“:F'gléfiamww A»
- . .—

M‘aim-u

n6 7,; mm.

':':-.

JroLZ.

. ribu'h‘gnmj -. :3- n 3

‘

'IIMI'VJ 0: .JII ”ﬁc-vﬁnuwv

3913mm 9.4-.

hid

“it.

‘5.— V

 

 

 

- M‘
""v'i VTLH \"

 

T; it 51's
3
0700 3
SHEIHVHQ
LIB RARY This is to certify that the
MICh Igan State dissertation entitled
U nwersuty

 

 

 

DETERMINANTS OF DESTINATION IMAGE

presented by

ASLI D. A. TASCI

has been accepted towards fulﬁllment
of the requirements for the

Ph.D. degree in PARK, RECREATION AND
TOURISM RESOURCES

 

Major Professor’s Signature
9/3/03

Date

MSU is an Aﬂinnab’vo Action/Equal Opportunity Institution

- v ~——v~

‘ —v——-~4~ _

PLACE IN RETURN BOX to remove this checkout from your record.
TO AVOID FINES return on or before date due.

MAY BE RECALLED with earlier due date if requested.
DATE DUE DATE DUE

 

ADATE DUE
m :2 78 2005

 

 

 

 

OCT 2 5 2005

 

' 04W0 5;.
APR 06 2010

 

 

@40810

 

 

 

 

 

6/01 c-JCIHC/DatoDuepGS—nts

 

 

 

DETERMINANTS OF DESTINATION IMAGE
By

Asli D. A. Tasci

A DISSERTATION
Submitted to
Michigan State University

in partial fulfillment of the requirements
for the degree of

DOCTOR OF PHILOSOPHY

Department of Park, Recreation and Tourism Resources

2003

as «6‘.
of mu
assist

destins
”ORE

Welds;

Stiet‘te
Image
“Flair
recent
in this
A.

destina
\lSlI 1C
“Umbe

ABSTRACT
DETERMINANTS OF DESTINATION IMAGE
By

Asli D. A. Tasci

As tourism has become an international multibillion-dollar industry, practitioners
as well as academics have become interested in assessing the factors affecting the success
of tourism destinations one of which is destination image management. In an effort to
assist destination marketing organizations, academics have studied many aspects of the
destination image constmct, including factors that have an impact on destination image.
However, studies on the possible antecedents of image have been limited and have
yielded divergent and inconclusive results.

The purpose of this study was to provide further and more robust evidence on
selected destination image antecedents by testing the relationships between destination
image and selected variables. Two models were proposed and tested in an effort to
explain and predict destination image, one for the general population and one for the
recent visitor segment of the population. Three sets of antecedent variables were included
in this study: (1) sociodemographics (race, gender, age, income, and state of residence),
(2) past travel behavior (overall travel experience; previous visitation to the study
destination; the frequency of visitation to the study destination; whether or not the last
visit to the study destination was the most recent; the season, the length of stay and the
number of activities participated in during the last visit to the study destination), and (3) a

methodological variable (the season of the survey).

mimic“
factor a:
nearing
impact of

II
but impr
strongest
"museum:
“I055 ti
Setting Sf
mmwg

image fa:

Actxxities'
lisnazion 5

Has
poSiilOned

IOun‘snc ac

the inﬂUen

Imam

A large-scale longitudinal dataset (N=21,1ll), collected through telephone
interviewing in 1996, 1997, 1998, 2001, and 2002 was used in this study. Exploratory
factor analysis was used to reduce the set of image measures to fewer and more
meaningful factors. Ordinary least squares regression analysis was used to identify the
impact of selected antecedents on destination image.

It was found that, over the ﬁve-year period, Michigan’s image remained positive,
but improved during the later years. Michigan’s scenic appeal was found to be its
strongest attribute (mean ranging between 8.17-8.27 on a 10-point scale), while
‘museums’ or ‘popularity’ were found to be the weakest attributes at different points
across the study period. Two image factors of Michigan were identiﬁed: The
Setting/Sense of Place (grand mean=7.75) and Activities/Things to do (grand
mean=6.93). Several antecedent factors were found to be signiﬁcant in explaining these
image factors in differentOdata periods; however, few of them were robust, meaning,
consistently signiﬁcant across different data periods. Visitation and race for the
Setting/Sense of Place Factor, age and visitors’ Illinois residence for the
Activities/Things to do Factor, and age, visitation, visitors’ gender and most recent
visitation for the Overall Image (mean image score) were found to be robust.

Based upon the study results, it is recommended that Michigan should be
positioned on its strongest attribute, its scenic appeal, while improving the image of its
touristic activity amenities. Promotional messages must reﬂect this position while taking
the inﬂuential sociodemographic variables, race, age and gender into consideration.

Limitations of the study and further research suggestions are also provided.

Copyright by
ASLI D. A. TASCI

2003

Tc

To those who always dare to see, be and lead ahead, starting with my DAD...

let
the formaum
Donald F H
during m} gr
mmmmt
Joseph D Fr.

Knights. it it

I 3p:
15m” and Er

“MOE. and

[miles and 3U!

ACKNOWLEDGMENTS

I would like to express my gratitude and appreciation to the many people who have played roles in
the formation and completion of this dissertation. First, I would like to thank my mentor and advisor, Dr.
Donald F. Holecek, for providing his expert knowledge and experience. guidance, and ﬁnancial support
during my graduate study, which made the completion of this project possible. I also extend sincere thanks
to the most experienced, supportive, caring and encouraging committee members I could ever have: Dr.
Joseph D. F ridgen, Dr. S. Tamer Cavusgil and Dr. Bonnie J. Knutson. Without their guidance, support and

insights, it would not be possible to accomplish this endeavor. -

I extend special thanks to my mentors and colleagues William Gartner, Muzaﬂ‘er Uysal, Yuksel
Ekinci, Ercan Sirakaya and Dogan Gursoy for their supportive fellowship that made my life easier. Thank
you for sharing your experiences. insights and suggestions and for keeping in touch with your encouraging

spirits!

I appreciate my family for playing a special role in completion of this project. I thank my parents.
Ismail and Bmine, for allowing me be who I am and believing in me, not to forget their ever lasting love,
sacriﬁce, and support. My uncle, Sakir Ay, deserves special thanks for his fatherly love and encouragement
throughout my higher education. I thank my dear and beloved brothers, sisters, nephews, nieces, cousins.

uncles and aunts for their care, support. love and inspiration. Sagolun, varolun!

Last but not least, I am grateful to have a few special friends who reminded me of my priorities
and kept me on my feet whenever I needed. I thank my faithful friends Ahmet Krzrlay and Salih Armagan
for their joyful friendship, support and inspirations throughout my graduate study. I thank my beloved
friend Guy M. Lee for reminding me of the pleasures of life during troubled times. I also thank my friends
Rana Yegéne, Nilgﬁn Dalkesen, Tune Uraz, Azlizam Aziz, Chang Huh, and Xin Li for their supportive
fellowship. I owe many thanks to all those colleagues and friends who touched and leﬁ a trace on my life; '

you are in my heart even though I could not mention your names.

LIST OF'
LIST OF 1

CHAPTEJ
DTRODI
Backgroun
D:
D:
Statement l
Study Obje
Dr

Central Ref
DEﬁmthns
Special Ter
Surnmary o

eﬂirtation
The impaCI
impaa
impact
”Wham
Soc

Pits

TABLE OF CONTENTS

LIST OF TABLES ............................................................................... x
LIST OF FIGURES .............................................................................. xiii
CHAPTER I
INTRODUCTION ................................................................................ 1
Background ......................................................................................... 2
Destination Image’s Impact on Tourist Behavior ................................... 3
Determinants of Destination Image ................................................... 4
Statement of the Problem .......................................................................... 6
Study Objectives .................................................................................... 9
Discussion of the Objectives ........................................................... 9
Objective 1. ..................................................................... 9
Objective 2. ..................................................................... 11
Objective 3. ..................................................................... 14
Objective 4. ..................................................................... 15
Objective 5. ..................................................................... 16
Central Research Hypotheses .................................................................... 17
Deﬁnitions of Related Concepts ................................................................. 18
Special Terminology ............................................................................... 19
Summary of the Content of the Dissertation .................................................... 20
CHAPTER I]
LITERATURE REVIEW ....................................................................... 21
Destination Image .................................................................................. 21
The impact of Destination Image on Pre-Visit Tourist Behavior Variables ................ 23
The impact of Destination Image on During Visit Tourist Behavior Variables ........... 27
The impact of Destination Image on Post-Visit Tourist Behavior Variables .............. 28
Determinants of Destination lrnage ............................................................. 29
Sociodemographics ...................................................................... 33 ‘
Past Travel Behavior .................................................................... 38
Overall Travel Experience .................................................... 38
Prior Visitation .................................................................. 39
Visitation-Related Variables .................................................. 43
The Frequency of Prior Visitation .................................. 43
The Length of Stay ................................................... 45
The Number of Activities ............................................ 45
The Season of the Visit ............................................... 46
Methodological Factors ................................................................. 46
Computer Assisted Telephone Interviewing (CATI) ......................................... 48
Factor Analysis ..................................................................................... 50
Regression Analysis ................................................................................ 53
Hypothesis Test .......................................................................... 56

vii

CHAPI
METH'
Study D
Stud} D
Study lr
Data C c
Respons
Data Pr:
Data .k'r

C IMPI

RESl'I.
Sample '.

EFA Tu

The R62

(”Han
SIMS“
511mm all

C.) L’f/TIIS (1'! In rnb-v

Implicatr

"U‘U

Multicollinearity and Singularity ...................................................... 60

CHAPTER III
METHODS ........................................................................................ 62
Study Design ........................................................................................ 62
Study Destination and Population ................................................................ 64
Study Instrument ................................................................................... 65
Data Collection Mode and Procedures .......................................................... 69
Response Rate ...................................................................................... 71
Data Preparation .................................................................................... 73
Data Analyses ...................................................................................... 86
CHAPTER IV
RESULTS AND DISCUSSION ................................................................ 93
Sample Proﬁles ..................................................................................... 96
The Mixed Sample Proﬁle .............................................................. 98
Comparison of Partial Image-Respondents with Full Image-
Respondents and General Population Model-Respondents ............... 105
Comparison of Full Image but not Model-Respondents with General
Population Model-Respondents .............................................. 108
Comparison of Different Races ............................................... 110
The Recent Visitor Group Proﬁle .................................................... 114
EFA Two-F actor Model ........................................................................... 120
Image Item Descriptions ............................................................... 12]
Comparison of Visitors’ and Non-visitors’ Images ................................. 124
Image Item Correlations ................................................................ 127
Results of Exploratory Factor Analyses .............................................. 128
The Regression Models ............................................................................ 135
General Population Model Variables ................................................. 136
General Population Model Test Results .............................................. 140
Recent Visitor Segment Model Variables ............................................ 149
Recent Visitor Segment Model Test Results ........................................ 155
CHAPTER V
SUMMARY AND CONCLUSIONS .......................................................... 164
Summary of Results ................................................................................ 164
Michigan’s Image ....................................................................... 164
Signiﬁcance of Demographics of the General Population ......................... 168
Signiﬁcance of Demographics of the Recent Visitor Segment .................... 170
Signiﬁcance of Past Travel Behavior of the General Population ................. 172
Signiﬁcance of Past Travel Behavior of the Recent Visitor Segment ............ 173
Signiﬁcance of the Season of the Survey. ............................................ 174
Models .................................................................................... 175
Implications of Results ............................................................................ 178
Positioning of Michigan ................................................................ ”8
Promotion of Michigan ................................................................. 179

viii

leliaiit

Suggesai

APPEN]
1996-19‘

APPENI
2001-”0l

REFERI

Product Improvement mMichrgan ............ 181

Limitations .......................................................................................... 1 82
The Nature, Consistency and Relevance of the Data ............................... 182
The Potential Bias from Sampling, Nonresponse and Recall ..................... 184
Generalizability .......................................................................... 1 86

Suggestions for Future Research ................................................................. 187

APPENDIX A

1996-1998 QUESTIONNAIRE ................................................................. 194

APPENDIX B

2001-2002 QUESTIONNAIRE ................................................................. 204

REFERENCES .................................................................................... 214

ix

LIST 1

Table I

Table I.

Table 3

Table 4
Table 5

Table 6

Table 7

Table 8

Table 9

Table 10

Table 11
Table 1:
Table 13
Table 14

Table 15

Table 16

Table 17

LIST OF TABLES

Table 1.

Table 2.

Table 3.

Table 4.

Table 5.

Table 6.

Table 7.

Table 8.

Table 9.

Table 10.

Table 11.
Table 12.
Table 13.
Table 14.

Table 15.

Table 16.

Table 17.
respondents. ............................................................................. I 07

Results of studies investigating the inﬂuence of sociodemographics on
destination image. ...................................................................... 35

Results of studies investigating the inﬂuence of prior visitation on
destination image. ....................................................................... 40

Different terminology provided for images prior to and aﬁer actual

visitation. ................................................................................ 42
The interview code scheme. .......................................................... 70
Michigan image measures. ............................................................ 76

The original measurement levels and codes of the variables included in the
general population model. ............................................................ 77

Modiﬁed codes for the nominal and ordinal variables and resulting multiple
dummy variables included in the general population model. ..................... 78

Original visitation variables and their response categories. ...................... 80

Original visitation variables, new dummy variables and the ultimate
visitation dummy variable. ............................................................ 80

Original visitation-related variables, their levels of measurement and

response categories. .................................................................... 83
The original and recoded categories of the visitation-related variables. ........ 83
Weighting ratios used for each study region. ....................................... 86
Sample Sizes for weighted and un-weighted datasets. ............................. 86
Steps of analyses used in the study. .................................................. 88
The descriptives and frequencies of all respondents’ demographics for

different time frames and subgroups. ................................................ 99
Comparison of full image-respondents and partial image-respondents. ........ 106

Comparison of general population model-respondents and partial image-

Iwklﬂ

Imkll

IDES:
TDkZS

ka24

kali

TDkZb

T&k27
ka23
ka29
ka30
Table 31

Table 3:

Table

‘w
3.)

T&k34
Table 35

Table 36

Table 18.

Table 19.

Table 20.

Table 21.

Table 22.

Table 23.

Table 24.

Table 25.

Table 26.

Table 27.

Table 28.

Table 29.

Table 30.

Table 31.

Table 32.

Table 33.

Table 34.

Table 3 5.

Table 36.

Comparison of general population model-respondents and full image but not
model-respondents. ..................................................................... 1 09

Racial distribution in the study region in the US Census 2000 data and

racial distribution in different groups utilized in the analyses of the study. . 11 1
Comparison of different races on selected variables. .............................. 113
The descriptives and frequencies of recent visitor respondents’ demographics
for different time frames and subgroups. ............................................ 115
Image items and corresponding notations. .......................................... 121
Image item descriptive statistics for different sampling periods. ................ 121
Results of the independent samples t-test on the differences between the
initial (1996-1997) and later (2001-2002) years of the study. ................. 122
Image item descriptive statistics for visitors and non-visitors. ................... 125
The results of the independent samples t-test on the differences between
visitors and non-visitors. ............................................................... 125
Image item descriptives for all visitors and all non-visitors. ..................... 127
Image item correlation matrix. ....................................................... 128
Factor analysis results for different data periods. .................................. 129
Factor analysis results for visitor and non-visitor groups. ........................ 131
Summary of factor analysis results. .................................................. 132
General population model variables’ descriptives and frequencies for the all-
years’ data. .............................. . ................................................ 137
General population model variables’ correlations. ................................. 138
General population model variables’ descriptive statistics for different data
periods. ................................................................................... 139
General population model test results for Factor I (The Setting/Sense of
Place). .................................................................................... 142
General population model test results for Factor II (Activities/Things to do). 143

ka4i

ka43

Iﬁk44

Table 37.

Table 38.

Table 39.

Table 40.

Table 41.

Table 42.

Table 43.

Table 44.

General population model test results for the Overall Image (mean image
score). ....................................................................................

Visitor model variables’ descriptives and frequencies. ............................
Recent visitor segment model variables’ correlations. ............................

Recent visitor segment model variables’ descriptive statistics across different
data periods. .............................................................................

