Description
The purpose of this paper is to examine competition between tourism destination brands in
terms of how they share travelers with each other
International Journal of Culture, Tourism and Hospitality Research
Generalized pattern in competition among tourism destinations
J ohn Dawes J enni Romaniuk Annabel Mansfield
Article information:
To cite this document:
J ohn Dawes J enni Romaniuk Annabel Mansfield, (2009),"Generalized pattern in competition among tourism destinations", International
J ournal of Culture, Tourism and Hospitality Research, Vol. 3 Iss 1 pp. 33 - 53
Permanent link to this document:
http://dx.doi.org/10.1108/17506180910940333
Downloaded on: 24 January 2016, At: 22:06 (PT)
References: this document contains references to 52 other documents.
To copy this document: [email protected]
The fulltext of this document has been downloaded 1821 times since 2009*
Users who downloaded this article also downloaded:
Sameer Hosany, Yuksel Ekinci, Muzaffer Uysal, (2007),"Destination image and destination personality", International J ournal of Culture,
Tourism and Hospitality Research, Vol. 1 Iss 1 pp. 62-81 http://dx.doi.org/10.1108/17506180710729619
Graham Hankinson, (2005),"Destination brand images: a business tourism perspective", J ournal of Services Marketing, Vol. 19 Iss 1 pp.
24-32 http://dx.doi.org/10.1108/08876040510579361
Saila Saraniemi, (2011),"From destination image building to identity-based branding", International J ournal of Culture, Tourism and
Hospitality Research, Vol. 5 Iss 3 pp. 247-254 http://dx.doi.org/10.1108/17506181111156943
Access to this document was granted through an Emerald subscription provided by emerald-srm:115632 []
For Authors
If you would like to write for this, or any other Emerald publication, then please use our Emerald for Authors service
information about how to choose which publication to write for and submission guidelines are available for all. Please visit
www.emeraldinsight.com/authors for more information.
About Emerald www.emeraldinsight.com
Emerald is a global publisher linking research and practice to the benefit of society. The company manages a portfolio of more than
290 journals and over 2,350 books and book series volumes, as well as providing an extensive range of online products and additional
customer resources and services.
Emerald is both COUNTER 4 and TRANSFER compliant. The organization is a partner of the Committee on Publication Ethics (COPE)
and also works with Portico and the LOCKSS initiative for digital archive preservation.
*Related content and download information correct at time of download.
D
o
w
n
l
o
a
d
e
d
b
y
P
O
N
D
I
C
H
E
R
R
Y
U
N
I
V
E
R
S
I
T
Y
A
t
2
2
:
0
6
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
Generalized pattern in competition among
tourism destinations
John Dawes, Jenni Romaniuk and Annabel Mans?eld
Abstract
Purpose – The purpose of this paper is to examine competition between tourism destination brands in
terms of how they share travelers with each other.
Design/methodology/approach – The study analyzes survey data from four international markets
(USA, UK, Japan and Singapore). The study examines the cross-purchasing of travel destinations. It
applies an established empirical generalization, the duplication of purchase law (DPL) to frame
hypotheses and contextualize results.
Findings – The overall results are consistent with the DPL. Destination brands share tourists with other
destinations generally in-line with the popularity of the competing destination. However, there are very
noticeable market partitions, most of which take two forms: destinations that are either geographically
close to each other, or close to the point of origin. Destination brands in these partitions share travelers
far more than they would be expected to, given their respective size.
Practical implications – Tourism marketers need to appreciate the broad nature of competition. A
speci?c destination brand competes with many other travel destinations, sharing customers more with
other broadly popular destinations and less with less popular destinations.
Originality/value – The analytical approach presented in this study provides a straightforward
benchmark for assessing the expected level of competition between particular tourist destinations,
given their respective overall popularity.
Keywords Brands, Law, Competitive strategy, Tourism, Consumer behaviour
Paper type Research paper
1. Introduction
Tourism is a major, and increasing, source of revenue for countries all around the world. At
present, tourism accounts for over 4 percent of global GDP. Travel and tourism together
account for approximately 10 percent of world GDP (WTTC, 2003) and recent forecasts are
for strong growth (WITC, 2008). The industry earns in excess of US$1.3 billion per day, and is
the biggest export earner worldwide. Travel and tourism, together with suppliers to the
industry, employs over 200 million people (Belau, 2003). With such large economic
contributions, most countries have a tourism marketing body, responsible for the marketing
of travel destinations. The marketing managers within these bodies direct substantive
dollars towards tourism marketing and research, with the aim of better understanding their
market, their competitive positioning and howthey can use this knowledge to attract tourists.
A fundamental question for marketers of tourist destination brands relates to the competitive
structure of the market: which other brands does my brand compete against more intensely?
Which other brands does my brand gain sales from, and to which other brands does my
brand lose sales to? We apply a well-established method that has resulted in empirical
generalizations about competitive structure in numerous market settings, to address these
questions. The method is to analyze cross-purchases, namely the proportion of one brand’s
DOI 10.1108/17506180910940333 VOL. 3 NO. 1 2009, pp. 33-53, Q Emerald Group Publishing Limited, ISSN 1750-6182
j
INTERNATIONAL JOURNAL OF CULTURE, TOURISM AND HOSPITALITY RESEARCH
j
PAGE 33
John Dawes and
Jenni Romaniuk are based
at Ehrenberg Bass Institute
for Marketing Science,
University of South
Australia, Adelaide,
Australia.
Annabel Mans?eld is based
at the Division of Business,
University of South
Australia, Adelaide,
Australia.
Received May 2008
Revised September 2008
Accepted November 2008
D
o
w
n
l
o
a
d
e
d
b
y
P
O
N
D
I
C
H
E
R
R
Y
U
N
I
V
E
R
S
I
T
Y
A
t
2
2
:
0
6
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
customer base who also buy another brand. The application of cross-purchase analysis has
been very successful in identifying systematic patterns, leading to the development of the
duplication of purchase law(Ehrenberg, 2000). The duplication of purchase law(DPL) is that
brands generally share customers with other brands in line with market share. This
proposition means that the largest driver in estimating which competitors that any brand’s
customers are buying is the market share of the competitor brand. The DPL not only provides
this qualitative information, but also quanti?es the level of competition, as will be explained
later in the paper. Additionally the DPL allows researchers and marketers to identify and
quantify exceptions (referred to as partitions) to the general pattern. Partitions occur if
certain brands share more or fewer customers than would be expected. Analyzing brands
within partitions helps identify the key factors that in?uence competition levels and their
relative importance. Analysis using the DPL has been successfully applied in many different
market settings: media (Goodhardt and Ehrenberg, 1969); packaged goods markets
(Ehrenberg and Goodhardt, 1970); durables (Colombo et al., 2000) and services (Bennett
and Ehrenberg, 2001; Bennett and Ehrenberg, 2002). Applications of the DPL do not include
tourist contexts but the approach offers a promising framework to examine competition
among tourist destinations.
This research has three objectives. The ?rst is to quantify the level of competition in the
long-haul tourist market, by analyzing aggregate-level data on consumer’s
cross-purchasing of travel destinations. By the term aggregate, we mean that our unit of
analysis is the brand, not the individual traveler. The information on each brand is
aggregated up from individual-level consumer behavior pertinent to that brand. The extent
of competition between the brands is then analyzed. There is still obviously individual-level
heterogeneity among the consumers within a particular market, but the results represent the
aggregate probability of certain travel behaviors, among consumers from a market of
interest.
The second aim of this research is to identify the key in?uences on destination brand
competition. The third aim is to then quantify those in?uences on brand competition to
ascertain which are more important. In other words, to ?nd out what country-level factors
associate with heightened or lessened competition between tourist destination brands, over
and above those factors relating to brand size. To do this, the study analyzes consumer’s
cross-purchasing of destination brands across multiple origin countries, and over different
time periods.
2. Background to the research
Recently, three factors dramatically in?uence the competitive environment for tourist travel
destinations. First, the international tourism market exhibits very high rates of growth. The
number of total international tourist arrivals worldwide grew 50 percent between 1990 and
2000 (Hotels, 2002); with more modest annual growth of 3 percent forecast for 2008 (WITC,
2007). Total receipts from the industry were estimated at e207 Billion in 1990 and e512 Billion
in 2000 (Hotels, 2002). Second, the economic bene?ts of tourism are coaxing many once
closed countries (e.g., China, Cuba, and Taiwan) to open their borders in an attempt to
obtain tourist dollars. For example, visitor numbers to Cuba grew from 200,000 per year in
the 1980s to two million by the year 2000 (Kolland et al., 2000). The explosive growth in
popularity of such destinations means that for most destination brands, overall market
growth is tempered by heightened competition. Third, tourists venture further from their
home country because new low-cost entrants offer lower airfares (e.g., Singh and Catlin,
2003); as well as the popularity of package holidays, rising incomes, faster aircraft (Liu,
2000) and the enhanced infrastructure and marketing efforts of new competitors. For
example, the proportion of trips to Western Europe by UK tourists declined between 1987
and 1997 while the proportion to the rest of the world increased (Liu, 2000). This trend
appears likely to continue, with long-haul travel forecasted to grow at a greater rate than
international travel generally (Law, 2004).
For marketers, a widening competitor context is occurring because of the increase in
long-haul travel and more destinations being available. More people are traveling more
PAGE 34
j
INTERNATIONAL JOURNAL OF CULTURE, TOURISM AND HOSPITALITY RESEARCH
j
VOL. 3 NO. 1 2009
D
o
w
n
l
o
a
d
e
d
b
y
P
O
N
D
I
C
H
E
R
R
Y
U
N
I
V
E
R
S
I
T
Y
A
t
2
2
:
0
6
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
often, over longer distances, and to potentially more destinations. Therefore a framework for
understanding competitors and competition that can be applied in this dynamic
environment seems timely. The unit of analysis that is particularly useful is brand-level –
for example, which other brands does brand A compete with the most?
2.1 Conceptual framework/de?nitions
The present study uses the term market to mean international travelers. Brands describes
countries as destinations (e.g., Kotler and Gertner, 2002), such as Germany, the United
States or South Africa. The study examines countries as the unit of analysis, although tourist
destinations can also take the form of regions (e.g., Western Europe); or cities (e.g., Paris).
Thepurchasingof travel destinationsis analogoustobrandbuying. For example, inpackaged
goods markets buyers make repeated purchases from a category and buy different brands
over cumulative occasions. In the travel context, an individual can buy from the category
multiple times by taking multiple travel trips over time (e.g., a holiday every year over a ?ve
year time period). Each of those holidays may be to one or more destinations, opening up the
potential for multi-brand buying or repeat brand buying. Therefore destinations compete with
each other for customers, in the same way as brands do in other markets.
The present study uses the termcompetition in the context of sharing customers. If a traveler
travels to destination A and also B, then A and B compete. Competition can be manifest in
several ways. The ?rst is competing for an allocation of the traveler’s time during a particular
trip – akin to their share of wallet during the trip. The second is competing for the traveler’s
choice over consecutive trips. For example, of the travelers who went to France on their last
trip, how many go again next time? This is competition for the traveler’s repeat-purchase.
Many ways exist to exhibit competition among brands but the present study uses the sharing
of customer’s travel destinations its focus.
2.2 Competitive market structure and duplication of purchase
An appreciation of how buyers of one brand also buy competitor brands over time is a key
aspect of understanding brand competition. Brands compete against other brands for the
purchases of consumers each time those consumers buy from the category. A particular
brand A can gain sales from competitors B, C, and D; and can lose sales to these
competitors. Which brand does the focal brand lose sales to, and fromwhich brand does the
focal brand gain sales from? Academic researchers have examined this issue for many
years, using a variety of methods. One method that is particularly appealing is to examine
howcustomers of each brand allocate their purchases over multiple occasions. This method
is readily understandable and poses minimal data requirements other than purchase
histories. As an example, we can consider this question: of the buyers of brand A, how many
also buy brand B, C, or D over a time period. If many buyers of A also buy B, but few of them
buy Dthen A competes more heavily with B than D. This type of approach to analysing brand
competition has been used in many applications, and with many nuances – but an
overriding theme is that competitor brand’s market share is a primary factor in estimating
how many brand buyers also buy those competitor brands. The duplication of purchase law
succinctly summarises this phenomena: brands share customers with other brands in-line
with the market share of those other brands (Ehrenberg, 2000, Chapter 9). We review
relevant literature on the DPL in the next section. Note that the termswitching is often used to
denote customers buying A then B. We use the terms switching, sharing and purchase
duplication interchangeably to communicate the same phenomenon of customers buying
multiple brands.
The earliest published study using the purchase duplication approach is by Goodhardt and
Ehrenberg (1969) that examines the extent to which viewers of certain television shows also
viewed other shows. Goodhardt and Ehrenberg (1969) report a consistent pattern: the
proportion of viewers of any television show A who also watched show B, C, D etc. was
predictable simply from the ratings of B, C, or D. Television viewing appeared to have a
strong stochastic component, with the market share (in this case, ratings) of the program
strongly in?uencing how many viewers switched to that program. Bass (1974, p. 29)
VOL. 3 NO. 1 2009
j
INTERNATIONAL JOURNAL OF CULTURE, TOURISM AND HOSPITALITY RESEARCH
j
PAGE 35
D
o
w
n
l
o
a
d
e
d
b
y
P
O
N
D
I
C
H
E
R
R
Y
U
N
I
V
E
R
S
I
T
Y
A
t
2
2
:
0
6
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
develops the theme of a pronounced stochastic component in brand switching and showed
that consumer switching between soft-drink brands was substantially in line with the market
share of those brands. Ehrenberg (2000) more fully expounds the phenomena in the
textbook Repeat Buying. The arithmetic formula is set out in Appendix 1.
Duplication of purchase-style analysis has been used in numerous applications from
frequently bought goods such as groceries (e.g., Bass, 1974; Ehrenberg and Goodhardt,
1970) and bottled wine (Romaniuk and Dawes, 2005). It has also been used in markets with
less frequent purchase patterns, such as electrical goods (Bennett, 2004) and motor
vehicles (Colombo et al., 2000). Service contexts have also been examined: quick service
restaurants (Bennett and Ehrenberg, 2002) retail store choice by consumers (Uncles and
Ehrenberg, 1990a) and mixed service/goods provision in a B-to-B setting (Ehrenberg,
1975b; Uncles and Ehrenberg, 1990b). The basic premise of the duplication of purchase law
also appears in the literature under other names. One name is the Aggregate Constant Ratio
Model which states ‘‘consumers select alternative products proportional to the market
shares of the alternatives’’ (Novak and Stangor, 1987, p. 84; see also Russell and Bolton,
1988, p. 230). The Aggregate Constant Ratio Model is used in several studies examining
competitive market structure (Bauer and Herrmann, 1995; Cooper and Inoue, 1996; Urban
et al., 1984). The term probabilistic independence (Hauser and Wernerfelt, 1989) is also
used to communicate the same idea as the DPL.
As mentioned above, tourist destination marketers should understand the competitive
structure of their market: of the travelers who visit us, what other destinations do they visit? Of
the travelers who visit rival destinations A, B, C, and so on, how many from each come to us?
These questions are explicitly considered using the duplication of purchase method of
analysis. Moreover, the DPL provides a useful framework and set of expectations within
which the results can be compared. If tourist destination brands follow the DPL this provides
a straightforward interpretation of the structure of competition between tourist brands. If
there are noticeable departures from the general pattern, those departures indicate speci?c
competitor destinations the marketing manager must pay close attention to.
One speci?c characteristic of travel destination purchasing occurs that differs from
purchasing in the types of markets analysed thus far using the purchase duplication
approach. That is, long-haul travel may encompass purchasing or visiting multiple
destinations on the same trip. By contrast, consumers in most other markets purchase tend
to purchase only one brand in a category at a time. This characteristic presents one
particular limitation to the study, namely that we cannot distinguish between partitions
formed from destinations that tend to be visited as part of a round trip, as opposed to
partitions based on consecutive purchases. However, as stated above, competition
between brands can be manifest as competition for repeat-purchase over consecutive trips,
or as competition for time spent during a particular trip (share-of-wallet). Therefore partitions
based on round trips can still represent heightened competition between particular brands
for the traveler’s spending. Also, partitions formed either from round trips or consecutive
purchases have a similar interpretation in relation to other brands outside the partition:
lowered competitive substitutability. Consider the scenario that destinations A and B tend to
be purchased together as part of a round trip, and C and D also tend to be purchased
together as a round trip. Purchase duplication analysis will indicate that these respective
pairs (A and B v. C and D) form two partitions and each partition shares fewer travelers with
the other partition than expected. Therefore the extent of competitive intensity across the
partitions is lower. This is the same interpretation that would arise if the combination of A and
B; or C and D, tended to be purchased together over consecutive trips. A round trip in which
the same destination is visited multiple times does not cause any bias in the present study
because we count the numbers of travelers to each destination, and do not consider how
many times they travel to that destination.
2.3 Outputs from purchase duplication analysis
This article uses the term purchase duplication analysis to mean an examination of the
purchases of pairs of competitive brands – how many people bought A and B; A and C, etc.
PAGE 36
j
INTERNATIONAL JOURNAL OF CULTURE, TOURISM AND HOSPITALITY RESEARCH
j
VOL. 3 NO. 1 2009
D
o
w
n
l
o
a
d
e
d
b
y
P
O
N
D
I
C
H
E
R
R
Y
U
N
I
V
E
R
S
I
T
Y
A
t
2
2
:
0
6
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
The DPL provides a framework for interpreting the results of purchase duplication analysis.
The different elements of the analysis each provide speci?c insights. We now discuss these
three elements in the context of destinations as brands:
2.3.1 Brand penetrations. How many travelers from the origin market have visited this brand
at least once. For example, how many British international tourists have visited France. This
example ?gure would represent the penetration ?gure for France, for the British traveler
market.
2.3.2 Brand duplications. How many people who traveled to brand A have also traveled to
brand B. The average duplication rate for each brand is derived from these ?gures. In other
words, what is the average proportion of travelers to any particular brand who also went to
brand B? According to the DPL, the proportion of people who traveled to destination A who
also traveled to B should be the same as the proportion of people who had traveled to C, D
etc. who then also traveled to B. Figure 1 shows an example of the way brand duplications
are usually represented.
2.3.3 Duplication co-ef?cient. This statistic re?ects the total amount of purchase duplication
in the market, allowing for the sizes of the brands. The duplication co-ef?cient is calculated
as the average of all purchase duplications divided by the average of all brand penetrations,
and is often denoted as D (Ehrenberg, 2000) – see Appendix 1 for details. The duplication
co-ef?cient provides the basis for calculating expected duplication levels for each brand.
