You are on page 1of 23

On Improving the Conceptual

Foundations of International Marketing


Research

International marketing research plays a vital role as firms ABSTRACT


expand globally. Yet limited attention has been paid to the con-
ceptual underpinnings of research needed to guide expansion.
This article focuses on these issues and specifically examines
the conceptual framework that guides research, the geographic
units of analysis, and the constructs. The authors suggest two
alternative approaches to organizing cross-cultural research:
the adapted etic and the linked emic.

As markets and marketing activities become more integrated


and global in scope, increasing interest has emerged among Susan P. Douglas and
both academics and practitioners in conducting research in C. Samuel Craig
multicountry settings. Although the process of conducting
international marketing research does not differ from domes-
tic marketing research, there are pitfalls and problems asso-
ciated with extending the scope of research beyond the con-
fines of a single country’s borders. Many of these problems
are readily apparent and are related to the need to arrange for
data collection in multiple locations, to coordinate activities
over substantial geographic distances, or to translate research
instruments into different languages. Underlying these obvi-
ous differences are more subtle issues that threaten the
integrity of the research and stem from differences in the
sociocultural context in which research is conducted.

Some of these issues have been discussed in the interna-


tional marketing and cross-cultural psychology literature
(Craig and Douglas 2005; Sekaran 1983; Van de Vijver and
Leung 1997; Van Raaij 1978). In particular, issues related to
measurement and scale equivalence and procedures to exam-
ine these have been extensively discussed (Steenkamp and
Baumgartner 1998; Van de Vijver and Leung 1997). However,
issues related to the underlying conceptual framework that
guides the research have received considerably less attention
(Nakata and Huang 2002). Yet if research is to expand knowl-
Submitted January 2005
edge of consumer behavior to guide management decision Accepted July 2005
making, greater awareness of the importance of examining Journal of International Marketing
the universality of theories and constructs anchored in a spe- © 2006, American Marketing Association
Vol. 14, No. 1, 2006, pp. 1–22
cific research paradigm or developed in relation to a specific ISSN 1069-031X (print)
geographic or sociocultural setting is essential. 1547-7215 (electronic)

1
The purpose of this article is to argue for a greater emphasis
on the early stages of research design in international mar-
kets and to suggest ways to improve international marketing
research. Attention needs to be paid to the assumptions that
underlie the theories and concepts used, particularly their
relevance and meaning in a given sociocultural context. The
equivalence of the constructs and the way they are opera-
tionalized in different sociocultural settings should be
assessed and accurate, and reliable research instruments
need to be designed to tap the meaning of a construct in a
given setting. All too frequently, cross-country research
begins with the third component (i.e., the research instru-
ment) without consideration of the underlying conceptual
framework and related constructs and their applicability in
other research contexts. The major issues and their resolu-
tion appear in Table 1, which also outlines the structure for
the article.

The most critical aspect of international marketing research


DEVELOPING THE is the development of the conceptual framework that guides
CONCEPTUAL FRAMEWORK data collection and hypothesis testing. An important issue is
the applicability of the original framework in different con-
texts or research settings. In some cases, the underlying
theory may not be salient. Even when the basic theory
applies, constructs may need to be modified or the nature of
relationships may differ from one research context to
another. Indigenous theories may suggest alternative frame-
works or concepts to be considered. This implies that the
original conceptual framework may need to be modified.
Similarly, construct measurement and the unit of analysis

Issues Resolution
Table 1. Conceptual Framework Refining the Conceptual Framework
Resolving Issues in •Relevance of theory •Decentering the research perspective
Multicountry Marketing
•Relevance of construct •Assessing the role of context
Research
•Expression of construct •Identifying the mediators and moderators
•Identifying construct bias

Unit of Analysis Refining the Unit of Analysis


•Relevance of unit •Alternative units
•Independence •Purposive selection
•Comparability of country •Degree of cultural interpenetration
•Heterogeneity within countries

Construct Improving Contruct Measurement


•Same operationalization •Preliminary research and pretesting
•Stimulus comparability •Use of emic/etic instruments
•Back transaction •Use of parallel translation
•Method and item bias

2 Susan P. Douglas and C. Samuel Craig


may require modification. The same data analysis would
typically be conducted as it was in the original country, but
with the possible addition of control variables to account for
contextual differences.

A common procedure in many marketing studies is to take a


conceptual model or research design used in one country Typical Procedure
(often the United States) and to extend or “replicate” the
study in another country without consideration of the appli-
cability or relevance of the research design in this new con-
text. This approach assumes that the theory and constructs
are universal, or “etic.” It assumes that they are relevant in
any other research context and not specific to the context in
which they were developed. This debate about the etic ver-
sus emic nature of theories and constructs is pervasive
throughout cross-cultural research in the social sciences
(Berry 1969; Triandis 1972). The emic school argues that con-
structs and theories are specific to and need to be developed
and studied within a clearly defined context. The etic school
argues that constructs and theories should be universal and
apply to all contexts. A principal danger of the etic philoso-
phy is that it will result in a pseudoetic perspective or bias
(Triandis 1972); that is, theories and constructs are assumed
to be universal when they are actually emic. Typically, a
pseudoetic research perspective is strongly anchored in the
domestic market context. The domestic market forms the
dominant frame of reference. It is assumed that research
questions and issues that are salient in that market and cul-
tural context are also salient in other countries.

Relevance of Theories. To avoid a pseudoetic bias, the rele-


vance of theory and constructs in each research context
needs to be assessed. A theory is typically developed within
a given socioenvironmental context (often the United States)
and thus reflects specific characteristics of that context
(Nakata and Huang 2002; Van Raaij 1978). For example,
although marketing orientation has been validated in devel-
oping countries, the antecedents have been found to differ. In
Ghana, centralization and formalization are important
antecedents of adoption of a marketing orientation and the
tendency to acquire, disseminate, and use information. This
has been attributed to the early stage in transition to a mar-
keting orientation. African managers tend to be authoritarian
and to endorse ascribed status, leading to greater centraliza-
tion. In turn, this centralization is necessary to stimulate
market-oriented behavior in the short run (Kuada and Buatsi
2005). Similarly, in Thailand, an innovation orientation is an
important factor associated with management acceptance of
a marketing orientation (Powpaka 1998). Thus, the real
understanding of market orientation comes not from finding
that it exists but in understanding its antecedents and the
factors that influence it in each market.