Recent visitor segment model test results for Factor I (The Setting/Sense of
Place). ....................................................................................

Recent visitor segment model test results for Factor 11 (Activities/Things to
do). ........................................................................................

Recent visitor segment model test results for the Overall Image (mean image
score). ....................................................................................

Summary of all model test results. ...................................................

xii

144

150

151

154

157

158

159

I76

LIST 0T 1
Hmml
ngel

HmueS

ngep
Rpm?
Hams
Hme
ﬁgmlg

Figure 1 1

LIST OF FIGURES

Figure 1. Tourist behavior inﬂuenced by destination image. .............................. 3
Figure 2. Determinants of destination image. ................................................ 4
Figure 3. A model of destination image, its determinants and effects. .................... 8
Figure 4. The causal relationship between visitation and destination image. ............ 12
Figure 5. The logical relationship between prior visitation, subsequent visitation and
destination image. ................................................................... 13
Figure 6. The study region (sampling frame). ............................................... 65
Figure 7. The ﬂowchart of the 1996-1998 questionnaire. .................................. 67
Figure 8. The ﬂowchart of the 2001-2002 questionnaires. ................................. 68
Figure 9. Samples, subsamples and subgroups utilized in different analyses. .......... 97
Figure 10. Two-Factor Michigan image model. .............................................. 133
Figure 11. A model of Michigan’s image and its determinants. ............................ 177

xiii

factors
and hid:
need for
has led
destinatic
of a destrr
Th

destination
has differs
Teiearchers
“isn't until
iImportance
academics a
ConsUmer be.
ﬁeld

Resea;
€50” to facil;
have SlUdied I

set .

CHAPTER I

INTRODUCTION

AS travel and tourism has grown into an international multibillion dollar industry,
factors affecting the success of this industry have become of interest to both academics
and industry practitioners. Both academics and industry practitioners have realized the
need for information on factors playing a role in the success of tourism destinations. This
has led to an increased number of investigations of the many facets of successful
destination marketing. One popular research focus has been the management of the image
of a destination.

The destination image construct can roughly be deﬁned as a mental picture of a
destination composed of how people visualize, think, and feel towards the destination. It
has different components, cognitive, affective and conative, that have intrigued many
researchers. The destination image construct was ﬁrst studied in the early 705, but it
wasn’t until the 905 that it gained momentum, which coincides with the realization of the
importance of destination image for successful destination promotion. Since both
academics and practitioners in the tourism ﬁeld have realized its strong effect on
consumer behavior, it has been a relatively well-studied venue of inquiry in the tourism
ﬁeld.

Researchers have studied various aspects of the destination image construct in an
effort to facilitate destination management endeavors of destination authorities. They
have studied the destination image construct in relation to its potential inﬂuence on

several tourist behavior variables; they have also investigated several potential

dc’.‘
int?
of 5
due
other

1171.95:

researc
determi

conduct.
question.
Objective
nil} be ldr

Summary (

determinant variables that could potentially have an inﬂuence on destination image. The
inﬂuence of destination image on several consumer behavior variables and the inﬂuence
of several determinant factors on destination image have been documented. However,
due to the complex nature of the destination image construct intertwined with several
other variables and concepts, there are several aspects of destination image yet to be
investigated.

This chapter will start with a brief discussion of the background containing the
research about the inﬂuence of destination image on consumer behavior followed by the
determinant variables that have an inﬂuence on destination image. The need for
conducting this study will be discussed, and the statement of the problem and research
questions will be provided. Study objectives along with a brief discussion of these
objectives will provide further explanation of the need for this study. Central hypotheses
will be identiﬁed followed by the deﬁnition of few concepts utilized in this study. A brief

summary of the subsequent chapters will also be provided at the end of this chapter.

Background
Destination image, “the sum of beliefs, ideas, and impressions that an individual
has of a destination” (Crompton 1979, p. 18), is an elusive and complex research
construct comprising many facets that has been the focus of the scientiﬁc inquiry. It has
been conceptualized as an overall (holistic) evaluation of a destination, as well as a
composite of conceptually different components, namely, cognitive, affective, and
conative (Baloglu & McCleary 1999; Chen & Kerstetter 1999; Echtner & Ritchie 1993;

Gartner 1993; Milman & Pizam 1995).

Destin

compo:
variable
that [fit

their b

making a:

Healing d'

Destination Image’s Impact on Tourist Behavior

Destination image, both as an overall evaluation and with its different
components, have been postulated to be inﬂuential on various consumer behavior
variables. As summarized in Figure 1, previous destination image studies have shown
that the image of a particular destination held by tourists has an inﬂuence on some of
their behaviors concerning not only before; but also during and after visiting the
destination (Britton 1979; Chen & Hsu 2000; Chen & Kerstetter 1999; Coshall 2000;
Court & Lupton 1997; Fakeye & Crompton 1991; Fridgen 1987; Gartner 1989; Gartner

1993; Milman & Pizam 1995; Murphy 1999). These studies will be reviewed in detail in

 
   
 

 

 

 

 

 

 

 

 

 

the next chapter.
Destination

Pre-visit behavior During visit behavior Post-visit behavior
Information search Time spent Word-of-mouth
Decision-making Enjoyment Recommendation
Planning time frame Satisfaction Revisit intentions
Destination choice Destination loyalty
Time planned to spend
Money planned to spend
Anticipation

 

 

 

Figure 1. Tourist behavior inﬂuenced by destination image.

Several researchers have studied destination image as an independent variable
inﬂuencing several consumer behavior variables, such as destination choice, decision-
making and satisfaction. Both holistic destination image (measured by one variable

treating destination image as an overall perception), as well as speciﬁc destination

dimer?
and d:
and a?

during

Court 6
Determi

researche
Strengthe:

managemg

role in the

E

Fig
,ureg
Deter

dimensions (measured by multiple variables focusing on individual destination attributes)
and destination image factors (a composite of two or more dimensions), both cognitive
and affective, were found to inﬂuence consumer behavior variables related to before,
during and after visitation of a destination (Chen & Hsu 2000; Chen & Kerstetter 1999;

Court & Lupton 1997; Ross 1993; Schroeder 1996).

Determinants of Destination Image

Due to destination image’s potential inﬂuence on several tourist behaviors,
researchers have been trying to identify the determinants that deﬁne, modify, and
strengthen this construct in an effort to help destination authorities in their image
management endeavors. Destination image studies have shown that several factors play a

role in the destination image formation process as summarized in Figure 2.

  
  
  
  
 

     
     

  
        
 
   
 

   
  
  
 

Information Information
sourcing sourcing from
from the Q Percerver-related \ autonomous

destination Determinants agents

 

Methodological
Determinants

 

 

Destination

Figure 2. Determinants of destination image.

As
sourcing
authorities
destination
popular 1
sociodemo
play a role
themselves

The
inﬂuenced 1
Inﬂuence 0:
& Fesenma
literarme C

prOpOSiIlOn"
Gum 197:,
have ShOwn
1992; 30m
remained PIC
SeVe:
charaeten'suc
regiden Ce Sec
1996‘ Chen 6

ofresearch O]

As it is summarized in Figure 2, these determinants include: (1) information
sourcing from the destination, such as promotional messages by the destination '
authorities, (2) information sourcing from the autonomous agents in between the
destination and the perceivers, such as news articles, educational materials, movies,
popular culture and word-of-mouth, (3) perceivers’ characteristics including
sociodemographics as well as past travel behavior, and (4) methodological factors that
play a role while measuring image, such as the methodologies used and researchers
themselves.

Therefore, past studies have treated destination image also as a dependent variable
inﬂuenced by several of the above-mentioned factors. A few researchers have Studied the
inﬂuence of visual materials on the images created by different types of people (MacKay
& Fesenmaier 1997; MacKay & Fesenmaier 2000; Smith & MacKay 2001); however, the
literature on the impact of destination-originated information has remained mostly
propositional (Alhemoud & Armstrong 1996; Bramwell & Rawding 1996; Gartner 1993;
Gunn 1972). The same is true for the impact of autonomous information; few researchers
have Shown the dramatic impact potential of autonomous information (Gartner & Shen
1992; Sonmez, Apostolopoulos, & Tarlow 1999), while others’ postulations have
remained propositional (Bramwell & Rawding 1996; Gunn 1972; Gartner 1993).

Several past studies have shown the inﬂuence of perceivers’ sociodemographic
characteristics including age, gender, household status, education, income, and
residence/geographic distance to the study destination (Alhemoud & Armstrong; Ahmed
1996; Chen & Kerstetter 1999; Hunt 1975; MacKay & Fesenmaier 1997). There is a lack

of research on other sociodemographic variables such as race.

Set
prexious ‘
dererminar
Crompton
1999. Veg
travel beh:
of the last

So:
destinatior
Ritchie 19
methodolo

literature 0

Alt!
POSSible de
behaﬁm. ti
theoretical
variab]eS p
StudieS hax
the“ result:

Q

same destin,

501315 fOcus

Several past travel behavior variables, including previous visitation, the amount of
previous visitation, and length of stay have also been documented as important
determinants of destination image by several researchers (Crompton 1979; Fakeye &
Crompton 1991; Hu & Ritchie 1993; Baloglu & McCleary 1999; Chen & Kerstetter
1999; Vogt & Andereck, in press). However, there is a lack of research on other past
travel behavior, such as overall travel experience, how recent the last trip was, the season
of the last visit and activities participated in while at the destination.

Some methodological factors have been recognized to have a possible impact on
destination image as measured by researchers (Dadgostar & Isotalo 1992; Echtner &
Ritchie 1993). However, there has been a lack of focus on measuring the impact of
methodological factors empirically, such as the timing of survey (Gartner 1986). The

literature on the determinants of image will also be reviewed in detail in the next chapter.

Statement of the Problem

Although destination image researchers have investigated the impact of several
possible destination image determinants, including sociodemographics and past travel
behavior, their efforts have not been comprehensive or conclusive enough to solidify a
theoretical background for ﬁiture researchers. There has been a lack of attention on some
variables potentially inﬂuential on destination image. In addition, results of previous
studies have remained not only divergent but also destination and time-speciﬁc since
these results have not been replicated by other researchers for other destinations or for the
same destinations over an extended period of time. Also, there has been a lack of studies

solely focusing on the determinants of destination image. Previous studies have provided

main 1
desiina

variabk

{051451 '
perceive?
and a me
detemline
model “7
tariables a
the results
gathered 0\
Thus
seleaed soc.
image. if so.
methodologic.
relative to the
Dre-diet destrna
developed for

deninar

ion and

”1056 Who visit

piecemeal information about the detemtinants of destination image after satisfying the
main goal of these studies, which has usually been measuring the image of a speciﬁc
destination. Moreover, previous studies have not attempted to identify the relative
magnitude of the determinant variables even when the impact of multiple determinant
variables was studied.

As depicted in Figure 3, the purpose of this study is to provide further and more
robust evidence of the relative impact of selected determinant variables, including
perceiver variables (selected sociodemographic characteristics and past travel behavior)
and a methodological variable (the season of the survey). The relative magnitude of these
determinant variables will be identiﬁed through application of a multiple regression
model with destination image as the dependent variable and selected determinant
variables as the independent variables. Robustness will be assured through validation of
the results on data ﬁ'om different time periods by using extensive longitudinal data
gathered over a ﬁve-year period.

Thus, the researchable questions emerging from the purpose of this study are: Are
selected sociodemographic and past travel behavior variables inﬂuential on destination
image, if so, what is each variable’s relative inﬂuence? Is the season of the survey, as a
methodological determinant, inﬂuential on destination image, if so, what is its inﬂuence
relative to the other determinant variables? In addition, in an effort to both explain and
predict destination image, can a parsimonious model of these determinant variables be
developed for the general population (including those who previously visited the study
destination and those who did not) and the recent visitor segment of the population (only

those who visited the study destination within the past 12 months)? Answers to these

 

>a—;.—.m 73:? Z— A_n._.—.<uv.y—.V.E>Z_
mrFZ<Z-—Z-m-.—.v-=

 

33 .356 o

azaea> eoaoeeoeoz 383

 

 

is as 2: mean 8:298 2:
an E 333253 853:8 .«o .3852 .
am; as 05 wars—o
5:95ch 3:6 2: a .86 mo 595‘— .
coca—:33
3:6 2: 3 am? as 2: mo :OmSm .
9: E82 $9: 2: 2 5.8558
beam 05 o. E? as 05 8: B 8523 o
eons—53v 32m
05 3 8357.3 Sta Co 35:62... .
noun—~32. beam 2: 2 55863 Sun o
85.595 325 555 o
m03~m§> Loss—Em _D>E.H “mun— Baum—om
25m vacuumed o
2:85
ow<
H850
8mm
8—3.5; uEaaconoeomoom vogue—om

.mwoobo 95 3523:83 at women: 223533 we ESE < .m 233m

 

 

.038. nous—Eon—
meoucuaﬁ ESQ: :ouoﬁmzmm
gauges—Booed “555.3...“
ﬁnes—does? “:on DEC.

 

 

 

 

 

 

5533 53.20 8322. 53 misc

 

eouamuueaa

55% 8 BEBE $.82
28% 2 3583 0:5.
865 835.80
253 DE: wEEBE
win—«8-5385

538 559585

 

 

 

5323 “Etta:—

 

 

 

 

>QDPm mum—h. Z— DHH<UFHmﬂ>Z—
mhz<7=§nmhma

owes:
cos—Eamon
3586:0qu
323—80522
/\ /\ \r/
gnome 358E580 8:2:er
3682.33 ‘ wows—orbzoocom ' 2: Eoc
Set @658 / Q Q \ @859.
558.85 cognac?—

       
    

  

queSIion

their ima

F:
were ider

I)

4»

'1!
v

Discussion 0

Obje.

mCOl'ne~ age

\

IO be inﬂUem

1501310 1997

questions will provide destination promoters with valuable information to be used for

their image management activities.

Study Objectives

From the researchable questions mentioned above, the following ﬁve objectives

were identiﬁed to frame this study:

1)

2)

3)

4)

5)

To investigate the impact of selected sociodemographic variables on
destination image,

To investigate the impact of selected past travel behavior variables on
destination image,

To measure the impact of the season of the survey on destination image,

To create a general population model of destination image and selected
determinant variables, including perceiver-related variables (selected
sociodemographics and past travel behavior) and the season of the survey,

To create a recent visitor segment model of destination image and selected
determinant variables, including perceiver-related variables (selected

sociodemographics and past travel behavior) and the season of the survey.

Discussion of the Objectives

Objective 1. Various sociodemographic variables, such as residence, gender,

income, age, the level of education, family life cycle and household size have been found

to be inﬂuential on tourist behavior (Bojanic 1992; Court & Lupton 1997; Dadgostar &

Isotalo 1992; Etzel & Woodside 1982; Gentry & Doering 1979; McQueen & Miller .

1985. Schu
& Pitts 19'-
inﬂuenced l
and docurr.
desrination
Crompton I
1997).

A re
from the 5'
destination r
researchers l
the results h
ditierenr age
ﬁnd any diff:
researchers h;
5; Fesenmaie}
There is a laci

Destin;
destination) M

of “ﬁlth coulc

e-‘Llsllng bOdV

destination ima

mile“Cr-Es The

1985; Schul & Crompton 1983; Snepenger, Meged, Snelling, & Worrall 1990; Woodside
& Pitts 1976). In the travel research literature, destination image is also postulated to be
inﬂuenced by various sociodemographic variables. Several researchers have investigated
and documented the relationships between various sociodemographic variables and
destination image (Alhemoud & Armstrong; Ahmed 1996; Chen & Kerstetter 1999;
Crompton 1979; Hunt 1975; Joppe, Martin, & Waalen 2001; MacKay & Fesenmaier
1997)

A few researchers have studied the impact of respondents’ residence or distance
from the study destination and found a signiﬁcant inﬂuence of this variable on
destination image (Hunt 1975; Crompton 1979; Walmsley & Young 1998). Some
researchers have studied age in relation to its inﬂuence on destination image; however,
the results have been divergent. Some researchers have reported image differences for
different age groups (Alhemoud & Armstrong 1996; Baloglu 2001) while others did not
ﬁnd any differences (MacKay & Fesenmaier 1997; Smith & MacKay 2001). Only a few
researchers have investigated the inﬂuence of gender (Chen & Kerstetter 1999; MacKay
& Fesenmaier 1997) and income (MacKay & Fesenmaier 1997) on destination image.
There is a lack of research on the race variable in the destination image literature.

