How do Brand Personality, Identification, and Relationship Length Drive Loyalty in Sports?

Structured Abstract Purpose – This study extends brand identification theory to the sports team context by testing the direct and indirect effects of a sports team’s personality, sports fans’ identification with the team, and the effect of the length of fans’ relationship with a team on their loyalty to it. Design – We conducted a quantitative study among ice hockey fans of one Finnish hockey team before play-off games. Data came from an online questionnaire generating 1,166 responses. Findings – We find that (1) identification with a team mediates the effects of brand personality on attitudinal loyalty and behavioral loyalty, (2) brand personality is a stronger driver of identification among newer fans, and (3) brand personality has a stronger influence on both attitudinal and behavioral loyalty among newer fans. These findings stress the importance of sports brand’s personality in driving fans’ identification with the team and their loyalty to it. Originality – The study develops and tests a new conceptual model on consumer loyalty in the sports team context. We shed light on how sports team personality affects its fans’ identification with the team and the formation of fan loyalty, from the perspective of fans’ relationship length.


Introduction
Branding and customer relationship management (CRM), along with the commercialization and professionalization of the sports industry, are factors central to the success of sports teams and their brands (Bauer et al., 2008;Ferrand et al., 2010). Recent research has shed ample light on how brand attributes and consumers' own identities affect buying decisions and loyalty (Aaker, 1997;Lam et al., 2013), which in turn lead to steadier spectator volumes and greater attention in the media from both advertisers and sponsors (Gladden and Funk, 2002). These positive effects of branding and identification are even more important during economic downturns. Although consumers' general interest in sports remained high during the last global recession (PricewaterhouseCoopers, 2011), the economic downturn had a significant effect on event ticket sales, resulting in reduced revenues for professional teams (Plunkett Research, 2013). The sport market industry, however, is still estimated to grow at an annual rate of 3.7 % and is expected to rise to $145 billion by 2015 (PricewaterhouseCoopers, 2011). These enormous revenues underscore the importance of understanding the factors that affect fans' motivation to attend games and remain loyal to a sports brand or event.
Prior research, however, falls short in two respects. Although previous research has highlighted the importance of brand personality and identification with a sports brand, detailed understanding is still lacking on (1) the role of identification between brand personality and loyalty and (2) the impact of relationship length on sports-consuming behavior (Kim and Trail, 2010;Mittal and Katrichis, 2000).
First, because growing evidence stresses the importance of brand identification in the quest for effective CRM (e.g., Malär et al., 2012;McDonald, 2010), this study examines the role of brand identification as a means of transforming sports consumers' brand affects into brand loyalty. Recent research links consumers' identification with a sports team to the teams' and sports brands' ability to exhibit unique and appealing personality traits that connect with consumers' self-identity (Carlson et al., 2009). This brand-self connection is essentially an emotionally charged construct that, therefore, is closely linked with fans' emotional connection with the team (see Bruner et al., 2014;De Backer et al., 2011), creating the basis for strong team-fan relationships. Previous research has also connected brand personality with behavioral outcomes (Funk and James, 2006;Wu et al., 2012). The factors that drive identification with a sports team are strong predictors of sports fans' consumption behavior (Decrop and Derbaix, 2010;Trail et al., 2000), such as continuing to participate in the team's activities even after long winless streaks (Bodet and Bernache-Assollant, 2011).
Although such loyalty consists of both attitudinal and behavioral aspects (Oliver, 1999), prior research has not specifically examined how identification mediates the effects of brand personality on both attitudinal and behavioral facets of loyalty.
To address the second limitation identified in prior research, our study tests the effects of brand personality on loyalty in the light of fan-team relationship length. Although McDonald et al. (2014) show that the tenure of a fan-team relationship is a strong predictor of relationship continuance, little evidence exists on its moderating effects on the relationships among brand personality, identification, and loyalty. Other studies have also highlighted the need for further research, suggesting that attitudinal and behavioral aspects of loyalty may not change congruently with relationship length (Funk and James, 2006;McDonald, 2010). Prior evidence also indicates that the effectiveness of brand personality decreases over the duration of the customer-brand relationship (Reimann et al., 2011;Richins and Bloch, 1986) and that brand identification and its antecedents may change over time in a non-uniform manner (Lam et al., 2013).
We aim to address these limitations in this study. We propose and test a conceptual framework that focuses on three key antecedents of loyalty: (1) sports brand personality, (2) identification with the team, and (3) relationship length. We test the conceptual model using online survey data collected from 1,166 Finnish ice hockey fans.
We begin by briefly reviewing the literature on brand personality, identification with a team, and customer loyalty. We then address how fan-team relationship length affects these constructs. Next, we describe the survey research methodology and present our findings. We then discuss our findings in light of the study hypotheses and explain how the findings could influence managerial practice. Finally, we present the study's limitations and suggest avenues for further research.

