Anda di halaman 1dari 20

J BUSN RES 191

1993:28:191-210

Measuring a Multi-Dimensional Construct:


Country Image

Ingrid M. Martin
University of Southern California

Sevgin Eroglu
Georgia State University

This study presents the procedures and results obtained in developing a scale to
measure the multi-dimensional construct of country image. This effort intends to
serve two purposes. First, it is likely to explicate many aspects of how product
evaluations are affected by country image and, therefore, has theoretical as well
as managerial implications. Second, a valid operational measure may help resolve
some of the methodological and conceptual issues raised in the area of country of
origin effects research. Tests of internal reliability and validity were conducted
across different countries and samples to assess the strength of the final 14-item
scale. The results imply that constructs used in international marketing research
are scalable and that they can meet precise measurement criteria.

Introduction
The growing literature on country image and country of origin effects to date has
indicated that industrial and consumer buyers develop stereotypical images of coun-
tries and/or their products, which subsequently affect their purchase decisions (see
Baughn and Yaprak, 1991, for a review). More specifically, a recent study by Han
(1989) identified two major functions for country image effects. First, buyers can
use country image in product evaluations when they are unable to detect the true
quality of a country’s products before purchase (halo function). As such, country
image indirectly affects brand attitudes through inferential beliefs. Second, as buy-
ers become more familiar with a country’s products, country image may help them
summarize their product beliefs and directly affect their brand attitudes (summary
function). In this capacity, the country image is found to stimulate buyers to think
more extensively about other product information as well (Hong and Wyer, 1989).
The above findings have important practical and theoretical implications. From
a managerial perspective, international marketers and public policy makers alike

Address correspondence to Ingrid M. Martin, Department of Marketing, Act-301, School of Business Adminis-
tration, University of Southern California, Los Angeles, CA 90089-1421.

Journal of Business Research 28, 191-210 (1993) 0148-2963/93/$6.OCl


0 1993 Elsevier Science Publishing Co., Inc.
655 Avenue of the Americas, New York, NY 10010
192 J BUSN RES
I. M. Martin and S. Eroglu
1993:28:191-210

need to understand country images within the context of their own offerings and
those of their competition. Research has shown, for example, that consumers’
negative product evaluations based on country images constitute significant market
barriers for companies from less developed countries (Schooler et al., 1987). The
image that consumers hold of South Africa where major boycotts have managed
to shut down the market for certain goods and services serves as an extreme example
of this.
Similarly, Johansson and Nebenzahl (1984) have found that multi-national firms
with foreign manufacturing operations may risk potential loss in brand image de-
pending on the country image of the sourcing country. Factors affecting country
image also are instrumental for the positioning strategies of firms that compete in
their domestic markets against foreign competitors (Hooley et al., 1988). In sum,
international managers need to assess the extent to which relevant country images
are favorable or unfavorable, if and how they affect product quality perceptions
and purchase decisions, and how they can be used to develop effective marketing
strategies.
From a theoretical standpoint researchers in the area of country image effects
have become increasingly sensitive to its theoretical and methodological dimensions
(e.g., Bilkey and Nes, 1982; Jaffe and Nebenzahl, 1984; Parameswaran and Yaprak,
1987). In their review of the country of origin literature, Baughn and Yaprak (1991)
conclude that these studies have contributed to international marketing research
by increasing concerns about the psychometric properties of cross-national mea-
sures such as country image in the following ways.
Presently, the measures used in the country of origin research stream are subject
to the same criticisms that are directed at research in cross-cultural consumer
behavior; namely, the shortage of valid and reliable measuring instruments (Davis
et al., 1981). Our literature survey to date indicates that there is no validated scale
for measuring country image per se. Furthermore, those that are currently being
used to tap country image effects seem questionable for two reasons.
First, from a conceptual perspective most of the scales presently used do not
clearly distinguish between the image objects; that is, whether it is country image
or product image that is being measured. The widely used Nagashima (1970, 1977)
scale is a case in point. The scale that is designed to measure the image of products
with a foreign country of origin includes items that also may capture country image.
A valid scale, however, requires a precise delineation of the construct’s domain.
If product attitudes are of interest, then the final scale should reflect measurement
of product-specific attributes (e.g., reliable/unreliable, expensive/inexpensive). If,
on the other hand, country image is being measured, the scale items should capture
country-relevant attributes (c.g., technically advanced/technically backward, cos-
mopolitan/ethnocentric). An accurate scale of country image needs to clearly spec-
ify the construct’s domain and to be exact concerning what is included as well as
what is excluded from the definition.
The second issue concerns the low reliability ratings of the existing scales used
in country image studies. Several researchers reported poor reliabilities in their
efforts to validate some of the popular scales used in country of origin research
(e.g., Narayana, 1981; Cattin et al., 1982). Their findings were supported by Jaffe
and Nebenzahl (1984) who concluded that existing image scales not only have low
reliability but also are not tested for internal consistency and stability.
J BUSN RES 193
Measuring a Multi-Dimensional Construct 1993:28:191-210

These measurement concerns present several problems for researchers. In the


event that conflicting measures are obtained, as pointed out by Bilkey and Nes
(1982) in their review of the country of origin research, the researcher is not sure
if the discrepancy is due to actual differences in country or product image or to
different measures. As yet, there is no validated instrument available to assess
country image without tapping into the image of products from the respective
country. The basic tenet of the country of origin research is that one’s country
image is reflected in higher evaluations of products originating in that country. It
is difficult to assess the extent and nature of country image impact on product
evaluations without an accurate instrument to measure it.
The objective of the present study is to describe the development and evaluation
of a multiple-item country image scale to help fill the need for better measures in
this area. In sum, there are two distinct reasons for examining and developing a
measure for country image. First, this effort is likely to explicate many aspects of
how product attributes and evaluations are affected by the images that consumers
have about their country of origin. This should have an impact on both the man-
agerial and theoretical issues discussed previously. Second, a valid operational
measure may help resolve some of the methodological and conceptual issues raised
in the area of country image research.
In the following pages, we first define the country image construct and identify
its relevant dimensions from a review of the literature. Next, we discuss the pro-
cedure used to generate scale items and establish the content validity of the con-
struct. This is followed by a discussion on assessing internal consistency and item
stability as well as the validation techniques used to fine tune the scale. Finally, a
discussion is presented on the relevancy of the final scale to both managers and
academicians.

