The influence of multiple combinations of celebrity endorsers on consumers' intentions to purchase a sports-related product

Abstract

Purpose

Endorsement research has focused primarily on determining the effectiveness of single endorsers, whereas marketing practice shows that companies usually engage multiple endorsers to promote their brands and products. As academic evidence for multiple endorsers is limited and extant findings are ambiguous, the purpose of this study is to determine the influence of different multiple endorser combinations on consumers' purchase intentions for a sports-related product endorsement and to identify whether endorser-product congruence and consumers' involvement with the endorsed product moderate endorsers' influences.

Design/methodology/approach

Two-hundred thirty-three useful responses were collected to an online experiment, and endorsers' direct and indirect influences on consumers' purchase intentions were analyzed in serial mediation analyses. Potential moderating effects of endorser-product congruence and consumers' involvement with the endorsed product were tested in moderated regression analyses.

Findings

The study's findings show that using multiple endorsers has an overall positive influence on consumers' purchase intentions, which is mediated by their attitudes toward the advertisements and the endorsed brand. Endorser-product congruence moderates an endorsement's effectiveness, whereas a moderating effect of consumers' product involvement was not supported.

Originality/value

This study adds to the extant body of endorsement research by confirming the overall effectiveness of using multiple endorsers to influence consumers' intentions to purchase the endorsed product. In addition, by showing that endorser-product congruence determines endorsers' effectiveness in a multiple endorser context, the study extends the current research perspective and provides practical implications for marketing professionals on how to combine multiple endorsers.

Citation

von Felbert, A. and Breuer, C. (2022), "The influence of multiple combinations of celebrity endorsers on consumers' intentions to purchase a sports-related product", Sport, Business and Management, Vol. 12 No. 4, pp. 440-458. https://doi.org/10.1108/SBM-03-2021-0030

Introduction

Companies frequently use product endorsements as part of their marketing strategies to promote their brands and products (Fink et al., 2014), and they engage celebrity endorsers to stand out in the highly competitive media environment (Braunstein-Minkove et al., 2011; McCormick, 2018). Research has shown that sports celebrities' endorsements are an effective marketing tactic in increasing consumers' brand perceptions and their purchase intentions (Felbert and Breuer, 2021) as they combine outstanding performance in their fields with increased media competency (Alexandris et al., 2012). However, endorsement practice and previous endorsement research have indicated that engaging an endorser always carries the risk that the endorser will misbehave, thereby negatively affecting the brand he or she is endorsing (Sato et al., 2015). For example, Knittel and Stango (2014) showed that the market value of Tiger Wood's sponsors decreased by more than 2% after the scandal because of his extramarital affairs. Therefore, engaging endorsers to promote a company's brand or product is “not necessarily a recipe for success” (Taylor, 2016, p. 167), as their selection is central for each endorsement's success in influencing advertising outcomes.

Previous studies have primarily used single-endorsement designs to determine endorsers' influences on advertising outcomes as they are less complex than multiple-endorsement designs and therefore easier to address in a research context. However, single endorsements are a rough simplification of real-life marketing practice as they disregard the reality that both endorsers and companies are usually involved in multiple endorsements (Chen et al., 2013). By simultaneously engaging multiple endorsers, companies seek to break the monotony of a single endorser and to leverage each endorsers' ability to attract the attention of the companies' target audiences (Rice et al., 2012). Nike's “Just Do It” campaign, for which the sportswear and apparel company engaged multiple sports celebrities, such as Colin Kaepernick and Shaquem Griffin (football), Serena Williams (tennis) and LeBron James (basketball), is a practical example of a multiple endorser strategy (Mejia, 2018). Rolex, the watch manufacturer, also engaged multiple endorsers, including Lindsey Vonn (alpine skiing), Mark Webber (formula 1) and Tiger Woods (golf) to communicate its brand philosophy and to influence consumers' brand perceptions.

However, only a few research studies have analyzed using multiple endorsers to determine their effectiveness in influencing advertising outcomes. Hsu and McDonald (2002), for example, argued that endorser-product congruence was an important factor in determining endorsements' effectiveness for the US “milk mustache” campaign. Moreover, Chen et al. (2013) found that, for multiple endorsements, a high level of endorser-product congruence positively affected consumers' brand perceptions, whereas Rice et al. (2012) indicated that, when consumers were highly involved with the endorsed product, endorser-product congruence moderated the influence that the number of endorsers had on consumers' attitudes toward the endorsed brand. However, Mowen and Brown (1981) questioned, early on, whether using multiple endorsers have an overall positive influence on advertising outcomes, arguing that consumers react less favorably to multiple endorsers than they due to a single endorser. In addition, Tripp et al. (1994) pointed out that consumers evaluated endorsers as more trustworthy and more of an expert in a single endorser context than they did in a multiple endorser context.

