Description
The relevant literature suggests that website designers should consider the needs of their
target groups. This study aims to show the importance of certain website content/applications as
perceived by specific user segments, and reveal a posteriori segments based on motivational factors for
reading user-generated content (UGC). The study then seeks to visualize the connections between
segments, their perceived importance of website applications, and further explanatory variables, by
applying correspondence analysis (CA). The authors show that creative usage of CA may give insight
into the varying contributions of certain variables through the exclusion of scale categories or segments.
International Journal of Culture, Tourism and Hospitality Research
Mapping segments accessing user-generated content and website applications in a joint space
Margit Kastner Brigitte Stangl
Article information:
To cite this document:
Margit Kastner Brigitte Stangl, (2012),"Mapping segments accessing user-generated content and website applications in a joint space",
International J ournal of Culture, Tourism and Hospitality Research, Vol. 6 Iss 4 pp. 389 - 404
Permanent link to this document:
http://dx.doi.org/10.1108/17506181211265103
Downloaded on: 24 January 2016, At: 22:20 (PT)
References: this document contains references to 53 other documents.
To copy this document: [email protected]
The fulltext of this document has been downloaded 475 times since 2012*
Users who downloaded this article also downloaded:
Caterina Presi, Charalampos Saridakis, Susanna Hartmans, (2014),"User-generated content behaviour of the dissatisfied service
customer", European J ournal of Marketing, Vol. 48 Iss 9/10 pp. 1600-1625 http://dx.doi.org/10.1108/EJ M-07-2012-0400
Margherita Pagani, Ronald E. Goldsmith, Charles F. Hofacker, (2013),"Extraversion as a stimulus for user-generated content", J ournal of
Research in Interactive Marketing, Vol. 7 Iss 4 pp. 242-256 http://dx.doi.org/10.1108/J RIM-11-2012-0052
Albert Barreda, Anil Bilgihan, (2013),"An analysis of user-generated content for hotel experiences", J ournal of Hospitality and Tourism
Technology, Vol. 4 Iss 3 pp. 263-280 http://dx.doi.org/10.1108/J HTT-01-2013-0001
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
)
Mapping segments accessing
user-generated content and website
applications in a joint space
Margit Kastner and Brigitte Stangl
Abstract
Purpose – The relevant literature suggests that website designers should consider the needs of their
target groups. This study aims to show the importance of certain website content/applications as
perceived by speci?c user segments, and reveal a posteriori segments based on motivational factors for
reading user-generated content (UGC). The study then seeks to visualize the connections between
segments, their perceived importance of website applications, and further explanatory variables, by
applying correspondence analysis (CA). The authors show that creative usage of CA may give insight
into the varying contributions of certain variables through the exclusion of scale categories or segments.
Design/methodology/approach – The authors collected 440 completed questionnaires in an online
survey. Of the 240 respondents who read UGC, the authors clustered motivational factors by applying a
vector quantization method, and then used CA to give insights into the importance of website
content/applications for certain segments. The paper explains howmatrices can be simpli?ed in order to
facilitate interpretation, and applies Rasch analysis to ensure the accuracy of this simpli?cation.
Findings – The results indicate that six segments exist with different motivations for accessing UGC:
enthusiasts, mavericks, tips and price optimizers, safety players, uncommercials, and avoiders. For
these different segments, the perceived importance of diverse website content/applications vary. The
authors show that interpretation may be simpli?ed, without the loss of substantive information, by
combining scale levels and excluding neutral categories. The Rasch analysis also supports combining
categories.
Research limitations/implications – The authors also show how the demonstration of certain effects
can be enhanced by animated graphics, and that these can then be embedded into PDF ?les. However,
embedding of animations only makes sense for digital articles or media in general; in a printed version,
the reader would need to be redirected to a website.
Practical implications – Social media website providers need to be aware that diverse segments
perceive the importance of content/applications differently, and designers should customize a website
accordingly. Finally, and in terms of methodology, this paper highlights how CA is valuable for
management presentations because it displays categorical data in an easy-to-read graph format.
Originality/value – No research has hitherto shed light on the connection between the perceived
importance of website content/applications and the motivational factors for accessing UGC. This paper
contributes to ?lling this gap.
Keywords User-generated content, Segmentation, Website design, Content analysis,
Topology-representing networks, Rasch, Web sites, Internet
Paper type Research paper
Introduction
The importance of social media and Web 2.0 sites is evidenced in the way search engines,
which are the number one online source in tourism, rank such sites highly. Rationales given for
the favorable positions of Web 2.0 sites in search engine results relate to their being frequently
updated due to large numbers of users, and their tendency to embed numerous hyperlinks
(Xiang and Gretzel, 2010). The opportunity for users to provide content on Web 2.0 sites
further increases the bulk of information available online. This prospect should encourage
DOI 10.1108/17506181211265103 VOL. 6 NO. 4 2012, pp. 389-404, Q Emerald Group Publishing Limited, ISSN 1750-6182
j
INTERNATIONAL JOURNAL OF CULTURE, TOURISM AND HOSPITALITY RESEARCH
j
PAGE 389
Margit Kastner is an
E-Developer at the Institute
for Tourism and Leisure
Studies, Vienna University
of Economics and
Business, Vienna, Austria.
Brigitte Stangl is a Project
Leader at the Institute for
Tourism and Leisure
Research, HTW Chur, Chur,
Switzerland.
Received March 2011
Revised June 2011
Accepted September 2011
The authors would like to
express their sincere gratitude
to Professor Josef Mazanec for
his supervision,
encouragement, and support in
understanding not only the
subject, but also the
importance of methods. The
authors also thank Professor
Mazanec for providing the time
and resources for them to learn
sophisticated methods of data
analyses, and wish him all the
best, with time to follow his
passion for research.
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
)
scientists to come up with new or improved applications aimed at structuring information,
enhancing the quality of search results, and generally handling information overload better
(Ricci, 2008). The increasing number of applications available forces website designers to be
aware of the needs of the target group of a speci?c website (Sullivan, 1997), because different
requirements appear due to users’ knowledge, experience, search strategies, goals, needs
(De Marsico and Levialdi, 2004), or diverse learning styles (Holtze, 2000, Stangl and
Dickinger, 2010). The design of a website should satisfy usability and content requirements of
its users without providing unnecessary applications/content to reduce excessive information
(Di Mascio and Tarantino, 2003). Because tourists are becoming more experienced in
searching information online and they increasingly know what they expect from a website
regarding content and applications offered, a satisfying site needs to adjust
content/applications based on its target groups’ requirements (Perfetti, 2001). This implies
that the tourism supply side needs to understand motivational factors of people accessing
user-generated content (UGC) as a source of information. Several researchers focus on
matters such as motivations for reading and providing UGC (e.g. Gretzel et al., 2007; Chau
et al., 2002). However, no research has hitherto shed light on the connection between the
perceived importance of website content/applications and the motivational factors for
accessing UGC. This paper contributes to ?lling this research gap.
With respect to motivations for reading UGC, the present research detects, pro?les, and plots
segments, along with users’ perceived importance of website content/applications, by
applying a correspondence analysis (CA). The creative usage of CA also allows for an
indication of the appropriateness of scale levels used, the effects of the middle category, and
the differences pertaining to two types of travelers (i.e. relaxation- and adventure-seekers).
The remainder of the paper is organized as follows: the next section presents relevant literature
about website design, followed by previous studies on motivational aspects for reading UGCor
electronic word-of-mouth (e-WOM). This includes information on sensation-seeking, which is
the underlying construct for differentiating between relaxation- and adventure-seekers. The
Methodology section includes questionnaire details and descriptions of methods used to
analyze the data. The Results section starts with the sample description, then moves to
detailing the characteristics of the segments’ pro?les. Here we also inspect and describe the
contributions of the segments to the dimensions in the CA, and examine the scale level. The
next section carries out a subset analysis excluding the middle category, and demonstrates the
resulting effects, and then compares the outcome of the scale level inspection to the results
fromthe Rasch analysis. Finally, we incorporate the type of traveler as an additional explanatory
variable before discussing the results and implications.
Theoretical background
Design considerations are essential for Web 2.0 sites where unstructured content published
by users for other users’ needs to be made accessible in a structured way and in a convenient
mode (Di Mascio and Tarantino, 2003). The design of a website comprises information
presentation and appearance, access navigation orientation, and informative content
architecture (De Marsico and Levialdi, 2004). All aspects must accord with the target users’
needs (Sullivan, 1997), which implies that the more accurately a site architecture matches the
mental model of users, the more users will be satis?ed. This premise is true for the content as
well as for applications provided on a website (Norman, 2002). Equally important is that
website elements communicate correct meanings to users (Mandel, 2002), thus, the interface
must translate applications and operations of a system in a clear way and complete users’
inquiries effectively and successfully (De Marsico and Levialdi, 2004). A tailored website
comprising appropriately adjusted content/applications allows the development of
long-standing communities of shared interests and sub-cultural identi?cation (Schmidt, 2007).
If website designers know which content/applications users perceive as important, and also
have knowledge about users’ motivation for visiting a certain site, they can tailor websites in
order to achieve an optimal design for each user group. Concerning motivation for
accessing Web 2.0 sites and reading e-WOM, the study by Goldsmith and Horowitz (2006)
shows that users are motivated by their desire to reduce risks, and ?nd pre-purchase
PAGE 390
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
)
information and low prices. Saving decision-making time and making better buying
decisions (Hennig-Thurau and Walsh, 2003), meeting and in?uencing other people (Lenhart
and Fox, 2006, Nardi et al., 2004), and ?nding hotels, restaurants, and things to do (Gretzel
et al., 2007) are further important reasons for accessing UGC; however, the importance
attached to each motivational factor may vary according to target users’ characteristics.
According to theories of optimum stimulation level (OSL), users differ in their level of
curiosity, variety-seeking, exploratory behavior, and risk-taking (Raju, 1980). Thus,
motivations for reading UGC may vary, based on the level of sensation-seeking; for
example, the degree of risk people are prepared to take on during a vacation (Zuckermann,
1979, 1994). People with a high need for sensation-seeking indicated that they did not
search for particular things and did not plan everything in advance in this study; they like to
be ?exible during a trip. On the other hand, lowsensation-seekers did not like surprises while
on vacation; they sought security and non-adventurous activities (Litvin, 2008; Mayo and
Jarvis, 1981). As a consequence of sensation-seeking, people search for different
information online; hence, knowledge of users’ sensation-seeking levels should allow
website designers to cater for the two extreme segments: low and high sensation-seekers
(called relaxation- and adventure-seekers for the remainder of this paper).
Based on previous literature, the present article aims to shed light on the relationship
between design aspects and individuals’ motivations to read UGC. We reveal
user-segments based on motivations for accessing UGC, and highlight differences
between segments pertaining to their perceived importance of content/applications on
websites. Finally, we demonstrate even more speci?c insights into design requirements by
differentiating between relaxation- and adventure-seekers.
Methodology
This research involved developing a standardized online questionnaire, with items for
motivational concepts adapted from previous studies by Goldsmith and Horowitz (2006),
Hennig-Thurau and Walsh (2003), and from a report by Gretzel et al. (2007). The
questionnaire comprises 19 items dealing with motivations for reading UGC (see the
Appendix), which are measured on a three-point-scale (1 ¼ strongly disagree, 2 ¼ neutral,
3 ¼ strongly agree). We also carried out an analysis of content/applications on travel-related
Web 2.0 platforms. This method consequently generated the following 16
content/applications, and the paper assesses their perceived importance on a ?ve-point
Likert scale (1 ¼ strongly disagree to 5 ¼ strongly agree):
1. videos;
2. photos;
3. interactive trip planner;
4. number of users;
5. number of reviews;
6. rating of reviews;
7. special deals;
8. comparing prices;
9. booking;
10. connecting with locals;
11. meeting travel buddies;
12. creating your own map;
13. subscription to a newsletter;
14. traveler forum;
VOL. 6 NO. 4 2012
j
INTERNATIONAL JOURNAL OF CULTURE, TOURISM AND HOSPITALITY RESEARCH
j
PAGE 391
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
)
15. destination description; and
16. attraction description.
The above numbers indicate the respective content/application in the plots. The
questionnaire includes items regarding demographics, travel and online searching
behavior, internet experience, and type of traveler (relaxation- or adventure-seeker). To
avoid respondent fatigue, we used none of the usual comprehensive sensation-seeking
scales (Hoyle et al., 2002; Zuckermann, 1996). Instead, we asked individuals to indicate
directly whether they were relaxation- or adventure-seekers on a ten-point rating scale. We
then framed the question with respect to the majority of travel they undertook.
For data analysis, in a ?rst step, we used topology-representing networks (TRNs), applying
the neural-gas algorithm by Martinetz and Schulten (1991). The algorithm is based on the
competitive learning principle, which means that typical cases are ‘‘rivalling to approximate
the frequency distribution of the empirical data’’ (Mazanec, 2001, p. 898). The application of
this vector quantization method, which combines the k-means objective function with
stochastic approximation, aims to identify unobserved heterogeneity with respect to
motivations for reading UGC. TRN is a further development of k-means clustering that
updates all prototypes simultaneously in training iterations, and implies that the number of
classes must be speci?ed in advance, and they are used as centroids or starting seeds for
the iteration process. We utilized the software TRN32 by Mazanec (2008, 2009) for
completing the a posteriori segmentation. In order to determine the correct number of
classes, TRN32 assists by displaying the weighted simple structure index (wSSI). The wSSI
is a heuristic measure with values from 0 to 1, which examines the contrasts between the
classes: the nearer the wSSI is to 1, the higher are the contrasts (Mazanec and Strasser,
2000). Furthermore, TRN32 allows us to determine the stability of a solution by providing a
replication procedure. This permits the presentation of the percentage of uncertainty
reduction (UR), which shows how often individual cases are assigned to the same cluster.
