Does Interest Alignment Between Hotels and Online Travel Agencies Intensify Review Manipulations? (Industrial Management & Data Systems)

Summary

This is the pre-published version of a research paper by Chen, Tang, Li and Fan in Industrial Management & Data Systems, 121(6), 1435-1456 (2021), asking whether closer interest alignment between online travel agencies and hotels intensifies review manipulation. Using panel data from a Chinese online travel agency, Ctrip, and the travel search engine Qunar.com, the authors apply a matching-based difference-in-difference approach. They find that the agency’s partner hotels had abnormally higher ratings than matched non-partner hotels when benchmarked against their search-engine ratings, and lower rating variance, which they say was likely caused by deleting extremely negative reviews. The authors conclude that this manipulation deteriorates hotels’ sales performance by damaging customer trust, more so for large-scale, expensive and prestigious hotels.

Summary drafted by a model from the document's text below and checked by script against that text before publication. It is a navigation aid, not a reading of what the document proves. Where AI is used

Full text

This is the Pre-Published Version.
© Emerald Publishing Limited. This AAM is provided for your own personal use only. It may not be
used for resale, reprinting, systematic distribution, emailing, or for any other commercial purpose
without the permission of the publisher.
The following publication Chen, Y., Tang, T., Li, Y., & Fan, D. (2021). Does interest alignment
between hotels and online travel agencies intensify review manipulations? Industrial Management
& Data Systems, 121(6), 1435-1456 and is available at https://doi.org/10.1108/
IMDS-11-2020-0657.

Does interest alignment between hotels and
online travel agencies intensify review
manipulations?

Abstract

Purpose - This study examines whether a higher interest alignment between online travel
agencies (hosting platform) and hotels (business owners) will intensify review manipula-
tion activities.
Design/methodology/approach - With a panel dataset collected from a Chinese online
travel agency and a travel search engine, we develop a matching-based difference-in-
difference approach to examine the presence of partnership-intensiﬁed review manipula-
tion.
Findings - We ﬁnd that the ratings of agency’s partner hotels (with a higher interest
alignment) is abnormally higher than those of matched non-partner hotels (with a lower
interest alignment), after benchmarked with their ratings on the search engine (without a
partnership business model). Further, our ananlysis results indicate that this partnership-
intensiﬁed manipulation deteriorates the hotel’s sales performance because of damaged
customer trust and satisfaction.
Originality/value - Previous studies implicitly assume that review manipulator is inde-
pedent from the hosting platform. This is the ﬁrst study examining the role of the hosting
platform in review manipulaitons.
Keywords Review manipulation; Online travel agency; Partnership model; Interest align-
ment; Hotel performance
Paper type Research paper

1. Introduction

Hotels are increasingly relying on the internet platforms provided by online travel agen-
cies (OTAs) as important sales channels. OTAs have also changed the way that consumers
acquire related information about target hotels. Most OTAs host a platform for consumers

1
Does interest alignment intensify review manipulations? 2

to share their experiences by writing online reviews. Literature has substantially inves-
tigated the effects of these reviews on consumer booking behavior (Casado-Díaz et al.
2020, Filieri & McLeay 2014, Zhao et al. 2015) and hotel business performance (Öğüt &
Onur Taş 2012, Xu et al. 2019, Ye et al. 2009). These studies generally concluded that
positive reviews can lead to favorable business outcomes.
The positive effect of online reviews is based on the assumption that the reviews ac-
curately reﬂect consumption experiences. However, recently, there is a growing concern
over review manipulation (Hu et al. 2011, 2012, Lee et al. 2018, Luca & Zervas 2016,
Mayzlin et al. 2014, Zhuang et al. 2018). Review manipulation is deﬁned by May-
zlin et al. (2014) as “reviewers with a material interest in consumer’s purchase decisions
may post reviews that are designed to inﬂuence consumers and resemble the reviews of
disinterested consumers”. Later, Zhuang et al. (2018) point out “Despite the efforts by
e-marketplace operators and review platforms to ﬁlter out fake reviews and the strength-
ening of guidelines and enforcement in various countries, manipulation of online reviews
has persisted and taken more varied forms”. Therefore, in this study, we extend the defini-
tion by Mayzlin et al. (2014) as “anyone with a material interest in consumer’s purchase
decisions may post, delete or modify reviews that are designed to inﬂuence consumers and
resemble the reviews of disinterested consumers”. Review manipulation has been found
on TripAdvisor (Mayzlin et al. 2014, Zhuang et al. 2018), Yelp (Luca & Zervas 2016),
and Twitter (Lee et al. 2018), and may mislead consumers’ decision-making and impede
review usefulness.
Previous studies implicitly assumed that review manipulators (e.g., hotels or their
rivals) were independent from the hosting platform. They have largely overlooked the
role of OTAs as the intermediary of hotel bookings and hosting platform for consumer
reviews nowadays. On the contrary, this research aimed to explore the role of the hosting
platform. In line with Mayzlin et al. (2014), our study does not attempt to distinguish
whether any review is fake or not, instead we empirically examine whether the operating
model of OTAs can be systematically intensifying review manipulation activities.
We drew from agency logic to explore this question. In the service supply chain of
the hotel industry, consumers (the principal) delegate hospitality activities to a hotel (the
agent) (Pavlou et al. 2007). From an agency perspective, such delegations facilitate the
agent to conduct opportunistic behaviors and gain self-interest, while such behaviors may
undermine the beneﬁt of the principal (Jensen & Meckling 1976). The agency problem
is primarily caused by a lack of interest alignment between agents and principals. We
conceptualize review manipulation as a type of opportunistic behavior, and argue that the
interests of OTAs are tied with hotels (through sales commissions), not with consumers.
Thus, the higher interest alignment between hotels and OTAs may intensify review ma-
nipulation and thus undermine consumers’ beneﬁt. Speciﬁcally, we aimed to answer three
research questions for both research and practice purposes:

1. Does a closer partnership (higher interest alignment) between hotels and OTAs in-
tensify manipulations in the hotel’s online review rating?

2. How does such partnership-intensiﬁed review manipulation affect hotels’ online
sales performance?
Does interest alignment intensify review manipulations? 3

3. Which types of hotels are more affected by the partnership-intensiﬁed review ma-
nipulation?

The blooming of Chinese online OTA markets provided a fertile ground to answer our
research questions. Using data collected from major OTAs in China, with a matching-
based difference-in-difference (DID) design, we capture an abnormally higher consumer
review rating in OTAs’ partner hotels (with high interest alignment) than the matched
non-partner hotels (with low interest alignment), after benchmarking to their ratings on
a major Chinese travel search engine, Qunar.com. This observatoin indicates that the
partnering business model of OTAs is likely intensifying review manipulation activities.
Further, we conduct a regression analysis for panel data of these hotels over seven weeks.
The results indicate that this partnership-intensiﬁed manipulation deteriorates the hotel’s
sales performance on the platform because of damaged customer trust and satisfaction.
Moreover, the performance deterioration is more salient for large-scale, expensive, and
prestigious hotels as a result of doubling the expectancy-violation effects, which implies
that people react more strongly to greater violation of their previous expectation.

