Professional Documents
Culture Documents
Kwon 2005
Kwon 2005
0 2005 SMAANZ
Galen T. Trail
University of Florida
Dean F. Anderson
Iowa State University
Hany H. Kwon is with the Sport Management, Recreation Management &Physical Education
Department at Florida State University. Galen T. Trail is with the Department of Tourism,
Recreation, & Sport Management at the University of Florida, and Dean F. Anderson is with
the Department of Health and Human Performance at Iowa State University. Email for Kwon:
kwon@coe.fsu.edu
256 Kwon. Trail & Anderson
It has been over a decade since Wann and Branscombe (1993) developed a scale
purported to measure fans' level of identification with their team. Considerable
effort has followed their work. Team identification has been shown to predict
cognitive (Trail, Anderson, & Fink, 2005; Trail, Fink, &Anderson, 2003), affective
(Madrigal, 1995; Matsuoka, Chelladurai, & Harada, 2003), conative (Melnick &
Warn, 2004; Warn, 2002; Wann & Branscombe, 1993), and behavioural (Laverie
& Amett, 2000) dimensions of sport spectatorship. Furthermore, many studies have
examined the construct of team identification itself and have developed scales that
attempted to measure the construct (e.g., Mahony, Madrigal, & Howard, 2000; Trail
& James, 2001; Trail, Fink et al., 2003).
These researchers have typically treated team identification as uni-
dimensional. However, recent research has suggested that there may be additional
points of attachment, rather than just attachment to the team (Funk, Mahony,
Nakazawa, & Hirakawa, 2000,2001; Funk, Mahony, & Ridinger, 2002; Kwon &
Armstrong, 2004; Robinson & Trail, 2005; Robinson, Trail, Dick, & Gillentine,
2005; Robinson, Trail, & Kwon, 2004; Trail, Robinson, Gillentine, & Dick, 2003).
Although Funk and colleagues have termed these concepts "motives", their ideas
regarding these constructs represent an attachment to a specific entity. On the other
hand, Trail and colleagues have framed these points of attachment within the identity
theory of Stryker (1968, 1980, 1994), under the guise of different role identities.
Along with this effort, recent research suggests that identification with the team, as
a unidimensional scale, may not be adequate to explain why people maintain loyalty
or why they spectate. Thus, one would expect that more points of attachment would
lead to more loyalty. Thus, sport marketers should be able to explain and predict
more about sport consumers' cognitive, affective, conative, and behavioural loyalty
with more dimensions of attachment.
to team and attachment to coach subscales), Robinson et al. (2004) noted good
convergent validity and internal consistency on five subscales: players, community,
sport, tour type, and charities.
the individual was solely a spectator or considered himself a fan. This indicated
that, in some instances, multiple points of attachment are relevant; especially when
examining spectators who may have low levels of team identification.
Furthermore, although Kwon et al. (2003) found that only four points
of attachment were distinct, they determined that three of those four points of
attachment (team, university, and sport) were statistically significant determinants
of five purchasing/wearing behaviours of sport-licensed apparel. This indicated that
there may be at least three distinct predictors of sport consumer behaviour.
In sum, the results of the Kwon et al. (2003) and the Robinson and Trail
(2005) studies indicated that a contraction of the number of dimensions in the PA1
might be possible. If this is true and there is a lack of discriminant validity in the
seven-factor model of PAI, then a more parsimonious model of attachment (i.e.,
a fewer number of dimensions) might be appropriate. However, it is not enough
to only test the discriminant validity of the model, predictive validity needs to be
examined as well. To test this, the relationships between the points of attachment
and sport consumers' cognitions, affect, conation, and/or behaviours need to be
evaluated. A comprehensive model should be retained if there is discriminant
validity and multiple dimensions explain variance in the dependent variables. A
more parsimonious model should be retained if there is no discriminant validity
among the points of attachment dimensions. If discriminant validity does exist in the
seven-factor model, but only a limited amount of additional variance is explained in
sport consumers' cognitions, affect, conation, and/or behaviour by the inclusion of
multiple dimensions, then a more parsimonious model again should be considered.
The purpose of this study was to examine the reliability and construct
validity of the Points of Attachment Index. However, we were primarily interested
in determining whether fewer subscales of the PA1 would satisfactorily predict
cognitive, affective, conative, and behavioural dimensions of sport spectatorship.