Yet measuring the stability is still essential, because results may depend on the starting
seeds selected (Mazanec and Strasser, 2000).
In a second step, we used the class labels of the detected segments as explanatory
variables in the CA. The group that did not read UGC at all we called ‘‘noreaders’’. Because
noreaders have no experience with UGCs as information sources, this group is an a priori
de?ned segment that differs from segments containing people who access UGC for diverse
purposes (such as hints on cooking, movie/book reviews or travel-related information). We
did not include noreaders in the cluster analysis that uses motivational reasons for reading
UGC as segmentation base. However, we did include noreaders in the CA as a further
segment. CA is a valuable technique for examining categorical data. Nominal or ordinal
scaled data is present in many ?elds of applied research. CA plots cross-tabulated data and
allows the easy grasp of patterns of numerical frequencies. CA also permits the inclusion of
additional explanatory variables. Theoretically, one can include as many explanatory
variables as desired; this is only a matter of interpretation. Furthermore, subset analyses
assist studying effects of missing and neutral responses (Greenacre and Pardo, 2005).
Generally, scales are always a sensitive issue in surveys (Clark and Schober, 1992) and
researchers must decide on the numbers of rating scale categories and whether to include
or exclude a middle alternative. Research provides mixed results concerning these issues,
and Bond and Fox (2007) state that the number of optimal response categories must be
tested empirically if an existing scale is applied to a new population or a new scale is
developed. Rugg and Cantril (1944) suggest that respondents without a clear opinion on a
topic are more likely to choose the middle alternative. Presser and Schuman (1980) show
similar biases in connection with the subject-involvement of respondents. A recent study by
Dolnicar and Gru¨ n (2010) shows that the level of agreement becomes much stronger if no
neutral position exists; hence, evasion behavior is diminished. Other researchers
demonstrate that combining ?ve ordered categories into three in the data increases the
test reliability for their sample (Zhu et al., 1997; Stone and Wright, 1994).
PAGE 392
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
)
To accomplish the analysis in the present study, we used the CA package in R (Nenadic and
Greenacre, 2007). An advantage of this package is that Greenacre’s standard biplot allows a
direct interpretation of the length of the arrows from the origin. The length of the arrows
re?ects the contribution of each variable. By analyzing the position of the variables in a CA
plot, we can determine the appropriateness of scale-levels, and researchers may decide
whether to combine categories or not. For a detailed description of CA, see Greenacre
(1993, 2007). Hair et al. (2010) also offer insights for the usefulness of CA. The present paper
focuses on plotting both row and column categories; it plots together identi?ed segments
with results of perceived importance regarding website content/applications. The
substantive variables requiring explanation comprise 16 content/applications. The
research explores plots in terms of appropriateness of the scale used and unambiguity of
interpretation, and carries out subset analyses excluding the middle alternative of the scale.
In a third step, we exemplarily compared CA results regarding scale levels with ?ndings
based on Rasch analysis. Rasch analysis is a probabilistic measurement model, which is
commonly used to examine scale-related issues (Rasch, 1960, 1980). To diagnose the
appropriateness of rating scales, a combination of criterions such as thresholds, probability
curves, and category frequencies should be used (Bond and Fox, 2007; Wright and
Masters, 1982). The Rasch parameters that re?ect the structure of the rating scale are the
thresholds (Andrich, 1978, 1998), in the sense that they are the boundaries between
categories, meaning that a ?ve-point Likert scale requires four boundaries (thresholds) to
separate the categories (Linacre, 2001). Problems occur if thresholds did not increase in a
logical, ordered manner and/or if low differences exist between the thresholds. As a rule of
thumb, the increase of thresholds should be between 1.4 and 5 logits (Linacre, 1999). With
the probability curve, one can visually inspect if thresholds are ordered and if the distances
between the thresholds are equal. In addition, category frequencies should be inspected
and, according to Linacre (1999), at least ten responses per category are required. When
the analysis indicates that problems exist concerning some categories, adjacent categories
should be merged in a thoughtful way in order to improve variable clarity (Linacre, 1999) and
the scale re-analyzed. The quality of this new scale should also be tested across different
groups of interest (Bond and Fox, 2007).
Finally, to analyze whether one can detect differences between adventure- and
relaxation-seekers, we included an additional explanatory variable in CA.
Results
Sample characteristics
In total, 440 questionnaires were usable for the purpose of this study. The data, gathered in
2009, consisted of 36.4 percent male and 63.6 percent female respondents. On average,
they were 24.9 years old, and approximately 80 percent were students. In terms of travel
experience, 14.5 percent perceived themselves as not very travel experienced, 75.9
percent said that they were experienced but not an expert, and 9.5 percent assessed
themselves to be travel experts. Concerning UGC usage, 240 persons read UGC entries.
The importance of the internet as a source of travel information is re?ected in the fact that
37.7 percent stated that they always used the internet to access travel information.
Thirty-three percent of the respondents used the internet often, 20.7 percent sometimes, 6.8
percent rarely, and only 1.4 percent did not use the internet at all. Regarding type of traveler,
the sample is divided into 145 relaxation- and 295 adventure-seekers, of which 51.7 percent
and 55.9 percent, respectively, read UGC.
Segments detected
The TRN clustering approach favored a six-segment solution, with a wSSI of 0.56. For a
six-cluster solution, a run of 50 replications results in a UR of 82.04 percent, while this is
slightly lower for a ?ve- and seven-cluster solution (78.01 percent and 81.51 percent,
respectively). However, combined considerations of UR, the size of the clusters, and
appropriateness of the results, suggest opting for six classes.
VOL. 6 NO. 4 2012
j
INTERNATIONAL JOURNAL OF CULTURE, TOURISM AND HOSPITALITY RESEARCH
j
PAGE 393
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
)
The results describe segments brie?y in descending order (by group size based on
prototype tables) and dedicate each concise labels. The description also covers pro?le
information received by cross-tabulating the segments with third variables included in the
questionnaire and by applying ANOVA. Only signi?cant results (i.e. gender, use of
travel-related UGC, travel experience, and travel partner) are presented in the pro?le
descriptions below:
B Enthusiasts (28.3 percent) – The largest segment, this represents people who stated that
they searched for UGC entries because they loved all aspects of e-WOM and enjoyed
reading UGC. They described themselves as being motivated by getting authentic and
reliable information – not offered by providers, but by travel buddies. They also liked trip
planning tools and they searched for contacts with other travelers as well as with locals.
Further reasons for accessing UGC included reducing risks, saving time, and comparing
prices or hotels. Approximately 80 percent were travel experienced but did not consider
themselves experts; 10.3 percent considered themselves expert travelers, and did a lot of
travel. In organizing a trip, 38.3 percent often or always and 60.3 percent sometimes or
rarely searched for travel-related UGC. Of the enthusiasts group, 67.6 percent were
female, and their preferred travel mate was their partner (48.5 percent), followed by
friends (33.8 percent).
B Motivated mavericks (22.1 percent) – This segment stated that they were motivated to
access UGC because they wanted to reduce poor decisions and get reliable
information. However, they said that they did not like contacting locals or other
travelers. Of this group, 83.0 percent traveled a lot but did not consider themselves
travel experts; while 7.5 percent did perceive themselves as expert travelers.
Approximately one-third rarely used travel-related UGC as a form of information, 43.4
percent used UGC sometimes, 15.1 percent used UGC often, and 7.5 percent used
UGC as a form of information for all travel. Although mavericks were loners in terms of
their motivation to read UGC, they primarily traveled with their friends (50.9 percent) or
partner (32.1 percent). Concerning gender, roughly 45 percent were male and 55
percent were female in this study.
B Tips and price optimizers (20.4 percent) – The so called ‘‘optimizers’’ indicated that they
appreciated tools which allowed them to compare prices, hotels, and destinations. They
were also interested in stories and descriptions as well as authentic photos of travel
buddies. The majority of optimizers were female (77.6 percent) in this study, and all
members of this segment used Web 2.0 platforms to search for travel-related information:
approximately one-third used UGC rarely, 36.7 percent used UGC sometimes, and 24.5
percent used UGCoften. Among optimizers, 12.2 percent perceived themselves as travel
experts. More than half traveled a lot but did not consider themselves travel experts; while
14.3 percent traveled only occasionally. Trips were primarily taken with friends (51.0
percent) or their partner (30.6 percent).
B Safety players (15.0 percent) – All but one member of this segment used travel-related
UGC to inform themselves about their journeys: nearly 30.6 percent rarely used UGC,
44.4 percent sometimes used it, and 16.7 percent often used UGC for travel planning.
The motivation to read UGC from travel buddies was reported to be in?uenced by the
wish to reduce the risk of making bad decisions. However, they were not motivated to
access UGC from other travelers for saving time or for networking purposes. The majority
of this segment traveled regularly (approximately 90 percent), half preferred going on
vacation with their partner, and approximately 40 percent with their friends. Males and
females were nearly equally distributed in this study.
B Uncommercial info searchers (7.1 percent) – Approximately 60 percent of the segment
labeled ‘‘uncommercials’’ traveled frequently and had already seen parts of the world.
This segment reported that they read travel-related UGC because they were eager to get
reliable and non-commercial information. They said that they liked hearing from other
travelers and searched for special tips and information that commercial providers did not
offer. In terms of looking for travel information on Web 2.0 sites, none in this group always
used UGC, approximately 11.8 percent often used UGC, 52.9 percent sometimes used it,
PAGE 394
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 29.4 percent rarely used it. This group was dominated by females (82.4 percent).
Approximately one-quarter did not mind traveling alone, while 41.2 percent traveled with
their friends.
B Avoiders (7.1 percent) – This segment said that they read UGC; but interestingly, they
were not very motivated to do so. They did not access travel-related Web 2.0 sites to
reduce risk, get reliable information, ?nd authentic information, have fun, or contact
others. Comparing prices somewhat motivated this segment; however, the strength of
motivation was not strong: 11.8 percent never used UGCas a source of travel information,
while 5.9 percent always searched for travel information on Web 2.0 sites, 23.5 percent
used UGCoften, 17.6 percent sometimes used UGC, and 41.2 percent rarely used it. This
group traveled the least, with 35.3 percent perceiving themselves as not travel
experienced, 29.4 percent traveling regularly but not considering themselves experts,
and 17.5 percent considering themselves experts. This cluster contained the lowest
percentage of people traveling with friends (17.6 percent). Most traveled with their
partner or family (53.0 percent), 17.6 percent traveled alone, and 11.8 percent traveled
with a tour group. In this study 52.9 percent were female.
We detected no signi?cant differences between the segments concerning age, occupation,
or internet experience. Furthermore, a one-factorial analysis of variance demonstrated that
differences existed regarding the importance of website applications; basically with
p-values of ,0.006. The research detected a 10 percent signi?cance level for ‘‘rating of
reviews’’ (p 2value ¼ 0:072) and ‘‘photos’’ (p 2value ¼ 0:090), while ‘‘number of users’’
was found to be not signi?cant.
The six segments identi?ed above constitute the substantive variable used for further
analyses with CA. As already mentioned in the methodology section, we added a seventh
segment comprising the group of people that did not read UGC at all (this segment is
henceforth called ‘‘noreaders’’).
Segments’ contribution to the dimensions
In order to visualize the results and to facilitate the interpretation of a matrix comparing all
seven segments with the importance of 16 website features, we estimated a CA. CA results
show that a two-dimensional solution is appropriate. Two dimensions represent 70.8 percent
of the inertia, while the third only accounts for approximately 10 percent. The scree plot and
eigenvalues also support two dimensions. In detail, eigenvalues of 6.8 and 2.5 ful?ll the
Kaiser criterion of being higher than 1 (Hair et al., 2010; Garson, 2008). The quality measure
provided by the CA procedure shows that safety players are represented best by the two
dimensions (97.0 percent), followed by enthusiasts (88.7 percent). The relative contribution
of safety players to the ?rst dimension is very high, at 86.8 percent, while the relative
contribution is only 10.2 percent for the second dimension. Enthusiasts (68.3 percent) also
contribute highly to the ?rst dimension. In contrast, the second dimension is mainly a
representation of noreaders (87.7 percent), but with an inertia of 18.9 percent, the second
dimension is far less important than the ?rst one (see Table I for contribution details).
Table I Segments’ contribution to the dimensions
Group
Overall
contribution
(%)
Contribution to
Dimension 1
(%)
Contribution to
Dimension 2
(%)
Safety players 97.0 86.8 10.2
Enthusiasts 88.7 68.3 20.4
Noreaders 87.8 0.1 87.7
Avoiders 33.1 23.2 10.0
Tips and price optimizers 26.0 25.8 0.2
Motivated mavericks 20.5 5.3 15.2
Uncommercial info searchers 15.5 0.0 15.5
VOL. 6 NO. 4 2012
j
INTERNATIONAL JOURNAL OF CULTURE, TOURISM AND HOSPITALITY RESEARCH
j
PAGE 395
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 1 visualizes the results described above by incorporating all information given.
