Tourism Management xxx (2012) 1e12
Contents lists available at SciVerse ScienceDirect
Tourism Management
journal homepage: www.elsevier.com/locate/tourman
Visual destination images of Peru: Comparative content analysis of DMO and
user-generated photographyq
Svetlana Stepchenkova*, Fangzi Zhan
Department of Tourism, Recreation and Sport Management, University of Florida, P.O. Box 118208, Gainesville, FL 32611-8208, USA
h i g h l i g h t s
< DMO and Flickr photos represented projected and perceived images of Peru.
< 20 destination attributes were identified.
< Representations of various regions of Peru were compared using geo-maps.
< Image maps were constructed using co-occurrences analysis of attributes.
< Hermeneutic circle of destination representation was partially supported.
a r t i c l e i n f o
a b s t r a c t
Article history:
Received 4 June 2012
Accepted 22 August 2012
With the arrival of new media and communication technologies in recent years, user-generated content
(UGC) on the internet has increasingly been considered a credible form of word-of-mouth. Social media
websites, such as Facebook, Flickr, and Panoramio, allow tourists to share their travel experiences with
others by uploading travel photos online, an activity that has gained popularity among internet users.
Unlike images created and projected by destination marketing organizations (DMOs), pictorial UGC
reflects users’ perceptions of a destination. This study compared images of Peru collected from a DMO’s
site and from Flickr, a photo-sharing website and identified statistical differences in several dimensions
of these images. The study visualized these differences by constructing maps representing “aggregated”
projected and perceived images of Peru, as well as maps of geographical distribution of the images.
Ó 2012 Elsevier Ltd. All rights reserved.
Keywords:
Content analysis
Destination image
Flickr
Peru
Photography
User-generated content
Visual research
1. Introduction
The importance of an attractive and unique destination image is
indisputable. Numerous studies document the efforts of destination marketing organizations (DMOs) in researching, building,
promoting, evaluating, and maintaining destination image. For
a long time, destination imagery was controlled almost exclusively
by a destination’s marketing organization (DMO). Materials
produced (or induced, in Gartner’s (1993) terminology) by DMOs
include brochures, guidebooks, postcards, video commercials, and
most recently, destination websites and other online materials.
Tourists, on the other hand, create their own materials about how
q An earlier draft of this study was presented at the 2011 Travel and Tourism
Research Association Annual Conference, London, Ontario, Canada.
* Corresponding author. Tel.: þ1 352 294 1652.
E-mail addresses: svetlana.step@ufl.edu (S. Stepchenkova), [email protected]
(F. Zhan).
they perceive destinations they visit by means of personal blogs,
reviews, photography, and videos. With the developments of Web
2.0 applications, tourists have been presented with previously
unimaginable opportunities to make their travel accounts truly
public through media-sharing websites and social networks.
Because non-promotional communications tend to affect destination images more than messages by DMOs and travel intermediaries (Connell, 2005), a situation in which DMOs lack direct control
over images that are “out there” challenges destination promotion.
DMOs need to know what images dominate the internet and
whether these images are consistent with the information projected by the destination itself, so that they can reinforce positive
images or counter unfavorable images, if necessary.
Destination image research has been conducted since the early
1970s (Hunt, 1975; Mayo, 1973) and comprises hundreds of studies
to date. However, meta-reviews of destination image research
(Gallarza, Saura, & Garcia, 2002; Pike, 2002; Stepchenkova & Mills,
2010; Tasci, Gartner, & Cavusgil, 2007) have shown preferences of
tourism researchers for assessing destination images using
0261-5177/$ e see front matter Ó 2012 Elsevier Ltd. All rights reserved.
http://dx.doi.org/10.1016/j.tourman.2012.08.006
Please cite this article in press as: Stepchenkova, S., & Zhan, F., Visual destination images of Peru: Comparative content analysis of DMO and usergenerated photography, Tourism Management (2012), http://dx.doi.org/10.1016/j.tourman.2012.08.006
2
S. Stepchenkova, F. Zhan / Tourism Management xxx (2012) 1e12
structured surveys of human respondents, mixed-methods
approach advocated by Echtner and Ritchie (1991, 1993), and,
more recently, textual data from blogs and websites. Much fewer
studies have addressed visual representations of destination
images.
This study comparatively analyzed pictorial materials produced
by DMOs and by destination visitors who post their photographs on
photo-sharing websites, such as Flickr. Peru was used as a case
study destination for this empirical research. The analytical
perspective for comparisons of the projected (DMO) and perceived
(Flickr) images of Peru was the Urry’s (1990) notion of hermeneutic
circle of representation. The main analytical approach employed in
this study was comparative content analysis and, to a lesser extent,
semiotic analysis. The study proposed and tested a methodology
that includes constructing maps of “aggregated” destination
images using the DMO-controlled and traveler-controlled sources.
The congruity of DMO and UGC representations of Peru was also
tested by constructing maps of geographical distributions of
collected pictorial materials.
2. Study background
2.1. Projected and perceived images
Photographs are means of “capturing” reality. From a DMO
perspective, photographs are the result of an “active signifying
practice in which media-makers select, structure, and shape what is
going to be registered on film and further alter and edit what is
eventually printed” (Hall, 1982: 64). From the travelers’ perspective, “slideshows and photographs are a common way to communicate personal trip experiences and perceived destination images”
(Schmallegger, Carson, & Jacobson, 2010: 245). In words of Albers
and James (1988: 136), “[p]hotographs are a medium through
which people relate to visual images and make them their own.” In
this paper, photos created and posted by travelers on photo-sharing
websites are termed user-generated content (UGC), and destination
images transmitted by the DMOs and UGC are referred to as projected and perceived images, respectively. Scholars have classified
UGC as having qualities of word-of-mouth promotion (Crotts,
Mason, & Davis, 2009; Yoo, Lee, Gretzel, & Fesenmaier, 2009) and,
similar to accounts by eyewitnesses, UGC tends to be trusted by
potential travelers. However, in research involving UGC, it is often
impossible to verify a user’s status with certainty (Lu &
Stepchenkova, 2012; O’Connor, 2008; Puri, 2007). Therefore, the
boundaries between projected (DMO) and perceived (UGC) images
may be blurred.
The relationship between projected and perceived images is not
entirely clear. Research guided by postcolonial theory (Said, 1978)
does not identify a difference between the two image types,
viewing destination images and their circulation as one hermeneutic circle of representation (Urry, 1990). In this circle, DMOs
project desirable destination images for potential tourists to
consume. Transmitted images spark the interest of potential tourists and help shape and reinforce the existing images that result
from previously acquired information from various sources, thus
affecting tourists’ perceptions of destinations. These perceptions
guide tourist’s gaze at a destination. Consciously or unconsciously,
tourists look for scenes and images that replicate their existing
perceptions. By capturing these images on camera, they close the
hermeneutic circle of destination representation.
Viewed from the lenses of the postcolonial theory, the notion of
hermeneutic circle supports the view that rather than endorsing
cultural understanding, tourism separates tourists and locals across
traditional cultural divisions. It is especially relevant with respect to
Western tourists (referred to as the Western ‘self’) visiting the Third
World destinations, i.e., the “mysterious, exotic, sensual, splendid,
cruel, despotic, sly, backward, and in decay/past its prime” (Caton &
Santos, 2008: 10, referencing Said, 1978) ‘other,’ thus, contrasting
the ‘self’ and the ‘other’ along a number of dimensions, or binaries,
such as “modern/ancient”, “advanced/primitive”, and “master/
servant,” to name just a few. Images perpetuated by travel advertisers often include depicting the ‘other’ as unchanged, unrestrained, and uncivilized (Echtner, 2002; Echtner & Prasad, 2003).
The unchanged aspect is thought to appeal to tourists’ search for
authentic, set in the past visual experiences, such as images of
archaeological ruins, old building, local residents in traditional
attire, etc. The unrestrained aspect reinforces the need for escape
and unleashing the ‘self’ in an exotic paradise of a destination. This
type of motivation was termed by Gray (1970) as ‘sunlust’ and is
often associated with the images of “sea, sun, and sand.” Finally, the
uncivilized aspect appeals to the need to explore and discover and
is often emphasized by the images of wild vegetation and indigenous people with body decorations (Caton & Santos, 2008).
