COMM 4841 Thesis EFFECTS OF A BRAND S COUNTRY OF ORIGIN ON CONSUMER BEHAVIOR IN BRAZIL

Transcription

1 COMM 4841 Thesis EFFECTS OF A BRAND S COUNTRY OF ORIGIN ON CONSUMER BEHAVIOR IN BRAZIL Aline Avigdor Altschul April. 24 th,

2 INTRODUCTION Consumers in developing countries often face the challenge of choosing between local and nonlocal, foreign products. The motives and implications of this choice-making process are worth researching, because companies can use such information to more effectively market their products. In particular, the marketing literature has considered how a brand s country of origin (CO), or the country with which a brand is associated, influences consumers perceptions of a brand s quality, and of their own social status or identity. Most of the primary research in the area of country of origin sought to understand risk-reducing biases used by Western consumers when evaluating products from, among others, less developed, and therefore, risky countries or regions (Batra et al., 2000: 84). Much of the existing research focuses on quality, examining CO as a halo construct that influences consumers believes about a product s quality (Han, 1989). However, recent research found that CO influences other aspects of consumers product evaluations, and that this relationship is affected by consumers level of national identification and ethnocentrism (Verlegh, 2007). More specifically, it is important to study the country of origin effect in developing countries. While some country of origin effects may be the same as in Western countries, others may be unique to the developing countries context. Unlike Western consumers, consumers in developing countries often look up to American and Western Europe consumption practices and lifestyles, trying to emulate those practices by purchasing brands they see on movies, TV, or on their trips overseas (Batra et al., 2000). Therefore, it is essential to understand consumer behavior in developing countries due to increasing exposure to local brands, and the increased importance of these emerging markets in the world economy. This thesis expands on the research of country of origin by analyzing the impact of CO on the perceptions of Brazilian consumers. Brazil is a developing country that lacks research on this topic. Specifically, I advance hypotheses regarding the influence of national identity and admiration for foreign lifestyles on consumers purchase 2

3 intentions and perceptions of quality. I also hypothesize the moderating effects of publicly consumed products and a product category that closely relates to national identity on both effects. CO studies of developing countries have focused on the implications of quality, status, and identity on consumer behavior, primarily in Turkey and India. These countries have similarities in terms of inequality, economic growth in the past decades, and emergence of the middle class. Turkey, for example, went through a shift from scarcity to market abundance in the 1990s, and the experience of Turkish consumers can offer a perspective on how the same shift may affect consumers in other developing countries (Belk 1988). On the other hand, India has a history of colonization, the last being by the British. As a result, many people argue that Indians still have an inferiority complex and wish to reproduce foreigners behaviors and purchase nonlocal products (Batra et al., 2000). Overall, studies in both India and Turkey exemplify the developing country context and provide a perspective on how country of origin research can be extended into emerging markets that currently lack research on this topic, such as Brazil. In this study, I tested four different hypotheses regarding the effects of country of origin on consumer behavior in Brazil. I suggest that, as their level of national identity increases, Brazilian consumers will be more likely to purchase domestic brands and perceive those brands as having better quality than foreign brands. I suggest that the opposite will happen as the level of admiration for foreign lifestyle increases. I also examined the extent to which the effects of admiration for foreign lifestyles and national identity are moderated by whether a product is privately or publicly consumed, and by whether the given product category is identified with Brazil. I tested these hypotheses using three different product categories toothpaste, athletic shoes, and coffee. Consumer surveys were administered online through Amazon Mechanical Turk and facebook.com. Overall, while some results were not statistically insignificant because of the limitations discussed later in this study, I was able to find very interesting results. I expected to find that both willingness 3

4 to purchase and perceived quality of the domestic product increases over the foreign product as the level of national identity increases. However, I was unable to find any statistically significant results to draw a conclusion. The same happened to the proposed effect of admiration for foreign lifestyles. However, I found that, as a product is publicly consumed and consumers show low admiration for foreign lifestyles, they are more likely to perceive the domestic brand to have better quality than the foreign brand, which is consistent with the hypothesis. Additionally, if the product category is identified with Brazil and the consumer has low national identity, perceived quality of the domestic product over the foreign product is higher. Also, when a product category is identified with Brazil and national identity is high, perceived quality of the domestic over the foreign product is lower. These findings are, interestingly, not consistent with the hypothesis. Lastly, I found that if the product is closely identified with Brazil and admiration for foreign lifestyles is low, willingness to purchase the domestic brand over the foreign brand is lower. However, if the product is closely identified with Brazil and admiration for foreign lifestyles is high, willingness to purchase the domestic brand over the foreign brand increases. These findings are also, interestingly, inconsistent with the proposed hypotheses. LITERATURE REVIEW AND HYPOTHESIS Marketing scholars argue that a brand s CO serves as an extrinsic cue (along with price and brand name) that supplements the use of intrinsic cues (perceptions of design, performance, etc.) (Batra et al., 2000: 84). In other words, consumers may use a brand s country of origin when drawing inferences about a product s quality. Additionally, consumers may use CO to draw conclusions about others social status and national identity. The research below has extended the discussion on CO and how it affects purchase decisions in both developed and developing countries. In developed countries, a stream of literature suggests that country image serves as a halo to 4