Conceptual Foundations of International Marketing Research 3


Relevance of Constructs. Similarly, the relevance of con-
structs in each research context needs to be carefully exam-
ined. Even when a general theory applies, a specific con-
struct may not be relevant or have the same meaning or
implications in another context. Just as in the case of
theories, constructs are identified within a specific socio-
environmental context, which differs in certain respects
from other contexts. For example, Nakata and Sivakumar
(2001) suggest that the implementation of marketing con-
cepts can be expressed in the generation, dissemination, and
utilization of different types of information in different
national cultures. For example, in cultures that are low on
individualism and uncertainty avoidance, there may be less
market intelligence dissemination and utilization but higher
information generation than in other types of cultures. This
suggests that the way a theory is operationalized and the
relative importance of different constructs can vary depend-
ing on the sociocultural setting. Therefore, researchers must
pay attention to such issues in the development and testing
of research hypotheses in different settings.

Expression of Constructs. Constructs may be relevant in


other contexts but not expressed in the same way. This is
most likely to occur when constructs are related to attitudes
or behaviors that are contextually embedded. For example,
trust may be expressed in different ways in different cul-
tures. Andaleeb and Anwar (1996) find that in a developing
country, Bangladesh, customers’ trust in an organization had
a significant impact on their trust in its salespeople. This was
attributed to the sociocultural context, in which people did
not inspire the same level of trust as did recognized and
established organizations. Similarly, in the Chinese business
environment, trust in a seller or salesperson is more a func-
tion of the seller’s company’s reputation or creditworthiness
than a seller’s individual product expertise or knowledge
(Armstrong and Yee 2001) Consequently, trust in the organi-
zation has been found to be associated with trust at the indi-
vidual level and to play a crucial role in facilitating the
industrial selling process. This implies that careful attention
must be paid to how a construct is construed in a given con-
text. Rather than assuming that it will necessarily be
expressed in the same way, it is important to examine the
underlying sociological context and consult local experts to
explicate the concept appropriately.

Decentering the Research Perspective. The extrapolation of a


Refining the Conceptual research design used in the home country to other contexts is
Framework likely to result in bias arising from the cultural perspective of
the researcher (Lee 1966). The researcher is likely to identify
relevant constructs or relationships to be studied in the light
of aspects and issues that are significant in his or her “home”
context. Often, the researcher begins from a base of knowl-

4 Susan P. Douglas and C. Samuel Craig


edge and experience from doing research in his or her own
culture in relation to issues that are highly salient in that cul-
ture. For example, the role of public policy or government
influence is more pervasive in many European countries
than in the United States and is frequently the focus of or
included in studies in Europe. For example, Hofstede (1980)
notes that scholars with Western backgrounds often are inter-
ested in issues related to values and attitudes. Such issues
may be of less significance to scholars in emerging markets,
who may be concerned with issues related to economic
growth or education. The researcher is likely to interpret
events, behavior, and data in the light of his or her own expe-
rience and store of knowledge.

Assessing the Role of Context. Because research is conducted


in multiple sociocultural environments that differ from one
another with regard to certain important characteristics, it is
important to consider how this is related to the behavior
studied. For example, small countries with high levels of
foreign trade provide a different context for the examination
of consumer ethnocentrism or attitudes toward foreign
products than do large industrialized countries, in which
foreign trade is a smaller percentage of gross national prod-
uct (Nijssen and Douglas 2004). In the former case, there may
be no domestic manufacturers for products such as automo-
biles or computers, resulting in a more favorable attitude
toward foreign products than in countries in which domestic
manufacturers for such products are a major source of
employment.

Identifying Mediators and Moderators. The role of mediating


and moderating factors in the conceptual framework must
also be identified. These are factors embedded in the socio-
cultural context that mediate or moderate the relationship
under examination. For example, in the study of the impact
of consumer ethnocentrism and of product evaluations on
intentions to purchase foreign products, feelings of animos-
ity toward a particular country affect the evaluation of prod-
ucts from that country. These feelings of animosity may stem
from previous acts of aggression or fears of economic domi-
nance. For example, Chinese people’s feelings of animosity
toward Japan, due to the massacre at Nanjing in the 1930s,
were found to affect their evaluations of and willingness to
buy Japanese products in the 1990s (Klein, Ettenson, and
Morris 1998).

Identifying Construct Bias. Use of a research instrument


developed in a single cultural setting in other research set-
tings can result in construct bias.1 If the same instrument is
used in another setting, bias will arise if the appropriate def-
inition of the construct differs across cultures or research set-
tings. For example, in developing the IGEN scale to measure

Conceptual Foundations of International Marketing Research 5


intergenerational communication and influence, an initial
set of 37 items that measured three dimensions of intergener-
ational transmission—consumer skills, preferences, and atti-
tudes toward marketer-supplied information—was devel-
oped (Viswanalthan, Childers, and Moore 2000). The scale
was then pretested on undergraduate students in the United
States, and it was reduced to 12 items. The reliability, dimen-
sionality, and validity of the scales were then tested on sam-
ples of alumni of MBA programs in the United States and
Thailand. Although the amount of pretesting was impres-
sive, it was conducted primarily in the United States. It
might have been desirable to allow for some emic input in
Thailand to ensure that some other domain aspect was not
underrepresented. This would seem to be particularly impor-
tant because in contrast to the United States, in Thailand,
there is a strong extended-family orientation, and respect for
authority and obedience are important values. Thus, the
influence of other family members, such as grandparents,
aunts, and uncles, might need to be included; in addition,
this research should have considered respect for senior fam-
ily members in learning and adopting consumer skills.

Because international marketing research is concerned with


UNIT OF ANALYSIS studying behavior in multiple settings or locations and mak-
ing implicit or explicit comparisons about differences and
similarities among locations, another important step is to
define the research setting or context and to select the rele-
vant setting or contexts to be studied. In essence, this implies
the need to define clearly the unit of analysis for the study.

In most international marketing studies, the country is used


Current Practice as the basic unit of analysis in the research design. The coun-
try becomes the spatial unit from which samples are drawn,
from which surveys or experiments are designed, and from
which inferences are made about similarities and differ-
ences. The country as a political and organizational entity
provides a practical and convenient unit for data collection.
Many secondary and industry data are available on a
country-by-country basis. In many cases, countries also pro-
vide linguistic entities or have an official language. In addi-
tion, because countries are customarily used as the unit of
analysis, findings can be related to and interpreted in the
light of previous research. However, although use of the
country to define the geographic boundaries of the research
setting is convenient, it has unintended consequences.

