“The nation which indulges toward another in an habitual hatred or an habitual fondness is in some degree a slave. It is a slave to its animosity or to its affection… The government sometimes participates in the national propensity, and adopts through passion what reason would reject; at other times it makes the animosity of the nation subservient to projects of hostility instigated by pride, ambition, and other sinister and pernicious motives.”

George Washington, US President (1776)

Animosity, defined as anger due to “previous or ongoing political, military, economic, or diplomatic events” (Klein et al. 1998, p. 90), is a concern for many nations (PEW 2007). Besides the wide range of political externalities associated with animosity, research has demonstrated that it affects consumer preferences and choices (Klein 2002). Surprisingly, research to date has not addressed whether and how animosity toward a country negatively affects consumers’ responses to brands associated with that country. Because brands vary in the degree to which they are associated with a country, animosity might have different effects on brands that are strongly stereotypic of a country and on brands that are more weakly associated with it. This research draws from the literature on prejudice and discrimination to assess whether the link between animosity and consumers’ attitudes and behaviors toward brands depends on the strength of stereotypic brand–country associations (BCAs). This proposed moderating effect is tested in two studies. In the first study, the perceived strength of stereotypic BCAs is measured for a series of existing brands. In the second study, a new brand's level of stereotypic association to a country is manipulated using different advertising messages and consumers' subsequent brand attitude and choice are measured.

1 Literature review

1.1 Country-of-origin and animosity effects on products and brands

Traditional country-of-origin (CoO) research has found that there is a relationship between consumers' image of a country and product judgments (e.g., Bilkey and Nes 1995; Gürhan-Canli and Maheswaran 2000; Maheswaran 1994). Focused more specifically on negative opinions of a country, the animosity model of foreign product purchase (Klein et al. 1998) and multiple studies since have documented that animosity negatively affects consumers’ willingness to buy a country's products (e.g., Klein 2002; Nijssen and Douglas 2004; Russell and Russell 2006). Animosity research has generally focused on the link between animosity and behavior, showing the effect of feelings of anger on purchase decisions and actual ownership of consumer durables. The finding in that context is that angry consumers do not denigrate a country's product; they simply do not want to purchase it even if the product itself is held in high esteem.

Although the original model of animosity did not find effects on consumer judgments and attitudes, there is some empirical evidence that animosity also relates to consumers' evaluative responses. For instance, in a study assessing Israeli Jewish consumers' opinions about Israeli–Arab products, Shoham et al. (2006) found that animosity equally impacted product quality judgments and willingness to buy. In another study, animosity towards China was significantly related to Taiwanese consumers’ attitudes toward TV programs imported from China (Yang and Tso 2007). Therefore, it is possible that, for products other than durable goods, animosity effects would extend to attitudes toward products and brands from that country (Pappu et al. 2007).

1.2 Brand–country associations: stereotypes, prejudice, and discrimination

The process driving CoO and animosity effects has been stated in straightforward terms, such as brand-related country associations (Klein et al. 1998) or interrelationships between country image and brand image (Han 1989). The CoO of a product/brand has been viewed as a stereotype, a schema, or knowledge structure that serves as a way for consumers to categorize a product, and that in turn influences their product evaluations (Maheswaran 1994). When such a schema is activated (Liu and Johnson 2005) or when little or ambiguous information is available (Maheswaran 1994), CoO-based stereotypes, positive or negative, transfer to consumers' judgments of products and brands associated with that country. The literature to date is scant about the processes that underline this transfer or the factors that may moderate this transfer. A turn to the social psychology literature on category-based stereotypes provides a rich basis for assessing how CoO-based categorization might impact responses to brands, and in particular how negative views of a country, such as animosity, affect prejudice (evaluative responses) and discrimination (behavioral responses) toward brands associated with that country.

Stereotypes refer to the semantic associations about a particular group; they are activated, often automatically, in the presence of a member or a symbolic equivalent of the stereotyped group (Blair 2002). Whether prejudice ensues, in the form of attitudinal responses toward a member or exemplar of the group, depends on people's personal beliefs toward the group (Devine 1989). Negative stereotypes and the emotions that they generate have also been linked to discriminatory behavioral tendencies (Cuddy et al. 2007). For instance, the research performed by Cuddy et al. (2007) demonstrated that anger toward a given group was directly related to active harm toward members of that group. Even though their research did not focus on countries, target groups included religious groups (e.g., Muslims) or cultural groups (e.g., Asians) and the anger component of their framework maps on directly to the definition of country animosity.

