Changing negative attitudes through sport sponsorship in Brazil: a perspective of sport involvement

Organizations have been visualizing sport through strategic lens, in particular, as a platform to improve their corporate image and brand positioning1. Nowadays, sponsorship is one of the marketing strategies most commonly used2. Particularly, Speed and Thompson3 argued that sport sponsorship has the potential to add value and influence consumer attitude toward the sponsoring brand, offering also the potential to generate a sustainable competitive advantage in the marketplace4,5. Additionally, Mason6 states that the key to the effectiveness of sponsorship is the understanding of how consumer attitudes are formed and changed. The same author6 explains that attitudes have a strong impact on consumer behavior. In this context, the understanding of the implications of sport sponsorships on attitude is crucial to the success of sponsorship initiatives. In another prism, constantly, organizations have to deal with crises and scandals that may create negative attitude from the consumer point of view. For instance, corruption has been a problem for a wide number of countries, which affects their institutions, labels and the attitude of their citizens. Facing these kinds of moments, when the brand is negatively assessed, organizations should plan marketing efforts in order to reverse this scenario or minimize the consequences of a Abstract

Organizations have been visualizing sport through strategic lens, in particular, as a platform to improve their corporate image and brand positioning 1 . Nowadays, sponsorship is one of the marketing strategies most commonly used 2 . Particularly, Speed and Thompson 3 argued that sport sponsorship has the potential to add value and influence consumer attitude toward the sponsoring brand, offering also the potential to generate a sustainable competitive advantage in the marketplace 4,5 . Additionally, Mason 6 states that the key to the effectiveness of sponsorship is the understanding of how consumer attitudes are formed and changed. The same author 6 explains that attitudes have a strong impact on consumer behavior. In this context, the understanding of the implications of sport sponsorships on attitude is crucial to the success of sponsorship initiatives.
In another prism, constantly, organizations have to deal with crises and scandals that may create negative attitude from the consumer point of view. For instance, corruption has been a problem for a wide number of countries, which affects their institutions, labels and the attitude of their citizens [7][8][9][10] . Facing these kinds of moments, when the brand is negatively assessed, organizations should plan marketing efforts in order to reverse this scenario or minimize the consequences of a Abstract This research assesses whether a sport sponsorship is able to change a negative attitude towards a brand. In this way, investigating the influence and potentialities of sport sponsorships on Brazilians consumers' attitude and on their consumption intentions. In addition, this paper also evaluates the effect of the sport sponsorship considering the level of sport involvement. An opinion research and a questionnaire were chosen to shape the data collection instrument. The questionnaire was segmented in scenarios, where, at first, negative information about the company was provided in order to create a negative evaluation of a fake brand. Then, the proper independent variable was inserted, being the sport sponsorship implanted through new sets of information (football, projects to the youth and sponsorship of a sport and athletes in an Olympic and Paralympic context). In front of each scenario, the participant was asked to express their attitudes and consumption intention regards the brand. In general, observing a cumulative effect, the findings provide evidence that the sport sponsorship is capable of change a negative attitude towards a brand. The descriptive results, the Wilcoxon Signed-Rank test and Spearman correlations supported the ideia that sport sponsorship can have this positive effect, having a similar positive trend in people with high or low sport involvement. However, the Olympic frame presented only a small effect in enhancing the attitude level in all involvement groups. In front of negative scenarios, lasting investments in different platforms of sport sponsorships seem like an effective strategy, affecting positively the purchase intention as well, with similar results regardless the levels of sport involvement, but a slight better performance of a high sport involvement group.
Keywords: Sport Sponsorship; Attitude; Sport Involvement; Marketing. possible crisis. Under these circumstances, the sport sponsorship might be a useful tool to change a negative evaluation, supporting a sponsoring organization to keep or achieve a positive attitude towards its brand.
For this reason, this study assesses if a sport sponsorship is able to change a negative attitude towards a brand, turning it into neutral or positive one. In this way, this research investigates the influence and potentialities of the sport sponsorships on the consumers' attitude towards a sponsor and on their consumption intentions. Additionally, this paper also evaluates the effect of the sport sponsorship considering the level of sport involvement.
Several points may justify this article, although the most important one is the lack of similar studies. Observing six important sports marketing journals (the Journal of Sport Management, Sport Management Review, European Sport Management Quarterly, International Journal of Sport Management, International Journal of Sports Marketing and Sponsorship and the Sport Marketing Quarterly), chosen based on a list of 12 most relevant journals in the field 11 , and also consulting the Brazilian online library platform, called "Periódicos Capes", no article or other academic work was found with the same objective, method/premises or culture investigated (country). The searches took as basis articles from 2000 to 2016, with the terms "sports marketing", "sponsorship" and/or "attitude" in the title (verified in English and Portuguese too).
In fact, studies have addressed the issue of sport sponsorship and negative attitudes through different frames 3,[12][13][14][15] . However, none have started their observations with negative attitudes towards the studied brands as a premise. Besides that, as emphasized by Low and Pyun 16 , there is also a lack of studies touching small-scale amateur sporting events benefits (e,g.: community events). Moreover, the effects of multiple sponsorships from a sponsoring brand need more investigations as well 17 . Another justification came from the massive concentration of marketing research on European countries and USA 15 . For this reason, Brazil is the focus of this investigation, with a diversified culture, sport involvement and the biggest economy in South America. The studies of Francisco et al. 18 , Fagundes et al. 19 and Mazzei et al. 20 also illustrate the necessity of more articles about sports marketing in Brazil.

