Research article

Journal of Vacation Marketing

1–14
Luxurious or economical? ª The Author(s) 2018
Reprints and permission:

An identification of tourists’ sagepub.co.uk/journalsPermissions.nav

DOI: 10.1177/1356766718757789
journals.sagepub.com/home/jvm
preferred hotel attributes using
best–worst scaling (BWS)

Bona Kim
Mount Saint Vincent University, Canada

Seongseop (Sam) Kim and Brian King

The Hong Kong Polytechnic University, Hong Kong

Cindy Yoonjoung Heo

Ecole hôtelière de Lausanne, HES-SO/University of Applied Sciences Western Switzerland, Switzerland

Abstract
This article explores consumer tendencies to opt for luxury or economy hotels by identifying their
most and least important selection attributes. The researchers investigate how sociodemographic
and behavioral characteristics influence traveler assessments of hotel attributes. In explaining con-
sumer hotel selection preferences, the researchers used an unconditional method—best–worst
scaling (BWS). Based on an analysis of responses from 397 luxury hotel customers and 351 economy
hotel customers in the United States, it was found that the two groups perceive hotel attributes
differently. Differentials were also identified on the basis of gender, income, and frequency of
purchase. While acknowledging that the task is complex, there is an urgent need to identify the
factors influencing hotel selection, because hoteliers need to attract new markets and also balance
this by retaining existing patrons. The findings extend existing literature by applying BWS to
the identification of hotel selection attributes.

Keywords
behavioral and sociodemographic characteristics, best–worst scaling, hotel selection attributes, hotel type

Introduction and the prioritization of future product develop-

ments. In the increasingly dynamic and compet-
Researchers have shown that product and service
itive hospitality environment, researchers should
features are the key influences on tourists’ hotel
acquire deep insights into how customers prior-
selection decisions (Pan et al., 2013). Hotel users
itize hotel attributes.
tend to pursue what may be described as opti-
According to random utility theory, the
mum utility, reflective of their preferences and
ranking of alternatives can be estimated by
values (Jannach et al., 2012; Lockyer, 2005).
determining an independent value scoring for
Tourism scholars have undertaken extensive
research on hotel selection attributes, often
focusing on the marketing and hotel customers’
behavior aspects (e.g. Li et al., 2013; Sohrabi Corresponding author:
Bona Kim, Business Administration and Tourism and
et al., 2012). From a practitioner perspective, Hospitality Management, Mount Saint Vincent University,
understanding hotel selection preferences can 166 Bedford Hwy, Halifax, NS, Canada.
inform the formulation of positioning strategies Email: [email protected]
2 Journal of Vacation Marketing XX(X)

each alternative and then using the scores to selection attributes that indicate customer pre-
generate a ranking system (Louviere et al., ferences using an unconditional approach con-
2013; Soufiani et al., 2012). The probability of sistent with hotel purchasing patterns. It is
a specified pair between a choice and an alter- important to compare different classes of hotel
native is relative to the distance between the two because of the differential customer expecta-
attribute levels on the latent utility scale (Flynn tions about service quality (Knutson et al.,
et al., 2007). This approach has been applied in 1993). This prompts such hotels to target differ-
developing a “best–worst scaling (BWS)” sta- ent markets.
tistical model. Thirdly, the researchers investigate whether
Previous researchers have adopted BWS as a the selection attributes apply to different hotel
choice-based measurement approach (Louviere types according to segmentation characteristics
et al., 2013; Scarpa et al., 2011). Among a set such as gender, income, and frequency of use.
of choices, a pair of response categories should These have been shown to be determinants of
be compared and discerned forcefully between salient customer selection (Spiggle and Sewall,
one another. Understanding the trade-offs that 1987). Most previous investigations concluded
customers make between the attributes associ- that demographic and behavioral factors influ-
ated with different properties can be particularly ence the perceived importance of hotel selection
informative for hotel chains that embrace a mul- attributes (e.g. Hart, 1993; McCleary et al., 1994;
tibrand strategy. Weaver and Oh, 1993). The present researchers
The many studies that have explored the consider such characteristics as potentially influ-
determinants of hotel selection using a variety ential variables when selecting a hotel and apply
of methodologies are indicative of the impor- BWS to choices about the most preferred hotel
tance of the topic (Masiero et al., 2015). While selection attributes.
hotel selection attributes have been identified
previously, researchers have not made use of the
random utility theory for decision-making, Literature review
though it is a potentially straightforward and
Hotel selection attributes and hotel users’
effective optimum utility concept. In other
words, the current authors note that researchers preferences
have given insufficient attention to the subtle Researchers have used a variety of methods to
distinctions affecting the relative importance explore the nature and significance of hotel
derived from the maximum utilities, notably the selection attributes (Chu and Choi, 2000; Dolni-
trade-offs that an individual must seek when car, 2002). Various studies have identified the
choosing between various options. Meanwhile factors or attributes that apply to hotel selection
it may be impossible to quantify differentials (Chu and Choi, 2000; Dolnicar, 2002; Masiero
between evaluations of various attributes. The et al., 2015). However, less attention has been
research results have been inconsistent, perhaps given to what influences and leads to satisfaction
due to an overconfidence among researchers that on the part of customers who are active social
the applicable ratings can be captured along an media users (Kim et al., 2016). The hotel choice
interval scale (Cohen, 2009). In practice, it is preferences of contemporary customers are par-
probable that respondents emanating from ticularly complex because of the diversity of util-
various cultural backgrounds have differential ity functions that they are seeking (Lockyer,
perceptions of the scale items and may subcon- 2005; Pan et al., 2013). A review of the previous
sciously adopt their own criteria (Cohen, 2009; literature shows five main variables that appear
Lee et al., 2008). to determine the preferences and selection beha-
The researchers use hotel selection attributes viors of hotel users: purpose of travel, gender,
and the BWS method to develop a comprehen- age, cultural differences, and accommodation
sive range of choice sets. The chosen approach types.
applies an innovative research method that Some studies have considered the purpose of
overcomes the previous overreliance on align- travel as a potential influence on hotel selection
ing selection attributes with perceived impor- (Griffin et al., 1997; Weaver and Oh, 1993;
tance. Secondly, the researchers examine the Wong and Lam, 2002). Weaver and Oh (1993)
relative importance of hotel selection attributes concluded that compared with their less frequent
for the luxury and economy categories. The counterparts, frequent business travelers attach
present investigation explores salient hotel greater importance to complementary facilities.
Kim et al. 3

