4.1. Descriptive Statistics
After sending the survey to a CBA fans’ online community, 400 respondents accepted our survey for 5 days. Among them, 12 were coaches, players, or bureau employees. The other 388 respondents were consumers of CBA games. A total of 15 responses were deleted for the monotonicity (same choice for more than 25 questions) in their answers which made correlation analysis impossible.
Among the 373 surveys, 194 (52%) respondents were male, and 179 (48%) were female. A total of 11 respondents (3%) were less than 18 years old, 223 (60%) were between 18 and 30 years old, 116 (31%) were between 31 and 50 years old, and 23 (6%) were more than 50 years old. In addition, 222 (59%) were not married, 7 (2%) were divorced, and the others were married. Regarding educational background, 336 respondents (90%) had a bachelor’s degree. A total of 245 (66%) of them had a monthly income more than the median income (about 700 dollars using purchasing power parity) in China in 2021 (National Bureau of Statistics of China). An amount of 269 (72%) respondents had their favorite team, and 104 (28%) of them had changed their favorite teams. In total, 299 (80%) respondents watched at least one CBA game on television, while 88 (24%) watched more than ten CBA games each year. Only 20 (5%) of them went to the arena to watch more than ten CBA games each year, and 258 (69%) have never watched any CBA game in the arena.
The respondents active in the fans’ online community were, in general, those who were young with higher degrees and incomes. They can be considered as the most active CBA fans at present and the most likely to stay loyal in the long term, even if most fans now can only watch most of the CBA games from TV broadcasts. The interview of our expert practitioners in the CBA confirmed that basketball fans’ consumption of CBA products (games, peripheric products, magazines, etc.) was much lower than NBA products. CBA games were generally lower in in-arena attendance, TV ratings, and online activity. Almost all teams are underfunded and supported by prominent local companies.
As shown in
Table 1, the score 4 is the most chosen for the importance of CSR and 5 for items concerning team identification and the clubs’ sustainable development. The scores 1 or 2 have been hardly chosen for any question. For almost all questions, positive answers (choice of 4 or 5 points) accounted for more than 60% of the total. Therefore, in concordance with other studies on CSR in professional sports [
58,
64,
69,
90], most consumers have a high level of CSR awareness and gave positive responses to the hypothetical CSR involvement, especially in items of CSR-related team identification. The only exception is the question concerning the purchase intention of peripheric products, where about 48% of respondents chose neutral or negative opinions (score 1, 2, or 3).
In consumer respondents’ interviews, most chose 4 for CSR items’ importance and 5 for team identification related to CSR efforts. The reason is as follows: “we know our team is not profitable. We did not give 5 for CSR’s importance because we did not think our favorite team had many resources to engage in it. However, if some CSR activities were done, they would enhance team identification”. Our interview with expert practitioners in CBA confirmed the lack of motivation for purchasing peripherical products: “in this era of high-speed media and logistics, the fans can easily buy the peripherical product of top clubs in the world, that is, NBA teams. We find it hard to develop peripherical products for only a few hardcore fans motivated to buy them.” This message revealed that most CBA fans were also NBA fans simultaneously. The CBA teams’ peripherical products could, at most, only be the second-best choice for them to show their identity in daily life.
For our H1 and H2, the respondents gave a direct validation in their answers. They expressed their positive responses to the hypothetical CSR efforts. For H1 and H2’s sub-hypotheses, H3 and H4, we used relational data analysis to assess them in the following subsection.
4.2. Relational Data Analysis
4.2.1. Reliability and Validity of Questionnaire for Factorial Analysis
Reliability tests were performed for each theoretical construct of our questionnaire with Cronbach’s alpha values. The results are shown in
Table 2. All constructs are far above the threshold value of 0.70 suggested by Hair JR, Black, Babin, and Anderson [
97]. This means that, considering the variation level of the collected responses, our sample size of 373 is large enough and obtains a consistent outcome for reproducible analysis results [
98].
For the structural validity of this questionnaire, the Bartlett test’s significance is smaller than 0.001, while the Kaiser–Meyer–Olkin score is 0.94, which is well acceptable according to the suggestion of Hair JR, Black, Babin, and Anderson [
97]. These results mean that the size of 373 of our sample is sufficient, and our questionnaire is suitable for factorial analysis.
4.2.2. Exploratory Factor Analysis
Without a predefined number of factors, the SPSS factorized our questionnaire into six constructs (
Table 3). However, the classification of our questionnaire’s items is quite different from the theoretical construct structure of CSR, team identification, and clubs’ sustainable development according to the design of our questionnaire.
Contrary to other factorial analyses in the same field [
64] with perfect correspondence between factorial classification and constructs’ structure in the questionnaire design, the CSR items in our study were not classified just as Carroll’s four-stage system [
32], while the team identification and clubs’ sustainable development were not separated apart from each other.