Attitude
Fishbein and ajzen 21 defined attitude as a learned predisposition to respond in a consistently favorable manner with respect to a given object. Around this predisposition, eagLy and chaiken 22 understood the concept as a psychological tendency that is expressed by evaluating a particular entity with some degree of favor or disfavour, being relatively stable and organized to react and manifest through opinions and acts towards a presented attitudinal object 23 . Triandis 24 explains attitude being composed of three factors: affective, cognitive and behavioral. The cognitive category is formed by thoughts, beliefs and perceptions about an attitudinal object. On the other side, the affective one refers to feelings and emotions associated with that object (evaluative connotation). The last one, behavioral, regards to intentions to act, such as a purchase decision 25 . In parallel, Torres 26 claims that attitude components are inter-related, but some inconsistencies may occur.
Furthermore, there is no common sense about how the structure of attitude should be observed. As reported by coeLho 27 , authors consider the concept as being three-dimensional, two-dimensional (cognitive and affective only) or even as onedimensional (just affective). For instance, PorTo and oLiveira-casTro 28 defend that the behavior can be predicted by affections deprived of a cognition process, insofar as the affective element represents an assessment towards an object 29 .
McLeod 30 brings that the Semantic Differential scale is a direct tools to measure attitudes. yusoFF and janor 31 explain that this scale usually consists of a series of 7-points or 5-points bipolar rating scales, with descriptive items containing two opposite adjectives in meaning in each item. The acceptance around this scale may be explained since the evaluative dimension of it has been mostly used by social psychologists as a measure of a person's attitude, because of this dimension reflects the affective aspect of the attitude 30 . As a matter of fact, marketing literature has acknowledged the use of social psychology to appreciate the complexity of the attitude-behaviour interactive process 32 .

Attitude modeling and change
Currently, ideas and theories about how an attitude can be changed or modeled are focused on the affective and cognitive elements. eagLy and chaiken 22 , for instance, detail the Balance Theory, which reveals that people prefer to have balance, order and harmony in their lives. Therefore, every time that there is an imbalance, people would change their attitudes (and/or behavior) in a way that would restore balance. Likewise, FesTinger 33 describes the Cognitive Dissonance Theory as a valuable framework to explain attitude changes. In line with the author, cognitive consistency is a desirable condition, where the inconsistencies among the cognitions generate feelings of displeasure. Thus, people are induced to reduce this feeling through amending their cognitions. Another theory, the Affective-Cognitive Consistency Theory of rosenberg 34 , examines the relationship between "attitudes" and beliefs, affirming that an unstable state occurs when an individual's attitudes toward an object and the knowledge about that object are inconsistent. To resume, the provision of new information (persuasion) that changes the cognitive component might cause an overall change in the attitude toward an object, bringing the knowledge and affect into harmony.
Additionally, edwards 35 remembers that attitudes formed on the basis of affect are more influenced by affective persuasions. However, attitudes formed via cognitions were equally influenced by affect and cognitive persuasions. As a whole, analyzing all these models and theories, the cognitive element is always preceded, at least in some level, by affective responses 36 . As a result, the affective component of the attitude should receive special attention in order to alter or model an attitude towards a brand. Especially, observing the thought that, when time is limited, feelings drive intuitive decision-making process, rather than logic or rationality 37 .

Attitude impact on purchase intention (PI)
Basically, an attitude towards a brand does not necessarily correspond with the consumption behavior related to it, characterizing a nondependent relation 38,39 . However, in line with kiM, ko and jaMes 39 , studies have shown that consumer's attitude is an important predictor of behavioral intentions. With this relation in mind, the intention to purchase a sponsor's product, for example, will increase if the sponsor develops a favorable attitude toward itself 40 . Additionally, other theories might be observed to clarify how attitude can influence intentions and behaviors. For example, the Reasoned Action Theory from ajzen and Fischbein 41 , describes that people act in accordance with their intentions and perceptions of control over their behavior, being intentions (such as to consume) influenced by attitudes toward the behavior 39 .
Additionally, the influence that attitude has on intention has been recognized as the final step in a process of image transfer 42 , which would be a measure of effectiveness of sponsorship 43 . In fact, purchase intention is also considered an indicator to measure sponsorship results 44 . In a different manner, sponsor awareness is an important piece in a consumer's decision to purchase a sponsor's product as well, but awareness alone is not capable of determine intentions to purchase, requiring a positive attitude towards those sponsors 44 .

