Introduction

Despite the gains girls and women have made in the sports and athletics context, they continue to be marginalized in many respects. For instance, women are under-represented as athletes on college campuses (Acosta and Carpenter 2006), and according to the Equity in Athletics Report (see http://www.ope.ed.gov/athletics/), they receive an inequitable share, relative to their proportion as student athletes, of the operating expenses. Women are also under-represented in coaching and leadership positions in virtually every sport context (Acosta and Carpenter 2006; Cunningham 2007). The disparities are not limited to the representation of women as athletes, coaches, and administrators—women are also represented in the media in stereotypical and sexist ways. The mass media often portrays sport as a masculine endeavor (Cramer 1994; Tuggle 1997), and women who participate in traditionally “feminine” sports, such as tennis or swimming, are featured more frequently than are their counterparts who take part in “masculine” sports, such as rugby (Fink and Kensicki 2002; Lumpkin and Williams 1991). Women who do receive media coverage are depicted as sex-symbols rather than “powerful, talented athletes” (Fink and Kensicki 2002, p. 331) and in supportive rather than participating roles (Cuneen and Sidwell 1998; Tuggle 1997).

More recent research has focused on women as athlete endorsers, a line of inquiry likely promulgated by the millions of dollars spent on celebrity endorsers each year (Broughton et al. 1999). Though athletes are used as endorsers in 11% of all television advertisements, women are only used 3% of the time (Turner et al. 1995). When women are present, they are depicted in stereotypical ways, in leisure time activities, or as participating in individual activities (Cuneen and Claussen 1999; Cuneen and Sidwell 1998; Lynn et al. 2002). These stereotypical depictions are also present in magazines devoted to women, such as Shape and Sports Illustrated for Women (Lynn et al. 2004). Several experimental studies have also shed light on the influence of gender. Peetz et al. (2004) found that (a) participants were more likely to identify a male celebrity in an advertisement and (b) male participants were more heavily influenced by male endorsers than they were by female endorsers, while the female participants were equally influenced by both male and female endorsers. Furthermore, Boyd and Shank (2004) found that men were more influenced by athlete endorsers than were women. Finally, Fink et al. (2004), in their study of female athlete endorsers, found that the attractiveness of the athlete and her expertise level both significantly contributed to participants’ view that she was an appropriate endorser for the event (i.e., a collegiate softball game), though the effects of expertise were significantly stronger than were those of attractiveness. In short, these studies illustrate that gender and gender stereotypes influence the use and effectiveness of athlete endorsers.

In the current investigation, we extend the athlete endorser line of research, and specifically the work of Fink et al. (2004), in several meaningful ways. First, the context of the study of Fink et al. (i.e., softball) may have contributed to the emphasis placed on expertise over attractiveness. In other, more “feminine” and “sex appropriate” sports, such as tennis (Kane and Parks 1990), different patterns may emerge. We examine this possibility. In a related way, while Fink et al. focused on the direct effects of attractiveness and skill level on subsequent attitudes toward the event, we extend their work by investigating the potential interactive influence of these two variables. The match-up hypothesis and associative learning theory were utilized to develop the hypotheses. Based on principles within these theories, we proposed that athlete attractiveness and expertise influence perceptions of the athlete as an appropriate endorser for the event. Further, we proposed that perceptions of appropriateness lead to positive attitudes toward the event and, ultimately, intentions to purchase tickets. In the next section we explain the theoretical foundation of the study and present hypotheses.

Theoretical Framework

According to the match-up hypothesis, endorsements are more effective when there is a match, or “fit,” between the individual endorser and the product being endorsed (Kamins 1990). Endorsers who are considered credible (Ohanian 1991) or appropriate (Till and Busler 1998) are more effective than are their counterparts. Such evaluations are complex, as they consist of the endorser’s expertise, attractiveness, and trustworthiness (Ohanian 1990), though limited support has been found for the latter dimension (Ohanian 1991). As an example, a professional athlete, because of her athletic acumen, might be seen as a more credible endorser of a sports drink than would a television actor.

Associative learning theory (Klein 1991) speaks more specifically to these dynamics. The theory suggests that different concepts can be linked in one’s mind to form an associated network of memory. Once connected, each concept is gathered every time the other concept is elicited (Anderson 1983; Till and Shimp 1998). When endorsers are used to market products, an individual’s experiences and attitudes about both are summoned and a link is developed. Over time, the product and endorser become part of one’s “association set,” and when either of the two is observed, the other immediately comes to mind (Till and Busler 2000). Most important to the match-up hypothesis, the strength of the association depends upon the fit, or shared schema, between the endorser and the product (Lynch and Schuler 1994; Till and Busler 2000).

