Anda di halaman 1dari 11

ARTICLE IN PRESS

Tourism Management 29 (2008) 1141–1151


www.elsevier.com/locate/tourman

An empirical structural model of tourists and places: Progressing


involvement and place attachment into tourism
Michael J. Gross, Graham Brown
School of Management, University of South Australia, Elton Mayo Building, City West Campus, North Terrace, Adelaide, SA 5000, Australia
Received 13 December 2006; accepted 17 February 2008

Abstract

Research which examines the relationship between involvement and place attachment has started to emerge in the leisure literature and
this paper reports the findings of a study, which applied these constructs within a tourism context. A survey was conducted with tourists
in five South Australian tourism regions that measured respondents’ attitudes towards tourism experiences in South Australia.
Involvement was conceptualised as a multidimensional construct consisting of the dimensions of attraction, centrality to lifestyle,
self expression and food & wine. Place attachment was also conceptualised as a multidimensional construct of place dependence and
place identity. The study developed and tested a structural model that was found to reliably and validly measure predictive relationships
between the constructs of involvement and place attachment. The findings indicated that the combined use of involvement and place
attachment is applicable in tourism. The implications for destination management are discussed.
r 2008 Elsevier Ltd. All rights reserved.

Keywords: Involvement; Place attachment; Tourism experiences; Structural equation modelling

1. Introduction activity or product and pleasure derived from participation or


use (Funk, Ridinger, & Moorman, 2004). The second is self
Tourism knowledge has advanced by applying theories expression, the degree to which participants express their self
developed in other disciplines and so it is surprising that concept or individuality through the situation or object of
some research streams have been developed, tested and study. Participation acts as a message of who the subjects are,
widely reported in leisure journals without attracting the and the situation or object acts as a vehicle through which
interest of tourism researchers. This is characteristic of the one projects and enhances one’s self image (Selin & Howard,
situation that exists with regard to research about 1988). The third is centrality to lifestyle, the extent to which
involvement, place attachment and, more recently, the participants’ social networks revolve around an activity, and
combined use of the two constructs. whether participants’ lifestyles are meaningfully impacted by
Involvement was developed in consumer behaviour and their participation (Havitz, Dimanche, & Bogle, 1994). Place
can be defined as: the perceived personal importance attachment was first developed in environmental psychology
and/or interest consumers attach to the acquisition, and is conceived as an affective bond or link between people
consumption, and disposition of a good, service, or an and specific places (Hidalgo & Hernandez, 2001). The place
idea (Mowen & Minor 1998, p. 64). When applied in leisure attachment construct typically consists of two dimensions: the
research, it has often consisted of three dimensions: the first first is place dependence, which refers to a functional
is attraction, conceived as the perceived importance of an attachment to a place, and the second is place identity, which
refers to a symbolic or affective attachment to a place
Corresponding author. Tel.: +61 8 8302 0350; fax: +61 8 8302 0512. (Backlund & Williams, 2003).
E-mail addresses: michael.gross@unisa.edu.au (M.J. Gross), The use of the place attachment and involvement
graham.brown@unisa.edu.au (G. Brown). constructs in combination has occurred only recently in

0261-5177/$ - see front matter r 2008 Elsevier Ltd. All rights reserved.
doi:10.1016/j.tourman.2008.02.009
ARTICLE IN PRESS
1142 M.J. Gross, G. Brown / Tourism Management 29 (2008) 1141–1151

leisure studies, and in the context of recreation. The pioneers camping, hiking, and climbing. Leisure tourists in the same
in this area have been Kyle and his co-scholars, who studied setting may seek more extrinsic recreational facilities and
involvement (Kerstetter, Confer, & Graefe, 2001; Kyle & would be more likely to access and pay for infrastructure
Chick, 2002; Kyle, Kerstetter, & Guadagnolo, 1999, 2002; and enhancements of the attraction such as accommoda-
Scott & Shafer, 2001) and place attachment (Kyle, Absher, tion, food and beverage outlets and other man-made
& Graefe, 2003b; Moore & Graefe, 1994; Moore & Scott, attractions. An understanding of the relationship in
2003) on separate research tracks until combining them in a different contexts is important in order to be able to
key 2003 study that measured the relationships between determine where recreation ends and tourism begins
leisure activity involvement and place attachment among (McKercher, 1996). The extent to which people exhibit
hikers on the Appalachian Trail (Kyle, Graefe, Manning, & different levels of involvement and place attachment with
Bacon, 2003a). The research reported in their paper sought destinations as the settings for activities may be valuable in
to examine the utility of combining involvement and place this determination. The body of literature combining
attachment in tourism. This is important to determine involvement and place attachment is described below.
because it cannot be assumed that factors associated with
recreational activities will be equally relevant in tourism
contexts where variables associated with the destination 2. Literature review
attributes may be more influential.
Considerations of the corresponding and contrasting The combined use of involvement and place attachment
qualities of activities undertaken in recreation and tourism was first used by Williams, Patterson, Roggenbuck, and
contexts have engaged the interest of the respective Watson (1992) in their study of visitors to four American
research communities. In the 1980s the work of Mieczkowski wilderness areas. They examined the emotional/symbolic
(1981) and Murphy (1985) helped to establish frameworks view of recreation places through an analysis of the
for understanding how tourism co-exists situationally with relationships between use history of the place, perceived
leisure and recreation. Their conceptual maps placed substitutability, sociodemographic and trip characteristics,
recreation entirely within the leisure domain, while tour- and sensitivity to wilderness impacts and levels of attach-
ism, due to its relationship with business travel, overlapped ment. Citing the specialisation literature as a source, they
with and extended beyond recreation and leisure (Fennell, incorporated some of Wellman, Roggenbuck, and Smith’s
2002; Poria, Butler, & Airey, 2003; Smith & Godbey, 1991; (1982) scale items into a wilderness attachment scale. They
Tribe, 1995). Hall and Page (1999), supported by Williams found that stronger place and wilderness attachment was
(2003), modified the model, extending recreation beyond associated with previous visits, rural residence, a setting
leisure in recognition of Stebbins’ (1982) work on serious focus, visiting alone, visiting on weekdays, hunting in the
leisure (Fig. 1). area, and sensitivity to site impacts and horse encounters.
While definitional debates of specific areas continue, a Place attachment was associated with a lack of non-
consensus has formed around at least one aspect of the wilderness substitutes, lower income, and lower education.
relationship: leisure travel undertaken for recreation Wilderness attachment was associated with membership in
purposes is a form of tourism, and is distinct from leisure wilderness and conservation organisations, visits to more
tourism. Taking the example of a park setting, recreational wilderness areas, a preference for longer visits, participa-
travellers may seek the intrinsic values of the park, and tion in nature study, and sensitivity to sight and sound
their behaviour will reflect those values through engage- intrusions in hiker encounters.
ment with the natural environment in such activities as Moore and Graefe (1994) used the conceptual frame-
works of activity specialisation and place attachment to
study recreation trail users in three American states,
Work
hypothesising that place identity with a particular setting
was a function of how long a person had been associated
Leisure with the site and how dependent they were on it (which was
itself a function of frequency of site visit). They found
predictive relationships that were moderated by frequency
Tourism of use, specifically that place identity was best predicted by
Business Leisure Recreation how long users had been associated with the trail, the
travel travel importance they ascribed to their trail activity, and their
level of place dependence. Level of place dependence was
best predicted by the distance between the trail and the
Serious
leisure
user’s home and user’s frequency of trail use. Their study
represented the first combined use of involvement and
place attachment in the leisure literature, although only the
Fig. 1. Relationship between leisure, recreation, and tourism (Source: Hall importance dimension of involvement was used. Graefe
& Page, 1999, p. 4). later extended this work in a conference paper with
ARTICLE IN PRESS
M.J. Gross, G. Brown / Tourism Management 29 (2008) 1141–1151 1143

