Вы находитесь на странице: 1из 14

This article was downloaded by: [Archives & Bibliothques de l'ULB]

On: 11 August 2015, At: 21:21


Publisher: Routledge
Informa Ltd Registered in England and Wales Registered Number: 1072954 Registered office: 5 Howick Place,
London, SW1P 1WG

Journal of Travel & Tourism Marketing


Publication details, including instructions for authors and subscription information:
http://www.tandfonline.com/loi/wttm20

Online Consumer Review Factors Affecting Offline


Hotel Popularity: Evidence from Tripadvisor
Karen L. Xie, Chihchien Chen & Shinyi Wu
Published online: 08 Jul 2015.

Click for updates

To cite this article: Karen L. Xie, Chihchien Chen & Shinyi Wu (2015): Online Consumer Review Factors Affecting Offline Hotel
Popularity: Evidence from Tripadvisor, Journal of Travel & Tourism Marketing, DOI: 10.1080/10548408.2015.1050538

To link to this article: http://dx.doi.org/10.1080/10548408.2015.1050538

PLEASE SCROLL DOWN FOR ARTICLE

Taylor & Francis makes every effort to ensure the accuracy of all the information (the Content) contained
in the publications on our platform. However, Taylor & Francis, our agents, and our licensors make no
representations or warranties whatsoever as to the accuracy, completeness, or suitability for any purpose of the
Content. Any opinions and views expressed in this publication are the opinions and views of the authors, and
are not the views of or endorsed by Taylor & Francis. The accuracy of the Content should not be relied upon and
should be independently verified with primary sources of information. Taylor and Francis shall not be liable for
any losses, actions, claims, proceedings, demands, costs, expenses, damages, and other liabilities whatsoever
or howsoever caused arising directly or indirectly in connection with, in relation to or arising out of the use of
the Content.

This article may be used for research, teaching, and private study purposes. Any substantial or systematic
reproduction, redistribution, reselling, loan, sub-licensing, systematic supply, or distribution in any
form to anyone is expressly forbidden. Terms & Conditions of access and use can be found at http://
www.tandfonline.com/page/terms-and-conditions
Journal of Travel & Tourism Marketing, 2015
2015 Taylor & Francis
ISSN: 1054-8408 print / 1540-7306 online
DOI: 10.1080/10548408.2015.1050538

ONLINE CONSUMER REVIEW FACTORS AFFECTING


OFFLINE HOTEL POPULARITY: EVIDENCE FROM
TRIPADVISOR
Karen L. Xie
Chihchien Chen
Shinyi Wu
Downloaded by [Archives & Bibliothques de l'ULB] at 21:21 11 August 2015

ABSTRACT. The business value of online consumer reviews has emerged in recent year as one of
utmost importance for hotel marketers. This study examines how online consumer reviews affect
ofine hotel popularity. Using time-series data of 56,284 hotel reviews posted for more than 1000
hotels listed on TripAdvisor, this paper estimates the effect of factors of online consumer review,
including quality, quantity, consistency, and recency, on the ofine hotel occupancy (i.e. how popular
the hotel is among consumers). The empirical evidence shows the relative effect of online consumer
review factors on ofine hotel popularity when controlling for other hotel characteristics. In particular,
the effect of review quality lasts for at least a couple of quarters, whereas that of other online consumer
review factors remains short-term. The ndings provide a managerial basis to improve the online
presence of hotels on social media platforms by strategically utilizing important review factors.

KEYWORDS. Online consumer review, ofine hotel popularity, hotel occupancy, social media
marketing

1. INTRODUCTION quality is often unknown before purchase


(Nelson, 1970; Pine & Gilmore, 1998). Gretzel
Technological advances over the past dec- and Yoo (2008) indicate that online reviews
ades have led to the proliferation of consumer offer quality information to reduce risk in pur-
reviews on social media platforms where con- chasing experience goods. Online reviews
sumers shop for goods. Consumer reviews are enable consumers to learn about the perception
posted of a wide range of products and services, of product quality and service satisfaction from
and they have become part of the decision- previous consumers without experiencing the
making process for many consumers goods themselves. Consumer reviews written
(Chevalier & Mayzlin, 2006). In particular, con- by previous buyers signal a certain level of
sumer reviews are important for learning about evaluation and feedback of experience goods,
experience goods such as hotel rooms as their thus providing an important reference for new

Karen L. Xie, PhD is Assistant Professor, Hospitality Management, Daniels College of Business,
University of Denver, Denver, CO, USA, USA (Email: Karen.Xie@du.edu).
Chihchien Chen, PhD, is Assistant Professor, College of Hotel Administration, University of Nevada, Las
Vegas, NV, USA (Email: Chih-Chien.Chen@unlv.edu).
Shinyi Wu, is Associate Professor, Information Systems, W. P. Carey School of Business, Arizona State
University, Tempe, AZ, USA (Email: shinyi.wu@asu.edu).
Address correspondence to: Karen L. Xie, Hospitality Management, Daniels College of Business,
University of Denver, Denver, CO, USA (Email: Karen.Xie@du.edu).
1
2 JOURNAL OF TRAVEL & TOURISM MARKETING

buyers to help them make informed decisions Despite the fact that the value to business of
and choose the products that best match their online consumer reviews has become one of the
preference. As a result, nowadays about three- most important research topics, empirical litera-
quarters of consumers take into account online ture still lags behind in three critical aspects that
consumer reviews when planning their travel motivate our study. First, previous literature has
itineraries (Xie, Zhang, & Zhang, 2014), and discussed online consumer review factors
nearly 50% of consumers visit an online review affecting business performance but most often
site for information connected with their online in a singular and fragmented fashion without
travel purchasing (Compete, Inc., 2006). Online accounting for the many aspects of online
consumer reviews have become an important review factors. For example, some studies only
source of information for consumers, substitut- investigate the review consistency to measure
ing for other forms of business-to-consumer the degree of disagreement in consumer opi-
(Jiang & Chen, 2007) and ofine word-of- nions (Godes & Mayzlin, 2004; Sun, 2012)
Downloaded by [Archives & Bibliothques de l'ULB] at 21:21 11 August 2015