Sport involvement and attitude
Involvement is defined as a state of motivation, arousal or interest toward an activity or product 45 . Specifically, sport involvement can be seen as the active interest in, engagement with, and commitment to a sport 46 . shank and beasLey 47 pointed out that involvement is a predictor of behaviours, considered also a variable for modelling the sport consumer behaviour. Based on MuLin, hardy and suTTon 48 and aLba 49 there are three basic elements of sport involvement: cognitive (e.g.: the knowledge about the sport), affective (e.g.: emotions around a team) and behavioural (e.g.: commitment to attend in a home game), where the consumer behaviour is generated by the first two forms. beasLey 50 indicated that elements of involvement may influence the cognitive and affective dimension of attitudes, where the construct can be understood as moderating the effect of attitude on intention to purchase 4 . As reported by ko et al. 44 , favourable purchase intentions were more likely to occur when consumers held a positive image (attitude) of the sponsoring companies and had a high level of sports involvement. Moreover, highly identified fans tend to have more favourable attitudes toward a sponsor and are more likely to consume the sponsor product 51 .
Studies about sport sponsorships have concluded that sport involviment positively leverages purchase intention 4,52,53 . According to gwinner 54 , the level of involvement with a sporting event will have a moderating effect on the sponsorship stimulus processing. In this way, more a person is involved with a sport greater is the probability of he/she to buy a product of a sponsor of that sport 55 . The work of shank and beasLey 47 also indicated that involvement was found to be positively associated with the ability to identify sport sponsors. Additionally, Turner and rao hiLL 56 brought that involvement is then influencing future attendance motivation, brand recall, brand recognition and/ or shaping brand and product category attitudes. Furthermore, the cognitive and affective elements of involvement are positively correlated with hours spent watching sports on tv, attending sport events and participating in sports 47 .
Indeed, as stated by aLexandris, Tsaousi and jaMes 57 , segmenting spectators according to their levels of sport activity involvement can facilitate desired sponsorship outcomes, after all, there is a positive orientation toward the sponsor who concede benefit on a consumer's favourite activity 52 . In the same way, when a consumer has a high involvement with an athlete and he/she is sponsored by a brand, for example, this celebrity endorsement leads to a favourable attitude toward that brand (biscaia et al. 40 ). Worth mentioning, about the Elaboration Likelihood Model (ELM), PeTTy and cacioPPo 58 indicate that attitudes may be formed by making less reflexive decisions, in other words, more involved individuals process information by an outlying route, more emotional.
On the other hand, other authors claim that attitudes towards the sponsor positively influence purchasing intent regardless of the level of consumer involvement with the event 42 . ToLedo and andrade 53 brought up that fan involvement may have positive and negative effects, for instance, when too high the involvement, it can create boycotts from rival fans over a sponsor´s product related to sponsoring a rival team. Besides that, PhaM 59 suggested that sport involvement may have an inverted-U effect, that is, as involvement to process the event reaches high levels, attention becomes more focused on the relevant sources of information (e.g. the game itself ) and away from irrelevant information such as sponsorship stimuli.

Sponsorship and attitude
Sport sponsorship is considered the investment in an event, athlete or team in order to ensure the sponsor access to commercial potentialities associated with sport properties 60 . kiM and choi 61 discuss that corporations pursue to sponsor a wide range of events and not-for-profit organizations aiming at different target, but sponsorship of sports events has been the most popular communication tool among companies. Generally, in terms of objectives, authors such as cornweLL and Maignan 62 and Fahy, FarreLLy and QuesTer 60 claim that an essential goal of the sponsorship is typically to create or maintain a positive attitude toward the sponsor.
To FarreLLy, QuesTer and burTon 63 , consumers' attitudes towards the brand involved in the sponsorship are a measure of the effectiveness of a marketing effort, specially because measuring awareness alone is not enough when evaluating the effectiveness of a sponsorship 4,64 . Likewise, zucco et al. 65 expousured that sponsorship has the capability of changing an attitude towards a brand, mainly if the costumer perceives a positive fit between the sponsor-property and the sponsoring brand 4 . sPeed and ThoMPson 3 bring that sponsorship may be used to communicate symbolic brand associations, as part of a process to change the image of the brand. Moreover, this strategy might be use to communicate functional brand associations to indicate superior performance. Yet, the researchers also propose that sponsors who are perceived as sincere in their sponsorship and motivated by philanthropy will achieve superior responses to their efforts. Then, consumers who perceive a sport sponsorship as an activity regards strictly to commercial objectives are less likely to develop positive responses toward the sponsor 58 .
As a matter of fact, sport sponsorship is characterized by increasing the frequency of the exposure of the message or brand of the sponsor 6 . However, the exposure degree also needs to be controlled and monitored. After all, when it is considered excessive, the repetition of the stimulus can change the attitude in a negative way 25 . Therefore, being brands affected by a mix of elements, the applied strategy around how the brand will be communicated to the consumer/ market is crucial. As an example, daLakas and Levin 51 concluded that sponsoring one property (e.g.: a football team) usually evokes positive attitudes from fans of that property, but may also elicit negative attitudes by fans of other properties. In this sense, sponsorship activities may lead also to various risks that in turn might decrease the effectiveness of this marketing communication vehicle. Scandals or underperformance within a team or athlete that is sponsored by a brand may also negatively affect the sponsor's brand image 66 .
Indeed, the introduction of negative information about a corporate sponsor negatively affects consumer's attitudes towards the sponsor and the consumption intention 13 . For instance, kiM and choi 61 verified a correlation between the judgment of the credibility of the sponsoring company and a subsequent effect on attitude toward that company brand. Taking into consideration this relation, new information or scenarios that negatively affect the image of the brand (its credibility in this case) cause damages on the attitude towards the organization. Under these circumstances, this study aims to assess in what extent a sport sponsorship strategy can change/minimize a negative attitude towards a brand or turn it into a positive one and the effects of it on the PI, considering also the differences in terms of sport involvement.
The experimental design is the main structure of this study, using a one-group pre-test/posttest variation to compare different effects of the independent variable. This research focuses on the population of Brazil, particularly, Brazilian citizens. The chosen sampling process was the non-probability sampling 67,68 , with characteristics of convenience 69 . This project was approved by the Coventry University Ethical Approval department in 2016, reference number: P42595. The minimum number of 216 samples was achieved based on the calculations and values described in Field 70 and Cohen 71 .