Brand personality and identification as sources of sports team loyalty
Brand personality creates the basis for the relationship between a consumer and an organization (Jafarnejad et al., 2012). It consists of human characteristics that can be connected with a specific brand (Aaker, 1997). Brand personality represents an instrumental driver of identification (Kuenzel and Halliday, 2010), which enables companies to improve consumers' preferences, attitudes, purchase intentions and brand loyalty (Mengxia, 2007). Tsiotsou (2012, p. 248) states that sports team brand personality is a "combination of the image of its administration and personnel, the image of the team (players and coaches on and off the field), and the image of its fans." A sports team's fans often have no difficulty describing the personality of the team and can do so almost as easily as they would describe the personality of someone familiar to them (Davis, 2012, p. 122).
According to prior research (Ross et al., 2008), Aaker's (1997) brand personality instrument may produce inadequately generalizable results in research contexts with distinctive features. Therefore, Tsiotsou (2012) introduced a brand personality instrument that better captures the distinctive features of sports teams' personalities; we thus apply this instrument in our study. The instrument consists of five personality traits of sports brands: competitiveness, prestige, morality, authenticity, and credibility. Competitiveness reflects consumers' perceptions of a team's ability to defeat competitors and achieve its goals. It consists of character descriptions such as proud, ambitious, dynamic, successful, winning, and triumphant. Prestige reflects a team's superiority and its general recognition in terms of achievements. This dimension is characterized by attributes such as being multitudinous, glorious, great, strong, and honorable. Morality reflects consumers' perceptions of a team's code of conduct. Character traits associated with it include being principled, cultural/cultivated, and ethical. Authenticity captures a team's uniqueness with traits such as being traditional, uncompromising, and radical. Credibility expresses that a team resonates with confidence and self-assurance and comprises traits such as being wealthy and influential.
Identification with a sports team is defined in terms of people's perceived sense of belonging to the team, and that sense of belonging is integral to how they define themselves (Ashforth and Mael, 1989). Identification represents a predisposition on the part of sports consumers to develop an emotional attachment to a team and demonstrate support for it (Bruner et al., 2014;De Backer et al., 2011;Donavan et al., 2005) and addresses behavior directed to a sports event and repurchase behavior (Amiot et al., 2013;Bee and Havitz, 2010). For a strongly identified person, being a fan is closely linked to self-identity and its congruence with the brand personality (Fetchko et al., 2013, p. 42). In this study, we examine the influence of sports brand personality (Tsiotsou, 2012) on loyalty in sports (Bauer et al., 2008), which is mediated by identification with the sports brand (Bhattacharya and Sen, 2003). The outcome variable in this study, namely loyalty, is a result of the interaction between an individual's internal sense of affection and external negative influences (Heere and Dickson, 2008). Dick and Basu (1994) contend that loyalty is the strength of the relationship between an individual's relative attitude and repurchase behavior. In the sports marketing context, behavioral loyalty has dominated both practice-oriented research (Plunkett Research, 2013;PricewaterhouseCoopers, 2011) and academic research (Bodet and Bernache-Assollant, 2011;Wu et al., 2012). However, Bauer et al. (2008) show that when examining loyalty to sports brands, the focus should be on sports consumers' beliefs, selfimages, and sense of belonging to a community. Therefore, a truly loyal customer is one who regularly purchases the brand's products as well as shows a strong attitudinal inclination to the brand (Caruana, 2002;Kaynak et al., 2008). For our purposes, we conceptualize attitudinal loyalty to represent fans' psychological commitment and their internal sense of attachment to a team (Bauer et al., 2008). Behavioral loyalty in a sports marketing context refers to an actual behavior directed toward a sports event (Bee and Havitz, 2010), which ranges from attending games and events (e.g., buying tickets, watching a team's games on television; Kaynak et al., 2008) to committing to engage in repurchase behavior (Oliver, 1999). We use the term "brand loyalty" interchangeably with "customer loyalty" (Funk and Pastore, 2000).
Research on relationships proposes that relationship length has an overall impact on sports-consuming behavior (Mittal and Katrichis, 2000). It is associated with the commitment and confidence of each partner (Spake and Megehee, 2010), consumers' identification with a sports team (Bhattacharya et al., 1995), and loyalty (Athanasopoulou, 2009;Wann et al., 2001, p. 62). Thus, relationship length is an integral part of any study dealing with relationships (Athanasopoulou, 2009). The studies of Lam et al. (2013), Reimann et al. (2011), and Richins and Bloch (1986) suggest that the importance of brand personality is greater in new consumer-brand relationships and diminishes as the relationships mature.
However, research on the effects of relationship length on the antecedents of loyalty is not concordant, though the topic has not been widely studied, especially in the context of sports. Figure 1 presents the conceptual framework outlined in the study. The model proposes that identification with a team mediates the effects of brand personality on loyalty. Such mediation occurs when (1) brand personality has a positive effect on identification with a team and loyalty, (2) identification is positively associated with loyalty, and (3) in the presence of identification, the relationship between brand personality and loyalty weakens (Baron and Kenny, 1986). The model also suggests that the duration of the fan-team relationship has a moderating effect on the relationship between brand personality and loyalty. In the next section, we discuss the rationale of the model and the development of our hypotheses.

