levin and jasper
TRANSCRIPT
8/8/2019 Levin and Jasper
http://slidepdf.com/reader/full/levin-and-jasper 1/14
Journal of Experimental Psychology: Applied1996, Vol. 2, No. 1,17-30
Copyright 1996by the American Psychological Association, Inc.107fr«l8X/96/$3.00
Experimental Analysis of Nationalistic Tendenciesin Consumer Decision Processes:Case of the M ultinational Product
Irwin P. Levin and J. D. JasperUniv ersity of Iowa
Participants evaluated pairs of hypothetical hybrid products varying in thepercentage of American workers employed in manufacturing each product(50% vs. 80%)and the company nationality (American vs. Japanese, German,or Taiwanese). For each pair, participants indicateda preference andreasonsfor their preference. Employing American workers was an important factor inthe preferences of all participants across all products (automobiles, personalcomputers, and clothing). However, preference for American companies w assignificant in most cases only for those scoring high on a scale of nationalism,as mediated by greater concern for supporting America and the Americaneconomy and lesser concern for perceived quality differences. Results arediscussed in terms of theories of consumer ethnocentrism and ways ofmarketing multinational products.
Would you be more apt to buy a product from a
company that employs mostly American workers
or from a company that employs mostly foreign
workers? For most of us, this would be a simple
question, but what if the company employing
mostly American workers is a foreign-based com-
pany and the one employing mostly foreign work-
ers is an American-owned company? These areexactly the kinds of questions that consumers face,
at least implicitly, in today's global marketplace. In
the experiments reported in this article, we explic-
itly posed such questions and addressed them with
the methods of experimental psychology.
Traditionally, as a number of studies will attest
(Bilkey & Nes, 1980; Gaedeke, 1973; Han, 1988;
Hong & Wyer, 1989, 1990; Levin & Jasper, 1995;
Levin, Jasper, Mittelstaedt, & Gaeth, 1993; Levin,
Johnson, & Jasper, 1993; Obermiller & Spangen-
Irwin P. Levin and J. D. Jasper, Department ofPsychology, University of Iowa.
We thank Gary Gaeth for helpful comments andWendy Forbes,Shelly K. Stein, Cheryl Nelson, and ScottPerlman for help in collecting and scoring data.
Correspondence concerning this article should beaddressed to Irwin P. Levin, Department of Psychology,University of Iowa, Iowa City, Iowa 52242. Electronicmail may be sent to [email protected].
berg, 1989; Schooler, 1971), consumers have had
relatively little trouble when using simple country-
of-origin cues as sources of information to differen-
tiate among and to make decisions about domestic
and foreign alternatives. For example, it might be
easy to assume that an American brand automo-
bile was designed, engineered, manufactured, and
assembled almost entirely in the United States.However, with the advent of hybrid or multina-
tional products, that "pure" relationship has been
somewhat obscured. Automobiles and other expen-
sive and complex products are now often designed
and engineered in one country and built in an-
other, thereby confusing their nationalistic identi-
ties. In fact, some foreign automakers have nur-
tured this lack of national clarity to sell cars in the
United States by advertising that they are made by
American workers. Not to be outdone, of course,
American manufacturersof automobiles have used
similar tactics to develop and market domesticbrands like Geo and Saturn (Ettenson & Gaeth,
1991).
The question that needs to be addressed through
research, but has thus far been largely ignored,
concerns what effect these multiple country-of-
origin cues have on consumer choice. The present
study was designed to investigate this issue by
independently manipulating twonationalistic cues:
17
8/8/2019 Levin and Jasper
http://slidepdf.com/reader/full/levin-and-jasper 2/14
18 LEVIN AND JASPER
company nationality (e.g., American or Japanese)
and percentage of American workers employed
(50% or 80%) in the manufacture of a particular
brand. We selected specific paired comparisons
that varied (a) the company nationality whileholding the percentage of American workers con-
stant, (b) the percentage of American workers
while holding the company nationality constant,
and (c) both factors either congruently or incongru-
ently. Of the 16 possible ways of pairing the four
distinct product descriptions formed by combining
tw o company nationalities and two percentages,
only 12 had the properties of interest. Six of these
were redundant with the other 6, leaving the 6
shown in Table 1. In each case, participants were
asked to indicate their degree of preference be-
tween the two products and to provide a writtenreason for their preference. Of particular interest
to us were those comparisons that created conflict
for consumers an d required a tradeoff between
cues such as the American company employing
fewer American workers than the foreign com-
pany.
W hat is unique about this methodology is that itallowed us to gain a detailed understanding of the
processes underlying such decisions. Asking partici-
pants to give reasons for their preferences pro-
vided on e vehicle for this understanding. However,
our knowledge would probably have been incom-plete without also considering individual differ-
ences, the most appropriate of which may be
consumer nationalism or ethnocentrism. As intro-
duced and defined by Shimp and Sharma (1987),
consumer ethnocentrism represents "the beliefs
held by American consumers about the appropri-
ateness, indeed morality, of purchasing foreign
made products" (p. 280). Shimp and Sharma
developed a scale to measure this construct
(adapted for the present study) and, along with
others, provided empirical evidence that country-
of-origin information w as more important in thepurchase considerations of those scoring higher on
their scale.1
W e wanted to know why and w e also
wanted to discover whether consumers high onethnocentrism were more sensitive to all country-of-origin cues? As useful as this construct may bein providing a theoretical underpinningfor examin-
ing individual differences in country-of-origin ef-
fects, w e decided that a more microscopic analysis
w as needed.