As previous research has determined the influences of multiple endorsers on consumers' attitudes toward the endorsers and the endorsed brand, the present study analyzes the direct and indirect influences of multiple combinations of endorsers on consumers' intentions to purchase an endorsed product. To derive practical implications for marketing professionals, the present study further determines the influences of endorser-product congruence and consumers' involvement with the endorsed product on their purchase intentions. Accordingly, this study seeks to address three main research questions: How do multiple endorsers affect consumers' purchase intentions? How does endorser-product congruence determine the effectiveness of using multiple endorsers? How does consumers' involvement with the endorsed product affect endorsers' influence in a multiple-endorser context?

Literature review

Celebrity endorsements

Product endorsements are contractually agreed and define the endorser's and the brand's rights and obligations, the consequences if one or the other party fails to fulfill them (e.g. the endorser misbehaves), the period of time during which the endorser is committed to promote a company's brand or product and the company's obligation to (usually financially) reimburse the endorser for the endorsement (Bergkvist and Zhou, 2016). Companies use product endorsements as part of their marketing strategy to influence consumers' brand perceptions and increase their intentions to purchase the endorsed product. Through an endorsement, consumers create connections between a conditioned stimulus and an unconditioned stimulus so that endorser traits are transferred from the endorser via the endorsement to the consumers (McCracken, 1989; Shimp et al., 1991). Previous studies have indicated that product endorsements have a positive influence on consumers' brand perceptions (Spry et al., 2011), their purchase intentions (Pradhan et al., 2014) and the overall market value of the product's company (Agnihotri and Bhattacharya, 2018).

To stand out in the highly competitive media environment, companies engage celebrity endorsers to promote their brands and products, as this type of endorser is well known to a large number of consumers and is associated with positive source characteristics (Erdogan, 1999). McCracken (1989) defined a celebrity endorser as an “individual who enjoys public recognition and who uses this recognition on behalf of a consumer good by appearing with it in an advertisement” (p. 310). Companies engage celebrity endorsers as they offer compelling product testimonials and are perceived to be trustworthy and having a high level of product expertise (Hovland et al., 1953; Ohanian, 1991). Endorsers' trustworthiness is defined as consumers' “degree of confidence in the communicator's intent to communicate the assertions he [or she] considers most valid” (Hovland et al., 1953, p. 21), and Till and Busler (2000) indicated that consumers are more likely to accept the message of a trustworthy endorser than that of a less trustworthy endorser. Moreover, Wang and Scheinbaum (2018) found that trustworthy endorsers are more effective in influencing consumers' (purchase) behavior than less trustworthy endorsers are. It was also shown that if an endorser is perceived to be an expert, his or her other source characteristics play a lesser role in determining his or her effectiveness in an endorsement context (Fink et al., 2004). Furthermore, McGuire (1985) and Ohanian (1990) pointed out that endorsers' physical attractiveness can also have a positive influence on consumers' brand perceptions and their purchase intentions. However, endorsers' visual appearance is rarely used as a selection criterion in marketing practice as it is highly subjective and controversially discussed (Fink et al., 2012; Ilicic et al., 2018).

More recent studies questioned the direct effects of endorsers' characteristics on advertising outcomes and argued that endorsers' influence on consumers' purchase intentions is mediated by their perceptions of the endorsement and the endorsed brand (Felbert and Breuer, 2021). Lee et al. (2017) showed that consumers' attitude toward an endorsement mediates endorsers' influence on consumers' brand perceptions, which is defined as “the extent to which consumers feel that [… the] firm has the knowledge or ability to fulfill its claims and whether the firm can be trusted to tell the truth” (Newell and Goldsmith, 2001, p. 235). Consumers' perceptions of the brand's expertise and trustworthiness determine their attitude toward the endorsed brand (Newell, 1993) and were shown to have a positive, direct influence on consumers' purchase intentions (Goldsmith et al., 2000).

Extant endorsement studies have indicated that different congruence relationships, for example the congruence between the endorsers and the endorsed product and the congruence between the targeted consumers and the endorsed product, determine endorsers' effectiveness in influencing advertising outcomes. According to Till and Busler (2000), endorser-product congruence influences consumers' perceptions of the endorsed brand, whereas Lee and Thorson (2008) found that it moderates an endorsement's influence on consumers' purchase intentions. As for the influence of the congruence between the targeted consumers and the endorsed product (also called consumers' product involvement), Celsi and Olson (1988) characterized it as the product's “perceived linkage between an individual's needs, goals, and values […] and their product knowledge” (p. 211). Early on, Mittal and Lee (1989) found a positive influence of consumers' product involvement on their brand perceptions, whereas Ko et al. (2008) indicated that consumers' involvement with the endorsed product positively affected their purchase intentions.