Destination image is a construct with cognitive (factual information about a
destination) and affective (feelings and attitude towards a destination) components, both
of which could be deﬁned by the perceivers’ sociodemographic characteristics. Since the
existing body of empirical studies investigating sociodemographic inﬂuences on
destination image has not been substantial, there is room for more investigation of such

inﬂuences. Therefore, sociodemographic variables (available in the secondary data set

10

used {01' Il'.

(3) age. (4

central res:
their imag:
inﬂuence c
are h}poth
lr}pothesis

focus on i
including. 11'
especially ft

1Images of th

Obje
inﬂuence 011
P1101 \‘isitanc
more accura;
Selby & Mon:
MaCKay (5; F
CromptOn 19
Prior “SI-tailor

Arong
of DrexiOUS Vls

Aﬂdereck‘ m p

used for the purposes of this study) investigated in this study include: (1) race, (2) gender,
(3) age, (4) total annual household income, and (5) respondents’ state of residence. The
central research question is: Do respondents’ sociodemographic characteristics inﬂuence
their images of a destination, if so, what is each sociodemographic variable’s relative
inﬂuence compared to other determinant variables? These sociodemographic variables
are hypothesized to have varying degrees of inﬂuence on destination image; if this
hypothesis is proven, this information can be used to specify which market variables to
focus on in directing the promotional messages for image management purposes
including, image formation, correction, and maintenance. The ﬁndings would be of value
especially for the destination authorities striving to change, improve or strengthen the

images of their destination.

Objective 2. Past travel behavior variables have been investigated for their
inﬂuence on destination image. Several researchers have investigated the inﬂuence of
prior visitation and arrived at different results. Some have observed that visitors hold
more accurate and positive images (Baloglu & McCleary 1999; Milman & Pizam 1995;
Selby & Morgan 1996), or more or better aﬂ’ective responses (Baloglu & McCleary 1999;
MacKay & Fesenmaier 1997); some have found mixed results (Ahmed 1996; Fakeye &
Crompton 1991; Hu & Ritchie 1993); yet others have found no signiﬁcant inﬂuence of
prior visitation on a destination’s image (Chen & Kerstetter 1999; Hunt 1975).

Along with visitation itself, some visitation-related variables, such as the amount
of previous visitation (Baloglu 2001; Fakeye & Crompton 1991; Schroeder 1996; Vogt &

Andereck, in press) and the length of stay at the destination (Fakeye & Crompton 1991;

ll

potent
some \
visit. a;
formatit
influenc:
inﬂuence
participat
season of
next ehapr
ln
invesrigare-t
to conclude
direetion of
delennined .
destination ir.

0n the Other

ﬁgure 4 The C

 

Schroeder 1996; Vogt & Andereck, in press) have also been investigated and found to
inﬂuence destination image.

There is a lack of research on other past travel behavior variables that could
potentially inﬂuence destination image. Factors, such as overall travel experience and
some visitation-related variables, such as how recent the last trip was, the season of the
visit, and the number of activities participated in at the destination could inﬂuence the
formation of the image of a destination. Although there is a lack of direct focus on the
inﬂuence of these variables on destination image, there is some evidence of a possible
inﬂuence of the overall travel experience (Schroeder 1996), the amount of activities
participated at the destination (Ashworth 1989; Fakeye & Crompton 1991), and the
season of the trip to the destination (Gartner 1986). This will be discussed further in the
next chapter.

In addition to prior visitation, several other past travel behavior variables were
investigated in this study. Although ﬁndings in the destination image literature drive one
to conclude that prior visitation could alter destination image in a dramatic way, the
direction of the causal relationship between destination image and visitation is yet to be
determined (Baloglu 2001; Tasci & Holecek 2002); as displayed in Figure 4, either
destination image or visitation could be the starting point and each one could be feeding

on the other.

  
 

 

    

Destination
Image

9

Figure 4. The causal relationship between visitation and destination image.

   

 

\Khie’n

imaee

\

However, in this study, it will be assumed that visitation improves destination
image. This assumption is driven by the logic arrived at by synthesizing the existing
literature. A few researchers have postulated that previous visits to a destination inﬂuence
subsequent visits by playing a reassuring role in the decision for further visitation (Court
& Lupton 1997; Gyte & Phelps 1989; McQueen & Miller 1985; Woodside 1980). As was
mentioned before, several researchers agree that destination image inﬂuences the
decision to visit a destination. Thus, if visitation brings subsequent visits, and destination
image inﬂuences visitation, then visitation improves destination image, which induces
more visitation as illustrated in Figure 5.

Thus, it is assumed in this study that visitation inﬂuences destination image,
which inﬂuences subsequent visits. However, the ﬁrst visit could be due to a positive

image formed through exposure to information prior to that ﬁrst visit.

  
  
 

   
 

 
    

Pﬁor

. . . Subsequent
Vrsrtatron

Visitation

   
 

Destination
Image

  
  

  

Destination
Image

       

Subsequent
Visitation

Visitation

         

Figure 5. The logical relationship between prior visitation, subsequent visitation and
destination image.

Past travel behavior variables (available in the secondary data set used for

the purposes of this study) investigated in this study include: (1) overall travel

13

expe
(3} F
destir
not It
nsit I
37311.
particr
selecte
if so.
Determ
informa
commur
from th.

Variables

experience, measured as the number of trips to any destination within the past 12 months,
(2) prior visitation to the study destination, (3) the frequency of visitation to the study
destination, measured as the number of visits within the past 12 months, (4) whether or
not the last visit to the study destination is the most recent trip, (5) the season of the last
visit to the study destination, (6) the length of stay at the study destination during the last
visit, measured as the number of nights spent, and (7) the number of activities
participated in at the study destination during the last visit. The central question is: Are
selected past travel behavior variables inﬂuential on respondents’ images of a destination,
if so, what is their relative inﬂuence among other selected determinant variables?
Determining the relative inﬂuence of past travel behavior variables would reveal valuable
information for destination marketers in terms of what to stress in their marketing
communications to improve a destination’s image. Destination marketers would beneﬁt
from the results on the degree of inﬂuence of visitation and other visitation-related
variables for weighing and improving the beneﬁts of familiarization tours and discounted

offers to gain positive image through visitation.

Objective 3. Although there has been a lack of research on the impact of
methodological factors on destination image, it is hypothesized that the methods
researchers use, such as qualitative rather than quantitative methods and the timing of
data collection could inﬂuence the results obtained about the image of a destination. One
methodological factor hypothesized to affect destination image, the season of the survey,
will be examined in this study. It is expected that the season of the survey will have an

impact on the destination image measured. This expectation is based on two assumptions:

l4

(1) differe

ditferent se
reﬂecting 0
thereby 39.“:
levels of tht
(2002') foun
satisfaction

The

determinant
selected det
Variable VOL;

and genera!”

Obje
image linen]
desinanon ir.
beha‘lm. and
explain and F
dESIlnation
PleﬁoUSl), V1.5
exﬁlmed

lha

mmOdOlOQica

(1) different seasons will connote different meanings about a destination, and (2)
different seasons will have different effects on an individual’s mood, the “state of mind
reﬂecting one’s feelings at any particular moment” (Sirakaya, Petrick, & Choi 2002),
thereby affecting her/his responses to destination image items. Studying the satisfaction
levels of the passengers of two different Caribbean cruises, Sirakaya, Petrick, and Choi
(2002) found an inﬂuence of individuals’ mood on their attribute evaluations and overall
satisfaction.

The central research question is: Is the season of the survey, as a methodological
determinant, inﬂuential on destination image, if so, what is its inﬂuence relative to other
selected detenninant variables? Identifying the possible effect of this methodological
variable would caution researchers about accounting for such an inﬂuence on their results

and generating remedial measures, such as calibrating their results for different seasons.

Objective 4. The purpose of this study is to contribute to the existing destination
image knowledge by providing a parsimonious (balancing simplicity and ﬁt) model of
destination image and its determinants, including selected sociodemographic, past travel
behavior, and a methodological variable. The research question is: In an effort to both
explain and predict destination image, can a parsimonious model of the best subset of
destination image determinants for the general population (including those who
previously visited the study destination and those who did not) be developed? It is
expected that the selected sociodemographics, past travel behavior and the

methodological variable have different degrees of impacts on a destination’s image.

 

Therefor

these set

E
where.

OV’T‘J’OWD

m ’1
'I\

A:
of these 1
terms of

marketing

or
brEhatior a
[he P0pula
monthst 1
resl-‘Ont’lEnt
inclUded in
as the “Um!
stud}. desm

deStinanOn‘

Therefore, destination image for the general population is proposed to be a function of

these selected variables as detailed below:

D1 = f(R, G, A, I, S, OTE, V, SS)
where,

D1 = Respondents’ image of the study destination

R = Respondents’ race

G = Respondents’ gender

A = Respondents’ age

I = Respondents’ total annual household income

S = Respondents’ state of residence

OTE = Respondents’ overall travel experience

V = Prior visitation to the study destination

SS = Survey season

As was mentioned before, the knowledge on the relative magnitude of the impacts
of these variables on destination image would be valuable for destination marketers in
terms of which characteristics of the target markets to focus on while directing their

marketing communications.

Objective 5. A similar model, including selected sociodemographics, past travel
behavior and a methodological variable, is also proposed for the recent visitor segment of
the population (those respondents who visited the study destination within the past 12
months). In the recent visitor segment model, the visitation variable is excluded since
respondents will all be recent visitors; however, additional visitation related variables are
included in this model: (1) the frequency of visitation to the study destination, measured
as the number of visits within the past 12 months, (2) whether or not the last visit to the
study destination is the most recent trip, (3) the season of the last visit to the study

destination, (4) the length of stay at the study destination during the last visit, measured

16

asr‘n

des-ti

TECE'

sari:

pop:

“he

as the number of nights spent, and (5) the number of activities participated in at the study
destination during the last visit.

The central research question is: In an effort to both explain and predict
destination image, can a parsimonious model of destination image determinants for the
recent visitor segment of the population be developed? The ﬁndings will reveal which
variables marketers should focus on to improve image of their destinations, thus, induce
more visitation. Therefore, destination image for the recent visitor segment of the
population is proposed to be a function of these selected variables as detailed below:

D1 = f(R, G, A, l, S, OTB, SS, VF, VR, VS, N, AC)
where,
D1 = Respondents’ image of the destination
R = Respondents’ race
G = Respondents’ gender
A = Respondents’ age
I = Respondents’ total annual household income
S = Respondents’ state of residence
OTE = Respondents’ overall travel experience
SS = Survey season
VF = Visitation frequency to the study destination
VR = Whether or not the last visit to the study destination is the most recent trip
VS = Season of the last visit to the study destination
N = Number of nights spent at the study destination during the last visit
AC = Number of activities participated in during the last visit to the study
destination

Central Research Hypotheses
Hypothesis 1. Selected sociodemographic variables, past travel behavior variables
and the season of the survey have varying degrees of inﬂuence on the destination image
held by the general population, including those who previously visited the study

destination and those who did not.

17

H:
and the Si
held by tl

study dest

ln 1
and comp
terminology
throughout

lma
desrinatiort
attractions~
dimensions a

lmag
COmfon that
(”Sually 3 Or
FaClorAnaI-‘S

Image

\

or dGSIlnathn

Hypothesis 2. Selected sociodemographic variables, past travel behavior variables
and the season of the survey have varying degrees of inﬂuence on the destination image
held by the recent visitor segment of the population, including those who visited the

study destination within the past 12 months.

Definitions of Related Concepts

In the destination image literature, the concepts of image dimension, image factor
and component are used interchangeably. For scientiﬁc parsimony, a uniform
terminology based upon the following deﬁnitions of these concepts will be used
throughout this dissertation.

Image Dimensions: Individual destination attributes that make up the image of a
destination, such as natural landscape, climate, transportation, historical and cultural
attractions, accommodation, eateries, and hospitality. Image attributes and image
dimensions are used interchangeably throughout this dissertation.

Image Factor: An image construct, such as natural resources, convenience, and
comfort that are composites of several image dimensions. Individual image dimensions
(usually 2 or more) are reduced to a factor through analysis tools, such as Exploratory
Factor Analysis or Conﬁrrnatory Factor Analysis.

Image Component: Image component is used when referring to cognitive,
affective or conative aspects of the destination image concept. The cognitive component
of destination image is the factual information about a destination; the affective
component comprises attitudes and feelings towards a destination; conative component

refers to actions or action intentions sourcing from image.

18

   

single tan.
destination

destinatior.

 

 

 

Gen|
refer to the
destination

\‘isi
set used in
mthin the 1
within the .
D01 “lthln
In the ima
measufes l
\151101’3 an;
L'leS’tlnaUOn

Visi

Overall Image: The overall or holistic assessment of a destination with one
single variable, extracted either by asking the respondents their overall assessment of the
destination or averaging their ratings of the multiple individual dimensions of a

destination

Special Terminology

General Population: Throughout this dissertation, general population is used to
refer to the study population inclusive of all people, including those who visited the study
destination and those who did not.

Visitors & Non-visitors: There are three types of visitors in the secondary data
set used in this study: (1) those who visited the study destination on their most recent trip
within the past 12 months, (2) those who visited the study destination in a less recent trip
within the past 12 months, and (3) those who visited the study destination sometime but
not within the past 12 months. To be all-inclusive, all of these visitor types are included
in the image measure analysis, including descriptives and factor analysis of image
measures. Hence, in the image analysis part, ‘visitors’ is used to refer to all three types of
visitors and ‘non—visitors’ is used to refer to those who have never visited the study
destination.

Visitors vs. Recent Visitors: In the recent visitor segment model analysis, only
the two groups of visitors mentioned above, namely the recent visitors who visited the
study destination within the past 12 months, are included in the analysis. The third group
of visitors mentioned above (those who visited the study destination sometime but not

within the past 12 months) is excluded from the analysis since there is no information in

19

the dataset

coined to C

In (
provided (‘
concept inc

role in the r:

    
  
   
 

methodoloc:
int‘estieate t
pepulatiort c

MY results

 

destination j
5m.“ results

the dataset about their trip to the study destination. Thus, the ‘recent visitors’ name is

coined to differentiate this group from ‘visitors’.

Summary of the Content of the Dissertation

In Chapter 1, the general background, reasoning and hypotheses of this study are
provided. Chapter 2 summarizes a review of the literature related to the destination image
concept including its inﬂuence on tourist behavior and determinant factors that play a
role in the image formation process. This chapter also includes a literature review on the
methodological aspects of this study. Chapter 3 contains the research methods utilized to
investigate the research questions of this study, including the study design, the study
population, data collection mode, data preparation and analysis tools. Chapter 4 contains
study results and discussion, including the proﬁle of mixed sample and visitor sample,
destination image measured, the impact of selected determinant variables along with the
general population and the recent visitor segment models. In Chapter 5, a summary of
study results along with managerial, theoretical and methodological implications are

provided, ending with future research suggestions.

20

 

Th;
relationshi;
namely. 5.
methodolog
existing ht
destination
organized z
and ('3) the
Variables i
travel be'
encompas
COliecu‘On

{3301’ an;

CHAPTER II

LITERATURE REVIEW

The purpose of this study is to build on the existing knowledge about the intricate
relationships between destination image and various possible determinant variables,
namely, selected respondent sociodemographics and past travel behavior and a
methodological variable (the season of the survey). This chapter provides a summary of
existing literature, conceptual and empirical, related to the relationships between
destination image and these possible determinant variables. The literature review is
organized around two subjects: (1) the concepts related to the subject matter of this study,
and (2) the methods utilized in this study. The conceptual review entails the four sets of
variables included in this study: (1) destination image (2) sociodemographics, (3) past
travel behavior, and (4) a methodological variable. The methodological review
encompasses a discussion of: (1) the computer-assisted telephone interview, data
collection method used in this study, and (2) the analysis tools utilized in this study (i.e.,

factor analysis and regression analysis).