While at a destination, tourists construct photographic images
as compositions of the most salient destination attributes, and this
practice is widespread on visual travel accounts (Albers & James,
1988; Day, 2002). Although it is plausible that potential visitors
are influenced by projected destination imagery, experiencing the
destination and producing accounts indicating that they have
“been there and seen that” means being exposed to the entire
visual universe of the ‘other’, which may not have been adequately
captured and represented in transmitted DMO images that shaped
tourists’ perceptions prior to visiting the destination. In such
a situation, the notion of the hermeneutic circle requires further
empirical confirmation, and the question of congruity emerges in
terms of DMO versus UGC photography, or projected versus
perceived images. This study aims to examine whether the concept
of the hermeneutic circle of representation has empirical support in
the context of online destination representations.
2.2. Analysis of pictorial materials
Destination photography, whether produced by DMOs or UGC,
communicates images that shape and re-shape travelers’ destination perceptions. There are two main components of a photograph,
content and composition (Albers & James, 1988). Content refers to
the appearances or signs captured in a photo in their totality. The
way in which these appearances are linked to each other and
presented to viewers constitutes the photo’s composition. Albers
and James (1988) posit that in the pictorial materials that are
analyzed in various domains, including tourism, redundancy in
content and composition is often found. This redundancy is not
simply random but is indicative of a convention and is used to
construct a meaning. The meaning of a photograph can be constructed from two perspectives, metonymic or metaphoric. From
the metonymic perspective, all signs presented on a photograph
stand for themselves and are interpreted at face value. The metaphoric perspective, in contrast, is concerned with what the image
signifies beyond mere appearances. All elements of a photograph
are treated as symbols that collectively allude to a meaning that lies
outside of the particular picture. A chrestomathic example of
a metaphoric “reading” would be a picture of a rose, which could be
classified as a plant or a flower or, in certain contexts, could
symbolize romance or passion (Strinati, 1995: 125). The chosen
perspective, metonymic or metaphoric, defines the methodological
approach for analysis.
The analytical treatment of photographs may be quantitative
and qualitative, with two broad groups of methodologies
employed, content analysis and semiotic analysis, respectively.
Content analysis has been used more often with textual content,
Please cite this article in press as: Stepchenkova, S., & Zhan, F., Visual destination images of Peru: Comparative content analysis of DMO and usergenerated photography, Tourism Management (2012), http://dx.doi.org/10.1016/j.tourman.2012.08.006
S. Stepchenkova, F. Zhan / Tourism Management xxx (2012) 1e12
but pictorial material can also be content-analyzed: in cultural
studies, imagery, including paintings, maps, videos, and even
landscapes, is often considered as a form of ‘text’. Content analysis
is attribute-based and is primarily concerned with quantitatively
describing the appearance of certain themes and attributes in the
collection of images, allowing the main focal themes in the pictures
to be identified and their frequencies, co-occurrence, clustering,
and other related issues to be recoded. Thus, content analysis
considers a photograph a metonym, and treats its elements (i.e.,
signs and manifestations) independently, as stand-alone attributes.
In other words, content analysis “breaks” a picture into a number of
attributes (or categories) guided by what is depicted on a photo and
takes these representations at face value. Classifying visual content
within the metonymic paradigm raises the issue of manifest versus
latent meaning, a familiar issue in content analysis studies of texts
(Krippendorff, 2004).
Albers and James (1988), in their study of the relationship
between photography, ethnicity, and travel, stated that in content
analysis, which is dominated by positivist paradigm with its
emphasis on quantification, several content categories have
a tendency to be significant. These are subjects (e.g., number of
people pictured, their age and gender), dress (e.g., everyday clothes
or festive attire), presentation (e.g., subjects pictured in an action or
as a formal portrait), and surroundings (e.g., studio, outdoor setting,
or tourist attraction). These authors also stated that co-occurrences
of focal elements in a photograph are an important factor in
describing a dataset. Once a dataset is quantitatively described, it
can be further analyzed from a temporal angle (when the pictures
were taken), geographic angle (where the images were taken), or
production perspective (who photographed and distributed the
images).
Up to date, there have been several studies that tested Urry’s
(1990) theory of “closed circle of representation” within the
content analysis framework. These studies applied content analysis
to photographs obtained by the tourists under the instruction of the
researcher, the technique called the visitor-employed photography
(VEP) (Garrod, 2008, 2009; MacKay & Couldwell, 2004). In a study
concerned with the image of Welsh seaside resort of Aberystwyth,
Garrod (2009) evaluated the visitors’ VEP photos and postcards of
the city and found that the two samples of images had common
features in terms of attractions captured in the images, locations
from which these attractions were depicted, as well as the overall
image composition. Another study by the same author (Garrod,
2008) compared the photos of visitors and residents of Aberystwyth; the interesting result was that both tourists and residents
adopted the same way of “reading” the destination. The author
speculated that these “readings” had been induced by the destination representations produced by the DMO. MacKay and
Couldwell (2004) compared photos of a national historic site in
the province of Saskatchewan, Canada, obtained using VEP, with
images used in promotional effort at that time. Each of the three
empirical studies confirmed to some degree the existence of Urry’s
circle of representation; however, they registered differences in
projected and perceived images as well.
As all texts, including visual images, imply certain meanings,
these meanings are not fixed and are a subject to reader’s interpretation. Even the most detailed and comprehensive content
analysis is unable to determine the “symbolic meaning of a specific
set of pictorial appearances” (Albers & James, 1988: 147). Semiotic
analysis considers the picture as a whole, and is concerned with
investigating how the content and composition of a picture
communicate intended messages through signs and symbols about
the place or the object they depict. As an approach to communication, which focuses on meaning and interpretation, semiotics
challenges the reductive transmission model that equates meaning
3
with “message”, i.e., content. The semiotic analysis “treats each
picture as a totality e marking the patterned relationships in its
content, connecting these to parallel and contrasting structures in
other pictures, and relating both to the written narratives that
accompany them” (Albers & James, 1988: 147). Signs do not just
“convey” meanings, but constitute a medium in which meanings
are constructed. It is important to note that, in semiotic analysis, the
meaning of a photograph by those who consume it can be different
from the intended meaning of its creator (MacKay & Couldwell,
2004). Thus, the semiotic approach can be characterized as highly
interpretive and dealing primarily with the latent content of the
photographs; however, as was noted by Duriau and Reger (2004),
the significance of latent content may be overestimates in certain
contexts.
Urry’s theory has been tested with semiotic analysis as well.
Guided by postcolonial discourse, Caton and Santos (2008)
analyzed photos made by students on a study abroad cruise trip
to a Third World country along the five dimensions: traditional/
modern, subject/object, master/servant, center/periphery, and
devious-lazy/moral-industrious. They concluded that the photos
complete the circle of representation and conform to the sociocultural ideologies of Western power and dominance. Markwick
(2001) analyzed Maltese postcard images and the contexts and
complexity of its consumption through the theoretical lenses of
tourist desire and motivation. The analysis included such
perspectives as “sun and sea”, “wanderlust”, authenticity, and
realism. The author identified the “complex circuits of consumption
and production” (Markwick, 2001: 435) and highlighted the
importance of the contexts in which these circuits operate. Finally,
Jenkins (2003) used a combination of quantitative and qualitative
approaches to determine whether Canadian backpackers in
Australia, the group, which is supposedly different from mass
tourists, followed the Urry’s notion. Photographs in brochures for
backpackers and backpackers’ own photography of Australia were
found to be part of a ‘spiral of representation’ through which the
iconic images of Australia were perpetuated.
In a discussion of the preferred methodological approach to
capture differences in projected and perceived images, all of abovementioned issues were considered. Content analysis was selected
as the primary method due to its ability to manage qualitative
material in a systematic, verifiable, and replicable way
(Krippendorff, 2004; Neuendorf, 2002). Although content analysis
does not allow the same complexity and richness of interpretation
as semiotic analysis does, it was judged adequate for the research
purposes: comparing DMO and UGC image attributes and
summarizing them into holistic visual representations of Peru. The
researchers aimed to construct “aggregated” DMO and UGC destination images based on the methodology described in Li and
Stepchenkova (2012), for which content analysis is a most suitable approach. It was also decided that once the similarities and
differences between DMO and UGC photos were identified, further
interpretation using the elements of semiotic approach would be
employed.
