Hedonic value is generally conceptualized as a person's emotional response to entertainment and experiential consumption [13, 14]. For example, in sports, several studies have found that fans feel positive effects after watching a sports event, such as enjoyment, excitement, and pleasure [11, 15]. In addition, those studies further demonstrated that the positive emotional response significantly affects fans' behaviors, including the intention to attend the game at a stadium, watch the game on television, and merchandise consumption [11, 15]. Similarly, the current study proposes that fans may attend a fan festival for a mega sports event to satisfy hedonic values, such as pleasure and excitement. Therefore, we have defined hedonic value as an overall assessment of positive emotions engendered by attending the fan festival.

Aesthetic value refers to an individual's desire to watch the game to enjoy the artistic beauty of athletes in motion [16]. For instance, fans may be motivated to watch sports that highlight aesthetic values (e.g., gymnastics) to enjoy the athletes' beautiful movements. Furthermore, evidence has indicated that the aesthetic aspect of sports motivates fans to watch the game [16]. It is a significant motivator for team support for women's professional sports [17]. With this in mind, we identified aesthetics as a key value that motivates fans to attend fan festivals for mega sports events.

Sports provide numerous opportunities for fans to interact with other fans [4, 11]. Research has found that when fans watch the game with others, the social experience enhances the enjoyment and helps those fans develop new social relationships with others [4, 5]. In particular, the research found that participation in sports and leisure activities reduces prejudice and alienation between different groups, thereby improving intergroup relations [5, 18]. Consequently, fans build intimate social relationships with others who watch the game together [16]. In this study, we conceptualized social value as a social experience that fans may experience by interacting with other fans through mega sports events. With this in mind, we propose that social value is strongly associated with fan festivals for mega sports events.

According to Schwartz and Bilsky [19], positive social justice values are built into moral systems. Values such as altruism, benevolence, kindness, and morality are important factors that determine individuals' moral actions and behaviors [8, 20, 21]. Social justice values are associated with various social interactions [22]. Sports fans attend fan festivals and interact with friends or strangers while watching the game. Social justice becomes particularly important when people interact with others they have never met before because they must redefine the social values (e.g., fairness and cooperation) associated with those social relations [23]. In this regard, because fan festivals enable fans to interact with others, the social consumption experience would help them obtain values related to social justice, such as fairness and cooperative spirit.

Study 1 consisted of three stages Construct. In the first stage, this study identified key values associated with a fan festival for a mega-sports event and generated an initial pool of items. The initial pool of items was derived from the relevant literature. Reviewing the existing literature resulted in 43 values (e.g., achievement, aesthetic, hedonism, and patriotism) and 78 items. After we eliminated redundant items, the initial pool included 33 items with five values: hedonic, aesthetic, emotional, social, functional, and social justice values [11-17].

In the second stage, we conducted an in-depth interview with 10 participants who had attended the 2010 World Cup Fan Festival and 10 who had attended the 2014 World Cup Fan Festival. Most spectators indicated they had mainly attended the fan festival for hedonic, aesthetic, emotional, and social values. In contrast, they found functional value (i.e., price) relatively unimportant. Thus, we removed the functional value dimension. Furthermore, ten expert panelists—including sports management professors, sports marketers, the World Cup Fan Festival promoters, and several Ph.D. students in sports management—examined 33 identified items together. After a careful review, these panelists suggested removing the emotional value because it may significantly overlap with the hedonic value (e.g., pleasure, enjoyment, and fresh feeling). As a result, this study identified four value dimensions (i.e., hedonic, aesthetic, social, and social justice) associated with the fan festival for the mega sports event. Each value dimension consisted of seven items.

In the third stage, 143 undergraduate students who participated in the 2018 World Cup Fan Festival during the final qualifying round for the Asia region completed the survey for the initial quantitative evaluation of the FFVS. All participants were enrolled in undergraduate- or graduate-level sports management classes at a large private university in Korea. Among participants, 50.3% (n = 72) were male, and the respondents ranged from 19 to 50 years (M = 21). Participants were asked to answer all items on a 5-point Likert-type scale (1 = Strongly Disagree; 5 = Strongly Agree). We tested the reliability of the scale using Cronbach's alpha (> .70) and Cattell's scree test [24]. The results indicated that Cronbach's alpha was .88 for hedonic value, .77 for aesthetic value, .92 for social value, and .89 for social justice value. In addition, the FFVS showed strong item-to-total correlations, ranging from .58 to .88 [25-28]. Furthermore, the average mean of inter-item corrections ranged from .68 for aesthetic value to .89 for social justice value. The item-to-total and mean inter-item correlations are reported in Table 1.