Measures
The measurements were based on previous studies that had considered the validity, internal reliability and accuracy of their items and models. The Semantic Differential scale (5 points) was chosen in order to measure attitude. In fact, the literature presents different numbers and meanings of items that would represent the construct 3,40,44,61,[72][73][74][75][76] . Notably, the work of sPeed and ThoMPson 3 , which directly deals with sport contexts, is highlighted in the light of it gathers valued items constantly employed by the other studies. For this reason, the construct attitude is measured using the items provided by these authors. Likewise, the purchase intention measurement was elaborated also following sPeed and ThoMPson 3 , keeping a logical coherence.
About sport involvement, according to aLexandris, Tsaousi and jaMes 57 , the construct could be observed by hours watching sports on the TV and by the fact that whether the participant practices any sportive activity. In addition, these points are aligned with other studies, like caPeLLa 46 . In this sense, involvement was measured based on these 2 pillars used by aLexandris, Tsaousi and jaMes 57 . Nevertheless, considering the paper of shank and beasLey 47 , sport involvement can be treated as a dichotomy variable (based on median, with low or high levels only). That study, also has provided parameters related to time watching sports on Tv and practicing sports activities, for instance, highly involved fans spend on average 7.5 hours per week watching sport content on TV (low = <3.2) and 6.5 hours/week exercising or playing some sport (low = <5.1). Therefore, this paper follows the dichotomy suggested by the authors.

Instrument (Questionnaire)
günTher 77 argues that making questions about what people do, think or their experiences around the past, present or future is one of the ways to study the human behavior. Thus, a questionnaire was structured based on a repeated measure design to attitude and purchase intention, in a scenario-based experimental survey. Certainly, this choice followed a thought that perceptions are influenced by the context in which they are found 17 . About that, the independent variable is the sport sponsorship in the form of sets of information (scenarios) described to the respondent. As the effect to be assessed, the dependent variable is the attitude towards the sponsoring brand and the purchase intention regards that brand's products.
The instrument was elaborated in 7 different sections. In the first one, it was described to the participants the objective of the study, with an Method agreement term to be accepted. The next section presented a fictitious brand of home appliances and TV, being positively contextualized with information around the 4 P's of the marketing mix 78 . Surely, this step aims to avoid any bias around the brand at the first moment, such as possible confused past experiences or conditioned responses 3 . In the sequence, the following section provided negative information about the brand and company, which configured the Scenario 1 (S1) of the study. In a short, middle and longterm perspectives, the brand is associated with negative allegations of corruption, mass layoffs and pollution, consequently, a negative attitude towards the fake brand was intended. After that, the attitude towards the brand was measured in order to identify which participants have developed negative attitudes.
The next section brought the first independent variable, being the sports sponsorship implanted through a new set of information, Scenario 2 (S2). The context refers only to football, where the brand is associated with sponsorship activities in short, middle and long terms. Particularly, the scenario correlated partnerships of the brand with clubs and national teams, which would have had a more commercial perception. The level of attitude and purchase intention was then evaluated through a 5 point Likert scale (pattern kept for the others scenarios). The following part, Scenario 3 (S3), described a new context, in which the brand was linked with sponsorships to football (in short, middle and long terms) one more time, but now supporting only projects to the youth, creating a more community perspective (less commercial). At the end, attitude and consumption intention were assessed.
After that, another set of information was presented to the respondent, Scenario 4 (S4). However, at this time, the brand is associated with a volleyball sponsorship (including athletes), in an Olympic and Paralympic framework. Similarly, the new context presented data in those time perspectives shown in the previous scenarios. At the end of the section, the participants were asked to express their attitudes and consumption intention regards the brand. Moreover, the final section asked few complementary questions about the respondent's sport involvement and socioeconomic indicators. Particularly, it is worth mentioning that, whereas the educational question is guided by a model offered by the Unesco 79 , the level of earning is based on an average figure provided by the Brazilian Institute of Geography and Statistics 80 .
In order to spread the study, the questionnaire was created in an online platform being submitted to a pre-test with some students and civil servants in Brazil (12 people). As pointed out by the Management Study Guide 81 , this action aimed to identify and correct possible misunderstandings and structural problems. On average, the instrument took 7 minutes to be fulfilled. A key point, the translation to the Brazilian Portuguese occurred through a back-translate method, adapted from biscaia et al. 40 . In resume, the questionnaire in English and a correspondent translation in Portuguese were assessed by a Brazilian specialist in consumer behaviour, with lasting and significant qualifications in English. After that, the translated instrument in Portuguese was back-translated to the original language by an English teacher and assessed by another expert in consumer behaviour studies.

Procedure
The questionnaires were collected from July 5 th to December 21 th , 2016, being promoted in all the 5 regions of Brazil. Basically, the participants were reached through social medias, professional networking groups (e.g.: Hospitals), corporative emails (e.g.: federal institutions) and mobile messages virally spread. The questionnaires were spread by a link with a brief message inviting people to take part in the research. In fact, targets were groups and networks, chosen by convenience, with national or regional relevance with different characteristics and backgrounds, such as age and level of education.

Data analysis
The data set was cleaned and analysed through the software IBM Statistics SPSS 22. Besides that, the sample was submitted to an inclusion criteria, that is: rightly answered questionnaires (misunderstandings or non sense answers were not considered) and participants who negatively assessed the brand in the first scenario of the instrument, that is, those ones who have marked 1 or 2 in the scale for attitude (point 3 was considered a neutral point in that scale and, above it, positive evaluations). Because of the inclusion criteria, purchase intention was only verified after Scenario 2.