Relevance as a Unit. Use of the country assumes that it pro-


vides the relevant context or socioeconomic setting for the
behavior studied. However, a more appropriate unit may be
cities or regions or a specific linguistic grouping. In some
cases, it is assumed that a country is synonymous with “cul-
ture,” which is then loosely viewed as implying “national

6 Susan P. Douglas and C. Samuel Craig


culture” (Hofstede 1980, 2001). Countries may also not be
the relevant unit for issues related to activities that span
national boundaries. As markets become increasingly inte-
grated at both the product-market and the macroenvironmen-
tal level, studies may examine issues such as the impact of
global media or the Internet on customer purchase and
information-seeking behavior, the emergence of regional or
global lifestyles, and the transfer of information or market
experience across national boundaries. For such studies, the
existence of country boundaries may have little or no rele-
vance, and the relevant spatial unit may be the world or
some other spatial configuration of markets. Recent work in
spatial segmentation has demonstrated that segments can be
identified across national boundaries (Ter Hofstede, Wedel,
and Steenkamp 2002). Locations are spatially dependent,
and this dependency crosses international borders.

Independence. Use of the country as the unit of analysis also


assumes that a country is an isolated or independent unit for
the purpose of the behavior studied. As Galton notes in his
remarks following E.B. Tylor’s presentation of his classic
paper on the cross-cultural method at the Royal Statistical
Society in 1889, it is typically impossible to obtain cross-
cultural sampling units that are independent of one another
(see Naroll 1970). Supposedly, culturally distinctive traits
have often spread among neighboring or historically linked
regions through diffusion or migration. This problem, which
was apparent more than 100 years ago, is even more signifi-
cant in today’s world in which countries are interlinked by
the spread of the Internet, e-mail, satellite media, and mas-
sive waves of migration coupled with increasing regional
economic integration.

Research participants, especially when they are students, are


likely to have traveled to other countries or, at a minimum, to
have been exposed to ideas and influences emanating from
other countries and cultures. Even those who have not trav-
eled are exposed to images and information about other
countries through television or print media. Consequently,
findings related to differences or similarities among coun-
tries (e.g., in relation to consumption or purchasing patterns,
attitudes toward different foreign or global brands) may
reflect factors such as consumer mobility and migration, the
spread of international retailers, and the promotional activi-
ties of international marketers rather than national or coun-
try characteristics.

Comparability of Countries. Countries may also not be com-


parable as units relative to factors that may affect the phe-
nomenon under study (e.g., size of the internal market, the
nature of the market infrastructure, linguistic or cultural
heritage, the degree of interconnectedness with other coun-

Conceptual Foundations of International Marketing Research 7


tries, geographic proximity). Such factors are key elements of
the market context and, as such, play an important role in
fashioning behavior in the marketplace. Consequently, the
comparability of countries with respect to such factors needs
to be carefully considered.

In some cases, a country is selected as an exemplar of a par-


ticular category. When attitudes or behavior are being com-
pared in individualistic versus collectivist societies, the
United States is often selected as an exemplar of an individ-
ualistic, Western society, whereas an Asian society, such as
Singapore, Hong-Kong, Thailand, or Taiwan, is often
selected as an exemplar of a collectivist society. Again, com-
parability with respect to other confounding factors (e.g.,
industrialization, political regime, education) that may
underlie or account for observed differences needs to be
assessed. For example, in examining how economic ideology
and industry type moderate the impact of market orientation
and relationship on performance, Sin and colleagues (2005)
compared Hong Kong and China because they were cultur-
ally similar but different in terms of economic ideology. As a
result, they find clear support for their proposition that eco-
nomic ideology moderates the link among market orienta-
tion, relationship market orientation, and business perform-
ance. Because they had matched countries on cultural
factors, these factors did not contaminate or account for the
observed relationships.

Heterogeneity Within Countries. There is also typically sub-


stantial variation in attitudes and behavior within countries.
In some cases, differences within countries may be greater
than differences between countries. For example, there are
vast regional differences within the People’s Republic of
China. Cui and Liu (2001) identify seven regional markets
with substantial variation in terms of conservatism and pref-
erences for foreign goods. Even in relation to concepts,
which are defined in terms of country or national units (e.g.,
national culture), there may be considerable heterogeneity
within countries. In assessing national identity using the
national identity (NATID) scale, Thelen and Honeycutt
(2004) found substantial differences between Soviet Rus-
sians aged 37–65 and contemporary Russians in various
dimensions of the NATID scale. Soviet Russians were less
concerned with cultural homogeneity, were more ethnocen-
tric, and, counter to the hypothesized attitude, were more
likely to view religion as part of their national identity than
contemporary Russians. This suggests the importance of
examining differences both within and among countries
because there may be considerable intracountry heterogene-
ity. Within-country heterogeneity is particularly relevant for
international market segmentation because it results in seg-

8 Susan P. Douglas and C. Samuel Craig


ments that cross national boundaries (see Steenkamp and Ter
Hofstede 2002; Ter Hofstede, Steenkamp, and Wedel 1999).

The limitations of the country as a unit of analysis have sev-


eral implications for the geographic units used in interna- Refining the Unit of Analysis
tional marketing research. First, it suggests that alternative
geographic units need to be considered, including hierarchi-
cal designs that incorporate multiple levels of units. Second,
when countries are used as a unit in the design, they should
be purposively selected with respect to factors, which may
be relevant or affect the behavior being studied. In addition,
the degree of cultural interpenetration and, thus, the con-
tamination of the unit need to be considered.

Alternative Units. In some international studies, use of more


tightly defined geographic units within countries (e.g., major
cities, urban areas, linguistic groups) is preferable. For exam-
ple, studies of global market segments, such as teens or
upscale consumers, might focus on major cities throughout
the world. However, this approach is subject to the risk that
there may be substantial differences among cities or other
units within countries. Conversely, in other cases, use of
units broader in scope than countries (e.g., regions) may be
the relevant unit. Finally, combinations of multiple levels
(e.g., groups of major cities within regions) might form the
relevant unit in the design.

Purposive Selection. When countries are used as a unit or


level in the research design, it is critical that they be purpo-
sively selected to be comparable, to reflect variance on char-
acteristics of interest, and to control for confounding effects.
This requires careful attention and consideration of
macroenvironmental or country characteristics that may be
related to or affect the behavior studied and the mechanism
by which these characteristics influence behavior.