The expected duplication between any pair of brands A and B is the penetration of B
multiplied by D. For example, if we calculate the duplication co-ef?cient for a market to be
1.5, this means the expected proportion of travelers to one brand that also travel to another
brand is 1.5 times the penetration of that other brand (as per Ehrenberg, 2000, Ch. 9). These
model-generated, expected duplications are useful because marked exceptions to them
indicate partitions in the market that can be further investigated.
2.3.4 Partitions. If a certain pair or group of brands exhibits quite different levels of purchase
duplication to their expected duplication values, this indicates heightened competitive
intensity between those brands. Heightened competitive intensity is called a partition
Figure 1 Portrayal of purchase duplication analysis and the DPL
VOL. 3 NO. 1 2009
j
INTERNATIONAL JOURNAL OF CULTURE, TOURISM AND HOSPITALITY RESEARCH
j
PAGE 37
D
o
w
n
l
o
a
d
e
d
b
y
P
O
N
D
I
C
H
E
R
R
Y
U
N
I
V
E
R
S
I
T
Y
A
t
2
2
:
0
6
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
(e.g., Kalwani and Morrison, 1977). Conversely, lower than expected sharing indicates the
brands do not compete much with each other at all, allowing for their size. The relative level
of competitive intensity can be quanti?ed. For example, suppose the duplication of
purchase analysis ?nds the expected proportion of people who have traveled to any
particular brand that also travel to brand B is 30 percent. But for a particular brand A, 60
percent of people who traveled to A also travel to B. This is interpretable as B having twice
the level of competitive substitutability with A as it should. Once partitions are identi?ed as
statistically signi?cant, and numerically large enough to be managerially signi?cant, the
possible reasons for the partition can be examined (e.g., location versus cultural reasons).
The marketer can then identify the dominant reasons for market partitions.
2.4 In?uences on brand competition
The previous section outlines the concept of partitions – pairs or clusters of brands that
compete more intensely, or less intensely, than expected. There are potentially many
characteristics of destination brands that could in?uence the formation of partitions. We
examine prior literature on market partitions in the next section to develop hypotheses. We
reiterate that we are constructing an aggregate-level model using the brand as the unit of
analysis and are not considering individual-level consumer variables such as perceptions,
perceived value, or bene?ts sought on the part of a speci?c traveler.
2.4.1 Functional differences. Market partitions are mostly attributable to functional
differences between certain groups of brands (e.g., Ehrenberg et al., 2004). For example,
coffee is partitioned into decaffeinated and regular, as well as ground and instant. The
functional differences between the coffee types lead to people buying brands within each of
the sub-categories. This means that instant coffee drinkers are more likely to buy other
brands of instant coffee over other brands of ground coffee (Grover and Rao, 1988).
A further example of market structure based on functional differences is from the car
industry: analysis of British and French car markets revealed that they were partitioned into
four functionally different groups: Luxury makes, Japanese makes, Other European makes
and Large Local makes. Cars within each group shared more customers with other brands in
the same group than would have been expected given their market share (see Colombo
et al., 2000). Partitioning based on functional differences is also identi?ed in the petrol
market, between leaded and unleaded petrol (Bennett et al., 2000) and in cigarettes
between ?ltered/non ?ltered and menthol/non menthol (Carter and Silverman, 2004).
Culture is a functional difference among tourism destinations that is often discussed. Culture
is de?ned as the ‘‘sum total of learned beliefs, values and customs that serve to direct the
consumer behavior. . .of a society’’ (Schiffman and Kanuk, 2004, p. 405). Consumers in
different countries exhibit marked differences in core values (Hofstede, 1994); and
numerous studies relate core values to consumer behavior differences. In the context of
competing destinations, many components of culture may in?uence destination choice,
therefore play a part in the competitive structure of the market (Moutinho, 1987). For
example, tourists could use the perceived cultural similarity of a destination as a decision
variable. Culture may also impact on the way tourists from different destinations select their
own holiday destinations. For example, there is a frequently discussed difference between
Japanese and Western cultures (Reisinger and Turner, 1999).
The impact of culture on the decision process may result in a departure from the DPL in the
tourism travel context. This outcome may occur in two ways. First, the DPL might simply not
hold due to cultural differences in destination choice. This will occur if we see the overall
pattern of brand sharing in-line with size holds for some markets and not others – for
example, if the pattern holds for UK and the USA, but not Japan and Singapore or vice versa.
Second, cultural differences could form the basis for market partitions. This outcome is
evident if destinations of similar culture share more customers than expected. For example, if
international tourists from mainland China exhibit a greater likelihood than expected to visit
another country such as Hong Kong or Taiwan. A limitation exists in assuming cultural
homogeneity within an origin country. The DPL posits aggregate-level relationships,
PAGE 38
j
INTERNATIONAL JOURNAL OF CULTURE, TOURISM AND HOSPITALITY RESEARCH
j
VOL. 3 NO. 1 2009
D
o
w
n
l
o
a
d
e
d
b
y
P
O
N
D
I
C
H
E
R
R
Y
U
N
I
V
E
R
S
I
T
Y
A
t
2
2
:
0
6
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
essentially the same as re?ecting the probability of behavior for a randomly drawn consumer
in the market of interest.
2.4.2 Geographic proximity. Geographic location or proximity can also underlie market
partitioning. For example, Sharp and Sharp (1997) examine the competitive structure of
retail department stores, and identify a partition between geographically close stores,
despite them having very different brand images. Since the location of a country (and its
neighbours) is a major, unchangeable characteristic, market partitions based on proximity
could occur. Two ways exist in which proximity could exert an in?uence on competitive
structure. First, destinations that are close to the home country (e.g., USA with Canada and
Mexico) could share more tourists than expected due to their closeness and accessibility.
Travelers to one are more likely to then travel to the other over time. This outcome is
consistent with the well-established phenomena of distance decay in air travel patterns
(Crotts, 2004). Second, destinations close to each other (regardless of distance from the
home destination) may be more likely to share visitors as there is a great chance they will
visited by the same tourist on the same trip (e.g., Australia and New Zealand from the
perspective of UK tourists). In this instance the two destinations are competing for share of
holiday time within a single trip.
The propositions to this point are as follows: An understanding of aggregate-level patterns in
competition among destination brands is desirable. A well-established method and
analytical framework from other markets exists for this purpose. Given the widespread
applicability of the DPL across numerous markets, we hypothesize the law will hold overall,
for tourist destinations. Cultural similarity or geographic proximity could give rise to
deviations from the overall pattern. These deviations from expected levels of
cross-purchasing or purchase duplication may be manifest as partitions or clusters,
identi?able as destination brands that share customers to a greater (or lesser) extent with
each other than with the rest of the market. The outcome of the investigation is to identify an
overall pattern of competition among destination brands and then to identify any exceptions
to the general pattern, along with possible reasons for those exceptions.
3. The study
The analysis for the study has three stages. Stage 1 identi?es if the DPL holds overall, and
whether the law is applicable over different countries. To do this we examine four different
international markets, namely the United States, United Kingdom, Singapore and Japan.
The term ‘‘market’’ means the pool of international travelers who reside in the respective
country. Over the four markets, there are marked differences in cultural values. This
assertion follows from two observations. The ?rst and obvious one is that two are Western
countries and two are Eastern. Of the two Eastern countries, Japan has a large population
and is one of the largest economies in the world. The other, Singapore, is a recently
established island state with a very vibrant trading economy and a small population. The
second observation is that there are marked cultural differences across the four countries,
based on the country-level scores for Hoftstede’s measures of cultural distance (for a
description of each facet see Hofstede, 1994).
The USA and the UK are quite similar culturally, and therefore we expect similarities in their
results – but still, replication across culturally similar markets provides con?dence that the
results are not just one-off ?ndings. In contrast to the UK and USA, the other two markets,
namely Singapore and Japan, are very different cultures to the other two countries – and to
each other. Therefore, their inclusion extends the potential for generalizable results (see
Table I).
Stage 2 involves close replications to see if the same pattern emerges two years later.
Examining results froma different sample of tourist consumers for two of the markets, namely
USA and Japan, accomplishes Stage 2. Stage 3 focuses on identifying and interpreting
deviations from the general pattern. These deviations indicate possible market partitions.
The scope of this analysis (four markets, and two time periods) offers credible scope for
identifying a potential empirical generalization in competitive structure among long-haul
destination brands.
VOL. 3 NO. 1 2009
j
INTERNATIONAL JOURNAL OF CULTURE, TOURISM AND HOSPITALITY RESEARCH
j
PAGE 39
D
o
w
n
l
o
a
d
e
d
b
y
P
O
N
D
I
C
H
E
R
R
Y
U
N
I
V
E
R
S
I
T
Y
A
t
2
2
:
0
6
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
3.1 Survey data
The data used for this analysis were collected as part of an ongoing brand tracking research
project for an international tourism body. An accredited international market research
provider collected the data. The recruitment procedure followed normal market research
industry quality standards. International travelers were the population of interest.
Respondents were randomly recruited via telephone, which means each individual
respondent was recruited individually from a sampling frame and all in the sampling frame
had an equal opportunity of being part of the survey.
Once contacted, respondents were then screened for suitability via survey questions. Only
those respondents who stated some probability of traveling outside of their home region
within four years of the date of the interview were retained in the sample. Therefore the
sample characteristics re?ect those likely to be engaged in long haul overseas travel, rather
than simply re?ecting the total adult population of each country. The demographic
breakdown the sample from each country is shown in Appendix 2.
As part of a larger questionnaire into brand and advertising awareness and brand
perceptions, respondents were asked, which countries or destinations outside , home
country . have you ever visited for a vacation/holiday or sightseeing? Only people who had
traveled to more than one destination were included in the analysis, because the analysis
examines cross-purchases of multiple brands. The exclusion of single-destination travelers
does not compromise the analysis. It only has the effect of lowering the penetration and
duplication levels for all brands. The strength of relationship between penetration and
purchase duplication, and the extent of partitioning, are not affected by excluding
single-destination travelers. Prior studies in contexts such as automobile purchasing
necessarily exclude single-purchase buyers (Colombo et al., 2000). Some destinations
might exhibit particular appeal among one-time-only travelers, and therefore enjoy a
competitive advantage that would not be apparent from this analysis. Prior research in other
markets indicates that infrequent buyers tend to unduly favor larger brands (e.g., Ehrenberg
et al., 2004).
This procedure provides analysis samples ranging for 41 percent for Japan, which had the
largest number of never traveled or single destination travelers, to 74 percent for Singapore,
which had the most multiple destination travelers. For each of the four markets, we analyze
the travel patterns of residents of that country by examining howmany respondents fromthat
country visit certain other countries, and of the travelers to country A, what proportion also
travel to country B, C, D and so on. A summary of the data is as follows (see Table II).
The study includes the creation of purchase duplication tables for each country-market. A
visual inspection of the purchase duplication tables will give an initial indication if the DPL
holds in this context. According to the DPL, the average proportion of travelers to any
country who also visit any other country should decline in line with the penetration of that
other country. Explication of this pattern is possible by simply ordering the brands in
descending size (popularity) by row and by column. Ordering data by size is very useful
procedure to show patterns in data (Ehrenberg, 1975a). Figure 1 shows this format and the
expected pattern. Due to space limitations we can only show one complete purchase
duplication table, but we show the results for all four markets in a series of graphs.
Table I Countries classi?ed according to Hoftstede’s measures of cultural difference
USA UK Singapore Japan
(score/100) (score/100) (score/100) (score/100)
Power – distance 40 35 74 54
Individualism 91 89 20 46
Masculinity 62 66 48 95
Uncertainty – avoidance 46 35 8 92
Long term orientation 29 25 48 80
Notes: High scores for these cultural aspects are in italics for clarity
PAGE 40
j
INTERNATIONAL JOURNAL OF CULTURE, TOURISM AND HOSPITALITY RESEARCH
j
VOL. 3 NO. 1 2009
D
o
w
n
l
o
a
d
e
d
b
y
P
O
N
D
I
C
H
E
R
R
Y
U
N
I
V
E
R
S
I
T
Y
A
t
2
2
:
0
6
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
Comparing the expected purchase duplication levels deriving from the duplication
co-ef?cient, to the actual levels serves to assess the overall level of model ?t. The
appropriateness of the DPL to the data is examined using three different measures. The ?rst
?t measure is mean absolute deviation (MAD) between the actual duplication, and the
model-generated estimates of duplication derived fromthe duplication co-ef?cient. MADis a
widely used ?t measure (e.g., Armstrong, 2001). For the UK market as an example, we take
the actual duplication for the proportion of those who traveled to France who also travel to
Germany, and compare this to the model-generated estimate. We repeat this comparison
over each combination of brands and calculate the mean absolute deviation between actual
and estimated ?gures, over all brands. Note that we use the MADsimply to communicate the
general extent to which the model provides estimates that match the empirical results. We
do not apply inferential tests to the MAD’s.
The second measure of ?t is the correlation between the model-generated purchase
duplications and the average purchase duplications for each brand. If a high correlation
exists between the model-based estimates and the actual data, the DPL represents
aggregate-level brand competition very well. Finally, we assess the statistical signi?cance of
market partitions using a permutational method outlined in Appendix 3.
The use of multiple ?t measures is useful, because any particular measure has a
shortcoming. The correlation co-ef?cient assesses the extent to which the two variables vary
together, but does not indicate if matched observations for two variables are actually
numerically close to each other. The MAD assesses the extent to which the values are
numerically close to each other, but does not provide a direct interpretation of the level of
co-variability or the statistical signi?cance of partitions. The statistical signi?cance of
particular brands that deviate from expected duplication levels is useful to properly identify
partitions, but does not necessarily inform us of the overall extent to which the DPL applies.
4. Results
4.1 Fit of the DPL
Consider a detailing of the results for the UK market, that is, the destinations of travelers from
the UK. The analysis focuses on the largest ten destinations. The destination brands are
ordered by size in both columns and rows. Interpret Table III as follows. Under the
penetration (Pen percent) column, we see 65 percent of international travelers from the UK
have traveled to France, 60 percent to Spain and so on. Looking across the rowfor France, of
the people who have traveled to France, 63 percent of those have also traveled to Spain, 36
percent also to the USA and so on to 19 percent having also traveled to Portugal. Overall, the
results show a strong in?uence of brand penetration on duplications, consistent with the
DPL. Of the people who have traveled to any destination in this list, more of them have also
traveled to France and Spain, the two largest destinations. Far fewer have also traveled to
Switzerland or Portugal, the two relatively less popular destinations for this market. That said,
we see variation in the individual cells down each column, which indicates market partitions.
Table II Summary of data
Analysis sample size
n ¼
Percent of
total sample
Sample of international travelers residing in. . .
UK 822 62
USA 434 43
Singapore 714 74
Japan 278 41
Additional samples to test generalizability of patterns over time
USA 491 47
Japan 318 53
VOL. 3 NO. 1 2009
j
INTERNATIONAL JOURNAL OF CULTURE, TOURISM AND HOSPITALITY RESEARCH
j
PAGE 41
D
o
w
n
l
o
a
d
e
d
b
y
P
O
N
D
I
C
H
E
R
R
Y
U
N
I
V
E
R
S
I
T
Y
A
t
2
2
:
0
6
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
We highlight statistically signi?cant deviations with an asterisk, and these partitions are
discussed shortly.
To more clearly show the purchase duplication pattern, the ?ndings include the results for
the UK, and the other three markets, in scatterplots (Figures 2–5). The X-axis represent the
penetration levels for each of the ten brands, and the Y-axis represent the proportion of
buyers of any destination brand who also traveled to the X-axis brand. We see visual
evidence of a strong relationship between penetration and brand duplication.
The analyses are for all four markets using the same method. Tables IV, V, VI and VII shows
the summary statistics. The ?ndings include a strong effect of brand size, consistent with the
DPL. The correlations between actual and predicted average sharing of tourists are 0.98,
0.68, 0.98, and 0.94 for the UK, USA, Singapore and Japan markets respectively. The mean
absolute deviations between observed and model-generated estimates of sharing range
Table III Duplication of purchase: results for UK market Wave 1 (top ten destinations)
% UK travelers who % Who also traveled to . . .
traveled to Pen % France Spain USA Italy BNL Greece Germany Austria Switzerland Portugal
France 65 63* 36 42 41 33 41 26 25 19
Spain 60 67* 35 38 33 39 30 19 19 20
USA 35 68* 61* 45 34 37 32 24 23 21
Italy 33 82 69 47 47 45 46* 36* 35* 23
BNL
a
32 83 63* 37 50 36 52 33 28 24
Greece 31 68* 75 41 48 37 35 27 25 23
Germany 31 85 59* 36 49 53 35 40* 33 21
Austria 20 84 58* 42 60* 51 41 61* 43* 25
Switzerland 18 87 63* 43 63* 50 43 56* 47* 24
Portugal 15 80 81 50 50 51 47* 43 33 29
Model-estimated
duplication (Pen £ 1.3) 87 81 46 45 42 42 42 27 25 20
Average duplication 78 64 43 49 45 40 45 32 30 23
Duplication coef?cient 1.3
MAD (estimated to actual
duplication) average ¼ 5 9 17 3 5 3 2 3 5 5 3
Notes:
a
¼ Benelux countries; * Indicates statistically signi?cantly different to expected level at p , 0.05
Figure 2 Scatterplot of brand penetrations and duplications (UK market)
PAGE 42
j
INTERNATIONAL JOURNAL OF CULTURE, TOURISM AND HOSPITALITY RESEARCH
j
VOL. 3 NO. 1 2009
D
o
w
n
l
o
a
d
e
d
b
y
P
O
N
D
I
C
H
E
R
R
Y
U
N
I
V
E
R
S
I
T
Y
A
t
2
2
:
0
6
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
from 1.0 to 8.2 percentage points. The purchase duplication coef?cients range from 0.9 to
1.3. This co-ef?cient, when multiplied by the penetration of each country estimates how
many, on average of the people who have traveled to any of a set of countries (A, B, C, etc.)
will have also traveled to country D. There is no inferential test for the purchase duplication
coef?cient, but the coef?cient produces estimates that correlate highly with the average
duplication levels for each brand at 0.68 to 0.98 (all p , 0.03 or lower). For the USA, UK and
Singapore the coef?cient is around 1.3, while for Japan the coef?cient is around 1. This
suggests Japanese tourists travel to a smaller range of destinations than USA, UK and
Singaporean Tourists. This result might be due to greater loyalty to single destinations or due
to taking fewer holidays overall.
