How to choose the endorser: An experimental analysis on the effects of fit and notoriety

The present study is focused on the endorser topic following two different paths: firstly, proposing an extension of the theoretical match-up model, enlarge it through two other potential types of consistency: the typicality fit and the imagery fit. Secondly, the present study aims verifies the applicability of the same framework to the emerging situation with a brand linked to a not well-known endorser (internal as the founder or external as a web influencer). An experimental 3*2 (fit typology*high/low notoriety) between subject analysis was conducted in the food service domain. It showed some interesting considerations. From a theoretical point of view, the first relevant finding is that endorsement might be assimilated to a co-branding strategy, confirming the match-up model as an effective theoretical framework in this domain as well, with significant differences among the three fit typologies investigated. The typicality fit reveals to be the less effective in increasing attitude and other behavioural effects on consumers in spite of the large adoption of this kind of fit by companies. Instead, the imagery fit, seems to be the most impactful in terms of positive word of mouth activation and viral communication activities, at the same level at the categorical one. Moreover, the categorical fit induces the wider range of positive effect on the dependent variables (attitudes, willingness to pay and willingness to buy). Another interesting contribution is that the presence of an appropriate fit (in particular the categorical one) is able to compensate the absence of endorser notoriety and, on the average, the usage of a very popular endorser from the same domain of * Associate Professor, University of Modena and Reggio Emilia, Department of Communication and Economics; e-mail: vgabrielli@unimore.it. ** Associate Professor, University of Modena and Reggio Emilia, Department of Communication and Economics; e-mail: ibaghi@unimore.it. The paper is the result of the joint effort of the authors. Copyright © FrancoAngeli This work is released under Creative Commons Attribution Non-Commercial NoDerivatives License. For terms and conditions of usage please see: http://creativecommons.org Veronica Gabrielli, Ilaria Baghi 58 the brand is not necessary more effective in comparison with a not well-known endorser form the same domain. This result is the peak of the present research from a managerial point of view, as it leads to consider the opportunity to support the emerging practices by which companies turn to not well-known people (disclosing the founder, or presenting some workers, or adopting a common consumer as an influencer). The endorser not well-known, but presented with an adequate story-telling might be the best choice: less onerous and more effective than a big unrelated celebrity. Keyword: endorser, fit, notoriety, food, consumer First submission: 31/12/2018, accepted: 20/10/2019


Introduction
The use of an endorser as a person effective in promoting a product or a brand is a long-dated practice that is becoming more and more diffused within fast growing economies (Zhao, Fan, 2004;Sun, 2010). The match between a brand and a person is significantly evolved during the time: the endorser is no more a person that simply promote the product, but she/he is a crucial element of a brand identity system. The most important contribution of the endorser is to anthropomorphize the brand in terms of personality traits and physical representation (Amos, Holmes, Stutton, 2008). Then, a brand becomes more and more distinguishable, well-known and less comparable to competitors. That is, nowadays the main goal adopting a testimonial is not only to attract the consumers' attention but rather to give to a brand some desirable and unique traits. This means that companies are very far from the initial practices; in the past, businesses usually selected endorsers for their popularity and their desirability as famous people. Several international ranks were arranged in order to suggest the most popular endorsers (i.e. Q ranking) and to evaluate each potential endorser. Some companies have realized that in this way endorsers usually obfuscate the brand as long as famous people were simultaneously testimonial for different brands. For this reason, some companies have started to adopt a different practice, using not well-known endorser (i.e. the founder, or an employee, a web influencer or common people- Biswas, 2008). Using a not well-known person as an endorser implies to build around this person a story, some characteristics effective in legitimating him/her on consumers' eyes, in order to finally exert a positive effect on the brand.
Companies must select not simply a testimonial but rather the best testimonial, regardless she/he is famous or not. Marketing literature agrees with the tendency to anthropomorphise the brand. Iasevoli (2003) even talks about a brand "genetic code" highlighting that the endorser is not a short-term communication activity, but rather a long-term investment.
In spite of these consideration, scant literature has been devoted to analyse the right way to select the appropriate endorser. In lights of the emerging phenomenon of not well-known endorsers (i.e. internal endorsers and influencer) the brand-endorser consistency becomes a crucial and more and more urgent issue (Bergkvist, Zhou, 2016;Wang et al., 2015).
The present research is the first attempt to shed light on this topic, following two different paths: firstly, proposing an extension of the theoretical match-up model, based on the assumption that there might be coherence between the brand and the endorser field. We propose to deepen the investigation of the boundaries of match-up model, enlarging the concept of coherence to two other potential types of consistency: the typicality fit and the imagery fit. Secondly, the present study aims to contribute to the comprehension of the phenomenon verifying its robustness in terms of the applicability of the same framework to the emerging situation with a brand linked to a not well-known endorser (internal as the founder or external as a web influencer). Finally, a second study is addressed to systematize the psychological multi-step process behind the effectiveness of different kind of brand-endorser fit.

Study 1: theoretical background and research hypothesis
Marketing literature on endorser has been rapidly developed since the last 30 years. The meta-analysis done by Amos, Holmes and Strutton (2008), based on 32 different studies, is a very useful systematization of the state-of-the art on this topic. The authors highlight theoretical foundations focused on the endorser credibility. It seems to be a very robust and consistent mainstream framework (Ohanian, 1991;De Sarbo, Harshman, 1985;Mc Guire, 1985;Ohanian, 1991;Pornpitakpan, 2003). Following this path, the endorser is able to affect attitudes and judgements on behalf of the brand by three antecedents: endorser attractiveness, credibility and experience. Each of these antecedents, in turn, is composed by different elements. Unfortunately, this framework assumes that the endorser is so popular to evoke reliability, specific competences and transparency on behalf of the brand. That means that the internal validity of this model is limited to those case with a well-known endorser. Moreover, this framework analyses the endorser on itself, indeed it considers the effectiveness of the endorser as an individual characteristic, regardless the linkage between a specific brand and the endorser. For these reason, it is not suitable for the present research. The second stream of research detected by Amos et al. (2008) meta-analysis is the match-up model (Kamis, 1990;Erdogan, 1999;Erdogan et al., 2001), based on two different theories: the one by Baker and Churchill (1977) and the other one by Freidman and Friedman (1979). The meta-analysis conducted by Knoll and Matthes suggested that the match-up model can be explained by the social adaptation theory (Kahle and Homer, 1985;Kamins, 1990) or by the schema-theory (Lynch and Schuler, 1994). The first one refers to the consumer proneness to consider information able to facilitate their adoption to the environment: the endorser might be such a kind of information, leveraging mental associations compatible with the consumer desired social representation. On the other hand, the schema theory considers the endorser as a set of informational nodes that might be integrated to the previous mental brand representation. The more the new information are consistent with the previous one, the more consumers would add them to her/his mental representation of the brand. The main contribution of the match-up model is to underline the opportunity to select the endorser not only for her/his own characteristics but rather for her/his consistency with the brand. Following the match-up model, companies might select not the best endorser at all, but rather the most appropriate endorser for themselves. Marketing studies based on the match-up model talk about "congruence" (Kamis, Lynch and Schuler, 1994), "fittingness" (Kanungo and Pang, 1973); "approprietness" (Solomon, Ashmore and Longo, 1992) or "consistency" (Walker, Langmeyer and Langmeyer, 1992). All of them assume that must be some degrees of fit between the brand and the endorser. The first analysis adopting the match-up model based on social adaptation theory argue that the physical attractiveness of the endorser is a key factor able to influence consumers' attitude toward the brand, intention to buy and believes on the brand for self-expressive products (as automobiles, perfumes, jewelleries). Unfortunately, these studies have produced conflicting findings as regards the product categories and the outcome variables (Till and Busler, 2000). Then, the match-up model stream of research based on schema theory has been subsequently focused on different trajectories of congruence between the brand and the endorser. Significant effects have been observed in the domain of categorical fit (Homer and Kahle, 1990;Till and Busler, 2000), but with some limitations stressed by Amos et al. (2008) meta-analysis. In particular, the authors call for a deeper analysis on other kinds of coherence in order to rule out different explanations for their results.