Research largely supports these tenets. For example, Till and Busler (2000) found that athletes were more effective endorsers of energy bars than actors, suggesting that the fit between the product and the endorser was the result of the athlete’s expertise regarding the characteristics of a good energy bar. Charbonneau and Garland (2006) found that four New Zealand athletes were all considered experts when endorsing an unbranded sports drink, yet mean ratings of other traits (i.e., trustworthiness, attractiveness) differed. In a study of young consumers, Veltri et al. (2003) showed that a majority of 10–14 year old consumers were more likely to be influenced by an athlete endorser when he/she was endorsing an athletic product. Finally, Till (2001) found that an athlete endorsers’ images can be tainted if they endorse an “inappropriate” product, particularly a product found to be a health risk. Obviously, product-endorser fit is key to the formation of positive attitudes and purchase intentions toward a product.

Antecedents

Though previous match-up research has focused on the endorsement of tangible goods, such as a sports drink, we depart from this trend by examining the effects of endorser characteristics on promoting an athletic event (see also Fink et al. 2004). In drawing from associative learning theory, we argue that the athlete’s skill level should be positively associated with perceptions of the appropriateness of that athlete as an endorser. This position is consistent with Ohanian (1990, 1991), who argued that expertise should result in greater endorser credibility. In this case, the link between the highly skilled athlete and the competitive athletic event should be strong. Fink et al. (2004) found evidence of this in their study of advertisements for a collegiate softball championship: athlete skill was significantly associated with perceptions that the athlete was an appropriate endorser for the event.

In addition to considering the effects of athlete expertise, we also examined the influence of attractiveness. Past research has shown that people generally have more positive attitudes toward attractive persons than they do toward their less attractive counterparts (Eagly et al. 1991; Jawahar and Mattsson 2005). Further, and consistent with associative learning theory, attractive persons are considered to be an appropriate endorser for products that relate to one’s physical appearance, such as hair products or perfume (Ohanian 1991). How then, does attractiveness factor into a discussion of promoting a women’s athletic event? Consistent with Werthein (2002), we argue that athlete attractiveness has become increasingly utilized in promoting and selling many women’s sports, especially “sex-appropriate” sports such as ice skating, swimming, and tennis. Indeed, research suggests that attractive and feminine athletes receive more press and more endorsements than do their counterparts, regardless of skill level (Spencer and McClung 2001). If this is the case, then it is possible that consumers may have formed an associated network linking “sex-appropriate” sport athletes with physical attractiveness. In other words, it is possible that people have come to associate some sports with the attractive athletes who participate in them (e.g., tennis and Maria Sharapova). Thus, inclusion of attractiveness is consistent with the associative learning perspective, such that perceptions of endorser appropriateness might be enhanced when the athlete is highly attractive.

Outcomes

Athletes who are considered an appropriate endorser for a product or event should elicit positive responses from the consumers. We consider two such responses: brand attitudes and purchase intentions. The purpose of using athlete and celebrity endorsers is to positively influence consumers’ attitudes toward the product itself—something that should be enhanced when the athlete is an appropriate match for the product. Till and Busler (2000) demonstrated evidence of this, as a close match between the endorser and product was found to positively influence attitudes toward the product. The positive attitudes toward the product should then result in desired behavioral manifestations, such as purchase intentions and behaviors (Madrigal 1995). Indeed, this reasoning is consistent with the theory of planned behavior (Ajzen 1991), whereby positive dispositions toward an action should result in people behaving accordingly. In the consumer behavior context, positive attitudes toward a sport event should result in people intending to attend that event (see Cunningham and Kwon 2003). This theorizing suggests a mediation perspective, where the positive relationship between athlete appropriateness and intentions to purchase tickets to the event should be mediated by positive attitudes toward that event.

Current Study

The current study was designed to examine the influence of an athlete endorsing a women’s athletic event. Following Till and Busler (2000) and Fink et al. (2004), we created a fictional athlete to endorse the National Collegiate Athletic Association (NCAA) National Tennis Championship in which the fictional athlete’s team was playing. Based on the review of the literature, we developed several hypotheses. Consistent with associative learning theory and the existing match-up hypothesis literature, we expected that athletes with high expertise would be perceived as a more appropriate endorser for the event than would those with low expertise (hypothesis 1). Similarly, attractive athletes were expected to be viewed as more appropriate endorsers of the event than were less attractive athletes (hypothesis 2). Finally, in line with our discussion of match-up outcomes, we expected athlete appropriateness to be positively associated with positive attitudes toward the event (hypothesis 3), which in turn, were expected to hold a positive relationship with intentions to purchase tickets to the event (hypothesis 4). Though our study was experimental in nature, we tested our hypotheses through observed path analysis to allow for simultaneous testing of the hypotheses (see MacCallum and Austin 2000).

