The international journal of tourism research tập 12, số 04, 2010 07 + 08

INTERNATIONAL JOURNAL OF TOURISM RESEARCH
Int. J. Tourism Res. 12, 307–320 (2010)
Published online 30 July 2009 in Wiley InterScience
(www.interscience.wiley.com) DOI: 10.1002/jtr.749

A Study of the Non-economic
Determinants in Tourism Demand
Vincent Cho*
Department of Management and Marketing, The Hong Kong Polytechnic University, Hung Hom,
Hong Kong
ABSTRACT
Traditionally, many studies have attempted
to use economic demand models. This paper
stresses on the influence of non-economic
factors on tourism demand. Some
researchers have suggested that tourists
from different origins have various cultural
and nationalistic backgrounds, and they
may interpret visual imagery and
experiences differently. Aligning with this
suggestion, we have investigated different
underlying factors of tourism demand from
four continents (Asia, the Americas, Europe
and Oceania). Statistical data are collected
from international organisations and 135
countries were covered. Our results showed
that there are differences and similarities
among the factors in determining the
tourism demand. Copyright © 2009 John
Wiley & Sons, Ltd.

Received 18 November 2008; Revised 25 June 2009; Accepted
30 June 2009

Keywords: tourism demand; non-economic
determinants; holistic approach.
INTRODUCTION

I

nternational tourism today has social, cultural and political significance, as well as
substantial economic benefits. In the last 50
years, tourism has emerged as one of the largest
and fastest growing industries in the world
(Eadington and Redman, 1991; WTO, 1992).
According to the World Tourism Organization
*Correspondence to: V. Cho, Department of Management
and Marketing, The Hong Kong Polytechnic University,
Hung Hom, Hong Kong.
E-mail:

(WTO), the number of international tourists
worldwide increased from 25 million in 1950
to 160 million in 1970, 429 million in 1990, 689
million in 2001, 846 million in 2006 and 1.6
billion by 2020. International tourism has experienced an overwhelming boom over the last
two decades and has now been called the
largest industry in the world. As a result of the
rise in the number of tourists, and the importance of the tourism sector for many countries
which have begun to channel their resources
into its development (Balaguer and CantavellaJorda, 2002), tourism demand analysis has

become increasingly important.
In general, the international tourism demand
model, which is based on classical economic
theory, is typically estimated as a function of
tourists’ income, tourism prices in a destination relative to those in the origin country,
tourism prices in the competing destinations
(i.e. substitute prices), exchange rates, transportation cost between destination and origin,
as well as dummy variables on various special
events and deterministic trends (e.g. Barry
and O’Hagan, 1972; Loeb, 1982; Stronge and
Redman, 1982; Uysal and Crompton, 1984;
Smeral, 1988; Di Matteo and Di Matteo, 1993;
Crouch, 1994; Lim, 1999; Croes, 2000; Vanegas
and Croes, 2000; Song et al., 2003; Chu, 2004; Li
et al., 2005; Song and Witt, 2006; Wong et al.,
2007; Chu, 2008; Song and Li, 2008). It postulates that factors of income and price are
likely to play a central role in determining the
demand for international tourism. As international tourism is generally regarded to be a
luxury commodity or service, it is not surprising that the study of such variables has
dominated past research.
There are three reasons why the discussed
economic framework needed to be extended.
First, from the consumers’ perspective, travelling overseas is one of the many options for
Copyright © 2009 John Wiley & Sons, Ltd.


308
them. Once a decision to travel has been made,
a consumer (tourist), faced with different alternatives, chooses a destination to maximise
utility. The tourist derives utility from spending time in a particular destination. The utility

stems from destinational attributes such as an
agreeable climate, beautiful scenery and/or
socio-cultural features. These attributes are
consumed along with other goods and services
available at the destination. The tourist’s utility
function represents the preferences for travelling abroad along with other goods and
services. This suggests that the choice of
destinations is a typical consumer choice
problem (Rugg, 1973; Divisekera, 1995). In this
vein, Naude and Saayman (2005) have devised
a utility function based on hotel capacity, air
distance, political stability, urbanisation rate,
etc. to estimate the tourist arrivals to Africa. It
was done using the regression analysis on a
cross-sectional data of five-year averages from
1996 to 2002.
Second, based on the theories of the
behaviour-intention model, including the
theory of planned behaviour and the theory of
reasoned actions, it states that the perceived
value and consequence of an action will affect
the behaviour of a person (Ajzen and Fishbein,
1980; Ajzen, 1991). Thus, the perceived image
of a destination will have an influence on the
intention and actions of a person (tourist) to
visit a destination. Empirically, Var et al. (1985)
showed that destination image of a convention
venue is directly proportional to the number
of delegates going to the convention.
Third, according to Sauran (1978), the main

difference between the economic and non-economic types of factors is that economic variables generally account for the total demand of
an origin country and that the role of non-economic variables has more to do with the types
of tourism. For instance, tourists in Thailand
may probably go for shopping and relaxation,
tourists in Europe may look for the historical
heritages. In this paper, we suggest to broaden
the investigation on the non-economic factors
based on the antecedent studies on destination
image to study tourism demand. This study
addresses the following research problems:
(1) to identify the potential factors influencing
the tourism demand;
Copyright © 2009 John Wiley & Sons, Ltd.

V. Cho
(2) to find out the significant underlying
factors of tourism demand; and
(3) to understand the tourism demand from
four continents (the Americas, Europe,
Asia and Oceania).
The organisation of this paper is as follows.
First, we review on the literature relating to the
potential antecedents of destination image and
formulate the framework for this study.
Second, we describe our data collection procedures and related analysis. Cross-sectional
data relating to tourism of 135 destination
countries are collected in this study. By applying regression and neural network analyses,
significant factors are sorted out. These factors
help to identify the most important factors
behind tourism demand. Interesting findings

and discussions are presented, and finally
there is a conclusion section.
LITERATURE REVIEW
According to Gearing et al. (1974), Ritchie and
Zins (1978) and Schmidt (1979), destination
image refers to an aggregated perception of
attributes which make the specific location
appealing as a potential destination to travellers. Leading image attributes identified are
nice climate, inexpensive goods and services,
safety, similar lifestyles, etc. To further understand the nature of destination image, we have
reviewed the literature as follows.
Gearing et al. (1974) have established an
overall measure of destination image for a
given region. These researchers proposed eight
factors including (i) accessibility of a region,
(ii) attitudes towards tourists, (iii) infrastructure of a region, (iv) price levels, (v) shopping
and commercial facilities, (vi) sport, recreation
and education facilities, (vii) natural beauty
and climate, and (viii) cultural and social characteristics. By combining the score relating to
the importance and actual perception of these
factors by tourists, an overall value of destination image can be derived.
Ritchie and Zins (1978) have conducted a
study on the importance of cultural and social
impact on destination image using a survey on
135 respondents. They identify four dimensions of cultural image of a tourism region:
Int. J. Tourism Res. 12, 307–320 (2010)
DOI: 10.1002/jtr


Non-economic Determinants in Tourism Demand

elements of daily life, remnants of the past,
good life and work habit.
Var et al. (1985) have studied the destination
image on convention tourism and found two
important factors that determine the number
of delegates. The first one is accessibility on
how close a convention is to the hometown of
a delegate, and the second one is the attractiveness of the convention location.
Lew (1987) attempted to group the factors
behind a destination image into three different
perspectives: (i) ideographic nature of a location focusing on its concrete description; (ii)
organisational nature stressing on the spatial,
capacity and temporal characteristics of a location; and (iii) cognitive nature describing the
perceptions and experience of tourists.
Getz (1993) applied the framework of destination image on business tourism. He compared the tourism business districts in Niagara
Falls (Ontario and New York) using underlying factors such as location, accessibility,
design, attractions and services. He concluded
that in order to be a good district for business
tourism, it should have three essential elements: (i) core attractions; (ii) central business
district functions; and (iii) supporting
services.
Utilising multidimensional scaling, Kim
(1998) determines the relative positions of five
well-known Korean national parks in terms of
selection criteria and the tourists’ psychological reception to the areas. He derived six
features namely seasonal and cultural attractiveness, clean and peaceful environment,
quality of accommodations and relaxing facilities, family-oriented amenities and safety,
accessibility and reputation, and entertainment
and recreational opportunities as the most
important factors influencing the destination

image.
Chen and Hsu (2000) measured the perceived image of South Korean tourists and
found that travel cost, destination lifestyle,
quality restaurants, freedom from language
barriers and availability of interesting places to
visit affects the destination choice behaviour of
a Korean tourist.
Recently, Russo and Borg (2002) used a case
study to analyse the destination image for cultural tourism in four European cities (Lyon,
Lisbon, Rotterdam and Turin). They found
Copyright © 2009 John Wiley & Sons, Ltd.

309
that these four cities, besides their own features to attract culture tourists, should pay
attention to those intangible elements, such as
transportation facilities, information centre
and quality of human capital in order to
enhance location attractiveness.
Getz and Brown (2006) explored the underlying factors for a region on wine tourism.
Using an extensive survey on the perception
on the importance of different features such as
‘the wine region is close to home’, and ‘the
region is popular with wine tourists like me’,
they found out there are five emerging factors:
(i) core wine product; (ii) core destination
appeal; (iii) core cultural product; (iv) variety;
and (v) tourist oriented, and that these factors
would define the image of a wine region.
Relating to the economic environment, Han
et al. (2006) found that price competitiveness is

an important factor influencing Americans
travelling to France, Italy and Spain, but not
the UK. However, as US expenditure rises, the
market shares of Spain and the UK decline,
while France and Italy benefit. Last, but not
least, Gallarza et al. (2002) presented an extensive review on destination image and proposed
a more comprehensive framework of destination image which contains cognitive elements,
time elements and distance elements.
RESEARCH FRAMEWORK
In our review on the literature, we classify these
attributes into five categories: (i) attitude
towards tourism; (ii) richness of tourism products/services; (iii) tourism support; (iv) environmental factors; and (v) economic factors. The
attitude of people in the destination towards
tourists and their social index are under the first
factor — attitude towards tourism. Richness of
tourism products/services includes the natural
and cultural heritage of a destination, and the
entertainment and recreational facilities in the
destination. Tourism support relates to adequacy of accommodation facilities, accessibility,
road network infrastructure and safety of a destination. The fourth category concerns the environmental factors such as seasonality of a
destination. The economic factors consist of
price levels of a destination as well as the gross
domestic product (GDP) of the source countries. Table 1 shows the summaries of related
Int. J. Tourism Res. 12, 307–320 (2010)
DOI: 10.1002/jtr