Destination Image
Destination image, which is “the sum of beliefs, ideas, and impressions that a
person has of a destination” (Crompton 1979, p. 18), is an intricate concept with links to
various possible determinants. Many researchers agree that destination image has two
components although they use different terminology for them: cognitive and affective

components. Gartner (1993), who is a widely cited destination image researcher,

21

delineates
attributes
behind the
from a de
Ganner's
towards a
destination
hierarchica
action com
State that
component
image defrr
empirical s
dE‘StlnaIIOn
beIICI' pred:
Des:
desnnanen
destination
l992; “Um
638;“ On S
and PTOmm
Benedetto. }

YOung l 995

delineates the cognitive component as a factual and intellectual evaluation of the known
attributes of a destination. According to him, human motives are the deﬁning force
behind the affective component of image; motives determine what people wish to obtain
from a destination, and thus, affect how people evaluate that destination. Therefore, in
Gartner’s contention, the affective component, comprised of attitudes and feelings
towards a destination, becomes relevant only when a potential visitor starts evaluating
destinations on his/her choice set. Gartner also maintains that these components are
hierarchically interrelated, and he adds another component to this hierarchy: conative, the
action component, visiting the destination in this case. Baloglu and McCleary (1992) also
state that cognitive and affective components are interrelated, and the affective
component depends on and is a ﬁrnction of the cognitive component. Although many
image deﬁnitions recognize both cognitive and affective components, they suggest that
empirical studies have largely focused on the perceptual or cognitive component of
destination image. They purport that these components should be measured separately to
better predict behavior.

Destination image is an important aspect in successful tourism development and
destination marketing. Some researchers relate this importance to the overall success of a
destination in tourism (Chen & Kerstetter 1999; Crompton 1979; Dadgostar & Isotalo
1992; Hunt 1975). Some researchers attribute destination image’s importance to its
effects on supply side aspects, namely, marketing related variables, such as positioning
and promotion (Baloglu & Brinberg 1997; Baloglu & McCleary 1999; Calantone,
Benedetto, Hakam, & Bojanic 1989; Chen & Kerstetter 1999; Fridgen 1987; Walmsley &

Young 1998). Still others relate the importance of destination image to its effect on

22

 

pie-trip, dt

remain pr:
likelihood
Vmabies,
Variables
recommf

destinati.

demand side aspects, such as tourist behavior, especially decision-making (Alhemoud &
Armstrong 1996; Baloglu & Brinberg 1997; Bramwell & Rawding 1996; Chen & Hsu
2000; Chen & Kerstetter 1999; Coshall 2000; Crompton 1979; Dadgostar & Isotalo 1992;
Dann 1996; Fakeye & Crompton 1991; Gartner 1993; Goodrich 1978a; Hunt 1975;
Fridgen 1987; MacKay & Fesenmaier 1997; Mayo 1973; Mayo & Jarvis 1981; Tapachai
& Wariszak 2000; Walmsley & Young 1998).

Destination image researchers have postulated that destination image is related to
pre—trip, during and post-trip behavior of tourists, although many of these relationships
remain propositional rather than empirical. Some of these pre-visit variables include
likelihood and intention to visit, planning time frame, anticipation, and decision-making
variables, such as destination choice, time and money to spend. Some of the during trip
variables include actual time spent at the destination, tourist enjoyment and satisfaction.
Finally, some of the post-trip variables include positive evaluation of the destination,
recommending the destination to the others (word-of-mouth), intentions to revisit the

destination, and destination loyalty.

The impact of Destination Image on Pre-Visit Tourist Behavior Variables
There are empirical ﬁndings that reﬂect on the relationship between destination
image and pre-visit variables. Chen and Hsu (2000) surveyed Koreans traveling to
overseas destinations at an international airport when they were about to depart, and they
found a signiﬁcant relationship between the ‘low travel cost’ dimension and tourists’ trip
planning time ﬁ'ames (i.e., how far ahead they start planning for the trip). Studying the

image of rural Pennsylvania held by university students, Chen and Kerstetter (1999)

23

 

'ma:
thei:
spet

felt ‘

found that respondents with intentions to travel to rural areas within 12 months were
more likely to have positive images of ‘tourism infrastructure’ and ‘natural amenity’
image dimensions than those respondents who did not intend to visit the area. Similarly,
Court and Lupton (1997), studying US respondents’ image of New Mexico, found that
certain dimensions positively inﬂuenced the likelihood of visiting New Mexico; ‘natural
and cultural amenities’ and ‘participative recreational opportunities’ increased the
likelihood of visiting the state.

Chen and Hsu (2000) also revealed that respondents who scored high on the
‘many interesting places to visit’ image dimension planned to spend much less time at
their destination than those who scored higher. They suggested that those who could not
spend enough time at the destination to enjoy all of its touristic opportunities might have
felt that there were more things to see and do, while those with an ample amount of time
might not have felt that way. On the other hand, those visitors who scored high on the
‘similar lifestyles’ image dimension planned to spend more time than those who scored
low on that dimension. They concluded that those who perceived the destination as
offering lifestyles similar to their own might have felt conﬁdent enough to plan to stay
longer than those who did not think so.

Chen and Hsu (2000) also found that Korean travelers’ images of overseas
destinations in terms of ‘similar lifestyles’, along with ‘plenty of quality restaurants’, and
‘no language barriers’ affected their budgeted travel costs. Those visitors who rated a
destination high on ‘similar lifestyles’ and ‘no language barriers’ planned to spend more
money than those who rated these image dimensions low; they concluded that this could

be due to their level of comfort when dealing with the businesses at a place with no

24

II—-_

    

responder.
the others

In.
mostly pro
potential It
the desnne
representa
that destt
deterrnint

they deri

discuss:
Of irnaé
amOng
(eihme

BUIIQ

language barriers. However, surprisingly, the effect of the ‘plenty of quality restaurants’
dimension on the budgeted travel cost was negative. They suggested that those
respondents with limited budgets might have paid more attention to this dimension than
the others did.

The literature on the relationship between destination image and anticipation is
mostly propositional. Researchers have proposed that the image of a destination gives a
potential tourist a special type of trial, a pre-taste, by representing the objective reality of
the destination (Chen & Kerstetter 1999; Fakeye & Crompton 1991). Depending on this
representation, tourists generate expectations about what their travel experience will be at
that destination (Britton 1979; Coshall 2000; Fridgen 1987). Such expectations may also
detemtine their travel choices about the destination, as well as the level of satisfaction
they derive from visiting (Coshall 2000).

Destination choice is one of the most oﬁen mentioned pre-trip variables in
discussions of the importance of destination image. Although the literature on the effect
of image on consumers’ destination choices remains propositional, there is a consensus
among researchers that destination image inﬂuences tourists’ travel destination choices
(Ahmed 1996; Alhemoud & Armstrong 1996; Bojanic 1991; Bramwell & Rawding 1996;
Britton 1979; Chen & Kerstetter 1999; Coshall 2000; Court & Lupton I997; Dann 1996;
Embacher & Buttle 1989; Fakeye & Crompton 1991; Fridgen 1987; Gartner 1989;
Gartner 1993; Goodrich 1978a; Hunt 1975; Joppe, Martin, & Waalen 2001; Mayo 1973;
Mayo & Jarvis 1981; Milman & Pizam 1995; Murphy 1999; Tapachai & Wariszak 2000).
It is postulated that image represents the destination in people’s minds whether it be

based on reality or not, and affects consumers decisions most of the time since consumers '

 

usually hal
reality (if?
d: C romptl
Wariszak 2
SO:
images of

Kerstetter I.

its positive

 

hand. state<
negative c0]
image musr
Armstrong
should has-e
expeﬁences
‘IBIamts'ell 4
Dub
destinaIIOn
Climate rmg
0n {OUTiSIS‘
images ma}
ITaVelers H
Who ”Walt

usually have a limited knowledge of destinations based on experience and objective
reality (Alhemoud & Armstrong 1996; Chen & Kerstetter 1999; Crompton 1979; Fakeye
& Crompton 1991; Gartner 1989; Gartner 1993; Goodrich 1978a; Hunt 1975; Tapachai &
Wariszak 2000).

Some researchers relate consumers’ destination choices to their positive overall
images of destinations (Chen & Kerstetter 1999; Milman & Pizam 1995). Chen and
Kerstetter (1999) postulate that tourists would choose a destination over others only when
its positive image aspects exceed its negative image aspects. Ahmed (1996), on the other
hand, states that for a destination to be chosen, it should be the least likely to have
negative consequences among other choices. Some researchers state that the destination’s
image must be not only positive but also strong to be chosen by travelers (Alhemoud &
Armstrong 1996; Hunt 1975; Ross 1993). Yet, another proposition is that destinations
should have distinctive images suggesting features different from tourists’ everyday life
experiences in order to be appealing to the tourists and to be chosen for their travel
(Bramwell & Rawding 1996).

Dubbing the image construct as a ‘conceptual appeal’, Hunt (1975) proposes that
destination image dimensions pertaining to resident population, natural environment, and
climate might be more inﬂuential than. recreational attractions and activities dimensions
on tourists’ destination choices. Hunt also implies that destinations with grossly exotic
images may not be chosen since such qualities might prove discomforting to potential
travelers. Hunt’s proposition is supported by ﬁndings of MacKay and Fesenmaier (1997),
who revealed that visuals depicting destinations with unique features caused anxiety in

subjects.

26

(JopP€~

(Gartner
de5tinati.
proxide t

the destin

The ir
Th:
\isit touns
along saith
Pursuit of r
found a rela
the Wet Tl’t
Variance in
Visitors “he
enl0}Tnent ti
DOSlIl\'el_\-_ Dr
found to be r
il’an’ance in

outStandl‘nQ d;

 

tourist infom

 

Other researchers base tourists’ destination choices on their needs and wants
(Joppe, Martin, & Waalen 2001) and beneﬁts that the destination is believed to offer
(Gartner 1989; Tapachai & Wariszak 2000). Speciﬁcally, Gartner stresses that the
destination choice depends on a “beneﬁt package, unique to the destination, expected to
provide the greatest intrinsic reward to the traveler” which is derived from the image of

the destination (1989, p. 16).

The impact of Destination Image on During Visit Tourist Behavior Variables

There are empirical ﬁndings on the impact of destination image on some during
visit tourist behaviors. Dadgostar and Isotalo (1992) found a moderate effect of image
along with distance, income, age and importance of cost on the length of time spent in
pursuit of recreational activities by tourists at near-home city destinations. Ross (1993)
found a relationship between backpacker visitors’ level of enjoyment and their images of
the Wet Tropics region of Northern Australia in terms of dimensions of ‘authenticity’,
‘variance in physical environment’, and ‘ﬁiendliness of local residents’ dimensions.
Visitors who rated these dimensions more positively also reported higher levels of
enjoyment than expected on their trip than those who rated these dimensions less
positively. Dimensions of ‘tourist information’ and ‘suitable accommodation’ were not
found to be related to tourists’ levels of enjoyment. Ross suggested that ‘authenticity’,
‘variance in physical environment’, and ‘friendliness of local residents’ might be
outstanding destination characteristics inﬂuencing tourists’ levels of enjoyment; whereas,
‘tourist information’ and ‘suitable accommodation’ might be taken for granted and be

expected to exist in any tourist destination.

27

destinati-o

 

of the anti
prior to \
\lSllOTS u
Armstrong
Fakeye ar
destinatior
Valle}; Te
expectatio'
these p60]
positixe ir

dOCumeme

The i

Ro
Nonhem .
reSandEm
TESidemS: ‘
dimenSiQn.
retommen

intend to I

Satisfaction is another during visit variable postulated to be inﬂuenced by
destination image. Satisfaction with a destination experience depends on the realization
of the anticipations and expectations generated based on the image of the destination held
prior to visitation; if the destination lives up to these expectations and anticipations,
visitors would experience satisfaction, if not, dissatisfaction would occur (Alhemoud &
Armstrong 1996; Britton 1979; Fakeye & Crompton 1991; Fridgen 1987; ROSS I993).
Fakeye and Crompton (1991) found that the expectations of non-visitors on some
destination dimensions exceeded what the actual visitors reported about the Rio Grande
Valley, Texas. Knowing that the result is likely to be a dissatisﬁed visitor whenever
expectations exceed performance, they cautioned that dissatisfaction would be likely if
these people visited the study destination. However, satisfaction due to realistically
positive images or dissatisfaction due to unrealistically positive images has not been

documented empirically.

The impact of Destination Image on Post-Visit Tourist Behavior Variables

Ross (1993), studying the backpacker visitors to the Wet Tropics region of
Northern Australia, found correlations between some destination image dimensions and
respondents’ evaluative responses. Those backpackers who scored high on ‘ﬁiendly local
residents’, ‘varied physical environment’, and ‘the authenticity of the destination’
dimensions were more likely to rate the destination as an ideal holiday destination and
recommend it to their family and ﬁiends. Ross also found that, if visitors had a positive
image of destination in terms of the ‘friendly locals’ dimension, they were more likely to

intend to revisit the destination. Other researchers have postulated that some dimensions,

28

such as
relatedi

i
DakO‘a‘
suPpOn I
image ’1
reponed
more trip

would boC

In 2
images for 1
complex ”2
determinants

Image
destination 0
selCCtEd b)’ lb
& Lupton l9‘.
formed throu
(Alhemoud 8;

Gunn 1972) a

McCleary 1999

such as ‘different cultural experience’ and ‘convenient transportation’ are signiﬁcantly
related to destination loyalty (cf. Joppe, Martin, & Waalen 2001, p. 523).

Schroeder (1996) found positive relationships between residents’ image of North
Dakota and both the likelihood of their recommending it as a travel destination and their
support of tourism development. Schroeder also found a relationship between residents’
image and their choices of recreation destinations. Residents with positive images
reported more trips within their state than those with less positive images, who reported
more trips outside their state. Schroeder suggests that improving the residents’ image

would boost the within-the-state travels of residents.

Determinants of Destination Image

In an effort to help destination authorities who are striving to portray positive
images for their destinations, researchers have tried to enhance their understanding of the
complex nature of the destination image construct, especially its formation and
determinants playing a role in this formation.

Image formation is deﬁned as the construction of a mental representation of a
destination on the basis of information cues delivered by the image formation agents
selected by the person (Alhemoud & Armstrong 1996; Bramwell & Rawding 1996; Court
& Lupton 1997; Gartner 1993). The image construct is purported to include perceptions
formed through exposure to commercial or non-commercial information sources
(Alhemoud & Armstrong 1996; Bojanic 1991; Court & Lupton 1997; Gartner 1993;
Gunn 1972) and through personal experience (Baloglu & Brinberg 1997; Baloglu &

McCleary 1999; Chen & Kerstetter 1999; Court & Lupton 1997; Crompton 1979; Dann

29

indrx
desri
Dest
than
Fese:
desti
due I

C0111."

dEStin,
“IEraI(
and 0
Inform

fiomfl

1996; Fakeye & Crompton 1991; Fridgen 1987; Hu & Ritchie 1993; MacKay &
Fesenmaier 1997; Ross 1993; Walmsley & Young 1998).

MacKay and Fesenmaier contend that destination image is a “composite of
individual inputs and marketer inputs” (1997, p. 559). However, other researchers make
destination image more than a function of inputs from marketers and individuals.
Destination marketers engage in promotional efforts to project positive images or to
change a negative image (Bramwell & Rawding 1996; Court & Lupton 1997; MacKay &
Fesenmaier 1997; MacKay & Fesenmaier 2000). However, the images projected by
destination marketers are not the same as those images received by the targeted markets
due to the following factors: (1) the original message can be altered by the very source
communicating this message (Bramwell & Rawding 1996), (2) the message can be
modiﬁed by the perceiver (Court & Lupton 1997), and most important of all, (3) these
destination-originated messages are not the only messages reaching the perceivers in the
image formation process. It is agreed that information about destination image formation
comes from a wider spectrum of sources than those for consumer products or other
services (Alhemoud & Armstrong 1996; Echtner & Ritchie 1991; Gartner 1993; Murphy,
1999; Selby & Morgan 1996).