Exploratory Factorial Analysis (principal component)
In reason of previous studies had a variety of ways and structures to assess attitude and purchase intention, and in order to confirm the factorial structure of the elaborated questionnaire 82 , an exploratory factorial analysis was run. All the 4 items for attitude were assessed considering their respective scenarios, therefore, evaluating whether those items were, in fact, representing one factor (attitude) per scenario. As a result, all items had high levels of correlation (R) between the elements of the same scenario. Additionally, the achieved Pearson's P coefficients were higher than the recommended 0,5 70 , being significant at p<0,05. In the same way, the determinants of the R-matrix showed values higher than 0,00001, which characterizes absence of multicollinearity in all scenarios 70 . The scenarios were also analyzed by their results in a KMO test. The four contexts reached values higher than 0.8, significant at p<0,05 70 . In the sequence, the Eigenvalues-Kaiser and the Jolliffe's criterion (retaining factors with Eigenvalues greater than 0.7) indicated that it was possible to extract one factor per scenario, when analysing the scenarios separately.
With regard to the purchase intention affirmatives, the same logic and tests were used. The results showed elements with high correlations (r>0.7 at p<0,05) and no apparent multicollinearity, having 1 factor extracted per scenario (Jolliffe's criterion) and KMO higher than 0.76 70 .

Reliability
About the reliability of the scales used, all the items for attitude and affirmatives for PI were evaluated. In short, the Cronbach's α indicated good/ acceptable values, above α>0.9 in all scenarios for all constructs 70,72 . Additionallly, the Alphas if item deleted were all beyond 0.8.

Sample
552 respondents participated in the study. However, only 54% of the total (302) respected the inclusion criteria. The results demonstrate a sample concentrated in 2 gender categories: male (53%) and female (46.4%). About the different ages of the respondents, the sample was concentrated in a zone between 26 to 36 years (45%), followed by people from 37 to 47 (22.8%) and 48 to 58 (17.2%) years old. Moreover, the questionnaire asked about the level of education of the participant, in which, the majority of participants have achieved a post-graduation level (55%), having as second position people with Bachelor degrees (25.8%). The research deduced the consumption power of the respondents observing their position related to the average earnings in Brazil. As a result, 77.8% of the respondents claimed receiving salaries (per capita) above or too much above Brazilian average. The last socioeconomic variable was the region where the respondent came from. Two percent of them were from abroad, 14.2% from the South, 3.3% from the North, Southeast represented 44.4%, Northeast more 7.3% and 28.8% of the participants were from the Midwest of Brazil. Indeed, 73.2% of the sample was concentrated on 2 areas of the country (Midwest and Southeast), which represents 8 states of the Union,

Results
including the three most important Brazilian cities: São Paulo, Rio de Janeiro and the federal capital, Brasília.

Descriptive results
From now on, the items and affirmatives used are described by codes as shown in the table below. Specifically, the five points in the elaborated scales for attitude were also grouped in accordance with the meaning of each item, that is, the first two points on the left side of the scale (1 and 2) represent the meaning on the same side (e.g.: Bad). The third point in all the scales measuring attitude (3) is considered a neutral representation of the participant's opinion. In this way, the last two points in the scales (4 and 5) represent the meaning on the right side of the item (e.g.: Good).
Below, the TABLE 2 presents and compares the figures for each item of the attitude measure related to the scenarios investigated, considering the valid sample as one unique group (not separating by involvement with sport), where each column represents one item in one scenario. As expected, in the first scenario (negative set of information) the majority of the participants evaluated the brand in the lowest point possible (1) considering all the provided items. The same table also presents the results regarding purchase intention and the created scenarios, in this case, contexts two, three capable of minimizing a negative evaluation. A point of attention, the third scenario seems to be more decisive than the second and fourth ones with regard to the observed level of improvement. Thus, metrics such as median and mode had clear improvements in a number of items and affirmatives scores after the gradual introduction of the independent variable. and four. In short, the figures offer evidence that both attitude towards the brand and correlated intention to consume its products had their scores positively improved by the sequential group of information provided through the questionnaire. Those contexts were able to enhance the general mean of every item and affirmative, or in a different prism, they were

Q1
The sport sponsorships made me more likely to use the sponsor's products. The sport sponsorships made me more likely to consider this company's products in a future purchase. 4 -Partially agree Q3 I am more likely to buy from the sponsor as a result of the sport sponsorships.

-Completely agree
Source: The author.

Specific results -Attitude
When the participants were asked to express their feelings and thoughts regards the brand, the first possibility referred to a 5 point scale guided by the opposite meanings bad and good (A1). In summary, taking into account the sample as just one group (with both sport involvement groups together), around 22.2% of the respondents who selected the points 1 or 2 in the first scenario pointed higher scores (above 2) in the second one. Certainly, this tendency was observed in the other scenarios, in reason of 57.6% of those who marked 1 or 2 in the context 1 selected a higher point in the third scenario as well. In the same way, more than 61.9% of the participants from the context 1 put their opinion above the point 2 in the last scenario. With relation to A2, A3 and A4, the scores followed the general trend of A1. About A3, although 33.8% of the sample from the second scenario selected points above 2 in the third, only 3% of the respondents with 1 or 2 in the third context moved beyond it in the fourth one, with the same occurring in A4. Based on the individual results, it is also possible to observe the rising of the scores around the points 4 and 5 of the scales (positive connotations). Notably, there were a salient and a slight improvement of these scores in the third and fourth context, respectively. Below, a graph demonstrates these achievements.

