Escolar Documentos
Profissional Documentos
Cultura Documentos
research-article2014
Research Article
Abstract
Most tourists love to share their holiday experiences with family, friends, and, increasingly, strangers, using the Internet. Such
word-of-mouth represents a highly influential information source for potential tourists and is therefore of great interest to
tourism marketing managers. This study aims to understand patterns among tourists when sharing holiday experiences, both
in terms of the communication channel they use and the kind of content they share. The findings contribute to a theoretical
understanding of word-of-mouth behavior by empirically showing that word-of-mouth is not a homogeneous activity. Rather,
results show that distinct segments of word-of-mouth behavior exist. Segments differ with regard to content shared (visual/
verbal) and channel used (offline/online). Two out of the five segments use only offline channels to share their experiences,
and the extent of visual content shared varies across segments. The article illustrates how these findings could be translated
into proactive marketing action aimed at instigating word-of-mouth behavior.
Keywords
market segmentation, electronic word-of-mouth, traditional word-of-mouth, bootstrap, cluster analysis
Introduction
Literature Review
The influence of word-of-mouth has been studied extensively in consumer decision making. This body of work is
characterized by a number of assumptions: (1) that word-of
mouth is a homogenous activity; (2) that electronic word-ofmouth is more influential than traditional word-of-mouth;
and (3) that word-of-mouth is predominantly verbal, not
visual. This article challenges these three assumptions. The
following research questions are investigated:
Research question 1: Is tourist word-of-mouth a homogeneous activity, or do different tourists use different
approaches when sharing their vacation experiences?
Research question 2: Do tourists use both electronic and
traditional word-of-mouth to communicate their experiences or does one form dominate?
Research question 3: Do tourists use both verbal and
visual word-of-mouth to communicate their experiences,
or does one form dominate?
To answer these three research questions, an analysis of heterogeneity of word-of-mouth patterns was conducted using
information provided by Australian tourists about how they
shared their holiday experiences. Findings from this study
contribute to theory by challenging current assumptions
about word-of-mouth. This has implications not only for
future research, but also for tourism marketing managers,
who may benefit from considering a larger set of marketing
activities to stimulate word-of-mouth than relying primarily
on social media.
Corresponding Author:
Amata Ring, Research Fellow, UQ Business School, The University of
Queensland, St Lucia, Brisbane, Queensland 4072, Australia.
Email: a.ring@uq.edu.au
482
related to the usage or characteristics of particular goods and services, or their sellers (Litvin, Goldsmith, and Pan 2008, p. 461).
Electronic resources such as email, the Internet, mobile phones,
instant messaging, and blogs have made sharing information and
opinions easier than ever (Allsop, Bassett, and Hoskins 2007).
Electronic word-of-mouth advances traditional word-of-mouth
in several ways: (1) the line of communication is expanded to
one-to-many, one-to-one, or many-to-many (Litvin, Goldsmith,
and Pan 2008). This (2) escalates the audience of both positive
and negative electronic word-of-mouth because it becomes
accessible globally (Chan and Guillet 2011; Ip, Lee, and Law
2012) and (3) allows sharing between strangers while traditionally word-of-mouth refers to opinions from friends and family.
(4) Electronic word-of-mouth can be spread over a variety of
channels, some of which allow anonymity (Benckendorff,
Sheldon, and Fesenmaier 2014). Furthermore, communicating
online (5) has overcome the perishability of traditional wordof-mouth because it can be collected and preserved (Goldsmith
2006).
These advantages of electronic word-of-mouth have led
to an increase in the importance of electronic word-of-mouth
compared to traditional word-of-mouth (Goldsmith 2006;
Litvin, Goldsmith, and Pan 2008; Sun etal. 2006).
Additionally, Gretzel and Yoo (2008) have shown the influence of electronic word-of-mouth on purchase decisions.