Research hypotheses and control variables
"Take in Figure 1 about here"

Effect of brand personality on customer loyalty
Consumers are more likely to identify with brands with distinctive and strong personalities (Bhattacharya and Sen, 2003). Previous research has also shown that consumers have an ideal self-image and strive to consume brands that will boost their self-esteem (Carlson et al., 2009). Furthermore, the stronger the link between brand personality and a consumer's self-image, the greater is the likelihood that the consumer will possess a meaningful attitude toward the brand and feel an emotional attachment to it (Bodet and Bernache-Assollant, 2011;Kuenzel and Halliday, 2010;Wu et al., 2012), leading to increased brand loyalty (Cheng et al., 2014;Huang et al., 2014;Malhotra, 1988). Evidence also indicates that brand personality directly influences attitudinal (Buil et al., 2013) and behavioral (Brakus et al., 2009) aspects of loyalty.
Sports teams are even more strongly linked to their brand personalities than other brands (Smith et al., 2006) because they are complex brands with unstable and intangible core products that can be described as emotionally charged, experiential, hedonic services. In addition, the interaction between a consumer and a sports team is a dynamic process that originates from social meaning and an individual's identity. Research has found that the fans who perceive close connections with their self-identity and the team's identity are more loyal to their team than fans with lesser self-team identity congruence (Stevens and Rosenberger, 2012). In support of this, other evidence shows that brand identity, consumers' self-identity, and consumers' likelihood to form positive attitudes toward the brand are positively related (Sirgy et al., 2008) and that brand identification conveys and enhances the transfer of brand effects from brand personality to attitudinal brand loyalty (He et al., 2012;Iglesias et al., 2011;Sirgy et al., 2008). Therefore, we expect a positive relationship between a sports team's personality and fans' attitudinal loyalty to the team, mediated by fans' identification with the team. Thus: H1a: Identification with the team mediates the effects of brand personality on attitudinal loyalty. Matsuoka et al. (2003) and Theodorakis et al. (2009) outline the direct relationship between fans' identification with a team and their intention to view the team's future games. Therefore, the most common means of improving behavioral loyalty relates to investing more in customer relationships and maintaining a distinguishable and appealing brand personality that bolsters a fan's identification with the team (Stevens and Rosenberger, 2012). Identified fans display a high degree of behavioral loyalty through their regular participation in the team's events and other activities and purchases of the team's licensed products. Therefore, a sports team's identity, fans' identification with the team, and their behavioral loyalty are all positively connected (Bauer et al., 2008;Wann et al., 2001, p. 59). He et al. (2012) and Carlson and Donavan (2013) further show that brand identification conveys the effects of brand personality on behavioral brand loyalty. Therefore, fans' identification with a sports team is a focal construct operating between the brand personality of a sports team and fans' behavioral loyalty to the team. Thus: H1b: Identification with the team mediates the effects of brand personality on behavioral loyalty.