Regarding the importance of website features, the length of the arrows illustrate that the
content/applications contributing most to the ?rst dimension are: ‘‘trip planners’’ (3d), ‘‘rating
of reviews’’ (6d), ‘‘special deals’’ (7d), ‘‘comparing prices’’ (8d), ‘‘connecting with locals’’
(10dd and 10a), ‘‘creating maps’’ (12d, 12a, and 12dd), and ‘‘subscription to newsletter’’
(13dd). ‘‘Photos’’ (2aa), ‘‘number of users’’ (4aa), ‘‘number of reviews’’ (5aa), and
particularly, neutral
responses regarding the importance of website content/applications
de?ne the second dimension (e.g. 6n, 15n, 16n). The mass is the weight given to a point and
is identi?ed by the different sizes of the circles. Generally, the bigger the circle, the higher
the mass, and the higher the population of the cell behind. Thus we can detect that
noreaders is a large segment, while uncommercials is quite a small one.
Before actually interpreting the results in terms of closeness of the importance values to the
segments, the focus now shifts to the ?ve-point Likert scale, which we used to measure
perceived importance of website content/applications.
Inspection of the scale level
Figure 1 shows that the categories strongly agree (aa) and agree (a) are quite close together,
as are strongly disagree (dd) and disagree (d). For instance, for the content ‘‘photos’’ (2),
2dd and 2d are displayed in close proximity, meaning they are closely related. If the
categories were signi?cantly different, they would be located further from each other in the
map. Scale levels are not displayed in an ordered way from dd on the left side to aa on right
side; this is true, for instance, for ‘‘destination description’’ (15), where d is located further left
than dd. The scale levels for ‘‘comparing prices’’ (8) are displayed in the following way:
d-n-dd-a-aa. These results demonstrate that a three-point scale (agree, neutral, and
disagree) would have been suf?cient. The smaller scale would not only visually improve the
graphic, but would also facilitate interpretation.
Figure 1 Importance of website content/applications measured on a ?ve-point-scale
PAGE 396
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 order to see if a three-point Likert scale is more appropriate, the categories aa and a are
combined, as are dd and d. Labels are changed accordingly (a ¼ agree, n ¼ neutral, and
d ¼ disagree). Compared to the ?ve-point Likert scale, results for the inertia for three
categories increase by 6.8 percent to 77.6 percent, and the graphic display of the contributions
of various categories is clearer (Figure 2). Using this format, the ?rst dimension is shown to be
mainly explained by safety players (87.3 percent), enthusiasts (77.5 percent), and optimizers
(42.0 percent), while noreaders (81.8 percent), uncommercials (32.4 percent) and mavericks
(27.8 percent) contribute almost only to the second. Avoiders contribute to both dimensions,
with 34.2 percent and 32.3 percent, respectively. A rough, parabolic left-to-right progression
from disagree via neutral to agree (left to right) is evident, which is the typical pattern of an
uni-dimensional scale (Camiz, 2005). To emphasize this pattern, the scale-levels are connected
by lines in Figure 2 and, as a showcase, the line for the content/application ‘‘connecting with
locals’’ (10) is presented in bold (Figure 2). The relation regarding explanatory variables is quite
intuitive. Enthusiasts who love all website content/applications are displayed on a diagonal line
opposite the segment avoiders, who read UGCalthough they did not perceive UGCimportant.
In contrast, noreaders are located nearest to the category n. Safety players were only motivated
by content/applications that reduced risky decisions; all the rest of the information offered on a
Web 2.0 site was not very interesting for this segment. It is apparent that they had a rather
contrary focus compared to enthusiasts and avoiders. Optimizers rated the importance of
networking and planning tools highly. Uncommercials perceived information provided by users
as important, and they did not want destination or attraction descriptions provided by
destination marketing organizations. Mavericks judged the importance of tools for booking
travel-related products low.
Inspection of the middle alternative
In order to determine whether the middle category n ( ¼ neutral) in?uenced the results, we
excluded this category in the next presentation by undertaking a subset analysis. A
comparison of the maps (a) and (b) in Figure 3 shows the consequences. By excluding n, the
inertia further increases by 5.4 percent to 83.0 percent. The segment noreaders shrinks
towards the center, while the contribution of the segments enthusiasts (87.9 percent),
Figure 2 Applications with combined scale-levels and its parabolic shape
VOL. 6 NO. 4 2012
j
INTERNATIONAL JOURNAL OF CULTURE, TOURISM AND HOSPITALITY RESEARCH
j
PAGE 397
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 3 Comparison with and without middle category: (a) CA including middle category,
(b) CA excluding middle category
PAGE 398
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
)
optimizers (54.6 percent) and safety players (92.7 percent) increases for the ?rst dimension
and decreases for the second to below 0.5 percent. On the second dimension, the
contribution of the segments mavericks (56.9 percent) and avoiders (51.0 percent) rises.
Some content/applications move towards the x-axis; thus, clearly contributing to the ?rst
dimension and no longer to the second (e.g. 5d, 6d, and 10a). Contributions by other
content/applications increase, which can easily be recognized by the length of the arrows.
Omitting the middle category reduces the contributions of the segment noreaders, and
simultaneously increases the contributions of other segments. Interpretation of the results is
facilitated because contributions become much more explicit, especially on the most
important ?rst dimension. In an animation, we may track the consequences of excluding the
middle alternative. A video showing the changes is provided (see http://tourism.wu.ac.at/ugc/
animation.htm), where the procedure to create the animation is suggested by Maier (2010).
CA scale level inspection results and Rasch analysis
The CA offers certain evidence regarding the inappropriateness of scale levels. The
following discussion presents patterns recognized in CA plots and the corresponding
?ndings through Rasch analysis. First, we have seen that the categories of some website
content/applications are not ordered using the ?ve-point Likert scale (e.g. ‘‘destination
description’’). Figure 1 obviates that 15d is displayed further left than 15dd, and the level of
agreement (15a) is positioned at the center of the CA plot, implying low contribution to the
dimensions. Findings through Rasch analysis acknowledge that thresholds do not increase
in a logical, ordered manner required by the Rasch measurement model from dd to aa:
21.56, 20.07, 20.11, and 1.74. The threshold fromlevel n to a (20.11) is less than that from
level d to n (20.07), implying that the scale is disordered. Inspection of the frequencies
indicates that 15aa has fewer than ten responses, which in turn does not ful?ll the
requirements stipulated by Linacre (1999). Furthermore, the shape of the distribution is
problematic, with nearly 55 percent of people either disagreeing or strongly disagreeing. We
mentioned earlier that low distances are evident between two categories in the CA plot; our
example was ‘‘photos’’, where 2dd and 2d are displayed close together and 2n is at the
center of the plot (see Figure 1). These ?ndings are supported by Rasch analysis, showing
low differences between the thresholds of these levels: 20.86, 20.25, 20.226, and 1.33.
Ideally, thresholds should be between 1.4 and 5 logits (Linacre, 1999); otherwise, the
category de?nition is too narrow or too many category options have been provided, which is
the case in our example. As another guideline, the frequencies suggest that categories dd
and d should be merged, because only four people strongly disagreed. Re-analyzing the
scale with only three categories shows that problems still exist with the scales because they
are not properly ordered with regards to the thresholds, and/or the corresponding logits are
below 1.4. Therefore, the exclusion of the middle category needs to be considered. Rasch
analysis supports the simpli?ed picture elaborated through the inspection of CA results.
Thus, evidence exists that different positions of the variables in the CA plot and the
corresponding scale levels gives an indication of the appropriateness of the scale.
In the next step, this more convenient depiction with only two scale levels (a ¼ agree and
d ¼ disagree) becomes the basis for including the type of traveler (relaxation- and
adventure-seeker) as an explanatory variable. This allows for further insights regarding the
perceived importance of website content/applications for two different types of travelers.
Including further explanatory variables
Figure 4 shows differences between the two types of travelers. In particular, the different
travel types of avoiders and safety players are plotted separately. The relaxation-seeking
avoiders appreciate ‘‘comparing prices’’ (7a) and ‘‘newsletters’ ’ (13a) more than
adventure-seeking avoiders. Furthermore, the ?rst group perceives ‘ ‘destination
descriptions’’ (15d) as being unimportant, while adventure-seeking avoiders perceive
‘‘photos’’ (2d), ‘‘reviews’’ (6d), and ‘‘attraction descriptions’’ (16d) to be unimportant.
VOL. 6 NO. 4 2012
j
INTERNATIONAL JOURNAL OF CULTURE, TOURISM AND HOSPITALITY RESEARCH
j
PAGE 399
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
)
Implications
Theoretical implications
A posteriori segmentation shows that heterogeneity pertaining to motivational aspects exists
for reading travel-related UGC. Differences are evident between the segments regarding the
frequency of accessing UGC for searching travel information, travel experiences, partners,
and gender. ANOVA demonstrates a difference between the segments in terms of the
perceived importance of website content/applications.
Enthusiasts, the largest segment, reported using Web 2.0 sites most often as travel
information sources. Thus, they ranked the importance of all website content/application
features as high. In contrast, avoiders hardly used UGC and reported thinking that only the
comparison of prices was a worthwhile reason for accessing it. Hence, the favorite
application for this segment is a price-comparison-tool. Optimizers were motivated by
special deals and authentic reports from other travelers, which is also re?ected in their
preferred content/applications; those which allow them to get in contact with travel buddies
or locals, to ?nd special deals, to plan the trip interactively, or to view photos and videos.
Mavericks were happy with all website content/applications but perceived features for
getting into contact with others as unimportant. Therefore, newsletters, for instance, were not
important to them. For uncommercials, review applications were most important. This
segment also searched in forums for travel information, and read destination and attraction
descriptions. For them, the availability of online booking was also perceived as important.
The motivation for safety players was to avoid risky decisions. Hence, they were interested in
review-applications, while all other features were perceived as unimportant. Pertaining to the
importance of content/applications, further differences are evident regarding type of
traveler, which seem to be intuitive, because adventure-seekers need different information
than people whose motivation for traveling is relaxation. These results con?rm ?ndings from
previous literature indicating that diverse groups of users desire different website designs
(e.g. Perfetti, 2001; Sullivan, 1997). Our ?ndings also show that noreaders have a neutral
position, which agrees with earlier studies that suggest that the middle alternative will be
chosen whenever a person has no clear opinion on a topic (Rugg and Cantril, 1944) or is
Figure 4 Importance of website regarding type of travelers
PAGE 400
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
)
uninvolved (Presser and Schuman, 1980). By opting for the neutral position, respondents
can avoid making a decision. For this reason, consideration should be given to the inclusion
of a ‘‘no opinion’’ or ‘‘not used’’ category when designing a questionnaire. In analyzing the
responses of those who did read UGC, the omission of the middle category would facilitate
interpretation without experiencing information loss; the use of a binary scale (agree and
disagree) is supported by the ?ndings of Dolnicar and Gru¨ n (2007). By applying the CA in a
creative way, this method is shown to be useful for detecting whether a higher-dimensional
scale confers additional insights – or if it is more appropriate to combine categories for the
sake of clarity without losing a lot of information. CA results of scale inspection are veri?ed by
?ndings through Rasch analysis.
The present study also demonstrates that CA is a worthwhile technique that permits
straightforward interpretations using Greenacre’s standard biplot (Greenacre, 1993). In this
plot the length of the arrows, or more generally, the distance of a point fromthe origin, can be
read as the contribution of a certain point ( ¼ variable). Subset analysis involves the
exclusion of categories, and which assists researchers and consultants to present complex
matrices that are tight and well arranged. An animation procedure proposed by Maier (2010)
seizes multimedia opportunities to further improve the demonstration of certain effects.
Managerial implications
From a managerial perspective, our ?ndings show a need to tailor a website according to
target customers’ needs. The site architecture or structure of a website must re?ect the
different requests of targeted segments (Perfetti, 2001). Beside the fact that websites
must cater to customers’ needs, an important decision website providers should consider
is how granulated a segment is determined. One can either use one large segmentation
base, such as the motivation to access UGC, or dig even deeper and take into account
an additional factor, such as in our case involving the level of sensation-seeking
(relaxation- and adventure-seekers). The more precise the segmentation undertaken, the
greater the opportunity to match site architecture and users’ mental models; however,
simultaneously, the amount of people attracted to the site consequently decreases.
Whether this is desired or not depends on whether the aim of the website is to target a
very speci?c niche market or to address a more general group of people. The biggest
segment revealed in this study is labeled enthusiasts, a segment that prefers authentic
information not provided by travel/service providers and applications that enable the user
to get in contact with travel buddies and locals. Other segments appreciate price
comparison tools or authentic photos or videos. Therefore, website providers initially must
de?ne their target group and detect what kind of information and which applications they
require. Websites can then be designed to incorporate information for the targeted user
group. Such a procedure should secure an appropriate design, resulting in satis?ed
users, which in turn leads to positive e-WOM and repeat usage of a website (DeLone and
McLean, 2003).
Limitations and future research
The relationship between patterns displayed in the CA plot and respective results of Rasch
analysis needs further investigation, including the inclusion of additional variables and the
use of various scales to allow generalizable conclusions (Bond and Fox, 2007). Another
limitation of the present study involves the sample size, which becomes quite small after
segmentation. However, the sample nevertheless allows for some preliminary insights into
different types of UGC users. Further, this study used a convenience sample, which
over-represented students; thus, follow-up studies including other cohorts are needed.
