MODELING FACTORS THAT INFLUENCE ONLINE
TRAVEL BOOKING
Michael Conyette
Okanagan School of Business, 7000 College Way, Vernon, BC, V1B 2N5, Canada
Keywords: B2C, Decision Process, Determinants, e-Commerce, Online Travel Booking, Theory of Reasoned Action.
Abstract: Data was collected from an online questionnaire completed by 1,198 respondents in 2008. Analysis of the
dataset involved, correlation analysis, exploratory factor analysis, and logistic regression. In the final model
building stage, a logistic regression model is generated containing key factors that lead to online travel
booking intention. These factors are a unique set of socio and psychographic variables that can be used to
more accurately predict website booking of travel products. The contribution to literature that this research
makes is that it appears to be one of only a few models available for predicting travel product booking. For
instance, this model predicts that consumers who previously booked specific travel products such as hotels
or airline tickets will have a greater intention to book other travel products online. This research study also
shows the relevance of the Theory of Reasoned Action to online travel but it goes further by enabling the
quantification of the strength of variables such as key beliefs, attitudes and subjective norms.
1 INTRODUCTION
The Internet and the tourism industry are the
contexts within which this research is based. A
virtual company such as a travel website operates by
providing access to its travel products and services
through the Internet, and both travel websites and
travel agents function within the tourism industry.
The focus of this research is to assess the
determinants of decision processes consumers
undertake when booking their leisure travel online.
However, it is recognized that consumers can use
online or offline aids in the travel planning process
(Conyette, 2010).
An online decision aid (ODA) is sometimes
nested within a travel website so that a consumer is
unaware they are using a sophisticated tool.
Consumers also consult a travel agent, that is, an
offline aid to assist them in travel planning. The
study will provide tourism marketers with some
understanding of leisure travelers and their behavior
in the travel planning process, thereby helping
marketers design suitable travel websites, online
tools, travel agency services and marketing
strategies. The knowledge deficit with online tools
used in travel includes a viable model that explains
and quantifies the interplay of beliefs, attitudes, prior
experience with travel agents and websites, social
support, and how these factors contribute to online
booking.
A proposed model containing these factors is used
to test key hypotheses through data collected with a
survey instrument. These hypotheses and
components point to online travel booking intention,
which is the primary interest. Quantitative data
analysis helps in understanding the ultimate factors
of online leisure travel booking intention.
Qualitative research consisting of interviews and
focus groups were first conducted in a related study
and they guided the development of the model and
inclusion of variables (Conyette, 2010).
This research project began with the initial
question of whether intelligent agents used in travel
planning compare with a travel agent that is highly
knowledgeable about both the product alternatives
available and the consumer’s tastes. Over the past
few years, technologies have advanced, and the
Internet has spawned numerous new travel business
models including travel search engines, online travel
agencies, and travel websites with varying levels of
sophistication and intelligent infrastructure. What
was once an advanced intelligent online tool existing
initially in artificial intelligent laboratories such as
SmartClient (Pu and Faltings, 2000), Heracles
(Ambite et al., 2002), Hamlet (Etzioni et al., 2003),
205
Conyette M..
MODELING FACTORS THAT INFLUENCE ONLINE TRAVEL BOOKING.
DOI: 10.5220/0003455902050210
In Proceedings of the International Conference on e-Business (ICE-B-2011), pages 205-210
ISBN: 978-989-8425-70-6
Copyright
c
 2011 SCITEPRESS (Science and Technology Publications, Lda.)

Theseus (Barish et al., 2000), INTRIGUE
(Ardissono, 2003), and other ODAs, is now
becoming more commonly used by consumers.
Intelligent tools can be found embedded in travel
websites such as Farecast’s airfare predictive
analytics tool, which is now incorporated into
Expedia’s website infrastructure. Consumers have
become more comfortable with Internet
technologies, and these technologies have advanced
so that they offer travelers more options and
assistance in a user-friendly, intuitive and interactive
way.