colleagues (Mowen, Graefe, & Virden, 1997) that com-


bined place attachment and activity involvement to create Attraction
a typology to understand visitor evaluations of recreation
experiences in a national recreation area in Virginia. They Place
Dependence
found that as attachment increased in intensity, evaluations
of setting and experiences were more positive. Their study Self Expression
represented the first combined use of the two dimensions of
place attachment (place dependence and place identity) Place
with the McIntyre (1989) modified CIP scale using the Identity
involvement dimensions of attraction, self expression and
Centrality
centrality to lifestyle.
In a 1998 study seeking to determine which approach
was most appropriate for segmenting trail users within a Fig. 2. Hypothesised structural model (Source: Kyle et al., 2003a).
particular geographic context, three segmentation ap-
proaches of activity type (hikers, bicyclists and horse
users), trail type (rail-trail and non-rail-trail), or an
interaction of activity type and trail type were tested on place attachment. The structural model showing the
(Mowen, Graefe, & Williams, 1998). The same dimensions hypothetical relationships is shown in Fig. 2. The involve-
of involvement and place attachment as in the Mowen et al. ment construct is represented by the dimensions of
(1997) study were employed and it was found that attraction, self expression, and centrality, and the place
differences in socio-demographic characteristics, visit attachment construct is represented by the dimensions of
patterns and visitor attitudes were best explained by place dependence and place identity.
activity type, and that horse users spent more time in their The findings indicated that the place identity dimension
activity than bicyclists and held the highest levels of activity was best predicted by the self expression and attraction
involvement and place attachment. dimensions, whereas the only predictor of place dependence
Bricker and Kerstetter (2000) studied whitewater recrea- was self expression. This was a significant study, as the
tionists, using involvement to measure levels of specialisa- authors’ structural modelling analysis has provided the
tion and levels of place attachment to a California river. basis for a number of subsequent studies along similar lines
A lifestyle dimension was added to place dependence and of enquiry, all of which have contributed insights into the
place identity as part of a three-dimensional place attach- underlying motivations for recreationists’ engagement in
ment construct. The study also included a centrality to specific leisure pursuits and visitation to specific recreation
lifestyle dimension as part of an involvement construct, settings. The structural model used in their study provided
which was measured by behavioural items of memberships the basis for the method and form of analysis employed in
to paddling clubs, subscriptions to whitewater magazines, the present study which sought to examine whether
and books related to whitewater sports. A relationship was respondents undertaking tourism-specific activities would
noted between dimension levels of specialisation and yield comparable results to those associated with recrea-
place attachment, as the authors found that centrality to tion-specific activities and settings such as hiking on trails.
lifestyle was one of five specialisation dimensions that Kyle and Graefe collaborated with Bricker and
confirmed their hypothesis that individuals possess a Wickham the following year (Kyle, Bricker, Graefe, &
range of specialisation. The place attachment lifestyle Wickham, 2004a) using variations of the same model to
dimension was measured by three items suggesting a deep further examine the relationship between involvement and
sense of attachment specific to the place and connected place attachment for hikers along the Appalachian Trail,
to an individual’s lifestyle. They contended in their boaters on a California river and anglers in New England.
findings that the lifestyle dimension constituted a viable Building on past findings that had shown that the potential
third dimension of place attachment, although acknow- for variation among activity and setting types was high,
ledging that it was not important for most whitewater they sought to compare the three samples to determine
recreationists. their respective performances in a predictive model. Again
Kyle et al. (2003a) conducted a key study of hikers on testing a model that suggested involvement as an ante-
the Appalachian Trail that included the use of a structural cedent of place attachment, they hypothesised that each
modelling method to examine place attachment’s dimension of involvement would have a positive and
antecedent processes that lead to recreationists’ attachment significant effect on the dimensions of place attachment.
to settings. It was hypothesised that hikers’ type of use The results illustrated that recreationists’ relationships with
(day use, overnight use, sectional use and hiking entire activities and settings indeed did vary among the three
length of the trail) would impact the relationship between sample groups. The overall results suggested that leisure
activity involvement and place attachment such that, as the involvement and the centrality dimension, in particular,
length of hikers’ visit along the trail increased, so too were a better predictor of the place identity dimension of
would be the strength of the effect of activity involvement place attachment than of place dependence.
ARTICLE IN PRESS
1144 M.J. Gross, G. Brown / Tourism Management 29 (2008) 1141–1151