Finally, future research should expand the topic from pre-trip planning to various stages of
traveling, and examine changes of perceived importance of website content/applications in
the various phases of a trip. Differences between website content/application use for large
holidays and short breaks might also be incorporated.
VOL. 6 NO. 4 2012
j
INTERNATIONAL JOURNAL OF CULTURE, TOURISM AND HOSPITALITY RESEARCH
j
PAGE 401
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
)
References
Andrich, D. (1978), ‘‘Aratingformulation for ordered response categories’’, Psychometrika, Vol. 43 No. 4,
pp. 561-73.
Andrich, D.A. (1998), ‘‘Thresholds, steps, and rating scale conceptualization’’, Rasch Measurement
Transactions, Vol. 12 No. 3, pp. 648-9.
Bond, T.G. and Fox, C.M. (2007), Applying the Rasch Model, Lawrence ErlbaumAssociates, Mahwah, NJ.
Camiz, S. (2005), ‘‘The Guttman effect: its interpretation and a new redressing method’’, Data Analysis
Bulletin, Vol. 5, pp. 7-34.
Chau, P.Y.K., Cole, M., Massey, A.P., Montoya-Weiss, M. and O’Keefe, R.M. (2002), ‘‘Cultural differences
in the online behavior of consumers’’, Communications of the ACM, Vol. 45 No. 10, pp. 138-43.
Clark, H.H. and Schober, M.F. (1992), ‘‘Asking questions and in?uencing answers’’, in Tanur, J.M. (Ed.),
Questions about Questions: Inquiries into the Cognitive Bases of Surveys, Russell Sage, New York, NY,
pp. 15-48.
De Marsico, M. and Levialdi, S. (2004), ‘‘Evaluating web sites: exploiting users’ expectations’’,
International Journal of Human-Computer Studies, Vol. 60 No. 3, pp. 381-416.
DeLone, W.H. and McLean, E.R. (2003), ‘‘The DeLone and McLean model of information systems
success: a ten-year update’’, Journal of Management Information Systems, Vol. 19 No. 4, pp. 9-30.
Di Mascio, T. and Tarantino, L. (2003), ‘‘Advanced visual interfaces: the focus is on the user’’,
The Knowledge Engineering Review, Vol. 18 No. 2, pp. 175-81.
Dolnicar, S. and Gru¨ n, B. (2007), ‘‘How constrained a response: a comparison of binary, ordinal and
metric answer formats’’, Journal of Retailing and Consumer Services, Vol. 14 No. 2, pp. 108-22.
Dolnicar, S. and Gru¨ n, B. (2010), ‘‘Translating: between survey answer formats’’, paper presented at the
Global Marketing Conference, 9-12 September, Tokyo.
Garson, D.G. (2008), ‘‘Correspondence analysis’’, Statnotes: Topics in Multivariate Analysis, available
at: http://faculty.chass.ncsu.edu/garson/pa765/statnote.htm (accessed May 20, 2010).
Goldsmith, R.E. and Horowitz, D. (2006), ‘‘Measuring motivations for online opinion seeking’’, Journal of
Interactive Advertising, Vol. 6 No. 2, pp. 1-16.
Greenacre, M. and Pardo, R. (2005), ‘‘Multiple correspondence analysis of a subset of response
categories’’, Economics Working Papers, available at: http://econpapers.repec.org/RePEc:upf:upfgen:881
(accessed September 2, 2010).
Greenacre, M.J. (1993), Correspondence Analysis in Practice, Academic Press, London.
Greenacre, M.J. (2007), Correspondence Analysis in Practice, Chapman & Hall/CRC, Boca Raton, FL.
Gretzel, U., Yoo, K.H. and Purifoy, M. (2007), ‘‘Trip advisor online travel review study: the role and
impacts of online travel review for trip planning’’, Laboratory for Intelligent Systems in Tourism, College
Station, TX.
Hair, J.F., Anderson, R.E., Tatham, R.L. and Black, W.C. (2010), Multivariate Data Analysis. A Global
Perspective, Prentice Hall, Upper Saddle River, NJ.
Hennig-Thurau, T. and Walsh, G. (2003), ‘‘Electronic word-of-mouth: motives for and consequences of
reading customer articulations on the internet’’, International Journal of Electronic Commerce, Vol. 8
No. 2, pp. 51-74.
Holtze, T.L. (2000), ‘‘Applying learning style theory to web page design’’, Internet Reference Services
Quarterly, Vol. 5 No. 2, pp. 71-80.
Hoyle, R.H., Stephenson, M.T., Palmgreen, P., Lorch, E.P. andDonohew, R.L. (2002), ‘‘Reliability and validity
of a brief measure of sensation seeking’’, Personality and Individual Differences, Vol. 32, pp. 401-14.
Lenhart, A. and Fox, S. (2006), ‘‘Bloggers – a portrait of the internet’s new storytellers’’, available at:
www.pewinternet.org/pdfs/PIP\%20Bloggers\%20Report\%20July\%2019\%202006.pdf (accessed
August 25, 2008).
Linacre, J.M. (1999), ‘‘Investigating rating scale category utility’’, Journal of Outcome Measurement,
Vol. 3 No. 2, pp. 103-22.
PAGE 402
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
)
Linacre, J.M. (2001), ‘‘Category, step and threshold: de?nitions & disordering’’, Rasch Measurement
Transactions, Vol. 15 No. 1, p. 794.
Litvin, S.W. (2008), ‘‘Sensation seeking and its measurement for tourism research’’, Journal of Travel
Research, Vol. 46, pp. 440-5.
Maier, M. (2010), ‘‘R-code for animation procedure to demonstrate changing effects’’, unpublished
code, Institute for Statistics and Mathematics, WU Vienna, Vienna.
Mandel, T. (2002), ‘‘User/system interface design’’, available at: www.theomandel.com/docs/mandel-
encyclopedia.pdf (accessed November 20, 2008).
Martinetz, T.M. and Schulten, K.J. (1991), ‘‘A neural-gas network learns topologies’’, Arti?cial Neural
Networks, North-Holland, Amsterdam, pp. 397-402.
Mayo, E.J. and Jarvis, L.P. (1981), The Psychology of Leisure Travel – Effective Marketing and Selling of
Travel Services, CBI Publishing Company, Boston, MA.
Mazanec, J.A. (2001), ‘‘Neural market structure analysis: novel topology-sensitive methodology’’,
European Journal of Marketing, Vol. 35 Nos 7/8, pp. 894-916.
Mazanec, J.A. (2008), ‘‘TRN32 for Windows’’, available at: www.wu.ac.at/itf/downloads/software/trn32
(accessed July 14, 2010).
Mazanec, J.A. (2009), ‘‘TRN2009’’, available at: www.wu.ac.at/itf/downloads/software/trn32 (accessed
July 14, 2010).
Mazanec, J.A. and Strasser, H. (2000), A Nonparametric Approach to Perceptions-based Market
Segmentation: Foundations, Springer, Vienna.
Nardi, B.A., Schiano, D.J. and Gumbrecht, M. (2004), ‘‘Blogging as social activity, or, would you let 900
million people read your diary?’’, Proceedings of the ACM Conference on Computer Supported
Cooperative Work, Chicago, IL, ACM, New York, NY, pp. 222-31.
Nenadic, O. and Greenacre, M. (2007), ‘‘Correspondence analysis in R, with two- and three-dimensional
graphics: the CA package’’, Journal of Statistical Software, Vol. 20 No. 3, pp. 1-13.
Norman, D.A. (2002), The Design of Everyday Things, Basic Books, New York, NY.
Perfetti, C. (2001), ‘‘Personas: matching a design to the users’ goals’’, available at: www.uie.com/
articles/personas (accessed November 18, 2008).
Presser, S. and Schuman, H. (1980), ‘‘The measurement of the middle position in attitude surveys’’,
Public Opinion Quarterly, Vol. 44 No. 1, pp. 70-85.
Raju, P.S. (1980), ‘‘Optimum stimulation level: its relationship to personality, demographics, and
exploratory behavior’’, Journal of Consumer Research, Vol. 1, pp. 272-82.
Rasch, G. (1960), Probabilistic Models for Some Intelligence and Attainment Tests, University of
Chicago Press, Chicago, IL.
Rasch, G. (1980), Probabilistic Models for Some Intelligence and Attainment Tests, University of
Chicago Press, Chicago, IL.
Ricci, F. (2008), ‘‘Information search and recommendation tools’’, paper presented at the JITT Workshop
Series: Tourism, Information Search and the Internet, Vienna.
Rugg, D. and Cantril, H. (1944), ‘‘The wording of questions’’, in Cantril, H. (Ed.), Gauging Public Opinion,
Princeton University Press, Princeton, NJ, pp. 23-50.
Schmidt, J. (2007), ‘‘Blogging practices. An analytical framework’’, Journal of Computer-Mediated
Communication, Vol. 12, pp. 1409-27.
Stangl, B. and Dickinger, A. (2010), ‘‘How communication modes determine website satisfaction’’,
Proceedings of ENTER – Information Communication Technologies in Tourism, Lugano, pp. 273-84.
Stone, M.H. and Wright, B.D. (1994), ‘‘Maximizing rating scale information’’, Rasch Measurement
Transactions, Vol. 8 No. 3, p. 386.
Sullivan, T. (1997), ‘‘All things web: the value of usability’’, available at: www.pantos.org/atw/35679.html
(accessed November 8, 2008).
Wright, B.D. and Masters, G. (1982), Rating Scale Analysis, MESA Press, Chicago, IL.
VOL. 6 NO. 4 2012
j
INTERNATIONAL JOURNAL OF CULTURE, TOURISM AND HOSPITALITY RESEARCH
j
PAGE 403
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
)
Xiang, Z. and Gretzel, U. (2010), ‘‘Role of social media in online travel information search’’, Tourism
Management, Vol. 31 No. 2, pp. 179-88.
Zhu, W., Updyke, W.F. and Lewandowski, C. (1997), ‘‘Post-hoc Rasch analysis of optimal categorization
of an ordered response scale’’, Journal of Outcome Measurement, Vol. 1 No. 4, pp. 286-304.
Zuckermann, M. (1979), Sensation Seeking: Beyond the Optimal Level of Arousal, Lawrence Erlbaum
Associates, Hinsdale, NJ.
Zuckermann, M. (1994), Behavioral Expressions and Biosocial Bases of Sensation Seeking, Cambridge
University Press, Cambridge, MA.
Zuckermann, M. (1996), ‘‘Itemrevisions in the Sensation Seeking Scale formV (SSS-V)’’, Personality and
Individual Differences, Vol. 20 No. 4, p. 515.
Appendix: Questionnaire items
About the authors
Margit Kastner received her doctorate from the Vienna University of Economics and
Business (WU Vienna). She has been employed by the Institute for Tourism and Leisure
Studies, WU Vienna, since February 2000, and at present is an eDeveloper (eLearning
Consultant) in the discipline of marketing. She lectures in marketing, international tourism,
and academic writing. Her research interests are centered on issues of e-learning, learning
and consumer behavior, market research in tourism, and quantitative methodology in
marketing and educational research.
Brigitte Stangl specialized in tourism and leisure studies as well as small business and
entrepreneurship, completing her PhD in 2010 with a dissertation entitled ‘‘User-based
website design in tourism with a special focus on web 2.0’’. Between 2006 and 2010 she
undertook duties as a Research and Teaching Assistant at the Institute for Tourism and
Leisure Studies, WU Vienna, and since 1 November 2010 she has been employed as Project
Leader at the Institute for Tourism and Leisure Research, HTW Chur. Her main research
interests lie in the areas of decision support systems, web design, Web 2.0, and virtual
realities. Brigitte Stangl is the corresponding author and can be contacted at:
[email protected]
Table AI Questionnaire items
Construct I access travel-related UGC because . . . References
Reduce risk . . . the chances of making a bad decision are reduced
. . . it helps me avoid a risky decision
. . . I can hear from people who have already taken a trip
. . . I want to make sure a trip is worth taking
Goldsmith and Horowitz (2006)
Time saving . . . the chances of making a bed decision are reduced
. . . it helps me avoid a risky decision
. . . I can hear from people who have already taken a tip
. . . I want to make sure a trip is worth taking
Hennig-Thurau et al. (2003)
Reliable
information
. . . I get information from someone who is not trying to sell me something
. . . I get special tips on what to see and do in a destination
. . . I want something more than travel service providers are offering
. . . I have not found suf?cient information in other sources of information
Goldsmith and Horowitz (2006),
Hennig-Thurau et al. (2003), Gretzel
et al. (2007)
Authentic
information
. . . I can watch authentic videos of the destination
. . . I can look at authentic photos of the destination
. . . I can read authentic reviews, journals, reports of the destination
Gretzel et al. (2007)
Trip preparation
tools
. . . I can use trip planning, which makes it easier to prepare my trip
. . . I can compare prices, destinations, hotels
Gretzel et al. (2007)
Fun and social
contact
. . . trip planning is fun
. . . I can connect with locals that way
. . . I can connect with other travelers
. . . I enjoy reading UGC
Gretzel et al. (2007)
PAGE 404
j
INTERNATIONAL JOURNAL OF CULTURE, TOURISM AND HOSPITALITY RESEARCH
j
VOL. 6 NO. 4 2012
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. Rob Law, Dimitrios Buhalis, Cihan Cobanoglu. 2014. Progress on information and communication technologies in hospitality
and tourism. International Journal of Contemporary Hospitality Management 26:5, 727-750. [Abstract] [Full Text] [PDF]
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_968655195.pdf
The relevant literature suggests that website designers should consider the needs of their
target groups. This study aims to show the importance of certain website content/applications as
perceived by specific user segments, and reveal a posteriori segments based on motivational factors for
reading user-generated content (UGC). The study then seeks to visualize the connections between
segments, their perceived importance of website applications, and further explanatory variables, by
applying correspondence analysis (CA). The authors show that creative usage of CA may give insight
into the varying contributions of certain variables through the exclusion of scale categories or segments.