Copyright © FrancoAngeli
This work is released under Creative Commons Attribution -Non-Commercial -NoDerivatives License.
For terms and conditions of usage please see: http://creativecommons.org Moreover, Biswas, in 2008, argued that all of the analysis had taken into consideration only the adoption of a well-known endorser. In 2018 Knoll and Matthes proposed another meta-analysis very useful in systematize the effects of endorsers. This meta-analysis confirms the effectiveness of endorsers on different levels: cognitive level (on cognition and awareness about the brand), affective level (attitude towards the brand) and conative level (intention to behave in favour of the brand). Knoll and Matthes meta-analysis also provides a complete list of moderators: most of them are separately related to the brand, or to the endorser, or to the ad promoting the brand with the endorser. The only variable focused on the linkage between the brand and the endorser is named "match". The meta-analysis corroborates the hypothesis of a significant and positive effect of a good match between the endorser and the brand. Notwithstanding this, this very recent meta-analysis showed that no studies considered not well-known endorsers (focusing on the presence of the endorser compared to the absence of it) and they lacked the opportunity to investigate different ways to build a good match between brand and endorser. The first study of the present research attempts to fill in the two gaps here above discussed: firstly, to demonstrate a possible extension of the match-up model considering other kinds of brand-endorser consistency; secondly, to verify the importance of a brand-endorser fit in the domain of not wellknown endorsers as well.
In order to extend the vision of the match-up model we follow the direction suggested by Till and Busler (2000), based on the associative memory model (Anderson, 2983;2013) and the schema theory (Lynch and Schuler, 1994). The associative memory model is a theoretical framework very useful in order to comprehend the match-up model, as it is focused on the abstract linkages between concepts and entities due to similarities and co-presence of them. In this particular domain, the endorser and the brand are informative nodes reciprocally linked, then they weave together their mental association in consumers' mind. This kind of linkage is very similar to a co-branding activity: it is a combination between two entities. The assumption is that the endorser is functional to a brand leveraging process (Keller, 2003), thus linking a brand to a different entity, companies try to transfer positive, unique and strong mental associations from the latter to the former. In this case we observe secondary mental associations and Keller (2003) stated that the entity might be another brand, a location or a person. Aaker (1996) included the endorser in his co-branding classification as an "energizer" for the brand, able to transfer vitality, interest and to highlight the brand. The endorser is, then, an entity able to transfer new associations to the brand, as a co-branding Copyright © FrancoAngeli This work is released under Creative Commons Attribution -Non-Commercial -NoDerivatives License.
For terms and conditions of usage please see: http://creativecommons.org activity. This parallelism suggests to get inspiration from co-branding literature empirical evidences (Simonin and Ruth, 1998;Blackett and Russell, 1999) in order to hypothesize some potential expansion of the match-up model applicable to the endorser topic. From this literature review, we consider other two potential sources of congruity, beside the categorical one, between the brand and the endorser. Iasevoli (2003) proposes an effective systematization of fit typologies in co-branding activities: similarity or categorical fit, linkage or imagery fit and typicality fit.
The similarity or categorical fit (Park et al., 1996;Muroma and Saari, 1996;Wang et al., 2015) refers to the affinity among functional characteristics of products involved in the co-branding activity. In particular, Boush and Loken (1991) get back to the similarity concept presented by Aaker and Keller (1990) referring to product categories. This kind of fit is very tight to the match-up model as it assumes that the two entities share the same conceptual category. It is a product category fit (Bhat and Reddy, 1997) as in the case of a co-branding strategy utilized to diversify the activity of a brand in order to rapidly land to another product category. This kind of fit is adaptable to the brand-endorser domain as a company might select a celebrity within the same category field: for example, a sailor man to promote yacht equipment.
The second typology of fit proposed by marketing literature is the linkage or imagery fit (Bhat and Reddy, 1997). This fit is based on the attribution of meanings and values that can be shared by the two brands involved as they are very similar for some characteristics. It is not necessary that the two brands belong to the same product category, but it is fundamental that they are consistent in terms of symbolic aspects as personality traits, country of origin or values. These elements might be of an extrinsic nature (as the country of origin) or of a symbolic and intrinsic one (as the personality traits). This kind of fit offers several interesting considerations as well. It is adaptable to the brand-endorser co-branding frame as all of the symbolic mental associations present in the two entities concept map might be reinforced by the co-branding initiative, as a sort of reciprocal contamination. Busacca and Bertoli (2012) addressed particular attention to this kind of fit. They named is as "celebrity fit". They applied this conceptual framework to a brand extension context: a clock brand that might go to the jewellery category linked to a very popular princess). They hypothesized that the imagery fit might have lightened the category fit between clocks and jewellery and the brand fit between the clock imagery and the new jewellery product line imagery. Their findings supported the idea of a positive influence of the imagery fit but their research is embedded in a different domain-brand extension). Nevertheless, it confirms the relevance of an imagery fit within a co-branding strategy and so it is a first example Copyright © FrancoAngeli This work is released under Creative Commons Attribution -Non-Commercial -NoDerivatives License.
For terms and conditions of usage please see: http://creativecommons.org of the potential contribution of this kind of fit in the field of endorser selection (regardless the diversification strategy). Therefore, it is a potential candidate to be considered in the present study. Finally, the third typology of fit is the tipicality one (Nedungadi and Hutchinson, 1985;Gurhan-Canli and Maheswaran, 1998). It is in presence of two brands that are prototypical of each of their field. In this case, we do not have two entities that are near in terms of product category, but they are very consistent in terms of positioning. For example, each brand is leader in its market. This kind of fit is used in case of linkage between products that are very distant but that can share positive associations as leadership and high performances. Applying this kind of fit within the endorsement context we can refer to those companies that select top actors, singers or show-men regardless any kind of categorical pertinence with the product/service managed by the brand.
The assumption of this kind of practice is that the product is as desirable and successful as the popular endorser (Guenzi and Borghini, 2003).
Summarizing, marketing literature confirmed the match-up model suitability in order to investigate co-branding strategies; co-branding literature provides evidences about the effectiveness of a categorical similarity between the two brands involved in a partnership and, parallel, it offers the suggestion of two other fit typologies: the typicality fit and the imagery one.
The assumption of the present research is that the partnership between a brand and an endorser is a specific kind of co-branding and, then, the abovementioned typologies of fit might be effective in this domain as well.
Notwithstanding this, it is expected a different level of efficacy by the three fit typologies. Each fit typology is based on a mental association shared by the brand and the endorser: in the categorical fit, it is the product domain; in the imagery fit, it is the imagery trait; in the typicality fit, it is the positioning of the brand and the endorser in the respective fields. These mental associations (Keller, 2003;Aaker, 1996) are informational nodes that compose, together with many others, a dense network around the central node of the brand in consumers' mind. Each mental association has a different origin (internal or external), proper characteristics (positivity, uniqueness and strength) and a different positioning in the mental network (primary associations and secondary associations). The three typologies of fit are based on shared mental associations with a different origin, characteristics and positioning. Then, the first hypothesis is that the three fits might have a different level of efficacy.
In particular, the categorical fit implies that the brand and the endorser belong to the same product category. The product category belongingness generally is a primary brand association, that is an informational node directly and strictly linked to the central node of the brand, as it is an essential Copyright © FrancoAngeli This work is released under Creative Commons Attribution -Non-Commercial -NoDerivatives License.
For terms and conditions of usage please see: http://creativecommons.org element of the product. This kind of information is really accessible to a consumer's mind, thus it is easier to remember and then very effective in influencing people evaluations and behavioral behavioral intentions (Fazio et al., 1989); a close link between two stimuli (e.g., product functionality and the job of the endorser) should make the development of an associative link easier (McSweeney and Bierley, 1984;Graeff, 1996). As regards the second typology of fit, it is based on an informational node that is not so essential as the categorical one; the imagery fit born from a mental association that is generally used by companies (and by endorser) in order to differentiate itself from competitors and it implies that this informational node is unusual respect to competitors. Marketing literature refers that, sometimes, this kind of mental associations are secondary ones, that is the origin is external to the company (for example the country of origin of a product that drives positive impressions) and it takes more time to communicate and build this kind of linkage to a brand. That's why generally this kind of informational node is more distant to the central node of the brand in comparison to the product category one. For this reason, the present study assumes that the imagery fit has a smaller magnitude of effect in comparison with the categorical fit. Finally, the third kind of fit, typicality one, is supposed to be the less effective. This kind of fit doesn't imply that brand and endorser belong to the same field and then they potentially interact from two mental starting points that are very distant each other; moreover, the positioning of the brand and of the endorser in the respective field is not necessary an objective and stable characteristic. An endorser might be the best singer for me but not for other people, for example, and it might not be the best forever (as for many athletes whose performances are not stable during the time). The perceptual field distance from the brand to the endorser and the subjectivity and volatility of the informational node constituent the typicality fit drive the expectation that this kind of fit is the less effective in influencing consumers. The different influence exerted by the three kinds of fit is expected on different level of dependent variable, at an affective level and at a conative one. As for the previous studies on co-branding activities, we hypothesized a significant effect on attitude toward the brand (Boush and Loken, 1991;Till and Busler, 2000;Gurhan-Canli and Maheswaran, 1998) on intention to buy (Tilla and Busler, 2000; Busacca and Bertoli, 2012) on the willingness to pay (Busacca and Bertoli, 2012) and on the intention to activate positive word of mouth on behalf of the brand (Boush and Loken, 1991).
H1a: The existence of a particular kind of fit between a brand and its endorser has a different positive effect on consumers' attitude toward the brand.

