INTERNATIONAL JOURNAL OF TOURISM RESEARCH

Int. J. Tourism Res. (2012)

Published online in Wiley Online Library
(wileyonlinelibrary.com) DOI: 10.1002/jtr.1899

Conjoint Analysis of Tourist Choice of

Hotel Attributes Presented in Travel
Agent Brochures
Rubén Huertas-Garcia1,*, Marta Laguna García2 and Carolina Consolación3
1
Department of Economics and Business Administration, Barcelona University, Barcelona, Spain
2
Department of Administration and Marketing Research, Valladolid University, Segovia, Spain
3
Department of Business Administration, Technical University of Catalonia, Barcelona, Spain

ABSTRACT to be a common source of information for tourists

(Bieger et al., 2000; Andereck, 2005; Molina and
The purposes of this study are to demonstrate
Esteban, 2006). The influence information pub-
how it is possible to determine which attributes
lished in brochures has on the decision-making
are the most important in the final choices of
process means brochures play an important role
tourists who use a travel agent brochure as a
in promoting tourist businesses and destinations.
source of information and how these attributes
Several studies analyse the influence various
influence perceived value in a pre-purchase
sources of information have on destination choice.
stage. We conduct the study in three phases: (i) a
Andereck (2005) argued that evaluations of bro-
qualitative study, (ii) an experiment using
chures by tourists influence interest in visiting a
choice-based conjoint analysis by means of a
destination, and these evaluations influence the
fractional factorial experimental design and
choice of destination as well. Hsieh and O’Leary
(iii) another experiment using a full factorial
(1993) reported that brochures are the third most
derived from the same design. Results suggest
common source of information used by travellers,
advertisement size, a hotel’s starred rating and
after interpersonal communication from relatives
price influence perceived value at this stage. The
and friends. In one of the most technologically
presence of a positive combined effect of price
advanced countries, Yamamoto and Gill (1999)
and advertising was found. Implications and
found that brochures are one of the two most
directions for future applications are offered,
important sources of information used by Japanese
focusing particularly on marketing services.
travellers who buy travel packages.
Copyright © 2012 John Wiley & Sons, Ltd.
More recently, Chiou et al. (2008) suggested
that traditional brochures greatly influence
Received 05 November 2010; Revised 29 May 2012; Accepted behaviour in verbalizer consumers, those who
11 June 2012 prefer to process written or verbal information
from sources such as relatives, friends or travel
Keywords: perceived value; choice-based conjoint
agents over pictorials. Ortega and Rodriguez
analysis; tourist brochure; experiments in blocks.
(2007) found that brochures are one of the most
important audio-visual formats for Spanish
tourists seeking destination information. Molina
INTRODUCTION and Esteban (2006) proposed a model of
brochure utility and studied its significance as

D
espite the increasing importance of the image generators. Clearly, the brochure is not
Internet as a communication and distri- an extinct product.
bution instrument, brochures continue Despite the importance of tourist brochures as
communication tools, little is known about their
*Correspondence to: Rubén Huertas-García, Department effectiveness. Some researchers call attention to
of Economics and Business Administration, Barcelona
University, Avda. Diagonal, 696, Barcelona, Spain. large amounts of money spent by tourism distri-
E-mail: [email protected] buters in brochure production but rarely
Copyright © 2012 John Wiley & Sons, Ltd.
 R. Huertas-Garcia, M. Laguna García and C. Consolación

demonstrate their usefulness or effectiveness in phase involves selection of destination, a condi-

meeting objectives (Hodgson, 1993; Andereck, tional decision because choice depends on length
2005). Little is known about the effect of tourist of holiday. Various socio-demographic factors
brochures as tools to promote destination and including income, number of family members,
tourism organizations. There is also little empir- age, environment and attitudes toward holidays
ical evidence concerning effects generated by lay- influence both decision phases. Having made a se-
outs presented to tourists on brochures, especially lection, the decision process continues with a
in the pre-purchase stage. The literature contains series of derivative decisions generally made prior
sufficient evidence to suggest that the way infor- to a trip: (i) accommodations, (ii) travel company,
mation is presented, the layout, influences con- (iii) transportation and (iv) changes to adjust
sumer evaluations and purchase intentions length of trip. The choice of accommodation is
(Munger and Grewal, 2001; Rewtrakunphaiboon often made at a later stage in the decision-making
and Oppewal, 2008). The purpose of this study is process for a tourist trip.
to demonstrate that it is possible to determine the The next stage involves searching for informa-
most important attributes people consider when tion about characteristics offered by service
choosing a hotel and using a travel agent companies and their prices. Consumers use both
brochure as a source of information. This paper internal and external information sources when
reports results from a two-step experiment where making purchase decisions. Consumers typic-
the design of the second was derived from the ally use internal information from memory as
results of the first. This process helps determine the first source and then move to external
what kind of information displayed by a hotel searches by using a variety of sources, a process
in the tourist brochures influences a consumer’s requiring considerable effort. Travellers often
perceived value and choice. visit several travel agencies or search for infor-
mation on the Internet. The two primary
categories of destination factors are related to
THE TOURIST CHOICE PROCESS AND (i) activities tourists can experience at a destin-
INFORMATION SEARCHES ation such as sports, dining, historic sites, night-
life, shopping and outdoor recreation, and
Choosing a holiday destination is not a simple (ii) other destination characteristics such as
decision because selection is derived from a set price, accessibility, climate, health and safety,
of independent choices. It is a complex, multi- residential attitudes toward tourists, language
faceted process in which choices are interrelated barriers, availability and quality of accommoda-
and evolve (Dellaert, et al., 1998). It is well estab- tion, and air quality (Dellaert et al., 1998).
lished in consumer behaviour literature that the All sources of travel information including
purchase decision process consists of five stages: brochures help travellers make informed deci-
problem recognition, information search, product sions. Planning and searching for information
options evaluation, purchase decision and post- offer tourists the opportunity to reduce uncer-
purchase support (Kotler, 2000). The purchase tainty risk and disappointing experiences (Roehl
process for a traveller begins when a consumer and Fesenmaier, 1992). Fodness and Murray
recognizes a need or unsolved problem, having (1999) suggested that for companies operating in
received external stimulus usually from tourist a competitive market, it is essential to understand
advertisements that trigger thoughts on making the search information process tourists follow.
a purchase. The more unfamiliar a destination is, the more
In the context of tourism services, the purchase time and effort must be spent on pre-purchase
process is much more complex because travel information searching (Andereck, 2005). Wicks
includes various services (e.g. destination, trans- and Schuett (1991) found that travellers who use
portation and hotel). Nicolau and Mas (2006) brochure information as an aid when planning
suggested that the process of choosing a tourist travel continue to use brochures as a guide or
destination involves two phases of sequential reference throughout a trip.
decisions. The first phase is formed by two Once the tourist gathers enough information,
chronological decisions, the decision of whether the next step in the purchase decision process is
to go on holiday and length of stay. The second evaluation. This is the third stage, involving
Copyright © 2012 John Wiley & Sons, Ltd. Int. J. Tourism Res. (2012)
DOI: 10.1002/jtr
Understanding Tourist Decisions