mouth (WOM) communication (Lewis & while others focus solely on review quantity to
Bridger, 2001) about the quality of service assess product awareness among consumers
providers. (Dellarocas et al., 2007; Godes & Mayzlin,
Given the prevalent use of online reviews by 2004). The number of empirical studies
consumers, the value of online consumer accounting for a holistic picture of the effects
reviews to business performance has emerged of online consumer review factors is limited.
in recent years as one of utmost importance to Second, previous studies mainly focus on
marketers. It is thus not surprising that online online performance measures such as online
consumer reviews have become the subject of product popularity on social media websites
extensive research (Cunningham, Smyth, Wu, (see for example Ghose, Ipeirotis, & Li, 2012
& Greene, 2010). A burgeoning academic for Travelocity; Luca, 2011 for Yelp; and Xie
body of literature has emerged to study how et al., 2014 for TripAdvisor), neglecting the
online consumer reviews affect ofine business ofine product popularity among consumers
performance (Moe & Trusov, 2011; Xie et al., (i.e. how many consumers actually purchase
2014). Specically, previous research has iden- the product). The ofine product popularity is
tied three primary factors that aggregate important because it justies the return on
online consumer review information and are investment (ROI) of the social media marketing
strategically important to business perfor- efforts and is directly related to the bottom line
mance: review quality, review consistency, of business performance (Aria, 2015). Despite
review quantity, and review recency. Review the industry trend toward recognizing the
quality, or the review rating, reects the level importance of the consumer voice, how online
of consumer satisfaction and is the focus of consumer reviews affect ofine product popu-
most empirical studies on product reviews larity remains poorly understood owing to the
(Clemons, Gao, & Hitt, 2006; Dellarocas, paucity of existing research on the topic.
Zhang, & Awad, 2007). Review consistency, Finally, previous social media studies have
measured as the standard deviation to the mean focussed mainly on one-off purchase items or
rating, captures the degree of disagreement short life-cycle information goods such as
among consumers (Godes & Mayzlin, 2004). movies (Dellarocas, Gao, & Narayan, 2010;
Review quantity, or the number of reviews, as Duan et al., 2008; Liu, 2006), books
a measure of the volume of discussions, signals (Chevalier & Mayzlin, 2006; Li & Hitt, 2008),
brand awareness and popularity of a product and software (Duan et al., 2008; Lee & Tan,
on a social media platform (Duan, Gu, & 2013), which is likely to be due to the avail-
Whinston, 2008; Zhu & Zhang, 2010). ability of data for these categories. However,
Finally, because each consumer review has its information goods are unique in that they have
time stamp at the time it is published online, product life cycles that are both short and fol-
review recency refers to the age of consumer low predictable exponential patterns (Moe &
reviews (TripAdvisor, 2015). Fader, 2001; Sawhney & Eliashberg, 1996).
Xie, Chen, and Wu 3

These products experience the greatest level of social media marketing perspective which
sales (and review activity) immediately after emphasizes a coordinated use of online consu-
launch. Very quickly after that, sales (and mer review factors to drive the ofine business
review activity) taper off dramatically. The dan- performance and to maximize the social media
ger of using such product categories is that marketing success.
results can be very sensitive to the stage of the In the following, we rst introduce related
product life cycle during which the researcher literature and develop our hypotheses. This is
collects the data. As a result, the product popu- followed by the description of our methodology.
larity observed can be more easily attributed to We then outline models to estimate, the results
changes due to the natural progression of the are presented, and research implications are dis-
product life cycle, and are less likely to be a cussed. We conclude by providing recommen-
result of changes in the consumer review envir- dations to hotel marketers and pointing out
onment. Studies that examine social media limitations of this study.
Downloaded by [Archives & Bibliothques de l'ULB] at 21:21 11 August 2015

exposure in relation to more mundane consumer


and business products such as hotels are rare.
This paper aims to bridge this knowledge 2. LITERATURE AND HYPOTHESES
gap by empirically examining the effect of
online consumer review factors on ofine The recent surge in online user-generated
hotel popularity. Our research question is there- reviews has evidently had a signicant inuence
fore: what is the relative effect of online con- on consumer purchasing behavior. It has
sumer review factors on ofine hotel become common practice to include these
popularity? Our study is based on a unique user-generated reviews alongside product
dataset of 56,284 individual TripAdvisor con- descriptions as a means for consumers to gather
sumer reviews of 1067 hotels across ve major- more information, form opinions, and make
destination cities in Texas, United States (US) purchasing decisions, as well as for companies
(i.e. Austin, Houston, San Antonio, Dallas, and to collect consumer data and make recommen-
Fort Worth). The uniqueness of our dataset is dations. Due to the high volume of user-gener-
two-fold. First, these novel data not only ated reviews that are posted, it becomes
enables us to investigate the direct effect of increasingly difcult for consumers to sift
consumer review factors within one self-con- through the reviews and identify a hotel of
tained platform but also prevent the inuence high quality. For that reason, most leading
of confounding factors from separate platforms online review websites such as Yelp, Amazon,
which have been used in most previous research and TripAdvisor tend to aggregate online con-
(Stephen & Galak, 2012). Second, previous sumer review information into four key factors,
media research focuses mostly on the aggre- namely review quality, review quantity, review
gated market level effect of consumer review consistency, and review recency, and highlight
factors, overlooking the critical phenomena these factors at the top of each hotel review
occurring at the dyadic individual hotel level. page (Figure 1). By referring to this aggregated
Our dataset allows us to look into the inuence information, consumers can easily access
of consumer review factors at a disaggregated recommended products from a vast amount of
individual hotel level. Using our dataset, we information available (Rajan, 2014).
exploit a blend of econometric models (xed
effects and random effects) to estimate the ef-
cacy of online consumer review factors such as 2.1 Review Quality
quantity, quality, and consistency in ofine
hotel popularity (i.e. hotel occupancy), control- Most social media platforms allow consu-
ling for the variation of differences across hotels mers to provide reviews and ratings to evalu-
and within time. This study sheds some light on ate their experience with a product (Chu,
the vague online review determinants of ofine 2009). According to Jeong and Jeon (2008),
hotel popularity and advocates an integrated ratings can be closely assimilated to an
4 JOURNAL OF TRAVEL & TOURISM MARKETING