For the items in our questionnaire, team identification and the clubs’ sustainable development are classified as the same factor (Factor 1). However, under a one-way ANOVA post-hoc test, comparing items of team identification (TI 1~TI 7) and the items of purchase intentions (CSD 1~CSD 3), there were always significant differences between these items (p < 0.01). Ethical and philanthropic responsibility items were heavily loaded into the same factor (Factor 2). In our interview with consumer respondents, we were told they had difficulty explaining the differences between ethical and philanthropic responsibilities. Cultural differences may cause this, e.g., in China, schooling activities and community security are entirely taken charge of by the state. The CBA clubs’ participation in these activities is generally considered as taking ethical responsibilities rather than philanthropic responsibilities.
4.2.3. Correlations between These Theoretical Constructs
With the mean and standard deviation of items’ answers within each theoretical construct, we obtained the correlation between different constructs as shown in
Table 4.
We found that answers were concentrated around 4 for all constructs. Standard deviations were always less than 0.5. For each construct of CSR, the correlation existed only between the neighboring stages, that is, , , and . Only the correlation between TI and the CSD is strong ().
The weak relationship between the different constructs of CSR and CSD made it impossible to verify a moderating role of TI between them (H4-1/-2/-3/-4). In fact, if we make the product of CSR constructs (ER, LR, EtR, and PR) and TI as new factors, the correlations between these new factors and CSD is just between 0.72 and 0.74 (, , , and ), lower than the direct correlation between TI and CS ().
If we take all CSR items as one single factor, the results are shown in
Table 5.
These results showed that the respondents held high expectations for CBA clubs’ CSR efforts and expressed their pride and intention to reward their favorite clubs’ kindness (
Table 3). The correlation remained weak between CSR expectations, the increase in pride, and purchasing intentions related to CSR efforts (
Table 4 and
Table 5). Therefore, we validated H3 of this study, which linked the team identification and the clubs’ sustainable development with a coefficient of correlation higher than 0.8. Nevertheless, the sub-hypothesis of H1-1/-2/-3/-4 and H2-1/-2/-3/-4 could not be validated in our study, for the correlation coefficients are generally lower than 0.4. This weak relationship between CSR and CSD
makes it impossible to verify the moderating effect of TI (H4). In fact,
, lower than the direct correlation between TI and CSD (
).
4.3. Structural Equation Model
To quantify the fitness of our research model, the SEM’s confirmatory factor analysis (CFA) indices are calculated with the maximum likelihood method and are shown in
Table 6.
According to the suggested values given by the work of Jung [
64], and according to the handbook of Hair JR, Black, Babin, and Anderson [
97], we found that the research model did not fit well with the results of our survey. Just as with the EFA step of our study, the inner structure of our collected sample data is different from our theoretical research model.
With structural path estimates, we can obtain the standard estimates of the principal paths as in
Figure 3.
Like the correlation between different constructs, only the regression weight between TI and the CSD is greater than 0.8. Other regression weights of the principal paths are near zero.
To understand the weak relationship between the evaluation of CSR’s importance and the TI and the CSD, we interviewed some consumer respondents about their general attitude toward their favorite clubs’ CSR involvement. We found that passionate fans were aware of the financial situation of their favorite CBA club. Most selected 4 instead of 5 in the CSR items because they considered their favorite clubs’ limited financial resources. However, if the club had engaged in more CSR, they would have been proud and deepened their identification with the club. For the items concerning the clubs’ sustainable development, their weak correlation with the CSR factors means that those who had attached the highest importance to CSR factors did not increase that much in their purchase intention and team loyalty. Expert CBA practitioners felt the same way about CSR involvement as ardent fans. On the one hand, they agree with the social values contained therein; on the other hand, they do not want the CSR activities to become a burden more significant than their club could bear.
4.4. Findings
For our H1 and H2, the respondents have already given a direct validation in their answers. They have generally expressed their positive reaction to the hypothetical CSR efforts. In our H3, the correlation between the TI and the CSD are validated with . However, the items in our questionnaire could not stand structural modelling analysis according to Carroll’s model (1991), probably because of the cultural and institutional differences between China and the US. The correlation between CSR awareness with TI (, regression weights < 0.3 and CSD , regression weights < 0.1) is weak, which means those who unreservedly highlight the importance of CSR efforts are not necessarily those who intend most to increase their support for the club as a reward to their favorite teams’ CSR efforts.
Following prudential principles, our research began with the framework and questionnaires of validated studies [
64,
73] in the developed market environment, but ended up with challenging its applicability to the CPSO market for relational data analysis. The questionnaire turns out to be in need of further adaptation due to the market and regime differences, e.g., the widely used indicator, Tobin’s q, has been found to be without a significant relationship with corporate performance in a study of Iranian manufacturing companies [
99]. The lack of structural validity of the theoretical constructs and the weak correlation between CSR and CSD prevent us from conducting in-depth analyses of the relationships between these theoretical constructs, e.g., as an investigation of moderating the role of internal control between earnings management, related party transactions, and corporate performance [
99].