2.3. Peru
The current research uses Peru as a case study. Peru is the third
largest country in Latin America. It is becoming increasingly
popular among American and European tourists due to the development and diversity of its travel offerings, which blend history,
indigenous cultures, breathtaking scenery, and tremendous
ecological diversity. With three geographic regions e the coast, the
highlands, and the jungle, Peru is famed for having the world’s
highest navigable lake (Lake Titicaca), the world’s highest peak in
a tropical area (Huascaran), and the world’s deepest canyon (Colca
Please cite this article in press as: Stepchenkova, S., & Zhan, F., Visual destination images of Peru: Comparative content analysis of DMO and usergenerated photography, Tourism Management (2012), http://dx.doi.org/10.1016/j.tourman.2012.08.006
4
S. Stepchenkova, F. Zhan / Tourism Management xxx (2012) 1e12
canyon). Additionally, Peru is famous for its Inca civilization, the
man-made world wonder Machu Picchu, the Nazca lines, and
colonial churches (Peru Tourism Bureau, 2012).
In addition to numerous tourism destinations within Peru, the
country offers a variety of experiences for tourists who enjoy ecoand outdoor adventures, festivals and carnivals. Peru is home to
20% of the bird species and 10% of the reptile species in the world,
and every year, more than 3000 festivals are celebrated throughout
the country (Peru Tourism Bureau, 2012). In 2011, TripAdvisor, the
world’s largest travel site, awarded Machu Picchu in Peru the “No. 3
Travelers’ Choice World Destination” distinction based on its
destination popularity and favorability. According to TripAdvisor,
Machu Picchu is a more desirable destination than Paris, France or
New York City in the United States (TripAdvisor, 2012).
Growing numbers of tourists are attracted to events and destinations in Peru. According to Peru’s Ministry of Foreign Commerce
and Tourism, during the period of 2005e2009, annual international
tourist arrivals were increasing on average by 8% a year, reaching
2.25 million of international tourists in 2010 (MINCETUR, 2011a).
Because tourism in Peru has high growth potential, Peru invests
heavily in its tourism facilities and infrastructure. According to
Peru’s official tourism site (www.peru.travel/en), the country
currently has 7600 lodgings that offer 131,600 rooms and 229,900
beds to tourists. Ten airports in Lima, Arequipa, Chiclayo, Pisco,
Pucallpa, Iquitos, Cusco, Trujillo, Tacna and Juliaca are ready for
international flights, and Peru has more than 78,000 km of
highways.
Following the improvements to Peru’s tourism facilities and
infrastructure, in 2010, PromPeru initiated a promotional campaign
in the U.S., the second largest market for tourism in Peru (18% of
total market share, second to Chile, Peru’s neighbor nation,
according to MINCETUR, 2011b). This promotional campaign aimed
to draw more tourists from the U.S. not only to Peru’s most famous
tourism destinations but also to all of Peru’s geographical regions.
The goal of this campaign was to position Peru as a destination with
diverse and rich tourism attractions, both natural and cultural.
3. Methodology
The study aimed to answer the following questions:
1. What are the dominant attributes of destination image of Peru,
as presented through DMO and UGC visual content? Is there
congruity between the DMO and UGC images with respect to
the frequencies of pictured destination attributes? These
questions are answered by applying the content analysis
methodology to the collected DMO and UGC images of Peru
and conducting the Chi-square tests on the attribute frequency
data.
2. What are the overall, or holistic, DMO and UGC destination
images of Peru? What are the major differences between
them? What destination attributes tend to be pictured together
in the images from each source? These questions are answered
by constructing “aggregated” maps of Peru images using
statistical analysis of co-occurrences of destination attributes.
3. Which geographical regions of Peru are most often represented
in the DMO and UGC images? Is there congruity between the
DMO and UGC images in terms of the geographical distribution
of various locations within the destination? This question is
answered by constructing DMO and UGC maps of geographical
regions represented in the collected photos.
3.1. Data collection
Projected images of Peru were defined as images represented by
the official Peru tourism website, www.peru.travel. The website
contains a photo gallery with images organized by 24 geographical
regions of Peru and by six thematic categories. Because all thematic
photos were included in the sets organized by region, only the
regional sets of images were collected. In all, 530 DMO photos were
downloaded.
The user-generated photography in this study was represented
by the images from Flickr (www.flickr.com). Flickr is a website that
provides hosting services for its members’ photos and videos, acts
as a photo sharing and online community portal, and is widely used
by bloggers to host images embedded in blogs and other social
media (www.wikipedia.org). Flickr was created in 2004 and
became widely popular, in part because it allowed users to instantly
post photos from phones and PCs (Terdiman, 2004). In 2011,
Yahoo!, which purchased the site in 2005, reported that Flickr had
51 million registered members, 80 million unique visitors, and
more than 6 billion images (www.wikipedia.org). Compared to
other popular photo-sharing websites, such as Picasa (picasa.
google.com), Facebook (www.facebook.com), and Panoramio
(www.panoramio.com), the authors found that the data on Flickr
were more accessible for the purposes of this research. Because of
its size, accessibility, and links to other social media, Flickr was
selected to represent the UGC universe of Peru images in this
research.
With respect to the Flickr sample, the data collection process
was more complex. The researchers decided to collect the most
recent images (those taken in 2010) and those tagged with the
words “Peru” and “travel.” The “travel” tag was thought to indicate
that images were posted by travelers rather than by residents of
Peru. The Flickr search engine allows users to view only the first
4000 search results; however, the number of relevant images for
2010 was in excess of 23,000. Therefore, the researchers divided the
year into seven periods to ensure that, for the purposes of sample
selection, they obtained access to all Flickr images for that year (see
Table 1). Because the number of photos within the seven periods
differed, the researchers sampled the photos proportionally from
each period, for a total sample size of 500. For sampling within each
period, a systematic random sampling protocol with a randomly
selected starting number was used, as described by Lohr (1999: 42).
To minimize technical errors in the data collection process, each
Flickr image was associated with four numbers: (1) the final sample
ID (from 1 to 500); (2) the number within a particular time period
(for example, for JanuaryeFebruary images, the number ranged
from 17 to 3448; this number was obtained using Lohr’s procedure); (3) the page number within a particular time period (for
JanuaryeFebruary images, the number ranged from 1 to 164);
and (4) the number of a particular photo on a page. These identification numbers were entered into an Excel file before downloading the photos to ensure a smooth data collection process.
Table 1
Flickr sample.
Number of photos
Sample
JaneFeb
MareApril
May 1steMay 15th
May 16theJune 30th
July 1steJuly 14th
July 15theAug 31st
SepeDec
Total
3464
74
2807
60
3687
79
3422
73
3606
77
2964
64
3405
73
23,355
500
Please cite this article in press as: Stepchenkova, S., & Zhan, F., Visual destination images of Peru: Comparative content analysis of DMO and usergenerated photography, Tourism Management (2012), http://dx.doi.org/10.1016/j.tourman.2012.08.006
S. Stepchenkova, F. Zhan / Tourism Management xxx (2012) 1e12
During preparation for data collection, it was noticed that the
number and order of the photos on the Flickr pages changed
slightly over time. Therefore, for each of the seven periods, the
data collection was conducted on a single day, which typically took
4e6 h. A few out of focus or low resolution images were discarded
during the sampling process and were replaced with the images
that immediately followed the discarded ones. The entire process of
data collection from Flickr was completed in two weeks. Table 1
provides a description of the Flickr sample.
3.2. Category development and data coding
The researchers employed content analysis for the selected
DMO (530) and Flickr (500) images. A set of image categories for
Peru was developed using a combined approach. Approximately
10% of the selected images were examined by both authors separately to identify the main destination attributes pictured in the
photos (Glaser & Strauss, 1967; Neuendorf, 2002). The seminal
paper by Echtner and Ritchie’s (1993) and research by Albers and
James (1988) provided the theoretical grounding for category
formation. In consecutive discussions, 20 categories were developed, that were believed to represent all essential features of the
image of Peru, including Nature & Landscape, Traditional Clothing,
Festivals & Rituals, and Wild Life (the full list is given in Table 2). All
categories were clarified and refined by the authors through a few
iterations of comparative coding and finally described in the coding
guidebook with pictorial examples. Each photo was regarded as
a single unit of content (Krippendorff, 2004; Neuendorf, 2002).
Typically, in the content analysis of textual materials, each unit of
content is classified into one and only one category (Weber, 1990).
However, because photos are complex entities that cannot be easily
and reliably broken down into smaller content units, each image
was coded into several categories (in practice, no more than 4
categories).