Methodology

Defining the Construct and Identifying Dimensions

The critical first step in the development of the scale is to specify the domain of
the construct of country image. This involves a comprehensive review of the lit-
erature related to country of origin as well as country image studies (e.g., Baughn
and Yaprak, 1991; Bilkey and Nes, 1982; Cattin et al., 1982; Dornoff et al., 1974;
Gaedeke, 1973; Hafhill, 1980; Han, 1989; Han and Terpstra, 1988; Hong and Wyer,
1989; Johansson et al., 1985; Johansson and Moinpour, 1977; Nagashima, 1970;
1977; Narayana, 1981; Parameswaran and Yaprak, 1987).
The review also included the literature on scale development and related issues
in both the psychology and marketing methodology areas (e.g., Campbell and
Fiske, 1959; Churchill, 1979; Frazier, 1983; Jaffe and Nebenzahl, 1984; Malhotra,
1981; Nagashima, 1970; 1977; Nunnally, 1978; Osgood et al., 19.57; Peter, 1979;
1981; Peter and Churchill, 1986; Wee, 1986; Wish et al., 1972; Zinkhan and Mu-
derrisoglu, 1982). On the basis of the literature in these areas as well as discussions
undertaken with international faculty and students, a conceptual definition was
developed for the domain of the country image construct. Accordingly, country
image was defined as the total of all descriptive, inferential and informational beliefs
one has about a particular country.
194 J BUSN RES
1993:2X:191-210
I. M. Martin and S. Eroglu

Again, it should be noted that this is completely different from one’s attitudes
toward products from a certain country. One’s country image can develop as a
result of a direct experience with the country, such as traveling to the country.
Alternatively, it can be influenced by outside sources of information, such as ad-
vertising or word of mouth communications. Last, it could be affected by inferences
(correct or incorrect) based on past experience such as opinions gained from using
products originating in that particular country.
In order to operationalize the definition, it was necessary to determine the
relevant underlying dimensions of this construct. Country image has been consis-
tently identified as a multi-dimensional concept (e.g., Jaffe and Nebenzahl, 1984;
Johansson and Moinpour, 1977; Han, 1989). Several analytic and multi-dimensional
scaling studies have indirectly implied some dimensions of country image across
various countries (Johansson and Moinpour, 1977; Johansson et al., 1985; Wish et
al., 1972).
Along with the above studies, an interdisciplinary review of the literature (e.g.,
international business, political sciences, economics, sociology) resulted in the iden-
tification of four relevant dimensions. These four dimensions used to define the
construct’s domain are (1) political, (2) economic, (3) technological, and (4) social
desirability. The first three dimensions are self-explanatory and the fourth dimen-
sion, social desirability, includes such factors as quality of life, standard of living,
and level of urbanization. Interestingly, the literature did not indicate culture or
cultural familiarity as an underlying dimension of the country image construct.
The format selected to measure country image in this study is the semantic
differential scale. The theoretical rationale for using the semantic differential scale
has been detailed by Osgood et al. (1957) as well as Churchill (1979). The most
commonly used device for measuring images of such concepts as brands or products,
stores, political candidates, countries, institutions, or ideas is the semantic differ-
ential scale. Using this scale device the researcher can measure, assess, and compare
the image of a concept or object with that of similar topics (Alreck and Settle,
1985). In addition, the semantic differential has enjoyed a great acceptance and
relevance in marketing research that is unmatched by any other scaling procedure
(Malhotra, 1981).
A two-phase procedure was used to generate the initial item pool. First, a short
questionnaire was developed to derive a pool of phrases and adjectives that describe
the image one has of any country. This instrument was administered to students
and faculty on two midwestern university campuses. The sample provided adjec-
tives, through free association, that were then combined with the second group
based on their usage frequency.
The second phase involved a focus group with eight doctoral students from
various international backgrounds. During the 2-hour session the participants dis-
cussed their beliefs and impressions of various countries ranging from Nigeria to
France to Korea to Yugoslavia. Once both sets of items were pooled together, all
duplicated items as well as items that were country-specific were deleted. The result
was a total of 60 bipolar word pairs.
The next step involved reducing the number of items to a usable subset with
the aid of five expert judges (three doctoral students and two faculty members in
international-related fields). The judges were provided with the definition of coun-
try image and were instructed to evaluate the 60 bipolar word pairs in terms of their
Measuring a Multi-Dimensional Construct .I BUSN RES 195
1993:28:191-210

compatibility with the given definition as well as making suggestions for better word-
ing of the items. An example of the instruction form is included in Appendix A.
The judges were asked to rate the bipolar word pairs using the following scale:
(1) clearly representative of a country’s image, (2) somewhat representative of a
country’s image, and (3) not at all representative of a country’s image. This pro-
cedure involved an independent evaluation by the five judges. Average interjudge
agreement and reliability was obtained for 29 of the 60 bipolar word pairs. The
Holsti procedure (1969) was used to determine an average interjudge agreement
of .848 and a reliability of .965. These 29 items were combined to form the initial
scale that was then tested (see Appendix B for the initial 29-item scale).

Purifying the Scale Items


The determination of how many items to have in a multi-dimensional scale is a
difficult problem. On the one hand, a multi-dimensional construct can be factorially
complex enough to require more than a single-item component to accurately capture
the concept. On the other hand, if too many items are included in the questionnaire,
then problems of fatigue and boredom may arise. Other scales in the literature
(e.g., Malhotra, 1981; Lundstrom and Lamont, 1976; Nagashima, 1970; 1977;
Zaichkowsky, 1985) were found to be shorter than the 29-item original scale.
Therefore, it was decided that as we continued to purify the scale we also would
allow for the deletion of scale items, if deemed necessary.
The reduction of the number of items in the scale was done through a commonly
used procedure (e.g., Lundstrom and Lamont, 1976; Malhotra, 1981; Wee, 1986)
that required further data collection using the 29-item scale. The sample used for
this purpose included 200 undergraduate and graduate students in the Colleges of
Business Administration and Communication Arts and Sciences on a midwestern
university campus. It is considered appropriate and common to use students as
subjects in scale developments in both marketing and psychology (e.g., Malhotra,
1981; Osgood et al., 1957; Zaichkowsky, 1985). A pre-test was conducted to de-
termine which country would be best to use in the initial test of the scale. The
country of Japan was found to be a good candidate because of its high familiarity
among the subjects.
The data were analyzed for internal consistency. Given the multi-dimensional
nature of the scale, though, it is not very meaningful to estimate an overall measure
of internal consistency (Peter, 1979). Therefore, first the coefficient alpha was
computed for each subset of scale items that made up a certain factor. The result
was a 21-item scale with item-total correlations ranging between .23 and .58. In-
ternal consistency for the scale was then tested by checking the reliability coefficient
alphas, which ranged from .77 to .81. These values were deemed reasonable when
compared with others obtained in the marketing studies previously mentioned as
well as the evaluation guidelines suggested by Peter (1979). Nunnally (1978) argues
that increasing reliability beyond .8 is unnecessary because at that level correlations
are attenuated very little by measurement error.
Principal component analysis was used to determine which variables had high
intraset correlations and low interset correlations for the country of Japan. This
methodology is common for exploratory scale development and it uncovered three
of the four factors that had been identified in the literature. Principal component
196 J BUSN RES I. M. Martin and S. Eroglu
1993:28:191-210