Multiple product endorsements

Multiple product endorsements are omnipresent in marketing practice and can be categorized as multiple type A endorsements and multiple type B endorsements (Hsu and McDonald, 2002; Um, 2008). Multiple type A endorsements are characterized by a single endorser promoting more than one brand or one product at the same time. However, this study's focus is on multiple type B endorsements which are determined by a single brand simultaneously using more than one endorser to promote its products (Erfgen et al., 2016; Rice et al., 2012). Practical examples of such multiple type B endorsements are Nike's “Just do it” campaign and Mercedes-Benz’ brand ambassador strategy, for which the automobile manufacturer engages Roger Federer (tennis), Susie Wolf (motorsport) and Jan Frodeno (triathlon), among others. Companies expect multiple type B endorsements to have an overall positive influence on consumers' brand perceptions and their purchase intentions (Hsu and McDonald, 2002; Rachita et al., 2013). However, Erdogan (1999) questioned, early on, whether using multiple endorsers has an overall positive influence on advertising outcomes, and Tripp et al. (1994) pointed out that the influence of multiple type B endorsements on advertising outcomes erodes because consumers perceive multiple endorsers – as compared to a single endorser – as less believable and less trustworthy. Moreover, Kim and Han (2016) showed that multiple type B endorsements can cause overshadowing effects through which consumers more likely remember the endorsers than the endorsed brand or product, resulting in an overall less effective product endorsement.

Research approach

Conceptual framework

For the purpose of this study, multiple type B endorsements with different endorser combinations (two sports celebrities, two non-sports-related celebrities, and a sports celebrity with a non-sports-related celebrity) were used to analyze their influence on consumers' purchase intentions and the potential moderating effects of endorser-product congruence and consumers' product involvement. The conceptual framework shown in Figure 1 was based on the work of Braunstein-Minkove et al. (2011) and the findings of Felbert and Breuer's (2021) serial mediation analyses for single product endorsements. Accordingly, the effectiveness of multiple type B endorsements is expected to be determined by endorsers' direct and indirect influences on consumers' purchase intentions (Amos et al., 2008). Endorsers' indirect influence on consumers' purchase intentions is determined by their attitudes toward the advertisements' (MacKenzie and Lutz, 1989) and the endorsed brand's (Felix and Borges, 2014) mediating effects. Moreover, consumers' attitude toward the advertisements is expected to directly influence their intentions to purchase the endorsed product, while their attitude toward the endorsers is expected to determine their attitude toward the endorsed brand (Erfgen et al., 2016). The conceptual framework is complemented by consumers' perceptions of endorsers' expertise, trustworthiness (Hovland et al., 1953; Ohanian, 1990) and attractiveness (Baker and Churchill, 1977; Kahle and Homer, 1985) characterizing their attitude toward the endorsers, whereas their attitude toward the endorsed brand is determined by their perceptions of the brand's expertise and trustworthiness (Newell and Goldsmith, 2001).

Endorser-product congruence and consumers' involvement with the endorsed product are expected to moderate the effectiveness of multiple type B endorsements. In accordance to the findings of Rachita et al. (2013) and Till and Busler (2000) who indicated that endorser-product congruence affected endorsers' influence on consumers' brand perceptions, this study hypothesizes for multiple type B endorsements moderating effects of endorser-product congruence on the relationship of consumers' attitude toward the endorsers, their attitude toward the advertisements and their attitude toward the endorsed brand with their intentions to purchase the endorsed product. Moreover, Rice et al. (2012) showed that consumers' product involvement moderated the influence an endorsement had on consumers' perceptions of the endorsed brand. Accordingly, this study tests the moderating effects of consumers' product involvement on endorsers' influences on consumers' attitude toward the advertisement, their attitude toward the endorsed brand and their purchase intentions for multiple type B endorsements.

Research hypotheses

The influence of single product endorsements on consumers' purchase intentions has been widely discussed and validated, whereas research for using multiple endorser combinations is still underrepresented in the academic literature, and evidence of their effects on advertising outcomes are limited (Thomas and Fowler, 2015). Accordingly, practical implications for marketing professionals on selecting endorsers, combining them effectively and predicting endorsers' influences on consumers' intentions to purchase the endorsed product are restricted. Therefore, the purpose of this study is to determine the influences of sports-related multiple type B endorsements, using different endorser combinations, on consumers' purchase intentions and to identify potential moderating effects of consumers' perceived endorser-product congruence and their involvement with the endorsed product.

Using multiple endorsers is expected to have an overall positive direct influence on consumers' purchase intentions. Moreover, it is expected that the effectiveness of multiple type B endorsements depends on attitudinal concepts that mediate endorsers' influences on consumers' brand perceptions and their purchase intentions. Following the findings of McCracken (1989) and Felbert and Breuer's (2021) for single product endorsements, a positive indirect influence of sports-related multiple type B endorsements on consumers' purchase intentions – through their perceptions of the endorsements and the endorsed brand – is expected. Therefore, the following initial research hypotheses are proposed:

H1a.