5 consumers product evaluations (Han, 1989). When evaluating unknown products, consumers may be unable to detect their true quality, therefore using country image to reach a conclusion (Huber and McCann, 1982). However, when consumers are familiar with a certain product, they use country image as a construct (Han, 1989). In other words, they use CO as a product characteristic that summarizes their beliefs about the product s attributes. Another stream of literature suggests that, in developed countries such as the United States and the Netherlands, consumers tend to indicate high levels of consumer ethnocentrism. The latter is defined as the consumer desire to protect the domestic economy, as they view the purchase of foreign, or imported goods, as wrong because it hurts the economy (Verlegh, 2007). Therefore, those consumers tend to rate domestic products positively and foreign ones negatively as a way to protect their country s economy (Verlegh, 2007). Overall, consumers in developed countries tend to use country of origin to infer quality and as a way to protect their economies through domestic consumption. However, results may change when one tests country of origin effects in developing countries, given differences in culture, history, and institutions. Research has shown that consumers in those countries also use CO to infer quality, but may also use it for status enhancing purposes and as a result of high levels of admiration of foreign lifestyles (Batra, et al., 2000). Prior country of origin research in developing countries shows that there is a positive relationship between perceived quality and purchase intention of nonlocal brands. Kumar et al. (2008) study of Indian consumers purchase intention toward a United States versus a local brand found that, in India, there is a positive relationship between perceived quality and purchase intention of American products, such as with the Levi s brand. Additionally, Indian consumers that have more positive attitudes toward American products tend to believe that local brands have inferior quality. These findings are in line with previous studies that show that, in terms of quality, consumers in developing countries have the tendency to evaluate foreign or U.S. products more positively when 5

6 compared to local products. That is because developed countries tend to have higher expertise in product development and lead innovations, more often than developing countries. That can be particularly true in the electronics product category with companies such as Apple and Microsoft. Additionally, another study conducted by Kinra in India suggests that, contrary to studies in developed countries, high levels of consumer ethnocentrism and national identification does not indicate preference for local brands. In fact, those consumers evaluate nonlocal products higher than local ones in aspects such as technology, quality, status, esteem, and higher credibility (Kinra, 2006). The reason that different results are found in developing countries is because even though consumers in those countries may hold the same ideals as consumers in developed countries (e.g., a high level of ethnocentrism), additional factors influence their preferences for local and nonlocal brands. Lastly, a stream of literature suggests that consumers in developing countries believe a brand s country of origin has implications for one s social status, and these beliefs may also affect their decisions when choosing between local and nonlocal brands. If they purchase a foreign brand product, for example, that may signal higher social status to other members of the society in which one lives. As mentioned above, consumers in emerging markets often praise consumer lifestyle in developed countries (Batra et al., 2000), whereas consumers in the latter often purchase local brands due to high levels of ethnocentrism (Verlegh, 2007). For consumers who have a greater admiration for lifestyles in developed countries, country of origin has a greater effect on the decision to purchase nonlocal brands. The same result was found for product categories that have high levels of social signaling value (Batra et al., 2000). In Turkey and Romania, consumers prominently favor foreign, mainly Western European products, as a way to enhance their social status since interpersonal relations are extremely important: In Turkey, interest in consumption is obvious not only in wants but also in the prominent display of, and talk about, products specially the status 6

7 brands (Ger et al., 1993: 105). In conclusion, results show that interest in consumption comes from the desire to display and talk about products, mainly those from foreign brands that signal high social status (Ger et al., 1993). Overall, the effects of country of origin in consumer behavior in Turkey, Romania, and India may be an example of purchase behavior in other developing economies (Belk, 1988), making it important to study country of origin effects in other emerging markets such as Brazil. THE BRAZILIAN SITUATION Overall, prior research shows that a brand s country of origin affects attributions made about quality and social status. National identity has also been documented to affect purchase behavior. This research has focused on the implications of those factors in developed countries such as the U.S, Netherlands (Verlegh, 2007), and developing countries such as Turkey, Romania (Ger et al., 1993), and India (Kinra, 2006); Kumar et al., 2008). However, this study proposes that similar factors will also affect consumer behavior in Latin America. More specifically, it is crucial to understand the consumer base in Brazil because of its large size and importance, given that lower social classes have rapidly moved socially and gained purchasing power in the past decade. Additionally, Brazil is the world s 7 th largest economy, what makes it important to study the effects of country of origin in Brazil. Like other countries in Latin America, Brazil has an export-oriented economy since colonization in During the period of industrial expansion between 1880 and 1930 in Europe and in the United States, different forms of production of raw materials expanded throughout Latin America, leading to a shift in class structure and political reorganization (Orlove, 1997). Consequently, this export-based growth allowed the wealthy to import many luxuries and the masses to replace local crafts such as textiles and pottery with cheap manufactured items form 7