310

V. Cho

Table 1. Underlying antecedents of destination image
Attitude
towards
tourism
Tourism
products/
services

Tourism
support

Environmental
factors
Economic
factors

Attitude towards tourists (Gearing et al., 1974; Ritchie and Zins, 1978; Schmidt, 1979;
Gallarza et al., 2002; Getz and Brown, 2006)
Social factors (Gearing et al., 1974; Ritchie and Zins, 1978; Schmidt, 1979)
Natural heritage (Lew, 1987; Getz and Brown, 2006; Getz, 1993; Gearing et al., 1974; Ritchie
and Zins, 1978; Schmidt, 1979)
Cultural heritage (Getz and Brown, 2006; Kim, 1996; Gearing et al., 1974; Ritchie and Zins,
1978; Schmidt, 1979)
Entertainment and recreational facilities
Shopping and relaxing facilities (Getz, 1993; Gearing et al., 1974; Ritchie and Zins, 1978;
Schmidt, 1979)
Sport and recreational facilities (Kim, 1998; Gearing et al., 1974; Ritchie and Zins, 1978;
Schmidt, 1979)
Accommodation (Kim, 1996)

Accessibility of a region (Russo and Borg, 2002; Var et al., 1985; Kim, 1998; Lew, 1987; Getz,
1993; Gearing et al., 1974; Ritchie and Zins, 1978; Schmidt, 1979; Gallarza et al., 2002)
Road network infra-structure of a region (Getz, 1993; Gearing et al., 1974; Ritchie and Zins,
1978; Schmidt, 1979)
Clean, peaceful and safe environment (Kim, 1998)
Tourism openness (Russo and Borg, 2002)
Seasonality and Climate (Lew, 1987; Kim, 1998; Gallarza et al., 2002)
Price levels (Han et al., 2006)

studies investigating the underlying factors
affecting destination image. We postulate that
these factors are also influential to tourism
demand.
As this study concerns tourism demand from
four different regions (Americas, Europe, Asia
and Oceania) instead of from individual countries, it is hard to estimate the GDP or the
exchange rates in those regions as a whole. Thus,
the GDP and the exchange rates are not considered in this study. Moreover, it is hard to compare
the consumer price index (CPI) of all 135 countries because different places have their own
preferences of goods and services as well as on
the weightings of those goods and services. Thus,
we also exclude the CPI in this study. Nevertheless, without the economic factors, our findings
would focus on the non-economic aspects and
would have limited implications.
DATA COLLECTION
In this study, we sourced reliable secondary
data from different international organisations
such as the WTO and World Development Indicators (WDI) from the World Bank.
Copyright © 2009 John Wiley & Sons, Ltd.

Initially, statistical data in 2005 from 214 countries and territories were collected. These
achieved data were reported in the yearbooks
from different international organisations in
2007. However, there are 29 countries or territories that did not have tourist arrival data
from four continents, they are Afghanistan,
Canary Islands, Ceuta, Channel Islands,
Cote d’Ivoire, Democratic Peoples Republic of
Korea, Democratic Republic of Timor-Leste,
Djibouti, Equatorial Guinea, Falkland Islands,
French Guiana, Gibraltar, Greenland, Guernsey, Isle of Man, Jersey, Liberia, Madeira Island,
Mauritania, Mayotte, Melilla, Netherlands
Antilles, Saint-Pierre and Miquelon, San
Marina, Solomon Islands, Somali Democratic
Republic, Turkmenistan and Western Sahara.
Also, there are 25 countries or territories that
did not have tourist arrival data from at least
one continent (mainly from Oceania), they are
Anguilla, Antigua and Barbuda, Argentina,
British Virgin Islands, Cape Verde, Republic of
the Congo, Curacao, Gambia, Guadeloupe,
Guyana, Haiti, Luxembourg, Martinique,
Mexico, Montserrat, Namibia, Norway, Qatar,
Reunion, Saba, Saint Helena, Saint Maarten,
Int. J. Tourism Res. 12, 307–320 (2010)
DOI: 10.1002/jtr


Non-economic Determinants in Tourism Demand
Saint Vincent and the Grenadines, and Sao
Tome and Principe. Owing to government

policy, 25 countries have not recorded some
variable data such as social index, tourism
openness index, etc. They are American Samoa,
Andorra, Aruba, Bermuda, Bonaire, Cayman
Islands, Cook Islands, Dominica, Eritrea,
French Polynesia, Grenada, Guam, Micronesia, Marshall Islands, New Caledonia, Niue,
North Mariana Island, Palau, Palestine, Puerto
Rico, Saint Kitts and Nevis, Saint Lucia, Turks
and Caicos Islands, Tuvalu, and United States
Virgin Islands. Those 79 countries or territories
were neglected from the data set, which means
135 countries in total remained (as indicated in
appendix A). The collected data of the 135
countries were analysed in order to sort out the
determinants of tourism demand.
Tourism demand statistics
Most destinations have used the same destination images or enticements to attract tourists
regardless of their country of origin (Bonn et
al., 2005). Previous research has suggested that
nationals of various geographic regions interpret visual imagery and experiences differently dependent on their country of origin
(Berlyne, 1977; Britton, 1979; Thurot and
Thurot, 1983). In order to investigate any differences among tourists from different origins,
we collect the data on tourism demand from
four continents: the Americas, Asia, Europe
and Oceania. The tourist arrival statistics of
135 countries in 2005 were reported in the
yearbook of tourism statistics published in
2007 by the WTO. The WTO is a leading international organisation in the field of tourism
and serves as a global forum for tourism policy
issues and a practical source of tourism

know-how.
Due to the functional form of the demand
model usually in terms of powers on those
underlying factors, we transformed the data
using the natural logarithm, which is a common
practice on most tourism demand studies. In
the last three decades, many studies have
assumed a multiplicative form of model made
linear by a logarithmic transformation of the
variables (Loeb, 1982; Stronge and Redman,
1982; Summary, 1983; Arbel and Ravid, 1985;
Witt and Martin, 1987; Poole, 1988; Croes, 2000;
Copyright © 2009 John Wiley & Sons, Ltd.

311
Vanegas and Croes, 2000; Song et al., 2003;
Song and Witt, 2006; Wong et al., 2007; Song
and Li, 2008).
For the analysis of the underlying noneconomic factors, we have captured the data
on social factors, natural and cultural heritage,
accessibility, road network infrastructure,
climate and distance from the origin continent
of those 135 countries in 2005. On the other
hand, data on safety and crime rate are only
available in a few countries and factors such as
lifestyle, government policy and intervention
cannot be easily measured in a numerical
sense. Hence, these factors were neglected in
this study. The details of the independent
variables are elaborated as follows.

Accessibility
Accessibility is a significant attribute to destination image (Gearing et al., 1974; Ritchie and
Zins, 1978; Schmidt, 1979; Var et al., 1985; Lew,
1987; Getz, 1993; Kim, 1998; Gallarza et al.,
2002; Russo and Borg, 2002). In this paper,
accessibility by air is proxy by the takeoffs
abroad of air carriers registered in the country.
The unit of registered carrier departures worldwide in this study is the number of carriers.
The data were collected from the WDI which
is the premier data source on the global
economy from the World Bank. It contains statistical data for over 550 development indicators and time series data from 1960 onwards
for over 220 countries and country groups with
populations of more than 1 million, as well as
for China and Taiwan. Natural logarithm was
applied before fitting into the demand model.
Accessibility by road is proxy by the total
road network which includes motorways,
highways, and main or national roads, secondary or regional roads, and all other roads in a
country. The unit of total road network in this
study is kilometre (km). The data were collected from the WDI. Natural logarithm was
used before fitting into the demand model.
Environmental condition
Kim (1998) has indicated that environmental
condition is a significant attribute to destination image. In this vein, we have included the
variable Carbon dioxide (CO2) emissions which
Int. J. Tourism Res. 12, 307–320 (2010)
DOI: 10.1002/jtr


312

V. Cho

are those stemming from the burning of fossil
fuels and the manufacture of cement. They
include carbon dioxide produced during consumption of solid, liquid, and gas fuels and gas
flaring. The unit of Carbon dioxide (CO2) emissions in this study is kilotons (kt). The data
were collected from the WDI. Natural logarithm was taken before fitting into the demand
model.
Travelling cost
In this study, we include the distance between
the origin region and the destination country,
which is a proxy for transport cost and effort
(Tremblay, 1989). Laber (1969) found that distance between ‘origin’ and ‘destination’ plays
a significant role as a determinant of tourism
demand. As our collected statistics only report
the tourism arrivals from a region — Asia,
Europe, the Americas or Oceania, thus we
need to manipulate the average distance of a
country from a region. First, we assume that
the distance between two countries is the distance between two countries’ capitals. From
the haversine formula as shown in Equation
(1) (Sinnott, 1984), let φs, λs; φf, λf be the geographical latitude and longitude of two points
respectively, and Δλ be the longitude difference. Hence, Δθ is the (spherical) angular difference/distance as follows,

Δq = 2 arcsin

⎛ 2 ⎛ φf − φs ⎞
⎜⎝ sin ⎜⎝ 2 ⎟⎠
⎛ Δl ⎞ ⎞

+ cos φs cos φf sin 2 ⎜
⎝ 2 ⎟⎠ ⎟⎠

(1)

The shape of the Earth more closely resembles a flattened spheroid with extreme values
for the radius of arc of 6335.437 km at the
equator (vertically) and 6399.592 km at the
poles, and having an average great-circle
radius of 6372.795 km (3438.461 nautical miles).
Using a sphere with a radius, r, of 6372.795 km,
thus results in an error of up to about 0.5% and
the distance between two points of the Earth is
equal to r Δ θ.
A matrix with 135 rows and 135 columns is
formed containing the distances among the 135
countries. Then we group the countries according to their continent and take the mean distance of the countries on the same continent.
Copyright © 2009 John Wiley & Sons, Ltd.

For instance, the distance from England to Asia
is calculated by averaging the distance between
London (England’s capital city) and all the 38
countries such as China, Hong Kong and Japan
in Asia using the locations of their capital cities.
Appendix A shows our list of countries (38
countries in Asia, 23 countries such as USA,
Canada and Mexico in the Americas, 7 countries such as Australia, New Zealand and Fiji
in Oceania, and 33 countries such as Hungary,
Latvia, Norway and the UK in Europe) in different origins for the manipulation of average
distance.

