Keywords

1 Introduction

Celebrities have been influencing popular culture and consumption patterns all over the world for many years. A watch, a dress, makeup, etc. used by a celebrity may become popular overnight. Due to their unique associations and the need to be managed professionally, Thomson (2006) states that celebrities may be considered as brands themselves. According to Euromonitor (2019) it is estimated that in the United States, 25% of all advertisements use celebrities. Additionally, celebrities “offer unique brand extension opportunities” (Kowalczyk & Royne, 2013: 211). Thus, it is important that celebrity brands should be the focus of academic studies.

Creating a brand is a costly process that requires an immense amount of money, time, and effort. Therefore, once such an investment is made and the brand is established in the market, it is common for firms to try and exploit this asset by extending the brand into other categories (Hem et al., 2003). According to a worldwide study by Nielsen (2015), about six people in 10 prefer to buy new products introduced by brands they are familiar to. Furthermore, as brand extensions reduce the risk of failure (Aaker, 1990; Hem et al., 2003; Thamaraiselvan & Raja, 2008) it is of no surprise that up to 95% of all new products entering the market are some type of extensions of established brands (Lye et al., 2001). As a result, for the past few decades, firms have been using brand extensions as a successful marketing strategy to obtain high profits from their established brands (Thorbjornsen, 2005).

While there are numerous academic studies on extensions pertaining to inanimate brands, a very limited number is on celebrity brand extensions. Additionally, existing literature focuses on celebrities majorly as endorsers but not as brands themselves, let alone examine celebrity brand extensions, in spite of the fact that there are many examples of celebrity brand extensions in practice (e.g., Rihanna’s beauty product line under her last name Fenty, Kim Kardashian’s fragrances, and Serena Williams’ clothing line). Indeed, celebrities are an important part of the market economy (Parmentier, 2010; Kowalczyk & Royne, 2013) who also act as businesses themselves selling all kinds of products and services (Preece, 2015). Attaining power and interest through the movies, TV shows, sports, music, and social media presence, they gain influence in popular culture and society (Schimmelpfennig & Hunt, 2020). These human brands interact with their external stakeholders such as critics and general public and the internal stakeholders such as professional agents to create their own brand identities and those products that the human brands create are strongly effective on their brand identities (Pluntz & Pras, 2020). However, there is a gap in literature on consumers’ intentions to purchase products that are celebrity brand extensions or what consumers’ attitudes are toward the celebrities’ brands that are extended to new product categories (Kennedy et al., 2021).

This study tries to fill the gap in celebrity brand literature by examining the feedback effects of celebrity brand extensions on attitudes toward the celebrity. Additionally, the effects of different levels of perceived authenticity of the celebrity and varying congruency levels of the extension with the celebrity on post-extension attitudes toward the celebrity, attitudes towards the extension, and purchase intention are examined.

The paper begins with framing the theoretical background for the key concepts of celebrity brand and brand extension in literature. Following the theoretical background, the relations between the variables are discussed in the hypotheses development section and the conceptual research model is presented. Subsequently, the selection of stimuli (celebrities and hypothetical extensions), questionnaire design, measures used in the study, and the sampling procedure are explained in the Methodology section. The paper ends with the results of the data analyses, discussion of the findings, managerial implications, conclusion, limitations, and future research recommendations.

2 Theoretical Background

As companies utilize other brands, persons, places, or events to link their brands to primary and secondary sources of brand equity (Keller, 1993; Uggla, 2006), celebrities have been known to be effective in advertising goods and services as endorsers (van Norel et al., 2014). Although companies have been benefiting from celebrities in promoting their products/services since the 1890s when actress Sarah Bernhardt appeared on posters for the French rice powder brand La Diaphane (Fleck et al., 2012: 651), treating celebrities as brands themselves is quite new. Thomson’s (2006: 105) study is one of the first in suggesting that humans, and thus celebrities, may also be considered as brands. Subsequently, celebrities have started managing their names as brands and utilizing brand extension strategies to capitalize on their established brand names (e.g., David Beckham’s bodywear line and Jamie Oliver’s cook books) (Moulard et al., 2015).