We initially used a single dimension of the PA1 (attachment to the team) as a
predictor. We then added two more dimensions (attachment to the university and
attachment to the sport) as predictors based on the findings of Kwon and Armstrong
(2004) and Robinson and Trail (2005). Finally, we added three more dimensions
(attachment to the coach, the players, and the level of sport), based on the findings
of Trail, Robinson, et al. (2003) and Robinson, Trail, and Kwon (2004). Through
these hierarchical regression equations, we were able to evaluate the amount of
variance that each set of points of attachment explained in the cognitive, affective,
conative, and behavioural dimensions of sport spectatorship.
Points of Attachment
Method
Participants
Participants were 154 male and 204 female students (N=358) enrolled in Health
and Human Performance classes at a large mid-westem university in the United
States. The students received class credit for completing the survey. The average
age of the respondents was 20.53 (SD = 2.54), and they had attended the school
for approximately two years on average. Most of the respondents (93.6%), self
identified as CaucasianlWhite and most (92.0%) also self identified as fans of the
target team (the university's football team) for 7.4 years on average.
Instrumentation
The questionnaire included six of the seven subscales (attachment to the team,
sport, university, player, level, and coach) of the Point of Attachment Index (Trail,
Robinson, et al., 2003). These six subscales have shown good reliability and
construct validity in past research (Robinson & Trail, 2005; Trail, Robinson, et al.,
2003). Each subscale of the PA1 had three items, which were anchored by a seven
point Likert-type scale ranging from 1 (strongly disagree) to 7 (strongly agree).
The questionnaire also included a cognitive loyalty measure (the BIRGing subscale;
Trail, Anderson, & Fink, 2005), an affective loyalty measure (satisfaction subscale;
Trail et al., 2005), a conative loyalty measure (Intentions for Sport Consumption
Behaviour Scale; Trail et al., 2005), and a behavioural loyalty measure (number
of games attended). All of the items in the dependent measures used a seven point
Likert-type scale from 1 (strongly disagree) to 7 (strongly agree), except for number
of games attended.
Data Analyses
A confirmatory factor analysis (CFA) of the PA1 (i.e., the six points of attachment
scales) was performed via structural equation modeling. Measures of fit included
Steiger's (1989) root-mean-square-error of approximation (RMSEA, a measure of
discrepancy per degrees of freedom), and the test of close fit (Browne & Cudeck,
1992). The RMSEA is thought to alleviate problems associated with model fit that
are not addressed by chi-square based statistics (Browne & Cudeck, 1992; Mulaik,
James, Van Alstine, Bennet, Lind, & Stilwell, 1989), thus those indices are not
included in the RAMONA statistical package. However, we have included the chi-
square value divided by the degrees of freedom as a frame of reference.
RMSEA values less than .05 indicate that a model has a close fit. Values of
.08 or less indicate reasonable fit, and models having RMSEA values higher than. 10
should not be considered (Browne & Cudeck, 1992). Hu and Bentler (1999) have
recently suggested that values less than .06 instead of .05 indicate that a model has
a close fit.
260 Kwon, Trail & Anderson
Results
The RMSEA for the confirmatory factor analysis on the six-factor PA1 showed
reasonable fit (RMSEA, E = .078; CI = .069, .087; pc,oS, < 0.001, x21df= 3591120 =
2.99). In addition, only 9.8% of the residuals were greater than .10 indicating that the
sample correlation matrix and the reproduced correlation matrix did not differ much.
Bagozzi and Yi (1988) suggested that a percentage greater than 10% indicated poor
model fit. Although the RMSEA value does not indicate good fit (only reasonable
fit), the six factors showed good convergent validity and reliability (Table 1). The
alpha coefficients ranged from .83 to .87 and the AVE (Average Variance Extracted)
values ranged from .635 to .725 for the six PA1 subscales. Furthermore, no construct
correlation between any two subscales was within two standard errors of unity
(Anderson & Gerbing, 1988), nor did any squared correlation exceed the AVE value
of any referent subscale (Fornell & Larker, 1981), thus indicating discriminant
validity (Table 2). The internal consistency and the convergent validity measures for
the three dependent scales were adequate as well (Table 3). The alpha coefficients
ranged from .71 to .87 and the AVE values ranged from .48 to .75.
Points of Attachment 261
Table 1: Factor Loadings ($,) Average Variance Extracted Values (AVE), Alpha Coefficients
(a), Means (M) and Standard Deviations (SD) for the Points of Attachment Index.
