The Motivation Scale for Sport Consumption: Assessment of the Scale's Psychometric Properties.

Trail, Galen T. ; James, Jeffrey D.

Galen T. Trial Iowa State University

Jeffrey D. James University of Illinois [a]

The purpose of this study was to develop the Motivation Scale for Sport Consumption (MSSC) to measure the motivations behind sport spectator consumption behavior. Previous efforts to develop scales to measure spectator motives have demonstrated weaknesses in content, criterion, and construct validity. Accordingly the content validity criterion validity construct validity and internal consistency of the MSSC were examined to determine whether the instrument accurately and appropriately measures the motivations of sport spectators. The weaknesses of previous scales are presented and then the psychometric properties of the MSSC are assessed. The results indicate that the MSSC does possess the psychometric properties requisite for accurately and reliably measuring the motivations of sport spectator consumption behavior.

Watching sporting events has a long tradition dating back to the first Olympics in 776 BC. Sport spectating also represents a predominant form of leisure behavior in contemporary society. In 1996, attendance at professional baseball, basketball, football and hockey games in America exceeded 110 million, generating over $2.74 billion in gate receipts alone (Badenhausen, Nikolov, Alkin, & Ozanian, 1997). America's largest corporations have also recognized the interest people have in watching and following sports. Sports represent a unique advertising vehicle through which companies can deliver a message to a specific target market. Consumer interest in sports has driven rights fees for sporting events up dramatically in recent years. For example, sports programming on national and regional broadcasts generated approximately $3.5 billion in advertising revenue in 1995, which is about 10% of television's annual sales ("TV Sports," 1996).

Despite the prominence of sport, little is known about the motives of individuals who are willing to invest financial, emotional, and temporal resources in following and watching sports. Individuals who invest their leisure time following sports may be thought of as sport spectators. Sport spectators may be further categorized along a continuum ranging from mere observers of a sporting event to highly committed fans (Sloan, 1989). Pooley (1978) suggested that observers watch a sporting event and then quickly forget what they have seen. Highly committed fans, however, continue their interest in the event or a team to the point that parts of every day are devoted to either the team or the sport itself. Observers enjoy the entertainment of a sporting event, while sport is an important part of the highly committed fan's life.

Previous research on sport consumption has focused primarily on the topic of sport demand. Studies have examined the effect of economic factors, promotions, and residual preference factors (e.g. scheduling of games, new stadia, accessibility, weather) on attendance at sporting events, and have studied the relationship between sociodemographic variables and watching sports (Baade & Tiehen, 1990; Greenstein & Marcum, 1981; Hansen & Gauthier, 1989; Schofield, 1983; Zhang, Pease, Hui, & Michaud, 1995; Zhang, Smith, Pease & Jambor, 1997).

Demand-based research contributes to an understanding of short-term, variable factors that influence decisions to attend or watch sporting events. Distinguishing between individuals who merely enjoy watching an event and those who think of sport as an important part of their life requires understanding the psychological motivations that influence sport consumption. A number of authors have hypothesized a wide array of motives to explain such behavior (Sloan, 1989; Zillmann, Bryant, & Sapolsky, 1989; Zillmann & Paulus, 1993), including aesthetics, catharsis, drama, entertainment, escape, social interaction, and vicarious achievement (see Trail, Anderson, & Fink, 2000, for a review of spectator motives). As Trail et al. suggested, most of the hypothesized motives are based on social and psychological needs. The next step in understanding the motives of sport spectators is to design an instrument that accurately assesses the motives of sport spectators.

Wann (1995), Milne and McDonald (1999), and Kahle, Kambara, and Rose (1996) have developed scales to measure the underlying motives of sports fans. Based on the existing conceptual literature within sport sociology (Sloan, 1989; Zillmann et al., 1989; Zillmann & Paulus, 1993), Wann (1995) developed the Sport Fan Motivation Scale (SFMS). Milne and McDonald developed an instrument to measure spectator and participant motives based on the work of Sloan (1989) and Maslow (1943), and Kahle et al. proposed a scale based on Kelman's (1958) functional theory of attitudinal influence. Before any scale may be utilized, however, it is imperative to establish the validity and reliability of such an instrument. An examination of the three scales reveals several psychometric limitations. Thus, the purpose of the present study was to create an instrument with the content validity, construct validity, reliability and criterion validity that measures the motives underlying sport spectator consumption behavior.

Limitations of Existing Scales

The development of measurement instruments is an important step in examining the motives behind why people watch or follow sports. Each of these three scales was proposed as a measure of sport spectator motives, but is limited in some aspect of their validity or reliability.

Sport Fan Motivation Scale

Wann's (1995) Sport Fan Motivation Scale (SFMS) was designed to "document empirically the motives [of sports fans] and establish the relative importance of each" (p. 378). Eight underlying factors, represented by 23 items that motivate fan behavior, were identified from the conceptual literature: eustress, self-esteem benefits, escape, entertainment, economic factors, aesthetic qualities, group affiliation, and family needs.