FIGURE 1. An Example of Online Consumer Review Factors on TripAdvisor

Review quantity
Downloaded by [Archives & Bibliothques de l'ULB] at 21:21 11 August 2015

Review consistency

Review quality
Review recency

overall service evaluation. In that sense, the than a lower quality one. Generally, if we
customer uses a single scale to express his or assume that consumers are rational, it is very
her judgment of the product experience. likely that they would purchase a product with a
Review quality serves as a simplied heuristic higher review quality indicated by other consu-
to instantly signal a hotels level of quality as mers than one with a lower review quality
agreed by individual consumers. Hence, the (Chen, Wu, & Yoon, 2004). There is therefore
review quality can be assimilated and used a strong reason to believe that the higher review
as a proxy of quality as perceived by previous quality indicated by previous consumers may
customers (Chen & Xie, 2008; Jiang & Chen, signal a desirable product quality and guest
2007). satisfaction which persuades subsequent consu-
According to the rational action theory mers to purchase the item (Liu, 2006). The
(Becker, 1976), a rational individual tends to superior review quality simply increases consu-
want good rather than bad. This assumption mers condence level when purchasing the
is widely used in social and economic behavior product (Ratnasingham, 1998) because the posi-
contexts. Rational action theory holds that con- tive WOM from previous reviewers mitigates
sumers, when making their purchase decisions, the potential risks and uncertainty associated
tend to choose a higher quality product rather with the purchase and ensures an informed
Xie, Chen, and Wu 5

decision, just like the decision previous products and state that less variability in the
reviewers have made (Xie et al., 2014). review quality could reduce risk and uncertainty
Products with a higher review quality are likely associated with the hotel quality perceived by
to be purchased by many subsequent reviewers readers and thus result in higher product perfor-
who have seen the previous consumer reviews, mance. Accordingly, we propose the following
and generally these products become more pop- hypothesis:
ular among consumers than those with a lower
review quality. We therefore propose the fol- H2: Review consistency of online consumer
lowing hypothesis (H): reviews has a negative effect on ofine
hotel popularity.
H1: Review quality of online consumer
reviews has a positive effect on ofine
hotel popularity.
Downloaded by [Archives & Bibliothques de l'ULB] at 21:21 11 August 2015

2.3 Review Quantity


Consumer reviews at the very least convey
2.2 Review Consistency the existence of the brand and thereby add it to
the choice set of the consumer (Clark,
According to the rational action theory Doraszelski, & Draganska, 2009). According
(Becker, 1976), rational consumers are risk- to the theory of social contagion (Sutherland,
averse. When making a purchase, consumers 1995), peer consumers are likely to follow the
are reluctant to accept a product of uncertain opinions of previous reviewers as a result of
quality rather than another product of a more pressure to conform to a peer group (Zhang,
certain (possibly even lower) quality (Zhu & Pan, Smith, & Li, 2009). As consumers post
Zhang, 2010). Faced with a wide range of their recommendations and opinions about a
review quality, it is impossible for subsequent product or service on social networking sites,
consumers to determine the true quality of a they attempt to persuade their peer consumers
product, especially if there is no textual review to see their point of view and, thus, inuence
to complement the review quality. Thus, incon- their decision-making (Chu & Kim, 2011).
sistent opinions should have a negative impact Thus, consumer reviews can lead consumers
on demand (Sun, 2012). As a statistical concept, to rationalize their purchasing decisions by
variance of reviews is a natural measure to reinforcing the idea that many other consumers
capture the heterogeneity or inconsistency in also bought or did not buy the same product or
consumer opinions. This concept of review con- services. Previous studies have lent support to
sistency is generally associated with uncertainty the presence of social contagion in online con-
about product quality (De Maeyer, 2012). A sumer reviews. For example, Moe and Trusov
wider range of review quality (i.e. higher varia- (2011) demonstrate that subsequent consumer
tion in review ratings) may indicate that pre- behavior is signicantly affected by reviews
vious consumers who write reviews have from previous consumers. In an experimental
inconsistent opinions about the product quality. setting, Schlosser (2005) nds that consumers
As Kirby (2000, p. E1) explains, one may not who have decided to post their opinions tend to
trust just one non-expert . . . but if 9 out of 10 negatively adjust their product evaluations
non-experts agree, it is probably worth buying. after reading negative reviews from previous
Zhu and Zhang (2010) point out that high var- reviewers. This indicates that consumer post-
iation carries great risk, while low variation ing behavior is affected by the inuence of
offers certainty. Ghose et al. (2012) investigate previously posted reviews. We therefore pro-
the impact of consumer reviews on a variety of pose the following hypothesis:
6 JOURNAL OF TRAVEL & TOURISM MARKETING

