Description
This paper has two main aims: to show responses like yes or very likely for inbound visitors
returning to a destination can lead to misleading and unreliable information; and to clarify the kind of
information that should be collected
International Journal of Culture, Tourism and Hospitality Research
Inadequate return questions: return when? Sometime?
Ching-Tang Wang Tzung-Cheng Huan Tang-Chung Kan
Article information:
To cite this document:
Ching-Tang Wang Tzung-Cheng Huan Tang-Chung Kan, (2012),"Inadequate return questions: return when? Sometime?", International
J ournal of Culture, Tourism and Hospitality Research, Vol. 6 Iss 1 pp. 44 - 53
Permanent link to this document:http://dx.doi.org/10.1108/17506181211206243
Downloaded on: 24 January 2016, At: 22:18 (PT)
References: this document contains references to 26 other documents.
To copy this document: [email protected]
The fulltext of this document has been downloaded 272 times since 2012*
Users who downloaded this article also downloaded:
Thomas Foscht, Karin Ernstreiter, Cesar Maloles, Indrajit Sinha, Bernhard Swoboda, (2013),"Retaining or returning?: Some insights for a
better understanding of return behaviour", International J ournal of Retail & Distribution Management, Vol. 41 Iss 2 pp. 113-134 http://
dx.doi.org/10.1108/09590551311304310
Lloyd C. Harris, (2010),"Fraudulent consumer returns: exploiting retailers' return policies", European J ournal of Marketing, Vol. 44 Iss 6 pp.
730-747http://dx.doi.org/10.1108/03090561011032694
Wu-Chung Wu, You-De Dai, Hsiou-Hsiang J ack Liu, (2012),"Effective information collection on international inbound visitors", International
J ournal of Culture, Tourism and Hospitality Research, Vol. 6 Iss 1 pp. 54-69http://dx.doi.org/10.1108/17506181211206252
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
:
1
8
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
Inadequate return questions: return when?
Sometime?
Ching-Tang Wang, Tzung-Cheng Huan and Tang-Chung Kan
Abstract
Purpose – This paper has two main aims: to show responses like yes or very likely for inbound visitors
returning to a destination can lead to misleading and unreliable information; and to clarify the kind of
information that should be collected.
Design/methodology/approach – Responses from Taiwan’s inbound visitors relating to returning are
examined to see what can be learned. Modeling is used to extract meaningful quantitative information
from data.
Findings – Modeling shows that survey responses about return are inconsistent. Although 95 percent of
non-visiting-friends-and-relations (VFR) leisure visitors indicate returning, this is not consistent with a
retention rate of 90 percent. A retention rate of 33 percent is consistent with the observation that 70
percent of person-visits are ?rst-visits. However, 33 percent retention is not consistent with over 95
percent of visitors returning. Conventional questions are yielding highly unreliable information and,
therefore, data collection should be changed.
Originality/value – Relations between vague questions and return trips have been established. This
research provides new evidence of the need for return data to include information allowing estimation of
volume and timing of return.
Keywords Return intentions, Taiwan, Inbound, Survey questions, Tourism management,
Social dynamics
Paper type Research paper
Introduction
In this research, the product of concern is a leisure visit to a foreign country that is not a
visiting-friends-and-relation (VFR) visit. Using common terminology the research is on
inbound leisure visiting that is not VFR. The caveat that the research is for leisure visits that
are not VFR is not continually repeated since, as the previous sentences suggest, repetition
can be tedious. However, periodic reference to leisure visits occurs to remind the reader of
the focus. This is done because, by de?nition, repeat purchase (visits) for visiting friends and
relation, for conferences and for business are driven by different mechanisms than
discretionary leisure travel.
Hsu and Kang (2007) cite the need to consider ideas regarding repeat visiting such as
raised in Huan et al. (2003) and Kozak et al. (2002). Hsu and Kang (2007) recognize much
research deals with return using response to questions like ‘‘do you plan to return?’’. Huan
et al. (2003), Kozak et al. (2002) and Beaman et al. (2002), hereafter HKB, focus attention on
a person’s yes-no response or response to a rating (e.g. not likely to highly likely) not
providing good information for understanding behavior. For example, very likely to return
does not clarify how likely or when return is being contemplated. Is returning sometime, say,
90 percent certain?
PAGE 44
j
INTERNATIONAL JOURNAL OF CULTURE, TOURISM AND HOSPITALITY RESEARCH
j
VOL. 6 NO. 1 2012, pp. 44-53, Q Emerald Group Publishing Limited, ISSN 1750-6182 DOI 10.1108/17506181211206243
Ching-Tang Wang is an
Associate Professor in the
Department of Sport
Management, National
Taiwan College of Physical
Education, Puzi City,
Taiwan. Tzung-Cheng Huan
is Dean of the College of
Management, National
Chiayi University, Chiayi
City, Taiwan. Tang-Chung
Kan is Head of the
Department of Travel
Management, National
Kaohsiung Hospitality
College, Kaohsiung City,
Taiwan.
Received September 2009
Revised April 2010
Accepted July 2010
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
:
1
8
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
If research is to build theory or to support planning, managing or marketing, are implications
of the research limited without addressing likelihood of returning in a time frame? With yes to
returning people who are 90 percent sure to return sometime in their life are lumped with
people 90 percent sure to return in the next couple of years. With such data, conclusions that
can be reached about volume of return in any year are limited. If yes or almost certainly are
taken to mean 90 percent or more likely to return but 50 percent or less is the return rate, is
theory sound? With a consumer product like dishwasher soap, some people may take a lot
longer to use some unit/quantity of the product than other people. Nevertheless, a
researcher can expect that most people currently using some brand of, for example,
dishwasher soap that is regularly purchased will continue using that dishwasher soap for
some time (e.g. for the next few years). Therefore, knowing about loyalty in making brand
choices has practical implications (re consumer products and brand see references in
Oliver, 1999; Sivadas and Baker-Prewitt, 2000; Haig, 2006). In the context just given, as of
2009, reason suggests intention to buy again, to return, should be addressing time frame
and/or likelihood of the next purchase.
Ideas in the previous paragraphs should provoke thinking about why intention-to-return data
are collected. Now that research relates satisfaction, loyalty and intention (Baker and
Crompton, 2000; Darnell and Johnson, 2001; Fakeye and Crompton, 1991; Gitelson and
Crompton, 1984; Hsu and Kang, 2007; Kozak, 2001; Oppermann, 1998a; Pritchard and
Howard, 1997; Yu¨ ksel, 2001), considering when respondents may return and likelihood of
return is appropriate. Consider that knowing an answer of yes means a 70 percent chance of
actually returning and a 50 percent chance of returning in two years. If this allows making
inferences about return ?ows, making estimates has practical value. One ?nds some
modeling ideas regarding repeating visitors (e.g. Baloglu and Erickson, 1998, 1999;
Oppermann, 1998b). However, HKB make arguments that some modeling by Markov
models is misleading. What can be learned from intention data such as those collected from
Taiwan’s inbound tourists (TIB data) is unclear. In fact, no research was found pursuing
whether creative thinking and modeling can yield models allowing useful inferences to be
made from TIB responses relating to return. Rather than looking at ideal data for analysis,
this research examines what can be done with responses relating to returning such as those
found in TIB data.
Objectives, hypotheses, methodology and terminology
Many countries collect intention and return data similar to those collected from inbound
visitors in Taiwan. Studies of returning acknowledge the importance of being a ?rst-time
visitor as opposed to a repeat visitor (e.g. HKB: Fakeye and Crompton, 1991). Examination
of surveys in which returning is considered (see citations and references in those) shows a
person is usually asked if they will or are likely to return and are expected to answer by
yes-no or a rating from very unlikely to very likely. In some research, a person is simply
asked if they are a ?rst-time visitor. In other cases, a respondent speci?es the number of
past visits ever or in a ?xed time frame. For TIB data used the time frame is the past three
years (i.e. in the past three years how many times have you visited Taiwan). In some data,
for example in the three years of TIB data used for model development, no information is
obtained on having visited prior to the time frame (data on ever visited are in a different TIB
survey). For TIB surveys without visited-ever information, ?rst-time visitors and repeat
visitors cannot be distinguished. For a visitor who has not come in the last three years, all
one knows is that they are ?rst-time or came to Taiwan more than three years prior to
becoming a TIB survey respondent.
One reason that this research examines existing data to see what can be learned is that
sound development of ideas on data collection is facilitated by understanding what can be
learned with data that have been collected. The general objective of this research is to learn
what useful information about the ?ow of returning over time can be extract fromdata like the
TIB data. The research objective is manifested in H1 and H2:
H1. Data with intention and return variables, such as in TIB data, will yield credible and
useful information on repurchase behavior of tourists.
VOL. 6 NO. 1 2012
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
:
1
8
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
H2. Analysis of what can be learned from data such as TIB data will yield speci?c
guidance/criteria for data collection that produces data that are useful in planning,
management and marketing.
Terminology: credible and useful
A reasonable basis for saying that a response to a question or responses to questions are
useful is that the information is good for testing hypotheses and/or they facilitate sound
decision making. However, if data have been used to test hypotheses and hypotheses are
established (i.e. as valid), how much testing is useful? With enough experiments, data from
one experiment will not con?rm some hypothesis (allow its acceptance). For example, for
signi?cance just at the 5 percent level about one replication in 20 of an experiment will result
in failure to con?rm an hypothesis. Developing theory involves more than repeating
experiments. Furthermore, in the context of a government using information, say the
government of Taiwan, collecting data should relate to using the information in planning,
management or marketing. In the case of asking if inbound tourists will return, knowing the
percent likely to return may appear to be of obvious value. However, assume year after year
the percent specifying returning is 96 percent or above (see Table I). Do you need the return
intention data year after year? Why ask about visits in the last three years (e.g. with
categories 1, 2, 3, or 4 visits and 5
þ
visits), assuming planning, management or marketing
use of responses, if proportions in categories change little from year to year? In fact for any
variable, if negligible change occurs over time and collecting data to measure change has a
signi?cant cost, why pay the collection price annually? Is money spent on collection well
used if information collected is not really needed? If information is not changing much or no
signi?cant contribution to meeting needs of academics or managers occurs for some other
reason, resources spent on the information can likely be put to better use.