Copyright © FrancoAngeli
This work is released under Creative Commons Attribution -Non-Commercial -NoDerivatives License.
For terms and conditions of usage please see: http://creativecommons.org In particular the categorical fit exerts the highest impact, then the imagery fit and, at a lower level the typicality fit. H1b: The existence of a particular kind of fit between a brand and its endorser has a different positive effect on consumers' intention to buy the brand. In particular the categorical fit exerts the highest impact, then the imagery fit and, at a lower level the typicality fit.
H1c: The existence of a particular kind of fit between a brand and its endorser has a different positive effect on consumers' willingness to pay for the brand. In particular the categorical fit exerts the highest impact, then the imagery fit and, at a lower level the typicality fit.
H1d: The existence of a particular kind of fit between a brand and its endorser has a different positive effect on consumers' intention to activate positive word of mouth in behalf of the brand. In particular the categorical fit exerts the highest impact, then the imagery fit and, at a lower level the typicality fit.
The first group of hypothesis is linked to the extension of the match-up model within the endorser context, extending the investigation of the different efficacy up to three kinds of fit: the categorical, the imagery and the typicality one.
As a precise intent of the present study is to comprehend if these kinds of fit have the same efficacy using not well-known endorsers (i.e. the company founder, or factory workers, not well-known influencer o common people), it is expected that the same different magnitude of fit typology will be observed regardless the notoriety of the endorser. In fact, following the assumptions of the match-up model, it is simply the presence of an entity consistent with the brand that induces a positive perceptual effect on mental associations referred to the brand itself. The linkage between a brand and an endorser is promoted regardless the different accessibility of the central node of the two entities on consumers' mind. In other words, the elements that make compatible a brand and an endorser overlook the notoriety of the two entities. That means that even an endorser which is not so well-known might be able to exert this kind of positive leverage effect on behalf of the brand, it is sufficient that the endorser possesses a mental association in common with the brand. There's no theoretical basis to hypothesize that in the domain of well-known endorsers should be observed a different hierarchy of fit typology effectiveness. In other words, we expected the same different effect of the three typologies of fit within not well-known endorsers. It is sufficient that each endorser, although it is not already known by consumers, has got at least the mental association founding the fit with the brand.

Copyright © FrancoAngeli
This work is released under Creative Commons Attribution -Non-Commercial -NoDerivatives License.
For terms and conditions of usage please see: http://creativecommons.org Notwithstanding this, the leveraging process (Keller, 2001) on a brand might be based not only on the presence of mental associations shared with the endorser, but on other mental associations composing the dense weft on consumers' mind as well. The popular endorser is more willing to activate several mental associations, activating some positive spill-over effect on the brand, because of its fame and previous notoriety, whereas the relatively unknown endorser drives very few mental associations, as he/she has probably got a less dense and smaller mental association network. For this reason, we hypothesized that, in spite of the different fit efficacy in the domain of not well-known endorser, the presence of a popular endorser parallely exerts a direct and positive effect on consumers' reactions, as for a co-branding with a very popular brand (Hillyer and Tikoo, 1995;Aaker, 1996;Blackett and Russel, 1999;Helmig, Huber and Leeflang, 2008).
H2a: The presence of a popular endorser has a significant positive effect on consumers' attitude toward the brand, in comparison with a not-well known endorser.
H2b: The presence of a popular endorser has a significant positive effect on consumers' intention to buy the brand, in comparison with a not-well known endorser.
H2c: The presence of a popular endorser has a significant positive effect on consumers' willingness to pay for the brand, in comparison with a notwell known endorser.
H2d: The presence of a popular endorser has a significant positive effect on consumers' positive word of mouth on behalf of the brand, in comparison with a not-well known endorser.