H3: Review quantity of online consumer 3. METHODOLOGY


reviews has a positive effect on ofine
hotel popularity. 3.1 Data and Measures
Our research context is TripAdvisor, a pro-
minent consumer review website on which the
2.4 Review Recency impact of online dissemination of opinions and
reviews is rapid and far-reaching (Litvin,
A plethora of consumer review data is gener-
Goldsmith, & Pan, 2008). TripAdvisor offers
ated, processed, and presented on the Internet on a
more than 150 million reviews and opinions
daily basis. Almost all social media platforms list
covering more than 810,000 hotels, bed-and-
consumer reviews on their webpages according to
breakfasts, and specialty lodging in the world
the recency of reviews, which refers to the age of
(TripAdvisor, 2015). It acts as a forum for
Downloaded by [Archives & Bibliothques de l'ULB] at 21:21 11 August 2015

the review (TripAdvisor, 2015). According to the


everyday consumers to air their personal opi-
serial position effect (Coleman, 2006), consumers
nions regarding service providers quality whilst
tent to recall or mainly take into account the last
also enabling them to read the recommendations
item in a list or sequence, rather than earlier items.
of fellow consumers (Jeacle & Carter, 2011). In
Particularly, the order in which consumer reviews
the meantime, hotel managers can regularly post
appear (and presumably the recentness) have been
their product and service information (Zhang
found to effect consumers purchase decisions and
et al., 2009) and also proactively induce their
perceived usefulness (Hu & Li, 2011). Usually,
consumers to spread the word about their pro-
recent reviews are listed before older reviews.
ducts online (Godes & Mayzlin, 2009).
Displaying upfront the data that matter to consumer
Therefore, TripAdvisor provides a good inter-
decision-making is relevant because consumers
active setting where we can observe consumers
pay more attention to recent, fresh information
information search behaviors and hotel man-
than older information. According to Murphy,
agers manipulation and feedback mechanism
Hofacker, and Mizerski (2006), a links location
strategies simultaneously.
on a web page is an important factor inuencing
Our study is based on a unique dataset of
visitors decision to click on a particular link. Using
56,284 hotel reviews for more than 1000 hotels
two eld experiments, they nd that the higher a
located in ve major hotel markets of Texas over
links position on a list of links, the greater the
26 quarters (2005 Quarter 12011 Quarter 2).
probability that visitors will click on that link. A
These niche data enable us to explore hotel
study by Ansari and Mela (2003) also suggests a
popularity in relation to consumer review factors
positive relationship between the serial position of
at individual hotel level over time. Table 1 pre-
a link in an email and recipients clicks on that link.
sents distributions of consumer review records
Similarly, Drze and Zufryden (2004) imply a
across our focal markets. Unlike most previous
positive relationship between a links serial posi-
tion and site visibility. That is, ceteris paribus,
consumers would pay most attention to product
reviews near the top of a review web page and TABLE 1. Distributions of Consumer Reviews
least attention to reviews near the bottom of the Across Hotel Markets (Texas, United States)
web page or email. Thus, considering that online
consumer reviews are typically posted in chrono- Consumer Reviews
Market
logical order, consumers are likely to be most
Reviews % of total Hotel % of total
inuenced by the recently posted online reviews.
We therefore propose the following hypothesis: Austin 9,814 17.43 162 15.18
Dallas 11,722 20.82 164 15.37
Fort Worth 2,896 5.14 89 8.34
H4: Review recency of online consumer Houston 13,158 23.36 345 32.33
reviews has a positive effect on ofine San Antonio 18,716 33.25 307 28.77
hotel popularity. Total 56,284 100 1067 100
Xie, Chen, and Wu 7

studies that tend to focus on either large or chain management response ratio is 0.09, indicating
hotels, our sample includes both large and small, that about nine hotel managers respond to every
and independent and chain hotels. 100 consumer reviews in a given quarter. On
Consumer reviews are auto-parsed from average, hotels in our sample are 17 years old
TripAdvisor review webpages using two crawlers with 189 guest rooms, between mid-market
developed by Ruby.1 Following previous litera- economy and full service, and with eight inter-
ture (Duan et al., 2008; Zhu & Zhang, 2010), we nal amenities.
collect individual reviewers hotel ratings
(ReviewQuality) distributed on a scale of 15, 1
representing terrible and 5 representing excel- 3.2 Model Specication
lent. Based on the individual reviews collected,
we calculate the standard deviation of review rat- Consumer reviews are time series for indivi-
ings (ReviewConsistency) and count the number dual hotels in our sample. This panel data struc-
Downloaded by [Archives & Bibliothques de l'ULB] at 21:21 11 August 2015

of consumer reviews received for each hotel ture makes it necessary to adopt appropriate
(ReviewQuantity). We also document the date econometric models to account for individual
and time each consumer review is posted in heterogeneity. We use two techniques to analyze
order to measure the review recency. the data, xed effects, and random effects, con-
In addition, we obtain multi-quarter, archi- trolling for unobservable variables across hotels
val data of hotel occupancy records from a or change over time but not across hotels (man-
research rm well-known for their hospitality agement strategy, hotel culture, etc.). The result-
data and research services (Trepp, LLC). The ing model is
hotel records contain information on the hotel
occupancy (logOccupancy).2 Besides the focal logOccupancyit Xit Sit Zi i
variable of hotel occupancy, we collect hotel t it
characteristics information, including the loga-
rithm of the average daily rate at the property where the dependent variable, logOccupancyit,
level (logADR), age of the hotel (HotelAge), is a measure of popularity of hotel i at quarter t.
number of guest rooms (HotelSize), hotel Xit is a set of consumer review factors, includ-
class (HotelClass), number of amenities ing ReviewQualityit, ReviewConsistencyit,
(HotelAmenity) and the ratio of the number ReviewQuantityit, and their lagged terms. We
of manager responses to the number of con- control for other variables, including time-var-
sumer reviews (ResponseRatio). iant social variables Sit (ResponseRatioit) and
We consider the intersection of consumer hotel characteristics Zi (logADRit, HotelAgeit,
reviews and the hotel occupancy records at a HotelSize i, HotelClassi, and HotelAmenityi).
quarterly level for each individual hotel in our The hotel and time dummies are represented
sample. The quarterly level aggregation of the by i and t. The error terms that have been
consumer review activities allows us to over- clustered at the hotel level are captured by it.
come the sparseness of social media activity
because consumer reviews for hotels tend to
be spaced out and rarely occur on multiple 4. RESULTS AND FINDINGS
days in the same quarter, as would be expected
for almost all hotel properties except for very We estimate review quality, review consis-
high-prole ones. tency, and review quantity in the ofine hotel
Table 2 presents variable denitions and popularity model and present the estimation
summary statistics. Hotels in our sample receive results in Table 3. By estimating both models
an average review rating of 3.48 out of 5 at a of xed effects and random effects, we examine
standard deviation of 0.99 in a given quarter. the effect of consumer review factors on hotel
The total number of consumer reviews varies popularity while accounting for hotel-specic
from 1 to 189, with an average number of 3.67 characteristics and social variables. Specically,
reviews in a given quarter. The average the xed effects estimation controls for all time-
Downloaded by [Archives & Bibliothques de l'ULB] at 21:21 11 August 2015