“Content analysis stands or falls by its categories” (Berelson,
1952: 147). Therefore, formal reliability study was conducted to
assess the soundness of the categories e whether they were
formulated in such a way that images could be unambiguously
coded into these categories for further statistical analysis. In each of
the DMO and Flickr datasets, 100 images were randomly selected.
The reliability coefficient for each category was calculated as the
5
percentage of photos for which the raters agreed on the presence or
absence of a particular category. Agreement was higher than 90% in
all but one category, Nature & Landscape for the DMO sample (87%).
Because each photo was coded into up to four categories, the
agreement between the coders was also assessed using fuzzy kappa
as an integrative indicator (Dou et al., 2007; Hagen, 2003). The
fuzzy kappa, in contrast to the “crisp” kappa (Cohen, 1960), is
applied when each unit of content can be coded into multiple
categories. The fuzzy kappa values were 0.814 and 0.809 for the
DMO and Flickr samples, respectively, which indicated ‘almost
perfect’ (0.81e1) and ‘substantial’ (0.61e0.80) agreement between
the coders (Landis & Koch, 1977: 165). After image-by-image
comparisons of the pictures on which the coders disagreed, the
guidelines were revised with respect to the Nature & Landscape and
Archaeological Sites categories. In order for the reader to have
a better understanding of how the data were coded, Table 3
provides descriptions of categories which frequency in the
combined DMO and Flickr sample is above 5% (pictorial examples
are not included because of the space constraints).
3.3. Construction of “aggregated” image maps of Peru
Each image in the dataset was associated with up to four
different categories or destination attributes. Preliminary data
scanning showed that the dataset included many somewhat similar
images, such as landscapes with archeological ruins or people
wearing traditional closing while engaging in everyday activities.
The purpose of constructing destination image maps was to obtain
a visual summary of the data and a better understanding of which
destination attributes tend to appear together in the photos. The
researchers had to decide whether DMO or travelers tended to
prefer certain types of images, i.e., combinations of destination
attributes, or whether these co-occurrences of attributes in the
photos simply happened by chance.
The approach of constructing DMO and Flickr “aggregated”
image maps of Peru followed a procedure specified in Li and
Stepchenkova (2012). Each destination attribute has a certain
probability of appearing in a DMO or Flickr image, calculated as the
ratio of the frequency of that attribute and the respective sample
size. If any two attributes a1 and a2 are independent of one another,
the number of their co-occurrences is a random variable fa1a2 that is
Table 2
Attribute frequencies: chi-square analysis.
Categories
DMO (N ¼ 530)
DMO (%)
Flickr (N ¼ 500)
Flickr (%)
Total (N)
Total (%)
Nature & Landscape
People
Archaeological Sites
Way of Life
Traditional Clothing
Architecture/Buildings
Outdoor/Adventure
Wild Life
Art Object
Tourism Facilities
Urban Landscape
Domesticated Animals
Plants
Festivals & Rituals
Leisure Activities
Food
Country Landscape
Transport/Infrastructure
Tour
Other
217
175
115
62
69
58
52
40
36
32
22
16
9
30
17
9
9
7
14
14
40.9
33.0
21.7
11.7
13.0
10.9
9.8
7.5
6.8
6.0
4.2
3.0
1.7
5.7
3.2
1.7
1.7
1.3
2.6
2.6
206
153
92
106
44
40
45
36
18
21
24
26
29
6
19
20
8
9
2
31
41.2
30.6
18.4
21.2
8.8
8.0
9.0
7.2
3.6
4.2
4.8
5.2
5.8
1.2
3.8
4.0
1.6
1.8
0.4
6.2
423
328
207
168
113
98
97
76
54
53
46
42
38
36
36
29
17
16
16
45
41.1
31.8
20.1
16.3
11.0
9.5
9.4
7.4
5.2
5.1
4.5
4.1
3.7
3.5
3.5
2.8
1.7
1.6
1.6
4.4
a
b
Chi-squarea
p-valueb
17.017
4.688
0.000
0.030
5.278
0.022
3.129
12.153
15.175
0.077
0.000
0.000
4.982
0.026
8.453
7.798
0.004
0.005
df ¼ 1 in all tests.
Results significant at 0.1 level are shown.
Please cite this article in press as: Stepchenkova, S., & Zhan, F., Visual destination images of Peru: Comparative content analysis of DMO and usergenerated photography, Tourism Management (2012), http://dx.doi.org/10.1016/j.tourman.2012.08.006
6
S. Stepchenkova, F. Zhan / Tourism Management xxx (2012) 1e12
Table 3
Description of image categories (Top 10).
Category and its share in the Description
combined sample (%)
Nature & Landscape
(N&L) e 41.1
People (PP) e 31.8
Archaeological Sites
(AS) e 20.1
Way of Life (WOL) e 16.3
Traditional Closing
(TC) e 11.0
Architecture/Buildings
(A) e 9.5
Outdoor/Adventure
(OA) e 9.4
Wild Life (WL) e 7.4
N&L refers to pristine nature with minor
human-related elements. N&L includes
mountains, deserts, forests, jungles, rivers and
lakes, beaches, etc. N&L may include
archeological remnants/sites and natural
attraction like grottos, caves, etc.
If people and wild or farm animals are
present in the photo but are not the focus of the
image (too small, barely distinguishable, or on
the periphery of the picture), then they are not
coded. If they are one of the foci, then they are
coded into their respective categories. Pictures
are not necessarily “panoramic” but they should
not be close-up shots either.
PP includes local people, Indians, travelers, etc.
People should be reasonably in the center of
the picture, or one of the image’s foci.
AS includes pre-Inca and Inca structures,
remnants of pre-Inca and Inca architecture,
Nazca lines, tombs and burial sites, pyramids,
temples, ceremonial centers, fortresses, etc.
AS may include elements/designs of these
structures or their decorative elements.
WOL refers to how local people live and reflects
their living conditions, housing, or everyday
activities. WOL may include single or multiple
houses and architectures that are for use of
residents of Peru: private dwellings, garages,
local stores and restaurants, small churches,
street scenes in urban residential or rural areas,
local market scenes in a countryside, etc. WOL
includes traditional activities like weaving
and making art and crafts. WOL includes
“modern” activities like fixing a car, doing farm
work, selling goods and souvenirs on the street,
cooking, etc. WOL is meant to represent natural
living conditions of local people. If a picture is
“staged”, it should not be coded as WOL.
TC refers to traditional dress or festival costumes
which reflect distinct culture of Peru and its
regions. In Peru, some native people
(for example, Yagua Indians of Amazon)
do not wear conventional clothing, normally
they wear grass dresses and jewelry made
of animals’ teeth or tree leaves. Images of
natives in their traditional attire are also
coded as TC.
A refers to both interior and exterior of Western
style/Spanish/modern architecture in urban and
rural areas. A includes churches, monasteries,
palaces, public buildings, etc. (Building coded as
A should not be for daily use of local residents,
compare to WOL category.) Close-up images of
buildings’ elements such as doors, windows,
or ornaments are coded as A. Architectural
structures with minimal street context are
also coded as A; however, in this case the
building should be the main focus of the photo.
Buildings coded as A can be tourist attractions,
for example, the Cathedral of Lima, a famous
tourist attraction.
Boating, kayaking, hiking, trekking, horse
riding, fishing, swimming, eco-tourism, etc.
Anything that features active lifestyle and
appreciation of nature is coded as OA. Images
of people in a hiking gear sitting on a cliff and
enjoying the view fall into this category.
WL includes birds, marine mammals,
Andean camelids, butterflies, primates, etc.
WL objects should be one of the foci of
the picture.
Table 3 (continued )
Category and its share in the Description
combined sample (%)
Art Object (AO) e 5.2
Tourism Facilities
(TF) e 5.1
AO includes folk art, museum art, modern art,
or street art, reflecting Peru ancient or modern
culture. Close-ups of arts and crafts produced
by local people are coded as AO.