Table 1. Correlations Between Social Desirability and the Other Three Factors of
Country Image
Political Economic Technological

Level of economic development ,344 .I3 ._51


(.Wl) (.065) (.0001)

Level of per capita income ,581 .49 .32


(.ml) (.OWl) (.0001)

Size of middle class .20 .18 .22


(.0058) (.OllO) (.002)

Existence of a welfare system .18 .Sl .04


(.0116) (.WOl) (.614)

Level of unemployment .Ol .02 .15


(.93) (.83) (.036)

Note; Numbers in parentheses are probability values for the respective correlations.

analysis with varimax rotation yielded three distinct and independent factors: (1)
political, (2) economic, and (3) technological. This provided supportive evidence
of construct validity for the scale.
The social desirability component did not emerge as a significant factor. The
literature postulates that this factor is described by such items as quality of life,
standard of living, and level of urbanization. It was hypothesized that such items
could have been captured in the three other factors of economic, political, and
technological aspects. For example, it could be argued that some of the items that
fall in the economic dimension (e.g., level of standard of living, stability of economic
environment, level of labor costs) all are very similar to the items in the social
desirability component. Similarly, level of urbanization very well may have been
captured in the technological dimension with such items as level of industrialization
and level of technological research.
Based on these findings and reasoning, we decided to conduct another set of
tests to determine if social desirability was truly captured by the three factors. The
next step was to identify the items in the scale that could be components of the
social desirability factor. These items were then correlated with the three factors
both as an individual factor of social desirability and as individual components of
the hypothesized factor. The five items identified included level of economic de-
velopment, level of per capita income, size of middle class, existence of a welfare
system, and level of unemployment.
First, the five items were individually correlated with the three factors and the
results can be seen in Table 1. The items with the highest correlations to one or
more of the factors included level of economic development, level of per capita
income, size of middle class, and existence of a welfare system. The last item, level
of unemployment, correlated only moderately with any of the technological factors
(r = .15, p < .04).
The second step in the analysis involved regressing each of the social desirability
items on each factor. The result was an adjusted R2 of .40, .44, and .34 for the
political, economic, and technological factors, respectively. This, along with the
moderately high correlations of the social desirability items, provided evidence that
this factor was indeed captured in the three factors. One last test was conducted
Measuring a Multi-Dimensional Construct J BUSN RES 197
1993:28:191-210

to ensure this point. The five social desirability items were combined into a factor
that was correlated with each of the other three factors. The result was moderate
and highly significant correlations between that factor and the political, economic,
and technological dimensions (r = .43, p < .OOOl; r = .50, p < .OOOl; r = .28,
p < .OOl, respectively). Based on these findings and the above reasoning, it was
concluded that the social desirability component was captured by the three emer-
gent dimensions.
Next, by using the criterion of meaningfulness recommended by Gorsuch (1983),
items that had factor loadings less than .30 or that loaded highly on more than one
factor were deleted, resulting in a scale with 14 items. The items that were removed
from the scale suggest that either these items did not explain a significant amount
of variation in the data or that these items were unique. The reliability coefficient
alphas for the three factors ranged from .56 to .71. The overall Cronbach’s alpha
for the new 1Qitem scale was found to be .95. Given these highly satisfactory
results, the scale was then tested using the 14 bipolar word pairs.

Testing the Revised Scale


The revised scale, (Appendix C), which consisted of 14 items, was tested to de-
termine if it was a reliable and valid measurement. This was necessary for two
reasons (Nunnally, 1978). First, the revised instrument was considerably shorter
than the original version. Second, determining if the scale provided consistent
results when different countries were used as image objects was necessary. If the
scale held up across various countries, then it could be considered a stable instru-
ment (Churchill, 1979).
This phase of scale testing included a new group of 230 undergraduate students
from 2 midwestern universities. The use of large samples during this phase of scale
construction is recommended by Nunnally (1978). At least 10 times as many subjects
as items should be used to ensure that the reported coefficients are not inflated.
In this phase, the United States was the image object. Confirmatory factor
analysis was used on the 1Citem scale with varimax rotation as the factors were
deemed independent. This resulted in the same three factors emerging, as previ-
ously mentioned. The majority of the item-to-total correlations for each of the
items was above .40, supporting the premise that they all relate to the single country
image construct. The Cronbach’s alpha for the entire scale was calculated at .925.
Additionally, the correlation between split halves was calculated and found to be
.78. The newly identified component scales are shown in Table 2 with factor loadings
ranging between .30 and .78.
The first factor contains five items that were related to the overall political climate
and characteristics of a country, in this instance the United States. The second
factor had five items that were related to the economic environment of a country.
The last set had four items that addressed the technological aspects and charac-
teristics of a country. The majority of the item-to-total correlations of these three
factors were above .40. The high item-to-total correlations were also reflected in
the low determinant of the correlation matrix of .0095. This measure shows little,
if any, variability in the factors. The second column of Table 2 displays, for each
item, a Cronbach’s alpha computed from the other items in the scale, ranging
between .81 and .84.
198 J BUSN RES I. M. Martin and S. Eroglu
1993:28:191-210