Suggesting that commonly occurring data collection is ineffective may seem cynical.
Actually, in searches as part of this research, no planning, management or marketing
justi?cation has been found for asking about number of visits in the last three years. Does
some rationale exist for three years? None has been found. Would asking about visits in the
past ?ve years or past two years be better? No literature has been found addressing this
matter. In fact, given increased tourist ?ow from the mainland to Taiwan, change in the
distribution of TIB responses about visits during the past three years will occur. Oppermann
(1998b) and Baloglu and Erickson (1998, 1999) address repeat frequency and data
requirements. However, Darnell and Johnson (2001) and Beaman et al. (2002) are the
articles found that have raised the matter of having adequate data to interpret change.
Consider what useful means in another context. Having information about how responses to
TIB survey questions play a role in speci?c planning, management and marketing decision
making would be useful. As of 2009, searching for such information is yielding no results.
What is the situation if planning, managing and marketing are being based on a 97 percent
intention to return rate and intention does not match reality? If intention is 97 percent and
reality is 90 percent, is that OK? What if with 97 percent intention, return turns out to be 50
percent for ?rst-time visitors? If marketing targets ?rst-time visitors based on keeping a high
Table I Summary information from general questions on intention to return
Frequency of visits during last
three years
Percent 2001 and
2003
Percent
2002
Percent intend return 2001 and 2003
by last three years
Year of last
visit
Percent
2002
0 75.0 77.25 96.2 First visit 77.25
1 15.3 13.27 97.1 1 11.10
2 5.1 4.35 98.1 2 3.54
3 1.3 1.37 100 3 1.37
4 þ 3.3 2.90 99.3 . 3 years 5.89
Notes: Since respondents are selected at random, every visit during a year must be viewed as having equal probability of selection. Therefore,
tabulation of records without considering likely number of visits yields results in person-visits. To get persons one could correct for expected
number of visits in a survey year using one-third of visits in the past three years when visits in the past three years is greater than one
PAGE 46
j
INTERNATIONAL JOURNAL OF CULTURE, TOURISM AND HOSPITALITY RESEARCH
j
VOL. 6 NO. 1 2012
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
:
1
8
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
return and return is not high, are resources being used well? Fromthe point of viewof making
good decisions, are data collected supporting making good decisions or evaluating what is
being achieved by, for example, resources spent on marketing? How do you measure
success of marketing when you do not know what percent of this year’s ?rst-time visitors, or
even what percent of all current year visitors, can be expected to return within, say, three
years? If you cannot predict and measure marketing performance and plan or manage
based on data being collected, logic suggests modifying data collection so results are more
useful.
In the last paragraph, ideas relating to data being useful and credible are both present. If
you collect intention to return data and over 95 percent intend to return based on survey
responses, do you assume that the 95 percent re?ects actual returning that will occur? If
returning ever is just, say, 60 percent, does intention correspond with reality? This research
addresses that matter.
People involved with scales of return (e.g. 7 ¼ very likely), may argue that asking yes-no
regarding intending to return is asking for problems because single item measures are not
as reliable as multiple item measures. Discussion is going on in the literature about the
problem of basing measurement on a single question (Boyd et al., 2005; Loo, 2002). The
scale questions below are hypothetical but follow the pattern of intention questions in the
literature. The idea is that a measure based on three questions relating to a concept will be
more accurate than a single measure. For some measurement problems, forming a scale
based on multiple questions is certainly appropriate. However, does one get better
information on intention to return by multiple questions and forming, for example, an additive
scale?
Answering questions such as those above by 1 (strongly disagree) to 5 (strongly agree) and
forming an additive scale may seem appropriate to many researchers who work with scales.
For an additive scale, Cronbach’s alpha may show acceptable or even high reliability.
However, one should not confuse reliability as measured by alpha with a measure conveying
an accurate understanding of behavior. Say that a value of 96 percent of ?rst-time visitors
stating yes to intention of returning corresponds to 50 percent, or even 70 percent returning
sometime later. Does a highly reliable mean of 4.5 have a less ambiguous meaning than a
mean of 97 percent for yes that is consistent from year to year and has standard error of
about 0.1 percent for any year of TIB data? How do respondents rate scale anchors? Does a
relation exist between scale values and a practically useful measure such as percent of
?rst-time visitors returning in, for example, two years? How accurate is the relation? Actually,
how do you calibrate a scale value without relatively accurate information on return?
Returning is observable. Of?cial in Taiwan and other countries record entry and exit of
inbound visitors using unique identi?ers (i.e. passport information). Is returning really
something that should be learned about by probing the psyche using a scale? Ratings
questions that approach returning obtusely do not give speci?c information about likelihood
or timeframe. Given some people actually return, logical reasoning suggests scale reliability
should not be confused with information being useful and credible for meeting quantitative
needs of planners, managers and marketers for numbers of visitors in?uenced by marketing
or services.
The position that HKB supports is that a person may be uncertain about the future but insight
gained by asking vague, obtuse and/or possibly misleading questions is not likely to provide
as good of information for quantitative analysis of behavior as questions about what is likely
to be done and when. For example, survey questions (e.g. see the following list) may not be
relevant when going somewhere involves negotiation (e.g. family trips or travel with friends).
Revisit intention scale questions (hypothetical):
1. This trip has shown me that X is a good place to come to.
2. Among alternative trips, coming back to X is a high priority.
3. I know I’d enjoy myself on a return trip to X.
VOL. 6 NO. 1 2012
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
:
1
8
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
Decrop (2001) and Huan and Beaman (2003a, b) discuss such matters. Asking a person
who goes somewhere that does not particularly interest her about satisfaction without asking
about why a trip was taken means analysis will contain variance that limits reaching clear
conclusions. For Taiwan’s inbound, a high percent are traveling with friends or family so what
they do and intent to return may have a limited relation to activities, satisfaction, meeting
expectations, etc. Given that decisions about planning, managing and marketing should be
based on visitor ?ows, their expenditures, etc. credibility and usefulness of data to support
such decisions should be assessed based on speci?c use of statistics in meeting decision
making needs.
Method
Analysis is based on modeling. Introducing the TIB data for analysis is a ?rst step. Then a
model compatible with the TIB data is formulated. The model is based on plausible ideas for
approximating how behavior occurs. Then analysis of TIB data occurs. The article provides
conclusions and implications of the research for data collection, for theory development and
practical consideration of tourist return.
Data
Taiwan Tourism Bureau (TTB) uses the exit survey of inbound visitors to understand the
motives, viewpoints, tendencies and consumption of the inbound visitors. Reports
(e.g. Taiwan Tourism Bureau, 2001, 2002, 2004) are prepared as reference material for
tourist related organizations to plan for international tourism. The survey is also a source for
estimating the amount of visitor expenditures.
Inbound survey data are collected at exit points from Taiwan using a questionnaire. Design
of the questionnaire and its administration is by contract with a university statistical institute
(see TTB annual reports in the references for details). Since 1995, sampling has always
involved quota sampling. During a year total interviews are about 5,000.
Questions asked vary somewhat from year to year. In 2000, 2001 and 2003 respondents
were asked: ‘‘do you think you will visit Taiwan again?’’ The answer allowed was yes or no.
Another question asked in various ways is times Taiwan was visited in the past three years. In
2000 and 2001 years responses are for 1 to 4 times and 5 þ (though English translation of 5
or more as ‘‘over 5 times’’ are used). In 2003 the respondents could report actual numbers of
visits in excess of 5. Unfortunately, in 2000, 2001, 2003 information that lets one know if a
person visited more than 3 years previously is not available. Information is available for 2002.
Other than to estimate the proportion of visitors who came more than three years ago, 2002
data are not used because they were collected by a different contractor than 2000, 2001 and
2003 data using a questionnaire structure that is structured somewhat differently from the
questionnaire for the data used.
Model formulation
Given one has a limited travel history for people, modeling must build on that. For the
moment, think of people who are ?rst time visitors. In terms of intention to return, intending or
not intending to return is what is known. Therefore, for TIB data for 2000, 2001 and 2003,
times visiting in the past three years is also a variable. For the years mentioned, one has no
information about visiting prior to the three-year horizon. Therefore, ?rst-time visitors and
those returning visitors who would respond that they came more than 3 years prior to giving
data are in one category. Regardless, initially let N
k
, k.0, be the number of people reporting
coming k times in last n years. Since one does not know travel history, think of this being a
group of people that is fed from new visitors. For k =3, think of people tending to come once
in three years. In stochastic modeling for annual visiting this corresponds to a Poisson
variable with a mean of a third. For such modeling, people coming once in the past three
years would have a certain probability of being people tending to come every year, every
two years, every three years, etc. (means of 1, 1/2, 1/3, etc.).
PAGE 48
j
INTERNATIONAL JOURNAL OF CULTURE, TOURISM AND HOSPITALITY RESEARCH
j
VOL. 6 NO. 1 2012
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
:
1
8
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
Equation 1 re?ects the idea that people enter N
k
from being ?rst time visitors and remain in
N
k
except for retention being less than 1 (r,1). Equation 2 just expresses equation 1 using
the sum of the series of r
t
:
N
k
¼ rh þ Sr
t
h
where t ¼ 2 to some large number ðsymbolically 1; infinityÞ:
ð1Þ
N
k
¼ Sr
t
h ¼ rh=ð1 2rÞ
where in the sum the exponent t ¼ 1 to 1:
ð2Þ
Retention r , 1 should be taken to imply that people in a group on average show annual
decreased return over time of 1-r. In other words, a 10 percent decrease from year to year of
r ¼ 0.9 re?ects decreased visiting from a group. Part of the decrease can be from people
stopping visiting because they die. However, in general terms this is a duration effect.