Induced and organic image formation agents are the two main sources of
destination image information that have been identiﬁed in the destination image
literature. Induced agents primarily include promotional materials from the destination,
and organic agents are distributed primarily by non-touristic and non-commercial
information sources, such as popular culture, education, general media, word-of-mouth

ﬁom family and friends and actual visitation (Gartner 1993; Gunn 1972). As stated by

30

 

Selby and

the skillti

informat
lbalog‘:
( l 997 )
comm
choice
in acti
genen
image
destir
belie:
the t

the 1

mot.

Selby and Morgan (1996), these agents are not mutually exclusive in practice because of
the skillful media relations of destination marketers.

Some researchers argue that familiarity with a destination through tourist
information (induced agents) plays an important role in tourists’ destination preferences
(Baloglu 2001; Court & Lupton 1997; Woodside & Lysonski 1989). Court and Lupton
(1997) recognize the importance of well-formulated and targeted marketing
communication, since, they postulate, tourist information inﬂuences tourists’ destination
choices by generating awareness and interest, which thus stimulates desire which results
in action. Nevertheless, Milman and Pizam (1995) argue that awareness by itself may not
generate interest and a purchase decision. They purport that awareness means having an
image of a destination, which may, at best, result in curiosity to learn more about the
destination. Unless the image is positive, visitation will not occur. Similarly, Gartner
believes that “awareness implies that an image of the destination exists in the mind(s) of
the decision makers,” which puts the destination into the ‘realizable opportunity set’ at
the time of the decision making; unless the destination has “a strong image for the types
of activities deemed important to the decision making group or individual” the
destination will be eliminated from the ‘opportunity set’ (1993, p. 196). Gartner also
postulates that the ‘opportunity set’ is reduced, step by step, to the ‘consideration set’, the
‘choice set’, the ‘evoked set’, and ﬁnally the ‘decision set’ by evaluating destination
attributes in terms of the expected returns, with the affective component of image being
operational throughout this evaluation process.

Some of the autonomous agents, namely news articles, educational materials,

movies, and popular culture, are postulated to be more inﬂuential on destination image

31

formation becaUS
does destination-
such as natural ar
Shen 1992; Albert

It is also is
destination image
purported to be th
and form images
Gartner 1903). As
the combination c
received image is c
formed through t
messages Perceise:
play a role in this in

of their relationship:

chapter.

Yet, another

researcher measurir.

pottrrlate that researl

 

to want. This co.

methods while measI

l
o

dlﬁelenl Ways duE

 

formation because they have higher credibility and ability to reach mass crowds than
does destination-originated information, especially when they depict a dramatic event,
such as natural and ‘humancaused’ disasters (Crompton 1979; Gartner 1993; Gartner &
Shen 1992; Alhemoud & Armstrong 1996; Sonmez, Apostolopoulos & Tarlow 1999).

It is also well accepted that induced and organic agents are not the only sources of
destination image detemtinants. The third source of destination image determinants is
purported to be the consumers (perceivers) who ﬁlter the information from these sources
and form images about travel destinations (Bramwell & Rawding 1996; Dann 1996;
Gartner 1993). As distinguished by Bramwell and Rawding (1996), a projected image is
the combination of messages and impressions created about a destination, while a
received image is consumers’ unique mental construct or representation of the destination
formed through their comprehension, understanding, and interpretation of these
messages. Perceivers’ sociodemographics, as well as past travel behavior, are assumed to
play a role in this image formation process, and therefore, have been investigated in terms
of their relationships with destination image, which will be discussed in detail later in this
chapter.

Yet, another source of destination image determinants is postulated to be the
researcher measuring the image held by the perceivers. Dadgostar and Isotalo (1992)
postulate that researchers impact a destination’s image by interpreting the data the way
they want. This could be a valid statement since different researchers use different
methods while measuring destination image, and they might interpret the same data in

different ways due to differing backgrounds and personal characteristics.

32

Thus.
destination
destination.
perceivers c
methodologi
Since some
investigated

section will

Sociodemn

Sint
sociodemo
Variables h
These \‘ari
TBSearch a
Sociodemc
(“Sldence

and more;

Thus, as depicted in Figure 3 in Chapter 1, there are 4 different sources of
destination image determinants: (1) commercial information sourcing from the
destination, (2) information sourcing from uncontrollable autonomous agents, (3)
perceivers’ characteristics, including sociodemographics and past travel behavior, and (4)
methodological factors that play a role while measuring the image held by the perceivers.
Since some sociodemographics, past travel behavior, and a methodological variable are
investigated in terms of their inﬂuence on destination image in this study, the following

section will provide a detailed review of studies dealing with these determinants.

Sociodemographics

Since different people demand and consume different products based on their
sociodemographic characteristics, such as age, gender, income, and residence, these
variables have been a focus of investigation in academic ﬁelds that deal with consumers.
These variables have been used as basic segmentation variables in consumer behavior
research as well as in tourism research. Several tourism researchers have included
sociodemographic characteristics in their studies to explain tourist behavior. Distance
(residence) and income have been found to be inﬂuential on the consumption of leisure
and recreation products (Court & Lupton 1997; Dadgostar & Isotalo 1992). The
importance of gender as another demographic variable has also been documented by
some tourism researchers (Gentry & Doering 1979; McQueen & Miller 1985; Schul &
Crompton 1983; Snepenger, Meged, Snelling, & Worrall 1990; Woodside & Pitts 1976).
Travelers’ age (Etzel & Woodside 1982; McQueen & Miller 1985; Schul & Crompton

1983), level of education (Schul & Crompton 1983), household size (Court and Lupton

33

1997), and family life cycle (Bojanic 1992; Dadgostar & Isotalo 1992) are other
sociodemographic variables that have been found to explain tourist behavior.

Ashworth (1989) argues that there is a relationship between travelers’
characteristics and the images they hold about a destination. Mayo and Jarvis state “no
two people see a destination in exactly the same way” (1981, p. 42). Gartner (1993)
suggests that choosing the appropriate image formation agents depends on not only
destination related variables, but also travel behavior and characteristics of the target
market, such as income, age, and education. Hunt (1975), on the other hand, while
recognizing the possible effect of distance (residence) on destination image, discounts the
importance of socio-economic factors. He accepts the possible systematic exclusion of
certain groups in the data he analyzed, yet proposes “brand or product image to be
independent of consumer socio economic class” (p. 2).

Some sociodemographic variables have been investigated for their inﬂuence on
destination image. Results from some of these studies are summarized in Table 1. Some
researchers have examined the relationships between the image of a destination and the
respondents’ distance from this destination, region or origin of residence, and home
country. Ahmed examined the differences in image due to distance from a destination,
recognizing that there might be “regional differences in taste preferences, value systems,
norms, attitudes, states of mind, and sub-cultures” even though both study and sample
destinations were within the US (1996, p. 42). Others realized the effect of distance in
terms of overall familiarity and knowledge with the destination (Crompton 1979;

Goodrich 1978b; Fakeye & Crompton 1991; Hunt 1975; Walmsley & Young 1998).

34

 

Table l. Resu

destt
f
ReseanhertSl

Crompton r 1§“’~‘

falteve 8;
Crompton t 1091

Ahmed l HUT) j

Alhemoud 8;
ARDSUOng 1' :ss

\‘j’almsle'x 5;
bung r lung

,' 'k. -
ll‘qu\ QR

Table I. Results of studies investigating the
destination image.

inﬂuence of sociodemographics on

 

 

MacKay (2001 )

destinations

_
Researcher(s) Study sampje Influential Area(s) of Difference(s)
Destination(s) Destination(s) Vanables
Hunt (1975) Rocky Mountain US Distance Further respondents could not
states-Colorado, differentiate between areas
Montana, Utah, within study regions as well
Wyoming as closer respondents
Crompton (1979) Mexico US Distance Further respondents had more
favorable images of
destination as a vacation
destination
Fakeye & Rio Grande Valley US Distance Infrastructure, food, and
Crompton (1991) in Texas friendly people
Ahmed (l 996) Utah US Region of Outdoor recreation resources,
residence outdoor recreation activities,
culture, liquor laws
and overall image
Alhemoud & Kuwait Kuwait (residents Age Older chose shopping,
Armstrong (1996) and foreigners) younger chose nightlife
Distance Foreigners chose cultural
attractions, local students
chose manufactured
attractions
Religion Foreigners were not aware of
national Islamic Museum
Walmsley & Local attractions in Sydney/ Australia Residence/ Evaluative schema frt
Young (1998) Sydney, US, UK, Distance international destinations but
New Zealand, Bali, not local
Hong Kong,
Singapore. Fiji.
Thailand
Chen & Kerstetter Pennsylvania. US US Gender Tourism infrastructure,
(1999) natural amenity
Household Tourism infrastructure,
status natural amenity
Education Natural amenity
Home country Tourism infrastructure.
atmosphere, natural amenity,
farm life
MacKay & Alberta, Canada. US and Taiwan Home country Number and interpretation of
F esenmaier dimensions
(2000)
MacKay & Riding Mountain Canada Age No difference
F esenmaier National Park, Marital status No difference
(1997) Manitoba/Canada Gender Holiday, atmosphere
Income Holiday and activity
Joppe, Martin, & Toronto Visitors Home country Attributes seen as important
Waalen (2001 ) and levels of satisfaction
Smith and Pictures of various Canada Age No difference in memory of

visual stimuli

 

35

\\'a
and linou'l
from 3 var
through it
common 5
local ones
promotion

C h
inﬂuence .
differences
(1996,) ide
l'lOWeVep
Fesenmaie
using visu
between t
dimensions
cautioned
distinctjye
held by Ja
intercepting
They fOUnc

germ” incr

Walmsley and Young predict that local images are based on personal experience
and knowledge through “long-term assimilation of place-related information gleaned
from a variety of everyday sources” while international images are more likely to develop
through induced information agents (1998, p. 66). They found that their proposed
common schema for evaluating places ﬁt international destination evaluations but not
local ones, implying that local level images are based on intimate factors rather than
promotional materials.

Chen and Kerstetter (1999) and Joppe, Martin, and Waalen (2001) tested the
inﬂuence of respondents’ home countries on destination image and found signiﬁcant
differences among respondents from different countries. Alhemoud and Armstrong
(1996) identiﬁed image differences between locals and foreigners. It should be noted,
however, that the foreigner population resided in the study destination. MacKay and
Fesenmaier (2000) also investigated the effect of home countries on destination image by
using visual representation without the name of the destination; they found differences
between the US and Taiwanese respondents in terms of the number of underlying
dimensions they identiﬁed and their interpretations of these dimensions. Thus, they
cautioned against using the same visuals consistently across different and culturally
distinctive segments. Yong and Gartner (in press) studied the perceptions of Seoul, Korea
held by Japanese, North American and European business and pleasure travelers by
intercepting them at the international airport in Seoul between May and August 2000.
They found signiﬁcant differences across respondents’ nationalities after controlling for

gender, income, and prior visitation to the destination.

36

destina:
income
amenity
educati
both n
cognitir
of age
demogi
Alhemt
respon-

destina‘

dilierer
dimeng
Fesenn
dlli‘cter
lattivitj
Chen 2

dCSllna'

Other sociodemographic variables studied in terms of their inﬂuence on
destination image include, age, gender, household status, marital status, education, and
income. Chen and Kerstetter (1999) found an inﬂuence of education on the ‘natural
amenity’ image factor of Pennsylvania. Baloglu (2001) has suggested that age and
education inﬂuence destination image by interacting with familiarity. He investigated
both main and interaction effects of these variables for their inﬂuence not only on
cognitive but also affective and overall images of Turkey. He found only the main effect
of age on the ‘attractions’ dimension. Recognizing the limitations of his study in terms of
demographic proﬁles of respondents, he recommended ﬁrture research on the issue.
Alhemoud and Armstrong (1996) also found differences between old and young
respondents while MacKay and Fesenmaier (1997) did not ﬁnd any inﬂuence of age on
destination image.

Chen and Kerstetter (1999) and MacKay and Fesenmaier (1997) found
differences between male and female respondents in terms of some image factors and
dimensions. Another sociodemographic variable found to be inﬂuential by MacKay and
Fesenmaier (1997) was income; depending on their income level, respondents had
different perceptions of Riding Mountain National park in terms of ‘holiday’ and
‘activity’ image factors. Their ﬁndings were not signiﬁcant for the marital status variable.
Chen and Kerstetter (1999) found an inﬂuence of respondents’ household status on

destination image.

37

Past Travel Behavior

The effect of past travel behavior on destination image is one of the most
discussed and tested aspects of destination image formation. The following section
contains a discussion of previous studies of the inﬂuence of past travel behavior
including: the amount of overall travel experience, prior visitation to the study
destination, the amount of prior visitation to a destination, the length of stay at a
destination, the number of activities participated in while at a destination, and the season

of the visit.

Overall Travel Experience

The impact of the amount of overall travel experience has not been studied
directly; however, implications of its role can be drawn from the existing destination
image literature. While studying residents’ image of North Dakota, Schroeder (1996)
found a relationship between destination image and the extent of residents’ travel
experience within their state. Residents who reported more out-of-state trips had less
positive images of North Dakota than those who reported more within-the-state trips.
Schroeder (1996) interpreted this ﬁnding as an effect of image rather than as a
determinant of image. In other words, he thought that those residents who had more out-
of-state trips did not travel within their state because of their poor images of North
Dakota. Therefore, recognizing that causality was not demonstrated, he suggests that
improving residents’ image would increase within-the-state travel of residents (p. 73).
However, this ﬁnding could also be interpreted as evidence for overall travel experience

being a possible determinant of destination image. Respondents with a narrow range of

38

destination
state more
have rated
have visitec
could come

desirable as

Pric

The
destination
However, 5
Hunt (1975
images of
distant res;
resPOI'tdentg
Sugé’ﬁ‘sted t
(:0an bet

Fakeye and

destination experiences might have a narrow point of reference, thus, comparing their
state more positively, while respondents with more diverse destination experiences might
have rated the state poorly, by comparing it with more popular tourist destinations they
have visited. Tourists who travel widely are more aware of what is available; also, they
could come to a point of saturation in terms of their perception of destinations as

desirable as they get more experience.

Prior Visitation

The impact of actual visitation is also a widely discussed and studied aspect of
destination image. Results from some of these studies are summarized in Table 2.
However, studies of the inﬂuence of previous visitation have arrived at different results.
Hunt (1975) and Chen and Kerstetter (1999) did not ﬁnd signiﬁcant differences between
images of visitors and non-visitors. Crompton (1979), on the other hand, found that
distant respondents rated Mexico more favorably as a vacation destination than did
respondents living in border communities. In a different study, Fakeye and Crompton
suggested that the reason for such differences to be due to “shallowness of much of the
contact between tourists and local culture...that is ‘pseudo-experience’” (1991, p.11).
Fakeye and Crompton (1991) compared the images of ﬁrst-time visitors, repeat visitors,
and non-visitors, and found image differences between visitors and non-visitors; non-
visitors had better images on two and worse images on three of ﬁve image factors of the

Rio Grande Valley in Texas.

