Main street retail districts or shopping centers? comparing the preferences of low-income consumers

This study investigated the attitudes of low-income consumers, comparing their preferences between high street retail districts and shopping centers. The interviews were conducted in three representative retail districts in low-income neighborhoods in the city of São Paulo. Although the consumers were more satisfied with shopping centers, they still indicated stronger patronage intention for street shopping districts. Ten attractiveness factors were assessed as influencing the responses regarding these two types of retail agglomerations. While shopping malls received better evaluation on environment, infrastructure, variety of stores and security, the street districts were judged to be better regarding access and value. With the increasing purchasing power, degree of discernment and mobility of low-income consumers in Brazil, we believe that street retail districts, without intense efforts for revitalization, will tend to enter a severe process of deterioration.


INTRODUCTION
hopping malls or main street retail districts?In Brazil, despite the rapid and steady expansion of shopping centers, with 417 units and sales of R$ 87 billion in 2010 according to the Brazilian Association of Shopping Centers (ABRASCE), traditional street retailer districts continue to account for the highest volume of sales and remain promising location alternatives, preferred by a large number of successful retail chains, such as Casas Bahia and Lojas Pernambucanas, as well as fast food chains like and Habib's.Street shopping districts and shopping centers are the two main types of retail agglomerations that compete for consumer preferences.With the increasing purchasing power of low-income people, shopping malls, previously targeted mainly at wealthier patrons, are now being developed to attract lower income shoppers.This study was carried out to investigate the preferences of this growing class of low-income consumers between these two retail formats.
Due to the cumulative attractiveness and synergy that various nearby stores create between each other, there is a tendency for retailers to agglomerate in certain geographic areas (Berman & Evans, 2007;Guy, 2007).These places can be called "retail agglomerations" (Teller 2010) and main street retail districts and shopping malls are the two main types that strive to attract consumers.This typology is based on the characteristics of how they are planned and constructed, and consequently how they are managed and integrated in the market as a whole.
In Brazil the contrasts between shopping centers and street retail districts to a certain extent reflect the socioeconomic polarization of society.In recent years, with high investments in the latest equipment, shopping centers have been continually upgraded.
Because they are planned agglomerations, they can present a well-balanced and complete mix of stores, besides offering security and aggregating services, such as entertainment.They have traditionally been aimed at middle-and high-income consumers.On the other hand, without planning and with little or no investment in urban equipment or in renovation of store installations, retail districts, despite their dynamism, are gradually deteriorating, and mainly serve low-income shoppers.But with the laudable expansion of the middle class ("Class C") and the increased disposable income of classes with lesser purchasing power, shopping centers are increasingly being built in lower income regions, thus threatening the survival of the traditional high street shopping districts located in these regions.
Low-income consumers typically dedicate close attention and time to decisions on allocation of resources, analyze spending in order of priority, research prices and try to organize their shopping to take maximum advantage of the available resources, while still having high consumption aspirations.In countries with high levels of inequality like Brazil, the gulf between the consumption habits of lower and higher income people is much wider than in developed countries.This means that companies that want to reach this market segment need to rethink various aspects of their business models (Chauvel & Mattos, 2008).
Besides their economic importance due to the high volume of sales, street retail districts play an important role in the life quality of urban centers.From an urbanistic standpoint, these agglomerations assure the vitality of cities and help to humanize the better integrate the urban social fabric (Loukaitou-Sideris, 2000).They are open systems that maintain an intense relationship of exchange and integration with the region where they are located, unlike shopping malls, which are isolated and closed systems that make the city less integrated, both from the urbanistic and social standpoints.Street shopping districts are more democratic spaces and are more accessible to consumers without their own transportation, thus better fitting the needs of low-income shoppers.The costs of opening a main street store tend to be lower than in a shopping center, so the former type of shopping venue tends to include more "mom and pop" stores, facilitating the development of small local enterprises.
The importance of traditional retail districts has been declining since the arrival of shopping centers in most cities, both in Brazil and other countries.However, public policymakers have been attaching renewed importance to these traditional districts as part of the movement to revitalize city centers in various countries, although this is still incipient in Brazil.For example, because of concern over the negative impact of the expansion of shopping malls in central regions, in the 1980s the British authorities placed sharp restrictions on these undertakings in city centers (Thomas, Bromley & Tallon, 2006).In American cities as well, the revitalization of downtown areas and neighborhood commercial districts receives large investments, by means of partnerships between the public and private sectors.These efforts are considered important incentives to improve the livability of cities (Geisman, 2004).This trend is motivated not only by the high volume of commercial transactions that generally occur in these places, but also due to recognition that the revitalization of these centers impedes deterioration of surrounding areas and the accompanying urban and social degradation, including crime (Loukaitou-Sideris, 2000).
BBR, Vitória, Special Edition BBR Conference, p. 154 -179, 2012 www.bbronline.com.brDespite the importance of these two main retail agglomerations and the intense competition waged between them, few studies have sought to understand and compare the preferences of consumers for each type of shopping experience.The study by Hart et al. (2007) compares the perceptions and patronage intention of consumers for shopping centers and retail districts in the United Kingdom.The excellent study by Teller (2008) Since low-income consumers are the main users of street shops, this study was carried out through interviews with 318 low-income consumers at three street shopping districts in São Paulo: São Miguel Paulista, Vila Nova Cachoeirinha and Capão Redondo.This article is organized in four sections besides this introduction: a review of the literature to provide the theoretical framework and more clearly define the investigation; explanation of the methodology, with description of the research procedures; description and analysis of the results; and our conclusions, together with some conceptual implications, managerial applications, limitations and recommendations for future research.