evaluation of product or service options offered and Urban 1986, Zeithaml 1988, Dodds et al.,
by various companies and examination of 1991, Ostrom and Iacobucci 1995, Teas and
product attributes such as price and brand. Agarwal 2000, Gallarza and Gil, 2006).
(Gupta et al., 2004). Zhou (1997) evaluated Price is often a conflictive element because it
the effectiveness of destination brochures plays a double role as an extrinsic product
requested by travellers, suggesting that of quality cue and a product choice monetary
those who read the brochure, 50% visited the constraint (Erickson and Johansson, 1985). These
destination and most consulted the brochure price roles can be labelled as informational and
during the visit. Andereck (2005) found that allocation roles (Rao and Sattler, 2007). Although
tourists who have no knowledge about a tour- the two roles of price are clear and derive from
ist destination demonstrate greater interest in disparate theoretical conceptions, measure-
information contained in brochures than those ments are difficult and tend to confuse the
who visited the destination previously. two when measured directly. In tourism
This study focuses on the evaluation of alter- literature, Naylor and Frank (2001) suggested
natives, the third stage of the buying process. It that price bundling increases perceptions of
is important to recognize that there exist vari- value for first-time holidaymakers. A study by
ous market segments with differing tastes and Rewtrakunphaiboon and Oppewal (2008) exam-
preferences; experiments require working with ines whether package information influences
groups that are as homogeneous as possible so intentions to visit and choice of beach holiday
differences in responses demonstrate combina- destinations. On the basis of extant research,
tions of factors rather than socio-demographic the authors assumed that the importance of an
respondent characteristics (Haaijer and Wedel, attribute increases when products display
2007). In this case, the target market chosen according to the attribute (Rewtrakunphaiboon
was students in their final year of university and Oppewal, 2008, p.129). Accordingly, find-
who wished to take a graduation trip for 4 days ings suggest that presenting price as a package
during the Easter holiday. The decision about heading (i.e. presenting packages with only
destination and duration, 4 days prior to Easter price) increases intentions to visit and choice of
day, was static, and participants searched for beach holiday destinations among students.
accommodations in travel agency brochures Services such as travel products are perceived
and on the Internet. as riskier purchases than goods (Zeithaml, 1988).
Because services are more difficult to evaluate
PERCEIVED VALUE AND CONSUMER and riskier to purchase, consumers use diverse
DECISIONS processes and cues for evaluation. Searching
for information is one heuristic used to reduce
Choosing a hotel room or a trip from a brochure risks and help travellers make decisions. Travel-
depends on the perceived utility people expect lers who have little information about a destin-
from the room or trip and the price. The market- ation from internal sources (i.e. memory,
ing literature defines the relationship between friends and relatives) use external sources such
perceived utility and price as perceived value. as brochures and the Internet (Andereck, 2005).
Zeithaml (1988, p.14) proposed a definition: ‘per- For purchasing services, extrinsic attributes
ceived value is the consumer’s overall assessment become cues when information on intrinsic attri-
of the utility of a product based on perceptions of butes is unavailable (Zeithaml 1988). Several
what is received and what is given’. What is researchers develop and test models of percep-
received and what is given represent trade-offs tions of value with particular emphasis on a
between give and get components. Benefit buyer’s extrinsic cues such as price and brand
components include intrinsic and extrinsic attri- name as indicators of quality and value. Dodds
butes, perceived quality and other relevant et al. (1991) conducted a study to examine direct
high-level abstractions, and sacrifice components and indirect relationships between price, brand
include price and other non-monetary costs. name and store name, and perceived quality,
Managers and academicians recognize a strong product value and willingness to buy. They
influence of perceived value on consumer behav- argue that price alone is the most important
iour, making it a priority for researchers (Hauser cue of perceived quality; adding store and brand
Copyright © 2012 John Wiley & Sons, Ltd. Int. J. Tourism Res. (2012)
DOI: 10.1002/jtr
 R. Huertas-Garcia, M. Laguna García and C. Consolación