Hotels, restaurants, resorts and spas, beaches
with chairs and umbrellas, golf courses, etc.
are coded as TF. TF also includes eco-lodges,
camps, and ski resorts. Images of hotel rooms
are coded as TF. Trolley-cars and tour boats at a
pier or on the river/lake are coded as TF.
binomially distributed, and its expected value and standard deviation can be calculated using probabilities of attributes a1 and a2
and respective DMO and Flickr sample sizes (Hogg, McKean, &
Craig, 2005). The authors tested a hypothesis that actual and
expected co-occurrences of a1 and a2 in the respective samples
were significantly different at the 0.05 significance level. The test
statistic, or z-score, was calculated as the difference between the
number of actual and expected co-occurrences of a1 and a2 divided
by the standard deviation of fa1a2. The test statistic was compared to
critical z-score from a normal (0, 1) distribution, which, for a nondirectional hypothesis at the significance level of 0.05, is 1.96
(Ritchey, 2007). Larger (in absolute value) z-scores would indicate
that the independence of the two images is unlikely. Large and
positive z-scores indicate a positive statistical association between
the two attributes, whereas large (in absolute value) and negative
z-scores indicate a negative association, meaning that respondents
tend not to include a particular pair of attributes in their photos.
4. Results
4.1. Attribute frequencies: chi-square analysis
The entire dataset was coded by the second author of the paper.
Chi-square analysis was conducted to statistically compare the
frequencies of the categories for the DMOs and Flickr samples to
determine congruity between the projected and perceived images
across all categories. For the top three categories, Nature & Landscape, People, and Archaeological Sites, no statistical differences
were detected between the DMO and Flickr samples. Statistical
differences were found for nine out of 20 categories (see Table 2).
Overall, the DMO pictures tended to feature more Traditional
Clothing, Festivals & Rituals, Art Objects, and Tour attributes,
whereas Flickr users posted more images depicting the Way of Life,
Plants, Domesticated Animals, and Food destination attributes. This
finding indicates that the travelers are more interested in how
Peruvian people live their everyday lives, whereas the DMO focuses
on promoting the distinctive Peruvian culture, traditions, and art.
4.2. “Aggregated” image maps: attribute co-occurrences analysis
The “aggregated” map of Peru constructed for the DMO sample
is shown in Fig. 1. Only 15 categories with frequencies of 3% or
higher (cf. O’Reilly, 1990) are included. The three most frequent
destination attributes, Nature & Landscape, People, and Archaeological Sites, are represented by larger shaded bubbles. Other 12
categories are represented by smaller white bubbles. The link
between bubbles, if present, is accompanied by two numbers: the
number of co-occurrences of the two attributes represented by the
bubbles and a respective z-score (explained in Section 3.3). Only
links with positive z-scores higher than 1.96, which indicate
a positive association between the two attributes, are shown. For
Please cite this article in press as: Stepchenkova, S., & Zhan, F., Visual destination images of Peru: Comparative content analysis of DMO and usergenerated photography, Tourism Management (2012), http://dx.doi.org/10.1016/j.tourman.2012.08.006
S. Stepchenkova, F. Zhan / Tourism Management xxx (2012) 1e12
7
Fig. 1. Destination image map of Peru: DMO.
example, the number 26 at the link between Nature & Landscape
and Tourism Facilities means that there were 26 photos in the DMO
sample that were coded into both of these categories. The expected
co-occurrence number is 13 (not shown). The number 3.61 in the
parentheses, the z-score, is greater than 1.96, meaning that the
actual co-occurrence is statistically larger than the expected one,
that is, these two categories had a tendency to appear together in
the DMO images. The solid-line links between the bubbles indicate
statistically significant co-occurrences of ten or higher. The dashed
lines indicate statistically significant co-occurrences with
frequencies less than ten. There are no links between the three
largest bubbles, meaning that the most frequent images of Nature &
Landscape, Archaeological Sites, and People were not positively
associated statistically. In fact, there is a negative association (not
shown) between the Archaeological Sites and People categories:
the actual and expected co-occurrence numbers are 13 and 38,
respectively, and the z-score is 4.2. This means that the DMO
images contain these two attributes much less frequently than
would be expected. Overall, it seems that the DMO map has two
clusters of attributes. The first one connects Nature & Landscape
with People through the Outdoor/Adventure, Tourism Facilities,
and Tour attributes. The second cluster is formed by the People,
Traditional Clothing, Way of Life, and Festivals & Rituals attributes.
The “aggregated” Flickr destination image map is somewhat
simpler. It depicts the 15 most frequent destination attributes of
Peru with frequencies of 3% and above (the category Other (6%) is
not included). The three most frequent attributes are Nature &
Landscape, People, and Way of Life, and the Archaeological Sites
attribute is number four. On the Flickr map, the link between the
Nature & Landscape and Archaeological Sites attributes is significant (z-score is 3.73), indicating that travelers tend to photograph
archaeological ruins in the natural setting. The People attribute
on the Flickr map has significant links to four other attributes,
indicating that travelers tend to photograph people in their
living environment, wearing traditional clothing and engaging in
outdoor and leisure activities. Through the Leisure Activities attribute, the category of People is connected with the Urban
LandscapeeArchitecture/Buildings cluster as well as the Tourism
Facilities attribute. To ensure high reliability, the authors did not
distinguish between locals and tourists when coding the photos,
which in some instances was very hard to do with certainty.
However, in aggregated terms, such information can be obtained
from the maps: people performing outdoor or leisure activities, or
pictured within groups are most often tourists, whereas images
connecting People with the Way of Life, Traditional Clothing, and
Festivals & Rituals categories most often involve locals or indigenous people.
There are a few “stand-alone” categories on each map. For the
DMO map, these are Art Object, and Wild Life. For example, the Art
Object attribute was depicted in 36 photos (images of arts and
crafts, as well as objects from museum collections). However,
these objects were rarely depicted in combination with another
destination attribute. For the Flickr map, the stand-alone attributes are Wild Life, Plants, Domesticated Animals, Food, and Art
Object.
4.3. Geo-maps: representation of Peru regions
Geographical distribution maps of the DMO and Flickr photos
were constructed using ARC GIS software. Although all DMO photos
were associated with a particular region, only 398 images in the
Flickr sample specified the location where the photo was taken by
means of an attached tag, picture title or short commentary. The
authors used these descriptions as well as their own familiarity
with the country (the first author had been to Peru) to assign
locations to some of the Flickr images. The final geo-maps (Fig. 3)
used 100% of the DMO images and 84.6% (423 photos) of the Flickr
images.
In the DMO sample photos were distributed somewhat evenly
between 24 regions of Peru. Each region had from 1% to 8% of all
photos with an average of 3%, with the exception of Cusco (13%). In
the Flickr sample, the provinces of Cusco (52%), Arequipa (13%),
Puno (10%), Lima (9%), and Ica (8%) were the most popular. These
provinces comprise the so-called Gringo Trail (http://adventures.
Please cite this article in press as: Stepchenkova, S., & Zhan, F., Visual destination images of Peru: Comparative content analysis of DMO and usergenerated photography, Tourism Management (2012), http://dx.doi.org/10.1016/j.tourman.2012.08.006
8
S. Stepchenkova, F. Zhan / Tourism Management xxx (2012) 1e12
Fig. 2. Destination image map of Peru: Flickr.
worldnomads.com/destination/168/itinerary/23.aspx), the route
through Peru that is popular with American tourists. The route
includes the following attractions and places of interest: (1) Lima e
the city, coast; (2) Ica e Paracas wildlife reserve, Nazca lines; (3)
Arequipa e Santa Catalina monastery, Colca canyon; (4) Puno e
Lake Titicaca, Uros, and Taquile islands; and (5) Cusco e the city,
Sacred Valley, and Machu Picchu.
5. Discussion
5.1. DMO and user-generated photography
The projected and perceived images of Peru were analyzed
using content analysis as the main technique, followed with
statistical comparisons of the obtained frequency data: chi-square
Fig. 3. Geo-maps: DMO and Flickr. The left-hand map depicts 24 Peru regions, with the names of their capitals. It shows geographical distribution of the DMO images. The righthand map is also divided into 24 regions, with regions’ names printed. The right-hand map shows geographical distribution of the Flickr photos. The color intensity indicates the
percentage of photos taken in that region in the respective sample.