There are several validity problems within the SFMS. One concern is the scale's content validity, the extent to which the items in the scale accurately represent the designated concepts. Wann identified the possible "motivations" of sport fans from previous research, but gave no indication as to how the scale items were generated, whether or not a panel of experts was used, or how the final list of items was selected. Another concern with the content validity of the SFMS lies in the wording of items. For example, one statement asks about "getting pumped up when watching a favorite team," but the question also included the terms "read" and "discuss." With several of the items it is unclear what was being measured -- watching, discussing, or reading about the favorite team. Wann also used the label "Economic" for one factor, when the items clearly ask solely about betting or wagering on sports not about economic influence. Labels should be used which accurately represent a factor and the items that make up a f actor.

In regard to construct validity, Wann claimed good convergent validity based on the goodness of fit index (GFI). There are, however, concerns about the processes that were used to generate the fit. Responses to the SFMS were initially submitted to a principal components analysis with rotation (type of rotation was not specified). The results yielded a seven-factor solution. Items from the Eustress and Self-esteem subscales loaded on the same factor in the initial analysis but were separated in subsequent analyses. Wann then submitted the same data set to a confirmatory factor analysis (CFA) and compared an eight-factor solution represented by the predetermined 23 items to the seven factor solution. The eight-factor model yielded a significantly better fit to the data than did the seven-factor model and resulted in an excellent fit to the data (CFI = .995). Because Gerbing and Anderson (1993) have noted several concerns about the CFI as a fit index, especially if no other fit indices are reported, we further assessed the construct validity of the SFMS by examining the common variance accounted for by each of the items and by calculating the average variance explained by the factors (Table 1)1. Only 17% of the items loaded below .707 on their assigned factors, indicating that a majority of the items had more common variance than unique variance. In addition, the AVE values for all subscales exceeded the 0.50 minimum standard (Fornell & Larcker, 1988), thus indicating that the constructs accounted for a greater amount of variance than the items. Wann then conducted a second study featuring a separate sample comprised of 144 students. A CFA was performed on the 23-item SFMS. Again, construct validity was claimed as the results indicated that the eight-factor model provided a good fit to the data (CFI = .999).

Another concern with the SFMS is that Wann did not directly examine the issue of discriminant validity to verify that the different constructs in the SFMS were truly distinct. Sloan (1989) noted that the theories underlying spectators' attraction to sporting events "aren't mutually exclusive and competitive. Any number of non-contradictory overlapping theories conceivably could be valid and may have additive or multiplicative influences on fan responses" (p. 202). At a minimum, the discriminant validity of the constructs should have been explicitly tested. For example, even though the eight-factor model provided a better fit to the data, items from the Eustress and Self-esteem subscales loaded on the same factor in the exploratory factor analysis. An appropriate test would have been to determine whether the construct correlation between the factors was within two standard errors of unity (Anderson & Gerbing, 1988), or whether the variance accounted for in either construct exceeded the shared variance of the two constructs (Fornell & Larcker 1981). The former cannot be determined from Wann's article because the standard errors were not reported, however neither construct's variance exceeded the shared variance, indicating that discriminant validity existed for these two subscales. While the remainder of the scales appears to be distinct from each other also, discriminant validity was not explicitly tested in either of Wann's (1995) two studies or more recently by Wann, Brewer, and Royalty (1999).

The reported reliability (internal consistency) of the SFMS has been adequate in the research when reported by Wann and colleagues (Wann, Brewer, & Royalty, 1999; Wann, Schrader, & Wilson, 1999). Specifically Cronbach's alpha reliability scores for the SFMS in Study 1 (Table 1) ranged from [alpha] = .63 to [alpha] = .89. [2]

To establish the criterion validity of an instrument it is important to assess both concurrent and predictive validity. Wann assessed concurrent validity by comparing the factors of the SFMS to two criteria: (1) level of identification with a favorite sports team (Wann & Branscombe 1993); and (2) general sports fanship (measured on a scale ranging from "not at all a sports fan" to "very much a sports fan"). Regarding team identification, Wann reported that a significant relationship existed between this scale and each of the eight SFMS factors (range: r = .14 for Economics to r .71 for Eustress and Self-esteem). The measure of general fanship was significantly correlated with each of the dimensions except Economics (range: r = .06 for Economics to r = .69 for Eustress). This indicated preliminary concurrent validity. Wann did not however, examine predictive validity, so criterion validity of the SFMS has not been established.

Thus, for the SFMS, concerns exist in the areas of content validity, discriminant validity, criterion validity, and to some extent convergent validity. By providing the first effort to operationalize potential motives for sport spectators, the SFMS provides a good starting point for building upon the positive attributes of the scale to develop a new instrument with better psychometric properties.

Motivations of the Sport Consumer(MSC)

Milne and McDonald (1999) suggested twelve motivation constructs (37 items) for spectators: (l)risk-taking; (2) stress reduction; (3) aggression; (4) affiliation; (5) social facilitation; (6) self-esteem; (7) competition; (8) achievement; (9) skill mastery; (10) aesthetics; (11) value development; (12) self-actualization (p. 23-26). One content validity concern with the MSC scale is that the instrument also measures motives for participation. Researchers have suggested that many of the motives for participation are consistent with the motives for spectators (Sloan, 1989; Zillmann, Bryan & Sapolsky, 1989). It should be noted though, that there are motives for participation that are not applicable to spectators. For example, skill mastery, risk-taking, and stress reduction are outcomes that participants may experience, but usually not spectators.