Table 2. Component Scale Items


Item-to-Total Alpha if
Description: Correlation Item Deleted

Factor One: Political Dimension


Democratic versus dictatorial system .713 .812
Capitalist versus communist system ,612 .x20
Civilian versus military system ,631 .818
Pro-western versus pro-communist ,708 .813
Free market versus centrally planned system ,573 ,823

Factor Two: Economic Dimension


Level of standard of living ,535 ,825
Stability of economic environment ,443 ,831
Quality of products ,359 ,836
Existence of a welfare system ,278 ,842
Level of labor costs ,414 ,834

Factor Three: Technological Dimension


Level of industrialization ,429 .832
Level of technological research ,247 .H4l
Level of literacy ,381 .H3S
Mass produced versus handcrafted products ,285 ,840

Total Scale Reliability Coefficient ,925

Construct Validation
The term construct validity is the degree to which a measure assesses the construct
it is purported to describe. This means the measure only can be assessed indirectly
as the construct (i.e., country image) is not observable (Peter, 1981). For example,
in this study country image was theoretically hypothesized to have four dimensions.
This hypothesis was derived from an evaluation of the interdisciplinary literature
on country image. The results of the factor analysis and further tests provided
significant support for three, not four, underlying dimensions. It was argued earlier
that the lack of the fourth dimension, social desirability, may be due to the fact
that it falls within the other three factors of country image and not as a separate
dimension. This was empirically verified using the social desirability items in the
original 29-item scale.
The next step was to hypothesize that country image has three underlying di-
mensions as identified in the factor analysis and that these three dimensions fit the
theoretical dimensions suggested by the literature. This fit could be interpreted as
supportive evidence of construct validity (Peter, 1981). More substantive evidence
of construct validity was then obtained via an examination of the content validity
and discriminant validity.
Construct validity concerns the adequacy of the domain of observables relating
to a construct and can be tested by analyzing how item scores correlate with each
other (Churchill, 1979; Nunnally, 1978). To do this, the lbitem scale was further
tested by collecting data using two other countries as image objects. The countries
of India and West Germany were chosen for this purpose mainly because pre-tests
showed these countries to be different with respect to at least two of the three
underlying dimensions of country image (economic and technological). The scale
Measuring a Multi-Dimensional Construct J BUSN RES 199
1993:28:191-210

Table 3. Evaluation of Construct Validity of Country Image for India


Item-to-Total Squared Multiple
Description: Correlation Correlation

Factor One: Political Dimension


Democratic versus dictatorial system s59 ,455
Capitalist versus communist system ,485 .448
Civilian versus military system ,482 .401
Pro-western versus pro-communist ,345 ,435
Free market versus centrally planned system ,324 ,390

Factor Two: Economic Dimension


Level of standard of living ,756 .710
Stability of economic environment ,718 .706
Quality of products ,589 ,528
Existence of a welfare system ,593 .477
Level of labor costs ,552 ,624

Factor Three: Technological Dimension


Level of industrialization ,734 ,721
Level of technological research .709 ,528
Level of literacy .686 .540
Mass produced versus handcrafted products ,724 .527

Total scale reliability coefficient .895

was then administered to a new group of 160 students. There were 80 students
who rated West Germany and 80 who rated India for country image.
The analysis gave highly satisfactory results (Tables 3 and 4). For both West
Germany and India, the three previously obtained (political, economic, and tech-
nological) factors emerged as a result of confirmatory factor analysis using varimax
rotation. The factor loadings varied between .52 and .85 for both countries. The
item-to-total correlations for the majority of the items in the scale were all greater
than 55 and .65 for India and West Germany, respectively. These favorable cor-
relations also were reflected in the very low determinants of the correlation matrices
of .00036 and .000055, for India and West Germany, respectively.
The squared multiple correlations obtained provide an index of the relative
extent to which each scale item contributes to capturing country image. This can
be seen where the majority of the squared multiple correlations were above .52
for India and .69 for West Germany. In addition, the Cronbach’s alphas for each
dimension were checked and were found in the range from .686 to .887 for the
country of India and from .581 to .761 for West Germany. This provided added
credence to support the conclusion that all the image items measure one singly
underlying construct, country image. Once this was established, the scale had to
be tested further for content and discriminant validity.
Content validity concerns the degree of representativeness of the items to the
construct. Open-ended interviews with representative respondents (Churchill,
1979) as well as expert judgments (Green and Tull, 1978) are deemed appropriate
procedures for checking scale content. In this case, generation of the initial item
pool employed both of these methods. Furthermore, the final scale was again
200 J BUSN RES I. M. Martin and S. Eroglu
1993:28:191-210

Table 4. Evaluation of Construct Validity of Country Image for West Germany


Item-to-Total Squared Multiple
Description: Correlation Correlation

Factor One: Political Dimension


Democratic versus dictatorial system .778 ,757
Capitalist versus communist system ,786 ,784
Civilian versus military system ,755 ,783
Pro-western versus pro-communist .73s .750
Free market versus centrally planned system .743 .730

Factor Two: Economic Dimension


Level of standard of living ,721 ,812
Stability of economic environment ,663 ,796
Quality of products ,664 ,356
Existence of a welfare system ,636 ,561
Level of labor costs .555 .444

Factor Three: Technological Dimension


Level of industrialization ,524 ,465
Level of technological research .649 ,687
Level of literacy .747 .710
Mass produced versus handcrafted products .44v .693

Total Scale Reliability Coefficient ,928

examined independently by the initial group of judges for scale content and
wording.
Discriminant validity needed to be established as the domain of the construct
is multi-dimensional (Campbell and Fiske, 1959). Discriminant validity exists to
the extent that one can empirically differentiate the construct from other similar
constructs and can indicate what is unrelated to the construct. Therefore, it was
necessary to test whether the new country image scale was significantly different
from the previously used Nagashima scale (1970, 1977).
The Nagashima scale (Appendix D) consists of 20 bipolar items measured on a
7-point scale. It is frequently used in international marketing research (e.g., Han
and Terpstra, 1988; Han, 1989; Narayana, 1981) and attempts to measure the image
of products with foreign country of origin, which, in turn is purported to reflect
one’s country image (Bilkey and Nes, 1982).
In order to make the comparison between the country image scale and that of
Nagashima, several steps were taken. First, a new group of 158 students was used
to rate West Germany (n = 79) and India (n = 79) with the Nagashima instrument.
Because discriminant validity requires that the two constructs do not correlate
highly with each other (e.g., Bagozzi, 1982; Campbell and Fiske, 1959; Peter, 1981)
correlations were calculated between each of the three dimensions of the country
image scale and the Nagashima scale for both India and West Germany (Table 5).
The correlations for each of the dimensions and the Nagashima scale were low
to moderate and non-significant for all the dimensions across both countries. Having
both countries tested across the two different scales using different student samples,
we found that there was discriminant validity between the country image scale and
.I BUSN RES
Measuring a Multi-Dimensional Construct 1993:28:191-210