By separating company nationality from employ-
ment of American workers and obtaining concur-
rent data on reasons for choice, we were able to
determine rather specifically how highly nationalis-
tic individuals differed from the rest. Furthermore,by including across experiments several differentforeign countries and products, we were able to
examine the extent to which country-of-origin
effect's depend on perceived quality differences
between products made by American an d foreigncompanies and the extent to which they depend on
perceived threats to American employment posed
by different countries.
Experiment 1
Participants in Experiment 1 were asked tochoose between automobiles made by American
and Japanese companies. Given recent media
attention to the ongoing economic battle between
these two countries, particularly in relation to car
manufacturing and sales, we thought that the
choice was quite appropriate. The main hypoth-
eses were that participants scoring high on a scale
of ethnocentrism would be (a) more sensitive to
American employment needs; (b) more concerned
with supporting American companies; and (c)
more apt to perceive that American products are
of superior quality.
Method
Paired comparisons. Sixty-three university stu-
dents (31 women and 32 men, virtually all Ameri-
can-born) participated in the experiment as part of
a course requirement. The initial cover story given
to participants read as follows:
Assume in each of six different situations that youare making a choice between two automobiles topurchase. Assume that the two cars within eachchoice pair are of comparable price and size andhave comparable features. The two d i f f e r only inthe nationality of the company making the car(either American or Japanese) and the percentage
1T he scale and the research using it, including the
present study, deal specifically with American ethnocen-trism or nationalism. Our assumption is that com parableresults would be obtained if analogs of this study weredone in other countries.
8/8/2019 Levin and Jasper
http://slidepdf.com/reader/full/levin-and-jasper 3/14
NATIONALISM IN CONSUMER DECISIONS 19
Table 1Six Paired Comparisons of Car Selection Choices
Choice A
Comparison no.
1
2
34
5
6
Companynationality
American
American
American
Japanese
American
American
Americanworkers (%)
80
50
80
80
80
50
Choice B
Companynationality
Japanese
Japanese
American
Japanese
Japanese
Japanese
Americanworkers (%)
80
50
50
50
50
80
of American workers employed in manufacturing
and/or assembling the car. You will find that this
percentage varies from car to car because of the
global nature of today's automobile industry. Some
parts of a car may be manufactured in one country,
other parts in another, and they might all be
assembled in yet a third country.
The six paired comparisons are shown in Table
1. The first two pairs varied the company national-
ity while the percentage of American workers was
held constant; the next twopairs varied the percent-
age of American workers while company national-
ity was held constant; and the last two pairs varied
both factors. Participants received the pairs in a
random order, each pair on a different page of the
response booklet, and were asked to indicate their
preferences by circling one of the numbers on a
6-point scale as illustrated below: "For Choice I (a
Japanese company with an 80% American work-
force) versus Choice J (an American company with
an 80% American workforce), the options are as
follows: 1—much prefer I, 2—somewhat prefer I,
3—slightly prefer I, 4—slightly prefer J, 5—some-what prefer J, 6—much prefer J."
Across participants, the left-right positions of
the two choices within a pair were counterbal-
anced. For purposes of standardization, responses
on the 6-point scale were scored such that a lownumber was associated with preference for anAmerican company over a Japanese company, or,
when company nationality was constant, a low
number w as associated with preference for 80%over 50% American workers employed. Use of a
numerical scale allowed us to assess the degree ofpreference for one type of car over another.
Reasons fo r choices. In addition to choosing
one of the two cars in each pair, participants were
required to indicate in their own words the main
reason or reasons for their choice. Responses were
classified by three judges according to the follow-ing categories:
(1) Employ Americans. Participants stated that
their choice employed or would employ more
American workers than the alternative.
(2) American economy. Participants mentioned
that their choice would keep more money in the
United States and/or help the economy.
(3) Japanese [American] quality. Participants
stated that their choice of a Japanese [American]
car was superior in quality to an American [Japa-
nese] alternative.
(4) American [Japanese] made (no further justi-
fication). Participants simply acknowledged that
their choice was American [Japanese] made and/or
originated from an American [Japanese] company.
No further justification w as offered.
(5) American made (support America). Partici-
pants acknowledged that their choice was Ameri-
can made and stated that it was important to
support one's own country.
(6) Company/worker match. Participants ar-
gued that their choice represented a match be-
tween company nationality and worker "majority,"
which might lead to abetter product.
(7) Employment diversity. Participants statedthat their choice of an equal mix of American an dnon-American workers w as better for everyone ina global economy.
(8) Idiosyncratic. Participants gave a reason that
was rare or uncategorizable.
We classified the reasons without knowledge of
the participants' level of nationalism (see below).
Average interjudge agreement between three inde-
pendent judges was 80%. Disagreements were
8/8/2019 Levin and Jasper
http://slidepdf.com/reader/full/levin-and-jasper 4/14
20 LEVIN AND JASPER
later resolved in each case by discussion among thethree judges.
Nationalism scale. At the end of each experi-mental session participants were given an attitude
survey designed to measure their nationalism-ethnocentrism. Nine of the 10 items in the surveywere taken from the 17-item scale of Consumers'Ethnocentric Tendencies (CETSCALE) devel-oped by Shimp and Sharma (1987). These nineitems correspond to numbers 1, 3, 5, 8, 11,13, 15,16, and 17 of the original scale. A 10th item , extentof agreement or disagreement with "Buy Americafirst" (used by Levin, Jasper, Mittelstaedt, &Gaeth, 1993), w as also included in the survey. T hesum of all 10 items defined our nationalism score;scores could range from 10 to 70. Cronb ach's alphameasured for this index w as .91.