There is a difference between a celebrity endorser and a celebrity brand extension. When the celebrity is an endorser, he/she gets paid for allowing the use of his/her image and associations in conjunction with a product whereas when the celebrity extends his/her (brand) name, he/she has “more control of his/her relationship with the product” and “incorporates” his/her name with the product directly (Kowalczyk & Royne, 2013: 212).

Celebrities have been considered mainly as endorsers of brands but not as brands themselves. Review of past literature shows that celebrity endorsement relies mainly on four different theories: associative-network memory theory (Krishnan, 1996; Till, 1998; Brandt et al., 2011; Spry et al., 2011; Dwivedi et al., 2015); source credibility model theory (Hovland & Weiss, 1951; Sternthal et al., 1978; Ohanian, 1990; Till & Busler, 1998; Goldsmith et al., 2000; Erdogan et al., 2001; Lafferty et al., 2002; Fleck et al., 2012); meaning transfer theory (McCracken, 1989; de Mooij, 2005; Fleck et al., 2012); and congruency/match-up theory (Kahle & Homer, 1985; Kamins 1990; Misra & Beatty, 1990; Erdogan, 1999; Till & Busler, 2000). However, when contemplating celebrities as human brands and celebrity brand extensions, in literature two theories are taken as basis, which are briefly explained below.

2.1 Meaning Transfer Theory

Meaning Transfer Model which was developed by McCracken in 1989 posits that celebrities develop a persona depending on their personalities and lifestyles and are assigned meanings by the culturally constituted society (Amos et al., 2008: 210). In the celebrity endorsement process initially a “meaning” or “association” transfer (de Mooij, 2005) from the celebrity to the product, and then from the product to the target audience occurs. The effectiveness of the celebrity depends on the associations he/she brings into the endorsement process. Correspondingly, one of the basic assumptions in brand extension studies is that meanings and associations are transferred from the parent brand to the extension (Aaker & Keller, 1990; Martin et al., 2005).

Prior research on brand extensions examine how this transfer affects attitudes toward the brand and/or extensions and demonstrate its effects on brand image, brand equity, and attitudes (Aaker & Keller, 1990; Boush & Loken, 1991; Martínez & Pina, 2003; Martínez and de Chernatony, 2004; Arslan & Altuna, 2010).

2.2 Congruency: Match-Up Hypothesis Theory

The Congruency/Match-Up Hypothesis claims that when there is high congruency between the celebrity and the endorsed product, the advertising message will be more effective (Kahle & Homer, 1985; Kamins, 1990; Misra & Beatty, 1990; Erdogan, 1999; Till & Busler, 2000). Earlier studies examined congruency between the celebrity endorser and the endorsed brand mostly in the context of physical attractiveness (Till & Busler, 2000). Subsequent research (e.g., Misra & Beatty, 1990; Kamins & Gupta, 1994), however, stressed the importance of using an overall level of congruency due to the fact that consumers assessed the congruency between a celebrity and a product based not only on a few factors but on intertwined multiple factors.

In the same vein, the congruency/match-up hypothesis also forms a basis for brand extension literature where congruency refers to the degree of match and/or consistency between the parent brand and the brand extension (Aaker & Keller, 1990).

2.3 Brand Extensions

The existing image of the parent brand is a critical issue in the acceptance of the extension because it encompasses all perceptions/associations a consumer holds about a brand in his/her memory (Keller, 1993). Additionally, empirical studies demonstrate that one of the most essential issues contributing to the success of the extension is the amount of congruency between the extension and the parent brand (Aaker & Keller, 1990; Park et al., 1991; Thamaraiselvan & Raja, 2008). Brand extensions, even if there is a high congruency between the parent brand and the extension, may negatively influence perceptions of the parent brand (Aaker, 2004; Martínez & Pina, 2003; Martínez & de Chernatony, 2004; Arslan & Altuna, 2010).