TABLE 5. Evaluation of Discriminant Validity of Country Image for West Germany


and India
West Germany India

Political dimension .18 .43


(.68) C.29)
Economic dimension .41 .28
632) t.501
Technological dimension .51 .34
C.19) C-191

Note: Numbers in parentheses are probability values for the respective correlations

the product image scale developed by Nagashima (1970). This verifies the hypoth-
esis that the new country image scale is unique and not simply a reflection of other
variables (Peter and Churchill, 1981).

Conclusions
Countries evoke different product images in consumers’ minds. However, because
country of origin effects vary across countries, samples, and products, the results
of this stream of research seem to lack consistency and generalizability. Among
the suggestions made to advance the state of the art in the country of origin literature
is more emphasis on measurement of the relevant constructs such as country image
(e.g., Jaffe and Nebenzahl, 1984; Cattin et al., 1982; Han, 1989).
To this end the present study was designed to develop and validate a multiple-
item scale for measuring the construct of country image as distinct from product
immge. The procedures recommended by Churchill’s (1979) paradigm for devel-
oping better measures were followed to capture the concept of country image. The
resulting 16item semantic differential scale was developed by using 4 different
data sets with students from 2 midwestern universities. Following a series of pro-
cedures for item generation, expert judges were used to select those items with
high content validity. The reliability of the scale was established by the use of
various indicators across different countries as well as the scale’s construct and
discriminant validities.
The present instrument is intended to be useful for practitioners and researchers
alike. From a practical perspective, there are situations where a quantitative yard-
stick on country image may be necessary for managers. In marketing a country in
a competitive industry such as tourism, it would be useful to have a baseline rating
of the target market’s opinion of the country to be promoted.
Similarly, when trying to attract foreign investment into a country, public and
private promoters would benefit from information on the perceptions that the
concerned parties have about a certain country. Under these and other similar
situations, when the “product” is a country, choosing the most appropriate mar-
keting strategy largely depends on the type of image held by the target ,market.
An example that we are now faced with is the dramatic change in eastern Europe.
With the radical change in the political, economic, and technological base of coun-
tries such as Poland, Romania, and Hungary, how will the perceptions that West-
erners hold of these countries affect their ability to compete in the world market?
202 J BUSN RES I. M. Martin and S. Eroglu
1993:2X:191-210

These countries should be interested in determining exactly what their respective


country image is with the Western world so that negative perceptions can be
changed to allow them to better compete in the marketplace.
This same concept can be applied to the changes occurring in western Europe
as they face 1992. The perception that they are a solid economic and political block
is critical for the success of their push to merge into a United States of Europe.
The country image scale can provide a quantitative yardstick for these countries
to determine the perception of other countries. This information then can be used
to adapt persuasive communication campaigns to provide others with the “correct”
image. If the image is positive, this should be stressed in an attempt to market the
country’s products and services.
The practical importance for marketers is that a favorable country image can
be used to sell inferior products only temporarily. For example, country image can
act as an information cue for consumers. This cue combines with an array of
information cues, both intrinsic and extrinsic to the product, to aid the consumer
in evaluating the product. This implies that there must be a match between country
image and the image that one has of a country’s products (Bilkey and Nes, 1982).
The theoretical interests of the scale development project are several. First,
country image has three, not four, underlying dimensions. Past literature as well
as research has asserted that social desirability is an underlying dimension of country
image, but based on extensive testing, it was determined that social desirability is
captured by the three factors of economic, political, and technological aspects.
Second, researchers would benefit from a validated instrument specifically de-
signed to measure country image. The scale would contribute to designing studies
that are more explanatory and/or predictive in nature rather than purely descriptive.
For example, the present scale would be applicable to determine if consumers’
evaluations of products from a country are consistent with their overall image of
that country. The scale also would help to explain how consumers develop the
country stereotypes they have by examining the scores for each of their dimensions.
Additionally, the country image scale could be used to understand if and how
different countries (as perceived by consumers) are likely to affect consumers’
evaluations of different product classes. Another possible application of the scale
would be to categorize consumers into several image segments (e.g., low, medium,
high) on their individual image scores for a single country. Alternatively, consumer
segments could be identified based on their image scores across countries.
It should be emphasized that this scale is strictly aimed at measuring one’s image
of a country and should not be used to assess image of or attitudes toward products
from a country. With respect to limitations of this scale, during the development
and validation of the scale only a few countries were used. Further tests need to
be carried out by using other countries as well as non-student samples to verify
the stability and the validity of the scale. Along with the above, it should be further
investigated as to whether social desirability continues to bc a component of the
three factors identified and verified in this study.
Future research could look at the ability to use measures of country image to
predict the probability of purchase behavior; that is, what are the stereotypes that
consumers have for countries that rank as planned economies with low standards
of living and low literacy and level of industrialization? The next interesting issue
is to determine if the stereotypes that form our country image also impact our
probability of buying a certain product from that country.
J BUSN RES 203
Measuring a Multi-Dimensional Construct 1993:2X:191-210

Appendix A
Judge Evaluation Form
JUDGE EVALUATION FORM

Thank you for agreeing to participate as a judge in this scaling project. This is a
crucial step in the development of an accurate scale of any concept. On the following
pages there are 60 word/phrases that are generated to represent the concept of
Country Image. In this study, for the purposes of scale development, the definition
of country image is:

One’s impressions of a country is based on a set of perceived ratings of the


country along various dimensions. These dimensions include economic, social,
cultural, geographical, technological and political characteristics which reflect
a perception of that particular country.

Your task is to judge each of the bipolar words/phrases as to how well they represent
the concept of Country Image as defined above. Each set of bipolar words/phrases
should be rated using the following scale:

a. Clearly representative of a country’s image.


b. Somewhat representative of a country’s image.
c. Not representative of a country’s image.