Cultural and natural heritages
Cultural and natural heritages are found to be
important attributes on destination images
(Gearing et al., 1974; Ritchie and Zins, 1978;
Schmidt, 1979; Lew, 1987; Kim, 1996; Getz and
Brown, 2006). The numbers of cultural and
natural world heritage were collected from the
United Nations Educational, Scientific and
Cultural Organization (UNESCO) World
Heritage Committee, which consists of representatives from 21 of the States Party to the
Convention elected by their General Assembly
for terms up to six years. It determines whether
a property is inscribed on the World Heritage
List which includes 644 cultural, 162 natural
and 24 mixed properties with outstanding universal value. Similar to the effect of taking
natural logarithm, Table 2, which is referenced
from Wikipedia, was used to transform the
heritage counting.
Seasonality and climate
Seasonality is a well-documented issue in the
literature, particularly in relation to cold-water
regions of Europe and North America (Aguiló
and Sastre, 1984; Snepenger et al., 1990; Donatos
and Zairis, 1991; Jeffrey, 1999; Kennedy, 1999;
Baum and Lundtorp, 2001). Although the
reasons for a significant variation in demand
are also well documented (the climate, institutional patterns like school or calendar holidays,
lifestyles, special events, etc.), there are a few
studies on tourism seasonality (Butler, 2001).
Belen-Gomez-Martin (2005) suggested that

weather and climate are significant in explaining tourism demand.
Int. J. Tourism Res. 12, 307–320 (2010)
DOI: 10.1002/jtr


Non-economic Determinants in Tourism Demand

313

Table 2. Transformation on the number of cultural and natural
heritages
Cultural heritage

Number
Level

1–4
1

5–19
2

20–29
3

30+
4

Natural heritage

Number
Level

1
1

2–3
2

4–9
3

10+
4

Tyndall Centre, which brings together scientists, economists, engineers and social scientists, who together are working to develop
sustainable responses to climate change, stored
the daily mean temperature data by country.
In advance, they calculated the monthly mean
temperature by taking the average value of
daily mean temperature in every month. We
can browse the monthly mean temperature
data from their website by country. The unit of
monthly mean temperature in this study is in
Celsius. The data were collected from the
Tyndall Centre for Climate Change Research.
Standard deviation of monthly mean temperature was calculated by taking the standard
deviation of monthly mean temperature in 12
months. If the standard deviation is large, this
would imply the destination country has a

wide spread of temperature within a year or a
clear seasonality.
Social Index
As identified by Gearing et al. (1974), Ritchie
and Zins (1978) and Schmidt (1979), social
factor is an important attribute on destination
image. Social Index, which is an aggregate
social index, combining the Human Development Index, Newspaper Index, Personal Computer Index, and Television Index, was collected
from the World Travel and Tourism Council.
Population
Population counts all residents regardless of
legal status or citizenship — except for refugees not permanently settled in the country of
asylum as they are generally considered part
of the population of their country of origin.
With higher population, which would mean a
larger coverage in area and more tourists,
hence we attempt to control this variable so as
to reveal the significance of the other influenCopyright © 2009 John Wiley & Sons, Ltd.

tial factors. The data were collected from the
WDI. Natural logarithm was used before fitting
it into the demand model. In sum, Table 3
highlights the above factors which would be
grouped into the relevant category as indicated
in Table 1.
ANALYSIS AND FINDINGS
Recently, Lim (1997), Morley (1996, 1998, 2000),
Turner et al. (1998) and Turner and Witt (2001)
surveyed more than 100 international tourism
demand studies that have attempted to model

the demand for tourism. Most studies are time
series econometric models such as the almost
ideal demand system and autoregressive distributed lag model estimated using multiple
least-squares regression, which is appropriate
for stationary time series data (Kulendran and
Witt, 2001). As this study tries to investigate
the tourism demand from another perspective
using the cross-sectional data from 135 countries, we applied regression and neural network
for the analysis to determine the most influential factors on tourism demand.
As the data are cross-sectional, common
time series models are not applicable. Regression is a simple yet robust linear analysis
method that is capable of identifying important factors, which are assumed to be linearly
related to the dependent variable. However, it
is not appropriate if the underlying relationship is non-linear. Because of this limitation,
neural network analysis, a well-developed
technique which would handle both linear and
non-linear data in artificial intelligence studies,
is also applied to check the reliability of the
results from linear regression. The results of
regression and neural network are shown in
Tables 4 to 7. Those significant factors in the
regression are sorted according to the weighting from the result of the neural network analysis and those insignificant factors are shown
Int. J. Tourism Res. 12, 307–320 (2010)
DOI: 10.1002/jtr


314

V. Cho

Table 3. Potential factors behind tourism demand
Influential factors

Independent variables

Demographic of the destination country
Accessibility by air (Russo and Borg, 2002; Var et al., 1985; Kim, 1998;
Lew, 1987; Getz, 1993; Gearing et al., 1974; Ritchie and Zins, 1978;
Schmidt, 1979; Gallarza et al., 2002)
Cultural heritage (Getz and Brown, 2006; Kim, 1996; Gearing et al.,
1974; Ritchie and Zins, 1978; Schmidt, 1979)
Natural heritage (Lew, 1987; Getz and Brown, 2006; Getz, 1993;
Gearing et al., 1974; Ritchie and Zins, 1978; Schmidt, 1979)
Environmental condition (Kim, 1998)
Infrastructure on road network (Getz, 1993; Gearing et al., 1974;
Ritchie and Zins, 1978; Schmidt, 1979)
Social factor (Gearing et al., 1974; Ritchie and Zins, 1978; Schmidt,
1979)
Seasonality and Climate (Lew, 1987; Kim, 1998; Gallarza et al., 2002)

Population
Registered aircraft departures
Cultural world heritage
Natural world heritage
Carbon dioxide emissions,
distance
Total road network
Social index
Average of monthly mean
temperature, Standard deviation

of monthly mean temperature

Table 4. Regression and neural network analyses on tourist arrival from Americas
Regression
(R2 = 0.727)
Tourist arrival from the Americas
Registered Aircraft Departures
Population
Social Index
Average distance from the Americas
Standard Deviation (Temp)
Roads, total network
CO2 Emissions
Natural world heritage
Cultural world heritage
Mean (Temp)

Standard coefficient

Significance

ANN
(Accuracy = 90.1)
Relative importance

0.374
0.395
0.342
−0.334
−0.252

0.186
0.142
0.088
−0.051
−0.050

0.000
0.000
0.000
0.000
0.046
0.086
0.316
0.148
0.488
0.581

0.298
0.251
0.235
0.223
0.173
0.080
0.116
0.122
0.069
0.007

at the end of the list. From Tables 4 to 7, the
significant factors from the regression are

rather consistent with those of high relative
importance in the neural network analysis. All
the insignificant factors have relative importance of less than 0.2 in the neural network
analysis (except the CO2 emission on the tourist
arrival from Americas). Hence, our results are
deemed to be reliable and trustworthy.
DISCUSSION
From the regression and data mining analyses
as shown in Tables 4 to 7, there are two common
Copyright © 2009 John Wiley & Sons, Ltd.

factors among the tourists from the four continents — distance from the origin (all betas are
negative and significant) and aircraft departure (all betas are positive and significant).
That is, most tourists prefer to visit proximal
countries with good accessibility. In this regard,
we suspect that travelling to proximal countries, which are usually associated with less
time and financial effort, would be a dominant
factor in selecting a destination to visit. It is
nice to have a mix of short and long haul trips
during a year. Usually, the number of short
hauls would be greater than the number of
long hauls for a normal traveller, unless
Int. J. Tourism Res. 12, 307–320 (2010)
DOI: 10.1002/jtr


Non-economic Determinants in Tourism Demand

315

Table 5. Regression and neural network analyses on tourist arrival from Asia
Regression
(R2 = 0.765)
Tourist arrival from Asia
Average distance from Asia
Population
Registered aircraft departures
CO2 emissions
Roads, total network
Social index
Natural world heritage
Cultural world heritage
Standard deviation (Temp)
Mean (Temp)

Standard coefficient

Significance

ANN
(Accuracy = 89.2)
Relative importance

−0.385
0.385
0.350
0.319
0.337
0.245
0.148

0.153
−0.049
0.050

0.000
0.000
0.000
0.018
0.002
0.005
0.010
0.025
0.541
0.553

0.316
0.302
0.288
0.225
0.210
0.145
0.099
0.081
0.102
0.036

Table 6. Regression and neural network analyses on tourist arrival from Europe
Regression
(R2 = 0.801)
Tourist arrival from Europe

CO2 emissions
Average distance from Europe
Mean (Temp)
Cultural world heritage
Standard deviation (Temp)
Registered aircraft departures
Social index
Population
Natural world heritage
Roads, total network

Standard coefficient

Significance

ANN
(Accuracy = 90.6)
Relative importance

0.436
−0.374
−0.188
0.172
−0.158
0.188
0.108
0.004
0.047
0.147

0.000
0.000
0.016
0.008
0.027
0.015
0.178
0.968
0.373
0.128

0.423
0.293
0.242
0.130
0.113
0.108
0.108
0.101
0.043
0.021

Table 7. Regression and neural network analyses on tourist arrival from Oceania
Regression
(R2 = 0.614)
Tourist arrival from Oceania
Registered aircraft departures
Average distance from Oceania
Social index
Population

CO2 emissions
Cultural world heritage
Mean (Temp)
Standard deviation (Temp)
Natural world heritage
Roads, total network
Copyright © 2009 John Wiley & Sons, Ltd.