A sample of Appalachian Trail hikers was used to The studies described above support the value of
examine place attachment’s effect on several variables combining involvement and place attachment as a method
purportedly related to the construct (Kyle, Graefe, & to examine the degree to which involvement in tourism
Manning, 2004c). The hypothesised direction of influence may be influenced by the places where the experiences are
was reversed, using the place attachment dimensions as gained. As the studies examined recreation activities and as
independent variables. Respondents were clustered into a predictive relationship between involvement and place
three groups by place dependence and place identity mean attachment has been established in leisure and recreation
scores, then the three groups were examined in relation to contexts, the purpose of the present study was to determine
the variables of demographics, visit motivations and if the relationship could be established in a tourism context
preferences, and activity involvement. Rather than analys- with tourism activities and settings.
ing the relationships using their 2003a structural model,
they used a combination of confirmatory factor analysis 3. Research method
(CFA), cluster analysis and one-way analysis of variance to
test the relationships. Variations among groups based on The objective of this study was to examine the utility of
level of attachment were observed for trip motivations, combining involvement and place attachment to measure
evaluations of setting conditions, evaluations of manage- destination-specific tourism experiences. Five locations in
ment actions, level of experience and level of involvement. South Australia were selected as the settings for the study.
This study served as further confirmation of the predictive Hypotheses for the measurement model were proposed (see
validity and reliability of the two-dimensional conceptua- Gross, Brien, & Brown, 2006) and a questionnaire, based
lisation of place attachment. on prior research from the involvement and place attach-
Kyle, Graefe, Manning, and Bacon (2004b) used a ment literature, was developed. It consisted of multiple-
reduced portion of their 2003a sample of Appalachian item scales using a five point Likert-type response format
Trail hikers to look at the effect of activity involvement (1 ¼ Strongly Disagree to 5 ¼ Strongly Agree). The
and place attachment on recreationists’ perceptions of rationale for the selection and number of items comprising
setting density. Using frameworks offered by social each dimension followed Kyle et al.’s (2003a) study, which
judgment and cognitive development theories, they had proved to be reliable and valid. The first section of the
hypothesised that hikers’ evaluations of setting density questionnaire was designed to measure the consumer
along the trail would be influenced by their degree of involvement construct for the dimensions of attraction,
activity involvement and place attachment. The two- centrality to lifestyle, and self expression. The second
dimensional place attachment construct was expanded to section was designed to measure the place attachment
include a third dimension of social bonding. The structural construct of place identity and place dependence. The third
model results indicated that only place identity and place section was designed to measure destination attributes.
dependence were significant predictors of respondents’ South Australia is strongly identified with food & wine and
perceptions of place density. While respondents scoring these attributes featured prominently in the pilot study
high on the place identity dimension were more inclined to (Gross, 2005).
report feeling crowded, respondents scoring high on the The main survey was conducted from November 2004
place dependence dimension were inclined to evaluate through May 2005 and 476 questionnaires were completed.
setting density more favourably. The target population consisted of adult tourists, staying at
A study of visitors to a national park in Taiwan, found least one night away from home and attending a Visitor
that both involvement and place attachment had positive Information Centre (VIC) or an Attraction (e.g. wine cellar
effects on perceived service quality and satisfaction door or resort) in one of five tourism regions of South
(Hwang, Lee, & Chen, 2005). This has been the only Australia; Adelaide, Barossa, Flinders and Outback,
involvement and place attachment combination study to Kangaroo Island, or McLaren Vale/Fleurieu.
include a specific tourism element, using a sample frame of Exploratory factor analysis (EFA) of the data generated
tourists and ‘visiting a national park’ as the involvement six dimensions. Tourism involvement was conceptualised
attitude object (personal communication, 2005). Building as a multidimensional construct consisting of attraction,
on Bricker and Kerstetter’s (2000) analysis, the authors centrality to lifestyle, and self expression. Place attachment
structured a similar instrument and, although it was not was also conceptualised as a multidimensional construct
reported, centrality did not survive the test of statistical consisting of place identity and place dependence. Food &
significance as one of the four involvement dimensions of wine was included as the destination-specific measure.
importance/pleasure, self expression/sign, risk probability, CFA was used to develop and test a measurement model
and risk consequence in the resulting theoretical structural for the six dimensions contained within the constructs of
model. Similarly, their findings did not support the lifestyle involvement and place attachment, substantiating that the
dimension found by Bricker and Kerstetter (2000) as part survey instrument resulted in a summated scale that reliably
of place attachment, although the other two place and validly measured the six separate dimensions. Internal
attachment dimensions of place dependence and place consistency of the scale’s dimensionality was established
identity were confirmed. through tests of reliability and validity (content, convergent,
ARTICLE IN PRESS
M.J. Gross, G. Brown / Tourism Management 29 (2008) 1141–1151 1145

err
7 q7centr
causal relations, it does show whether the causal assump-
err
6
err
q6centr 0.83
0.78
tions embedded in a model match a sample of data (Bollen,
12 q12centr
0.38
0.37 err q11centr
0.47
0.59
1989, p. 4). Since Reisinger and Turner (1999) noted that
11
0.27
0.51 err
10 q10centr
0.64
0.50
Centrality to SEM had not yet been applied widely in the tourism
err Lifestyle
15
err
q15centr
0.80 discipline, its use in recent years in establishing the predictive
0.21 8 q8centr
err q14centr
0.53
nature of conceptual relationships has been increasing in
14 0.74
tourism contexts (Gursoy, Jurowski, & Uysal, 2002; Gursoy
err q5attrac
5
err q1attrac
0.80
0.77
& Rutherford, 2004; Kang, Jeon, Lee, & Lee, 2005; Lee &
1
err
2 q2attrac
0.76
Back, 2006; Quan & Wang, 2004; Ryan & Huyton, 2000;
0.78
Attraction 0.67
err
3 q3attrac
0.63 Swanson & Horridge, 2004; Yoon, Gursoy, & Chen, 2001;
err q4attrac
4
err
0.58
Yoon & Uysal, 2005; Zins, 2002). Generally, a structural
9 q9centr