8 Europe and North America (Orlove, 1997: 2). After the Great Depression in 1929, demand for Latin America products dropped significantly, causing countries like Brazil to manufacture the products previously imported from developed regions, but not with the same knowledge, capacity and quality. The period thereafter was marked by strong state participation in the economy and leftist dictatorships that gave importance to the growing working class (Orlove, 1997). Throughout the twentieth century, Latin America faced increased wealth and income inequalities. The elite in those countries, be it in the capital of Argentina, Brazil, Peru, or Chile, imported grand pianos, European wines and liquors, traveled abroad, and sent their children to European schools and believed that in doing so, they were sharing the same activities and attitudes of the international gentry (Orlove, 1997: 5, Beezley 1987: 14). In other words, the elite had access to imported goods and consumption, looking up to foreign practices, and differentiating themselves from and leaving the masses with little accessibility to those products. In The Allure of the Foreign, Orlove argues that it is important to understand the social and cultural implications of export-oriented economies that emphasize affordability of imports. In other words, foreignness is an attribute of the imported goods that merits consideration, like their affordability and cost. (Orlove, 1997: 7) The notions of modernity in Latin America that persist today provide context to the study of country of origin effects in consumer behavior in the region, such as in Brazil. It is a break from the traditions of the past, a valuation of the present, an openness toward experimentation in seeking to construct a future different from both. Modernity appears almost always as an international movement, but Latin American participation in it was problematic, since the centers of modernity lay in Europe and the United States. (Orlove, 1987: 8). Lastly, it is important to study the effects of a brand s country of origin in Brazil because of the history of cultural influence and because of the country s significant participation in the global 8

9 economy. With the increase of the middle class in the past decade, Brazilians have, more than ever, access to foreign products and trips abroad. With the current exposure to international tourists through large sporting events, to the English language and the influential role of the TV, Brazilians are more susceptible to foreign lifestyle influence. The results of such study will provide valuable information to both local and nonlocal brands that currently sell in the country, and to those that wish to enter Brazil. NATIONAL IDENTITY AND PURCHASE BEHAVIOR OF BRAZILIAN CONSUMERS Given Brazil s history of cultural influence, I aimed to understand whether national identity is powerful enough to drive purchase intention of local brands. National identification is rooted in consumers need to enhance group and self-esteem, and based on a desire for positive social identity (Verlegh, 2007). More specifically, nationality is part of a consumer s identity, and can be expressed in the purchase of domestic products, since they have important social and cultural connotations (Askegaard and Ger, 1998). This concept can be confounded with consumer ethnocentrism, which reflects the desire of consumers to protect domestic economy, [ ] viewing the purchase of imported products as wrong, because it hurts the domestic economy (Verlegh, 2007). Since both concepts of national identity and consumer ethnocentrism can overleap, I measured both in the data collection so that I did not confound them. However, given that social influence tightly relates to national identity, I focused on analyzing the latter in the hypotheses. Furthermore, the same effect also pertains to national identification and perceived quality. The stream of literature in national identity also argues that this factor influences consumer s perception of a product s quality according to whether the product is local or nonlocal. The higher the level of national identity, the more favorable consumers are to domestic products and more 9

10 likely they are to perceive them to have better quality than foreign goods. Therefore, I hypothesize: H1a: As a consumer s level of national identity increases, their willingness to purchase domestic brands over foreign brands increases. H1b: As a consumer s level of national identity increases, their perception of quality of domestic brands over foreign brands increases. Brazil has a history of admiration for foreign lifestyles. As previously mentioned, wealthy and middle class individuals now purchase foreign goods for several reasons: export oriented economy, willingness to consume the same modern and high quality brands as Western Europeans and Americans, and the desire to emulate their lifestyles. As discussed, previous research has also found that in developing countries such as India, Turkey and Romania, consumers exhibit high levels of admiration of foreign lifestyles. They are also more likely to perceive a foreign brand to have better quality and are more willing to purchase the foreign brand (Batra et al., 2000; Ger et al., 1993). Thus, CO effects should be stronger when consumers have high levels of admirations of foreign lifestyles. Therefore, I tested the following in Brazil: H2a: As a consumer s level of admiration of economically developed countries lifestyles increases, their willingness to purchase foreign brands over domestic brands increases. H2b: As a consumer s level of admiration of economically developed countries lifestyles increases, the perceived quality of foreign over domestic brands increases. Additionally, the reason many consumers prefer non-local brands is because they believe those products have a reference group appeal (Batra et al., 2000). In other words, a consumer in a developing country thinks that nonlocal brands appeal to certain social groups and will influence his 10