LITERATURE REVIEW
The literature review is organized into three parts.First, we describe the two types of retail agglomerations in more detail.Then we explain the satisfaction and patronage intention constructs, which reflect the responses of the consumers interviewed.Finally, we present the descriptive statistics of these responses.

Retail Agglomerations
Retailers often try to locate their stores near other retail establishments.This tendency for concentration leads to the formation of retail agglomerations (Berman & Evans, 2007, Levy & Weitz, 2009).A retail agglomeration is thus a set of stores near each other, in which the concentration of stores strengthens the attraction of the group of stores.The synergy among these stores benefits consumers through greater convenience and choice while the stores gain by the greater flow of shoppers (Berman & Evans, 2007, Levy & Weitz, 2009).
The two main types of retail agglomerations are unplanned ones, represented by groups of streets in cities where commercial establishments are concentrated, and planned ones, including commercial centers and shopping centers of various sizes and formats.The first grouping is formed spontaneously and evolves without any organizational entity to manage or order it, while the second type is conceived, designed, constructed and launched as a collective retail entity with central coordination (Levy & Weitz, 2009).In this study we compare the attractiveness of these two types of agglomerations in the eyes of consumers.
Shopping malls first emerged in the United States in the early 1950s and the idea spread to Europe at the end of that decade (Anselmsson, 2006, Levy & Weitz, 2009).In Brazil, as in other countries, the development of shopping centers was spurred by the spread of mass car ownership, increasing the traffic congestion in cities, and by the migration of populations with higher purchasing power to areas farther from city centers.The difference between these two agglomerates results from the way they are planned and built and how they are managed and integrated.These aspects have a great impact on the marketing mix -the mix of stores, type and quality of products, accessibility, price range and variety of services offered (Teller, 2008).These different characteristics directly and indirectly influence the capacity of the retail agglomerate to adapt to consumers' needs and hence to determine their attractiveness and competitiveness (Teller, Reutterer & Schnedlitz, 2008).
With the rapid growth of Brazilian cities in the second half of last century, new unplanned retail agglomerates arose, to meet the buying needs of the newly populated regions (Parente, 2000).The typical pattern in Brazil was for most low-income migrants to concentrate on the periphery of large metropolitan areas (Lacerda, Zancheti & Diniz, 2000).
This emerging consumer market wound up favoring the development of substantial retail conglomerates associated with public transportation systems, with a supply of products and services demanded by these new markets (Parente, 2000).The limited mobility of these people, due to the general lack of car ownership, strengthened the option for local commerce.
Despite the great changes in cities and the organization of commerce, street shopping districts still are very important for cities due to the large volume of transactions that occur there.This importance justifies movements to revitalized downtown areas in various countries, including Brazil.The study of an unplanned retail agglomeration offers a diagnosis of the current situation of the respective area, identifying the problems and opportunities, besides evidencing the importance of public and/or private investments for revitalization of these districts (Loukaitou-Sideris 2000;Hutchinson & Loukaitou-Sideris, 2001).Anselmsson (2006) showed the importance of understanding the factors that determine customer satisfaction with the different types of retail agglomerations.According to Devgan & Kaur (2010), the supply of a varied range of stores and services is a prerequisite for a pleasant shopping experience.The understanding of these attractiveness factors is important for the retail sector to offer a set of benefits and to continually adjust them in response to transformations in the consumer market (Johnson & Raveendran, 2009;Guy 1998).

