Description
This study focuses on the emotional aspects of destinations and employs two different
scales for capturing the affective component of city destination image. The aim of this paper is not only to
measure the emotions assigned to different European cities, but also to compare these two
instruments/scales by means of Procrustes analysis.
International Journal of Culture, Tourism and Hospitality Research
Measuring the resemblance between pictorial and verbal city image spaces
Ilona Pezenka Christian Buchta
Article information:
To cite this document:
Ilona Pezenka Christian Buchta, (2012),"Measuring the resemblance between pictorial and verbal city image spaces", International J ournal
of Culture, Tourism and Hospitality Research, Vol. 6 Iss 4 pp. 326 - 339
Permanent link to this document:
http://dx.doi.org/10.1108/17506181211265068
Downloaded on: 24 January 2016, At: 22:20 (PT)
References: this document contains references to 43 other documents.
To copy this document: [email protected]
The fulltext of this document has been downloaded 257 times since 2012*
Users who downloaded this article also downloaded:
Ulrike Gretzel, Yeong-Hyeon Hwang, Daniel R. Fesenmaier, (2012),"Informing destination recommender systems design and evaluation
through quantitative research", International J ournal of Culture, Tourism and Hospitality Research, Vol. 6 Iss 4 pp. 297-315 http://
dx.doi.org/10.1108/17506181211265040
Sangwon Park, Daniel R. Fesenmaier, (2012),"Non-response bias in internet-based advertising conversion studies", International J ournal of
Culture, Tourism and Hospitality Research, Vol. 6 Iss 4 pp. 340-355 http://dx.doi.org/10.1108/17506181211265077
Sara Dolnicar, Friedrich Leisch, (2012),"One legacy of Mazanec: binary questions are a simple, stable and valid measure
of evaluative beliefs", International J ournal of Culture, Tourism and Hospitality Research, Vol. 6 Iss 4 pp. 316-325 http://
dx.doi.org/10.1108/17506181211265059
Access to this document was granted through an Emerald subscription provided by emerald-srm:115632 []
For Authors
If you would like to write for this, or any other Emerald publication, then please use our Emerald for Authors service information about
how to choose which publication to write for and submission guidelines are available for all. Please visit www.emeraldinsight.com/
authors for more information.
About Emerald www.emeraldinsight.com
Emerald is a global publisher linking research and practice to the benefit of society. The company manages a portfolio of more than
290 journals and over 2,350 books and book series volumes, as well as providing an extensive range of online products and additional
customer resources and services.
Emerald is both COUNTER 4 and TRANSFER compliant. The organization is a partner of the Committee on Publication Ethics (COPE) and
also works with Portico and the LOCKSS initiative for digital archive preservation.
*Related content and download information correct at time of download.
D
o
w
n
l
o
a
d
e
d
b
y
P
O
N
D
I
C
H
E
R
R
Y
U
N
I
V
E
R
S
I
T
Y
A
t
2
2
:
2
0
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
Measuring the resemblance between
pictorial and verbal city image spaces
Ilona Pezenka and Christian Buchta
Abstract
Purpose – This study focuses on the emotional aspects of destinations and employs two different
scales for capturing the affective component of city destination image. The aimof this paper is not only to
measure the emotions assigned to different European cities, but also to compare these two
instruments/scales by means of Procrustes analysis.
Design/methodology/approach – The authors collected measurements on two different scales (verbal
and pictorial) for capturing the emotional (affective) component of destination image – both based on
Russell’s circumplex model of affect – in two independent surveys.
Findings – Signi?cant differences were found between the multidimensional scaling (MDS) results of
both scales. Because the two samples match in terms of demographics and psychographics, the
differences of the perceptual spaces are likely due to the form of stimuli (pictures compared to verbal
items) presented. The results indicate that pictures are easy to use, but they also are subject to broader
interpretations. In contrast, verbal items are more concise, but perhaps respondents may ?nd it harder
to assign them to cities (especially negative emotions).
Research limitations/implications – The paper contributes to the literature by suggesting a
methodology for detailed analysis of the differences in measurements, and introducing implications that
apply primarily to researchers once new measurement methods are introduced. The limitations of this
research relate to the sample (convenience sample): the respondents were solely Austrians, and thus
only one culture was represented.
Originality/value – Although several other researchers suggest measuring both the cognitive and the
affective image, very few studies incorporate both aspects in evaluating destination image. In contrast,
this study applies different scales to incorporate emotions in destination image measurement and
demonstrates the applicability of the scales in the tourism context.
Keywords Destination image, Emotions, Circumplex model of affect, Procrustes statistics,
Multidimensional scaling, Tourism, Place identity
Paper type Research paper
Introduction
Most destination image studies to date emphasise the cognitive component (beliefs,
knowledge, and experience) of destination image and totally ignore emotions and feelings,
which are referred to as the ‘‘affective’’ aspects. Because destination choice is believed to
be dominated by intangible factors (Walmsley and Young, 1998; Beerli and Mart? ´n, 2004) –
such as, for example, emotions – city destination marketers should place more importance
on the emotional perception of their city.
The importance of pictures as emotional elements of marketing communication strategies is
demonstrated by the fact that in most tourism brochures, more than 75 per cent of the content
is pictorial (Jenkins, 1999). Especially in tourism, pictures often serve as means to evoke a ?rst
impression of what one might expect on their journey or destination (Berger et al., 2007).
Therefore, this study concentrates on the emotional aspects of destinations, and employs two
different scales for capturing the affective component of city destination image.
PAGE 326
j
INTERNATIONAL JOURNAL OF CULTURE, TOURISM AND HOSPITALITY RESEARCH
j
VOL. 6 NO. 4 2012, pp. 326-339, Q Emerald Group Publishing Limited, ISSN 1750-6182 DOI 10.1108/17506181211265068
Ilona Pezenka is a
Researcher and Christian
Buchta is a Researcher and
Lecturer, both at the
Institute for Tourism and
Leisure Studies, Vienna
University of Economics
and Business, Vienna,
Austria.
Received March 2011
Revised June 2011
Accepted September 2011
This work was supported by the
Jubila¨ umsfonds der
Oesterreichischen
Nationalbank.
D
o
w
n
l
o
a
d
e
d
b
y
P
O
N
D
I
C
H
E
R
R
Y
U
N
I
V
E
R
S
I
T
Y
A
t
2
2
:
2
0
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
Because human information processing is largely based on images (Zaltman, 1997), and as
Kosslyn et al. (1990) state, two-thirds of information reaches the brain through the visual
system, this study also uses pictorial stimuli. The picture scale is indented to facilitate
expression of emotions and feelings evoked by cities. Also, the present researchers assume
that the less boring and more diversi?ed survey procedure of visual-based scales lowers
respondents’ fatigue, and thus ensures a higher response rate (Herzig, 1991).
Theory
Destination image
The scienti?c community widely agrees that destination image plays a key role in tourist
behaviour (Moutinho, 1987; Gartner, 1989; Chon, 1990; Goodall, 1988). Researchers have
suggested that destinations that have a positive image are more likely to be considered in
the destination selection process (Goodall, 1988).
Regarding the de?nition and operationalisation of destination image, we may identify several
streams of research. Some scientists (e.g. Milman and Pizam, 1995) view image solely as a
holistic concept, as demonstrated by the de?nition of Crompton (1979), who states that
destination image is the ‘‘sum of beliefs, ideas, and impressions that a person has of a
destination’’. Others postulate that it consists of different components – cognitive aspects
(beliefs, knowledge, and experience) and affective aspects (feelings), which all form the
global image. This classi?cation is useful when analysing the structure of destination image
and ?nally predicting consumer behaviour (Bagozzi and Burnkrant, 1985), and it also helps
to analyse factors in?uencing destination image formation and to reveal causal relationships.
The approach is also consistent with the theory of attitudes (Ajzen and Fishbein, 1980).
According to Mazanec (1978), images can be distinguished from attitudes by their having a
more distinctive emotional component. In contrast to emotions, attitudes can be stored for
long periods of time, their connection to volition and action is not particularly strong and
direct, and ?nally, arousal is unnecessary in connection with attitudes, but it is an important
part of emotions (Bagozzi et al., 1999, p. 188).
Although several authors suggest measuring both cognitive and the affective image
components, very few studies incorporate both aspects in evaluating destination image.
Mazanec (2010) uses an affect-centred interpretation of destination image when studying
the supply-side induced (projected) image, by analysing the connotations attached to
country names in textual web content. Baloglu and McCleary (1999) and Baloglu and Love
(2005) apply Russell’s (1980) circumplex model of affect to assess survey-based affective
image components.
The lack of destination image studies employing scales to capture feelings and emotions
may be due to the complex nature of emotions and their measurement.
Emotions
Bagozzi et al. (1999, p. 184) describe the term ‘‘affect’’ as ‘‘an umbrella for a set of more
speci?c mental processes including emotions, moods, and (possibly) attitudes’’. Here, the
authors refer to emotions whenever it comes to the affective aspects of destination image.
The plethora of theories of emotions leads to different approaches of measurement (see
Bagozzi et al., 1999 and Richins, 1997, for comprehensive reviews). In marketing research,
the empirical measurement of emotions mostly relies on self-reports (questionnaires), with
the number of items ranging up to 180 (Bagozzi et al., 1999). Several authors have tried to
identify the basic dimensions of emotions (Plutchik, 1962; Izard, 1977; Russell and
Mehrabian, 1977; Holbrook, 1986), and have successfully demonstrated the circular
arrangement of these base emotions (Plutchik, 1997; Russell, 1989).
This paper employs one of the most widely studied multidimensional approaches for
capturing affective responses: Russell’s (1980) circumplex model of affect. The model
suggests that the meaning attributed to environments can be de?ned by two orthogonal
bipolar dimensions:
VOL. 6 NO. 4 2012
j
INTERNATIONAL JOURNAL OF CULTURE, TOURISM AND HOSPITALITY RESEARCH
j
PAGE 327
D
o
w
n
l
o
a
d
e
d
b
y
P
O
N
D
I
C
H
E
R
R
Y
U
N
I
V
E
R
S
I
T
Y
A
t
2
2
:
2
0
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
1. pleasure/valence (pleasant/unpleasant); and
2. arousal (arousing/sleepy).
Any affective quality can be de?ned as a combination of the two base components, and
therefore can be presented in the affective space. In conclusion, Russell (1980) identi?es
and places eight affect concepts in a circular order around the two-dimensional bipolar
space (see Figure 1). Two additional dimensions (i.e. exciting/gloomy and
distressing/relaxing) help to de?ne the quadrants of the affective space, but do not form
independent dimensions. Due to its simplicity, Russell’s (1980) scale has been successfully
employed by a number of authors (e.g. Baloglu and McCleary, 1999).
Research method
Research design
Regarding the selection of city destinations, 12 European cities – Amsterdam, Barcelona,
Berlin, Budapest, Hamburg, London, Munich, Paris, Prague, Venice, Vienna and Zurich – were
used for both studies. These cities were chosen because they rank among the most visited city
break destinations by Austrian tourists travelling within Europe (cf. TourMIS – a touristic
marketing information systemthat provides electronic tourismstatistics; see www.tourmis.info).