Sociocultural settings may be purposively selected on the


basis of some key contextual variables that are assumed to
affect variation in the patterning or frequency of the behavior
or attitude studied. For example, in examining differences in
service encounters in fast foods and grocery retailing, Keil-
lor, Hult, and Kandemir (2004) selected eight countries that
represented diverse positions on Hofstede’s five measures of
national culture. This allowed them to generalize their
results to a wide range of different and culturally disparate
types of countries.

Degree of Cultural Interpenetration. Another important issue


is to determine the degree of cultural interpenetration and,
thus, the extent to which the country is truly an isolated or
independent unit (Naroll 1970). Cultural interpenetration

Conceptual Foundations of International Marketing Research 9


can occur in two ways: (1) It may reflect the extent to which
people from a country have traveled to other countries and
thus are familiar with the mores, artifacts, and behavior of
other cultures, and (2) it may reflect the extent to which a
society is exposed to other cultures through mass media, the
Internet, foreign tourism, and other communication links.
For example, when applying the NATID scale in a given
country or examining the degree of consumer ethnocentrism,
researchers should consider and examine the extent to
which respondents have traveled to other countries and have
been exposed to other cultures and whether they watch for-
eign news programs. At a minimum, the degree of exposure
to other countries can be used as a covariate in the analysis.
If the covariate is not significant, greater confidence can be
placed on the obtained scale values. A significant covariate
would indicate that the construct is different for respondents
with external contacts, though this may be a function of self-
selection.

Research instruments that provide accurate and reliable


CONSTRUCT MEASUREMENT measurement tools in multiple research settings need to be
developed. Often these settings differ in terms of language,
levels of literacy, educational systems, and rules of social
interaction. These factors affect respondent comprehension,
suitability of different response formats, differential
response styles, and the nature of the interviewee–
interviewer interaction. Differences in the research infra-
structure also affect the efficacy of different modes of data
collection and administration procedures.

As in the case of conceptual models and constructs, there is


Current Procedures often a tendency, particularly in replication studies, to adopt
research instruments used in the original, or base, study,
appropriately translated when necessary into the language of
the other research context. If the instrument “works” and
exhibits acceptable levels of internal reliability, it is consid-
ered to provide an adequate measure. Typically, little atten-
tion is given to its appropriateness in another setting or to
whether it covers all aspects of the construct to be measured.
In addition, because the instrument exists, little considera-
tion is given to whether an alternative instrument might pro-
vide a more accurate and reliable measure of the construct.

The dangers inherent in this approach are particularly evi-


dent in the case of attitudinal scales. The typical procedure
is to take a scale developed in one sociocultural setting, to
back-translate it, and then to test the translated scale in
another setting on the basis of measures of internal reliabil-
ity, such as Cronbach’s alpha or factor analysis. Items that
exhibit low levels of reliability are eliminated. In some cases,
external measures of validity (e.g., product-related pre-

10 Susan P. Douglas and C. Samuel Craig


dictability) are examined. Rarely is nomological validity (i.e.,
relation to other similar constructs) considered.

Application of a common research instrument in diverse


research settings assumes that the construct is equally salient
and can be measured in the same way in any research setting.
It also assumes equal familiarity with stimuli and research
instruments and the absence of method or item bias across
research settings. Finally, it assumes that a faithful transla-
tion or replication of the research instrument provides the
most effective measure of the construct.

Same Operationalization. As we noted in our discussion of


the conceptual framework, constructs may not be expressed
in the same way in different cultures or research settings,
insofar as the range or type of behavior and attitudes covered
may vary. Thus, use of the same research instrument will not
be appropriate. In addition, a key feature of cross-cultural
research is that it is conducted at multiple sites in multiple
contexts. Consequently, as we noted previously, there may be
factors specific to a given context that affect the hypothe-
sized relationship. Insofar as contexts may differ at both the
macro and/or the micro level, their impact on the attitudes
and behavior studied needs to be considered. For example,
in examining service encounters in a diverse range of coun-
tries, Keillor, Hult, and Kandemir (2004) selected the fast-
food and grocery industries because they assumed that these
would provide relatively similar experiences in terms of
basic want and need fulfillment. However, given the range of
countries examined (i.e., from developing countries, such as
India and Morocco, to highly developed countries, such as
Sweden and the United States, which have sophisticated
retail infrastructures), it seems likely that the context in
which these services were provided may well have differed
and influenced perceptions of the provision of services.
Whereas the impact of the servicescape was examined in
terms of ambient design, functional design, and social ele-
ments, other factors related to the service organization also
could have been considered.

Stimulus Commonality. Use of a common research instru-


ment also assumes equal familiarity with visual and verbal
stimuli (e.g., verbal rating scales, such as Likert scales or
paired comparisons, or visual stimuli, such as photographs)
as well as comparability in visual and verbal stimuli. This
may not always be the case, especially when research is con-
ducted in radically different sociocultural environments (see
Deregowski and Serpell 1971). In addition, colors and color
combinations have been found to have different associations
in different countries (Madden, Hewett, and Roth 2000). This
suggests that different combinations of color can evoke dif-
ferent meanings in different cultures and that a thorough

Conceptual Foundations of International Marketing Research 11


understanding of colors and how color combinations are per-
ceived in each location is needed when the same stimuli are
compared in different countries.

Back Translation. The use of back translation is the best


known and most popular method in educational testing and
psychology (Brislin 1980; Hambleton 1993, 1994). Following
this procedure, a questionnaire is translated from the initial
or source language by a bilingual translator, who is a native
speaker of the target language, into the language in which the
translation is being made. A bilingual who is a native
speaker of that language then translates this version back
into the original or source language. The two versions are
then compared in the source language to check for errors and
the quality of the translation.

Back translation begins with the assumption of a “source”


questionnaire or language. Evaluation of the translation is
made in the source language and assumes that a faithful or
literal translation reflecting the terms and phraseology used
in the source language is required (Werner and Campbell
1970). Consequently, this raises the issue of the extent to
which such structure and terms dominate the questionnaire.
Not only are some phrases or terms nonexistent and difficult
to render in other languages, but in some cases, it may also
be preferable to translate the material into phrases that are
equivalent in intent and meaning.