11 or her social status in the eyes of those groups. However, these effects should be dependent on whether products are publicly consumed, given that only products that one observes another consumer use can influence his or her evaluations. Thus, in order to measure how attitude varies according to social status and admiration of foreign lifestyles, it is important to understand behavioral changes when a product or brand is publicly consumed compared to when a product is privately consumed. This hypothesis has not yet been tested in other studies of developing countries. Therefore, this study will contribute to the overall country of origin literature and more specifically about the latter in Brazil. I therefore hypothesized: H3: The effects proposed in hypothesis 1 and hypothesis 2 will be moderated by whether or not the product is publicly or privately consumed, such that the effects are stronger when a product is publicly consumed. Lastly, it is interesting to study how the effects of national identification and admiration of foreign lifestyles might differ for a product category that is strongly associated with Brazil versus a category that does not have this association. When a product category is associated with Brazil, one would expect to see stronger effects of national identification and weaker effects of admiration of foreign lifestyle on the consumers perceptions of quality and purchase intention of local brands. As previously mentioned, national identity can be expressed through the purchase of domestic products, as they may have important social and cultural connotations (Askegaard and Ger, 1998). Therefore, it is expected that, the more consumers identify themselves with the country, the more likely they will be to consume a product that falls into a category that is identified with their country. Similarly, one would expect a weaker effect on high levels of admiration of foreign lifestyles, given that consumers would choose not to purchase a brand if that is highly associated with their own 11

12 country. Therefore, I hypothesized: H4a: The effects proposed in hypothesis 1 will be moderated by the extent to which consumers perceive the product category to be identified with Brazil. The effects will be stronger the more the product category is identified with Brazil. H4b: The effects proposed in hypothesis 2 will be moderated by the extent to which consumers perceive the product category to be identified with Brazil. The effects will be weaker the more the product category is identified with Brazil. METHODOLOGY In this research, I collected data through two studies. The first study tested the proposed relationships between national identity (hypotheses 1a & 1b) and admiration of foreign lifestyles (hypotheses 2a & 2b), and their effects on purchase intention and perceived quality. It also tested the moderating effect of private vs. public consumption (hypothesis 3). The second study also tested hypotheses 1a, 1b, 2a & 2b, as well as the moderating effect of the national identity of the product category (hypothesis 4a & 4b). I planned to collect data by conducting online surveys with Brazilian consumers through Amazon Mechanical Turk (MTurk). The latter is a crowdsourcing marketplace that serves as an information gathering pool. Originally, I intended to use this system because it has proven to be a valuable tool for data collection when the surveys considered the domain of consumption and include screening questions that measure attention and language comprehension (Goodman et al., 2012). Across the two studies, four hundred Brazilian respondents were expected to complete a short online survey in exchange for 51 cents. A Portuguese version of the questionnaire (translated from the English version shown on Exhibit 1) was used given that most 12

13 Brazilians are not fluent in the English language. There was also no age or gender restriction in either of the studies. STUDY 1 DESIGN AND PROCEDURE Study 1 focused on testing the proposed relationships between national identity (hypotheses 1a & 1b) and admiration of foreign lifestyles (hypotheses 2a & 2b), and their effects on purchase intention and perceived quality. It also tested the moderating effects of public versus private consumption proposed in hypothesis 3. Thus, study 1 utilized national identity (continuous), admiration of foreign lifestyles (continuous), product category (binary; publicly vs. privately consumed), and product origin (binary; Brazilian vs. American) as the factors of interest. Two hundred Brazilian respondents were expected to participate in this study. First, they were randomly shown one of four advertisements (please refer to exhibits 3-6 for the English version of the ads used in study 1). The ads varied in whether or not they utilized a fictitious American or Brazilian brand name, and whether the product category presented was publicly or privately consumed athletic shoes (publicly consumed) or toothpaste (privately consumed). Respondents were then asked questions about perceived brand quality, perceived brand origin, and intent to purchase the advertised product. Please refer to exhibit 1 for a chart that indicates the survey questions. Respondents were then asked to answer items relating to national identity, consumer ethnocentrism, admiration of lifestyles in economically developed countries, and demographics. As previously mentioned, national identity and consumer ethnocentrism were both included to ensure that these factors were not confounded. 13

14 The inclusion of two different product categories (toothpaste and athletic shoes) allowed me to test how consumer behavior is influenced by reference group appeal and social status (hypothesis 3). Evaluations of athletic shoes (publicly consumed) were compared to evaluations of toothpaste (privately consumed) to test hypothesis 3. Originally, within each of these product categories, fifty respondents were expected to answer questions about a fictitious Brazilian brand, and the other fifty about a fictitious American brand. The table on the right shows how study participants were supposed to be divided Total respondents: 200 Athletic Shoes Number of respondents per survey US Brand Toothpaste Brazilian Brand between conditions. STUDY 2 DESIGN AND PROCEDURE Similar to study 1, study 2 tested the levels of national identity (hypotheses 1a & 1b), levels admiration to foreign lifestyles (hypotheses 2a & 2b), and their effects on purchase intention and perceived quality. Thus, study 2 utilized a similar design and procedure to study 1 with one notable exception. Rather than examining the moderating role of public vs. private consumption, study 2 tested whether purchase intention and perceived quality changed when presented with a product category that closely relates to national identity (hypotheses 4a & 4b). To test these hypotheses, this study used two privately consumed product categories. As with study 1, respondents were asked to view one of four advertisements that varied on brand name (fictitious American vs. Brazilian brand) and product category (strongly vs. weakly linked to national identity). Please refer to exhibits 7-10 for the English version of the ads used in study 2. Respondents were then asked questions about perceived brand quality, perceived brand origin, and intent to purchase the advertised product. Following the ad questions, respondents 14