Responses: Satisfaction and Patronage Intention
The basic characteristics or attributes of a particular agglomerate determine the perception of consumers in its respect.These perceptions influence aspects such as satisfaction, frequency of visits and patronage intention, among others.Satisfaction and patronage intention influence customers' behavior and the performance of retail agglomerations (Anselmsson, 2006).
The most widely accepted and utilized concept of customer satisfaction is based on the disconfirmation paradigm of Oliver (1980).According to this paradigm, satisfaction happens when the performance obtained from a product exceeds the consumer's expectations.This concept rests on the comparison between expectations and performance.In the context of retailing, defining expectations with respect to a store or shopping experience and evaluating them in light of the retailer's performance is a complex task.Various studies focused on the determinants of customer satisfaction have recognized the subjective aspects of consumers' evaluation of shopping experiences and have used perceived performance as a construct to explain satisfaction levels (Dabholkar, Shepherd & Thorpe, 2000).
The satisfaction of customers with a commercial agglomeration can be considered as resulting both from a cognitive and affective reaction, related to the personal assessment of are important for each business and targeted customer group.The total set of experiences is expressed and triggered in a highly diversified manner (Anselmsson, 2006;Theler 2008).
The concept of patronage intention, initially developed to understand the relation of consumers with individual stores, includes concepts such as store choice and shopping frequency (Pan & Zinkhan, 2006).The strength of the patronage intention depends on the degree of perceived attractiveness of a certain retail agglomeration and the likelihood of choosing it over other shopping alternatives.Retailer actions and strategies can influence the shopping experience and patronage intention of a store or group of stores (Anselmsson, 2006;Teller, 2008).Baker et al. (2002) proposed a comprehensive store choice model to explain patronage intention, which integrates theories from cognitive and environmental psychology with the proposal of perceived merchandise value of Zeithaml (1988), in which buying decisions are based on perceptions of product quality and price.
In their study on store choice, Baker et al. (2002) present the value variable as a function of the perception of price, quality and shopping experience costs.According to their can influence the satisfaction with the shopping experience and patronage intention of a store or agglomeration (Anselmsson, 2006;Teller, 2008).

Factors That Influence Attractiveness
Many variables influence the purchasing behavior of consumers.We considered the following independent variables based on the theoretical framework: access, store variety, value, parking, sales staff service, infrastructure and image (Berman & Evans, 2007: Dunne & Lusch, 2010;Levy & Weitz, 2009;Teller, 2008).Preliminary research among low-income women in São Paulo, through in-depth interviews, confirmed the relevance of these factors in influencing their choices of retail agglomerates.This preliminary investigation also explored the existence of other particular aspects of the Brazilian context that influence these consumers' preferences.Two other factors -security and social prejudice -emerged from this exploratory research and were incorporated in the questionnaire to obtain the quantitative data.
The serious problem of security that characterizes the daily lives of Brazilian citizens emerged during the preliminary interviews, when the respondents stressed the importance of this factor as an influence on their shopping location choices.Some respondents clearly identified shopping malls as more secure than street retail districts.Another interesting factor that emerged from the exploratory survey, although less explicitly, was a feeling of discomfort some consumers expressed regarding shopping centers, due to perceived discrimination, since these shopping venues are targeted at consumers with a higher socioeconomic level.Coincidentally, the respondents who expressed this feeling were blacks, due to the attitudes of shopping center guards, perhaps a reflection of the still strong racial prejudice in Brazilian society.The quotations below, from two respondents in the preliminary phase, illustrate this unfortunate feeling of discomfort: At the shopping center, the guards give us mistrustful looks (...) I entered a store and he kept glancing at me, so I went and asked him why he was watching me like that, and he said it was his job (...), but he didn't need to act this way (...), so I stopped shopping there.
I went with a group of friends, there were more boys than girls, and the guards followed us around (...) as if we were thieves.
A more varied and balanced mix of stores is important for the success of a retail agglomeration, by increasing consumers' interest and buying, since consumers are more likely to buy in more attractive places (Bellenger et al., 1977;Wakefield & Baker, 1998;El-Adly, 2007;Yiu, 2009).Consumers' perception regarding the diversity and complementarity of the store mix has a considerable effect on their preferences (Dunne & Lusch, 2010, Levy & Weitz, 2009).Unattractive stores can have a negative impact on the entire venue's image and attractiveness, reducing the frequency of visits and sales.As an integrated system, a retail agglomeration needs to be attractive and satisfy consumers in different stages of the buying process (Ingene, 1984;Severin et al., 2001;Teller, 2008;Yiu, 2009).
The ambience of the stores and the venue as a whole also has a strong influence on the shopping experience and emotional and subconscious reactions of consumers, through stimuli, such as music, lighting, temperature and architecture, as well as the interaction with other shoppers (Slatten et al., 2009).To survive and remain attractive, retail agglomerations need to offer a minimum structure that allows access to shoppers using public transportation and those with their own cars, as well as comfort items, such as restrooms and parking lots, among others (Ratchford, 2004, Arenntze & Timmermans, 2001).The design and layout of stores are also aspects that can influence the shopping experience and increase the time spent in the store or center (Baker et al., 2002;Bearden, 1977, Severin et al., 2001).Another factor that exercises a strong influence on the shopping experience satisfaction is the quality of the relationship with the store and the service offered by the sales staff (Berman & Evans, 2007;Dunne & Lusch, 2010;Teller, 2008).
In choosing where to shop, a key factor is the perception of value of different retail venues, planned or not.The value perception results from comparison of the cost-benefit ratio from the choice made, including the sacrifices perceived as necessary, including nonfinancial ones (Zeithaml, 1988).Baker et al. (2002) used the value construct to measure the relation between consumer perception of store environment cues and patronage intention.Value perception goes beyond products, because it is also related to other points that satisfy psychological needs, even subconscious ones, such as socialization, information and selfgratification, among others (Baker et al., 2002).
Image can be defined as the way the retail agglomeration is defined in the shopper's mind, partly by the functional qualities and partly by psychological attributes (Martineau, 1958).This definition encompasses both attributes or factors with objective and rational characteristics, such as variety and price, and more subjective and emotional aspects, such as feelings from psychological attributes like belonging and acceptance, which add meaning to the shopping experience (Lindquist, 1975).A shopping venue's image is not just the sum of the individual images of each of its stores, but also of other overall characteristics, such as infrastructure and atmosphere.
Studies of consumers' retail buying behavior also recognize the influence of age and income on their responses to the shopping environment (Baker, 2002;Pan & Zinkhan, 2006;Kumar, 2008;Konus, 2008).Based on the possible impact of these two demographic variables, we also investigated the influence of these two factors as moderating variables.