2 HYPOTHESES
A parsimonious model consisting of seven key
hypotheses was assessed using logistic regression
analysis. The components of the model emerged
from theoretical frameworks consisting primarily of
the Theory of Reasoned Action (TRA). TRA
proposes that a person's beliefs influence their
attitudes which in turn affect their behavior as
measured by behavioural intention (BI). One key
element of TRA is that behavioral intention has been
found to predict actual behaviour. Also, BI results
from both attitudes toward a behavior and subjective
social norms toward that behaviour (Fishbein,
1967).The qualitative research conducted prior to
model development suggested that TRA was a
suitable and significant theoretical framework for
understanding travel product purchases (Conyette,
2010).
The research hypotheses to be tested are as
follows:
H1b. Consumers who have more positive beliefs
about online travel booking will have a more
positive attitude toward online travel booking than
consumers who have less positive beliefs about
online travel booking.
H1f. Consumers who have more social support for
online travel booking will perceive more social
acceptance of online travel booking than consumers
who have less social support.
H1g. A consumer’s perceptions of the extent to
which significant referents approve of Internet use
for online travel booking will positively affect
prediction intention to use the Internet for travel
booking.
H1h. Consumers with more prior experience with
the Internet and Internet travel will have more
positive beliefs about online travel booking than do
consumers who have less prior experience with the
Internet.
H1i. Consumers who have more positive beliefs
about travel agents will have lesser intention to
purchase travel online than do consumers who have
less positive beliefs about travel agents.
H1j. Consumers who have more prior experience
with the Internet and Internet travel will have greater
intention to purchase travel online than do
consumers who have less prior experience with the
Internet.
H1k. Consumers with a more positive attitude
toward online travel booking have greater intention
to purchase travel products online than consumers
who have a less positive attitude.
3 SURVEY INSTRUMENT
An online survey questionnaire was used to
determine how the various factors affect travel
planning and purchasing decisions. Respondents
were invited by various businesses that expressed an
interest in the research including The Prestige Hotels
& Resorts, Budget Car Rentals, The Kettle Valley
Steam Railway, The Fintry Queen boat charters, and
DiscoverTheIslands.com.
Thirty five questions were asked to assess prior
experience with computers and the Internet,
purchasing patterns online and offline, beliefs and
attitudes about travel agents and travel websites,
knowledge of travel and involvement with it,
motivations for using the Internet, and various
demographics. A total of 1300 surveys were
submitted. One hundred and two surveys were
deleted, as responses were not complete, leaving
1198 completed surveys for data analysis.
Some of the tests used on the data collected from
the survey instrument include the following:
• Factor analysis, which simplifies the data by
reducing the information contained in a large
number of variables into a smaller number of subsets
or factors. This helps identify the main factors.
• Pearson chi-square test of independence and
logistic regression to determine which variables are
most strongly associated with the intention to book
online.
Pre-testing the questionnaire was important to
validate the instrument. After about 250 surveys
were collected the data was analyzed to assess the
survey instrument and determine whether any
changes were needed. Despite the small proportion
of unfinished surveys referred to earlier, there were
no significant gaps in responses to indicate that
questions were unclear to respondents or that
ICE-B 2011 - International Conference on e-Business
206

respondents were skipping a particular question.
Question items seemed easy to read and understand,
meaningful to participants and sufficiently detailed.
Directions provided in the questionnaire appeared to
be helpful as well.
4 DATA ANALYSIS
Data analysis was performed using both SPSS 17.0,
and Stata 10 software. Statistical analysis was
conducted using Pearson’s chi-square test of
independence, logistic regression analysis,
Spearman correlation analysis, and factor analysis.
A 95% confidence interval was used to determine
the level of statistical significance for tests.
The data was assessed for normality, linearity and
homoscedasticity. Multivariate normality was not
evident with most variables. Transformations of
these variables did not improve normality;
furthermore, the data needed to be simplified in
order to make comparisons easier. Therefore,
categories of these variables were merged when
needed to more evenly distribute the data and reflect
a meaningful distinction between categories in
practical terms without limiting interpretations.
5 STATISTIC CALCULATIONS
Hypotheses were tested using logistic regression.