As mentioned before, although celebrities play key roles in media and the entertainment business and commonly extend their name into new categories (Moulard et al., 2015), barely few academic studies are on celebrity brand extensions. Hence, this study aims to contribute to existing literature by focusing on the neglected issue of celebrities as brands themselves and celebrity brand extensions.

2.4 Development of Hypotheses

2.4.1 Attitude Toward the Celebrity

Past research indicate that extensions have diluting effects on the parent brand (e.g., Loken & Roedder-John, 1993; Aaker, 2004; Martínez & Pina, 2003; Martínez & de Chernatony, 2004; Arslan & Altuna, 2010) no matter whether the extension has high congruency with the parent brand or not (Ries & Trout, 1986). Extrapolating this finding to celebrity brands, it can be argued that celebrity brand extensions will have a negative effect on post-extension attitudes toward the celebrity as well.

  • H1: Celebrity brand extensions will have a deteriorating effect on attitudes toward the celebrity.

2.4.2 Authenticity

Turner (2013) states that for a celebrity brand to be successful he/she needs to be authentic. For this reason, authenticity has been selected as one of the independent variables of this study. It should be noted that in this study the concept is undertaken not as the authenticity of the self (“felt perception”), but how others perceive the authenticity of another (“perceived authenticity”) as described by Guillen et al. (2016: 3).

Napoli et al. (2014: 1094) define brand authenticity as “a subjective evaluation of genuineness ascribed to a brand by consumers” and present it as a significant predictor of purchase intention. Similarly, celebrity authenticity is defined as the perception that a celebrity behaves according to “his/her true self” (Ilicic & Webster, 2016) and is a significant predictor of purchase intention. Equally, Raphael and Lam (2019) state that authenticity is very important in consumer behavior as it affects consumers’ perceptions of a brand (human/product) and impacts purchase decisions. Kowalczyk and Pounders (2016) present empirical evidence on the positive relationship of authenticity with emotional attachment, positive word of mouth, and purchase likelihood. Likewise, in their study on relational authenticity, Ilicic and Webster (2014) found that authenticity is a significant predictor of positive brand attitudes and purchase intentions. Therefore, it is expected that the higher the perceived celebrity authenticity, the more positive consumers’ attitudes toward the celebrity, the extensions, and purchase intention. Accordingly, the following hypotheses are developed:

  • H2: The effect of celebrity brand extension on attitudes toward the celebrity will be more favorable when the authenticity of the celebrity is high as compared to moderate or low levels.

  • H3: Consumers’ attitude toward the brand extensions will be more favorable when the authenticity of the celebrity is high as compared to moderate and low levels.

  • H4: Consumers’ purchase intentions toward the extensions will be more favorable when the authenticity of the celebrity is high as compared to moderate and low levels.

  • H5: Perceived authenticity of the celebrity will have a positive significant effect on post-extension attitudes toward the celebrity.

  • H6: Perceived authenticity of the celebrity will have a positive significant effect on purchase intention of the extension.

2.4.3 Perceived Congruency of Extension with Celebrity

Congruency is defined as “the fit between a brand and any other entity, which may be another brand, an individual, a new product category, or an event” (Mishra et al., 2015: 1161). In brand extension literature, congruency denotes how appropriate the consumer believes that the brand extension is to the parent brand (Smith & Park, 1992), either denoting similarity to the product category and/or usage situation of the parent brand (e.g., Dove soap bar/shampoo) or relevance to the parent brand’s image/associations (e.g., Harley Davidson motorcycles/leather boots) (Boush & Loken, 1991; Broniarczyk & Alba, 1994). Similarly, celebrity endorsement studies indicate that as the congruency between the celebrity and the product being endorsed increases, a more positive response is achieved from consumers toward the advertisement and purchase intention (Kahle & Homer, 1985; Kamins, 1990; Misra & Beatty, 1990).

Likewise, in brand extension studies, it is evidenced that higher levels of congruency of the extension with the parent brand have elevated effects on evaluations of the extension and/or the parent brand (Aaker & Keller, 1990; Boush & Loken, 1991; Park et al., 1991; Loken & Roedder-John, 1993; Broniarczyk & Alba 1994; Gürhan-Canli & Maheswaran, 1998; Maoz & Tybout, 2002). Thus the following hypotheses are proposed:

  • H7: The effect of celebrity brand extensions on attitudes toward the celebrity will be more favorable the higher the perceived congruency between the extension and the celebrity.