METHODOLOGY
In this work we drew particular inspiration from previous works that have compared the two types of retail agglomerations, specifically those of Teller (2008) and Teller, Reuterer & Schnedlitz (2008).The questionnaire was enriched by the addition of the security and social prejudice constructs, resulting from our exploratory research through in-depth interviews with 15 low-income consumers.Our goal with this exploratory step was not only to identify peculiar factors in the Brazilian context, but also to validate the selection of the factors suggested in the literature.
The questionnaires were pretested with consumers of the same profile as the sample and in a similar setting as that where the main data collection took place.After the pretest, we engaged in a discussion with six retail specialists to finalize the questionnaire (Appendix).We (1998), Ingene (1984), Reinartz & Kumar (1999), Ruiz et al. (2003), Severin et al. (2001), Van Kenhove et al. (1999), Wakefield & Baker (1998), Woodside & Trappey (1992) and Burns & Warren (1995).For the prejudice construct, we adapted the scale developed by Brumbaugh & Rosa (2009).The appendix presents the scales adopted in this study.
For treatment of the quantitative data we carried out descriptive analyses together with exploratory factor analysis (AFE).The descriptive analyses allowed us to start understanding the characteristics of the sample and to confirm that the participants indeed came from lowincome households.Our aim in this exploratory factor analysis was to validate the scales used (convergent and divergent validity) and to extract the variables used in the modeling for the multivariate analysis.To assess the main factors that affect the satisfaction and patronage intention for street retail districts and shopping centers, we carried out multivariate regression analysis.This type of analysis provides a robust test of the association between independent and dependent variables in cross-section studies (Hair, 2008).
We also tested the "income" and "age" variables as moderators of the relations between the attractiveness factors and the two dependent variables.The moderation effect occurs by means of an interaction between the moderating variable and the independent variables.This influence can attenuate the positive or negative effect of each factor, reverse the sign of the relation of the attractiveness factors with the satisfaction or patronage intention for the two shopping venues, or also accentuate the effect of the factors on the dependent variables.The moderation test was carried by the slope difference test, as found in previous articles (Dawson & Richter, 2006).
Finally, we conducted cluster analysis to identify the existence of consumer groups with similar characteristics in buying behaviors.

RESULTS
We interviewed 110 consumers at the São Miguel Paulista (SMP) shopping district, 112 at Vila Nova Cachoeirinha (VNC) and 96 at Capão Redondo (CR), for a total of 318.The sample had the following characteristics: 78% with up through nine years of schooling; 86% with monthly household income up to R$ 1,750; and average age of 36 years.There were no significant differences in the customer profile of the three retail venues.
Regarding means of transportation, the results in Table 1 show the existence of substantial differences in the percentages of access by car and on foot.Of the respondents, 17% came to the street shopping district by car, while 38% used a car to go to their preferred shopping center.Another marked difference was access by walking: 32% of the respondents arrived on foot at the venue of the interviews, but only 2% said they went to a shopping center on foot.The percentages for the two types of shopping agglomerations can be seen in Table 1.
The percent of customers arriving on foot at Capão Redondo is much greater than for the other two.This result can be explained by its smaller dimension in comparison with the other two, which restricts its attraction power.