Figure 3 Scatterplot of brand penetrations and duplications (USA market)
Figure 4 Scatterplot of brand penetrations and duplications (Singapore market)
VOL. 3 NO. 1 2009
j
INTERNATIONAL JOURNAL OF CULTURE, TOURISM AND HOSPITALITY RESEARCH
j
PAGE 43
D
o
w
n
l
o
a
d
e
d
b
y
P
O
N
D
I
C
H
E
R
R
Y
U
N
I
V
E
R
S
I
T
Y
A
t
2
2
:
0
6
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
We conduct the same analysis for USA and Japan data gathered two years later, to test the
over-time stability of these patterns. The sampling regime and questioning techniques were
the same, but the individual respondents were different. The results (in Table VIII) show the
overall pattern holds in both countries on a different set of data, with similar duplication
Figure 5 Scatterplot of brand penetrations and duplications (Japan market)
Table V Duplication of purchase (top ten destinations in each market)
USA Wave 1 Mex Fra Car Ger Can UK Ita Swi Aus Haw Average MAD
Correlation
(Act. v Est.)
Average duplication 47 47 27 45 20 34 39 32 26 16
Model-estimated duplication
(Pen £ 1.3) 61 39 39 37 33 33 31 23 19 18
MAD (estimated to actual duplication) 14 8 12 7 13 1 8 9 8 2
Duplication coef?cient 1.3 8.2 0.68
Notes: MAD ¼ Mean Absolute Deviation across all pairs of duplicated destinations; Act: ¼ actual duplications; Est: ¼ model-estimated
duplications
Table IV Duplication of purchase (top ten destinations in each market)
UK Wave 2 Fr Sp USA Ita BNL Gre Ger Aus Swi Por Average MAD
Correlation
(Act. v Est.)
Average duplication 78 64 43 49 45 40 45 32 30 23
Model- estimated duplication
(Pen £ 1.3) 87 81 46 45 42 42 42 27 25 20
MAD (estimated to actual duplication) 9 17 3 5 3 2 3 5 5 3
Duplication coef?cient 1.3 5.0 0.98
Notes: MAD ¼ Mean Absolute Deviation across all pairs of duplicated destinations; Act: ¼ actual duplications; Est: ¼ model-estimated
duplications
PAGE 44
j
INTERNATIONAL JOURNAL OF CULTURE, TOURISM AND HOSPITALITY RESEARCH
j
VOL. 3 NO. 1 2009
D
o
w
n
l
o
a
d
e
d
b
y
P
O
N
D
I
C
H
E
R
R
Y
U
N
I
V
E
R
S
I
T
Y
A
t
2
2
:
0
6
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
coef?cients in both years for both markets and correlations between estimated and
predicted average sharing ?gures above 0.93.
4.1.1 A systematic deviation. A systematic deviation occurs in the general pattern of results.
The model-estimated duplications are markedly over-estimated for high-penetration
destinations. The greatest observed deviations occur for brands that have penetrations
higher than 50 percent. Although this seems signi?cant, the overestimation of predicted
duplications is a known ?aw of the duplication of purchase model, known to generally occur
for brands with penetrations greater than 50 percent (Ehrenberg, 2000; Sharp and Sharp,
1997). The fact the same ?aw occurs in this study provides additional con?dence that the
same underlying patterns found in other types of markets also hold for tourism destinations.
4.2 Market partitions
Brands can share more (or fewer) customers than expected given their respective size.
From a competition perspective, this means that certain brands compete more intensely or
less intensely with each other – so-called partitions (Kalwani and Morrison, 1977). Partitions
have important strategic implications. They can re?ect emerging sub-categories that break
away to form new markets and present opportunities for new brand launches or positioning
directions. Furthermore, if a marketer is managing a partitioned brand, they need to pay
special attention to the other brands in the partition, because the marketing activity of those
competitors can have more impact than other brands of the same size. Therefore knowing
about, and understanding market partitions can increase marketing effectiveness by
streamlining resources towards the most important competitive issues.
This research identi?es partitions by identifying which individual countries deviate markedly
from the expected duplication ?gure. Referring back to Table III for illustration, the study
identi?es a partition by subtracting the estimated duplication level for Austria (27 percent)
from the actual duplication ?gure for travelers to Switzerland who also travel to Austria (47
percent), which is a deviation of 20. In other words, UK travelers who have been to
Switzerland are 20 percentage points more likely than expected to also have traveled to
Austria – given the overall popularity of Austria and the general extent to which destinations
share customers in this market. This particular partition is statistically signi?cant at p , 0.05.
Table VI Duplication of purchase (top ten destinations in each market)
Singapore Wave 1 Mal Tha Bal Aus HK Chi Tai USA Jap Kor Average MAD
Correlation
(Act. v Est.)
Average duplication 82 59 55 39 47 33 24 23 25 14
Model-estimated duplication
(Pen £ 1.2) 101 63 62 41 39 27 20 19 18 12
MAD (estimated to actual duplication) 19 4 8 2 8 6 4 4 7 2
Duplication coef?cient 1.2 6.0 0.98
Notes: MAD ¼ Mean Absolute Deviation across all pairs of duplicated destinations; Act: ¼ actual duplications; Est: ¼ model-estimated
duplications
Table VII Duplication of purchase (top ten destinations in each market)
Japan Wave 1 Fr Sp USA Ita BNL Gre Ger Aus Swi Por Average MAD
Correlation
(Act. v Est.)
Average duplication 37 27 25 26 19 23 23 16 16 13
Model-estimated duplication ¼ ðPen £ 1.4) 37 27 24 24 22 22 22 18 15 15
MAD (estimated to actual duplication) 0 0 1 2 3 1 1 2 1 2
Duplication coef?cient 0.9 1.0 0.94
Notes: MAD ¼ Mean Absolute Deviation across all pairs of duplicated destinations; Act: ¼ actual duplications; Est: ¼ model-estimated
duplications
VOL. 3 NO. 1 2009
j
INTERNATIONAL JOURNAL OF CULTURE, TOURISM AND HOSPITALITY RESEARCH
j
PAGE 45
D
o
w
n
l
o
a
d
e
d
b
y
P
O
N
D
I
C
H
E
R
R
Y
U
N
I
V
E
R
S
I
T
Y
A
t
2
2
:
0
6
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
T
a
b
l
e
V
I
I
I
C
o
m
p
a
r
i
s
o
n
o
f
r
e
s
u
l
t
s
o
v
e
r
m
u
l
t
i
p
l
e
y
e
a
r
s
:
J
a
p
a
n
a
n
d
U
S
A
J
a
p
a
n
W
a
v
e
1
J
a
p
a
n
W
a
v
e
2
U
S
A
W
a
v
e
1
U
S
A
W
a
v
e
2
A
B
C
A
B
C
A
B
C
A
B
C
D
e
s
t
P
e
n
%
A
v
g
D
u
p
E
s
t
D
u
p
D
e
s
t
P
e
n
%
A
v
g
D
u
p
E
s
t
D
u
p
D
e
s
t
P
e
n
%
A
v
g
D
u
p
E
s
t
D
u
p
D
e
s
t
P
e
n
%
A
v
g
D
u
p
E
s
t
D
u
p
H
a
w
4
0
3
7
3
7
H
a
w
5
3
4
8
5
4
M
e
x
4
8
4
7
6
1
F
r
a
n
4
9
6
1
6
3
U
S
A
3
0
2
7
2
7
U
S
A
3
7
3
5
3
8
F
r
a
3
1
4
7
3
9
U
K
4
1
4
7
5
2
K
o
r
2
6
2
5
2
4
G
u
a
3
2
2
8
3
2
C
a
r
3
1
2
7
3
9
I
t
a
4
0
4
9
5
1
H
K
2
6
2
6
2
4
H
K
2
8
3
1
2
9
G
e
r
3
0
4
5
3
7
M
e
x
3
4
3
6
4
3
G
u
a
2
4
1
9
2
2
F
r
a
2
3
3
1
2
3
C
a
n
2
6
2
0
3
3
G
e
r
2
7
3
8
3
5
F
r
a
2
4
2
3
2
2
S
i
n
g
2
3
2
0
2
3
U
K
2
6
3
4
3
3
C
a
n
2
5
2
8
3
2
S
i
n
2
4
2
3
2
2
A
u
s
2
2
2
0
2
3
I
t
a
2
5
3
9
3
1
S
p
a
2
0
2
5
2
6
A
u
s
2
0
1
6
1
8
K
o
r
1
9
1
9
2
0
S
w
i
1
8
3
2
2
3
C
a
r
i
2
0
2
0
2
6
U
K
1
7
1
6
1
5
I
t
a
1
8
2
3
1
9
A
u
s
1
5
2
6
1
9
B
N
L
1
6
2
3
2
1
T
a
i
1
6
1
3
1
5
U
K
1
7
2
2
1
7
H
a
w
1
5
1
6
1
8
S
w
i
1
5
2
3
2
0
D
0
.
9
1
.
0
1
.
3
1
.
3
C
o
r
r
.
0
.
9
7
0
.
9
3
0
.
9
7
0
.
9
7
M
A
D
6
.
4
7
.
1
8
.
2
3
.
5
N
o
t
e
s
:
D
¼
D
u
p
l
i
c
a
t
i
o
n
c
o
e
f
?
c
i
e
n
t
;
C
o
r
r
:
¼
c
o
r
r
e
l
a
t
i
o
n
b
e
t
w
e
e
n
a
v
e
r
a
g
e
d
u
p
l
i
c
a
t
i
o
n
(
A
v
g
D
u
p
)
a
n
d
e
s
t
i
m
a
t
e
d
d
u
p
l
i
c
a
t
i
o
n
(
E
s
t
D
u
p
)
;
M
A
D
¼
M
e
a
n
A
b
s
o
l
u
t
e
D
e
v
i
a
t
i
o
n
b
e
t
w
e
e
n
a
c
t
u
a
l
d
u
p
l
i
c
a
t
i
o
n
s
a
n
d
m
o
d
e
l
-
e
s
t
i
m
a
t
e
d
d
u
p
l
i
c
a
t
i
o
n
s
PAGE 46
j
INTERNATIONAL JOURNAL OF CULTURE, TOURISM AND HOSPITALITY RESEARCH
j
VOL. 3 NO. 1 2009
D
o
w
n
l
o
a
d
e
d
b
y
P
O
N
D
I
C
H
E
R
R
Y
U
N
I
V
E
R
S
I
T
Y
A
t
2
2
:
0
6
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
Continuing the examination of the UK market, we see Spain, USA and Greece share fewer
travelers with France than expected. Similarly, several countries share fewer travelers with
Spain than expected: USA, Benelux, Germany, Austria and Switzerland (that said, over half
the travelers to any of these countries still also visit Spain). There are other noticeable
partitions, with heightened sharing of travelers between Germany, Austria, Switzerland and
Italy. These partitions are all statistically signi?cant and are 20 percent (or more) different to
the expected ?gure, which indicates they are worthy of managerial attention.
In Table IX we show major partitions across the four markets. Most involve destinations
sharing travelers more than expected, but we also see that certain brands share travelers far
less than expected (all exhibit differences to expected values at p , 0.05). We discuss the
partitions that appear to be based on geographic proximity ?rst. The effect of proximity can
occur where the countries in the partition are both or all close to the origin market; or where
the partitioned countries are comparatively far fromthe origin market but close to each other
(designated as ‘‘1’’ or ‘‘2’’ respectively in Table IX). Proximity is associated with heightened
sharing, in some cases to a very large extent. We also note that tourist market partitions
occur that are based on destinations sharing a similar culture and language. For example,
Austria, Switzerland and Germany share 38 percent more tourists from the UK market than
expected (Italy is also in this partition). However, these destinations also share geographic
proximity to each other. The combination of geographical and cultural similarity makes it
dif?cult to say which is the better explanation for the partition.
Consider the destinations that share travelers less with each other than expected. China and
France, for example, exhibit lower than expected sharing of USA travelers. This may be due
to the long distance between each country. Instances of lower levels of sharing occur that
could build from cultural factors. A very noticeable example is from the Singapore market:
Singaporeans who travel to India are far less likely to also travel to China (3 percent
compared to 28 percent expected). This proposal follows from the cultural heterogeneity in
Singapore, speci?cally, that this city-state has a signi?cant Indian population.
Finally, in relation to culture and its possible effects, we proposed earlier that the impact of
culture might be exhibited through a poorer ?t of the DPL. We can make some exploratory
observations about this issue from examining the overall results for the two Western markets
with the two Eastern markets. Inspection of the graphs in Figures 2–5 shows the correlations
between penetration and duplications are higher for UK and USA (r ¼ 0:89 and 0.83)
Table IX Major partitions
Market Wave Countries in partition Actual dup. Est. dup. % Diff. Type of partition
Japan 1 and 2 HK – Sing 37 24 54 1
1 and 2 Korea – Taiwan 38 20 90 1
1 and 2 France – UK 55 15 258 2
1 and 2 Hawaii – Guam 33 22 50 2
USA 1 Australia – New Zealand 25 14 80 2
1 France – Italy 42 30 40 2
1 Germany – France 35 25 240 2
1 China – Japan 23 12 92 2
1 China – France 15 25 240 3
UK 1 Austria – Germany – Italy – Switzerland 51 37 38 1
1 Spain, USA, Greece (share less with): France 67 87 220 3
1 Germany, Austria, USA, Switz, BLX (share less
with): Spain 58 78 225 3
Singapore 1 India – China 3 28 283* 3
1 Bali – Malaysia 43 18 240 1
1 Malaysia – India 35 10 142 3
China – Hong Kong 35 18 98 2
Notes: * Classi?cation of Partitions: 1 ¼ Partition between countries both or all close to origin country; 2 ¼ Partition between neighbouring
countries comparatively far from origin country but close to each other; 3 ¼ Unknown. All these partitions are statistically signi?cant at p
, 0.05; Actual dup. is the average of the observed duplications between the respective countries; Est. dup. is the model-generated
estimate of what the average duplications between those respective countries should be, given their respective popularity
VOL. 3 NO. 1 2009
j
INTERNATIONAL JOURNAL OF CULTURE, TOURISM AND HOSPITALITY RESEARCH
j
PAGE 47
D
o
w
n
l
o
a
d
e
d
b
y
P
O
N
D
I
C
H
E
R
R
Y
U
N
I
V
E
R
S
I
T
Y
A
t
2
2
:
0
6
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
compared to Singapore and Japan (r ¼ 0:79 and 0.53). These ?ndings support the
suggestion that competitive market structure in the tourism context may be altered to some
degree by the culture of the home market.
5. Summary and implications
This research examines four different international tourist origin markets: Singapore, Japan,
the USA, and the UK. The results show that the pattern of competition between tourist
destination brands is strongly in?uenced by overall market share, consistent with the DPL.
For markets where multi-year data is available, the duplication co-ef?cient parameters and
correlations between actual and estimated purchase duplications are stable over time. The
results show that destinations share travelers with other destinations approximately in-line
with the size of those other destinations. The study estimates the proportion of travelers to A
who also have traveled to B to be between 0.9 and 1.3, multiplied by the penetration (percent
ever traveled to) of B. This means the overall popularity of the destination is a primary
determinant of brand competition.
The major exceptions to this general pattern tend to comprise either:
B destinations that are an appreciable distance from the origin country but close to each
other; and
B destinations close geographically to the origin country
These partitions based on functional differences are broadly consistent with ?ndings in other
markets (e.g., Bennett et al., 2000; Carter and Silverman, 2004; Colombo et al., 2000; Grover
and Rao, 1988). Partitions that involve lowered levels of sharing between destinations are
identi?able. In one instance, cultural heterogeneity in the origin market may be a possible
explanation. These ?ndings might seem obvious in hindsight. However, no previous studies
identify them in the context of destination brands (note, an earlier version of this research
was presented, with considerably less detail, at the 2003 EMAC conference).
The ?ndings complement existing frameworks or approaches for understanding
competition. For example, substantial work examines the speci?c competitive factors that
drive tourist demand. This work includes physiography and tourism superstructure (Crouch
and Ritchie, 1999; Enright and Newton, 2005); as well as price levels and marketing
expenditure (e.g., O’Hagan and Harrison, 1984). Those factors drive overall market share –
a destination with more favorable levels of these factors will be more attractive to visitors.
This study shows that if a destination brand gains more market share, that brand will gain at
the expense of many other destination brands – in-line with their current overall popularity
levels. The extent of gain from speci?c competitor brands will be moderated by cultural
similarities and geographic proximity.
The results are also a very useful complement to analyses of competitive structure based on
brand positioning. The rationale for brand positioning is to create a perception of difference
that in turn drives preference (e.g., Kapferer, 1995, p. 95). Techniques such as
multi-dimensional scaling or correspondence analysis are often used to identify a brand’s
position (Carroll et al., 1986; Hoffman and Franke, 1986; Kennedy et al., 1996). Brands
closer in perceptual space are interpreted as closer substitutes, and are considered to be
more direct competitors to each other (e.g., Day et al., 1979, p. 14). The present study
highlights that similarity in brand positioning may not be the key factor that drives
competition. In contrast, brands tend to share more customers with larger brands (and those
in close geographic proximity), and these larger competitor brands may or may not be
positioned along similar lines. An emphasis on interpreting brand competition based on
positioning analysis could therefore lead tourism marketers to ignore major competitors
(larger share brands) in favor of brands that look similar in the eyes of consumers. The
?ndings of the present study suggest tourism marketers should pay more attention to other
larger destinations rather than other destinations that have similar brand positions. This shift
in outlook can increase marketing effectiveness and allocation of resources. For example,
interpreting competition based on Duplication of Purchase outputs could broaden the
de?nition of target market for a destination. Such a broadened targeting approach would
PAGE 48
j
INTERNATIONAL JOURNAL OF CULTURE, TOURISM AND HOSPITALITY RESEARCH
j
VOL. 3 NO. 1 2009
D
o
w
n
l
o
a
d
e
d
b
y
P
O
N
D
I
C
H
E
R
R
Y
U
N
I
V
E
R
S
I
T
Y
A
t
2
2
:
0
6
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
contrast to an approach based on brand positioning which would be to target travelers who
have been to other destinations that are perceptually positioned close to one’s own
destination. The present study suggests that the travelers to one’s own market are more likely
to have previously been to any other large/overall popular destination, not necessarily those
with brand perceptual positions close to one’s own. Likewise, the present study suggests
that the travelers who visit one’s own market are far more likely to then visit other large
destinations rather than destinations with similar brand image. The major exceptions are for
geographically proximate destinations, where there will be more sharing compared to
estimated levels based on the size effect in the DPL.
In terms of marketing opportunities, a destination can usually expect to share more tourists
with countries in the same vicinity than expected, given the overall popularity of the group of
destinations. Whether this sharing is viewed as a competitive threat, or a marketing
opportunity will depend on the country partitions and the relationship between destinations.