0.09
equation model is a complex composite statistical hypoth-
0.51
err
19 q19expre
0.76
0.49
esis. It consists of two main parts: the measurement model
err q18expre
18
err q20expre
0.72
0.68
0.29 that represents a set of observable variables as multiple
20 Self 0.07
err
17 q17expre
0.65
Expression indicators of a smaller set of latent variables, and the path
0.67 0.37
0.29 err
16 q16expre
0.61 model that describes relationships of dependency between
err q33lifst 0.08
33
the latent variables (McDonald & Ho, 2002). In this study,
err 0.18
35
err
q35lifst 0.79
0.81 0.26
the observable variables were represented by the scale items
q34lifst
34 Food &
-0.75 err
31 q31lifst
0.80
Wine 0.26 generated by the survey of tourists, and the latent variables
0.51
err
30 q30lifst were represented by the four involvement dimensions of
err
26 q26depen 0.94 0.31 attraction, centrality to lifestyle, self expression, and food &
0.24
err
27 q27depen 0.90
0.86
Place wine, and the two place attachment dimensions of place
err

0.47
25
err
q25depen
0.72
Dependence
0.23 dependence and place identity. The term structural model
28 q28depen
is reserved in this study for the composite structural
err q21ident 0.86
21
err q22ident 0.91
equation model, i.e. the combined measurement and path
22 Place 0.68
err
23 q23ident
0.85
Identity models.
0.42
err
29 q29lifst SEM was performed using the AMOS 5 statistical
program on the 32 scale items represented by the
Fig. 3. Measurement model (Source: Gross et al., 2006).
involvement and place attachment constructs. Maximum
likelihood (ML) estimation was used to estimate all
models. ML is regarded as considerably more robust than
and discriminant). The measurement model showing other estimation methods such as generalised least squares
coefficients in standardised form is presented in Fig. 3, or weighted least squares (Olsson, Foss, Troye, & Howell,
where ovals represent latent variables, rectangles represent 2000), given a sufficiently large sample size of 200 or more
observed variables, and circles represent measurement error (Tabachnick & Fidell, 2001, p. 74). Another guideline for
associated with observed variables. A six-factor measure- stable ML estimation is that sample size should have a
ment model of centrality to lifestyle (8 items), attraction ratio of at least 10:1 or 15:1 to the number of observed
(6 items), self expression (6 items), food & wine (4 items), variables (Thompson, 2000). These conditions for sample
place dependence (4 items), and place identity (4 items) was size were satisfied for the 32-item structural model with a
developed. A list of the survey scale items is provided in sample size of 476. The path diagram used to illustrate the
Appendix A. relationship between the two constructs of involvement
and place attachment was based on the Kyle et al. model
3.1. Structural equation modelling shown in Fig. 2.
The model fitting process followed the approach
Having established a reliable and valid measurement advocated by Byrne (2001, pp. 79–88) to assess model fit.
model in the preceding work, the next logical step in the The method used for examining the relationships between
present study to test the predictive relationship between latent variables was inspection of the standardised coeffi-
involvement and place attachment in a tourism context was cients for the regression paths (Tabachnick & Fidell, 2001,
to proceed to the full structural model. Structural equation p. 673). The statistical significance of parameter estimates
modelling (SEM) is a method of multivariate statistical for the structural model was examined through the critical
analysis that is simultaneously capable of measuring the ratio (c.r.) test statistic. That is, the c.r.’s for paths between
concepts generated by EFA, and the paths of the the latent constructs in the structural model should fall
hypothesised relationships between the concepts (Klem, outside the threshold range of 71.96. Non-significant
2000). It is a statistical inference method that tests the parameters, with the exception of error variances, can be
assumption that parameters are not just descriptive mea- considered unimportant to the model, and in the interest of
sures of association, but rather that they reveal a predictive scientific parsimony, they should be deleted from the model
relationship. While the method does not necessarily discover (Byrne 2001, p. 76).
ARTICLE IN PRESS
1146 M.J. Gross, G. Brown / Tourism Management 29 (2008) 1141–1151

3.2. Hypothesised structural model improvement in fit between the independence model and the
hypothesised model.
Consistent with the study objective, and based on the Post hoc model modifications were undertaken in an
CFA findings, one a priori hypothesis was proposed for attempt to develop a better fitting, more parsimonious
the structural model. The hypothesis was drawn from structural model. All c.r.’s for regression weights, covar-
a combination of theoretical, practical, and methodological iances and variances for the structural model fell within the
considerations as described below. acceptable range, with the exception of three regression
The hypothesis proposed for the structural model was paths: attraction-place identity, self expression-place
that (H1) each dimension of involvement would have a dependence, and self expression-place identity. These three
positive and significant effect on each dimension of regression paths were therefore deleted.
place attachment. In other words, the four dimensions In an effort to identify further areas of misfit, the
of centrality to lifestyle (8 items), attraction (6 items), standardised residuals and modification indices (MI) were
self expression (6 items), food & wine (4 items) would examined. An examination of the standardised residuals
positively predict the two dimensions of place dependence identified one possible item that was a candidate for either
(4 items), and place identity (4 items). The involvement and respecification or deletion. This was item 30 ‘‘The
place attachment literature in a leisure and recreation distinctive food of the region is something that attracted
context, primarily represented by Kyle et al., has examined me here’’, which displayed multiple covariance discrepan-
the relationship between involvement and place attachment cies well exceeding a threshold value of 72.58, which is
using the involvement constructs as predictor variables considered to be large (Byrne, 2001, p. 89). The MI were
in structural models (Kyle et al., 2003a, 2004a; Kyle, examined to determine what, if any, action should be taken
Graefe, Manning, & Bacon, 2004d; Kyle & Mowen, 2005). on item 30. Examination of the regression weights
These studies have produced findings that suggest, in suggested that substantial parameter improvements
the present study context, that as respondents’ scores on (MI ¼ 51.27, par change ¼ .660) could be made by deleting
the dimensions of involvement (attraction, centrality to item 30 from the food & wine factor, and the item was
lifestyle, self expression and food & wine) increase, so deleted. Deletion of this item also had the effect of
too will their scores on the dimensions of place attachment removing the correlation of the errors between item 30 in
(place dependence and place identity). If present study the food & wine dimension and item 29 ‘‘The distinctive
findings supported H1, this would suggest that similar lifestyle of the region is something that attracted me here’’,
positive and significant relationships between involvement in the place identity dimension, an error correlation that
and place attachment might exist in a tourism context had been established during development of the measure-
to those that have been reported in the leisure and ment model. Configuration of the hypothesised structural
recreation literature. The four involvement dimensions model had positioned this error correlation as one between
were used as independent (exogenous) variables and the independent and dependent variables; however, the error
two place attachment dimensions were used as dependent correlation would have been removed anyway on theore-
(endogenous) variables in the hypothesised structural tical grounds, as such a correlation is theoretically
model. untenable in a structural model (DeShon, 1998). Turning
attention to the MI covariance section, the MI’s represent-
ing error covariances were examined, and no substantial
4. Results model fit improvements were indicated for allowing any
further error terms to correlate. Following this, model
Analysis of the AMOS output indicated that estimation was considered complete with a final 31-item
the independence model that tests the hypothesis that structural model that was identified and that showed an
all variables were uncorrelated was easily rejectable, acceptable fit with the data.
w2 (chi-square) ¼ 9512 (n ¼ 476, df ¼ 496, po.01). The All of the measures in the final structural model
hypothesised model was tested next and reasonable possessed acceptable psychometric properties. The differ-
support was found for the hypothesised model, w2 ¼ 1291 ence between the initial model and the final model in the
(n ¼ 476, df ¼ 440, po.01), comparative fit index w2 was 114, the difference in the degrees of freedom (DF)
(CFI) ¼ .906; root mean square error of approximation was 26 and po0.01. Therefore, model estimation resulted
(RMSEA) ¼ .064. A w2-difference test indicated a significant in an improved model fit as demonstrated by Table 1.