The extant literature on prejudice and discrimination thus suggests that if consumers harbor animosity toward a particular country, they are likely to express prejudice against brands that are stereotypic of the country (i.e., their attitudes toward the brand would be more negative), and to discriminate against those brands by not consuming them.

1.3 The moderating role of the strength of stereotypic associations

Research has shown that prejudice and discrimination are driven not just by the existence of a stereotype but also by the degree to which a particular object is associated with the category. For instance, in the context of racial stereotypes, prejudice, which reflects attitudinal responses, does not apply equally to all members of a category. Livingston and Brewer (2002) found that Black Americans who had stronger Negroid facial features triggered more automatic negative evaluations than those with less Negroid features. Another characteristic of stereotypes is that they are activated with little cognitive activity and irrespective of their accuracy (Cunningham et al. 2004). Thus, a prejudiced response might be triggered if a particular object or person is perceived to be a member of a given category, even when that is in fact not true (Smith and DeCoster 2000).

In the context of animosity toward a country, this suggests the need to account for the strength of the stereotypic association between brand and country, a distinction that has not been addressed in previous research. Some brands are iconic, having a strong stereotypic association with their CoO (Holt 2004), but others may have weaker linkages to their CoO. Complicating this issue is the fact that consumers are often oblivious to brands' actual origins and may in fact have incorrect perceptions of where a brand originates (Samiee et al. 2005). In their study of a large sample of US consumers, Samiee et al. (2005) found that, on average, respondents correctly identified the origins of only one-third of the 84 brands they tested. BCAs can be affected by cues such as a brand name's foreign pronunciation or spelling (Leclerc et al. 1994). Thus researchers must account for consumers' perceptions of the origin of brands, which may be different from their actual origin.

2 Research questions

The extant research suggests that animosity toward a country should negatively affect consumers’ opinions of brands associated with that country (Klein et al. 1998; Pappu et al. 2007). However, based on social–psychological research on prejudice and discrimination, this relationship should depend on the degree to which a brand activates the country representation (i.e., the strength of the stereotypic BCA; Livingston and Brewer 2002). Brands that are strongly stereotypical of a country would suffer more from any animosity that consumers might harbor toward that country than brands that are less stereotypical of that country.

Addressing limitations of previous research where the CoO of products is made salient to respondents with survey questions specifically stating the country (e.g., “I do not like the idea of owning Japanese products”) or experimental manipulations that openly indicate the origin of a hypothetical brand, the level of stereotypic BCA is measured in study 1 and then manipulated in study 2. The first study uses cross-sectional data to test the link between country animosity and consumers’ attitudes toward brands that vary in terms of stereotypic BCA. The second study experimentally manipulates the strength of the brand–country stereotypic association to provide causal evidence of its moderating effect on the linkage between animosity and consumers’ attitudinal and behavioral responses to a brand.

3 Study 1

3.1 Hypotheses

Because brands vary in the degree to which they activate a country representation, it is proposed that the relationship between animosity toward a country and attitudes toward brands is moderated by the strength of the stereotypic BCA. In other words, animosity towards a country should be directly and negatively related to attitudes toward brands that are strongly stereotypical of that country. Animosity toward a country should not relate to attitudes toward brands if they are weakly associated with that country. For brands that have intermediate levels of association with the country, the relationship between country animosity and brand attitudes should be moderated by the strength of the BCA.

3.2 Research context

The context for the research is the US. Many US businesses have recognized that anti-Americanism is a profound concern (Kuehn 2005) and indeed anti-American feelings negatively impact US international business interests across many sectors (O’Brien 2005). Polls indicate that, across the globe, consumers' opinions of American multinational brands are shaped by US foreign policy (GMI 2005) but academic research to date has not assessed this relationship.

The studies were conducted in France, a country where anti-Americanism is the most pronounced of any allied nation (Lacorne 2005) and where it extends beyond the general construct of animosity to include cultural and institutional components (Roger 2005). France's animosity towards the US gained prominence in the mid-1850s when the concept of Americanism and its negative connotations manifested among the French elite (Lacorne 2005). This form of animosity became even more prevalent during the last century as the French elite perceived US cultural products, especially movies, as being mass-produced for uncultivated audiences and as a threat to local culture (Kadmi-Cohen 1930).

Criteria were established to select a set of brands in a range of categories where consumers have a choice (e.g., no monopolies). The set would include both leading European and US brands that would be familiar to the population. And of course, brands should vary in the strength of their stereotypic association with the US, in other words, in their level of perceived Americanness (GMI 2005, Samiee et al. 2005).