15 answered questions on national identity, consumer ethnocentrism, admiration of lifestyles in economically developed countries, and demographics. The product categories used for this study were home-brewed coffee (strongly linked to national identity) and toothpaste (weakly linked to national identity). By dividing the survey into two different product categories, I was able to measure how respondents reacted to privately consumed products with different degrees of identification with Brazil, removing any reference group or social status effects. One category represents an everyday product (toothpaste), and the other category, coffee, is strongly linked to the Brazilian national identity. In the 19 th and 20 th century, the Southeast region of Brazil specialized in coffee production and its exportation, which made this product known for its high quality worldwide. No previous study of country of origin in developing countries has tested this kind of hypothesis, Total respondents: 200 Number of respondents per survey US Brand Brazilian Brand which makes it interesting to test this product category Coffee Toothpaste influence in Brazil. Lastly, similarly to Study 1, within each of these product categories, fifty respondents were expected to answer questions about a fictitious Brazilian brand, and the other fifty about a fictitious American brand. After setting up the survey on qualtrics.com (survey software tool) and publishing both studies on Amazon MTurk, I encountered unexpected difficulties. I was only able to get five respondents within a one-week period, since the fact that studies were set up to target only Brazilian consumers brought limitations to the response rate. After looking into other publishing tools, I decided to use the social media website facebook.com to reach a broader Brazilian audience. According to Statista, Brazil is the world s third largest facebook.com country user, with approximately 70.5 million active users. In order to reach as wide of an audience as possible, I posted the survey link on my Facebook profile wall, messaged 15

16 friends individually asking them to fill out the survey, and had friends post on their wall about the survey. I also published the link on several Brazilian facebook.com groups, such as the BRASA (Brazilian Student Association) and Conferência ENE groups. After another week, I was able to get 137 responses. 67 people responded Study 1, and 70 people responded Study 2, as shown on the tables above. Total respondents: 67 Study 1 Number of respondents per survey US Brand Athletic Shoes Toothpaste Total respondents: 70 Brazilian Brand Study 2 Number of respondents per survey US Brand Coffee Toothpaste Brazilian Brand Because of time constraints regarding the nature of this academic project, I was unable to look into other options to reach the ideal 400 respondents. If time permitted, I would have travelled to Brazil and surveyed respondents personally in shopping malls in the city of São Paulo, SP Brazil, as this has proven to be an effective data collection method in previous studies, such as Verlegh (2007). Alternative collective methods would also allow me to collect a more representative sample of Brazilian consumers, since the technique I used is more likely to yield a sample with higher income and education levels. This is represented by the high mean of education level (μ = 3.66) and job status (μ = 3.29) in the sample. MEASURES The items used in my scales were drawn from scales previously validated in studies of the country of origin literature. The dependent variables used were perceived quality and willingness to purchase the brand. The independent variables in the study were national identity (Verlegh, 2007), admiration of foreign lifestyles (Batra et al., 2000), brand origin (Batra et al., 2000), and consumer ethnocentrism (Shimp & Sharma, 1987). Please refer to Exhibit 1 for survey questions. 16

17 To test willingness to purchase, respondents answered a single item scale, I intent to buy this product. To test perceived quality, there was a two-item scale, This brand is of inferior quality, and This brand is better than average. As for perceived brand origin, respondents answered the degree to which I consider this brand to be a Brazilian brand. Admiration of foreign lifestyles was measured in a single item scale, To what extent do you yourself admire the lifestyles of people who live in the United States? As for national identity, consumers rated the degree to which they strongly agree or strongly disagree (7-point scale) with the following statements: Being Brazilian means a lot to me, I am proud to be Brazilian, and I don t feel any ties with Brazil. Consumer ethnocentrism was also measured in the studies to avoid confounding. As a five-item scale, consumers rated the degree to which, It s not right to purchase foreign-made products, Brazilians should not buy foreign-made products, because this hurts Brazilian businesses and causes unemployment, A real Brazilian should always buy Brazilian-made products, I always prefer Brazilian products over foreign, American products, and I prefer buying American products than Brazilian products. Finally, most items used a continuous scale from 1 to 7 (1=strongly agree 7 = strongly agree), unless otherwise stated on Exhibit 1. Consumer demographics questions varied from openended (age) to four level (marital status and income), five level (education level), and seven level scales (job status). The means and standard deviation of the independent variables of interest are also shown on Exhibit 1. The variables brand origin, public/private good, and national identity of product categories were all manipulated in the study. They were coded binomially based on the ad shown to the respondent. Lastly, some of the scales required reverse coding to facilitate the analysis in terms of the direction of the effects described in the hypotheses. 17