Some studies have focused on business traveler customer segments evidently attach differential
hotel selection attributes (Griffin et al., 1997; importance to salient attributes. Sohrabi et al.’s
Wong and Lam, 2002). Griffin et al. (1996) iden- (2012) study on Iran extracted core hotel selec-
tified certain prominent factors in the mid-priced tion factors, namely, hotel staff and service,
and luxury hotel categories, comprising mainly recreational information, cleanliness and room
service-oriented features such as bell service, comfort, expenditure, room facilities, parking,
concierge service, and prearranged check-in ser- promenade and comfort, security, and network
vice. Wong and Lam’s (2002) Hong Kong study service. Shanahan (2003) highlighted the attri-
identified several selection attribute criteria, with butes when advertising a motel, with cleanliness
price being the most crucial, followed by star identified to be the most important, followed by
rating, location, brand, and room type. Business low price and nonsmoking room. The least
travelers prioritized star rating, while leisure tra- important factors were parking and special func-
velers focused more on room rate. Lockyer tional rooms such as pet friendly and conference
(2002) investigated hotel selection attributes rooms.
from the guest and management perspectives and Mehta and Vera (1990) listed 26 attributes
found that business travelers prefer facility- that were regarded as important by individual
related attributes, whereas management attaches hotel customers in Singapore across several mar-
high importance to service and staff-oriented ket segments. The key attributes were identified
attributes. as cleanliness, security, overall service, location,
Demographics such as gender (Hart, 1993; and check-in service. However, the prioritization
McCleary et al., 1994) as well as age cycle stage of attributes varied by customer segment, with
(Ananth et al., 1992; Callan and Bowman, 2000) those staying at higher-class hotels having
can influence perceptions of hotel selection attri- greater expectations about service and amenities
butes. McCleary et al. (1994) identified 12 attri- and being more willing to pay higher prices
butes and concluded that females prioritize (Griffin et al., 1997; Knutson, 1988). The study
security facilities, room service, and price. Male concluded that upscale and budget hotel selec-
customers attach greater importance to business tion diverge because of differing customer needs.
services and facilities and to special room fea- In attempting to identify the most decisive
tures. Focusing on female travelers, Hart (1993) factors in making decisions about the type of
identified security, convenient location, clean accommodation, Chen et al. (2017) deployed
room, reasonable cost, and a workout facility as multinomial logit and nested logit models. Their
the priorities. study concluded that spending is the most signif-
It is evident that age influences the impor- icant factor when considering a diverse set of
tance attached to selection attributes (Ananth accommodation types ranging from international
et al., 1992; Callan and Bowman, 2000). Accord- tourist hotels, tourist hotels, ordinary hotels, and
ing to Ananth et al. (1992), younger travelers youth hostels, to B&Bs. Tanford et al. (2012)
attach importance to value for money, location, compared customers of limited-service hotels
special discounts, temperature control, fire and those of full-service hotels and found differ-
alarms, and bed mattress. Callan and Bowman ent selection factors including utility, green,
(2000) identified value for money as the most brand, amenity, image, and price. They con-
significant selection factor for mature British tra- cluded that price was the most important attri-
velers, followed by safety and security and bute for limited-service hotel customers, whereas
location. utility and price were of equally great importance
Some studies highlighted that cultural differ- for full-service hotel customers.
ences may influence the common core attributes Chu and Choi (2000) identified six hotel
impacting hotel selection (Li et al., 2013; Mehta selection attributes, namely, quality of service,
and Vera, 1990). Particular traveler groups evi- business facility, value, room, food, and security.
dently attach greater importance to certain hotel Most of these attributes were considered to be
attributes. Li et al. (2013) conducted interaction important by both business and leisure travelers,
analyses to identify selection preferences. They though business travelers attach greater impor-
concluded that service is a core attribute for busi- tance to business facilities than their leisure tra-
ness travelers from Asia, Europe, and North veler counterparts. All six attributes are
America. Room quality is the most salient factor considered to be important for selection of
for European business travelers and for couples higher-priced hotels compared with their lower-
from North America and Oceania. Hotel priced equivalents. This suggests that the
4 Journal of Vacation Marketing XX(X)