name information has moderate effects. If a price dominant criterion for evaluating profiles, esti-
changes from low to high, perceived value mated by aggregate multi-nomial logit models
increases and at a point decreases significantly, (Louviere et al., 2000; Street and Burgess, 2007).
impacting willingness to buy negatively. The Implicit models are compensatory and decom-
negative effect of price on value perceptions positional, enabling analyses of the importance
reduces if brand or store name is provided. Price of each product’s characteristics, starting with a
has a positive influence on perceived quality but customer’s stated preferences. In this study, we
negative effects on perceived value and willing- demonstrate the process of designing a choice-
ness to buy. based conjoint analysis and how it determines
Perceived value also depends on the frame- the value of extrinsic cues that, for a market seg-
work under which customers conduct evalua- ment, are used to choose a hotel room from a
tions (Hempel and Daniel, 1993), defined as a tourist brochure. We use one of the least com-
dynamic variable. Perceptions of value change mon models, ranking, based on a one-sixth frac-
if an assessment is made before or after purchase tion from a 28 full factorial design in eight two-
(Woodruff 1997) and whether it is assessed size blocks in which the second profile of each
before sale, at the moment of sale, at the moment block is a mirror image of the first.
of use or after use (Moliner et al., 2005). Some
authors consider other factors that alter
Choice-based conjoint analysis
perceived value. Lee et al. (2007) reported that
value depends on a consumer’s characteristics The process of evaluating the choice of accom-
and the type of product under consideration. modation from a brochure operates within
Value is a latent construct not observed directly discrete choice models (CMs). CM is a family
(Teichert and Shehu, 2007); most studies in this of survey-based methodologies for modelling
area use scales to estimate the latent component. preferences for goods in which goods are
We use stated preferences to estimate the value described by their attribute levels. Respondents
or utility generated by varying profiles in the are presented with alternative descriptions of
pre-purchase stage. products differentiated by attributes and levels
and rank the various alternatives, rating them
or choosing those they prefer (Hanley et al.,
METHODOLOGY 2001). The form of evaluation (e.g. lexicographic,
elimination by aspect, economic and attribute
Because perceived value and utility are unob- screening) that predominates among respon-
servable, they require scales or proxy variables dents in driving such selections remains elusive
to estimate them (Teichert and Shehu, 2007). (Scarpa and Rose, 2008).
According to Klein (1990), there are two meth- Choice model based on choosing the best
ods to identify and measure consumer needs option has several advantages over traditional
and desires: asking directly or deducing the ratings and rankings. Choice is often the behav-
motives from other kinds of data. The advan- iour of ultimate interest in a decision process
tages of indirect methods include establishing because it estimates behaviours accurately.
robust causal relationships between attributes Models estimated from choice allow direct pre-
and customer evaluations. A popular device, diction of choices, avoiding the need for conjoint
choice-based conjoint analysis, is an indirect simulators (Elrod et al., 1992). Its primary disad-
method used to obtain measurement of prefer- vantage is that it works poorly with small
ences (Haaijer and Wedel, 2007). During conjoint samples because the process of choice contains
analysis’ evolution over the last 40 years, there minimal information concerning consumer
was a shift in the types of responses used for preferences. A chosen option indicates which
analysis from ranking, to ratings, and finally to alternative is the most preferred, but it does not
choosing the best profiles (Elrod and Chrzan, provide information about other alternatives
2007). The most frequent criterion for assessing not selected. A choice process implicitly entails
a choice set was ratings of full profiles, evaluated consideration of multiple alternatives, but only
individually (Wittink and Cattin, 1989). Since one option is chosen; in a ranking process, all
the late 1980s, the choice prevailed as the alternatives are considered (Elrod and Chrzan,
Copyright © 2012 John Wiley & Sons, Ltd. Int. J. Tourism Res. (2012)
DOI: 10.1002/jtr
Understanding Tourist Decisions

2007). We propose a ranking model adjusted and tendency to rent a room. The theory of
by a proportional odds ordinal logistic model reasoned action posits behaviour intentions
(Train, 2009). The decision to use this model influences behaviours (Ajzen and Madden,
was based on two objectives. Estimations made 1986). Social psychology research suggests
by using all ranking information are more effi- intentions are the best predictor of behaviours
cient, and efficiency is greatest when a full pro- because they allow individuals to incorporate
file design is used. This improvement in all relevant factors independently that influ-
efficiency allows more acceptable results with ence behaviours (Fishbein and Ajzen, 1975).
reduced samples. Ranking is consistent with Several studies examine the relationship
consumer behaviour in both economy theory between purchase intentions and purchase
and with respect to the nature of preferences behaviours for non-durable goods (Dowling
representing an ordered relationship (Frank, and Staelin, 1994). The degree to which people
2009). A typical choice-based conjoint analysis express preferences is a reasonable predictor of
is characterized by several stages: purchase behaviours.

• Identification of relevant attributes of goods Qualitative research. Investigating the first two
to be valued. The use of literature reviews, stages of the experimental process of conjoint
consultation with experts and qualitative selection, we began with qualitative research
techniques such as focus groups establishes to identify the attributes considered relevant
these attributes. in the perceived value of choosing a hotel
• Assignment of a feasible, realistic range of displayed in a travel agent’s brochure. We evalu-
plausible levels for attributes. Qualitative ated aggregate utility generated by the sample of
research or literature searches establish ap- students from varying combinations of attri-
propriate levels. butes offered by advertisements in a brochure.
• Choose an experimental design for gathering The fieldwork for the qualitative research was
data. The experimental design is based on carried out in Segovia, Spain from November
statistical theory to combine levels of the attri- 2006 to February 2007. The use of university stu-
butes into alternative profiles presented to dent samples for experimental research attracted
respondents. Whereas full factorial designs the attention of many researchers recently
often generate an impractically large number (Gallarza and Gil, 2006; Rewtrakunphaiboon
of combinations for evaluation, fractional and Oppewal, 2008). We used a focus group
factorial designs and block designs reduce with three smaller groups of eight students each
the number of combinations but with a con- based on a chosen destination of Punta Cana,
comitant loss of power. Santo Domingo. The sessions were conducted
• Profiles identified by the experimental design with two travel agency brochures, asking open-
are grouped into choice sets and presented to ended questions and recording answers. After
respondents. Profiles can be presented to the sessions, we coded and classified answers
respondents individually, in respondent pairs to reveal relevant attributes. The data were
or in groups. coded using inductive category coding, consist-
• Measurement of preferences. Individual pre- ing of labelling factors repeatedly found in text
ferences are discovered in choice-modelling (Spiggle 1994). Similar processes are used in
surveys by asking respondents to rank content analysis in the service literature (Tax
options, score them or choose those they et al., 1998), and the method is used extensively
prefer most. These methods of measuring in studies of consumer behaviour to identify
preferences correspond with variations of relationships within text (Spiggle, 1994).
modelling approaches. The attributes gathered in the focus groups
• Estimation procedure. The model is adjusted were as follows: the image and aesthetics of
using Ordinary Least Square (OLS) regression the illustrations in the advertisement were an
or maximum likelihood estimation. important factor; aerial or panoramic images
were preferred, although well-lit, nocturnal
In this study, preference differences between photos were also viewed favourably; the size
hotel listings indicated differences in value of the illustrations and the hotel’s starred rating
Copyright © 2012 John Wiley & Sons, Ltd. Int. J. Tourism Res. (2012)
DOI: 10.1002/jtr
 R. Huertas-Garcia, M. Laguna García and C. Consolación