In an earlier application (Levin, Johnson, &Jasper, 1993), this derived scale w as validated for apopulation comparable to the present one byshowing that the nationalism score w as signifi-cantly higher for owners of Am erican cars than forowners of foreign cars. In fact, the same relation-ship held for Canadians when the appropriatesubstitution of the word Canadian for Americanw as made on all scale items.
For the purposes of the present experiment,participants were divided into three groups on thebasis of nationalism score. Scores for the low
nationalism group (n = 20) ranged from 18 to 34;
scores for the medium nationalism group (n - 22)
ranged from 35 to 45; and scores for the highnationalism group (n = 21 ) ranged from 46 to 67.
Results
The main data are the preference ratings for
each paired comparison. These ratings were ana-lyzed statistically to determine w hether each nation-alism group had a significant preference for one ofthe alternative s in a pair (i.e., their ratings differedfrom neutral, 3.5, the midpoint on our scale) andto assess whether the three nationalism groupsdiffered from each other. Table 2 gives the meanpreference rating for each choice shown separatelyfor low, medium, and high nationalism groups.Table 3 provides counts of the most frequentlyreported reasons for making choices and helpsexplain the pattern of preference ratings observedin Table 2.
We can draw tw o important conclusions from
the preference data. The first is that participants at
all levels of nationalism w ere eq ually responsive tothe percentage of American workers employed,strongly favoring companies that employed agreater percentage of American workers. This is
clearly shown in the two comparisons w here com-pany nationality w as constant and percentage ofAmerican w orkers varied. In each choice situation,
the mean preference rating for each group w as
Table 2Mean Paired Comparison Preference Ratings for Each Nationalism Group
in Experiment 1
Type ofcomparison
Nationalism group
Low Medium High
% Americans constant and country variable
80% American vs. 80% Japanese 3.0050% American vs. 50% Japanese 3.20
2.953.05
1.52**1.88**
% Country constant, Am ericans variable
80% American vs. 50% American 1.78** 2.00**80% Japanese vs. 50% Japanese 1.90** 2.09**
1.67**1.79**
i Country and Americans variable
80% American vs. 50% Japanese50% American vs. 80% Japanese
2.55*4.65**
2.41**4.50**
1.33**3.33
Note. Ratings were scored from 1 to 6, with lower numb ers representing preference for the brandlisted first in each comparison. Neutra l rating = 3.5 on scale.* p < .05 when compared with neutral. **p < .01 when compared with neutral.
8/8/2019 Levin and Jasper
http://slidepdf.com/reader/full/levin-and-jasper 5/14
NATIONALISM IN CONSUMER DECISIONS 21
Table 3Most CommonReasons for Choices in Experiment 1
Response
category
Employ Americans
American made (nofurther justification)
American made
(support America)
American economy
Japanese quality
American(80%)vs.American
(50%)
LN MN HN
16 17 19
0 1 0
0 0 11 3 10 0 0
Japanese(80%) vs.Japanese
(50%)
LN MN HN
19 20 21
0 0 0
0 0 0
2 0 40 0 0
American(80%) vs.
Japanese(80%)
LN MN HN
0 0 0
6 7 7
3 5 9
3 1 7
5 7 0
American(50%) vs.
Japanese(50%)
LN MN HN
0 0 0
8 8 7
0 3 5
4 4 105 5 0
American(80%) vs.
Japanese(50%)
LN MN HN
12 15 17
5 3 9
0 3 14 3 44 2 0
American(50%) vs.
Japanese(80%)
LN MN HN
15 16 6
2 1 4
0 0 2
4 1 6
3 5 0
Note. Some respondents indicated tw o reasons for a choice. Both reasons were included in the above tabulation. LN = lownationalism; MN = medium nationalism; HN = high nationalism.
significantly different from neutral in the direction
of preference for the higher percentage of Ameri-
can workers, and the groups did not differ signifi-
cantly from each other. Moreover, "employ Ameri-
cans" was cited as a reason often and with equal
regularity by all three nationalism groups.
The second conclusion is that those participants
classified as high on nationalism were more respon-
sive to American companies in their choices than
were those medium and low on nationalism. As
seen in Table 2, the high nationalism group was the
only group to show a significant preference, a
preference for American companies, when percent-
age of American workers was constant and com-
pany nationality varied. The ratings for this group,
in fact, were significantly different from those
of the medium and low nationalism groups
combined, both when percentage of American
workers = 80 and when percentage of American
workers = 50,F(l, 61) = 22.18 and 26.52, respec-
tively,/? < .01, in each case.
The reasons for these choices also differed
among groups. "Japanese quality" wasmentioned
22 times by the low and medium groups and notonce by the high nationalism group. Instead, indi-
viduals high on nationalism were concerned about
supporting the United States and keeping money
out of the hands of foreign competition. "Ameri-
can made" and "American economy," as indicated
in Table 3, were cited much more frequently by
this group than by the other two groups. Interest-
ingly, "American quality" was seldom mentioned
as a reason in these or, for that matter, in any of
the choices.
Thus far,wehave discussed only those compari-
sons that varied one cue while holding the other
constant. The same effects identified above were
also evidenced in the remaining choice pairs where
participants were forced to weight and integrate
the two cues. When the cues were "congruent," in
that the American company employed a higher
percentage of American workers than the Japa-
nese company, all groups (but especially the high
nationalism group) preferred the American brand
and cited "employ Americans" as the most com-
mon reason. However, when the cues were "incon-
gruent," in that the American company employed
the lower percentage of American workers, the
high nationalism group was the only group not to
prefer the Japanese brand and was the only group
to cite "American economy" and "American-
made" as often as "employ Americans." The
difference between the ratings of these partici-
pants and those medium and low on nationalism
combined in both situations was again statistically
significant, F(l, 61) = 18.15 and 12.89, respec-tively,/) < .01, in each case.