We collected data from two separate samples. Both samples were instructed to assign the
respective stimuli to 12 European cities. For the ?rst sample (471 respondents), pictorial
stimuli were selected from the International Affective Picture System (IAPS; Lang et al., 2008).
According to Russell (1980), these stimuli are standardised on the basis of ratings of pleasure
and arousal. The second sample (367 respondents) was instructed to assign verbal items
(taken from Russell and Pratt, 1980). In order to test the homogeneity of the samples, we
collected additional information (demographics, travel experience, preferences). The surveys
were conducted through two self-administered online questionnaires (Sample 1: June 2009;
Sample 2: August 2009). Because we employed convenience sampling in both studies, the
URL address for the online surveys was publicised through e-mail distributions (WU survey
mailing list) and the respondents were solely Austrian tourists.
Sample 1 (pictures)
Respondents of Sample 1 were instructed to judge whether or not the emotions conveyed by
the presented pictures correspond to the cities in study. The instructions stressed
Figure 1 Russell’s circumplex model of affect
PAGE 328
j
INTERNATIONAL JOURNAL OF CULTURE, TOURISM AND HOSPITALITY RESEARCH
j
VOL. 6 NO. 4 2012
D
o
w
n
l
o
a
d
e
d
b
y
P
O
N
D
I
C
H
E
R
R
Y
U
N
I
V
E
R
S
I
T
Y
A
t
2
2
:
2
0
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
respondents to assign the pictures on the basis of the feelings they evoked, rather than on
their actual content. The photographs that formed the affective stimuli of the scale were
selected from the IAPS (Lang et al., 2008) through a multi-step process. The IAPS is a
database of more than 1,000 emotionally evocative colour photographs which are used in
experimental investigations of emotions worldwide. The pictorial stimuli are standardised on
the basis of ratings of pleasure (valence) and arousal by using the self-assessment manikin
(SAM; Bradley and Lang, 1994). The resulting normative scores (mean values and standard
deviations) for males, females and both sexes combined are available for each picture of the
IAPS. Because the IAPS is based on the dimensional view of affect according to Russell
(1980) and the circumplex model was already adapted in a tourism destination context
(e.g. Baloglu and McCleary, 1999; Baloglu and Love, 2005), it seemed to ?t well with the
present study.
In a ?rst step, we excluded apparently offensive pictures as well as low-quality pictures, and
also eliminated pictures judged differently by sex. Preliminary results show that these
pictures cause great variance in assignment patterns. In a second step, empirical
card-sorting studies, using WebSorte, a web-based software application, were conducted
to further reduce photographs. For the ?rst study the affective space of the selected 245
IAPS images was subdivided into three groups of emotional valence, with 36 images for the
subgroup ‘‘unpleasant’’, 112 images for the subgroup ‘‘neutral’’ and 97 images for the
subgroup ‘‘pleasant’’. Three different sorting tasks using three different samples
(unpleasant, n ¼ 19; neutral, n ¼ 18; pleasant, n ¼ 17) were performed, and respondents
were asked to sort the stimuli into the eight aforementioned emotional categories with an
additional ‘‘no classi?cation’’ option. Images with the highest category membership values
for the ‘‘no classi?cation’’ option were eliminated, and images with the ten highest
percentages of assignment in each of the categories were selected for further evaluation. A
second study, with 124 respondents sorting all 92 remaining images into the nine categories
was conducted. Again, pictures with high percentages in the ‘‘no classi?cation’’ category
were excluded from further analysis. The resulting categorisation was then compared with
the positions of the original IAPS normative values. Images with ambiguous category
membership were excluded. Because the distribution of the remaining 36 pictures was not
equal or balanced over all nine affective categories, ?ve of the 245 preliminary survey
images were added, so that ?nally, each of the eight relevant categories were represented
by the best ?tting three images to design a nonverbal scale for assessing the affective
component of a city image.
Sample 2 (verbal items)
Sample 2 was instructed to assign verbal items (taken from Russell and Pratt, 1980) to 12
European cities. The original pool of 40 verbal items (?ve items per category) was reduced in
such a way that the number of items representing one category was consistent with the
number of photographs per category (Study 1). Three experts judged the terms with respect
to their adequacy to describe certain emotions evoked by city tourism destinations and
eliminated those unfamiliar or rarely used in everyday language. Finally, 16 verbal items
representing eight affective qualities were used (Figure 2)
In order to test the homogeneity of the two samples (pictorial or verbal items), several tests
were performed. x
2
tests (sex, education) show no signi?cant x
2
values. Regarding age
and travel experience, the one-way ANOVA did not provide signi?cant F-values. The
distribution of the preference ranks was tested using the Mann-Whitney test – where again,
no signi?cant differences were detected between the two samples. Table I shows the
characteristics of the two samples and the results of the homogeneity tests.
Methodology
In both studies participants were asked to associate stimuli (pictorial or verbal items) with the
same objects of interest (cities). In a ?rst step, we assumed that the relative frequencies of
stimulus-object pairs were a meaningful empirical measure of similarity between objects and
stimuli, S(X, Y). In a second step, we sought a joint representation of objects and stimuli in a
VOL. 6 NO. 4 2012
j
INTERNATIONAL JOURNAL OF CULTURE, TOURISM AND HOSPITALITY RESEARCH
j
PAGE 329
D
o
w
n
l
o
a
d
e
d
b
y
P
O
N
D
I
C
H
E
R
R
Y
U
N
I
V
E
R
S
I
T
Y
A
t
2
2
:
2
0
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
low-dimensional space. The data samples are independent, and therefore two independent
representations exist. Finally, in a third step, we compared multidimensional scaling (MDS)
ordination results by means of Procrustes analysis in order to ?nd an optimal
superimposition of the two con?gurations.
Table I Demographics and homogeneity of the samples
Verbal Pictures
n Per cent n Per cent x
2
/F/U p
Total number of respondents 367 471
Abandonment rate 21.8 44.8
Sex 1.267 0.146
Male 115 31 165 35
Female 252 69 306 65
Age
Mean 26.20 26.38 0.13 0.719
Standard deviation 7.44 7.21
Education 5.093 0.278
Elementary school 0.3 1.3
Secondary education 3.3 3.2
A levels 72.2 70.3
University 24.3 25.3
Travel experience
a
1.40 0.277
Mean 4.35 4.27
Standard deviation 1.07 1.10
Preferences 14.42
Top three cities Barcelona, London, Paris Barcelona, London, Paris
Least preferred cities Munich, Vienna, Zurich/Budapest Vienna, Budapest, Munich
Note:
a
Short trips þ vacation trips
Figure 2 Verbal items per category used
PAGE 330
j
INTERNATIONAL JOURNAL OF CULTURE, TOURISM AND HOSPITALITY RESEARCH
j
VOL. 6 NO. 4 2012
D
o
w
n
l
o
a
d
e
d
b
y
P
O
N
D
I
C
H
E
R
R
Y
U
N
I
V
E
R
S
I
T
Y
A
t
2
2
:
2
0
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
In the MDS literature, the problem outlined for the second step is known as the unfolding
problem (Borg and Groenen, 2005, Coombs, 1950). Technically, this problem may be
considered a special case of MDS where, in the present context, the object-object and
stimulus-stimulus proximities are missing. However, and especially when values are missing
systematically, unfolding solutions are known to be prone to degeneracies (see the
discussion in Borg and Groenen, 2005). In order to avoid this problem, we suggest replacing
the missing values with estimates from the data on the one hand, and basing them on
external information on the other.
First, we assume that the unknown proximities between objects (cities) can be approximated
by the joint frequency that two objects are associated with the same stimulus by the same
respondent. More formally, we have all the ingredients of a cross-tabulation of two binary
variables, fromwhich we can compute some measure of similarity (for example, see Cox and
Cox, 2001). We obtained good results using the Jaccard coef?cient (Jaccard, 1908) in order
to replace the missing object-object proximities.
Next, we suggest replacing the missing stimulus-stimulus proximities with (Euclidean)
distances between the known (hypothetical) locations of the presented pictorial (or verbal)
stimuli in the IAPS (circumplex) space. Unfortunately, the literature does not report positions
of the verbal items in the circumplex space (Russell and Pratt, 1980). The best we can do
under these circumstances is to presume that the verbal items are located at their theoretical
positions. This is similar to using an unbiased prior in Bayesian model estimation (see Rossi
et al., 2005). In doing so, we assume the verbal items to be located at the grid points of the
circumplex model that correspond with an item’s category (see also the discussion
regarding Figure 3). Further, the object-object similarities above must be transformed to
distances in the range of the IAPS space (or vice versa).
The important point here is that we postulate that the circumplex model is complete and
unbiased. Consequently, the MDS problems to be solved in the third step must be restricted
Figure 3 Joint pictorial and verbal measurement map
VOL. 6 NO. 4 2012
j
INTERNATIONAL JOURNAL OF CULTURE, TOURISM AND HOSPITALITY RESEARCH
j
PAGE 331
D
o
w
n
l
o
a
d
e
d
b
y
P
O
N
D
I
C
H
E
R
R
Y
U
N
I
V
E
R
S
I
T
Y
A
t
2
2
:
2
0
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
to two dimensions, and their solutions must be transformed to ?t the MDS representation of
the circumplex model. More precisely, the MDS con?guration of a speci?c mode of
measurement (pictorial or verbal) must be ?tted to the model representation of the same
mode. This third step is presented in more detail below. Further, the model representations of
both modes represent the same space for a joint representation of the different approaches
to measurement in order to be meaningful. Finally, we restrict the MDS solutions to metric,
which ensures that the distances between stimuli (see above) are preserved. Accordingly,
we obtain a mapping of the cities into the IAPS-calibrated space of emotions.
Another, more substantial, reason why we do not suggest estimating the stimulus-stimulus
proximities from the data is that in general, we cannot expect the objects to be
representative of the measurement space. The assignment frequencies may bias the
stimulus-stimulus relations. However, in an exploratory setting, pooling the data sets would
be recommended.
Last, we seek transformations of MDS solutions that make themcomparable. This is known in
the literature (Borg and Groenen, 2005; Gower, 1971) as the Procrustes problem, which can
be stated more formally as follows: approximate (?t) the MDS con?guration Y (the testee) to X
(the target) so that:
kY 2f ð:Þk
2
! min !
s:t: f ðs; t; T ; XÞ ¼ sXT þ t1
0
;
where s is a (scalar) dilation factor, T is an orthogonal transformation satisfying TT
0
¼ T
0
T ¼
1 (a re?ection and rotation matrix) and t is a translation (vector). Note that f(.) de?nes a
transformation of the coordinate system of Y, and therefore of any points that were not used
in the model ?tting. Therefore, although our MDS solutions contain both objects and stimuli,
we use only one or the other for transformation. More precisely, let X(Y) denote the
con?guration of objects and stimuli in the pictorial (verbal) MDS solution, and X
0
(Y
0
) the
con?guration of the stimuli in the circumplex model, respectively. As outlined above, we
need to determine the transformations: X !X
0
and Y !ðY
0
!X
0
) in order to obtain a joint
representation of the pictorial and verbal measurements. Remember that X
0
and Y
0
were
assumed to share the same space (coordinate system). Therefore, only the magnitudes of
the scales can differ, so we must ?t Y
0
to X
0
subject to t ¼ ð0; 0Þ and T ¼ 1.