Method

Participants

The participants were undergraduate students (N = 184) enrolled in sport management and journalism classes in three large mid-western universities. The sample was mostly White (n = 159, 86.4%) and consisted of a relatively even mix of men (n = 105, 57.1%) and women (n = 78, 42.4%). The mean age was 20.76 years (SD = 2.60).

Procedures

We employed a 2 × 2 full factorial design such that participants were randomly assigned to one of four conditions: high attractiveness-high expertise (n = 45; 23 men, 22 women), high attractiveness-low expertise (n = 51; 27 men, 23 women, 1 did not provide gender information), low attractiveness-high expertise (n = 44; 27 men, 17 women), and low attractiveness-low expertise (n = 44; 28 men, 16 women). The advertisements included a close range photo of a woman and read: “_____ University NCAA Tennis. Come see MARY ENDRIES this week at the NCAA Women’s Tennis Championship.” The university connection differed by the school in which the data were collected. The caption was left of the athlete’s face. The attractive athlete had long blonde hair, a lean face, and light make-up, while the less attractive athlete had shorter brown hair, a slightly heavier face, and no make-up. The high expertise conditions included the following sentences, in addition to the caption above: “Come see #1 tennis player, Mary Endries this week at the NCAA Tennis Championship” and “Mary Endries is: Last year’s NCAA Champion, Voted #1 NCAA player 2002, Voted #1 division player 2001 & 2002, Voted #1 _______ University player 2000, 2001, 2002.” In the low-expertise conditions, these phrases were absent.

The study was conducted during class time, and participation was voluntary. We provided the materials (the one-page advertisement and the subsequent 2 pages with measures) to participants face down. A brief introduction of the study ensured in which participants were told that they were being employed to ascertain the effectiveness of advertisements for women’s sports. Participants had 30 s to view the advertisement. Then they completed a 2-page questionnaire. Pilot testing revealed that the manipulation was successful.

Measure

The questionnaire used in the study ascertained demographic variables (age, race, gender) as well as the variables of interest in the study. The mean of the items represented the final score for each measure. Reliability estimates (Cronbach’s alpha) for each measure in the questionnaire were calculated and are reported below.

Athlete-event Appropriateness

The appropriateness of the athlete as an endorser for the event was measured using an adapted version of Till and Busler’s (2000) five item scale: “I think the athlete is an appropriate endorser of the NCAA Championships,” “I think the combination of the athlete and the NCAA championships goes together well,” “I think the combination of the athlete and the NCAA championships is a good fit,” “I think the athlete is an effective endorser of the NCAA championships,” and “I think the athlete is a suitable endorser of the NCAA championships.” Items were measured on a 9-point Likert-type scale ranging from 1 (strongly disagree) to 9 (strongly agree), and the reliability estimate was high (α = .96).

Attitudes

Similar to Till and Busler (2000), attitudes toward the event were assessed using three semantic differential scales in response to the following phrase, “In general, how do you feel about the event?” The three scales were “strongly dislike–strongly like,” “negative–positive” and “unfavorable–favorable.” All items were measured on a 9-point scale. The reliability estimate for the measure was high (α = .91).

Purchase Intentions

Till and Busler’s (2000) items (3) were used to measure intentions to purchase a ticket. Participants responded to the following, “How likely is it that you would consider purchasing a ticket to this event?” The phrase was anchored by 9-point semantic differential scales with endpoints “definitely would not–definitely would,” “unlikely–likely,” and “improbable–probable.” There was a high reliability estimate for the measure (α = .97).

Manipulation Checks

To ensure the efficacy of the manipulations, we used Ohanian’s (1990) scale to assess the attractiveness and expertise of the athlete. Items from both scales were preceded by the phrase, “the athlete in the advertisement is...” and utilized 9-point semantic differential scales. For the attractiveness scale, the endpoints were “unattractive–attractive,” “not classy–classy,” “ugly–beautiful,” “plain–elegant,” and “not sexy–sexy.” For the expertise scale, the endpoints were “not an expert–an expert,” “inexperienced–experienced,” “unknowledgeable–knowledgeable,” “unqualified–qualified,” and “unskilled–skilled.” Reliability coefficients for the attractiveness (α = .97) and expertise (α = .95) measures were high.