TABLE 2. Description of Variables

Description Mean Standard deviation Median Minimum Maximum

ReviewQuality Average review rating of a hotel: 1 for terrible, 2 for poor, 3 for average, 4 for 3.48 1.19 3.86 1 5
very good, and 5 for excellent in a given quarter
ReviewConsistency Standard deviation of review ratings of a hotel in a given quarter 0.99 0.62 2 0 2.83
ReviewQuantity Number of consumer reviews in a given quarter 3.67 6.62 2 1 189
logADR Logarithm of the ADR of a hotel in a given quarter 4.53 0.01 4.60 2.82 5.97
ResponseRatio Number of management responses compared with number of consumer reviews 0.09 0.26 0 0 1
in a given quarter
HotelClass Dummy variable with the value of 5 for a luxury hotel, 4 for an above average 2.72 1.05 3 0 5
hotel with some outstanding features and a broad range of services, 3 for a
full service hotel, 2 for a mid-market economy hotel, and 1 for a budget
traveler hotel
HotelAmenity Number of internal amenities such as indoor swimming pool, free high-speed 7.90 2.54 8 0 13
Internet, tness center, wheelchair access, and pets allowed
HotelAge Number of years since the opening of a hotel 17.43 12.65 13 1.00 63
HotelSize Number of guest rooms 189.41 184.91 132 5 1,840
JOURNAL OF TRAVEL & TOURISM MARKETING

Notes. ADR: average daily rate.


Xie, Chen, and Wu 9

TABLE 3. Estimates of Consumer Review Effects on Hotel Popularity

Hotel popularity (logOccupancy)

(1) Fixed effects (2) Random effects

Estimate Robust p-value Estimate Robust p-value

ReviewQualityit 0.011** (0.011) 0.046* (0.079)


ReviewConsistencyit 0.000 (0.584) 0.024 (0.541)
ReviewQuantityit 0.003** (0.033) 0.043** (0.048)
ReviewQualityit-1 0.020*** (0.000) 0.044*** (0.000)
ReviewConsistencyit-1 0.001 (0.222) 0.035 (0.355)
ReviewQuantityit-1 0.002 (0.216) 0.079 (0.311)
ReviewQualityit-2 0.009** 0.025 0.053** (0.020)
Downloaded by [Archives & Bibliothques de l'ULB] at 21:21 11 August 2015

ReviewConsistencyit-2 0.000 (0.829) 0.024 (0.565)


ReviewQuantityit-2 0.000 (0.663) 0.037 (0.563)
logADRit 0.446*** (0.000) 0.188*** (0.000)
ResponseRatioit 0.021 (0.248) 1.376 (0.180)
HotelClassi - - 0.025*** (0.000)
HotelAmenityi - - 0.003 (0.176)
HotelAgeit 0.032*** (0.000) 2.034*** (0.000)
HotelSizei 0.002*** (0.000) 0.078*** (0.000)
Constant 2.941*** (0.000) 3.309*** (0.000)
Hotel xed effects Yes Yes
Time xed effects Yes Yes
Adjusted R-squared 0.256 0.167
Number of year-quarter 26 26
Diagnostic tests
Hausman test for model preference Null hypothesis: Difference in coefcient is not systematic
Chi2 = 3.670; Prob > Chi2 = 0.042
BreuschPagan test Null hypothesis: Presence of homoskedasticity (or constant variance)
Chi2 = 5.860; Prob > Chi2 = 0. 000
Lagram-Multiplier test Null hypothesis: There is no serial correlation.
F = 0.214; Prob > F = 0.660

Note. ADR: average daily rate; *** p < 0.01; ** p < 0.05; * p < 0.1.

invariant differences between the hotels so that H1 and H3. In other words, the popularity of
the estimated coefcients of the xed-effects hotels becomes higher the higher the number of
models cannot be biased because of omitted positive ratings (to inform consumers of the hotel
time-invariant characteristics. Unlike xed quality) and the higher the number of consumer
effects, random effects estimation assumes that reviews from reviewers (to create online buzz).
differences across hotels may have some inu- However, the effect of ReviewConsistency on
ence on hotel popularity and should be mea- hotel popularity (0.000, p = 0.584) is not signi-
sured. We then run a Hausman test to decide cant, which means consistency of reviewer opi-
which is the preferred model for xed effects nions would not signicantly inuence hotel
and random effects respectively. In addition, we popularity. It is thus likely that quality and quan-
conduct a few additional diagnostic tests for tity of consumer reviews become primary consu-
robustness check purposes. mer-generated information to determine the hotel
Column (1) in Table 3 shows the results of occupancy rate, suppressing other information
mean effects estimation using xed effects estima- sources such as review consistency. Interestingly,
tion. The regression estimates that hotel popular- the effect of review quality carries over to at least
ity becomes higher with an increase in a couple of quarters (0.020, p < 0.001 for
ReviewQualityit (0.011, p = 0.011) and ReviewQualityit-1 and 0.009, p = 0.025 for
ReviewQuantityit (0,003, p = 0.033), supporting ReviewQualityit-2), whereas the effect of review
10 JOURNAL OF TRAVEL & TOURISM MARKETING