However, some advantages of electronic word-of-mouth are
also perceived as disadvantages by others. While credibility
is regularly attributed to electronic word-of-mouth (Bickart
and Schindler 2001; Bronner and de Hoog 2010), others
argue that online word-of-mouth may be perceived as
untrustworthy or inaccurate because it is created by strangers
whose credibility may not be easily established (Tham, Croy,
and Mair 2013). In an investigation of the different dimensions of trustworthiness for online channels, Dickinger
(2011) shows that tourists doubt the ability of users who generate electronic word-of-mouth to provide high-quality
information. The differences between traditional and electronic word-of-mouth, especially with regards to credibility,
led Tham, Croy, and Mair (2013) to conclude that traditional
word-of-mouth is still important. Tan and Tangs (2013)
findings also show that one cannot generally conclude that
either traditional or electronic word-of-mouth is more
important.
Consequently, while the emergence of and increase in
electronic word-of-mouth has to be acknowledged, investigating traditional and electronic word-of-mouth simultaneously seems to represent a more complete picture of total
word-of-mouth behavior.
483
Ring et al.
where the verbal aspect also dominates (Goldsmith 2006;
Hennig-Thurau etal. 2004; Litvin, Goldsmith, and Pan 2008;
Sun etal. 2006). Consequently, a substantial amount of tourism-related research has been conducted on text-driven blogs
(e.g., Pan, MacLaurin, and Crotts 2007; Zehrer, Crotts, and
Magnini 2011) and online reviews (e.g., Gretzel and Yoo
2008; Sparks and Browning 2011; Vermeulen and Seegers
2009). Despite the dominance of visual experiences in tourism (Feighey 2003), in the context of word-of-mouth,
researchers have turned their attention to the visual representation of a travel experience only recently (e.g., Lo etal.
2011; Stepchenkova and Zhan 2013).
The only study that explicitly focuses on the importance
of the combination of visual and verbal content in tourismrelated word-of-mouth is by Lee and Tussyadiah (2011).
Based on the dual-coding theory (Paivio 1990), they argue
that both visual (e.g., photographs) and verbal means (e.g.,
written and spoken language) constitute a tourists mental
representation of a destination. Lee and Tussyadiah found
that in the context of an online travel community, the combination of text and photo is perceived as the most influential
source of information, and is also the combination of content
most frequently contributed to the online travel community.
However, Bronner and de Hoog (2011) show that not everyone contributes both types of information in the context of
posting reviews.
In line with Lee and Tussyadiah (2011), this study focuses
on the use of both visual and verbal content in wordof-mouth; an area that has not yet been extensively studied.
Methodology
Data
Survey data were collected from 1003 adult Australian residents using an Australian research-only online panel company that recruits respondents both online and through
traditional avenues such as mail and intercepts. Respondents
were paid a small amount of money as compensation for
their efforts. Although well-maintained online panels have
been shown to lead to equally representative samples
(Dolnicar, Laesser, and Matus 2009), it is possible that online
data collection in this particular study will lead to an overestimation of respondents who use electronic media to communicate about their holiday experiences. This is not
problematic, however, because the primary aim of the study
is to explore patterns of word-of-mouth behavior, not to correctly predict the segment size of people who display those
patterns.
Respondents were asked how often they go on holiday
and which travel information sources they normally use for
making their decisions. Respondents were also asked to
think of their last domestic holiday and indicate where they
spent it, how far it was from home, how they got there, how
long the holiday was, what the accommodation was, how
937 (100%)
618 (66%)
67 (7%)
70 (7%)
381 (41%)
380 (40%)
138 (15%)
25 (3%)
484
Sample Description
Data Analysis
485
Ring et al.
clusters solutions. Then, the predictions were repeated based
on the cluster centroids of bootstrap sample B. The outcome
of this step was a set of predictions for all 2-to-10 clusters
solutions. One pair of predictions for every observation in
the original data set was obtained for all numbers of clusters
solutions (2-to-10). The first prediction was based on the
solution from bootstrap sample A, while the second was
based on the solution from bootstrap sample B. These pairs
of predictions were the basis for the stability analysis.