While interesting in their own right, the above
results can be aggregated at a more formal level.
Information integration theory (Anderson, 1981)
provides a model-based analysis of how national-
ism influenced the tradeoffs that were made be-
tween company nationality and the percentage of
American workers employed. The procedure is
8/8/2019 Levin and Jasper
http://slidepdf.com/reader/full/levin-and-jasper 6/14
22 LEVIN AND JASPER
relatively simple and involves estimating the impor-
tance assigned to the two cues by each nationalism
group. As described more fully in the Appendix
and as shown in Table 4, its results can be
summarized as follows: (a) The weight given topercentage of American workers was large and
approximately equal across the three nationalism
groups, (b) For those individuals high on national-
ism, the two cues, percentage of American workers
and company nationality, were of about equal
importance, (c) For those low and medium on
nationalism, company nationalitywas of little or no
consequence.
Discussion
The unique feature of Experiment 1 is thatcountry-of-origin information, long thought to be
important to consumers but heretofore considered
as unidimensional, was divided into likely compo-
nents reflecting the character of today's multina-
tional or hybrid product. Our experimental design
and analyses allowed us to separate and assess for
the first time the independent effects of company
nationality and employment of American workers
and to compare the effects for individuals differing
on consumer nationalism-ethnocentrism. Asking
participants to give reasons following each choice
provided us with additional insights into these
matters.We had anticipated that those respondents scor-
ing high on nationalism would be more sensitive to
American employment needs than those scoring
low. This was indeed the case in earlier work
(Levin &Jasper, 1995) when percentage of Ameri-
can workers was combined with additional price
an d quality cues; in fact, we found that fo r high
nationalism respondents, an d only for these respon-
Table 4
Comparison of E f f e c t Sizes Across Nationalism
Groups in Experiment 1
Nationalism group
Effect
Company nationality(Equation A3)
Percentage of Americanworkers (Equation A4)
Low
0.20
2.10
Medium
0.09
2.09
High
2.34
2.00
Note. Appendix provides detailed description of how entrieswere derived.
dents, percentage of American workers became
more important as they progressed from a large
number of purchase options to a final choice.
Nevertheless, in Experiment 1, when percentage
of American workers was pitted directly againstcompany nationality and no other cues were of-
fered, employing American workers was an impor-
tant factor in the preference ratings of respondents
across all levels of nationalism.
Our second conjecture was that those scoring
high on nationalism would be more responsive to
American companies. This assertion was sup-
ported. Interestingly, high nationalism respon-
dents not only assigned greater importance to the
company nationality cue than did other respon-
dents, but they were also at least as responsive to
the distinction between American and Japanesebrands as they were to a change between 50% and
80% American workers. This was especially evi-
dent in the two choices varying both factors.
Company nationality "added to" percentage of
American workers in one case and "canceled out"
percentage of American workers in the other.
Finally, we hypothesized that those high on
nationalism would be more likely to acknowledge
"American quality" as a reason for preferring
American brands. Clearly this was not the case.
American quality was rarely mentioned; rather,
those high on nationalism were motivated in their
choices by a desire to support America and theAmerican economy as well as American workers
and were apt to cite "American made" as the
reason for their choices without need for furtherjustification. "Japanese quality,"on the other hand,
w as apt to be given as a reason for choices by low
nationalism respondents.
Experiment 2
Of course, several questions naturally arose
concerning the generality of results found in Experi-
ment 1. For example, would the effects of companynationality and percentage of American workers
on consumer preference be the same for products
other than automobiles and company nationalities
other than American and Japanese? Also, would
American consumer nationalism as measured by
our scale play the same role with other products
and with other countries? To answer these ques-
tions, additional countries and products represent-
ing varying levels of departure from American
8/8/2019 Levin and Jasper
http://slidepdf.com/reader/full/levin-and-jasper 7/14
NATIONALISM IN CONSUMER DECISIONS 23
versus Japanese automobiles were selected and
used in Experiment 2.Specifically, participants were asked to choose
between American andJapanese personal comput-
ers, Am erican and German automobiles, or Ameri-can and Taiw anese clothing. Thus, the American-
Japanese comparison w as extended to a newproduct, the American-foreign car comparisonw as extended to a new country, and, finally, a less
developed country w as introduced in a realisticproduct setting. On the basis of the findings from
Experiment 1, we hypothesized that across allcomparisons low and high nationalism respon-dents would be equally sensitive to Americanemployment needs, but that high nationalism re-spondentswould be more concerned than lowwith
supporting American companies. In addition, onthe basis of earlier work showing more favorableevaluations of products from countries with greatereconomic development (Bilkey & Nes, 1982; Gae-deke, 1973), w e hypothesized that for both low andhigh nationalism respondents perceived differ-
ences in q uality would favor American companiesin American-T aiwanese choices more than inother choices. We thought, for example, thatparticipants in this condition m ight be more apt tomention "American quality" as a reason whenchoosing an A merican over a T aiwanese company.