Pre-test
In order to test our hypothesis, we selected as a stimulus the opening of a new gourmet Irish steakhouse. It was a fictitious brand. We simulated that this steakhouse chain was going to open several points of sale in Italy (all of the respondents were Italians). The experimental stimulus was a flyer representing an endorser announcing the next opening of the Irish gourmet steakhouse. We conducted a pre-test in order to select some appropriate endorsers. Our intent was to identify three well-known endorsers with the same level of notoriety: one characterized for her/his prominence on the food domain (categorical fit); another selected for her/his similarity with the brand in terms of Copyright © FrancoAngeli This work is released under Creative Commons Attribution -Non-Commercial -NoDerivatives License.
For terms and conditions of usage please see: http://creativecommons.org country of origin (an Irish women/men-imagery fit) and the last one as a topperformer in her/his domain far from food (typicality fit). We managed a brainstorming on 15 people in order to elicit 9 potential well-known endorsers, three for each typology of fit. Those popular endorsers will be later used for a pre-test in order to select the right stimuli for study 1. Participants of the brainstorming were similar to the sample composition of the main study: not only students, but adult workers and purchase responsible as well. We recruited these participants through students, within their friends, parents and neighbours. Each participant belongs to a different family group. Then, we proposed the 9 different potential endorsers emerged by the brainstorming to a sample of 20 Italian consumers (45% males, 55% females) asking to respondents to rate on a 7-point Liker scale how much they think these popular people were pertinent in terms of image-country of origin (How much do you mind this celebrity is Irish?), in terms of category (How much do you mind this celebrity is involved in food?) and in terms of typicality (How much do you mind this celebrity is the best in her/his domain?). We control for the level of notoriety and pleasantness of each celebrity as well. We check for the homogeneity in terms of popularity and pleasantness, respectively, through two different questions: How much do you know this celebrity?; How much do you like this celebrity?
Pre-test revealed three celebrities suitable for the main study: they were significantly different in terms of category, image (Irish country of origin) and typicality fit. For the categorical fit emerged the popular chef Joe Bastianich (food category= 6.35 vs Irish imagery = 1.52 vs typicality-top performing = 4.78 1 ). As regards the (Irish) imagery fit, the popular singer Bono Vox emerged (food category = 1.30 vs Irish imagery = 5.09 vs typicality-top performing = 4.70 2 ). As regards the typicality (or performance) fit, pre-test revealed the Olympic athletic champion Usain Bolt (food cateogry = 2.17 vs Irish imagery = 1.09 vs typicality-top performing = 6.26 3 ). Manipulation check, and in particular the test of contrasts, confirm that each fit is significantly related to an endorser, in comparison with the others celebrities (Joe Bastianich food category = 6.35 vs Bono Vox food category = 1.30 vs Bolt food category = 2.17; p < .05; Bono Vox imagery fit = 5.09 vs Joe Bastianich imagery fit =1.53 vs Bolt imagery fit =1.09; p < .05; Bolt typicality fit = 6.26 vs Bono Vox typicality fit = 4.70 vs Bastianich typicality fit = 4.78; p < .05). All of the celebrities were equals in terms of notoriety (Joe Bastianich M awareness = 5.83 vs Bono Vox M awareness = 6.17 vs Usain Bolt M awareness = 6.26 4 ) and pleasantness (Joe Bastianich M like = 4.13 vs Bono Vox M like = 4.65 vs Usain Bolt M like = 5.13 5 ). Then we selected three not well-known potential fictitious endorsers, consistent with the three kinds of fit: the chef Daniel Kastlevic; the Irish singer Rohan O'kelley and the Olympic athlete Mark Ukhov, able to be used for the experimental set with unknown endorsers. For these endorsers, we didn't measure the level of fit, as the informational node shared with the brand will be manipulated in the scenario description within the main study; we only check for the same level of notoriety and pleasantness in order to rule out any confounding effect due to the name sounding. As expected, the three not wellknown endorsers were equals in terms of notoriety (