Calculation of Adjusted Rand Indices. The third step formed the
heart of the stability analysis. The Rand index was used to
judge on the stability of the 2-to-10 clusters solutions (Rand
1971). The Rand index expresses agreements as a proportion
of agreements and disagreements between two solutions. As
an example, consider the predictions from the pair of bootstrap samples A and B for the two-cluster solution. For every
pair of observations, it was determined (1) if the two observations were assigned to the same cluster based on the predictions from bootstrap sample A, or not; and (2) if they were
assigned to the same cluster based on the predictions from
bootstrap sample B, or not. Agreements are all pairs of observations either assigned to the same cluster in both (1) and (2),
or not assigned to the same cluster in both (1) and (2). Disagreements are all pairs of observations that were either put
into the same cluster in (1), but into different clusters in (2),
or put into the same cluster in (2), but into different clusters
in (1). An index value closer to one indicates a higher agreement between the two predictions based on A and B. Perfect
agreement is represented by a Rand index of 1. Finally, the
Rand index was adjusted for chance (adjusted Rand index by
Hubert and Arabie 1985; for computation see also Steinley
2004). For the pair of bootstrap samples A and B, this computation of the adjusted Rand index was repeated for all
numbers of cluster solutions (2-to-10), resulting in one
adjusted Rand index for each of the 2-to-10 clusters
solutions.
For the stability analyses, 50 such pairs of bootstrap samples were drawn. Consequently, for each of the 2-to-10 clusters solution, 50 adjusted Rand indices were obtained.
Selection of Optimal Number of Clusters.In the final step,
these adjusted Rand indices were used to determine the optimal number of clusters. For each of the 2-to-10 clusters solutions, the distribution of the adjusted Rand indices was
compared. The closer the average of the 50 adjusted Rand
indices for a given number of clusters to one, the more stable
the solution.
Figure 1 summarizes the results of the stability analysis. It
shows the boxplots of the adjusted Rand indices (y axis) calculated from the 50 pairs of bootstrap samples for 2-to-10
clusters (x axis). The adjusted Rand index lies between 0 (no
agreement between a pair of cluster solutions) and 1 (perfect
agreement between a pair of cluster solutions). For every
prespecified number of clusters (2-to-10), the shown
486
Note: Bars are in color only if the difference between the clusters mean and the overall mean for this variable is either at least half of the overall mean or
at least a tenth of the total maximum for this variable. Because the total maximum is one for all binary variables in this analysis, the latter absolute cut-off
value equals .1 for all variables. These cut-off values could have been defined in a more or less restrictive way. Note that the percentages given are based
on the total sample size of 940. The fifth segment, offline verbalists, are 21% (200 respondents).
Results
Research Question 1: Is Tourist Word-of-Mouth a
Homogeneous Activity?
Figure 2 profiles the segments: colored bars indicate the percentage of people in each segment who engage in a specific
activity, and the red horizontal lines represent the sample
average. To easily identify variables that contribute the most
to the distinct profile of a specific cluster, variables that do
not show a substantial deviation from the total sample mean
are shown in white (Dolnicar and Leisch 2014). Writing
emails and emailing photos best distinguishes between the
four segments. In segments 1 and 3, all respondents emailed
photos, whereas in segments 2 and 4 nobody did. In segments 3 and 4, all respondents sent (verbal) emails, whereas
in clusters 1 and 2 nobody did. All segments liked to show
photos (offline) of their holidays, but segment 4 did to a considerably lesser extent.
In total, five word-of-mouth segments are identified. Four
emerged from the data-driven segmentation analysis and the
fifth was extracted a priori. Note that all segments share their
experiences in conversations. For the a priori selected
487
Ring et al.
Table 2. Multinomial Logistic Regression Results.
Article I.