Table 1
Summary of structural model fit statistics

Model Chi-square DF Chi-square/DF p-Value CFI RMSEA 90% C.I. RMSEA

Initial SEM 1291 440 2.934 .000 .906 .064 (.060, .068)
Final SEM 1177 414 2.843 .000 .912 .062 (.058, .066)
ARTICLE IN PRESS
M.J. Gross, G. Brown / Tourism Management 29 (2008) 1141–1151 1147

4.1. Structural model direct effects All significant paths were positive except attraction-
place dependence, which was negative. The strongest
An effect of model estimation was to remove three predictor of either place attachment dimension was
non-significant regression paths, reducing the number of centrality to lifestyle. As tourism experiences became more
inter-construct relationships for analysis from eight to five. central to respondents’ lifestyle, their functional and
The three non-significant regression paths were: emotional attachment to the tourism regions visited
increased. Food & wine was also a positive predictor of
 attraction-place identity (standardised coefficient .02, both place attachment dimensions. As respondents’ level of
c.r. .254) involvement with the food & wine features of their tourism
 self expression-place dependence (standardised coeffi- experiences increased, their functional and emotional
cient .08, c.r. 1.095) attachment to the tourism regions visited also increased.
 self expression-place identity (standardised coefficient The sole negative relationship between attraction-place
.07, c.r. .907) dependence indicated that as respondents’ attraction levels
to tourism experiences increased, their functional attach-
The remaining five significant regression paths were: ment to the tourism regions visited decreased.

 centrality to lifestyle-place dependence (standardised 4.2. Final structural model


coefficient .48, c.r. 5.780)
 centrality to lifestyle-place identity (standardised coef- The final structural model, showing coefficients in
ficient .36, c.r. 5.934) standardised form, is shown in Fig. 4. The final configura-
 attraction-place dependence (standardised coefficient tion of the latent constructs was of the involvement
.28, c.r. 4.087) construct consisting of the four dimensions of centrality
 food & wine-place dependence (standardised coefficient to lifestyle (8 items), attraction (6 items), self expression
.20, c.r 3.828) (6 items), food & wine (3 items), and the place attachment
 food & wine-place identity (standardised coefficient .19, construct consisting of the two dimensions of place
c.r. 3.331) dependence (4 items), and place identity (4 items).

0.68
err
7 q7centr
0.60
err
6 q6centr 0.83
0.22
err
0.78
12 q12centr 0.47
0.38 0.35
err 0.59
0.37 11 q11centr
0.25
0.51 err 0.50
0.27 10 q10centr
0.41
Centrality to
0.64
err
15 q15centr 0.80
Lifestyle 0.48
0.64
err
0.21 8 q8centr 0.53 0.16
0.52
0.28
err
err
q14centr 0.36 0.72 q28depen 28
14 0.75
0.86
err
0.64 0.73
Place q25depen
0.80
25
0.90
err
5 q5attrac Dependence 0.94 q27depen err
27
0.60 0.80 0.89
err err
1 q1attrac 0.77 q26depen 26
0.57 0.76
err
2 q2attrac 0.78
0.61
err
q3attrac
0.67 Attraction -0.28 Residual
3 0.63
0.40 1
err 0.58
4 q4attrac 0.65
0.34
Residual
err
9 q9centr 2
0.09
0.58 0.49 0.18 0.31
err err
19 q19expre 0.55 q29lifst 29
0.52 0.76 0.72
0.50 err 0.85 err
18 q18expre
0.46
0.72 Place 0.91
q23ident
0.82 23
err
q20expre 0.68 Identity q22ident err
20
0.42 0.06
0.65
Self 0.86
0.74 22
err err
17 q17expre Expression q21ident 21
0.44 0.66
0.29 err
16 q16expre 0.62
0.38
err
33 q33lifst
0.19
0.70 0.20
err
35 q35lifst 0.83
0.64 0.19
err 0.80
q34lifst
34
0.53 0.73
Food &
-0.45 err
31 q31lifst Wine

Fig. 4. Final structural model.