  • H8: Consumers’ attitudes toward the brand extensions will be more favorable the higher the perceived congruency between the extension and the celebrity.

  • H9: Consumers’ purchase intentions to extensions will be more favorable the higher the perceived congruency between the extension and the celebrity.

  • H10: Perceived congruency of the extension with the celebrity will have a positive significant effect on post-extension attitudes toward the celebrity.

  • H11: Perceived congruency of the extension with the celebrity will have a positive significant effect on purchase intention of the extension.

3 Methodology

The research was performed in two stages. In Stage 1 “Pretests,” real celebrities but hypothetical extensions were used. Three pretests were conducted for stimuli selection. The first pretest involved the generation of celebrities familiar to the subjects. The generation of celebrity names was restricted to actors/actresses of local television serials. The rationale for this is the increasing popularity of local TV series industry in Turkey in the past decade.

3.1 Selection of Celebrities

The first pretest involved 25 graduate business students of Marmara (n = 14) and Istanbul (n = 11) Universities where three open-ended questions were asked (i.e., “Which television serial actors/actresses do you find highly (moderately/low) authentic?”). A total of 55 celebrities according to varying authenticity levels (23 = High; 20 = Moderate; 12 = Low) were generated.

A second pretest was conducted for rating the generated celebrities in terms of their authenticity levels where a total of 97 undergraduate business students of Marmara (n = 51) and Istanbul (n = 46) Universities participated. Aslı Enver (high authenticity); Merve Boluğur (moderate authenticity); and Hande Erçel (low authenticity) were the celebrities who received the highest ratings in each level of authenticity (high, moderate, low) and thus were selected as suitable celebrities.

3.2 Selection of Hypothetical Extensions

The final pretest was carried with 18 graduate business students to generate and select the hypothetical extensions to be used in the main study. Participants were instructed to generate suitable hypothetical extensions with different congruency levels for each celebrity selected. Following the collection of the results, a class discussion was conducted where each hypothetical extension generated was evaluated in terms of its congruency level (High/Low) with the celebrity. The hypothetical extensions that showed relevance to all three celebrities were: skin care products (high congruency) and laptop computers (low congruency) and hence were selected for the main study.

In Stage 2, for the main study, data were collected from undergraduate business students of Marmara University and Istanbul University using a 3 (authenticity: High/Moderate/Low) x 2 (congruency: High/Low) factorial experimental design (see Fig. 1).

Fig. 1
figure 1

Conceptual Model (Experimental Design). Source: Developed by authors

Full size image

3.3 Questionnaire Design and Measures

Six different structured questionnaires, each of which was based on one of six different scenarios (authenticity level: Aslı EnverHigh; Merve BoluğurModerate; Hande ErçelLow and congruency level: skin care productHC; laptop computerLC), were used to collect the data. Hence, a between-subjects experimental design was conducted where each participant answered questions regarding one celebrity (high/moderate/low authenticity) and one hypothetical extension (high/low congruency).

In the first part of the questionnaire, the respondents were asked to evaluate the celebrity for familiarity, authenticity, and the respondents’ (initial) attitudes toward the celebrity. In the second part of the questionnaire the scenario was presented (i.e., Imagine that Aslı Enver has created a new skin care product line and has introduced it to the market. Answer the questions below keeping in mind Aslı Enver and these skin care products). Following the scenario, the questions on congruency, attitudes toward the brand extension, purchase intention, and (final) attitudes of respondents toward the celebrity were asked. Lastly, questions regarding the demographic characteristics of the respondents took place.