Online viualists vs. offline verbalists
Sociodemographics
Activities
Sociodemographics
Activities
Travel characteristics
Sociodemographics
Activities
Sociodemographics
Activities
Travel characteristics
B (SE)
Odds Ratio
Intercept
Retireda
Bushwalks
Whale/dolphin watching
Charter boat/cruise
Visiting a spa/getting a massage
General sightseeing
Having picnics/barbecues
Eating out in reasonably priced places
Booked accommodation on the Internet
Traveled to destination by four-wheel drive
Traveled to destination by bus
Traveled alone
Size of travel party
Was a special vacation (once every few years)b
3.03 (0.96)**
1.45 (0.49)**
0.88 (0.32)**
1.12 (0.44)*
0.89 (0.42)*
1.10 (0.47)*
0.98 (0.41)*
0.89 (0.30)**
1.01 (0.38)**
0.66 (0.32)*
1.25 (0.53)*
1.41 (0.71)*
1.14 (0.53)*
0.06 (0.03)*
0.82 (0.39)*
4.26
2.41
0.33
2.44
3.00
2.66
2.43
0.36
1.93
3.49
4.09
0.32
1.06
0.44
Intercept
Retireda
Studenta
Bushwalks
Charter boat/cruise
Visiting wildlife parks/zoos/aquariums
Relax/doing nothing
Number of international trips per year
Traveled alone
A regular break (multiple times a year)b
0.34 (0.78)
1.19 (0.42)**
1.62 (0.57)**
0.65 (0.29)*
0.79 (0.38)*
0.65 (0.32)*
0.66 (0.31)*
0.47 (0.18)*
0.92 (0.42)*
0.75 (0.31)*
3.29
5.05
1.91
2.20
1.92
0.52
0.63
0.40
0.47
Intercept
Retireda
Bushwalks
Visiting wildlife parks/zoos/aquariums
Having picnics/barbecues
Eating out in upmarket restaurants
Booked accommodation on the Internet
Use information from tourist info centerc
Traveled to destination by plane
Traveled to destination by bus
Traveled around at destination by plane
3.42 (0.83)***
0.83 (0.42)*
0.88 (0.29) **
0.86 (0.33)**
0.79 (0.27)**
0.73 (0.28)**
0.88 (0.27)**
0.52 (0.25)*
1.03 (0.33)**
1.66 (0.62)**
1.00 (0.50)*
2.30
2.40
2.35
2.20
2.07
2.42
1.67
2.80
5.24
2.73
Intercept
Female
Bushwalks
Visit botanic or other public gardens
Going for scenic walks/drives
Visiting friends and relatives
Eating out in upmarket restaurants
Relax/doing nothing
Someone else in travel party booked accommodation
Traveled to destination by plane
Traveled to destination by bus
Traveled around at destination by plane
Times went out for lunch
Number of domestic trips per year
3.31 (0.96)***
0.87 (0.30)**
0.78 (0.33)*
1.08 (0.35)**
0.75 (0.34)*
0.64 (0.32)*
0.84 (0.32)**
0.97 (0.36)**
1.22 (0.52)*
1.20 (0.36)***
2.00 (0.66)**
1.39 (0.54)**
1.09 (0.48)*
0.08 (0.03)*
2.38
2.18
0.34
2.12
1.90
2.32
0.38
0.30
3.33
7.36
4.03
0.34
1.08
*p < .05, **p < .01, ***p < .001. No significant results were found for education (reference category: university education), age, how often newspapers are read
(sociodemographics), going camping, playing tennis, visiting museums or art galleries, going on guided tours (activities), book accommodation: on the phone, at arrival, via travel
agent, use hotel brochures as information source,c use exhibitions/fairs as information sourcec (information and booking), get around at destination by: walking, bus, ferry,
taxi, rented car (transport), vacation taken on weekend or during the week, type of accommodation stayed at (reference category: private property where no payment was
required), traveled with children, length of the trip, times went out for coffee (travel characteristics).
a
Full-time employed was used as the reference category.
b
A typical annual vacation (once a year) was used as the reference category.
c
Asked in general (In general, which information sources do you use to help you with your vacation destination choice?). All other trip characteristics refer to the last
Australian (domestic) vacation.
488
Online
Visualists (14%)
Offline
Visualists (24%)
Interactive
Sharers (26%)
Online
Verbalists (14%)
Offline
Verbalists (21%)
80
91
91
54
14
70
85
86
44
13
81
97
96
62
15
71
91
89
60
12
65
80
76
45
9
Note: Results are significant at the 0.05 level, after correcting for multiple testing (Holm 1979).
489
Ring et al.
increase by eight percent. Online verbalists shared their
experiences to an average extent, with the exception of sharing with colleagues, where they show a higher than average
level. They also were more active on vacations than the
offline verbalists, but to a lesser extent than were the other
three segments (see Table 3).