Specific results -PI
The first measure of PI (Q1) asked the level of agreement of the participant considering whether the sport sponsorship(s) quoted in the scenario made he/she more likely to use the sponsor's products. Based on the results assuming the sample as just one group as well, it was clear a progressive tendency in the "agree level" (points 4 and 5 of the scale) between the contexts, from an initial 21.5% (S2) to 32.8% (S3) and, then, 38.1% (S4). In Q2 and Q3, the obtained results show the same behavior of Q1, including a decreasing tendency around the scores 1 and 2 of the scale ("disagree level"). In general, the scenario 4 brought better positive scores ("agree level") than the other contexts. On the other hand, when compared the percentages of the lowest point of rejection (point 1) against the first one of agreement (point 4), the third scenario provided the highest positive change, around 10%.

Specific results -Involvement with Sports
Firstly, participants were asked about their three favorite sports in order of importance. A list of 23 different sports was achieved, with some items representing categories of sports/activities. In absolute numbers, football was the first favorite sport for the majority (40.4%), followed by volleyball (second favourite, 19.5%). As the third option, most of the participants claimed that they did not have a third favorite sport (22.2%). Even considering that even more than 90% of the sample mentions a favorite sport, the proper level of sport involvement was observed based on two pillars: time watching sport on Tv and practicing any sport per week. Taking into consideration the median, the group was divided in low (below the median) or high (equal or above) involvement. In terms of watching sports, the medium was a point between 2 or 3 hours per week, whereas 3 or 4 hours was the medium related to time practicing sports.
Comparing with shank and beasLey 47 , there was an alignment around the results, for instance, considering the low involvement figures (less than 3.2 hours per week watching on Tv and 5.1 practicing sports). Observing the two pillars simultaneously, assuming that high involvement is only achieved by points equal or above the medium in both pillars, the sample presented 86 cases characterized as high involved with sports (approximately 30%). involvement groups together (low and high).
Nevertheless, observing the figures of the high involvement sample it is possible to affirm that the scenario 4 was not able whatsoever to provide statistically a significant positive (or less negative) change comparing with the previous one (z<1,96 at p>0,05). Checking the mean of the scores of the construct (A1+A2+A3+A4) per scenario in the high involvement group, it is clear a gradual improvement (from S1 =1.29, S2 = 1.89, S3 = 2.92 to S4 = 3.05), but the last context did not have a relevant statistical result for that group. Similarly, the low involvement group also did not have statistical significance in some elements of attitude in the scenario 4. Then, even having better attitudes scores regards A3 and A4 (means from S1 = 1.28, S2 = 1.80, S3 = 2.75 to S4 = 2.85), the test was not able to claim a significant positive effect of the sponsorship variable on that group considering the last context. The figure below illustrates the gradual evolution of the means of attitude as an unique construct per scenario, separating it into two groups (high and low sport involvement).

Statistical tests and analysis
Firstly, the scores of the Kurtosis and Skewness, Kolmogorov-Smirnova and Shapiro-Wilk tests did not support the use of parametric tests. Consequently, non-parametric tests were used in this study. At this moment, it is essential to emphasize that a parametric test would be more powerful than the non-parametric only if the parametric assumptions would have been met 70,72 , which is not the case in this dataset.

Wilcoxon Signed-Rank test
Therefore, to compare two related conditions, a Wilcoxon Signed-Rank test (WSRt), appropriate to repeated measures 70 , was run in pairs considering all the attitude items and scenarios. Notably, the results of the test support the previous notion obtained in the descriptive statistic, that is, in general, there was a significant positive difference (z>1,96 at p<0,05, Monte Carlo Sig.) in the levels of the attitude items, taking into consideration the scenarios elaborated and both sport Discussion TABLE 3 -WSRt results for attitudes items per scenario, separated by sport involvement.

A4 S4
x A4 S3  In another perspective, the findings of the test brought a pattern related to the overall tendency of the effect of the scenarios on the attitude items. Considering a general group (with both groups of involvement), the scenarios two and four were able to improve the level of the construct, but most of the respondents (mode) in these contexts kept equal scores comparing with the previous context. Essentially, the scenario three was the unique that not only improved the attitude towards the brand (less 1 and 2 points) but also was the frame where the majority of the subjects have marked more positively their scores (from the point 3 to 5 of the scale). When analysing this pattern per sport involvement group, in both cases the same results were observed.

Both
With regard to PI, the WSRt also reveals significant variations in the levels of agreement of the affirmatives when the scenarios were compared. In a word, when the sample was analysed as one group, the values of Q1, Q2 and Q3 had higher positive variances in the scenario 3. Separating it by the level of sport involvement, the findings were basically the same.
About the effect size regards the WSRt, calculations were done following the formula provided by FieLd 70 . Point of attention, the formula used has considered different number of subjects per group of sport involvement (total sample/both = 302, low = 216 and high = 86), which meant different dividers. To sum up, a r value of 0.1 means a small effect of the intervention, r=0.3 a medium one and 0.5 represents large effect 71 . The  TABLE 4 below brings the results, on average per construct, considering the different contexts and sport involvement groups.