For the study, existing reliable and valid scales were used. All of the constructs in the study were measured using 5-point Likert-type scales (1 = Definitely Disagree; 5 = Definitely Agree). Familiarity was measured with a 4-item scale: three items from Felix and Borges (2014); one item from Bergkvist et al. (2016). Authenticity was measured using the 3-item scale of Moulard et al. (2015). To measure the respondents’ attitudes toward the celebrity before and after introducing the hypothetical extension, the same attitude scale was used (6-item scale: 5 items from MacKenzie and Lutz (1989); one item from Fleck et al. (2012)). Perceived congruency was measured using 5-items: 4 items from Choi and Rifon (2012); one item from Dwivedi et al. (2015). The respondents’ attitudes toward the extension were measured using the 5-item scale of MacKenzie and Lutz (1989) which was adopted and modified for the brand extension context. Purchase intention for the hypothetical extensions was measured by adopting the 3-item scale of Baker and Churchill (1977). As the celebrities were all female and the hypothetically extended high congruency product was skin care products, the purchase intention items were worded as “I would think of buying the skin care products of ... for myself or for others (friends, sisters, girlfriend, etc.).”

3.4 Sampling

The sample consisted of 438 sophomore and senior undergraduate business students of Marmara and Istanbul Universities where convenience sampling was used. The use of student samples is usually a controversial issue as it may hinder the generalizability of the findings in extension studies (Peterson, 2001). Nevertheless, student samples in studies using experimental designs are preferred by many scholars (e.g., Ahluwalia & Gürhan-Canli, 2000; Chen & Liu, 2004; van Riel & Ouwersloot, 2005) as they allow for the control of extraneous variables which is an essential concern in experimental designs (Buil et al., 2009). Furthermore, when student samples are used, homogeneity is achieved concerning age, educational level, and areas of interest which ensures internal validity (Enis et al., 1972; Ferber, 1977; Lau & Phau, 2007).

Each student was randomly exposed to one of six different scenarios. The majority of the students were between 21 and 22 years of age (54.8%), women (55.9%), senior students (54.3%) with an income level of 501–3000 TL (55%). Additionally, the familiarity score for Aslı Enver was (x̄ = 3.53, SD = 0.96), for Merve Boluğur (x̄ = 3.24, SD = 0.92), and for Hande Erçel (x̄ = 2.90, SD = 1.01) indicating that the respondents were familiar with the celebrities used in the study (Table 1).

Table 1 Number of subjects in each experimental condition
Full size table

4 Results

The scales were checked for reliability, validity, and dimensionality. For each construct used in the study, exploratory factor analysis (EFA) using principal components analysis and varimax rotation were conducted where all of the constructs were found to be unidimensional and all item loadings were highly above the cut-off point (λ ≥ 0.50) except for one item of the familiarity scale which was excluded from further analysis. To test for reliability and validity, confirmatory factor analysis (CFA) was conducted for each construct independently. Standardized factor loadings of each item were greater than 0.50 and significant, indicating convergent validity of each construct (Fornell and Larcker, 1981). For the three-item scales, Cronbach’s alpha (α) values were between 0.770 and 0.946, composite reliability (CR) scores between 0.775 and 0.946 and average variance extracted (AVE) scores were between 0.537 and 0.854. All α indices were ≥ 0.60, CR ≥ 0.70, and AVE ≥ 0.50 showing that all scales showed sufficient reliability and convergent validity (Nunnaly, 1978; Hair et al., 2014).

For the initial attitudes toward the celebrity, congruency, attitudes toward the extension, and final attitudes toward the celebrity constructs, the α, CR, and AVE were all above the cutoff levels (α: 0.933–0.960; CR: 0.932–0.960; AVE: 0.664–0.825). Additionally, for the constructs all standardized factor loadings were above 0.50, indicating high reliability and convergent validity. Goodness of fit indices for all constructs used in the study (χ2/df = 0.0832–3601; CFI = 1000–0.993; GFI = 1000–0.977; RMSEA = 0.000–0.077) were all at acceptable levels (Hair et al., 2014).

4.1 Manipulation Checks

Manipulation checks were conducted for authenticity (ANOVA) and congruency (t-tests) experimental conditions (Table 2).