customers of different hotel types apply different

standards toward levels of service and quality of
facilities (Jeong and Jeon, 2008; Musante et al.,
2009). According to recent statistics about the
US hotel market (Gallup, 2014), luxury hotel
customers spend an average of US$910 on their
hotel stay, whereas economy hotel customers
typically spend US$176. This indicates that the
class of hotel is a key determinant for segmenting
target markets that range from economy to lux- Figure 1. Conceptual framework.
ury, based on the needs and preferences of their
customers.
According to adaptation-level theory, individ- The BWS method
ual judgments can be explained by considering Numerous researchers who used rating scales to
the heaviness of several weights (Helson, 1948). explore the importance of attributes have not pro-
Standards are the key influence in judging the vided respondents with an opportunity to under-
target weights. Four- and five-star hotels are gen- take subtle trade-offs between pairs of attributes
erally more luxurious and provide higher-level and have underestimated individual idiosyncra-
services and facilities than their one- and two- sies caused by cultural differences or verbal ambi-
star equivalents (Griffin et al., 1997). Accord- guities (Lee et al., 2008). In practice, respondents
ingly, five-star hotel users will pay more than who are asked to make paired comparisons in
those who opt for their one-star counterparts survey settings are unable to analyze all possible
(Guillet and Law, 2010). User standards and pur- pairs of multiple comparisons between items.
chasing patterns vary according to hotel type, However, using BWS allows the most important
thereby influencing the selection criteria.
attributes to be explained by asking respondents to
Previous researchers theorized the role of
select one most important and one least important
hotel types in determining hotel users’ purchas-
item from a choice set by making a trade-off
ing patterns as their standard of judgment (Grif-
between the attributes (Cohen, 2009).
fin, et al., 1997; Knutson, 1988; Knutson et al.,
The statistical model underlying BWS assumes
1993). Knutson et al. (1993) identified that
that the relative choice probability of a specified
expectations about services and room amenities
pair relates to the distance between the two attri-
relate to categories such as economy, mid-price,
bute levels on the latent utility scale (Flynn et al.,
and luxury. Griffin et al. (1997) confirmed that
2007). This experimental method tested customer
hotel guests anticipate higher standards as prices
increase. Although previous researchers have judgments in choice modeling (Auger et al., 2006;
shown that selection standards vary according Goodman et al., 2005). Its base theory explains
to hotel type, this study proposes that purchasing how customers undertake an orderly ranking of
patterns are influenced by hotel type. both the most important and the least important
As mentioned previously, hotel selection attri- items on a list (Flynn and Marley, 2014).
bute studies concluded that differentials are attri- The BWS explores how individuals evaluate
butable to sociodemographic, psychological, and an object that includes several attributes by
behavioral variables. Conventional sociodemo- choosing both the top- and the bottom-ranked
graphic variables include gender, income, age, items as valid and reliable selections (Louviere
and education level, while behavioral indicators et al., 2013; Scarpa et al., 2011). The BWS model
include behavioral pursuits (Kotler and Arm- involves a cognitive process by allowing respon-
strong, 2010). In the tourism context, hotel dents to make repeated selections of the best and
operators of different classes provide users with the worst in sets of items. This in turn exhibits
distinct physical features and services, and these the largest perceptual differences on an underly-
criteria form a basis for categorizing hotels from ing continuum of their judgments by providing a
economy to luxury (Knutson, 1988). In consid- best–worst (BW) score (Finn and Louviere,
ering customer selection, the present study 1992). It has rarely been used, despite being one
deploys the balanced incomplete block design of the most useful experimental scaling methods.
(BIBD) and BWS methods as the respective In the present study, it is used to identify the
choice set and scale method. Figure 1 presents hotel selection attributes that have the most
the proposed conceptual framework. marked influence on final customer choices.
Kim et al. 5

The BWS method has been deployed in adoption of the BWS method led to the identifi-
diverse settings outside the hospitality and tour- cation of hotel selection determinant attributes.
ism domain (Auger et al., 2006; Flynn et al., The exploration of key hotel selection determi-
2007; Loureiro and Arcos, 2012). Selection of nant attributes proceeded in two ways. First, a
food and wine may be regarded as sitting both variety of attributes were identified based on a
within and beyond the hospitality field. The review of the literature on hotel selection attri-
diverse cultural traits that apply to food and wine butes. The review extended to articles published
consumption in different national and local set- between 1984 and 2013 in eight leading hospi-
tings and the variety of intrinsic and extrinsic tality and tourism-related journals (e.g. Ananth
product attributes have stimulated researchers’ et al., 1992; Callan and Bowman, 2000; Chow
interest (Bernabéu et al., 2012; Casini et al., et al., 1995; Griffin et al., 1997; Hart, 1993;
2009; Chrysochou et al., 2012). Some cultural Lewis, 1984; Li et al., 2013; Lockyer, 2002;
tourism studies explored the macro environment McCleary et al., 1994; Saleh and Ryan, 1992;
by applying BWS to estimating willingness to Shanahan, 2003; Sohrabi et al., 2012; Weaver
pay or customer preferences (Louviere et al., and Oh, 1993; Wong and Lam, 2002). The search
2013; Scarpa et al., 2011). generated a total of 34 key hotel selection attri-
A conjoint analysis can be applied to calculate butes. During the next stage of the research, a
the relative importance of several attributes (Kim review of comments generated by customers on
et al., 2016; Park et al., 2010). This approach to Trip Advisor was used to generate some key
design identifies a preferred level for each attri- attributes that are considered by customers when
bute, by combining the preferred levels of each selecting hotels. Researchers compiled a list of
attribute within the wider attribute set. To ensure 10 hotel selection attributes that were consis-
its applicability, a conjoint analysis requires the
tently confirmed as the most influential by con-
formation of a small number of combination sets
sidering the levels of preference on the hotel
in designing a set of attribute profiles. In consid-
selection attributes. The following “top 10” attri-
ering the relative merits of the alternative
butes were identified: location, price, cleanli-
approaches, BWS is unique in identifying
ness, security, room comfort and decor,
trade-offs between preferences by comparing all
parking, restaurant and food quality, service,
of the different attributes at the same level. The
bed, and quietness.
preferred levels of the various attributes are
In the second step, the attributes were com-
determined in the minds of the respondents and
pared with each other to identify between indi-
importance levels are recorded on an optimal and
objective basis as the “best or worst” preferred vidual items the best or worst as trade-off
attributes. decisions. In designing the choice sets, all items
Consistent with the primary study objectives, appearing over an equal number of times were
BWS is better suited for identifying the most included to allow for all possible comparisons
important hotel selection attributes in a consid- based on multinomial logit model (Louviere and
eration of two distinct groupings—luxury and Woodworth, 1983). The researchers applied
economy hotel customers. It is anticipated that BIBD in designing the BWS questionnaire.
the results may generate distinct features that The BIBD is the most widely used approach
align with customers’ demographic and beha- for conducting counting-based analyses to orga-
vioral characteristics. While the authors note that nize a series of choice sets (Auger et al., 2006;
a few tourism scholars have adopted BWS, the Cohen, 2009; Flynn and Marley, 2014). A large
approach has been absent in the tourism litera- number of items can be usefully compared in
ture notably as a means of trade-off to investigate order to obtain the full rank of all items in a small
hotel customers’ hotel selection attributes. From number of subsets (Auger et al., 2006; Cohen,
the foregoing, the potential to understand their 2009). It controls the number of times that each
preferences by identifying the attributes that pair is compared since the total number of sub-
affect their behaviors is evident (Cohen, 2009). sets and the number of items in a subset grows by
increasing the number of comparisons. The
design is rooted on a Latin Square design with
Study design and data analysis n items arranged by n rows and n columns. The
A procedure involving multiple BWS steps was items for each row and column are in different
designed and followed in order to identify the positions and are indicative of a block or a choice
most salient hotel selection attributes. The initial set (Weller and Romney, 1988).
6 Journal of Vacation Marketing XX(X)