18 RESULTS The hypotheses in the study concerned the effects of a brand s country of origin on consumer behavior, such as willingness to purchase and perceived quality, moderated by whether the product is privately or publicly consumed, and whether the product category closely relates to national identity, such as coffee. On both studies, I conducted regression analyses and performed spotlight analyses to further understand the observed effects. To test this type of directional hypotheses, it is optimal to see if the sign of the estimated coefficient is in the hypothesized direction and is statistically significant by looking at its p-value (Batra et al., 2000). Hypotheses about moderating effects can be tested via the statistical significance of the appropriate interaction term (cf. Baron & Kenny, 1986). I started the analysis by running a manipulation check on brand country of origin to determine whether indicating in the ads that the brand was Brazilian vs. American did indeed result in differences in the extent to which respondents believed the brand to be more or less Brazilian. I ran a t-test that compared the means of Brazilian Brand Origin (BBO) between the experimental conditions that presented consumers with an ad describing a Brazilian product and those that presented an ad which referenced an American product. To measure perceived BBO, respondents were asked to rate on a scale (1 = strongly disagree, 7 = strongly agree) whether they agreed that the product shown in the ad was Brazilian. Based on the number of people that responded to the survey, the manipulation check was statistically significant (t = , p <.05). Consistent with expectations, respondents that read the Brazilian ad believed that the brand was more Brazilian (M = 4.0) than respondents that saw the American ad (M = 2.6). Even though the manipulation check was statistically significant, it did not produce as large of an effect as I expected. Across the Brazilian 18

19 conditions, BBO only averaged a 4.0 on a 7.0-point scale, indicating that consumers still perceived the advertised brand to be as much American as Brazilian. Moving forward, in order to test hypotheses 1 & 2, I conducted linear regression analyses. More specifically, for hypotheses 1a & 2a, I conducted a regression of consumer s willingness to purchase the brand as a function of (a) the level of national identity (hypothesis 1a) or the level of admiration of foreign lifestyles (hypothesis 2a), (b) Brazilian brand (BB; 1=Brazilian ad; 0=American ad), and (c) their interaction. To test hypotheses 1b and 2b, I conducted a similar regression, expect I substituted the dependent variable to be perceived quality. In evaluating the results, I looked at the significance and direction of the coefficient of the interaction terms. Values and direction of the estimated regression coefficients are shown in Exhibit 2. Model 1 indicates the results for the regression analyses used to test hypothesis 1. Model 2 indicates the results for regression analyses used to test hypothesis 2. After running the regressions for both hypotheses 1&2 with data from both Study 1 and Study 2, I found in model 1 that none of the interaction coefficients were statistically significant (all p-values > 0.345). The limitations of the model, such as small sample size, discrepancy between the brand s intended origin and respondents perceptions of brand origin, and limited research time period, may be reasons for such outcome. If the coefficients for the interaction terms were positive and significant, it would support hypotheses H1a & H1b in that as a consumer s level of national identity (BNI) increases, both their willingness to purchase (WTP) and perception of quality (PQ) of domestic brands over foreign brands increases, respectively. However, contrary to our expectations in hypotheses H1a, the coefficient of the interaction term is negative (B = , t = -0.11, p = 0.91). If significant, this would indicate that at higher levels of national identity, respondents are less willing to purchase when they are told that the advertised product is Brazilian vs. American. Furthermore, as in hypothesis 1a, the interaction between national identity and brand origin in 19

20 hypothesis 1b is negative (B = , t = -0.95, p = 0.345), suggesting that at higher levels of national identity, respondents perceive that quality of the advertised product is lower when they are told it is Brazilian vs. American. Even though neither of the interaction terms coefficients is significant, model 1 showed other interesting findings. The more a respondent identifies high levels of Brazilian national identity, the more he/she is willing to purchase the product, since the coefficient of the main effect is 0.093, though this result is not significant. The coefficient for the main effect of national identity that tested H1b was also positive (B =.017), consistent with our expected pattern. If significant, this finding would suggest that respondents with a higher national identity perceive the quality of the advertised product to be higher. To test hypotheses 2a & 2b, we ran the aforementioned regression analyses, but substituting the national identity variable for admiration for economic developed countries (AEDC), The coefficient for the interaction term was not significant in either model (both p-values > 0.697). In hypothesis 2a, if the coefficient of the interaction term was significant, it would show that as a consumer s level of admiration of economically developed countries lifestyles increases, their willingness to purchase domestic over foreign brands increases, since the coefficient is positive (B = 0.070, t = 0.39, p = 0.697). This is inconsistent with the hypothesized effect, which could be due to the limitations of this study. Furthermore, the coefficient for the interaction term in H2b was negative (B = , t = -0.20, p = 0.843). Therefore, if significant, it would be consistent with the predicted hypothesis; as a consumer s level of admiration of economically developed countries lifestyles increases, they perceive that the quality of the American brand is higher than that of the Brazilian brand. However, there are other interesting results in model 2. The main effect of AEDC had a positive coefficient in both WTP (B =.135) and PQ (B = 0.177), although only the latter was marginally significant. This would indicate that the higher AEDC, the higher the WTP and the PQ of the product. 20