Standard coefficient

Significance

ANN
(Accuracy = 88.3)
Relative importance

0.571
−0.264
0.231
0.287
−0.283
0.145
−0.086
−0.042
0.010
0.087

0.000
0.000
0.036

0.027
0.093
0.104
0.538
0.664
0.891
0.616

0.325
0.223
0.203
0.093
0.113
0.107
0.062
0.052
0.052
0.036

Int. J. Tourism Res. 12, 307–320 (2010)
DOI: 10.1002/jtr


316
he/she is travelling on business or for other
purposes.
Registered aircraft departure is a significant
and important factor to tourists from the four
continents. We can see that the accessibility of
countries is the main concern to most international tourists. According to the World Bank

Group, there are about 48.84% (11 816 848 aircrafts) of total aircraft takeoffs, domestic or
abroad, registered in the Americas in 2004. We
can see that tourists from the Americas are
likely to travel by aircraft. This figure is coherent to our finding, registered aircraft departures is a significant factor to tourists from the
Americas. Geographically, Oceania is separated from other continents by Ocean. If
Oceania tourists travel to other continents,
they need to travel by air or ship. Oceania tourists are much more likely to go to countries
that have airline connections from their home
countries. By investing in infrastructure on
airport facilities, a country seems to secure a
niche position for tourists from the Americas
and Oceania. For European tourists, they
usually take the aircrafts for their travel. Nevertheless, the total road network of a country
is not a significant factor except for those Asian
tourists (as shown in Tables 4 to 7). This is due
to the fact that most travellers visit the major
cities of a country which are usually well connected by air. Thus, as a traveller, one does not
need to worry about the details of its road
network within a country.
Social factor is found to be another important factor for tourists from Americas, Asia
and Oceania. This implies social culture, which
embeds the culture of hospitality, is an important issue in attracting tourism. People in a
sociable country would act hospitable, that is,
the reception and entertainment of guests,
visitors or strangers, with liberality and
goodwill.
Regarding the cultural and natural heritages,
our analyses found that cultural heritage, as
indicated in Tables 5 and 6, is a significant
factor to Asian and European tourists, and

referring to Table 5, countries with natural
heritages seem to attract Asian tourists. It is
rather reasonable to say that Europe tourists
are concerned about cultural world heritage.
Countries in Europe had 52% (315 heritages)
of cultural heritages in the world during 2007
Copyright © 2009 John Wiley & Sons, Ltd.

V. Cho
(UNESCO World Heritage Committee). Actually, Europeans are proud of their own culture.
They pay much attention and put a lot of effort
into protecting their cultural world heritages.
Even in travel, they also view cultural heritage
as the most significant factor in selecting a
country to visit.
For seasonality, tourists from the Americas
and Europe like to visit those destinations with
mild variations of temperature or seasonal
changes (refer to Tables 4 and 6). According to
our result, large temperature variation or distinctive seasons in a country have a negative
impact on the total tourist arrivals in a year.
Usually people may like to visit a place during
its best season in terms of its climate and
natural beauty. For example, if a country only
has warm weather for a few months in summer,
tourist may likely visit there during these few
months. As a result, there is a heavy slump of
tourist arrivals in other seasons. On the other
hand, as from Table 6, the average temperature
of a country has a significant and negative

impact on European tourist arrivals. That is,
the lower the average temperature of a country,
the number of tourists arriving from Europe
would be higher. This would be because most
Europeans like to go skiing.
Concerning the environmental condition,
CO2 emission, as shown in Tables 5 and 6, is a
positive and significant factor for both European and Asian tourists in selecting a country
to visit. Certainly, a person may not like to visit
a country that is heavily polluted. Given the
fact that most people do not recognise the
figures of CO2 emission of a country, instead,
they usually have perceptions on whether a
country is a fabulous one or not. Naturally,
some tourists may like to visit a country that
is economically strong, which is somehow
positively associated with its CO2 emission.
This is due to the consumption of gasoline in
the production and transportation process. In
those economically renowned cities which are
usually characterised by busy traffic, the CO2
emissions from automobiles are also significant. In this regard, we suspect that both European and Asian tourists would like to visit
some countries that are economically strong.
This would explain why Hong Kong, Singapore and other economically developed countries attract so many Asians and Europeans
Int. J. Tourism Res. 12, 307–320 (2010)
DOI: 10.1002/jtr


Non-economic Determinants in Tourism Demand
every year. However, the CO2 emissions in

those places are also relatively high. In sum,
this paper has contributed to the literature by
including the non-economic factors to study
tourism demand and identified factors of
tourism demand from different continents.
LIMITATION AND DIFFICULTIES
The availability of tourist arrival data from the
Middle East is much lower than other continents. A lot of countries do not further differentiate their tourist arrival from the Middle
East; some of them will mark those arrivals
from the Middle East as ‘other area’. Therefore,
tourist arrival data from the Middle East are
not suitable to be included in this study.
We may think that the factors behind location attractiveness may vary according to the
different purposes of tourists. Most people
travel abroad on holiday, to visit their friends
or family members, to study and for business
(UNWTO). At the beginning, we wanted to
further subdivide the tourist arrival data,
according to their purpose, as a dependent
variable. However, those data were not available in UNWTO.
The aircraft departure, which may not be a
truly exogenous variable, would limit the
interpretation of this study. In some sense,
tourism demand is determined by aircraft
departure; nevertheless, it may also be true
that it determines the aircraft departure.
Somehow, aircraft departure is affected by the
government policy in a country. According to
Bieger and Wittmer (2006), aircraft departure
often is limited by geographic restriction, or by

ensuring local environment is maintained.
Some governments would promote tourism
intensively by various policies such as exempting visa requirement and sales tax rebate.
These policies would probably inject a higher
number on aircraft departure and bring along
high number of tourists. Idealistically, government policy, if it can be quantified, would be
used as an instrumental variable to elaborate
the impact of aircraft departure to tourism
demand in a more accurate way. In the future
study, we should identify some instrumental
variables, which would influence both aircraft
departure and tourism demand, and make
a better estimate of the impact of aircraft
Copyright © 2009 John Wiley & Sons, Ltd.

317
departure to tourism demand. Right now, this
is a limitation in our study.
On the other hand, it would make more
sense if the future study could estimate the
model using panel data approach with units
being the individual source markets. By doing
this, the problem associated with the aggregate
models may disappear and the GDP and price
variable (CPI) would be considered in the
model. Right now, the demand models, which
focus on the non-economic factors, are not
comparable with the demand models in the
literature.
CONCLUSION

This paper has collected data from 135 countries and investigated the non-economic underlying factors on tourism demand from four
continents. Rigorous data pre-processing has
been done so as to make the data normalised.
Both the regression analysis and neural
network have shown consistent results and
our findings showed there are different significant factors based on tourism demand from
different continents. We have contributed to
the tourism theories in two aspects. First,
besides those well-studied economic factors,
the non-economic factors are also significant
to tourism demand. Second, the underlying
factors of tourism demand are different for
tourists from different origins.
In terms of practice, government and tourism
bureaus in different countries and territories
view tourism as an important industry; they
invest lots of resources so as to attract more
tourist arrivals. This may include the preservation and restoration of their heritage, build
tourist-related infrastructure (e.g. airport, road,
rail and wharf), provide attractive conditions
to enterprises building theme parks and hotels
or promote their location through marketing
campaigns around the world. This paper has
identified that people from different regions
have their own preference in selecting a destination. While Europeans and Asians like to
visit a destination for its cultural heritage,
Asians also prefer a destination with a natural
heritage. Tourists from the Americas like to
visit a proximal country with a sociable environment that would be accessible by air. In
general, investments on building accessibility

Int. J. Tourism Res. 12, 307–320 (2010)
DOI: 10.1002/jtr


318
by air, maintaining heritage would secure
tourism demand in a destination. Last, but not
least, tourism operators should pay attention
to the origins of tourists and take care of their
preferences.
REFERENCES
Aguiló E, Sastre A. 1984. La Medición de la Estacionalidad Del Turismo: el caso de Baleares. Estudios
Turísticos 81: 79–88.
Ajzen I. 1991. The theory of planned behavior. Organizational Behavior and Human Decision Processes
50(2): 179–211.
Ajzen I, Fishbein M. 1980. Understanding Attitudes
and Predicting Social Behavior. Prentice-Hall:
Englewood Cliffs, NJ.
Arbel A, Ravid SA. 1985. On recreation demand: a
time series approach. Applied Economics 17:
979–990.
Balaguer J, Cantavella-Jorda M. 2002. Tourism as
a long-run economic growth factor: the Spanish
case. Applied Economics 34: 877–884.
Barry K, O’Hagan J. 1972. An econometric study of
British tourist expenditure in Ireland. Economic
and Social Review 3(2): 143–161.
Baum T, Lundtorp S. 2001. Seasonality in Tourism.
Pergamom: Oxford.
Belen-Gomez-Martin M. 2005. Weather, climate

and tourism, a geographical perspective. Annals
of Tourism Research 32(3): 571–591.
Berlyne D. 1977. Psychological aesthetics. In Handbook of Cross-Cultural Psychology, Vol. 3, Triandis
H, Lamberts W (eds). Ally Bacon: Boston;
323–326.
Bieger T, Wittmer A. 2006. Air transport and tourism
next term — perspectives and challenges for
destinations, airlines and governments. Journal
of Air Transport Management 12(1): 40–46.
Bonn MA, Sacha MJ, Mo D. 2005. International
versus domestic visitors: an examination of
destination image perception. Journal of Travel
Research 43(3): 294–301.
Britton R. 1979. The image of the third world in
tourism marketing. Annals of Tourism Research 6:
318–329.
Butler R. 2001. Seasonality in tourism: issues and
implications. In Seasonality in Tourism, Baum T,
Lundtorp S (eds). Pergamom: Oxford; 5–
21.
Chen JS, Hsu CHC. 2000. Measurement of Korean
tourists’ perceived images of overseas destinations. Journal of Travel Research 38: 441–416.
Chu FL. 2004. Forecasting tourism demand: a cubic
polynomial approach. Tourism Management 25(2):
209–218.
Copyright © 2009 John Wiley & Sons, Ltd.

V. Cho
Chu FL. 2008. A fractionally integrated autoregressive
moving average approach to forecasting tourism

demand. Tourism Management 29(1): 79–88.
Croes RR. 2000. Anatomy of Demand in International
Tourism: The Case of Aruba. van Gorcum: Assen,
The Netherlands.
Crouch G. 1994. The study of international tourism
demand: a survey of practice. Journal of Travel
Research 32(4): 41–55.
Di Matteo L, Di Matteo R. 1993. The determinants of
expenditures by Canadian visitors to the United
States. Journal of Travel Research Spring: 34–42.
Divisekera S. 1995. Economics of international travel
and tourism: an Australian perspective. PhD thesis
in Economics, La Trobe University: Australia.
Donatos G, Zairis P. 1991. Seasonality of foreign
tourism in the Greek island of Crete. Annals of
Tourism Research 18: 515–519.
Eadington WO, Redman M. 1991. Economics of
tourism. Annals of Tourism Research 18: 41–56.
Gallarza MG, Saura IG, Garcia HC. 2002. Destination image: towards a conceptual framework.
Annals of Tourism Research 29(1): 56–78.
Gearing CE, Swart WW, Var T. 1974. Establishing
a measure of touristic attractiveness. Journal of
Travel Research 12(4): 1–8.
Getz D. 1993. Planning for tourism business districts. Annals of Tourism Research 20: 583–600.
Getz D, Brown G. 2006. Critical success factors for
wine tourism regions: a demand analysis. Tourism
Management 27: 146–158.
Han Z, Durbarry R, Sinclair MT. 2006. Modeling
US tourism demand for European destinations.
Tourism Management 27: 1–10.