2. Theoretical background and hypotheses
2.1. Online reviews and review manipulation
Online reviews play an increasingly critical role in consumers’ purchasing decisions. Pre-
vious studies typically reported a positive correlation between online review ratings and
business performance in various industries, such as bookstores, hotels and restaurants
(Chevalier & Mayzlin 2006, Mauri & Minazzi 2013, Öğüt & Onur Taş 2012, Ye et al.
2009, 2011). Therefore, stakeholders are tempted to promote the online reviews of their
products or services to improve their business. Based on a qualitative study with 20 hotel
managers in southern Sweden, Gössling et al. (2018) concluded that there exists a wide
range of review manipulation strategies in practice. Moreover, this paper postulated that,
as hotel managers compete increasingly over online reputation, they will eventually ﬁnd
that engaging in manipulation is the most rational choice.
There are several challenges in detecting review manipulation and quantifying their
effect. First, because of social (un)desirability, it is almost impossible to directly observe
review manipulation. However, empirical and technical methods were developed recently
to capture review manipulation (Hu et al. 2011, 2012, Lee et al. 2018, Luca & Zervas
2016, Mayzlin et al. 2014, Zhuang et al. 2018). By examining the difference in review
ratings for the same hotel on Expedia.com and TripAdvisor.com, Mayzlin et al. (2014) and
Zhuang et al. (2018) captured empirical evidence about review manipulation on a travel
review site - TripAdvisor. Luca & Zervas (2016) studied reviews on the review website
Yelp and concluded that suspicious reviews has grown signiﬁcantly over time. Lee et al.
(2018) studied movie Tweets before and after the movie’s release date and found that
the proportion of positive Tweets exhibited a signiﬁcant drop on the movie’s release day.
Based on these observations, they concluded that the movie industry might be actively
managing online sentiment in a strategic manner.
Second, the effect of review manipulation is not static, but evolving over time and
Does interest alignment intensify review manipulations? 4

dependent on speciﬁc contexts. Using data from Expedia and TripAdvior, Zhuang et al.
(2018) argued that review manipulating has an initially positive effect on sales and there-
after a negative effect as manipulation becomes more frequent and intensive. Some stud-
ies (Jin Ma & Lee 2014, Reimer & Benkenstein 2016) also show that as manipulation
becomes more prevailing and consumers become more vigilant, suspicious reviews can
have a greater adverse effect on consumers’ perceptions.
However, there are several important research gaps in the literature to be addressed.
First, most of the studies discussed above implicitly assumed that hotels are indepen-
dent of the review hosting platform, while the role of OTAs in review manipulation has
been largely overlooked. The case presented by Gössling et al. (2018) implies that some
practices of OTAs may facilitate review manipulation of their partner hotels: the partner-
ship between Shangri-La and TripAdvisor has signiﬁcantly increased the number of page
views and consumer reviews, hotel ratings, and market share. In addition, a recent study
by Hunold et al. (2020), using data collected from Booking.com, Expedia, and Kayak,
revealed that OTAs alter their search results to discipline hotels for price differences on
competing channels, and therefore reduce search quality for consumers. Thus, our re-
search applies a fresh perspective to systematically and quantitatively explore the role of
the hosting platform in review manipulation.
In addition, our research extends existing methods of detecting review manipulation
(examining either the average or ratio of review ratings (Hu et al. 2011, 2012, Lee et al.
2018, Luca & Zervas 2016, Mayzlin et al. 2014, Zhuang et al. 2018)). The robustness
check of this study indicates that the variance of customer ratings can also help detect
review manipulation. Finally, instead of using a proxy for online sales (Ye et al. 2011,
Zhuang et al. 2018), the number of actual bookings collected from the OTAs allows us
to investigate the direct effect of review manipulation on online business performance. In
the following section, we draw from agency theory to conceptualize the role of OTAs and
explain why manipulations can be intensiﬁed in OTA partnered hotels.

2.2. Partnership-intensiﬁed review manipulation on OTAs: an agency
perspective
Agency theory investigates occasions in which one party (the principal) delegates author-
ity to another party (the agent) (Jensen & Meckling 1976). In principal-agent relations,
both the principal and agent are self-interested and seek to maximize their own proﬁts
by individually interpreting their contract relations (Fleisher 1991). This theory has been
widely applied to explain the opportunistic behavior conducted by either principals or
agents motivated by their own beneﬁts (Bosse & Phillips 2016).
The literature has discussed the agency problems that arise in a business-to-business
(B2B) context, where ﬁrms delegate sales, marketing, and advertising activities to agents.
The agents may take advantage of the delegating ﬁrm’s resources (e.g., brand equity and
customer loyalty) to gain self-interest with guile (Zhang et al. 2015). Another smaller
stream of literature uses the agency theory framework to understand the relations between
product/service providers and consumers (Pavlou et al. 2007, Singh & Sirdeshmukh 2000,
Tan & Lee 2015). In this business-to-consumer (B2C) context, the buyer (consumer)
Does interest alignment intensify review manipulations? 5

delegates responsibility to a seller (business) to deliver products/services, and the seller
acts on behalf of the buyer (Pavlou et al. 2007). In this relation, out of consideration for
short-term revenue-related goal, the business (agent) may be opportunistic and provide
misleading information to induce consumers (principal) to purchase, thereby causing the
agency problem.
Traditional hospitality transactions ﬁt the B2C context. The hotel and consumer are
opposing forces and have diverged interests: hotels seek the highest revenue, while con-
sumers seek the maximum surplus. As a result of such diverged interests and information
asymmetry, hotels may conduct opportunistic behaviors aimed at reaching a deal with
a shorter time and higher price. Review manipulation is one typical example of oppor-
tunistic behavior that provides false information to consumers and affects their buying
decisions.
When OTAs are involved as an intermediary in hotel booking and a hosting plat-
form for customer reviews, would review manipulation be intensiﬁed or inhibited? Our
hypothesized answer to this question is that the manipulations would be intensiﬁed as
the interest alignment between the hotel and OTA becomes stronger. In an outcome-
based contract, ﬁrms reward their sales forces based on measured performance outcomes
(Ekanayake 2004), such as sales volume. Such risk- and revenue-sharing mechanisms are
often used in contracts as incentives and reward strategies to stimulate short-term sales
(Norrman 2008). For major OTAs, such as Ctrip and Fliggy, two types of hotels are listed
on their platforms: partner hotels and non-partner hotels. Partner hotels typically pay a
higher commission for each room sale. For example, to become a strategic partner hotel
of Ctrip, hotels must pay a commission as high as 15% to 25%, compared with 8% to
10% for typical non-partner hotels (Liu 2019). This indicates a higher interest alignment
between OTAs and their partner hotels, compared with non-partner hotels. Thus, OTAs
are incentivized by higher commissions from their partner hotels, and their partner hotels
try to make up for higher commissions with more room sales. Intuitively, higher hotel
review ratings will lead to more room sales. Therefore, both the OTAs and their partner
hotels are tempted to boost or help boost the partner hotels’ review ratings. By doing so,
they expect to promote these partner hotels’ room sales and ultimately increase their own
beneﬁts.
Thus, the higher interest alignment (sales commission) between OTAs and their part-
ner hotels likely tempts OTAs to loosen their vigilance in the review manipulation of these
hotels. Their partner hotels may be emboldened because they believe that the partnered
OTA will be more lenient toward their behavior, which leads to an additional intensity
of review manipulation. Based on this discussion, we postulate that partnership relations
will intensify review manipulation:
Hypothesis 1. A higher level of interest alignment between hotels and OTAs is associ-
ated with a higher review manipulation for the hotels (i.e., partnership-intensiﬁed review
manipulation).
Does interest alignment intensify review manipulations? 6