Only one of the above choices that best represents your opinion of the bipolar
words/phrases and its ability to reflect the concept of Country Image, should be
circled. The following example illustrates the task of selecting representative con-
cepts of Country Image.

1. Extroverted Introverted
People People

a. Clearly representative b. Somewhat rep. of @ Not rep. of a


of a country’s image. a country’s image country’s image

2. Backward Advanced
Economy Economy

@ Clearly representative b. Somewhat rep. of c. Not rep. of a


of a country’s image. a country’s image. country’s image

Please remember that this is a weeding out process. It is important that you realize
that most or all may seem to clearly represent the concept. However, we are
interested in the fine distinctions between the a, b, c categories.
204 J BUSN RES
1993:28:191-210
I. M. Martin and S. Eroglu

Appendix B

Original 29-Item Country Image Scale

Country Image Scale

This is a survey to find out what a person thinks about a certain country. To
measure this, we will ask you to rate the country that appears at the top of the
page against a series of descriptors by placing a check (J) on the scale from one
to seven that best reflects your judgment. There are no right or wrong answers.
We are only interested in how YOU perceive the country. Consider the following
example:

By placing a check (J) in the middle, it would mean that you feel that the country
that appears at the top of the page is neither pro-Western or pro-Communist.

pro-Western --: ~. -. J: -. -: -: pro-Communist


(I) (2)’ (3)’ (4) (5)’ (6) (7)

However, if you feel that the country is extremely pro-Western you would mark
the scale in the following manner:

pro-Western J: pro-Communist
(1) (2):(3):(4):0:(h):(7):

Or, if you feel that it is slightly pro-Communist you would mark the scale in this
manner:

pro-Western _: -: -. -. ~. -. J. __: pro-Communist


(I) (2) (3) ’ (4)’ (5) ’ (6) (7)

Please ask if you have any questions.

(Country Name)

1) economically economically
developed (1):(2):(3):(4):(5):(h):(7):
underdeveloped
2) democratic ~. ~. -. -. -. ~. _. dictatorial
system (1)’ (2)’ (3)’ (4)’ (5)’ (6)’ (7)’ system
3) civilian ~. ~. -. -. -. military
government -.(I)’ -.(2)’ (3)’ (4)’ (5)’ (6)’ (7)’ government
4) extended -. . -. . -. . -. . -. ~. . _. . nuclear
family (1) (2) (3) (4) (5) (6) (7) family
5) large -. -. -. -. -. ~. ~. small
population (I)’ (2)’ (3)’ (4)’ (5)’ (6). (7)’ population
6) high per low per
capita (I)’ (2)’ (3)’ (4)’ (5)’ (6)’ (7)’ capita
income income
Measuring a Multi-Dimensional Construct J BUSN RES
1993:28:191-210
205

7) exporter of -. -. -. -. importer of
industrial -.(1)’ -.(2)’ -.(3)’ (4)’ (5)’ (6)’ (7)’ industrial
products products

8) free market -. -. -. centrally


system ~.(1)’ -.(2)’ -.(3)’ -.(4)’ (5)’ (6)’ (7)’ planned system

9) culturally -. ~. _. -. -. ~. culturally
diverse -.(1)’ (2)’ (3)’ (4)’ (5)’ (6)’ (7)’ uniform

10) stable _.-. -. -. unstable


economic -.(1)’ -.-.
(2)’ (3)’ (4)’ (5)’ (6)’ (7)’ economic
environment environment

11) exporter of -. -. _. -. -. _. importer of


agricultural -.(1)’ (2)’ (3)’ (4)’ (5)’ (6). (7)’ agricultural
products products

12) existence of -.-. -. -. existence of


a large -.(1)’ -.-.
(2)’ (3)’ (4) * (5)’ (6)’ (7)’ a small middle
middle class class

13) large land -. ~. _. -. -. _. small land


mass -.(1)’ (2)’ (3)’ (4)’ (5)’ (6)’ (7)’ mass

14) high level of -. -. ~. _. -. ~. _. low level of


technological (1)’ (2)’ (3)’ (4)’ (5)’ (6)’ (7)’ technological
research research

15) mass produced -. -. -. -. handcrafted


products -.(1)’ ~.(2)’ -.(3)’ (4)’ (5) * (6)’ (7)’ products

16) high literacy -. -. -. low literacy


rates -.(1)’ -.(2)’ -.(3)’ -.(41’ (5)’ (6)’ (7)’ rates

17) exporter of -._. -. -. importer of


raw materials -.(1)’ -.-.
(2)’ (3)’ (4)’ (5)’ (6)’ (7)’ raw materials

18) pro-Western -.(1)’ -.-.


(2)’ (3)’ -.-.
(4)’ (5)’ -.(6)’ -: pro-Communist
(7)

19) high labor (2)’


-.(1)’ ~._. (3)’ -.-.
(4)’ (5)’ -.6 -: low labor
costs (7) costs

20) existence of a (2)’


-.(1)’ -.-. (3)’ -._.
(4)’ (5)’ -.(6)’ -: lack of a
welfare system (7) welfare system

21) production of -._. -. -. production of


high quality (2)’
-.(1)’ -.-. (3)’ (4)’ (5)’ (6)’ (7)’ low quality
products products

22) high standard -. -. ~. -.-. ~. -. low standard


of living (1)’ (2)’ (3)‘ (4)’ (5)’ (6)’ (7)’ of living

23) stable -. ~. -. -. unstable


political -.(1). ~.(2)’ -.(3)’ (4)’ (5)’ (6)’ (7)’ political
environment environment
206 J BUSN RES
1993:28:191-210
I. M. Martin and S. Eroglu

24) large social --: ---: ~. ~. ~. small social


class (1) (2) (3)‘(4):(5): (6)’ (7)’ class
differences differences

25) exporter of -: -. -: -: _: _. _. importer of


consumer (1) (2)’ (3) (4) (5) (6)’ (7). consumer
products products

26) capitalist -: -: -. ~. ~. _. _. communist


system (1) (2) (3)’ (4)’ (5)’ (6)‘ (7)’ system

27) high pop- -: -: -. -. ~. ~. _. . high population


ulation in- (1) (2) (3) ’ (4)’ (9’ (6) 6’) out-migration
migration rate rate

28) predominantly -: -. -: predominantly


industrialized (1) (2)’ (3) (4):(5): (6):(7): non-industrialized

29) high unemploy- _: _: _: -: --: _: _: low unemployment


ment rate (1) (2) (3) (4) (5) (6) (7) rate

Appendix C
Revised II-Item Country Image Scale
Country Image Scale

This is a survey to find out what a person thinks about a certain country. To
measure this, we will ask you to rate the country that appears at the top of the
page against a series of descriptors by placing a check (J) on the scale from one
to seven that best reflects your judgment. There are no right or wrong answers.
We are only interested in how YOU perceive the country. Consider the following
example:

By placing a check (J) in the middle, it would mean that you feel that the country
that appears at the top of the page is neither pro-Western or pro-Communist.

pro-Western -: -: -: J: _: _: _: pro-Communist
(1) (2) (3) (4) (5) (6) (7)

However, if you feel that the country is extremely pro-Western you would mark
the scale in the following manner:

pro-Western J. . pro-Communist
(1).(2):(3):(4):&q:(7).