People switch destinations, travel less or do not travel for various reasons (e.g. see
Oppermann, 1998a). Given the data available having duration effects associated with
changes in categories (Beaman et al., 2002) does not work because the data does not allow
model estimation. Nevertheless, allowing retention accommodates variability in behavior
that is modeled directly in more complicated models. An aggregate duration effect (e.g. r) is
a way of approximating changing behavior becoming more or less frequently. In a model in
which age is a variable, mortality can explain some decrease in coming. However, just as
many linear models are used to approximate reality, this model is proposed as a reasonable
model to approximate return behavior.
Confusion can arise because in exit surveys N
k
is not unique people. The unit of N
k
is person
trips (e.g. see note with Table I). For ?rst-time visitors, numbers of persons equals number of
visits. However, if X ?rst-time visitors become members of a k times in three years group, the
X ?rst-time visits can only be expected to make h visits per year (h¼X/k). In general for visits
in n years, think about an average of k visits per year as having k =m/n with m as a response
for visits in n years. X new visitors with k visits for year will only yield h ¼ X/k visits per year
with retention of 1(r ¼ 1). In other words, ?rst-time visitors who return and feed repeat
groups and have retention of r, satisfy equation 3.
X
k
¼ r kh
k
were variables are as specified above
ð3Þ
Analysis and results
Challenges exist in applying the model because data are needed to make estimates. Let N
r
be
the number of ?rst-time visitors, say N, minus ?rst-time visitors not intending to return visitors.
One challenge is determining how to partition ?rst-time visitors intending to return, N
r
, into
categories, k. However, for TIB data being used N is not known because people coming more
than three years prior to responding to the inbound survey are not distinguished from?rst-time
visitors. In other words, how to de?ne h
k
based on N
r
offers a challenge.
To pursue this challenge, Table I gives data on the variable visits in the past three years and for
some other variables. Note again that for visits in the past three years, ?rst-time and repeat
visitors who have not come in the past three years are in the same category. However, in 2002
and some other years of TIBdata one has information such as year of last visit or reason for last
visit. For example, for 2002 only about 6 percent indicated a last visit more than three years in
the past. Accepting 6 percent, about 70 percent of person-visits are ?rst visits.
For intention to return, data can be deceptive because people’s intentions do not necessarily
correspond with behavior. Consider, whether logically about 70 percent of person-visits
should be ?rst-time visits given other information. To avoid confusion resulting from working
in percents, the following presents results for 1,000 respondents. If about 15 percent of
visitors report coming once in the past 3 years this amounts to 150 person-visits. For the 5
percent reporting coming twice, 50 person-visits occur. For the over 3 percent reporting
coming 4þ times in the past 3 years, 30 person-visits arise. Based on arguments above
VOL. 6 NO. 1 2012
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
:
1
8
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
(equation 3), 3 £ 1000 £ 15:3% ¼ 459 ?rst-time visitors being people coming back are
required to generate 153 person-visits returning once in 3 years. However, given that when
people enter the 1-per-3-year-return category (i.e. 1 visit in last 3 years), they have a
retention rate of r, equation 2 implies that the number of 1-per-3-year-return people includes
people who have been in the group for years (made a ?rst visit 2, 3, etc. years earlier).
Given that for over 96 percent of person-visits (for about 960 person-visits) the intention is to
return, r ¼ 0.9 might be reasonable. In that case, by equation 2, one has a multiplier of 9
(r=ð1 2rÞ ¼ 0:9=ð1 20:9Þ). Therefore, to sustain 153 person-visits, only 51 ?rst-time visitors
(¼ 3 £ 1:7%£ 1000 ¼ 3 £ 15:3=9 £ 1000) become 1-per-3-year-return visitors. Applying the
same logic to 2, 3 and 4 þ visits in the last 3 years, less than 150 of ?rst-time visitors are
allocated to return categories. This is inconsistent with having 700 ?rst-time visitors. To be
consistent nearly all of the 700 ?rst-time visitors would be allocated to return categories.
Apparently, 0.9 for retention is much too large.
If retention of 0.9 is too high, what if r ¼ 1=3? Then the multiplier drops from 9 to 2
(¼ ð1=3Þ=ð1 21=3Þ). For 1-per-3-year-return, the ?rst-time visits allocated becomes 230
(¼ 3 £ 15:3=2 £ 1000). If one believes that many people reporting visiting 1 time in the past 3
years should actually be recognized as in a 1-per-4-year-return or 1-per-5-year-return
category, 230 is low. Assume that based on better data one was really con?dent that a r ¼ 1/3
allocation to categories resulted in 500 of 700 ?rst-time visitors being in return categories.
Then, only about a third of ?rst-visits are not allocated. With r ¼ 1/3, a plausible assertion is
that about a third of declarations of return were never based on serious thought. Does r ¼ 1/3
yield a good model? Unfortunately, r ¼ 1/3 and about a third of people declaring they will
return are not really thinking about returning, suggests responses about intentions give a
highly misleading impression of intention to return. If you cannot get a somewhat accurate idea
of actual return from responses and really do not have data to justify a model, what good are
the intention data other than for showing they are not much good?
Given the model formulated is reasonable, TIB data give an inconsistent picture of behavior.
Yes, the model is an approximation. Nevertheless, formulating a better model that can be
estimated with variables such as occur in the TIB data does not seem possible. Using a
reasonable model, retention that is consistent with responses is too high for the number of
?rst-time visitors returning. But, a retention (r ¼ 1/3) that allocates ?rst-time visitors
consistent with data on visits in the past three years implies over 50 percent stating intent to
return do not return. Possibilities are: that the model shows responses are not consistent with
behavior, or behavior is complex enough that behavior cannot be modeled realistically
without more data (i.e. other variables). Whether inconsistency or complexity present
problems, logically one can conclude data are not adequate for understanding return
behavior. Therefore, H1 is rejected and H2 is accepted.
Discussion
This research did not begin with the expectation of rejecting H1 and of accepting H2. Because
many countries collect data with intention to return questions similar to TIB questions and
many research projects use such questions, the expectation was that with clever modeling a
fairly accurate quantitative picture of return behavior would be derived that would support
planning, managing and marketing. Logic suggested that since statisticians and researchers
developed data collection, analysis had gone on showing that the questions being used in
surveys would yield useful and credible information. That, for example, TIB data would not
yield a consistent picture of behavior seemed unlikely. However, proceeding with this research
led to accepting ideas in HKB articles. Evidence from this research supports accepting that
intention to return questions need to be modi?ed to provide good information on return
behavior for theory building and for practical application.
A practical implication of this research is that when planning, management and marketing is
best supported by quantitative information, different data than are collected for TIB and
similar surveys are needed. Scales that give no idea about when return will occur may
support academic research. However, when a survey yields 90
þ
percent intending to return
PAGE 50
j
INTERNATIONAL JOURNAL OF CULTURE, TOURISM AND HOSPITALITY RESEARCH
j
VOL. 6 NO. 1 2012
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
:
1
8
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
but modeling suggests fewer than 50 percent return, a problem exists for quantitative
analysis. Is anything learned from yes-no responses that is adequately accurate for use in
decision making when decisions should be based on knowing visitor ?ows or changes in
visitor ?ows? What about using scales with multiple items? If multiple items yield a high
reliability in forming a scale (high alpha) and one has no idea what, say, 4.5 means in terms
of returning ever or in a fewyears, what is the value of the scale being reliable? Nevertheless,
whether one thinks about the model introduced here or models that are more elaborate
(e.g. see HKB), models are useless without data to support their use (data allowing model
estimation). Data that allowestimating models that predict numbers that can be checked are
needed to assess how well models approximate behavior. For example, the model
introduced might yield a relatively good approximation to behavior if one knew the degree to
which intention statements overstate likelihood of behavior. Having data to know if a model is
a reasonable approximation to reality is a problem for Taiwan and for any data collection
yielding TIB-like data. The important practical matter is having data that allowunderstanding
of what is occurring over time. In other words, better data are needed.
Appropriate scale of measurement of return intention
Up to this point this article is critical of questions being asked in surveys but suggestions of
questions needed to address likelihood of behavior occurring at a given time have not
occurred. To be positive, logical consideration of matters raised suggests allowing a
respondent to indicate a most likely behavior/action and its likelihood. A simple approach is
asking a respondent to check a response or responses. The following example allows for
specifying not returning or specifying an action and its likelihood. A possible question
structure for better information on returning to a destination:
Below, specify that you will not return OR select the most likely action (A) and the likelihood of that
action (B).
[ ] I’ll not return
(A) I’ll return [ ] next year [ ] within 2 years [ ] within 5 years [ ] within 10 years [ ] eventually
(B) Certainty of the action is [ ] 90%
þ
, [ ] 75% to 90% [ ] 40% to 75% [ ] 20% to 40% [ ] Less
Because this research suggests the need to move to a more quantitative approach to return,
choices exist. Inferring quantitative answers (e.g. numbers or percent returning in a time
frame) from responses such as ratings without getting speci?c information time frame for
returning may be an option. Maybe a clever model can be developed. However, based on
this research, no way is seen for making inferences from ratings without likelihood or time
frame information on returning. Movement of numbers of people in time frames are what
allowmaking ?owestimates to use to plan for arrivals or to assess the impact of programs for
which funding is to be allocated. Will using the questioning strategy suggested above work?