Since the BIBD assumes that the occurrence The basis for survey participation was the data-
of items and the comparison of set sizes are con- base pool, namely, US hotel users. The survey
stant, it minimizes the chance that respondents was administered in October and November
respond to aspects of the design and make unin- 2014 and generated a total of 748 responses for
tended assumptions about the objects (Flynn and data analysis purposes (i.e. 397 from luxury
Marley, 2014). The method enables each item to hotels and 351 from economy hotels).
appear in every possible position at the same For data analysis purposes, it was necessary to
number of times in the same number of choice transform each choice set of attributes to the
sets (Louviere and Woodworth, 1983). A large original number of attributes. The occurrences
number of items can be compared in order to of best and worst choices for each item were
obtain the full rank of all items in a small number cumulated into frequencies. The data consisted
of subsets (Cohen, 2009). of a combination of best and worst frequencies
In the present study, the BWS method was from each set. On the 10 choice sets, each attri-
constructed using hotel selection attribute choice bute can be chosen three times as the best/most
sets. A BWS model contains three or more items important or as the worst/least important.
in a set and the method allows participants to The BW scores indicate the total worst score
pick the two attributes that are the furthest on the was subtracted from its total best score counter-
underlying dimension of their interest by picking part, and each attribute may range from þ3 to –3.
the best/most important and the worst/least The average BW (ABW) scores were calculated
important attributes. Respondents provide easy by dividing the total BW scores by the number of
and quick responses to each set of experimental respondents and the frequency of each attribute
cases (Goodman et al., 2005). The researchers in the design of the choice sets in descending
are well-placed to analyze the importance of numerical order. The attribute rankings were
each choice. obtained by the BW score and listed according
Drawing upon the 10 hotel selection attri- to the ABW scores. The standard scores for each
butes, the researchers designed 10 subsets that attribute were calculated as follows:
include three attributes. Each was repeated three
times in choice set. The design of the 10 choice Average BW Score ¼
sets took account of the need for a manageable ðCount Best or Most  Count Worst or LeastÞ
length for the questionnaire (Cohen, 2009). The an
composition of the BIBD choice sets for hotel where the count best (most) is the total number of
selection attributes was as follows: (1,2,4), attributes chosen as the most important; count
(2,3,5), (3,4,6), (4,5,7), (5,6,8), (6,7,9), (7,8,10), worst (least), the total number of attributes cho-
(1,8,9), (2,9,10), and (1,3,10). Each attribute was sen as the least important; a, the pair frequency
labeled somewhere between 1 and 10 of a total of of each attribute; and n, the number of total
10. Therefore, the BIBD for 10 attributes is respondents.
regarded as (b, r, k, ), where b is the number
of choice sets (10), r is the repetition per level
(3), k is the number of items in each choice set
(3), and  is the pair frequency (1).
Results
Third, to discern two distinct levels of cus- According to the respondent demographic pro-
tomers who might stay at luxury or economy files, in the luxury hotel data set 40.6% of the
hotels, two screening questions sought their respondents were male. The most prominent age
views about hotel stays undertaken during the range was 21–40 years (51.1%) and older than 51
previous year, including the applicable class. years (33%). Over half (58.2%) of the respon-
Further involvement was confined to respon- dents possessed an associate or bachelor’s
dents whose experiences met the applicable cri- degree. Income levels were evenly distributed
teria. The respondents were assigned to two from the category “less than US$40,000” to
categories: luxury hotels (4/5 star) or economy “US$80,000–99,999.” Over half of the respon-
hotels (1/2 star). The BWS questionnaire proce- dents (about 51%) had stayed at luxury hotels
dure begins with the selection of a best/most on two or three occasions. Most reported that
important hotel attribute and a worst/least attri- they usually stay at a hotel priced in the range
bute in a set of options including identified attri- “US$151–US$400.” Of the economy hotel data
butes. To conduct the study, the researchers set, 32.8% were male and almost half (44.8%)
recruited an online survey company—Qualtrics. were aged 21–40. Of the respondents, 33.9%
Kim et al. 7

Table 1. Attributes considered by luxury and economy hotel users.

Luxury hotel data set (N ¼ 397) Economy hotel data set (N ¼ 351)

Total Total BW ABW Total Total BW ABW

Attribute best worst score score Attribute best worst score score

Cleanliness 768 85 683 0.573 Cleanliness 672 36 636 0.604

Bed 630 213 417 0.350 Bed 529 162 367 0.349
Service 521 197 324 0.272 Price 535 202 333 0.316
Room comfort and 420 306 114 0.096 Security 450 232 218 0.207
decor
Security 396 306 90 0.076 Service 408 224 184 0.175
Price 461 403 58 0.049 Room comfort and 304 369 65 0.062
decor
Location 409 422 13 0.011 Location 323 397 74 0.070
Quietness 242 444 202 0.170 Quietness 221 425 204 0.194
Restaurant and its food 127 676 549 0.461 Parking 69 676 607 0.576
quality
Parking 63 883 820 0.688 Restaurant and its food 52 752 700 0.665
quality
BW: best–worst; ABW: The average best–worst.
Note: Positive signs on the BW and ABW scores indicate that the total number of attributes chosen as the most important is
greater than the total number of attributes chosen as the least important. Negative signs indicate vice versa.