quantity decays quickly in the next quarter (0.002, factors on ofine hotel popularity was examined.
p = 0.216 for ReviewQuantityit-1 and 0.000, An emerging research stream has started to look
p = 0.663 for ReviewQuantityit-2). at social media platforms featuring user-gener-
Column (2) in Table 3 presents the random ated content and consumer WOM. However, this
effects of our hotel popularity model. The stream of research mainly focuses on one-off
results of review factor estimations of random purchase items or short life cycle information
effects in Column (2) are also qualitatively goods (Chevalier & Mayzlin, 2006; Dellarocas
similar to those in Column (1). Hence, our et al., 2010; Duan et al., 2008; Liu, 2006),
results in Table 3 are quite robust with differ- neglecting experience products with a stable pro-
ent estimation methods. Although both results duct life cycle, such as hotels. Our ndings
of the xed effects estimation in Column (1) address industry concerns regarding the undi-
and the random effects estimations in Column sclosed mechanism of online consumer reviews
(2) are qualitatively similar, our proposed by shedding light on specic online consumer
Downloaded by [Archives & Bibliothques de l'ULB] at 21:21 11 August 2015

model using xed effects ts the data better review factors that would contribute to ofine
than random effects based on the Hausman test hotel popularity. We explore multiple aspects of
(Prob > Chi2 = 3.670, Prob > Chi2 = 0.042) consumer review information to examine their
(Greene, 2003). effects on ofine hotel popularity. By integrating
In order to check the robustness of our xed the multifaceted data we extracted from consu-
effects estimation results, we conduct several mer reviews in an empirical setting, this study
diagnostic tests. First, we use a BreuschPagan advances our understanding of the connection
test to check for heteroskedasticity. As shown in between the multidimensional facets of review
Table 3, the BreuschPagan test result content and business performance-based ofine
(Chi2 = 5.860, Prob > Chi2 = 0. 000) does not hotel popularity.
support the null hypothesis, suggesting there is Our analysis suggests that review quality and
preferred heteroskedasticity. This nding indi- quantity can signicantly elevate the ofine
cates that our use of xed-effects estimation is popularity of hotels, supporting H1 and H3. In
warranted. Second, we use the Lagram- particular, the inuence of the review quality
Multiplier test to check for the presence of serial can last for at least two quarters while the
correlation of the errors. Serial correlation effects of other online consumer review factors
causes the standard errors of the estimated such as review quantity and consistency taper
regression coefcients to be smaller than they off quickly within about one quarter. This nd-
actually are and inates the estimation R- ing partially supports H4. Additionally, hotels
square. Although it does appear to arise natu- frequently receive very diverse opinions from
rally in time-series data, one would want to look consumers about the quality of service and pro-
carefully at the data and the model specication ducts. Consistency of opinions, however, seems
before assuming that it is present (Stock & not be an inuencer of hotel popularity. This
Watson, 2007). The result of the Lagram- nding does not lend support to H2.
Multiplier test fails to reject the null hypothesis These ndings provide important implica-
(F = 0.214; Prob > F = 0.660), the conclusion tions for practicing managers and research in
being that the data does not have rst-order this eld. Specically, we demonstrated that an
autocorrelation. Our diagnostic tests show that increase in consumers review quality regarding
results in Table 3 are quite robust. their hotels increase ofine hotel popularity.
This relationship reinforces the importance of
product and service quality of hotels. Despite
5. CONCLUSION AND the popularity of social media in the hotel
IMPLICATIONS industry, the key to hotel popularity remains
outstanding service quality. Therefore, hotel
Consumer reviews are at the forefront of e- businesses should try to focus on continuous
commerce in the hospitality industry. In this improvement of service quality and encourage
paper the effect of online consumer review positive reviews from guests immediately after
Xie, Chen, and Wu 11

they check out or even during their stay. In TripAdvisor can provide more customized mar-
addition, hotel managers can provide incentives keting plans to hotel businesses planning to
to encourage customers to write reviews and increase their popularity among consumers.
particularly reward frequent reviewers. For example, instead of distributing resources
Developing a customer review system that con- to all review functions, social media websites
nects to the property management system such such as TripAdvisor can work with each hotel
as MICROS OPERA, and targeting customers property to personalize the hotel review page by
accordingly at the time they check in and out at highlighting primary review factors that may
an establishment might be a trend in the future. inuence the hotel popularity.
Furthermore, review quantity drives ofine
hotel popularity. A direct implication of this
nding is that a hotel can take advantage of 6. LIMITATIONS AND FUTURE
network effects in terms of online buzz. Hotels STUDIES
Downloaded by [Archives & Bibliothques de l'ULB] at 21:21 11 August 2015

should correctly identify their need of social


media given that the social media strategy is This study is not without its limitations.
most effective in producing an increase in While we have taken the rst step with regard
brand awareness through the sheer number of to several research directions in learning about
consumer reviews. Therefore, a hotel business hotel popularity, this study may prompt future
successful social media campaign can expand researchers to extend the research. First, we are
its consumer base by strategically encouraging condent about the quality of our dataset
more consumer reviews to be written speci- because it captured the actual consumer review
cally for that hotel. Hotel businesses can then behaviors online with a social media platform
target this customer group and solicit hotel for ve major markets in the state of Texas.
reviews from them in order to create more However, markets in Texas may not necessarily
social buzz through subsequent consumers represent the US hotel lodging industry.
who would see the previous reviews and ulti- Scholars are encouraged to collect more repre-
mately drive hotel popularity. sentative samples to justify the estimated effects
Finally, the recency of online reviews also of online consumer review factors. Second, we
matters. In this study it was found that, on have initiated efforts to analyze consumer
average, the quality of reviews in a recent per- review data to inform business decisions. With
iod (i.e. extending over two quarters) inuences the advanced of technology nowadays, real-time
ofine hotel popularity. Vast amounts of consu- data analytics are imperative in the hospitality
mer review data are generated, processed, and industry. Managers concerned with relative
presented on social media platforms on a daily popularity determinants can utilize our study
basis. Collecting timely data that matter to con- and exercise caution when employing social
sumer decision-making is relevant for hotel media strategies. We expect more future
businesses. For example, besides asking for research to emerge that can provide up-to-date
consumer reviews when guests check out or perspectives by integrating actual behavior data
during their stay, hotel businesses can use an collection and advanced data analytics techni-
automated guest response and communication ques in order to optimize business strategy in
system to send out an email to text message the hospitality industry.
guests asking for consumer reviews. This
email or text message should thank guests for
visiting and remind them that if they enjoyed NOTES
their stay it would be extremely valuable if they
clicked the link provided to write a review of 1. Fully automated parsing refers to the approach
used to collect information from a website. We developed
the hotel.
two crawlers using Ruby (1) to download automatically
Overall, our research reveals the importance the web pages of hotel reviews and other hotel information
of unobtrusive social media marketing. Based from TripAdvisor, and (2) to remove the HTML format-
on our ndings, social media platforms such as ting from the text and then convert it into an XML le that
12 JOURNAL OF TRAVEL & TOURISM MARKETING