ARTICLE IN PRESS
1148 M.J. Gross, G. Brown / Tourism Management 29 (2008) 1141–1151

5. Discussion as the degree to which participants express their self-


concept or individuality through the situation or object of
The involvement construct consisting of the four study (Selin & Howard, 1988), this meant that respondents
dimensions of centrality to lifestyle (8 items), attraction did not conceive their tourism experiences as channels for
(6 items), self expression (6 items), food & wine (3 items), expressing themselves through either functional or emo-
was hypothesised in the present study to predict the place tional attachment to the five surveyed regions.
attachment construct consisting of the two dimensions
of place dependence (4 items), and place identity (4 items). 5.1. Limitations
One a priori hypothesis was proposed for the structural
model: The model developed in this study relied on data
gathered from visitors in the state of South Australia.
Hypothesis (H1). : Each dimension of involvement would The generalisability of the findings is bounded due to the
have a positive and significant effect on each dimension of geographic limits imposed by South Australia as the sole
place attachment. study site. Future research applying the study method in
other destinations will help to establish the generalisability
This hypothesis was partially supported by the data. of the model.
The broad interpretation of the final structural model is The tourism industry is highly seasonal, and the impact
that, in a regional South Australian context, involvement of seasonal demand variation is one of the dominant policy
in tourism experiences was a positive predictor for both and operational concerns of tourism interests in both the
dimensions of place attachment for the involvement public and private sectors (Baum & Lundtorp, 2001). The
dimensions of centrality to lifestyle and food & wine. As main study sampling of tourists was conducted from
the centrality to lifestyle dimension is conceived as the November 2004 through May 2005, which corresponds to
extent to which participants’ social networks revolve the Australian seasons of summer and autumn. It is
around an activity, and whether participants’ lifestyles possible that different sample profiles would be derived
are meaningfully impacted by their participation (Havitz from other seasons. Further, more longitudinal, data
et al., 1994), this meant that the more the respondents gathering in future research will help to understand what,
perceived tourism experiences as representing a social if any, effect seasonality has on results.
component and playing a central role in their lifestyle, the
greater the functional and emotional attachment they had 6. Conclusions and implications
to that region. Similarly, as the food & wine dimension is
conceived as the relevance that food & wine have in tourism This study has examined whether research that has been
experiences in the places visited, this meant that the more developed in leisure and recreation contexts could be
the respondents perceived food & wine as relevant to their established in a tourism context with tourism activities and
tourism experiences in one of the regions visited, the settings. In so doing, it has attempted to overcome some of
greater the attachments they had to that region. the barriers that exist between the two research commu-
Attraction was a negative predictor for place dependence nities. It has demonstrated the viability of combining
only, and had no statistically significant predictive relation- involvement and place attachment in a tourism context and
ship for place identity. As the attraction dimension is has shown the importance of centrality to lifestyle as a
conceived as the perceived importance of an activity or determinant of tourism outcomes. The model pioneered in
product and pleasure derived from participation or use a recreation context by Kyle et al. (2003a) has been shown
(Funk et al., 2004), this meant that the greater the to be applicable in a tourism context. It has the flexibility to
importance that respondents perceived for tourism experi- accommodate variations in different destinations. Impor-
ences in a particular region, the smaller the functional tantly, it can be used to help understand how tourists
attachment they had to that region, and that they had no respond in different settings and how they evaluate
significant emotional attachment to the region. Based on different dimensions of their destination experience. As
previous studies, the inverse relationship between attraction Kelly (2002) noted, in order to accommodate the increasing
and place dependence was not an unexpected finding; importance of leisure travel in peoples’ lifestyles, the aim
however, the lack of a significant relationship between should be to identify aspects which are unique or
attraction and place identity was unexpected. particularly characteristic of a destination, and then match
Self expression had no statistically significant predictive product, service and experience provision accordingly. This
relationship for either dimension of place attachment. may engage a wide range of destination attributes that can
Based on previous studies, this lack of relationship was be connected with a variety of preferences of individuals or
an unexpected finding. The standardised coefficient groups of tourists.
between self expression and place dependence was .08 These results suggest that the structural model metho-
(non-significant), and the standardised coefficient between dological approach is transferable, and that involvement
self expression and place identity was .07 (also non- and place attachment aspects of tourism can be examined
significant). As the self expression dimension is conceived in a similar way to experiences gained in a recreation
ARTICLE IN PRESS
M.J. Gross, G. Brown / Tourism Management 29 (2008) 1141–1151 1149