39

Table 2. Results of studies investigating the inﬂuence of prior visitation on destination

 

 

 

image.
Researcher(s) Study Destination(s) Sample Differences Between
Destination(s) Visitors & Non-visitors
Crompton (1979) Mexico US Different regions on
different dimensions
Fakeye & Crompton (1991) Rio Grande Valley in US Social opportunities and
Texas attractions. natural and
cultural amenities,
accommodations and
transportation, infrastructure
foods and friendly people,
bars and evening
entertainment
Hu & Ritchie (1993) Hawaii, Greece, Canada Different dimensions for
Australia. F rance. different countries in
China different contexts
Ahmed (1996) Utah US Outdoor recreation
resources. culture, nightlife
and liquor laws, overall
image
Milman & Pizam (1995) Central Florida US Several product. and
environment-related
dimensions
Selby & Morgan (1996) A seaside resort in Visitors to Prejudice in cleanliness
Barry Island/ South surrounding
Wales area
Baloglu & McCleary Turkey, Egypt. Greece. US Several dimensions, more
(1999) Italy differences in affective
dimensions
Chen & Kerstetter (1999) Pennsylvania. US US No difference
MacKay & Fesenmaier Riding Mountain Canada Activity, familiarity,
(1997) National Park, holiday, atmosphere
Manitoba/Canada Familiars evaluated visuals

affectively, unfamiliars
cognitively

 

Some researchers have found the effect of visitation on different components of

image. Both Baloglu (2001) and Baloglu and McCleary (1999) found differences

between visitors and non-visitors in terms of cognitive and affective image components

and overall image. Baloglu and McCleary (1999) compared the images of four

40

Mediterranean I
image of these
differences beu
\isited Turkey
did not. Ahrr
xisirors. Like
responses of
Familiar my.
picture) , _. 1
Mo:

have used
differences
mlglmdin
image On

Press, Vc

I”“flers

phﬁ\'ps t

Mediterranean countries and concluded that visitation altered all components of the
image of these destinations although they did not test for the significance of the
differences between visitors and non-visitors. Baloglu (2001) found that respondents who
visited Turkey had more positive responses for all components of image than those who
did not. Ahmed (1996) also found differences in overall images of visitors and non-
visitors. Likewise, MacKay and Fesenmaier (1997) have found differences between the
responses of the familiar and unfamiliar respondents among focus group participants.
Familiar respondents, being able “to see and project beyond the visual stimuli (actual
picture). provided more affective and emotional descriptions” (p. 558).

Most of the studies investigating the impact of visitation on destination image
have used mixed samples including both visitors and non-visitors and have noted
differences between the images of these two groups. A few other studies have used
longitudinal data with pre-trip, during trip, and/or post-trip measurements of destination
image on the same sample (Dann 1996; Pearce 1982; Phelps 1986; Vogt & Andereck, in
press; Vogt & Stewart 1998). Pearce (1982) compared pre-trip and post-trip images of
travelers to two Mediterranean countries and found image change for both countries.
Phelps (1986) has also found variations between the pre and post-trip images of travelers.

Dann (1996) measured pre-trip and during-trip images of travelers to Barbados
and found differences for both cognitive and affective components of image as well as
overall image. Vogt and Andereck (in press) studied the pre-trip, during-trip, and post-
trip images of motorist travelers to Arizona; they found increases in both the cognitive
component (knowledge level) and the affective component (desirability level) of image

as the trip progressed. The differences between these three level images (pre-trip, during

41

and Posi'mp) u

 

component. The"
due to high deg:
siewm (1998) 31

In SPiIe C
among destinallor
images of it due
between imagCS P
can be seen in Tab

rabi3D' "

Researche

(19m and 031
Goodnch (1978)
Fakeye & Crompton
ROSS (1993)
Selim & Morgan l 1‘”-
\

 

Although G
organic. many Olhf
Since it includes ]
Crompton 1979; F
Morgan 1996). Fal

organic image. and

Complex because it

 

and post-trip) were signiﬁcant for the cognitive component but not for the affective
component. They postulated that the affective component did not change much, maybe
due to high destination desirability felt by the travelers at the onset of a trip. Vogt and
Stewart (1998) also found similar results.

In spite of the variation across results of different studies, it is well-accepted
among destination image researchers that visiting a destination results in more realistic
images of it due to ﬁrst hand experience. Some researchers have tried to distinguish
between images prior to and after visitation by coining different terminology for them as
can be seen in Table 3.

Table 3. Different terminolog provided for images prior to and after actual visitation.

 

 

Researcher(s) Image before visitation Image after visitation
Gunn (1972) and Gartner (1993) Induced Orgamc
Goodrich (1978) Secondary Primary
Fakeye & Crompton (1991) Organic & Induced Complex
Ross (1993) Ideal Actual
Selby & Morgan (1996) Naive Re-evaluated

 

Although Gunn (1972) and Gartner (1993) considered image through visitation as
organic, many others considered image through visitation as a separate type of image
since it includes perception of the actual product (Alhemoud & Armstrong 1996;
Crompton 1979; Fakeye & Crompton 1991; MacKay & Fesenmaier I997; Selby &
Morgan 1996). Fakeye and Crompton (1991) separated this type of image from Gunn’s
organic image, and deemed it to be a third level, or ‘complex’ image. They assert that it is
complex because it allows a more differentiated outlook and truer comprehension of the
destination rather than simple stereotyping, especially if the visitor spends “enough time
there to be exposed to the destination's varying dimensions through developing contacts

and establishing relationships” (p. 11). Their assertion has also been widely accepted by

other researcl
Fesenmaier l‘.
Realiz
(1995) recon
would be be
Fakeye and
segments “'2
based on vi

strategies.

Visit

Not
investigated
mamglu 20
and the ler
exemplify St
the number
damnation i

0” dEStinatic

other researchers (Chen & Hsu 2000; Chen & Kerstetter 1999; Lubbe 1998; MacKay &
Fesenmaier 1997; Milman & Pizam 1995).

Realizing the impact of actual visitation on destination image, Milman and Pizam
(1995) recommend the use of different promotional tactics to induce visitation, which
would be beneﬁcial especially for the destinations with unrealistically negative images.
Fakeye and Crompton (1991) provide similar recommendations to induce visitation of
segments with different previous experiences. Ahmed (1996) suggests segmentation
based on visitors and non-visitors to develop appropriate positioning and promotion

strategies.

Visitation-Related Variables

Not only visitation itself, but also some visitation-related variables have been
investigated for their inﬂuence on destination image. The frequency of prior visitation
(Baloglu 2001; Fakeye & Crompton 1991; Schroeder 1996; Vogt & Andereck, in press)
and the length of stay (Fakeye & Crompton 1991; Vogt and Andereck, in press)
exemplify some of the visitation-related variables that have been investigated. Although
the number of activities and the season of visitation have not been studied directly by
destination image researchers, there are implications about the impact of these variables

on destination image within the existing destination image literature.
The Frequency of Prior Visitation

While comparing visitors’ and non-visitors’ images of four Mediterranean

destinations, Baloglu and McCleary (1999) deleted multiple visit respondents from the

43

b:

CI

dd
IA

cha

analysis since they believed that multiple visits would play an important role in
evaluations of a destination. Baloglu (2001) found an effect of the amount of prior
visitation on the affective component of image; respondents who had multiple visits to
Turkey had different image ratings from the others, including both visitors and non-
visitors. Multiple visit respondents gave higher ratings to ‘the quality of experience’ and
‘relaxing’ image dimensions as well as the overall image. Yong and Gartner (in press)
compared the images of Seoul, Korea held by business and pleasure travelers from Japan,
North America and Europe; they found that Japanese business travelers, who had the
most experience with Seoul, were more interested in functional attributes of Seoul,
attributes that made their trips routine, safe and productive rather than exciting and
adventurous. They concluded that as the amount of experience increases, a strong organic
image resistant to change forms; and, in this process, the features of the destination
become more tangible.

While comparing the images of non-visitors, first-time visitors and repeat visitors
of the Rio Grande Valley in Texas, Fakeye and Crompton (1991) did not ﬁnd such
differences between ﬁrst time and repeat visitors and concluded that the majority of the
change in destination image occurs during the ﬁrst visit, leaving little alterations for the
following visits. Their conclusion was supported by the results of Vogt and Andereck (in
press), who studied during-trip image change of Arizona travelers, and found that the
image change was larger for the ﬁrst-time visitors than for repeat visitors. Schroeder
(1996) also found that residents with more positive images reported more trips within
North Dakota than those with less positive images who reported more trips outside their

state.

44

The

Anot
length of st
Crompton (
better than
attractions
stay allows
realistic im
Andereck (-
alfeaive co
cOmponent

greater ima;

The

Ash
about a des
0f Studies i
VlSllOrs par
ﬁndings of
found that
stores fOr
De0p1e~ fa

acrixities e1

The Length of Stay

Another visitation-related variable studied by destination image researchers is the
length of stay (Fakeye & Crompton 1991; Vogt and Andereck, in press). Fakeye and
Crompton (1991) found that long stayers (over 8 weeks) rated the Rio Grande Valley
better than short stayers (8 weeks or less) on two factors: ‘social opportunities and
attractions’ and ‘infrastructure, foods and friendly people’. They concluded that longer
stay allows travelers to experience better social integration, thereby providing more
realistic images void of stereotypes about these aspects of a destination. Vogt and
Andereck (in press) found no signiﬁcant relationship between the length of stay and the
affective component of image; however, the relationship was signiﬁcant for the cognitive
component in an unexpected way. Repeat visitors who had short stays (1-3 days) had

greater image change than those who stayed longer (4-7 days or 8-30 days).

The Number of Activities

Ashworth (1989) argues that there is a relationship between images travelers hold
about a destination and activities they hope to engage in at the destination. There is a lack
of studies investigating the relationship between the activities or the amount of activities
visitors participate in at a destination and their images of the destination. However, the
ﬁndings of Fakeye and Crompton (1991) have implications for such a relationship. They
found that both ﬁrst-time and repeat visitors who stayed longer (over 8 weeks) had better
scores for ‘social opportunities and attractions’ and ‘infrastructure, foods and friendly
people’ factors. It can be postulated that long-stayers must have participated in more

activities encompassing the aspects of these factors, such as visiting attractions, attending

45

events, dining at restaurants, sightseeing, driving for pleasure, etc. In other words, the
longer visitors stay, the more things they do, thus, the more differentiated images they

have.

The Season of the Visit

As was mentioned before, there is a lack of focus on the relationship between
image of a destination held by visitors and the season of their visit to the destination.
However, there is an implication for the impact of this variable in Gartner’s (1986) study,
in which he investigates the temporal inﬂuences on destination image. He stresses that
tourism is both seasonal and depends on the environment, and he contends that the
seasons of participating in an activity at a destination might inﬂuence the image of this

activity.

Methodological Factors

The methodology literature suggests that, based on their sociodemographic
characteristics, research subjects’ responsiveness and truthfulness are affected by
methodological factors, such as research techniques, data collection modes, and forms of
questions; such effects are reported to be more signiﬁcant when trying to discover or
describe attitudes, especially consumer attitudes or socially stigmatized behaviors, such
as drug or alcohol use (Aquilino 1994; Churchill & Peter 1984; Ervin & Gilmore 1999;
Gazel, Schwer, & Daneshvary 1998; Grube 1997; Herzog Rodgers, & Kulka 1983; Kaye
& Johnson 1999; Krysan, Schuman, Scott, & Beatty 1994; McAuliffe, Geller, LaBrie,

Paletz, & Foumier 1998; Rockwood, Sangster, & Dillman 1997; Sniderman & Grob '

46

1996; Sudman & Bradbum 1973; Turner, Ku, & Rogers 1998; Wright, Aquilino, &
Supple 1998).

Few destination image researchers have touched upon the impact of
methodological factors on destination image. Dadgostar and Isotalo (1992) realized the
impact of the researcher through subjective data interpretation. Recognizing the
multicomponent nature of the destination image construct, Echtner and Ritchie (1993)
recommended the use of a multimethod approach, in which a comprehensive list of
structured and open-ended questions would be used not only in the scale development
stage but also the data collection stage. Several researchers have followed their
recommendations and incorporated qualitative techniques in their research (Bramwell &
Rawding, 1996; Choi, Chan, & Wu, 1999; Dann, 1996; Joppe, Martin, & Waalen, 2001;
Lubbe, 1998; MacKay & Fesenmaier, 1997; Milman & Pizam, 1995; Murphy, 1999;
Selby & Morgan, 1996; Tapachai & Waryszak, 2000). Some researchers have also tested
the applicability of new data analysis tools on destination image research, such as
Multidimensional Scaling (Goodrich 1978b; Gartner 1989; MacKay and Fesenmaier
2000), Repertory Grid Analysis (Embacher and Buttle 1989), and Correspondence
Analysis (Calantone, et al. 1989).

However, there is a lack of empirical studies testing the inﬂuence of such
methodological factors on the ﬁndings of destination image research. Gartner (1986)
attempted to investigate the impact of timing of the survey on destination image. He
recognizes possible temporal ﬂuctuations in the destination image construct due to the
seasonality aspect of tourism, which, he cautions, could result in a bias in the product’s

positioning. He also realizes that this possibility is commonly ignored by researchers. He

47

surveyed a stratiﬁed random sample of US households about their impressions of
activities and attractions in Colorado, Montana, Utah, and Wyoming; the same research
was conducted twice within a four-month interval. Of the 52 comparative analyses
between two time periods and four states, only two resulted in signiﬁcance. In the
second-wave survey, Utah was rated more impressive than others in the ‘nightlife’
attribute and Wyoming was rated more impressive than others on the ‘boating’ activity.
He speculated that a strong brand image might be rendering temporal ﬂuctuations
unimportant while cautioning about the short time interval of the research design and

recommending further investigation of this subject matter.

Computer Assisted Telephone Interviewing (CATI)

CATI has been used as a data collection mode since the early 19705 (Fink 1983).
In this system, telephone numbers are stored in a computer and the computer is
programmed to select and dial a number either randomly or in an order (Babbie 1998).
The numbers are generated randomly to avoid the sampling bias due to unlisted numbers
(Babbie I998). The questions are usually programmed in such a way to account for the
skip patterns in a questionnaire; that is, the appropriate questions are selected by the
computer rather than the interviewer (Babbie 1998). Once the interview is completed, the
data are stored in a data ﬁle, ready for cleaning and analysis; in other words, no data
entry is involved (Babbie 1998).

Some CATI soﬁware programs have systems that enable much of the work to be
done electronically, such as questionnaire writing, call management, quota controlling,

call disposition monitoring, interviewer assistance, reporting. and analysis (Churchill

48

1999). Thus, CATI has advantages over traditional survey techniques, such as low cost,
wide distribution, higher response rates, less-interviewer-related bias, fast and easier
questionnaire completion, and easier follow up (Churchill 1999). However, it also has
disadvantages, such as inability to use visual aids, difficulty in developing rapport with
the respondents, difﬁculty in handling long questionnaires, and difficulty in respondent
screening (Churchill 1999).

In spite of random digit dialing (RDD), sampling bias is purported to be possible
in CATI surveys. McAuliffe, Geller, LaBrie, Paletz, and Foumier (1998) suggest that
telephone surveys reveal more precise estimates (due to random digit dialing) than face-
to-face surveys based on the same number of interviews. However, they also purport that
telephone surveys reveal less accurate data than face-to-face interviewing due to the
systematic exclusion of people who live in households without telephones, and lower
response rates. Herzog, Rodgers, and Kulka (1983) found that telephone surveys have a
tendency to underrepresent older adults who are less likely to participate in a telephone
survey even though they are more likely to have a telephone than younger adults.

Aquilino and Lo Sciuto (1990) revealed that the telephone survey with RDD had
high response rates but also exhibit a bias towards higher socioeconomic status with
higher income, education, and likelihood to be married and employed especially among
the black group; this effect of the survey mode was not this signiﬁcant for whites.
Sniderman and Grob (1996) also found that mode effects were stronger for minorities
than whites, and the amount of mode effect was larger for blacks than for whites.
Aquilino conducted a ﬁeld experiment to measure “the impact of interview mode on

respondents’ willingness to reveal sensitive or socially undesirable information” (1994, p.

49

211). The researcher compared three interview modes: (1) self-administered
questionnaires, (2) interviewer-administered in-person interviews, and (3) telephone
interviews. The researcher found that since telephone interviewing provided the least
amount of anonymity, it decreased respondents’ willingness to reveal sensitive

information.

Factor Analysis

Factor analysis facilitates the summarizing or reducing of many variables into
fewer manageable factors by identifying the underlying structure among variables and
grouping correlated variables together (Babbie 1998; Churchill 1990; Fraenkel & Wallen
1996; Hair, Anderson, Tatham, & Black 1998). In data summarization, factor analysis
identiﬁes relationships among a group of variables or respondents; while in data
reduction, it identiﬁes a few underlying factors to be substituted for a larger number of
original variables and used as input variables in further multivariate analyses (Hair et al.
1998). It is different from regression analysis since it tries to provide the maximum
explanation for the variables rather than predict a criterion variable (Hair et al. 1998).
Thus, a high level of correlation between variables, greater than 0.30, is desired in factor
analysis, while high correlation can cause unstable matrix inversion in regression analysis
(Hair et al. 1998).