We assess the goodness of ?t (sums of squares) of a transformation using permutation tests
(Jackson, 1995; Schneider and Borlund, 2007). This involves sampling the distribution of the
?t under random permutations of the correspondence mapping between points in X and
points in Y. The probability value is then the percentage of samples with a ?t smaller or equal
than observed.
The analysis was done using the R environment for statistical computing (R Development
Core Team, 2005), in particular, package SMACOF (De Leeuw and Mair, 2009) for
multidimensional scaling, which is based on the scaling by majorising a complex function
approach (De Leeuw and Heiser, 1980). For Procrustes transformations and permutation
testing, we used our own implementations. However, see package VEGAN (Oksanen et al.,
2010) which provides, amongst others, an implementation of Procrustes transformations,
Protest and the Mantel test.
Results
Initially, we show that the suggested approach is reliable. To illustrate, we computed the
average IAPS (verbal) object scores across respondents. Because respondents could
associate an object with more than one (redundant) stimulus, the scores must be interpreted
with caution. For example, from a high score, we cannot tell whether the emotional
perception of a city was intense (more than one item from the same category was
associated) or multifaceted (more than one category was associated).
Figures 4 and 5 show that the MDS solutions reproduce the scores very well. In Figure 4 the
con?guration of the pictorial scores (grey circles) was ?tted to the object con?guration of the
PAGE 332
j
INTERNATIONAL JOURNAL OF CULTURE, TOURISM AND HOSPITALITY RESEARCH
j
VOL. 6 NO. 4 2012
D
o
w
n
l
o
a
d
e
d
b
y
P
O
N
D
I
C
H
E
R
R
Y
U
N
I
V
E
R
S
I
T
Y
A
t
2
2
:
2
0
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
Figure 4 Fitted pictorial scores
Figure 5 Fitted verbal scores
VOL. 6 NO. 4 2012
j
INTERNATIONAL JOURNAL OF CULTURE, TOURISM AND HOSPITALITY RESEARCH
j
PAGE 333
D
o
w
n
l
o
a
d
e
d
b
y
P
O
N
D
I
C
H
E
R
R
Y
U
N
I
V
E
R
S
I
T
Y
A
t
2
2
:
2
0
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
pictorial measurements. The largest deviations occur for Venice, Vienna, and Barcelona. The
probability that the ?t could be due to chance is insigni?cant (0.001, n ¼ 1; 000). For the
verbal measurements (black dots), we see in Figure 5 that the ?t is slightly worse (but still
signi?cant at 0.001, n ¼ 1; 000) compared to the pictorial case. The largest deviations occur
for Hamburg, Budapest, and again, Vienna. Finally, returning to the aforementioned
ambiguities with scores, scaling the response variables to sumto 1 (0) across stimuli, we did
not observe signi?cant differences with respect to Figures 4 and 5. This suggests that a
change in the range of the scores only occurred which was compensated by the dilation
factor s. However, closer inspection of the data reveals that individual responses are
redundant and multifaceted.
We conclude that the approach in constructing proximities and their mapping into a lower
dimensional space via metric MDS is reliable in the sense that the observed overall
(average) city positions are preserved.
We nowcheck the ?t of pictorial and verbal stimuli to their known (theoretical) positions in the
measurement space (see Figures 6 and 7). In Figure 6 the black dots indicate the locations
(mean scale values) of the pictures in the IAPS space. The grey circles indicate the ?tted
locations of the MDS con?guration. The image codes and the categories they belong to are
indicated by the labels. Closer inspection of the frequency data suggests, as earlier
discussed, that the deviations re?ect differences in stimulus usage by respondents. For
example, the comparatively less-used images 5920 (0.166) and 5940 (0.105) from the
‘‘arousing’’ category shift towards the periphery; or the more frequently used image 9590
(0.454) fromthe ‘‘unpleasant’’ category shifts towards the centre. Thus, shifts can be seen as
the result of distortions in usage of stimuli ‘‘on’’ objects, and the obvious conclusion is that
stimuli might not be representative either. However, we cannot be sure, because no variation
exists in the object sets (see discussion above). In the verbal case (see Figure 7), we had to
?ll in the reference positions due to lack of external data. Thus, the items are collapsed to the
ideal locations of their category (black dots). For example, the ‘‘arousing’’ item from the
Figure 6 Fit of pictorial stimuli
PAGE 334
j
INTERNATIONAL JOURNAL OF CULTURE, TOURISM AND HOSPITALITY RESEARCH
j
VOL. 6 NO. 4 2012
D
o
w
n
l
o
a
d
e
d
b
y
P
O
N
D
I
C
H
E
R
R
Y
U
N
I
V
E
R
S
I
T
Y
A
t
2
2
:
2
0
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
‘‘arousing’’ category is a (signi?cantly) less-used city attribution (compared to ‘‘active’’ and
‘‘alive’’). Overall, the pictorial measurements (SSE ¼ 12:93) seem to be more biased than
the verbal ones (SSE ¼ 8:39).
In Figure 3 all the ?gures presented so far are superimposed into a joint map of the pictorial
(black dots) and verbal (grey circle) measurements. The categories are indicated by the
(convex) hulls of their members. Ideally, these regions would be near their theoretical
centres, at (22, 2 2), (0, 2 2), . . . (2, 2). The centres de?ne an ideal division of the
measurement space into convex regions. The empirical one is indicated by the Voronoi
tessellation of the centre points of the hulls (solid grey lines) plus the origin (dashed grey
lines). For example, London is predominantly perceived as ‘‘unpleasant’’, but also as the
location in the centre area indicates, multifaceted and thus possibly ambivalent. Under
verbal measurement the cities shift signi?cantly in the ‘‘pleasant’’ and/or ‘‘arousing’’
direction. Table II shows that the category ‘‘intensity’’ is higher than in the pictorial setting;
that is, in the same category, more terms are associated than pictures. The most pertinent
difference between pictorial and verbal perception occurs for Venice, where the dominance
on the ‘‘arousal’’ dimension is reversed.
For an overall comparison of the two modes of measurements, we computed the Spearman
rank correlation between the (relative) city category frequencies (see Table II) and its
probability value under two regimes of permutation sampling. Under the ?rst, the cities are
?xed in order to assess the correspondence of the categories across samples; under the
second, the categories are ?xed in order to assess the correspondence of the cities. This
measure is stricter than varying objects and categories simultaneously. The correlation is
positive, but low (0.113), and both probability values are 0 (for n ¼ 10; 000 samples). Thus, a
signi?cant difference exists between the pictorial and verbal ‘‘perception’’ of the cities.
Comparing the total observed stimulus usage of pictures (0.364) and terms (0.224) to the
more detailed ?gures in Table II do not seemto contradict such an assumption. Furthermore,
Figure 7 Fit of verbal stimuli
VOL. 6 NO. 4 2012
j
INTERNATIONAL JOURNAL OF CULTURE, TOURISM AND HOSPITALITY RESEARCH
j
PAGE 335
D
o
w
n
l
o
a
d
e
d
b
y
P
O
N
D
I
C
H
E
R
R
Y
U
N
I
V
E
R
S
I
T
Y
A
t
2
2
:
2
0
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
in 72 (of 96) cases, the percentage of pictures associated is equal or larger than the
percentage of terms.
The stress values for the pictorial and verbal MDS solutions are 0.0832 and 0.0772, as
reported by package SMACOF; or 0.1289 and 0.0918 using Kruskal’s measure (see Kruskal
1964a, b). The former values are both signi?cant (0, n ¼ 100) under sampling of the distance
matrix from the uniform distribution (scaled to the observed range).
Discussion and limitations
This research ?rst shows that the proposed method of constructing proximities and mapping
into a lower dimensional space via metric MDS is reliable, in the sense that observed overall
(average) city positions are preserved. We then checked the ?t of pictorial and verbal stimuli
to their known (theoretical) positions in the measurement space according to Russell’s
circumplex model of affect. Results show a ?t of the empirical data to the theoretical
circumplex model of affect. Thus, this model offers a sound explanation of different classes
of emotions evoked by city destinations. To test whether this model is applicable to
destinations in general, replication studies should be carried out.
Regarding the comparison of the ordinations, results indicate a signi?cant difference
between the pictorial and verbal ‘‘perception’’ of the cities. We try to explain this result thusly:
verbal associations may re?ect the dominant sentiments towards a city formed from
processing all of one’s impressions; while visual associations may indicate that the set of
visual impressions is actually diverse. Another explanation might be that pictures are richer
in content than words; and therefore, a comparatively larger number of pictures ?ts one’s
image of a city. These explanations can be supported by the dual-coding theory (Sadoski
Table II Percentage stimulus attribution £ 100, categories £ cities £ ðpictorial; verbalÞ
Arousing Distressing Exciting Gloomy Pleasant Relaxing Sleepy Unpleasant
Amsterdam Pictorial 30 37 50 36 43 40 49 49
Verbal 7 16 19 15 20 39 38 18
Barcelona Pictorial 36 27 47 33 42 38 33 41
Verbal 11 23 23 17 16 45 56 25
Berlin Pictorial 27 38 46 51 40 26 38 52
Verbal 17 22 20 20 13 38 37 18
Budapest Pictorial 21 35 17 54 48 37 56 46
Verbal 10 15 23 20 18 31 17 18
Hamburg Pictorial 18 23 32 38 32 28 35 30
Verbal 5 17 20 12 18 32 27 15
London Pictorial 18 42 41 44 36 30 44 43
Verbal 14 23 23 25 11 46 47 20
Munich Pictorial 16 9 43 28 51 57 39 14
Verbal 4 20 23 17 25 30 30 16
Paris Pictorial 25 28 39 41 62 38 55 37
Verbal 14 18 25 17 10 45 46 20
Prague Pictorial 19 31 21 54 53 38 60 37
Verbal 9 17 22 12 17 39 28 18
Venice Pictorial 13 17 9 32 43 21 48 32
Verbal 8 12 28 16 15 44 33 22
Vienna Pictorial 17 17 48 43 77 61 52 24
Verbal 8 23 29 18 20 46 47 23
Zurich Pictorial 14 8 42 28 44 53 44 13
Verbal 4 18 24 17 23 28 26 16
PAGE 336
j
INTERNATIONAL JOURNAL OF CULTURE, TOURISM AND HOSPITALITY RESEARCH
j
VOL. 6 NO. 4 2012
D
o
w
n
l
o
a
d
e
d
b
y
P
O
N
D
I
C
H
E
R
R
Y
U
N
I
V
E
R
S
I
T
Y
A
t
2
2
:
2
0
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
and Paivio, 2001), which also assumes that differences in the qualities of verbal and
nonverbal units exist, and that their connection is important.
To conclude, our results do not showa clear advantage of one method (scale) over the other;
the study only indicates that differences exist. A combination of both methods can be
recommended in order to combine the advantages of the two scales.