2.3. Effect of partnership-intensiﬁed review manipulation
Review manipulation is considered an opportunistic and unethical business practice
because it generates false or misleading information that interferes with consumers’
decision-making (Jin Ma & Lee 2014). Studies in consumer behavior reveal that unethi-
cal business practices greatly affect consumers’ evaluation of the company and purchase
intentions (Brown & Dacin 1997, Folkes & Kamins 1999, Jin Ma & Lee 2014, Mohr &
Webb 2005, Sen & Bhattacharya 2001). A survey conducted by Brown & Dacin (1997)
demonstrated that consumers’ knowledge about a ﬁrm can inﬂuence their beliefs and at-
titudes toward the ﬁrm’s products. Experiments of Folkes & Kamins (1999) showed that
unethical behavior elicits consumers’ negative attitude toward a company. Later, a study
of Mohr & Webb (2005) also conﬁrmed that a company’s lack of social responsibility
will lead to consumers’ lower valuations and purchase intentions. Moreover, customers
are sensitive and diagnostic to potential negative practices by ﬁrms (Sen & Bhattacharya
2001). These perceptions can seriously damage customer trust, which consequently un-
dermines buying intention.
When hotels and platforms are tempted to boost review ratings, they may also cause
customer suspicion. Prior research has found that consumers evaluate the credibility of
online reviews in terms of various factors, such as grammar and mechanics (Ketron 2017)
and reviewer identity (Akhtar et al. 2019). The power of review valence highly depends
on consumers’ perceptions of review credibility (Filieri 2015, 2016, Filieri et al. 2018,
Jin Ma & Lee 2014, Reimer & Benkenstein 2016). Moreover, Ahmad & Sun (2018)
found that psychological discomfort caused by distrust with reviews also strongly affects
consumer purchase behavior. Experiment results of Zhuang et al. (2018) revealed that
customer suspicion lead to a decrease in willingness to book. Apart from this, industrial
data (Womply Research 2020), also shows that lodging places with a star rating between
3.5 and 4.5 earned more than any other ratings group, including the 5 rated. This indicates
that consumers’ trust can decrease when the rating is too positive.
Meanwhile, customers’ perceived satisfaction is inﬂuenced by their perception and
expectation. A study of Mauri & Minazzi (2013) showed a positive correlation between
the valence of the review and the hotel service expectation. If customers do not sense a re-
view manipulation and make a booking, this manipulated rating can inﬂate the customers’
expectations. This may increase the likelihood of having an expectation-perception dis-
crepancy that disconﬁrms the pre-purchase expectation. The disconﬁrmation leads to
customer dissatisfaction (Anderson & Sullivan 1993), which will in turn be reﬂected in
customers’ negative post-purchase reviews (online or ofﬂine) and behaviors, such as re-
luctance to repurchase or complain to others.
Based on the discussion above, we argue that partnership-intensiﬁed review manip-
ulation can undermine customer trust if customers identify the manipulation. Moreover,
inﬂated ratings can expand the gap between customer expectation and perception, which
worsens perceived service satisfaction. Both mechanisms will deteriorate the hotel’s sub-
sequent performance. Therefore, we posit that:
Hypothesis 2. Partnership-intensiﬁed review manipulation deteriorates hotels’ subse-
quent online business performance.
Does interest alignment intensify review manipulations? 7

2.4. Moderating role of hotel characteristics
H2 hypothesizes that partnership-intensiﬁed review manipulation worsen hotel perfor-
mance because of damaged customer trust and satisfaction. It is essential to explore the
circumstances in which a hotel is more vulnerable or resilient to this deteriorating effect.
We further postulate that the hotel’s reputation could be a liability, that makes the hotel
more vulnerable to partnership-intensiﬁed review manipulation.
A ﬁrm’s reputation creates consumer expectation about its product/service quality
(Shapiro 1983). Organizational theorists ﬁnd that highly reputed ﬁrms suffer more market
penalties for product defects because expectations about product quality are more likely
to be violated (Rhee & Haunschild 2006). In the hotel industry, customer dissatisfaction is
considered a defect in the service, which can be caused by discrepancy between expecta-
tion and perception. If online reviews are boosted for OTAs’ partner hotels, these hotels’
guests may encounter a mismatch between the service quality and review rating, which
results in an expectancy violation. The expectancy-violation effect in the communication
literature points out that people react more strongly to violations under circumstances in
which such violations are perceived to be less likely to occur (Bond et al. 1992, Bur-
goon & Le Poire 1993). This indicates that customers of high-reputation hotels may react
more strongly because of reputational liability (Rhee & Haunschild 2006). Speciﬁcally,
hotels’ reputations can be viewed as an implicit promise to potential customers to of-
fer good services, with a quality level commensurate with their word-of-mouth (Rhee &
Haunschild 2006). This implies that the better the hotel’s reputation, the greater the ex-
tent to which consumers expect the hotel’s service to be consistent with their online review
rating. When service quality does not match review rating, customers of high-reputation
hotels respond more negatively toward the hotel. Similarly, marketing research has shown
that different market segments react differently to expectancy violations (Heath & Chat-
terjee 1995). Experiments of Simonson & Tversky (1992) found that losing quality is
generally more aversive in the higher-end market than the lower-end market.
In addition to creating expectations about product/service quality, a ﬁrm’s reputation
also generates consumer expectations about ﬁrms’ behavior (Shapiro 1983). A good rep-
utation is often associated with trust (Doney & Cannon 1997, Walsh & Beatty 2007) and
integrity (Bick et al. 2003). Review manipulation is considered an unethical business prac-
tice that is less likely to occur with high-reputation hotels. If customers became aware of
review manipulation after their stay, customers of high-reputation hotels may react more
strongly because of the expectancy-violation effect discussed above. Next, we use hotel
size, price, and prestige as proxy indicators for reputation, and examine their moderating
effects on the manipulation-performance relation.
First, scholars associate organization size with organization reputation because larger
organizations enjoy greater name recognition than do smaller organizations (Williams &
Barrett 2000). Consumers tend to perceive that a large hotel is more trustworthy and
hospitable than a small hotel (Arifﬁn & Maghzi 2012). Thus, consumers may be more
disappointed if they learn that large hotels are manipulating their online reviews:
Hypothesis 3. The relation between partnership-intensiﬁed review manipulation and
hotel performance is more negative when the hotel size is larger.
Does interest alignment intensify review manipulations? 8

Second, the hotel price should be associated with the hotel reputation expected by
customers because such price premiums should be a signal of good service quality. Ya-
couel & Fleischer (2012) found that reputable hotels enjoy a price premium. Consumer
trust may be more seriously undermined if they are cheated by a hotel that is expensive.
In addition, pre-purchase price perceptions are correlated with pre-purchase expectations
of service quality (Voss et al. 1998). The raised expectations of higher hotel price would
be more easily violated because of review manipulation, leading to undermined customer
satisfaction and subsequent hotel sales:
Hypothesis 4. The relation between partnership-intensiﬁed review manipulation and
hotel performance is more negative when the hotel price is higher.
Finally, the star rating of a hotel can be viewed as a recognition of a third party (e.g.,
governments and certiﬁcation bodies) of the service quality of the hotel. Thus, the star
rating is considered a reputation-based quality signal for consumers (Abrate et al. 2011).
Consumers expect a better experience for their stay at a ﬁve-star hotel than at a three-star
hotel, which increases the likelihood of expectancy violation if the hotel’s online review is
manipulated. Consumers would be more disappointed if they became aware of the review
manipulation of a high-star-rated hotel, leading to a stronger deteriorating effect on future
sales:
Hypothesis 5. The relation between partnership-intensiﬁed review manipulation and
hotel performance is more negative when the hotel has a higher star rating.
澳Figure 1 contains a summary of our hypotheses and their directionality.

濣濴瀅瀇瀁濸瀅瀆濻濼瀃激 濛瀂瀇濸濿瀠瀆澳濢瀁濿濼瀁濸澳
濜瀁瀇濸瀅濸瀆瀇澳濔濿濼濺瀁瀀濸瀁瀇澳 濛濄澳 濛濅澳
濼瀁瀇濸瀁瀆濼濹濼濸濷澳濥濸瀉濼濸瀊澳 濕瀈瀆濼瀁濸瀆瀆澳
濵濸瀇瀊濸濸瀁澳濛瀂瀇濸濿瀆澳濴瀁濷澳
澾澳 濠濴瀁濼瀃瀈濿濴瀇濼瀂瀁澳 激澳 濣濸瀅濹瀂瀅瀀濴瀁濶濸澳
濢濧濔瀆澳
濔濿濼濺瀁瀀濸瀁瀇澳濵濸瀇瀊濸濸瀁澳 濵濸瀇瀊濸濸瀁澳濛瀂瀇濸濿瀆澳濴瀁濷澳
濛瀂瀇濸濿瀆澳濴瀁濷澳濢濧濔瀆澳 濢濧濔瀆澳
激澳

濣瀅瀂瀋瀌澳濩濴瀅濼濴濵濿濸瀆濍澳
激澳濦濼瀍濸澳澻濛濆澼澳
濛瀂瀇濸濿瀠瀆澳濥濸瀃瀈瀇濴瀇濼瀂瀁澳 澳 激澳濣瀅濼濶濸澳澻濛濇澼澳
激澳濦瀇濴瀅澳濥濴瀇濼瀁濺澳澻濛濈澼澳
澳澳

F IGURE 1: Research Model with Hypothese

3. Data and measurements
To test the hypotheses above, we collected data from one major online travel agency,
Ctrip.com, and one major travel search engine, Qunar.com. Ctrip is the largest online
travel agency in China, for which commission is a major part of its hotel booking service.
Does interest alignment intensify review manipulations? 9

On Ctrip, only users who have booked a hotel through this platform within one year
can post reviews. In addition, after posting reviews, users are not allowed to modify or
delete them. In contrast, Qunar is a major Chinese travel search engine focusing on price
comparison. Its revenue from hotel searching services mainly derives from advertising.
In addition, for robustness checks, we also collected data from another major OTA in
China, Fliggy.com, whose business model of hotel booking services is very similar to that
of Ctrip.