In a sports-related multiple type B endorsement, consumers' attitude toward the endorsers has a positive, direct influence on consumers' purchase intentions.

H1b.

In a sports-related multiple type B endorsement, endorsers' positive influence on consumers' purchase intentions is mediated by their attitudes toward the advertisements and the endorsed brand.

Additional analyses are needed to determine more systematically the effectiveness of multiple sports-related type B endorsements on consumers' purchase intentions, especially with respect to consumers' perceived endorser-product congruence. Therefore, three different endorser combinations (two sports celebrities, two non-sports-related celebrities and a sports celebrity with a non-sports-related celebrity), with an expected variation in their congruence with the endorsed product, were defined and their influences on consumers' purchase intentions analyzed (Liang and Lin, 2018). The importance of endorser-product congruence in marketing practice has been widely discussed for single endorsements (Till and Busler, 2000), and initial findings in the multiple endorser context indicate that a high level of endorser-product congruence can positively affect consumers' perceptions of the endorsers and the endorsements (Hsu and McDonald, 2002; Rice et al., 2012). However, the influence of endorser-product congruence on consumers' purchase intentions is still unclear, practical implications are missing and further research is needed (Thomas and Fowler, 2015). According to the findings of single product endorsements, a positive, moderating effect of endorser-product congruence is hypothesized for the influence of the endorsers, the endorsements and the endorsed brand on consumers' purchase intentions for a sports-related multiple type B endorsement:

H2a.

In a sports-related multiple type B endorsement, endorser-product congruence has a positive moderating effect on the relationship between consumers' attitude toward the endorsers and their purchase intentions.

H2b.

In a sports-related multiple type B endorsement, endorser-product congruence has a positive moderating effect on the relationship between consumers' attitude toward the advertisements and their purchase intentions.

H2c.

In a sports-related multiple type B endorsement, endorser-product congruence has a positive moderating effect on the relationship between consumers' attitude toward the endorsed brand and their purchase intentions.

Moreover, Lee and Koo (2015) opined that further research is needed to determine more precisely the influence of consumers' involvement with the endorsed product on advertising outcomes in a multiple endorser context. Extant findings for single endorsements showed that consumers' product involvement had a positive influence on their attitude toward the endorsed brand and their intentions to purchase the endorsed product. Rice et al. (2012) showed that, for multiple endorsements, consumers' product involvement moderated endorsers' influence on consumers' brand perceptions. However, Tellis (1998) indicated that a product endorsement can have a negative influence on consumers' brand perceptions and their purchase intentions if consumers are highly involved with the endorsed product as, over time, they have built an own, increased level of product expertise. Therefore, it is expected that highly involved consumers perceive celebrity endorsers as less trustworthy than do consumers with a low level of product involvement such that a negative moderating influence of consumers' product involvement is expected for a sports-related multiple type B endorsement:

H3a.

In a sports-related multiple type B endorsement, consumers' product involvement has a negative moderating effect on the relationship between consumers' attitude toward the endorsers and their purchase intentions.

H3b.

In a sports-related multiple type B endorsement, consumers' product involvement has a negative moderating effect on the relationship between consumers' attitude toward the endorsers and their attitude toward the advertisements.

H3c.

In a sports-related multiple type B endorsement, consumers' product involvement has a negative moderating effect on the relationship between consumers' attitude toward the endorsers and their attitude toward the endorsed brand.

Pre-study

A pre-study with 100 participants, of whom 52% were men and 48% were women, with an average age of 34.2 years, was conducted to choose an adequate sports-related product category for the main experiment. A sports-related product category was chosen, as a significant variation in consumers' perceived endorser-product congruence was expected for sports celebrity endorsers (high level of congruence) and non-sports-related celebrity endorsers (low level of congruence) (Lee and Koo, 2015; Liang and Lin, 2018). Moreover, a sports-related product category was deemed appropriate to complement the extant body of endorsement research as previous studies have mainly used non-sports-related product categories, such as Ilicic and Webster (2011) using soft drinks and Erfgen et al. (2016) using a knife sharpener, to determine endorsers' influences on advertising outcomes. Participants' familiarity with ten preselected, sports-related product categories was surveyed with a single item on a 7-point differential scale, from highly unfamiliar to highly familiar (Roy and Mishra, 2018), to identify a sports-related product category of which consumers would have well-established memory structures. The results of the survey question led to sports shoes, a mid-priced, durable consumer good with little product or financial risk, being selected for the main experiment.