International Journal of Culture, Tourism and Hospitality Research
Mapping segments accessing user-generated content and website applications in a joint space
Margit Kastner Brigitte Stangl
Article information:
To cite this document:
Margit Kastner Brigitte Stangl, (2012),"Mapping segments accessing user-generated content and website applications in a joint space",
International J ournal of Culture, Tourism and Hospitality Research, Vol. 6 Iss 4 pp. 389 - 404
Permanent link to this document:
http://dx.doi.org/10.1108/17506181211265103
Downloaded on: 24 January 2016, At: 22:20 (PT)
References: this document contains references to 53 other documents.
To copy this document: [email protected]
The fulltext of this document has been downloaded 475 times since 2012*
Users who downloaded this article also downloaded:
Caterina Presi, Charalampos Saridakis, Susanna Hartmans, (2014),"User-generated content behaviour of the dissatisfied service
customer", European J ournal of Marketing, Vol. 48 Iss 9/10 pp. 1600-1625 http://dx.doi.org/10.1108/EJ M-07-2012-0400
Margherita Pagani, Ronald E. Goldsmith, Charles F. Hofacker, (2013),"Extraversion as a stimulus for user-generated content", J ournal of
Research in Interactive Marketing, Vol. 7 Iss 4 pp. 242-256 http://dx.doi.org/10.1108/J RIM-11-2012-0052
Albert Barreda, Anil Bilgihan, (2013),"An analysis of user-generated content for hotel experiences", J ournal of Hospitality and Tourism
Technology, Vol. 4 Iss 3 pp. 263-280 http://dx.doi.org/10.1108/J HTT-01-2013-0001
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
)
Mapping segments accessing
user-generated content and website
applications in a joint space
Margit Kastner and Brigitte Stangl
Abstract
Purpose – The relevant literature suggests that website designers should consider the needs of their
target groups. This study aims to show the importance of certain website content/applications as
perceived by speci?c user segments, and reveal a posteriori segments based on motivational factors for
reading user-generated content (UGC). The study then seeks to visualize the connections between
segments, their perceived importance of website applications, and further explanatory variables, by
applying correspondence analysis (CA). The authors show that creative usage of CA may give insight
into the varying contributions of certain variables through the exclusion of scale categories or segments.
Design/methodology/approach – The authors collected 440 completed questionnaires in an online
survey. Of the 240 respondents who read UGC, the authors clustered motivational factors by applying a
vector quantization method, and then used CA to give insights into the importance of website
content/applications for certain segments. The paper explains howmatrices can be simpli?ed in order to
facilitate interpretation, and applies Rasch analysis to ensure the accuracy of this simpli?cation.
Findings – The results indicate that six segments exist with different motivations for accessing UGC:
enthusiasts, mavericks, tips and price optimizers, safety players, uncommercials, and avoiders. For
these different segments, the perceived importance of diverse website content/applications vary. The
authors show that interpretation may be simpli?ed, without the loss of substantive information, by
combining scale levels and excluding neutral categories. The Rasch analysis also supports combining
categories.
Research limitations/implications – The authors also show how the demonstration of certain effects
can be enhanced by animated graphics, and that these can then be embedded into PDF ?les. However,
embedding of animations only makes sense for digital articles or media in general; in a printed version,
the reader would need to be redirected to a website.
Practical implications – Social media website providers need to be aware that diverse segments
perceive the importance of content/applications differently, and designers should customize a website
accordingly. Finally, and in terms of methodology, this paper highlights how CA is valuable for
management presentations because it displays categorical data in an easy-to-read graph format.
Originality/value – No research has hitherto shed light on the connection between the perceived
importance of website content/applications and the motivational factors for accessing UGC. This paper
contributes to ?lling this gap.
Keywords User-generated content, Segmentation, Website design, Content analysis,
Topology-representing networks, Rasch, Web sites, Internet
Paper type Research paper
Introduction
The importance of social media and Web 2.0 sites is evidenced in the way search engines,
which are the number one online source in tourism, rank such sites highly. Rationales given for
the favorable positions of Web 2.0 sites in search engine results relate to their being frequently
updated due to large numbers of users, and their tendency to embed numerous hyperlinks
(Xiang and Gretzel, 2010). The opportunity for users to provide content on Web 2.0 sites
further increases the bulk of information available online. This prospect should encourage
DOI 10.1108/17506181211265103 VOL. 6 NO. 4 2012, pp. 389-404, Q Emerald Group Publishing Limited, ISSN 1750-6182
j
INTERNATIONAL JOURNAL OF CULTURE, TOURISM AND HOSPITALITY RESEARCH
j
PAGE 389
Margit Kastner is an
E-Developer at the Institute
for Tourism and Leisure
Studies, Vienna University
of Economics and
Business, Vienna, Austria.
Brigitte Stangl is a Project
Leader at the Institute for
Tourism and Leisure
Research, HTW Chur, Chur,
Switzerland.
Received March 2011
Revised June 2011
Accepted September 2011
The authors would like to
express their sincere gratitude
to Professor Josef Mazanec for
his supervision,
encouragement, and support in
understanding not only the
subject, but also the
importance of methods. The
authors also thank Professor
Mazanec for providing the time
and resources for them to learn
sophisticated methods of data
analyses, and wish him all the
best, with time to follow his
passion for research.
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
)
scientists to come up with new or improved applications aimed at structuring information,
enhancing the quality of search results, and generally handling information overload better
(Ricci, 2008). The increasing number of applications available forces website designers to be
aware of the needs of the target group of a speci?c website (Sullivan, 1997), because different
requirements appear due to users’ knowledge, experience, search strategies, goals, needs
(De Marsico and Levialdi, 2004), or diverse learning styles (Holtze, 2000, Stangl and
Dickinger, 2010). The design of a website should satisfy usability and content requirements of
its users without providing unnecessary applications/content to reduce excessive information
(Di Mascio and Tarantino, 2003). Because tourists are becoming more experienced in
searching information online and they increasingly know what they expect from a website
regarding content and applications offered, a satisfying site needs to adjust
content/applications based on its target groups’ requirements (Perfetti, 2001). This implies
that the tourism supply side needs to understand motivational factors of people accessing
user-generated content (UGC) as a source of information. Several researchers focus on
matters such as motivations for reading and providing UGC (e.g. Gretzel et al., 2007; Chau
et al., 2002). However, no research has hitherto shed light on the connection between the
perceived importance of website content/applications and the motivational factors for
accessing UGC. This paper contributes to ?lling this research gap.
With respect to motivations for reading UGC, the present research detects, pro?les, and plots
segments, along with users’ perceived importance of website content/applications, by
applying a correspondence analysis (CA). The creative usage of CA also allows for an
indication of the appropriateness of scale levels used, the effects of the middle category, and
the differences pertaining to two types of travelers (i.e. relaxation- and adventure-seekers).
The remainder of the paper is organized as follows: the next section presents relevant literature
about website design, followed by previous studies on motivational aspects for reading UGCor
electronic word-of-mouth (e-WOM). This includes information on sensation-seeking, which is
the underlying construct for differentiating between relaxation- and adventure-seekers. The
Methodology section includes questionnaire details and descriptions of methods used to
analyze the data. The Results section starts with the sample description, then moves to
detailing the characteristics of the segments’ pro?les. Here we also inspect and describe the
contributions of the segments to the dimensions in the CA, and examine the scale level. The
next section carries out a subset analysis excluding the middle category, and demonstrates the
resulting effects, and then compares the outcome of the scale level inspection to the results
fromthe Rasch analysis. Finally, we incorporate the type of traveler as an additional explanatory
variable before discussing the results and implications.
Theoretical background
Design considerations are essential for Web 2.0 sites where unstructured content published
by users for other users’ needs to be made accessible in a structured way and in a convenient
mode (Di Mascio and Tarantino, 2003). The design of a website comprises information
presentation and appearance, access navigation orientation, and informative content
architecture (De Marsico and Levialdi, 2004). All aspects must accord with the target users’
needs (Sullivan, 1997), which implies that the more accurately a site architecture matches the
mental model of users, the more users will be satis?ed. This premise is true for the content as
well as for applications provided on a website (Norman, 2002). Equally important is that
website elements communicate correct meanings to users (Mandel, 2002), thus, the interface
must translate applications and operations of a system in a clear way and complete users’
inquiries effectively and successfully (De Marsico and Levialdi, 2004). A tailored website
comprising appropriately adjusted content/applications allows the development of
long-standing communities of shared interests and sub-cultural identi?cation (Schmidt, 2007).
If website designers know which content/applications users perceive as important, and also
have knowledge about users’ motivation for visiting a certain site, they can tailor websites in
order to achieve an optimal design for each user group. Concerning motivation for
accessing Web 2.0 sites and reading e-WOM, the study by Goldsmith and Horowitz (2006)
shows that users are motivated by their desire to reduce risks, and ?nd pre-purchase
PAGE 390
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
)
information and low prices. Saving decision-making time and making better buying
decisions (Hennig-Thurau and Walsh, 2003), meeting and in?uencing other people (Lenhart
and Fox, 2006, Nardi et al., 2004), and ?nding hotels, restaurants, and things to do (Gretzel
et al., 2007) are further important reasons for accessing UGC; however, the importance
attached to each motivational factor may vary according to target users’ characteristics.
According to theories of optimum stimulation level (OSL), users differ in their level of
curiosity, variety-seeking, exploratory behavior, and risk-taking (Raju, 1980). Thus,
motivations for reading UGC may vary, based on the level of sensation-seeking; for
example, the degree of risk people are prepared to take on during a vacation (Zuckermann,
1979, 1994). People with a high need for sensation-seeking indicated that they did not
search for particular things and did not plan everything in advance in this study; they like to
be ?exible during a trip. On the other hand, lowsensation-seekers did not like surprises while
on vacation; they sought security and non-adventurous activities (Litvin, 2008; Mayo and
Jarvis, 1981). As a consequence of sensation-seeking, people search for different
information online; hence, knowledge of users’ sensation-seeking levels should allow
website designers to cater for the two extreme segments: low and high sensation-seekers
(called relaxation- and adventure-seekers for the remainder of this paper).
Based on previous literature, the present article aims to shed light on the relationship
between design aspects and individuals’ motivations to read UGC. We reveal
user-segments based on motivations for accessing UGC, and highlight differences
between segments pertaining to their perceived importance of content/applications on
websites. Finally, we demonstrate even more speci?c insights into design requirements by
differentiating between relaxation- and adventure-seekers.
Methodology
This research involved developing a standardized online questionnaire, with items for
motivational concepts adapted from previous studies by Goldsmith and Horowitz (2006),
Hennig-Thurau and Walsh (2003), and from a report by Gretzel et al. (2007). The
questionnaire comprises 19 items dealing with motivations for reading UGC (see the
Appendix), which are measured on a three-point-scale (1 ¼ strongly disagree, 2 ¼ neutral,
3 ¼ strongly agree). We also carried out an analysis of content/applications on travel-related
Web 2.0 platforms. This method consequently generated the following 16
content/applications, and the paper assesses their perceived importance on a ?ve-point
Likert scale (1 ¼ strongly disagree to 5 ¼ strongly agree):
1. videos;
2. photos;
3. interactive trip planner;
4. number of users;
5. number of reviews;
6. rating of reviews;
7. special deals;
8. comparing prices;
9. booking;
10. connecting with locals;
11. meeting travel buddies;
12. creating your own map;
13. subscription to a newsletter;
14. traveler forum;
VOL. 6 NO. 4 2012
j
INTERNATIONAL JOURNAL OF CULTURE, TOURISM AND HOSPITALITY RESEARCH
j
PAGE 391
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
)
15. destination description; and
16. attraction description.
The above numbers indicate the respective content/application in the plots. The
questionnaire includes items regarding demographics, travel and online searching
behavior, internet experience, and type of traveler (relaxation- or adventure-seeker). To
avoid respondent fatigue, we used none of the usual comprehensive sensation-seeking
scales (Hoyle et al., 2002; Zuckermann, 1996). Instead, we asked individuals to indicate
directly whether they were relaxation- or adventure-seekers on a ten-point rating scale. We
then framed the question with respect to the majority of travel they undertook.
For data analysis, in a ?rst step, we used topology-representing networks (TRNs), applying
the neural-gas algorithm by Martinetz and Schulten (1991). The algorithm is based on the
competitive learning principle, which means that typical cases are ‘‘rivalling to approximate
the frequency distribution of the empirical data’’ (Mazanec, 2001, p. 898). The application of
this vector quantization method, which combines the k-means objective function with
stochastic approximation, aims to identify unobserved heterogeneity with respect to
motivations for reading UGC. TRN is a further development of k-means clustering that
updates all prototypes simultaneously in training iterations, and implies that the number of
classes must be speci?ed in advance, and they are used as centroids or starting seeds for
the iteration process. We utilized the software TRN32 by Mazanec (2008, 2009) for
completing the a posteriori segmentation. In order to determine the correct number of
classes, TRN32 assists by displaying the weighted simple structure index (wSSI). The wSSI
is a heuristic measure with values from 0 to 1, which examines the contrasts between the
classes: the nearer the wSSI is to 1, the higher are the contrasts (Mazanec and Strasser,
2000). Furthermore, TRN32 allows us to determine the stability of a solution by providing a
replication procedure. This permits the presentation of the percentage of uncertainty
reduction (UR), which shows how often individual cases are assigned to the same cluster.