were associate degree holders and the most pre- considered room comfort and decor to be impor-
valent income level was “less than US$40,000” tant when selecting a hotel, whereas economy
(39.6%). Nearly 75% of the respondents had cohorts do not attach importance to this aspect.
stayed at economy hotels more than once, though It was commonly shown in both hotel classes,
fewer than three times. About 89% of the respon- wherein four attributes were insignificant,
dents paid between US$51 and US$150 for a namely, “location,” “quietness,” “parking,” and
hotel room. “restaurant and food quality.”
Table 1 illustrates the rankings of the hotel The researchers identified gender-based hotel
selection attributes between the luxury and econ- selection attributes for luxury and economy
omy hotel data sets. A total best score consists of hotels. The plots reported in Figure 2 provide a
the number of times that each respondent identi- graphic illustration of the ABW score by show-
fied the attribute as the most important, whereas ing the significant attributes in the black column
a total worst score is the number of times that the (ABW > 0) and the insignificant attributes in the
attribute was selected as the least important. As white column (ABW < 0). Graphical presenta-
illustrated in Table 1, luxury and economy hotel tion indicates ABW scores by highlighting the
users perceive the importance of selection attri- most and the least important attributes.
butes differently. “Cleanliness” was the most Figure 2 indicating ABW scores illustrates
important attribute in luxury hotels, followed important attributes from the perspective of male
by “bed,” “service,” “room comfort and decor,” and female luxury and economy hotel cohorts.
“security,” and “price.” The most important attri- The three most significant attributes indicated
bute in economy hotels was cleanliness, followed by both male and female luxury hotel users were
by bed, price, security, and service. cleanliness, bed, and service. Location was only
There is evidently a differential gap between regarded as important by males, whereas this was
luxury and economy hotel user perceptions about not the case for females. On the other hand, secu-
selection attributes. For example, service was rity was regarded as important by females,
reported as the third most important attribute for whereas this was not the case for males. Given
luxury cohorts, compared to the fifth place for the analysis of gender-based attributes from the
economy ones. In addition, price was considered economy hotel data set, the three most important
more important by economy cohorts than by attributes were cleanliness, followed by bed and
luxury ones, while room comfort and decor was price. Interestingly, location was only an impor-
only considered to be an important attribute in tant attribute among the male economy hotel
the case of luxury cohorts. Luxury cohorts users and not among females.
8 Journal of Vacation Marketing XX(X)

Figure 2. The average best–worst (ABW) scores for luxury and economy hotels by gender.

The next analysis involved segmenting the economy hotel users across the five categories
importance of hotel selection attributes on of income.
the basis of their incomes. Figure 3 presents the Next, the various hotel attributes were cate-
results of the ABW scores for the data sets of gorized on the basis of frequency of use. Luxury
luxury hotels. Luxury cohorts earning more than hotel users who stayed once a year considered
US$60,000 and a lower-income group consid- price to be the second most important, whereas
ered location distinctively. The perceived those who stayed more than twice considered bed
importance level of bed and service were differ- to be the second most important. The third most
ent between groups. The ranking of perceived important factor across all categories was ser-
importance for price was widely dispersed vice. Price was an important factor for those who
across all the income categories. As incomes used luxury hotels once, twice, five times, or
drop, price increases in importance when choos- even more, whereas it was unimportant for those
ing a luxury hotel. However, its ranking of per- who stayed three and four times. Location was
ceived importance when selecting a luxury hotel considered to be important for only two cate-
drops as incomes rise. Luxury hotel users with gories, namely, traveling two times and five
lower incomes perceive room comfort and times or more. Interestingly, those who stayed
decor as less important, but higher-income at hotels on more than five occasions a year were
respondents perceived this attribute as very more likely to consider price as relatively signif-
important. icant and location as relatively insignificant com-
Customers in the economy hotel data set con- pared with other frequent users.
sidered cleanliness to be the most salient attri- In the economy hotel data set, the results
bute, irrespective of their income. The showed that they considered price to be rela-
respondents who were on lowest income consid- tively important across all categories. It is inter-
ered price as the second most important factor, esting to note that the most frequent hotel users
whereas those on higher incomes considered who travel more than five times a year consid-
security as the second or third most important ered room comfort and decor and location to be
factor when selecting an economy hotel. Only significant, whereas users in other categories
three categories who had lower incomes and the considered these two factors as insignificant. In
highest income considered location to be unim- addition, respondents who stayed more than five
portant. Restaurant and food quality were the times considered only service as unimportant.
least important factor of all categories, followed Figure 4 displays important hotel attributes and
by parking. Figure 3 presents the important the ABW values of luxury and economy hotel
attributes and the ABW values of luxury and users according to the frequency of use.
Kim et al. 9

Figure 3. The average best–worst (ABW) scores by income level.

Discussion and conclusions (Hart, 1993). In the current study, bed was iden-
tified as the second salient attribute, followed by
Consistent with previous research, the present
service, security, and price, although their impor-
investigation confirmed that cleanliness is the
tance levels were regarded differentially by lux-
most influential hotel selection determinant
ury and economy hotel users. In addition, room
(Lockyer, 2002; Shanahan, 2003). It is useful to condition was a significant determinant for lux-
compare the results with previous identifications ury hotel customers, though not among economy
of location as the most important factor (Aksoy hotel customers. The results show a consensus
and Ozbuk, 2017; Lewis, 1984; McCleary et al., that luxury hotel users attach greater emphasis
1994) together with service (Xue and Cox, to room ambience than their economy hotel
2008). Several significant factors were consistent counterparts (Lewis, 1984).
with those identified in other studies, namely, According to the results of the present study,
value for money (Callan and Bowman, 2000; the four least considered attributes were location,
Chen et al., 2017; Wong and Lam, 2002), interior quietness, restaurant, and parking. These differ
decor (Saleh and Ryan, 1992), and security issue from the findings of previous studies that have
10 Journal of Vacation Marketing XX(X)

Figure 4. The average best–worst (ABW) scores by frequency of hotel use.