Two additional concerns with the MSC scale are that neither discriminant nor convergent validity was examined in the construction of the scale. The primary problem with the fit of Milne and McDonald's model is their use of exploratory factor analysis. The initial twelve subscales demonstrated good reliability, and should have been used to compute a confirmatory factor analysis at this point. Instead, an exploratory factor analysis (principal axis factoring) was computed, with the factors constrained to be orthogonal. As a result, the twelve subscales were reduced to a four-factor solution. Consequently, 78% of the loadings of the individual items on the new factors explained more unique variance than common variance (i.e., loadings below .707), indicating that they did not represent the factors well. The unique variance was responsible for a greater amount of the observed variance than was common variance in 75% of the subscales. As is evident in Table I, only one of the subscale's AVE values exceeded .50, i ndicating that Milne and McDonald's factors had more specific variance and measurement error than common variance. [1]

Milne and McDonald also computed several CFA's to generate goodness of fit indices for each subscale and claimed acceptable fit for each of their models although the RMR values ranged from .04 to .07, their AGFI values from .77 to .88, and their NFI values from .89 to .96. Bagozzi and Yi (1988) have noted that models that have AGFI and NFI values greater than .90 indicate meaningful models, but often misrepresent the model fit because of the effects of a large sample size. Although the NFI values were generally acceptable, the AGFI values indicated that the models could be improved considerably. In addition, considering the sample size of Milne and McDonald's data set (N = 1611) these fit indices were inflated. Thus, although discriminant validity was implicit in the MSC scale due to the orthogonal rotation during the exploratory factor analysis, convergent validity is not evident due to the low loadings of the items and the questionable fit indices.

The reliabilities of Milne and McDonald's original subscales ([alpha] = .74 to .93) were within acceptable ranges (see Table 1). However, Milne and McDonald then combined several of these subscales and did not recalculate the new internal consistency values, which certainly would have changed some of the alpha values. Thus, there are no known reliability estimates.

Milne and McDonald reported that they examined predictive validity, however this was not the case if we are using Malhotra's (1992) definition. Milne and McDonald measured the relationships between their four factors and various fan behaviors (e.g. watching, listening, reading, etc.) at the same time, not in the future as Malhotra indicated. Several of the relationships could represent concurrent validity, however their support of the claimed criterion validity was limited (i.e. no references for their statistical procedures).

In general, for the MSC scale, content validity was good on the subscales that were applicable to spectator sport motivation, that is, if the subscales specific to participation motivation were removed. Unfortunately, there is little evidence for convergent validity or criterion validity. Overall, the usefulness of this scale lies in its original items and categories from which a new instrument could be developed with better psychometric properties.

Fan Attendance Motivations (FAM)

Kahle et al. (1993) examined the "individual differences in psychological motivation for attending college football games" (p. 51). They focused on level of emotional attachment to a team, and the importance of winning and camaraderie as motives for attending games. Kahle et al. did not seek to identify the motives of spectators per se. Drawing from Kelman's (1958) theory of attitudinal influence, they tested the extent to which attendance at college football games was motivated by a desire for a self-expressive experience, camaraderie, and a love of the game. The components tested may well be important motives for sport spectator consumption behavior, but other potential motives were not assessed because they were not relevant to Kelman's theory. Additional concerns with FAM scale are that the internal consistency (reliability) of the whole scale was very low ([alpha] = .64) and 5 out of the 7 subscales had alphas below .64 (Table I). In addition, neither construct validity nor criterion validity was discus sed and results were not presented in their paper that would allow these measures to be calculated by the reader. Although Kelman's (1958) theory of attitudinal influence may be useful in examining fan motivation, the FAM scale is not psychometrically sound enough to warrant continued use.

Each of the spectator/fan motivation scales mentioned above has one or more areas of concern regarding different aspects of validity and/or reliability. The scales do, however, provide a foundation from which to develop a psychometrically sound instrument for assessing the motives of sport spectators. Drawing on the strengths of Wann's SFMS and the MSC scale proposed by Milne and McDonald, a new instrument to measure the motivations of sport spectators relative to sport consumption behavior is presented.

Development of the Motivation Scale for Sport Consumption (MSSC)

The Motivation Scale for Sport Consumption (MSSC) was developed from a review of the literature and also from the evaluation of the Wann (1995) and Milne and McDonald (1999) scales. The motives included in the MSSC were based on the sport sociology literature (Sloan, 1989; Zillmann et al., 1989; Zillmann & Paulus, 1993), and were consistent with the motives identified by previous research. Trail et al. (2000) hypothesized that there were nine factors representing motives for following sports: achievement, acquisition of knowledge, aesthetics, drama/eustress, escape, family, physical attractiveness of participants, the quality of the physical skill of the participants and social interaction. These are the motives that the MSSC measures. Items were generated for the nine motives, with special attention given to wording items to avoid measurement confusion (as noted above). Experts in the field were asked to evaluate the entire scale and based upon the feedback received, several of the items were revised prior to administration of the MSSC.