separated the data into records (the review) and elds (the the craft beer industry. Journal of Management
data in each review) in an automated fashion using a Information Systems, 23(2), 149171. doi:10.2753/
precoded computer program on the local machine. We MIS0742-1222230207
used the crawlers to retrieve all available user-generated Coleman, A. (2006). Dictionary of psychology (2nd ed., p.
content information for the designated hotels. For each 688). Oxford: Oxford University Press.
hotel, we obtained all of the posted reviews. Each con- Compete, Inc. (2006). Embracing consumer buzz creates
sumer review was analyzed and selected review features measurement challenges for marketers. Retrieved April
were recorded. 26, 2015, from http://class.classmatandread.net/am1/
2. In order to account for the nonlinearities and to Buzz.pdf
smooth skewed distributions (Greene, 2003), we took the Cunningham, P., Smyth, B., Wu, G., & Greene, D. (2010).
logarithm of the hotel occupancy. Similar variable trans- Does TripAdvisor makes hotels better? University
formation has also been applied on the average daily rate. College Dublin School of Computer Science and
For purposes of brevity, the nonlinearity and skewness of Informatics Report. Retrieved January 1, 2013, from
each variable is not shown but is available upon request. http://www.csi.ucd.ie/content/does-tripadvisor-makes-
Downloaded by [Archives & Bibliothques de l'ULB] at 21:21 11 August 2015

hotels-better
De Maeyer, P. (2012). Impact of online consumer reviews
REFERENCES on sales and price strategies: A review and directions
for future research. Journal of Product & Brand
Ansari, A., & Mela, C. F. (2003). E-customization. Management, 21(2), 132139. doi:10.1108/
Journal of Marketing Research, 40(2), 131145. 10610421211215599
doi:10.1509/jmkr.40.2.131.19224 Dellarocas, C., Gao, G., & Narayan, R. (2010). Are con-
Aria. (2015). Boosting the bottom line: 3 Years with sumers more likely to contribute online reviews for hit
Denton, TX. Aria Case Study Series. Retrieved March or niche products? Journal of Management Information
17, 2015, from http://www.ariaagency.com/work/boost Systems, 27(2), 127158. doi:10.2753/MIS0742-
ing-the-tourism-bottom-line-3-years-of-hotel-occu 1222270204
pancy-tax-revpar-average-daily-rates-and-average-occu Dellarocas, C., Zhang, X. M., & Awad, N. F. (2007).
pancy-in-denton-texas Exploring the value of online product reviews in forecast-
Becker, G. S. (1976). The economic approach to human ing sales: The case of motion pictures. Journal of
behavior. Chicago, IL: University of Chicago Press. Interactive Marketing, 21(4), 2345. doi:10.1002/
Chen, P. Y., Wu, S. Y., & Yoon, J. (2004, December dir.20087
1215). The impact of online recommendations and Drze, X., & Zufryden, F. (2004). Measurement of online
consumer feedback on sales. Proceedings of the visibility and its impact on Internet trafc. Journal of
International Conference on Information Systems, Interactive Marketing, 18(1), 2037. doi:10.1002/dir.10072
ICIS, Washington, DC (pp. 711724). Duan, W., Gu, B., & Whinston, A. B. (2008). The
Chen, Y., & Xie, J. (2008). Online consumer review: dynamics of online word-of-mouth and product sales
Word-of-mouth as a new element of marketing com- An empirical investigation of the movie industry.
munication mix. Management Science, 54(3), 477491. Journal of Retailing, 84(2), 233242. doi:10.1016/j.
doi:10.1287/mnsc.1070.0810 jretai.2008.04.005
Chevalier, J. A., & Mayzlin, D. (2006). The effect of word Ghose, A., Ipeirotis, P. G., & Li, B. (2012). Designing rank-
of mouth on sales: Online book reviews. Journal of ing systems for hotels on travel search engines by mining
Marketing Research, 43(3), 345354. doi:10.1509/ user-generated and crowd-sourced content. Marketing
jmkr.43.3.345 Science, 31(3), 493520. doi:10.1287/mksc.1110.0700
Chu, S. C. (2009). Determinants of consumer engagement Godes, D., & Mayzlin, D. (2004). Using online conversations
in electronic word-of-mouth in social networking sites to study word-of-mouth communication. Marketing
(Doctoral Dissertation). University of Texas, Austin. Science, 23(4), 545560. doi:10.1287/mksc.1040.0071
Chu, S.-C., & Kim, Y. (2011). Determinants of consumer Godes, D., & Mayzlin, D. (2009). Firm-created word-of-
engagement in electronic word-of-mouth (eWOM) in mouth communication: Evidence from a eld test.
social networking sites. International Journal of Marketing Science, 28(4), 721739. doi:10.1287/
Advertising, 30(1), 4775. mksc.1080.0444
Clark, C. R., Doraszelski, U., & Draganska, M. (2009). Greene, W. H. (2003). Econometric analysis (5th ed.).
The effect of advertising on brand awareness and per- Upper Saddle River, NJ: Prentice Hall.
ceived quality: An empirical investigation using panel Gretzel, U., & Yoo, K. H. (2008). Use and impact of
data. Quantitative Marketing and Economics, 7, 207 online travel reviews. In D. P. OConnor, D. W.
236. doi:10.1007/s11129-009-9066-z Hpken, & D. U. Gretzel (Eds.), Information and com-
Clemons, E. K., Gao, G., & Hitt, L. M. (2006). When munication technologies in tourism (pp. 3546).
online reviews meet hyper-differentiation: A study of Vienna: Springer.
Xie, Chen, and Wu 13