stressed the importance that is attached to these The findings for the luxury and economy
attributes in hotel selection (Aksoy and Ozbuk, hotel categories are consistent with the previ-
2017; Ananth et al., 1992; Lockyer, 2002). Such ously held view that high-end hotel customers
inconsistent results may be attributed to different seek higher-quality service (Walls et al., 2011)
methodologies, noting that previous studies and low-end customers are price sensitive (Chen
employed Likert-type scales which do not under- et al., 2017; Senior and Morphew, 1990). This
take rankings on the basis of preferences. Likert- study has confirmed that service is more impor-
type scales do exhibit a tendency to one-sided tant in luxury hotel selection, while price is more
responses (i.e. “yes sayers” and “no sayers”). important in economy hotel selection. The find-
By way of contrast, the BWS approach facilitates ings can derive a reasonable level of support
trade-offs by allowing respondents to compare from previous studies which indicated that lux-
attributes and assists researchers to discriminate ury hotel customers expect to experience profes-
between the levels of importance for various sional and proactive services, and space and
attributes (Cohen, 2009). furnishings can be assured in luxury hotel
Kim et al. 11

settings (Walls et al., 2011). Interestingly, it was when choosing a hotel, while they did not attach
found that luxury hotel users were the only importance to service quality. The results differ
respondent category to attach importance to when the frequency of use and the level of prop-
room comfort and decoration. On the other hand, erty are analyzed.
customers who chose economy hotels expect a
minimum level of room facilities or service, per-
Implications, limitations, and suggestions
haps because of their budgetary constraints
(Chen et al., 2017).
for further research
The gender-based segmentation that formed This study contributes to knowledge by deploy-
part of this study indicates that males emphasize ing a novel research method that identifies hotel
location, whereas females did not do so, regard- selection attributes and has various scholarly
less of the category of hotel. Female luxury hotel merits. First, although BWS has been widely
users attached importance to security, whereas applied over the past 15 years across a variety
male luxury hotel users did not do ao. Both male of domains, hospitality applications have been
and female economy hotel users considered rare. Unlike previous studies, this investigation
security to be a priority. introduced a forceful ranking method that per-
Regarding respondent levels of income, high- mits trade-offs between attributes. By identifying
income users of luxury hotels tended to stress on six salient hotel selection attributes, the results
bed quality, while luxury hotel users on low should facilitate a better understanding of per-
income prioritized service quality. A distinct pat- suasive attributes that may guide hotel selection.
tern was evident in price and room condition Second, this study has explored how attri-
among those on different income levels. The butes are considered differentially by compar-
results support the findings of previous research- ing them on the basis of hotel type. Hotel class
ers that higher-income customers attach less is evidently a relevant criterion for providing
importance to price and more importance to levels of customers’ purchasing patterns (Grif-
room-related attributes (Chen et al., 2017; Tan- fin et al., 1997; Knutson et al., 1993). However,
ford et al., 2012). The location of their hotel was relatively few studies investigated the theoreti-
more important for high-income luxury hotel cal aspects of hotel class. The few studies
users than for low-income earners. Economy served as criteria for assessing satisfaction
hotel users with high incomes view security as and/or dissatisfaction among hotel attributes
important, whereas users of economy hotels on (Dolnicar, 2002). The present study provides
middle-level incomes viewed hotel location as a an advance by identifying selection attributes
significant attribute. With the exception of park- in the context of two different luxury and econ-
ing facilities, the least significantly rated attri- omy hotels.
butes were restaurant related. Third, this study shows the importance of
Consistent with their usage patterns, less fre- intrinsic features in the backgrounds of individ-
quent luxury hotel users considered price as an ual customers, thereby complementing the
important attribute, while most luxury hotel users extrinsic consumption patterns that have been
attached more significance to the bed. This is demonstrated in previous customers’ behavior
consistent with the study by Chu and Choi research. For example, customers had differing
(2000) which concluded that their different pur- perceptions of the four attributes when making
poses of travel prompted business travelers to selections on the basis of gender, income level,
give greater consideration to business facilities and frequency of hotel use. Since hotel selec-
than their leisure travel counterparts. Price was tion attributes appear to differ according to tar-
an important issue for luxury hotel users who get market characteristics or hotel class, future
stay once or twice and even five times or more. researchers may seek more in-depth under-
Less frequent luxury hotel users evidently con- standing of customer diversity in different tar-
sider price to be important, whereas more fre- get markets.
quent hotel users show less price sensitivity. It The research findings have a number of impli-
was, however, found that very frequent luxury cations for practitioners. First, customers have
hotel users consider price to be important. Econ- been shown to base their hotel selection by con-
omy hotel users who use hotels more frequently sidering different attributes, with cleanliness and
show greater price sensitivity. Interestingly, most bed being rated as the most important. These two
frequent economy hotel users regard room dec- attributes were highlighted by all respondents,
oration, comfort, and hotel location as significant regardless of demographics or behavioral
12 Journal of Vacation Marketing XX(X)

characteristics. As an outcome for practitioners, In aligning with the preferences of different