Study 1: procedure
The six potential endorsers (three well-known and three not well-known) revealed by the pre-test phase were finally used in the main study in order to compose the experimental set. Only one endorser appeared in the fictitious flyer showed to each respondent (between-subject design). Each respondent was asked to imagine to receive at her/his home a flyer representing the endorser announcing the next opening of the Quinn's new gourmet Steakhouse in Italy. The ad claim was "A true chef/Irish man/champion (depending on the fit stimulated by the experimental set-respectively categorical fit/imagery fit/typicality fit) chooses the Irish steakhouse Quinn's".
The experimental design was a 3 (fit: category vs imagery vs typicality) * 2 (endorser notoriety high vs low) between subject design. We used the experimental design as we intended to precisely measure the strength of the hypothesized linkages and to give a strong internal validity to our findings (Campbell, 1957;Amos et al. 2008). After having selected the stimuli, a sample of 138 Italian consumers participated to the main study (42% males, 58% females). They were randomly assigned to one of the six experimental design (between-subject). In lights of the limitation of previous studies, using only students as respondents (Amos, 2008), we decided to conduct the analysis among actual consumers. By doing so, we aimed at avoiding the probability that the relatively demographic and psychographic homogeneity within the student sample would have facilitated the hypothesized relations (Lynch, 1999).
Respondents were randomly recruited linking the questionnaire to different blogs and Facebook pages not related to food, ranging from several domains in order to facilitate the presence of different ages and food habits. We controlled for the internal homogeneity in each set through a post-test during the data analysis.
Each respondent was asked to reply to 9 different items corresponding to the dependent variables. The only variable measured with a single item was the willingness to pay (WTP). The others were measured following the scales cited in table 2: attitude toward the brand (ATB: pleasant/unpleasant; attractive/unattractive; good/bad; beautiful/ugly), intention to buy (ITB: Are you willing to go to this restaurant in the next month?/How much do you think is probably that you try this new restaurant?), willingness to pay for a dinner (WTP: How much are you willing to pay for a dinner in this steakhouse?from €15 to €45), and willingness to activate positive word of mouth (WOM: Do you intend to recommend this restaurant to your friends/family?/ I am willing to positive talk about this restaurants to my friends/family). 18 respondents didn't ultimate the task, so we excluded them from the analysis. The final sample is composed by 120 consumers.  Finally, we insert questions referred to the manipulation check. We control for the existence of each fit and for the level of notoriety and pleasantness of the endorser. In the case of not well-known endorsers we didn't check for the degree of the existence of the fit (as respondents didn't previously know the endorser), but rather we ask if they remember the presence of the specific characteristics of the endorser (to be a chef, to be an Irish man, to be an Olympic athlete). We used the same questions as the pre-test. At the end of the questionnaire we check for potential covariates: age, gender, food habits and meat consumption habits for each respondent.
The same homogeneity has been demonstrated by data analysis on food habits (to have lunch/dinner out of home (χ2 = 13.075; df = 15; sig > .05) and on meat consumption habits (χ2 = 29.71; df = 20; sig > .05). This is an important starting point in order to rule out any other explanation for our findings.
The manipulation check related to fit perception by respondents confirmed the adequacy of the stimuli. In particular, the well-known endorsers were significantly different in terms of category, image (Irish country of origin) and typicality fit. The categorical fit is confirmed for the popular chef Joe Bastianich (food category = 4.63 vs Irish imagery = 1.53 vs typicalitytop performing = 1.25 6 ). As regards the (Irish) imagery fit, it is confirmed for the popular singer Bono Vox emerged (food category = 1.55 vs Irish imagery = 4.20 vs typicality-top performing = 1.31 7 ). As regards the typicality (or performance) fit, pre-test confirmed the adequacy of the Olympic athletic champion Usain Bolt (food category = 2.22 vs Irish imagery = 2.73 vs typicality-top performing = 5.81 8 ). Manipulation check, and in particular the test of contrasts, confirm that each fit is significantly related to an endorser, in comparison with the others celebrities (Joe Bastianich food category = 4. Attitude toward the brand: the analysis of variance (ANOVA) showed an effect of fit typology on attitude toward the brand at the limit of the significance (F (1,120 = 2,98; p = .05). Descriptive statistics, in particular contrast analysis with LSD, better showed this result: the two fits with the higher impact are the categorical and the imagery one. In fact, contrary to our expectations, the difference of the effects of these two fit on attitude toward the brand is not statistically significant (Category Fit M ATB = 4.80 vs Imagery Fit M atteitude = 4.38; p > .05), while the effect of the typicality fit on attitude toward the brand is significantly lower than the categorical one (Typicality Fit M ATB = 4.03 vs Category Fit M ATB = 4.80; p < .05). Instead, the contrast between Typicality Fit M ATB = 4.03 vs Imagery Fit M ATB = 4.38 is not significant (p > .05). H1a is partially confirmed as the categorical fit has got the great impact on attitude toward the brand, and the typicality fit is the worst one, but the imagery fit didn't show a clear difference in comparison to them. 7 T-test for coupled samples demonstrated the significant difference between the imagery fit and the other two: Vox category fit vs Vox imagery fit p < .05; Vox imagery fit vs Vox typicality fit p < .05 8 T-test for coupled samples demonstrated the significant difference between the typicality fit and the other two: Bolt category fit vs Bolt typicality fit p < .05; Bolt imagery fit vs Bolt typicality fit p < .05

Copyright © FrancoAngeli This work is released under Creative Commons Attribution -Non-Commercial -NoDerivatives License.
For terms and conditions of usage please see: http://creativecommons.org As regards the second block of hypothesis, the endorser notoriety exerts a significant and positive effect on attitude toward the brand (F (1,120) = 4.63; p < .05). In fact, the sets with well-known endorsers registered a higher level of attitude toward the brand (High notoriety M ATB = 4.68 vs Low notoriety M ATB = 4.13; p < .05). H2a is confirmed.
Intention to buy: the analysis of variance (ANOVA) din't show any main effect due to the typology of fit (F (1,120 = 2.36; p > .05) as the mean values registered by this dependent variable are not statistically different between the categorical fit and the imagery one. As for the attitude toward the brand, the only fit that had a value for the intention to buy significantly lower is the typicality one (Categorical Fit M ITB = 3.98 vs Imagery Fit M ITB = 4.05 vs Typicality Fit M ITB = 3.32; p > .05 9 ). Therefore, H1b is partially confirmed as the categorical and the imagery fit exert the same magnitude of effect on intention to buy, while the typicality fit plays the lower impact on this dependent variable.
As regards the second block of hypothesis, the endorser notoriety didn't exert a significant and positive effect on intention to buy (F (1,120) = 2.92; p > .05). In fact, the sets with well-known endorsers didn't registered a different level of intention to buy in comparison with those with not well-known endorsers (High notoriety M ITB = 4.04 vs Low notoriety M ITB = 3.52; p > .05). H2b is not confirmed.
Willingness to pay: the analysis of variance (ANOVA) showed a significant main effect of the fit typology on willingness to pay for a dinner in a Quinn's steakhouse (F (1,120 = 5.42; p < .05). H1c is confirmed. Descriptive statistics showed that, as expected, the categorical fit is the most appropriate to induce a positive effect on willingness to pay (Categorical Fit M wtp = € 24.37 vs Imagery Fit M wtp = € 21.62 vs Typicality Fit M wtp = 19.75; p < .05 10 ), consistently with the classical match-up model; followed, at a short distance, the fit evoking the country of origin; the typicality fit is the lowest. Therefore, H1c is confirmed.
As regards the endorser notoriety, it didn't exert any significant effect on willingness to pay (F (1,120) = 2.52; p > .05). Then H2c is not confirmed. Descriptive statistics confirmed that the endorser notoriety hadn't any effect 9 The analysis of contrsts with LSD showed the following significance indexes: Word of mouth: the analysis of variance (ANOVA) didn't show a significant effect of fit typology on willingness to activate a positive word of mouth on behalf of the brand (F (1,120 = 2.76; p > .05). Descriptive statistics showed a similar effect induced by the categorical and the imagery fit (Category Fit M wom= 4.16; Imagery Fit M wom = 4.12; p > .05) in comparison with the typicality fit that is statistically lower (Typicality Fit M wom = 3.35 vs Category Fit M wom = 4.16; p < .05; Typicality Fit M wom = 3.35 vs Imagery Fit M wom = 4.12; p < .05). H1d is partially supported.
As regards the second block of hypothesis, the endorser notoriety exerts a significant and positive effect on word of mouth proneness (F (1,120) = 4.20; p < .05). In fact, the sets with well-known endorsers registered a higher level of disposition to activate a positive word of mouth on behalf of the brand (High notoriety M wom = 4.20 vs Low notoriety M wom = 3.55; p < .05). H2d is confirmed Tab. 3 and 4 summarize the average values registered for each dependent variable for the different dependent variables.  Moreover, the effects of typology fit and those due to the endorser notoriety on the dependent variable are not interacting each other. That means that the different efficacy of brand-endorser fit parallel runs beside the endorser notoriety. In fact, we checked for the absence of interaction. None of the dependent variables registered such an interaction effect 11 . This is consistent with our theoretical assumptions.