Segment 5 (offline verbalists) shared holiday experiences only in conversations. Because this segment was chosen as the reference category, the negative intercepts for all
segments (see Table 2) indicate that an observation where all
continuous variables are 0 and the values for all categorical
variables are their reference categories is most likely to
belong to Segment 5. Additional information from Table 3
shows that offline verbalists shared their experiences with
the smallest number of people and were also the least active
during their vacations.
In summary, the results for Research Question 1 (Is tourist
word-of-mouth a homogeneous activity, or do different tourists use different approaches when sharing their vacation
experiences?) suggests that word-of-mouth is not a homogeneous activity; rather, distinct segments with regards to different types of word-of-mouth behavior exists.
Discussion
Results indicate that there is substantial heterogeneity in how
word-of-mouth is expressed. Traditional and electronic
word-of-mouth stand side by side, and both visual and verbal
word-of-mouth play a role in sharing holiday experiences.
Different people display different tendencies of using combinations of word-of-mouth modes and channels.
These findings contribute to the theoretical understanding
of word-of-mouth in several ways. They empirically support
the notion that, while electronic media is increasingly important, traditional word-of-mouth continues to play a major
role in sharing holiday experiences (Tham, Croy, and Mair
2013). Additionally, the findings provide empirical evidence
for the importance of a relatively new stream of research
studying the role of visual stimuli as part of word-of-mouth
behavior in tourism (Lee and Tussyadiah 2011; Lo etal.
2011; Stepchenkova and Zhan 2013) and point to the potential of simultaneous use of verbal and visual content (Lee and
Tussyadiah 2011).
This study is the first to demonstrate the extent of heterogeneity among tourists with respect to different ways of sharing holiday experiences. This finding has practical
implications, because it implies that tourism marketers may
want to not only focus on social media to stimulate and
orchestrate word-of-mouth activity but may want to think
about other strategies to achieve the same aim for different
tourist groups. Several authors (e.g., Schmallegger and
Carson 2008; Litvin, Goldsmith, and Pan 2008) have argued
that providing incentives for tourists to write blogs based on
their experience can positively influence how other tourists
(such as their friends and family) perceive a destination
image. Similar incentives could be developed for tourists
who prefer other ways of sharing their holiday experiences
or who communicate predominantly visual or textual information. For example, online visualists (an active segment
often found on cruises, sightseeing, picnicking, or fourwheel driving) could be offered free photo shoots on cruises
or in typical four-wheel driving hot spots. Sending free digital copies of these photographs to members of this segment
is likely to lead to those photos being actively shared online.
It could be argued, of course, that people will not share photos taken by others. But traditionally, this was the way it was
done: for many decades it was the postcarda photo taken
by a professionaland a few lines of text that were used by
tourists to share their holiday experiences. Today, it is common practice for tourists to share photos taken by tour operators such as Contiki and TopDeck Tours on social media. The
postcard might be a dying tradition, but similar mechanisms
could well be embraced by todays tourists, and may have
substantial benefits for the tourism destination or provider. A
professionally taken picture makes the destination appear
more attractive, allowing the tourist destination or provider
to portray itself in line with its positioning, while at the same
time improving the boasting potential for the tourist. Also,
it may offer a way of getting around not being permitted to
take pictures at sites of cultural significance such as museums and sacred locations.
Another example of how these findings could be translated
into marketing ideas stimulating word-of-mouth would be for
tour operators or destinations who build their positioning on
490
Funding
The author(s) disclosed receipt of the following financial support
for the research, authorship, and/or publication of this article: This
research was funded by the Australian Research Council under the
Discovery Grant Scheme (project number DP110101347).