Hu, Y., & Li, X. (2011). Context-dependent product eva- Nelson, P. (1970). Information and consumer behavior.
luations: An empirical analysis of Internet book Journal of Political Economy, 78(2), 311329.
reviews. Journal of Interactive Marketing, 25(3), Pine, J. B., & Gilmore, J. H. (1998). Welcome to the experi-
123133. doi:10.1016/j.intmar.2010.10.001 ence economy. Harvard Business Review, 76(4), 97106.
Jeacle, I., & Carter, C. (2011). In TripAdvisor we trust: Rajan, R. (2014). How to be number one on TripAdvisor.
Rankings, calculative regimes and abstract systems. RethinkHotels. Retrieved July 1, 2014, from http://
Accounting, Organizations and Society, 36(45), www.rethinkhotels.com/how-to-be-number-one-on-tri
293309. doi:10.1016/j.aos.2011.04.002 padvisor/#comments
Jeong, M., & Jeon, M. M. (2008). Customer reviews of Ratnasingham, P. (1998). The importance of trust in elec-
hotel experiences through Consumer Generated Media tronic commerce. Internet Research: Electronic
(CGM). Journal of Hospitality Marketing & Networking Applications and Policy, 8(4), 313321.
Management, 17(1), 121138. doi:10.1080/ doi:10.1108/10662249810231050
10507050801978265 Sawhney, M. S., & Eliashberg, J. (1996). A parsimonious
Jiang, B. J., & Chen, P. Y. (2007). An economic analysis of model for forecasting gross box-ofce revenues of
Downloaded by [Archives & Bibliothques de l'ULB] at 21:21 11 August 2015

online product reviews and ratings (Working Paper). motion pictures. Marketing Science, 15(2), 113131.
Pittsburgh, PA: Tepper School of Business, Carnegie doi:10.1287/mksc.15.2.113
Mellon University. Schlosser, A. (2005). Posting versus lurking:
Kirby, C. (2000, January). Everyones a critic: Web sites Communicating in a multiple audience context.
hope online reviews of products lead to online buying. Journal of Consumer Research, 32(2), 260265.
San Francisco Chronicle, 22, E1. doi:10.1086/jcr.2005.32.issue-2
Lee, Y.-J., & Tan, Y. (2013). Effects of different types of Stephen, A. T., & Galak, J. (2012). The effects of tradi-
free trials and ratings in sampling of consumer soft- tional and social earned media on sales: A study of a
ware: An empirical study. Journal of Management microlending marketplace. Journal of Marketing
Information Systems, 30(3), 213246. doi:10.2753/ Research, 49(5), 624639. doi:10.1509/jmr.09.0401
MIS0742-1222300308 Stock, M., & Watson, J. (2007). Introduction to econo-
Lewis, D., & Bridger, D. (2001). The soul of the new metrics (2nd ed.). Addison-Wesley Series in
consumer: Authenticity What we buy and why in Economics. Addison-Wesley.
the new economy. Clerkenwell: Nicholas Brealey Sun, M. (2012). How does the variance of product ratings
Publishing. matter? Management Science, 58(4), 696707.
Li, X., & Hitt, L. (2008). Self-selection and information doi:10.1287/mnsc.1110.1458
role of online product reviews. Information Systems Sutherland, S. (1995). Macmillan dictionary of psychol-
Research, 19(4), 456474. doi:10.1287/isre.1070.0154 ogy. London: Macmillan Press.
Litvin, S. W., Goldsmith, R. E., & Pan, B. (2008). TripAdvisor. (2015). TripAdvisor for business. Retrieved
Electronic word-of-mouth in hospitality and tourism August 2, 2015, from http://www.tripadvisor.com/
management. Tourism Management, 29(3), 458468. Owners-t2
doi:10.1016/j.tourman.2007.05.011 Xie, K. L., Zhang, Z., & Zhang, Z. (2014). The business
Liu, Y. (2006). Word of mouth for movies: Its dynamics value of online consumer reviews and management
and impact on box ofce revenue. Journal of response to hotel performance. International Journal
Marketing, 70(3), 7489. doi:10.1509/jmkg.70.3.74 of Hospitality Management, 43, 112. doi:10.1016/j.
Luca, M. (2011). Reviews, reputation, and revenue: The ijhm.2014.07.007
case of Yelp.com (Working Paper No. 12-016). Zhang, L., Pan, B., Smith, W. W., & Li, X. (2009). An
Cambridge, MA: Harvard Business School. exploratory study of travelers use of online reviews
Moe, W., & Trusov, M. (2011). The value of social and recommendations: A qualitative approach.
dynamics in online product ratings forums. Journal of Information Technology & Tourism, 11(2), 157167.
Marketing Research, 48(3), 444456. doi:10.1509/ doi:10.3727/109830509789994775
jmkr.48.3.444 Zhu, F., & Zhang, X. (2010). Impact of online consumer
Moe, W. W., & Fader, P. S. (2001). Modeling hedonic reviews on sales: The moderating role of product and
portfolio products: A joint segmentation analysis of consumer characteristics. Journal of Marketing, 74(2),
music compact disc sales. Journal of Marketing 133148. doi:10.1509/jmkg.74.2.133
Research, 38(3), 376385. doi:10.1509/
jmkr.38.3.376.18866 SUBMITTED: August 29, 2014
Murphy, J., Hofacker, C., & Mizerski, R. (2006).
FINAL REVISION SUBMITTED:
Primacy and recency effects on clicking behavior.
Journal of Computer-Mediated Communication, 11 March 17, 2015
(2), 522535. Article 7. doi:10.1111/jcmc.2006.11. ACCEPTED: May 8, 2015
issue-2 REFEREED ANONYMOUSLY

Вам также может понравиться