Item AVE a M SD
Attachment to the team ,701 ,870 4.56 1.37
I consider myself to be a "real" fan of the basketball team .826 4.93 1.54
I would experience a loss if I had to stop being a fan of .801 4.15 1.63
the team
Being a fan of the basketball team is very important to me .883 4.56 1.47
-- --- --
Item fl AVE a M SD
BIRGing .686 .87 4.54 1.23
I would like to let others know about my association with .804 4.74 1.31
this team
I would like to publicize my connection with this team .83 1 4.34 1.41
I would like to tell others about my association with this team .849 4.62 1.33
Satisfaction .478 .71 4.90 1.05
I was satisfied with my decision to attend .705 5.21 1.27
I was satisfied with my spectating experience .732 4.90 1.31
I was satisfied with the game experience .633 4.68 1.24
Conative loyalty .754 .84 4.43 1.23
Iam more likely to attend future games .869 4.90 1.45
Iam more likely to purchase the team's merchandise .905 4.26 1.32
I am more likely to buy (team name) clothing .829 4.11 1.38
Number of home games attended 3.89 2.62
Results of the first multiple regression analysis (Table 4), indicated that the first block
(attachment to the team) explained a significant amount of the variance in BIRGing
(R2 = .430). The second block (attachment to the university and attachment to the
sport) explained a significant, although small, amount of additional variance (AR2
= .026). The third block (attachment to the player, attachment to the level of sport,
and attachment to the coach) also explained a significant, albeit small, amount of
additional variance in BIRGing, (AR2= -019).
Points of Attachment 263
In the second multiple regression analysis (Table 5), the first block (attachmentto the
team) explained a significant amount of the variance in satisfaction (R2= .263). The
second block (attachmentto the university and attachment to the sport) explained a
significant, but small, amount of additional variance (AR2= .036). The third block
(attachment to the player, attachment to the level of sport, and attachment to the
coach) did not explain a significant amount of additional variance in satisfaction.
In the third multiple regression analysis, the first block (attachmentto team)
explained a significant amount of the variance in conations (Table 6; R2= .340). The
second block (attachment to the university and attachment to the sport) explained
a significant, although small, amount of additional variance (AR2= -019).The third
block (attachment to the player, attachment to the level of sport, and attachment
to the coach) also explained a significant, although small, amount of additional
variance in conations, (AR2= .024).
In the fourth multiple regression analysis, the first block (attachment to the
team) explained a significant amount of the variance in attendance (Table 4; R2 =
.241). Neither the second block (attachmentto the university and attachment to the
sport) nor the third block (attachmentto the player, attachment to the level of sport,
and attachment to the coach) explained a significant amount of additional variance
in attendance.
264 Kwon, Trail &Anderson
Team
Step 2
Team
University
sport
Step 3
Team
University
sport
Player
Level
Coach
Discussion
The primary focus of this study was to examine the validity and reliability ofthe Points
of Attachment Index subscales. Specifically, we wanted to determine whether fewer
subscales of the PA1 would satisfactorily predict cognitive, affective, conative, and
behavioural dimensions of sport spectatorship. The PA1 had reasonable fit statistics
and the six dimensions indicated good convergent and discriminant validity. Thus,
solely based on the psychometric properties of the PAI, there is no reason to reduce
the number of dimensions.
The attachment to the team dimension explained a significant and large
amount of variance in all four sport spectator aspects (cognition, affect, conation,
and behaviour) ranging from 24% in attendance to 43% in BIRGing. The addition
of the attachment to the university and the attachment to the sport dimensions of the
PA1 to the regression equation contributed significant,but small, amounts of variance
to BIRGing, satisfaction, and conative loyalty, but not to attendance behaviour;
however the effect was due solely to attachment to the university, as attachment to
the sport was never a significanat predictor. Considering that the largest amount of
additional variance explained was only 3.6%(in Satisfaction),one might question the
added value of the above two subscales. Furthermore, when the final three subscales
(attachment to theplayers, level of sport, and coach) were entered into the equation,
266 Kwon, Trail &Anderson
team, with the university, and with the particular sport were significant determinants
of purchasing and wearing behaviours of licensed sport-apparel. It also peripherally
supported the findings of Robinson and Trail (2005) that identification with team,
sport, and university are perhaps the primary aspects of points of attachment.