Please cite this article in press as: Stepchenkova, S., & Zhan, F., Visual destination images of Peru: Comparative content analysis of DMO and usergenerated photography, Tourism Management (2012), http://dx.doi.org/10.1016/j.tourman.2012.08.006
S. Stepchenkova, F. Zhan / Tourism Management xxx (2012) 1e12
tests of Peru destination attributes that appeared on the photos and
co-occurrences analysis on these attributes. Using the frequency
counts of attribute co-occurrences, the authors summarized the
DMO and Flickr photographs as “aggregated” maps and also constructed maps showing geographical distribution of images for
each sample. These quantitative analyses were supplemented by
elements of semiotic analysis when the researchers identified
“iconic” images of Peru. The notion of “iconic” is understood here as
being widely recognized and reproduced across a range of media
(cf. Hariman & Lucaites, 2008). The following discussion of the
study results, which attempts to answer the question whether
the projected and perceived images of Peru differ or converge in the
same hermeneutic circle of destination representation, uses the
postcolonial theory (Said, 1978) and the theory of tourist motivation (Gray, 1970) as reference points. The discussion is focused on
(1) the most frequent destination attributes of Peru identified in
this study; (2) attributes where significant differences were found;
and (3) the images of those Peru regions that were the most heavily
represented in the photos, namely, Lima, Ica, Arequipa, Puno, and
Cusco.
By the combined DMO and Flickr count, three most frequent
attributes of the Peruvian image were Nature & Landscape, People,
and Archaeological Sites. There were no statistical differences
between the DMO and Flickr samples with respect to frequencies of
these three attributes (Table 2). However, the maps (Figs. 1 and 2)
indicate that travelers do tend to picture archaeological sites as part
of the Peruvian natural landscape. The iconic image of Machu Picchu (Cusco region) e the sweeping vistas of archaeological ruins in
a mountainous landscape e was repeatedly “produced, projected,
perceived, propagated and perpetuated” (Jenkins, 2003: 324) in
both DMO (38 photos) and Flickr (43 photos) samples. In several
cases, even the angle from which the complex was photographed
was the same (Exhibit 1a and b). Interestingly, images of the Nazca
lines (Ica region), which are heavily promoted by the DMO, made
a distinctive group in both samples but lacked the exactness of
reproduction on the Flickr side. The authors would speculate that
the reasons include the enormous size of the Nazca lines, the need
of special equipment to capture them on a photograph, and lack of
observation towers in the area. While there are tours that offer
a plane ride over the Nazca lines, such tours are expensive.
Nevertheless, due to their distinctive shapes and colors, collectively, the Nazca lines can be considered the iconic images of Peru.
Images of Machu Picchu, the Inca Trail (also present in both
samples with obvious similarities in content and composition), and
the Nazca lines appeal to tourist desire of ‘wanderlust’, which was
defined by Gray (1970: 93e94) as “the desire to exchange the
known for the unknown, to leave things familiar and to go to see
different places, people, and cultures or relics of the past in places
famous for monuments. and contributions to society.” These
9
images also evoke the notion of unchanged (Echtner & Prasad,
2003); however, ancient ruins are balanced in both samples by
the photos of more modern buildings and architecture, transportation and infrastructure, tourism facilities, etc. As follows from
the frequency analysis of relevant categories (Table 2), the DMO and
Flickr users perceive the country’s place along the “modern/
ancient” dimension somewhat similarly.
The category frequency counts (Table 2) point out that in the
Flickr photos, the Way of Life, Domesticated Animals, Food, and
Plants destination attributes are overrepresented, whereas the
Traditional Closing, Festivals & Rituals, Art Object, and Tour attributes are underrepresented, compared to the respective categories
in the DMO sample. This finding indicates that travelers are more
interested in how Peruvian people live their everyday lives (Fig. 2),
whereas the DMO focuses on promoting the distinctive Peruvian
culture, traditions, and art (Fig. 1). A closer look at the DMO photos
coded into the People, Way of Life and Traditional Clothing categories, that make a distinctive triangle on both maps, often reveal
“picture perfect” local people clad in clean bright clothes or people
doing traditional crafts in a scenic landscape. Flickr users, on the
other hand, are less concerned with perfection, their images are
more down-to-earth and realistic; they often feature Peruvian
people in contemporary dress and in everyday life situations and
surroundings.
Markwick (2001) points to the tourist’s desire to experience
a commonplace, everyday, and authentic life of the local people.
These experiences can be “front stage”, i.e., highly visible, taking
place primarily in the public areas, and “back stage”, i.e., confined to
the private spaces. Since travelers, in general, do not belong to the
“social fabric” of the host community, their experiences with
authentic are primarily “front stage.” A good example of the “front
stage” treatment is DMO’s photos that picture tourists and local
hosts doing farm work, milking a cow, or trekking with llamas on
the Inca trail (Exhibit 2a). These images intend to give visitors
a glimpse in the everyday life of Peruvian people and demonstrate
the role of traditional animals like llama in Peruvian culture. (On
a side note, in the Flickr sample, llama and alpaca are more likely to
be photographed solely, as symbols of Peru and destination image
icons.) The desire for authentic, “back stage” perspective of the host
community in the Flickr sample become apparent through the
images of closed doors of Peruvian homes, people sitting next to
their houses with the doors open, and even people inside their
houses. Exhibit 2b presents an Uru woman cooking in her own
home on Uros islands. Uros is a group of floating islands on Lake
Titicaca (Puno region) made of dried totora reeds that can be
reached by a 20-min boat ride from Puno. Colorful images of the
islands in the morning sun, when the first boats arrive to Uros, local
people in very distinctive dress are an iconic theme with similarities in both samples. It is interesting to observe how “back stage”
Exhibit 1. a &b. Machu Picchu: DMO and Flickr.
Please cite this article in press as: Stepchenkova, S., & Zhan, F., Visual destination images of Peru: Comparative content analysis of DMO and usergenerated photography, Tourism Management (2012), http://dx.doi.org/10.1016/j.tourman.2012.08.006
10
S. Stepchenkova, F. Zhan / Tourism Management xxx (2012) 1e12
Exhibit 2. a&b. In search for authenticity: DMO and Flickr.
experiences become “front stage” when Uru people demonstrate to
tourists how the islands are made and let them have a peek inside
their houses.
Despite the obvious interest of the travelers in the everyday
activities of Peruvian people, there were few photographs in the
Flickr sample that depicted both tourists and locals in the process of
meaningful interaction, for example, on the market, in the streets,
at a festival, or on a farm. The maintained distance on the part of the
travelers was suggestive of the Western tourist viewing the ‘other’
as different and alien. Pictures that captured poor living conditions
of the local people reinforced the “under-developed” aspect of the
destination. However, no binary master/servant was evoked in the
pictures, nor any photograph in the data depicted local people in
sexually suggestive poses and roles, a finding that runs contrary to
what has been reported in research on the Third World
destinations.
The “sunlust” aspect of tourist motivation is typically associated
with mass tourism and all-inclusive “sun and sea” vacation; it is
focused on the “body” and often interpreted as the desire to escape
to the tropical paradise and rejuvenate the self. While Peru does
make one of the best surfing spots near Lima, its shoreline is not
particularly welcoming for bathing. Therefore, the bodily dimension of “sunlust” is transformed as a search for outdoor activities
and adventure: hiking, trekking, biking, mountaineering, rafting,
windsurfing, and eco-tourism. As Outdoor/Adventure, Leisure
Activities, or Tourism Facilities attributes associated with the
“sunlust” dimension of the image are concerned, no differences
between the samples were found.
There are large disparities in the numbers of DMO and Flickr
photos coded into the Festivals & Rituals category, with 30 (5.7%)
and six (1.2%), respectively. The official travel website of Peru
promotes 140 annual festivals in various regions of the country. The
timeframe of many festivals is short, often a day or two. During the
peak season of May to July (w50% of all photos on Flickr), only 40
festivals (29%) around the country are scheduled. Of these 40, only
15 are held in the provinces comprising the Gringo Trail. Thus,
disparities between the two samples in the number of photos in the
Festivals & Rituals category point to a “scheduling issue”: tourists’
chances of attending a festival or carnivals during a trip to Peru are
not very high, unless their primary goal is to attend festivals. In
terms of promotion of this colorful Peruvian tradition, the DMO
should consider focusing on one or two festivals in the Arequipa,
Puno, or Cusco regions during the shoulder or off-peak seasons.
Peru is a country that is famed for being home to a diversity of
animals and plants. Photos of birds on ocean cliffs in the Parakas
reserve (Ica region) or condors in the Colca canyon (Arequipa
region) are very recognizable images in both samples and can be
termed iconic. Although the frequency of the Plants category was
significantly higher in the Flickr sample (6.6%) than in the DMO
sample (1.7%), overall, the total counts were relatively low, with no
iconic images. There is also a distinctive group of Flickr photographs capturing nicely prepared restaurant food (e.g., ceviche) and
plentiful of fruit and vegetables on the street markets; however, the
food aspect of the destination is not fully developed in the DMO
sample. Finally, the DMO and Flickr photographs also differed
significantly in the Other category: the lower counts in the DMO
sample indicate that the DMO photos are more structured and
planned, whereas the Flickr photos experience a larger amount of
“noise” (i.e., irrelevant material). This finding was somewhat
expected.