Maybe such questioning will yield good results. Research is needed. This research just
highlights the need to experiment with getting better information.
Modeling, measurement and data
This research has not pursued modeling with the level of sophistication of HKB and work
they cite. The reason is that data like the TIB data do not support sophisticated analysis.
Even using TIB data to examine return intentions by nation of origin and other variables was
dropped because results (e.g. cross tabulations) were accompanied by warnings that
chi-squares might not be reliable because too many cells had expected frequencies under 5
(e.g. 40 percent and more).
An important question for research is how many observations are needed and what
variables are required to start to develop a quantitative understanding of return behavior.
Thinking about this matter prompted recall of data being collected by customs and
immigration (C&I) organizations in various countries. Taiwan, Malaysia, the USA, etc. collect
passport, trip purpose, entry-exit dates, and other information. Passport numbers allow
linking data across visits. Creating trip records for a person means one can have census like
data for an interval of time. If on entry, when having visited previously is not clear, a person is
asked if they ever visited previously. As years of tracking data accumulate, developing
VOL. 6 NO. 1 2012
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
:
1
8
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
sophisticated models for segments such as leisure visitors could occur. Having hundreds of
thousands or even millions of records is reasonable so sophisticated models can be
developed. Without loading much new data collection on C&I, C&I data may prove to be a
key resource for planning, managing and marketing. Models based on their data and
special studies can yield accurate information to which special studies can be linked. For
example, integrating some sample based TIB-like collection into data collection by C&I
(using existing automation) is a way to calibrate and/or check special surveys.
Implications of research for theory and practice
Maybe, cases exist in tourismresearch in which scales like 23 ¼ not very likely to return and
þ3 ¼ very likely to return are of value. For consumer products like laundry soap, repurchase
and brand-switching relates to how long an average consumer takes to use a quantity of a
product. Continued use of the product, at least in a reasonably long time frame is not a given
for tourism products. Since tourism products are not like laundry/dish soap and many other
consumer products (Gitelson and Crompton, 1984), special analysis considerations must
occur. Given that this research suggests that over half of Taiwan’s inbound visitors stating
very likely to return do not return and given one has no idea when return will occur, is
conventional intention-to-return information good for developing theory or making planning,
management and marketing decisions?
Realistically, for planning, management and marketing, should one use change in a scale
value computed for a mean of a single or multiple scale to make decisions about programs
working or actions needed? Such measurement of change is some what like measuring
change in temperature without knowing what temperatures correspond to either boiling or
freezing. If you measured in Celsius, Fahrenheit and Kevin, you would get three different rates
of change. Is one correct? No. All are correct but a rate only means something when correctly
associated with the scale being used. And, those scales are only de?ned when two points on
them have speci?c values. A thermometer is only calibrated when, for example, freezing and
boiling are associated with appropriate values. Making decisions when one has no idea of
volume of return because, for example, one does not know about return once, twice or three
times in the next three years is making decisions without relating a measure to ?ows/volumes
that affect such matters as transportation capacity, accommodation needed and viability of
service businesses. If one is going to collect data, the target should be having information
elucidating volumes of visitors and impacts relating to goals (e.g. supporting regional
development). Current theory and methodology is not suf?cient to support decision making.
Yes, this research largely focuses attention on matters that need to be addressed by theory
development so that practical matters can be analyzed. However, an important implication is
that resources are being poorly expended in much research on tourist loyalty and return.
Data collected are not appropriate to develop an understanding of the dynamics of return.
The critical matter documented is the need for a change in research practices and data
collection practices regarding intentions of tourists to return.
References
Baker, D.A. and Crompton, J.L. (2000), ‘‘Quality, satisfaction and behavioral intentions’’, Annals of
Tourism Research, Vol. 27 No. 3, pp. 785-803.
Baloglu, S. and Erickson, R.E. (1998), ‘‘Destination loyalty and switching behavior of travelers: a Markov
analysis’’, Tourism Analysis, Vol. 2 No. 2, pp. 119-28.
Baloglu, S. and Erickson, R.E. (1999), ‘‘Reply to comment on destination loyalty and switching behavior
of travelers: a Markov analysis’’, Tourism Analysis, Vol. 3 No. 2, pp. 117-8.
Beaman, J.G., Huan, T.C. and Kozak, M. (2002), ‘‘Estimating a Markov model that incorporates ?rst visit
decisions and varying repeat frequency’’, Tourism Analysis, Vol. 6 No. 2, pp. 81-97.
Boyd, B.K., Gove, S. and Hitt, M.A. (2005), ‘‘Construct measurement in strategic management research:
illusion or reality?’’, Strategic Management Journal, Vol. 26, pp. 239-57.
Darnell, A.C. and Johnson, P.S. (2001), ‘‘Repeat visits to attractions: a preliminary economic analysis’’,
Tourism Management, Vol. 22 No. 2, pp. 119-26.
PAGE 52
j
INTERNATIONAL JOURNAL OF CULTURE, TOURISM AND HOSPITALITY RESEARCH
j
VOL. 6 NO. 1 2012
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
:
1
8
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
Decrop, A. (2001), ‘‘The antecedents and consequences of vacationers’ dis/satisfaction: tales from the
?eld’’, in Mazanec, J.A., Crouch, G.I., Ritchie, J.R.B. and Woodside, A.G. (Eds), Consumer Psychology
of Tourism, Hospitality and Leisure, Vol. 2, Cabi Publishing, Wallingford, pp. 333-47.
Fakeye, P.C. and Crompton, J.L. (1991), ‘‘Image differences between prospective, ?rst-time, and repeat
visitors to the Lower Rio Grande Valley’’, Journal of Travel Research, Vol. 30 No. 2, pp. 10-16.
Gitelson, R.J. and Crompton, J.L. (1984), ‘‘Insights into the repeat vacation phenomenon’’, Annals of
Tourism Research, Vol. 11 No. 2, pp. 199-217.
Haig, M. (2006), Brand Royalty: Howthe World’s Top 100 Brands Thrive and Survive, Kogan Page, London.
Hsu, C.H.C. and Kang, S.K. (2007), ‘ ‘CHAID-based segmentation: international visitors’ trip
characteristics and perceptions’’, Journal of Travel Research, Vol. 46 No. 2, pp. 207-16.
Huan, T.C. and Beaman, J.G. (2003a), ‘‘Individual and dynamics of social interaction in information search
in discretionary travel decision’’, in Crouch, G.I., Perdue, R.R., Timmermans, H. and Uysal, M. (Eds),
Consumer Psychology of Tourism, Hospitality and Leisure, Vol. 3, La Trobe University, Cabi Publishing,
Wallingford, pp. 149-59.
Huan, T.C. and Beaman, J.G. (2003b), ‘‘Contexts and dynamics of social interaction and information
search in decision-making for discretionary travel’’, Tourism Analysis, Vol. 8 Nos 2-4, pp. 177-82.
Huan, T.C., Kozak, M. and Beaman, J.G. (2003), ‘‘Implications of models of repeat travel for the concept
of destination loyalty’’, Leisure and Society, Vol. 26 No. 1, pp. 183-207.
Kozak, M. (2001), ‘‘Repeaters’ behavior at two distinct destinations’’, Annals of TourismResearch, Vol. 28
No. 3, pp. 785-808.
Kozak, M., Huan, T.C. and Beaman, J.G. (2002), ‘‘A systematic approach to non-repeat and repeat
tourism with measurement and destination loyalty concept implications’’, Journal of Travel and Tourism
Marketing, Vol. 12 No. 4, pp. 19-38.
Loo, R. (2002), ‘‘A caveat on using single-item versus multiple-item scales’’, Journal of Managerial
Psychology, Vol. 17 No. 1, pp. 68-75.
Oliver, R.L. (1999), ‘‘Whence consumer loyalty’’, Journal of Marketing, Vol. 63, pp. 33-44.
Oppermann, M. (1998a), ‘‘Destination loyalty and switching behavior of travelers revisited’’, Tourism
Analysis, Vol. 3 No. 1, pp. 115-6.
Oppermann, M. (1998b), ‘‘Destination threshold potential and the law of repeat visitation’’, Journal of
Travel Research, Vol. 37 No. 2, pp. 131-7.
Pritchard, M.P. and Howard, D.R. (1997), ‘‘The loyal traveler: examining a typology of service
patronage’’, Journal of Travel Research, Vol. 35 No. 4, pp. 2-10.
Sivadas, E. and Baker-Prewitt, J.L. (2000), ‘‘An examination of the relationship between service quality,
customer satisfaction and store loyalty’’, International Journal of Retail and Distribution Management,
Vol. 28 No. 2, pp. 73-82.
Taiwan Tourism Bureau (2001), 2000 Annual Survey Report on Visitors Expenditure and Trend in Taiwan,
Taiwan Tourism Bureau, Taipei.
Taiwan Tourism Bureau (2002), 2001 Annual Survey Report on Visitors Expenditure and Trend in Taiwan,
Taiwan Tourism Bureau, Taipei.
Taiwan Tourism Bureau (2004), 2003 Annual Survey Report on Visitors Expenditure and Trend in Taiwan,
Taiwan Tourism Bureau, Taipei.
Yu¨ ksel, A. (2001), ‘‘Managing customer satisfaction and retention: a case of holiday destinations,
Turkey’’, Journal of Vacation Marketing, Vol. 7 No. 2, pp. 153-68.
Corresponding author
Tang-Chung Kan can be contacted at: [email protected]
VOL. 6 NO. 1 2012
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
:
1
8
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
This article has been cited by:
1. Wu?Chung Wu, You?De Dai, Hsiou?Hsiang Jack Liu. 2012. Effective information collection on international inbound visitors.