However, in both of these cases, team identification was the primary component
andlor predictor.
There has been no research as far as we know that has looked at all of
the points of attachment together and the relationship to different aspects of sport
spectatorship or sport fandom. Thus, our results cannot be compared to any previous
research in this regard. Regardless, a small but significant amount of additional
variance was explained in only two of the dependent variables.
Since at least four of the six-dimensions of the PA1 are reliable and valid,
the issue of the usefulness (predictive ability) of multiple dimensions becomes
important. Two arguments could be made here. The first being that the amount of
additional variance explained by the combination of identificationwith the university
and identification with the sport explained barely more than a "tivial" amount of
variance in these four dependent variables. Furthermore, when the additional three
dimensions were added they were not significant in half of the dependent variables.
Thus, including multiple dimensions of the PA1 in a survey may not be worth the
extra time necessary for the respondents to answer the additional 6 to 15 items in
certain situations. In addition, a more parsimonious model (i.e., fewer dimensions
-probably the four that have been shown here to be predictive) may be easier to
interpret and evaluate for researchers. Practitioners would benefit because of ease
of data collection and applicability.
The counter argument is this. There is theoretical value in, and statistical
support for, a multidimensional model. Just because the increase in variance
explained in the dependent variables is small, it does not mitigate the value of
the added prediction, especially considering the correlations between the different
points of attachment. Three of five dimensions combined (i.e., other than attachment
to a team) did explain a significant amount of additional variance (up to 4.5%)
in cognitive, affective, and conative aspects of sport spectatorship. Furthermore,
fewer dimensions may not be appropriate in all situations. Each sport organisation
has its own unique situation. Some organisations may have a great star player and
thus the player attachment subscale might be more applicable in those situations
and might explain more variance in cognitive, affective, conative and behavioural
loyalty. Some organisations have a great coach and thus the attachment to the
coach subscale would be applicable in those situations. In the present data set, the
university analysed did not have an extremely popular player or a great coach. This
could have had, and probably did have, a significant impact on the relevance of
those two subscales to the respondents.
268 Kwon, Trail & Anderson
Recommendationsand Summary
In sum, the reliability and construct validity of the six-dimension PA1 was
encouraging. The attachment to the team subscale explained a significant and
meaningful amount of variance in BIRGing, satisfaction, conative loyalty, and
attendance behaviour. Three of the other subscales significantly explained only
a minimal amount of additional variance. However, academics and practitioners
should be careful in eliminating the other points of attachment until fbther research
is done. There may be instances in which the other attachment subscales, especially
attachment to university or to a specific sport, might be applicable. The same may
apply to attachment to player or coach in specific instances. Further research should
assist in these determinations.
Points of Attachment 269
References
Anderson, J. C. & Gerbing, D. W. (1988). Structural equation modeling in practice: Areview
and recommended two-step approach. Psychological Bulletin, 103(3), 41 1-423.
Bagozzi, R. P., & Yi, Y. (1988). On the evaluation of structural equation models. Journal of
the Academy of Marketing Science, 16,7494.
Browne, M. W., & Cudeck, R. (1992). Alternative ways of assessing model fit. Sociological
Methods and Research, 21,230-258.
Cohen, J. (1 988). Statisticalpoweranalysisfor the behavioral sciences (2nd ed.). Hillsdale,
NJ: Lawrence Erlbaum.
Fomell, C., & Larcker, D. F. (198 1).Evaluatingstructuralequation models withunobservable
variables and measurement error. Journal of Marketing Research, 18,39-50.
Funk, D. C., Mahony, D. F., Nakazawa, M., & Hirakawa, S. (2000). Spectator motives:
Differentiating among objects of attraction in professional football. European Journal
for Sport Management, 7,5 1-67.
Funk, D. C., Mahony, D. F., Nakazawa, M., & Hirakawa, S. (2001). Development of the
Sport Interest Inventory (SII): Implications for measuring unique consumer motives
at team sporting events. International Journal of Sports Marketing & Sponsorship,
3,291-316.
Funk, D. C., Mahony, D. F., & Ridinger, L. L. (2002). Characterizing consumer motivation
as individual difference factors: Augmenting the sport interest inventory (SII) to
explain level of sport. Sport Marketing Quarterly, 11,3343.