Heightened sharing also provides opportunities for neighbouring countries to collaborate on
tourism promotion ventures, particularly in regions with a number of small countries. For
example, Yong et al. (1989) discuss this approach in the context of Singapore, as an
outcome of a Delphi study investigating expert’s perceptions about what is needed for the
future development of that country’s tourism industry.
The present study applies the DPL to multiple market settings and multiple time periods,
which strengthens generalizability of the ?ndings. The duplication of purchase-style analysis
has an additional bene?t for the tourism research community, based on the fact that the law
was originally developed in the contexts of television viewing and packaged goods, then
extended to broader applications, and now to tourism. The approach of taking an empirical
regularity from one context to other contexts opens up opportunity to apply other
fundamental patterns in buyer behavior to tourism. For example, does the famous double
jeopardy pattern in brand loyalty (e.g., Ehrenberg et al., 1990) apply to tourismdestinations?
While the present study does ?nd some indication of market partitions that could be based
on cultural factors, further research is necessary to disentangle these from the primary
market share and location effects.
5.1 Limitations and future research
The key limitations of this research relate to the data. First, we cannot identify growth trends
for speci?c brands because data on visits were not provided in temporal order. This should
be an area for future research with data within a speci?ed time frame analysed, or repeat
behavior (last holiday and then holiday prior to that). Second, we cannot do cross-country
comparisons for all the individual destinations from each origin market. Further analysis with
larger samples and a larger number of destinations fromeach origin market would provide a
clearer picture of competitive structure. In addition, analysis using larger numbers of
destinations could better distinguish between culture and proximity effects by including
more sets of countries that are culturally similar but geographically dispersed (eg, Spain and
Argentina).
This study examines aggregate-level relationships, and these results need not hold for any
individual consumer, or consumer segment. However this study appears to be the ?rst
examination of cross-purchasing in a tourism destination context and an aggregate-level
analysis is appropriate. The study is able to tentatively, at least, indicate howheterogeneity in
the origin market can affect competitive market structure, in the case of lower levels of
sharing between India and China for Singaporean travelers. Subsequent studies could
examine heterogeneity in more detail by conducting purchase duplication analysis over
various consumer segments.
This research area could also progress by examining tourist travelers’ cross-purchases
according to whether they occurred on the same trip or on different trips, which would
distinguish between competitiveness and complementarity. For example, if travelers tend to
visit brands A and B on the same trip then these two brands might simultaneously compete
against, and also complement each other. The two brands compete for a share of the
VOL. 3 NO. 1 2009
j
INTERNATIONAL JOURNAL OF CULTURE, TOURISM AND HOSPITALITY RESEARCH
j
PAGE 49
D
o
w
n
l
o
a
d
e
d
b
y
P
O
N
D
I
C
H
E
R
R
Y
U
N
I
V
E
R
S
I
T
Y
A
t
2
2
:
0
6
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
traveler’s time and expenditure within the same trip, but could also complement each other
by together offering a more attractive overall destination than either one on its own.
References
Agresti, A. (2002), Categorical Data Analysis, 2nd ed., Wiley, Hoboken, NJ.
Armstrong, J.S. (2001), ‘‘Evaluating forecasting methods’’, in Armstrong, J.S. (Ed.), Principles of
Forecasting. A Handbook for Researchers and Practitioners, Kluwer Academic Publishers, Boston, MA.
Bass, F.M. (1974), ‘‘The theory of stochastic preference and brand switching’’, Journal of Marketing
Research, Vol. 11, February, pp. 1-20.
Bauer, H.H. and Herrmann, A. (1995), ‘‘Market demarcation: theoretical framework and results of an
empirical investigation of the German car market’’, European Journal of Marketing, Vol. 29 No. 11,
pp. 18-34.
Belau, D. (2003), ‘‘New threats to employment in the travel and tourism industry’’, International Labour
Organisation, Geneva.
Bennett, D. (2004), ‘‘Mapping the dynamic Chinese marketplace – the case of TV’s’’, ANZMAC
Conference, Wellington, Victoria University of Wellington, New Zealand.
Bennett, D. and Ehrenberg, A.S.C. (2001), ‘‘A lot can be revealed by a little data: two purchase analysis
of fast food buying’’, ANZMAC Conference, Massey University, Auckland, New Zealand.
Bennett, D., Ehrenberg, A. and Goodhardt, G. (2000), ‘‘Two purchase analysis of brand loyalty among
petrol buyers’’, ANZMAC Conference, Grif?th University, Gold Coast, Queensland.
Bennett, D.R. and Ehrenberg, A.S.C. (2002), ‘‘The way Australians buy fast food’’, Proceedings of the
European Marketing Academy Conference (EMAC 2002). Braga, Portugal, May.
Carroll, J.D., Green, P.E. and Schaffer, C.M. (1986), ‘ ‘Interpoint distance comparisons in
correspondence analysis’’, Journal of Marketing Research, Vol. 23, August, pp. 271-80.
Carter, J. and Silverman, F. (2004), ‘‘An empirical approach to market partitioning: application to the
cigarette market’’, Journal of Targeting, Measurement and Analysis for Marketing, Vol. 12 No. 4,
pp. 366-78.
Colombo, R., Ehrenberg, A. and Sabavala, D. (2000), ‘‘Diversity in analyzing brand-switching tables:
the car challenge’’, Canadian Journal of Marketing Research, Vol. 19, pp. 23-36.
Cooper, L.G. and Inoue, A. (1996), ‘‘Building market structures from consumer preferences’’, Journal of
Marketing Research, Vol. 33, August, pp. 239-306.
Crotts, J. (2004), ‘‘The effect of cultural distance on overseas travel behaviors’’, Journal of Travel
Research, Vol. 43, August, pp. 83-8.
Crouch, G.I. and Ritchie, J.R.B. (1999), ‘‘Tourism, competitiveness, and societal prosperity’’, Journal
of Business Research, Vol. 44, pp. 137-52.
Day, G.S., Shocker, A.D. and Srivastava, R.K. (1979), ‘‘Customer-oriented approaches to identifying
product-markets’’, Journal of Marketing, Vol. 43, Fall, pp. 8-19.
Ehrenberg, A.S.C. (1975a), Data Reduction, Wiley, London and New York, NY.
Ehrenberg, A.S.C. (1975b), ‘‘The structure of an industrial market: aviation fuel contracts’’, Industrial
Marketing Management, Vol. 4, pp. 273-85.
Ehrenberg, A.S.C. (2000), ‘‘Repeat-buying: facts, theory and applications’’, Journal of Empirical
Generalisations in Marketing Science, Vol. 5, pp. 392-770.
Ehrenberg, A.S.C. and Goodhardt, G.J. (1970), ‘‘A model of multi-brand Buying’’, Journal of Marketing
Research, Vol. 7, February, pp. 77-84.
Ehrenberg, A.S.C., Goodhardt, G. and Barwise, T.P. (1990), ‘‘Double Jeopardy revisited’’, Journal of
Marketing, Vol. 54, July, pp. 82-91.
Ehrenberg, A.S.C., Uncles, M.D. and Goodhardt, G.G. (2004), ‘‘Understanding brand performance
measures: using Dirichlet benchmarks’’, Journal of Business Research, Vol. 57 No. 12, pp. 1307-25.
PAGE 50
j
INTERNATIONAL JOURNAL OF CULTURE, TOURISM AND HOSPITALITY RESEARCH
j
VOL. 3 NO. 1 2009
D
o
w
n
l
o
a
d
e
d
b
y
P
O
N
D
I
C
H
E
R
R
Y
U
N
I
V
E
R
S
I
T
Y
A
t
2
2
:
0
6
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
Enright, M.J. and Newton, J. (2005), ‘‘Determinants of tourism destination competitiveness in Asia
Paci?c: comprehensiveness and universality’’, Journal of Travel Research, Vol. 43, pp. 339-50.
Goodhardt, G.J. and Ehrenberg, A.S.C. (1969), ‘‘Duplication of television viewing between and within
channels’’, Journal of Marketing Research, Vol. 6, May, pp. 169-78.
Grover, R. and Rao, V. (1988), ‘‘Inferring competitive market structure based on a model of interpurchase
intervals’’, International Journal of Research in Marketing, Vol. 5, pp. 55-72.
Hauser, J.R. and Wernerfelt, B. (1989), ‘‘The competitive implications of relevant-set/response analysis’’,
Journal of Marketing Research, Vol. 26, pp. 391-405.
Hoffman, D.L. and Franke, G.R. (1986), ‘‘Correspondence analysis: graphical representation of
categorical data in marketing research’’, Journal of Marketing Research, Vol. 23, August, pp. 213-27.
Hofstede, G. (1994), Cultures and Organizations: Intercultural Cooperation and its Importance for
Survival, Harper Collins Publishers, London.
Hotels (2002), ‘‘Hotels global update: ?nance’’, Reed Business, available at: www.hotelsmag.com
(accessed February 2008).
Kalwani, M.U. and Morrison, D.G. (1977), ‘‘A parsimonious description of the Hendry System’’,
Management Science, Vol. 23 No. 5, pp. 467-77.
Kapferer, J-N. (1995), Strategic Brand Management – New Approaches to Creating and Evaluating
Brand Equity, Kogan Page Limited, London.
Kennedy, R., Riquier, C. and Sharp, B. (1996), ‘‘Practical applications of correspondence analysis to
categorical data in market research’’, Journal of Targeting, Measurement and Analysis for Marketing,
Vol. 5 No. 1, pp. 56-70.
Kolland, F., Milian, J., Scheibenstock, M. and Schonbauer, H. (2000), ‘‘The short waves of the
product-cycle of tourism: Cuba’s tourism between de-coupling and integration in to the world market’’,
Management, Marketing and the Political Economy of Travel and Tourism, University of Northumbria,
Newcastle.
Kotler, P. and Gertner, D. (2002), ‘‘Country as brand, product and beyond: a place marketing and brand
management perspective’’, Journal of Brand Management, Vol. 9 Nos 4-5, pp. 249-61.
Law, J. (2004), ‘‘The long and the short of it’’, Travel Trade Gazette, Vol. 2, July.
Liu, Z. (2000), ‘‘The locational advantage of tourism destinations – determinants, dimensions and
dynamics’’, Marketing and the Political Economy of Travel and Tourism, University of Northumbria,
Newcastle.
Moutinho, L. (1987), ‘‘Consumer behavior in tourism’’, European Journal of Marketing, Vol. 21 No. 10,
pp. 5-44.
Novak, T.P. and Stangor, S. (1987), ‘‘Testing competitive market structures: an application of weighted
least squares methodology to brand switching data’’, Marketing Science, Vol. 6 No. 1.
O’Hagan, J.W. and Harrison, M.J. (1984), ‘‘Market shares of US tourist expenditure in Europe:
an econometric analysis’’, Applied Economics, Vol. 16, pp. 919-31.
Reisinger, Y. and Turner, L. (1999), ‘‘A cultural analysis of japanese tourists: challenges for tourism
marketers’’, European Journal of Marketing, Vol. 33 Nos 11/12, pp. 1203-27.
Romaniuk, J. and Dawes, J. (2005), ‘‘Loyalty to price tiers in purchases of bottled wine’’, Journal of
Product & Brand Management, Vol. 14 No. 1, pp. 57-64.
Russell, G.J. and Bolton, R.N. (1988), ‘‘Implications of market structure for elasticity structure’’, Journal
of Marketing Research, Vol. 25, August, pp. 229-41.
Schiffman, L.G. and Kanuk, L.L. (2004), Consumer Behavior, Pearson Prentice Hall, Englewood Cliffs,
NJ.
Sharp, B. and Sharp, A. (1997), ‘‘Positioning and partitioning’’, 26th European Marketing Academy
Conference, Warwick Business School, The University of Warwick, Coventry.
Singh, S. and Catlin, J. (2003), ‘‘Low-cost airlines ?ying high’’, Marketing Week, April, pp. 34-5.
VOL. 3 NO. 1 2009
j
INTERNATIONAL JOURNAL OF CULTURE, TOURISM AND HOSPITALITY RESEARCH
j
PAGE 51
D
o
w
n
l
o
a
d
e
d
b
y
P
O
N
D
I
C
H
E
R
R
Y
U
N
I
V
E
R
S
I
T
Y
A
t
2
2
:
0
6
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
Uncles, M.D. and Ehrenberg, A.S.C. (1990a), ‘‘The buying of packaged goods at US retail chains’’,
Journal of Retailing, Vol. 66 No. 3, pp. 278-96.
Uncles, M.D. and Ehrenberg, A.S.C. (1990b), ‘‘Industrial buying behavior: aviation fuel contracts’’,
International Journal of Research in Marketing, Vol. 7 No. 1, pp. 56-68.
Urban, G.L., Johnson, P.L. and Hauser, J.R. (1984), ‘‘Testing competitive market structures’’, Marketing
Science, Vol. 3 No. 2.
World Travel and Tourism Council (WITC) (2003), Blueprint for New Tourism, Vol. 2005, World Travel and
Tourism Council, London, available at, www.wttc.org/eng/About_WTTC/Blueprint_for_New_Tourism/
(accessed February 2008).
World Travel and Tourism Council (2007), Tourism Satellite Accounting, available at:
/www.wttc.org/blueprint/WTTCBlueprintFinal.pdf (accessed February 2008).
World Travel and Tourism Council (2008), ‘‘Continued growth signalled for travel and tourism industry’’,
March, available at: www.wttc.travel/eng/Tourism_News/Press_Releases/Press_Releases_2008
(accessed February 2008).
Yong, Y.W., Keng, K.H. and Leng, T.L. (1989), ‘‘A Delphi forecast for the Singapore tourism industry:
future scenario and marketing implications’’, European Journal of Marketing, Vol. 23, pp. 15-26.
Appendix 1
The DPL is summarized with the simple formula:
b
XY
= b
X
¼ Db
Y
(Ehrenberg and Goodhardt, 1970, p. 78).
In words, this formula states that the proportion of brand X buyers who also buy brand Y is
directly proportional to the penetration of brand Y (where b
x
and b
Y
are the penetrations of
brands X and Y respectively, b
xy
is the proportion of X buyers who buy Y, and D is the
average amount of cross-purchasing or purchase duplication across all pairs of brands in
the market).
The duplication coef?cient, as stated in the body of the paper is average of purchase
duplications divided by average penetration. Arithmetically, for four brands A to D it is:
ðb
AB
þ b
AC
þ b
AD
þ b
BA
þ b
BC
. . . þ b
DC
Þ = 12
*
ðb
A
þ b
B
þ b
C
þ b
D
Þ = 4
ð
*
For a 4 £ 4 brandmatrix less 4 entries on diagonalsÞ:
Where b
A
. . . .b
D
is the penetrations for the brands respectively and b
AB
, b
AC
, b
AD
. . .b
DC
is
the duplications for each pair of brands in the market respectively.
Appendix 2. Sample characteristics
USA – 55 percent male, 45 percent female. 70 percent were aged 46 or over. 55 percent
worked full time and 30 percent had an annual income before tax of over $100,000. 66
percent were married while 38 percent had children living in the household.
UK – 41 percent male and 59 percent female. 44 percent were aged between 26 and 45
years, with a further 21 percent aged between 46 and 55 years. 44 percent worked full time
and 21 percent were retired; 30 percent were A or B social class. 67 percent were married
and 62 percent had children living at home.
Singapore – 57 percent male and 43 percent female. 54 percent were aged between 25 and
46 years, but a further 33 percent were under 25 years. 32 percent worked full time while 37
percent worked part-time and 23 percent had an average income before tax of over
S$6,000. 49 percent were married and 67 percent reported having children living in the
home.
Japan – 24 percent male and 76 percent female; 55 percent were aged 25-44 years; 37
percent were in professional, of?ce or administrative work while 36 percent were
housewives; 20 percent earned over 9 million Yen before tax. 83 percent were married
and 68 percent had children living in the home. The larger proportion of females in the Japan
sample re?ected a heightened propensity to engage in travel or decide on travel on the part
of Japanese females.
PAGE 52
j
INTERNATIONAL JOURNAL OF CULTURE, TOURISM AND HOSPITALITY RESEARCH
j
VOL. 3 NO. 1 2009
D
o
w
n
l
o
a
d
e
d
b
y
P
O
N
D
I
C
H
E
R
R
Y
U
N
I
V
E
R
S
I
T
Y
A
t
2
2
:
0
6
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
Appendix 3. Method of assessment for the statistical signi?cance of partitions
Each respondent in this study could report traveling to two or more destinations. This
characteristic makes the data unsuitable to analyse using methods such as log-linear
analysis (e.g., Agresti, 2002). Therefore, the statistical signi?cance of partitions was
assessed using a permutational method. A Visual Basic program was constructed to
repeatedly draw random samples from an extremely large population. Each sample was of
the same size, number of brands, brand penetration and overall duplication levels, as each
of the respective surveys. The purchase duplications between speci?c brands were
recorded for each sample. Over a series of 10,000 iterations, this procedure revealed the
probability of observing brand duplications of a particular magnitude (e.g., deviations),
given the respective sizes of the brands. Brands that exhibited a deviation that was less than
5 percent likely to occur due to random sampling variation were judged to be statistically
signi?cant.
Corresponding author
John Dawes can be contacted at: [email protected]
VOL. 3 NO. 1 2009
j
INTERNATIONAL JOURNAL OF CULTURE, TOURISM AND HOSPITALITY RESEARCH
j
PAGE 53
To purchase reprints of this article please e-mail: [email protected]
Or visit our web site for further details: www.emeraldinsight.com/reprints
D
o
w
n
l
o
a
d
e
d
b
y
P
O
N
D
I
C
H
E
R
R
Y
U
N
I
V
E
R
S
I
T
Y
A
t
2
2
:
0
6
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
This article has been cited by:
1. Naomi Gruneklee, Sharyn Rundle-Thiele, Krzysztof Kubacki. 2016. What can social marketing learn from Dirichlet theory patterns in a
physical activity context?. Marketing Intelligence & Planning 34:1, 41-60. [Abstract] [Full Text] [PDF]
2. Margaret Faulkner, Oanh Truong, Jenni Romaniuk. 2014. Uncovering generalized patterns of brand competition in China. Journal of Product
& Brand Management 23:7, 554-571. [Abstract] [Full Text] [PDF]
D
o
w
n
l
o
a
d
e
d
b
y
P
O
N
D
I
C
H
E
R
R
Y
U
N
I
V
E
R
S
I
T
Y
A
t
2
2
:
0
6
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
doc_573225353.pdf
The purpose of this paper is to examine competition between tourism destination brands in
terms of how they share travelers with each other
International Journal of Culture, Tourism and Hospitality Research
Generalized pattern in competition among tourism destinations
J ohn Dawes J enni Romaniuk Annabel Mansfield
Article information:
To cite this document:
J ohn Dawes J enni Romaniuk Annabel Mansfield, (2009),"Generalized pattern in competition among tourism destinations", International
J ournal of Culture, Tourism and Hospitality Research, Vol. 3 Iss 1 pp. 33 - 53
Permanent link to this document:
http://dx.doi.org/10.1108/17506180910940333
Downloaded on: 24 January 2016, At: 22:06 (PT)
References: this document contains references to 52 other documents.