Table 2 Manipulation test results
Full size table

Table 2 shows that for the three celebrities, significant differences exist regarding authenticity levels (F = 64,903, p < 0.001) verifying the first manipulation (Aslı EnverHigh; Merve BoluğurModerate; Hande ErçelLow). The results of the t-tests indicate that there are significant differences between skin care products and laptop computers concerning congruency levels for the whole sample (t = 13.609, p < 0.001, skin care productsHC: x̄ = 3.40, SD = 1.02; laptop computersLC: x̄ = 2.12, SD = 0.96) and for each celebrity (Aslı EnverHigh: t = 8.126, p < 0.001; Merve BoluğurModerate: t = 9.500, p < 0.001; Hande ErçelLow: t = 6.870, p < 0.001) indicating that both of the manipulations were effective.

4.2 Tests of Hypotheses

As seen in Table 3, the celebrity’s introduction of an extension results in a statistically significant decline in the attitudes of the respondents toward the celebrity, regardless of the authenticity level of the celebrity nor the congruency level of the extension (t = 7.21, p < 0.001). However, when the extension shows high congruency with the celebrity, the difference between the initial and final attitudes of the respondents toward the celebrity is not significant despite the slight decline (t = 1.14, p > 0.05); whereas the difference is substantial when congruency is low (t = 8.78, p < 0.001) indicating that high congruency between the celebrity and the extension does not affect the attitudes of the consumers toward the celebrity but incongruency causes a negative effect.

Table 3 Variations in attitude toward celebrity (Initial, Final, Variation)
Full size table

There is a significant difference between the high and low congruency conditions (t = 6.14, p < 0.001). Considering the authenticity levels, for all three celebrities, there is a significant decline in the attitude of the respondents toward the celebrity once an extension is introduced, regardless of the congruency level (F = 13.21, p < 0.001). Hence H1 finds support. It should be noted that the decline is harsher for the High (3.74 → 3.19) authenticity condition as compared to the Moderate (3.07 → 2.91) and Low (2.38 → 2.27) authenticity conditions (F = 13.21, p < 0.001) (see Fig. 2).

Fig. 2
figure 2

Variations in attitudes towards the celebrity when an extension is introduced (Pre-/Post-extension). Source: Own work/based on the data analysis

Full size image

In addition to the authenticity treatment when the congruency treatment is taken into account, similar results are achieved. Specifically, when the extension has high congruency with the celebrity (see Table 3), although there is a decline in post-extension attitude toward the celebrity, the amount of decline is less (AtC variation = 0.41–0.03) as compared to the low congruency conditions (AtC variation = 0.68–0.27). For the high congruency condition, the difference in attitude toward the celebrity is insignificant when the authenticity of the celebrity is Low (LHC: AtC variation = −0.03, p = 0.49). Moreover, as the authenticity level of the celebrity increases, the decline in post-extension attitude toward the celebrity is more severe for both of the congruency conditions (LLC: AtC variation = 0.27, MLC: AtC variation = 0.54, HLC: AtC variation = 0.68; LHC: AtC variation = −0.03, MHC: AtC variation = −0.19, HHC: AtC variation = 0.41). Hence H1, H2, and H7 are supported.

According to the varying levels of perceived authenticity of celebrities and congruency of the extensions with the celebrities, the differences in attitudes toward the extension and purchase intention were tested using a series of ANOVA and t-tests (Table 4).

Table 4 Effects of authenticity and congruency on attitudes towards the extension and purchase intention
Full size table

As seen in Table 4, taking into consideration the levels of congruency of the extension with the celebrity, significant differences exist in the attitudes toward the extension (t = 7.08, p < 0.001) and purchase intention (t = 6.06, p < 0.001) where the more congruent the extension, the higher the attitude toward the extension and purchase intention.

When only the authenticity experimental condition is considered, significant differences exist in attitudes toward the extension (F = 19.50, p < 0.001) and purchase intention (F = 10.07, p < 0.001) where the higher the level of authenticity of the celebrity, the higher the attitudes toward the extension and purchase intention. When the authenticity of the celebrity increases and the congruency of the extension with the celebrity is high, the attitude toward the extension and purchase intention are also higher as compared to lower conditions (see Table 4).

