction and Validation of the CETSCALE.” Journal of Marketing the areas of Internet marketing and customer relationship man-
Research 24 (August): 280-289. agement (CRM). She has articles published in the Journal of
Simmons, Carolyn J. and John G. Lynch. 1991. “Inference Effects With- International Marketing, the Journal of Product and Brand Man-
out Inference Making? Effects of Missing Information on Discount-
agement, and the Journal of Organizational Computing and
ing and Use of Presented Information.” Journal of Consumer
Research 17 (March): 477-491. Electronic Commerce.
Srivastava, Rajendra K. and Allan D. Shocker. 1991. “Brand Equity: A
Prospective on Its Meaning and Measurement.” Working Paper No. Shan-Ling Pan (pansl@comp.nus.edu.sg), Ph.D., is an assistant
91-124. Marketing Science Institute, Cambridge, MA. professor and the coordinator of the Knowledge Management
Steenkamp, Jan-Benedict E. M., Frenkel ter Hofstede, and Michel Wedel.
Laboratory (http://kmlab.comp.nus.edu.sg) in the Department of
1999. “A Cross-National Investigation Into the Individual and
National Cultural Antecedents of Consumer Innovativeness.” Jour- Information Systems of the School of Computing at the National
nal of Marketing 63 (April): 55-69. University of Singapore. He received his MBA degree (1993)
 and Hans C. M. van Trijp. 1991. “The Use of LISREL in Validat- from the University of Texas at San Antonio, his MA degree
ing Marketing Constructs.” International Journal of Research in (1996) from the University of London School of Oriental and
Marketing 8:283-299.
Stern, Barbara B., George M. Zinkhan, and Morris B. Holbrook. 2002.
African Studies (SOAS), and his Ph.D. degree (2000) from the
“The Netvertising Image: Image Communication Model (NICM) and University of Warwick, United Kingdom. His primary research
Construct Definition.” Journal of Advertising 11 (3): 15-28. focuses on the recursive interaction of organizations and infor-
Tse, David K. and Gerald J. Gorn. 1992. “An Experiment on the Salience mation technology (enterprise systems), with particular empha-
of Country-of-Origin in the Era of Global Brands.” Journal of Inter- sis on organizational issues such as work practices, cultures,
national Marketing 1 (1): 57-76.
Wheatley, John J. and John S. A. Chiu. 1977. “The Effects of Price, Store structures, decision making, change, and strategy implemen-
Image, and Product and Respondent Characteristics on Perceptions tation. His research work has been published in Institute of Elec-
of Quality.” Journal of Marketing Research 14 (May): 181-186. trical and Electronics Engineers (IEEE) Transaction on
Whitelock, J. and D. Chung.1989. “Cross-Cultural Advertising: An Engineering Management, the Journal of the American Society
Empirical Study.” International Journal of Advertising 8 (3): 291-
310.
for Information Systems and Technology, Communications of the
Wilkie, William. 1986. Consumer Behavior. New York: John Wiley. Association for Computing Machinery, Information and Organi-
Yip, George S. 1995. Total Global Strategy: Managing for Worldwide zation, the Journal of Strategic Information Systems, the Euro-
Competitive Advantage. Englewood Cliffs, NJ: Prentice Hall. pean Journal of Information Systems, and Decision Support
Zhang, Y. 1996. “Chinese Consumers’ Evaluation of Foreign Products: Systems.
The Influence of Culture, Product Types and Product Presentation
Format.” European Journal of Marketing 30 (12): 50-68.
Zinkhan, George M. and Rudy Hirschheim. 1992. “Truth in Marketing Rudy Setiono (rudys@comp.nus.edu.sg) received his Bachelor
Theory and Research: An Alternative Perspective.” Journal of Mar- of Science degree from Eastern Michigan University and M.Sc.
keting 56 (April): 80-88. and Ph.D. degrees from the University of Wisconsin–Madison in
 and Penelope J. Prenshaw. 1994. “Good Life Images and Brand 1984, 1986, and 1990, respectively. He has been with the
Name Associations: Evidence From Asia, America, and Europe.” In
Advances in Consumer Research, Vol. 21. Eds. Chris Allen and
National University of Singapore since 1990 and he is currently
Deborah Roedder John. Provo, UT: Association for Consumer an associate professor in the School of Computing. His research
Research, 496-500. interests include linear programming, nonlinear optimization,
and neural networks. He is a senior member of IEEE and serves
as an associate editor of IEEE Transactions on Neural Networks.
ABOUT THE AUTHORS His publications have appeared in the SIAM Journal on Control
and Optimization, the European Journal of Operational
Ming-Huei Hsieh (hmh@saturn.yzu.edu.tw), Ph.D., is an assis- Research, the Journal of Optimization Theory and Applications,
tant professor in the Department of International Business, Yuan IEEE Transactions on Neural Networks, IEEE Transactions on
Ze University, Taiwan. She has obtained her Ph.D. degree (2002) Data and Knowledge Engineering, IEEE Transactions on Sys-
from Warwick Business School of the University of Warwick, tems Man and Cybernetics, Neural Computation, Neurocomput-
United Kingdom. Prior to her Ph.D. study, she conducted several ing, Connection Science, Management Science, and many other
market research projects for major multinationals, including reputable journals.
General Motors, Procter & Gamble, Bayer, Johnson & Johnson,

You might also like