Factor analysis can be performed in an exploratory or a conﬁrmatory approach. In
the exploratory approach, the researcher explores the data without any preconceived

theories about the underlying structure of the data; while in the conﬁrmatory approach,

50

the researcher has preconceived hypotheses about the number and type of underlying
factors in the data and tests for these hypotheses (Hair et al. 1998).

Factor analysis applies two types of analysis, common and component. Common
factor analysis identiﬁes the common features shared by the entire variable set while the
component analysis (principal component analysis) summarizes the maximum amount of
original information in the entire set of variables (measured as variance) through a
minimum number of factors possible (Hair et al. 1998). In principal component analysis,
the ﬁrst factor extracted explains the largest amount of variance in the original variables
and following factors explain smaller and smaller amounts of variance (Aﬁﬁ 1984, Hair
et al. 1998). Thus, the initial (unrotated) factor solution reveals the best linear
combination of the variables; the ﬁrst factor is the best summary of these relationships
since it has the highest variance of the original variables (Hair et al. 1998). The second
factor has to be orthogonal to the ﬁrst factor, so it is extracted from the variance leftover
after the ﬁrst factor is extracted, and so on (Hair et al. 1998).

The initial unrotated solution achieves data reduction but may not enable adequate
interpretation since the factor loadings it provides (the correlations between individual
variables to their underlying factor) may not have a meaningful pattern (Hair et al. 1998).
Therefore, the factors must be rotated (usually orthogonally, at 90 degrees) to retain more
meaningful and simpler factor solutions; during this rotation, the variances are
redistributed between the factors, thus reducing ambiguities of the initial factor solution
(Hair et al. 1998). The Varimax rotation is commonly used since it restricts the factors to
being uncorrelated by forcing the factor loadings to approach the limits of 0 and :1 as

closely as possible (Hair et al. 1998). Variables with high loadings (i.e. close to i 1)

51

indicate a good correlation, positive or negative, between an individual variable and the
corresponding factor (Hair et al. 1998).

As a rule of thumb, variable loadings of i 0.50 are considered as practically
signiﬁcant to be kept in a factor although it can be smaller depending on the sample size
(Hair et al. 1998). The factors are interpreted by looking at the variables with the highest
loadings since they have the highest correlation with the factor (Hair et al. 1998).

In deciding the number of factors, researchers most commonly use the ‘latent root
criterion’ in which the factors with latent roots (Eigenvalues greater than 1) are kept since
they explain the variance of at least one variable included in that factor (Aaker 1981;
Bernstein 1988; Hair et al. 1998). The internal stability of each factor is determined by
using the Cronbach’s alpha; usually a Cronbach’s alpha of 0.70 is considered as
substantially stable (Hair et al. 1998).

Upon determining the number of factors to be retained, factor scores (a composite
measure of all variables included in the factor) are calculated for each factor and for each
case (Hair et al. 1998). The function of factor scoring is to mark the relative position of
each individual on each factor (Rossi, Wright, & Anderson 1983). This is similar to score
standardization in which the original scores are transformed into a new set of scores with
a mean of 0 and variance of 1. (See Fakeye and Crompton (1991) and Court and Lupton
(1997) for a discussion of this method.) Eventually, these factor scores are used as new
input variables in further analyses (Hair et al. 1998).

Hair et al. (1998) recognize that factor stability depends on the sample size and
the number of cases for each variable included in the factor; they recommend the use of

the largest samples possible for developing parsimonious factor models. They also

52

recommend a check on the robustness of the factor solution by splitting the sample
randomly into two subsets and applying factor analysis on each (Hair et al. 1998).

Factor analysis is criticized by some researchers for it can produce meaningless
factors by putting disparate variables together, and it does not involve statistical tests,

thus rendering the hypotheses useless (Babbie 1998).

Regression Analysis

Regression analysis is used to explain or predict human-related phenomena and
their likely outcomes by identifying the existing relationships among certain variables
(Fraenkel & Wallen 1996; Hair et al. 1998; Hamilton 1992). If a substantial relationship
exists, it is then possible to predict the value of a dependent variable (criterion variable)
when the value of the independent variable (predictor variable) is known, and vice versa
(Babbie 1998; Fraenkel & Wallen 1996; Hair et al. 1998; Hamilton 1992; Pindyck &
Rubinfeld I981). The basic mathematical description of this relationship is:

Y = ﬂX + c
where, Y is a random variable (stochastic), ﬂ is the estimated coefﬁcient, X is ﬁxed
(nonstochastic) variable, and e is a random error term (Hair et al. 1998; Hamilton 1992;
Pindyck & Rubinfeld 1981). Error terms (residuals or prediction error) are the differences
between the predicted values and actual values of the dependent variable, which may
arise from several factors in the research process; error terms are assumed to be
uncorrelated, normally distributed with a zero expected value and constant variance
(homoscedasticity) for all observations (Hair et al. 1998; Hamilton 1992; Pindyck &

Rubinfeld 1981).

53

Ordinary Least Squares (OLS) is a widely used regression method of ﬁtting linear
models to data (Hamilton 1992; Pindyck & Rubinfeld 1981). It includes “a system of
techniques for describing sample data and extending conclusions to a larger population”
(Hamilton 1992, p. 30). The least squares criterion in regression creates a ‘line of best ﬁt’
by minimizing the sum of squared distances between the actual data point and the
estimated data point on the straight line hypothetically drawn among the actual data
points (Hamilton 1992; Pindyck & Rubinfeld 1981). This criterion allows estimation of
“regression parameters, which are best linear unbiased and (under some conditions)
maximum likelihood”; however, it is sensitive to outliers in the data (Pindyck &
Rubinfeld 1981, p. 4).

In social research, a criterion variable is usually inﬂuenced by several predictor
variables (Babbie 1998). Inclusion of more variables in the prediction must cause smaller
errors of prediction since more explanatory information is fed into the prediction
(Fraenkel & Wallen 1996). When more than one predictor variable is included in the
estimation, then it becomes multiple regression, in which the criterion variable is being
predicted by using “the best combination of two or more predictor variables” (Fraenkel &
Wallen 1996, p. 313).

Regression is different from correlation for several reasons. First, regression
involves an assumption of causality between variables while correlation primarily
measures the degree of linear association between two variables (Pindyck & Rubinfeld
1981). In regression, the dependent variable is assumed to be random (stochastic) with a
probability distribution while independent variables are assumed to have ﬁxed values

(nonstochastic) if measured repeatedly (Pindyck & Rubinfeld 1981). There is no such

54

distinction in correlation, and both variables are assumed to be random (Pindyck &
Rubinfeld 1981).

Regression analysis involves some assumptions, or more precisely requirements,
namely, simple random sampling, no nonsampling errors, and continuous interval data.
Since these requirements are rarely satisﬁed in social research, interpretation of the
results of regression analysis should be done cautiously (Babbie 1998). More speciﬁcally,
excluded variables, nonlinear relationships between dependent and independent variables,
error terms with a nonconstant variance, correlated and nonnorrnal error terms, and
outliers in the data are some regression assumptions and factors to which regression
analysis is sensitive (Hair et al. 1998; Hamilton 1992; Pindyck & Rubinfeld 1981).
Violation of these assumptions and/or overlooking the factors to which regression
analysis is sensitive can render the results of a regression biased or invalid.

Normality is a fundamental assumption in multivariate analysis since it is required
to use t and f statistics; violation of this assumption would render the results of statistical
tests invalid (Hair et al. 1998). Therefore, visual checks, such as histograms and normal
probability plots along with statistical tests, such as the Shapiro-Wilks test and
Kolmogorov-Smimov test are used to make sure this assumption is not violated in the
data (Hair et al. 1998). Various data transformation methods, such as square root and
logarithm are employed to remedy the problem of nonnorrnality in the data; these
techniques also can remedy the problem of heteroscedasticity, in which the variance of
the dependent variable concentrates in only a few independent variables (Hair et al.

1998).

55

The Central Limit Theorem is postulated to protect against the failure of
normality in sampling distributions of large samples (Hamilton 1992; Pindyck &

Rubinfeld 1981). According to this theorem,

a. the sampling distribution of the mean becomes approximately normal, regardless of
the shape of the variable’s frequency distribution.

b. the sampling distribution will be centered around the variable’s population mean u.

c. the standard deviation of the sampling distribution, called its standard error,
approaches oNn. the variable’s population standard deviation divided by the square
root of the sample size (Hamilton 1992, p. 27).

In other words, as sample size gets larger, the sample more closely resembles the
study population, thus rendering the results of the statistical tests accurate or valid even if
the normality assumption is violated (Hamilton 1992; Pindyck & Rubinfeld 1981).
However, very large samples (1,000 or more observations) may result in statistical
signiﬁcance for any relationship test, thus researchers should use ‘criterion of practical
signiﬁcance’ in this case (Hair et al. 1998, p. 165).

The use of continuous variables is a requirement for regression analysis, but, in
many cases, categorical variables, such as respondents’ gender are relevant to the
analysis; in these cases, the categorical variables are incorporated into analysis using
replacement variables called dummy variables with the dichotomy of O and 1 (Hair et al.
1998). For a variable with n categories, n-l dummy variables are created and

incorporated in the regression model to avoid the singularity problem (Hair et al. 1998).

Hypothesis Test

Coefﬁcients (b) are the estimates derived from the sample for the population

parameters concerning the independent variables (Pindyck & Rubinfeld 1981). A positive

56

estimate indicates a positive relationship between the dependent and independent
variables (Hamilton 1992).

Since independent variables are measured in different units, regression
coefficients need to be standardized to make it possible to compare the degree of impact
of independent variables on the dependent variable. Standardized coefficients (beta
coefficients or B) indicate the change in the dependent variable for each unit change in
one independent variable when all other independent variables are held constant (Hair et
al. 1998; Hamilton 1992). Standardized coefficients (beta coefficients) can be used to
make comparisons among independent variables in terms of the magnitude of their effect
on the dependent variable (Hair et al. 1998; Hamilton 1992).

Since the predicted values of the dependent variable are different from the real
values, researchers also calculate the standard error of estimate, an indicator of prediction
error, which provides the likelihood of predicted value’s being inaccurate, the bigger the
standard error of estimate the more inaccurate the prediction (Fraenkel & Wallen 1996;
Hair et al. 1998).

The t-statistic is calculated by dividing a coefficient by its associated standard
error and is used to test if a coefficient is signiﬁcantly different from zero for the whole
population, taking the variation in the sample into consideration; in general, a t-test value
bigger than two indicates statistical signiﬁcance; and the greater the value, the more
statistically signiﬁcant the related estimated coefficient (Hamilton 1992; Pindyck &
Rubinfeld 1981).

The p-value (or) provides the signiﬁcance level for the two-way test of a

coefficient being different from zero (Pindyck & Rubinfeld 1981). Speciﬁcally, it “equals

57

the estimated probability of obtaining these sample results, or results more favorable to
H1, if the sample were drawn randomly from a population where H0 is true” (Hamilton
1992,p.44)

Model ﬁt statistics, namely, R2, the f statistic, and a (p-value), are calculated as
measures of the model ﬁt to the data. These values are used to test the null hypothesis
that there is no linear relationship between the dependent variable and the independent
variables in the population (Hair et al. 1998; Hamilton 1992; Pindyck & Rubinfeld 1981).
In other words, these values are used to test multiple equivalent null hypotheses that all of
the population partial regression coefficients are 0, and the population value for multiple
R2 is 0.

The coefficient of determination (r2 for simple regression and R2 for multiple
regression) is a measure of the goodness of the ﬁt for the regression model; it indicates
the percentage of variation in the dependent variable explained by independent variables
(explained variance divided by total variance), in other words, the degree of predictability
of the dependent variable on the basis of the independent variables (Babbie 1998;
Fraenkel & Wallen 1996; Hair et al. 1998; Hamilton 1992; Pindyck & Rubinfeld 1981).
The R2 value ranges between the values of 0 and l; the closer R2 to the value of 1, the
better the model ﬁts the data (Babbie 1998; Fraenkel & Wallen 1996; Hair et al. 1998;
Pindyck & Rubinfeld 1981).

There are some arguments about the use of R2 as a criterion of goodness of ﬁt.
First, selected variables may not be good explanatory variables in the regression equation
(Pindyck & Rubinfeld 1981). There might be other more important factors that explain

the dependent variable. Second, it is argued in literature that a low R2 is common in

58

models of consumer behavior (Court and Lupton, 1997; Bass 1975). Achen (1982) and
Court and Lupton (1997) argue that in regression models, the R2 is not a good indicator of
causally strong relationships since the sample is the deﬁning factor of the variances, not
the underlying relationship between the variables (Court and Lupton 1997; Achen 1982).

Pindyck and Rubinfeld assert that

(i)n time series studies...one often obtains high values of R2 simply because any
variable growing over time is likely to do a good job of explaining the variation of any
other variable growing over time. In cross-section studies. on the other hand, a lower R2
may occur even if the model is a satisfactory one, because of the large variation across
individual units of observation which is inherently present in the data (1981, p. 64).

The f-statistic is used to test hypotheses regarding groups of parameters
(Hamilton 1992; Pindyck & Rubinfeld 1981). The f—statistic is the ratio of two mean
squares, the regression mean square and the residual mean square (Hair et al. 1998). Both
t and f procedures assume errors to be normally distributed (Hair et al. 1998; Hamilton
1992). The numerical distribution of the f-statistic is known, and if the calculated f is
smaller than the critical I value, the null hypothesis is accepted (Hair et al. 1998; Pindyck
& Rubinfeld 1981). The a (p-value) provides the signiﬁcance level for the test of multiple
regression equation, in other words, the signiﬁcance level for the one-way test for the null
hypothesis that the R2 is equal to zero (Hair et al. 1998; Pindyck & Rubinfeld 1981).

For validation of the regression model, the split sample technique can be used
since it is usually costly to draw a new sample from the population (Hair et al. 1998). In
this technique, the original sample is divided into two subsamples, one for estimation of

the regression model and one for testing the regression equation (Hair et al. 1998).

59

Multicollinearity and Singularity

Multicollinearity and singularity are two conditions that can occur in correlation
matrices; both of these conditions can render parts of the correlation matrices unstable,
which can lead to biased or invalid results (Hair et al. 1998; Pindyck & Rubinfeld 1981).
Multicollinearity means a high correlation between two or more variables in a matrix
(Hair et al. 1998; Pindyck & Rubinfeld 1981). Singularity occurs when the value of a
variable is a linear combination of others (Hair et al. 1998). Despite the differences in the
nature of multicollinearity and singularity conditions, they cause similar problems in
multivariate analyses, by causing an unstable matrix inversion or termination of the
analysis altogether (Hair et al. 1998).

Detection of these conditions is done through the examination of tolerance
measures and correlation matrices (Hair et al. 1998). If the correlation matrix reveals high
correlations (usually 0.90 and above) between three or more independent variables, it is a
sign of multicollinearity (Hair et al. 1998). Tolerance is a statistic used to determine how
much the independent variables are in linear relation with one another, that is,
multicollinear; in other words, tolerance values are a measure of independence among
independent variables (Hair et al. 1998). The tolerance of an independent variable is the
proportion of its variability not explained through its linear relationships with one or
more of the other independent variables included in the model (Hair et al. 1998). Similar
to R2, its values range between 0 and 1. If the tolerance value of a variable is small, it
means it is linearly related to one or more of the other independent variables, thus
rendering the data to be multicollinear (Hair et al. 1998). As a rule of thumb, the cutoff

for tolerance values is 0.10; variables with tolerance values of less than 0.19 would have

60

a substantial correlation, a correlation equal to or above 0.90 (Hair et al. 1998). In case of
multicollinearity, the researcher can either delete one or more of the correlated
independent variables, or proceed with the model for prediction only, without interpreting

the coefficients (Hair et al. 1998).

61

CHAPTER III

METHODS

The goal of this study was to identify the determinants of destination image and
develop a parsimonious model of destination image and its determinants. Due to the
costly nature of image studies in real life tourism settings, a secondary data set was
utilized for the purposes of this study. This chapter delineates the methods used in
collecting these data and the procedures followed in conducting the investigation of
determinants to destination image formation. The chapter is divided into seven sections:
(1) study design, (2) study destination and population, (3) the study instrument including
operationalization of concepts, (4) data collection mode and procedures, (5) response
rate, (6) data preparation, and (7) data analyses. The ﬁrst ﬁve sections describe the data
source, The Michigan Regional Travel Market Survey (MRTMS). Sections 6 and 7 detail
the variables included in the image model and analysis procedures, statistics, and

computer programs used to test the proposed hypotheses.