Note
1. The nonsignificant variables in these five models were as follows: income, frequency of watching TV (sociodemographics), visit the beach, visit farms, swimming, snow activities,
horse riding, cycling, hiking/climbing, exercise, golf, fishing,
diving, surfing, four-wheel driving, adventure activities (e.g.,
bungee jumping), other water activities (e.g., sailing), attend
concert, theatre or other performing arts, visit history/heritage
491
Ring et al.
buildings or sites, experience aboriginal arts/crafts, attend festivals/fairs, visit amusement/theme parks, going to markets,
visit industrial tourism heritage, visit attractions for kids, visit
pubs, shopping, watch movies, visit casinos, attend an organized sporting event (activities), use as information source (in
general): destination information brochures, brochures from
tour operators, information from travel agent, ads in newspapers, travel guides/books, information from friends and relatives, radio, TV, Internet, slide nights (information sources and
booking), get to destination by rented car, train, get around at
destination by other water transport (e.g., private boat), rented
car. Get to destination by car and get around at destination by
car were not included in the multinomial logistic regression
model because of their strong correlation with other transport
variables (transport), traveled only with partner, traveled with
friends, packaged vs. independent trip, times eating out for
breakfast, times eating out for dinner (travel characteristics).
References
Allsop, Dee T., Bryce R. Bassett, and James A. Hoskins. (2007).
Word-of-Mouth Research: Principles and Applications.
Journal of Advertising Research, 47 (4): 398-411.
Andereck, Kathleen L., and Linda L. Caldwell. (1994). The
Influence of Tourists Characteristics on Ratings of Information
Sources for an Attraction. Journal of Travel & Tourism
Marketing, 2 (2/3): 171-90.
Anderson, Eugene W. (1998). Customer Satisfaction and Word of
Mouth. Journal of Service Research, 1 (1): 5-17.
Arndt, Johan. (1967). Role of Product-Related Conversations
in the Diffusion of a New Product. Journal of Marketing
Research, 4 (3): 291-95.
Australian Bureau of Statistics. (2014). Census for a Brighter
Future. http://www.abs.gov.au/census (accessed August 1,
2014).
Benckendorff, Pierre J., Pauline J. Sheldon, and Daniel R.
Fesenmaier. (2014). Social Media and Tourism. In Tourism
Information Technology, 2nd edition. Oxford: CABI, pp. 12351.
Bickart, Barbara, and Robert M. Schindler. (2001). Internet
Forums as Influential Sources of Consumer Information.
Journal of Interactive Marketing, 1 (3): 31-40.
Bieger, Thomas, and Christian Laesser. (2004). Information
Sources for Travel Decisions: Toward a Source Process
Model. Journal of Travel Research, 4 (4): 357-71.
Bronner, Fred, and Robert de Hoog. (2010). Consumer-Generated
versus Marketer-Generated Websites in Consumer Decision
Making? International Journal of Market Research, 52 (2):
231-48.
Bronner, Fred, and Robert de Hoog. (2011). Vacationers and
eWOM: Who Posts, and Why, Where, and What? Journal of
Travel Research, 50 (1): 15-26.
Chan, Nga Ling, and Basak Denizci Guillet. (2011). Investigation
of Social Media Marketing: How Does the Hotel Industry in
Hong Kong Perform in Marketing on Social Media Websites?
Journal of Travel & Tourism Marketing, 28 (4): 345-68.
Day, Jonathon, Liping Cai, and Laurie Murphy. (2012). Impact
of Tourism Marketing on Destination Image: Industry
Perspectives. Tourism Analysis, 17 (3): 273-84.
Dibb, Sally, and Lyndon Simkin. (2008). Market Segmentation
Success: Making It Happen! New York: Routledge.
492
Author Biographies
Amata Ring is a Research Fellow at the University of Queensland.
She has undertaken her doctoral studies at the Vienna University of
Economics and Business in Austria. Her research interests are tourism and international marketing, specifically market segmentation
and heterogeneity within consumer perceptions.
Aaron Tkaczynski is a lecturer in tourism and events at the
University of Queensland. Dr Tkaczynskis research interests center
on destination market segmentation, seasonality, and events tourism.
Sara Dolnicar holds degrees in Psychology and Business
Administration. She is a Research Professor in Tourism at the
University of Queensland. Saras key research areas are market segmentation methodology measurement in the social sciences. She has
applied her work mainly to tourism, but also a range of social marketing topics (environmental volunteering, foster care and public acceptance of water alternatives).