Or, if you feel that it is slightly pro-Communist you would mark the scale in this
manner:
Measuring a Multi-Dimensional Construct J BUSN RES
1993:28:191-210
207

pro-Western J. pro-Communist
o:(2):(3):(4):(5):(6).(7):
Please ask if you have any questions.

(Country Name)

1) economically -: -: -: --: _: _: -: economically


developed (1) (2) (3) (4) (5) (6) (7) underdeveloped

2) democratic -: -: -: _: _: -: -: dictatorial
system (1) (2) (3) (4) (5) (6) (7) system

3) mass produced _: _: _: -: -: -: -: handcrafted


products (1) (2) (3) (4) (5) (6) (7) products

4) civilian . -. -. military
government (1)’ (2)’ (3)’ (4)’ 0’ (6)’ 6’)’ government

5) predominantly -: -: -: _: _: _: -: predominantly
industrialized (1) (2) (3) (4) (5) (6) (7) non-industrialized

6) high labor -: -: -: -: -: -: -: low labor


costs (1) (2) (3) (4) (5) (6) (7) costs

7) high literacy -: -: -: _: _: -: -: low literacy


rates (1) (2) (3) (4) (5) (6) (7) rates

8) free market _: _: _: -: -: -: -: centrally


system (1) (2) (3) (4) (5) (6) (7) planned system

9) existence of -: -: -: _: _: -: -: lack of a
welfare system (1) (2) (3) (4) (5) (6) (7) welfare system

10) stable -. -. -. -. ~. -. -. unstable


economic (1)’ (2)’ (3)’ (4)’ (5)’ (6)’ (7)’ economic
environment environment

11) exporter of -: -: -: _: -: -: -: importer of


agricultural (1) (2) (3) (4) (5) (6) (7) agricultural
products products

12) production of -: ~. _. -. -. ~. -: production of


high quality (1) (2)’ (3)’ (4)’ (5)’ (6)’ (7) IOW quality
products products

13) high standard -: -: -: . low standard


of living (1) (2) (3) O’(5)’ (6)’ 0’ of living

14) high level of -: f low level of


technological (1) (2)’ u)’ (4)’ (5)’ (6)’ (7)’ technological
research research
208 J BUSN RES
1993:28:191-210
I. M. Martin and S. Eroglu

Appendix D
The Nagashima Product Image Scale

(Country Name)

1) Expensive -: -: -: _. --: _: -: Inexpensive


(I) (2) (3) (4)’ (5) (6) (7)

2) Reasonably -: -: -: --: ~. ~. -: Unreasonably


Priced (I) (2) (3) (4) (5)’ (6)’ (7) Priced

3) Reliable -: ~. ~. -: _: -. -: Unreliable
(I) (2)’ (3)’ (4) (5) (6)’ (7)

4) Luxury items Necessary Items


(1):(2):(3):(4):(5):(h):(7):

5) Heavy Industry -: -: -: -: -: -: -: Light Manufacturing


Products (1) (2) (3) (4) (5) (6) (7) Products

6) Careful and . -. -. . Not so careful


meticulous (I)’ (2). (3)’ (4)’ (5)’ (6)’ (7) meticulous
workmanship workmanship

7) Technically . . . . . . : Technically
Advanced (1) (2) (3) (4) (5) (6) (7) Backward

8) Mass Produced : Hand Made


(I)’ (2)’ (3)’ (4)’ (5)’ (6)’ (7)
9) World Wide . . . . .. -. . -. . Mostly Domestic
Distribution (1) (2) (3) (4) (5) (6) (7) Distribution

10) Inventive . Imitative


(1):(2):(3):(4):(5):(6):(7).

11) Pride of -: -: -: -: _: _: _: Not much pride


ownership (1) (2) (3) (4) (5) (6) (7) of ownership

12) Much . . . : -: Little


Advertising (1) (2) (3) (4) (5) (6) (7) Advertising

13) Recognizable -: -: -: -: -: -: -: Unrecognizable


brand names (1) (2) (3) (4) (5) (6) (7) brand names

14) Large choice -: ~. ~. -: -: ~. -: Limited choice


of size & (1) (2)’ (3)’ (4) (5) (6)’ (7) of size &
model model

15) More concerned-. -. -. ~. ~. ~. -: More concerned


with outward (I)’ (2)’ (3)‘ (4)’ (5)’ (6)’ (7) with performance
appearance

16) Clever use . Not clever


of color (1):(2):(3):(4):(5):(6):(7). use of color
J BUSN RES 209
Measuring a Multi-Dimensional Construct 1993:28:191-210

17) More for -. -. -. -. ~. _: More for


young people (1): (2)’ (3) ’ (4)’ (5) ’ (6)’ (7) old people
18) More for men _: _: -. -.-. -. -. . More for women
(1) (2) (3)’ (4)’ (5)’ (6)’ (7)
19) Upper class ._____:
_: -: -: -: -: -: Lower class
(1) (2) (3) (4) (5) (6) (7)
20) Exclusive _. . -.-. . . -.-. . . -. . _: Common
(1) (2) (3) (4) (5) (6) (7)