Method
A different group of participants from the samepopulation as Experiment 1 served as participantsfor each product-country combination: n = 40 forAmerican versus Japanese personal computers(PCs); n = 37 for American versus German auto-mobiles; and n = 39 for American versus T aiwan-ese clothing. For each group of participants, the
tasks, procedure, and instructions were exactly thesame as in Experiment 1, except for appropriatesubstitution of product and country names.
Selection of participants varying in consumerethnocentrism, however, w as different in Experi-ment 2 than in Experiment 1. Our scale w asadministered to 758 students from elementarypsychology courses in a group testing session thatincluded a variety of other scales and surveys.Students w ho scored in the upp er or bottom thirdson the nationalism scale were later recruited for
the experiment. These constituted our high andlow nationalism groups; the middle level was not
included in Experiment 2 because responses inExperiment 1 by low and medium level partici-pants were no t appreciably different.
2
Results
Table 5 shows the mean preference rating foreach choice in Experiment 2, as well as thepreference ratings for the high and low national-
ism groups from Experiment 1. This allows acomparison between the various products andcountries from both experiments. As in Experi-ment 1, the preference data w ere analyzed statisti-cally to determine w hether each nationalism grouphad a significant preference for one of the alterna-tives in a pair and to assess whether the high and
low nationalism groups for each product-countrycombination differed from each other.The results shown in Table 5 can be summarized
as follows: (a) Preference for companies employ-ing a higher percentage of American w orkers w asfound for all products, all country pairings, and forparticipants scoring both high and low on thenationalism scale (but more so for those scoringhigh), (b ) Preference for American companies w asfound primarily in the high nationalism group, butw as also significant for the low nationalism groupconsidering American versus Taiwanese clothing,(c) Consumer nationalism had its greatest impact
on preferences for Am erican companies over Japa-nese companies; in particular, only in the American-Japanese car condition w as there a significantdifference between high and low nationalism groupswhen considering the tradeoff between an Ameri-can company employing a lower percentage ofAmerican workers and a non-American companyemploying a higher percentage of American workers.
Paralleling Experiment 1, a comparison w as alsomade of effect sizes across groups following thederivations in the Appendix. T he relevant values
2W e recognize that the different ways of classifying
participants in Experiments 1 and 2 could lead todifferential arousal of nationalistic feelings that in turncould affect scores on the scale. In Experiment 1, thenationalism scale is given after participants m ake choicesbetween American and Japanese cars; in Experiment2,the scale is given several week s before the experimentaltask. Nevertheless, as we will show, the two experimentsproduce sim ilar results that do not appear to be affected
by whether the scale w as administered first or last.
8/8/2019 Levin and Jasper
http://slidepdf.com/reader/full/levin-and-jasper 8/14
24 LEVIN AND JASPER
Table 5
Mean Paired-Comparison Preference Ratings in Experiments 1 and 2 for Cars,
Personal Computers (PCs), and Clothing
American-Japanese American-Japanese American-German American-Taiwanese
cars PCs cars clothing
Type o f L N H N L N H N L N HN LN HN
comparison (n = 20 ) (n = 21 ) (n = 20) (« = 20) (n = 18) (n = 19) (n = 20) (n = 19)
% Am erican constant, country varies
American (80%)-
Non-American (80%)American (50%)-
Non-American (50%)
3.00
3.20
1.52*
1.88*
3.00
2.85
1.85*
2.10*
3.56
3.78
2.05*
237*
2.50*
2.85*
1.74*
1.68*
Country constant, % Americans varies
American (80%)-
American (50%)
Non-American (80%)-Non-American (50%)
1.78*
1.90*
1.67*
1.79*
2.60*
2.40*
1.70*
2.30*
2.22*
3.22
1.95*
2.00*
2.32*
2.60*
1.68*
1.84*
Both country and % Americans vary
American (80%)-Non-American (50%)
American (50%)-
Non-American (80%)
2.55*
4.65*
1.33*
3.33
2.80
4.00
1.70*
3.30
3.22
3.94
1.95*
3.89
2.20*
3.85
1.67*
3.95
Note. Bold type indicates significant difference between high and low nationalism groups,/) < .05. Neutral rating = 3.5. LN = low
nationalism; HN = high nationalism.* p < .05 wh en compared with neutral.
from Experiment 1, as well as the values from
Experiment 2, are given in Table 6. Results,supplementing those from Table 5, can b e summa-rized as follows: Although the effect of percentageof Am erican w orkers was generally high across allcountry and product comparisons, differences be-tween the high and low nationalism groups on thesize of the effect were greatest for American-German cars and least for American-Japanesecars and PCs. In contrast, differences between thehigh and low nationalism groups on the size of the
company nationality effect w ere greatest for Ameri-
can-Japanese comparisons. Among high natio nal-ism respondents, only those considering Am erican-Japanese products gave at least as much w eight tocompany nationality as to percentage of Americanworkers.
Again, the reasons for choices help explain thedata in Tables 5 and 6. Table 7 provides frequencycounts of the most common reasons given by eachgroup for each paired comparison in Experiment2, and Table 8 summarizes the reasons data for
Table 6
Comparison of Effect SizesAcross Nationalism Groups in Experiments 1 and 2 for Cars,Personal Computers (PCs), and Clothing
American-Japanese American-Japanese American-German American-Taiwanese
cars PCs cars clothing
Effect
Company nationalityAmerican workers (%)
LN
(n = 20)
0.202.10
HN
(n = 21)
2.342.00
LN
(n = 20)
0.201.20
HN
(n = 20)
2.001.60
LN
(n = 18)
0.160.72
HN
(n = 19)
1.171.94
LN
(n = 20)
0.951.65
HN
(n = 19 )
1.382.28
Note. LN = lownationalism; HN = high nationalism.