Jeffrey D. 1999. An analysis of the nature, causes
and marketing implications of seasonality in the
occupancy performance of English hotels. Tourism
Economics 5: 69–91.
Kennedy E. 1999. Seasonality in Irish tourism.
Tourism Economics 5: 25–47.
Kim H. 1996. Perceptual mapping of attributes and
preferences: an empirical examination of hotel
F&B products in Korea. International Journal of
Hospitality Management 15(4): 373–91.
Kim H. 1998. Perceived attractiveness of Korean
destinations. Annals of Tourism Research 25(2):
340–361.
Kulendran N, Witt S. 2001. Cointegration versus
least squares regression. Annals of Tourism Research
28(2): 291–311.
Laber G. 1969. Determinants of international travel
between Canada and the United States. Geographical Analysis 1(4): 329–336.
Lew AA. 1987. A framework of tourist attraction
research. Annals of Tourism Research 14: 553–
575.
Int. J. Tourism Res. 12, 307–320 (2010)
DOI: 10.1002/jtr


Non-economic Determinants in Tourism Demand
Li G, Song H, Witt SF. 2005. Recent developments
in econometric modeling and forecasting. Journal
of Travel Research 44: 82–99.
Lim C. 1997. An econometric classification and

review of international tourism demand models.
Tourism Economics 3(1): 69–81.
Lim C. 1999. A meta-analytic review of international
tourism demand. Journal of Travel Research 37:
273–284.
Loeb PD. 1982. International travel to the United
States: an econometric evaluation. Annals of
Tourism Research 9: 7–20.
Morley C. 1996. A comparison of three methods for
estimating tourism demand models. Tourism
Economics 2(3): 223–234.
Morley C. 1998. A dynamic international demand
model. Annals of Tourism Research 25(1): 70–84.
Morley C. 2000. Demand modeling methodologies
integration and other issues. Tourism Economics
6(1): 5–19.
Naude WA, Saayman A. 2005. Determinants of
tourist arrivals in Africa: a panel data regression
analysis. Tourism Economics 11(3): 365–391.
Poole M. 1988. Forecasting methodology. Occasional
Paper No. 3, Bureau of Tourism Research:
Australia.
Ritchie JRB, Zins M. 1978. Culture as determinant
of the attractiveness of a tourism region. Annals
of Tourism Research 5(2): 252–267.
Rugg D. 1973. The choice of journey destinations: a
theoretical and empirical analysis. Review of Economic Statistics 20: 64–71.
Russo AP, Borg J. 2002. Planning considerations for
cultural tourism: a case study of four European
cities. Tourism Management 23: 631–637.

Sauran A. 1978. Economic determinants of tourism
demand: a survey, the tourist. Review 3(1): 2–4.
Schmidt CJ. 1979. The guided tour: insulated adventure. Urban Life 7(4): 441–467.
Sinnott RW. 1984. Virtues of the haversine. Sky and
Telescope 68(2): 159.
Smeral E. 1988. Tourism demand, economic theory
and econometrics: an integrated approach. Journal
of Travel Research 26(4): 38–42.
Snepenger D, Hauser B, Snepenger M. 1990. Seasonality of demand. Annals of Tourism Research 17:
429–441.

Copyright © 2009 John Wiley & Sons, Ltd.

319
Song H, Li G. 2008. Tourism demand modeling and
forecasting — a review of recent research. Tourism
Management 29(2): 203–220.
Song H, Witt S. 2000. Tourism Demand Modeling
and Forecasting: Modern Econometric Approaches.
Pergamon: Amsterdam.
Song H, Wong KF, Chon KS. 2003. Modeling and
forecasting the demand for Hong Kong tourism.
International Journal of Hospitality Management
22(4): 435–451.
Stronge WB, Redman M. 1982. Tourism in Mexico:
an empirical analysis. Annals of Tourism Research
9: 21–35.
Summary RM. 1983. The role of tourism in the
economic growth and development of Kenya and
Barbados: a comparative study. PhD dissertation,

University of Illinois: Urbana-Champaign.
Thurot J, Thurot G. 1983. The ideology of class and
tourism confronting the discourse of advertising.
Annals of Tourism Research 10: 173–189.
Tremblay P. 1989. Pooling international tourism in
Western Europe. Annals of Tourism Research 16:
477–491.
Turner L, Reisinger Y, Witt S. 1998. Tourism demand
analysis using structural equation modeling.
Tourism Economics 4(4): 301–323.
Turner LW, Witt SF. 2001. Forecasting tourism
using univariate and multivariate structural time
series models. Tourism Economics 7(2): 135–
147.
Uysal M, Crompton JL. 1984. The demand for international tourist flows to Turkey. Tourism Management 5: 288–297.
Vanegas MS, Croes RR. 2000. Evaluation of demand:
US tourists to Aruba. Annals of Tourism Research
27(4): 946–963.
Var T, Cesario F, Mauser G. 1985. Convention
tourism modeling. Tourism Management 6(2):
194–204.
Witt SF, Martin CA. 1987. Econometric models for
forecasting international tourism demand. Journal
of Travel Research 25(3): 23–30.
Wong KF, Song H, Witt SF, Wu DC. 2007. Tourism
forecasting: to combine or not to combine?
Tourism Management 28(4): 1068–1078.
World Tourism Organization. 1992. World Tourism
Trends to the Year 2000. WTO: Madrid.

Int. J. Tourism Res. 12, 307–320 (2010)
DOI: 10.1002/jtr


320

V. Cho

Appendix A
The collected data are of the following 135 countries.
Continent

Countries

Americas

Bahamas, Barbados, Belize, Bolivia, Brazil, Canada, Chile, Colombia, Costa Rica, Dominican
Republic, Ecuador, El Salvador, Guatemala, Honduras, Jamaica, Nicaragua, Panama,
Paraguay, Peru, Trinidad and Tobago, United States of America, Uruguay, Venezuela.
Armenia, Azerbaijan, Bahrain, Bangladesh, Bhutan, Brunei Darussalam, Cambodia, China,
Cyprus, Georgia, Hong Kong (China), India, Indonesia, Iran, Israel, Japan, Jordan,
Kazakhstan, Korea, Kuwait, Kyrgyz Republic, Lao People’s Democratic Republic, Macao,
Malaysia, Maldives, Mongolia, Morocco, Nepal, Oman, Pakistan, Philippines, Saudi Arabia,
Singapore, Sri Lanka, Syria Arab Republic, Thailand, Tunisia, Vietnam.
Albania, Austria, Belgium, Bulgaria, Croatia, Czech Republic, Denmark, Estonia, Finland,
France, Germany, Greece, Hungary, Iceland, Ireland, Italy, Latvia, Lithuania, Malta,
Netherlands, Poland, Portugal, Republic of Moldova, Romania, Russian Federation, Slovak
Republic, Slovenia, Spain, Sweden, Switzerland, The Former Yugoslav Rep. of Macedonia,
Ukraine, UK.
Australia, Fiji, New Zealand, Papua New Guinea, Samoa, Tonga, Vanuatu.

Algeria, Benin, Botswana, Burkina Faso, Burundi, Cameroon, Central African Republic, Chad,
Egypt, Ethiopia, Gabon, Ghana, Guinea-Bissau, Kenya, Lesotho, Great Socialist People’s
Libyan Arab Jamahiriya, Madagascar, Malawi, Mali, Mauritius, Mozambique, Niger,
Nigeria, Rwanda, Senegal, Seychelles, Sierra Leone, South Africa, Sudan, Swaziland, Togo,
Uganda, United Republic of Tanzania, Zambia.

Asia

Europe

Oceania
Africa

Copyright © 2009 John Wiley & Sons, Ltd.

Int. J. Tourism Res. 12, 307–320 (2010)
DOI: 10.1002/jtr


INTERNATIONAL JOURNAL OF TOURISM RESEARCH
Int. J. Tourism Res. 12, 321–333 (2010)
Published online 28 October 2009 in Wiley InterScience
(www.interscience.wiley.com) DOI: 10.1002/jtr.753

Tourists’ Information Search:
the Differential Impact of Risk
and Uncertainty Avoidance
Vanessa Ann Quintal1,*, Julie Anne Lee2 and Geoffrey N. Soutar2
School of Marketing, Curtin University of Technology, Bentley, Western Australia, Australia
2

Business School, University of Western Australia, Crawley, Western Australia, Australia

1

ABSTRACT
While studies in and out of tourism contexts
have explored risk and/or uncertainty
avoidance’s impact on information search,
few have clarified whether the two
constructs impact differentially on
information search. To examine this issue,
data were collected from large online panels
in Australia, China and Japan. The risk and
uncertainty avoidance scales were reliable,
had convergent and discriminant validity
and were invariant across the three country
samples. As expected, uncertainty avoidance
was positively related to the extent of
information search in all three country
samples, whereas risk avoidance was not.
This suggests that the constructs are distinct
and may impact at different stages of
decision-making. Copyright © 2009 John
Wiley & Sons, Ltd.
Received 22 April 2008; Revised 4 September 2009; Accepted
25 September 2009

Keywords: consumer decision-making; risk
and uncertainty avoidance; information
search.

INTRODUCTION

C

onsumer decision-making is uncertain
and risky as it involves choices that may
or may not deliver expected benefits.