Hu, L., & Bentler, P. M. (1999). Cutoff criteria for fit indexes in covariance structure
analysis: Conventional criteria versus new alternatives. Structural Equation Modeling,
6, 1-55.
James, J. D. & Trail, G.T. (in press). The relevance of team identification to sport
consumption behavior intentions. International Sports Journal, 9(1).
Kwon, H., & Armstrong, K. L. (2004). An exploration of the construct of psychological
attachment to a sport team among college students: A multi-dimensional approach.
Sport Marketing Quarterly, 1 3 , 9 4 103.
Kwon, H. H., Anderson, D. A., & Trail, G. T. (2003, June). The relationship between
sport consumers' points of attachment and licensed merchandise consumption.
Paper presented at the Annual Conference of the North American Society of Sport
Management, Ithaca, New York.
Laverie, D. A., & Arnett, D. B. (2000). Factors affecting fan attendance: The influence of
identity salience and satisfaction. Journal of Leisure Research, 32,225-246.
Madrigal, R. (1995). Cognitive and affective determinants of fan satisfaction with sporting
event attendance. Journal of Leisure Research, 27,205-227.
Mahony, D. F., Madrigal, R., & Howard, D. (2000). Using the Psychological Commitment
to Team (PCT) scale to segment sport consumers based on loyalty. Sport Marketing
Quarterly, 9, 15-25.
Matsuoka, H., Chelladurai, P., & Harada, M. (2003). Direct and interaction effects of
team identification and satisfaction on intention to attend games. Sport Marketing
Quarterly, 12,244-253.
270 Kwon, Trail & Anderson
Melnick, M. J., & Wann, D. L. (2004). Sport fandom influences, interests and behaviors
among Norwegian University Students. International Sports Journal, 8 (Winter),
1-13.
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, 1 0 5 , 4 3 W 5 .
Nunally, J. C., & Bemstein, I. H. (1994). Psychometric theory (3rd ed.). New York,
McGraw-Hill.
Robinson, M., & Trail, G. T. (2005). Relationships among spectator gender, motives and
points of attachment in selected intercollegiate sports. Journal of Sport Management,
19,58-80.
Robinson, M., Trail, G. T. Dick, R., & Gillentine,A. (2005). Fans vs. spectators:An analysis
of those who attend intercollegiate football games. Sport Marketing Quarterly. 14,
43-53.
Robinson, M., Trail, G. T., & Kwon, H. (2004). Motives and points of attachment of
professional golf spectators. Sport Management Review, 7, 161-1 85.
Steiger, J. H. (1989). EZPATH: A supplementaly module for SYSTAT and SYGRAPH.
Evanston, IL: SYSTAT.
Stryker, S. (1968). Identity salience and role performance: the relevance of symbolic
interaction theory for family research. Journal of Marriage and the Family, 30,
558-564.
Stryker, S. (1980). Symbolic interactionism: A social structure version. Menlo Park, CA:
BenjaminICummings.
Stryker, S. (1994). Identity theory: Its development, research base, and prospects. Studies
in Symbolic Interaction, 16,9-20.
Trail, G. T., Anderson, D. F., & Fink, J. S. (2003, November). Consumer satisfaction and
identity theory: A model of sport spectator conative loyalty. Paper presented at the
Inaugural Conference of The Sport Marketing Association, Gainesville, Florida.
Trail, G. T., Anderson, D. F., & Fink, J. S. (2005). Consumer satisfaction and identity
theory: A model of sport spectator conative loyalty. Sport Marketing Quarterly 14,
98-1 12.
Trail, G. T., Fink, J. S., & Anderson, D. F. (2003). Sport spectator consumption behavior.
Sport Marketing Quarterly, 12,8-17.
Trail, G. T., & James, J. D. (2001). The motivation scale for sport consumption: Assessment
of the scale's psychometric properties. Journal of Sport Behavior, 24, 108-127.
Trail, G. T., Robinson, M. J., Gillentine, A., & Dick, R. (2003). Motives and points of
attachment: Fans versus spectators in intercollegiate athletics. Sport Marketing
Quarterly, 12,2 17-227.
Wann, D. L. (2002). Preliminary validation for a measure of assessing identification
as a sports fan: The sports fandom questionnaire. International Journal of Sports
Management, 3,103-1 15.
Wann, D. L., & Branscombe, N. R. (1993). Sports fan: Measuring degree of identification
with their team. International Journal of Sport Psychology, 24, 1-17.