8/8/2019 Levin and Jasper
http://slidepdf.com/reader/full/levin-and-jasper 9/14
NATIONALISM IN CONSUMER DECISIONS 25
Table 7
Most CommonReasons fo r Choices in Experiment 2
Response
category
American(80%) vs.
American(50%)
LN HN
Japanese(80%) vs.
Japanese(50%)
LN HN
American vs. Japanese
Employ AmericansAmerican made (no further
justification)American made (support
America)American economyJapanese quality
14 19
0 6
0 0
3 20 0
American
(80%) vs.American
(50%)
LN HN
13 17
0 0
0 0
3 0
1 1
German
(80%) vs.German(50%)
LN HN
American(80%) vs.
Japanese(80%)
LN HN
American(50%) vs.
Japanese(50%)
LN HN
American(80%) vs.
Japanese(50%)
LN HN
American(50%) vs.
Japanese(80%)
LN HN
personal computers
0 0
4 11
3 1
3 5
3 1
American
(80%) vs.German(80%)
LN HN
0 0
5 9
4 2
2 54 1
American
(50%) vs.German(50%)
LN HN
13 15
3 14
1 1
3 1
3 1
American
(80%) vs.German(50%)
LN HN
8 9
2 4
1 2
3 4
4 1
American
(50%) vs.German(80%)
LN HN
Ame rican vs. German cars
Employ AmericansAmerican made (no further
justification)American made (support
America)American economyGerman quality
9 17
1 4
0 03 20 0
American(80%) vs.American
(50%)
LN HN
12 18
0 0
0 01 2
3 0
Taiwanese(80%) vs.
Taiwanese(50%)
LN HN
0 0
4 8
1 50 3
9 1
American(80%) vs.
Taiwanese(80%)
LN HN
0 0
2 9
3 41 38 1
American(50%) vs.
Taiwanese(50%)
LN HN
6 12
2 8
2 42 48 0
American(80%) vs.
Taiwanese(50%)
LN HN
7 14
3 3
0 1
1 5
7 1
American(50%) vs.
Taiwanese(80%)
LN HN
American vs. Taiwanese clothing
Employ AmericansAmerican m ade (no further
justification)American made (support
America)American economy
17 19
1 1
0 0
5 3
16 17
0 0
0 00 2
0 0
8 11
1 45 4
0 0
9 10
2 64 4
15 15
5 9
1 2
5 2
15 13
2 3
0 1
1 5
Note. LN = lownationalism; HN = high nationalism.
both experiments. Together these data show that
"employ Americans" was the most frequently
given reason for choices followed in order by
"American made," "American economy," and
"American made (support America)." With a few
notable exceptions, these reasons were given more
often by those scoring high on the nationalism
scale than by those scoring low. On the other hand,
"foreign quality" was cited more often by low
nationalism participants than by high, as was
"American quality," even though this was not a
high frequency response for any group. Quality
was mentioned only once in the American-
Taiwanese clothing condition.
8/8/2019 Levin and Jasper
http://slidepdf.com/reader/full/levin-and-jasper 10/14
26 L EVIN AN D JASPER
aI
o
№g 8'S 1€ §.
p
1
3?S
_P
§ 7£,oo
1
3 V
a n
Rep
cey
3
S
a
rt 00 (S 00 O 00
3 S
£ §-O O .-c
-HO 00
00O-H
oo moo
oo "•> •* ON
O I/I O <S O
I1&iiIf
S S00 O I- c«1 O O
LN =ow no
8/8/2019 Levin and Jasper
http://slidepdf.com/reader/full/levin-and-jasper 11/14
NATIONALISM IN CONSUMER DECISIONS 27
A convenient w ay of seeing how the reasons help
to elucidate the choices, especially when they
differ across conditions, is to compare the sum-
mary of effect sizes (see Table 6) with the summary
of reasons data (see Table 8). This comparisonleads to a number of intriguing observations. First,
the fact that the percentage of American workers
effect w as generally larger than the company
nationality effect is mirrored by the fact that
"employ Americans" was the most frequent rea-
son for choices in all groups. Second, the percent-
age of American workers effect w as generally
greater for high nationalism than for low national-
ism respondents, with the exception that it was aslarge for low as for high nationalism respondents
considering American-Japanese cars. "Employ
Americans" as a reason for choices follows asimilar pattern (except for American-Taiwanese
clothing). Third, the company nationality effect
w as greater for high than for low nationalism
respondents in all conditions. This parallels the
fact that "American made" and "American
economy" were cited much more frequently byhigh than by low nationalism respondents. How-
ever, there were of course a few exceptions.
Finally, the company nationality effect exhibited
by low nationalism respondents w as generally quite
low, except in the American-Taiwanese clothing
condition. This w as also the only condition inwhich low nationalism respondents failed to cite
foreign quality as a reason for their choices. Thus,
where company nationality has a reduced effect, itseems to be due in part to the perception of
superior quality of the products made by foreigncompanies.
Discussion
Experiment 2 was conducted largely to test thegenerality of the results from Experiment 1, across
variations in country and product comparisons. Ingeneral, Experiment 2 replicated Experiment 1,and, for the most part, the hypotheses were sup-
ported. For all country and product comparisons,
those scoring high on the nationalism scale were
more apt than those scoring low to favor American
over foreign companies and to cite "American
made" and "American economy" as their reasons.
Also, across groups there was a reliable tendency
to prefer companies that employed a higher per-
centage of American workers and to give "employ
Americans" as the most frequent reason for choice.