Discussion study 1
Findings supported the opportunity to extend the choice of an endorser to other kind of fit, in particular the imagery one, while, consistently with our expectations, the typicality fit is the one which always exerts the lower impact on the dependent variables. However, in some cases, the categorical fit and the imagery one have the same impact. This is one reason to better investigate the phenomenon. Moreover, the analysis confirms that the effects due to fit typology are the same in case of a well-known endorser in comparison with a not well-known one.
These results confirm the opportunity to enlarge the match-up model within the brand-endorser relationship including new kind of fit and to extend the boundaries of this effect over the notoriety. The assumption of the study was that the efficacy of brand-endorser fit is due to the ability to induce 11 no interaction effect between the fit typology and the endorser popularity on attitude pay (F(1,120) = .220; p > .05); no interaction effect on the willingness to pay (F(1,120) = .350; p > .05); no interaction effect on the willingness to pay (F(1,120) = .690; p > .05); no interaction effect on the intention to activate positive word of mouth (F(1,120) = .235; p > .05).

Copyright © FrancoAngeli This work is released under Creative Commons Attribution -Non-Commercial -NoDerivatives License.
For terms and conditions of usage please see: http://creativecommons.org a perception of consistency within their relationship, based on schema similarities. After having tested the robustness of fit even in the domain of brandendorser relationship, the study proceeded with the aim to explicitly demonstrate the psychological process behind the observed effects, in order to rule out any other potential explanation. For this reason, we arranged a second study. The contribution of this second step is: to explicitly measure the different perception of consistency between the brand and the endorser induced by different typologies of fit, in order to point-out the cognitive effect of perceived similarity between bend and endorser, then to position this variable into a processual model able to show the psychological process managed by consumers facing a certain brand-endorser partnership: from the perceptual leverage, throughout the affective level (attitude towards the brand), up to the conative one (as a proxy of final behavioural effects). Moreover, the processual model is intended to better clarify some effects registered in study 1 that are not so clear, probably because they are not direct effects but, rather fully mediated by the perceived consistency. Indeed, in study 2 a multi-step mediation model is tested.

Study 2: theoretical background and research hypothesis
The second study originates from the intent to demonstrate that the different effect of fit typologies is related to the different ability to induce a perception of a certain degree of consistency between the brand and the endorser. The assumption is that consumers look for a degree of similarity between entities that are jointly presented as they need to save cognitive resources and to avoid cognitive dissonance. Within co-branding activities there are several studies demonstrating the relevance of perceived consistency between different products (Aaker and Keller, 1990) and brands (Arnett et al., 2010).
Associative learning principles (McSweeney and Bierley, 1984;Shimp et al., 1991) support the idea that a positive transfer between associated entities will be stronger when the two objects are closely linked. In lights of the assumption of the first study, we expected that the fit typology between brand and endorser differently induce a perception of coherence. In particular, the categorical fit is expected to be the strongest way to foster a perception of congruity as it is based on a functional fit (Trimble and Rifo, 2006), then on a characteristic that is objectively end deeply embedded in the brand (and in the endorser). Instead, the imagery fit is based on an image-based congruity, that is on image-based features that constitutes a lighter and less prominent Copyright © FrancoAngeli This work is released under Creative Commons Attribution -Non-Commercial -NoDerivatives License.
For terms and conditions of usage please see: http://creativecommons.org connection. Notwithstanding this, the imagery fit is a trait visibly shared by brand and endorser as it is usually made prominent in communication activities (i.e. the highlighting of the country of origin). On the contrary, the typicality fit is based on a subjective and less homogeneous trait; to be a top performer in a domain might imply very different characteristics and traits in comparison with a performer in a different domain. Summarizing, the similarity between a brand and an endorser in case of a categorical fit is based on an intrinsic and objective shared characteristic; in case of an imagery fit it is based on a secondary and objective characteristic, while in case of a typicality fit it is based on a subjective quality (not a stable characteristic. Based on this conceptual distinction, we hypothesized that the different effects of fit typologies is reflected by a multi-step psychological process: the brand-endorser fit induce a different degree of brand-endorser perceived coherence that, in turn, affect consumers' reactions.
H4: The existence of a particular kind of fit between a brand and its endorser has a different positive effect on consumers' perceived consistency. In particular the categorical fit exerts the highest impact, then the imagery fit and, at a lower level, the typicality fit.
Linking this hypothesis to the previous ones, emerges that the perceived consistency of fit between the endorser and the product is expected to mediate the impact of the typology of fit on consumers' reactions. In fact, study 2 has an incremental ambition respect to the study 1, aiming at demonstrating the psychological process managed by consumers. For this reason, the dependent variables investigated in the study 1 are hereafter positioned with a processual perspective: the typology of fit influence the perceived consistency that, in turn, impact on the affective level (attitude toward the brand) and, then the conative one, thus the intention to behavioral outcomes.
As regards the causal link from perceived consistency and attitude toward the brand, many researchers argue that consumers positively evaluate similarity and congruency between entities and it promotes positive evaluations toward brands or products linked together (Aaker and Keller, 1990;Boush and Loken, 1991;Dawar, 1996;Herr, Farquhar, and Fazio, 1996). Moreover, a wide set of contributions have showed a positive effect of congruence in terms of perceived brand image (Gwinner and Eaton, 1999) and brand credibility (Rifon et al., 2004;Becker-Olsen and Hill, 2006). Endorser/product compatibility influences consumer choice through associative learning (Shimp et al., 1991). This is consistent with information integration theory (Anderson, 1981), the schema theory of learning (Schmidt, 1975), and the balance theory (Heider, 1946(Heider, , 1958Mowen and Brown, 1981) which suggest that prior attitudes will be integrated with the new congruent information Copyright © FrancoAngeli This work is released under Creative Commons Attribution -Non-Commercial -NoDerivatives License.
For terms and conditions of usage please see: http://creativecommons.org influencing positive evaluation and response, as consumers usually adopt a consistent organization of their mental structure in order to avoid internal tension. Kamins, in 1990, stressed the consumers' willingness to adopt incremental mental process: they rely on adaptive significance, integrating previous information with consistent new ones. Boush and Loken (1991) gave a deep explanation of this phenomenon: a strong perceived coherence in a co-branding activity facilitates the consumer decisional process because the consumer can adopt a bookeeping mental account approach. The bookeeping process occurs when people add new and favourable information (referred to the co-branded product) to the previous ones available in their mind before the linkage. That is, when consumers observe an endorser that is strictly linked to a brand, in a consistent way, they are more prone to add positive information and beliefs to their previous attitude to the brand. From this premise, it derives the following hypothesis: H5: The perception of consistency between the product and the endorser mediates the effect of different kind of fit on consumers' attitude toward the product In the marketing and consumers' behavior literature, there is significant evidence that attitude is the best predictor of intention, which in turn is the best predictor of behavior (Ajzen and Fishbein, 1975). Therefore, a favorable attitude toward an act, product, or brand should be related with positive behavior, while a negative attitude should restrain an individual from pursuing such behavior (Ruvio et al., 2008). Even if some evidence suggests that the attitude-behavior link is often weak (Schlegelmilch et al., 1996), the theory of reasoned action assumes that individuals are rational and make systematic use of the information available to them (Fishbein and Ajzen, 1975); that is, the individual's attitude affects his or her behavioral intention (Eastlick and Lotz, 1999;Szymanski and Hise, 2000). The linkage between attitude and behavioral intentions is demonstrated in the domain of co-branding as well (Becker-Olsen et al., 2006;Gupta and Pirsch, 2006;Pracejus and Olsen, 2004). Descending from previous hypothesis and moving throughout a mental process step, in light of attitude-behavior linkage, we formulated our last hypothesis.
H6a: Attitude toward the brand positively mediates the effect of perceived consistency on intention to buy the brand.
H6b: Attitude toward the brand positively mediates the effect of perceived consistency on willingness to pay for the brand.