For each hypothesis, the Pearson chi-square test of
independence with an alpha of 0.05 was firstly used
to assess if there was independence between each
predictor and corresponding response variable. After
each association test was conducted, some variables
were kept and others dropped based on statistical
criteria. This is followed by univariate logistic
regression tests using a level of significance of 0.05
to determine whether the independent variable in the
model is significantly related to the outcome
variable. The decision to keep predictor variables at
this stage was made primarily based on the
likelihood test but also the Wald test. Finally, a
model was built for each hypothesis by selecting
variables for the multivariable analysis using a
stepwise method to explain the remaining predictors
for the response variable of each hypothesis. The
importance of each variable included in the model
was verified through an examination of the Wald
test statistic. Evidence of interactions in the data was
tested and no interaction was found between
variables. Therefore, initially there are seven
models, one for each H1 hypothesis.
After this hypothesis testing phase, a Final Model
(the focus of this paper) predicting online travel
booking intention was built using a stepwise logistic
regression method by selecting specific variables for
multivariable analysis.
Predictor variables retained from testing
hypotheses H1g, H1i, H1j and Hik will comprise the
elements used in the Final Model to be developed
using logistic regression.
5.1 Beliefs Affecting Attitudes
Using the statistical approach outlined above,
Hypothesis H1b is supported with predictor
variables for the attitude ‘desirable’ consisting of
‘beliefs’ ‘convenient’, ‘safe’, ‘easy’, ‘enjoyable’ and
‘convenience importance’. It is also supported with
predictor variables for the attitude ‘positive’
consisting of ‘beliefs’ ‘convenient’, ‘safe’, ‘easy’,
and ‘enjoyable’.
5.2 Social Support Impacts Social
Acceptance
Odds ratios indicate hypothesis H1f is supported
with the predictor variable ‘my friends or family
encourage me to purchase travel products via the
Internet’.
5.3 Social Acceptance Affects Online
Booking Intention
Hypothesis H1g is supported with the predictor
variable ‘some of my friends or family buy travel
products’ as demonstrated by the odds ratios.
5.4 Prior Experience with Internet
Influences Beliefs
Based on odds ratios generated, hypothesis H1h is
supported with predictor variables prior experience
purchasing specific travel products online such as
‘destination tour/attraction tickets’ and ‘airline
tickets’, and ‘leisure travel purchased online in the
past 12 months’. Four models developed to test other
belief variables support the hypothesis as well
(Conyette, 2010).
5.5 Travel Agent Beliefs Influences
Online Booking
Hypothesis H1i is supported with predictor variables
MODELING FACTORS THAT INFLUENCE ONLINE TRAVEL BOOKING
207

beliefs, ‘convenient’ and ‘expensive’ likely to
influence online booking intention.
5.6 Prior Experience Influences Online
Booking Intention
Hypothesis H1j is supported with various predictor
variables measuring prior experience.
5.7 Attitudes Affect Online Booking
Intention
Hypothesis H1k is supported with attitude predictor
variables ‘positive’ and ‘desirable’.
5.8 Final Model
As stated earlier, the final model building process
involves determining which variables best predict
online travel booking intention. The Final Model
includes retained variables resulting from the tests of
hypotheses H1g, H1i, H1j and Hik since they
contribute directly to online travel booking intention.
In the Final Model odds ratios may be interpreted to
gauge the relative importance of predictors and their
predictive ability.
Thus in the Model, Table 1, the dependent or
response variable is online booking intention as
related to the survey question, “How likely is it that
you will book or purchase any travel product
through the Internet within the next six months?”
Categories were merged so that three remain,
1 = highly likely, 2 = likely, and 3 = somewhat
likely.
Furthermore, 13 independent or predictor
variables are as follows: Some of my friends or
family buys travel products on the Internet, Belief
that booking with a travel agent is
‘convenient/inconvenient’, Belief that booking with
a travel agent is ‘expensive’, Having access to the
Internet from places other than home or work,
Length of time a person has been using the Internet,
Number of leisure trips taken in the past year, The
attitude that it is positive to book with a travel
website, The attitude it is desirable to book with a
travel website, Prior experience purchasing five
specific travel products online such as, ‘destination
tour/attraction tickets’, ‘hotels or accommodation’,
‘airline tickets’, ‘car rentals’, ‘long-distance train
tickets’.
At this stage model building was a simple task
since all variables have already been merged where
necessary, and assessed using a Chi-square test for
independence, and univariate analysis using a level
of significance of 0.05. One variable was dropped
through an examination of the Wald test statistic.