5.2. Methodological issues and further research
From a methodological perspective, this exploratory study
examines the feasibility of deriving “aggregated” destination
images from pictorial content on the web. The proposed approach
of constructing maps of large pools of visual data allows comparisons of the destination’s projected and perceived images. The
approach makes it possible to identify “common” or “typical” visual
images by placing the most frequent destination attributes and
significant links between them on the map. For example, a “typical”
DMO image of Peru would be one that pictured people in traditional or festive costumes participating in a festival (Fig. 1). To the
authors’ best knowledge, this is the first study to apply a methodology of mapping pictorial data in destination image research.
The results in this study were obtained using quantitatively
oriented content analysis as an analytical tool. Studies on photo
elicitation showed that interpretations of image producers and
receivers are often different and that the intent of the producers of
images cannot be reliably gauged (MacKay & Couldwell, 2004).
Quantitative content analysis, which treats a photograph as
a metonym rather than a metaphor, circumvents this issue and
derives as much meaning from the photos as can be reliably obtained through the category coding and follow-up statistical
procedures. As Pullman and Robson (2007) pointed out, analytically, photographs are analogous to verbal content. Therefore, three
“rules” for developing categories apply: categories should be (1)
exhaustive; (2) exclusive (i.e., they should not overlap); and (3)
analytically interesting. Rule (2) is not mandatory, but if it is not
met, it may complicate the statistical analysis (Weber, 1990). For the
Please cite this article in press as: Stepchenkova, S., & Zhan, F., Visual destination images of Peru: Comparative content analysis of DMO and usergenerated photography, Tourism Management (2012), http://dx.doi.org/10.1016/j.tourman.2012.08.006
S. Stepchenkova, F. Zhan / Tourism Management xxx (2012) 1e12
purposes of this research, the authors consider a picture a unit of
content that can have several prominent features. The authors were
most interested in the “overlap” of these features because this
overlap allowed the authors to calculate the co-occurrences of
these features and to construct “aggregated” destination images.
Thus, each picture was treated as a survey respondent who
provides up to four answers about destination’s most prominent
features, for example, Nature & Landscape; Archaeological
Site, People, Traditional Clothing. A large enough number of
co-occurrences of two particular attributes allow a conclusion of
their statistical connection in the photos. With respect to rule (1),
the relatively small number of images that were classified into the
Other category indicates that the coding scheme was indeed
exhaustive. With respect to rule (3), the developed categories allow
the interpretation of the photos along the dimensions identified by
the relevant theories.
This research used a relatively large sample size for the study of
visuals (cf. Garrod, 2009; Jenkins, 2003; Jutla, 2000): 530 DMO
images and 500 Flickr images. For instance, Garrod’s (2009) study
used a sample of 164 photographs and an unspecified but
“comparable” number of postcards. Jenkins (2003) used 17 travel
brochures with an unspecified number of photos. In this research,
the authors required a larger sample size to reliably estimate the
probabilities of various destination features appearing in the
images. Li and Stepchenkova (2012), who constructed perceptual
maps of images of the U.S. as perceived by Chinese travelers, used
a sample size of 1600 but with the verbal data. When deciding on
a sample size in content analysis, there is always a trade-off
between accurate and reliable frequency data for statistical analysis and the manpower needed to complete the study. In this study,
however, once the categories had been developed and tested
(a lengthy procedure indeed), the coding itself was speedy and
relatively simple.
The researchers also note that the DMO image pool is a “census”
of all photographs from the photo gallery page on the DMO website,
whereas the Flickr pool of images is a sample. The authors took
great care to randomize the sample (see Section 3.1) to obtain an
image pool that was representative of all photos on the Flickr site.
In retrospect, this study could have adopted a modified approach,
similar to the one described by Tussyadiah and Fesenmaier (2009):
these researchers selected only those YouTube videos about New
York City that had received at least one comment from viewers. The
adoption of a similar requirement for the Flickr photos might have
resulted in a sample with stronger persuasive power. For future
research a data collection procedure that combines both randomization and commentary features should be seriously considered.
Another direction of future research is the construction of “aggregated” destination image maps as perceived by travelers from
countries with different “cultural proximity” to a destination
(Kastenholz, 2010; MacKay & Fesenmaier, 2000). Various degrees of
cultural distance can be ensured by using profiles of UGC users that
indicate the country of residence. Knowledge of the relationship
between visitors’ cultural backgrounds and their perceptions of
a destination will help destination DMOs to more effectively
market their tourism offers in culturally different target markets.
6. Conclusion
This empirical study has methodological, theoretical, and practical implications. Due to an increasing relevance of visual
communication and a growing number of visualizations in online
media, quantitative generalization about these visuals is an
important methodological question. This study conducted
a comparative analysis of DMO and Flickr images of Peru to assess
the level of congruity between the projected and perceived images
11
of this destination. The study found a number of similarities and
differences in representation of the destination, thus, lending
partial support to Urry’s (1990) theory of hermeneutic circle of
destination representation. Some of the binaries of the postcolonial
discourse were detected, pointing out to the cultural distance still
existing between Western tourists and the destination. Overall, the
DMO side tends to present a well-rounded image of Peru by giving
“voice” to all of the country’s regions and focuses on the natural
beauty of the country, its archaeological heritage, customs, traditions, and art. While on the Flickr side all of these themes are
present, the UGC photography also reflects tourists’ strong interest
in how Peruvian people live their lives and their everyday activities.
To answer this call, DMO may consider expanding on existing
ethnographical attractions or engaging in new ones, especially in
the regions with the heaviest visitation volumes. Currently, Peru
provinces of Cusco, Arequipa, Puno, Lima, and Ica are more heavily
presented by UGC exemplified by Flickr photographs and may
contribute disproportionally in shaping the overall destination
image of Peru.
Acknowledgments
The authors would like to express their gratitude to Dr. Andrei
Kirilenko for his help with creating geographical maps of Peru
images using the ARC GIS software.
References
Albers, P. C., & James, W. R. (1988). Travel photography: a methodological approach.
Annals of Tourism Research, 15, 134e158.
Berelson, B. (1952). Content analysis in communication research. Glencoe, IL: Free
Press.
Caton, K., & Santos, C. A. (2008). Closing the hermeneutic circle? Photographic
encounters with the other. Annals of Tourism Research, 35(1), 7e26.
Cohen, J. (1960). A coefficient of agreement for nominal scales. Educational and
Psychological Measurement, 20(1), 37e46.
Connell, J. (2005). Toddlers, tourism, and tobermory: destination marketing issues
and television-induced tourism. Tourism Management, 26(5), 763e776.
Crotts, J. C., Mason, P. R., & Davis, B. (2009). Measuring guest satisfaction and
competitive position in the hospitality and tourism industry: an application
of stance-shift analysis to travel blog narratives. Journal of Travel Research,
48(2), 139.
Day, J. (2002). Image selection in destination positioning: a new approach. Journal of
Vacation Marketing, 8(2), 177e186.
Dou, W., Ren, Y., Wu, S., Chen, Y., Bloyet, D., & Constans, J.-M. (2007). Fuzzy kappa
for the agreement measure of fuzzy classifications. Neurocomputing, 70(4e6),
726e734.
Duriau, V. J., & Reger, R. K. (2004, March 10e12). Choice of text analysis software in
organization research: Insight from a multi-dimensional scaling (MDS) analysis.
Paper presented at the 7th International Conference on the Textual Data
Statistical Analysis, Louvain-la-Neuve, Belgium.
Echtner, C. M. (2002). The content of third world tourism marketing: a 4A approach.
International Journal of Tourism Research, 4, 413e434.
Echtner, C. M., & Prasad, P. (2003). The context of the third world tourism
marketing. Annals of Tourism Research, 30, 660e682.
Echtner, C. M., & Ritchie, J. R. B. (1993). The measurement of destination image: an
empirical assessment. Journal of Travel Research, 31(4), 3e13.
Echtner, C. M., & Ritchie, J. R. B. (1991). The meaning and measurement of destination image. The Journal of Tourism Studies, 2(2), 2e12.