were determining factors; five-star establish- reported that the hotel listing’s position was
ments displayed in large photographs were irrelevant, we incorporated it anyway given
preferred; the structure of the buildings was the influence product placement has on prefer-
very important; low, open buildings were ences (Cox, 1970; Chevalier, 1975). To deter-
preferred to high, square buildings, although mine high and low prices, we calculated the
there was no consensus concerning the size of average ranks of high and low prices from a
the hotel; accessibility and proximity to the sample of Santo Domingo hotels published in
beach and restaurants were important; the three brochures.
starred rating was indicative of quality, but no
importance was given to the ICTE’s Q (Spanish
tourist quality symbol); subjects preferred the Quantitative research. For quantitative research,
price to be on the same page as the advertise- we propose a choice-based conjoint experi-
ment that described the hotel, although price ment of a sample of undergraduate students
range was not one of the most important who travel in groups. The underlying choice
attributes. A wide range of hotel services was surveys were the statistical design of the
preferred but not essential. No remarks were experiment, used to allocate all combinations of
made or attention paid to room amenities (e.g. factors and their levels to form a set of alterna-
air conditioning) or a hotel’s surroundings (e.g. tives to be used on the survey. An experimental
topography). design is the systematic arrangement of profiles
Not all attributes must or can be considered; in matrices of coded values researchers use to
they must be chosen realistically and in a way describe attribute levels representing hypothet-
that is appropriate to the situation. Gustafsson ical alternatives of marketing options in the
et al. (1999) suggested a list of rules for choosing choice set. It is not an easy task; Kuhfeld et al.
attributes when evaluating service quality: (1994) pointed out that the best designs are
choose attributes that are important when the discovered when the researcher uses both
interviewees are buying, ones that can be modi- human design skills and computerized searches.
fied and ones used to compare with competitors. To design the experiment, we considered eight
Because there is no consensus in the literature factors: three quantitative variables (starred
about the elements encompassing each of the ratings, advertisement size and prices) and five
positive and negative dimensions of value qualitative variables (names, descriptions of
(Woodruff, 1997), eight relevant attributes were hotel size, brochure illustrations, positions in
chosen for evaluation: the pamphlet and positions on the page). All
variables were coded as vectors (
1, 1); for
• Name (one short name ‘Majestic Punta Cana’ factor 1, name, a positive sign indicated the long
and one long name ‘Majestic Colonial Punta name ‘Majestic Colonial Punta Cana Beach
Cana Beach Resort Golf Casino & Spa’); Resort Golf Casino & Spa’, and a negative sign
• Hotel’s starred rating (four or five stars); indicated the short name ‘Majestic Punta Cana’
• Phrase used to describe the hotel size (24 and so on. Table 1 shows the factors and their
buildings with three floors, suites and junior codes. A full factorial design would have
suites; 659 suites and junior suites); required a 28 factorial experiment (i.e. 256
• Picture in the brochure (a general view of the profiles in the choice set). To avoid saturation
tourist building or a view of the swimming of information to respondents, we used a IV
pool); resolution design, 28
4
IV , a fraction one-sixth of
• Size of the hotel listing (quarter or half page); the 28 full design. Resolution levels determine
• Position in the brochure (right or left page); the degree of confusion generated by the experi-
• Position on the page (top or bottom); ment among factors and interactions between
• Prices (a low price of €277 or a high price them. A IV resolution design avoids confusion
of €499) between main effects and second-order interac-
tions, but two-factor interactions are mutually
The first five variables were gathered from confused. A complete overview of the design
the qualitative research, and the last three were of the experiment can be found in Box et al.
added by the researchers. Although subjects (2005) and Myers and Montgomery (2002).
Copyright © 2012 John Wiley & Sons, Ltd. Int. J. Tourism Res. (2012)
DOI: 10.1002/jtr
Understanding Tourist Decisions

Table 1. Proposed variables and their codification

Variable
1 +1
A Name Majestic Punta Cana Majestic Colonial Punta Cana
Beach Resort Golf Casino & Spa
B Hotel rating 4 5
(number of starts)
C Hotel size 24 buildings of three floors 659 suites and junior suites
with suites and junior suites
D Picture of advertisement View of building View of swimming pool
E Hotel listing size 1/4 of page 1/2 of page
F Right or left location Left Right
G Position on page On the bottom On the top
H Price 3 days, full-board, service €277 €499
and airport; transfers included

The experimental design required a design In many cases, it is inappropriate to propose

generator. A 2k
p fractional factorial design a choice set of 16 profiles either because of the
necessitates the selection of p independent load of work involved in assessing the rank of
design generators. A reasonable criterion to 16 profiles or because of the difficulties with
select generators is the one with the highest pos- practical application. It is possible to divide
sible resolution. Myers and Montgomery (2002) the profile set proposed by the researcher into
presented a selection of tables of 2k
p fractional smaller blocks. Confusion is a design technique
factorial designs with the highest resolution for for arranging a full or fractional factorial into
factors below 11 (k ≤ 11). For an experiment of blocks; the researcher multiplies two or more
IV , the generators are E = BCD, F = ACD,
28
4 factors (A  B = AB) as a criterion to separate
G = ABC and H = ABD. The complete design the design into smaller blocks. When the model
is shown in Table 2. is estimated, the researcher does not know

Table 2. Fractional factorial design

Number Hotel Picture Advertisement Right or left Position on Price
Variables Name of stars size advertisement size location page 3 days
Experiment A B C D E = BCD F = ACD G = ABC H = ABD
1
1
1
1
1
1
1
1
1
2 1
1
1
1
1 1 1 1
3
1 1
1
1 1
1 1 1
4 1 1
1
1 1 1
1
1
5
1
1 1
1 1 1 1
1
6 1
1 1
1 1
1
1 1
7
1 1 1
1
1 1
1 1
8 1 1 1
1
1
1 1
1
9
1
1
1 1 1 1
1 1
10 1
1
1 1 1
1 1
1
11
1 1
1 1
1 1 1
1
12 1 1
1 1
1
1
1 1
13
1
1 1 1
1
1 1 1
14 1
1 1 1
1 1
1
1
15
1 1 1 1 1
1
1
1
16 1 1 1 1 1 1 1 1