Main experiment

The main experiment used a fictitious sports shoe brand (Erfgen et al., 2016; Rice et al., 2012) and fictitious endorsers to avoid participants' preconceived notions about real brands and real endorsers (Braunstein-Minkove et al., 2011; Ilicic and Webster, 2011). Ensuring that participants did not have preexisting perceptions of the brand or the endorsers enhanced the study's internal validity (Sato et al., 2016). Accordingly, textual descriptions of the fictitious brand and the fictitious endorsers that were provided before the endorsement stimuli were presented indicated that the brand is well known for its sports-related products and that the endorsers are famous celebrities in their respective professions. All endorsement stimuli were similar in design and structure and varied only in the combination of endorsers used. The control group saw product advertisements without reference to endorsers. To give participants adequate time to perceive the endorsement stimuli, endorsements were presented for at least five seconds before participants were allowed to proceed with the survey. After a general introduction to the survey, which included information about the European Data Protection Regulation, participants were randomly assigned to one of only three multiple type B endorsement conditions or the control group. Participants were exposed to and surveyed about a single endorsement condition to avoid priming and anchoring effects. In the multiple endorser conditions, two sports celebrities, two famous movie actors, or a sports celebrity and a famous movie actor endorsed a pair of fictitious sports shoes.

After seeing the endorsement stimuli, participants used a 7-point differential scale to rate their attitude toward the endorsers with five items (Kim and Na, 2007; Ohanian, 1990) on endorsers' expertise, trustworthiness and attractiveness. Next, participants used a 5-point differential scale to rate their attitude toward the advertisements with four items (Mitchell and Olson, 1981), a 7-point Likert scale to rate their attitude toward the endorsed brand with four items on the brand's expertise and trustworthiness (Newell and Goldsmith, 2001), a 9-point differential scale to rate three items for endorser-product congruence (Speed and Thompson, 2000) and a 7-point differential scale to rate three items for the dependent variable of consumers' purchase intentions (Koernig and Boyd, 2009) (Table 1). Thereafter, consumers used a 5-point differential scale to rate twenty items on their involvement with the endorsed product (Zaichkowsky, 1985). As only well-proven and statistically validated theoretical constructs were applied, different scales and scale endpoints were used. Next, sociodemographic factors like gender, age, educational qualification, profession, income and monthly sports-related expenditures were surveyed, and participants responded to a set of control questions. A voucher lottery was offered at the end of the survey to increase participation.

The main experiment was conducted in German to account for participants' language preferences. All questions, items and scale endpoints were translated from English to German, the wording was tested with a group of non-participants and minor edits were made. A second-order confirmatory factor analysis was used to determine whether the variables applied adequately represent the underlying theoretical constructs (Morata-Ramírez and Holgado-Tello, 2012). For further analyses, items were parceled (unweighted average) for each construct (Matsunaga, 2008), and Cronbach's alpha values were calculated to test for internal reliability (Cronbach, 1951; Hair et al., 2006). Items that measure consumers' involvement with the endorsed product were summed (Zaichkowsky, 1985). Next, (moderated) serial mediation analyses were performed to determine the endorsers' direct and indirect influences on consumers' purchase intentions and potential moderating effects of endorser-product congruence and consumers' involvement with the endorsed product.

As a common method variance (CMV) bias would distort the analyses' findings, countermeasures were applied before the main experiment was conducted (Chang et al., 2010; Lindell and Whitney, 2001). Different scales and scale endpoints were used, and the order of construct items were randomized, if applicable (Podsakoff et al., 2003). It was also controlled in retrospect for a CMV bias, since a single-sourced research design with a self-reported online questionnaire was used (Chang et al., 2010). The Harman single-factor test and the marker variable test indicated an overall low likelihood of a common method bias (Harman, 1976).

Results of the main experiment

Data preparation and data sample

The 257 participants – students, employees and people in their respective social environment – in the main experiment were recruited using the network of a major German university and their responses were collected in an online survey (survey tool: Unipark). Before factor analyses were conducted, data cleaning was performed, and 24 responses (9.3%) were removed from the data sample because of logical inconsistencies or because participants failed at least one of the control questions. Accordingly, the final data sample consisted of 233 useful responses with 56–60 responses in each condition; the data were analyzed using the statistical software SPSS Statistics, including Hayes's PROCESS extension tool (Hayes, 2018), and SPSS Amos Graphics. One hundred ten of the 233 responses were male (47.2%) and 123 were female (52.8%), and the average age was 36.7 years. Most respondents had an academic or equivalent degree (79.0%), and their median monthly net income, which 78.5% reported voluntarily, was 2,501–3,000 EUR/month. Participants reported that they spend an average of 12.0% (equal to approximately 350 EUR) of their monthly net income on sports-related consumer goods, supporting the supposition that sports shoes are an adequate, sports-related durable product category.