*Correspondence to: Dr. V. A. Quintal, School of Marketing, Curtin University of Technology, Kent Street, Bentley
6102, Western Australia, Australia.
E-mail:

The tourism industry is very vulnerable to
direct and indirect events, over which they
often have little control (commonly known as
crises), that threaten travellers’ assurance and
safety. Since 2000, a number of events have
negatively impacted on tourism around the
globe (Law, 2006; Kozak et al., 2007) and have
influenced the way potential tourists respond
to financial, performance, psychological, social,
physical and time risk, as well as to the uncertainty associated with travel (Simpson and
Siguaw, 2008).
Consumers averse to risk and uncertainty
are likely to engage in risk and uncertaintyreducing strategies, such as looking for quality
assurances (Sweeney et al., 1999) and searching
extensively for information (Vogt and Fesenmaier, 1998). This is likely to be more pronounced in tourism because of its ‘intangible
and experiential nature’ (Sirakaya and Woodside, 2005, p. 816), which leads people to ‘search
for information and move back and forth
between search and decision-making stages’

(Jun et al., 2007, p. 267). However, it is difficult
to untangle the differential influence of risk
and uncertainty avoidance, as researchers have
often used the terms interchangeably. The
present research attempted to shed some light
on this issue by examining differences between
risk and uncertainty avoidance and seeing
how the two constructs impact on tourists’
information search. First, we clarify some conceptual and operational differences between
risk and uncertainty to help integrate the fragmented literature in these two areas. Then, we
examine the potentially distinct influences risk
and uncertainty avoidance have on the extent
of information search undertaken by tourists
using data collected in three countries.
Copyright © 2009 John Wiley & Sons, Ltd.


322
A LITERATURE REVIEW
Risk avoidance versus
uncertainty avoidance
Risk and uncertainty can be distinguished by
the probabilities of their outcomes. Risk exists
when the probabilities of outcomes are known,
while uncertainty exists when the probabilities
of outcomes are not known (Knight, 1948;
Savage, 1954). This distinction underlies many
conceptual risk and uncertainty avoidance
definitions. For instance, Weber and Bottom
(1989, p. 128) defined risk avoidance (RA) as

‘whether, ceteris paribus, a decision maker has
a tendency to be attracted or repelled by alternatives that he or she perceives as more risky
over alternatives perceived as less risky’, while
Hofstede (1991, p. 113) and other researchers
(e.g. Reisinger and Turner, 2003) defined uncertainty avoidance (UA) as ‘the extent of feeling
threatened by uncertain or unknown situations’. People in high UA cultures have lower
tolerance for ambiguity (Hofstede, 2001) and
are more likely to prefer structures that make
events more easily interpretable and predictable (Reisinger and Turner, 1999). In contrast,
people in low UA cultures are relatively more
comfortable with ambiguity and are more
likely to seek novelty and convenience (Lee
et al., 2007).
In differentiating between risk and uncertainty avoidance, Hofstede (2001, p. 148)
argued UA is not the same as RA, even though
many people ‘interpreted “uncertainty avoidance” as “risk avoidance” — for example, in
business decisions’. He observed that UA is
related to structure and escape from ambiguity
and not necessarily to RA. Indeed, people may
engage in risky behaviour in order to reduce
ambiguity, ‘such as starting a fight with a
potential opponent rather than sitting back
and waiting’ (Hofstede, 2001, p. 148).
In spite of the distinction made between
risk and uncertainty avoidance, much of the
research has used these constructs interchangeably (Hofstede, 2001). This has led to some
confusion, with cases of RA being attributed to
UA and vice versa. For instance, Steenkamp
et al. (1999, p. 59) described consumers from
high uncertainty-oriented cultures as ‘resistant

to change from established patterns and . . .
Copyright © 2009 John Wiley & Sons, Ltd.

V. A. Quintal, J. A. Lee and G. N. Soutar
focused on risk avoidance and reduction’. They
examined UA using Hoppe’s (1990) updated
country ratings that extensively validated
Hofstede’s (1980) results and found innovativeness was lower in UA cultures. Bao et al.
(2003) used Hofstede and Bond’s (1984, p. 419)
UA definition to define risk aversion as ‘the
extent to which people feel threatened by
ambiguous situations, and have created beliefs
and institutions that try to avoid these’. Despite
their conceptual definition of RA being UA,
they measured risk aversion using a scale
adapted from Raju (1980) and concluded that
risk aversion appeared to contribute to different decision-making styles in the USA and
China. Similarly, Money and Crotts (2003, p.
191) suggested that UA was ‘a measure of
intolerance for risk’.
The conceptual distinction between risk and
uncertainty is especially important, as the constructs are likely to be correlated. This implies
that at least some people who avoid risk may
also avoid uncertainty and vice versa. Thus,
when researchers only measure one of the constructs, the influence of the other may be incorrectly attributed to the measured construct,
especially in cases where the theoretical reasoning confuses the two constructs, such as in
previous examples. This is especially problematic when researchers use culture level constructs, such as Hofstede’s (1980) country level
uncertainty avoidance index (UAI) to predict
individual level behaviour, as people within a
country differ widely in their tolerance for risk

and uncertainty. In the current paper, we
clarify the issue by examining the influence
risk and uncertainty avoidance have on the
extent of information search, which is theoretically related to only one of the avoidance constructs, at an individual level. To address this
objective, we used tourists from three countries that differ on Hofstede’s country level
UAI scores.
Information search
Information search is an important aspect of
tourism decision-making; as such decisions are likely to be a high cost and highinvolvement purchase (Bonn et al., 1998) and
the search process is often seen as an enjoyable
part of the travel experience. Information
Int. J. Tourism Res. 12, 321–333 (2010)
DOI: 10.1002/jtr


Tourists’ Information Search
search in a travel context has been studied in
relation to the amount of searching (Fodness
and Murray, 1997; Bai et al., 2004; Öörni, 2004;
Lehto et al., 2006); the sources of information
used (Fodness and Murray, 1997; Chen and
Gursoy, 2000; Cai et al., 2004; Kerstetter and
Cho, 2004; Mourali et al., 2005; Lehto et al.,
2006), including online searching (Bai et al.,
2004; Jang, 2004; Luo et al., 2004; Öörni, 2004;
Beldona, 2005; Lehto et al., 2006); the search
process itself (Fodness and Murray, 1997;
Bieger and Laesser, 2004; Gursoy and McCleary,
2004; Kim et al., 2006; Pan and Fesenmaier,
2006); situational factors (Fodness and Murray,

1997; Gursoy and Chen, 2000; Bieger and
Laesser, 2004; Luo et al., 2004); consumer
involvement (Cai et al., 2004; Gursoy and
McCleary, 2004; Lehto et al., 2006); demographic differences (Fodness and Murray,
1997; Luo et al., 2004; Kim et al., 2006); and
cultural differences (Webster, 1992; Reisinger
and Turner, 1998; Chen and Gursoy, 2000;
Gursoy and Chen, 2000; Money and Crotts,
2003; Litvin et al., 2004). Most of these studies
have focused on destination choice (Chen and
Gursoy, 2000; Money and Crotts, 2003; Cai et
al., 2004; Kerstetter and Cho, 2004; Luo et al.,
2004; Lehto et al., 2006) and the most commonly
used sources seem to include the Internet, brochures and pamphlets, family and friends and
travel agents (Chen and Gursoy, 2000; Gursoy
and Chen, 2000; Kerstetter and Cho, 2004).
Information search and risk and
uncertainty avoidance
While considerable research has examined the
influence UA has on aspects of information
search, it has usually examined the sources of
information used, rather than the extent of
information search, across countries that differ
on Hofstede’s (1980) UAI scores. For instance,
Money and Crotts (2003) and Litvin et al. (2004)
compared people from a high UA culture
(Japan) with people from lower UA cultures
(Germany and the USA). They found that
people from the higher UA culture were more
likely to select travel information sources

related to the channel (e.g. travel agent) over
personal or mass media sources, compared
with people from lower UA cultures. In the
current paper, we argue that an individual’s
Copyright © 2009 John Wiley & Sons, Ltd.

323
level of UA, rather than RA, influences the
extent of their information search.
People are likely to react differently to situations with inherent ambiguity, depending on
their tolerance for uncertainty. The early stages
of decision-making are more uncertain, prompting people to look for information on available
products or services that might fulfil recognised
needs (Comegys et al., 2006). During the information search stage, people are unlikely to consider precise probabilities of potential outcomes,
which would infer the existence of risk. The
calculation of probabilities is more likely to
come at a later evaluation stage when only a few
alternatives are compared. At the evaluation
stage, the probability of a loss from one alternative can be weighed against the probability of a
loss from other alternatives. Consequently, we
suggest people’s level of UA is more likely to be
related to the extent of their information search
than RA. It is expected people with higher UA,
will search more sources of information than
will people with lower UA, holding RA constant. Conversely, it can be inferred that RA,
will not influence the search for information at
this early stage, holding UA constant, which
suggests:
Hypothesis 1: UA will be positively associated with extent of information search,
holding RA constant.

Hypothesis 2: RA will not be associated
with extent of information search, holding
UA constant.
METHODOLOGY
The present study assessed the differential
influences RA and UA had on the extent of
potential tourists’ information search. Samples
were obtained in three countries that vary on
Hofstede’s (2001) national UAI. This country
index is high in Japan (UAI score = 92), medium
in Australia (51) and low in China (30). Samples
in each country were obtained from commercial online panels and chosen to reflect the
populations’ gender and age characteristics.
However, the Japanese sample was also
screened to only include people who had
travelled internationally in the last five years
or who intended to travel internationally in the
Int. J. Tourism Res. 12, 321–333 (2010)
DOI: 10.1002/jtr


324
next five years. This international travellers’
sample was added to see whether RA and UA
functioned in a similar manner for more experienced travellers. It was expected that international travellers would be less risk and
uncertainty avoidant than the general population, as such, Japan (a high UAI culture) was
chosen. The questionnaires in Australia were
administered in English, while those in China
and Japan were translated into Mandarin
and Japanese respectively. The procedures

used followed the translation-back-translation
method recommended by Brislin (1976).
The survey was electronically administered
to adult members of large online consumer
panels in Australia (State of Victoria), three
major Chinese cities (Beijing, Shanghai and
Guangzhou) and three major Japanese cities
(Tokyo, Osaka and Nagoya). The sample sizes
were 200 in Australia, 443 in China and 342 in
Japan, which reflected response rates of 93%,
77% and 57% respectively. The samples approximated their respective populations on gender
(55% male in Australia, 49% male in China and
56% male in Japan). The median age was 43
years in Australia, 33 years in China and 42
years in Japan. Both age and gender were
reasonably representative in each country (e.g.
Australian median age = 37, China = 33 and
Japan = 43), according to the CIA World Factbook
(CIA, 2006). Annual household incomes in
excess of A$60 000 were reported by 50% of
Australian respondents, 61% of Japanese
respondents but by less than half in China.
RA
Donthu and Gilliland’s (1996) three-item scale
was used to measure RA as the scale had
reasonable measurement properties in past
research (e.g. α = 0.78) and is relevant to purchase behaviour. The RA items were measured
on a seven-point Likert-type scale ranging
from strongly disagree (1) to strongly agree (7).
The items were averaged to produce a

composite measure of RA for each country
sample.
UA
Yoo and Donthu’s (2002) five-item scale was
used to measure UA as this scale had reasonCopyright © 2009 John Wiley & Sons, Ltd.