3.1. Data collection
Our research design required comparison between an OTA (Ctrip/Fliggy) and a search
engine (Qunar). The data collection process started by selecting the 10 most popular Chi-
nese cities for tourism listed on Ctrip: Beijing, Shanghai, Guangzhou, Shenzhen, Nanjing,
Hangzhou, Chengdu, Xiamen, Qingdao and Sanya. We then obtained a list of 21,103 ho-
tels listed on all three websites (Ctrip, Qunar, and Fliggy) on January 3, 2019. This
procedure excluded the hotels that were exclusively represented by a sole agent.
We collected the hotels’ initial overall ratings, individual ratings from each user, and
basic information. We then deleted the inactive hotels with an overall rating of less than 1
or a total number of reviews less than 10. This procedure removed hotels with an abnormal
status or a number of reviews that did not provide sufﬁcient reference information for
customers. This left us with a total of 15,887 hotels.
Given that hotel sales on the OTA are not publicly available, we developed a crawler
using Python to routinely monitor the Ctrip website to identify the latest sales condition.
If a customer accessed the webpage of a hotel, Ctrip would notify with a pop-up window
with the information ‘Last booking: X minutes ago’. We accessed all the hotels’ Ctrip
pages every minute and tracked whether the information was reset to ‘Last booking: one
minute ago’. If so, we considered that the hotel had one sale in the last minute. In addition,
we tracked the change of time-variant factors, including the individual hotel ratings and
lowest price offering. This data collection process ran from January 5 to February 22,
2019. We then organized the data and developed a seven-week panel dataset.

3.2. Manipulation measurement: a DID approach
To measure the partnership-intensiﬁed review manipulation, we adopted a matching-
based DID design to catch and quantify the rating difference between the matched partner
and non-partner hotels. We ﬁrst classiﬁed the hotels into two groups: partner hotels and
non-partner hotels (based on each hotel’s category marked as "partner hotel" or "non-
partner hotel" on Ctrip.com). Our search found 2,838 partner hotels and 13,049 non-
partner hotels from Ctrip. Following this, we employed a matching method to control
the confounding effects of hotels’ observable characteristics in the treated (partner hotels)
and control (non-partner hotels) groups. For example, ﬁve-star hotels may simultaneously
have a higher likelihood of being a partner hotel and a higher review rating. Speciﬁcally,
we matched each partner hotel with a group of non-partner hotels based on the following
criteria:

1. the hotels were in the same city
Does interest alignment intensify review manipulations? 10

2. the hotels had the same star rating (e.g., four-star or ﬁve-star hotels)

3. the non-partner hotel had an offered price within ±10% of the price of the partner
hotel.

Criterion 1 ensured that the matched hotels were in the same city, which mitigated the
variation caused by geographical locations, such as service level standards . Criterion 2
ensured that the matched hotels had the same star rating. The hotels with the same star
rating had similar physical attributes and quality signals. Criterion 3 controlled the price
variation in the matched groups because price shapes consumer expectations, which could
affect evaluation of service quality (Kopalle & Lehmann 2006). There are two approaches
to control hotel price: (1) select one non-partner hotel for each partner hotel with the
closest offered price (one-to-one nearest-neighborhood matching) or (2) select a group
of non-partner hotels whose prices are within a certain range of the partner hotel’s price
(one-to-many matching). A common complaint about nearest-neighborhood matching is
that it can suffer from discarding a large number of observations and lead to reduced
power (Stuart 2010). Thus, we used the one-to-many approach, which selected non-
partner hotels whose prices were within ±10% of the partner hotel. We considered that the
selected non-partner hotels that met all three criteria should have similar characteristics to
the matched partner hotel in terms of location, star rating, price and customer patterns.
The ﬁrst difference in review rating between the treated (partner hotels, i) and control
(non-partner hotels, j) groups is speciﬁed in Equation (1):

Dif f erence of Review Rating on Ctripi
(1)
=Review Rating on Ctripi − M edian Review Rating on Ctripj

where Review Rating on Ctripi and M edian Review Rating on Ctripj are the rating
of partner hotel i on Ctrip and the median of all ratings of the matched non-partner hotel
group j on Ctrip, respectively.
However, the ﬁrst difference calculated from Equation (1) may still be confounded
with unobservable factors. For example, the observed difference in Equation (1) may
merely be a result of hotels’ abundant promotion budgets. A hotel with a high promotion
budget may simultaneously have a higher likelihood of being a partner hotel and have
more fake or ‘promotional’ online reviews. We argue that these unobservable characteris-
tics of partners and non-partners would be reﬂected on other major platforms, regardless
of whether there is a partnership or not, such as the largest Chinese travel search engine,
Qunar. If the manipulated review ratings are driven by unobservable factors, such as ex-
tra marketing budget, they should be observable on other non-partner platforms as well.
Therefore, we benchmarked the ﬁrst difference with the rating difference of the same
hotels observed in Qunar, which is calculated in Equation (2).

Dif f erence of Review Rating on Qunari
(2)
=Review Rating on Qunari − M edian Review Rating on Qunarj

We based on following assumptions to select Qunar as the benchmark: 1) Commission is
not applicable to hotel sales through search engine, thus revenue of search engine is not
Does interest alignment intensify review manipulations? 11

TABLE I
Descriptive statistics of review ratings for partner hotels and non-partner hotels on Ctrip
and Qunar (partner hotels vs. non-partner hotels, N=2,838 pairs)
Partner Hotel Non-partner Hotel
Ctrip Qunar Ctrip Qunar

Mean Rating 4.599 4.553 4.302 4.318
Median of Rating 4.600 4.600 4.400 4.400
Variance of Rating 0.052 0.087 0.220 0.203

directly tied with hotel sales. Therefore, the partnership-intensiﬁed review manipulation
(not all manipulations) should be insigniﬁcant on Qunar. 2) User behaviours should be
similar on the two platforms. We veriﬁed this assumption by examining demographics
of Ctrip and Qunar users using Baidu Index (Baidu 2020). Baidu Index is provided by
Baidu, which is currently the largest search engine in China. Based on the behavioral data
generated on Baidu, Baidu Index provides insight on online users searching for speciﬁc
keywords and thereby helps companies understand their target markets. Figures 2, 3, and
4 in the Appendix show that Ctrip and Qunar users are similar in terms of gender, age,
and habits. This evidence mitigates the concern that user groups are signiﬁcantly different
between these two websites.
The review rating difference calculated in Equation (2) reﬂects the unobservable dif-
ference between the two groups of hotels, while the effects of partnership are not included
because Qunar has the same relationship with both groups of hotels.
Finally, we used Equation (1) (including both effects of unobservable factors and part-
nership) minus Equation (2) (including effects of unobservable factors only) to isolate the
effects of partnership. Speciﬁcally, we calculated the partnership-intensiﬁed review ma-
nipulation for a partner hotel i with Equation (3):