A Kaiser-Meyer-Olkin value of 0.95 and a significant Bartlett's test of sphericity (p < 0.01) indicated that the underlying data were sufficient to conduct further factor analyses (Kaiser and Rice, 1974). A second-order Confirmatory Factor Analysis was performed to test whether items adequately represented the theoretical constructs (Morata-Ramírez and Holgado-Tello, 2012), followed by serial mediation analyses (PROCESS models #4 and #6) to determine endorsers' direct and indirect influences on consumers' purchase intentions for the sports-related multiple type B endorsements and the control group. Thereafter, moderated serial mediation analyses were performed to determine potential moderating effects of endorser-product congruence (PROCESS model #88) and consumers' product involvement (PROCESS model #84) (Figure 2). As correlation coefficients of the latent constructs and variance inflation factors were below recommended thresholds, multicollinearity was unlikely to distort the analyses' findings (Hair et al., 2006).

Second-order confirmatory factor analysis

A second-order CFA (Figure 3) was conducted to reduce the underlying level of complexity and account for the limited sample size, as a larger number of variables is explained by a smaller number of latent constructs (Luo, 2011). Convergent and discriminant validity were established (Table 2), and the model fit indices applied indicated that the model fit the data well (CMIN/df = 2.18, RMSEA = 0.07, CFI = 0.97). After three construct items (Att_Ad_II, C_Expert_IV, and C_Trust_IV) were removed because of insufficient factor loadings, the remaining standardized regression coefficients were all highly significant (p < 0.01), and factor loadings above 0.79 indicated that the items represent the underlying constructs well. Moreover, model factors were tested for internal consistency (α > 0.87) as well as for construct reliability (average variance extracted >0.70).

(Moderated) serial mediation analyses

Standardized path coefficients were analyzed to determine the influential effects of the multiple type B endorsements (PROCESS model #6) and the control group (PROCESS model #4) on consumers' intentions to purchase the endorsed product (Table 3). The results of the analysis revealed that sports-related multiple type B endorsements had, independent of the endorser combination used, a positive, indirect influence through consumers' attitudes toward the advertisements and the endorsed brand on their purchase intentions. Group comparisons further indicated that the influence of multiple type B endorsements on consumers' intentions to purchase the endorsed product was greater compared to the influence of the product advertisement without reference to an endorser.

Conducted serial mediation analyses confirmed the conceptual framework for sports-related multiple type B endorsements to a large extent. However, the direct influence of attitude toward the endorsers on consumers' purchase intentions was not significant, so H1a was not supported. Instead, it was consumers' attitude toward the advertisements and toward the endorsed brand that mediated endorsers' influence on consumers' purchase intentions. Moreover, consumers' perceptions of the advertisement stimuli directly influenced their purchase intentions, resulting in a 49% explained variance (R² = 0.49). The 95% confidence interval (CI) of the total indirect effect of endorsers' influences on consumers' purchase intentions was different from zero (indirect effect = 0.21, standard error (SE) = 0.05, 95% CI = [0.11, 0.30]), so H1b was supported. The mediating effect of consumers' perceptions of the endorsers on their intentions to purchase the endorsed product – through attitude toward the advertisements and attitude toward the endorsed brand – was the weakest, isolated indirect effect (indirect effect = 0.03, SE = 0.01, 95% CI = [0.01, 0.05]), whereas the mediating effect of endorsers' influence on consumers' purchase intentions through their attitude toward the advertisements (indirect effect = 0.09, SE = 0.04, 95% CI = [0.03, 0.17]) and through their attitude toward the endorsed brand (indirect effect = 0.09, SE = 0.03, 95% CI = [0.04, 0.14]) was significantly stronger.

Subsequently, moderated serial mediation analyses were conducted to determine potential moderating effects of endorser-product congruence and consumers' product involvement on their purchase intentions. The first moderated serial mediation analysis (PROCESS model #88) found that endorser-product congruence had a significant moderating influence on the relation between consumers' attitude toward the advertisements and their purchase intentions. Therefore, H2b was supported. However, H2a and H2c were not supported, as the direct influence of attitude toward the endorsers on consumers' purchase intentions and the moderating effect of endorser-product congruence on the relationship between consumers' attitude toward the endorsed brand and their purchase intentions were both not significant. The moderated mediation index for the moderated influence of attitude toward the advertisements on consumers' purchase intentions was above zero, the 95% CI was different from zero and the 51% explained variance of consumers' purchase intentions (R² = 0.51) was slightly higher than in the simple serial mediation regression. The moderating effect of endorser-product congruence on the relationship between consumers' attitude toward the advertisements and their purchase intentions increased with the level of congruence (Figure 4). Further analyses revealed that participants perceived endorser-product congruence for the multiple endorser condition with two sports celebrities (average congruence score: 6.0) similar to the congruence of the multiple endorser condition using a sports celebrity and a well-known movie actor (average congruence score: 5.8), whereas the endorsement condition with two well-known movie actors showed a significantly lower level of endorser-product congruence (average congruence score: 4.5).