3.3 Method

In order to avoid contamination of responses by collecting consumption and animosity measures concurrently, data were collected at two points in time. In phase one, an invitation to participate in a “study about brands” was circulated electronically to 5,000 students at the two participating public universities. Of those, 731 (239 males) fully completed the survey, thus providing a response rate of 14.6%. In this first online survey, participants responded to a series of questions about 14 brands, which included half US and half non-US brands and which covered a range of product categories. For each brand, participants indicated their general attitude (overall how do you feel about brand X: 1 = very negative, 7 = very positive). The strength of the stereotypic BCA was measured in the second phase of the study, with respondents rating how much they perceived each brand as American (1 = not at all, 7 = very).

One month later, participants in the first wave were contacted again to partake in a “study of how French people feel about the US.” Participation was encouraged through the use of a retail gift certificate. A total of 291 (91 males) completed the survey, corresponding to a response rate of 39.8%. The survey included a measure of animosity toward the US (α = 0.76) adapted from the research of Klein et al. (1998) and previous studies of animosity toward the US (Russell and Russell 2006): respondents rated their degree of agreement with 12 statements (1 = not at all to 7 = very much), some of which reverse-scored (see Appendix).

3.4 Descriptive statistics

Animosity toward the US ranged from 1.5 to 7 with a mean of 4.74 (SD = 0.83), a distribution similar to previous animosity research (Klein 2002). As seen in the first column of Table 1, stereotypic BCAs varied in strength. Perceptions of Americanness of US brands were all above the scale mid-point (all ts(726) > 10.93, p < .05), and perceptions of Americanness of non-US brands were all below the scale mid-point (all ts(725) > 5.29, p < .05). Four US brands scored very high on perceived Americanness and were qualified as having a strong stereotypic US association, where a main effect of animosity is expected without an interaction. Four non-US brands scored low on perceived Americanness and can thus be qualified as having weak stereotypic US association, for which brand attitudes should be unrelated to animosity toward the US and also unaffected by its interaction with the strength of the BCA. For the remaining brands, the relationship between animosity and brand attitudes should be moderated by the strength of the stereotypic BCA (i.e., the brands' perceived Americanness).

Table 1 Multiple regression results - study 1

3.5 Relationship between animosity and brand attitudes

A series of stepwise multiple regressions was conducted for each brand on the brand attitude dependent measure. In the first step, animosity and strength of the BCA were entered to test the main effect hypotheses, and in the second step, the interaction term was added to test the moderation hypotheses. Results appear in Table 1. As expected, animosity is directly and negatively related to attitudes toward brands that are strongly stereotypic of the US, with the exception of Levi's. The interaction between animosity and strength of the BCA in the regressions for the four non-US brands with weak stereotypic US association is not significant, as predicted.

For the six middle-of-the range brands, the significance and direction of the interaction terms support the prediction that the relationship between animosity and brand attitudes is moderated by the strength of the stereotypic BCA. In two cases, a main effect of animosity is also present such that animosity in general is negatively related to opinions about these brands, in addition to the significant interaction with the strength of the BCA. Interestingly, the analyses also reveal a main effect in the negative direction of the strength of the BCA for Heinz and Nescafé, indicating that, the more consumers perceive these brands as American, the more negative they feel toward them.

3.6 Discussion

This first study provides empirical evidence that animosity toward a country is negatively related to attitudes toward brands that are strongly stereotypic of that country but not toward those that are weakly associated with the country. In line with theories of prejudice (Blair 2002), anti-American views harbored by French consumers are negatively related to their attitudes toward brands perceived as strongly stereotypically American. The strength of the stereotypic BCA moderates the relationship between animosity and brand attitudes: animosity toward the US has a stronger negative relationship with attitudes toward brands the more the brands are perceived as American and this effect occurs whether or not the brands actually are American.

4 Study 2

Building on the correlational findings of study 1, study 2 was conducted to provide causal evidence of the moderating role of the strength of the brand–country stereotypic association on the linkage between animosity and consumers' responses to a brand. It was designed to address some of the limitations of study 1, which relied on existing brands and their associations and a sample of university students. The focus was a brand without prior associations (i.e., an unknown brand in the market) and the level of BCA was manipulated experimentally using different advertisements. Expanding beyond attitudinal effects, study 2 also measures behavioral responses (brand choice) to assess whether animosity triggers discriminatory responses as well as prejudiced ones.