Method

Participants and Measures

The MSSC was administered to season ticket holders for a major league baseball team. A stratified random sample of respondents was drawn from the pool of season ticket accounts based on seating level: (1) lower, (2) middle, and (3) upper level seating. At each seating level the percentage of season ticket accounts for each price category was calculated. A corresponding percentage of accounts were then selected from each level and each price category. Previous research has drawn from student populations (Kahie et al., 1996; Wann, 1995) and the general population (Milne & McDonald, 1999). In order to get a true test of an instrument measuring spectator motives, it was important to administer the MSSC to a sample that was definitely interested in sports in general and in this case one specific team, as evidenced by the willingness to purchase tickets and attend games.

Surveys were mailed to 275 season ticket holders, along with a letter explaining the purpose of the study. Complete data were collected from 203 season ticket holders, for a response rate of 73.8%. The questionnaire used in this study was comprised of multiple sections, of which one pertained to fan motives. This section contained 27 items representing the nine factors (see Table 2). The item statements were measured on a 7-point scale indicating level of agreement ranging from 1 (strongly disagree) to 7 (strongly agree). In addition to the MSSC, respondents were asked to complete a three-item scale measuring their level of identification with the team ([alpha] = 0.84), indicate the number of games that they had attended to date (M=5.23), and complete a single item self-rating of fanship with the team (M=7.15). The latter item was measured on a 9-point scale ranging from "not a fan at all" (1) to "an extremely loyal fan" (9). These additional measures were included to assess concurrent validity.

Data Analysis

The current study assessed a model in which nine latent variables or constructs were represented by the measures comprising that factor. Each factor was represented by three items. The RAMONA Covariance Structure Modeling (CSM) technique, available in the SYSTAT 7.0 (1997) statistical package, was used to compute a confirmatory factor analysis. RAMONA implements the McArdle and McDonald (1984) Reticular Action Model (RAM) for path analysis with latent and manifest variables (SYSTAT 7.0, 1997). The measures of fit used in the current study were Steiger's (1989; Steiger & Lind, 1980) root-mean-square-error of approximation (RMSEA, a measure of discrepancy per degrees of freedom), and the test of close fit (Browne & Cudeck, 1992). Goodness-of-fit indices (GFI), such as a comparative fit index (CFI) and chi-square statistics, often reflect the size of a sample or the number of parameters rather than the adequacy of the model (Browne & Cudeck, 1992). The RMSEA is thought to alleviate problems associated with mode l fit that are not addressed by GFI or chi-square statistics (Browne & Cudeck, 1992; Mulaik, James, Van Alstine, Bennet, Lind, & Stilwell, 1989), thus those indices are not included in the statistical package.

The RMSEA is bounded by zero on the lower end and will only be zero if the model fits exactly. Values less than .05 indicate that a model has a close fit, values of .08 or less would indicate reasonable fit, and RMSEA values higher than .10 should not be considered. Since the RMSEA is a point estimate, it is also suggested that a 90% confidence interval be calculated and reported to show the level of confidence that the model would fit well within the population (Browne & Cudeck, I 992). [3]

Results

Overall Model Fit

The results of the analysis indicated that the model fit the data reasonably well. The RMSEA point estimate for the nine-factor model was 0.057, and the confidence interval around the point estimate was relatively small (CI = .047, .066). [4] The test for close fit (pclose = .124) was not rejected, indicating that the model fit this population well (Browne & Cudeck, 1992). The chi-square divided by the degrees of freedom value was 1.63, below the "rule of thumb" value of 2.00 used in social science research.

Construct Validity - Convergent Validity

A fundamental aspect of construct validity is the determination of whether a scale's items each contribute to its underlying theoretical construct. Specifically, convergent validity is evidenced if each indicator's loading on its posited underlying construct is greater than twice its standard error, i.e. significant (Anderson & Gerbing, 1988). In the MSSC, although each of the items loaded significantly on its specified factor (t values ranged from 6.47 to 88.03), six of the items loaded below .707 on their assigned factors (Table 2). This indicated that these items' unique variance was greater than their common variance. In other words, measurement error and specific variance were responsible for a greater amount of observed variance than was common variance. Fornell and Larcker (1981) suggested that the average variance extracted (AVE) provides another indicator of the overall convergent validity of a subscale. This measure indicates the amount of variance explained by the construct relative to the amount of variance that may be attributed to measurement error and should exceed .50. Table 2 shows that all but one of the subscale AVE values exceeded .50, ranging from .51 (Escape) to .74 (Achievement). The Family Needs subscale was slightly below the designated value (AVE = .48). Another measure of the internal quality of a construct's indicators is the residual matrix. [5] Residuals greater than .10 generally indicate a problem with the model (Bagozzi & Yi, 1988). Only 5% of the residuals were moderately high (36 residuals were between .10 and .19). Items with the highest residuals were associated with the Family Needs and Drama factors. Two items specifically accounted for 33.3% of the high residuals (Drama 1 and Family3). In sum, there is considerable evidence of convergent validity based on t values, values of the factor loadings, AVE values and residuals.

Construct Validity - Discriminant Validity

Another dimension of construct validity, discriminant validity, was used to determine the extent to which the constructs were unique. According to Anderson and Gerbing (1988), the correlation between any two constructs should not be within two standard errors of unity. If this confidence interval includes 1.0, the two constructs are not distinct. The correlations among latent constructs and their standard errors are shown in Table 3. Although some of the correlations between the constructs were fairly high, none failed the initial test of discriminant validity.