context. Tourism researchers have been slow to apply perceptions of the location can then inform marketing
the techniques that have been used by Kyle and his strategies through targeted promotional programs. By
co-scholars. They have performed a service for the natural monitoring changes over time, knowledge of changing
resource research and management community through involvement and place attachment trends can be captured
their recreation-oriented body of work. They have helped and incorporated into strategic planning. Destination
recreation attraction managers to better understand the marketers will benefit more by fully understanding the
nature and intensity of meanings recreationists ascribe to role of place attachment in the consumption behaviour of
specific leisure environments. The model used in their tourists, and this will be assisted by further testing of the
2003a study has yielded understanding in the areas of model in other destinations.
variation among groups of recreationists (hikers, boaters The intent and hope for the present study is that it will
and anglers) (Kyle et al., 2004a), perceptions of setting initiate dialogue and research interest in the tourism
density (Kyle et al., 2004b), and predictors of behavioural community around the importance of the relationship
loyalty (Kyle et al., 2004d). The current study has between involvement and place attachment in tourism
presented evidence that a comparable model can be a contexts. We believe that the potential exists for tourism
useful tool in a similar effort in the service of tourism researchers to extract benefits from the proposed method
destination managers. We argue that the tourism research that will assist the marketing efforts of tourism destination
and management community should take an active interest managers in ways similar to those that recreation
in this approach. Naturally, the research agendas and researchers have been able to achieve for the natural
practical outcomes sought would reflect the specific needs resource management community. The present study
of tourism, and the results of the current study provide represents a beginning of that effort.
methodological encouragement toward those ends.
The type of knowledge generated by SEM analysis may Acknowledgement
have implications for destination marketers for whom it is
critical to be able to distinguish those attitudes that are This research is an outcome of a Ph.D. project funded
substitutable from those that are perceived to be unique to by the Sustainable Tourism Cooperative Research
a particular place. By understanding more about the extent Centre, established by the Australian Commonwealth
to which tourism experiences are rooted in place, tourism Government.
destination managers may be able to determine the role of
involvement dimensions in the marketing of destinations.
For example, identification of a strong involvement in Appendix A. Supplementary materials
food & wine in a South Australian setting in the present
study suggests to local destination managers that infra- Supplementary material associated with this article
structure, development strategies, and marketing efforts can be found in the online version at doi:10.1016/j.
would be well-directed towards acknowledging and culti- tourman.2008.02.009.
vating the high levels of involvement that tourists have
towards food & wine. References
Operators should use their knowledge of respondents’
involvement and place attachment to ensure that the Backlund, E. A., & Williams, D. R. (2003). A quantitative synthesis of
product and service provision of the tourism location are place attachment research: investigating past experience and place
attachment. Paper presented at the Northeastern Recreation Research
aligned with visitors’ expectations, wants, and needs. High Symposium (pp. 320–325). Bolton Landing, New York.
levels of tourist involvement and place attachment Baum, T., & Lundtorp, S. (2001). Seasonality in tourism: An introduction.
constitute an asset for the destination. Operators need to In T. Baum, & S. Lundtorp (Eds.), Seasonality in tourism (pp. 1–4).
ensure that the tourism experiences on offer and delivered Amsterdam: Pergamon.
Bollen, K. A. (1989). Structural equations with latent variables. New York:
are consistent with the recognition and nurturing of that
Wiley.
asset. Bricker, K. S., & Kerstetter, D. L. (2000). Level of specialization and place
Although comparison of results among the five surveyed attachment: An exploratory study of whitewater recreationists. Leisure
regions was outside of the scope of this paper, it would be Sciences, 22(4), 233–257.
expected that levels of respondents’ involvement would Byrne, B. M. (2001). Structural equation modeling with AMOS: Basic
vary by region. For example, respondents’ involvement concepts, applications, and programming. Mahwah, New Jersey:
Lawrence Erlbaum.
levels for a dimension such as food & wine would DeShon, R. P. (1998). A cautionary note on measurement error
reasonably be expected to vary between a wine region corrections in structural equation models. Psychological Methods,
such as Barossa and a nature region such as Flinders & 3(4), 412–423.
Outback. Destination managers and operators should Fennell, D. A. (2002). Recreation and tourism. In Ecotourism programme
periodically (e.g. annually) survey their tourist populations planning (Chapter 1, pp. 1–10). Wallingford, UK: CABI Publishing.
Funk, D. C., Ridinger, L. L., & Moorman, A. M. (2004). Exploring
and use the information to better understand the level of origins of involvement: Understanding the relationship between
interest that tourists have in the product and service consumer motives and involvement with professional sport teams.
provision in their particular location. The consumer Leisure Sciences, 26(1), 35–61.
ARTICLE IN PRESS
1150 M.J. Gross, G. Brown / Tourism Management 29 (2008) 1141–1151