Copyright © 2012 John Wiley & Sons, Ltd. Int. J. Tourism Res. (2012)
DOI: 10.1002/jtr
 R. Huertas-Garcia, M. Laguna García and C. Consolación

whether values correspond to the combined maximum-a standard deviation of €112.6, an

effects of two factors or to individual compo- average trim of €349.6 and a median of €304.
nents. Another design that produces econom- Regarding sample size, we used the resource
ical and more precise designs for main factors equation proposed by Mead (1988). In this
is the two-size block arrangement. We used a equation, we required at least n runs in blocks
design called mirroring where the second of size nb:
profile is arranged with all factors with the sign
changed. This design was chosen because it n qðq þ 3Þ
n¼ þ þ nlof þ npe (1)
facilitates practical application of the experi- nb 2
ment. For building this arrangement, Box et al.
(2005) proposed a three-block generator based where q is the number of variables, nlof is a
on two-factor interactions. For example, small number of degrees of freedom (typically
B1 = AB, B2 = AC and B3 = AD. Table 3 shows 5 to 10) for estimating higher-order terms and
the reassignment of the 16 profiles into blocks to check for lack of fit and npe is a small number
of two. of degrees of freedom (typically 5 to 15) for
estimating pure error. In our case, nb = 2, q = 8,
taken as nlof + npe = 15, the same value used by
Field work
Gilmour and Trinca (2006). The resource equa-
Once the eight-block experiment was config- tion suggested a minimum sample of 118
ured, it was shown to a number of potential profiles. However, instead of the minimum,
clients to evaluate each experiment by using a we used eight blocks in which two profiles
ranking scale. The profiles were constructed were replicated eight times (i.e. 128 profiles).
following the design of the hotel listing shown Each block of the two profiles was shown to
in the tourist brochures, combining images and eight students. A sample of 64 students from
texts (Vriens et al., 1998). The scenario that Valladolid University (Spain) who were choos-
frames the experiment was the choice of a hotel ing a destination for their end-of-studies trip
in Santo Domingo for a graduation trip. An participated in the experiment by answering
example scenario with seven profiles is shown questionnaires during class time. Fieldwork
in Appendix A. was conducted between February and March
Some authors such as Hempel and Daniel 2007. Descriptive statistics show that 58.5%
(1993) criticize laboratory experiments employed were female and 41.5% were male. More than
for drawing inferences from experimental situa- 38.4% had previous international travel experi-
tions in which individual choices are manipulated ence. Respondents reported an average age of
by frames imposed by a researcher. Subjects are 22.3 years, and the destinations preferred were
often charged with solving problems in a labora- the Caribbean (76.2%), the Canary Islands
tory in which respondents intensify sensitivity to (13.3%) and European cultural cities (9.5%).
a researcher’s instructions, making biassed deci- The random utility model provides the eco-
sions in comparison with the decisions subjects nomic theory framework for analysing the data
make on their own in a real environment. To from a ranking exercise. This model is derived
reduce the risks posed by a laboratory, we used under the assumption of utility-maximizing
familiarity formats similar to a travel agent’s behaviour by the decision maker. Under the
brochure. Interviews were conducted during the assumption of independent and identically
pre-purchase process when students were organ- distributed random error with an extreme value
izing trips. We used a mix of profiles, some distribution, Beggs et al. (1981) developed a
affected and others placebo. To administer the rank-order logit model capable of using all
questionnaire within a frame of reference that information contained in a survey where alter-
facilitates interviewee decisions, each question- natives are fully ranked by respondents. Speci-
naire contained seven hotel listings: two experi- fication is based on repeated application of the
mental and five non-exposed advertisements. probability expression given in Equation (2)
The dimensions of the frame of reference were until a full ranking of all the alternatives is
seven prices with a range of 285 Euros-235 Euros obtained. This model involves considering
was the minimum price and 520 Euros the each of the choices made by the respondents
Copyright © 2012 John Wiley & Sons, Ltd. Int. J. Tourism Res. (2012)
DOI: 10.1002/jtr
 Table 3. Blocked design
Number of Picture Advertisement Right or Position
Understanding Tourist Decisions

Variables Name stars Hotel size advertisement size left location on page Price 3 days Block variables

Copyright © 2012 John Wiley & Sons, Ltd.

Block Experiment A B C D E = BCD F = ACD G = ABC H = ABD B1 = AB B2 = AC B3 = AD
1 2 1
1
1
1
1 1 1 1
1
1
1
15
1 1 1 1 1
1
1
1
1
1
1
2 13
1
1 1 1
1
1 1 1 1
1
1
4 1 1
1
1 1 1
1
1 1
1
1
3 6 1
1 1
1 1
1
1 1
1 1
1
11
1 1
1 1
1 1 1
1
1 1
1
4 9
1
1
1 1 1 1
1 1 1 1
1
8 1 1 1
1
1
1 1
1 1 1
1
5 7
1 1 1
1
1 1
1 1
1
1 1
10 1
1
1 1 1
1 1
1
1
1 1
6 12 1 1
1 1
1
1
1 1 1
1 1
5
1
1 1
1 1 1 1
1 1
1 1
7 3
1 1
1
1 1
1 1 1
1 1 1
14 1
1 1 1
1 1
1
1
1 1 1
8 16 1 1 1 1 1 1 1 1 1 1 1
1
1
1
1
1
1
1
1
1 1 1 1

Int. J. Tourism Res. (2012)

DOI: 10.1002/jtr
 R. Huertas-Garcia, M. Laguna García and C. Consolación

as independent. Each of the choices is called a odds model (or cumulative logit model) by
pseudo-observation because each complete using the PLUM process from SPSS.
ranking is only one observation formed by
multiple pseudo-observations depending on RESULTS AND DISCUSSION
the number of choice sets (Train, 2009). The
model considers that the first option is chosen The results of estimation are shown in Table 4.
among seven profiles; afterwards, the decision The ordinal logistic regression model has a de-
maker chooses the second option among the gree of adjustment usual in this type of study
remains six profiles and so on. The probability (Cox and Snell’s pseudo R2 = 0.220;
2 log-likeli-
of any ranking of an alternative made by indi- hood = 160.768, p < 0.001) (Louviere et al., 2000).
vidual i can be expressed as a sequence of To determine whether the model offers adequate
probability multiplications: predictions, we used two log-likelihood values,
comparing two log-likelihood values for the
2 3 intercept-only model and the final model with
J 6 7 the predictors. The chi-square statistic indicates
Y 6 expðVijÞ 7
PiðUi1 > Ui2 > ⋯ > UijÞ ¼ 6 7 whether the model offers significant improve-
6PJ 7
j¼1 4 5 ment over the baseline intercept-only model.
expðVikÞ
k¼j The dependent variable had only five of the
seven possible values, meaning the two exposed
(2) hotel listings were chosen between the first and
Pq fifth orders. Hence, the ordinal model estimates
where Vij ¼ b0 þ i¼1 bi xi þ e , Pi (Uij) only four cut-off values that separate the five
measures the probability of this chosen order, response categories. Parameter estimates are
bi are the values of the slope of the vector for shown in Table 4, where four of the 11 effects
each main factor and e is an error term. An are significant. These are B (starred ratings),
ordinal logistic regression model was used to H (size of hotel listing), E (prices) and B2 (the
adjust the full profile data in the proportional blocking variable). Whereas a variable in