The moderating role of relationship length
Relationship length has an overall effect on all relationships and a positive direct effect on several relational constructs, such as brand identification (Reimann et al., 2011), satisfaction, trust, and commitment (Verhoef et al., 2002). However, evidence is less consistent on how relationship duration moderates the associations between these relational constructs. For example, Bolton (1998) shows that relationship duration positively affects the relationship between satisfaction and retention, whereas Raimondo et al. (2008) find that relationship duration has a negative effect on the satisfaction-loyalty relationship. Ranaweera and Menon (2011) demonstrate a negative effect on the relationship between satisfaction and word of mouth, and Bartikowsky et al. (2011) show a similar effect on the reputation-loyalty relationship. Other evidence indicates that relationship length has little, if any, influence on the relationships between perceived value and customer loyalty or trust and purchase intention (Karjaluoto et al., 2012;Raimondo et al., 2008;Verhoef et al., 2002).
Recent studies have shown that in longer relationships, consumers' preference for closeness is weaker (Mende et al., 2013) and the effectiveness of brand personality decreases along with relationship length (Reimann et al., 2011). Levin et al. (2006) explain this by proposing that that all relationships are bound to change over time, along with the change in how individuals process the relationship-related information. Prior research suggests that in the beginning of relationships, consumers are more inclined to view the brand positively and show their commitment to it than later in the relationship (Ranaweera and Menon, 2011). This may be due to the company's efforts to project itself in a positive manner as well as consumers' ego-centric reasons to justify the made choice. In addition, in new relationships individuals use cognitive cues available to them to categorize another party into existing groups or stereotypes (Kim et al., 2004). In these occasions, consumers' perceptions of and attitudes toward a brand are formed from information about the brand's and other customers' demographical characteristics, because actual behavior-or experience-based information is not available (see Meyerson et al., 1996). Therefore, in new consumer-brand relationships, consumers are likely to emphasize more brand image-related information that helps them characterize the brand, thus enabling the evaluation of its congruence with one's own positive self-image and the formation of brand attitude. However, as the relationships mature and consumers accumulate actual experience-based information through several episodes with the brand and other customers using the brand, they are able to form individual schemas about the brand, which are less connected with the brand's demographics or personality (see Levin et al., 2006). This prior evidence brings us to propose that the brand personality-brand loyalty relationship is moderated by relationship length. We expect the associations among brand personality, brand identification, and brand loyalty to be stronger in newly formed customer relationships than in older ones (e.g., Lam et al., 2013, Reimann et al., 2011, Richins and Bloch, 1986. In newly formed relationships, brands can convey effective positive emotions to consumers; however, this effect wears off as the relationship matures. Funk and James (2006) and Mende et al. (2013) further stress that the effect of relationship length may differ between product and service contexts and types of relationships, highlighting the need for the present study.
Taken together, we expect relationship length to moderate the relationships between brand personality and identification and between brand personality and loyalty in the sports context, such that in newer fan-team relationships the effects of brand personality are stronger. On this basis, we propose the following hypotheses: H2a: The direct effect of brand personality on identification is stronger in newer fan relationships than in older fan relationships.
H2b: The total influence of brand personality on attitudinal loyalty is stronger in newer fan relationships than in older fan relationships.
H2c: The total influence of brand personality on behavioral loyalty is stronger in newer fan relationships than in older fan relationships.

Control Variables
We controlled for age and season ticket ownership, which could possibly affect the outcome constructs in our model. Fan age is an important variable in the sports industry, as the industry's attention is mostly focused on fans between the ages of 18 and 34, even though approximately 60% of fans are age 35 or older (Luker, 2012). According to an ESPN Sports Poll (Luker, 2012), interest in sports among older age groups is constantly growing, and they also are willing to spend more money on sports. Season ticket ownership is generally related to fans' activity in terms of attending games. Fans with season tickets are typically considered the most valuable to an organization (McDonald, 2010).