However, unlike Experiment 1, Experiment 2revealed differences between high and low nation-
alism respondents in the size of the percentage ofAmerican workers effect. Although the size of thepercentage of American workers effect was thesame for high and low nationalism respondents
choosing between American and Japanese cars, it
w as greater for high nationalism respondents than
for low nationalism respondents in other condi-
tions. One likely reason for this is that mass media
attention to foreign competition for jobs has cen-
tered particularlyon American- versus Japanese-
made cars, thus sensitizing even low nationalism
individuals to this issue.
Finally, the reasons data show a differencebetween American-Taiwanese and other compari-
sons. Whereas lownationalism respondents were
apt to cite superior foreign quality for American-
Japanese and American-German comparisons,
such was not the case for American-Taiwanese
comparisons. These results are consistent withHong and Wyer's (1989) observation that a coun-
try's reputation can greatly affect inferences ofproduct quality. However, it should be noted that
although persons in the American-Taiwanese con-
dition did not cite superior foreign quality, they
also did not cite superior American quality.
General Discussion
Previous research in this area has demonstrated
that as a unified cue country-of-origin plays animportant role in consumer choice. In hybridproducts, however, country-of-origin is itself amulticomponent factor. In this study, w e focused
on two likely components: company nationality
and the employment of American workers. Byasking participants to choose between products
varying on these cues, w e were able for the first
time to separate and assess their independenteffects. By using different country and product
combinations across experiments and by asking
participants to provide reasons for their choices,
w e were able to achieve a deeper understandingofnationalistic tendencies in consumer decision pro-cesses.
In general, our results support the contention of
Shimp and Sharma (1987) that American con-
sumer ethnocentrism is an important construct
8/8/2019 Levin and Jasper
http://slidepdf.com/reader/full/levin-and-jasper 12/14
28 LEVIN AND JASPER
underlying responsiveness to country-of-origin cues.However, differences in response as a function ofconsumer nationalism or ethnocentrism dependon whether company nationality or the employ-
ment of Americans is mentioned, and they alsovary across countries and, to a lesser extent, acrossproducts. For example, media attention to compe-tition from Japan, especially in the automobileindustry, appears to have brought ou t increasedconcern for the plight of the American workerwhen choosing between American and Japaneseproducts. In a similar vein, concern among someconsumers about the superiority of foreign prod-ucts applies to Japanese and German but not to
Taiwanese products. Shimp and Sharma's conclu-sion that consumers scoring high on their scale are
more sensitive to country-of-origin information isthus subject to the caveat that there may bedifferences in consumer choices and the reasonsfor those choices as a function of the specificproducts and countries being compared.
W e also found that the reasons for specificchoices vary as a function of the type of cuemanipulated within choice options. When theoptions withina pair differ in percentage of Ameri-can workers employed, reasons for choice centeraround the desirability of employing Americans.When the options within a pair differ in companynationality, reasons for choosing American or
non-American companies depended on the particu-lar non-American country. In the American-Japanese and American-German comparisons,forexample, reasons for choosing the American com-pany relate to supporting the American economywhile reasons for choosing the foreign companyrelate to perceived quality differences. This w asnot the case for American-Taiwanese compari-sons where the perception of superior foreignquality w as absent.
Of course, it can be argued that participants inthe present study were making only hypothetical
choices with unrealistic information. Interestingly,however, since w e started this line of research,laws have been enacted to provide new car buyerswith more complete information about the multina-tional character of the product being considered.Furthermore, we now have evidence that scores onour nationalism-ethnocentrism scale are relatedto both hypothetical and real choices (Levin,Johnson, & Jasper, 1993) and to choices withvarying numbers of cues (Levin & Jasper, 1995),
thus lending credence to the present findings.
Taken together, these findings seem to suggest thatcurrent practices that inform potential consumersthat a product, even on e from a foreign company,
was made by a high percentage of Americanworkers are extremely effective and, from a market-ing standpoint, should continue.
At a more general level, the present results haveimplications for all forms of communication inwhich considerations of nationalism may play arole, including political messages as well as con-sumer reports. For example, politicians wanting to
promote "buy American first" should stress boththe choice of American companies and the employ-ment of American workers, and they should beaware that appeals have different effects on differ-
ent segments of the population.In closing, w e must emphasize that methodolo-
gies that include additional measures beyond purechoices, rankings or ratings appear to provide a
more complete understanding of consumers' pref-erences and decisions. By including in our designboth the concurrent generation of reasons for eachpreference and an a priori measure of individualdifferences, w e were able to gain insight into theprocesses and motives underlying consumers' reac-tions to new hybrid, multinational products.
References
Anderson, N. H. (1981). Foundations of information
integration theory. Ne w York: Academic Press.
Bilkey, W. J., &Nes, E. (1982). Country-of-origin effects
on product evaluation. Journal of International Busi-
ness Studies,13 , 89-99.
Birnbaum, M. H., Wong, R., & Wong, L. K. (1976).
Combining information from sources that vary in
credibility.Memory & Cognition, 4, 330-336.
Ettenson, R., & Gaeth, G. (1991). Consumer percep-
tions of hybrid (bi-national) products. Journal of
Consumer Marketing, 8, 13-18.
Gaedeke, R. (1973). Consumer attitudes toward prod-
ucts made in developing countries. Journal of Retail-ing, 49, 13-24.
Han, C. M. (1989). Halo or summary construct? Journal
of Marketing Research, 26, 222-229.
Hong, S., & Wyer, R. S. (1989). Effects of country-of-
origin and product-attribute information on product
evaluation: An information processing perspective.