Correlation -Attitude and PI.
With the purpose of to confirm a correlation between attitude and PI, a Spearman' correlation (proper for non-parametric dataset) was tested. FieLd 70 put emphasis on the fact that the correlation coefficient is commonly used to measure the effect size as well. The figures achieved report a significant correlation at <0.01, in a bivariate correlation, related to all items and affirmatives. In the same way, all the relations demonstrated positive correlations. Moreover, analyzing the results through general means per scenario, the spearman's r coefficient in the scenario 2 reached an approximated mean of .39 (TABLE  5), having a clear difference when considered the level of sport involvement. For the scenarios 3 and 4, the values were around .55 and .61, respectively. Point of attention, only in the scenario four there was not a relevant difference in the involvement groups, including an unexpected slight higher effect size in the low involvement group. In this sense, based on cohen 71 , in general the second scenario might have its coefficient representing medium effects (correlation), whereas the others are interpreted as with large effects. Taking into account the symbolic and affective content of marketing initiative such as these sponsorships described in the contexts 84,85 , the results corroborate with the statements of kiM, ko and jaMes 39 , sPeed and ThoMPson 3 and aLexandris, Tsaousi and jaMes 57 , which defend that sponsorships when perceived more philanthropic than commercial likely conquest positive attitude and enhance a willingness to consume the sponsor's product. Based on kiM and choi 61 , the highest scores of the scenario 3 can also be explained by the importance of actions socially demanded, in accordance with concepts like CSR (Corporate Social Responsibility), which could influence both cognitive and affective components of attitude, having then significant positive effects regardless levels of sport involvement 42 .
In contrast, the Olympic frame presented only a small effect in enhancing the attitude level in all involvement groups. In statistical terms, the effect of the context four was not significant when separating the involvement clusters. Certainly, this fact was bigger in the high involvement group, possibly, in reason of the inverted U-effect quoted by PhaM 59 , the recent over-commercialization of the Olympic/ Paralympic environment (excessive exposure 25 ) and the proximity of the data collection with the Rio 2016 Games (submerged in corruption allegations and delays problems in Brazil). Worth mentioning, the low influence of S4 would be explained also by the fact that, during the data collection period, the publicity around Rio 2016 did not effectively affected the Brazilian general public in terms of affective persuasion, an essential element to leverage attitude 35,36,83 .
To resume, all the sponsorship scenarios were able to show some positive effects on the levels of attitude in all groups of sport involvement, however, the context four could not provide figures statistically significant in two of them. The general group (not separated by levels of involvement) initially (S1) showed a mean score of attitude towards the brand around 1.25, achieving 2.9 in the last scenario, with median and mode of 3, where the differences between the contexts were statistically significant. Considering the standard deviations, points 4 in that scale of attitude could be seen in S4, with similar movements in terms of PI. These figures permit the claim that the independent variables were capable of minimize a negative evaluation of the brand, turn it into a "neutral" attitude in the end.