As seen in Table 4, for all three authenticity conditions, significant differences were achieved for attitude toward the extension and purchase intention taking into consideration high/low congruency levels where the higher the authenticity and congruency levels, the higher the attitude toward the extension and purchase intention. Moreover, the results show that in all three situations the means are higher for attitude toward the extension as compared to purchase intention. Thus H3, H4, H8, and H9 are supported.

To test H5, H6, H10, and H11 multiple regression analyses were conducted (see Tables 5, 6, and 7). The results of the regression analyses (Table 5) show that for the whole sample, authenticity and congruency have significant effects on post-extension attitudes toward the celebrity where congruency (β = 0.490) has a higher effect compared to authenticity (β = 0.418). On the other hand, congruency (β = 0.600) has a dominant effect on purchase intention as compared to authenticity (β = 0.132) where both have significant effects. Thus H5, H6, H10, and H11 are supported.

Table 5 Regression Analysis Results (Whole Sample)
Full size table
Table 6 Effects of authenticity and congruency on post-extension attitudes toward the celebrity (Experimental conditions)
Full size table
Table 7 Effects of authenticity and congruency on purchase intention (Experimental conditions)
Full size table

The regression analyses were repeated for the varying experimental conditions (Tables 6 and 7).

Both authenticity and congruency have significant effects on the post-extension attitudes toward the celebrity where for all situations the effect of congruency is higher compared to authenticity except for the situation where congruency is low and authenticity is high. Thus, H5 and H10 are supported for the varying experimental conditions as well.

The dominant effect of congruency on purchase intention indicates that when consumers are faced with a new celebrity brand extension in the market, their intention to purchase depends mainly on whether the extended product is congruent with the celebrity or not; and not on the perceived authenticity of the celebrity. Thus, H6 finds no support but H11 does for the various experimental conditions.

5 Discussion

Celebrities “are easily recognized and attract attention” all over the world and therefore have positive effects on the promotion and sales of brands (Choi & Rifon, 2012: 647). Studies on celebrities as human brands and celebrity brand extensions are extremely limited. This study makes a contribution to extant literature by considering a celebrity as a human brand and examining celebrity brand extensions regarding the effects of authenticity and congruency on evaluations of the celebrity/brand extension and purchase intention.

In view of the findings of this study, several issues deserve attention. Firstly, there is a significant deterioration in attitudes toward the celebrity when a celebrity extends his/her name into a new product category. However, this deterioration is much more severe for the celebrity who is perceived as having high authenticity. This may be due to the belief that when a celebrity is successful, the consumer expects that he/she should concentrate on what he/she does best (acting in this study) and not direct his/her attention to other areas. This result is quite similar to the findings of Kowalczyk and Royne (2013) where it was found that the higher the attachment of consumers to the celebrity, the lower their attitudes toward the extension. This finding is also compatible with the results of prior brand extension studies where it was found that the dilution of image is greater for the more prestigious parent brand as compared to the less prestigious parent brand (e.g., Martínez & de Chernatony, 2004; Arslan & Altuna, 2010).

Secondly, this study provides evidence that the higher the congruency between the celebrity and the brand extension, the better the evaluations of consumers on attitudes toward the celebrity, attitudes toward the extension, and purchase intention. This result is parallel to past findings on brand extension and also celebrity endorsement studies where the higher the congruency between the parent brand and the extension (the endorser and the endorsed brand), the more positive the attitudes toward the extension (the endorsed product/advertisement) (e.g., MacKenzie & Lutz, 1989; Martínez & de Chernatony, 2004; Arslan & Altuna, 2010) and the greater the purchase intention (e.g., Kahle & Homer, 1985, Misra & Beatty, 1990; Kamins, 1990). As expected, these findings validate the use of congruency/match-up theory for celebrity brand extensions.

Thirdly, to the knowledge of the authors, this study is the first to test the effects of varying perceived authenticity levels of celebrities on attitudes toward the celebrity/brand extension and purchase intention. Specifically, the higher the perceived authenticity of the celebrity, the harsher the deterioration of attitudes toward the celebrity and the extension and the lower the level of purchase intention. This finding is interesting and needs further research attention.