Validation of the Theoretical Constructs
Since there were equal questions to measure the attractiveness of the street retail districts and shopping centers, and our research interest was to compare the factors and their weight in determining the attractiveness, we carried out exploratory factor analysis (EFA) of the constructs referring to the responses given by the respondents.This procedure also served to statistically validate the theoretical construct measures on the questionnaire.We performed the same EFA for the dependent variables.After finalizing the analysis of the factor model and naming the factors, it was possible to compare the averages of the factors between street retail district and shopping center, besides comparing the influence of the factors on satisfaction and patronage intention, for the two types of retail agglomeration in question.
The EFA showed that the factor model was adequate for the sample data for the retail districts, because the commonality extracted from the factors, in at least 90% of the questions, was greater than 0.60, the KMO index was 0.8, and the Bartlett test was significant the 1% The attractiveness factors based on the EFA carried out for the questions on the two types of retail venues, in order of variance explained, with the variance percentage in parentheses, are: value (9%), prejudice (8%), access (8%), environment (7%), security (7%), parking (6.5%), sales staff service (6%), store variety (6%) and image (5%).As the first relevant results of this study, we can mention that the theoretical constructs used in studies of other countries were corroborated in this survey conducted for retail venues with focus on low-income consumers.Another point was the variance explained individually by the nine factors, which indicates that none of them individually had high power to explain the variation of the data of the sample.This can indicate the need for retailers to pay attention to all the factors.
We carried out the same validation procedure as for the constructs composing the attractiveness factors for the dependent variables as well.The EFA indicated the factor model was adequate for the sample data, because the commonality extracted from the factors was greater than 0.65, the KMO was 0.73 and the result of the Bartlett test was significant at the 1% level.The variance explained by the factors of the final model was 71%, above the threshold level indicated by other authors (Hair, 2009;Maroco, 2007).Although the differences between the results for the three districts were not statistically significant, the satisfaction and patronage intention were highest for São Miguel Paulista (SMP) and lowest for Capão Redondo (CR), suggesting that districts with a larger number of stores and products are more attractive.
Table 3 presents the average evaluations of the attractiveness factors for the two types of retail agglomerations.The shopping centers received more positive assessments in the factors store variety, comfort and services, environment, safety and parking, at p-value ≤ 0.05.
This result explains the better overall satisfaction assessment for shopping centers.In contrast, the better evaluation of the street districts in the value (benefit/cost) and access factors can explain the higher patronage intention for this type of venue.The greatest difference in the assessments between the two types of retail agglomerations was the infrastructure factor (comfort and services), which was very unfavorable for the street districts (1.6), in contrast to the positive evaluation of shopping centers (4.3).The descriptive analyses, EFA and tests of the means performed and discussed so far allow distinguishing the scores given to satisfaction and patronage intention of each retail agglomeration and shed light on the attractiveness dimensions in the eyes of the respondents.
However, they are not suitable to make any inference about the level of influence of the descriptive variables and attractiveness factors generated by the EFA on the consumers' responses (satisfaction and patronage intention).For this purpose, below we present the results of a multivariate regression analysis in which we tested the influence of the attractiveness factors on the consumers' responses.

Regression Analysis
To test the intensity of the association of the attractiveness factors with satisfaction and patronage intention, we performed multivariate regression (MR).The independent variables were the eight attractiveness factors found in the literature review and the two new dimensions discovered in the preliminary research (security and prejudice), while the dependent variables were consumer satisfaction with street retail districts (SRD_SAT) and shopping centers (SC_SAT) and patronage intention for retail districts (SRD_PI) and shopping centers (SC_PI).
Before performing the association analysis, we carried out a correlation analysis of the attractiveness factors to assess the interdependence of the independent variables.As expected, due to the use of the EFA in the initial step, the matrix showed low pairwise correlation between the factors for the street districts and shopping centers.The four final models presented sufficiently large and significant F-statistics, with p-values ≤ 1%.The adjusted R 2 values were acceptable for the MR models for the purpose of testing the association of the variables.The results of the four multiple regression models are presented in Table 4.As shown in Table 4, value and image were the two factors that stood out most as significant in all four regression models.This is a natural result, since these two factors reflect, in a certain way, the general assessment of consumers of the shopping venues.Some factors were particularly significant for the street retail districts.The environment factor, for example, was only significant as an influence on the satisfaction for the street venues, with this having the highest coefficient among all the betas in the four models.This result suggests the importance of improving urban equipment to enhance consumer satisfaction.The store variety factor was significant in the satisfaction and patronage intention for the retail districts.This result suggests that these venues, due to their format and mix of stores more aimed at lower income shoppers, manage to meet the needs of this consumer segment more fully.The influence of the access factor on the satisfaction and patronage intention for the retail districts can be due to the importance of public transportation for low-income shoppers with limited mobility because of low car ownership.The security factor also appears to exercise a significant influence on the street shopping venues, reflecting consumers' strong concern over the high level of insecurity that characterizes the streets in large Brazilian metropolitan areas.
BBR, Vitória, Special Edition BBR Conference, p. 154-179, 2012 www.bbronline.com.br In turn, the infrastructure factor exercises a significant influence on the patronage intention for shopping centers.This response can indicate how consumers value the greater comfort offered by shopping centers: they are typically air conditioned and protected from the sun and rain, with restrooms, food courts and rest areas.The prejudice factor was also significant and negatively associated with customer satisfaction with shopping centers.This finding is particularly relevant because it indicates that low-income shoppers suffer negative experiences due to prejudice at shopping centers.In contrast, the street retail districts are more democratic spaces, as revealed by the lack of responses indicating prejudice or snobbery at these venues.
We also conducted moderation tests for income by estimating four new models with a new set of independent variables.We did the same for age moderation tests.The interaction was operationalized by multiplying the income variable (five income levels) and age variable (numerical variable) by each attractiveness factor.The moderation tests were carried out by the slope difference test, as found in previous articles (Dawson & Richter, 2006).
None of the eight new models presented stronger explanatory power than the four previous models, estimated without the moderator variables.The R 2 values of the MR models with the inclusion of the income and age moderator variables were similar to those of the models without this inclusion, and the F-statistics declined, although they maintained their level of significance.Therefore, the inclusion of the moderator variables did not improve the general parameters of the models, indicating that the income and age variables do not exercise a moderating influence on the associations found in the four previous models without moderation.