Yet measuring the stability is still essential, because results may depend on the starting
seeds selected (Mazanec and Strasser, 2000).
In a second step, we used the class labels of the detected segments as explanatory
variables in the CA. The group that did not read UGC at all we called ‘‘noreaders’’. Because
noreaders have no experience with UGCs as information sources, this group is an a priori
de?ned segment that differs from segments containing people who access UGC for diverse
purposes (such as hints on cooking, movie/book reviews or travel-related information). We
did not include noreaders in the cluster analysis that uses motivational reasons for reading
UGC as segmentation base. However, we did include noreaders in the CA as a further
segment. CA is a valuable technique for examining categorical data. Nominal or ordinal
scaled data is present in many ?elds of applied research. CA plots cross-tabulated data and
allows the easy grasp of patterns of numerical frequencies. CA also permits the inclusion of
additional explanatory variables. Theoretically, one can include as many explanatory
variables as desired; this is only a matter of interpretation. Furthermore, subset analyses
assist studying effects of missing and neutral responses (Greenacre and Pardo, 2005).
Generally, scales are always a sensitive issue in surveys (Clark and Schober, 1992) and
researchers must decide on the numbers of rating scale categories and whether to include
or exclude a middle alternative. Research provides mixed results concerning these issues,
and Bond and Fox (2007) state that the number of optimal response categories must be
tested empirically if an existing scale is applied to a new population or a new scale is
developed. Rugg and Cantril (1944) suggest that respondents without a clear opinion on a
topic are more likely to choose the middle alternative. Presser and Schuman (1980) show
similar biases in connection with the subject-involvement of respondents. A recent study by
Dolnicar and Gru¨ n (2010) shows that the level of agreement becomes much stronger if no
neutral position exists; hence, evasion behavior is diminished. Other researchers
demonstrate that combining ?ve ordered categories into three in the data increases the
test reliability for their sample (Zhu et al., 1997; Stone and Wright, 1994).
PAGE 392
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
)
To accomplish the analysis in the present study, we used the CA package in R (Nenadic and
Greenacre, 2007). An advantage of this package is that Greenacre’s standard biplot allows a
direct interpretation of the length of the arrows from the origin. The length of the arrows
re?ects the contribution of each variable. By analyzing the position of the variables in a CA
plot, we can determine the appropriateness of scale-levels, and researchers may decide
whether to combine categories or not. For a detailed description of CA, see Greenacre
(1993, 2007). Hair et al. (2010) also offer insights for the usefulness of CA. The present paper
focuses on plotting both row and column categories; it plots together identi?ed segments
with results of perceived importance regarding website content/applications. The
substantive variables requiring explanation comprise 16 content/applications. The
research explores plots in terms of appropriateness of the scale used and unambiguity of
interpretation, and carries out subset analyses excluding the middle alternative of the scale.
In a third step, we exemplarily compared CA results regarding scale levels with ?ndings
based on Rasch analysis. Rasch analysis is a probabilistic measurement model, which is
commonly used to examine scale-related issues (Rasch, 1960, 1980). To diagnose the
appropriateness of rating scales, a combination of criterions such as thresholds, probability
curves, and category frequencies should be used (Bond and Fox, 2007; Wright and
Masters, 1982). The Rasch parameters that re?ect the structure of the rating scale are the
thresholds (Andrich, 1978, 1998), in the sense that they are the boundaries between
categories, meaning that a ?ve-point Likert scale requires four boundaries (thresholds) to
separate the categories (Linacre, 2001). Problems occur if thresholds did not increase in a
logical, ordered manner and/or if low differences exist between the thresholds. As a rule of
thumb, the increase of thresholds should be between 1.4 and 5 logits (Linacre, 1999). With
the probability curve, one can visually inspect if thresholds are ordered and if the distances
between the thresholds are equal. In addition, category frequencies should be inspected
and, according to Linacre (1999), at least ten responses per category are required. When
the analysis indicates that problems exist concerning some categories, adjacent categories
should be merged in a thoughtful way in order to improve variable clarity (Linacre, 1999) and
the scale re-analyzed. The quality of this new scale should also be tested across different
groups of interest (Bond and Fox, 2007).
Finally, to analyze whether one can detect differences between adventure- and
relaxation-seekers, we included an additional explanatory variable in CA.
Results
Sample characteristics
In total, 440 questionnaires were usable for the purpose of this study. The data, gathered in
2009, consisted of 36.4 percent male and 63.6 percent female respondents. On average,
they were 24.9 years old, and approximately 80 percent were students. In terms of travel
experience, 14.5 percent perceived themselves as not very travel experienced, 75.9
percent said that they were experienced but not an expert, and 9.5 percent assessed
themselves to be travel experts. Concerning UGC usage, 240 persons read UGC entries.
The importance of the internet as a source of travel information is re?ected in the fact that
37.7 percent stated that they always used the internet to access travel information.
Thirty-three percent of the respondents used the internet often, 20.7 percent sometimes, 6.8
percent rarely, and only 1.4 percent did not use the internet at all. Regarding type of traveler,
the sample is divided into 145 relaxation- and 295 adventure-seekers, of which 51.7 percent
and 55.9 percent, respectively, read UGC.
Segments detected
The TRN clustering approach favored a six-segment solution, with a wSSI of 0.56. For a
six-cluster solution, a run of 50 replications results in a UR of 82.04 percent, while this is
slightly lower for a ?ve- and seven-cluster solution (78.01 percent and 81.51 percent,
respectively). However, combined considerations of UR, the size of the clusters, and
appropriateness of the results, suggest opting for six classes.
VOL. 6 NO. 4 2012
j
INTERNATIONAL JOURNAL OF CULTURE, TOURISM AND HOSPITALITY RESEARCH
j
PAGE 393
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
)
The results describe segments brie?y in descending order (by group size based on
prototype tables) and dedicate each concise labels. The description also covers pro?le
information received by cross-tabulating the segments with third variables included in the
questionnaire and by applying ANOVA. Only signi?cant results (i.e. gender, use of
travel-related UGC, travel experience, and travel partner) are presented in the pro?le
descriptions below:
B Enthusiasts (28.3 percent) – The largest segment, this represents people who stated that
they searched for UGC entries because they loved all aspects of e-WOM and enjoyed
reading UGC. They described themselves as being motivated by getting authentic and
reliable information – not offered by providers, but by travel buddies. They also liked trip
planning tools and they searched for contacts with other travelers as well as with locals.
Further reasons for accessing UGC included reducing risks, saving time, and comparing
prices or hotels. Approximately 80 percent were travel experienced but did not consider
themselves experts; 10.3 percent considered themselves expert travelers, and did a lot of
travel. In organizing a trip, 38.3 percent often or always and 60.3 percent sometimes or
rarely searched for travel-related UGC. Of the enthusiasts group, 67.6 percent were
female, and their preferred travel mate was their partner (48.5 percent), followed by
friends (33.8 percent).
B Motivated mavericks (22.1 percent) – This segment stated that they were motivated to
access UGC because they wanted to reduce poor decisions and get reliable
information. However, they said that they did not like contacting locals or other
travelers. Of this group, 83.0 percent traveled a lot but did not consider themselves
travel experts; while 7.5 percent did perceive themselves as expert travelers.
Approximately one-third rarely used travel-related UGC as a form of information, 43.4
percent used UGC sometimes, 15.1 percent used UGC often, and 7.5 percent used
UGC as a form of information for all travel. Although mavericks were loners in terms of
their motivation to read UGC, they primarily traveled with their friends (50.9 percent) or
partner (32.1 percent). Concerning gender, roughly 45 percent were male and 55
percent were female in this study.
B Tips and price optimizers (20.4 percent) – The so called ‘‘optimizers’’ indicated that they
appreciated tools which allowed them to compare prices, hotels, and destinations. They
were also interested in stories and descriptions as well as authentic photos of travel
buddies. The majority of optimizers were female (77.6 percent) in this study, and all
members of this segment used Web 2.0 platforms to search for travel-related information:
approximately one-third used UGC rarely, 36.7 percent used UGC sometimes, and 24.5
percent used UGCoften. Among optimizers, 12.2 percent perceived themselves as travel
experts. More than half traveled a lot but did not consider themselves travel experts; while
14.3 percent traveled only occasionally. Trips were primarily taken with friends (51.0
percent) or their partner (30.6 percent).
B Safety players (15.0 percent) – All but one member of this segment used travel-related
UGC to inform themselves about their journeys: nearly 30.6 percent rarely used UGC,
44.4 percent sometimes used it, and 16.7 percent often used UGC for travel planning.
The motivation to read UGC from travel buddies was reported to be in?uenced by the
wish to reduce the risk of making bad decisions. However, they were not motivated to
access UGC from other travelers for saving time or for networking purposes. The majority
of this segment traveled regularly (approximately 90 percent), half preferred going on
vacation with their partner, and approximately 40 percent with their friends. Males and
females were nearly equally distributed in this study.
B Uncommercial info searchers (7.1 percent) – Approximately 60 percent of the segment
labeled ‘‘uncommercials’’ traveled frequently and had already seen parts of the world.
This segment reported that they read travel-related UGC because they were eager to get
reliable and non-commercial information. They said that they liked hearing from other
travelers and searched for special tips and information that commercial providers did not
offer. In terms of looking for travel information on Web 2.0 sites, none in this group always
used UGC, approximately 11.8 percent often used UGC, 52.9 percent sometimes used it,
PAGE 394
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 29.4 percent rarely used it. This group was dominated by females (82.4 percent).
Approximately one-quarter did not mind traveling alone, while 41.2 percent traveled with
their friends.
B Avoiders (7.1 percent) – This segment said that they read UGC; but interestingly, they
were not very motivated to do so. They did not access travel-related Web 2.0 sites to
reduce risk, get reliable information, ?nd authentic information, have fun, or contact
others. Comparing prices somewhat motivated this segment; however, the strength of
motivation was not strong: 11.8 percent never used UGCas a source of travel information,
while 5.9 percent always searched for travel information on Web 2.0 sites, 23.5 percent
used UGCoften, 17.6 percent sometimes used UGC, and 41.2 percent rarely used it. This
group traveled the least, with 35.3 percent perceiving themselves as not travel
experienced, 29.4 percent traveling regularly but not considering themselves experts,
and 17.5 percent considering themselves experts. This cluster contained the lowest
percentage of people traveling with friends (17.6 percent). Most traveled with their
partner or family (53.0 percent), 17.6 percent traveled alone, and 11.8 percent traveled
with a tour group. In this study 52.9 percent were female.
We detected no signi?cant differences between the segments concerning age, occupation,
or internet experience. Furthermore, a one-factorial analysis of variance demonstrated that
differences existed regarding the importance of website applications; basically with
p-values of ,0.006. The research detected a 10 percent signi?cance level for ‘‘rating of
reviews’’ (p 2value ¼ 0:072) and ‘‘photos’’ (p 2value ¼ 0:090), while ‘‘number of users’’
was found to be not signi?cant.
The six segments identi?ed above constitute the substantive variable used for further
analyses with CA. As already mentioned in the methodology section, we added a seventh
segment comprising the group of people that did not read UGC at all (this segment is
henceforth called ‘‘noreaders’’).
Segments’ contribution to the dimensions
In order to visualize the results and to facilitate the interpretation of a matrix comparing all
seven segments with the importance of 16 website features, we estimated a CA. CA results
show that a two-dimensional solution is appropriate. Two dimensions represent 70.8 percent
of the inertia, while the third only accounts for approximately 10 percent. The scree plot and
eigenvalues also support two dimensions. In detail, eigenvalues of 6.8 and 2.5 ful?ll the
Kaiser criterion of being higher than 1 (Hair et al., 2010; Garson, 2008). The quality measure
provided by the CA procedure shows that safety players are represented best by the two
dimensions (97.0 percent), followed by enthusiasts (88.7 percent). The relative contribution
of safety players to the ?rst dimension is very high, at 86.8 percent, while the relative
contribution is only 10.2 percent for the second dimension. Enthusiasts (68.3 percent) also
contribute highly to the ?rst dimension. In contrast, the second dimension is mainly a
representation of noreaders (87.7 percent), but with an inertia of 18.9 percent, the second
dimension is far less important than the ?rst one (see Table I for contribution details).
Table I Segments’ contribution to the dimensions
Group
Overall
contribution
(%)
Contribution to
Dimension 1
(%)
Contribution to
Dimension 2
(%)
Safety players 97.0 86.8 10.2
Enthusiasts 88.7 68.3 20.4
Noreaders 87.8 0.1 87.7
Avoiders 33.1 23.2 10.0
Tips and price optimizers 26.0 25.8 0.2
Motivated mavericks 20.5 5.3 15.2
Uncommercial info searchers 15.5 0.0 15.5
VOL. 6 NO. 4 2012
j
INTERNATIONAL JOURNAL OF CULTURE, TOURISM AND HOSPITALITY RESEARCH
j
PAGE 395
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 1 visualizes the results described above by incorporating all information given.