4.1 Hypotheses

Based on previous research on prejudice, animosity toward a country is expected to generate more negative attitudes toward a new brand if the brand is perceived as having a strong stereotypic association with the country but not if it is not associated with the country. In other words, the strength of the stereotypic BCA should moderate the relationship between animosity and attitudes toward a new brand. Furthermore, based on research on discrimination which has shown that group-based stereotypes can produce active and harmful behavioral reactions toward members of that group (Cuddy et al. 2007), people who harbor high animosity toward a country are expected to reject a new brand that is strongly associated with that country, in the form of lesser choice. Instead, a new brand that is weakly associated with a country may benefit from active facilitation from high-animosity consumers, in the form of increased choice (Cuddy et al. 2007). Thus, the strength of the stereotypic BCA should moderate the relationship between animosity and choice of a new brand such that there is no difference in choice amongst low-animosity consumers whether the BCA is strong or weak but there is lower choice when the BCA is strong and consumer harbor high animosity toward the country.

4.2 Method

The study was also conducted in France, but unlike study 1, participants were members of a consumer panel representative of the general population. The data were collected via a professional European marketing research company that maintains a national panel of 100,000 consumers. Invitations to participate in an online study were sent to 80 panel members aged 20–35, an age range selected to match the product used in the experiment (a non-alcoholic energy drink). Participants were invited to an online survey consisting of two separate studies and offered a gift certificate as an incentive.

In the first part of the study, participants were asked to comment on a print advertisement for a new beverage brand, called V, that was about to be introduced in France. One version of the advertisement was designed to create a strong stereotypic association between the brand and the US: the main visual was a skyline of New York City with the Statue of Liberty and the verbal message stated that “V is soon arriving from America.” The other version provided a weak BCA, containing a modern skyline photo that could be perceived as an American city (it was in fact Melbourne) and stating that “V is soon arriving from afar”.

In both conditions, participants were asked to take a few minutes to observe the ad. To understand underlying beliefs toward the message in each condition, cognitive responses were elicited. Cognitive responses reflect the processes and reactions of the respondent (Wright 1973). Participants could list up to five “thoughts (they) had while (they) were looking at the ad.” These thoughts were then coded for CoO mentions.

Participants were asked if they had heard of the V brand before. They were then informed that “V provides people who lead a busy lifestyle with a boost of energy and vitamins in one non-alcoholic drink” and that “it will soon be available in France and we would like to know (their) opinions about it.” Attitude toward the brand was assessed on a five-item semantic differential scale (negative–positive, unfavorable–favorable, strongly dislike–strongly like, bad–good, low quality–high quality, uncool–cool, α = 0.96). The strength of the stereotypic BCA was measured by asking how much they perceived the brand as American (1 = not at all, 5 = very). Attitude toward the advertisement was measured on a three-item semantic differential scale (negative–positive, unfavorable–favorable, strongly dislike–strongly like, α = 0.87). The order of all these measures was randomized to avoid order effects. In concluding this section, respondents were told that they would be entered into a lottery to receive a free pack of beverages of their choice. They were asked to select an assortment of 15 bottles from five brands, four existing brands and V, to comprise their pack.

The second part of the study was presented as a “study of how French people feel about the US” and included the same measures as used in study 1, in particular the 12-item measure of animosity (α = 0.80). Demographic information was collected last. Participants received a debriefing email message following completion of the study.

4.3 Descriptive statistics

Seventy-seven (22 males) completed the online study. The mean age was 24.5 (SD = 3.55). Seven participants were removed because they had seen the V brand before (it is currently sold in Australia and New Zealand). The ad manipulations successfully created different levels of BCA: the brand was more strongly associated with the US in the US condition (4.21 vs. 3.64, t(68) = 2.34, p < 0.05), even though there were no differences in overall attitudes toward the ad (2.36 vs. 2.53, t(68) = 0.70, n.s.). The US condition also generated a greater proportion of thoughts related to the US (40.82%) than the country-neutral condition (14.28%; χ 2(1) = 5.85, p < 0.05). The distribution of animosity scores was similar to study 1 (M = 4.51, SD = 0.78).

4.4 Relationship between animosity and brand attitudes and choice

Participants were separated into low- and high-animosity groups based on a median split and brand attitudes were analyzed in a 2 (ad condition: weak vs. strong stereotypic association) × 2 (animosity: low vs. high) ANOVA with attitude toward the ad as a covariate. The analysis reveals a significant interaction (F(1, 63) = 4.21, p < 0.05) with no main effects (both Fs (1, 63) < 2.56, n.s.). As expected, the effect of animosity on brand attitude is moderated by the level of stereotypic BCA. As illustrated in Fig. 1, for consumers whose animosity is low, there is no difference in brand attitudes whether the BCA is strong or weak (2.90 vs. 3.01) but when animosity is high, brand attitudes are lower when the BCA is strong than when it is weak (2.59 vs. 2.96).