Journal of Consumer Research, 16, 175-187.
Hong, S., &Wyer, R. S. (1990). Determinantsof product
evaluation: Effects of the time interval between knowl-
8/8/2019 Levin and Jasper
http://slidepdf.com/reader/full/levin-and-jasper 13/14
NATIONALISM IN CONSUMER DECISIONS 29
edge of a product's country of origin and information
about its specific attributes. Journal of Consumer
Research, 17 , 277-288.
Levin, I. P., & Gaeth, G. J. (1988). Howconsumer are
affected by the framing of attribute information be-fore and after consuming the product. Journal of
Consumer Research, 15 , 374-378.
Levin, I. P., &Jasper, J. D. (1995). Phased narrowing:A
new process tracing method for decision making.
Organizational Behavior and Human Decision Pro-
cesses, 64, 1-8.
Levin, I. P., Jasper, J. D., Mittelstaedt, J. D., & Gaeth,
G. J. (1993). Attitudes toward "Buy America First"
an d preferences for American and Japanese cars: A
different role for country of origin information. Ad -
vances in Consumer Research, 20, 625-629.
Levin, I. P., Johnson, J. D., & Jasper, J. D. (1993,
December). The role of nationalism in the consumer
choice process: Comparisons between the U.S. and
Canada. Proceedings of the Fourth Symposium on
Cross-Cultural Consumer and Business Studies, Ka-
huku (Oahu), Hawaii.
Levin, I. P., Kim, K. J., &Cony, F. A. (1976). Invariance
of the weight parameter in information integration.
Memory & Cognition, 4, 43-47.
Obermiller,C, &Spangenberg, E. (1989).Exploring the
effects of country of origin labels: An informationprocessing framework. Advances in Consumer Re-
search, 16 , 454-459.
Schooler, R. D. (1971). Bias phenomena attendant to
the marketingof foreign goods in the U.S. Journal of
International Business Studies, 4, 71-80.
Shanteau, J. (1988). Consumer impression formation:
The integration of visual and verbal information. In S.
Hecker & D. W. Stewart (Eds.), Nonverbal communi-
cation in advertising (pp. 42-57). Lexington, MA:
Heath.
Shimp, T. A., & Sharma, S. (1987). Consumer ethnocen-
trism: Construction and validation of the CETSCALE.
Journal of Marketing Research, 24, 280-289.
Troutman, C. M., & Shanteau, J. (1976). Do consumers
evaluate products by adding or averaging attribute
information? Journal of Consumer Research, 3,101-106.
Appendix follows on next page.
8/8/2019 Levin and Jasper
http://slidepdf.com/reader/full/levin-and-jasper 14/14
30 LEVIN AND JASPER
Appendix
Averaging Model
The following averaging model assumes thateach respondent assigns separate subjective scale
values to American companies and Japanese com-
panies (vamer
and Vjp n
, respectively) and to compa-
nies that employ 80% American workers and 50%
American workers (vgo
and v50
, respectively). In
evaluating a brand described by both its national
origin and the percentage of American workers
employed, the integrated impression R is de-
scribed as
R =
w%v
%
w%
where vm
and v%
represent the subjective scale
values for company nationality and the percentage
of American workers, respectively, andwm
and w%
represent the importance weights assigned to the
two cues.
The model can be applied to each of the six
comparisons employed in the experiment by taking
the difference between the average ratings for the
two options in a pair. The comparisons in which
both factors vary are illustrated below. For Ameri-can (80%) versus Japanese (50%),
_ "Warner + H^VgQ _ WmV'
№n+ W
%V
50
-w %
~VJ
Pn)
Wco
(Al)
For American (50%) versus Japanese (80%),
+ W%V80
~Vipn) ~V
80)
. (A2)
percentage of American workers cancels out, andwe obtain
^ '
B y subtracting Equations Al and A2, the term
fo r company nationality cancels out and we obtain
2w%(v
80- v
50)
WC
(A4)
B y adding Equations Al and A2, the term for
In the present case, a weighted sum model
(Equations Al and A2 without the denominatorterms) would have served the same purpose as the
weighted average model used here. However, we
chose the averaging form because of previous
support of averaging over adding in similar situa-
tions that included formal model tests (Anderson,
1991; Birnbaum, Wong, & Wong, 1976; Levin &
Gaeth, 1988; Shanteau, 1988;Troutman & Shan-
teau, 1976). In either form of the model, the actual
importance of a factor is the product of its weight
and the difference in scale values across levels of
that factor (Levin, Kim, & Corry, 1976). Equation
A3 is a constant times the importance of company
nationality; Equation A4 is the same constant
times the importance of the percentage of Ameri-
ca n workers. Thus, Equations A3 and A4 provide
the means with which to compare the relative
importance of each factor across the three nation-
alism groups. The results are shown in Table 4 with
the data from Table 2expressed as absolute differ-
ences from the neutral value on the rating scale.
For example, consider the data in the bottom
two rows of Table 2 for the high nationalism group.
RI + R2, expressed as deviations from 3.5, the
neutral point on the scale, equals (3.5 - 1.33)+
(3.5 - 3.33), which equals 2.34. /? t - R2, alsoexpressed as deviations from 3.5, equals
(3.5 - 1.33) - (3.5- 3.33) which equals 2.00.Because th e unit of this scale is arbitrary, th e
proportionality constant can be assigned the value
of 1, and the above resultants can be taken as the
desired effect sizes.
Received October 14,1994
Revision received July 21,1995
Accepted August 11,1995 •