Table 4. Estimated effects from the experimental design 28
4

IV

Effect Estimation Wald Significance

Order = 1
1.004 3.562 0.059
Order = 2 0.106 0.041 0.839
Order = 3 1.140 4.632 0.031
Order = 4 2.211 15.842 0.000
A 0.175 0.333 0.564
B 0.917 8.842 0.003
C
0.456 2.254 0.133
D 0.038 0.015 0.901
E = BCD 0.924 8.950 0.003
F = ACD 0.363 1.431 0.232
G = ABC
0.324 1.143 0.285
H = ABD
0.661 4.682 0.030
B1 = AB = CG = DH = EF
0.476 2.453 0.117
B2 = AC = BG = DF = EH 0.803 6.826 0.009
B3 = AD = BH = CF = EG 0.174 0.332 0.565
Pseudo R-square
Cox and Snell 0.220
Nagelkerke 0.225
McFadden 0.066
Diagnostic
2 log- 160.768 0.001
likelihood

Copyright © 2012 John Wiley & Sons, Ltd. Int. J. Tourism Res. (2012)
DOI: 10.1002/jtr
Understanding Tourist Decisions

statistical design experiment can be inter- effects from the derivative model. Values are
preted singly if there is no evidence that it is similar to the previous analysis.
confused with another, B2 (the blocking vari- H (prices) had a negative impact on evalu-
able) is confused with several second-order ation by the sample. This result is in accord
interactions, B2 = AC = VG = DF = HE; inter- with the interpretation of price as sacrifice; the
pretation is, therefore, imprecise. We suspect lower the price, the greater the customer’s
that the value of B2 responds significantly to evaluation. Other more highly valued effects
the interaction between H (size of hotel listing) were as follows: B (starred ratings), E (hotel
and E (prices). listing size) and HE (interaction between price
To verify these results, we considered the and size of hotel listing). B (starred ratings)
analysis of only three factors, those which and E (advertisement size) had a positive influ-
demonstrated significance after the first experi- ence on the subjects’ evaluation. Respondents
ment of the three main effects (B, H and E). used both factors as external value cues; the
Because only three factors were significant, a higher the starred ratings and the bigger the size
full factorial of 23 required eight profiles drawn of the advertisement, the greater its evaluation.
from the previous design using 16 profiles. This Moreover, the joint presence of HE (interaction
is possible because any fractional factorial between price and size of hotel listing) increased
design of resolution R may contain complete positive valuation. The dual role of price as
factorials in any sub-group of R
1 variables. sacrifice and extrinsic cue of service quality is
Because the design is resolution R IV, it was appreciated when price interacts with a large
possible to configure a complete factorial advertisement. Thus, a negative evaluation repre-
design with these three variables. This design sented by high price accommodation is offset by
allows estimation of main effects and interac- a large advertisement. Likewise, the positive
tions without any confusion patterns and, conse- influence generated by a large advertisement
quently, allowed direct interpretation. Ordinal increases with price.
logistic regression was used to evaluate
significance of the effects. Table 5 shows the
model-fitting information and the estimated CONCLUSIONS

Customer evaluation and decision making are

Table 5. Estimated effects from the experimental primary targets in all areas of business, particu-
design 23
larly in tourism management. To understand
Effect Estimation Wald Significance customer needs and determine the primary
Order = 1
0.850 3.771 0.052 factors of perceived value, a researcher can ask
Order = 2 0.247 0.331 0.565 subjects directly or deduce desires through
Order = 3 1.249 7.962 0.005 indirect methods such as experimentation. In
Order = 4 2.281 23.423 0.000 this study, we demonstrate the process of imple-
B 0.907 8.756 0.003 menting a choice-based conjoint analysis and
E 0.859 7.894 0.005 how this analysis can be used as a powerful tool
H
0.663 4.760 0.029 for evaluating preferences. These designs are
BH 0.149 0.246 0.620 particularly suitable for subjects related to
BE
0.110 0.133 0.716 demand for introduction of new products, line
HE 0.827 7.314 0.007 extensions (Louviere et al., 2000) or design of a
BHE
0.298 0.979 0.322
new hotel listings in a tourist brochure. In
Pseudo R-square addition to providing an evaluation of the
Cox and Snell 0.170 weight of main attributes, it also shows the
Nagelkerke 0.177 weight of interactions of several attributes.
McFadden 0.058 Value perceptions depend on a frame of
Diagnostic
2 100.845 0.000 reference in which consumers make evalua-
log- tions. Employing an experimental design in a
likelihood context of pre-purchase with a static frame of
reference, we examined the main factors of
Copyright © 2012 John Wiley & Sons, Ltd. Int. J. Tourism Res. (2012)
DOI: 10.1002/jtr
 R. Huertas-Garcia, M. Laguna García and C. Consolación