A more rigorous test of discriminant validity was also applied in which the average variance extracted (AVE) for each construct was evaluated. Fornell and Larcker (1981) suggested that the AVE for each construct should be greater than the squared correlation between that construct and any other. Failing this test indicates that the variance accounted for by a construct explains less variance than that explained by that construct's correlation with another, thus denoting a lack of discrimination. None of the squared correlations exceeded the AVE values for any of the constructs, indicating that discriminant validity was good.

Reliability

Cronbach's alpha coefficients represented each factor's internal consistency. The alpha coefficient for the overall scale was 0.87. One factor fell slightly short of the 0.70 cutoff recommended by Nunnally (1978): Family ([alpha] = .68). Table 2 shows that the remaining alpha values for the factors ranged from .72 (Escape) to .89 (Achievement).

Criterion Validity

Criterion validity examines the extent to which a measurement scale performs as expected in relation to some external variables considered to be meaningful criteria, and can take two forms: concurrent and predictive validity. Concurrent validity was assessed by comparing the factors of the MSSC to three criteria: (1) level of identification with a favorite sports team (a three item scale with [alpha] = .85); (2) general fanship of the team (one item); and (3) number of games attended. All subscales except Physical Attraction (see Table 4) showed a significant correlation with the team identification scale (range: r = .23 for Drama to r = .71 for Achievement). The measure of general team fanship was significantly correlated with each of the dimensions except Physical Attraction (range: r = .21 for Escape to r = .51 for Achievement). In addition, three subscales correlated significantly with the number of games attended: Acquisition of Knowledge, r = .24; Aesthetics r =.25; Physical Skill, r = .23 (see Table 4 for all correlations). The significant correlations between the motives and the criterion variables provide good evidence of the MSSC's concurrent validity.

Predictive validity was assessed by comparing the factors of the MSSC to four criteria from data that were collected at the end of the season (approximately two months later). The four criteria were: (1) general fanship of the team (same as in the initial survey); (2) the loyalty of the fan (a five item scale with a = .86); (3) change in team merchandise consumption (a three item scale with a = .83); and (4) change in media consumption (a three item scale with [alpha] = .69). All subscales except Drama and Physical Attraction were significantly correlated with being a fan of the team at the end of the season (range: r = .19 for Escape to r = .39 for Achievement). Six of the nine subscales were significantly correlated with continued loyalty (range: r = .21 for Aesthetics to r = .40 for Achievement). Five scales were significantly correlated with an increase in team merchandise consumption (range: r = .19 for Physical Attraction to r = .39 for Achievement). Lastly, five scales were significantly correlated wi th an increase in media consumption about the team (range: r = .19 for Aesthetics to r = .50 for Achievement) (see Table 5 for all correlations). Again these significant correlations indicate that these subscales were predictive of fanship, loyalty, future merchandise consumption and future media consumption. In terms of the content validity, criterion validity, construct validity, and internal consistency, the MSSC demonstrated the best psychometric properties overall to accurately and reliably measure motivations of sport spectator consumption behavior.

Discussion

The purpose of the current study was to develop and evaluate the Motivation Scale for Sport Consumption (MSSC). The MSSC improved upon the content of Wann's (1995) Sport Fan Motivation Scale and Milne and McDonald's (1999) Motivations of Sport Consumers scale by incorporating the best aspects of both and by establishing content validity similar to that of the earlier scales. The MSSC bettered the two scales by accurately assessing nine motives that have been hypothesized to drive spectator consumption behavior (Sloan, 1989; Zillmann, Bryan & Sapolsky, 1989; Trail et al., 2000), including two new factors (acquisition of knowledge and physical attraction of participants) that represent unique constructs not measured by previous scales.

The results of the current study demonstrate that the SFMS, Milne and McDonald's (1999) scale, and Kahle et al.'s (1996) scale are lacking in certain psychometric properties and are not valid and reliable measures of spectator motives. The MSSC, however, showed good construct validity as indicated by the close fit of the model to the data (RMSEA = 0.47, test for close fit was not rejected, pclose = .124), and also by the convergent validity (the AVE values for all but one subscale exceeded .50) and discriminant validity (no squared correlations exceeded the AVE values for any construct). The internal consistency of the subscales was very good, with the alpha values for all but one factor exceeding the .70 cutoff. The significant correlations between the motives and three criterion variables (level of identification with a sports team, general fanship, and number of games attended) show that the MSSC does have criterion validity. Significant correlations between the motives and data collected two months after the initial survey also confirmed the predictive validity of the scale. The results show that the MSSC has demonstrated the best psychometric properties overall to accurately and reliably measure motivations of sport spectator consumption behavior.

The development of the MSSC advances the study of sport spectators by providing a tool for measuring the psychological motivations that influence sport consumption. Previous research has focused to a large extent on the demand for sport based on economics, promotions, and residual preference factors, and the relationship between demographic variables and an interest in watching sports. The MSSC will allow academics and practitioners to better understand the impact of psychological motives (e.g., need for achievement, escape, desire for drama, social interaction, etc.) on attendance at sporting events, purchase of merchandise, and other consumptive behavior. Identifying psychological motives will also allow researchers to advance our understanding of why people make a commitment (i.e., become loyal) to a specific sport or team.