Data collection
We designed a questionnaire that included measures for the model constructs, two battling, team commitment, winning team, dynamic, middle-Finnish (in terms of geography).
The majority of the fan base comes from the region near its home town and central Finland.
The team has approximately 36,000 Facebook fans, which to some extent reflects the number of the total active fan base. Ice hockey is the most popular spectator sport in Finland, and its fan base is large compared with other competitive sports, a factor justifying an analysis of that fan base.
In two weeks, the questionnaire was completed 1,166 times. The total number of visitors to the questionnaire was 1,977, which resulted in an effective response rate of 59%.
We tested possible non-response bias by comparing the first 25% of respondents with the last 25% with respect to the study constructs and background variables. We found no significant differences.
We used validated scales to measure the model constructs. The items were first translated from the original language (English) to Finnish. We tested the accuracy of translation with back translation, comparing the differences between the original source items and items that were translated back into English. No notable differences emerged, so we deemed the translation accurate. To minimize common method bias, the items within the questionnaire were mixed and respondents' identities kept confidential. In addition, Harman's (1967) one-factor test showed the presence of measurement model factors rather than a general factor and that the largest factor did not account for a majority of variance (12.7%).
Following Podsakoff et al. (2003), we ran a partial least squares (PLS) model with a common method factor with indicators that included all the principal constructs, and we calculated each indicator's variance as substantively explained by the principal construct. This analysis shows that average variance explained variance of the indicators (0.617), while the average method-based variance was 0.013. Given the magnitude of method variance, common method bias is unlikely to be of serious concern in this study.
The questionnaire was screened by four university professors and five fans of the team before publication, which resulted in minor modifications to the wording of some items. A seven-point Likert scale (1 = strongly disagree, 7 = strongly agree) served as a basic measurement scale for the model constructs. The survey comprised 50 questions: 24 measured the team's personality constructs, seven measured identification with the team, six measured attitudinal loyalty, and four measured behavioral loyalty (see Appendix for the items).

Measurement
To measure team personality, the study employed the instrument of sports team personality developed by Tsiotsou (2012). Respondents were asked to answer according to the image they had in their minds about the team at that moment. We measured identification with the team with the Sport Spectator Identification Scale developed by Wann and Branscombe (1993). The scale for measuring attitudinal loyalty was based on the Attitudinal Loyalty to Team Scale (Heere and Dickson, 2008). To measure behavioral loyalty, we applied the instrument suggested by Bauer et al. (2008). However, as Bauer et al.'s (2008) scale contains items closely related to our identification measure, such as following the team in the media and purchasing club-related merchandise, we excluded these items from our behavioral loyalty measure to ensure discriminant validity.
Regarding the moderator variable, we measured relationship length in line with Raimondo et al. (2008) with the question, "How long have you been supporting the team?" on a four-point scale (1 = less than two years, 2 = 2 to 5 years, 3 = 6 to 9 years, 4 = 10 years or more). For the purposes of the study, we recoded relationship length into a dichotomous variable. The groups contained respondents who reported being fans for five years or less (responses "1" and "2") and more than 6 years (responses "3" and "4"). We measured age on a seven-point scale (1 = below 18, 7 = 65 or over). We assessed season ticket ownership with a two-item scale (1 = does not hold a season ticket, 2 = holds a season ticket).
The majority of respondents were men (73%); this is typical in a sports context, in which men usually outnumber women in terms of attendance at games and events (Theodorakis et al., 2012). A study by Statistics Finland (2014) shows that 50% of Finnish men had participated in a sport event within the past 12 months while women's participation ratio was only 34%. A majority of the respondents (74.5%) had been supporting the team for more than 10 years. The descriptors for the control variables were as follows: The two largest age groups were 26 to 35 (32%) and 18 to 25 (24%). Around one-fifth (19%) of the respondents were between 36 and 45 years of age. Close to 20% (18.6%) held season tickets.
Of the season ticket holders, almost half (45.2%) had held tickets for two to five years, and around one-third (36.8%) had held them for more than five years. In terms of background variables, the sample is in line with the team's own research of its fan base.

Results
To test our hypotheses, we analyzed the data using SmartPLS (Ringle et al., 2005). All the constructs in the model are reflective in nature. Exploratory factor analysis of the items confirmed the unidimensionality of all the constructs and revealed five distinct factors for brand personality and two loyalty dimensions. Furthermore, the analysis showed that the items associated with identification formed their own factor and did not demonstrate high cross-loadings with the loyalty items. We used composite measures for the five dimensions measuring brand personality so as to reduce the number of items in the next stage of the analysis (see Brakus et al., 2009). In the confirmatory phase, we ran the PLS algorithm to test for validity of the measurement model. The factor loadings were high (≥ 0.611). The internal reliability of each scale is significant, as all constructs presented high composite reliabilities (≥ 0.854) and Cronbach's alpha demonstrated reliability equal to or greater than 0.774, well above the recommended value of 0.70 (Nunnally, 1978). Fornell and Larcker's (1981) criterion, which is based on the premise that a latent variable should better explain variance of its own indicators than variance of other latent variables, offered strong support for discriminant validity. Table 1 shows the crosscorrelation matrix in which the square root of the average variance extracted is compared with correlations between the latent variable and all other latent constructs.