value perception in the process of choosing a before purchasing is extremely useful for busi-
hotel from a travel agent brochure. Subjects nesses and managers. A travel agent brochure
used a reduced number of attributes to infer may be giving more prominence to elements that
perceived quality and the sacrifice necessary do not determine perceived value of a tourist,
to acquire it. In this case, only three of the eight making communications ineffective and not
attributes were significant: advertisement size contributing to a consumer’s decision. However,
(quarter or half page) and the hotel’s starred the validity of the attributes of perceived value
rating (4 or 5) as quality attributes, and price disappears with purchase. Fisher et al. (1994)
indicating sacrifice. For this market segment argued that when consumers evaluate perceived
of students in their final year of university, bro- service quality after purchase, they rarely
chure illustration was not significant in spite of mention the criteria used for evaluation before
the results from qualitative research, neither purchase; when they do, such criteria are relegated
were position on page, nor name of to a lower hierarchic level to the one occupied
establishment, nor left versus right location, before purchase.
although results are in line with the qualitative This study has some limitations. Participants
research. Hotel size occupies an intermediate were from one university, disallowing statis-
position of importance; an increase in sample tical inferences from the sample to a popula-
size is necessary to verify the variable’s influ- tion. Some variables such as room quality and
ence on selection. hotel location were ignored as potentially rele-
Attributable to the second analysis using a vant attributes. The laboratory environment
full factorial design, it was possible to detect that afforded internal validity because of envir-
the second inductive factor of perceived quality onment control compromised external validity
consisting of the joint presence of two vari- because of an artificial venue. The sample size
ables, advertisement size (quarter or half page) was also a limiting factor because it did not
and price. The two joint variables demonstrate allow us to determine the significance of some
the dual role of price (Dodds et al., 1991), as a primary variables (e.g. hotel size), although the
sacrifice people must make to buy goods or as model using ordinal logistic regression was
an extrinsic cue of service quality, when it inter- proposed because of its suitability to fit small
acts with a large advertisement. Interaction samples. Designing this experiment, we
between price and hotel listing size showed a transformed a full fractional model into a
positive result when negative was expected, fractional one; the resulting design was orga-
suggesting some kind of compensatory effect. nized into blocks of two. This design only
These findings have important implications for allows the estimation of main factors without
managers, specifically that a higher price interactions. The design could be improved
expected to generate negative utility is offset by by taking advantage of multiple runs and
a larger advertisement to produce a positive using a fractional factorial with higher reso-
effect on a tourist’s final decision. The increased lution such as R V, although there is the
cost of a larger advertisement could be recouped disadvantage of the cost of designing 32 bro-
by charging a higher price for the trip and chures instead of eight. The model helped us
hotel room. get closer to customers, identifying opinions
This study considers perceived value attributes and preferences; it was a simple procedure that
in a pre-purchase stage, a stage traditionally facilitated sequential research that enabled us
understudied. Knowing which aspects or attri- to continuously improve knowledge of the
butes determine perceived value of the product subject.

Copyright © 2012 John Wiley & Sons, Ltd. Int. J. Tourism Res. (2012)
DOI: 10.1002/jtr
Understanding Tourist Decisions

Appendix A
Example of experimental travel agent brochure

Copyright © 2012 John Wiley & Sons, Ltd. Int. J. Tourism Res. (2012)
DOI: 10.1002/jtr
 R. Huertas-Garcia, M. Laguna García and C. Consolación

REFERENCES Fisher S, Clemons D, Scott, R, Woodruff D, Schumann

W, Burns MJ.1994. Comparing consumers’ recall of
Ajzen I, Madden TJ. 1986. Prediction of goal-directed prepurchase and postpurchase product evaluation
behavior: attitudes, intentions, and perceived experiences. Journal of Consumer Research 20(2):
behavioral control. Journal of Experimental Social 548–560.
Psychology 22(5): 453–474. Fodness D, Murray B. 1999. A model of tourist infor-
Andereck KL. 2005. Evaluation of a tourist brochure. mation search behaviour. Journal of Travel Research
Journal of Travel & Tourism Marketing 18(2): 1–13. 37(3): 220–230.
Beggs S, Cardell S, Hausman J. 1981. Assessing the Frank RH. 2009. Microeconomics and behavior. Mc
potential demand for electric cars. Journal of Graw-Hill: New York
Econometrics 17(1): 1–19. Gallarza MG, Gil I. 2006. Value dimensions,
Bieger T, Laesser C, Gallen S. 2000. Segmenting perceived value, satisfaction and loyalty: an inves-
travel situations on the basis of motivation and tigation of university students’ travel behaviour.
information-collection by the traveller. Revue de Tourism Management 27(3): 437–452.
Tourisme 2: 54–64. Gilmour SG, Trinca LA. 2006. Response surface
Box EP, Hunter JS, Hunter WG. 2005. Statistics for experiments on processes with high variation. In Re-
experimenters: design, innovation, and discovery. sponse Surface Methodology and Related Topics, Khuri
2ndedn . Wiley: Hoboken. AI (ed.). World Scientific Publishing Co: New York;
Chevalier M. 1975. Increase in sales due to in-store 19–46.
display. Journal of Marketing Research 12(4): 426–431. Gupta A, Bo-Chiuan S, Walter Z. 2004: An empirical
Chiou WB, Wan CS, Lee HY. 2008. Virtual experi- study of consumer switching from traditional to
ence vs brochures in the advertisement of scenic electronic channels: a purchase-decision process
spots: How cognitive preferences and order perspective. International Journal of Electronic
effects influence advertising effects on consumers. Commerce 8(3): 131–161.
Tourism Management 29(1): 146–150. Gustafsson A, Ekdahl F, Bergman B. 1999. Conjoint
Cox KK. 1970. The effect of shelf space upon sales analysis: a useful tool in the design process. Total
of branded products. Journal of Marketing Research Quality Management 10(3): 327–343.
7(1): 55–58. Haaijer R, Wedel M. 2007. Conjoint choice experi-
ments: general characteristics and alternative
Dellaert B, Ettema D, Lindh C. 1998. Multi-faceted
model specifications. In Conjoint Measurement.
tourist travel decisions. a constraint-based concep-
Methods and Applications, 4thedn , Gustafsson A,
tual framework to describe tourist’s sequential
Herrmann A, Huber F (eds). Springer: Berlin;
choices of travel components. Tourism Management
199–229.
19(4): 313–320.
Hanley N, Mourato S, Wright RE. 2001. Choice
Dodds WB, Monroe, KB, Grewal D. 1991. Effects of modelling approaches: a superior alternative for
price, brand and store information on buyers’ environmental valuation?. Journal of Economic
product evaluations. Journal of Marketing Research Surveys 15(3): 435–462
28(4): 307–319.
Hauser JA, Urban GL. 1986. The value priority
Dowling G, Staelin R. 1994. A model of perceived hypotheses for consumer budget plans Journal of
risk and intended risk-handling activity. Journal Consumer Research 12(4): 446–462.
of Consumer Research 21(1): 119–134. Hempel DJ, Daniel HZ. 1993. Framing dynamics:
Elrod T, Chrzan K. 2007: The value of extent-of- measurement issues and perspectives. Advances
preference information in choice-based conjoint in Consumer Research 20: 273–279.
analysis. In Conjoint Measurement. Methods and Hodgson P. 1993. Tour operator brochure design
Applications, 4thedn , Gustafsson A, Herrmann research revisited. Journal of Travel Research 32(1):
A, Huber F (eds). Springer: Berlin; 133–144. 50–52.
Elrod T, Louviere JJ, Davey KS. 1992. An empirical Hsieh S, O’Leary L. 1993. Communication channels
comparison of ratins-based and choice-based to segment pleasure travelers. Journal of Travel &
conjoint models. Journal of Marketing Research 29(3): Tourism Marketing 2(2/3): 57–76.
368–377 Klein R. 1990. New techniques for listening to the
Erickson, GM, Johansson, JK. 1985. The role of price voice of the customer. Applied Marketing Science,
in multi-attribute product evaluations. Journal of Inc., Second Symposium on Quality Function Deploy-
Consumer Research 17: 195–199. ment, June 18–19, 1990. Cited by Cohen L. 1995.
Fishbein M, Ajzen I. 1975. Belief, attitude, intention Quality Function Deployment: How to Make QFD Work
and behavior: an introduction to theory and research. for You. Addison-Wesley Publishing Company,
Addison-Wesley: New York. Reading, Massachusetts, USA.
Copyright © 2012 John Wiley & Sons, Ltd. Int. J. Tourism Res. (2012)
DOI: 10.1002/jtr
Understanding Tourist Decisions