V. A. Quintal, J. A. Lee and G. N. Soutar
able measurement properties in past research
(e.g. α = 0.88) and is based on Hofstede’s (1980)
UA items. The UA items were also measured
on a seven-point Likert-type scale ranging
from strongly disagree (1) to strongly agree (7).
Again, the items were averaged to produce a
composite measure of UA for each country
sample.
Extent of information search
Respondents were asked whether or not they
used 14 sources of information when making
four types of tourism decisions (tourism
destinations, pricing, accommodation and
flight decisions). The information sources
included personal sources (relatives, friends,
travel agents, tour operators), traditional travel
sources (travel guidebooks, newspapers, TV
travel programmes, travel magazines, airline
telephones, tourism office brochures) and
web sources (Internet, airline websites, online
newsletters, travel agent websites) and were
adapted from prior tourism information
studies (e.g. Fodness and Murray, 1997; Chen

and Gursoy, 2000; Gursoy and Chen, 2000).
FINDINGS
The reliability and validity of the risk and
uncertainty avoidance scales as well as their
measurement invariance across the three
country samples are discussed first. Next, the
results of the Rasch analysis that was used to
obtain a unidimensional measure of the extent
of search across the 14 sources of information
are reported. Finally, the influence risk and
uncertainty avoidance had on the extent of
information search is outlined.
The measurement properties of the risk and
uncertainty avoidance scales
The AMOS 16 software package (Arbuckle,
2007) was used to estimated the RA and UA
scales; first as a pooled sample and then separately in the three country samples. The goodness of fit indices for the RA and UA scales are
shown in Table 1, while the parameter estimates of the RA and UA items, as well as the
means and standard deviations of the summed
scores, are shown in Table 2.
Int. J. Tourism Res. 12, 321–333 (2010)
DOI: 10.1002/jtr


Tourists’ Information Search

325

Table 1. The RA and UA scale goodness of fit indices
Risk avoidance

Goodness of fit indices
χ2
d.f.
p value
RMSEA
NNFI
GFI
CFI
Uncertainty avoidance
Goodness of fit indices
χ2
d.f.
p value
RMSEA
NNFI
GFI
CFI

Australia
n = 200

China
n = 443

Japan
n = 342

Pooled sample
n = 985

1.80
1
>0.05
0.06
0.99
1.00
0.99

1.38
1
>0.05
0.03
1.00
1.00
1.00

0.26
1
>0.05
0.01
1.00
1.00
1.00

0.03
1
>0.05
0.01
1.00
1.00

1.00

Australia
n = 200

China
n = 443

Japan
n = 342

Pooled sample
n = 985

3.52
2
>0.05
0.06
0.99
1.00
0.99

3.97
2
>0.05
0.05
1.00
1.00
1.00

0.26
2
>0.05
0.01
1.00
1.00
1.00

1.30
2
>0.05
0.01
0.99
1.00
0.99

CFI, comparative fit index; d.f., degrees of freedom; GFI, goodness of fit index; NNFI, non-normed fit index; RA, risk
avoidance; RMSEA, root mean square error of approximation; χ2, chi-square; UA, uncertainty avoidance.

As there were only three items in the RA
scale, model fit could not be estimated because
there were no degrees of freedom. However,
an examination of the results suggested two of
the error variances were very similar and could
be constrained to be equal to provide the
degrees of freedom needed to examine the RA
construct’s measurement properties. As can be
seen in Table 1, the chi-square statistic and the
other indices suggested that the modified
three-item RA model had a good fit to the

pooled sample and to the data in each of the
three country samples.
The chi-square statistic in the pooled sample
was significant for the five-item UA model
(χ2 = 31.06; d.f. = 5; p ≤ 0.001). However, an
examination of the modification indices suggested that the poor fit was because of a correlation between two of the error terms.
Consequently, the item with the lower standardised estimate (instructions for operations are
important) was excluded and the model was
re-estimated. As can be seen in Table 1, the
four items fitted the data well. Further, the correlation between the five-item scale and the
Copyright © 2009 John Wiley & Sons, Ltd.

revised four-item scale was 0.99, which suggests nothing would be lost by using the fouritem scale (Thomas et al., 2001).
As can be seen in Table 2, the scales had
acceptable reliabilities in each case, with the
RA scale’s reliabilities ranging from 0.72 to
0.79 and the UA scale’s reliabilities from 0.87
to 0.88. Further, the magnitude, direction and
statistical significance of the estimated parameters were consistent, which suggests that convergent validity can be assumed (Steenkamp
and van Trijp, 1991). Convergent validity was
also examined by computing average variance
extracted (AVE) scores for each construct,
which should be equal to or greater than 0.50
(Fornell and Larcker, 1981). The RA and UA
scales for the pooled sample had AVE scores
of 0.56 and 0.62 respectively, while the AVE
scores for the two scales in the three country
samples ranged from 0.53 to 0.64, suggesting
that the scales had convergent validity in each
country.

Discriminant validity was investigated in
two ways. First, the correlations between the
RA items and the UA items were computed.
Int. J. Tourism Res. 12, 321–333 (2010)
DOI: 10.1002/jtr


326

V. A. Quintal, J. A. Lee and G. N. Soutar

Table 2. The RA and UA scale item parameter estimates, means and standard deviations
Risk avoidance items

Australia
n = 200

China
n = 443

Japan
n = 342

Pooled sample
n = 985

I would rather be safe than sorry.

0.83
5.23

(1.53)

0.79
5.39
(1.55)

0.73
4.52
(1.38)

0.78
5.05
(1.54)

I want to be sure before I purchase anything.

0.60
5.24
(1.56)

0.67
5.49
(1.52)

0.55
4.72
(1.34)

0.65
5.17

(1.51)

I avoid risky things.

0.75
4.55
(1.64)

0.77
5.39
(1.49)

0.76
4.39
(1.42)

0.78
4.88
(1.57)

0.77

0.79

0.72

0.78

Australia
n = 200

China
n = 443

Japan
n = 342

Pooled sample
n = 985

It is important to have instructions spelled out in detail
so I always know what I am expected to do.

0.77
4.60
(1.60)

0.75
5.32
(1.39)

0.79
4.99
(1.21)

0.76
5.06
(1.40)

It is important to closely follow instructions and

procedures.

0.81
4.92
(1.50)

0.81
5.02
(1.40)

0.84
4.80
(1.12)

0.82
4.92
(1.33)

Rules and regulations are important because they tell me
what is expected of me.

0.81
4.79
(1.48)

0.83
5.09
(1.43)

0.73

4.68
(1.13)

0.80
4.89
(1.36)

Standardised work procedures are helpful.

0.76
5.03
(1.47)

0.81
5.26
(1.37)

0.78
4.91
(1.19)

0.79
5.09
(1.34)

0.87

0.88

0.87

0.87

Reliability
Uncertainty avoidance items

Reliability

Note: Parameter estimates are shown in bold, mean scores are shown in the second row in each case and standard deviations are shown in parentheses.

All of the correlations were small (less than
0.50) and well below 0.80, which has been
suggested as the level at which discriminant
validity issues become problematic. Second,
Fornell and Larcker’s (1981) discriminant
validity test was undertaken. As was noted
earlier, the AVE scores for the RA and UA
constructs for the pooled data were 0.56
and 0.63 respectively. Because both values
exceeded the square of the correlation between
the constructs (0.46) in the pooled data,
discriminant validity was supported. The
highest squared correlation in the three
Copyright © 2009 John Wiley & Sons, Ltd.

country samples was 0.49, while the lowest
AVE score was 0.53, suggesting that discriminant validity could also be assumed in each
country sample.
The measurement invariance of the risk and
uncertainty avoidance scales

To see whether the RA and UA scales were
equivalent across the three country samples,
configural invariance and metric invariance
were examined. Steenkamp and Baumgartner
(1998) suggested that configural and at least
partial metric invariance are necessary for the
Int. J. Tourism Res. 12, 321–333 (2010)
DOI: 10.1002/jtr


Tourists’ Information Search

327

Table 3. Assessment of measurement invariance and latent mean differences across the pooled sample from
Australia, China and Japan
Model specification
RA

χ2

d.f.

M1: Configural invariance* 0.00
M2: Full metric invariance 10.79
Model specification
UA
M1: Configural invariance
M2: Full metric invariance

0
6

Models
compared

Δχ2

—
—
M2 vs M1 10.79

χ2

d.f.

Models
compared

7.75
18.24

6
12

—
—
M2 vs M1 10.49

Δχ2

p value

RMSEA

NNFI

—
0.10

—
0.03

—
0.99

p value

RMSEA

NNFI

0.01
0.11

0.02
0.02

1.00
0.99

GFI

CFI

—
—
0.99 0.99
GFI

CFI

1.00 1.00
0.99 1.00

χ2/d.f.
—
1.80
χ2/d.f.
1.29
1.52

* There are no degrees of freedom in this case as there are only 3 items in the scale and the error variances have been
freed to allow the cross-country comparison to be made
CFI, comparative fit index; d.f., degrees of freedom; GFI, goodness of fit index; NNFI, non-normed fit index; RA, risk
avoidance; RMSEA, root mean square error of approximation; χ2, chi-square; UA, uncertainty avoidance.

analysis that was used in the present study.
These forms of measurement invariance are
nested models as each test is nested in the preceding model. As one of the goals of the study

was to test the basic meaning and structure of
the RA and UA measures across countries,
partial metric invariance was seen as the
minimal requirement.
Although the results for each country sample
reported in Table 2 provided indications of
configural invariance (the first level of measurement invariance), the pooled sample with
mean structure is the most efficient and appropriate means of testing measurement invariance (Wang and Waller, 2006). Consequently,
the measurement invariance of the RA and UA
scales was examined using the pooled data
from Australia, China and Japan and the results
obtained are shown in Table 3. As can be seen
in the Table, the goodness of fit indices for the
multiple group RA model were acceptable.
These results, coupled with the fact that the
hypothesised loadings were all significant in
the pooled sample, suggested that the RA scale
had configural invariance. The metric invariance of the RA scale (M2), in which the loadings were set to be invariant across the three
country samples, was examined. As can be
seen in Table 3, the increase in the chi-square
statistic between M1 and M2 (Δχ2(6) = 10.79)
was not significant at the 10% level, suggesting
that the RA scale had full metric invariance.
As can also be seen in Table 3, the goodness
of fit indices for the multiple group UA model
Copyright © 2009 John Wiley & Sons, Ltd.