Fig. 1
figure 1

Brand attitude—study 2

A similar pattern of effects emerged in the behavioral response, actual choice. The proportions of V cans selected in the gift pack were compared across conditions and the results show that the effect of animosity on brand choice is also moderated by the strength of the BCA: whereas there is no difference in choice proportions amongst low-animosity consumers (33.33% in both conditions; χ 2(1) = 1.00, n.s.), choice differs significantly amongst high-animosity consumers (55.56% in weak association condition vs. 17% in strong association condition; χ 2(1) = 4.66, p < 0.05). This choice pattern is depicted in Fig. 2. In addition to documenting the expected negative effect of animosity on brand choice when consumers harbor strong animosity toward the country and the brand is strongly associated with that country, the results reveal that animosity has a facilitative behavioral effect (greater choice) if the new brand is weakly associated with the country toward which they have animosity.

Fig. 2
figure 2

Brand selection percentage—study 2

4.5 Discussion

Study 2 shows that creating a strong stereotypic association with a country for a new brand ignites animosity effects and triggers both prejudice, in terms of negative attitudes, and discrimination, in terms of reduced choice, toward the brand. When consumers who harbor high animosity toward a country are exposed to a new brand that is strongly associated with that country, their attitudes toward the new brand are negatively impacted and their choice of that brand is reduced. In contrast, consumers who have low animosity toward the country do not respond any differently, attitudinally or behaviorally, to the new brand, whether it has a strong or weak association with the country. Building on the cross-sectional evidence from study 1, this experiment's results provide causal evidence that the strength of the stereotypic association between a brand and a country moderates the relationship between animosity and brand attitudes and behaviors toward the brand.

In addition, the study reveals that consumers who harbor high animosity toward a country exhibit more approach behavior toward brands that are not associated with that country. Thus, in addition to evidence of discrimination toward brands that are strongly stereotypic of a country, there is evidence of facilitation toward brands that are weakly stereotypic of the country.

5 General discussion

Two studies provide correlational and causal evidence that the link between animosity and consumers’ attitudes and behaviors toward brands depends on the strength of stereotypic BCA. Extending existing knowledge about animosity and CoO effects, the research shows that animosity toward a country can trigger both prejudice, in terms of negative attitudes, and discrimination, in terms of reduced choice, toward brands that are strongly associated with a given country. The focus on a single target country, the US, and a single country of respondents, France, leaves open the need to extend the research to other contexts where animosity may take on different forms. For instance, opinions of other countries may be more or less covert, and these different levels of accessibility may affect whether and how brand prejudice and discrimination occur (Devine 1989). Consumers' direct experience of a country and its people may also affect the nature of their animosity and these individual differences might further moderate animosity's impact on brand prejudice and discrimination.

The research adds to the growing evidence of moderators of the impact of country-specific constructs such as animosity. Hong and Kang (2006) found that animosity does not impact consumers’ opinions of products if the country from which they emanate is well-known for producing these types of products. Yet, the results of this research indicate that this pattern does not hold at the brand level because the salience of a brand's origin, either inherent to a brand or conveyed via advertising messages, in essence, acts as a catalyst of animosity effects. This discrepancy offers opportunities for future research in other contexts to assess whether a high match between a country's image and a product category would override animosity effects (Roth and Romeo 1992).

There are clear managerial implications for the finding that brands may suffer from animosity effects merely due to consumers’ perceptions of the brands’ image. As advocated in previous research, whenever brands originate from a country towards which people harbor negative feelings, they should be promoted in a way that does not obviously associate them with that country (Ettenson and Klein 2000). More specifically, this research substantiates concerns about anti-Americanism (Katzenstein and Keohane 2006) and demonstrates the need for US and non-US companies alike to carefully consider how they brand themselves to avoid undesirable negative associations.

One possible strategy for brands to avoid animosity-related prejudice and discrimination is to actively engage in corrective processes, such as counter-stereotypic association training (Kawakami et al. 2007). Giving people the opportunity to be aware of and to correct for perceived influences could remove the automaticity of the stereotypic association and reduce both the activation of a bias (Devine 1989) and the application of the bias (Kawakami et al. 2007). Because previous research has shown that corrective processes dissipate over time, strategies to induce such processes must be relentless especially for those brands that have strong stereotypic associations with a country.