Data Analysis

We conducted two analysis of variance procedures to assess the efficacy of the experimental manipulations. Means, standard deviations, and bivariate correlations were then computed for all variables. Hypotheses 1–4 were then tested through observed path analysis, using AMOS 7.0 (Arbuckle 2006). We followed Marsh et al. (2004b) recommendations to test for interactions among the variables (i.e., the conditions: attractiveness and expertise). Further, given that men and women respond differently to athlete endorsers (Peetz et al. 2004), we included participants’ gender in model, including paths from gender to appropriateness, attitudes, and purchase intentions. We interpreted the root mean square error of approximation (RMSEA) and comparative fit index (CFI) to assess model fit. RMSEA values less than .06 and CFI values greater than .95 are generally considered as indicative of close model fit (Hu and Bentler 1998, 1999; but see also Marsh et al. 2004a, for alternative arguments).

Results

Manipulation Checks

Persons in the two high attractiveness conditions perceived the athlete as significantly more attractive (M = 7.35, SD = 1.36) than did persons in the low attractiveness conditions (M = 2.65, SD = .93), F(1, 182) = 737.82, p < .001. Similarly, persons in the high expertise conditions rated the athlete as more skilled (M = 6.53, SD = 1.87) than did persons in the low expertise conditions (M = 5.39, SD = 2.38), F(1, 180) = 12.74, p < .001. Both of these results suggest that the manipulation was successful.

Descriptive Statistics

Means and standard deviations are presented in Table 1, while bivariate correlations are presented in Table 2. In interpreting the results, readers should remain cognizant that attractiveness was coded as 0 = low attractiveness, 1 = high attractiveness, while expertise was coded as 0 = low expertise, 1 = high expertise. From a direct effects perspective, persons who viewed an attractive athlete perceived her as a more appropriate endorser for the event (r = .36, p < .001), had more positive attitudes toward the event (r = .24, p < .01), and had greater purchase intentions (r = .21, p < .01). Expertise, on the other hand, was not related to any of those outcomes. Perceptions of endorser appropriateness were significantly associated with purchase intentions (r = .36, p < .001), as were positive attitudes toward the event (r = .67, p < .001).

Table 1 Means and standard deviations.
Full size table
Table 2 Means, standard deviations, and bivariate correlations.
Full size table

Hypothesis Testing

Observed path analysis was used to test the study hypotheses. The model provided a close fit to the data: χ2 (df = 9, n = 184) = 7.06, p = .63; RMSEA (90% CI:.00,.07) = .01; CFI = 1.00. Figure 1 provides an illustrative summary of the findings. The model explained 24% of the variance in appropriateness of the athlete as an endorser, 13% of the variance in positive attitudes toward the event, and 48% of the variance in purchase intentions.

Fig. 1
figure 1

Illustrative summary of hypothesized model.

Full size image

The control variable, participant gender, was negatively associated with the appropriateness of the athlete (β = −.21, p < .01), meaning that men viewed the athlete as a more appropriate endorser for the event than did women.

Hypothesis 1 predicted that expertise would be positively related to perceptions of the athlete as an appropriate endorser for the event, while Hypothesis 2 made similar predictions for attractiveness. Both of these hypotheses were supported (β = .34, p < .001, and β = .55, p < .001, respectively). Note, however, that these effects were qualified by a significant expertise × attractiveness interaction (β = −.32, p < .01). The nature of the interaction is presented in Fig. 2. Results indicated that, among the athletes with high levels of expertise, there were no differences in perceived appropriateness between the highly attractive and less attractive athlete. However, among the athletes with low expertise, the attractive athlete was perceived as a more appropriate endorser than was the less attractive athlete.

Fig. 2
figure 2

Effects of attractiveness and expertise on athlete-event appropriateness.

Full size image

Hypothesis 3, which predicted a positive relationship between athlete appropriateness and positive attitudes toward the event, was supported (β = .36, p < .001).

Support was also demonstrated for Hypothesis 4, as positive attitudes toward the event were positively related to intentions to purchase tickets to that event (β = .60, p < .001).