Analysis of
"Please insert Table 1 about here"

Direct effects
We first examined direct effects, followed by indirect and multi-group moderating effects. To assess the significance of each estimated path, we applied a standard bootstrapping procedure with 5,000 re-samples consisting of the same number of cases as in the original sample (Hair et al., 2013, p. 132). Table 2 presents the results of the PLS estimation for the direct effects. The R 2 values of the model are moderate or large. The model's predictive relevance (Q 2 ) of the endogenous constructs is medium to large (0.15 ≤ Q 2 ≤ 0.35), providing support for the model's predictive relevance.
"Take in Table 2 about here" The direct effects (Table 2) show that brand personality is strongly associated with identification (β = 0.556, p < 0.01) and its f 2 effect size is large (f 2 > 0.35) in producing the R 2 of identification. The q 2 effect size also indicates that brand personality has a large role in producing the Q 2 predictive relevance for identification. The findings also show that identification with a team is strongly associated with attitudinal loyalty (β = 0.698, p < 0.01).

The mediating role of identification with a team
To assess H1 regarding the mediating effect of identification, we used the bootstrapping method to test mediation (Preacher and Hayes, 2008). This test is superior to Baron and Kenny's (1986) approach or the Sobel test and is perfectly suited for PLS-SEM (Hair et al., 2013, p. 223) because PLS-SEM applies nonparametric bootstrapping and makes no assumptions about the distribution of the data or the sampling distribution of the statistics (Hair et al., 2013, p. 223). Thus, we tested the significance of the indirect effects and assessed the strength of the mediation by bootstrapping the sampling distribution (5,000 bootstrap samples, no sign changes) and by calculating the variance accounted for (VAF) value (Hair et al., 2013, pp. 223-229;Preacher and Hayes, 2008). Table 3 reports the results of this analysis.
"Take in Table 3 about here" The size of the indirect effect is significant (p < 0.01). The strength of the mediation shows that the effect of brand personality on attitudinal loyalty is partially mediated (VAF = 0.665) by identification, in support of H1a. In turn, the effect of brand personality on behavioral loyalty is fully mediated by identification (VAF = 0.892), in support of H1b.

The moderating role of relationship length
In testing H2a-H2c, we conducted a multi-group moderation test. The data confirm H2a, which proposes that the direct effect of brand personality on identification is stronger in newer fan relationships. As Table 2 shows, the effect is stronger in short fan relationships (β = 0.711, p < 0.01) than in long fan relationships (β = 0.501, p < 0.01). We also find support for H2b and H2c, which test the total influence of brand personality on loyalty in the light of short and long fan relationships. Table 3 shows that the total effect of brand personality on attitudinal loyalty is significantly stronger in short relationships (0.685) than in longer relationships (0.487). Similarly, the total effect of brand personality on behavioral loyalty is stronger in short fan relationships (0.433) than in long fan relationships (0.254).

Research issues
Winning and team performance are no longer the most important factors determining consumer attraction and behavior (Wu et al., 2012). For this reason, it is vital to examine factors beyond success to help sports brand managers adopt brand management strategies that will secure steady and long-term sales revenue (Kaynak et al., 2008). A differentiated and appealing brand personality is a necessity for a sports organization to achieve its financial goals (Kuenzel and Halliday, 2010). However, brand management and customer retention in the sports context possess special features (Stevens and Rosenberger, 2012) which, on many levels, lack empirically established knowledge. Additional evidence is especially required on how brand personality contributes to the extent to which consumers identify with the brand and become loyal to the sports team. Similarly, evidence is scant on how the length of fans' relationship with a sports team moderates the relationship between brand personality and loyalty. Against this background, this study constructs and tests a new conceptual model on consumer loyalty in the sports team context. It shows the exact effects of sports team personality on fans' identification with the team and on fan attitudinal and behavioral loyalty, in the light of fans' relationship length.
This study is among the first to examine the moderating effects of relationship length on the relationship between brand personality and team loyalty (Heere and Dickson, 2008). In addition, this study adds to existing knowledge by linking sports team personality (Tsiotsou, 2012) with the extent to which fans identify with the team (Wann and Branscombe, 1993).
The study found support for all five proposed hypotheses (see Table 4). Table 4 about here" First, the data provide support for the hypotheses dealing with the mediating role of identification between brand personality and attitudinal loyalty (H1a) and brand personality and behavioral loyalty (H1b). Our results are consistent with previous literature that links brand personality with identification (e.g., Carlson et al., 2009;Kuenzel and Halliday, 2010), identification with attitudinal loyalty (e.g., Bodet and Bernache-Assollant, 2011;Matsuoka et al., 2003), identification with behavioral loyalty (Laverie and Arnett, 2000), and brand personality with loyalty (Buil et al., 2013). The findings show that identification is more strongly linked with an attitudinal type of loyalty than behavioral loyalty (see Bruner et al., 2014;De Backer et al., 2011). The strong connection between these two emotion-based constructs (identification and attitudinal loyalty) is well suited with the view that sports teams are emotionally charged, hedonic brands. This leads us to purport that strong self-brand identification plays a key role in the formation of "truly loyal" fans-that is, fans with strong emotional attachment to the team (vs. those showing only behavioral loyalty). These fans are the most likely to regularly purchase the team's products and participate in the team's activities even when the team faces hard times, such as losing streaks or economic downturns (see Bauer et al., 2008;Caruana, 2002;Kaynak et al., 2008).