To copy this document: [email protected]
The fulltext of this document has been downloaded 1821 times since 2009*
Users who downloaded this article also downloaded:
Sameer Hosany, Yuksel Ekinci, Muzaffer Uysal, (2007),"Destination image and destination personality", International J ournal of Culture,
Tourism and Hospitality Research, Vol. 1 Iss 1 pp. 62-81 http://dx.doi.org/10.1108/17506180710729619
Graham Hankinson, (2005),"Destination brand images: a business tourism perspective", J ournal of Services Marketing, Vol. 19 Iss 1 pp.
24-32 http://dx.doi.org/10.1108/08876040510579361
Saila Saraniemi, (2011),"From destination image building to identity-based branding", International J ournal of Culture, Tourism and
Hospitality Research, Vol. 5 Iss 3 pp. 247-254 http://dx.doi.org/10.1108/17506181111156943
Access to this document was granted through an Emerald subscription provided by emerald-srm:115632 []
For Authors
If you would like to write for this, or any other Emerald publication, then please use our Emerald for Authors service
information about how to choose which publication to write for and submission guidelines are available for all. Please visit
www.emeraldinsight.com/authors for more information.
About Emerald www.emeraldinsight.com
Emerald is a global publisher linking research and practice to the benefit of society. The company manages a portfolio of more than
290 journals and over 2,350 books and book series volumes, as well as providing an extensive range of online products and additional
customer resources and services.
Emerald is both COUNTER 4 and TRANSFER compliant. The organization is a partner of the Committee on Publication Ethics (COPE)
and also works with Portico and the LOCKSS initiative for digital archive preservation.
*Related content and download information correct at time of download.
D
o
w
n
l
o
a
d
e
d
b
y
P
O
N
D
I
C
H
E
R
R
Y
U
N
I
V
E
R
S
I
T
Y
A
t
2
2
:
0
6
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
Generalized pattern in competition among
tourism destinations
John Dawes, Jenni Romaniuk and Annabel Mans?eld
Abstract
Purpose – The purpose of this paper is to examine competition between tourism destination brands in
terms of how they share travelers with each other.
Design/methodology/approach – The study analyzes survey data from four international markets
(USA, UK, Japan and Singapore). The study examines the cross-purchasing of travel destinations. It
applies an established empirical generalization, the duplication of purchase law (DPL) to frame
hypotheses and contextualize results.
Findings – The overall results are consistent with the DPL. Destination brands share tourists with other
destinations generally in-line with the popularity of the competing destination. However, there are very
noticeable market partitions, most of which take two forms: destinations that are either geographically
close to each other, or close to the point of origin. Destination brands in these partitions share travelers
far more than they would be expected to, given their respective size.
Practical implications – Tourism marketers need to appreciate the broad nature of competition. A
speci?c destination brand competes with many other travel destinations, sharing customers more with
other broadly popular destinations and less with less popular destinations.
Originality/value – The analytical approach presented in this study provides a straightforward
benchmark for assessing the expected level of competition between particular tourist destinations,
given their respective overall popularity.
Keywords Brands, Law, Competitive strategy, Tourism, Consumer behaviour
Paper type Research paper
1. Introduction
Tourism is a major, and increasing, source of revenue for countries all around the world. At
present, tourism accounts for over 4 percent of global GDP. Travel and tourism together
account for approximately 10 percent of world GDP (WTTC, 2003) and recent forecasts are
for strong growth (WITC, 2008). The industry earns in excess of US$1.3 billion per day, and is
the biggest export earner worldwide. Travel and tourism, together with suppliers to the
industry, employs over 200 million people (Belau, 2003). With such large economic
contributions, most countries have a tourism marketing body, responsible for the marketing
of travel destinations. The marketing managers within these bodies direct substantive
dollars towards tourism marketing and research, with the aim of better understanding their
market, their competitive positioning and howthey can use this knowledge to attract tourists.
A fundamental question for marketers of tourist destination brands relates to the competitive
structure of the market: which other brands does my brand compete against more intensely?
Which other brands does my brand gain sales from, and to which other brands does my
brand lose sales to? We apply a well-established method that has resulted in empirical
generalizations about competitive structure in numerous market settings, to address these
questions. The method is to analyze cross-purchases, namely the proportion of one brand’s
DOI 10.1108/17506180910940333 VOL. 3 NO. 1 2009, pp. 33-53, Q Emerald Group Publishing Limited, ISSN 1750-6182
j
INTERNATIONAL JOURNAL OF CULTURE, TOURISM AND HOSPITALITY RESEARCH
j
PAGE 33
John Dawes and
Jenni Romaniuk are based
at Ehrenberg Bass Institute
for Marketing Science,
University of South
Australia, Adelaide,
Australia.
Annabel Mans?eld is based
at the Division of Business,
University of South
Australia, Adelaide,
Australia.
Received May 2008
Revised September 2008
Accepted November 2008
D
o
w
n
l
o
a
d
e
d
b
y
P
O
N
D
I
C
H
E
R
R
Y
U
N
I
V
E
R
S
I
T
Y
A
t
2
2
:
0
6
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
customer base who also buy another brand. The application of cross-purchase analysis has
been very successful in identifying systematic patterns, leading to the development of the
duplication of purchase law(Ehrenberg, 2000). The duplication of purchase law(DPL) is that
brands generally share customers with other brands in line with market share. This
proposition means that the largest driver in estimating which competitors that any brand’s
customers are buying is the market share of the competitor brand. The DPL not only provides
this qualitative information, but also quanti?es the level of competition, as will be explained
later in the paper. Additionally the DPL allows researchers and marketers to identify and
quantify exceptions (referred to as partitions) to the general pattern. Partitions occur if
certain brands share more or fewer customers than would be expected. Analyzing brands
within partitions helps identify the key factors that in?uence competition levels and their
relative importance. Analysis using the DPL has been successfully applied in many different
market settings: media (Goodhardt and Ehrenberg, 1969); packaged goods markets
(Ehrenberg and Goodhardt, 1970); durables (Colombo et al., 2000) and services (Bennett
and Ehrenberg, 2001; Bennett and Ehrenberg, 2002). Applications of the DPL do not include
tourist contexts but the approach offers a promising framework to examine competition
among tourist destinations.
This research has three objectives. The ?rst is to quantify the level of competition in the
long-haul tourist market, by analyzing aggregate-level data on consumer’s
cross-purchasing of travel destinations. By the term aggregate, we mean that our unit of
analysis is the brand, not the individual traveler. The information on each brand is
aggregated up from individual-level consumer behavior pertinent to that brand. The extent
of competition between the brands is then analyzed. There is still obviously individual-level
heterogeneity among the consumers within a particular market, but the results represent the
aggregate probability of certain travel behaviors, among consumers from a market of
interest.
The second aim of this research is to identify the key in?uences on destination brand
competition. The third aim is to then quantify those in?uences on brand competition to
ascertain which are more important. In other words, to ?nd out what country-level factors
associate with heightened or lessened competition between tourist destination brands, over
and above those factors relating to brand size. To do this, the study analyzes consumer’s
cross-purchasing of destination brands across multiple origin countries, and over different
time periods.
2. Background to the research
Recently, three factors dramatically in?uence the competitive environment for tourist travel
destinations. First, the international tourism market exhibits very high rates of growth. The
number of total international tourist arrivals worldwide grew 50 percent between 1990 and
2000 (Hotels, 2002); with more modest annual growth of 3 percent forecast for 2008 (WITC,
2007). Total receipts from the industry were estimated at e207 Billion in 1990 and e512 Billion
in 2000 (Hotels, 2002). Second, the economic bene?ts of tourism are coaxing many once
closed countries (e.g., China, Cuba, and Taiwan) to open their borders in an attempt to
obtain tourist dollars. For example, visitor numbers to Cuba grew from 200,000 per year in
the 1980s to two million by the year 2000 (Kolland et al., 2000). The explosive growth in
popularity of such destinations means that for most destination brands, overall market
growth is tempered by heightened competition. Third, tourists venture further from their
home country because new low-cost entrants offer lower airfares (e.g., Singh and Catlin,
2003); as well as the popularity of package holidays, rising incomes, faster aircraft (Liu,
2000) and the enhanced infrastructure and marketing efforts of new competitors. For
example, the proportion of trips to Western Europe by UK tourists declined between 1987
and 1997 while the proportion to the rest of the world increased (Liu, 2000). This trend
appears likely to continue, with long-haul travel forecasted to grow at a greater rate than
international travel generally (Law, 2004).
For marketers, a widening competitor context is occurring because of the increase in
long-haul travel and more destinations being available. More people are traveling more
PAGE 34
j
INTERNATIONAL JOURNAL OF CULTURE, TOURISM AND HOSPITALITY RESEARCH
j
VOL. 3 NO. 1 2009
D
o
w
n
l
o
a
d
e
d
b
y
P
O
N
D
I
C
H
E
R
R
Y
U
N
I
V
E
R
S
I
T
Y
A
t
2
2
:
0
6
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
often, over longer distances, and to potentially more destinations. Therefore a framework for
understanding competitors and competition that can be applied in this dynamic
environment seems timely. The unit of analysis that is particularly useful is brand-level –
for example, which other brands does brand A compete with the most?
2.1 Conceptual framework/de?nitions
The present study uses the term market to mean international travelers. Brands describes
countries as destinations (e.g., Kotler and Gertner, 2002), such as Germany, the United
States or South Africa. The study examines countries as the unit of analysis, although tourist
destinations can also take the form of regions (e.g., Western Europe); or cities (e.g., Paris).
Thepurchasingof travel destinationsis analogoustobrandbuying. For example, inpackaged
goods markets buyers make repeated purchases from a category and buy different brands
over cumulative occasions. In the travel context, an individual can buy from the category
multiple times by taking multiple travel trips over time (e.g., a holiday every year over a ?ve
year time period). Each of those holidays may be to one or more destinations, opening up the
potential for multi-brand buying or repeat brand buying. Therefore destinations compete with
each other for customers, in the same way as brands do in other markets.
The present study uses the termcompetition in the context of sharing customers. If a traveler
travels to destination A and also B, then A and B compete. Competition can be manifest in
several ways. The ?rst is competing for an allocation of the traveler’s time during a particular
trip – akin to their share of wallet during the trip. The second is competing for the traveler’s
choice over consecutive trips. For example, of the travelers who went to France on their last
trip, how many go again next time? This is competition for the traveler’s repeat-purchase.
Many ways exist to exhibit competition among brands but the present study uses the sharing
of customer’s travel destinations its focus.
2.2 Competitive market structure and duplication of purchase
An appreciation of how buyers of one brand also buy competitor brands over time is a key
aspect of understanding brand competition. Brands compete against other brands for the
purchases of consumers each time those consumers buy from the category. A particular
brand A can gain sales from competitors B, C, and D; and can lose sales to these
competitors. Which brand does the focal brand lose sales to, and fromwhich brand does the
focal brand gain sales from? Academic researchers have examined this issue for many
years, using a variety of methods. One method that is particularly appealing is to examine
howcustomers of each brand allocate their purchases over multiple occasions. This method
is readily understandable and poses minimal data requirements other than purchase
histories. As an example, we can consider this question: of the buyers of brand A, how many
also buy brand B, C, or D over a time period. If many buyers of A also buy B, but few of them
buy Dthen A competes more heavily with B than D. This type of approach to analysing brand
competition has been used in many applications, and with many nuances – but an
overriding theme is that competitor brand’s market share is a primary factor in estimating
how many brand buyers also buy those competitor brands. The duplication of purchase law
succinctly summarises this phenomena: brands share customers with other brands in-line
with the market share of those other brands (Ehrenberg, 2000, Chapter 9). We review
relevant literature on the DPL in the next section. Note that the termswitching is often used to
denote customers buying A then B. We use the terms switching, sharing and purchase
duplication interchangeably to communicate the same phenomenon of customers buying
multiple brands.
The earliest published study using the purchase duplication approach is by Goodhardt and
Ehrenberg (1969) that examines the extent to which viewers of certain television shows also
viewed other shows. Goodhardt and Ehrenberg (1969) report a consistent pattern: the
proportion of viewers of any television show A who also watched show B, C, D etc. was
predictable simply from the ratings of B, C, or D. Television viewing appeared to have a
strong stochastic component, with the market share (in this case, ratings) of the program
strongly in?uencing how many viewers switched to that program. Bass (1974, p. 29)
VOL. 3 NO. 1 2009
j
INTERNATIONAL JOURNAL OF CULTURE, TOURISM AND HOSPITALITY RESEARCH
j
PAGE 35
D
o
w
n
l
o
a
d
e
d
b
y
P
O
N
D
I
C
H
E
R
R
Y
U
N
I
V
E
R
S
I
T
Y
A
t
2
2
:
0
6
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
develops the theme of a pronounced stochastic component in brand switching and showed
that consumer switching between soft-drink brands was substantially in line with the market
share of those brands. Ehrenberg (2000) more fully expounds the phenomena in the
textbook Repeat Buying. The arithmetic formula is set out in Appendix 1.
Duplication of purchase-style analysis has been used in numerous applications from
frequently bought goods such as groceries (e.g., Bass, 1974; Ehrenberg and Goodhardt,
1970) and bottled wine (Romaniuk and Dawes, 2005). It has also been used in markets with
less frequent purchase patterns, such as electrical goods (Bennett, 2004) and motor
vehicles (Colombo et al., 2000). Service contexts have also been examined: quick service
restaurants (Bennett and Ehrenberg, 2002) retail store choice by consumers (Uncles and
Ehrenberg, 1990a) and mixed service/goods provision in a B-to-B setting (Ehrenberg,
1975b; Uncles and Ehrenberg, 1990b). The basic premise of the duplication of purchase law
also appears in the literature under other names. One name is the Aggregate Constant Ratio
Model which states ‘‘consumers select alternative products proportional to the market
shares of the alternatives’’ (Novak and Stangor, 1987, p. 84; see also Russell and Bolton,
1988, p. 230). The Aggregate Constant Ratio Model is used in several studies examining
competitive market structure (Bauer and Herrmann, 1995; Cooper and Inoue, 1996; Urban
et al., 1984). The term probabilistic independence (Hauser and Wernerfelt, 1989) is also
used to communicate the same idea as the DPL.
As mentioned above, tourist destination marketers should understand the competitive
structure of their market: of the travelers who visit us, what other destinations do they visit? Of
the travelers who visit rival destinations A, B, C, and so on, how many from each come to us?
These questions are explicitly considered using the duplication of purchase method of
analysis. Moreover, the DPL provides a useful framework and set of expectations within
which the results can be compared. If tourist destination brands follow the DPL this provides
a straightforward interpretation of the structure of competition between tourist brands. If
there are noticeable departures from the general pattern, those departures indicate speci?c
competitor destinations the marketing manager must pay close attention to.
One speci?c characteristic of travel destination purchasing occurs that differs from
purchasing in the types of markets analysed thus far using the purchase duplication
approach. That is, long-haul travel may encompass purchasing or visiting multiple
destinations on the same trip. By contrast, consumers in most other markets purchase tend
to purchase only one brand in a category at a time. This characteristic presents one
particular limitation to the study, namely that we cannot distinguish between partitions
formed from destinations that tend to be visited as part of a round trip, as opposed to
partitions based on consecutive purchases. However, as stated above, competition
between brands can be manifest as competition for repeat-purchase over consecutive trips,
or as competition for time spent during a particular trip (share-of-wallet). Therefore partitions
based on round trips can still represent heightened competition between particular brands
for the traveler’s spending. Also, partitions formed either from round trips or consecutive
purchases have a similar interpretation in relation to other brands outside the partition:
lowered competitive substitutability. Consider the scenario that destinations A and B tend to
be purchased together as part of a round trip, and C and D also tend to be purchased
together as a round trip. Purchase duplication analysis will indicate that these respective
pairs (A and B v. C and D) form two partitions and each partition shares fewer travelers with
the other partition than expected. Therefore the extent of competitive intensity across the
partitions is lower. This is the same interpretation that would arise if the combination of A and
B; or C and D, tended to be purchased together over consecutive trips. A round trip in which
the same destination is visited multiple times does not cause any bias in the present study
because we count the numbers of travelers to each destination, and do not consider how
many times they travel to that destination.
2.3 Outputs from purchase duplication analysis
This article uses the term purchase duplication analysis to mean an examination of the
purchases of pairs of competitive brands – how many people bought A and B; A and C, etc.
PAGE 36
j
INTERNATIONAL JOURNAL OF CULTURE, TOURISM AND HOSPITALITY RESEARCH
j
VOL. 3 NO. 1 2009
D
o
w
n
l
o
a
d
e
d
b
y
P
O
N
D
I
C
H
E
R
R
Y
U
N
I
V
E
R
S
I
T
Y
A
t
2
2
:
0
6
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
The DPL provides a framework for interpreting the results of purchase duplication analysis.
The different elements of the analysis each provide speci?c insights. We now discuss these
three elements in the context of destinations as brands:
2.3.1 Brand penetrations. How many travelers from the origin market have visited this brand
at least once. For example, how many British international tourists have visited France. This
example ?gure would represent the penetration ?gure for France, for the British traveler
market.
2.3.2 Brand duplications. How many people who traveled to brand A have also traveled to
brand B. The average duplication rate for each brand is derived from these ?gures. In other
words, what is the average proportion of travelers to any particular brand who also went to
brand B? According to the DPL, the proportion of people who traveled to destination A who
also traveled to B should be the same as the proportion of people who had traveled to C, D
etc. who then also traveled to B. Figure 1 shows an example of the way brand duplications
are usually represented.
2.3.3 Duplication co-ef?cient. This statistic re?ects the total amount of purchase duplication
in the market, allowing for the sizes of the brands. The duplication co-ef?cient is calculated
as the average of all purchase duplications divided by the average of all brand penetrations,
and is often denoted as D (Ehrenberg, 2000) – see Appendix 1 for details. The duplication
co-ef?cient provides the basis for calculating expected duplication levels for each brand.