P artnership_intensif ied review manipulationi
=Equation 1 − Equation 2
=Dif f erence of Review Rating on Ctripi
(3)
− Dif f erence of Review Rating on Qunari
=(Review Rating on Ctripi − M edian Review Rating on Ctripj )−
(Review Rating on Qunari − M edian Review Rating on Qunarj )

4. Analysis model and results
Table I provides the descriptive statistics of review ratings of matched partner hotels
and non-partner hotel. Table II presented the paired t-test and non-parametric Wilcoxon
signed-rank (WSR) test to examine H1. The results indicates the manipulation intensity
is signiﬁcantly larger than zero (the null effect). Thus, H1 is supported.
H2 to H5 examined the effects of partnership-intensiﬁed review manipulation on hotel
online performance. We developed a hotel-week panel dataset and conducted regression
analyses to test these hypotheses. The dependent variable was hotel online performance,
Does interest alignment intensify review manipulations? 12

TABLE II
Testing results of partnership-intensiﬁed review manipulation (partner hotels vs. non-
partner hotels, N=2,838 pairs)
Paired t-test WSR Sign Rank test

Mean t p Positive: Negative Z p
H1 0.175 28.644 0.000 1.407 27.656 0.000

which was measured by the hotel’s weekly sales. The sales data were based on the book-
ing information collected from Ctrip’s website introduced in Section 3.1. We used the
number of bookings multiplied by the average price of the hotel in one week to obtain the
weekly sales volume. Hotel size was measured by the number of rooms offered by the ho-
tel (De Jorge & Suárez 2014). Hotel price was measured by the average price offered by
the hotel in the week. Hotel star rating was rated by the CNTA (China National Tourism
Administration) in an ordinal manner, ranging from one star (the lowest standard) to ﬁve
stars (the highest standard).
We used control variables to increase the robustness of the analysis. First, we included
the one-week lag value of the dependent variable (lag hotel performance) to create a dy-
namic panel data structure, which effectively controlled for the prior performance of the
hotel. We included hotel age because customers may favor new facilities provided by
new hotels. Further, Coenders, Espinet & Saez (2003) stressed the importance of hotel
attributes related to facilities and amenities for business and leisure travelers. Thus, we
controlled the hotel types by indicating whether a hotel was a business hotel or leisure
hotel, based on the labels indicated by the hotel’s Ctrip webpage. We included the num-
ber of room types and number of service types because consumers may favor hotels with
additional services provided, which was also indicated on the hotels’ Ctrip websites. We
included the number of holidays in each week because sales may be higher on holidays.
Finally, we included a dummy variable of observed time (week) and city of the hotel to
control for effects relating to time and location. We performed natural logarithm trans-
formation for hotel performance, lag hotel performance, hotel size, hotel price, and hotel
age to correct for the skewness of these variables. The panel data analysis model was
speciﬁed as follows,
ln(HotelP erf ormanceit ) =β0 + β1 M anipulationIntensityi,t−1 +
β2 ln(Sizei ) + β3 ln(P ricei,t−1 ) + β4 P restigei +
β5 ln(HotelP erf ormancei,t−1 ) + β6 ln(Agei )+
(4)
β7 BusinessHoteli + β8 ResortHoteli +
β9 RoomT ypesi + β10 ServiceT ypesi +
β11 Holidayst + W eek + City + u,
where i indicates hotel, t denotes the week of observation, and u is the error term. We
performed ordinary least squares to estimate the coefﬁcients β in the speciﬁed model.
Table III presents the descriptive statistics and correlations of the variables. The max-
imum variance inﬂation factor was 2.42, which suggested that multicollinearity was not a
serious concern in our analysis.
Does interest alignment intensify review manipulations? 13

Table IV presents the results from the regression analysis. Model 1 included all the
control variables and the standardized direct effects of the moderators. The adjusted R2
was 75.38%, which suggested that these variables created satisﬁed controls. Model 2
examined H2 by adding the standardized variable of partnership-intensiﬁed review ma-
nipulation. The coefﬁcient of manipulation intensity was signiﬁcantly negative (−0.021,
p <0.01). Adding the variable increased the goodness-of-ﬁt of Model 2 compared with
Model 1 (F =22.3, p <0.01). Thus, H2 was supported.
Model 3 examined H3 by adding the interaction term between standardized partnership-
intensiﬁed review manipulation and hotel size. The coefﬁcient of the interaction term was
signiﬁcantly negative (-0.013, p <0.01). Adding the variable increased the goodness-of-
ﬁt of Model 3 compared with Model 2 (F =7.50, p <0.01). Thus, H3 was supported.
Model 4 examined H4 by adding the interaction term between standardized partnership-
intensiﬁed review manipulation and hotel price. The coefﬁcient of the interaction term
was signiﬁcantly negative (-0.009, p <0.01). Adding the variable increased the goodness-
of-ﬁt of Model 4 compared with Model 2 (F =4.620, p <0.05). Thus, H4 was supported.
Model 5 examined H5 by adding the interaction term between standardized partnership-
intensiﬁed review manipulation and hotel prestige (star rating). The coefﬁcient of the in-
teraction term was signiﬁcantly negative (-0.053, p <0.01). Adding the variable increased
the goodness-of-ﬁt of Model 5 compared with Model 2 (F =17.17, p <0.01). Thus, H5
was supported.
Does interest alignment intensify review manipulations?
TABLE III
Descriptive statistics and correlations (N=16,743 hotel-week observations)
Variables Mean Standard Deviation 1 2 3 4 5 6 7 8 9 10 11
1 Hotel performance 8.918 1.098
2 Lag performance 8.987 1.114 0.831
3 Hotel size 4.554 0.831 0.452 0.458
4 Hotel price 5.705 0.638 0.658 0.614 0.272
5 Hotel prestige (star) 0.305 1.068 0.154 0.165 0.314 0.189
6 Hotel age 1.576 0.864 0.089 0.097 0.330 0.108 0.431
7 Business hotel 0.529 0.499 0.215 0.228 0.348 0.177 0.184 0.107
8 Resort hotel 0.203 0.402 0.321 0.324 0.347 0.415 0.259 0.200 0.058
9 Number of room types 10.253 4.293 0.157 0.150 0.085 0.096 0.068 0.165 -0.040 0.159
10 Number of service types 3.341 .694 0.413 0.419 0.525 0.502 0.297 0.230 0.422 0.539 0.136
11 Number of holidays 2.825 1.856 -0.005 -0.087 -0.001 0.118 -0.002 -0.003 -0.001 0.002 0.002 0.001
12 OTA review manipulation intensity 0.175 0.793 -0.052 -0.035 -0.021 -0.045 -0.048 -0.015 -0.024 0.010 0.012 -0.020 0.048

14
TABLE IV

Does interest alignment intensify review manipulations?
Regression analysis of the impacts of review manipulation by OTA (N=16,743)
Dependent variable: Hotel performance at week t
Model 1 Model 2 Model 3 Model 4 Model 5
Independent variables (t-1) Coef. S.E. p Coef. S.E. p Coef. S.E. p Coef. S.E. p Coef. S.E. p
Lag performance 0.606 0.006 0.000 0.606 0.006 0.000 0.606 0.006 0.000 0.606 0.006 0.000 0.606 0.006 0.000
Hotel size 0.155 0.006 0.000 0.155 0.006 0.000 0.156 0.006 0.000 0.155 0.006 0.000 0.155 0.006 0.000
Hotel price 0.274 0.006 0.000 0.273 0.006 0.000 0.273 0.006 0.000 0.274 0.006 0.000 0.273 0.006 0.000
Hotel prestige (star) -0.009 0.005 0.053 -0.010 0.005 0.035 -0.011 0.005 0.023 -0.011 0.005 0.027 -0.015 0.005 0.003
Hotel age -0.036 0.006 0.000 -0.036 0.006 0.000 -0.035 0.006 0.000 -0.035 0.006 0.000 -0.036 0.006 0.000
Business hotel 0.025 0.010 0.014 0.025 0.010 0.015 0.026 0.010 0.012 0.025 0.010 0.014 0.025 0.010 0.013
Resort hotel -0.046 0.013 0.001 -0.044 0.013 0.001 -0.043 0.013 0.001 -0.044 0.013 0.001 -0.045 0.013 0.001
Number of room types 0.007 0.001 0.000 0.007 0.001 0.000 0.007 0.001 0.000 0.007 0.001 0.000 0.007 0.001 0.000
Number of service types -0.023 0.004 0.000 -0.024 0.004 0.000 -0.024 0.004 0.000 -0.024 0.004 0.000 -0.024 0.004 0.000
Number of holidays -0.016 0.003 0.000 -0.015 0.003 0.000 -0.015 0.003 0.000 -0.015 0.003 0.000 -0.015 0.003 0.000
OTA review manipulation intensity -0.021 0.005 0.000 -0.023 0.005 0.000 -0.022 0.005 0.000 -0.034 0.006 0.000
Manipulation × Hotel size -0.013 0.005 0.006
Manipulation × Hotel price -0.009 0.004 0.032
Manipulation × Hotel prestige -0.053 0.013 0.000
Adjusted R2 75.38% 75.41% 75.42% 75.41% 75.43%