it is suggested that marketers promoting both customer segments, hotels should target more
luxury hotel chains and independent economy accurately and build stronger relationships. Hotel
hotels should emphasize these two guest room promotions should customize the alignment
features to appeal customers. For example, when between prevailing messaging and the guest
communicating with prospective customers and experience and remodel hotel facilities that align
providing reservation opportunities through hotel with priority attributes. To guarantee their
websites or social media, hoteliers could stress longer-term success and positioning, hotels
on cleanliness and a good night’s sleep (the bed). should communicate effectively with customers
Awareness training should also be enhanced about their priority selection attributes and tailor
among sales and reservations staff. Noting that the product offer to their needs.
some hotel companies do emphasize the quality Given the explicit focus of the current study
and comfort of their beds, housekeepers could be on the hotel sector, its scope—to view any simi-
reminded about the importance of tidiness for larities or dissimilarities in the results of a hotel
guests to enhance the perception of bed. The sector—has been limited to a single method.
provision of high-quality bedding could be fur- There might be merit in extending the scope to
ther emphasized as an aspect of providing guests other sectors with a view to identifying any sec-
with a good night’s sleep. tor wide prevalence in tourism. Future studies
Second, the research findings imply that dif- are needed to compare the results using differ-
ferent classes of hotel should undertake differen- ent methods and explore other types of accom-
tiated marketing strategies and campaigns. modation such as condominiums, B&Bs, even
Managers of luxury hotels may place an empha- focusing on aspects of the shared economy such
sis on the hardware aspect such as room facilities as Airbnb.
and decor as well as service aspect, whereas their Secondly, it is noted that the study has applied
economy hotel counterparts should stress on a limited number of attributes and choice sets to
price competitiveness, noting that economy hotel two classes of hotel. In addition, the researchers
respondents were rate sensitive. Above all, the have mainly investigated hotel selection attri-
results indicate a need for differentiation when butes in accordance with customer demographics
managing different classes of hotel. A hotel room and behaviors. It is noted that the present study
is not an undifferentiated “commodity.” This has explored the preferences of US hotel custom-
principle applies equally to hotel chains with an ers when selecting hotels across two classes and
extensive portfolio of properties across different that future studies might explore applications in
class or star levels and also for independent prop- different countries or tourism destinations.
erties that operate within competitive local Finally, it is noted that future researchers might
markets. investigate whether hotel selection attributes dif-
Third, hotel management should consider all fer on the basis of psychological traits such as
target markets, taking into account their diverse perceived value (e.g. hedonic vs. utilitarian) or
demographics and any critical attributes. cultural background.
This research found that clients have different
perceptions of the importance of five attri- Declaration of conflicting interests
butes—service quality, security level, price com- The author(s) declared no potential conflicts of
petitiveness, hotel location, and room comfort interest with respect to the research, authorship,
and its decor—on the basis of gender, income, and/or publication of this article.
and the frequency of hotel use. Noting the dis-
parate preferences, the findings indicate that
Funding
marketing and management activities should be
tailored to the needs of particular market seg- The author(s) received no financial support for
ments. Hotels can present customized messages the research, authorship, and/or publication of
to particular client groups. For example, female this article.
customers expressing concern about security in
both hotel categories and by male customers at References
economy hotels may benefit from some subtle Aksoy S and Ozbuk MY (2017) Multiple criteria deci-
messaging about prevailing security arrange- sion making in hotel location: Does it relate to post
ments both inside the hotel and rooms and also purchase consumer evaluations? Tourism Manage-
in the surrounding areas. ment Perspectives 22: 73–81.
Kim et al. 13

Ananth M, DeMicco FJ, Moreo PJ, et al. (1992) Mar- Flynn TN and Marley AAJ (2014) Best–worst scaling:
ketplace lodging needs of mature travelers. The theory and methods. In: Hess S, Daly A and Elgar
Cornell Hotel and Restaurant Administration E (eds) Handbook of Choice Modelling. pp.
Quarterly 33(4): 12–24. 178–199.
Auger P, Devinney TM, and Louviere JJ (2006) Using Gallup (2014) From economy to luxury, what matters
best–worst scaling methodology to investigate con- most to hotel guests. Business Journal (5 Septem-
sumer ethical beliefs across countries. Journal of ber 2014). Available at: http://www.gallup.com/
Business Ethics 70(3): 299–326. businessjournal/175568/economy-luxury-matters-
Bernabéu R, Dı́az M, Olivas R, et al. (2012) Consumer hotel-guests.aspx (accessed 24 April 2017).
preferences for wine applying best–worst scaling: a Goodman S, Lockshin L and Cohen E (2005) Best–
Spanish case study. British Food Journal 114(9): worst scaling: a simple method to determine drinks
1228–1250. and wine style preferences. Proceedings of the 2nd
Callan RJ and Bowman L (2000) Selecting a hotel and International Wine Marketing Symposium, 2005:
determining salient quality attributes: a prelimi- 16p. In: International Wine Marketing Symposium,
nary study of mature British travelers. The Interna- 2nd. Sonoma, 2005.
tional Journal of Tourism Research 2(2): 97–118. Griffin RK, Shea L and Weaver P (1997) How busi-
Casini L, Corsi AM and Goodman S (2009) Consumer ness travelers discriminate between mid-priced
preferences of wine in Italy applying best–worst and luxury hotels: an analysis using a longitudinal
scaling. International Journal of Wine Business sample. Journal of Hospitality & Leisure Market-
Research 21(1): 64–78. ing 4(2): 63–75.
Chen CM, Tsai YC and Chiu HH (2017) The decision- Guillet BD and Law R (2010) Analyzing hotel star
making process of and the decisive factors in ratings on third-party distribution websites. Inter-
accommodation choice. Current Issues in Tourism national Journal of Contemporary Hospitality
20(2): 111–119. Management 22(6): 797–813.
Chow KE, Garretson JA and Kurtz DL (1995) An Hart W (1993) What women want. Cornell Hotel and
exploratory study into the purchase decision pro- Restaurant Administration Quarterly 34(5): 10–11.
cess used by leisure travelers in hotel selection. Helson H (1948) Adaptation-level as a basis for a
Journal of Hospitality & Leisure Marketing 2(4): quantitative theory of frames of reference. Psycho-
53–72. logical Review 55(6): 297.
Chrysochou P, Krystallis A, Mocanu A, et al. (2012) Jannach D, Gedikli F, Karakaya Z, et al (2012)
Generation Y preferences for wine: an exploratory Recommending hotels based on multi-dimensional
study of the US market applying the best–worst customer ratings. In: Information and communica-
scaling. British Food Journal 114(4): 516–528. tion technologies in tourism 2012. Springer, pp.
Chu RK and Choi T (2000) An importance- 320–331.
performance analysis of hotel selection factors in Jeong M and Jeon M (2008) Customer reviews of hotel
the Hong Kong hotel industry: a comparison of experiences through consumer generated media
business and leisure travellers. Tourism Manage- (CGM). Journal of Hospitality & Leisure Market-
ment 21(4): 363–377. ing 17(1/2): 121–138.
Cohen E (2009) Applying best–worst scaling to wine Kim B, Kim S and Heo C (2016) Analysis of satisfiers
marketing. International Journal of Wine Business and dissatisfiers in online hotel reviews on social
Research 21(1): 8–23. media. International Journal of Contemporary
Dolnicar S (2002) Business travelers’ hotel expec- Hospitality Management 28(9): 1915–1936.
tations and disappointments: a different perspec- Kim S, Chung JY, Petrick J, et al (2016) Determina-
tive to hotel attribute importance investigation. tion of preferred performing arts tourism products
Asia Pacific Journal of Tourism Research 7(1): using conjoint analysis. Journal of Vacation Mar-
29–35. keting 24(1): 1–18.
Finn A and Louviere JJ (1992) Determining the appro- Kotler P and Armstrong G (2010) Principles of mar-
priate response to evidence of public concern: the keting, 13th Global ed. Englewood Cliffs: Prentice
case of food safety. Journal of Public Policy & Hall.
Marketing 11(2): 12–25. Knutson BJ (1988) Hotel services and room amenities
Flynn TN, Louviere JJ, Peters TJ, et al. (2007) Best– in the economy, mid-price and luxury market seg-
worst scaling: what it can do for health care ments: what do frequent travelers expect? Journal
research and how to do it. Journal of Health Eco- of Hospitality & Tourism Research 12(2):
nomics 26(1): 171–189. 259–264.
14 Journal of Vacation Marketing XX(X)