Future research should be done to replicate the findings of the current study and to continue developing the MSSC. The overall performance of the scale was good, but there is room for improvement. For example, the alpha coefficients for two of the items in the Family subscale were below the desired .70 cut-off. Additional work is needed to refine and strengthen these items. An important contribution that the scale makes to the study of sport spectators is the valid use with the desired target group. Previous scale development has been conducted primarily with student samples that are not necessarily sport spectators. The validity and reliability of the MSSC was established using season ticket holders of a Major League Baseball team. The MSSC should be administered to other spectators, particularly those who simply enjoy observing sport (e.g., people who purchase single game tickets), to assess the ecological validity of the scale for measuring motives of observers.

A final contribution the MSSC offers to the study of consumptive behavior is the potential to determine whether people follow different sports for similar or different motives. Future research should test the MSSC in other sports besides professional baseball to determine whether there are motives that influence sport consumption in general, or if there are different motives that are connected more strongly to one sport than another. Research that seeks to explain the reason(s) observers and fans are attracted to different sports would advance the knowledge of personal and situational factors that help explain spectator motivation and identify the reasons sports spectators enjoy watching sporting events.

(a.) The authors contributed equally to this manuscript.

Footnotes

(1.) Since AVE values were not reported by Wann (1995) nor by Milne and McDonald (1999), the values shown in Table 1 were calculated by averaging the squared multiple correlations derived from the reported standardized loadings.

(2.) It should be noted that Wann did not report reliabilities for the constructs in Study 2, just the range (.59 for Family to .94 for Economic). Consequently, it was not possible to compare the results of the current study to Study 2.

(3.) The RMSEA is not necessarily influenced by adding additional parameters to improve model fit as are some measures of fit. The RMSEA does not use a comparison to the null model, it takes into account the benefits of a parsimonious model, it is not obscured by the incorporation of n into the equation and a confidence interval for the point estimate can be calculated.

(4.) The sample discrepancy function value, which represents the difference between the sample correlation matrix and the reproduced correlation matrix, was low (FML = 1.618) as was the population discrepancy function (FO = 0.865; CI = 0.655, 1.103).

(5.) Items in the residual matrix represent the difference between corresponding correlations from the observed and reproduced correlation matrices. Values in the residual matrix should lie between zero and one. Residuals that are closer to zero (less than . 10) provide an indication that a sample fits the model; residuals greater than .10 generally indicate that there are problems with a model (Bagozzi & Yi, 1988).

References

Anderson, J. C., & Gerbing, D. W. (1988). Structural equation modeling in practice: A review and recommended two-step approach. Psychological Bulletin, 103, 411-423.

Baade, R. A., & Tiehen, L. J. (1990). An analysis of Major League Baseball attendance, 1969--1987. Journal of Sport and Social Issues, 14, 14-32.

Badenhausen, K., Nikolov, C., Alkin, M., & Ozanian, M. K. (1997, June 17). Sports, more than a game: An in-depth look at the raging bull market in sports franchises. Financial World, 166(6), 40-62.

Bagozzi, R. P., & Yi, Y. (1988). On the evaluation of structural equation models. Journal of the Academy of Marketing Science, 16, 74-94.

Browne, M. W., & Cudeck, R. (1992). Alternative ways of assessing model fit. Sociological Methods and Research, 21, 230-258.

Fornell, C., & Larcker, D. F. (1981). Evaluating structural equation models with unobservable variables and measurement error. Journal of Marketing Research, 18, 39-50.

Gerbing, D. W., & Anderson, J. C. (1993). Monte Carlo evaluations of goodness-of-fit indices for structural equation models. In K. A. Bollen & J. S. Long (Eds.), Testing Structural Equation Models (pp. 1-9). Newbury Park, CA: Sage Publications.

Greenstein, T. N., & Marcum, J. P. (1981). Factors affecting attendance of major league baseball: Team performance. Review of Sport and Leisure, 6(2), 21-34.

Hansen, H., & Gauthier, R. (1989). Factors affecting attendance at professional sport events. Journal of Sport Management, 3, 15-32.

Kahle, L. R., Kambara, K. M., & Rose, G. M. (1996). A functional model of fan attendance motivations for college football. Sport Marketing Quarterly, 5(4), 51-60.

Kelman, H. C. (1958). Compliance, identification, and internalization: Three processes of attitude change. Journal of Conflict Resolution, 2, 51-60.

Malhotra, N. K. (1992). On the construct validity of intrinsic sources of personal relevance. Journal of Business Research, 25, 143-147.

Maslow, A. H. (1943). A theory of human motivation. Psychological Review, 50, 370-396.

McArdle, J. J., McDonald, R. P. (1984). Some algebraic properties of the Reticular Action Model for moment structures. British Journal of Mathematical & Statistical Psychology, 37, 234-251.

Milne, G. R., & McDonald, M. A. (1999). Sport marketing: Managing the exchange process. Sudbury, MA: Jones and Bartlett Publishers.

Mulaik, S. A., James, L. R., Van Alstine, J., Bennett, N., Lind, S., & Stilwell, C.D. (1989). Evaluation of goodness-of-fit indices for structural equation models. Psychological Bulletin, 105, 430-445.