The results of this study are useful for market researchers. New scales for measuring
destination image, as well as a methodology for a detailed analysis of the differences in
measurements are introduced. This paper also provides several reasons for the importance
of the inclusion of emotions in the context of destination image measurement.
Limitations of this study are related to the sample and the research design. Because we
used convenience sampling, the results cannot be generalised. Respondents were solely
Austrians, thus only one culture was represented; therefore, this study should be replicated
in other countries and cultures. The experimental design of the study also entails limitations
regarding the comparability of the scales. The sample homogeneity was tested in terms of
demographics and psychographics, but ultimately, reliability could be improved using a
panel design where intra-subject comparisons are possible.
Future work could reduce the number of categories presented to respondents into a
balanced subset; for example, those on the axes or at the corners of the circumplex model.
Further, ineffective stimuli should be replaced with new, more effective ones, and the number
of stimuli per category could be reduced. This could especially improve the picture-based
measurement approach as presentational capabilities usually are limited (24 pictures do not
?t well on one page, and having to navigate through a multi-page layout makes the
association task a lengthier, more onerous process). Last but not least, other research
designs could be applied, and future research could focus on the use of pictorial stimuli in
combination with verbal items.
References
Ajzen, I. and Fishbein, M. (1980), Understanding Attitudes and Predicting Social Behavior, Prentice Hall,
New York, NY.
Bagozzi, P. and Burnkrant, R. (1985), ‘‘Attitude organization and the attitude-behavior relation: a reply to
Dillon and Kumar’’, Journal of Personality and Social Psychology, Vol. 49, pp. 47-57.
Bagozzi, R.P., Gopinath, M. and Nyer, P.U. (1999), ‘‘The role of emotions in marketing’’, Journal of the
Academy of Marketing Science, Vol. 27, pp. 184-206.
Baloglu, S. and Love, C. (2005), ‘‘A cognitive-affective positioning analysis of convention cities:
an extension of the circumplex model of affect’’, Tourism Analysis, Vol. 9, pp. 299-308.
Baloglu, S. and McCleary, K.W. (1999), ‘‘A model of destination image formation’’, Annals of Tourism
Research, Vol. 26, pp. 868-97.
Beerli, A. and Mart? ´n, J. (2004), ‘‘Factors in?uencing destination image’’, Annals of Tourism Research,
Vol. 31, pp. 657-81.
Berger, H., Denk, M., Dittenbach, M., Merkl, W. and Pesenhofer, A. (2007), ‘‘Quo vadis homo turisticus?
Towards a picture-based tourist pro?ler’’, in Sigala, M. (Ed.), Information and Communication
Technologies in Tourism, ENTER Conference 2007, Ljubljana, Springer, Vienna, pp. 87-96.
Borg, I. and Groenen, P. (2005), Modern Multidimensional Scaling: Theory and Applications, 2nd ed.,
Springer, New York, NY.
Bradley, M.M. and Lang, P.J. (1994), ‘‘Measuring emotion: the self-assessment manikin and the
semantic differential’’, Journal of Behavioral Therapy and Experimental Psychiatry, Vol. 25, pp. 49-59.
Chon, K. (1990), ‘‘The role of destination image in tourism: a reviewand discussion’’, Revue de Tourisme,
Vol. 2, pp. 2-9.
Coombs, C.H. (1950), ‘‘Psychological scaling without a unit of measurement’’, Psychological Review,
Vol. 57, pp. 145-58.
VOL. 6 NO. 4 2012
j
INTERNATIONAL JOURNAL OF CULTURE, TOURISM AND HOSPITALITY RESEARCH
j
PAGE 337
D
o
w
n
l
o
a
d
e
d
b
y
P
O
N
D
I
C
H
E
R
R
Y
U
N
I
V
E
R
S
I
T
Y
A
t
2
2
:
2
0
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
Cox, T.F. and Cox, M.A.A. (2001), Multidimensional Scaling, 2nd ed., London.
De Leeuw, W.J. and Heiser, J. (1980), ‘‘Multidimensional scaling with restrictions on the con?guration’’,
in Krishnaiah, P.R. (Ed.), Multivariate Analysis, North-Holland, Amsterdam, pp. 501-22.
De Leeuw, J. and Mair, P. (2009), ‘‘Multidimensional scaling using majorization: SMACOF in R’’, Journal
of Statistical Software, Vol. 31 No. 3, pp. 1-30, available at: at: www.jstatsoft.org/v31/i03/ (accessed
23 June 2010).
Gartner, W. (1989), ‘‘Tourism image: attribute measurement of state tourism products using
multidimensional scaling techniques’’, Journal of Travel Research, Vol. 28, pp. 16-20.
Goodall, B. (1988), ‘‘How tourists choose holidays’’, in Goodall, B. and Ashworth, G. (Eds), Marketing in
the Tourism Industry, Routledge, London.
Gower, J.C. (1971), ‘‘Statistical methods of comparing different multivariate analyses of the same data’’,
in Hodson, F.R., Kendall, D.G. and Tautu, P. (Eds), Mathematics in the Archaeological and Historical
Sciences, Edinburgh University Press, Edinburgh, pp. 138-49.
Herzig, O.A. (1991), Markenbilder, Markenwelten – Neue Wege in der Imageforschung,
Service-Fachverlag der Wirtschaftsuniversita¨ t Wien, Vienna.
Holbrook, M.B. (1986), ‘‘Emotion in the consumption experience: toward a new model of the human
consumer’’, in Peterson, R., Hoyer, W. and Wilson, W. (Eds), The Role of Affect in Consumer Behavior, DC
Heath, Lexington, MA.
Izard, C.E. (1977), Human Emotions, Plenum Press, New York, NY.
Jaccard, P. (1908), ‘‘Nouvelles recherches sur la distribution ?orale’’, Bulletin de la Socie´ te´ Vaudoise des
Sciences Naturelles, Vol. 44, pp. 223-70.
Jackson, D.A. (1995), ‘‘PROTEST: a procrustean randomization test of community environment
concordance’’, E
´
coscience, Vol. 2 No. 3, pp. 297-303.
Jenkins, O. (1999), ‘‘Understanding and measuring tourist destination image’’, International Journal of
Tourism Research, Vol. 1, pp. 1-15.
Kosslyn, S., Segar, M.C., Pani, J. and Hillger, L.A. (1990), ‘‘When is imagery used? A diary study’’,
Journal of Mental Imagery, Vol. 14, pp. 131-52.
Lang, P.J., Bradley, M.M. and Cuthbert, B.N. (2008), ‘‘International Affective Picture System (IAPS):
affective ratings of pictures and instruction manual’’, Technical Report A-8, University of Florida,
Gainesville, FL.
Mazanec, J.A. (1978), Strukturmodelle des Konsumverhaltens, Wirtschaftsuniversita¨ t, Vienna.
Mazanec, J.A. (2010), ‘‘Tourism-receiving countries in connotative Google Space’’, Journal of Travel
Research, Vol. 49, pp. 501-12.
Milman, A. and Pizam, A. (1995), ‘‘The role of awareness and familiarity with a destination: the central
Florida case’’, Journal of Travel Research, Vol. 33, pp. 21-7.
Moutinho, L. (1987), ‘‘Consumer behavior in tourism’’, European Journal of Marketing, Vol. 21, pp. 3-44.
Plutchik, R. (1962), The Emotions: Facts, Theories and a New Model, Random House, New York, NY.
Plutchik, R. (1997), Circumplex Models of Personality and Emotions, American Psychological
Association, Washington, DC.
RDevelopment Core Team(2005), R: ALanguage andEnvironment for Statistical Computing, RFoundation
for Statistical Computing, Vienna, available at: www.R-project.org (accessed 16 July 2010).
Richins, M.L. (1997), ‘‘Measuring emotions in the consumption experience’’, Journal of Consumer
Research, Vol. 24, pp. 127-46.
Rossi, P.E., Allenby, G. and McCulloch, R. (2005), Bayesian Statistics and Marketing, Wiley, New York, NY.
Russell, J.A. (1980), ‘‘A circumplex model of affect’’, Journal of Personality and Social Psychology,
Vol. 39 No. 6, pp. 1161-78.
Russell, J.A. (1989), ‘‘Measures of emotion’’, Emotion: Theory, Research and Experience, Volume 4: The
Measurement of Emotions, Academic Press, San Diego, CA, pp. 83-112.
PAGE 338
j
INTERNATIONAL JOURNAL OF CULTURE, TOURISM AND HOSPITALITY RESEARCH
j
VOL. 6 NO. 4 2012
D
o
w
n
l
o
a
d
e
d
b
y
P
O
N
D
I
C
H
E
R
R
Y
U
N
I
V
E
R
S
I
T
Y
A
t
2
2
:
2
0
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
Russell, J.A. and Mehrabian, A. (1977), ‘‘Evidence for a three-factor-theory of emotions’’, Journal of
Research in Personality, Vol. 11, pp. 273-94.
Russell, J.A. and Pratt, G. (1980), ‘‘A description of the affective quality attributed to environments’’,
Journal of Personality and Social Psychology, Vol. 38 No. 2, pp. 311-22.
Sadoski, M. and Paivio, A. (2001), Imagery and Text: A Dual Coding Theory of Reading and Writing,
Lawrence Erlbaum Associates, Mahwah, NJ.
Schneider, J.W. and Borlund, B. (2007), ‘‘Matrix comparison, Part 2: measuring the similarity between
proximity measures of ordination results by use of the Mantel and Procrustes statistics’’, Journal of the
American Society for Information Science and Technology, Vol. 58 No. 11, pp. 1596-609.
Walmsley, D. and Young, M. (1998), ‘‘Evaluative images and tourism: the use of perceptual constructs to
describe the structure of destination images’’, Journal of Travel Research, Vol. 36, pp. 65-9.
Zaltman, G. (1997), ‘‘Rethinking market research: putting people back in’’, Journal of Marketing
Research, Vol. 34, pp. 424-37.
Further reading
Mazanec, J.A. (2010), ‘‘European tourist cities in connotative Google space’’, in Mazanec, J.A. and
Wo¨ ber, K. (Eds), Analysing International City Tourism, Springer, Vienna, pp. 191-210.
Corresponding author
Ilona Pezenka can be contacted at: [email protected]
VOL. 6 NO. 4 2012
j
INTERNATIONAL JOURNAL OF CULTURE, TOURISM AND HOSPITALITY RESEARCH
j
PAGE 339
To purchase reprints of this article please e-mail: [email protected]
Or visit our web site for further details: www.emeraldinsight.com/reprints
D
o
w
n
l
o
a
d
e
d
b
y
P
O
N
D
I
C
H
E
R
R
Y
U
N
I
V
E
R
S
I
T
Y
A
t
2
2
:
2
0
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
This article has been cited by:
1. Carl Marcussen. 2014. Multidimensional scaling in tourism literature. Tourism Management Perspectives 12, 31-40. [CrossRef]
D
o
w
n
l
o
a
d
e
d
b
y
P
O
N
D
I
C
H
E
R
R
Y
U
N
I
V
E
R
S
I
T
Y
A
t
2
2
:
2
0
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
doc_812377260.pdf
This study focuses on the emotional aspects of destinations and employs two different
scales for capturing the affective component of city destination image. The aim of this paper is not only to
measure the emotions assigned to different European cities, but also to compare these two
instruments/scales by means of Procrustes analysis.