Discussion -Key Findings
Firstly, it is worthy of mention the characteristics of the sample in terms of age and educational level, whose majority might then be related to a relevant cognitive capacity. At the same time, in reason of the high number of participants who were in somehow involved with sports (around 90%), it may also be assumed a certain emotional relation with it, low or high. In fact, being football and volleyball the first two sports more quoted as favorites, the choices previously made around the sports into the scenarios were reinforced. In this way, the questionnaire was appealing to emotional connections as well, which could have had an impact in terms of persuasion 83 and fan identification 51 . Consequently, both affective and cognitive components of attitude seem to be demanded by the scenarios in the instrument.
In general, observing the cumulative effect of the independent variables, the findings provide evidence that the sport sponsorship is capable of change (minimize) a negative attitude towards a brand. The descriptive results, WSRt and the Spearman correlations supported the idea that the sport sponsorship can have this positive effect, having a similar positive trend in people with high or low sport involvement. Clearly, the scenario 2 (sponsorship to football with a commercial connotation) improved the marks of attitude, obtaining an approximated large effect size of the phenomenon (WSRt) for all levels of involvement. Nevertheless, the scenario 2 (first moment of assessment of the construct) had the lowest scores, which raises questions about the sponsorships only to football clubs and national teams. Worth mentioning, the image awareness offered by this kind of sponsor property, alone, does not seem to be enough to an effective attitude change, where corporate sponsors should go beyond it and develop more effective marketing strategies geared to enhance consumers' positive perception of a sponsors' image 44 .
However, the magnitude of the effect from the scenario 3 is highlighted, insofar as the context got a considerable greater effect size (WSRt) and almost 1 point higher in terms of mean for attitude comparing with S2 for both involvement groups. In addition, the sport sponsorship when presented as supporting youth communities showed a significant potential to change a negative attitude towards a sponsor. For instance, the attitude mean jumped from around 1.2 (S1) to 2.8 (S3), being these figures in somehow similar regarding the 2 different When observing the data through levels of sport involvement, the group considered with high involvement presented better results, with the mean score of attitude coming from 1.29 to 3.05 in S4 reaching then a absolute neutral attitude status, having differences statistically significant and large effect size only until S3 (mean of 2.92), but median and mode of attitude from 1 to 3 in the scale elaborated. On the other hand, the low involvement cluster began from 1.28 to close to 3 (2.85) in S4. In the same way, the differences found were statistically significant only until S3 (mean of 2.75 and a cumulative large effect size). In a word, it would be understood that the group got a "neutral" attitude status using approximated mean numbers and considering also the general median and mode of attitude which came from 1 to 3 in S4.
In the light of sport sponsorship being able to change negative attitudes, some theories might help to interpret the findings. First, when facing the potential positive information coming from the sport sponsorships, the participants initially at the scenario 2 had their mental harmony between the elements of attitude affected (a set of negative thoughts/feelings facing a supposed positive set of information). Based on the Balance Theory 22 , Dissonance Cognitive Theory 33 and Affective-Cognitive Consistency Theory 34 , in this kind of situation, the respondent tend to change one attitude element (unfavourable thought or feeling) to restore the harmony, in this case, turning part of the negative thought or feeling into a neutral or positive one as an effect of the sportive frame.
Regards PI, the results presented a high correlation with attitude. Besides that, Spearman correlations and the WSRt revealed significant effect sizes of the experiment around attitude on the affirmatives, regardless sport involvement. Due to this fact, even with that negative starting point, it might be claimed that the sport sponsorship, affecting the attitude, also had a positive impact on PI. Notably, the observed evolution trends on the attitudes measures had the same general movements in terms of PI. For sure, the Spearman's coefficient showed a significant correlation of the constructs, which gradually got higher along the contexts. Consequently, once the sport sponsorship changed the negative attitude towards the brand it was implicit a simultaneous positive effect on the PI in all sport involvement groups. These findings corroborate with previous studies indicating that sport sponsorships could affect PI 13,39,52,61 .
In front of negative scenarios, lasting investments in different platforms of sport sponsorships seem like an effective strategy. As a whole, the results achieved through the research offer indicators that the sport sponsorship, when used in a macro and lasting strategy, is able to change (or minimize) a negative attitude towards a sponsor, affecting positively the PI, with similar results in people with high or low levels of sport involvement, but a slight better performance of the high sport involvement sample. In other words, similar to the findings of zardini FiLho 5 about consumer behaviour in Brazil, sponsors using sports as competitive differentials can improve the attitude towards them or, possibly, create a barrier against negative scenarios. Nonetheless, the data presented here brought worrisome indicators (not statistical significances in changing attitude) when looking at an "Olympic effect" in high or low involvement group of people analysed separately.
In short, the findings of this research are important to marketing and sport managers, especially Brazilians, insofar as they reinforce sport sponsorship as a strategic tool which influences attitude and PI. Mainly, considering the negative starting point, the results provide significant evidence that the sport can support organizations to face crisis situations, as a prevention and/or recovery strategy 86 , even regarding people with no high interests around sports. Likewise, considering the necessity to maximize financial investments, the results can indicate resources (like sports properties to be sponsored) that are more worthy to invest, in a Resource-based view 60 . In particular, the research provided more arguments to those interested in the Brazilian sponsorship market, which can be used during negotiation/selling times.
As a conclusion, all the sponsorship scenarios were able to show some positive effects on the levels of attitude in all groups of sport involvement; however, the context four could not provide figures statistically significant in two of them. Observing the general group (not separated by levels of involvement) and its figures, it is possible to claim that the independent variables were capable of minimize a negative evaluation of the brand, turn it into a "neutral" attitude in the end. Similarly, when observing the group considered with high sport involvement, it had better results, reaching an absolute neutral attitude status. On the other hand, the low involvement cluster would be understood as achieving a "neutral" attitude status using approximated mean numbers and considering also the general median and mode of attitude.
The study also provided evidence that the sport sponsorship can change negative attitudes towards a sponsor, regardless the level of sport involvement. Consequently, the findings are important to marketing and sport managers insofar as they reinforce sport sponsorship as a strategic tool which influences attitude and PI. The research also reports a positive implication of the sport sponsorship on the purchase intention, presenting significant correlation between attitude and PI in all sportive scenarios and sport involvement groups.
In reason of the negative starting point, the results also provided final evidences that the sport can support organizations to face crisis situations in Brazil, as a prevention and/or recovery strategy 86 , even regarding people with no high interests around sports, but with a slight better performance of the high sport involvement sample. Besides that, the statistical tests demonstrated a general positive and relevant impact of the scenarios on the attitude items. Comparing the contexts, when related to supporting projects for the "community" (scenario 3), the sponsorship to football was able of provide the most significant positive effect in all involvement groups analysed. In fact, S3 showed even better results than the following one, the sponsorship to an Olympic/Paralympic sport, in terms of improving the attitude scores, being the results not statistical significant in the last context in both high and low sport involvement groups. In this way, football seems to, in fact, play an important cultural and entertainment role to Brazilians.
Based on the abovementioned findings, it would be recommended to focus marketing strategies on sport sponsorships with clear influence on the affective element of attitude and PI, as seems to be the case of S2 and S3, creating a platform for affective persuasion. Moreover, it would be interesting actions which could improve the level of sport involvement and the awareness around the sponsoring brand, such as: special prices of products and sport tickets, the development and implementation of thematic website, CSR activities, regular fan meetings and autograph sessions with sportive celebrities 42 .

Limitations and future suggestions
One limitation of this research refers to the lack of development about the level of involvement around the products related to the fake brand, specially, in terms of purchase decisions. Another restriction about the results obtained, the sample profile (in terms of income and level of education) was not aligned with characteristic of the vast majority of the Brazilian people, a country with a huge issue of social equality. For future studies, it would be necessary to analysed the impact of the independent variables in a gender perspective. In addition, future studies better segmenting the profile of the sample in terms of income and education, with a bigger number of participants, may achieve different results.