Copyright © FrancoAngeli
This work is released under Creative Commons Attribution -Non-Commercial -NoDerivatives License.
For terms and conditions of usage please see: http://creativecommons.org

78
H6c: Attitude toward the brand positively mediates the effect of perceived consistency on the intention to activate positive word of mouth in behalf of the brand. Figure 1 represents the hypothesized conceptual process.

Study 2: procedure
A sample of 213 participants (50% male; mean age 38) took part in an on-line experiment. All procedures are consistent with Study 1 unless otherwise specified. In Study 2 only the typology of fit was manipulated using the same fictitious endorsers presented in Study 1: Chef Daniel Kastlevic (category fit); Irish singer Rohan O'Kelley (imagery fit); the Olympic athlete Mark Ukhov (typology fit). In this second study, we didn't distinguish in not well-known and well-known endorsers as these characteristics is not significantly related to the different effect of fit typologies. This result of study 1 allowed us to investigate only one kind of endorser. We decided to use not well-known endorsers because we think that this domain is more interesting in a theoretical and managerial point of view. Participants were randomly assigned to one of three experimental conditions and, after reading the stimuli (the same used in Study 1), they were asked to answer to the same dependent variable scales used in Study 1. Respondents were also asked to evaluate the perceived endorser/product consistency (Congruent/Not congruent; Consistent/Inconsistent; Compatible/Not compatible; Goes together/Does not go together; Bigné-Alcaniz et al., 2012; .98). At the end of the questionnaire, they were asked to answer to some questions about their food habits and meat consumption habits.

Copyright © FrancoAngeli
This work is released under Creative Commons Attribution -Non-Commercial -NoDerivatives License.
For terms and conditions of usage please see: http://creativecommons.org
The same homogeneity has been demonstrated by data analysis on food habits (to have lunch/dinner out of home (χ2 = 12,.075; df = 16; sig > .05) and on meat consumption habits (χ2 = 32.31; df = 19; sig > .05). This is an important starting point in order to rule out any other explanation for our findings.
Perceived consistency: the analysis of variance (ANOVA) showed a main effect of fit typology on perception of consistency between the brand and its endorser (F = 1,212 = 13.44; p < .05). Descriptive statistics showed that the most positive effect on perceived consistency is caused by the categorical fit (Category Fit M consistency = 4.61) in comparison with the typicality fit (Imaginary Fit M consistency = 3.86) and the typology fit (Typicality Fit M consistency = 3.31). In conclusion, this result showed that, as expected, the categorical fit is able to exert a more positive effect on perceived consistency that the imagery fit and the typicality one. H4 is supported.
Attitude toward a brand: the analysis of variance (ANOVA) showed a significant effect of fit typology on attitude toward a brand as a dependent variable (F (1,213 = 4.14; p < .05). H1 a is confirmed in the study 2.
Intention to buy: the analysis of variance (ANOVA) showed a main effect of the typology of fit on this dependent variable (F (1,213 = 4.14; p < .05). Descriptive statistics showed that the fit with the most positive impact on purchase intentions was the categorical one (Categorical Fit M ITB = 4.18 vs Imagery Fit M ITB = 3.91 vs Typicality Fit M ITB = 3.35; p < 0,05). H1 b is supported in the study 2.
Willingness to pay: the analysis of variance (ANOVA) showed a significant main effect of the fit typology on willingness to pay for a dinner in a Quinn's steakhouse (F(1, 213) = 5.66; p < .05). Once again, the categorical fit is the most appropriate to induce a positive effect on willingness to pay (Categorical Fit M wtp = € 24.19 vs Imagery Fit M wtp = € 20.83 vs Typicality Fit M wtp = € 20.71; p< .05). H1c is supported in the study 2.
Positive word of mouth: the analysis of variance (ANOVA) showed a main effect of fit typology on willingness to activate a positive word of mouth on behalf of the brand (F (1,213) = 3.69; p < .05). Descriptive analysis showed that the categorical fit is the one that induced the higher consumers'  To test the mediation model proposed a mediation analysis using PRO-CESS model 6 (Hayes, 2012;Muller et al., 2005) with 10,000 bootstrap estimation resamples and 95% confidence intervals was conducted to verify the expectation that perception of consistency between the endorser and the product enhances attitude toward the brand (H5) that improves intention to buy, positive word of mouth and willingness to pay (H6 a, b, c). The categorical fit increases consumers' perception of consistency (H4) that causes an improvement of attitude toward the brand that positive affects all the conative outcomes (intentions to behave). This analysis was therefore conducted on a sample of 213 evaluations. The mean of all items for each construct was used in the analysis. The indirect effect of the typology of fit on intention to buy, positive word of mouth and willingness to pay was mediated by the enhancement of consistency perception that improves attitude toward the brand (indirect effect on intention to buy = .21, CI from = .11 to = .32; indirect effect on positive word of mouth = .11, CI from = .04 to = .19; indirect Copyright © FrancoAngeli This work is released under Creative Commons Attribution -Non-Commercial -NoDerivatives License. For terms and conditions of usage please see: http://creativecommons.org effect on willingness to pay = .49, CI from = .20 to = .88). All the path coefficients in the model were consistent with the hypotheses. The effect of the categorical fit on intention to buy and positive word of mouth was fully mediated by perception of consistency and attitude toward the brand while the same effect on willingness to pay was partially mediated ( Table 6). The fully mediation demonstrated in the second study contributes to explain and clarify some not so clear and net tendencies registered in the study 1, and it sheds light on the mental process behind the brand-endorser link efficacy reflecting the multi-step informational process: from a perspective level, to an affective one up to the intention to behave (conative outcomes).
In order to rule out potential alternative explanations, we estimated our model again including several covariates. We controlled for the effect of two sociodemographic characteristics (i.e., age and gender) and food habits and meat consumption. Neither food habits nor meat consumption habits had any significant effect on any of the variables included in the model. Age and gender were also not significantly associated with any of our conceptual variables.