Investigating the Existence of Segments of Consumers
To better understand the behavior of consumers regarding retail agglomerations, we conducted a cluster analysis according to Malhotra (2004) and Maroco (2007), with the aim of investigating the existence of different segments of consumers among the interviewees.In this respect, we sought to identify segmentation for satisfaction and patronage intention for each of the retail agglomerations based on the attractiveness factors listed in Table 3. From the hierarchical cluster analysis we identified two clusters.To assess the characteristics of these clusters and identify if there are significant differences between them, we used the K-means cluster analysis technique from the centroids of the attractiveness factors.
BBR, Vitória, Special Edition BBR Conference, p. 154 -179, 2012 www.bbronline.com.brWe found two groups, one with 146 individuals and the other with 172 individuals, identified respectively as "more demanding" and "less demanding".These names were based on the means of each of the three most representative differentiation factors between the two types of shopping venues -value, environment and security -identified based on the ANOVA results, due to the high variability between the clusters (indicated by the cluster mean square) and low variability within the groups (indicated by the error mean square).The demographic variables were not significant to characterize the two groups.
The value factor obtained an average of 3.5 for the individuals belonging to the less demanding group, while it was 4.34 for the more demanding shoppers (inter-group variability of 92.33 and the intra-group variability of 0.74).For the environment factor, the average was 2.48 in the less demanding group and 3.95 for the more demanding group (inter-group variability of 169.62 and intra-group variability of 0.87).Finally, the average for the security factors was 1.86 for the more demanding shoppers and 3.41 among the less demanding ones (inter-group variability of 188.33 and intra-group variability of 0.94).Regarding the satisfaction with the retail districts, the more demanding consumers had a lower perception of satisfaction than the less demanding ones (SRD_SAT more demanding = 3.5; SRD_SAT less demanding = 4.2).The satisfaction variables for shopping centers and the patronage intention for retail districts and shopping centers did not present significant differences (SC_SAT more demanding = 4.0 and SC_SAT less demanding = 4.14; SRD_PI more demanding = 4.44 and SC_PI less demanding = 4.82).
For the street retail districts, these results can be worrying, because they appear to suggest that more demanding consumers, because they are less satisfied with streets hooping venues, are more likely to migrate to shopping centers.