This was the variable related to the question, “About
how much time do you use the Internet each week
for any reason other than work?”
The Model as a whole yielded a log likelihood of
-709.05 and an R2 of 16.46%. As Hilbe (2009)
indicates the proportional odds model assumes
equality of slopes among response levels or
categories, so that the odds ratios pertaining to
1 = ‘highly likely’ to book apply as well to the
categories of 2=‘likely’, and 3= ‘somewhat likely’.
A notable predictor of online booking intention was
social acceptance as expressed in the survey
statement, “Some of my friends or family buys
travel products on the Internet”. The social influence
element of the Theory of Reasoned Action seems
critical in explaining consumers’ intention to book
travel online.
Another important predictor is a ‘positive’ attitude
toward booking online. A person’s attitudes are
strongly influenced by groups to which he or she
belongs so it is not surprising to see these two
variables emerging as key predictors together in this
model.
The role of consumers’ attitudes that this Model
reveals is supported with literature even when some
of the literature does not explicitly refer to travel
purchases. When consumers have affirmative
feelings and attitudes about the online medium and
using technology in general (Dabholkar, 1996),
(Dabholkar and Bobbitt 2001), (Li and Chen, 2009),
(Morrison et al., 2001) and have positive perceptions
about the financial benefits of booking online, they
are more likely to be online bookers. This is
especially the case if they are aware of other people
who booked online and if they have been using the
Internet for longer periods of time. Online
information sources from other consumers are
regarded as critical with experience products such as
travel products (Bei et al., 2004). Findings from the
Web User Survey also reveal that online purchasing
increases incrementally with online experience
(Georgia Institute of Technology, 1998).
The Model reveals that consumers who previously
booked travel products such as ‘destination
tour/attraction tickets’ demonstrate a greater
intention to book travel products online. It could be
there is a hierarchical structure of vacation planning
and purchasing where travelers book certain travel
products before others. One conclusion is that early
in the planning process travelers reduce uncertainty
by taking care of core elements of travel such as
transportation and accommodations (Beldona,
ICE-B 2011 - International Conference on e-Business
208

2003). In addition, according to one researcher, once
accommodation has been booked, the vacation
itinerary is relatively predetermined and fixed
(Hyde, 2008).
Survey respondents’ attitude of desirability
toward online travel booking is shaped by their
beliefs that online travel booking is safe using their
credit card, easy, enjoyable, and they highly regard
the importance of its convenience. A desirable
attitude coupled with a positive one are the key
variables leading to online travel booking intention.
Respondents believe it is more expensive and
inconvenient booking with an agent than a website.
Also, it is more enjoyable and easier booking with a
website.
Furthermore, in the Model, the variable “Internet
access other ~2” recorded an odds ratio of 1.589
meaning that the expected odds of booking travel
online (‘highly likely’ to book) is almost 1.6 times
greater among respondents indicating they had
access to the Internet asides from home or work,
than respondents who said they do not have such
access, controlling for all other factors in the model.
With categorical variables Stata creates k indicator
variable sets. The procedure is to omit the first group
of variables so it acts as a baseline for other
categories to help understand their odds ratios. Other
key variables from hypotheses tests H1g, H1i, H1j
and Hik are significant in this Final Model.
6 CONCLUSIONS
The predictive Final Model contains socio and
psychographic variables of the traveler’s decision
making process when booking travel products, and it
quantifies the strength of these variables through the
interpretation of odds ratios as indicated above. Such
a quantitative model for predicting travel products is
unique in travel and e-business literature and thus it
is a valued contribution.
In addition to uncovering these predictors, the
probability of group membership in each category
(online booking intention where, 1 = highly likely,
2 = likely, and 3 = somewhat likely) was determined
using logistic regression. When one combines these
probabilities with demographic parameters
(Conyette, 2011), it gives a website operator
valuable information for targeting consumers.
This study holds important strategic implications
for the travel industry, and the following are offered
to travel website operators and travel agents so they
will prosper in the new marketplace.
Some implications deal with consumers’ beliefs
and attitudes. Websites should note the perceptions
online consumers have that booking with a travel
website is positive. Understanding what particular
attributes of online shopping make people feel this
way is important. It could be that online bookers
regard the savings in time and money as a positive
benefit. Yet, these savings are not always
Table 1: Final Model.
realized by using a website, so the online retailer
should consider how to convey the belief that
savings will result. However, many online bookers
are more motivated by the rich information available
on the Web and the convenience of accessing it.