International Journal of Culture, Tourism and Hospitality Research
Measuring the resemblance between pictorial and verbal city image spaces
Ilona Pezenka Christian Buchta
Article information:
To cite this document:
Ilona Pezenka Christian Buchta, (2012),"Measuring the resemblance between pictorial and verbal city image spaces", International J ournal
of Culture, Tourism and Hospitality Research, Vol. 6 Iss 4 pp. 326 - 339
Permanent link to this document:
http://dx.doi.org/10.1108/17506181211265068
Downloaded on: 24 January 2016, At: 22:20 (PT)
References: this document contains references to 43 other documents.
To copy this document: [email protected]
The fulltext of this document has been downloaded 257 times since 2012*
Users who downloaded this article also downloaded:
Ulrike Gretzel, Yeong-Hyeon Hwang, Daniel R. Fesenmaier, (2012),"Informing destination recommender systems design and evaluation
through quantitative research", International J ournal of Culture, Tourism and Hospitality Research, Vol. 6 Iss 4 pp. 297-315 http://
dx.doi.org/10.1108/17506181211265040
Sangwon Park, Daniel R. Fesenmaier, (2012),"Non-response bias in internet-based advertising conversion studies", International J ournal of
Culture, Tourism and Hospitality Research, Vol. 6 Iss 4 pp. 340-355 http://dx.doi.org/10.1108/17506181211265077
Sara Dolnicar, Friedrich Leisch, (2012),"One legacy of Mazanec: binary questions are a simple, stable and valid measure
of evaluative beliefs", International J ournal of Culture, Tourism and Hospitality Research, Vol. 6 Iss 4 pp. 316-325 http://
dx.doi.org/10.1108/17506181211265059
Access to this document was granted through an Emerald subscription provided by emerald-srm:115632 []
For Authors
If you would like to write for this, or any other Emerald publication, then please use our Emerald for Authors service information about
how to choose which publication to write for and submission guidelines are available for all. Please visit www.emeraldinsight.com/
authors for more information.
About Emerald www.emeraldinsight.com
Emerald is a global publisher linking research and practice to the benefit of society. The company manages a portfolio of more than
290 journals and over 2,350 books and book series volumes, as well as providing an extensive range of online products and additional
customer resources and services.
Emerald is both COUNTER 4 and TRANSFER compliant. The organization is a partner of the Committee on Publication Ethics (COPE) and
also works with Portico and the LOCKSS initiative for digital archive preservation.
*Related content and download information correct at time of download.
D
o
w
n
l
o
a
d
e
d
b
y
P
O
N
D
I
C
H
E
R
R
Y
U
N
I
V
E
R
S
I
T
Y
A
t
2
2
:
2
0
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
Measuring the resemblance between
pictorial and verbal city image spaces
Ilona Pezenka and Christian Buchta
Abstract
Purpose – This study focuses on the emotional aspects of destinations and employs two different
scales for capturing the affective component of city destination image. The aimof this paper is not only to
measure the emotions assigned to different European cities, but also to compare these two
instruments/scales by means of Procrustes analysis.
Design/methodology/approach – The authors collected measurements on two different scales (verbal
and pictorial) for capturing the emotional (affective) component of destination image – both based on
Russell’s circumplex model of affect – in two independent surveys.
Findings – Signi?cant differences were found between the multidimensional scaling (MDS) results of
both scales. Because the two samples match in terms of demographics and psychographics, the
differences of the perceptual spaces are likely due to the form of stimuli (pictures compared to verbal
items) presented. The results indicate that pictures are easy to use, but they also are subject to broader
interpretations. In contrast, verbal items are more concise, but perhaps respondents may ?nd it harder
to assign them to cities (especially negative emotions).
Research limitations/implications – The paper contributes to the literature by suggesting a
methodology for detailed analysis of the differences in measurements, and introducing implications that
apply primarily to researchers once new measurement methods are introduced. The limitations of this
research relate to the sample (convenience sample): the respondents were solely Austrians, and thus
only one culture was represented.
Originality/value – Although several other researchers suggest measuring both the cognitive and the
affective image, very few studies incorporate both aspects in evaluating destination image. In contrast,
this study applies different scales to incorporate emotions in destination image measurement and
demonstrates the applicability of the scales in the tourism context.
Keywords Destination image, Emotions, Circumplex model of affect, Procrustes statistics,
Multidimensional scaling, Tourism, Place identity
Paper type Research paper
Introduction
Most destination image studies to date emphasise the cognitive component (beliefs,
knowledge, and experience) of destination image and totally ignore emotions and feelings,
which are referred to as the ‘‘affective’’ aspects. Because destination choice is believed to
be dominated by intangible factors (Walmsley and Young, 1998; Beerli and Mart? ´n, 2004) –
such as, for example, emotions – city destination marketers should place more importance
on the emotional perception of their city.
The importance of pictures as emotional elements of marketing communication strategies is
demonstrated by the fact that in most tourism brochures, more than 75 per cent of the content
is pictorial (Jenkins, 1999). Especially in tourism, pictures often serve as means to evoke a ?rst
impression of what one might expect on their journey or destination (Berger et al., 2007).
Therefore, this study concentrates on the emotional aspects of destinations, and employs two
different scales for capturing the affective component of city destination image.
PAGE 326
j
INTERNATIONAL JOURNAL OF CULTURE, TOURISM AND HOSPITALITY RESEARCH
j
VOL. 6 NO. 4 2012, pp. 326-339, Q Emerald Group Publishing Limited, ISSN 1750-6182 DOI 10.1108/17506181211265068
Ilona Pezenka is a
Researcher and Christian
Buchta is a Researcher and
Lecturer, both at the
Institute for Tourism and
Leisure Studies, Vienna
University of Economics
and Business, Vienna,
Austria.
Received March 2011
Revised June 2011
Accepted September 2011
This work was supported by the
Jubila¨ umsfonds der
Oesterreichischen
Nationalbank.
D
o
w
n
l
o
a
d
e
d
b
y
P
O
N
D
I
C
H
E
R
R
Y
U
N
I
V
E
R
S
I
T
Y
A
t
2
2
:
2
0
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
Because human information processing is largely based on images (Zaltman, 1997), and as
Kosslyn et al. (1990) state, two-thirds of information reaches the brain through the visual
system, this study also uses pictorial stimuli. The picture scale is indented to facilitate
expression of emotions and feelings evoked by cities. Also, the present researchers assume
that the less boring and more diversi?ed survey procedure of visual-based scales lowers
respondents’ fatigue, and thus ensures a higher response rate (Herzig, 1991).
Theory
Destination image
The scienti?c community widely agrees that destination image plays a key role in tourist
behaviour (Moutinho, 1987; Gartner, 1989; Chon, 1990; Goodall, 1988). Researchers have
suggested that destinations that have a positive image are more likely to be considered in
the destination selection process (Goodall, 1988).
Regarding the de?nition and operationalisation of destination image, we may identify several
streams of research. Some scientists (e.g. Milman and Pizam, 1995) view image solely as a
holistic concept, as demonstrated by the de?nition of Crompton (1979), who states that
destination image is the ‘‘sum of beliefs, ideas, and impressions that a person has of a
destination’’. Others postulate that it consists of different components – cognitive aspects
(beliefs, knowledge, and experience) and affective aspects (feelings), which all form the
global image. This classi?cation is useful when analysing the structure of destination image
and ?nally predicting consumer behaviour (Bagozzi and Burnkrant, 1985), and it also helps
to analyse factors in?uencing destination image formation and to reveal causal relationships.
The approach is also consistent with the theory of attitudes (Ajzen and Fishbein, 1980).
According to Mazanec (1978), images can be distinguished from attitudes by their having a
more distinctive emotional component. In contrast to emotions, attitudes can be stored for
long periods of time, their connection to volition and action is not particularly strong and
direct, and ?nally, arousal is unnecessary in connection with attitudes, but it is an important
part of emotions (Bagozzi et al., 1999, p. 188).
Although several authors suggest measuring both cognitive and the affective image
components, very few studies incorporate both aspects in evaluating destination image.
Mazanec (2010) uses an affect-centred interpretation of destination image when studying
the supply-side induced (projected) image, by analysing the connotations attached to
country names in textual web content. Baloglu and McCleary (1999) and Baloglu and Love
(2005) apply Russell’s (1980) circumplex model of affect to assess survey-based affective
image components.
The lack of destination image studies employing scales to capture feelings and emotions
may be due to the complex nature of emotions and their measurement.
Emotions
Bagozzi et al. (1999, p. 184) describe the term ‘‘affect’’ as ‘‘an umbrella for a set of more
speci?c mental processes including emotions, moods, and (possibly) attitudes’’. Here, the
authors refer to emotions whenever it comes to the affective aspects of destination image.
The plethora of theories of emotions leads to different approaches of measurement (see
Bagozzi et al., 1999 and Richins, 1997, for comprehensive reviews). In marketing research,
the empirical measurement of emotions mostly relies on self-reports (questionnaires), with
the number of items ranging up to 180 (Bagozzi et al., 1999). Several authors have tried to
identify the basic dimensions of emotions (Plutchik, 1962; Izard, 1977; Russell and
Mehrabian, 1977; Holbrook, 1986), and have successfully demonstrated the circular
arrangement of these base emotions (Plutchik, 1997; Russell, 1989).
This paper employs one of the most widely studied multidimensional approaches for
capturing affective responses: Russell’s (1980) circumplex model of affect. The model
suggests that the meaning attributed to environments can be de?ned by two orthogonal
bipolar dimensions:
VOL. 6 NO. 4 2012
j
INTERNATIONAL JOURNAL OF CULTURE, TOURISM AND HOSPITALITY RESEARCH
j
PAGE 327
D
o
w
n
l
o
a
d
e
d
b
y
P
O
N
D
I
C
H
E
R
R
Y
U
N
I
V
E
R
S
I
T
Y
A
t
2
2
:
2
0
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
1. pleasure/valence (pleasant/unpleasant); and
2. arousal (arousing/sleepy).
Any affective quality can be de?ned as a combination of the two base components, and
therefore can be presented in the affective space. In conclusion, Russell (1980) identi?es
and places eight affect concepts in a circular order around the two-dimensional bipolar
space (see Figure 1). Two additional dimensions (i.e. exciting/gloomy and
distressing/relaxing) help to de?ne the quadrants of the affective space, but do not form
independent dimensions. Due to its simplicity, Russell’s (1980) scale has been successfully
employed by a number of authors (e.g. Baloglu and McCleary, 1999).
Research method
Research design
Regarding the selection of city destinations, 12 European cities – Amsterdam, Barcelona,
Berlin, Budapest, Hamburg, London, Munich, Paris, Prague, Venice, Vienna and Zurich – were
used for both studies. These cities were chosen because they rank among the most visited city
break destinations by Austrian tourists travelling within Europe (cf. TourMIS – a touristic
marketing information systemthat provides electronic tourismstatistics; see www.tourmis.info).