Knutson BJ, Stevens P, Patton M, et al. (1993) Con- price acceptability judgements for user fees using a
sumers’ expectations for service quality in econ- conjoint analysis. Tourism Management 31(2):
omy, mid-price and luxury hotels. Journal of 202–212.
Hospitality & Leisure Marketing 1(2): 27–43. Saleh F and Ryan C (1992) Client perceptions of
Lee JA, Soutar G and Louviere J (2008) The best– hotels: a multi-attribute approach. Tourism Man-
worst scaling approach: an alternative to agement 13(2): 163–168.
Schwartz’s values survey. Journal of Personality Scarpa R, Notaro S, Louviere J, et al. (2011) Exploring
Assessment 90(4): 335–347. scale effects of best/worst rank ordered choice data
Lewis RC (1984) The basis off hotel selection. Cornell to estimate benefits of tourism in alpine grazing
Hotel and Restaurant Administration Quarterly commons. American Journal of Agricultural Eco-
25(1): 54–69. nomics 93(3): 813–828.
Li G, Law R, Vu HQ, et al. (2013) Discovering the Senior M and Morphew R (1990) Competitive strate-
hotel selection preferences of Hong Kong inbound gies in the budget hotel sector. International Jour-
travelers using the Choquet Integral. Tourism Man- nal of Contemporary Hospitality Management
agement 36: 321–330. 2(3): 3–9.
Lockyer T (2002) Business guests’ accommodation Shanahan KJ (2003) The degree of congruency
selection: the view from both sides. International between roadside billboard advertisements and
Journal of Contemporary Hospitality Management sought attributes of motels by US drive tourists.
14(6): 294–300. Journal of Vacation Marketing 9(4): 381–395.
Lockyer T (2005) Understanding the dynamics of the Sohrabi B, Vanani IR, Tahmasebipur K, et al. (2012)
hotel accommodation purchase decision. Interna- An exploratory analysis of hotel selection factors: a
tional Journal of Contemporary Hospitality Man-
comprehensive survey of Tehran hotels. Interna-
agement 17(6): 481–492.
tional Journal of Hospitality Management 31(1):
Loureiro ML and Arcos FD (2012) Applying best–
96–106.
worst scaling in a stated preference analysis of
Spiggle S and Sewall MA (1987) A choice sets model
forest management programs. Journal of Forest
of retail selection. The Journal of Marketing 51(2):
Economics 18(4): 381–394.
97–111.
Louviere J, Lings I, Islam T, et al. (2013) An intro-
Azari H, Parks D and Xia L (2012) Random utility
duction to the application of (Case 1) best–worst
theory for social choice. In: Advances in Neural
scaling in marketing research. International Jour-
Information Processing Systems, pp. 126–134.
nal of Research in Marketing 30(3): 292–303.
Tanford S, Raab C and Kim YS (2012) Determinants
Louviere JJ and Woodworth G (1983) Design and anal-
of customer loyalty and purchasing behavior for
ysis of simulated consumer choice or allocation
full-service and limited-service hotels. Interna-
experiments: an approach based on aggregate data.
Journal of Marketing Research 20(4): 350–367. tional Journal of Hospitality Management 31(2):
Masiero L, Heo CY and Pan B (2015) Determining 319–328.
guests’ willingness to pay for hotel room attributes Walls A, Okumus F, Wang Y, et al. (2011) Under-
with a discrete choice model. International Journal standing the consumer experience: an exploratory
of Hospitality Management 49: 117–124. study of luxury hotels. Journal of Hospitality Mar-
McCleary KW, Weaver PA and Lan L (1994) Gender- keting & Management 20(2): 166–197.
based differences in business travelers’ lodging Weaver PA and Oh HC (1993) Do American business
preferences. Cornell Hotel and Restaurant Admin- travelers have different hotel service requirements?
istration Quarterly 35(2): 51–58. International Journal of Contemporary Hospitality
Mehta SC and Vera A (1990) Segmentation in Management 5(3): 16–21.
Singapore. The Cornell Hotel and Restaurant Weller SC and Romney AK (1988) Systematic Data
Administration Quarterly 31(1): 80–87. Collection, Vol. 10. Newbury Park, CA: Sage.
Musante MD, Bojanic DC and Zhang J (2009) An Wong KK and Lam C (2002) Predicting hotel choice
evaluation of hotel website attribute utilization and decisions and segmenting hotel consumers: a com-
effectiveness by hotel class. Journal of Vacation parative assessment of a recent consumer based
Marketing 15(3): 203–215. approach. Journal of Travel & Tourism Marketing
Pan B, Zhang L and Law R (2013) The complex mat- 11(1): 17–33.
ter of online hotel choice. Cornell Hospitality Xue XH and Cox C (2008) Hotel selection criteria and
Quarterly 54(1): 74–83. satisfaction levels of the Chinese business traveler.
Park J, Elis G, Kim S, et al. (2010) An investigation of Journal of China Tourism Research 4(3/4):
camping tourists’ perceptions of social equity and 261–281.