Method and Item Bias. Use of a research instrument devel-


oped in another cultural setting can generate method bias
due to factors such as differential response style (i.e., social
acquiescence and extreme ratings), differential familiarity
with response formats or stimuli, and interviewer effect (Van
de Vijver and Leung 1997). Bias due to differential response
style or response formats, such as extreme rating styles, is
relatively well documented (Baumgartner and Steenkamp
2001). Statistical procedures can be used to assess the degree
of bias. As we noted previously, this is especially likely to
result in bias when stimuli developed in Western cultures
are applied in Asian or emerging-market cultures.

A typical procedure is to take items developed in a base


country in which research has been originally conducted
(often the United States), translate them into the relevant lan-
guage, and assume that these will tap the relevant construct.
If measurement equivalence is obtained, it is assumed that
the construct is equivalent. However, there is a danger that
this is not the case. For example, in taking data from a study
on the effectiveness of television commercials with high and
low information content in the United States and Korea,
Myers and colleagues (2000) tested measurement equiva-
lence for three measures: attitudes toward the advertisement,

12 Susan P. Douglas and C. Samuel Craig


attitude toward the brand, and buyer intent. Two items were
found to be nonequivalent, which suggests that this could
reflect problematic operationalization. Therefore, the testing
of construct equivalence in this manner is desirable to
ensure reliable and valid measures.

The dangers inherent in using a research instrument devel-


oped in one cultural context or research setting in another Improving Construct
context or setting suggest the desirability of first conducting Measurement
preliminary research to assess the degree of bias associated
with using an existing research instrument and mode of
administration. When examination of the construct or con-
ceptual model suggests that these are expressed in a different
way in different cultural settings or contexts, research instru-
ments that tap specific elements or constructs need to be
developed. When instruments are directly transposed to
another setting, parallel translation helps provide a means to
decenter a research instrument. Equivalent procedures can
be used with visual stimuli. Finally, use of monotrait, multi-
method procedures (triangulation) provides a means to
assess potential method bias and to ensure that instruments
are valid and reliable in each specific research or cultural
context.

Preliminary Research and Pretesting. Preliminary research


and pretesting of research instruments in each cultural con-
text is critical to detect potential sources of bias and to
develop instruments specifically adapted to a given cultural
setting or research environment. Here, qualitative research
can be used to probe potential method bias and to construct
bias associated with a given type of instrument. In-depth
interviews and focus groups can be particularly helpful in
the diagnosis of problems associated with comprehension of
questions and sources of item bias.

In some cases, preliminary research is undertaken to develop


items relevant to a given cultural context (or to assess the
appropriateness of items developed in one context in
another). For example, in a study of new product perform-
ance in Taiwan and South Korea, Song and colleagues (1997)
examined existing research and scales that had previously
been designed to measure the constructs used in the study.
These scales were then refined on the basis of a series of 36
in-depth case studies and focus group studies conducted in
the context of Taiwanese and South Korean new product
development. Team members were asked to elicit and define
salient constructs to assess conceptual equivalence, to deter-
mine whether a list of items tapped these constructs, and to
modify the items where appropriate. A list of constructs and
measurement items was submitted to two panels of academic
experts, and a draft questionnaire was then drawn up and
pretested in both South Korea and Taiwan.

Conceptual Foundations of International Marketing Research 13


Prior experience with a particular type of research instru-
ment (e.g., Likert scaling, semantic differential) in a given
setting can also help provide insights into potential response
bias. Again, instruments should be pretested, and multiple
instruments or triangulation (Hui and Triandis 1983; Trian-
dis, McClusker, and Hui 1990) should be used to identify
potential bias and to develop improved and more reliable
instruments.

Use of Emic/Etic Instruments. When preliminary research


suggests that certain constructs are not salient or are
expressed in different ways in different cultures, use of a
supposedly etic instrument will result in bias. If culture-
specific elements of a conceptual model or constructs are
identified, development of a hybrid emic/etic instrument or
a purely emic or culture-specific instrument is necessary. In
the former case, the core elements of a scale developed for a
single country or context may be used to form the etic com-
ponent of the scale. Scales developed in a single market or
cultural context often include domains or aspects of a con-
struct that are noncentral and context specific. Therefore,
extraction of core items improves predictive ability (Baum-
gartner and Steenkamp 2001). In other cases, culture-specific
instruments may be developed. In examining brand person-
ality in Japan and Spain, Aaker, Benet-Martínez, and Garol-
era (2001) developed an instrument that incorporates meas-
ures specific to each market, replicating analogous measures
in the United States. Some elements were found to be the
same, but others differed. In the case of attitudinal scales,
when a construct is found to have different dimensions in
different cultures and contexts, emic or culture-specific
items that tap culture-specific dimensions should be identi-
fied. These can then be included in the item pool and meas-
urement instrument, resulting in a hybrid emic/etic measure.

Use of Parallel Translation. Use of parallel or committee


translation is increasingly recommended in educational psy-
chology and other social sciences (Hambleton 1993, 1994). A
questionnaire is first forward- and/or back-translated. A
committee of bilingual translators and experts then compare
versions of the questionnaire to check for errors of meaning.
The adequacy of alternative translations is discussed and
equivalence of meaning is examined until agreement is
reached on a final version. This technique has been criti-
cized on the grounds that there is no formal control over the
accuracy of the translation. It appears to be a useful approach
in international marketing research, in which a freer and less
exact translation is often desirable (Bontempo 1993; Hamble-
ton 1993, 1994). It also has the advantage of using people of
different backgrounds and perspectives in a co-operative
effort to improve the accuracy of the translation and assess
its accuracy. Often, this approach is useful in decentering the

14 Susan P. Douglas and C. Samuel Craig


research instrument and in developing one that is better
adapted to a specific culture (Werner and Campbell 1970).

Two alternative iterative approaches can be identified to


arrive at a solution that provides comparability without A FRAMEWORK FOR CROSS-
ignoring emic elements. These differ in terms of the starting CULTURAL STUDIES
or base point. The first, “the adapted etic approach,” begins
with a base culture or context and attempts to adapt the con-
ceptual model developed in the base culture to other con-
texts, taking explicitly into account differences in the con-
text being studied and their implications for the research
design. The second, “the linked emic model,” takes the local
context as its starting point, but at multiple research sites.
Input from each site is then incorporated into the overall
conceptual framework and research design.