Discussion

The purpose of the study was to examine the influence of athlete endorser characteristics on subsequent consumer attitudes and behavioral intentions. In drawing from associative learning theory, we proposed that, within the context of a traditionally “feminine” women’s sport event, both the expertise and attractiveness of the athlete would influence perceptions of that athlete as an appropriate endorser of the event. Our results were in line with these predictions, thereby supporting previous match-up hypothesis literature (e.g., Fink et al. 2004; Ohanian 1991). Interestingly, however, the main effects were qualified by an expertise × attractiveness interaction (for similar findings, see Charbonneau and Garland 2006). While we found that the most attractive and most expert endorser will produce the greatest effect, our results also suggest that deficits in the area of expertise can be off-set by higher levels of attractiveness. Thus, our results showed that a less expert endorser may still be effective as long as she is attractive. These findings lend support to the growing research literature that suggests that women athletes are rewarded more for their physical attributes than they are for their physical ones (Spencer and McClung 2001; see also Eagly et al. 1991). In other words, all other things being equal, attractive people are perceived to possess more positive characteristics than those who are less attractive. This would seem to be especially the case in tennis (the context of the current study), as opposed to sports that have traditionally had a less “feminine” emphasis, such as softball or rugby.

It would be interesting, however, to determine if this finding would also hold true for male athletes endorsing athletic events. That is, would the same interaction be found; would a less skilled male athlete be considered a good fit if he were attractive? Perhaps women’s sport, and especially women’s tennis, poses a special case. The Women’s Tennis Association (WTA) has a long tradition of utilizing sex appeal in their promotional campaigns (Werthein 2002). The match-up hypothesis suggests that the strength of an association set in one’s mind depends upon the match between the product and the endorser. However, perhaps there is a longitudinal effect on the match-up hypothesis. That is, an attractive (yet less expert) endorser may not, at first, create a strong link with an athletic event in a consumer’s mind, but over time and repeated exposure, associative learning theory would suggest the link can be forged. Because promoters of women’s sport often attempt to exploit female athletes’ sex appeal, athlete attractiveness, expertise, and women’s tennis may all be part of the same association set in consumers’ minds.

Similar to the findings of Fink et al. (2004), perceptions of athlete appropriateness as an endorser were related to positive attitudes toward the event and all were related to purchase intentions. The interaction of attractiveness and expertise explained a large amount of variance in fit (18%), and the model itself explained nearly 50% of the variance in purchase intentions. Thus, it appears that who is chosen as an endorser of a women’s tennis event could be quite vital to attendance. Just as not all celebrity endorsers are successful in marketing a product or service (Mittelstaedt et al. 2000), our results suggest that not all athletes are equal in promoting their sporting events.

Limitations and Directions for Future Research

Using a fictitious endorser and event provided greater internal validity; however, it also limited the study’s generalizability to actual endorsers and events. Consistent with Till and Busler (2000), “we chose to sacrifice some ecological validity (by using a fictitious rather than a real athlete), but gained construct validity by minimizing spurious confounds and statistical conclusion validity by minimizing within-group variation with our manipulation” (p. 5). While the results suggest attractiveness and expertise interact to produce the highest level of appropriateness for a women’s tennis event, future studies should attempt to use real athletes and events. However, care must be taken to minimize confounding variables in such studies (e.g., athlete personality, race, etc.).

Future studies should also include male athletes as endorsers of their sport to determine whether differences exist across male and female sporting events. Other demographic variables (e.g., age, race) can be manipulated to determine their effects on athlete-event fit. For example, many of the National Basketball Association’s marquee players are African American; thus, race may play a role in consumers’ ideas of perceived fit.

Athlete characteristics beyond demographic variables may also enhance perceptions of athlete-event appropriateness. For example, an athlete’s persona, familiarity, or likeability may influence fit with various events. Lindsay Davenport and Serena Williams often vied for the number one ranking in women’s tennis making them both “experts” as endorsers of their sporting events; however, they exhibited very different personas both on and off the court that could influence consumers’ perceived fit at different events (e.g., Wimbledon versus the United States Open).

Taking the match-up hypothesis even one step farther, it would be interesting to note how different consumer characteristics interact with endorser characteristics to influence athlete/event fit. For example, the WNBA has quite a diverse fan base in terms of gender, age, sexual orientation, and income level (Drehs 2001; Sportsfolio WNBA Enthusiast 2005); thus, if targeting different aspects of this fan base (e.g., men versus women), will consumer characteristics interact with endorser characteristics to determine the best fit? While a few researchers have attempted to determine the effect of consumer gender on athlete endorser effectiveness (e.g., Boyd and Shank, 2004; Peetz et al. 2004), certainly other characteristics could impact perceived fit.