International Journal of Culture, Tourism and Hospitality Research 6:1, 54-69. [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
:
1
8
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
doc_342126463.pdf
This paper has two main aims: to show responses like yes or very likely for inbound visitors
returning to a destination can lead to misleading and unreliable information; and to clarify the kind of
information that should be collected
International Journal of Culture, Tourism and Hospitality Research
Inadequate return questions: return when? Sometime?
Ching-Tang Wang Tzung-Cheng Huan Tang-Chung Kan
Article information:
To cite this document:
Ching-Tang Wang Tzung-Cheng Huan Tang-Chung Kan, (2012),"Inadequate return questions: return when? Sometime?", International
J ournal of Culture, Tourism and Hospitality Research, Vol. 6 Iss 1 pp. 44 - 53
Permanent link to this document:http://dx.doi.org/10.1108/17506181211206243
Downloaded on: 24 January 2016, At: 22:18 (PT)
References: this document contains references to 26 other documents.
To copy this document: [email protected]
The fulltext of this document has been downloaded 272 times since 2012*
Users who downloaded this article also downloaded:
Thomas Foscht, Karin Ernstreiter, Cesar Maloles, Indrajit Sinha, Bernhard Swoboda, (2013),"Retaining or returning?: Some insights for a
better understanding of return behaviour", International J ournal of Retail & Distribution Management, Vol. 41 Iss 2 pp. 113-134 http://
dx.doi.org/10.1108/09590551311304310
Lloyd C. Harris, (2010),"Fraudulent consumer returns: exploiting retailers' return policies", European J ournal of Marketing, Vol. 44 Iss 6 pp.
730-747http://dx.doi.org/10.1108/03090561011032694
Wu-Chung Wu, You-De Dai, Hsiou-Hsiang J ack Liu, (2012),"Effective information collection on international inbound visitors", International
J ournal of Culture, Tourism and Hospitality Research, Vol. 6 Iss 1 pp. 54-69http://dx.doi.org/10.1108/17506181211206252
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
:
1
8
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
Inadequate return questions: return when?
Sometime?
Ching-Tang Wang, Tzung-Cheng Huan and Tang-Chung Kan
Abstract
Purpose – This paper has two main aims: to show responses like yes or very likely for inbound visitors
returning to a destination can lead to misleading and unreliable information; and to clarify the kind of
information that should be collected.
Design/methodology/approach – Responses from Taiwan’s inbound visitors relating to returning are
examined to see what can be learned. Modeling is used to extract meaningful quantitative information
from data.
Findings – Modeling shows that survey responses about return are inconsistent. Although 95 percent of
non-visiting-friends-and-relations (VFR) leisure visitors indicate returning, this is not consistent with a
retention rate of 90 percent. A retention rate of 33 percent is consistent with the observation that 70
percent of person-visits are ?rst-visits. However, 33 percent retention is not consistent with over 95
percent of visitors returning. Conventional questions are yielding highly unreliable information and,
therefore, data collection should be changed.
Originality/value – Relations between vague questions and return trips have been established. This
research provides new evidence of the need for return data to include information allowing estimation of
volume and timing of return.
Keywords Return intentions, Taiwan, Inbound, Survey questions, Tourism management,
Social dynamics
Paper type Research paper
Introduction
In this research, the product of concern is a leisure visit to a foreign country that is not a
visiting-friends-and-relation (VFR) visit. Using common terminology the research is on
inbound leisure visiting that is not VFR. The caveat that the research is for leisure visits that
are not VFR is not continually repeated since, as the previous sentences suggest, repetition
can be tedious. However, periodic reference to leisure visits occurs to remind the reader of
the focus. This is done because, by de?nition, repeat purchase (visits) for visiting friends and
relation, for conferences and for business are driven by different mechanisms than
discretionary leisure travel.
Hsu and Kang (2007) cite the need to consider ideas regarding repeat visiting such as
raised in Huan et al. (2003) and Kozak et al. (2002). Hsu and Kang (2007) recognize much
research deals with return using response to questions like ‘‘do you plan to return?’’. Huan
et al. (2003), Kozak et al. (2002) and Beaman et al. (2002), hereafter HKB, focus attention on
a person’s yes-no response or response to a rating (e.g. not likely to highly likely) not
providing good information for understanding behavior. For example, very likely to return
does not clarify how likely or when return is being contemplated. Is returning sometime, say,
90 percent certain?
PAGE 44
j
INTERNATIONAL JOURNAL OF CULTURE, TOURISM AND HOSPITALITY RESEARCH
j
VOL. 6 NO. 1 2012, pp. 44-53, Q Emerald Group Publishing Limited, ISSN 1750-6182 DOI 10.1108/17506181211206243
Ching-Tang Wang is an
Associate Professor in the
Department of Sport
Management, National
Taiwan College of Physical
Education, Puzi City,
Taiwan. Tzung-Cheng Huan
is Dean of the College of
Management, National
Chiayi University, Chiayi
City, Taiwan. Tang-Chung
Kan is Head of the
Department of Travel
Management, National
Kaohsiung Hospitality
College, Kaohsiung City,
Taiwan.
Received September 2009
Revised April 2010
Accepted July 2010
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
:
1
8
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
If research is to build theory or to support planning, managing or marketing, are implications
of the research limited without addressing likelihood of returning in a time frame? With yes to
returning people who are 90 percent sure to return sometime in their life are lumped with
people 90 percent sure to return in the next couple of years. With such data, conclusions that
can be reached about volume of return in any year are limited. If yes or almost certainly are
taken to mean 90 percent or more likely to return but 50 percent or less is the return rate, is
theory sound? With a consumer product like dishwasher soap, some people may take a lot
longer to use some unit/quantity of the product than other people. Nevertheless, a
researcher can expect that most people currently using some brand of, for example,
dishwasher soap that is regularly purchased will continue using that dishwasher soap for
some time (e.g. for the next few years). Therefore, knowing about loyalty in making brand
choices has practical implications (re consumer products and brand see references in
Oliver, 1999; Sivadas and Baker-Prewitt, 2000; Haig, 2006). In the context just given, as of
2009, reason suggests intention to buy again, to return, should be addressing time frame
and/or likelihood of the next purchase.
Ideas in the previous paragraphs should provoke thinking about why intention-to-return data
are collected. Now that research relates satisfaction, loyalty and intention (Baker and
Crompton, 2000; Darnell and Johnson, 2001; Fakeye and Crompton, 1991; Gitelson and
Crompton, 1984; Hsu and Kang, 2007; Kozak, 2001; Oppermann, 1998a; Pritchard and
Howard, 1997; Yu¨ ksel, 2001), considering when respondents may return and likelihood of
return is appropriate. Consider that knowing an answer of yes means a 70 percent chance of
actually returning and a 50 percent chance of returning in two years. If this allows making
inferences about return ?ows, making estimates has practical value. One ?nds some
modeling ideas regarding repeating visitors (e.g. Baloglu and Erickson, 1998, 1999;
Oppermann, 1998b). However, HKB make arguments that some modeling by Markov
models is misleading. What can be learned from intention data such as those collected from
Taiwan’s inbound tourists (TIB data) is unclear. In fact, no research was found pursuing
whether creative thinking and modeling can yield models allowing useful inferences to be
made from TIB responses relating to return. Rather than looking at ideal data for analysis,
this research examines what can be done with responses relating to returning such as those
found in TIB data.
Objectives, hypotheses, methodology and terminology
Many countries collect intention and return data similar to those collected from inbound
visitors in Taiwan. Studies of returning acknowledge the importance of being a ?rst-time
visitor as opposed to a repeat visitor (e.g. HKB: Fakeye and Crompton, 1991). Examination
of surveys in which returning is considered (see citations and references in those) shows a
person is usually asked if they will or are likely to return and are expected to answer by
yes-no or a rating from very unlikely to very likely. In some research, a person is simply
asked if they are a ?rst-time visitor. In other cases, a respondent speci?es the number of
past visits ever or in a ?xed time frame. For TIB data used the time frame is the past three
years (i.e. in the past three years how many times have you visited Taiwan). In some data,
for example in the three years of TIB data used for model development, no information is
obtained on having visited prior to the time frame (data on ever visited are in a different TIB
survey). For TIB surveys without visited-ever information, ?rst-time visitors and repeat
visitors cannot be distinguished. For a visitor who has not come in the last three years, all
one knows is that they are ?rst-time or came to Taiwan more than three years prior to
becoming a TIB survey respondent.
One reason that this research examines existing data to see what can be learned is that
sound development of ideas on data collection is facilitated by understanding what can be
learned with data that have been collected. The general objective of this research is to learn
what useful information about the ?ow of returning over time can be extract fromdata like the
TIB data. The research objective is manifested in H1 and H2:
H1. Data with intention and return variables, such as in TIB data, will yield credible and
useful information on repurchase behavior of tourists.
VOL. 6 NO. 1 2012
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
:
1
8
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
H2. Analysis of what can be learned from data such as TIB data will yield speci?c
guidance/criteria for data collection that produces data that are useful in planning,
management and marketing.
Terminology: credible and useful
A reasonable basis for saying that a response to a question or responses to questions are
useful is that the information is good for testing hypotheses and/or they facilitate sound
decision making. However, if data have been used to test hypotheses and hypotheses are
established (i.e. as valid), how much testing is useful? With enough experiments, data from
one experiment will not con?rm some hypothesis (allow its acceptance). For example, for
signi?cance just at the 5 percent level about one replication in 20 of an experiment will result
in failure to con?rm an hypothesis. Developing theory involves more than repeating
experiments. Furthermore, in the context of a government using information, say the
government of Taiwan, collecting data should relate to using the information in planning,
management or marketing. In the case of asking if inbound tourists will return, knowing the
percent likely to return may appear to be of obvious value. However, assume year after year
the percent specifying returning is 96 percent or above (see Table I). Do you need the return
intention data year after year? Why ask about visits in the last three years (e.g. with
categories 1, 2, 3, or 4 visits and 5
þ
visits), assuming planning, management or marketing
use of responses, if proportions in categories change little from year to year? In fact for any
variable, if negligible change occurs over time and collecting data to measure change has a
signi?cant cost, why pay the collection price annually? Is money spent on collection well
used if information collected is not really needed? If information is not changing much or no
signi?cant contribution to meeting needs of academics or managers occurs for some other
reason, resources spent on the information can likely be put to better use.