Regarding the importance of website features, the length of the arrows illustrate that the
content/applications contributing most to the ?rst dimension are: ‘‘trip planners’’ (3d), ‘‘rating
of reviews’’ (6d), ‘‘special deals’’ (7d), ‘‘comparing prices’’ (8d), ‘‘connecting with locals’’
(10dd and 10a), ‘‘creating maps’’ (12d, 12a, and 12dd), and ‘‘subscription to newsletter’’
(13dd). ‘‘Photos’’ (2aa), ‘‘number of users’’ (4aa), ‘‘number of reviews’’ (5aa), and
particularly, neutral
responses regarding the importance of website content/applicationsde?ne the second dimension (e.g. 6n, 15n, 16n). The mass is the weight given to a point and
is identi?ed by the different sizes of the circles. Generally, the bigger the circle, the higher
the mass, and the higher the population of the cell behind. Thus we can detect that
noreaders is a large segment, while uncommercials is quite a small one.
Before actually interpreting the results in terms of closeness of the importance values to the
segments, the focus now shifts to the ?ve-point Likert scale, which we used to measure
perceived importance of website content/applications.
Inspection of the scale level
Figure 1 shows that the categories strongly agree (aa) and agree (a) are quite close together,
as are strongly disagree (dd) and disagree (d). For instance, for the content ‘‘photos’’ (2),
2dd and 2d are displayed in close proximity, meaning they are closely related. If the
categories were signi?cantly different, they would be located further from each other in the
map. Scale levels are not displayed in an ordered way from dd on the left side to aa on right
side; this is true, for instance, for ‘‘destination description’’ (15), where d is located further left
than dd. The scale levels for ‘‘comparing prices’’ (8) are displayed in the following way:
d-n-dd-a-aa. These results demonstrate that a three-point scale (agree, neutral, and
disagree) would have been suf?cient. The smaller scale would not only visually improve the
graphic, but would also facilitate interpretation.
Figure 1 Importance of website content/applications measured on a ?ve-point-scale
PAGE 396
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 order to see if a three-point Likert scale is more appropriate, the categories aa and a are
combined, as are dd and d. Labels are changed accordingly (a ¼ agree, n ¼ neutral, and
d ¼ disagree). Compared to the ?ve-point Likert scale, results for the inertia for three
categories increase by 6.8 percent to 77.6 percent, and the graphic display of the contributions
of various categories is clearer (Figure 2). Using this format, the ?rst dimension is shown to be
mainly explained by safety players (87.3 percent), enthusiasts (77.5 percent), and optimizers
(42.0 percent), while noreaders (81.8 percent), uncommercials (32.4 percent) and mavericks
(27.8 percent) contribute almost only to the second. Avoiders contribute to both dimensions,
with 34.2 percent and 32.3 percent, respectively. A rough, parabolic left-to-right progression
from disagree via neutral to agree (left to right) is evident, which is the typical pattern of an
uni-dimensional scale (Camiz, 2005). To emphasize this pattern, the scale-levels are connected
by lines in Figure 2 and, as a showcase, the line for the content/application ‘‘connecting with
locals’’ (10) is presented in bold (Figure 2). The relation regarding explanatory variables is quite
intuitive. Enthusiasts who love all website content/applications are displayed on a diagonal line
opposite the segment avoiders, who read UGCalthough they did not perceive UGCimportant.
In contrast, noreaders are located nearest to the category n. Safety players were only motivated
by content/applications that reduced risky decisions; all the rest of the information offered on a
Web 2.0 site was not very interesting for this segment. It is apparent that they had a rather
contrary focus compared to enthusiasts and avoiders. Optimizers rated the importance of
networking and planning tools highly. Uncommercials perceived information provided by users
as important, and they did not want destination or attraction descriptions provided by
destination marketing organizations. Mavericks judged the importance of tools for booking
travel-related products low.
Inspection of the middle alternative
In order to determine whether the middle category n ( ¼ neutral) in?uenced the results, we
excluded this category in the next presentation by undertaking a subset analysis. A
comparison of the maps (a) and (b) in Figure 3 shows the consequences. By excluding n, the
inertia further increases by 5.4 percent to 83.0 percent. The segment noreaders shrinks
towards the center, while the contribution of the segments enthusiasts (87.9 percent),
Figure 2 Applications with combined scale-levels and its parabolic shape
VOL. 6 NO. 4 2012
j
INTERNATIONAL JOURNAL OF CULTURE, TOURISM AND HOSPITALITY RESEARCH
j
PAGE 397
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 3 Comparison with and without middle category: (a) CA including middle category,
(b) CA excluding middle category
PAGE 398
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
)
optimizers (54.6 percent) and safety players (92.7 percent) increases for the ?rst dimension
and decreases for the second to below 0.5 percent. On the second dimension, the
contribution of the segments mavericks (56.9 percent) and avoiders (51.0 percent) rises.
Some content/applications move towards the x-axis; thus, clearly contributing to the ?rst
dimension and no longer to the second (e.g. 5d, 6d, and 10a). Contributions by other
content/applications increase, which can easily be recognized by the length of the arrows.
Omitting the middle category reduces the contributions of the segment noreaders, and
simultaneously increases the contributions of other segments. Interpretation of the results is
facilitated because contributions become much more explicit, especially on the most
important ?rst dimension. In an animation, we may track the consequences of excluding the
middle alternative. A video showing the changes is provided (see http://tourism.wu.ac.at/ugc/
animation.htm), where the procedure to create the animation is suggested by Maier (2010).
CA scale level inspection results and Rasch analysis
The CA offers certain evidence regarding the inappropriateness of scale levels. The
following discussion presents patterns recognized in CA plots and the corresponding
?ndings through Rasch analysis. First, we have seen that the categories of some website
content/applications are not ordered using the ?ve-point Likert scale (e.g. ‘‘destination
description’’). Figure 1 obviates that 15d is displayed further left than 15dd, and the level of
agreement (15a) is positioned at the center of the CA plot, implying low contribution to the
dimensions. Findings through Rasch analysis acknowledge that thresholds do not increase
in a logical, ordered manner required by the Rasch measurement model from dd to aa:
21.56, 20.07, 20.11, and 1.74. The threshold fromlevel n to a (20.11) is less than that from
level d to n (20.07), implying that the scale is disordered. Inspection of the frequencies
indicates that 15aa has fewer than ten responses, which in turn does not ful?ll the
requirements stipulated by Linacre (1999). Furthermore, the shape of the distribution is
problematic, with nearly 55 percent of people either disagreeing or strongly disagreeing. We
mentioned earlier that low distances are evident between two categories in the CA plot; our
example was ‘‘photos’’, where 2dd and 2d are displayed close together and 2n is at the
center of the plot (see Figure 1). These ?ndings are supported by Rasch analysis, showing
low differences between the thresholds of these levels: 20.86, 20.25, 20.226, and 1.33.
Ideally, thresholds should be between 1.4 and 5 logits (Linacre, 1999); otherwise, the
category de?nition is too narrow or too many category options have been provided, which is
the case in our example. As another guideline, the frequencies suggest that categories dd
and d should be merged, because only four people strongly disagreed. Re-analyzing the
scale with only three categories shows that problems still exist with the scales because they
are not properly ordered with regards to the thresholds, and/or the corresponding logits are
below 1.4. Therefore, the exclusion of the middle category needs to be considered. Rasch
analysis supports the simpli?ed picture elaborated through the inspection of CA results.
Thus, evidence exists that different positions of the variables in the CA plot and the
corresponding scale levels gives an indication of the appropriateness of the scale.
In the next step, this more convenient depiction with only two scale levels (a ¼ agree and
d ¼ disagree) becomes the basis for including the type of traveler (relaxation- and
adventure-seeker) as an explanatory variable. This allows for further insights regarding the
perceived importance of website content/applications for two different types of travelers.
Including further explanatory variables
Figure 4 shows differences between the two types of travelers. In particular, the different
travel types of avoiders and safety players are plotted separately. The relaxation-seeking
avoiders appreciate ‘‘comparing prices’’ (7a) and ‘‘newsletters’ ’ (13a) more than
adventure-seeking avoiders. Furthermore, the ?rst group perceives ‘ ‘destination
descriptions’’ (15d) as being unimportant, while adventure-seeking avoiders perceive
‘‘photos’’ (2d), ‘‘reviews’’ (6d), and ‘‘attraction descriptions’’ (16d) to be unimportant.
VOL. 6 NO. 4 2012
j
INTERNATIONAL JOURNAL OF CULTURE, TOURISM AND HOSPITALITY RESEARCH
j
PAGE 399
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
)
Implications
Theoretical implications
A posteriori segmentation shows that heterogeneity pertaining to motivational aspects exists
for reading travel-related UGC. Differences are evident between the segments regarding the
frequency of accessing UGC for searching travel information, travel experiences, partners,
and gender. ANOVA demonstrates a difference between the segments in terms of the
perceived importance of website content/applications.
Enthusiasts, the largest segment, reported using Web 2.0 sites most often as travel
information sources. Thus, they ranked the importance of all website content/application
features as high. In contrast, avoiders hardly used UGC and reported thinking that only the
comparison of prices was a worthwhile reason for accessing it. Hence, the favorite
application for this segment is a price-comparison-tool. Optimizers were motivated by
special deals and authentic reports from other travelers, which is also re?ected in their
preferred content/applications; those which allow them to get in contact with travel buddies
or locals, to ?nd special deals, to plan the trip interactively, or to view photos and videos.
Mavericks were happy with all website content/applications but perceived features for
getting into contact with others as unimportant. Therefore, newsletters, for instance, were not
important to them. For uncommercials, review applications were most important. This
segment also searched in forums for travel information, and read destination and attraction
descriptions. For them, the availability of online booking was also perceived as important.
The motivation for safety players was to avoid risky decisions. Hence, they were interested in
review-applications, while all other features were perceived as unimportant. Pertaining to the
importance of content/applications, further differences are evident regarding type of
traveler, which seem to be intuitive, because adventure-seekers need different information
than people whose motivation for traveling is relaxation. These results con?rm ?ndings from
previous literature indicating that diverse groups of users desire different website designs
(e.g. Perfetti, 2001; Sullivan, 1997). Our ?ndings also show that noreaders have a neutral
position, which agrees with earlier studies that suggest that the middle alternative will be
chosen whenever a person has no clear opinion on a topic (Rugg and Cantril, 1944) or is
Figure 4 Importance of website regarding type of travelers
PAGE 400
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
)
uninvolved (Presser and Schuman, 1980). By opting for the neutral position, respondents
can avoid making a decision. For this reason, consideration should be given to the inclusion
of a ‘‘no opinion’’ or ‘‘not used’’ category when designing a questionnaire. In analyzing the
responses of those who did read UGC, the omission of the middle category would facilitate
interpretation without experiencing information loss; the use of a binary scale (agree and
disagree) is supported by the ?ndings of Dolnicar and Gru¨ n (2007). By applying the CA in a
creative way, this method is shown to be useful for detecting whether a higher-dimensional
scale confers additional insights – or if it is more appropriate to combine categories for the
sake of clarity without losing a lot of information. CA results of scale inspection are veri?ed by
?ndings through Rasch analysis.
The present study also demonstrates that CA is a worthwhile technique that permits
straightforward interpretations using Greenacre’s standard biplot (Greenacre, 1993). In this
plot the length of the arrows, or more generally, the distance of a point fromthe origin, can be
read as the contribution of a certain point ( ¼ variable). Subset analysis involves the
exclusion of categories, and which assists researchers and consultants to present complex
matrices that are tight and well arranged. An animation procedure proposed by Maier (2010)
seizes multimedia opportunities to further improve the demonstration of certain effects.
Managerial implications
From a managerial perspective, our ?ndings show a need to tailor a website according to
target customers’ needs. The site architecture or structure of a website must re?ect the
different requests of targeted segments (Perfetti, 2001). Beside the fact that websites
must cater to customers’ needs, an important decision website providers should consider
is how granulated a segment is determined. One can either use one large segmentation
base, such as the motivation to access UGC, or dig even deeper and take into account
an additional factor, such as in our case involving the level of sensation-seeking
(relaxation- and adventure-seekers). The more precise the segmentation undertaken, the
greater the opportunity to match site architecture and users’ mental models; however,
simultaneously, the amount of people attracted to the site consequently decreases.
Whether this is desired or not depends on whether the aim of the website is to target a
very speci?c niche market or to address a more general group of people. The biggest
segment revealed in this study is labeled enthusiasts, a segment that prefers authentic
information not provided by travel/service providers and applications that enable the user
to get in contact with travel buddies and locals. Other segments appreciate price
comparison tools or authentic photos or videos. Therefore, website providers initially must
de?ne their target group and detect what kind of information and which applications they
require. Websites can then be designed to incorporate information for the targeted user
group. Such a procedure should secure an appropriate design, resulting in satis?ed
users, which in turn leads to positive e-WOM and repeat usage of a website (DeLone and
McLean, 2003).
Limitations and future research
The relationship between patterns displayed in the CA plot and respective results of Rasch
analysis needs further investigation, including the inclusion of additional variables and the
use of various scales to allow generalizable conclusions (Bond and Fox, 2007). Another
limitation of the present study involves the sample size, which becomes quite small after
segmentation. However, the sample nevertheless allows for some preliminary insights into
different types of UGC users. Further, this study used a convenience sample, which
over-represented students; thus, follow-up studies including other cohorts are needed.