The second moderated serial mediation analysis (PROCESS model #84) was performed to determine potential moderating effects of consumers' product involvement on the relationship between consumers' attitude toward the endorsers and their attitude toward the advertisements, their attitude toward the endorsed brand and their purchase intentions. As endorsers' direct influence on consumers' purchase intentions was not significant, H3a was not supported. H3b and H3c were also not supported, as the moderated mediation indices of both indirect effects and their 95% CI contained zero. This was supported by further analyses for which consumers were categorized in up to five independent groups, each with its distinct level of product involvement, and also indicated nonsignificant moderation effects of consumers' product involvement. Thus, the effectiveness of a sports-related multiple type B endorsement in influencing consumers' purchase intentions was – at least in this study – not significantly affected by consumers' level of product involvement.

Discussion

This study contributes to the extant endorsement literature along the three articulated research questions and provides practical implications for marketing professionals. The study's findings show that using multiple endorsers can increase consumers' intentions to purchase an endorsed product. Moreover, the study confirms the relevance of attitudinal concepts in the multiple endorser context and shows that a multiple type B endorsement should always be embedded in a company's overall brand(-building) strategy if it is to unfold its full potential because consumers' attitudes toward the endorsements and the endorsed brand mediate endorsers' influence on their purchase intentions. Therefore, marketing professionals should carefully account for consumers' perceptions of the endorsements and the endorsed brand.

The study's findings show that, in a multiple-endorser context, endorsers' congruence with the endorsed product is critical for the endorsements' effectiveness, as it moderates endorsers' influence on consumers' purchase intentions. The moderated serial mediation analysis also indicates that it is not a single endorser's congruence with the endorsed product but the congruence of the combination of endorsers used that determines the effectiveness of a multiple type B endorsement. A sports-related endorsement using two sports celebrities is highly effective in influencing consumers' purchase intentions, as consumers perceive them as highly congruent with the endorsed product. Thus, it supports the findings of Liang and Lin (2018), who pointed out that athlete-product congruence, in contrast to celebrity-product congruence, has a positive influence on consumers' purchase intentions. However, the present study finds that consumers perceive a sports-related multiple type B endorsement with a sports celebrity and a movie actor as endorsers as similar congruent, indicating that consumers assess endorser-product congruence in a multiple endorser context in terms of all the endorsers used. Therewith, the study's findings indicate spillover effects between endorsers if at least one endorser is highly congruent with the endorsed product and shows that selecting appropriate endorsers is a central consideration of marketing practice. Marketing professionals can take advantage of spillover effects between endorser by, for example, selecting a mixed combination of endorsers with varying levels of endorser-product congruence. In this way, they have an endorser combination that affects consumers' purchase intentions while simultaneously being able to target different consumer groups (Hsu and McDonald, 2002). However, an endorsing company can also decide to target potentially fewer consumers by engaging only sports-related celebrity endorsers who are all highly congruent with the endorsed product and therewith having a significant moderating influence on consumers' intentions to purchase the endorsed product. Multiple endorsers, none of whom are particularly congruent with the endorsed product, have been shown to be less effective in influencing consumers' purchase intentions, as such combinations of endorsers are less congruent with the endorsed product.

The study's findings also show that consumers' involvement with the endorsed product does not significantly determine the effectiveness of a sports-related multiple type B endorsement on consumers' purchase intentions, indicating that companies are able to address consumers independent of their extant level of product involvement. Thus, highly involved consumers can be targeted with a multiple type B endorsement as they perceive endorsers as highly familiar and potentially identify themselves with them (Thomas and Fowler, 2015). As for consumers with a low level of product involvement, they perceive celebrity endorsers as a trustworthy source of information with a high anticipated level of product expertise and, thus, a high level of source credibility (Lee and Koo, 2015). Therewith, the study's findings help to simplify a company's endorsement practices by showing that a brand can target existing consumers and new consumers with the same multiple type B endorsement without having to specify it to each targeted group of consumers.

Limitations and directions of future research

This study relies on simplifications of complex, real-life product endorsements, which limit the findings' generalizability. Only a single product (category), certain types of endorsers and a limited number of multiple endorser combinations were used. Using a fictitious study design to eliminate participants' prior perceptions of the brand and the endorser (Ilicic and Webster, 2011; Lee and Thorson, 2008) increases the study's internal validity but diminishes its external validity (Kim et al., 2017). Consequently, the study's generalizability is limited and its practical significance is restricted (Lee and Koo, 2015). The study's generalizability is further constrained as the experiment's data have been gathered in a geographically confined area with a socially restricted group of survey participants.