21 Moving forward, study 1 tested hypothesis 3, which, despite the limitations of the model, had some statistically significant results. Hypothesis 3 tested whether the effects proposed on hypotheses 1&2 were moderated by whether or not the product was publicly or privately consumed (0 = private, 1 = public), such that the effects are stronger when a product is publicly consumed. This would entail a significant three-way interaction. However, neither of the three-way terms was significant. Model 3 indicates results for hypothesis 3 for the effects proposed in hypothesis 1, and model 4 indicates results for the same hypothesis, but for the effects proposed in hypothesis 2. Yet, if the limitations of the models were smaller, there would be a statistically significant interaction on the moderating effect of publicly vs. privately consumed product category on admiration for foreign lifestyles and brand origin regarding perceived quality (hypothesis 2b). The interaction coefficient was negative (B = , t = , p = 0.157), which indicates that when a product is publicly consumed (tennis shoes) and consumers show low admiration for foreign lifestyles, they are more likely to perceive the domestic brand to have better quality than the foreign brand. This is consistent with our hypothesis given that all the variables in the study were constructed in a way that higher values indicate higher PQ and WTP of domestic Brazilian brands. Please refer to the graph above for an illustration of the interaction. Even though the latter was the only marginally significant three-way interaction in hypothesis 3, models 3 and 4 revealed other interesting two-way interactions. In terms of the results in model 3, the interaction between brand national identity and publicly consumed products were not statistically significant (p-value: 0.189). However, if the sample size were larger, this interaction would likely have a significant effect on willingness to purchase (B = ) and perceived quality 21

22 (B = ), since the current p-values of the interactions are less than 0.2. This finding would suggest that as BNI increases, WTP and PQ for the publicly consumed product over the privately consumed product decreases. Also, willingness to purchase (B = 1.483) and perceived quality (B = 2.998) are higher for the publicly consumed product than the privately-consumed product, although only the latter main effect is marginally significant. Finally, as BNI increases, WTP (B = 0.192) and PQ (B = 0.265) towards the advertised product increase, although only the latter main effect is marginally significant. As for the moderating effect of publicly consumed products on model 4, the interaction between BB and PCP was marginally significant in both willingness to purchase (B = 3.826) and perceived quality (B = 3.202), suggesting that when a product is Brazilian (e.g. Brazilian tennis shoes), both willingness to purchase and perceived quality of the publicly consumed product over the privately consumed product increases. Additionally, the interaction between admiration for foreign lifestyle and publicly consumed product on perceived quality is statistically significant with a positive coefficient (B = 0.689, t = 2.405, p = 0.018). This entails that as admiration for foreign lifestyles increases, the perceived quality of a publicly consumed product over the privately consumed product increases. Lastly, the main effect of publicly consumed product category was marginally significant in WTP (B = , t = 1.819, p = 0.071) and statistically significant in PQ (B = , t = , p = 0.014). With a publicly consumed vs. privately consumed product, WTP decreases by and PQ by Moreover, study 2 tested hypothesis 4, which had some statistically significant results despite the limitations of the model. Hypothesis 4 tested the following: (a) the effects proposed in hypothesis 1 will be moderated by the extent to which consumers perceive the product category to be identified with Brazil. The effects will be stronger the more the product category is identified with Brazil, and (b) the effects proposed in hypothesis 2 will be moderated by the extent to which consumers perceive the product category to be identified with Brazil. The effects will be weaker the 22

23 more the product category is identified with Brazil. Model 5 indicates results for hypothesis 4 for the effects proposed in hypothesis 1, and model 6 indicates results for the same hypothesis, but for the effect proposed in hypothesis 2. Hypothesis 4a had statistically significant results in relation to the perceived quality of a product (B = , t = -2.37, p = ), although in the opposite direction of the proposed hypothesis. If the product category is Brazilian (coffee) and the consumer has low national identity, PQ of the domestic product over the foreign product is higher. However, when a product is Brazilian (coffee) and BNI is high, PQ of the domestic over the foreign product is lower. Please refer to the graph for the illustration of the proposed relationships. The regression of the same analysis on non-coffee product category and willingness to purchase was neither statistically nor marginally significant. Hypothesis 4b also had statistically significant results in relation to the three-way interaction of willingness to purchase on a product category that is closely related to Brazil (coffee), brand origin, and admiration for foreign lifestyles. The coefficient of the interaction is positive (B = 1.126, t = 2.37, p = 0.019). If the product is Brazilian (coffee) and AEDC is low, WTP the domestic brand over the foreign brand is lower. However, if the product is Brazilian (coffee) and AEDC is high, WTP of the domestic brand over the foreign brand increases. Please refer to the graph for the illustration of the relationships proposed. The results are not in line with the proposed 23