The expected duplication between any pair of brands A and B is the penetration of B
multiplied by D. For example, if we calculate the duplication co-ef?cient for a market to be
1.5, this means the expected proportion of travelers to one brand that also travel to another
brand is 1.5 times the penetration of that other brand (as per Ehrenberg, 2000, Ch. 9). These
model-generated, expected duplications are useful because marked exceptions to them
indicate partitions in the market that can be further investigated.
2.3.4 Partitions. If a certain pair or group of brands exhibits quite different levels of purchase
duplication to their expected duplication values, this indicates heightened competitive
intensity between those brands. Heightened competitive intensity is called a partition
Figure 1 Portrayal of purchase duplication analysis and the DPL
VOL. 3 NO. 1 2009
j
INTERNATIONAL JOURNAL OF CULTURE, TOURISM AND HOSPITALITY RESEARCH
j
PAGE 37
D
o
w
n
l
o
a
d
e
d
b
y
P
O
N
D
I
C
H
E
R
R
Y
U
N
I
V
E
R
S
I
T
Y
A
t
2
2
:
0
6
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
(e.g., Kalwani and Morrison, 1977). Conversely, lower than expected sharing indicates the
brands do not compete much with each other at all, allowing for their size. The relative level
of competitive intensity can be quanti?ed. For example, suppose the duplication of
purchase analysis ?nds the expected proportion of people who have traveled to any
particular brand that also travel to brand B is 30 percent. But for a particular brand A, 60
percent of people who traveled to A also travel to B. This is interpretable as B having twice
the level of competitive substitutability with A as it should. Once partitions are identi?ed as
statistically signi?cant, and numerically large enough to be managerially signi?cant, the
possible reasons for the partition can be examined (e.g., location versus cultural reasons).
The marketer can then identify the dominant reasons for market partitions.
2.4 In?uences on brand competition
The previous section outlines the concept of partitions – pairs or clusters of brands that
compete more intensely, or less intensely, than expected. There are potentially many
characteristics of destination brands that could in?uence the formation of partitions. We
examine prior literature on market partitions in the next section to develop hypotheses. We
reiterate that we are constructing an aggregate-level model using the brand as the unit of
analysis and are not considering individual-level consumer variables such as perceptions,
perceived value, or bene?ts sought on the part of a speci?c traveler.
2.4.1 Functional differences. Market partitions are mostly attributable to functional
differences between certain groups of brands (e.g., Ehrenberg et al., 2004). For example,
coffee is partitioned into decaffeinated and regular, as well as ground and instant. The
functional differences between the coffee types lead to people buying brands within each of
the sub-categories. This means that instant coffee drinkers are more likely to buy other
brands of instant coffee over other brands of ground coffee (Grover and Rao, 1988).
A further example of market structure based on functional differences is from the car
industry: analysis of British and French car markets revealed that they were partitioned into
four functionally different groups: Luxury makes, Japanese makes, Other European makes
and Large Local makes. Cars within each group shared more customers with other brands in
the same group than would have been expected given their market share (see Colombo
et al., 2000). Partitioning based on functional differences is also identi?ed in the petrol
market, between leaded and unleaded petrol (Bennett et al., 2000) and in cigarettes
between ?ltered/non ?ltered and menthol/non menthol (Carter and Silverman, 2004).
Culture is a functional difference among tourism destinations that is often discussed. Culture
is de?ned as the ‘‘sum total of learned beliefs, values and customs that serve to direct the
consumer behavior. . .of a society’’ (Schiffman and Kanuk, 2004, p. 405). Consumers in
different countries exhibit marked differences in core values (Hofstede, 1994); and
numerous studies relate core values to consumer behavior differences. In the context of
competing destinations, many components of culture may in?uence destination choice,
therefore play a part in the competitive structure of the market (Moutinho, 1987). For
example, tourists could use the perceived cultural similarity of a destination as a decision
variable. Culture may also impact on the way tourists from different destinations select their
own holiday destinations. For example, there is a frequently discussed difference between
Japanese and Western cultures (Reisinger and Turner, 1999).
The impact of culture on the decision process may result in a departure from the DPL in the
tourism travel context. This outcome may occur in two ways. First, the DPL might simply not
hold due to cultural differences in destination choice. This will occur if we see the overall
pattern of brand sharing in-line with size holds for some markets and not others – for
example, if the pattern holds for UK and the USA, but not Japan and Singapore or vice versa.
Second, cultural differences could form the basis for market partitions. This outcome is
evident if destinations of similar culture share more customers than expected. For example, if
international tourists from mainland China exhibit a greater likelihood than expected to visit
another country such as Hong Kong or Taiwan. A limitation exists in assuming cultural
homogeneity within an origin country. The DPL posits aggregate-level relationships,
PAGE 38
j
INTERNATIONAL JOURNAL OF CULTURE, TOURISM AND HOSPITALITY RESEARCH
j
VOL. 3 NO. 1 2009
D
o
w
n
l
o
a
d
e
d
b
y
P
O
N
D
I
C
H
E
R
R
Y
U
N
I
V
E
R
S
I
T
Y
A
t
2
2
:
0
6
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
essentially the same as re?ecting the probability of behavior for a randomly drawn consumer
in the market of interest.
2.4.2 Geographic proximity. Geographic location or proximity can also underlie market
partitioning. For example, Sharp and Sharp (1997) examine the competitive structure of
retail department stores, and identify a partition between geographically close stores,
despite them having very different brand images. Since the location of a country (and its
neighbours) is a major, unchangeable characteristic, market partitions based on proximity
could occur. Two ways exist in which proximity could exert an in?uence on competitive
structure. First, destinations that are close to the home country (e.g., USA with Canada and
Mexico) could share more tourists than expected due to their closeness and accessibility.
Travelers to one are more likely to then travel to the other over time. This outcome is
consistent with the well-established phenomena of distance decay in air travel patterns
(Crotts, 2004). Second, destinations close to each other (regardless of distance from the
home destination) may be more likely to share visitors as there is a great chance they will
visited by the same tourist on the same trip (e.g., Australia and New Zealand from the
perspective of UK tourists). In this instance the two destinations are competing for share of
holiday time within a single trip.
The propositions to this point are as follows: An understanding of aggregate-level patterns in
competition among destination brands is desirable. A well-established method and
analytical framework from other markets exists for this purpose. Given the widespread
applicability of the DPL across numerous markets, we hypothesize the law will hold overall,
for tourist destinations. Cultural similarity or geographic proximity could give rise to
deviations from the overall pattern. These deviations from expected levels of
cross-purchasing or purchase duplication may be manifest as partitions or clusters,
identi?able as destination brands that share customers to a greater (or lesser) extent with
each other than with the rest of the market. The outcome of the investigation is to identify an
overall pattern of competition among destination brands and then to identify any exceptions
to the general pattern, along with possible reasons for those exceptions.
3. The study
The analysis for the study has three stages. Stage 1 identi?es if the DPL holds overall, and
whether the law is applicable over different countries. To do this we examine four different
international markets, namely the United States, United Kingdom, Singapore and Japan.
The term ‘‘market’’ means the pool of international travelers who reside in the respective
country. Over the four markets, there are marked differences in cultural values. This
assertion follows from two observations. The ?rst and obvious one is that two are Western
countries and two are Eastern. Of the two Eastern countries, Japan has a large population
and is one of the largest economies in the world. The other, Singapore, is a recently
established island state with a very vibrant trading economy and a small population. The
second observation is that there are marked cultural differences across the four countries,
based on the country-level scores for Hoftstede’s measures of cultural distance (for a
description of each facet see Hofstede, 1994).
The USA and the UK are quite similar culturally, and therefore we expect similarities in their
results – but still, replication across culturally similar markets provides con?dence that the
results are not just one-off ?ndings. In contrast to the UK and USA, the other two markets,
namely Singapore and Japan, are very different cultures to the other two countries – and to
each other. Therefore, their inclusion extends the potential for generalizable results (see
Table I).
Stage 2 involves close replications to see if the same pattern emerges two years later.
Examining results froma different sample of tourist consumers for two of the markets, namely
USA and Japan, accomplishes Stage 2. Stage 3 focuses on identifying and interpreting
deviations from the general pattern. These deviations indicate possible market partitions.
The scope of this analysis (four markets, and two time periods) offers credible scope for
identifying a potential empirical generalization in competitive structure among long-haul
destination brands.
VOL. 3 NO. 1 2009
j
INTERNATIONAL JOURNAL OF CULTURE, TOURISM AND HOSPITALITY RESEARCH
j
PAGE 39
D
o
w
n
l
o
a
d
e
d
b
y
P
O
N
D
I
C
H
E
R
R
Y
U
N
I
V
E
R
S
I
T
Y
A
t
2
2
:
0
6
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
3.1 Survey data
The data used for this analysis were collected as part of an ongoing brand tracking research
project for an international tourism body. An accredited international market research
provider collected the data. The recruitment procedure followed normal market research
industry quality standards. International travelers were the population of interest.
Respondents were randomly recruited via telephone, which means each individual
respondent was recruited individually from a sampling frame and all in the sampling frame
had an equal opportunity of being part of the survey.
Once contacted, respondents were then screened for suitability via survey questions. Only
those respondents who stated some probability of traveling outside of their home region
within four years of the date of the interview were retained in the sample. Therefore the
sample characteristics re?ect those likely to be engaged in long haul overseas travel, rather
than simply re?ecting the total adult population of each country. The demographic
breakdown the sample from each country is shown in Appendix 2.
As part of a larger questionnaire into brand and advertising awareness and brand
perceptions, respondents were asked, which countries or destinations outside , home
country . have you ever visited for a vacation/holiday or sightseeing? Only people who had
traveled to more than one destination were included in the analysis, because the analysis
examines cross-purchases of multiple brands. The exclusion of single-destination travelers
does not compromise the analysis. It only has the effect of lowering the penetration and
duplication levels for all brands. The strength of relationship between penetration and
purchase duplication, and the extent of partitioning, are not affected by excluding
single-destination travelers. Prior studies in contexts such as automobile purchasing
necessarily exclude single-purchase buyers (Colombo et al., 2000). Some destinations
might exhibit particular appeal among one-time-only travelers, and therefore enjoy a
competitive advantage that would not be apparent from this analysis. Prior research in other
markets indicates that infrequent buyers tend to unduly favor larger brands (e.g., Ehrenberg
et al., 2004).
This procedure provides analysis samples ranging for 41 percent for Japan, which had the
largest number of never traveled or single destination travelers, to 74 percent for Singapore,
which had the most multiple destination travelers. For each of the four markets, we analyze
the travel patterns of residents of that country by examining howmany respondents fromthat
country visit certain other countries, and of the travelers to country A, what proportion also
travel to country B, C, D and so on. A summary of the data is as follows (see Table II).
The study includes the creation of purchase duplication tables for each country-market. A
visual inspection of the purchase duplication tables will give an initial indication if the DPL
holds in this context. According to the DPL, the average proportion of travelers to any
country who also visit any other country should decline in line with the penetration of that
other country. Explication of this pattern is possible by simply ordering the brands in
descending size (popularity) by row and by column. Ordering data by size is very useful
procedure to show patterns in data (Ehrenberg, 1975a). Figure 1 shows this format and the
expected pattern. Due to space limitations we can only show one complete purchase
duplication table, but we show the results for all four markets in a series of graphs.
Table I Countries classi?ed according to Hoftstede’s measures of cultural difference
USA UK Singapore Japan
(score/100) (score/100) (score/100) (score/100)
Power – distance 40 35 74 54
Individualism 91 89 20 46
Masculinity 62 66 48 95
Uncertainty – avoidance 46 35 8 92
Long term orientation 29 25 48 80
Notes: High scores for these cultural aspects are in italics for clarity
PAGE 40
j
INTERNATIONAL JOURNAL OF CULTURE, TOURISM AND HOSPITALITY RESEARCH
j
VOL. 3 NO. 1 2009
D
o
w
n
l
o
a
d
e
d
b
y
P
O
N
D
I
C
H
E
R
R
Y
U
N
I
V
E
R
S
I
T
Y
A
t
2
2
:
0
6
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
Comparing the expected purchase duplication levels deriving from the duplication
co-ef?cient, to the actual levels serves to assess the overall level of model ?t. The
appropriateness of the DPL to the data is examined using three different measures. The ?rst
?t measure is mean absolute deviation (MAD) between the actual duplication, and the
model-generated estimates of duplication derived fromthe duplication co-ef?cient. MADis a
widely used ?t measure (e.g., Armstrong, 2001). For the UK market as an example, we take
the actual duplication for the proportion of those who traveled to France who also travel to
Germany, and compare this to the model-generated estimate. We repeat this comparison
over each combination of brands and calculate the mean absolute deviation between actual
and estimated ?gures, over all brands. Note that we use the MADsimply to communicate the
general extent to which the model provides estimates that match the empirical results. We
do not apply inferential tests to the MAD’s.
The second measure of ?t is the correlation between the model-generated purchase
duplications and the average purchase duplications for each brand. If a high correlation
exists between the model-based estimates and the actual data, the DPL represents
aggregate-level brand competition very well. Finally, we assess the statistical signi?cance of
market partitions using a permutational method outlined in Appendix 3.
The use of multiple ?t measures is useful, because any particular measure has a
shortcoming. The correlation co-ef?cient assesses the extent to which the two variables vary
together, but does not indicate if matched observations for two variables are actually
numerically close to each other. The MAD assesses the extent to which the values are
numerically close to each other, but does not provide a direct interpretation of the level of
co-variability or the statistical signi?cance of partitions. The statistical signi?cance of
particular brands that deviate from expected duplication levels is useful to properly identify
partitions, but does not necessarily inform us of the overall extent to which the DPL applies.
4. Results
4.1 Fit of the DPL
Consider a detailing of the results for the UK market, that is, the destinations of travelers from
the UK. The analysis focuses on the largest ten destinations. The destination brands are
ordered by size in both columns and rows. Interpret Table III as follows. Under the
penetration (Pen percent) column, we see 65 percent of international travelers from the UK
have traveled to France, 60 percent to Spain and so on. Looking across the rowfor France, of
the people who have traveled to France, 63 percent of those have also traveled to Spain, 36
percent also to the USA and so on to 19 percent having also traveled to Portugal. Overall, the
results show a strong in?uence of brand penetration on duplications, consistent with the
DPL. Of the people who have traveled to any destination in this list, more of them have also
traveled to France and Spain, the two largest destinations. Far fewer have also traveled to
Switzerland or Portugal, the two relatively less popular destinations for this market. That said,
we see variation in the individual cells down each column, which indicates market partitions.
Table II Summary of data
Analysis sample size
n ¼
Percent of
total sample
Sample of international travelers residing in. . .
UK 822 62
USA 434 43
Singapore 714 74
Japan 278 41
Additional samples to test generalizability of patterns over time
USA 491 47
Japan 318 53
VOL. 3 NO. 1 2009
j
INTERNATIONAL JOURNAL OF CULTURE, TOURISM AND HOSPITALITY RESEARCH
j
PAGE 41
D
o
w
n
l
o
a
d
e
d
b
y
P
O
N
D
I
C
H
E
R
R
Y
U
N
I
V
E
R
S
I
T
Y
A
t
2
2
:
0
6
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
We highlight statistically signi?cant deviations with an asterisk, and these partitions are
discussed shortly.
To more clearly show the purchase duplication pattern, the ?ndings include the results for
the UK, and the other three markets, in scatterplots (Figures 2–5). The X-axis represent the
penetration levels for each of the ten brands, and the Y-axis represent the proportion of
buyers of any destination brand who also traveled to the X-axis brand. We see visual
evidence of a strong relationship between penetration and brand duplication.
The analyses are for all four markets using the same method. Tables IV, V, VI and VII shows
the summary statistics. The ?ndings include a strong effect of brand size, consistent with the
DPL. The correlations between actual and predicted average sharing of tourists are 0.98,
0.68, 0.98, and 0.94 for the UK, USA, Singapore and Japan markets respectively. The mean
absolute deviations between observed and model-generated estimates of sharing range
Table III Duplication of purchase: results for UK market Wave 1 (top ten destinations)
% UK travelers who % Who also traveled to . . .
traveled to Pen % France Spain USA Italy BNL Greece Germany Austria Switzerland Portugal
France 65 63* 36 42 41 33 41 26 25 19
Spain 60 67* 35 38 33 39 30 19 19 20
USA 35 68* 61* 45 34 37 32 24 23 21
Italy 33 82 69 47 47 45 46* 36* 35* 23
BNL
a
32 83 63* 37 50 36 52 33 28 24
Greece 31 68* 75 41 48 37 35 27 25 23
Germany 31 85 59* 36 49 53 35 40* 33 21
Austria 20 84 58* 42 60* 51 41 61* 43* 25
Switzerland 18 87 63* 43 63* 50 43 56* 47* 24
Portugal 15 80 81 50 50 51 47* 43 33 29
Model-estimated
duplication (Pen £ 1.3) 87 81 46 45 42 42 42 27 25 20
Average duplication 78 64 43 49 45 40 45 32 30 23
Duplication coef?cient 1.3
MAD (estimated to actual
duplication) average ¼ 5 9 17 3 5 3 2 3 5 5 3
Notes:
a
¼ Benelux countries; * Indicates statistically signi?cantly different to expected level at p , 0.05
Figure 2 Scatterplot of brand penetrations and duplications (UK market)
PAGE 42
j
INTERNATIONAL JOURNAL OF CULTURE, TOURISM AND HOSPITALITY RESEARCH
j
VOL. 3 NO. 1 2009
D
o
w
n
l
o
a
d
e
d
b
y
P
O
N
D
I
C
H
E
R
R
Y
U
N
I
V
E
R
S
I
T
Y
A
t
2
2
:
0
6
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
from 1.0 to 8.2 percentage points. The purchase duplication coef?cients range from 0.9 to
1.3. This co-ef?cient, when multiplied by the penetration of each country estimates how
many, on average of the people who have traveled to any of a set of countries (A, B, C, etc.)
will have also traveled to country D. There is no inferential test for the purchase duplication
coef?cient, but the coef?cient produces estimates that correlate highly with the average
duplication levels for each brand at 0.68 to 0.98 (all p , 0.03 or lower). For the USA, UK and
Singapore the coef?cient is around 1.3, while for Japan the coef?cient is around 1. This
suggests Japanese tourists travel to a smaller range of destinations than USA, UK and
Singaporean Tourists. This result might be due to greater loyalty to single destinations or due
to taking fewer holidays overall.