Kotler P. 2000. Marketing Management. The Millen- Roehl W, Fesenmaier D. 1992. Risk perceptions and
nium Edition. Prentice Hall Inc: New Jersey. pleasure travel: an exploratory analysis. Journal of
Kuhfeld WF, Tobias RD, Garratt M. 1994. Efficient Travel Research 30(4): 17–26.
experimental design with marketing applications. Scarpa R, Rose JM. 2008. Design efficiency for non-
Journal of Marketing Research 31(4): 545–557 market valuation with choice modeling: how to
Lee C-K, Yoon Y-S, Lee S-K. 2007. Investigating the measure it, what to report and why. The Australian
relationships among perceived value, satisfaction Journal of Agricultural and Resource Economics 52(3):
and recommendations: the case of Korean DMZ. 253–282.
Tourism Management 28(1): 204–214. Spiggle S. 1994. Analysis and interpretation of qualita-
Louviere JJ, Hensher, DA, Swait JD. 2000. Stated Choice tive data in consumer research. Journal of Consumer
Methods. Cambridge University Press: Cambridge. Research 21(3): 491–503.
Mead R. 1988. The design of experiments. Cambridge:
Street D, Burgess L. 2007. The Construction of Optimal
Cambridge University Press; cited by Gilmour SG,
Stated Choice Experiments: Theory and Methods.
Trinca L A. 2006. Response surface experiments on
Wiley: Hoboken, NJ.
processes with high variation. In Response Surface
Tax S, Brown SW, Chandrashekaran M. 1998.
Methodology and Related Topics, Khuri AI (ed.). World
Customer evaluation of service complaint experi-
Scientific Publishing Co: New York.
ences: implications for relationship marketing.
Molina A, Esteban A. 2006. Tourism brochures:
Journal of Marketing 62(April): 60–76.
usefulness and image. Annals of Tourism Research
33(4): 1036–1056. Teas RK, Agarwal, S. 2000. The effects of extrinsic
Moliner MA, Rodríguez RM, Callarisa Ll J, Sánchez product cues on consumer’s perceptions of
J. 2005. Dimensionalidad del valor percibido quality, sacrifice and value. Journal of the Academy
global de una compra. Revista Española de Investiga- of Marketing Science 28(2): 278–290.
ción de Marketing 9(2): 135–157. Teichert T, Shehu, E. 2007. Evolutionary conjoint. In
Munger JL, Grewal D. 2001. Effects of alternative Conjoint Measurement. Methods and Applications,
price promotional methods on consumers’ evalu- 4thedn , Gustafsson A, Herrmann A, Huber F
ation and purchase intentions. Journal of Product (eds). Springer: Berlin; 295–320.
& Bran Management 10(3): 185–197. Train K. 2009. Discrete Choice Methods with
Myers RH, Montgomery DC. 2002. Response Surface Simulation. 2ndedn , Cambridge University Press:
Methodology. John Wiley & Sons, Inc: New Jersey. Cambridge.
Naylor G, Frank, KE. 2001. The effect of price bundling Vriens M, Loosschilder, G, Rosbergen E, Wittink D.
on consumer perceptions of value. Journal of Services 1998. Verbal versus realistic pictorial representa-
Marketing 15(4): 270 – 281. tions in conjoint analysis with design attributes.
Nicolau JL, Mas FJ. 2006. Elección de la duración Journal of Product Innovation Management 15(5):
vacacional: una aproximación con modelos de 455–467.
recuento. Revista Europea de Dirección y Economía Wicks BE, Schuett MA. 1991. Using travel brochures
de la Empresa 15(2): 99–116. to target frequent travelers and “big-spenders”.
Ortega E, Rodriguez B. 2007. Information at tourism Journal of Travel & Tourism Marketing 2(3): 77–90.
destinations. Importance and cross-cultural differ- Wittink DR, Cattin P. 1989. Commercial use of
ences between international and domestic tourist. conjoint analysis: an update. Journal of Marketing
Journal of Business Research 60(2): 146–152. 53(3): 91–96.
Ostrom A, Iacobucci D. 1995. Consumer trade-offs
and the evaluation of services. Journal of Marketing Woodruff BR. 1997. Customer value: the next source
59(1): 17–28. for competitive advantage. Journal of the Academy
Rao, VR, Sattler, H. 2007. Measurement of price of Marketing Science 25(2): 139–153.
effects with conjoint analysis: separating informa- Yamamoto D, Gill AM. 1999. Emerging trends in
tional and allocative effects of price. In Conjoint Japanese package tourism. Journal of Travel Research
Measurement. Methods and Applications, 4thedn , 38(2): 134–143.
Gustafsson A, Herrmann A, Huber F (eds). Zeithaml VA. 1988. Consumer perceptions of price,
Springer: Berlin; 31–46. quality and value: a means-end model and synthe-
Rewtrakunphaiboon W, Oppewal H. 2008. Effects of sis of evidence. Journal of Marketing 52(3): 2–22.
package holiday information presentation on Zhou Z. 1997. Destination marketing: measuring the
destination choice. Journal of Travel Research 47(2): effectiveness of brochures. Journal of Travel & Tourism
127–136. Marketing 6(3): 143–158.

Copyright © 2012 John Wiley & Sons, Ltd. Int. J. Tourism Res. (2012)
DOI: 10.1002/jtr