"Take in
The result that brand personality is strongly linked to identification suggests that a team possessing unique personality traits influences fans' identification with the team in a positive way (Carlson et al., 2009). Although previous evidence suggests that identification is a strong construct in explaining fans' participation in team-related activities (Stevens and Rosenberger, 2012), we add to this knowledge by showing that identification with a team is a central mediator in transferring the effects of brand personality to the two types of loyalty.
Specifically, we show that identification is a stronger mediator (full mediation) for the effects of brand personality on behavioral loyalty than for the effects of brand personality on attitudinal loyalty (partial mediation). This highlights that a sports team's brand personality affects attitudinal and behavioral aspects of loyalty differently. Brand personality affects attitudinal loyalty both directly and indirectly through identification, whereas the effect on behavioral loyalty goes only through identification.
Second, our results provide support for all three moderating effects hypotheses. They show that in newer fan-team relationships, brand personality has a stronger effect on identification and that this effect on both attitudinal and behavioral loyalty is also stronger.
This finding is similar to that of Reimann et al. (2011), Richins and Bloch (1986), and Lam et al. (2013). The rationale behind this is that in newly formed relationships, brands can convey positive emotions to consumers, which wears off as the relationships mature. This suggests that because new fans lack actual experience-based information about the team, they use other available cues to form perceptions of and attitudes toward the team and other fans (Kim et al., 2004;Meyerson et al., 1996). In technology context, Venkatesh, Thong and Xu (2012, p. 163) confirm this by stating that "as experience increases, the attractiveness of the novelty that contributes to the effect of hedonic motivation on technology use will diminish and consumers will use the technology for more pragmatic purposes, such as gains in efficiency or effectiveness." In longer relationships, factors other than brand personality, such as habits, inertia and subjective norms, are more likely to explain identification with the team and loyalty.
Finally, we show that the covariates age and season ticket ownership are negatively related to attitudinal loyalty, age is negatively related to behavioral loyalty, and season ticket ownership is positively related to behavioral loyalty. These findings indicate, in line with our assumptions (see also Luker, 2012), that age still plays a role in loyalty; in our study, younger respondents scored higher on both attitudinal and behavioral loyalty. We expected season ticket ownership to positively affect both loyalty dimensions, but the results reveal that the opposite is true for attitudinal loyalty, implying that season ticket holders have weaker attitudinal loyalty than those who buy single tickets for games. Although the path coefficient was significant, the association between the two was relatively weak (-0.129). This rather unexpected finding might be explained by routine behavior: attending games becomes a habit for season ticket holders, which has an impact on how they rate their emotional aspects of loyalty. Another explanation (provided by the team manager) is that half the season tickets of the team are owned by local companies, which circulate the tickets among their employees.
In this situation, the employees often visit games for reasons other than just the team, such as seeing friends and meeting colleagues and customers.