Nunnally, J. C. (1978). Psychometric Theory (2nd ed). New York: McGraw-Hill.

Pooley, J. C. (1978). The sport fan: A social psychology of misbehavior. Sociology of Sport Monograph Series. Calgary: CAPHER

Schofield, J. A. (1983). Performance and attendance at professional team sports. Journal of Sport Behavior, 6, 196-206.

Sloan, L. R. (1989). The motives of sports fans. In J.H. Goldstein (Ed.), Sports, games, and play: Social and psychological viewpoints (2nd ed., pp. 175-240). Hillsdale, NJ: Lawrence Erlbaum Associates.

Steiger, J. H. (1989). EZPATH: A supplementary module for SYSTAT and SYGRAPH. Evanston, IL: SYSTAT.

Steiger, J. H., & Lind, J. C. (1980, May). Statistically-based tests for the number of common factors. Paper presented at the annual Spring meeting of the Psychometric Society, Iowa City, IA.

Trail, G. T., Anderson, D. F., & Fink, J. (2000). A theoretical model of sport spectator consumption behavior. International Journal of Sport Management, 3, 154-180.

TV sports: The $3.5 billion ticket. (1996, May 13). Broadcasting and Cable, 126(21), 34.

Wann, D. L. (1995). Preliminary validation of the sport fan motivation scale. Journal of Sport and Social Issues, 19, 377-396.

Wann, D. L., & Branscombe, N. R. (1993). Sports fans: Measuring degree of identification with their team. International Journal of Sport Psychology 24, 1-17.

Wann, D. L., Brewer, K. R., & Royalty, J. L. (1999). Sport fan motivation: Relationships with team identification and emotional reactions to sporting events. International Sports Journal, 3(Summer), 8-18.

Wann, D. L., Schrader, M. P., & Wilson, A. M. (1999). Sport fan motivation: Questionnaire validation, comparisons by sport, and relationship to athletic motivation. Journal of Sport Behavior, 22, 114-139.

Zhang, J. J., Pease, D. G., Hui, S. C., & Michaud, T. J. (1995). Variables affecting the spectator decision to attend NBA games. Sport Marketing Quarterly, 4(4), 29-39.

Zhang, J. J., Smith, D. W., Pease, D. G., & Jambor, E. A. (1997). Negative influence of market competitors on the attendance of professional sport games: The case of a minor league hockey team. Sport Marketing Quarterly, 6(3), 34-40.

Zillmann, D., Bryant, J., & Sapolsky, B. S. (1989). The enjoyment of watching sports contests. In J. G. Goldstein (Ed.), Sports, games, and play: Social and psychological viewpoints (2nd ed., pp. 241-278). Hillsdale, NJ: Lawrence Erlbaum Associates.