15
Does interest alignment intensify review manipulations? 16

5. Robustness checks
In this section, we performed robustness checks to strengthen the validity of our ﬁndings.

5.1. Alternative detection of partnership-intensiﬁed review manipulation:
variance
Generally, there are two possible ways to manipulate review ratings: adding positive re-
views and deleting negative reviews. Previous studies have found that the adverse effect
of negative reviews is greater than the favorable effect of positive reviews (Lappas, Sabnis
& Valkanas 2016). In addition, deleting negatvie reivews is more disguised and difﬁcult to
be suspected (Zhuang et al. 2018). In practice, it is very risky for ﬁrms to conduct review
manipulation. Once discovered, their reputation and credibility will be damaged. More-
over, adding positive reviews is physically much more difﬁcult than deleting negative
reviews, as it involves creating fraudulent accounts, feigning transaction records (as only
guests can leave reviews), and posting fake comments. Therefore, compared with adding
favorable reviews, deleting negative reviews is a more disguised and feasible long-term
strategy to boost online ratings.
If this is the case, it will lead to lower variance observed in online customer ratings
for partner hotels than for non-partner hotels. This distinguishes partnership-intensiﬁed
manipulation from other types of review manipulation, which can only be conducted by
adding positive/negative (positive by business owners or negative by competitors) reviews
(users are typically not allowed to delete reviews to maintain the integrity of the review
system). Adding fake reviews typically leads to more variance because the purpose of do-
ing so is to change (elevate or decrease) the current overall rating and only extreme values
(compared with normal review ratings) can serve this purpose effectively. Therefore, we
posited that a lower variance would be observed in the customer review ratings for OTAs’
partner hotels.
We conducted a paired t-test and WSR test to determine the difference in the variance
of customer review ratings between matched hotels (Ctrip partner hotels versus matched
non-partner hotels). The test t-statistic and p-value of the paired t-test were -5.503 and
0.000, respectively. The z- and p-values of the WSR test were -4.865 and 0.000, respec-
tively. The average difference in the variance of the review ratings of the matched Ctrip
partner and non-partner hotels was -0.0319. These results indicate a lower variance in
customer ratings for Ctrip partner hotels than for non-partner hotels. This test provides
additional evidence to indicate the intensiﬁed manipulated reviews for partner hotels.

5.2. Alternative data source: Fliggy
Given that we examined our hypotheses with data collected from Ctrip, some may argue
that partnership-intensiﬁed review manipulation may depend on platform-level unobserv-
able variables (e.g., tendency to choose partner hotels). Thus, we examined the sensitivity
of our main analysis by replacing the data from Ctrip with data from Fliggy (the bench-
mark data from Qunar unchanged). Fliggy is the second-largest OTA in China—second
only to Ctrip in number of active users. We created 2,007 pairs of matched Fliggy partner
hotels and non-partner hotels for this robustness analysis.
Does interest alignment intensify review manipulations? 17

The paired t-test about the rating difference in the matched Fliggy partner hotels and
non-partner hotels, benchmarked with Qunar, indicated a statistically signiﬁcant differ-
ence, with a t-statistic and p-value of 4.363 and 0.000, respectively. The average rating
difference was 0.028. The paired t-test about the variance observed in customer ratings
for Fliggy partner hotels and non-partner hotels also indicated a statistically signiﬁcant
difference, with a t-statistic and p-value of -6.858 and 0.000, respectively. The average
difference in the variance was -0.54, which was similar to the results observed in Ctrip.
These two tests provide additional evidence to support H1.

6. Conclusions
Online reviews of hotel services are critical in consumer purchasing decisions. How-
ever, manipulated reviews may be harmful. Although review manipulation aims to boost
sales, it may backﬁre. Our study captured an abnormally higher rating for the OTA part-
nered hotels, which indicates that the higher interest alignment between the hotels and
OTA was intensifying the review manipulation of these hotels. This conclusion was fur-
ther strengthened by the ﬁnding that the variance in review ratings of partner hotels was
signiﬁcantly lower than that of non-partner hotels, which was likely caused by deleting
extremely negative reviews.
After that, we examined the effect of the partnership-intensiﬁed review manipulation
on hotels’ sales performance. With scenario-based experiments, the previous literature
(Reimer & Benkenstein 2016) has found that suspicious reviews have negative effects on
business performance. We conﬁrmed this notion with ﬁeld data collected from realistic
business processes. Our results suggest that partnership-intensiﬁed review manipulation
deteriorates hotels’ business performance by undermining customer trust and widening
the gap between customer expectation and perception. Moreover, this deterioration of
performance is more salient for large-scale, expensive, and prestigious hotels.

6.1. Theoretical contributions
This research contributes to the literature on online customer reviews and review manip-
ulation. The previous literature has largely focused on the role of business owners and
ignored the role of the hosting platform in review manipulation. Our research differs
from previous studies by exploring the role of OTAs as the intermediary between hotels
and consumers. Nowadays, hotel booking transactions increasingly rely on the platform
provided by OTAs, yet it was unclear how the interaction between hotels and OTAs could
affect the intensity of review manipulation. We entered this discourse by investigating
whether the interest alignment between hotels and OTAs can facilitate review manipu-
lation. OTAs perform sales and promotion activities on behalf of hotels, and perform
information search activities on behalf of consumers. Ideally, OTAs should be fair and
protect the interests of both opposing parties (hotels and consumers). However, this may
not be true in practice (Hunold et al. 2020). The aligned interest between hotels and OTAs
may disrupt this balance, and subsequently facilitate opportunistic behaviors against con-
sumers. Speciﬁcally, by capturing the abnormally high rating of OTA partnered hotels
Does interest alignment intensify review manipulations? 18

(compared with non-partners), we provide evidence that such interest alignment through
sales commissions can facilitate review manipulation.
This research also contributes to the agency theory. It is widely suggested in agency
studies that interest alignment between the principal (i.e., buyer) and agent (i.e., supplier)
can reduce the agent’s opportunistic behavior (Eisenhardt 1989). Interest alignment re-
solves the goal conﬂicts between the two parties, so that the supplier is more likely to
act on behalf of the buyer’s interests. However, few studies have discussed how this op-
portunistic behavior may be affected when an intermediary is involved in the transaction.
We have found that the interest ties between the intermediary and agent (i.e., hotels) will
intensify opportunistic behavior toward consumers. This is likely to occur because the ma-
nipulation can increase the short-term sales of these hotels, which subsequently increases
the commission income of the OTA. The agent-intermediary (i.e., hotel-OTA) partnership
may intensify the myopic goal and short-term gain, yet compromise the long-term beneﬁt
of the hotel. This view is consistent with prospect theory, which hypothesizes that decision
makers favor instant gain, rather than future gain (Tversky & Kahneman 1979). These
ﬁndings call for a rethink of the merits of interest alignment in controlling transaction
costs in outsourcing relationships.