Gallarza, M. G., Saura, I. G., & Garcia, H. C. (2002). Destination image e towards
a conceptual framework. Annals of Tourism Research, 29(1), 56e78.
Garrod, B. (2009). Understanding the relationship between tourism destination
imagery and tourist photography. Journal of Travel Research, 47(3), 346e358.
Garrod, B. (2008). Exploring place perception: a photo-based analysis. Annals of
Tourism Research, 35(2), 381e401.
Gartner, W. C. (1993). Image formation process. Journal of Travel & Tourism
Marketing, 2(3), 191e216.
Glaser, B. G., & Strauss, A. L. (1967). The discovery of grounded theory: Strategies for
qualitative research. Chicago, IL: Aldine Publishing Company.
Gray, H. P. (1970). International travel e International trade. Lexington Heath:
Lexington Books.
Hagen, A. (2003). Fuzzy set approach to assessing similarity of categorical maps.
International Journal of Geographical Information Science, 17(3), 235e249.
Hariman, R., & Lucaites, J. L. (2008). Public identity and collective memory in U.S.
iconic photography. In L. Olson, C. Finnegan, & D. Hope (Eds.), Visual rhetoric: A
Please cite this article in press as: Stepchenkova, S., & Zhan, F., Visual destination images of Peru: Comparative content analysis of DMO and usergenerated photography, Tourism Management (2012), http://dx.doi.org/10.1016/j.tourman.2012.08.006
12
S. Stepchenkova, F. Zhan / Tourism Management xxx (2012) 1e12
reader in communication and American culture (pp. 175e198). Los Angeles: Sage
Publications, Inc.
Hall, S. (1982). The rediscovery of ideology: return of the repressed in media
studies. In M. Gurevitch, T. Bennett, J. Curran, & J. Woollacott (Eds.), Culture,
society, and the media (pp. 56e90). London: Methuen.
Hogg, R. V., McKean, J. W., & Craig, A. T. (2005). Introduction to mathematical
statistics (6th ed.). Upper Saddle River, NJ: Pearson Prentice Hall.
Hunt, J. D. (1975). Image as a factor in tourism development. Journal of Travel
Research, 13(3), 1e7.
Jenkins, O. H. (2003). Photography and travel brochures: the circle of representation. Tourism Geographies, 5(3), 305e328.
Jutla, R. S. (2000). Visual image of the city: tourists’ versus residents’ perceptions of
Simla, a hill station in North India. Tourism Geographies, 2(4), 404e420.
Kastenholz, E. (2010). “Cultural proximity” as a determinant of destination image.
Journal of Vacation Marketing, 16(4), 313e322.
Krippendorff, K. (2004). Content analysis: An introduction to its methodology
(2nd ed.). Thousand Oaks, CA: Sage Publications.
Landis, J. R., & Koch, G. G. (1977). The measurement of observer agreement for
categorical data. Biometrics, 33, 159e174.
Lohr, S. L. (1999). Sampling: Design and analysis. Duxbury Press.
Li, X., & Stepchenkova, S. (2012). Chinese outbound tourists’ destination image of
America: part I. Journal of Travel Research, 51(3), 250e266.
Lu, W., & Stepchenkova, S. (2012). Ecotourism experiences reported online: classification of satisfaction attributes. Tourism Management, 33(3), 702e712.
MacKay, K. J., & Couldwell, C. M. (2004). Using visitor-employed photography to
investigate destination image. Journal of Travel Research, 42(4), 390e396.
MacKay, K. J., & Fesenmaier, D. R. (2000). An exploration of cross-cultural
destination image assessment. Research note. Journal of Travel Research,
38(4), 417e423.
Markwick, M. (2001). Postcards from Malta: image, consumption, context. Annals of
Tourism Research, 28(2), 417e438.
Mayo, E. J. (1973). Regional images and regional travel behavior. Research for changing
travel patterns: Interpretation and utilization. Paper presented at the Travel
Research Association 4th Annual Conference, Sun Valley, Idaho.
MINCETUR. (2011a). Boletin Mensual de Turismo e Año I e N 01. Last retrieved on
17.08.12 from. http://www.sociedadhotelesdelperu.org/estadisticas/Boletin%
20Mensual%20de%20Turismo%20Ano%20I%20%20N%20001_140111.pdf.
MINCETUR. (2011b). Informe de Evaluación del Plan Operativo Institucional e Año
2010. Last retrieved on 17.08.12 from. http://www.mincetur.gob.pe/newweb/
portals/0/EVALUACION_DEL_POI_2010.pdf.
Neuendorf, K. A. (2002). The content analysis guidebook. Thousand Oak, CA: Sage.
O’Connor, P. (2008). User-generated content and travel: a case study on tripadvisor.com. In Information and communication technologies in tourism: Proceedings
of the international information and communication technologies in tourism
conference. Austria, 2008 (pp. 47e58). New York, NY: Springer Wein.
O’Reilly, M. D. (1990). Free elicitation of descriptive adjectives for tourism image
assessment. Journal of Travel Research, 28(4), 21e26.
Peru Tourism Bureau. (2012). About Peru. Last retrieved 28.05.12, from. http://www.
visitperu.com/aboutperu.htm.
Pike, S. (2002). Destination image analysis: a review of 142 papers from 1973 to
2000. Tourism Management, 23(5), 541e549.
Pullman, M. E., & Robson, S. K. A. (2007). Visual methods: using photography to
capture customers’ experiences with design. Cornell Hotel and Restaurant
Administration Quarterly, 48(2), 121e144.
Puri, A. (2007). The web of insights: the art and practice of webnography. International Journal of Market Research, 49(3), 387e408.
Ritchey, F. J. (2007). Statistical imagination (2nd ed.). Boston, MA: McGraw-Hill.
Said, E. (1978). Orientalism. New York: Random House.
Schmallegger, D., Carson, D., & Jacobsen, D. (2010). The use of photographs
on consumer generated content websites: practical implications for
destination image analysis. In N. Sharda (Ed.), Tourism informatics:
Visual travel recommender systems, social communities, and user interface design
(pp. 243e260). Hershey, PA: Information Science Reference.
Stepchenkova, S., & Mills, J. E. (2010). Destination image: a meta-analysis of
2000e2007 research. Journal of Hospitality Marketing and Management, 19(6),
575e609.
Strinati, D. (1995). An introduction to theories of popuar culture. London: Routledge.
Tasci, A. D. A., Gartner, W. C., & Cavusgil, S. T. (2007). Conceptualization and operationalization of destination image. Journal of Hospitality & Tourism Research,
31(2), 194e223.
Terdiman, D. (12.09.2004). Photo site a hit with bloggers. Last Retrieved on 29.04.12
at. http://www.wired.com/culture/lifestyle/news/2004/12/65958.
TripAdvisor. (2012). TripAdvisor unveils 2011 travelers’ choice destinations awards.
Last retrieved 28.05.12, from. http://www.tripadvisor.com/PressCenter-i4665c1-Press_Releases.html.
Tussyadiah, I. P., & Fesenmaier, D. R. (2009). Mediating tourist experiences: access to
places via shared videos. Annals of Tourism Research, 36(1), 24e40.
Urry. (1990). The tourist gaze. London: Sage Publications.
Weber, R. P. (1990). Basic content analysis (2nd ed.). Newbury Park, CA: Sage.
Yoo, K. H., Lee, Y., Gretzel, U., & Fesenmaier, D. R. (2009). Trust in travel-related
consumer generated media. Information and Communication Technologies in
Tourism, 2009, 49e59.
Svetlana Stepchenkova, Ph.D., is an assistant professor at
the Department of Tourism, Recreation, and Sport
Management at the University of Florida and an associate
director of Eric Friedheim Tourism Institute. She is interested in quantitative assessment of destination image
using qualitative data. She is also interested in influence of
media messages on image formation, and destination
image as a factor in explaining destination choice. Svetlana
studies applications of information technologies in travel
and tourism, particularly virtual travel communities,
destination websites, and user-generated content as
a means of obtaining a competitive advantage in destination marketing and management.
Fangzi Zhan received her M.S. degree in Tourism at the
University of Florida in 2012. She has been involved in
a number of research projects as a graduate student. Her
research interests are visual representations of destination
images and social media research.
Please cite this article in press as: Stepchenkova, S., & Zhan, F., Visual destination images of Peru: Comparative content analysis of DMO and usergenerated photography, Tourism Management (2012), http://dx.doi.org/10.1016/j.tourman.2012.08.006