The adapted etic model begins with the assumption that the
theory or conceptual framework used in the research design The Adapted Etic Model
is pancultural—that is, it applies in all contexts or settings—
but that some adaptation to specific local research context
may be necessary. Typically, research has been conducted
and validated in a single research context or setting, and the
researcher attempts to examine the extent to which the
model and related research findings can be generalized to
other research contexts and settings.

An important first step is to explicate the underlying theory


and conceptual framework in multiple sociocultural settings.
This entails an examination of whether the assumptions that
underlie the theory apply and are relevant in other sociocul-
tural settings. For example, theories and research related to
consumerism or consumer satisfaction may have little rele-
vance in emerging or transitional economies, in which a con-
sumerist society has not yet emerged and problems related to
the supply or availability of goods are paramount.

When the relevance of a theory or conceptual framework to a


given sociocultural setting has been assessed, the next step is
to examine the relevant constructs and hypotheses to deter-
mine whether they are equally valid and relevant in all set-
tings. This typically requires obtaining input either directly
by conducting research in other settings or indirectly by
using other researchers’ experience in a given setting. Local
researchers with expertise in the area can be asked to adapt
constructs to the local research context or identify culture-
specific concepts related to the topic studied. For example,
in developing his universal values, Schwartz (1992) asked
collaborators to examine the source list and identify any
other values that might be relevant in their country/culture.

An alternative approach that Hui and Triandis (1983) suggest


is to build alternative measures of concepts explicitly into

Conceptual Foundations of International Marketing Research 15


the research design. This approach begins with a construct
that is assumed to be etic. Different ways of operationalizing
or measuring the construct in different cultures are then
developed. Comparisons based on emic measurements of the
etic construct in different contexts are conducted to develop
a derived etic construct. This is an iterative, divergent
approach that begins with an etic-based conceptual frame-
work and gradually moderates the constructs and relation-
ships in the framework on the basis of successive studies that
examine modifications in different contexts. The wider the
range of contexts studied, the more likely it is that modifica-
tions will be made in the original conceptual model.

The adapted etic approach is likely to result in an emphasis


on similarities rather than on differences. Even when modifi-
cations are made to the original etic construct to fit the spe-
cific context, the emphasis is on retaining the essence of that
construct and the underlying assumption that it is pancul-
tural or universal. The cultural perspective of the base coun-
try in which research was first conducted is likely to domi-
nate. Therefore, little attention is likely to be paid to
identifying emic-specific elements or constructs. This
approach is likely to result in an emphasis on similarities
rather than on differences and to facilitate comparison across
countries and contexts. Modifications are made for differ-
ences in the context or setting, but explicit identification and
in-depth examination of emic-specific constructs are less
likely. In particular, the perspective of the researcher from
the base country is likely to prevail.

To incorporate culture-specific elements and effectively


The Linked Emic Model build a broader knowledge base, a hybrid, or “linked,” emic
model is suggested. Emic research initially begins in multi-
ple sites. A group of researchers, including at least one from
each of the contexts being studied, begin by agreeing on the
scope of common parameters of the research and the key
research questions. This first step may create some difficulty
insofar as each researcher may view the key research ques-
tions differently. Definitions or understanding the meanings
of terms may be particularly problematic. In many cases,
English is the common language in which the research is dis-
cussed and articulated. However, translation and interpreta-
tion of the meaning associated with key terms such as lead-
ership, trust, and so on, in different languages may already
give rise to issues of comparability and equivalence, which
need to be discussed and resolved.

For example, in the early stages of an international research


study, qualitative research should be undertaken to assess
the relevance of the theories and constructs initially hypoth-
esized to be related to the research questions being studied
and to identify how these constructs are expressed and

16 Susan P. Douglas and C. Samuel Craig


whether any emic constructs relevant to the study need to be
considered. This should include a literature search not only
in the base language and country in which the study was ini-
tially conceived but also in all other countries and contexts
that the study covers. It is particularly important to conduct
this research in the language of the other cultural settings
because this may reveal alternative theories or constructs
that are salient to the research issue in that country or con-
text and also may help broaden the overall perspective of the
study.

After consensus has been reached on the key research ques-


tions under investigation, each researcher develops a
research design. This sets up a conceptual framework for the
study that considers characteristics specific to each research
context. This should also articulate how each element of the
model (i.e., constructs and concepts) will be operationalized
or, more specifically, how it will be defined and construed in
each local context.

At this point, researchers from all contexts discuss differ-


ences and similarities in these localized models and the role
of contextual factors in accounting for these differences and
similarities. This discussion may result in the identification
of (1) elements, concepts, and relationships that appear to be
common across contexts; (2) concepts that are common but
are operationalized in different ways in different contexts;
(3) relationships among constructs that differ across con-
texts; and (4) constructs that are unique to a given context.
Further discussion is then aimed at moving toward greater
harmonization of conceptual models across contexts. This
may be achieved by including elements that are unique to
one context in other contexts or by amalgamating or combin-
ing constructs from different contexts to develop multifac-
eted constructs that are then applied to all contexts. When it
is feasible, a overarching model may be developed that incor-
porates all the elements from the local, context-specific mod-
els and delineates both etic and emic elements. Some theo-
retical rationale related to the impact of contextual factors
should also be developed to account for the emic compo-
nents and relationships within the model. For example,
group-related variables may play a more critical role in
explaining behavior in collectivist societies than in individu-
alistic societies.

This type of emic approach begins with an emphasis on local


knowledge and a local perspective. As a result, it tends to
emphasize the unique features that characterize each context
rather than pancultural or etic aspects; in other words, the
emic approach interprets these elements within the local
context. This approach avoids potential ethnocentric or
pseudoetic bias, insofar as researchers from each local con-

Conceptual Foundations of International Marketing Research 17


text are responsible for building the research design and
interpreting findings from each local context. Coordination
at subsequent phases of research may also be facilitated
when coordination and integration of findings across
research contexts are conducted by a multicultural team of
researchers from different backgrounds, each of whom con-
tributes local and general knowledge.

Adopting either an adapted etic or a linked emic approach to


MANAGERIAL IMPLICATIONS conducting international marketing research has several
important implications for management. When conducting
international marketing research, it is important to involve
local management in the research effort. If research is
designed and executed centrally, the purpose and implica-
tions of the research may not be clear to local management,
resulting in misunderstanding, miscommunication, and a
failure to act on the research findings. The involvement of
local management also helps ensure that models and con-
cepts are appropriately adapted to the local research envi-
ronment and that emic, or country-specific elements, are
included in the research design.