We collected data from two separate samples. Both samples were instructed to assign the
respective stimuli to 12 European cities. For the ?rst sample (471 respondents), pictorial
stimuli were selected from the International Affective Picture System (IAPS; Lang et al., 2008).
According to Russell (1980), these stimuli are standardised on the basis of ratings of pleasure
and arousal. The second sample (367 respondents) was instructed to assign verbal items
(taken from Russell and Pratt, 1980). In order to test the homogeneity of the samples, we
collected additional information (demographics, travel experience, preferences). The surveys
were conducted through two self-administered online questionnaires (Sample 1: June 2009;
Sample 2: August 2009). Because we employed convenience sampling in both studies, the
URL address for the online surveys was publicised through e-mail distributions (WU survey
mailing list) and the respondents were solely Austrian tourists.
Sample 1 (pictures)
Respondents of Sample 1 were instructed to judge whether or not the emotions conveyed by
the presented pictures correspond to the cities in study. The instructions stressed
Figure 1 Russell’s circumplex model of affect
PAGE 328
j
INTERNATIONAL JOURNAL OF CULTURE, TOURISM AND HOSPITALITY RESEARCH
j
VOL. 6 NO. 4 2012
D
o
w
n
l
o
a
d
e
d
b
y
P
O
N
D
I
C
H
E
R
R
Y
U
N
I
V
E
R
S
I
T
Y
A
t
2
2
:
2
0
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
respondents to assign the pictures on the basis of the feelings they evoked, rather than on
their actual content. The photographs that formed the affective stimuli of the scale were
selected from the IAPS (Lang et al., 2008) through a multi-step process. The IAPS is a
database of more than 1,000 emotionally evocative colour photographs which are used in
experimental investigations of emotions worldwide. The pictorial stimuli are standardised on
the basis of ratings of pleasure (valence) and arousal by using the self-assessment manikin
(SAM; Bradley and Lang, 1994). The resulting normative scores (mean values and standard
deviations) for males, females and both sexes combined are available for each picture of the
IAPS. Because the IAPS is based on the dimensional view of affect according to Russell
(1980) and the circumplex model was already adapted in a tourism destination context
(e.g. Baloglu and McCleary, 1999; Baloglu and Love, 2005), it seemed to ?t well with the
present study.
In a ?rst step, we excluded apparently offensive pictures as well as low-quality pictures, and
also eliminated pictures judged differently by sex. Preliminary results show that these
pictures cause great variance in assignment patterns. In a second step, empirical
card-sorting studies, using WebSorte, a web-based software application, were conducted
to further reduce photographs. For the ?rst study the affective space of the selected 245
IAPS images was subdivided into three groups of emotional valence, with 36 images for the
subgroup ‘‘unpleasant’’, 112 images for the subgroup ‘‘neutral’’ and 97 images for the
subgroup ‘‘pleasant’’. Three different sorting tasks using three different samples
(unpleasant, n ¼ 19; neutral, n ¼ 18; pleasant, n ¼ 17) were performed, and respondents
were asked to sort the stimuli into the eight aforementioned emotional categories with an
additional ‘‘no classi?cation’’ option. Images with the highest category membership values
for the ‘‘no classi?cation’’ option were eliminated, and images with the ten highest
percentages of assignment in each of the categories were selected for further evaluation. A
second study, with 124 respondents sorting all 92 remaining images into the nine categories
was conducted. Again, pictures with high percentages in the ‘‘no classi?cation’’ category
were excluded from further analysis. The resulting categorisation was then compared with
the positions of the original IAPS normative values. Images with ambiguous category
membership were excluded. Because the distribution of the remaining 36 pictures was not
equal or balanced over all nine affective categories, ?ve of the 245 preliminary survey
images were added, so that ?nally, each of the eight relevant categories were represented
by the best ?tting three images to design a nonverbal scale for assessing the affective
component of a city image.
Sample 2 (verbal items)
Sample 2 was instructed to assign verbal items (taken from Russell and Pratt, 1980) to 12
European cities. The original pool of 40 verbal items (?ve items per category) was reduced in
such a way that the number of items representing one category was consistent with the
number of photographs per category (Study 1). Three experts judged the terms with respect
to their adequacy to describe certain emotions evoked by city tourism destinations and
eliminated those unfamiliar or rarely used in everyday language. Finally, 16 verbal items
representing eight affective qualities were used (Figure 2)
In order to test the homogeneity of the two samples (pictorial or verbal items), several tests
were performed. x
2
tests (sex, education) show no signi?cant x
2
values. Regarding age
and travel experience, the one-way ANOVA did not provide signi?cant F-values. The
distribution of the preference ranks was tested using the Mann-Whitney test – where again,
no signi?cant differences were detected between the two samples. Table I shows the
characteristics of the two samples and the results of the homogeneity tests.
Methodology
In both studies participants were asked to associate stimuli (pictorial or verbal items) with the
same objects of interest (cities). In a ?rst step, we assumed that the relative frequencies of
stimulus-object pairs were a meaningful empirical measure of similarity between objects and
stimuli, S(X, Y). In a second step, we sought a joint representation of objects and stimuli in a
VOL. 6 NO. 4 2012
j
INTERNATIONAL JOURNAL OF CULTURE, TOURISM AND HOSPITALITY RESEARCH
j
PAGE 329
D
o
w
n
l
o
a
d
e
d
b
y
P
O
N
D
I
C
H
E
R
R
Y
U
N
I
V
E
R
S
I
T
Y
A
t
2
2
:
2
0
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
low-dimensional space. The data samples are independent, and therefore two independent
representations exist. Finally, in a third step, we compared multidimensional scaling (MDS)
ordination results by means of Procrustes analysis in order to ?nd an optimal
superimposition of the two con?gurations.
Table I Demographics and homogeneity of the samples
Verbal Pictures
n Per cent n Per cent x
2
/F/U p
Total number of respondents 367 471
Abandonment rate 21.8 44.8
Sex 1.267 0.146
Male 115 31 165 35
Female 252 69 306 65
Age
Mean 26.20 26.38 0.13 0.719
Standard deviation 7.44 7.21
Education 5.093 0.278
Elementary school 0.3 1.3
Secondary education 3.3 3.2
A levels 72.2 70.3
University 24.3 25.3
Travel experience
a
1.40 0.277
Mean 4.35 4.27
Standard deviation 1.07 1.10
Preferences 14.42
Top three cities Barcelona, London, Paris Barcelona, London, Paris
Least preferred cities Munich, Vienna, Zurich/Budapest Vienna, Budapest, Munich
Note:
a
Short trips þ vacation trips
Figure 2 Verbal items per category used
PAGE 330
j
INTERNATIONAL JOURNAL OF CULTURE, TOURISM AND HOSPITALITY RESEARCH
j
VOL. 6 NO. 4 2012
D
o
w
n
l
o
a
d
e
d
b
y
P
O
N
D
I
C
H
E
R
R
Y
U
N
I
V
E
R
S
I
T
Y
A
t
2
2
:
2
0
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
In the MDS literature, the problem outlined for the second step is known as the unfolding
problem (Borg and Groenen, 2005, Coombs, 1950). Technically, this problem may be
considered a special case of MDS where, in the present context, the object-object and
stimulus-stimulus proximities are missing. However, and especially when values are missing
systematically, unfolding solutions are known to be prone to degeneracies (see the
discussion in Borg and Groenen, 2005). In order to avoid this problem, we suggest replacing
the missing values with estimates from the data on the one hand, and basing them on
external information on the other.
First, we assume that the unknown proximities between objects (cities) can be approximated
by the joint frequency that two objects are associated with the same stimulus by the same
respondent. More formally, we have all the ingredients of a cross-tabulation of two binary
variables, fromwhich we can compute some measure of similarity (for example, see Cox and
Cox, 2001). We obtained good results using the Jaccard coef?cient (Jaccard, 1908) in order
to replace the missing object-object proximities.
Next, we suggest replacing the missing stimulus-stimulus proximities with (Euclidean)
distances between the known (hypothetical) locations of the presented pictorial (or verbal)
stimuli in the IAPS (circumplex) space. Unfortunately, the literature does not report positions
of the verbal items in the circumplex space (Russell and Pratt, 1980). The best we can do
under these circumstances is to presume that the verbal items are located at their theoretical
positions. This is similar to using an unbiased prior in Bayesian model estimation (see Rossi
et al., 2005). In doing so, we assume the verbal items to be located at the grid points of the
circumplex model that correspond with an item’s category (see also the discussion
regarding Figure 3). Further, the object-object similarities above must be transformed to
distances in the range of the IAPS space (or vice versa).
The important point here is that we postulate that the circumplex model is complete and
unbiased. Consequently, the MDS problems to be solved in the third step must be restricted
Figure 3 Joint pictorial and verbal measurement map
VOL. 6 NO. 4 2012
j
INTERNATIONAL JOURNAL OF CULTURE, TOURISM AND HOSPITALITY RESEARCH
j
PAGE 331
D
o
w
n
l
o
a
d
e
d
b
y
P
O
N
D
I
C
H
E
R
R
Y
U
N
I
V
E
R
S
I
T
Y
A
t
2
2
:
2
0
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
to two dimensions, and their solutions must be transformed to ?t the MDS representation of
the circumplex model. More precisely, the MDS con?guration of a speci?c mode of
measurement (pictorial or verbal) must be ?tted to the model representation of the same
mode. This third step is presented in more detail below. Further, the model representations of
both modes represent the same space for a joint representation of the different approaches
to measurement in order to be meaningful. Finally, we restrict the MDS solutions to metric,
which ensures that the distances between stimuli (see above) are preserved. Accordingly,
we obtain a mapping of the cities into the IAPS-calibrated space of emotions.
Another, more substantial, reason why we do not suggest estimating the stimulus-stimulus
proximities from the data is that in general, we cannot expect the objects to be
representative of the measurement space. The assignment frequencies may bias the
stimulus-stimulus relations. However, in an exploratory setting, pooling the data sets would
be recommended.
Last, we seek transformations of MDS solutions that make themcomparable. This is known in
the literature (Borg and Groenen, 2005; Gower, 1971) as the Procrustes problem, which can
be stated more formally as follows: approximate (?t) the MDS con?guration Y (the testee) to X
(the target) so that:
kY 2f ð:Þk
2
! min !
s:t: f ðs; t; T ; XÞ ¼ sXT þ t1
0
;
where s is a (scalar) dilation factor, T is an orthogonal transformation satisfying TT
0
¼ T
0
T ¼
1 (a re?ection and rotation matrix) and t is a translation (vector). Note that f(.) de?nes a
transformation of the coordinate system of Y, and therefore of any points that were not used
in the model ?tting. Therefore, although our MDS solutions contain both objects and stimuli,
we use only one or the other for transformation. More precisely, let X(Y) denote the
con?guration of objects and stimuli in the pictorial (verbal) MDS solution, and X
0
(Y
0
) the
con?guration of the stimuli in the circumplex model, respectively. As outlined above, we
need to determine the transformations: X !X
0
and Y !ðY
0
!X
0
) in order to obtain a joint
representation of the pictorial and verbal measurements. Remember that X
0
and Y
0
were
assumed to share the same space (coordinate system). Therefore, only the magnitudes of
the scales can differ, so we must ?t Y
0
to X
0
subject to t ¼ ð0; 0Þ and T ¼ 1.