A noteworthy finding is that, consistent with past research, perceived celebrity authenticity is a predictor of purchase intention (Ilicic & Webster, 2014, 2016; Kowalczyk & Pounders, 2016) only when the experimental conditions are not taken into account where this effect becomes insignificant when the sample is split according to the experimental conditions. Furthermore, results show that when the consumer cannot find any congruency between the celebrity and the brand extension, his/her perception of the authenticity of the celebrity is what positively affects his/her attitude toward the celebrity. Further examination concerning these relationships will be thought provoking.

One final interesting finding of this study is that, consumers’ attitudes toward the celebrity and the brand extension are much higher than their purchase intention for the extended product. This suggests that even if consumers have high attitudes toward celebrity brand extensions, this does not mean that their sale is warranted. This result may be a productive venue for further research.

6 Managerial Implications

Nowadays interest in celebrities is rising exponentially. In using celebrities as part of a company’s marketing/branding strategies one of the most critical decisions is the selection of a suitable celebrity. The same concern also exists for celebrities who manage their name just like a company. Celebrity brand extensions have also become a major industry all over the world. Hence there is a need for a better understanding of human brands, in particular celebrity brands, and the effects of celebrity brand extensions on attitudes toward the celebrity, the brand extension, and purchase intention. The findings of this study may shed light on the critical factors that should be tailored deliberately not only for the extension’s success but also to protect and/or enhance the attitudes toward the celebrity. The findings of this study indicate that for celebrities who are perceived as having high authenticity it is a better idea not to introduce any brand extensions as it causes a decline in the attitudes toward the celebrity. Despite all the risks, if marketing managers and/or the celebrity insist on extending the name of the celebrity into a new category, it is advised that it should be made sure that the extension has high congruency with the celebrities’ persona. Marketing managers are advised to find creative ways to convey messages that the celebrity is highly authentic. Additionally, choosing product categories that are highly congruent with the celebrity brand is highly crucial.

7 Conclusion

Although, there is vast research on celebrity endorsements, the studies focusing on celebrities as human brands and examining the products and services launched by the celebrities as extensions of their self-brands are highly limited. This paper aims to fill this gap by examining primarily the feedback effects of celebrity brand extensions on attitudes towards the celebrity. Secondly, the effects of different levels of perceived authenticity of the celebrity and the congruency levels between the extension and the celebrity on post-extension attitudes toward both the celebrity and the extension and purchase intentions are examined. Results reveal a significant deterioration in attitudes toward the celebrity when a celebrity extends his/her name into a new product category particularly for the celebrities with higher authenticity. Moreover, the evaluations of consumers on attitudes toward both the celebrity and the extension and purchase intention improves in situations when the congruency between the celebrity and the brand extension is higher. Similarly, when the authenticity level of the celebrity increases, the deterioration of attitudes toward the celebrity and the extension is harsher where purchase intention is also decreased. According to the results of the study, perceived celebrity authenticity and congruency between the celebrity and the extension are predictors of purchase intention.

8 Limitations and Future Research Directions

The main limitation of the study is the use of a student sample. However, it should be noted that data were collected from two of the most crowded universities in Istanbul. Conducting the study on students has resulted in a sample made up of millenials. Young people may have more positive feelings toward celebrities as compared to older people. Therefore, replicating the study for other ages or generational cohorts and on different cultures would provide further evidence. Another limitation is the use of a non-probable sampling method, convenience sampling, which restrains the generalizability of the results.

To increase external validity real celebrities but hypothetical extensions were presented to the subjects. Even though the extensions were generated in the pretests, in future studies the use of real extensions may allow more representative results. Celebrities used in the experimental design, who were actresses appearing in Turkish television serials. This is an important limitation as it decreases the generability of the results. Also, as only female celebrities were included in the study, the attitude toward the celebrity may differ according to the gender of the respondents. Lastly, the inclusion of perceived risk and involvement as additional variables in the research model would be worthwhile to enhance the findings and may provide deeper insights.