Suggesting that commonly occurring data collection is ineffective may seem cynical.
Actually, in searches as part of this research, no planning, management or marketing
justi?cation has been found for asking about number of visits in the last three years. Does
some rationale exist for three years? None has been found. Would asking about visits in the
past ?ve years or past two years be better? No literature has been found addressing this
matter. In fact, given increased tourist ?ow from the mainland to Taiwan, change in the
distribution of TIB responses about visits during the past three years will occur. Oppermann
(1998b) and Baloglu and Erickson (1998, 1999) address repeat frequency and data
requirements. However, Darnell and Johnson (2001) and Beaman et al. (2002) are the
articles found that have raised the matter of having adequate data to interpret change.
Consider what useful means in another context. Having information about how responses to
TIB survey questions play a role in speci?c planning, management and marketing decision
making would be useful. As of 2009, searching for such information is yielding no results.
What is the situation if planning, managing and marketing are being based on a 97 percent
intention to return rate and intention does not match reality? If intention is 97 percent and
reality is 90 percent, is that OK? What if with 97 percent intention, return turns out to be 50
percent for ?rst-time visitors? If marketing targets ?rst-time visitors based on keeping a high
Table I Summary information from general questions on intention to return
Frequency of visits during last
three years
Percent 2001 and
2003
Percent
2002
Percent intend return 2001 and 2003
by last three years
Year of last
visit
Percent
2002
0 75.0 77.25 96.2 First visit 77.25
1 15.3 13.27 97.1 1 11.10
2 5.1 4.35 98.1 2 3.54
3 1.3 1.37 100 3 1.37
4 þ 3.3 2.90 99.3 . 3 years 5.89
Notes: Since respondents are selected at random, every visit during a year must be viewed as having equal probability of selection. Therefore,
tabulation of records without considering likely number of visits yields results in person-visits. To get persons one could correct for expected
number of visits in a survey year using one-third of visits in the past three years when visits in the past three years is greater than one
PAGE 46
j
INTERNATIONAL JOURNAL OF CULTURE, TOURISM AND HOSPITALITY RESEARCH
j
VOL. 6 NO. 1 2012
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
:
1
8
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
return and return is not high, are resources being used well? Fromthe point of viewof making
good decisions, are data collected supporting making good decisions or evaluating what is
being achieved by, for example, resources spent on marketing? How do you measure
success of marketing when you do not know what percent of this year’s ?rst-time visitors, or
even what percent of all current year visitors, can be expected to return within, say, three
years? If you cannot predict and measure marketing performance and plan or manage
based on data being collected, logic suggests modifying data collection so results are more
useful.
In the last paragraph, ideas relating to data being useful and credible are both present. If
you collect intention to return data and over 95 percent intend to return based on survey
responses, do you assume that the 95 percent re?ects actual returning that will occur? If
returning ever is just, say, 60 percent, does intention correspond with reality? This research
addresses that matter.
People involved with scales of return (e.g. 7 ¼ very likely), may argue that asking yes-no
regarding intending to return is asking for problems because single item measures are not
as reliable as multiple item measures. Discussion is going on in the literature about the
problem of basing measurement on a single question (Boyd et al., 2005; Loo, 2002). The
scale questions below are hypothetical but follow the pattern of intention questions in the
literature. The idea is that a measure based on three questions relating to a concept will be
more accurate than a single measure. For some measurement problems, forming a scale
based on multiple questions is certainly appropriate. However, does one get better
information on intention to return by multiple questions and forming, for example, an additive
scale?
Answering questions such as those above by 1 (strongly disagree) to 5 (strongly agree) and
forming an additive scale may seem appropriate to many researchers who work with scales.
For an additive scale, Cronbach’s alpha may show acceptable or even high reliability.
However, one should not confuse reliability as measured by alpha with a measure conveying
an accurate understanding of behavior. Say that a value of 96 percent of ?rst-time visitors
stating yes to intention of returning corresponds to 50 percent, or even 70 percent returning
sometime later. Does a highly reliable mean of 4.5 have a less ambiguous meaning than a
mean of 97 percent for yes that is consistent from year to year and has standard error of
about 0.1 percent for any year of TIB data? How do respondents rate scale anchors? Does a
relation exist between scale values and a practically useful measure such as percent of
?rst-time visitors returning in, for example, two years? How accurate is the relation? Actually,
how do you calibrate a scale value without relatively accurate information on return?
Returning is observable. Of?cial in Taiwan and other countries record entry and exit of
inbound visitors using unique identi?ers (i.e. passport information). Is returning really
something that should be learned about by probing the psyche using a scale? Ratings
questions that approach returning obtusely do not give speci?c information about likelihood
or timeframe. Given some people actually return, logical reasoning suggests scale reliability
should not be confused with information being useful and credible for meeting quantitative
needs of planners, managers and marketers for numbers of visitors in?uenced by marketing
or services.
The position that HKB supports is that a person may be uncertain about the future but insight
gained by asking vague, obtuse and/or possibly misleading questions is not likely to provide
as good of information for quantitative analysis of behavior as questions about what is likely
to be done and when. For example, survey questions (e.g. see the following list) may not be
relevant when going somewhere involves negotiation (e.g. family trips or travel with friends).
Revisit intention scale questions (hypothetical):
1. This trip has shown me that X is a good place to come to.
2. Among alternative trips, coming back to X is a high priority.
3. I know I’d enjoy myself on a return trip to X.
VOL. 6 NO. 1 2012
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
:
1
8
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
Decrop (2001) and Huan and Beaman (2003a, b) discuss such matters. Asking a person
who goes somewhere that does not particularly interest her about satisfaction without asking
about why a trip was taken means analysis will contain variance that limits reaching clear
conclusions. For Taiwan’s inbound, a high percent are traveling with friends or family so what
they do and intent to return may have a limited relation to activities, satisfaction, meeting
expectations, etc. Given that decisions about planning, managing and marketing should be
based on visitor ?ows, their expenditures, etc. credibility and usefulness of data to support
such decisions should be assessed based on speci?c use of statistics in meeting decision
making needs.
Method
Analysis is based on modeling. Introducing the TIB data for analysis is a ?rst step. Then a
model compatible with the TIB data is formulated. The model is based on plausible ideas for
approximating how behavior occurs. Then analysis of TIB data occurs. The article provides
conclusions and implications of the research for data collection, for theory development and
practical consideration of tourist return.
Data
Taiwan Tourism Bureau (TTB) uses the exit survey of inbound visitors to understand the
motives, viewpoints, tendencies and consumption of the inbound visitors. Reports
(e.g. Taiwan Tourism Bureau, 2001, 2002, 2004) are prepared as reference material for
tourist related organizations to plan for international tourism. The survey is also a source for
estimating the amount of visitor expenditures.
Inbound survey data are collected at exit points from Taiwan using a questionnaire. Design
of the questionnaire and its administration is by contract with a university statistical institute
(see TTB annual reports in the references for details). Since 1995, sampling has always
involved quota sampling. During a year total interviews are about 5,000.
Questions asked vary somewhat from year to year. In 2000, 2001 and 2003 respondents
were asked: ‘‘do you think you will visit Taiwan again?’’ The answer allowed was yes or no.
Another question asked in various ways is times Taiwan was visited in the past three years. In
2000 and 2001 years responses are for 1 to 4 times and 5 þ (though English translation of 5
or more as ‘‘over 5 times’’ are used). In 2003 the respondents could report actual numbers of
visits in excess of 5. Unfortunately, in 2000, 2001, 2003 information that lets one know if a
person visited more than 3 years previously is not available. Information is available for 2002.
Other than to estimate the proportion of visitors who came more than three years ago, 2002
data are not used because they were collected by a different contractor than 2000, 2001 and
2003 data using a questionnaire structure that is structured somewhat differently from the
questionnaire for the data used.
Model formulation
Given one has a limited travel history for people, modeling must build on that. For the
moment, think of people who are ?rst time visitors. In terms of intention to return, intending or
not intending to return is what is known. Therefore, for TIB data for 2000, 2001 and 2003,
times visiting in the past three years is also a variable. For the years mentioned, one has no
information about visiting prior to the three-year horizon. Therefore, ?rst-time visitors and
those returning visitors who would respond that they came more than 3 years prior to giving
data are in one category. Regardless, initially let N
k
, k.0, be the number of people reporting
coming k times in last n years. Since one does not know travel history, think of this being a
group of people that is fed from new visitors. For k =3, think of people tending to come once
in three years. In stochastic modeling for annual visiting this corresponds to a Poisson
variable with a mean of a third. For such modeling, people coming once in the past three
years would have a certain probability of being people tending to come every year, every
two years, every three years, etc. (means of 1, 1/2, 1/3, etc.).
PAGE 48
j
INTERNATIONAL JOURNAL OF CULTURE, TOURISM AND HOSPITALITY RESEARCH
j
VOL. 6 NO. 1 2012
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
:
1
8
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
Equation 1 re?ects the idea that people enter N
k
from being ?rst time visitors and remain in
N
k
except for retention being less than 1 (r,1). Equation 2 just expresses equation 1 using
the sum of the series of r
t
:
N
k
¼ rh þ Sr
t
h
where t ¼ 2 to some large number ðsymbolically 1; infinityÞ:
ð1Þ
N
k
¼ Sr
t
h ¼ rh=ð1 2rÞ
where in the sum the exponent t ¼ 1 to 1:
ð2Þ
Retention r , 1 should be taken to imply that people in a group on average show annual
decreased return over time of 1-r. In other words, a 10 percent decrease from year to year of
r ¼ 0.9 re?ects decreased visiting from a group. Part of the decrease can be from people
stopping visiting because they die. However, in general terms this is a duration effect.