Finally, future research should expand the topic from pre-trip planning to various stages of
traveling, and examine changes of perceived importance of website content/applications in
the various phases of a trip. Differences between website content/application use for large
holidays and short breaks might also be incorporated.
VOL. 6 NO. 4 2012
j
INTERNATIONAL JOURNAL OF CULTURE, TOURISM AND HOSPITALITY RESEARCH
j
PAGE 401
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
)
References
Andrich, D. (1978), ‘‘Aratingformulation for ordered response categories’’, Psychometrika, Vol. 43 No. 4,
pp. 561-73.
Andrich, D.A. (1998), ‘‘Thresholds, steps, and rating scale conceptualization’’, Rasch Measurement
Transactions, Vol. 12 No. 3, pp. 648-9.
Bond, T.G. and Fox, C.M. (2007), Applying the Rasch Model, Lawrence ErlbaumAssociates, Mahwah, NJ.
Camiz, S. (2005), ‘‘The Guttman effect: its interpretation and a new redressing method’’, Data Analysis
Bulletin, Vol. 5, pp. 7-34.
Chau, P.Y.K., Cole, M., Massey, A.P., Montoya-Weiss, M. and O’Keefe, R.M. (2002), ‘‘Cultural differences
in the online behavior of consumers’’, Communications of the ACM, Vol. 45 No. 10, pp. 138-43.
Clark, H.H. and Schober, M.F. (1992), ‘‘Asking questions and in?uencing answers’’, in Tanur, J.M. (Ed.),
Questions about Questions: Inquiries into the Cognitive Bases of Surveys, Russell Sage, New York, NY,
pp. 15-48.
De Marsico, M. and Levialdi, S. (2004), ‘‘Evaluating web sites: exploiting users’ expectations’’,
International Journal of Human-Computer Studies, Vol. 60 No. 3, pp. 381-416.
DeLone, W.H. and McLean, E.R. (2003), ‘‘The DeLone and McLean model of information systems
success: a ten-year update’’, Journal of Management Information Systems, Vol. 19 No. 4, pp. 9-30.
Di Mascio, T. and Tarantino, L. (2003), ‘‘Advanced visual interfaces: the focus is on the user’’,
The Knowledge Engineering Review, Vol. 18 No. 2, pp. 175-81.
Dolnicar, S. and Gru¨ n, B. (2007), ‘‘How constrained a response: a comparison of binary, ordinal and
metric answer formats’’, Journal of Retailing and Consumer Services, Vol. 14 No. 2, pp. 108-22.
Dolnicar, S. and Gru¨ n, B. (2010), ‘‘Translating: between survey answer formats’’, paper presented at the
Global Marketing Conference, 9-12 September, Tokyo.
Garson, D.G. (2008), ‘‘Correspondence analysis’’, Statnotes: Topics in Multivariate Analysis, available
at: http://faculty.chass.ncsu.edu/garson/pa765/statnote.htm (accessed May 20, 2010).
Goldsmith, R.E. and Horowitz, D. (2006), ‘‘Measuring motivations for online opinion seeking’’, Journal of
Interactive Advertising, Vol. 6 No. 2, pp. 1-16.
Greenacre, M. and Pardo, R. (2005), ‘‘Multiple correspondence analysis of a subset of response
categories’’, Economics Working Papers, available at: http://econpapers.repec.org/RePEc:upf:upfgen:881
(accessed September 2, 2010).
Greenacre, M.J. (1993), Correspondence Analysis in Practice, Academic Press, London.
Greenacre, M.J. (2007), Correspondence Analysis in Practice, Chapman & Hall/CRC, Boca Raton, FL.
Gretzel, U., Yoo, K.H. and Purifoy, M. (2007), ‘‘Trip advisor online travel review study: the role and
impacts of online travel review for trip planning’’, Laboratory for Intelligent Systems in Tourism, College
Station, TX.
Hair, J.F., Anderson, R.E., Tatham, R.L. and Black, W.C. (2010), Multivariate Data Analysis. A Global
Perspective, Prentice Hall, Upper Saddle River, NJ.
Hennig-Thurau, T. and Walsh, G. (2003), ‘‘Electronic word-of-mouth: motives for and consequences of
reading customer articulations on the internet’’, International Journal of Electronic Commerce, Vol. 8
No. 2, pp. 51-74.
Holtze, T.L. (2000), ‘‘Applying learning style theory to web page design’’, Internet Reference Services
Quarterly, Vol. 5 No. 2, pp. 71-80.
Hoyle, R.H., Stephenson, M.T., Palmgreen, P., Lorch, E.P. andDonohew, R.L. (2002), ‘‘Reliability and validity
of a brief measure of sensation seeking’’, Personality and Individual Differences, Vol. 32, pp. 401-14.
Lenhart, A. and Fox, S. (2006), ‘‘Bloggers – a portrait of the internet’s new storytellers’’, available at:
www.pewinternet.org/pdfs/PIP\%20Bloggers\%20Report\%20July\%2019\%202006.pdf (accessed
August 25, 2008).
Linacre, J.M. (1999), ‘‘Investigating rating scale category utility’’, Journal of Outcome Measurement,
Vol. 3 No. 2, pp. 103-22.
PAGE 402
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
)
Linacre, J.M. (2001), ‘‘Category, step and threshold: de?nitions & disordering’’, Rasch Measurement
Transactions, Vol. 15 No. 1, p. 794.
Litvin, S.W. (2008), ‘‘Sensation seeking and its measurement for tourism research’’, Journal of Travel
Research, Vol. 46, pp. 440-5.
Maier, M. (2010), ‘‘R-code for animation procedure to demonstrate changing effects’’, unpublished
code, Institute for Statistics and Mathematics, WU Vienna, Vienna.
Mandel, T. (2002), ‘‘User/system interface design’’, available at: www.theomandel.com/docs/mandel-
encyclopedia.pdf (accessed November 20, 2008).
Martinetz, T.M. and Schulten, K.J. (1991), ‘‘A neural-gas network learns topologies’’, Arti?cial Neural
Networks, North-Holland, Amsterdam, pp. 397-402.
Mayo, E.J. and Jarvis, L.P. (1981), The Psychology of Leisure Travel – Effective Marketing and Selling of
Travel Services, CBI Publishing Company, Boston, MA.
Mazanec, J.A. (2001), ‘‘Neural market structure analysis: novel topology-sensitive methodology’’,
European Journal of Marketing, Vol. 35 Nos 7/8, pp. 894-916.
Mazanec, J.A. (2008), ‘‘TRN32 for Windows’’, available at: www.wu.ac.at/itf/downloads/software/trn32
(accessed July 14, 2010).
Mazanec, J.A. (2009), ‘‘TRN2009’’, available at: www.wu.ac.at/itf/downloads/software/trn32 (accessed
July 14, 2010).
Mazanec, J.A. and Strasser, H. (2000), A Nonparametric Approach to Perceptions-based Market
Segmentation: Foundations, Springer, Vienna.
Nardi, B.A., Schiano, D.J. and Gumbrecht, M. (2004), ‘‘Blogging as social activity, or, would you let 900
million people read your diary?’’, Proceedings of the ACM Conference on Computer Supported
Cooperative Work, Chicago, IL, ACM, New York, NY, pp. 222-31.
Nenadic, O. and Greenacre, M. (2007), ‘‘Correspondence analysis in R, with two- and three-dimensional
graphics: the CA package’’, Journal of Statistical Software, Vol. 20 No. 3, pp. 1-13.
Norman, D.A. (2002), The Design of Everyday Things, Basic Books, New York, NY.
Perfetti, C. (2001), ‘‘Personas: matching a design to the users’ goals’’, available at: www.uie.com/
articles/personas (accessed November 18, 2008).
Presser, S. and Schuman, H. (1980), ‘‘The measurement of the middle position in attitude surveys’’,
Public Opinion Quarterly, Vol. 44 No. 1, pp. 70-85.
Raju, P.S. (1980), ‘‘Optimum stimulation level: its relationship to personality, demographics, and
exploratory behavior’’, Journal of Consumer Research, Vol. 1, pp. 272-82.
Rasch, G. (1960), Probabilistic Models for Some Intelligence and Attainment Tests, University of
Chicago Press, Chicago, IL.
Rasch, G. (1980), Probabilistic Models for Some Intelligence and Attainment Tests, University of
Chicago Press, Chicago, IL.
Ricci, F. (2008), ‘‘Information search and recommendation tools’’, paper presented at the JITT Workshop
Series: Tourism, Information Search and the Internet, Vienna.
Rugg, D. and Cantril, H. (1944), ‘‘The wording of questions’’, in Cantril, H. (Ed.), Gauging Public Opinion,
Princeton University Press, Princeton, NJ, pp. 23-50.
Schmidt, J. (2007), ‘‘Blogging practices. An analytical framework’’, Journal of Computer-Mediated
Communication, Vol. 12, pp. 1409-27.
Stangl, B. and Dickinger, A. (2010), ‘‘How communication modes determine website satisfaction’’,
Proceedings of ENTER – Information Communication Technologies in Tourism, Lugano, pp. 273-84.
Stone, M.H. and Wright, B.D. (1994), ‘‘Maximizing rating scale information’’, Rasch Measurement
Transactions, Vol. 8 No. 3, p. 386.
Sullivan, T. (1997), ‘‘All things web: the value of usability’’, available at: www.pantos.org/atw/35679.html
(accessed November 8, 2008).
Wright, B.D. and Masters, G. (1982), Rating Scale Analysis, MESA Press, Chicago, IL.
VOL. 6 NO. 4 2012
j
INTERNATIONAL JOURNAL OF CULTURE, TOURISM AND HOSPITALITY RESEARCH
j
PAGE 403
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
)
Xiang, Z. and Gretzel, U. (2010), ‘‘Role of social media in online travel information search’’, Tourism
Management, Vol. 31 No. 2, pp. 179-88.
Zhu, W., Updyke, W.F. and Lewandowski, C. (1997), ‘‘Post-hoc Rasch analysis of optimal categorization
of an ordered response scale’’, Journal of Outcome Measurement, Vol. 1 No. 4, pp. 286-304.
Zuckermann, M. (1979), Sensation Seeking: Beyond the Optimal Level of Arousal, Lawrence Erlbaum
Associates, Hinsdale, NJ.
Zuckermann, M. (1994), Behavioral Expressions and Biosocial Bases of Sensation Seeking, Cambridge
University Press, Cambridge, MA.
Zuckermann, M. (1996), ‘‘Itemrevisions in the Sensation Seeking Scale formV (SSS-V)’’, Personality and
Individual Differences, Vol. 20 No. 4, p. 515.
Appendix: Questionnaire items
About the authors
Margit Kastner received her doctorate from the Vienna University of Economics and
Business (WU Vienna). She has been employed by the Institute for Tourism and Leisure
Studies, WU Vienna, since February 2000, and at present is an eDeveloper (eLearning
Consultant) in the discipline of marketing. She lectures in marketing, international tourism,
and academic writing. Her research interests are centered on issues of e-learning, learning
and consumer behavior, market research in tourism, and quantitative methodology in
marketing and educational research.
Brigitte Stangl specialized in tourism and leisure studies as well as small business and
entrepreneurship, completing her PhD in 2010 with a dissertation entitled ‘‘User-based
website design in tourism with a special focus on web 2.0’’. Between 2006 and 2010 she
undertook duties as a Research and Teaching Assistant at the Institute for Tourism and
Leisure Studies, WU Vienna, and since 1 November 2010 she has been employed as Project
Leader at the Institute for Tourism and Leisure Research, HTW Chur. Her main research
interests lie in the areas of decision support systems, web design, Web 2.0, and virtual
realities. Brigitte Stangl is the corresponding author and can be contacted at:
[email protected]
Table AI Questionnaire items
Construct I access travel-related UGC because . . . References
Reduce risk . . . the chances of making a bad decision are reduced
. . . it helps me avoid a risky decision
. . . I can hear from people who have already taken a trip
. . . I want to make sure a trip is worth taking
Goldsmith and Horowitz (2006)
Time saving . . . the chances of making a bed decision are reduced
. . . it helps me avoid a risky decision
. . . I can hear from people who have already taken a tip
. . . I want to make sure a trip is worth taking
Hennig-Thurau et al. (2003)
Reliable
information
. . . I get information from someone who is not trying to sell me something
. . . I get special tips on what to see and do in a destination
. . . I want something more than travel service providers are offering
. . . I have not found suf?cient information in other sources of information
Goldsmith and Horowitz (2006),
Hennig-Thurau et al. (2003), Gretzel
et al. (2007)
Authentic
information
. . . I can watch authentic videos of the destination
. . . I can look at authentic photos of the destination
. . . I can read authentic reviews, journals, reports of the destination
Gretzel et al. (2007)
Trip preparation
tools
. . . I can use trip planning, which makes it easier to prepare my trip
. . . I can compare prices, destinations, hotels
Gretzel et al. (2007)
Fun and social
contact
. . . trip planning is fun
. . . I can connect with locals that way
. . . I can connect with other travelers
. . . I enjoy reading UGC
Gretzel et al. (2007)
PAGE 404
j
INTERNATIONAL JOURNAL OF CULTURE, TOURISM AND HOSPITALITY RESEARCH
j
VOL. 6 NO. 4 2012
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. Rob Law, Dimitrios Buhalis, Cihan Cobanoglu. 2014. Progress on information and communication technologies in hospitality
and tourism. International Journal of Contemporary Hospitality Management 26:5, 727-750. [Abstract] [Full Text] [PDF]
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_968655195.pdf