6.2. Managerial implications
This research presents practical implications to hotels and OTAs to construct a better and
healthier hospitality industry. First, our results provide evidence that partnership-intensi-
ﬁed review manipulation will lead to a negative (not positive) effect on sales. Managers
should not assume that consumers can be easily manipulated. Moreover, our results indi-
cate that partnership-intensiﬁed review manipulation lead to a 0.2% decrease on hotels’
online sales. Given Ctrip’s gross merchandise volume jumped 30 percent year-on-year to
hit ¥725 billion in 2018 (Zhu 2019), the magnitude of the effect on total online sales
of Ctrip alone will be measured in billions. Moreover, the robustness check of this study
indicates that, in addition to the average rating, the variance of customer ratings can also
help detect review manipulation.
This evidence can also be used to educate hotel managers to avoid excess review
manipulation. If the idea that review manipulation can lead to higher sales with little
cost prevails among hotel managers, the manipulation behavior will enter managers into
a "prisoner’s dilemma"—the manager may consider that they are at a disadvantage if not
manipulating reviews when others are doing so. However, our solid evidence can be
an important reference indicating that hotels who are manipulating their reviews will be
punished by the market. The hospitality industry will be healthier if the idea that review
manipulation is harmful is widely accepted.
Last, OTAs should not lessen their monitoring of their partner hotels and allow review
manipulation to prevail on their platform. The image of independence and fairness is vital
for OTAs to gain consumer trust, given that they perform information search activities for
consumers. Intensiﬁed manipulations will hurt consumer trust in both the hotel and OTA,
which will subsequently undermine their sales performance.
Does interest alignment intensify review manipulations? 19

6.3. Limitations
There are several limitations of this study. First, our detection of review manipulation
was based on empirical evidence, rather than direct observation. Second, the number of
bookings was used as a proxy for room sales, instead of the number of rooms booked.
This may have caused some measuring errors. In addition, this message may also subject
to manipulation. However, as long as the way of such manipulation is consistent, it will
not impact our qualitative conclusions. In the future, there are certain research directions
worth exploring. For example, it would be interesting to examine online and ofﬂine data
together over the same time period to ﬁnd out interplay of online reviews and ofﬂine
customer behaviour. Also, it would be interesting and challenging to examine the role of
OTAs and their partner hotels in partnership-intensiﬁed review manipulation. The power
asymmetry between OTAs and these hotels and its correlation with manipulation intensity
may help answer this question.

References
Abrate, G., Capriello, A. & Fraquelli, G. (2011), ‘When quality signals talk: Evidence
from the turin hotel industry’, Tourism Management 32(4), 912–921.

Ahmad, W. & Sun, J. (2018), ‘Modeling consumer distrust of online hotel reviews’, In-
ternational Journal of Hospitality Management 71, 77–90.

Akhtar, N., Ahmad, W., Siddiqi, U. I. & Akhtar, M. N. (2019), ‘Predictors and outcomes
of consumer deception in hotel reviews: The roles of reviewer type and attribution of
service failure’, Journal of Hospitality and Tourism Management 39, 65–75.

Anderson, E. W. & Sullivan, M. W. (1993), ‘The antecedents and consequences of cus-
tomer satisfaction for ﬁrms’, Marketing science 12(2), 125–143.

Arifﬁn, A. A. M. & Maghzi, A. (2012), ‘A preliminary study on customer expectations
of hotel hospitality: Inﬂuences of personal and hotel factors’, International Journal of
Hospitality Management 31(1), 191–198.

Baidu (2020), ‘Baidu index.’. https://index.baidu.com.

Bick, G., Jacobson, M. C. & Abratt, R. (2003), ‘The corporate identity management pro-
cess revisited’, Journal of Marketing Management 19(7-8), 835–855.

Bond, C. F., Omar, A., Pitre, U., Lashley, B. R., Skaggs, L. M. & Kirk, C. (1992), ‘Fishy-
looking liars: deception judgment from expectancy violation.’, Journal of personality
and social psychology 63(6), 969.

Bosse, D. A. & Phillips, R. A. (2016), ‘Agency theory and bounded self-interest’,
Academy of Management Review 41(2), 276–297.

Brown, T. J. & Dacin, P. A. (1997), ‘The company and the product: Corporate associations
and consumer product responses’, Journal of marketing 61(1), 68–84.
Does interest alignment intensify review manipulations? 20

Burgoon, J. K. & Le Poire, B. A. (1993), ‘Effects of communication expectancies, actual
communication, and expectancy disconﬁrmation on evaluations of communicators and
their communication behavior’, Human communication research 20(1), 67–96.

Casado-Díaz, A. B., Andreu, L., Beckmann, S. C. & Miller, C. (2020), ‘Negative online
reviews and webcare strategies in social media: effects on hotel attitude and booking
intentions’, Current Issues in Tourism 23(4), 418–422.

Chevalier, J. A. & Mayzlin, D. (2006), ‘The effect of word of mouth on sales: Online
book reviews’, Journal of marketing research 43(3), 345–354.

Coenders, G., Espinet, J. & Saez, M. (2003), ‘Predicting random level and seasonality of
hotel prices: a latent growth curve approach’, Tourism Analysis 8(1), 15–31.

De Jorge, J. & Suárez, C. (2014), ‘Productivity, efﬁciency and its determinant factors in
hotels’, The Service Industries Journal 34(4), 354–372.

Doney, P. M. & Cannon, J. P. (1997), ‘An examination of the nature of trust in buyer–seller
relationships’, Journal of marketing 61(2), 35–51.

Eisenhardt, K. M. (1989), ‘Agency theory: An assessment and review’, Academy of man-
agement review 14(1), 57–74.

Ekanayake, S. (2004), ‘Agency theory, national culture and management control systems’,
Journal of American academy of business 4(1/2), 49–54.

Filieri, R. (2015), ‘What makes online reviews helpful? a diagnosticity-adoption frame-
work to explain informational and normative inﬂuences in e-wom’, Journal of Business
Research 68(6), 1261–1270.

Filieri, R. (2016), ‘What makes an online consumer review trustworthy?’, Annals of
Tourism Research 58, 46–64.

Filieri, R., Hofacker, C. F. & Alguezaui, S. (2018), ‘What makes information in online
consumer reviews diagnostic over time? the role of review relevancy, factuality, cur-
rency, source credibility and ranking score’, Computers in Human Behavior 80, 122–
131.

Filieri, R. & McLeay, F. (2014), ‘E-wom and accommodation: An analysis of the fac-
tors that inﬂuence travelers? adoption of information from online reviews’, Journal of
Travel Research 53(1), 44–57.

Fleisher, C. S. (1991), ‘Using an agency-based approach to analyze collaborative fed-
erated interorganizational relationships’, The Journal of applied behavioral science
27(1), 116–130.

Folkes, V. S. & Kamins, M. A. (1999), ‘Effects of information about ﬁrms’ ethical and
unethical actions on consumers’ attitudes’, Journal of consumer psychology 8(3), 243–
259.
Does interest alignment intensify review manipulations? 21

Gössling, S., Hall, C. M. & Andersson, A.-C. (2018), ‘The manager’s dilemma: a con-
ceptualization of online review manipulation strategies’, Current Issues in Tourism
21(5), 484–503.

Heath, T. B. & Chatterjee, S. (1995), ‘Asymmetric decoy effects on lower-quality versus
higher-quality brands: Meta-analytic and experimental evidence’, Journal of Consumer
Research 22(3), 268–284.

Hu, N., Bose, I., Gao, Y. & Liu, L. (2011), ‘Manipulation in digital word-of-mouth: A
reality check for book reviews’, Decision Support Systems 50(3), 627–635.

Hu, N., Bose, I., Koh, N. S. & Liu, L. (2012), ‘Manipulation of online reviews: An
analysis of ratings, readability, and sentiments’, Decision support systems 52(3), 674–
684.

Hunold, M., Kesler, R. & Laitenberger, U. (2020), ‘Hotel rankings of online travel agents,
channel pricing and consumer protection.’, Marketing Science 39(1), 92–116.