We assess the goodness of ?t (sums of squares) of a transformation using permutation tests
(Jackson, 1995; Schneider and Borlund, 2007). This involves sampling the distribution of the
?t under random permutations of the correspondence mapping between points in X and
points in Y. The probability value is then the percentage of samples with a ?t smaller or equal
than observed.
The analysis was done using the R environment for statistical computing (R Development
Core Team, 2005), in particular, package SMACOF (De Leeuw and Mair, 2009) for
multidimensional scaling, which is based on the scaling by majorising a complex function
approach (De Leeuw and Heiser, 1980). For Procrustes transformations and permutation
testing, we used our own implementations. However, see package VEGAN (Oksanen et al.,
2010) which provides, amongst others, an implementation of Procrustes transformations,
Protest and the Mantel test.
Results
Initially, we show that the suggested approach is reliable. To illustrate, we computed the
average IAPS (verbal) object scores across respondents. Because respondents could
associate an object with more than one (redundant) stimulus, the scores must be interpreted
with caution. For example, from a high score, we cannot tell whether the emotional
perception of a city was intense (more than one item from the same category was
associated) or multifaceted (more than one category was associated).
Figures 4 and 5 show that the MDS solutions reproduce the scores very well. In Figure 4 the
con?guration of the pictorial scores (grey circles) was ?tted to the object con?guration of the
PAGE 332
j
INTERNATIONAL JOURNAL OF CULTURE, TOURISM AND HOSPITALITY RESEARCH
j
VOL. 6 NO. 4 2012
D
o
w
n
l
o
a
d
e
d
b
y
P
O
N
D
I
C
H
E
R
R
Y
U
N
I
V
E
R
S
I
T
Y
A
t
2
2
:
2
0
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
Figure 4 Fitted pictorial scores
Figure 5 Fitted verbal scores
VOL. 6 NO. 4 2012
j
INTERNATIONAL JOURNAL OF CULTURE, TOURISM AND HOSPITALITY RESEARCH
j
PAGE 333
D
o
w
n
l
o
a
d
e
d
b
y
P
O
N
D
I
C
H
E
R
R
Y
U
N
I
V
E
R
S
I
T
Y
A
t
2
2
:
2
0
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
pictorial measurements. The largest deviations occur for Venice, Vienna, and Barcelona. The
probability that the ?t could be due to chance is insigni?cant (0.001, n ¼ 1; 000). For the
verbal measurements (black dots), we see in Figure 5 that the ?t is slightly worse (but still
signi?cant at 0.001, n ¼ 1; 000) compared to the pictorial case. The largest deviations occur
for Hamburg, Budapest, and again, Vienna. Finally, returning to the aforementioned
ambiguities with scores, scaling the response variables to sumto 1 (0) across stimuli, we did
not observe signi?cant differences with respect to Figures 4 and 5. This suggests that a
change in the range of the scores only occurred which was compensated by the dilation
factor s. However, closer inspection of the data reveals that individual responses are
redundant and multifaceted.
We conclude that the approach in constructing proximities and their mapping into a lower
dimensional space via metric MDS is reliable in the sense that the observed overall
(average) city positions are preserved.
We nowcheck the ?t of pictorial and verbal stimuli to their known (theoretical) positions in the
measurement space (see Figures 6 and 7). In Figure 6 the black dots indicate the locations
(mean scale values) of the pictures in the IAPS space. The grey circles indicate the ?tted
locations of the MDS con?guration. The image codes and the categories they belong to are
indicated by the labels. Closer inspection of the frequency data suggests, as earlier
discussed, that the deviations re?ect differences in stimulus usage by respondents. For
example, the comparatively less-used images 5920 (0.166) and 5940 (0.105) from the
‘‘arousing’’ category shift towards the periphery; or the more frequently used image 9590
(0.454) fromthe ‘‘unpleasant’’ category shifts towards the centre. Thus, shifts can be seen as
the result of distortions in usage of stimuli ‘‘on’’ objects, and the obvious conclusion is that
stimuli might not be representative either. However, we cannot be sure, because no variation
exists in the object sets (see discussion above). In the verbal case (see Figure 7), we had to
?ll in the reference positions due to lack of external data. Thus, the items are collapsed to the
ideal locations of their category (black dots). For example, the ‘‘arousing’’ item from the
Figure 6 Fit of pictorial stimuli
PAGE 334
j
INTERNATIONAL JOURNAL OF CULTURE, TOURISM AND HOSPITALITY RESEARCH
j
VOL. 6 NO. 4 2012
D
o
w
n
l
o
a
d
e
d
b
y
P
O
N
D
I
C
H
E
R
R
Y
U
N
I
V
E
R
S
I
T
Y
A
t
2
2
:
2
0
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
‘‘arousing’’ category is a (signi?cantly) less-used city attribution (compared to ‘‘active’’ and
‘‘alive’’). Overall, the pictorial measurements (SSE ¼ 12:93) seem to be more biased than
the verbal ones (SSE ¼ 8:39).
In Figure 3 all the ?gures presented so far are superimposed into a joint map of the pictorial
(black dots) and verbal (grey circle) measurements. The categories are indicated by the
(convex) hulls of their members. Ideally, these regions would be near their theoretical
centres, at (22, 2 2), (0, 2 2), . . . (2, 2). The centres de?ne an ideal division of the
measurement space into convex regions. The empirical one is indicated by the Voronoi
tessellation of the centre points of the hulls (solid grey lines) plus the origin (dashed grey
lines). For example, London is predominantly perceived as ‘‘unpleasant’’, but also as the
location in the centre area indicates, multifaceted and thus possibly ambivalent. Under
verbal measurement the cities shift signi?cantly in the ‘‘pleasant’’ and/or ‘‘arousing’’
direction. Table II shows that the category ‘‘intensity’’ is higher than in the pictorial setting;
that is, in the same category, more terms are associated than pictures. The most pertinent
difference between pictorial and verbal perception occurs for Venice, where the dominance
on the ‘‘arousal’’ dimension is reversed.
For an overall comparison of the two modes of measurements, we computed the Spearman
rank correlation between the (relative) city category frequencies (see Table II) and its
probability value under two regimes of permutation sampling. Under the ?rst, the cities are
?xed in order to assess the correspondence of the categories across samples; under the
second, the categories are ?xed in order to assess the correspondence of the cities. This
measure is stricter than varying objects and categories simultaneously. The correlation is
positive, but low (0.113), and both probability values are 0 (for n ¼ 10; 000 samples). Thus, a
signi?cant difference exists between the pictorial and verbal ‘‘perception’’ of the cities.
Comparing the total observed stimulus usage of pictures (0.364) and terms (0.224) to the
more detailed ?gures in Table II do not seemto contradict such an assumption. Furthermore,
Figure 7 Fit of verbal stimuli
VOL. 6 NO. 4 2012
j
INTERNATIONAL JOURNAL OF CULTURE, TOURISM AND HOSPITALITY RESEARCH
j
PAGE 335
D
o
w
n
l
o
a
d
e
d
b
y
P
O
N
D
I
C
H
E
R
R
Y
U
N
I
V
E
R
S
I
T
Y
A
t
2
2
:
2
0
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
in 72 (of 96) cases, the percentage of pictures associated is equal or larger than the
percentage of terms.
The stress values for the pictorial and verbal MDS solutions are 0.0832 and 0.0772, as
reported by package SMACOF; or 0.1289 and 0.0918 using Kruskal’s measure (see Kruskal
1964a, b). The former values are both signi?cant (0, n ¼ 100) under sampling of the distance
matrix from the uniform distribution (scaled to the observed range).
Discussion and limitations
This research ?rst shows that the proposed method of constructing proximities and mapping
into a lower dimensional space via metric MDS is reliable, in the sense that observed overall
(average) city positions are preserved. We then checked the ?t of pictorial and verbal stimuli
to their known (theoretical) positions in the measurement space according to Russell’s
circumplex model of affect. Results show a ?t of the empirical data to the theoretical
circumplex model of affect. Thus, this model offers a sound explanation of different classes
of emotions evoked by city destinations. To test whether this model is applicable to
destinations in general, replication studies should be carried out.
Regarding the comparison of the ordinations, results indicate a signi?cant difference
between the pictorial and verbal ‘‘perception’’ of the cities. We try to explain this result thusly:
verbal associations may re?ect the dominant sentiments towards a city formed from
processing all of one’s impressions; while visual associations may indicate that the set of
visual impressions is actually diverse. Another explanation might be that pictures are richer
in content than words; and therefore, a comparatively larger number of pictures ?ts one’s
image of a city. These explanations can be supported by the dual-coding theory (Sadoski
Table II Percentage stimulus attribution £ 100, categories £ cities £ ðpictorial; verbalÞ
Arousing Distressing Exciting Gloomy Pleasant Relaxing Sleepy Unpleasant
Amsterdam Pictorial 30 37 50 36 43 40 49 49
Verbal 7 16 19 15 20 39 38 18
Barcelona Pictorial 36 27 47 33 42 38 33 41
Verbal 11 23 23 17 16 45 56 25
Berlin Pictorial 27 38 46 51 40 26 38 52
Verbal 17 22 20 20 13 38 37 18
Budapest Pictorial 21 35 17 54 48 37 56 46
Verbal 10 15 23 20 18 31 17 18
Hamburg Pictorial 18 23 32 38 32 28 35 30
Verbal 5 17 20 12 18 32 27 15
London Pictorial 18 42 41 44 36 30 44 43
Verbal 14 23 23 25 11 46 47 20
Munich Pictorial 16 9 43 28 51 57 39 14
Verbal 4 20 23 17 25 30 30 16
Paris Pictorial 25 28 39 41 62 38 55 37
Verbal 14 18 25 17 10 45 46 20
Prague Pictorial 19 31 21 54 53 38 60 37
Verbal 9 17 22 12 17 39 28 18
Venice Pictorial 13 17 9 32 43 21 48 32
Verbal 8 12 28 16 15 44 33 22
Vienna Pictorial 17 17 48 43 77 61 52 24
Verbal 8 23 29 18 20 46 47 23
Zurich Pictorial 14 8 42 28 44 53 44 13
Verbal 4 18 24 17 23 28 26 16
PAGE 336
j
INTERNATIONAL JOURNAL OF CULTURE, TOURISM AND HOSPITALITY RESEARCH
j
VOL. 6 NO. 4 2012
D
o
w
n
l
o
a
d
e
d
b
y
P
O
N
D
I
C
H
E
R
R
Y
U
N
I
V
E
R
S
I
T
Y
A
t
2
2
:
2
0
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
and Paivio, 2001), which also assumes that differences in the qualities of verbal and
nonverbal units exist, and that their connection is important.
To conclude, our results do not showa clear advantage of one method (scale) over the other;
the study only indicates that differences exist. A combination of both methods can be
recommended in order to combine the advantages of the two scales.