24 hypothesis. The regression for the non-coffee product category was not significant, neither was the regression of the same analysis on perceived quality. Just like hypothesis 3, this study revealed interesting results in the two-way interactions of models 5 and 6. In model 5, there is a statistically significant interaction between brand origin and the identity of the product category to which it belongs (B = 8.219, t = 2.415, p = 0.017). All else equal, a Brazilian brand will be perceived as higher quality when the product category it belongs to is also closely identified with Brazil (e.g. coffee). Additionally, Brazilian Product category (BP) would have a marginally significant effect on perceived quality if the sample size were larger. Since it has a negative coefficient (B = ), it would entail that as the product category is coffee, perceived quality of the product decreases. The same effect was found regarding WTP (B = ), but this variable is not statistically significant. As for model 6, the same interaction between BB and BP was significant, but with regards to willingness to purchase and the effects were in the opposite direction. As a product is Brazilian, willingness to purchase the product when the category is Brazilian (coffee) decreases over the product of which category is non-brazilian. Additionally, if the sample size were larger, the interaction between AEDC and BP would be marginally significant (B = ), and would entail that, as AEDC increases, the consumer is less willing to purchase the Brazilian (coffee) product over the non-brazilian product category. Lastly, admiration for economically developed countries has a significant and positive effect on perceived quality (B = 0.249), and marginally significant positive effect on WTP (B = 0.218), indicating that as admiration for foreign lifestyles increases, so does perceived quality and willingness to purchase the product. 24

25 ADDITIONAL ANALYSES As mentioned in the methodology section, this study had some limitations brought about the small sample size of 137 respondents instead of the desired 400 respondents. In the analysis section, I discussed a few regression analyses that would be statistically significant if the sample increased, assuming that a p-value of less than 0.20 would be marginally significant with a larger sample size. I decided not to use the p-value of less than 0.30 (since the sample size would be approximately three times the achieved sample) because there would be a bigger margin for approximation errors. Another limitation to the model comes from the manipulation check, in which I found that, although the regression between BB - Brazilian Brand (the original brand of the product) and BBO Brazilian Brand Origin (whether the perceived brand origin was Brazilian) was statistically significant at t = ), there was still some discrepancy between perceived brand origin and intended brand origin. This is demonstrated by the fact that, on average, respondents rated perceived Brazilian brand origin as 4.0, out of a 7.0-point scale, indicating that they still perceived the advertised brand to be as much American as Brazilian. A possible explanation for this discrepancy is the names given to each brand, mainly to the toothpaste and coffee brands. Roots and Shiny White are too American, English names that cannot be easily pronounced or related to the Portuguese language, at least not as easily as the tennis shoes brand Mile. In future research, I would manipulate the names of the brands more closely and create names that are similar amongst both languages. This way, I would avoid misinterpretations and make the effects stronger. Because of the limitations described above, I performed an additional analysis on H1 and H2 since neither variable had consistent results with the study design. Instead of running regressions with the binary BB (Brazilian Brand) variable, I used the continuous BBO (Brazilian Brand Origin) variable, as the latter measures the perceived brand origin of respondents. In terms of hypothesis 1a, the results would be marginally significant if the sample size were larger (B = 0.098, t = 1.454, p = 25

26 0.148). If significant, this would indicate that at higher levels of national identity, 7 Hypothesis 1(a) effect respondents would be more willing to purchase when they are told the product is WTP Brazilian vs. American, which is in line with the hypothesis. However, this shows Low BBO High BBO Low BNI High BNI a different effect than the one found in the regression with the binary BB variable. More specifically, as a product is perceived to be American (low BBO), consumers with low BNI are more willing to purchase the product as those with high Brazilian national identity. However, as the product is perceived to be Brazilian, consumers with high national identity were more willing to purchase the product as those with low national identity. Please refer to the graph for an illustration of the effects. In terms of the effect of national identity on perceived quality, the interaction would also be marginally significant if the sample size was larger (B = 0.091, t = 1.543, p = 0.125). The results, illustrated by the graph, would entail that at higher levels of national identity, respondents perceive the quality of the domestic product to be better than the foreign product. More specifically, foreign brands are perceived to have better quality than domestic brand by consumers with low national identity. However, domestic brands are perceived to have better quality over foreign brands when consumers have high national identity. These results are in the same direction as expected in hypothesis 1b. With regards to hypothesis 2a & 2b, the results were neither statistically not marginally significant (p-values > 0.531). 26