CONCLUSIONS, LIMITATIONS AND FUTURE RESEARCH AVENUES
The purpose of this investigation was to gain a better understanding of the A particularly noteworthy result is that while the shopping centers were better evaluated on the satisfaction factor, the street retail districts received a more favorable rating in patronage intention.This result indicates that the latter type of retail venue still retains a certain level of preference as a local shopping option, despite the greater satisfaction generated by shopping centers.This higher level of satisfaction with shopping centers can be explained by the better assessment they received in store variety and other factors that make the shopping experience more agreeable, such as factors linked to environment and infrastructure: comfort and services, environment, security and parking.On the other hand, the better result on patronage intention can be explained by aspects of a more rational nature that were better rated for street retail districts, such as ease of access and value, or a better benefit/cost ratio.
The greatest difference in the evaluations between the two types of retail agglomerations occurred for the infrastructure factor (comfort and services), which was very unfavorable for street districts (1.6), in contrast with the positive evaluation of shopping centers (4.3).This result points to the need for store owners to create merchants' associations to coordinate efforts, along with government authorities, to improve the urban equipment (sidewalks, lighting, parking) as well as the security at these retail venues.
In the regression analysis, the four final models presented results for the F-statistic and adjusted R 2 values that were acceptable for multiple regression models with the aim of testing the association of variables.The results of the four MR models are presented in Table 4.
Value and image were the two factors that stood out as significant in all four regression models.This is a natural result, because these two factors reflect, in a certain form, an overall evaluation of consumers of the agglomerations investigated.
Different factors were respectively significant for the shopping center models and the street retail district models.The factors that were particularly significant for the models assessing the street districts were access, security, environment and store variety.On the other hand, the infrastructure and prejudice factors were more significant in the shopping center models, with the prejudice factor having a negative relationship with consumer satisfaction with shopping centers.This finding is especially relevant because it indicates that low-income consumers suffer negative experiences due to prejudice at shopping centers, in contrast to street retail districts, which are more democratic spaces and for which no negative responses were recorded regarding prejudice.These results reinforce the need to contextualize studies of the perception of consumers and their behavior when the focus is on consumers with lower purchasing power.
The results of the moderation tests indicated that the variables "income" and "age" do not moderate the relations of the attractiveness factors and the two dependent variables.The absence of a moderating effect of the income variable can be explained by the fact that the sample was restricted to low-income consumers.An influence of income level would likely occur in studies of consumers with heterogeneous economic characteristics.
The results taken together indicate a serious vulnerability of street retail districts in relation to shopping malls.Not only did shopping centers receive a better overall satisfaction rating, but this bias was much stronger among the more demanding consumers, who expressed a negative perception of street retail districts.The only advantages of these venues over shopping centers were associated with better access and better benefit/cost ratio.Among the disadvantages, the standout was infrastructure.The economic improvements that have been occurring in Brazil with respect to low-income segments are not only increasing the mobility of this segment, due to greater car ownership, but they are also making this segment more discerning.In this process, the advantage of ease of access of street districts is declining while the dissatisfaction over their lack of infrastructure is climbing.The construction of shopping centers in low-income regions is accelerating this process of deterioration of street retail districts.
Previous studies have identified the relevance that unplanned retail agglomerations have in the dynamics of their surrounding areas and of cities as a whole.In Brazil these retail agglomerations are democratic arenas that permit the greater insertion of low-income people in the consumer market, because in general they are closer to shoppers' homes and better manage to meet the needs of this consumer segment.To prevent street retail districts from facing an accelerated process of decadence, with the consequent social deterioration in their surrounding areas, it is essential to develop public policies to restore the attractiveness of these agglomerations, by means of direct public investments in urban equipment (such as sidewalks, lighting, rest areas, public restrooms) and to improve their security, as well as BBR, Vitória, Special Edition BBR Conference, p. 154-179, 2012 www.bbronline.com.brthrough tax incentives to private investments for these purposes and the establishment of effective public-private partnerships to improve these shopping venues.

Limitations and Suggestions for Further Research
The main limitation of this study is its restricted scope, since the sample was limited to shoppers at three street retail districts located in low-income regions of the city of São Paulo.
Therefore, the results reflect the specific local context investigated.Future studies with larger and more diversified samples and in other regions would provide a richer insight into the phenomena studied here and allow greater generalization of the findings.We also recommend comparative surveys among consumers of different economic levels to identify how income affects the assessments and shopping behavior at these two types of retail agglomerations.
Since the interviews were conducted in the main streets of the three retail districts, we did not manage to implement a probabilistic sampling method.Also, there may have been a response bias because no interviews were conducted at shopping centers.We therefore suggest new studies be formulated to overcome these limitations, by including larger samples in a greater and more diversified number of regions.The interviews can also be conducted at a "neutral" site -a place that is neither a street retail district nor a shopping center.
The identification of moderating and mediating variables and the evaluation of their impact can shed more light on the satisfaction and patronage intention of consumers in relation to retail agglomerations.A more heterogeneous sample would allow performing moderation tests with variables such as income, ethnicity (color/race), car ownership and shopping motivation (utilitarian or hedonistic).The value perception (from the standpoint of benefit/cost) could also be investigated as a mediator variable between the attractiveness factors and buying behavior, as already tested in other retail contexts (Machleit et al., 2000;Eroglu et al., 2005).Based on the attractiveness factors that were significant in this study, such as image and value, the use of experiments would be useful to give more robustness to the results obtained here and would also be an alternative to measure the moderation and mediation relations.
In the preliminary interviews, the consumers indicated that stores crowded with and/or shoppers?Although this environment factor has received attention in other countries (Machleit et al. 2000;Eroglu et al., 2005, Brandão et al., 2010), there have been none in Brazil.
We hope this work serves to increase awareness of public policymakers of the importance of preserving street retail districts.We also hope that the academic community will feel motivated to undertake further studies to obtain a better understanding of the determinants of the vitality of street shopping venues, by investigating successful revitalization projects and offering innovative solutions to prevent their decline, thus serving to increase awareness of public officials of this theme.
, carried out in Vienna, Austria, investigates the satisfaction and intention to buy of consumers between shopping malls and street retail districts.But in a search of Brazilian academic magazines and the annals of events sponsored by the National Association of Graduate Study and Research in Administration (ANPAD) over the past five years, we did not find any similar comparative studies for Brazil.The importance of the theme and the scarcity of information on the peculiarities of Brazilian reality motivated the development of this investigation, to better understand the phenomena of consumer preferences for different types of retail agglomerations.More specifically, our aim was to answer the following questions: What are the preferences and buying behavior of low-income consumers at street shopping districts?What factors explain the satisfaction and buying intention of consumers in choosing between shopping malls and street retail districts?How is this perception of attractiveness configured in different consumer segments?What should proprietors of street shops do to face the growing competition from shopping centers?The general objective of this study is to develop a better understanding of the preference of low-income consumers for these two retail agglomerations.The specific aims are: (1) to evaluate and compare, among low-income consumers in the city of São Paulo, the levels of satisfaction and patronage intention for retail districts and shopping centers; (2) to estimate and compare the assessments of consumers regarding the various attributes of these two shopping experiences; (3) to identify and compare the evaluation of the attributes that affect the relative attractiveness of street shopping districts and shopping malls; (4) to evaluate how the variations of each of these attributes (independent variables) are associated with the attractiveness variables (dependent variables: satisfaction and patronage intention) of these two types of retail agglomerations; (5) to identify the existence of different segments of consumers with homogeneous behaviors and characteristics in relation to these agglomerations; (6) to encourage the academic community to carry out new investigations on this important theme; and (7) to encourage the development of public policies to reverse the decline of high street retail districts.
the experiences at the shopping venue.Consumers who are satisfied with the experience will tend to remain faithful.The managers of individual stores and of retail agglomerates as a whole (be it shopping center managers or members of local merchant associations) need to understand how and to what intensity different attraction factors exercise on consumer responses.For this purpose, retailers need to know what customer satisfaction determinants applied the questionnaires in person to 318 consumers at three important retail venues (unplanned agglomerations) mainly patronized by low-income shoppers in the city of São Paulo.The interviews were with women because they make up the majority of shoppers at such retail centers.Since the aim was to compare the respondents' behavior between these two shopping formats, we applied a filter in selecting the respondents, only including those who stated they shopped both at the street shopping district where the interviews were being conducted and a shopping mall as well.We chose three retail venues located in low-income regions of the city, by means of a judgment sample, seeking to represent retail agglomerations of varied sizes and different regions.In this respect, we took advantage of the data from a previous study of street shopping districts conducted in the city of São Paulo byParente,   Miotto & Barki (2008).The regions chose were: São Miguel Paulista, a large shopping area (some 700 stores) in the eastern part of the city; Vila Nova Cachoeirinha, a medium-size venue (480 stores) in the northern part of the city; and Capão Redondo, a small venue (150 stores), in the city's southern zone.BBR, Vitória, Special Edition BBRConference, p. 154-179, 2012  www.bbronline.com.brWe operationalized the constructs of the two dependent variables and eight independent variables suggested by the literature based on the scales prepared by Teller, who in turn relied on the studies ofAlzubaidi et al. (1997), Arentze & Timmermans (2001),Arentze et al. (2005), Baker (2002),Bearden (1977), Bellenger et al. (1977), Dellaert et al.