The results of this study are useful for market researchers. New scales for measuring
destination image, as well as a methodology for a detailed analysis of the differences in
measurements are introduced. This paper also provides several reasons for the importance
of the inclusion of emotions in the context of destination image measurement.
Limitations of this study are related to the sample and the research design. Because we
used convenience sampling, the results cannot be generalised. Respondents were solely
Austrians, thus only one culture was represented; therefore, this study should be replicated
in other countries and cultures. The experimental design of the study also entails limitations
regarding the comparability of the scales. The sample homogeneity was tested in terms of
demographics and psychographics, but ultimately, reliability could be improved using a
panel design where intra-subject comparisons are possible.
Future work could reduce the number of categories presented to respondents into a
balanced subset; for example, those on the axes or at the corners of the circumplex model.
Further, ineffective stimuli should be replaced with new, more effective ones, and the number
of stimuli per category could be reduced. This could especially improve the picture-based
measurement approach as presentational capabilities usually are limited (24 pictures do not
?t well on one page, and having to navigate through a multi-page layout makes the
association task a lengthier, more onerous process). Last but not least, other research
designs could be applied, and future research could focus on the use of pictorial stimuli in
combination with verbal items.
References
Ajzen, I. and Fishbein, M. (1980), Understanding Attitudes and Predicting Social Behavior, Prentice Hall,
New York, NY.
Bagozzi, P. and Burnkrant, R. (1985), ‘‘Attitude organization and the attitude-behavior relation: a reply to
Dillon and Kumar’’, Journal of Personality and Social Psychology, Vol. 49, pp. 47-57.
Bagozzi, R.P., Gopinath, M. and Nyer, P.U. (1999), ‘‘The role of emotions in marketing’’, Journal of the
Academy of Marketing Science, Vol. 27, pp. 184-206.
Baloglu, S. and Love, C. (2005), ‘‘A cognitive-affective positioning analysis of convention cities:
an extension of the circumplex model of affect’’, Tourism Analysis, Vol. 9, pp. 299-308.
Baloglu, S. and McCleary, K.W. (1999), ‘‘A model of destination image formation’’, Annals of Tourism
Research, Vol. 26, pp. 868-97.
Beerli, A. and Mart? ´n, J. (2004), ‘‘Factors in?uencing destination image’’, Annals of Tourism Research,
Vol. 31, pp. 657-81.
Berger, H., Denk, M., Dittenbach, M., Merkl, W. and Pesenhofer, A. (2007), ‘‘Quo vadis homo turisticus?
Towards a picture-based tourist pro?ler’’, in Sigala, M. (Ed.), Information and Communication
Technologies in Tourism, ENTER Conference 2007, Ljubljana, Springer, Vienna, pp. 87-96.
Borg, I. and Groenen, P. (2005), Modern Multidimensional Scaling: Theory and Applications, 2nd ed.,
Springer, New York, NY.
Bradley, M.M. and Lang, P.J. (1994), ‘‘Measuring emotion: the self-assessment manikin and the
semantic differential’’, Journal of Behavioral Therapy and Experimental Psychiatry, Vol. 25, pp. 49-59.
Chon, K. (1990), ‘‘The role of destination image in tourism: a reviewand discussion’’, Revue de Tourisme,
Vol. 2, pp. 2-9.
Coombs, C.H. (1950), ‘‘Psychological scaling without a unit of measurement’’, Psychological Review,
Vol. 57, pp. 145-58.
VOL. 6 NO. 4 2012
j
INTERNATIONAL JOURNAL OF CULTURE, TOURISM AND HOSPITALITY RESEARCH
j
PAGE 337
D
o
w
n
l
o
a
d
e
d
b
y
P
O
N
D
I
C
H
E
R
R
Y
U
N
I
V
E
R
S
I
T
Y
A
t
2
2
:
2
0
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
Cox, T.F. and Cox, M.A.A. (2001), Multidimensional Scaling, 2nd ed., London.
De Leeuw, W.J. and Heiser, J. (1980), ‘‘Multidimensional scaling with restrictions on the con?guration’’,
in Krishnaiah, P.R. (Ed.), Multivariate Analysis, North-Holland, Amsterdam, pp. 501-22.
De Leeuw, J. and Mair, P. (2009), ‘‘Multidimensional scaling using majorization: SMACOF in R’’, Journal
of Statistical Software, Vol. 31 No. 3, pp. 1-30, available at: at: www.jstatsoft.org/v31/i03/ (accessed
23 June 2010).
Gartner, W. (1989), ‘‘Tourism image: attribute measurement of state tourism products using
multidimensional scaling techniques’’, Journal of Travel Research, Vol. 28, pp. 16-20.
Goodall, B. (1988), ‘‘How tourists choose holidays’’, in Goodall, B. and Ashworth, G. (Eds), Marketing in
the Tourism Industry, Routledge, London.
Gower, J.C. (1971), ‘‘Statistical methods of comparing different multivariate analyses of the same data’’,
in Hodson, F.R., Kendall, D.G. and Tautu, P. (Eds), Mathematics in the Archaeological and Historical
Sciences, Edinburgh University Press, Edinburgh, pp. 138-49.
Herzig, O.A. (1991), Markenbilder, Markenwelten – Neue Wege in der Imageforschung,
Service-Fachverlag der Wirtschaftsuniversita¨ t Wien, Vienna.
Holbrook, M.B. (1986), ‘‘Emotion in the consumption experience: toward a new model of the human
consumer’’, in Peterson, R., Hoyer, W. and Wilson, W. (Eds), The Role of Affect in Consumer Behavior, DC
Heath, Lexington, MA.
Izard, C.E. (1977), Human Emotions, Plenum Press, New York, NY.
Jaccard, P. (1908), ‘‘Nouvelles recherches sur la distribution ?orale’’, Bulletin de la Socie´ te´ Vaudoise des
Sciences Naturelles, Vol. 44, pp. 223-70.
Jackson, D.A. (1995), ‘‘PROTEST: a procrustean randomization test of community environment
concordance’’, E
´
coscience, Vol. 2 No. 3, pp. 297-303.
Jenkins, O. (1999), ‘‘Understanding and measuring tourist destination image’’, International Journal of
Tourism Research, Vol. 1, pp. 1-15.
Kosslyn, S., Segar, M.C., Pani, J. and Hillger, L.A. (1990), ‘‘When is imagery used? A diary study’’,
Journal of Mental Imagery, Vol. 14, pp. 131-52.
Lang, P.J., Bradley, M.M. and Cuthbert, B.N. (2008), ‘‘International Affective Picture System (IAPS):
affective ratings of pictures and instruction manual’’, Technical Report A-8, University of Florida,
Gainesville, FL.
Mazanec, J.A. (1978), Strukturmodelle des Konsumverhaltens, Wirtschaftsuniversita¨ t, Vienna.
Mazanec, J.A. (2010), ‘‘Tourism-receiving countries in connotative Google Space’’, Journal of Travel
Research, Vol. 49, pp. 501-12.
Milman, A. and Pizam, A. (1995), ‘‘The role of awareness and familiarity with a destination: the central
Florida case’’, Journal of Travel Research, Vol. 33, pp. 21-7.
Moutinho, L. (1987), ‘‘Consumer behavior in tourism’’, European Journal of Marketing, Vol. 21, pp. 3-44.
Plutchik, R. (1962), The Emotions: Facts, Theories and a New Model, Random House, New York, NY.
Plutchik, R. (1997), Circumplex Models of Personality and Emotions, American Psychological
Association, Washington, DC.
RDevelopment Core Team(2005), R: ALanguage andEnvironment for Statistical Computing, RFoundation
for Statistical Computing, Vienna, available at: www.R-project.org (accessed 16 July 2010).
Richins, M.L. (1997), ‘‘Measuring emotions in the consumption experience’’, Journal of Consumer
Research, Vol. 24, pp. 127-46.
Rossi, P.E., Allenby, G. and McCulloch, R. (2005), Bayesian Statistics and Marketing, Wiley, New York, NY.
Russell, J.A. (1980), ‘‘A circumplex model of affect’’, Journal of Personality and Social Psychology,
Vol. 39 No. 6, pp. 1161-78.
Russell, J.A. (1989), ‘‘Measures of emotion’’, Emotion: Theory, Research and Experience, Volume 4: The
Measurement of Emotions, Academic Press, San Diego, CA, pp. 83-112.
PAGE 338
j
INTERNATIONAL JOURNAL OF CULTURE, TOURISM AND HOSPITALITY RESEARCH
j
VOL. 6 NO. 4 2012
D
o
w
n
l
o
a
d
e
d
b
y
P
O
N
D
I
C
H
E
R
R
Y
U
N
I
V
E
R
S
I
T
Y
A
t
2
2
:
2
0
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
Russell, J.A. and Mehrabian, A. (1977), ‘‘Evidence for a three-factor-theory of emotions’’, Journal of
Research in Personality, Vol. 11, pp. 273-94.
Russell, J.A. and Pratt, G. (1980), ‘‘A description of the affective quality attributed to environments’’,
Journal of Personality and Social Psychology, Vol. 38 No. 2, pp. 311-22.
Sadoski, M. and Paivio, A. (2001), Imagery and Text: A Dual Coding Theory of Reading and Writing,
Lawrence Erlbaum Associates, Mahwah, NJ.
Schneider, J.W. and Borlund, B. (2007), ‘‘Matrix comparison, Part 2: measuring the similarity between
proximity measures of ordination results by use of the Mantel and Procrustes statistics’’, Journal of the
American Society for Information Science and Technology, Vol. 58 No. 11, pp. 1596-609.
Walmsley, D. and Young, M. (1998), ‘‘Evaluative images and tourism: the use of perceptual constructs to
describe the structure of destination images’’, Journal of Travel Research, Vol. 36, pp. 65-9.
Zaltman, G. (1997), ‘‘Rethinking market research: putting people back in’’, Journal of Marketing
Research, Vol. 34, pp. 424-37.
Further reading
Mazanec, J.A. (2010), ‘‘European tourist cities in connotative Google space’’, in Mazanec, J.A. and
Wo¨ ber, K. (Eds), Analysing International City Tourism, Springer, Vienna, pp. 191-210.
Corresponding author
Ilona Pezenka can be contacted at: [email protected]
VOL. 6 NO. 4 2012
j
INTERNATIONAL JOURNAL OF CULTURE, TOURISM AND HOSPITALITY RESEARCH
j
PAGE 339
To purchase reprints of this article please e-mail: [email protected]
Or visit our web site for further details: www.emeraldinsight.com/reprints
D
o
w
n
l
o
a
d
e
d
b
y
P
O
N
D
I
C
H
E
R
R
Y
U
N
I
V
E
R
S
I
T
Y
A
t
2
2
:
2
0
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
This article has been cited by:
1. Carl Marcussen. 2014. Multidimensional scaling in tourism literature. Tourism Management Perspectives 12, 31-40. [CrossRef]
D
o
w
n
l
o
a
d
e
d
b
y
P
O
N
D
I
C
H
E
R
R
Y
U
N
I
V
E
R
S
I
T
Y
A
t
2
2
:
2
0
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
doc_812377260.pdf