Conclusions
Findings of the present study, that simulates a linkage between an Irish steakhouse next opening in Italy and an endorser with a certain degree of notoriety and fit with the brand, offers some interesting theoretical and managerial considerations.
From a theoretical point of view, the first relevant finding is that endorsement might be assimilated to a co-branding strategy, confirming the matchup model as an effective theoretical framework in this domain as well, strengthening its robustness and enlarging its boundaries. Following this model, the endorser might be a person consistent with the brand, not simply a very popular celebrity regardless her/his compatibility with the brand. The present research corroborates the match-up model and was intended to try to extend this model proposing two other kinds of fit: the typicality one and the imagery one. The former indicates an endorser which is a best performer in her/his domain as the brand in its market; the latter implies to select an endorser which is very similar than the brand in terms of values, personality traits or other symbolic meanings (in this case the country of origin). Then, another interesting finding of the present study is to have compared the efficacy of these kind of fit. In absolute, the typicality fit reveals to be the less effective in increasing attitude and other behavioural effects on consumers compared to the imagery fit and the categorical one. In general, the typicality fit is one of the most adopted by companies: they usually select celebrities looking for very popular and best performing endorsers (celebrities, top models, sports champions, actors), regardless any kind of similarity with the brand. Results of the present study induce to believe that is not an advisable choice, as consumers are not prone to positively react to a simply famous endorser without any coherence element with the brand in order to consider the linkage between these two entities as plausible and desirable. Without

Copyright © FrancoAngeli
This work is released under Creative Commons Attribution -Non-Commercial -NoDerivatives License.
For terms and conditions of usage please see: http://creativecommons.org this consistency, it doesn't happen any sort of transfer of positive mental associations from the endorser to the brand (leveraging process). That is, the choice of a testimonial who share with the brand some meanings and mental associations exert a positive effect on word of mouth activation, promoting viral communication spreading. It is important to highlight that, in this study, we selected an exterior consistency, manipulating the country of origin (Irish for the endorser as for the brand), which is a secondary mental association (Keller, 2001). We selected this kind of consistency as it is simple to manipulate it (in particular in presence of a not well-known endorser) and to communicate it with a flyer. In lights of our results it might be very interesting to activate an imagery fit working on deepener elements (as personality traits). Even if the imagery fit doesn't result as effective as the categorical one for all of the dependent variables, it deserves a lot of attention in particular for those companies that intend to promote their brands with viral and unconventional communication strategies. Summarizing, the imagery fit emerges as a valid candidate to complete the match-up theoretical model. However, the categorical fit appears as the most effective on attitude and willingness to pay for the service. One particularly surprising result is that the average price that respondents indicate within the experimental set with a categorical fit is 20% more than the price in presence of a typicality fit. This finding corroborates the match-up model, confirming that the categorical fit induces the wider range of positive effect on the dependent variables (attitudes, willingness to pay and intention to buy). The present study significantly contributes to the comprehension of the match-up model demonstrating that it is so effective to operates regardless the endorser popularity. That consideration further confirms the psychological mental process behind the match-up model, as the categorical fit is immediate and it is very simple to be detected and perceived. After having demonstrated the different degree of impact of fit typologies, the study 2 allowed to systematize the results and to represent the psychological process managed by consumers facing a brandendorser partnership. The second study confirm the hypothesis that the different origin, strength and positioning of mental associations shared by a brand and its endorser induce different levels of perceived consistency between these two entities. This perceived consistency are based on elements shared by brand schema and the endorser schema in consumers' mental representation and activates a bookkeeping mental process able to foster the consumer attitude toward the brand and, then, to activate the intention to behave in favour of the brand in some different ways (intention to buy, willingness to pay and word of mouth). The processual path corroborated by the Copyright © FrancoAngeli This work is released under Creative Commons Attribution -Non-Commercial -NoDerivatives License.
For terms and conditions of usage please see: http://creativecommons.org second study provides an evidence of the conceptual framework systematized by Knoll and Matthes (2018), contributing to comprehend different ways to build brand-endorser match, testing some paths only described by their meta-analysis (i.e. the linkage between affective level and conative one), and enlarging the domain of the phenomenon by the inclusion of not well-known endorsers. Indeed, it is a significant improvement in comprehending the psychological process behind the brand-endorser efficacy, demonstrating that the different efficacy of fit typologies doesn't act in a direct way, but rather its effects are described as a mediation model: from the cognitive level (the perceived consistency) through the affective level (attitude towards the brand) up to the conative level. From a managerial point of view, the selection of an endorser whose domain is compatible with the brand (in our example a chef for a next opening steakhouse chain) is advisable to sustain the entrance on a new context. The endorser legitimates the new brand on consumers' eyes. Companies which have selected such a kind of fit (for example sports equipment producers with and athlete endorser) take comfort to this result. The opportunity of this kind of choice is further strengthen by our results within the experimental set with a not well-known endorser. Our hypothesis hadn't formulated any expectation about the existence of an interactive effect between fit category and endorser notoriety, because of theoretical foundations about the possibility to foster internal consistency between two entities regardless the notoriety and the opportunity to apply the same mental leverage in the domain of not wellknown endorsers. Notwithstanding this, data analysis had the chance to investigate this aspect as well, and they showed very interesting considerations. The endorser who is not well-known, as long as it is qualified as pertinent in terms of categorical domain, is able to activate the best reactions in terms of attitude, intention to positive word of mouth and willingness to pay. This effect is possible as the categorical fit acts parallel to the notoriety. The most surprising result is that these two effects are actually parallel and independent, they do not interact each other. That means that the presence of an appropriate fit (in particular the categorical one) is able to compensate the absence of endorser notoriety and that, on the average, the adoption of a very popular endorser from the same domain of the brand is not necessary more effective in comparison with a not well-known endorser form the same domain. In fact, the experimental set with a not well-known chef showed the same results than the set with the Olympic champion Bolt, despite his popularity, and not ever the results in the set with Joe Bastianich were significantly higher than those with the not well-known chef. This result is the peak Copyright © FrancoAngeli This work is released under Creative Commons Attribution -Non-Commercial -NoDerivatives License.
For terms and conditions of usage please see: http://creativecommons.org of the present research from a managerial point of view, as it leads to consider the opportunity to support the emerging practices by which companies turn to not well-known people (disclosing the founder, or presenting some workers, or adopting a common consumer as an influencer). The categorical coherence is one of the simplest mental association to be built. It is rapidly perceivable to consumers, facing a communication activity that describe the endorser as an expert in the brand field (as a chef in our example). This is a great opportunity for companies, as it advices to not devote so much investments to obtain the right to link the brand to an endorser selected only on the basis of her/his popularity. Moreover, this kind of celebrities are usually over-prominent with several brands at the same time. The endorser not well-known, but presented with an adequate story-telling might be the best choice: less onerous and more effective than a big unrelated celebrity.
The present studies have got some limitations: first of all, we investigate the phenomenon only within one domain (food service). It might be interesting to test the robustness of the mental process within other contexts as well. Secondly, we manipulate the imagery fit with an external trait (country of origin); it might be of some interest to deepen the analysis using an internal imagery trait (i.e. personality trait). It could be difficult in case of not wellknown endorsers, but it might be implemented with a careful description of the stimuli. The investigation of the hierarchy between categorical fit and imagery one deserves to be improved, as the imagery fit has got more chance to be distinctive for a brand in comparison with the categorical one. Further research might deepener the present studies following these paths.