F
Notes: The significance of the coefficients and F-test are represented as follows: *≤0.01; **≤0.05;***≤0.10.The DV (dependent variables) are: RD_SAT = satisfaction with street retail districts, SC_SAT = satisfaction with shopping centers, RD_PI = patronage intention for street retail districts, SC_PI = patronage intention for shopping centers.The dependent variables are the attractiveness factors generated by the EFA.
phenomenon of attractiveness of street shopping districts located in low-income neighborhoods in the city of São Paulo, by means of the perception of consumers, in relation to the attractiveness of shopping centers.The findings can help to construct a theoretical reference on the responses of consumers to this important type of unplanned retail venue.The in-depth interviews conducted during the exploratory phase of the study not only confirmed the relevance of the eight factors identified in the literature review of studies conducted in other countries, it revealed security and prejudice as two other possible factors that influence the attractiveness of retail agglomerations to low-income consumers.The BBR, Vitória, Special Edition BBR Conference, p. 154-179, 2012 www.bbronline.com.brresults show the variance explained individually by each of the 10 attractiveness factors is similar, indicating that none of them in isolation have high explanatory power of the variation of the sample data, suggesting these factors exercise a synergetic effect on consumers' preferences.
shoppers and/or products can signal both attractive offers and discomfort in the shopping experience.A new line of research could investigate how these two crowding factors affect the perception of value, prices, satisfaction and patronage intention of consumers in different types of retail agglomerations.How do consumers react to stores overloaded with products BBR, Vitória, Special Edition BBRConference, p. 154 -179, 2012  www.bbronline.com.br

Table 1 : Way of arriving at the street shopping district/shopping center
The variance explained by the final model's factors was 62.6%, above the threshold value indicated byHair (2009) and Maroco (2007).

Table 2
presents the average satisfaction and patronage intention measures for the two types of agglomerations and for each of the three street retail districts.Of particular interest is that shopping centers were better assessed regarding satisfaction while the street retail districts received a more favorable evaluation regarding patronage intention.This result indicates that street shopping venues retain a certain preference for local shopping despite the greater satisfaction generated by shopping malls.The t-tests showed that the difference between the satisfaction and patronage intention metrics for the two retail agglomeration types was significant, with p-value ≤ 0.01.BBR, Vitória, Special Edition BBRConference, p. 154 -179, 2012  www.bbronline.com.br