These bookers find the use of travel agents
inconvenient and probably inefficient. Offline
consumers often want the assistance and interaction
only a human could provide. Travel agencies are
best advised to make their services convenient,
efficient, useful and very personable.
Travel agents could be trained on the usage of
intelligent online tools and combine the assistance
these tools provide with the unique aspects of advice
that comes from a human touch (human
intelligence). Agents could think about inimitable
aspects of human knowledge, intelligence and
reasoning that cannot be currently provided by
online intelligent tools. Combining these methods in
novel and powerful ways will exceed the
expectations of consumers. Planning long haul or
complex trips seems to be the strength of travel
Ordered logistic regression Number of obs = 856
LR chi2(27) = 279.47
Prob > chi2 = 0.0000
Log likelihood = -709.04845 Pseudo R2 = 0.1646
------------------------------------------------------------------------------
Online booking intent| Odds Ratio Std. Err. z P>|z| [95% Conf. Interval]
-------------+----------------------------------------------------------------
Some of my friends _2 1.490089 .3075721 1.93 0.053 .9942923 2.233111
Some of my friends _3 2.635803 .5708602 4.47 0.000 1.724092 4.029633
Some of my friends _4 3.846541 1.003958 5.16 0.000 2.306239 6.415587
Belief convenientA_2 .8059347 .2195409 -0.79 0.428 .4725286 1.374585
Belief convenientA_3 .6573546 .1672914 -1.65 0.099 .3991864 1.082489
Belief convenientA_4 .6292642 .1552033 -1.88 0.060 .3880525 1.020412
Belief convenientA_5 .4359062 .1128951 -3.21 0.001 .262386 .724178
Belief convenientA_6 .5405856 .1592311 -2.09 0.037 .3034873 .9629159
Belief convenientA_7 .5648519 .1587146 -2.03 0.042 .3256568 .9797359
Belief expensiveA_2 1.037616 .2842878 0.13 0.893 .6064888 1.775212
Belief expensiveA_3 1.395234 .3590141 1.29 0.196 .8425983 2.310328
Belief expensiveA_4 1.234324 .3131459 0.83 0.407 .7507253 2.029447
Belief expensiveA_5 1.625916 .3987905 1.98 0.048 1.005363 2.629501
Internet access other~2 1.589834 .2889482 2.55 0.011 1.113392 2.270154
How long using net_2 .7402478 .1117285 -1.99 0.046 .5506833 .9950671
N
umber of trips~2 .6097605 .1305857 -2.31 0.021 .4007438 .9277945
Destination tour _1 .5910556 .1000929 -3.11 0.002 .4241122 .8237131
Hotels_1 .6086096 .122005 -2.48 0.013 .4108682 .9015193
Airline tickets_1 .5602512 .1240503 -2.62 0.009 .3630033 .8646794
Car rentals_1 .6281079 .1085975 -2.69 0.007 .4475738 .8814626
Long-distance train_1 .367288 .1489853 -2.47 0.014 .165855 .8133637
Attitude positiveW_2 1.654819 .3965239 2.10 0.036 1.034636 2.646754
Attitude positive W_3 1.968378 .5129192 2.60 0.009 1.181143 3.280305
Attitude positiveW_4 1.934779 .4825798 2.65 0.008 1.186647 3.154577
Attitude desirableW_2 1.188753 .25725 0.80 0.424 .7778375 1.816747
Attitude desirableW_3 1.6572 .3936764 2.13 0.033 1.040321 2.639868
Attitude desirableW_4 1.749352 .4169593 2.35 0.019 1.096459 2.791012
-------------+----------------------------------------------------------------
/cut1 -.0922823 .352925 -.7840026 .599438
/cut2 1.092407 .3544086 .3977792 1.787035
MODELING FACTORS THAT INFLUENCE ONLINE TRAVEL BOOKING
209

agencies (Law et al., 2004) and so consumers
desiring these vacations should be the targets of
agencies.
Davis et al., (1989) state that perceived usefulness
has a direct influence on behavioral intention.