In addition, research needs to be initiated in the local coun-


try context both to identify country-specific concepts and
elements and to probe country-specific problems and issues.
Again, local management is best placed to identify such
issues and should have the capability of initiating and com-
missioning such research. Even when management is inter-
ested in developing a global strategy, it may still be impor-
tant to account for local differences.

Ultimately, the appropriate approach for conducting interna-


tional marketing research depends to a large degree on the
organizational structure of the company. A highly central-
ized company in which corporate headquarters exercises
strong control over local operations will find that an adapted
etic approach is best suited to its organizational structure.
Conversely, a decentralized company in which local country
management plays an important role is likely to find a linked
etic approach more consistent with its decision-making
structure.

Regardless of the approach adopted in conducting interna-


tional marketing research, it is highly desirable to develop
teams that are composed of researchers from different coun-
tries and different cultural backgrounds. In particular, capa-
bility in multiple languages is essential to understand the
difficulties of translation and to eliminate potential bias that
results from a single cultural background. Although such
multicultural research teams are difficult to organize, they
are an important prerequisite to the generation of sound
international marketing research.

18 Susan P. Douglas and C. Samuel Craig


Finally, research that is conceptually well grounded pro-
vides sounder information for decision making. Therefore, it
is important to pay careful attention to the theoretical under-
pinnings of research and to ensure that these are relevant to
the problem and context at hand. Sound research that can
effectively guide management decision making will be gener-
ated only if there is a strong conceptual framework.

International marketing research presents several unique


design and methodological issues. With few exceptions, CONCLUSIONS
these have received relatively little attention in the market-
ing literature. Whereas methodological issues in the analysis
of multicountry data have been extensively discussed, issues
related to overall research design and, specifically, the rele-
vance of theories and equivalence of constructs and concepts
in multicountry research have been largely neglected.

An underlying theme of this article has been the importance


of decentering the research perspective, that is, removing the
influence of a dominant culture or philosophy in the devel-
opment and execution of the research design. Although this
issue has been extensively discussed in relation to transla-
tion (Werner and Campbell 1970), the effects are less well
recognized in relation to theories and constructs and in the
design of the conceptual framework for international market-
ing studies. All too frequently, theories or constructs devel-
oped in relation to one sociocultural context or setting are
used in the study of another, which can result in bias in the
definition and measurement of constructs and in subsequent
stages of research. No matter how accurate and refined the
analytic procedures, if the underlying conceptual model is
flawed, the findings and their interpretation will, at best, be
biased.

If future research in international marketing is to provide


valid results and contribute to the advancement of knowl-
edge, greater attention must be paid to the early stages of
research design and, in particular, to the relevance and
salience of the underlying conceptual framework in each
research setting or context. Only then will data be generated
from which valid and meaningful inferences can be drawn
that can guide management decision making.

1. According to Van de Vijver and Leung (1997), equivalence


can be viewed as distinct from bias. Equivalence is the NOTE
level at which a comparison is made (i.e., the equivalence
of a construct across contexts or cultures or of the meas-
urement units, the scale, or the score). In contrast, bias is
any factor that jeopardizes the validity of the comparison.

Conceptual Foundations of International Marketing Research 19


Aaker, Jennifer, Veronica Benet-Martínez, and Jordi Garolera
REFERENCES (2001), “Consumption Symbols as Carriers of Culture: A Study of
Japanese and Spanish Brand Personality Constructs,” Journal of
Personality and Social Psychology, 81 (3), 492–508.
Andaleeb, Syed Saad and Syed Ferhat Anwar (1996), “Factors
Influencing Customer Trust in Salespersons in a Developing
Country,” Journal of International Marketing, 4 (4), 35–52.
Armstrong, Robert W. and Siew Min Yee (2001), “Do Chinese Trust
Chinese? A Study of Chinese Buyers and Sellers in Malaysia,”
Journal of International Marketing, 9 (3), 63–86.
Baumgartner, Hans and Jan-Benedict E.M. Steenkamp (2001),
“Response Styles in Marketing Research: A Cross-National Inves-
tigation,” Journal of Marketing Research, 38 (May), 143–56.
Berry, J.W. (1969), “On Cross-Cultural Comparability,” Interna-
tional Journal of Psychology, 4 (2), 119–28.
Bontempo, R. (1993), “Translation Fidelity of Psychological
Scales,” Journal of Cross-Cultural Psychology, 24 (2), 149–66.
Brislin, Robert W. (1980), “Translation and Content Analyses of
Oral and Written Material,” in Handbook of Cross-Cultural Psy-
chology, Vol. 1, H.C. Triandis and J.W. Berry, eds. Boston: Allyn
& Bacon, 389–444.
Craig, C. Samuel and Susan P. Douglas (2005), International Mar-
keting Research, 3d ed. London: John Wiley & Sons.
Cui, Geng and Qiming Liu (2001), “Emerging Market Segments in a
Transitional Economy: A Study of Urban Consumers in China,”
Journal of International Marketing, 9 (1), 84–106.
Deregowski, J.B. and Robert Serpell (1971), “Performance on a
Sorting Task: A Cross-Cultural Experiment,” International Jour-
nal of Psychology, 6, 273–81.
Hambleton, R.K. (1993), “Translating Achievement Tests for Use in
Cross-National Studies,” European Journal of Psychological
Assessment, 9, 57–68.
——— (1994), “Guidelines for Adapting Educational and Psycho-
logical Tests: A Progress Report,” European Journal of Psycho-
logical Assessment, 10, 229–44.
Hofstede, Geert (1980), Culture’s Consequences: International Dif-
ferences in Work-Related Values. Beverly Hills, CA: Sage
Publications.
——— (2001), Culture’s Consequences: Comparing Values, Behav-
iors, Institutions and Organizations Across Cultures. Thousand
Oaks, CA: Sage Publications.
Hui, C. Harry and Harry C. Triandis (1983), “Multistrategy
Approach to Cross-Cultural Research: The Case of Focus of Con-
trol,” Journal of Cross-Cultural Psychology, 14 (March), 65–83.
Keillor, Bruce D., G. Tomas M. Hult, and Destan Kandemir (2004),
“A Study of Service Encounters in Eight Countries,” Journal of
International Marketing, 12 (1), 9–35.
Klein, Jill G., Richard Ettenson, and Marlene D. Morris (1998),
“The Animosity Model of Foreign Product Purchase: An Empiri-
cal Test in the People’s Republic of China,” Journal of Marketing,
62 (January), 89–100.