Zillmann, D., & Paulus, P.B. (1993). Spectators: Reactions to sports events and effects on athletic performance. In R. N. Singer, M. Murphey & L. K. Tennant (Eds.), Handbook of Research on Sport Psychology (pp. 600-619). New York, NY: McMillan. Cronbach's alpha coefficients and AVE values for Wann's SFMS (1995), Milne & McDonald's (1999) and Kahle et al.'s (1996) scales. Wann's SFMS (1995) Milne & McDonald (1999) (1996) Subscale [alpha] AVE Subscale [alpha] AVE Sport-Based Needs .36 Aesthetics .81 .60 Stress Release .88 Drama .89 .78 Skill Mastery .79 Economics .84 .65 Aesthetics .83 Entertainment .85 .69 Mental Well Being Needs .55 Escape .85 .70 Self-esteem .93 Family .63 .61 Self-Actualization .92 Group Affiliation .72 .51 Value Development .85 Self-esteem .78 .78 Social Needs .38 Social Facilitation .84 Affiliation .84 Personal Needs .23 Achievement .82 Risk-Taking .90 Aggression .85 Competition .72 Wann's SFMS (1995) Kahle, Kambara, & Rose (1996) Subscale Subscale [alpha] AVE Aesthetics Internalization .80 NA Drama Self-expressive experience .56 NA Economics Camaraderie .72 NA Entertainment Compliance .53 NA Escape Obligation .64 NA Family Self-defining experience .61 NA Group Affiliation Identification with winning .59 NA Self-esteem Item Loadings ([beta]), Confidence Intervals (CI), Standard Errors (SE), t-values (t), Cronbach's Alphas ([alpha]) and Average Variance Explained (AVE) for the Motivation Scale for Sport Consumption Factor and Item [beta] CI SE t Achievement I feel like I have won when the team wins .906 .871-.941 .021 42.30 I feel a personal sense of achievement when the team does well .858 .816-.899 .025 34.03 I feel proud when the team plays well .808 .759-.857 .030 27.15 Knowledge .80 .59 I regularly track the statistics of specific players .761 .692-.830 .042 18.14 I usually know the team's win/loss record .660 .580-.741 .049 13.56 I read the box scores and team statistics regularly .867 .806-.928 .037 23.54 Aesthetics I appreciate the beauty inherent in the game .912 .877-.948 .022 42.23 There is a certain natural beauty to the game .853 .810-.895 .026 32.68 I enjoy the gracefulness associated with the game .783 .730-.837 .032 24.14 Drama I enjoy the drama of a "one run" game .796 .729-.863 .041 19.50 I prefer a "close" game rather than a "one-sided" game .804 .738-.871 .040 19.92 A game is more enjoyable to me when the outcome is not decided until the very end .684 .605-.762 .048 14.30 Escape Games represent an escape for me from my day-to-day activities .824 .759-.889 .040 20.78 Games are a great change of pace from what I regularly do .759 .688-.830 .043 17.64 I look forward to the games because they are something different to do in the summer .509 .409-.609 .061 8.36 Family I like going to games with my family .936 .855-1.02 .050 18.84 I like going to games with my spouse .499 .398-.600 .061 8.14 I like going to games with my children .557 .461-.653 .058 9.53 Factor and Item [alpha] AVE Achievement .89 .74 I feel like I have won when the team wins I feel a personal sense of achievement when the team does well I feel proud when the team plays well Knowledge .80 .59 I regularly track the statistics of specific players I usually know the team's win/loss record I read the box scores and team statistics regularly Aesthetics .88 .72 I appreciate the beauty inherent in the game There is a certain natural beauty to the game I enjoy the gracefulness associated with the game Drama .80 .58 I enjoy the drama of a "one run" game I prefer a "close" game rather than a "one-sided" game A game is more enjoyable to me when the outcome is not decided until the very end Escape .72 .51 Games represent an escape for me from my day-to-day activities Games are a great change of pace from what I regularly do I look forward to the games because they are something different to do in the summer Family .68 .48 I like going to games with my family I like going to games with my spouse I like going to games with my children Physical Attraction .78 .69 I enjoy watching players who are physically attractive .603 .520-.686 .05 11.93 The main reason that I watch is because I find the players attractive .967 .910-1.02 .035 27.84 An individual player's "sex appeal" is a big reason why I watch .794 .730-.858 .039 20.49 Physical Skills .75 .53 The physical skills of the players are something I appreciate .675 .596-.754 .048 14.10 Watching a well-executed athletic performance is something I enjoy .729 .657-.800 .044 16.75 I enjoy a skillful performance by the team .771 .705-.838 .040 19.11 Social .78 .54 Interacting with other fans is a very important part of being at games .773 .701-.844 .044 17.74 I like to talk to other people sitting near me during the games .714 .636-.791 .047 15.10 Games are great opportunities to socialize with other people .724 .647-.800 .047 15.55 Correlations among Factors, Standard Errors (in parentheses) Factor Achievement Knowledge Aesthetics Drama Acquisition of knowledge .185 [*] (.080) Aesthetics .339 [*] .463 [*] (.071) (.067) Drama .309 [*] .249 [*] .311 [*] (.077) (.082) (.077) Escape .656 [*] .023 .329 [*] .319 [*] (.055) (.088) (.077) (.082) Family .355 [*] .075 .353 [*] .338 [*] (.073) (.082) (.073) (.077) Physical attraction .194 [*] -.014 .147 [*] -.115 (.075) (.080) (.076) (.081) Physical Skills of players .465 [*] .257 [*] .622 [*] .411 [*] (.071) (.084) (.059) (.078) Social interaction .451 [*] .334 [*] .376 [*] .388 [*] (.071) (.081) (.075) (.079) Factor Escape Family Attraction Skills Acquisition of knowledge Aesthetics Drama Escape Family .406 [*] (.076) Physical attraction .164 [*] .062 (.081) (.078) Physical Skills of players .455 [*] .630 [*] .160 [*] (.077) (.065) (.082) Social interaction .479 [*] .312 [*] .083 .519 [*] (.075) (.080) (.083) (.074) Note: (*.)Correlations significant at the .05 level Assessment of Concurrent Validity for the MSSC Pearson Correlation Fan of the team Number of games attended Achievement .509 [**] .134 Acquisition of knowledge .274 [**] .241 [**] Aesthetics .369 [**] .253 [**] Drama .216 [**] .061 Escape .215 [**] -.011 Family .327 [**] .034 Physical attraction .076 .034 Physical skill .449 [**] .233 [**] Social interaction .223 [**] .116 Pearson Correlation Team identification Achievement .712 [**] Acquisition of knowledge .380 [**] Aesthetics .437 [**] Drama .232 [**] Escape .445 [**] Family .291 [**] Physical attraction .132 Physical skill .489 [**] Social interaction .393 [**] Note: (*.)Correlation is significant at the 0.05 level (2-tailed). (**.)Correlation is significant at the 0.01 level (2-tailed). Assessment of Predictive Validity for the MSSC Pearson Correlation Fan of Loyalty to Increase in Increase in media the Team the team merchandise consumption purchasing Achievement .394 [**] .395 [**] .393 [**] .496 [**] Acquisition of knowledge .289 [**] .009 .164 .068 Aesthetics .244 [**] .205 [*] .237 [**] .191 [*] Drama -.018 .000 -.003 .140 Escape .185 [*] .326 [**] .168 .377 [**] Family .221 [*] .286 [**] .129 .128 Physical attraction .064 .159 .194 [*] .166 Physical skill .248 [**] .247 [**] .195 [*] .249 [**] Social interaction .365 [**] .319 [**] .221 [*] .325 [**]