Implications for managers
We identified three important managerial implications from our findings. First, to differentiate brands from their competitors, managers should acknowledge their own brands' features and the strengths of those features and then shape them in a meaningful and targetoriented way. This also applies to sports marketing and, specifically, sports team management. Recognizing a team's personality traits in the same way as fans perceive them would offer managers a mechanism with which to differentiate the team from competitors. In the eyes of fans, a meaningful and consistent team personality offers an opportunity to create a closer relationship with the team. If the personality traits are meaningful and appear to strongly influence identification with the team, it is reasonable to continue emphasizing this personality. If the personality is inappropriate, it is possible to mold it into a more appropriate form through marketing activities.
In this study, the team's three strongest personality traits in the eyes of identified fans were competitiveness (M = 5.72), authenticity (M = 5.58), and morality (M = 5.48). The trait competitiveness is also one of the five personality traits of the JYP team, mentioned in their marketing materials as a "winning team." This emphasizes the importance of a "winning team" image in branding, and along with promoting team authenticity and morality, it can form a good basis for the team's marketing plan. By emphasizing brand personality in the sports team's marketing communications, managers could boost the extent to which fans identify with the team. At the same time, the communication involved would clarify the consistent and meaningful team image in the eyes of fans and lead to stronger customer loyalty. Marketing activities that project a strong and uniform brand personality constitute an effective means of CRM because individuals tend to attach themselves more tenaciously to brands with a strong and meaningful brand personality (Carlson et al., 2009).
Second, our findings suggest that the relationship between identification and customer loyalty is strong. We found that identified fans more often participated in the team's activities and their attitudes toward the relationship were stronger and more positive than those of less engaged fans, which foretells sustained future revenue flows. The managerial implication of this finding is that it is vital to support processes that facilitate fans' identification with the team to ensure the formation of loyalty. Therefore, sports teams should invest in communicating a brand image that corresponds to their fans' self-or ideal self-images. The findings also show that identification has a strong positive effect on attitudinal type of loyalty. Because this type of emotionally charged loyalty forms the strongest premises for lasting consumer-brand relationships (see Bauer et al., 2008;Caruana, 2002;Kaynak et al., 2008), our study urges sports teams to emphasize communication and activities that enhance the fans' emotional connection with the team.
Third, our findings regarding the moderating role of relationship length are important for managers, in that we show that brand personality is a strong driver of both identification and loyalty for newer fans. This implies that identification and loyalty in longer fan relationships are more effectively retained and deepened by factors other than the team's brand image, whereas among newer fans, the team image plays a strong role in the formation of loyalty. Therefore, sports teams' managers need to be aware of these differences between new and existing loyal fans and account for them in marketing communication by differentiating messages to new fans from messages to fans with longer relationships with the team.

Limitations and further research
This research contributes to the understanding of brand identification and brand loyalty formation in the sports context, but it also has limitations, which offer avenues for further research. The first limitation pertains to external validity. Although the sample in this study was representative of the fan base of a typical Finnish ice hockey team, considerations about the team's personality and brand identification were team related. The second limitation relates to other factors that may influence the model. Although we controlled for differences in age and season ticket ownership in the model, other factors, such as the economic situation in the relevant market, may also have an impact on outcome variables. Third, the nonexperimental research design limits claims of causality. Fourth, most of the fans reported being in the relationship with the team for more than 10 years and only a few for fewer than two years, which limits the contribution of the effect of relationship age on the conceptual model. Finally, although we minimized the likelihood of common method bias and showed that it was not present in the data, such bias can only be ruled out in full by using varied data sources or a longitudinal study design.
Future research could extend the model with additional antecedents to brand loyalty, such as by investigating the effects of a team's social media presence and its impact on fan loyalty and examining the effects of team's success on the model. Another worthy area of study would be to examine further how season ticket ownership affects the direct and indirect relationships in the model. Research could compare season ticket ownership in terms of a firm's season tickets versus personally acquired season tickets. Research should also replicate the study with other sports teams in other fields and in other markets and then test the model using moderators such as team success and economic situation, income, and, perhaps, respondent personality.    The team is imbued with acceptable rules of conduct. n/a 5.77 1.01 The team develops and promotes the growth of culture (e.g., educational, art and science programs). n/a 4.57 1.39

List of Figures
The team behaves in accordance with standards for right practice.
n/a 5.67 1.08 Authenticity 0.879 5.58 1.22 The team is unique. n/a 5.80 1.40 The team is faithful to internal rather than external ideas, and retains its spirit and character despite external forces. n/a 5.80 1.07 The team has been around for many years and handed down legends (e.g., players) and customs (e.g., team anthem).
n/a 5.55 1.32 The team adheres to principal positions. n/a 5.61 1.00 The team is open to changes and new ideas. n/a 5.12 1.30 Credibility 0.811 4.94 1.33 The team is trustworthy and reliable.
n/a 5.77 1.00 The team is rich and financially independent.
n/a 4.65 1.78 The team has the capacity and power to exert effects on others and their decisions (e.g., on the administration of the national league).