People switch destinations, travel less or do not travel for various reasons (e.g. see
Oppermann, 1998a). Given the data available having duration effects associated with
changes in categories (Beaman et al., 2002) does not work because the data does not allow
model estimation. Nevertheless, allowing retention accommodates variability in behavior
that is modeled directly in more complicated models. An aggregate duration effect (e.g. r) is
a way of approximating changing behavior becoming more or less frequently. In a model in
which age is a variable, mortality can explain some decrease in coming. However, just as
many linear models are used to approximate reality, this model is proposed as a reasonable
model to approximate return behavior.
Confusion can arise because in exit surveys N
k
is not unique people. The unit of N
k
is person
trips (e.g. see note with Table I). For ?rst-time visitors, numbers of persons equals number of
visits. However, if X ?rst-time visitors become members of a k times in three years group, the
X ?rst-time visits can only be expected to make h visits per year (h¼X/k). In general for visits
in n years, think about an average of k visits per year as having k =m/n with m as a response
for visits in n years. X new visitors with k visits for year will only yield h ¼ X/k visits per year
with retention of 1(r ¼ 1). In other words, ?rst-time visitors who return and feed repeat
groups and have retention of r, satisfy equation 3.
X
k
¼ r kh
k
were variables are as specified above
ð3Þ
Analysis and results
Challenges exist in applying the model because data are needed to make estimates. Let N
r
be
the number of ?rst-time visitors, say N, minus ?rst-time visitors not intending to return visitors.
One challenge is determining how to partition ?rst-time visitors intending to return, N
r
, into
categories, k. However, for TIB data being used N is not known because people coming more
than three years prior to responding to the inbound survey are not distinguished from?rst-time
visitors. In other words, how to de?ne h
k
based on N
r
offers a challenge.
To pursue this challenge, Table I gives data on the variable visits in the past three years and for
some other variables. Note again that for visits in the past three years, ?rst-time and repeat
visitors who have not come in the past three years are in the same category. However, in 2002
and some other years of TIBdata one has information such as year of last visit or reason for last
visit. For example, for 2002 only about 6 percent indicated a last visit more than three years in
the past. Accepting 6 percent, about 70 percent of person-visits are ?rst visits.
For intention to return, data can be deceptive because people’s intentions do not necessarily
correspond with behavior. Consider, whether logically about 70 percent of person-visits
should be ?rst-time visits given other information. To avoid confusion resulting from working
in percents, the following presents results for 1,000 respondents. If about 15 percent of
visitors report coming once in the past 3 years this amounts to 150 person-visits. For the 5
percent reporting coming twice, 50 person-visits occur. For the over 3 percent reporting
coming 4þ times in the past 3 years, 30 person-visits arise. Based on arguments above
VOL. 6 NO. 1 2012
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
:
1
8
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
(equation 3), 3 £ 1000 £ 15:3% ¼ 459 ?rst-time visitors being people coming back are
required to generate 153 person-visits returning once in 3 years. However, given that when
people enter the 1-per-3-year-return category (i.e. 1 visit in last 3 years), they have a
retention rate of r, equation 2 implies that the number of 1-per-3-year-return people includes
people who have been in the group for years (made a ?rst visit 2, 3, etc. years earlier).
Given that for over 96 percent of person-visits (for about 960 person-visits) the intention is to
return, r ¼ 0.9 might be reasonable. In that case, by equation 2, one has a multiplier of 9
(r=ð1 2rÞ ¼ 0:9=ð1 20:9Þ). Therefore, to sustain 153 person-visits, only 51 ?rst-time visitors
(¼ 3 £ 1:7%£ 1000 ¼ 3 £ 15:3=9 £ 1000) become 1-per-3-year-return visitors. Applying the
same logic to 2, 3 and 4 þ visits in the last 3 years, less than 150 of ?rst-time visitors are
allocated to return categories. This is inconsistent with having 700 ?rst-time visitors. To be
consistent nearly all of the 700 ?rst-time visitors would be allocated to return categories.
Apparently, 0.9 for retention is much too large.
If retention of 0.9 is too high, what if r ¼ 1=3? Then the multiplier drops from 9 to 2
(¼ ð1=3Þ=ð1 21=3Þ). For 1-per-3-year-return, the ?rst-time visits allocated becomes 230
(¼ 3 £ 15:3=2 £ 1000). If one believes that many people reporting visiting 1 time in the past 3
years should actually be recognized as in a 1-per-4-year-return or 1-per-5-year-return
category, 230 is low. Assume that based on better data one was really con?dent that a r ¼ 1/3
allocation to categories resulted in 500 of 700 ?rst-time visitors being in return categories.
Then, only about a third of ?rst-visits are not allocated. With r ¼ 1/3, a plausible assertion is
that about a third of declarations of return were never based on serious thought. Does r ¼ 1/3
yield a good model? Unfortunately, r ¼ 1/3 and about a third of people declaring they will
return are not really thinking about returning, suggests responses about intentions give a
highly misleading impression of intention to return. If you cannot get a somewhat accurate idea
of actual return from responses and really do not have data to justify a model, what good are
the intention data other than for showing they are not much good?
Given the model formulated is reasonable, TIB data give an inconsistent picture of behavior.
Yes, the model is an approximation. Nevertheless, formulating a better model that can be
estimated with variables such as occur in the TIB data does not seem possible. Using a
reasonable model, retention that is consistent with responses is too high for the number of
?rst-time visitors returning. But, a retention (r ¼ 1/3) that allocates ?rst-time visitors
consistent with data on visits in the past three years implies over 50 percent stating intent to
return do not return. Possibilities are: that the model shows responses are not consistent with
behavior, or behavior is complex enough that behavior cannot be modeled realistically
without more data (i.e. other variables). Whether inconsistency or complexity present
problems, logically one can conclude data are not adequate for understanding return
behavior. Therefore, H1 is rejected and H2 is accepted.
Discussion
This research did not begin with the expectation of rejecting H1 and of accepting H2. Because
many countries collect data with intention to return questions similar to TIB questions and
many research projects use such questions, the expectation was that with clever modeling a
fairly accurate quantitative picture of return behavior would be derived that would support
planning, managing and marketing. Logic suggested that since statisticians and researchers
developed data collection, analysis had gone on showing that the questions being used in
surveys would yield useful and credible information. That, for example, TIB data would not
yield a consistent picture of behavior seemed unlikely. However, proceeding with this research
led to accepting ideas in HKB articles. Evidence from this research supports accepting that
intention to return questions need to be modi?ed to provide good information on return
behavior for theory building and for practical application.
A practical implication of this research is that when planning, management and marketing is
best supported by quantitative information, different data than are collected for TIB and
similar surveys are needed. Scales that give no idea about when return will occur may
support academic research. However, when a survey yields 90
þ
percent intending to return
PAGE 50
j
INTERNATIONAL JOURNAL OF CULTURE, TOURISM AND HOSPITALITY RESEARCH
j
VOL. 6 NO. 1 2012
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
:
1
8
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
but modeling suggests fewer than 50 percent return, a problem exists for quantitative
analysis. Is anything learned from yes-no responses that is adequately accurate for use in
decision making when decisions should be based on knowing visitor ?ows or changes in
visitor ?ows? What about using scales with multiple items? If multiple items yield a high
reliability in forming a scale (high alpha) and one has no idea what, say, 4.5 means in terms
of returning ever or in a fewyears, what is the value of the scale being reliable? Nevertheless,
whether one thinks about the model introduced here or models that are more elaborate
(e.g. see HKB), models are useless without data to support their use (data allowing model
estimation). Data that allowestimating models that predict numbers that can be checked are
needed to assess how well models approximate behavior. For example, the model
introduced might yield a relatively good approximation to behavior if one knew the degree to
which intention statements overstate likelihood of behavior. Having data to know if a model is
a reasonable approximation to reality is a problem for Taiwan and for any data collection
yielding TIB-like data. The important practical matter is having data that allowunderstanding
of what is occurring over time. In other words, better data are needed.
Appropriate scale of measurement of return intention
Up to this point this article is critical of questions being asked in surveys but suggestions of
questions needed to address likelihood of behavior occurring at a given time have not
occurred. To be positive, logical consideration of matters raised suggests allowing a
respondent to indicate a most likely behavior/action and its likelihood. A simple approach is
asking a respondent to check a response or responses. The following example allows for
specifying not returning or specifying an action and its likelihood. A possible question
structure for better information on returning to a destination:
Below, specify that you will not return OR select the most likely action (A) and the likelihood of that
action (B).
[ ] I’ll not return
(A) I’ll return [ ] next year [ ] within 2 years [ ] within 5 years [ ] within 10 years [ ] eventually
(B) Certainty of the action is [ ] 90%
þ
, [ ] 75% to 90% [ ] 40% to 75% [ ] 20% to 40% [ ] Less
Because this research suggests the need to move to a more quantitative approach to return,
choices exist. Inferring quantitative answers (e.g. numbers or percent returning in a time
frame) from responses such as ratings without getting speci?c information time frame for
returning may be an option. Maybe a clever model can be developed. However, based on
this research, no way is seen for making inferences from ratings without likelihood or time
frame information on returning. Movement of numbers of people in time frames are what
allowmaking ?owestimates to use to plan for arrivals or to assess the impact of programs for
which funding is to be allocated. Will using the questioning strategy suggested above work?