20 Susan P. Douglas and C. Samuel Craig


Kuada, John and Seth N. Buatsi (2005), “Market Orientation and
Management Practices in Ghanaian Firms: Revisiting the THE AUTHORS
Jaworski and Kohli Framework,” Journal of International Mar-
keting, 13 (1), 58–88. Susan P. Douglas is Paganelli-
Lee, J.A. (1966), “Cultural Analysis in Overseas Operations,” Har- Bull Professor of Marketing and
vard Business Review, 44 (March–April), 106–111. International Business (e-mail:
Madden, Thomas J., Kelly Hewett, and Martin S. Roth (2000), sdouglas@stern.nyu.edu), and
“Managing Images in Different Cultures: A Cross-National Study C. Samuel Craig is Catherine
of Color Meanings and Preferences,” Journal of International and Peter Kellner Professor and
Marketing, 8 (4), 90–107. Director of the Entertainment,
Mullen, Michael R. (1995), “Diagnosing Measurement Equivalence Media and Technology Program
in Cross-National Research,” Journal of International Business (e-mail: scraig@stern.nyu.edu),
Studies, 26 (3), 573–96. Stern School of Business, New
Myers, Matthew B., Roger J. Calantone, Thomas J. Page Jr., and York University.
Charles R. Taylor (2000), “An Application of Multiple-Group
Causal Models in Assessing Cross-Cultural Measurement Equiv-
alence,” Journal of International Marketing, 8 (4), 108–121.
Nakata, Cheryl and Yili Huang (2002), “Progress and Promise: The
Last Decade of International Marketing Research,” AMA Educa-
tors’ Proceedings: Enhancing Knowledge Development in Mar-
keting, Vol. 13, William J. Kehoe and John H. Lindgren Jr., eds.
Chicago: American Marketing Association, 306–307.
——— and K. Sivakumar (2001), “Instituting the Marketing Con-
cept in a Multinational Setting: The Role of National Culture,”
Journal of the Academy of Marketing Science, 29 (3), 255–75.
Naroll, Raoul (1970), “The Culture-Bearing Unit in Cross-Cultural
Surveys,” in The Handbook of Method in Cultural Anthropology,
Raoul Naroll and R. Cohen, eds. New York: Natural History
Press, 721–65.
Nijssen, Edwin J. and Susan P. Douglas (2004), “Examining the
Animosity Model in a Country with a High Level of Foreign
Trade,” International Journal of Research in Marketing, 21 (1),
23–38.
Powpaka, Samart (1998), “Factors Affecting the Adoption of Mar-
ket Orientation: The Case of Thailand,” Journal of International
Marketing, 6 (1), 33–55.
Schwartz, S.H. (1992), “Universals in the Content and Structure of
Values: Theoretical Advances and Empirical Tests in 20 Coun-
tries,” in Advances in Experimental Social Psychology, Vol. 25,
M. Zanna, ed. Orlando, FL: Academic Press, 1–65.
Sekaran, Uma (1983), “Methodological and Theoretical Issues and
Advancements in Cross-Cultural Research,” Journal of Interna-
tional Business Studies, 15 (Fall), 61–73.
Sin, Leo Y.M., Alan C.B. Tse, Oliver H.M. Yau, Raymond P.M.
Chow, and Jenny S.Y. Lee (2005), “Market Orientation, Relation-
ship Marketing Orientation, and Business Performance: The
Moderating Effects of Economic Ideology and Industry Type,”
Journal of International Marketing, 13 (1), 36–57.
Song, X. Michael, Mitzi M. Montoya-Weiss, and Jeffrey B. Schmidt
(1997), “The Role of Marketing in Developing Successful New
Products in South Korea and Taiwan,” Journal of International
Marketing, 5 (3), 47–70.

Conceptual Foundations of International Marketing Research 21


Steenkamp, Jan-Benedict E.M. and Hans Baumgartner (1998),
“Assessing Measurement Invariance in Cross-National Con-
sumer Research,” Journal of Consumer Research, 25 (June), 78–
90.
——— and Frenkel Ter Hofstede (2002), “International Market Seg-
mentation: Issues and Outlook,” International Journal of
Research in Marketing, 19 (3), 185–213.
Ter Hofstede, Frenkel, Jan-Benedict E.M. Steenkamp, and Michel
Wedel (1999), “International Market Segmentation Based on
Consumer-Product Relations,” Journal of Marketing Research, 36
(February), 1–17.
———, Michel Wedel, and Jan-Benedict E.M. Steenkamp (2002),
“Identifying Spatial Segments in International Markets,” Market-
ing Science, 21 (Spring), 160–77.
Thelen, Shawn T. and Earl D. Honeycutt Jr. (2004), “Assessing
National Identity in Russia Between Generations Using the
National Identity Scale,” Journal of International Marketing, 12
(2), 58–81.
Triandis, Harry C. (1972), The Analysis of Subjective Culture. New
York: John Wiley & Sons.
———, Christopher McClusker, and C. Harry Hui (1990), “Multi-
method Probes of Individualism and Collectivism,” Journal of
Personality and Social Psychology, 59 (November), 1006–1020.
Van de Vijver, Fons and Kwok Leung (1997), Methods and Data
Analysis for Cross-Cultural Research. Thousand Oaks, CA: Sage
Publications.
Van Raaij, W. Fred (1978), “Cross-Cultural Research Methodology
as a Case of Construct Validity,” in Advances in Consumer
Research, Vol. 5, Keith Hunt, ed. Ann Arbor, MI: Association of
Consumer Research, 693–701.
Viswanathan, Madhubalan, Terry L. Childers, and Elizabeth S.
Moore (2000), “The Measurement of Intergenerational Commu-
nication and Influence on Consumption: Development, Valida-
tion and Cross-Cultural Comparison of the IGEN Scale,” Journal
of the Academy of Marketing Science, 28 (3), 406–424.
Werner, O. and D. Campbell (1970), “Translating, Working Through
Interpreters and the Problem of Decentering,” in A Handbook of
Method in Cultural Anthropology, R. Naroll and R. Cohen, eds.
New York: American Museum of Natural History, 398–420.

22 Susan P. Douglas and C. Samuel Craig

You might also like