Jensen, M. C. & Meckling, W. H. (1976), ‘Theory of the ﬁrm: Managerial behavior,
agency costs and ownership structure’, Journal of ﬁnancial economics 3(4), 305–360.

Jin Ma, Y. & Lee, H.-H. (2014), ‘Consumer responses toward online review manipula-
tion’, Journal of Research in Interactive Marketing 8(3), 224–244.

Ketron, S. (2017), ‘Investigating the effect of quality of grammar and mechanics (qgam)
in online reviews: The mediating role of reviewer crediblity’, Journal of Business Re-
search 81, 51–59.

Kopalle, P. K. & Lehmann, D. R. (2006), ‘Setting quality expectations when entering a
market: What should the promise be?’, Marketing Science 25(1), 8–24.

Lappas, T., Sabnis, G. & Valkanas, G. (2016), ‘The impact of fake reviews on online visi-
bility: A vulnerability assessment of the hotel industry’, Information Systems Research
27(4), 940–961.

Lee, S.-Y., Qiu, L. & Whinston, A. (2018), ‘Sentiment manipulation in online platforms:
An analysis of movie tweets’, Production and Operations Management 27(3), 393–
416.

Liu, Z. (2019), ‘Online travel agency leading company operates stably.’.
https://m.gelonghui.com/p/246981.

Luca, M. & Zervas, G. (2016), ‘Fake it till you make it: Reputation, competition, and yelp
review fraud’, Management Science 62(12), 3412–3427.

Mauri, A. G. & Minazzi, R. (2013), ‘Web reviews inﬂuence on expectations and pur-
chasing intentions of hotel potential customers’, International Journal of Hospitality
Management 34, 99–107.
Does interest alignment intensify review manipulations? 22

Mayzlin, D., Dover, Y. & Chevalier, J. (2014), ‘Promotional reviews: An empirical inves-
tigation of online review manipulation’, American Economic Review 104(8), 2421–55.

Mohr, L. A. & Webb, D. J. (2005), ‘The effects of corporate social responsibility and
price on consumer responses’, Journal of consumer affairs 39(1), 121–147.

Norrman, A. (2008), ‘Supply chain risk-sharing contracts from a buyers’ perspec-
tive: content and experiences’, International Journal of Procurement Management
1(4), 371–393.

Öğüt, H. & Onur Taş, B. K. (2012), ‘The inﬂuence of internet customer reviews on the
online sales and prices in hotel industry’, The Service Industries Journal 32(2), 197–
214.

Pavlou, P. A., Liang, H. & Xue, Y. (2007), ‘Understanding and mitigating uncertainty in
online exchange relationships: A principal-agent perspective’, MIS quarterly pp. 105–
136.

Reimer, T. & Benkenstein, M. (2016), ‘When good wom hurts and bad wom gains: The
effect of untrustworthy online reviews’, Journal of Business Research 69(12), 5993–
6001.

Rhee, M. & Haunschild, P. R. (2006), ‘The liability of good reputation: A study of product
recalls in the us automobile industry’, Organization Science 17(1), 101–117.

Sen, S. & Bhattacharya, C. B. (2001), ‘Does doing good always lead to doing better?
consumer reactions to corporate social responsibility’, Journal of marketing Research
38(2), 225–243.

Shapiro, C. (1983), ‘Premiums for high quality products as returns to reputations’, The
quarterly journal of economics 98(4), 659–679.

Simonson, I. & Tversky, A. (1992), ‘Choice in context: Tradeoff contrast and extremeness
aversion’, Journal of marketing research 29(3), 281–295.

Singh, J. & Sirdeshmukh, D. (2000), ‘Agency and trust mechanisms in consumer satisfac-
tion and loyalty judgments’, Journal of the Academy of marketing Science 28(1), 150–
167.

Stuart, E. A. (2010), ‘Matching methods for causal inference: A review and a look for-
ward’, Statistical science: a review journal of the Institute of Mathematical Statistics
25(1), 1.

Tan, J. C. K. & Lee, R. (2015), ‘An agency theory scale for ﬁnancial services’, Journal of
Services Marketing .

Tversky, A. & Kahneman, D. (1979), ‘Prospect theory: An analysis of decision under
risk’, Econometrica 47(2), 263–291.
Does interest alignment intensify review manipulations? 23

Voss, G. B., Parasuraman, A. & Grewal, D. (1998), ‘The roles of price, performance, and
expectations in determining satisfaction in service exchanges’, Journal of marketing
62(4), 46–61.

Walsh, G. & Beatty, S. E. (2007), ‘Customer-based corporate reputation of a service
ﬁrm: scale development and validation’, Journal of the academy of marketing science
35(1), 127–143.

Williams, R. J. & Barrett, J. D. (2000), ‘Corporate philanthropy, criminal activity, and
ﬁrm reputation: Is there a link?’, Journal of Business Ethics 26(4), 341–350.

Womply Research (2020), ‘How do online reviews impact revenue for lodging places?’.
https://www.womply.com/impact-of-online-reviews-on-small-business-revenue/.

Xu, Y., Zhang, Z., Law, R. & Zhang, Z. (2019), ‘Effects of online reviews and managerial
responses from a review manipulation perspective’, Current Issues in Tourism pp. 1–16.

Yacouel, N. & Fleischer, A. (2012), ‘The role of cybermediaries in reputation building and
price premiums in the online hotel market’, Journal of Travel Research 51(2), 219–226.

Ye, Q., Law, R. & Gu, B. (2009), ‘The impact of online user reviews on hotel room sales’,
International Journal of Hospitality Management 28(1), 180–182.

Ye, Q., Law, R., Gu, B. & Chen, W. (2011), ‘The inﬂuence of user-generated content on
traveler behavior: An empirical investigation on the effects of e-word-of-mouth to hotel
online bookings’, Computers in Human behavior 27(2), 634–639.

Zhang, J. J., Lawrence, B. & Anderson, C. K. (2015), ‘An agency perspective on service
triads: Linking operational and ﬁnancial performance’, Journal of Operations Manage-
ment 35, 56–66.

Zhao, X. R., Wang, L., Guo, X. & Law, R. (2015), ‘The inﬂuence of online reviews to
online hotel booking intentions’, International Journal of Contemporary Hospitality
Management .

Zhu, S. (2019), ‘Ctrip reports revenue and gmv growth in 2018.’.
https://www.meadin.com/yj/101733.html.

Zhuang, M., Cui, G. & Peng, L. (2018), ‘Manufactured opinions: The effect of manipu-
lating online product reviews’, Journal of Business Research 87, 24–35.
Does interest alignment intensify review manipulations? 24

Appendix: Demographics of Ctrip and Qunar users by Baidu index

Demographics of Ctrip and Qunar users using Baidu Index are illustrated in Figures 2, 3,
and 4, which indicate that user behaviours are similar on these two platforms.

80
Ctrip Qunar
58.3 60

Percentage of users
60

41.7 40
40

0
Male Female

F IGURE 2: Gender distributions of Ctrip and Qunar users.

80
Ctrip Qunar

Percentage of users
60
51.3
43.8
39.4
40
29.3

20
8.4 8.6 8.7
4.9 2.4 3.2
0
0–19 20–29 30–39 40–49 50+
Age

F IGURE 3: Age distributions of Ctrip and Qunar users.

1
Ctrip Qunar

100

50
Percentage of users

F IGURE 4: Hobby distributions of Ctrip and

0
Video Movies Travel IT tech. Software Education Health Reading Finance Domestic
games and and services
Does interest alignment intensify review manipulations?
1 Qunar users.

music training

File and source

File: chen-tang-li-fan-2021-imds-hotels-online-travel-agencies-aam.pdf
Size: 272,823 bytes
SHA-256: 5272e188fc5cd335e8d302922dc1584eafc601750d4e3e56dbd4453bc13e6f8c
Our copy: chen-tang-li-fan-2021-imds-hotels-online-travel-agencies-aam.pdf
Original: No public link identified.