This study aimed to validate and confirm the structure through the confirmatory factor analysis (CFA). Three hundred ninety-nine (n = 399) sports fans who attended the fan festival for the World Cup match between South Korea and Sweden were recruited for and participated in this study. Among participants, 59.4% (n = 237) were male, and the average age was 30 (ranging from 18 to 59). This construct used seven items for hedonic, social, and social justice values dimensions and six for aesthetic value dimensions.

In stage 2, we performed various statistical analyses to examine the reliability and validity of the FFVS. To examine the validity of the FFVS scale, we used structural equation modeling (SEM) and tested the composite reliability and the AVE. The results of Cronbach's alpha ranged from .81 (for aesthetic value) to .95 (for hedonic value), indicating high internal validity. Specifically, the CFA was employed to examine the dimension construction of the first-order factor and second-order factor model of the FFVS. To evaluate the fit of the proposed model of sample 2, we examined the CFA (maximum likelihood method) using AMOS 21.0. We employed the cutoff criteria for fit indices recommended by Hu and Bentler [29-38] in all our model evaluations.

A CFA was conducted to examine the psychometric properties of the measures. Specifically, this study used the indices of the root mean square error of approximation (RMSEA), the standardized root mean square residual (SRMR), and the comparative fit index (CFI). The results indicated that the first-order factor model suffers from low fit (x² = 1884.91; df = 246; x²/df = 7.66; RMSEA = .13; SRMR = .10; CFI = .77) because of some inadequate fit value in the model. This study used the modification indices (MIs) as criteria for eliminating items. Consequently, we deleted three items for the hedonic, aesthetic, and social values. In addition, we deleted four items for the social justice value. After deleting these items, the modified measurement model yielded an acceptable model fit (x² = 196.76; df = 71; x²/df = 2.77; RMSEA = .07; SRMR = .04; CFI = .97). The loading ranged between .63 (AV 5) and .95 (AV 2). All items loading on model 1 were statistically significant (p < .001).

The model 2 CFA (second-order factor model) was completed with the 14 items of the modified FFVS. Model 2 yielded an acceptable model fit (x² = 206.27; df = 73; x²/df = 2.83; RMSEA = .07; SRMR = .05; CFI = .97), and the loading ranged from .63 (AV 5) to .95 (AV 2). The second-order factor models were more economical than the first [30]. In addition, as shown by the chi-square difference test (Δx²), model 1 was also more significant than model 2 (Δx² = 9.51, df = 2, p < .01). All measures loaded significantly on their respective factors.

Model 1 yields a better fit, and by considering the results of CFA, we concluded that Model 1 is more robust than Model 2. Therefore, the comparison between Model 1 and Model 2 is reported in Table 2.

The final factor loadings model is presented in Table 3. Thus, we used Model 1 to test the construct reliability and discriminated validity. To examine the construct reliability of Model 1 (i.e., first-order factor; four-factor model), we measured composite reliability (CR) and Cronbach's alpha value. The CR values were greater than .70 (hedonic value = .94; aesthetic value = .83; social value = .87; social justice value = .87) [32], indicating good overall construct reliability. In addition, the results showed that Cronbach's alpha was greater than .70 for each factor, indicating high levels of internal consistency and reliability (hedonic value =.95; aesthetic value = .81; social value = .85; social justice value = .86). The AVE values were also greater than .50 (hedonic value = .77; aesthetic value = .62; social value = .55; social justice value = .69) [39]. We used a correlation analysis among the value dimensions to test for discriminant validity. The estimated correlations between factors of the FFVS were not excessively high (e.g., < .85) [37]. This showed a reasonably appropriate level of factor correlations among FFVS factors; the squared correlation between those two constructs was also less than the AVE value for each construct. Thus, the validity and reliability of the FFVS were established. The mean and standard deviation scores for each dimension are shown in Table 4.