were acceptable. Again, these results, coupled
with the fact that the hypothesised factor loadings were all significant in the pooled sample,
suggested that the UA scale had configural

invariance. The metric invariance of the UA
scale (M2) was also tested. As can be seen in
Table 3, the increase in the chi square statistic
between M1 and M2 (Δχ2(6) = 10.49) was not
significant at the 10 % level, suggesting that the
UA scale also had full metric invariance.
The acceptance of configural invariance suggested that the basic meaning and structure
was similar for both scales across the three
country samples (Wang and Waller, 2006), and
the full metric invariance of the RA and UA
scales suggested the scales had the degree
of measurement invariance necessary for
meaningful comparisons of cross-country
differences (Steenkamp and Baumgartner,
1998). Consequently, this issue was examined
in the subsequent analysis.
The extent of information search
A descriptive analysis of the sources of
information was conducted, prior to
undertaking a Rasch analysis (Andrich, 1988)
to develop a unidimensional extent of
information search scale. As was noted earlier,
respondents were asked whether they consulted
14 different sources of information for four
types of tourism decisions (tourism destinations,
pricing, accommodation and flight decisions).
The sources of information reflected four
Int. J. Tourism Res. 12, 321–333 (2010)
DOI: 10.1002/jtr

328
personal sources, six traditional sources and
four web-based sources, and the frequency
of their use in each country can be seen in
Table 4.
Chi-square tests demonstrated some
similarities in the sources used by the majority
of respondents in each country sample. For
destination decisions, all three source categories
were used by a majority of respondents in each
country. For the other decisions (pricing,
accommodation and flights), web-based
sources were more common and traditional
sources less common. However, for Chinese
respondents, personal sources were commonly
used for all decisions except flights.
Despite these similarities in patterns of
information sources used, most of the
frequencies for the sources differed significantly
across the three countries, as can be seen in
Table 4 (based on a calculation of percentage
difference Z scores). Very few of the sources
were consulted in similar proportions across
the three country samples, as only two of the
traditional sources for destination decisions
(newspapers and tourism office brochures)
and one traditional (travel guidebooks) and
one web-base source for accommodation
decisions were the same across the three

country samples.
Some of these differences support prior
research. For instance, Australian and Chinese
respondents used personal sources, such as
relatives and friends more often than the more
experienced Japanese respondents. This supported Moutinho’s (1987) and Urbany et al.’s
(1989) suggestion that experienced travellers
depend less on personal information sources.
Further, Chinese and Japanese respondents
were more likely than Australians to consult
certain printed information sources, such as
travel guidebooks and travel magazines, which
was consistent with Chen’s (2000), Money and
Crotts’ (2003) and Nishimura et al.’s (2006)
results. Finally, web sources were used intensively in each country, particularly for destination, pricing and flight information. However,
the Internet was consulted significantly more
often in Australia than in China or Japan.
As the purpose of this paper was to examine
the extent of search, rather than the types of
sources consulted, the 56 items (14 sources × 4
decisions) were analysed using the Rasch
Copyright © 2009 John Wiley & Sons, Ltd.

V. A. Quintal, J. A. Lee and G. N. Soutar
model to see whether a unidimensional order
existed (Andrich, 1988). A few of the information
search items (e.g. consulting tourism office
brochures for destination decisions) did not fit the
Rasch model as their associated chi-square
statistics were significant well beyond the p ≤

0.001 level (Soutar and Cornish-Ward, 1997).
The information search items with a poor fit to
the Rasch model were removed iteratively
to improve overall fit. After removing five
information search items (tourism office brochures for destination, pricing, and flight decisions,
relatives living in the region and airline websites
for accommodation decisions), an acceptable fit
was obtained (χ2 = 124.6; p ≤ 0.10). This
suggested that the remaining items fitted the
Rasch model and that the 51 sources of information could be combined to form an extent of
information search construct.
Risk and uncertainty avoidance’s effects on
information search
The two hypotheses were examined using the
AMOS 16 program’s multiple-group structural
equation modelling procedure. UA and RA
were modelled as correlated exogenous variables and regressed on the extent of search
score developed through the Rasch analysis.
As can be seen in Table 5, the structural models
were equivalent across the three country
samples (χ2 = 2.07, p ≤ 0.71; RMSEA = 0.01;
AGFI = 0.99). That is, the impact the two constructs had on information search was similar
in the three countries. As expected, UA had a
significant positive effect on the extent of information search in Australia (b = 0.27, p ≤ 0.01),
China (b = 0.18, p ≤ 0.01) and Japan (b = 0.17,
p ≤ 0.01), while RA was unrelated to the extent
of information search in each sample (Australia = −0.06, not significant (ns); China = 0.09, ns;
Japan = 0.01, ns), supporting Hypothesis 1 and
Hypothesis 2. In addition, the correlations
between UA and RA in the three country

samples were all significant at the 0.001 level
(Australia = 0.58; China = 0.72; Japan = 0.55).
If we had examined RA and UA individually, we may have attributed the effect of UA
to RA, at least in China. The correlation
between RA and the extent of information
search was significant for China (r = 0.22, p ≤
0.001), but not for Australia (r = 0.10, ns) or
Int. J. Tourism Res. 12, 321–333 (2010)
DOI: 10.1002/jtr


Copyright © 2009 John Wiley & Sons, Ltd.

59
60a
60a
32
13ab
69a
80
34a
43ab
48a

Traditional sources
Travel guidebooks
Newspapers
TV travel programs
Travel magazines
Airline telephones

Tourism office brochures

Web sources
Internet
Airline websites
Online newsletters
Travel agent websites
71a
24b
42a
43a

68a
59a
59a
69
19a
63a

66
60a
55b
43

China

68a
27ab
35b
60

73a
53a
51
61
11b
62a

53
38
49ab
32

Japan

79
60
42a
62ab

30a
37a
21
20
29ab
45a

32
29
56a

29

Australia

65a
37a
39a
54a

45
44a
38
50
30a
46a

62
52
70
50

China

Pricing

62a
40a
20
62b

32a
23
13
32
22b
58

18
12
49a
9

Japan

80
30a
43a
60a

44a
38a
26a
24
11a
57

46
48a
52a
25a

Australia

59a
27a
36a
49b

43a
33a
32a
48
25
42a

55
50a
62
40

China

63a
24a
17
54ab

41a
25
18

40
17a
44a

22
21
47a
19a

Japan

Accommodation

71
79
36
59

20a
22a
13a
16a
40a
26a

22
17
53a
18

Australia

40a
58a
28
37a

28
23a
21
29
64
25a

32
25
47a
26

China

Flights

38a
61a
12
42a

15a
10

8a
12a
38a
17

7
8
39
11

Japan

Note: Multiple sources could be consulted for the same decision. Frequencies that share the same subscript across the three countries are not significantly different
from one another at the .05 level.

78
65a
44a
22

Personal sources
Relatives/friends visiting
Relatives/friends living
Travel agents
Tour operators

Australia

Destination

Table 4. Sources of information consulted: The frequency of information source use for travel decisions on destinations, price, accommodation and
flights (%)

Tourists’ Information Search
329

Int. J. Tourism Res. 12, 321–333 (2010)
DOI: 10.1002/jtr


330

V. A. Quintal, J. A. Lee and G. N. Soutar
Table 5. Standardized path coefficients and correlations
Paths
Hypothesis 1: UA → Info search
Hypothesis 2: RA → Info search

Australia
0.27**
−0.06

Model fit statistics
CMIN
p value
RMSEA
AGFI
Correlations
RA
n

China
0.18**
0.09

Japan
0.17**
0.01

2.07
0.71
0.01
0.99
Australia
UA
0.58**
200

China
UA
0.72**
443

Japan
UA
0.55**
342

* p < 0.05; ** *p < 0.01; *** p < 0.001.
AGFI, adjusted goodness of fit index; CMIN, chi-square; RA, risk avoidance;

RMSEA, root mean square error of approximation; UA, uncertainty avoidance.

Japan (r = 0.09, ns). Thus, this result is better
explained by the higher correlation between
UA and RA for China, rather than the distinct
influence of RA.
CONCLUSIONS
The present study clarified the differences
between RA and UA by assessing the crosscountry applicability of existing scales and
their differential effects in the information
search phase of travellers’ decision-making
processes. The results suggested that the RA
and UA scales were unidimensional and reliable, and had convergent, discriminant and
predictive validity. Both scales had full metric
equivalence across the three countries in which
data were collected, allowing an examination
of cross-cultural differences. Further, the
results illustrated a remarkable amount of similarity in the effect of UA and RA on the extent
of information search, despite differences in
the actual sources of information consulted
across countries. As expected, the UA construct
was positively related to the extent of information search in all three country samples,
whereas the RA construct was not. This provides some evidence that the RA and UA constructs are distinct and are likely to operate in
different stages in travellers’ decision-making
processes.
Copyright © 2009 John Wiley & Sons, Ltd.

UA appears to operate in the early stages of
decision-making when general information is
sought (e.g. Dawar et al., 1996; Money and

Crotts, 2003; Litvin et al., 2004) and the probabilities of outcomes are unclear, while RA does
not. Because acquiring more information over
an extended period of time can help uncertainty-avoidant consumers to reduce general
anxieties associated with travel, destination
management organisations may need to formulate diverse and sustained messages about
their products and services to potential consumers (Drollinger et al., 2006). Future research
is needed to assess the expected influence RA
has in the later evaluation stage of decisionmaking, when specific alternatives with probable outcomes are being compared.
Some general and more specific recommendations can be made by examining the frequency of information used in Table 4. For
instance, it appears that a very wide variety of
sources are consulted for destination decisions
in all three country samples, suggesting that
destination marketers need to consider
presenting consistent information about their
destination in a wide variety of information
sources. Information about other travel decisions, such as pricing, accommodation and
flights, are more likely to be sought from personal sources, such as travel agents and webbased sources. The importance of personal
Int. J. Tourism Res. 12, 321–333 (2010)
DOI: 10.1002/jtr