Maybe such questioning will yield good results. Research is needed. This research just
highlights the need to experiment with getting better information.
Modeling, measurement and data
This research has not pursued modeling with the level of sophistication of HKB and work
they cite. The reason is that data like the TIB data do not support sophisticated analysis.
Even using TIB data to examine return intentions by nation of origin and other variables was
dropped because results (e.g. cross tabulations) were accompanied by warnings that
chi-squares might not be reliable because too many cells had expected frequencies under 5
(e.g. 40 percent and more).
An important question for research is how many observations are needed and what
variables are required to start to develop a quantitative understanding of return behavior.
Thinking about this matter prompted recall of data being collected by customs and
immigration (C&I) organizations in various countries. Taiwan, Malaysia, the USA, etc. collect
passport, trip purpose, entry-exit dates, and other information. Passport numbers allow
linking data across visits. Creating trip records for a person means one can have census like
data for an interval of time. If on entry, when having visited previously is not clear, a person is
asked if they ever visited previously. As years of tracking data accumulate, developing
VOL. 6 NO. 1 2012
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
:
1
8
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
sophisticated models for segments such as leisure visitors could occur. Having hundreds of
thousands or even millions of records is reasonable so sophisticated models can be
developed. Without loading much new data collection on C&I, C&I data may prove to be a
key resource for planning, managing and marketing. Models based on their data and
special studies can yield accurate information to which special studies can be linked. For
example, integrating some sample based TIB-like collection into data collection by C&I
(using existing automation) is a way to calibrate and/or check special surveys.
Implications of research for theory and practice
Maybe, cases exist in tourismresearch in which scales like 23 ¼ not very likely to return and
þ3 ¼ very likely to return are of value. For consumer products like laundry soap, repurchase
and brand-switching relates to how long an average consumer takes to use a quantity of a
product. Continued use of the product, at least in a reasonably long time frame is not a given
for tourism products. Since tourism products are not like laundry/dish soap and many other
consumer products (Gitelson and Crompton, 1984), special analysis considerations must
occur. Given that this research suggests that over half of Taiwan’s inbound visitors stating
very likely to return do not return and given one has no idea when return will occur, is
conventional intention-to-return information good for developing theory or making planning,
management and marketing decisions?
Realistically, for planning, management and marketing, should one use change in a scale
value computed for a mean of a single or multiple scale to make decisions about programs
working or actions needed? Such measurement of change is some what like measuring
change in temperature without knowing what temperatures correspond to either boiling or
freezing. If you measured in Celsius, Fahrenheit and Kevin, you would get three different rates
of change. Is one correct? No. All are correct but a rate only means something when correctly
associated with the scale being used. And, those scales are only de?ned when two points on
them have speci?c values. A thermometer is only calibrated when, for example, freezing and
boiling are associated with appropriate values. Making decisions when one has no idea of
volume of return because, for example, one does not know about return once, twice or three
times in the next three years is making decisions without relating a measure to ?ows/volumes
that affect such matters as transportation capacity, accommodation needed and viability of
service businesses. If one is going to collect data, the target should be having information
elucidating volumes of visitors and impacts relating to goals (e.g. supporting regional
development). Current theory and methodology is not suf?cient to support decision making.
Yes, this research largely focuses attention on matters that need to be addressed by theory
development so that practical matters can be analyzed. However, an important implication is
that resources are being poorly expended in much research on tourist loyalty and return.
Data collected are not appropriate to develop an understanding of the dynamics of return.
The critical matter documented is the need for a change in research practices and data
collection practices regarding intentions of tourists to return.
References
Baker, D.A. and Crompton, J.L. (2000), ‘‘Quality, satisfaction and behavioral intentions’’, Annals of
Tourism Research, Vol. 27 No. 3, pp. 785-803.
Baloglu, S. and Erickson, R.E. (1998), ‘‘Destination loyalty and switching behavior of travelers: a Markov
analysis’’, Tourism Analysis, Vol. 2 No. 2, pp. 119-28.
Baloglu, S. and Erickson, R.E. (1999), ‘‘Reply to comment on destination loyalty and switching behavior
of travelers: a Markov analysis’’, Tourism Analysis, Vol. 3 No. 2, pp. 117-8.
Beaman, J.G., Huan, T.C. and Kozak, M. (2002), ‘‘Estimating a Markov model that incorporates ?rst visit
decisions and varying repeat frequency’’, Tourism Analysis, Vol. 6 No. 2, pp. 81-97.
Boyd, B.K., Gove, S. and Hitt, M.A. (2005), ‘‘Construct measurement in strategic management research:
illusion or reality?’’, Strategic Management Journal, Vol. 26, pp. 239-57.
Darnell, A.C. and Johnson, P.S. (2001), ‘‘Repeat visits to attractions: a preliminary economic analysis’’,
Tourism Management, Vol. 22 No. 2, pp. 119-26.
PAGE 52
j
INTERNATIONAL JOURNAL OF CULTURE, TOURISM AND HOSPITALITY RESEARCH
j
VOL. 6 NO. 1 2012
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
:
1
8
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
Decrop, A. (2001), ‘‘The antecedents and consequences of vacationers’ dis/satisfaction: tales from the
?eld’’, in Mazanec, J.A., Crouch, G.I., Ritchie, J.R.B. and Woodside, A.G. (Eds), Consumer Psychology
of Tourism, Hospitality and Leisure, Vol. 2, Cabi Publishing, Wallingford, pp. 333-47.
Fakeye, P.C. and Crompton, J.L. (1991), ‘‘Image differences between prospective, ?rst-time, and repeat
visitors to the Lower Rio Grande Valley’’, Journal of Travel Research, Vol. 30 No. 2, pp. 10-16.
Gitelson, R.J. and Crompton, J.L. (1984), ‘‘Insights into the repeat vacation phenomenon’’, Annals of
Tourism Research, Vol. 11 No. 2, pp. 199-217.
Haig, M. (2006), Brand Royalty: Howthe World’s Top 100 Brands Thrive and Survive, Kogan Page, London.
Hsu, C.H.C. and Kang, S.K. (2007), ‘ ‘CHAID-based segmentation: international visitors’ trip
characteristics and perceptions’’, Journal of Travel Research, Vol. 46 No. 2, pp. 207-16.
Huan, T.C. and Beaman, J.G. (2003a), ‘‘Individual and dynamics of social interaction in information search
in discretionary travel decision’’, in Crouch, G.I., Perdue, R.R., Timmermans, H. and Uysal, M. (Eds),
Consumer Psychology of Tourism, Hospitality and Leisure, Vol. 3, La Trobe University, Cabi Publishing,
Wallingford, pp. 149-59.
Huan, T.C. and Beaman, J.G. (2003b), ‘‘Contexts and dynamics of social interaction and information
search in decision-making for discretionary travel’’, Tourism Analysis, Vol. 8 Nos 2-4, pp. 177-82.
Huan, T.C., Kozak, M. and Beaman, J.G. (2003), ‘‘Implications of models of repeat travel for the concept
of destination loyalty’’, Leisure and Society, Vol. 26 No. 1, pp. 183-207.
Kozak, M. (2001), ‘‘Repeaters’ behavior at two distinct destinations’’, Annals of TourismResearch, Vol. 28
No. 3, pp. 785-808.
Kozak, M., Huan, T.C. and Beaman, J.G. (2002), ‘‘A systematic approach to non-repeat and repeat
tourism with measurement and destination loyalty concept implications’’, Journal of Travel and Tourism
Marketing, Vol. 12 No. 4, pp. 19-38.
Loo, R. (2002), ‘‘A caveat on using single-item versus multiple-item scales’’, Journal of Managerial
Psychology, Vol. 17 No. 1, pp. 68-75.
Oliver, R.L. (1999), ‘‘Whence consumer loyalty’’, Journal of Marketing, Vol. 63, pp. 33-44.
Oppermann, M. (1998a), ‘‘Destination loyalty and switching behavior of travelers revisited’’, Tourism
Analysis, Vol. 3 No. 1, pp. 115-6.
Oppermann, M. (1998b), ‘‘Destination threshold potential and the law of repeat visitation’’, Journal of
Travel Research, Vol. 37 No. 2, pp. 131-7.
Pritchard, M.P. and Howard, D.R. (1997), ‘‘The loyal traveler: examining a typology of service
patronage’’, Journal of Travel Research, Vol. 35 No. 4, pp. 2-10.
Sivadas, E. and Baker-Prewitt, J.L. (2000), ‘‘An examination of the relationship between service quality,
customer satisfaction and store loyalty’’, International Journal of Retail and Distribution Management,
Vol. 28 No. 2, pp. 73-82.
Taiwan Tourism Bureau (2001), 2000 Annual Survey Report on Visitors Expenditure and Trend in Taiwan,
Taiwan Tourism Bureau, Taipei.
Taiwan Tourism Bureau (2002), 2001 Annual Survey Report on Visitors Expenditure and Trend in Taiwan,
Taiwan Tourism Bureau, Taipei.
Taiwan Tourism Bureau (2004), 2003 Annual Survey Report on Visitors Expenditure and Trend in Taiwan,
Taiwan Tourism Bureau, Taipei.
Yu¨ ksel, A. (2001), ‘‘Managing customer satisfaction and retention: a case of holiday destinations,
Turkey’’, Journal of Vacation Marketing, Vol. 7 No. 2, pp. 153-68.
Corresponding author
Tang-Chung Kan can be contacted at: [email protected]
VOL. 6 NO. 1 2012
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
:
1
8
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
This article has been cited by:
1. Wu?Chung Wu, You?De Dai, Hsiou?Hsiang Jack Liu. 2012. Effective information collection on international inbound visitors.
International Journal of Culture, Tourism and Hospitality Research 6:1, 54-69. [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
:
1
8
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
doc_342126463.pdf