References
Alreck, P. L., and Settle, R. B., The Survey Research Handbook, Irwin Series in Marketing.
1985.
Bagozzi, R. P., A Field Investigation of Causal Relations among Cognitions, Affect, In-
tentions, and Behavior. Journal of Marketing Research 19 (November 1982): 562-584.
Baughn, C. C., and Yaprak, A., Mapping the Country of Origin Literature: Recent De-
velopments and Emerging Research Avenues, in Product and Country Image: Current
Perspective. N. Papadopoulos and L. Heslop, eds.,
Bilkey, W. J., and Nes, E., Country of Origin Effects in Product Evaluations. Journal of
International Business Studies 10 (Spring/Summer 1982): 89-99.
Campbell, D. T., and Fiske, D. W., Convergent and Discriminant Validation by the
Multitrait-Multimethod Matrix. Psychological Bulletin 56 (1959): 81-105.
Cattin, P., Jolibert, A., and Lohnes, C., Cross-Cultural Study of Made-in Concepts. Journal
of International Business Studies 10 (Winter 1982): 131-141.
Churchill, G. A., A Paradigm for Developing Better Measures of Marketing Constructs.
Journal of Marketing Research 16 (February 1979): 64-73.
Davis, H., Douglas, S., and Silk, A., Measure Unreliability: A Threat to Cross National
Marketing Research. Journal of Marketing 45 (2) (1981): 98-109.
Dornoff, R. J., Tankersley, C. B., and White, G. P. Consumers’ Perceptions of Imports.
Akron Business and Economic Review 5 (Summer 1974): 26-29.
Frazier, G. L., On The Measurement of Interfirm Power in Channels of Distribution. Journal
of Marketing Research 20 (May 1983): 158-166.
Gaedeke, R., Consumer Attitudes Toward Products “Made-In” Developing Countries.
Journal of Retailing 49 (2) (1973): 13-24.
Gorsuch, R. L., Factor Analysis, 2nd ed., Lawrence Erlbaum and Associates, Hillsdale,
NJ. 1983.
Green, P., and Tull, D. S., Research for Marketing Decisions, Prentice-Hall, Englewood
Cliffs, NJ. 1978.
Hafhill, D. S., Multinational Marketing Strategy: Implications for Attitudes Toward Country
of Origin. Management International Review 20 (4) (1980): 26-30.
Han, C. M., Country Image: Halo or Summary Construct? Journal of Marketing Research
26 (May 1989): 222-229.
Han, C. M., and Terpstra, V., Country of Origin: Effects for Uni-National and Bi-National
Products. Journal of International Business Studies 16 (Summer 1988): 235-256.
Holsti, 0. R., Content Analysis for the Social Sciences and Humanities, Addison-Wesley
Publishing, Reading, MA. 1969.
Hong, S., and Wyer, R. S., Effects of Country of Origin and Product Attribute Information
on Product Evaluation. Journal of Consumer Research 16 (1989): 175-187.
J BUSN RES 1. M. Martin and S. Eroglu
lYY3:28:lYl-210

Hooley, G., Shipley, D., and Krieger, N., A Method for Modeling Consumers’ Perceptions
of Country of Origin. Znternational Marketing Review (Autumn 1988): 67-76.
Jaffe, E. D., and Nebenzahl, I. D., Alternative Questionnaire Formats for Country Image
Studies. Journal of Marketing Research 21 (November 1984): 463-471.
Johansson, J. K., and Moinpour, R., Objective and Perceived Similarity of Pacific Rim
Countries. Columbia Journal of World Business 12 (4) (Winter 1977): 65-76.
Johansson, J. K., Douglas, S. P., and Nonaka, I., Assessing the Impact of Country of Origin
on Product Evaluations: A New Methodological Perspective. Journal of Marketing Re-
search 22 (November 1985): 373-381.
Lundstrom, W. J., and Lamont, L. M., The Development of a Scale to Measure Consumer
Discontent. Journal of Murketing Reseurch 13 (November 1976): 373-381.
Malhotra. N., A Scale to Measure Self-concepts, Person-concepts, and Product-concepts.
Journal of Marketing Research 18 (November 1981): 456-464.
Nagashima, A., A Comparison of Japanese and U.S. Attitudes Toward Foreign Products.
Journal of Marketing 34 (July 1970): 68-74.
Nagashima, A., A Comparative Made-In Product Image Survey Among Japanese Busi-
nessmen. Journul of Marketing 41 (July 1977): 95-100.
Narayana, C., Aggregate Images of American and Japanese Products: Implications on
International Marketing. Columbia Journal of World Business 16 (2) (Summer 1981):
31-35.
Nunnally, J., Psychometric Theory, McGraw-Hill, New York. 1978.
Osgood, C. E., Succi, G. J., and Tannenbaum, P. M., The Measurement of Meaning,
University of Illinois Press, Champaign-Urbana, IL. 1057.
Parameswaran, R.. and Yaprak, A., A Cross-National Comparison of Consumers’ Research
Measures. Journal of lnternutional Business Studies 15 (Winter 1987): 35-49.
Peter, J. P., Reliability: A Review of Psychometric Basics and Recent Marketing Practices.
Journal of Marketing Research 16 (February 1979): 6-17.
Peter. J. P., Construct Validity: A Review of Basic Issues and Marketing Practices, Journal
of Marketing Research, May, 133-45, 1981.
Peter, J. P., and Churchill, G. A., Relationships Among Research Design Choices and
Psychometric Properties of Rating Scales: A Meta-Analysis. Journul of Marketing Re-
search 18 (May 1986): 133-141.
Schooler, C., Wildt, A., and Jones, J., Strategy Development for Manufactured Exports
of Third World Countries to Developed Countries. Journal of Global Marketing 1 (l-2)
(1987): 53-67.
Wee, C. H., Shopping Area Image: Its Factor Analytic Structure and Relationships with
Shopping Trips and Expenditure Behavior. Advances in Consumer Research Proceedings
14 (1986): 48-52.
Wish, M., Deutsch, M., and Biener, L., Differences in Perceived Similarity of Nations, in
Multidimensional Scaling: Theory and Application in the Behavioral Sciences. A.K. Rom-
ney. ed., Seminar Press, New York. 1972.
Zaichkowsky, J. L., Measuring the Involvement Construct. Journal of Consumer Research
12 (December 1985): 341-352.
Zinkhan, G. M., and Muderrisoglu, A., Involvement, Familiarity, Cognitive Differentiation,
and Advertising Recall: A Test of Convergent and Discriminant Validity. Advances in
Consumer Reseurch Proceedings 10 (1982): 356-361.

Anda mungkin juga menyukai