Perceived usefulness and ease-of-use are important
(Dabholkar, 1996) and therefore marketers should
take these into consideration in the design of user-
friendly websites.
Attitudes, which are essentially a person’s mental
state of readiness (Zimbardo and Ebbesen, 1970)
have been used to predict and explain behavior in
many research studies (Trafimow and Sheeran,
2004); (Davis et al., 1989); (Fishbein, 1967); (Lord,
2004). Consumer attitudes change when people learn
by looking, listening or reading (Conyette, 2010).
Marketers can become sensitive to the varied
reasons underlying the attitude in question. This
study confirms the importance of attitudes and
subjective norms in TRA and quantifies the strength
of these in the Final Model.
REFERENCES
Bei, Chen & Widdows (2004). Consumers’ Online
Information Search Behavior and the Phenomenon of
Search vs. Experience Products. Journal of Family
and Economic Issues, 25(4), (Winter), 449-467.
Beldona, S. (2003, July). Examining the structure of
online pleasure travel planning. Paper presented at I-
CHRIE 2005 in Las Vegas, NV.
Conyette, M. (2010). “Determinants of Online Leisure
Travel Planning Decision Processes: A Segmented
Approach (Unpublished doctoral dissertation,
University of Newcastle, Newcastle)”.
Conyette, M. (2011). Demographics for Segmentation in
Online Travel, International Journal of Trade,
Economics and Finance, 2 (1), 93-98.
Dabholkar, P. A (1996). Consumer evaluations of new
technology-based self-service options: an investigation
of alternative models of service quality. International
Journal of Research in Marketing, 13 (1), 29-51.
Dabholkar, P. A, Bobbitt, L. M. (2001). Integrating
attitudinal theories to understand and predict use of
technology-based self-service: The Internet as an
illustration. International Journal of Service Industry
Management, 12 (5), 423-450.
Davis, F. D., Bagozzi, R. P., & Warshaw, P. R. (1989).
User acceptance of computer technology: A
comparison of two theoretical models. Management
Science, 35, 982–1003.
Fishbein, M. & Raven B. H. (1962). The AB scales: an
operational definition of belief and attitude. Human
Relations, 15, 35-44.
Fishbein, M. (1967). Readings in attitude theory and
measurement. New York, John Wiley & Sons, Inc.
Georgia Institute of Technology: Graphics, Visualization,
and Usability Center (GVU). (1998). GVU’s tenth
WWW survey. Atlanta: Author.
Hilbe, J. (2009). Logistic regression models, Norwell,
MA, Chapman & Hall/CRC Press.
Hyde, K. F. (2008). Tourist information processing and
touring planning theory. Unpublished manuscript.
Jarvelainen, J., Puhakainen, J., (2004). Distrust of one's
own web skills: A reason for offline booking after
online information search. Electronic Markets London
14, (4), 333-343.
Law, R., Leung, K. & Wong, J. (2004). The impact of the
internet on travel agencies, International Journal of
Contemporary Hospitality Management 16, (2), 100-
107.
Li, Y.M., Chen, C. W. (2009). A synthetical approach for
blog recommendation: Combining trust, social
relation, and semantic analysis. Expert Systems with
Applications, 36, 6536-6547.
Lord, C. G. (2004). The role of exemplar stability in
attitude consistency and attitude change. In Haddock,
G., & Maio, G. R. (Eds.), Contemporary Perspectives
on the Psychology of Attitudes (pp. 299-323). Hove,
UK: Psychology Press.
Morrison, A. M., Jing, S., O’Leary, J. T., Cai, L. A.
(2001). Predicting usage of the Internet for travel
bookings: An exploratory study. Information
Technology & Tourism, 4, 15-30.
Trafimow, D. & Sheeran, P. (2004). A theory about the
translation of cognition into affect and behavior. In
Haddock, G., & Maio, G. R. (Eds.), Contemporary
Perspectives on the Psychology of Attitudes (pp. 57-
75). Hove, UK: Psychology Press.
Zimbardo, P. & Ebbesen, E. B. (1970). Influencing
attitudes and changing behavior. Reading, Addison-
Wesley Publishing Company.
ICE-B 2011 - International Conference on e-Business
210