Figure 3 Scatterplot of brand penetrations and duplications (USA market)
Figure 4 Scatterplot of brand penetrations and duplications (Singapore market)
VOL. 3 NO. 1 2009
j
INTERNATIONAL JOURNAL OF CULTURE, TOURISM AND HOSPITALITY RESEARCH
j
PAGE 43
D
o
w
n
l
o
a
d
e
d
b
y
P
O
N
D
I
C
H
E
R
R
Y
U
N
I
V
E
R
S
I
T
Y
A
t
2
2
:
0
6
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
We conduct the same analysis for USA and Japan data gathered two years later, to test the
over-time stability of these patterns. The sampling regime and questioning techniques were
the same, but the individual respondents were different. The results (in Table VIII) show the
overall pattern holds in both countries on a different set of data, with similar duplication
Figure 5 Scatterplot of brand penetrations and duplications (Japan market)
Table V Duplication of purchase (top ten destinations in each market)
USA Wave 1 Mex Fra Car Ger Can UK Ita Swi Aus Haw Average MAD
Correlation
(Act. v Est.)
Average duplication 47 47 27 45 20 34 39 32 26 16
Model-estimated duplication
(Pen £ 1.3) 61 39 39 37 33 33 31 23 19 18
MAD (estimated to actual duplication) 14 8 12 7 13 1 8 9 8 2
Duplication coef?cient 1.3 8.2 0.68
Notes: MAD ¼ Mean Absolute Deviation across all pairs of duplicated destinations; Act: ¼ actual duplications; Est: ¼ model-estimated
duplications
Table IV Duplication of purchase (top ten destinations in each market)
UK Wave 2 Fr Sp USA Ita BNL Gre Ger Aus Swi Por Average MAD
Correlation
(Act. v Est.)
Average duplication 78 64 43 49 45 40 45 32 30 23
Model- estimated duplication
(Pen £ 1.3) 87 81 46 45 42 42 42 27 25 20
MAD (estimated to actual duplication) 9 17 3 5 3 2 3 5 5 3
Duplication coef?cient 1.3 5.0 0.98
Notes: MAD ¼ Mean Absolute Deviation across all pairs of duplicated destinations; Act: ¼ actual duplications; Est: ¼ model-estimated
duplications
PAGE 44
j
INTERNATIONAL JOURNAL OF CULTURE, TOURISM AND HOSPITALITY RESEARCH
j
VOL. 3 NO. 1 2009
D
o
w
n
l
o
a
d
e
d
b
y
P
O
N
D
I
C
H
E
R
R
Y
U
N
I
V
E
R
S
I
T
Y
A
t
2
2
:
0
6
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
coef?cients in both years for both markets and correlations between estimated and
predicted average sharing ?gures above 0.93.
4.1.1 A systematic deviation. A systematic deviation occurs in the general pattern of results.
The model-estimated duplications are markedly over-estimated for high-penetration
destinations. The greatest observed deviations occur for brands that have penetrations
higher than 50 percent. Although this seems signi?cant, the overestimation of predicted
duplications is a known ?aw of the duplication of purchase model, known to generally occur
for brands with penetrations greater than 50 percent (Ehrenberg, 2000; Sharp and Sharp,
1997). The fact the same ?aw occurs in this study provides additional con?dence that the
same underlying patterns found in other types of markets also hold for tourism destinations.
4.2 Market partitions
Brands can share more (or fewer) customers than expected given their respective size.
From a competition perspective, this means that certain brands compete more intensely or
less intensely with each other – so-called partitions (Kalwani and Morrison, 1977). Partitions
have important strategic implications. They can re?ect emerging sub-categories that break
away to form new markets and present opportunities for new brand launches or positioning
directions. Furthermore, if a marketer is managing a partitioned brand, they need to pay
special attention to the other brands in the partition, because the marketing activity of those
competitors can have more impact than other brands of the same size. Therefore knowing
about, and understanding market partitions can increase marketing effectiveness by
streamlining resources towards the most important competitive issues.
This research identi?es partitions by identifying which individual countries deviate markedly
from the expected duplication ?gure. Referring back to Table III for illustration, the study
identi?es a partition by subtracting the estimated duplication level for Austria (27 percent)
from the actual duplication ?gure for travelers to Switzerland who also travel to Austria (47
percent), which is a deviation of 20. In other words, UK travelers who have been to
Switzerland are 20 percentage points more likely than expected to also have traveled to
Austria – given the overall popularity of Austria and the general extent to which destinations
share customers in this market. This particular partition is statistically signi?cant at p , 0.05.
Table VI Duplication of purchase (top ten destinations in each market)
Singapore Wave 1 Mal Tha Bal Aus HK Chi Tai USA Jap Kor Average MAD
Correlation
(Act. v Est.)
Average duplication 82 59 55 39 47 33 24 23 25 14
Model-estimated duplication
(Pen £ 1.2) 101 63 62 41 39 27 20 19 18 12
MAD (estimated to actual duplication) 19 4 8 2 8 6 4 4 7 2
Duplication coef?cient 1.2 6.0 0.98
Notes: MAD ¼ Mean Absolute Deviation across all pairs of duplicated destinations; Act: ¼ actual duplications; Est: ¼ model-estimated
duplications
Table VII Duplication of purchase (top ten destinations in each market)
Japan Wave 1 Fr Sp USA Ita BNL Gre Ger Aus Swi Por Average MAD
Correlation
(Act. v Est.)
Average duplication 37 27 25 26 19 23 23 16 16 13
Model-estimated duplication ¼ ðPen £ 1.4) 37 27 24 24 22 22 22 18 15 15
MAD (estimated to actual duplication) 0 0 1 2 3 1 1 2 1 2
Duplication coef?cient 0.9 1.0 0.94
Notes: MAD ¼ Mean Absolute Deviation across all pairs of duplicated destinations; Act: ¼ actual duplications; Est: ¼ model-estimated
duplications
VOL. 3 NO. 1 2009
j
INTERNATIONAL JOURNAL OF CULTURE, TOURISM AND HOSPITALITY RESEARCH
j
PAGE 45
D
o
w
n
l
o
a
d
e
d
b
y
P
O
N
D
I
C
H
E
R
R
Y
U
N
I
V
E
R
S
I
T
Y
A
t
2
2
:
0
6
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
T
a
b
l
e
V
I
I
I
C
o
m
p
a
r
i
s
o
n
o
f
r
e
s
u
l
t
s
o
v
e
r
m
u
l
t
i
p
l
e
y
e
a
r
s
:
J
a
p
a
n
a
n
d
U
S
A
J
a
p
a
n
W
a
v
e
1
J
a
p
a
n
W
a
v
e
2
U
S
A
W
a
v
e
1
U
S
A
W
a
v
e
2
A
B
C
A
B
C
A
B
C
A
B
C
D
e
s
t
P
e
n
%
A
v
g
D
u
p
E
s
t
D
u
p
D
e
s
t
P
e
n
%
A
v
g
D
u
p
E
s
t
D
u
p
D
e
s
t
P
e
n
%
A
v
g
D
u
p
E
s
t
D
u
p
D
e
s
t
P
e
n
%
A
v
g
D
u
p
E
s
t
D
u
p
H
a
w
4
0
3
7
3
7
H
a
w
5
3
4
8
5
4
M
e
x
4
8
4
7
6
1
F
r
a
n
4
9
6
1
6
3
U
S
A
3
0
2
7
2
7
U
S
A
3
7
3
5
3
8
F
r
a
3
1
4
7
3
9
U
K
4
1
4
7
5
2
K
o
r
2
6
2
5
2
4
G
u
a
3
2
2
8
3
2
C
a
r
3
1
2
7
3
9
I
t
a
4
0
4
9
5
1
H
K
2
6
2
6
2
4
H
K
2
8
3
1
2
9
G
e
r
3
0
4
5
3
7
M
e
x
3
4
3
6
4
3
G
u
a
2
4
1
9
2
2
F
r
a
2
3
3
1
2
3
C
a
n
2
6
2
0
3
3
G
e
r
2
7
3
8
3
5
F
r
a
2
4
2
3
2
2
S
i
n
g
2
3
2
0
2
3
U
K
2
6
3
4
3
3
C
a
n
2
5
2
8
3
2
S
i
n
2
4
2
3
2
2
A
u
s
2
2
2
0
2
3
I
t
a
2
5
3
9
3
1
S
p
a
2
0
2
5
2
6
A
u
s
2
0
1
6
1
8
K
o
r
1
9
1
9
2
0
S
w
i
1
8
3
2
2
3
C
a
r
i
2
0
2
0
2
6
U
K
1
7
1
6
1
5
I
t
a
1
8
2
3
1
9
A
u
s
1
5
2
6
1
9
B
N
L
1
6
2
3
2
1
T
a
i
1
6
1
3
1
5
U
K
1
7
2
2
1
7
H
a
w
1
5
1
6
1
8
S
w
i
1
5
2
3
2
0
D
0
.
9
1
.
0
1
.
3
1
.
3
C
o
r
r
.
0
.
9
7
0
.
9
3
0
.
9
7
0
.
9
7
M
A
D
6
.
4
7
.
1
8
.
2
3
.
5
N
o
t
e
s
:
D
¼
D
u
p
l
i
c
a
t
i
o
n
c
o
e
f
?
c
i
e
n
t
;
C
o
r
r
:
¼
c
o
r
r
e
l
a
t
i
o
n
b
e
t
w
e
e
n
a
v
e
r
a
g
e
d
u
p
l
i
c
a
t
i
o
n
(
A
v
g
D
u
p
)
a
n
d
e
s
t
i
m
a
t
e
d
d
u
p
l
i
c
a
t
i
o
n
(
E
s
t
D
u
p
)
;
M
A
D
¼
M
e
a
n
A
b
s
o
l
u
t
e
D
e
v
i
a
t
i
o
n
b
e
t
w
e
e
n
a
c
t
u
a
l
d
u
p
l
i
c
a
t
i
o
n
s
a
n
d
m
o
d
e
l
-
e
s
t
i
m
a
t
e
d
d
u
p
l
i
c
a
t
i
o
n
s
PAGE 46
j
INTERNATIONAL JOURNAL OF CULTURE, TOURISM AND HOSPITALITY RESEARCH
j
VOL. 3 NO. 1 2009
D
o
w
n
l
o
a
d
e
d
b
y
P
O
N
D
I
C
H
E
R
R
Y
U
N
I
V
E
R
S
I
T
Y
A
t
2
2
:
0
6
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
Continuing the examination of the UK market, we see Spain, USA and Greece share fewer
travelers with France than expected. Similarly, several countries share fewer travelers with
Spain than expected: USA, Benelux, Germany, Austria and Switzerland (that said, over half
the travelers to any of these countries still also visit Spain). There are other noticeable
partitions, with heightened sharing of travelers between Germany, Austria, Switzerland and
Italy. These partitions are all statistically signi?cant and are 20 percent (or more) different to
the expected ?gure, which indicates they are worthy of managerial attention.
In Table IX we show major partitions across the four markets. Most involve destinations
sharing travelers more than expected, but we also see that certain brands share travelers far
less than expected (all exhibit differences to expected values at p , 0.05). We discuss the
partitions that appear to be based on geographic proximity ?rst. The effect of proximity can
occur where the countries in the partition are both or all close to the origin market; or where
the partitioned countries are comparatively far fromthe origin market but close to each other
(designated as ‘‘1’’ or ‘‘2’’ respectively in Table IX). Proximity is associated with heightened
sharing, in some cases to a very large extent. We also note that tourist market partitions
occur that are based on destinations sharing a similar culture and language. For example,
Austria, Switzerland and Germany share 38 percent more tourists from the UK market than
expected (Italy is also in this partition). However, these destinations also share geographic
proximity to each other. The combination of geographical and cultural similarity makes it
dif?cult to say which is the better explanation for the partition.
Consider the destinations that share travelers less with each other than expected. China and
France, for example, exhibit lower than expected sharing of USA travelers. This may be due
to the long distance between each country. Instances of lower levels of sharing occur that
could build from cultural factors. A very noticeable example is from the Singapore market:
Singaporeans who travel to India are far less likely to also travel to China (3 percent
compared to 28 percent expected). This proposal follows from the cultural heterogeneity in
Singapore, speci?cally, that this city-state has a signi?cant Indian population.
Finally, in relation to culture and its possible effects, we proposed earlier that the impact of
culture might be exhibited through a poorer ?t of the DPL. We can make some exploratory
observations about this issue from examining the overall results for the two Western markets
with the two Eastern markets. Inspection of the graphs in Figures 2–5 shows the correlations
between penetration and duplications are higher for UK and USA (r ¼ 0:89 and 0.83)
Table IX Major partitions
Market Wave Countries in partition Actual dup. Est. dup. % Diff. Type of partition
Japan 1 and 2 HK – Sing 37 24 54 1
1 and 2 Korea – Taiwan 38 20 90 1
1 and 2 France – UK 55 15 258 2
1 and 2 Hawaii – Guam 33 22 50 2
USA 1 Australia – New Zealand 25 14 80 2
1 France – Italy 42 30 40 2
1 Germany – France 35 25 240 2
1 China – Japan 23 12 92 2
1 China – France 15 25 240 3
UK 1 Austria – Germany – Italy – Switzerland 51 37 38 1
1 Spain, USA, Greece (share less with): France 67 87 220 3
1 Germany, Austria, USA, Switz, BLX (share less
with): Spain 58 78 225 3
Singapore 1 India – China 3 28 283* 3
1 Bali – Malaysia 43 18 240 1
1 Malaysia – India 35 10 142 3
China – Hong Kong 35 18 98 2
Notes: * Classi?cation of Partitions: 1 ¼ Partition between countries both or all close to origin country; 2 ¼ Partition between neighbouring
countries comparatively far from origin country but close to each other; 3 ¼ Unknown. All these partitions are statistically signi?cant at p
, 0.05; Actual dup. is the average of the observed duplications between the respective countries; Est. dup. is the model-generated
estimate of what the average duplications between those respective countries should be, given their respective popularity
VOL. 3 NO. 1 2009
j
INTERNATIONAL JOURNAL OF CULTURE, TOURISM AND HOSPITALITY RESEARCH
j
PAGE 47
D
o
w
n
l
o
a
d
e
d
b
y
P
O
N
D
I
C
H
E
R
R
Y
U
N
I
V
E
R
S
I
T
Y
A
t
2
2
:
0
6
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
compared to Singapore and Japan (r ¼ 0:79 and 0.53). These ?ndings support the
suggestion that competitive market structure in the tourism context may be altered to some
degree by the culture of the home market.
5. Summary and implications
This research examines four different international tourist origin markets: Singapore, Japan,
the USA, and the UK. The results show that the pattern of competition between tourist
destination brands is strongly in?uenced by overall market share, consistent with the DPL.
For markets where multi-year data is available, the duplication co-ef?cient parameters and
correlations between actual and estimated purchase duplications are stable over time. The
results show that destinations share travelers with other destinations approximately in-line
with the size of those other destinations. The study estimates the proportion of travelers to A
who also have traveled to B to be between 0.9 and 1.3, multiplied by the penetration (percent
ever traveled to) of B. This means the overall popularity of the destination is a primary
determinant of brand competition.
The major exceptions to this general pattern tend to comprise either:
B destinations that are an appreciable distance from the origin country but close to each
other; and
B destinations close geographically to the origin country
These partitions based on functional differences are broadly consistent with ?ndings in other
markets (e.g., Bennett et al., 2000; Carter and Silverman, 2004; Colombo et al., 2000; Grover
and Rao, 1988). Partitions that involve lowered levels of sharing between destinations are
identi?able. In one instance, cultural heterogeneity in the origin market may be a possible
explanation. These ?ndings might seem obvious in hindsight. However, no previous studies
identify them in the context of destination brands (note, an earlier version of this research
was presented, with considerably less detail, at the 2003 EMAC conference).
The ?ndings complement existing frameworks or approaches for understanding
competition. For example, substantial work examines the speci?c competitive factors that
drive tourist demand. This work includes physiography and tourism superstructure (Crouch
and Ritchie, 1999; Enright and Newton, 2005); as well as price levels and marketing
expenditure (e.g., O’Hagan and Harrison, 1984). Those factors drive overall market share –
a destination with more favorable levels of these factors will be more attractive to visitors.
This study shows that if a destination brand gains more market share, that brand will gain at
the expense of many other destination brands – in-line with their current overall popularity
levels. The extent of gain from speci?c competitor brands will be moderated by cultural
similarities and geographic proximity.
The results are also a very useful complement to analyses of competitive structure based on
brand positioning. The rationale for brand positioning is to create a perception of difference
that in turn drives preference (e.g., Kapferer, 1995, p. 95). Techniques such as
multi-dimensional scaling or correspondence analysis are often used to identify a brand’s
position (Carroll et al., 1986; Hoffman and Franke, 1986; Kennedy et al., 1996). Brands
closer in perceptual space are interpreted as closer substitutes, and are considered to be
more direct competitors to each other (e.g., Day et al., 1979, p. 14). The present study
highlights that similarity in brand positioning may not be the key factor that drives
competition. In contrast, brands tend to share more customers with larger brands (and those
in close geographic proximity), and these larger competitor brands may or may not be
positioned along similar lines. An emphasis on interpreting brand competition based on
positioning analysis could therefore lead tourism marketers to ignore major competitors
(larger share brands) in favor of brands that look similar in the eyes of consumers. The
?ndings of the present study suggest tourism marketers should pay more attention to other
larger destinations rather than other destinations that have similar brand positions. This shift
in outlook can increase marketing effectiveness and allocation of resources. For example,
interpreting competition based on Duplication of Purchase outputs could broaden the
de?nition of target market for a destination. Such a broadened targeting approach would
PAGE 48
j
INTERNATIONAL JOURNAL OF CULTURE, TOURISM AND HOSPITALITY RESEARCH
j
VOL. 3 NO. 1 2009
D
o
w
n
l
o
a
d
e
d
b
y
P
O
N
D
I
C
H
E
R
R
Y
U
N
I
V
E
R
S
I
T
Y
A
t
2
2
:
0
6
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
contrast to an approach based on brand positioning which would be to target travelers who
have been to other destinations that are perceptually positioned close to one’s own
destination. The present study suggests that the travelers to one’s own market are more likely
to have previously been to any other large/overall popular destination, not necessarily those
with brand perceptual positions close to one’s own. Likewise, the present study suggests
that the travelers who visit one’s own market are far more likely to then visit other large
destinations rather than destinations with similar brand image. The major exceptions are for
geographically proximate destinations, where there will be more sharing compared to
estimated levels based on the size effect in the DPL.
In terms of marketing opportunities, a destination can usually expect to share more tourists
with countries in the same vicinity than expected, given the overall popularity of the group of
destinations. Whether this sharing is viewed as a competitive threat, or a marketing
opportunity will depend on the country partitions and the relationship between destinations.