Analyzing the effectiveness of multiple endorsers for additional types of endorsements like social media advertisements could further increase the understanding of endorsers' influences on advertising outcomes (Lee et al., 2017; McCormick, 2018). For social media advertisements, researchers could use the interaction between consumers and endorsers (e.g., consumers liking or commenting endorsers' posts) as an estimation of consumers' involvement with the endorser and the endorsed product and analyze the influence on consumers' purchase behavior. Moreover, cross-cultural comparisons including high- versus low-power-distance countries could extend the current body of multiple endorsement research. Analyzing consumers' real purchase behavior instead of their purchase intentions would further strengthen the understanding of endorsements' effectiveness in a more realistic context.

Figures

Figure 1

Conceptual framework

Figure 2

(Moderated) serial mediation models

Figure 3

Second-order confirmatory factor analysis

Figure 4

Moderating effect of endorser-product congruence

Table 1

Research constructs and Cronbach's alpha values

Construct		Items	Scale	Source	Cronbach's alpha
Attitude toward the endorser	Source expertise	Expert ↔ Not an expert	7-point	Kim and Na (2007), Ohanian (1990)	α = 0.99
		Experienced ↔ Inexperienced	Differential
		Knowledgeable ↔ Unknowledgeable	Scale
		Qualified ↔ Unqualified
		Skilled ↔ Unskilled
	Source trustworthiness	Dependable ↔ Undependable	7-point	Kim and Na (2007), Ohanian (1990)	α = 0.99
		Honest ↔ Dishonest	Differential
		Reliable ↔ Unreliable	Scale
		Sincere ↔ Insincere
		Trustworthy ↔ Untrustworthy
	Source attractiveness	Attractive ↔ Unattractive	7-point	Kim and Na (2007), Ohanian (1990)	α = 0.99
		Classy ↔ Not Classy	Differential
		Beautiful ↔ Ugly	Scale
		Elegant ↔ Plain
		Sexy ↔ Not sexy
		Bad ↔ good	5-point	Mitchell and Olson (1981)	α = 0.87
Attitude toward the		Irritating ↔ Not irritating	Differential
Advertisement		Dislike ↔ Like	Scale
		Uninteresting ↔ Interesting
Attitude toward the Brand		The XYZ corporation has a great amount of expertise	7-point	Newell and Goldsmith (2001)	α = 0.93
	Corporate	The XYZ corporation is skilled in what they do	Likert
	Expertise	The XYZ corporation has great experience	Scale
		The XYZ corporation does not have much experience
		I Trust the XYZ corporation	7-point	Newell and Goldsmith (2001)	α = 0.89
	Corporate	The XYZ corporation makes truthful claims	Likert
	Trustworthiness	The XYZ corporation is honest	Scale
		I do not believe what the XYZ corporation tells me
Consumers' purchase intentions		Unlikely ↔ Likely	7-point	Koernig and Boyd (2009), Till and Busler (2000), Yi (1990)	α = 0.97
		Impossible ↔ Possible	Differential
		Improbable ↔ Probable	Scale

Table 2

Descriptive statistics and correlation of variables

	Scale	Mean	SD	AVE	CR	Factor Correlations
						A_Endorser	A_Advertisement	A_Brand	Purchase Intention
A_Endorser	7-point	2.70	1.65	0.90	0.95	1.00
A_Advertisement	5-point	2.69	1.02	0.70	0.87	0.18	1.00
A_Brand	7-point	4.12	1.11	0.80	0.89	0.35	0.48	1.00
Purchase intention	7-point	2.90	1.66	0.93	0.98	0.18	0.67	0.59	1.00

Note(s): A = attitude; SD = standard deviation; AVE = average variance extracted; CR = composite reliability

Table 3

Standardized path coefficients

		Two sports celebrity endorsers	A sports celebrity and a non-sports-related celebrity endorser	Two non-sports-related celebrity endorsers	Control group (no endorser reference)
Regression model (according to Hayes)		model #6	model #6	model #6	model #4
Group size		n = 60	n = 59	n = 56	n = 58
Standardized path coefficients
A_Endorser	→ A_Advertisement	0.84***	0.81***	0.70***	n/a
	→ A_Brand	0.74***	0.62***	n.s	n/a
	→ purchase intention	n.s	n.s	n.s	n/a
A_Advertisement	→ A_Brand	0.26**	n.s	0.46***	0.54***
	→ purchase intention	0.78***	0.72***	0.64***	0.71***
A_Brand	→ purchase intention	0.36**	0.80***	0.67***	0.35**

Note(s): ***: p < 0.01; **: p < 0.05; *: p < 0.10; A = Attitude; n.s. = not significant; n/a = not available