Gross, M. J. (2005). Tourism industry perceptions of lifestyle tourism: an Lee, C. K., & Back, K. J. (2006). Examining structural relationships among
exploratory analysis. Refereed paper presented at the Council for perceived impact, benefit, and support for casino development based on
Australian University Tourism and Hospitality Education (CAUTHE) 4 year longitudinal data. Tourism Management, 27(3), 466–480.
Conference (pp. 1–24). Alice Springs, Northern Territory. McDonald, R. P., & Ho, M. H. R. (2002). Principles and practice in
Gross, M. J., Brien, C., & Brown, G. (2006). Examining the relationship reporting structural equation analyses. Psychological Methods, 7(1),
between tourism involvement and place attachment. Refereed paper 64–82.
presented at the Council for Australian University Tourism and McIntyre, N. (1989). The personal meaning of participation: Enduring
Hospitality Education (CAUTHE) Conference (pp. 486–503). Mel- involvement. Journal of Leisure Research, 21(2), 167–179.
bourne, Victoria. McKercher, B. (1996). Differences between tourism and recreation in
Gursoy, D., Jurowski, C., & Uysal, M. (2002). Resident attitudes: A parks. Annals of Tourism Research, 23(3), 563–575.
structural modeling approach. Annals of Tourism Research, 29(1), Mieczkowski, Z. T. (1981). Some notes on the geography of tourism: a
79–105. comment. Canadian Geographer, 25(2), 186–191.
Gursoy, D., & Rutherford, D. G. (2004). Host attitudes toward tourism: Moore, R. L., & Graefe, A. R. (1994). Attachments to recreation settings:
An improved structural model. Annals of Tourism Research, 31(3), The case of rail-trail users. Leisure Sciences, 16(1), 17–31.
495–516. Moore, R. L., & Scott, D. (2003). Place attachment and context:
Hall, C. M., & Page, S. J. (1999). The geography of tourism and recreation: Comparing a park and a trail within. Forest Science, 49(6), 877–884.
Environment, place and space. London: Routledge. Mowen, A. J., Graefe, A. R., & Virden, R. J. (1997). A typology of place
Havitz, M. E., Dimanche, F., & Bogle, T. (1994). Segmenting the adult attachment and activity involvement. Paper presented at the North-
fitness market using involvement profiles. Journal of Park and eastern Recreation Research Symposium (pp. 89–92). Bolton Landing,
Recreation Administration, 12(3), 38–56. New York.
Hidalgo, M. C., & Hernandez, B. (2001). Place attachment: conceptual Mowen, A. J., Graefe, A. R., & Williams, D. R. (1998). An assessment of
and empirical questions. Journal of Environmental Psychology, 21(3), activity and trail type as indicators of trail user diversity. Journal of
273–281. Park and Recreation Administration, 16(1), 80–96.
Hwang, S. N., Lee, C., & Chen, H. J. (2005). The relationship among Mowen, J. C., & Minor, M. (1998). Consumer behavior (5th ed.). Upper
tourists’ involvement, place attachment and interpretation satisfaction Saddle River, New Jersey: Prentice-Hall.
in Taiwan’s national parks. Tourism Management, 26(2), 143–156. Murphy, P. E. (1985). Scope and nature of tourism. In Tourism: A
Kang, I., Jeon, S., Lee, S., & Lee, C. K. (2005). Investigating structural community approach (Chapter 1, pp. 3–16). London: Routledge.
relations affecting the effectiveness of service management. Tourism Olsson, U. H., Foss, T., Troye, S. V., & Howell, R. D. (2000). The
Management, 26(3), 301–310. performance of ML, GLS, and WLS estimation in structural equation
Kelly, I. (2002). Australian regional tourism handbook: Industry solu- modeling under conditions of misspecification and nonnormality.
tions 2001. Cooperative Research Centre for Sustainable Tourism. Structural Equation Modeling, 7(4), 557–595.
/www.crctourism.com.auS Accessed 14 April 2005. Poria, Y., Butler, R., & Airey, D. (2003). Revisiting Mieczkowski’s
Kerstetter, D., Confer, J., & Graefe, A. (2001). An exploration of the conceptualization of tourism. Tourism Geographies, 5(1), 26–38.
specialization concept within the context of heritage tourism. Journal Quan, S., & Wang, N. (2004). Towards a structural model of the tourist
of Travel Research, 39(3), 267–274. experience: An illustration from food experiences in tourism. Tourism
Klem, L. (2000). Structural equation modeling. In L. G. Grimm, & P. R. Management, 25(3), 297–305.
Yarnold (Eds.), Reading and understanding more multivariate statistics Reisinger, Y., & Turner, L. (1999). Structural equation modeling with
(pp. 227–260). Washington, DC: American Psychological Association. Lisrel: Application in tourism. Tourism Management, 20(1), 71–88.
Kyle, G. T., Absher, J. D., & Graefe, A. R. (2003b). The moderating role Ryan, C., & Huyton, J. (2000). Aboriginal tourism: A linear structural
of place attachment on the relationship between attitudes toward fees relations analysis of domestic and international tourist demand.
and spending preferences. Leisure Sciences, 25(1), 33–50. International Journal of Tourism Research, 2(1), 15–29.
Kyle, G., Bricker, K., Graefe, A., & Wickham, T. (2004a). An Scott, D., & Shafer, C. S. (2001). Recreation specialization: A critical look
examination of recreationists’ relationships with activities and settings. at the construct. Journal of Leisure Research, 33(3), 319–343.
Leisure Sciences, 26(2), 123–142. Selin, S. W., & Howard, D. R. (1988). Ego involvement and leisure
Kyle, G., & Chick, G. (2002). The social nature of leisure involvement. behavior: A conceptual specification. Journal of Leisure Research,
Journal of Leisure Research, 34(4), 426–448. 20(3), 237–244.
Kyle, G., Graefe, A., & Manning, R. (2004c). Attached recreationists: Smith, S. L. J., & Godbey, G. C. (1991). Leisure, recreation and tourism.
Who are they? Journal of Park and Recreation Administration, 22(2), Annals of Tourism Research, 18(1), 85–100.
65–84. Stebbins, R. A. (1982). Serious leisure: A conceptual statement. Pacific
Kyle, G., Graefe, A., Manning, R., & Bacon, J. (2003a). An examination Sociological Review, 25(2), 251–272.
of the relationships between leisure activity involvement and place Swanson, K. K., & Horridge, P. E. (2004). A structural model for souvenir
attachment among hikers along the Appalachian Trail. Journal of consumption, travel activities, and tourist demographics. Journal of
Leisure Research, 35(3), 249–273. Travel Research, 42(4), 372–380.
Kyle, G., Graefe, A., Manning, R., & Bacon, J. (2004b). Effect of activity Tabachnick, B. G., & Fidell, L. S. (2001). Using multivariate statistics (4th
involvement and place attachment on recreationists’ perceptions of ed.). Boston: Allyn & Bacon.
setting density. Journal of Leisure Research, 36(2), 209–231. Thompson, B. (2000). Ten commandments of structural equation
Kyle, G., Graefe, A., Manning, R., & Bacon, J. (2004d). Predictors of modeling. In L. G. Grimm, & P. R. Yarnold (Eds.), Reading and
behavioral loyalty among hikers along the Appalachian Trail. Leisure understanding more multivariate statistics (pp. 261–283). Washington,
Sciences, 26(1), 99–118. DC: American Psychological Association.
Kyle, G. T., Kerstetter, D. L., & Guadagnolo, F. B. (1999). The influence Tribe, J. (1995). Introduction. In: The economics of leisure and tourism:
of outcome messages and involvement on participant reference price. environments, markets and impacts. (Chapter 1, pp. 1–17). Oxford, UK:
Journal of Park and Recreation Administration, 17(3), 53–75. Butterworth-Heinemann.
Kyle, G. T., Kerstetter, D. L., & Guadagnolo, F. B. (2002). Market Wellman, J. D., Roggenbuck, J. W., & Smith, A. C. (1982). Recreation
segmentation using participant involvement profiles. Journal of Park specialization and norms of depreciative behavior among canoeists.
and Recreation Administration, 20(1), 1–21. Journal of Leisure Research, 14(4), 323–340.
Kyle, G., & Mowen, A. J. (2005). An examination of the leisure Williams, D. R., Patterson, M. E., Roggenbuck, J. W., & Watson, A. E.
involvement–agency commitment relationship. Journal of Leisure (1992). Beyond the commodity metaphor: Examining emotional and
Research, 37(3), 342–363. symbolic attachment to place. Leisure Sciences, 14(1), 29–46.
ARTICLE IN PRESS
M.J. Gross, G. Brown / Tourism Management 29 (2008) 1141–1151 1151

Williams, S. (2003). Tourism, recreation and leisure. In Tourism and Yoon, Y., & Uysal, M. (2005). An examination of the effects of
recreation (Chapter 1, pp. 1–24). Harlow, Essex, England: Pearson motivation and satisfaction on destination loyalty: A structural model.
Education. Tourism Management, 26(1), 45–56.
Yoon, Y., Gursoy, D., & Chen, J. S. (2001). Validating tourism Zins, A. H. (2002). Consumption emotions, experience quality and
development theory with structural equation modeling. Tourism satisfaction: A structural analysis for complainers versus non-
Management, 22(4), 363–372. complainers. Journal of Travel and Tourism Marketing, 12(2–3), 3–18.

Anda mungkin juga menyukai