Heightened sharing also provides opportunities for neighbouring countries to collaborate on
tourism promotion ventures, particularly in regions with a number of small countries. For
example, Yong et al. (1989) discuss this approach in the context of Singapore, as an
outcome of a Delphi study investigating expert’s perceptions about what is needed for the
future development of that country’s tourism industry.
The present study applies the DPL to multiple market settings and multiple time periods,
which strengthens generalizability of the ?ndings. The duplication of purchase-style analysis
has an additional bene?t for the tourism research community, based on the fact that the law
was originally developed in the contexts of television viewing and packaged goods, then
extended to broader applications, and now to tourism. The approach of taking an empirical
regularity from one context to other contexts opens up opportunity to apply other
fundamental patterns in buyer behavior to tourism. For example, does the famous double
jeopardy pattern in brand loyalty (e.g., Ehrenberg et al., 1990) apply to tourismdestinations?
While the present study does ?nd some indication of market partitions that could be based
on cultural factors, further research is necessary to disentangle these from the primary
market share and location effects.
5.1 Limitations and future research
The key limitations of this research relate to the data. First, we cannot identify growth trends
for speci?c brands because data on visits were not provided in temporal order. This should
be an area for future research with data within a speci?ed time frame analysed, or repeat
behavior (last holiday and then holiday prior to that). Second, we cannot do cross-country
comparisons for all the individual destinations from each origin market. Further analysis with
larger samples and a larger number of destinations fromeach origin market would provide a
clearer picture of competitive structure. In addition, analysis using larger numbers of
destinations could better distinguish between culture and proximity effects by including
more sets of countries that are culturally similar but geographically dispersed (eg, Spain and
Argentina).
This study examines aggregate-level relationships, and these results need not hold for any
individual consumer, or consumer segment. However this study appears to be the ?rst
examination of cross-purchasing in a tourism destination context and an aggregate-level
analysis is appropriate. The study is able to tentatively, at least, indicate howheterogeneity in
the origin market can affect competitive market structure, in the case of lower levels of
sharing between India and China for Singaporean travelers. Subsequent studies could
examine heterogeneity in more detail by conducting purchase duplication analysis over
various consumer segments.
This research area could also progress by examining tourist travelers’ cross-purchases
according to whether they occurred on the same trip or on different trips, which would
distinguish between competitiveness and complementarity. For example, if travelers tend to
visit brands A and B on the same trip then these two brands might simultaneously compete
against, and also complement each other. The two brands compete for a share of the
VOL. 3 NO. 1 2009
j
INTERNATIONAL JOURNAL OF CULTURE, TOURISM AND HOSPITALITY RESEARCH
j
PAGE 49
D
o
w
n
l
o
a
d
e
d
b
y
P
O
N
D
I
C
H
E
R
R
Y
U
N
I
V
E
R
S
I
T
Y
A
t
2
2
:
0
6
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
traveler’s time and expenditure within the same trip, but could also complement each other
by together offering a more attractive overall destination than either one on its own.
References
Agresti, A. (2002), Categorical Data Analysis, 2nd ed., Wiley, Hoboken, NJ.
Armstrong, J.S. (2001), ‘‘Evaluating forecasting methods’’, in Armstrong, J.S. (Ed.), Principles of
Forecasting. A Handbook for Researchers and Practitioners, Kluwer Academic Publishers, Boston, MA.
Bass, F.M. (1974), ‘‘The theory of stochastic preference and brand switching’’, Journal of Marketing
Research, Vol. 11, February, pp. 1-20.
Bauer, H.H. and Herrmann, A. (1995), ‘‘Market demarcation: theoretical framework and results of an
empirical investigation of the German car market’’, European Journal of Marketing, Vol. 29 No. 11,
pp. 18-34.
Belau, D. (2003), ‘‘New threats to employment in the travel and tourism industry’’, International Labour
Organisation, Geneva.
Bennett, D. (2004), ‘‘Mapping the dynamic Chinese marketplace – the case of TV’s’’, ANZMAC
Conference, Wellington, Victoria University of Wellington, New Zealand.
Bennett, D. and Ehrenberg, A.S.C. (2001), ‘‘A lot can be revealed by a little data: two purchase analysis
of fast food buying’’, ANZMAC Conference, Massey University, Auckland, New Zealand.
Bennett, D., Ehrenberg, A. and Goodhardt, G. (2000), ‘‘Two purchase analysis of brand loyalty among
petrol buyers’’, ANZMAC Conference, Grif?th University, Gold Coast, Queensland.
Bennett, D.R. and Ehrenberg, A.S.C. (2002), ‘‘The way Australians buy fast food’’, Proceedings of the
European Marketing Academy Conference (EMAC 2002). Braga, Portugal, May.
Carroll, J.D., Green, P.E. and Schaffer, C.M. (1986), ‘ ‘Interpoint distance comparisons in
correspondence analysis’’, Journal of Marketing Research, Vol. 23, August, pp. 271-80.
Carter, J. and Silverman, F. (2004), ‘‘An empirical approach to market partitioning: application to the
cigarette market’’, Journal of Targeting, Measurement and Analysis for Marketing, Vol. 12 No. 4,
pp. 366-78.
Colombo, R., Ehrenberg, A. and Sabavala, D. (2000), ‘‘Diversity in analyzing brand-switching tables:
the car challenge’’, Canadian Journal of Marketing Research, Vol. 19, pp. 23-36.
Cooper, L.G. and Inoue, A. (1996), ‘‘Building market structures from consumer preferences’’, Journal of
Marketing Research, Vol. 33, August, pp. 239-306.
Crotts, J. (2004), ‘‘The effect of cultural distance on overseas travel behaviors’’, Journal of Travel
Research, Vol. 43, August, pp. 83-8.
Crouch, G.I. and Ritchie, J.R.B. (1999), ‘‘Tourism, competitiveness, and societal prosperity’’, Journal
of Business Research, Vol. 44, pp. 137-52.
Day, G.S., Shocker, A.D. and Srivastava, R.K. (1979), ‘‘Customer-oriented approaches to identifying
product-markets’’, Journal of Marketing, Vol. 43, Fall, pp. 8-19.
Ehrenberg, A.S.C. (1975a), Data Reduction, Wiley, London and New York, NY.
Ehrenberg, A.S.C. (1975b), ‘‘The structure of an industrial market: aviation fuel contracts’’, Industrial
Marketing Management, Vol. 4, pp. 273-85.
Ehrenberg, A.S.C. (2000), ‘‘Repeat-buying: facts, theory and applications’’, Journal of Empirical
Generalisations in Marketing Science, Vol. 5, pp. 392-770.
Ehrenberg, A.S.C. and Goodhardt, G.J. (1970), ‘‘A model of multi-brand Buying’’, Journal of Marketing
Research, Vol. 7, February, pp. 77-84.
Ehrenberg, A.S.C., Goodhardt, G. and Barwise, T.P. (1990), ‘‘Double Jeopardy revisited’’, Journal of
Marketing, Vol. 54, July, pp. 82-91.
Ehrenberg, A.S.C., Uncles, M.D. and Goodhardt, G.G. (2004), ‘‘Understanding brand performance
measures: using Dirichlet benchmarks’’, Journal of Business Research, Vol. 57 No. 12, pp. 1307-25.
PAGE 50
j
INTERNATIONAL JOURNAL OF CULTURE, TOURISM AND HOSPITALITY RESEARCH
j
VOL. 3 NO. 1 2009
D
o
w
n
l
o
a
d
e
d
b
y
P
O
N
D
I
C
H
E
R
R
Y
U
N
I
V
E
R
S
I
T
Y
A
t
2
2
:
0
6
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
Enright, M.J. and Newton, J. (2005), ‘‘Determinants of tourism destination competitiveness in Asia
Paci?c: comprehensiveness and universality’’, Journal of Travel Research, Vol. 43, pp. 339-50.
Goodhardt, G.J. and Ehrenberg, A.S.C. (1969), ‘‘Duplication of television viewing between and within
channels’’, Journal of Marketing Research, Vol. 6, May, pp. 169-78.
Grover, R. and Rao, V. (1988), ‘‘Inferring competitive market structure based on a model of interpurchase
intervals’’, International Journal of Research in Marketing, Vol. 5, pp. 55-72.
Hauser, J.R. and Wernerfelt, B. (1989), ‘‘The competitive implications of relevant-set/response analysis’’,
Journal of Marketing Research, Vol. 26, pp. 391-405.
Hoffman, D.L. and Franke, G.R. (1986), ‘‘Correspondence analysis: graphical representation of
categorical data in marketing research’’, Journal of Marketing Research, Vol. 23, August, pp. 213-27.
Hofstede, G. (1994), Cultures and Organizations: Intercultural Cooperation and its Importance for
Survival, Harper Collins Publishers, London.
Hotels (2002), ‘‘Hotels global update: ?nance’’, Reed Business, available at: www.hotelsmag.com
(accessed February 2008).
Kalwani, M.U. and Morrison, D.G. (1977), ‘‘A parsimonious description of the Hendry System’’,
Management Science, Vol. 23 No. 5, pp. 467-77.
Kapferer, J-N. (1995), Strategic Brand Management – New Approaches to Creating and Evaluating
Brand Equity, Kogan Page Limited, London.
Kennedy, R., Riquier, C. and Sharp, B. (1996), ‘‘Practical applications of correspondence analysis to
categorical data in market research’’, Journal of Targeting, Measurement and Analysis for Marketing,
Vol. 5 No. 1, pp. 56-70.
Kolland, F., Milian, J., Scheibenstock, M. and Schonbauer, H. (2000), ‘‘The short waves of the
product-cycle of tourism: Cuba’s tourism between de-coupling and integration in to the world market’’,
Management, Marketing and the Political Economy of Travel and Tourism, University of Northumbria,
Newcastle.
Kotler, P. and Gertner, D. (2002), ‘‘Country as brand, product and beyond: a place marketing and brand
management perspective’’, Journal of Brand Management, Vol. 9 Nos 4-5, pp. 249-61.
Law, J. (2004), ‘‘The long and the short of it’’, Travel Trade Gazette, Vol. 2, July.
Liu, Z. (2000), ‘‘The locational advantage of tourism destinations – determinants, dimensions and
dynamics’’, Marketing and the Political Economy of Travel and Tourism, University of Northumbria,
Newcastle.
Moutinho, L. (1987), ‘‘Consumer behavior in tourism’’, European Journal of Marketing, Vol. 21 No. 10,
pp. 5-44.
Novak, T.P. and Stangor, S. (1987), ‘‘Testing competitive market structures: an application of weighted
least squares methodology to brand switching data’’, Marketing Science, Vol. 6 No. 1.
O’Hagan, J.W. and Harrison, M.J. (1984), ‘‘Market shares of US tourist expenditure in Europe:
an econometric analysis’’, Applied Economics, Vol. 16, pp. 919-31.
Reisinger, Y. and Turner, L. (1999), ‘‘A cultural analysis of japanese tourists: challenges for tourism
marketers’’, European Journal of Marketing, Vol. 33 Nos 11/12, pp. 1203-27.
Romaniuk, J. and Dawes, J. (2005), ‘‘Loyalty to price tiers in purchases of bottled wine’’, Journal of
Product & Brand Management, Vol. 14 No. 1, pp. 57-64.
Russell, G.J. and Bolton, R.N. (1988), ‘‘Implications of market structure for elasticity structure’’, Journal
of Marketing Research, Vol. 25, August, pp. 229-41.
Schiffman, L.G. and Kanuk, L.L. (2004), Consumer Behavior, Pearson Prentice Hall, Englewood Cliffs,
NJ.
Sharp, B. and Sharp, A. (1997), ‘‘Positioning and partitioning’’, 26th European Marketing Academy
Conference, Warwick Business School, The University of Warwick, Coventry.
Singh, S. and Catlin, J. (2003), ‘‘Low-cost airlines ?ying high’’, Marketing Week, April, pp. 34-5.
VOL. 3 NO. 1 2009
j
INTERNATIONAL JOURNAL OF CULTURE, TOURISM AND HOSPITALITY RESEARCH
j
PAGE 51
D
o
w
n
l
o
a
d
e
d
b
y
P
O
N
D
I
C
H
E
R
R
Y
U
N
I
V
E
R
S
I
T
Y
A
t
2
2
:
0
6
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
Uncles, M.D. and Ehrenberg, A.S.C. (1990a), ‘‘The buying of packaged goods at US retail chains’’,
Journal of Retailing, Vol. 66 No. 3, pp. 278-96.
Uncles, M.D. and Ehrenberg, A.S.C. (1990b), ‘‘Industrial buying behavior: aviation fuel contracts’’,
International Journal of Research in Marketing, Vol. 7 No. 1, pp. 56-68.
Urban, G.L., Johnson, P.L. and Hauser, J.R. (1984), ‘‘Testing competitive market structures’’, Marketing
Science, Vol. 3 No. 2.
World Travel and Tourism Council (WITC) (2003), Blueprint for New Tourism, Vol. 2005, World Travel and
Tourism Council, London, available at, www.wttc.org/eng/About_WTTC/Blueprint_for_New_Tourism/
(accessed February 2008).
World Travel and Tourism Council (2007), Tourism Satellite Accounting, available at:
/www.wttc.org/blueprint/WTTCBlueprintFinal.pdf (accessed February 2008).
World Travel and Tourism Council (2008), ‘‘Continued growth signalled for travel and tourism industry’’,
March, available at: www.wttc.travel/eng/Tourism_News/Press_Releases/Press_Releases_2008
(accessed February 2008).
Yong, Y.W., Keng, K.H. and Leng, T.L. (1989), ‘‘A Delphi forecast for the Singapore tourism industry:
future scenario and marketing implications’’, European Journal of Marketing, Vol. 23, pp. 15-26.
Appendix 1
The DPL is summarized with the simple formula:
b
XY
= b
X
¼ Db
Y
(Ehrenberg and Goodhardt, 1970, p. 78).
In words, this formula states that the proportion of brand X buyers who also buy brand Y is
directly proportional to the penetration of brand Y (where b
x
and b
Y
are the penetrations of
brands X and Y respectively, b
xy
is the proportion of X buyers who buy Y, and D is the
average amount of cross-purchasing or purchase duplication across all pairs of brands in
the market).
The duplication coef?cient, as stated in the body of the paper is average of purchase
duplications divided by average penetration. Arithmetically, for four brands A to D it is:
ðb
AB
þ b
AC
þ b
AD
þ b
BA
þ b
BC
. . . þ b
DC
Þ = 12
*
ðb
A
þ b
B
þ b
C
þ b
D
Þ = 4
ð
*
For a 4 £ 4 brandmatrix less 4 entries on diagonalsÞ:
Where b
A
. . . .b
D
is the penetrations for the brands respectively and b
AB
, b
AC
, b
AD
. . .b
DC
is
the duplications for each pair of brands in the market respectively.
Appendix 2. Sample characteristics
USA – 55 percent male, 45 percent female. 70 percent were aged 46 or over. 55 percent
worked full time and 30 percent had an annual income before tax of over $100,000. 66
percent were married while 38 percent had children living in the household.
UK – 41 percent male and 59 percent female. 44 percent were aged between 26 and 45
years, with a further 21 percent aged between 46 and 55 years. 44 percent worked full time
and 21 percent were retired; 30 percent were A or B social class. 67 percent were married
and 62 percent had children living at home.
Singapore – 57 percent male and 43 percent female. 54 percent were aged between 25 and
46 years, but a further 33 percent were under 25 years. 32 percent worked full time while 37
percent worked part-time and 23 percent had an average income before tax of over
S$6,000. 49 percent were married and 67 percent reported having children living in the
home.
Japan – 24 percent male and 76 percent female; 55 percent were aged 25-44 years; 37
percent were in professional, of?ce or administrative work while 36 percent were
housewives; 20 percent earned over 9 million Yen before tax. 83 percent were married
and 68 percent had children living in the home. The larger proportion of females in the Japan
sample re?ected a heightened propensity to engage in travel or decide on travel on the part
of Japanese females.
PAGE 52
j
INTERNATIONAL JOURNAL OF CULTURE, TOURISM AND HOSPITALITY RESEARCH
j
VOL. 3 NO. 1 2009
D
o
w
n
l
o
a
d
e
d
b
y
P
O
N
D
I
C
H
E
R
R
Y
U
N
I
V
E
R
S
I
T
Y
A
t
2
2
:
0
6
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
Appendix 3. Method of assessment for the statistical signi?cance of partitions
Each respondent in this study could report traveling to two or more destinations. This
characteristic makes the data unsuitable to analyse using methods such as log-linear
analysis (e.g., Agresti, 2002). Therefore, the statistical signi?cance of partitions was
assessed using a permutational method. A Visual Basic program was constructed to
repeatedly draw random samples from an extremely large population. Each sample was of
the same size, number of brands, brand penetration and overall duplication levels, as each
of the respective surveys. The purchase duplications between speci?c brands were
recorded for each sample. Over a series of 10,000 iterations, this procedure revealed the
probability of observing brand duplications of a particular magnitude (e.g., deviations),
given the respective sizes of the brands. Brands that exhibited a deviation that was less than
5 percent likely to occur due to random sampling variation were judged to be statistically
signi?cant.
Corresponding author
John Dawes can be contacted at: [email protected]
VOL. 3 NO. 1 2009
j
INTERNATIONAL JOURNAL OF CULTURE, TOURISM AND HOSPITALITY RESEARCH
j
PAGE 53
To purchase reprints of this article please e-mail: [email protected]
Or visit our web site for further details: www.emeraldinsight.com/reprints
D
o
w
n
l
o
a
d
e
d
b
y
P
O
N
D
I
C
H
E
R
R
Y
U
N
I
V
E
R
S
I
T
Y
A
t
2
2
:
0
6
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
This article has been cited by:
1. Naomi Gruneklee, Sharyn Rundle-Thiele, Krzysztof Kubacki. 2016. What can social marketing learn from Dirichlet theory patterns in a
physical activity context?. Marketing Intelligence & Planning 34:1, 41-60. [Abstract] [Full Text] [PDF]
2. Margaret Faulkner, Oanh Truong, Jenni Romaniuk. 2014. Uncovering generalized patterns of brand competition in China. Journal of Product
& Brand Management 23:7, 554-571. [Abstract] [Full Text] [PDF]
D
o
w
n
l
o
a
d
e
d
b
y
P
O
N
D
I
C
H
E
R
R
Y
U
N
I
V
E
R
S
I
T
Y
A
t
2
2
:
0
6
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
doc_573225353.pdf