Study Design
This is an associational study in which secondary data are used to investigate the
existing relationships between destination image and several possible determinant
variables. More speciﬁcally, it is a causal-comparative study (also referred to as ex post
facto); the causes of differences that already exist between or among groups of

individuals are to be determined in this study (Fraenkel & Wallen 1996). There is no

62

attempt to manipulate any variables as in experimental studies; the differences already
exist (measured) and the variables cannot be manipulated (such as gender and age).

For the relationship between destination image and respondents’
sociodemographic variables, causal direction is clear per se. For example, the image
differences between different gender groups must be caused by the gender membership
since destination image cannot determine the gender of a person. However, the causal
direction between visitation and destination image is not so obvious. Visitation might
cause better images of a destination or better images might result in visitation to a
destination. Determining this direction would require an experimental design in which
subjects’ pre and post visitation images are observed and compared.

Several variables, such as visitation itself and other visitation-related variables
could have been manipulated to deﬁne the causal direction; however, since data from a
cross-sectional survey of random samples across several years are used, such
manipulation is not possible. Because of the lack of causal direction between destination
image and several determinant variables involved in the analyses of this study, this study
can also be considered as correlational study, which is also associational in nature
(Fraenkel & Wallen 1996).

The data were collected through a quantitative survey design. Speciﬁcally, it is a
longitudinal household survey conducted since 1996 by the Travel, Tourism &
Recreation Resource Center at Michigan State University. The main objective of this
survey is to provide Michigan tourism providers and decision makers with timely and
accurate data on the status of tourism in Michigan. Initially, the study was ﬁrnded through

Travel Michigan, the state’s major public tourism agency conducting tourism promotion

63

in and for Michigan, the Agricultural Experiment Station at Michigan State University
(MSU), and the MSU Office of the Provost. During later years, various sponsors
contributed to ﬁrnding the survey, which was modiﬁed to address individual sponsors’

data requirements.

Study Destination and Population

As stated before, the main goal of the Michigan Regional Travel Market Survey
(MRTMS) has been to provide market intelligence for the tourism industry in Michigan;
therefore, its questions pertain to various aspects of travel and tourism in Michigan. The
study population is the resident households in Illinois, Indiana, Michigan, Ohio,
Wisconsin, and Ontario, Canada (see Figure 6). This region is considered to be the
primary market generating tourism in Michigan. The US. Census Bureau reported that,
in 1995, the ﬁve states in this region generated almost 90 % of travelers in Michigan
(US. Department of Transportation, Bureau of Transportation Statistics, 1995).
Minnesota was included in the earlier years of MRTMS, but excluded ﬁ'om the sampling
frame beginning in 1999; to achieve comparability across all years, Minnesota
respondents were excluded ﬁ'om the dataset for the purposes of this study.

Telephone numbers were purchased from Survey Sampling Inc., which generated
a random pool of phone numbers by employing random digit dialing (RDD). Some of
these numbers were found to be disconnected, fax numbers or business numbers, and
were removed from the sampling frame. Respondents were screened for age since only

resident adults, 18 or older, were to be interviewed. To achieve a random sample, the

64

interviewers asked to speak to the “adult over 17 years old who will have the next

birthday.”

 

Figure 6. The study region (sampling frame).

Study Instrument
The questionnaire has been rather lengthy, and its length and content have varied
depending on the sponsors and the nature of their interests. It was mainly designed to
proﬁle the respondents’ most recent pleasure trip within the past 12 months. If this trip

was taken to Michigan, this proﬁle also represented the proﬁle of the most recent

65

pleasure trip to Michigan. If not, the respondents were asked if they took any pleasure
trips to Michigan within the past 12 months, and their most recent pleasure trip to
Michigan was proﬁled using the same questions. Thus, if the respondents’ most recent
pleasure trip was to another destination and they also took a pleasure trip to Michigan
within the past 12 months, these respondents were asked the same trip proﬁling questions
twice, ﬁrst in the “Most Recent Pleasure Trip Block” and then in the “General Michigan
Pleasure Trip Block”. See Figure 7 and 8 for the ﬂowcharts of the instruments and
Appendix A and B for the actual instruments.

Only pleasure trips were proﬁled, which were deﬁned as “any overnight or day
trip to a place at least 50 miles from home that was made for enjoyment, including
vacations, weekend getaways, shopping trips, trips to a second home, and trips to visit
ﬁiends or relatives.”

The initial version of the survey included 164 questions focusing on promotion
awareness of Michigan in comparison with neighboring and competitor states,
advertising inﬂuence, pleasure trip proﬁle and/or Michigan pleasure trip proﬁle,
Michigan image, travel intentions and sociodemographic questions (see Figure 7 and
Appendix A).

Later versions comprised 140 individual questions on average, with a core set,
including key variables pertaining to pleasure trips, Michigan pleasure trips and
demographics, as well as a periodically rotated set of questions. With this rotation, the
purpose was to capture trends, for example, trip intentions during major holidays, such as

Thanksgiving, and Labor Day (see Figure 8 and Appendix B).

66

Have you traveled in the past 12 months? (1)

No /\ Yes

Introductory Block (2-8)
Promotional Awareness and Response Block (9-16)
Michigan Image Block (17-35)

Have you taken a pleasure trip to any destination in the past 12 months? (36)

No Yes

How many pleasure trips? (37)

Attractions Block (3 8-5 1 )

Most Recent Pleasure Trip Proﬁle Block (52-85)

What was the main destination of this trip? (86)
Non-Ml

/\Ml
t

Have vou taken a pleasure trip to MI in the past 12 months? (87)

4/\

 

 

No Yes
v l Ml Pleasure Trip Block (90-124)
Have you ever taken a pleasure trip to MI? (88) ¢
/\ inﬂuence Block (125-135)
V
No Yes +
* Ml Pleasure Trip History Block (136-139)

When was the last time? (89)

y

Ml Travel Expectations Block (140-142)

i

Ml Trip Volume Block (143-148)

 

 

$ Personal/ Household Characteristics Block (149-163)

Quit (164)

Figure 7. The ﬂowchart of the 1996-1998 questionnaires.

67

Introductory Block-Vehicle ownership, attitudes about gasoline price increase, air travel (l-IO)
Michigan Image Block (1 1-13)
Pleasure Trip Preferences Block (14-15)

In the past 12 months, have you taken any day or overnight pleasure trips to any destination? (16)

No Yes
How many pleasure trips? (17)

Most Recent Pleasure Trip Profile Block (18-52)

What was the main destination of this trip? (53)
Non-Ml

‘/\MI
J,

Have you taken a pleasure trip to MI in the past 12 months? (54)

 

 

No Yis
l Ml Pleasure Trip Block (56-91)
——i> Have you ever taken a pleasure trip to MI? (55) j

Was this the ﬁrst pleasure trip? (92)

No Yes

l

How manv pleasure trips to M1 within the past 12 months? (93)

 

y

> MI Travel Expectations Block (94-101)

 

 

Internet Block (102-105)

Personal/Household Characteristics/Volunteerism Block ( 106-135)

l

Quit (136)
Figure 8. The ﬂowchart of the 2001-2002 questionnaires.

68

Some items, including image measures were previously generated and tested for
reliability and validity by Certec Inc., a private research and consulting company that was
a research partner for the ﬁrst three years that The Michigan Regional Travel Market
Survey was conducted.

Although the questionnaire was a lengthy one, due to the skip pattern used
throughout the questionnaire, respondents did not have to answer all the questions most
of the time (see Figure 7 and 8). The length of interview varied depending on whether or
not the interviewee took a pleasure trip to any destination during the past 12 months,
whether or not this trip was to Michigan, and whether or not they took any pleasure trip
to Michigan within the past 12 months if their most recent trip was not to Michigan.

Thus, if respondents did not take any pleasure trip to any destination within the
past 12 months, the interview was completed in few minutes. But if their most recent
pleasure trip was to another destination and they also took a pleasure trip to Michigan
within the past 12 months, the interview took about twenty minutes or so depending on
the speed of the conversation between the interviewer and the respondent. The average

time required to complete the questionnaire is about 12 minutes.

Data Collection Mode and Procedures
A CATI-Lab (Computer Assisted Telephone Interviewing Laboratory) was
utilized for this survey. The lab had six stations that were connected to the main server at
the Travel, Tourism & Recreation Resource Center. The questionnaire was electronically

programmed for each interviewing station with the StatPac software language (StatPac

69

1995). Every time an interview was completed, the data were transmitted to the main
server.

Interviews were conducted on weekday evenings and weekend aﬁemoons,
between 6 and 10 pm on weekdays, 12 and 4 pm. on Saturdays, and 2-6 pm. on
Sundays. No interviews were conducted on Fridays due to peoples’ tendency to go out on
this day of the week.

Phone numbers called were coded upon the ﬁrst trial using the coding scheme
presented in Table 4. If the status of a phone number changed during subsequent calls,
codes were modiﬁed accordingly. If an interview was completed, at the end of the
interview, the computer coded the phone number as (+) automatically and did not allow
any subsequent callbacks. Those phone numbers coded as N, U, F, B, T, H, and Q also
were not called again. The answering machine, no response, busy and those who asked to
be called back were called up to three times (this was increased to ﬁve times starting in
2000). After 3/5 trials, the old numbers were discarded, and a brand new set of numbers
was uploaded to the interview stations.

Table 4. The interview code scheme.

 

 

Codes Terminology Explanations

+ Completed Interview is completed.

C Call Back The respondent asked to be called back.

M Answering Machine Answering machine picked up.

R No Answer There is no answer.

2 Busy The line is busy.

N Not In Service The line is disconnected.

U Business Business number.

F Fax Machine Fax machine number.

B Language Barrier The respondent cannot speak or understand English.

T Terminated The respondent terminated the interviewer by letting the interviewer
know beforehand.

H Hang Up The respondent hung up without letting the interviewer know
beforehand.

Q Qualiﬁed But Refused The respondent was eligible but refused to participate in the study.

 

7O

Graduate students from the Department of Park, Recreation & Tourism Resources
at Michigan State University were hired for supervision of the CATI-Lab; and at least
one supervisor was always present when the interviews were conducted. Interviewers
were usually undergraduate female students at Michigan State University.

Interviewers received both oral and practical training for up to four hours before
conducting a real interview. They received detailed information about the sponsors and
purpose of the study, deﬁnitions and concepts used in the survey, the code of conduct in
the lab, and were oriented to effective interviewing techniques. They practiced
interviewing with other trainees and subsequently observed experienced interviewers
conduct real interviews. Their performance was monitored both in terms of work
productivity and adherence to rules and procedures. They were given ﬁrst oral then
written warnings in case of unsatisfactory performance.

Interviewer turnover was high, especially towards the end of the school terms
because of the interviewers tendency to go home during school breaks. This caused some
reduction in number of interviews completed. However, data quality standards were
maintained since the supervisors started hiring and training new interviewers before the
end of each school semester and never allowed interviewers to conduct actual interviews

before they were fully trained.

Response Rate
The average response rate was 44 % when the partially completed interviews
were included. The formula of this response rate was: Response Rate = (Fully Completed

Interviews + Partially Completed Interviews)/[Sample Size- (Not in Service + Businesses

71

+ Fax Machines)] x 100. In partially completed interviews, respondents did not answer
all the questions due to either one of the following reasons: (1) they hung up without
letting the interviewer know in advance, (2) they terminated upon telling the interviewer,
(3) the phone line was disconnected due to technical problems.

The response rate was 35% when only fully completed interviews were included
in the calculation. All refusals (about 29 % of the eligible respondents), answering
machine, no answer, and busy numbers were considered to be nonresponses in
calculating this response rate. Thus, the formula of this response rate was: Response Rate
= (Fully Completed Interviews)/[Sample Size- (Not in Service + Businesses + Fax
Machines)] x 100. This is the most conservative method of calculating the survey
response rate (Dillman, 1978; Frey, 1989; Lavrakas, 1993).

During the ﬁrst year of the study, the presence of nonresponse bias was
investigated. Some of the eligible numbers (C, M, R, Z), which were not reached within
the ﬁrst three attempts, were selected and called up to three additional times. Q (qualiﬁed
but refused), T (terminated), and H (hung up) categories were not included due to the
ethical reasons. Those respondents reached on these additional call backs (six-attempt
group) were assumed to have similar characteristics to overall non-respondents in the
study (Court & Lupton 1997; Jain, Pinson, & Ratchford 1982). These respondents
(N=173) were compared to a randomly selected subsample of the three-attempt
respondents (N=173) to appraise nonrespondents’ characteristics (Miller & Smith 1983).

By using t-tests and Chi-square tests, the two groups were compared on 84 key
variables, including demographic characteristics, attitude items, and trip proﬁle items,

only three of which were found statistically signiﬁcant at the .05 level. These included:

72

(1) on average, the six-attempt group rated the desirability of Ontario as a pleasure trip
destination on a ten-point scale more highly than did the three-attempt group (6.7 vs. 5.2),
(2) on average, the six-attempt group lived in households containing fewer persons than
did the three-attempt group (2.6 vs. 3.1), and (3) the six-attempt group visited a state or
national park on their most recent pleasure trip in Michigan at a higher rate than the
three-attempt group did (43% vs. 28%).

Given that signiﬁcant differences were found for only three of 84 items examined,
non-response bias would not appear to be present in this data set. However, only ‘non-
contact’ non-response bias was evaluated in this case leaving open the question of non-
response bias among those who were contacted but refused to participate in the survey.
As noted, it was not possible to contact those who refused since to do so could be

considered harassment and unethical.

Data Preparation

The study’s main focus was on the most recent pleasure trips taken to Michigan
by the residents of Illinois, Indiana, Michigan, Minnesota, Ohio, Wisconsin, and Ontario,
Canada in the past 12 months. The study’s purpose was to proﬁle respondents’ most
recent pleasure trips taken within the past 12 months; if this trip was not to Michigan,
then the purpose was to proﬁle their most recent pleasure trips to Michigan within the
past 12 months. This has been uniform in the survey across all study years. Also,
Michigan image measurement items are located at the beginning of the questionnaire so
that Michigan image held by both those who visited Michigan within past 12 months and

those who did not can be measured.

73

However, there is a difference between the periods of 1996-1998 and 2001-2002
in terms of screening respondents (see Figure 7 and 8). In the 1996-1998 questionnaire,
the ﬁrst question asked if the respondents traveled in the past 12 months. If they said yes,
then they were asked the Michigan image questions; if they said no, they were routed to
the demographic questions located at the end of the questionnaire. Thus, for respondents
to be asked Michigan image questions, they had to have taken at least one pleasure trip to
any destination within the past 12 months. On the other hand, in the 2001-2002 period,
respondents were not screened for a pleasure trip within the past 12 months before being
asked the Michigan image questions. Therefore, image respondents in this period include
all respondents, including recent travelers and non-travelers. Since the vast majority (80.8
%) of respondents in the initial years of the study reported taking a pleasure trip during
the past 12 months, this screening difference is unlikely to be a source of bias in the
analyses conducted for this study.

Since the questionnaire was a lengthy one, some respondents hung up or
terminated the call before they answered all the questions on the survey. Others could not
or would not respond to all items in the questionnaire. Therefore, although the total
number of cases in the dataset is 21,111, the case number for each variable is different in
the data set. For this reason, some respondents were not included in some analyses.
Different analyses were performed on different numbers of cases as will be described and
explained later in this chapter. The case numbers for different analyses are provided in
Chapter 4.

First, some respondents were deleted during the factor analysis of image data

since some respondents terminated the call or hung up before they reached the Michigan

74

image measures. In addition, some respondents failed to provide a legitimate rating for
some image measures. Although “don’t know” and “don’t want to answer” response
categories were accepted by interviewers, in the factor analysis conducted, an actual
rating number was required across all image measures; otherwise, the case was dropped.
Those cases having responses, such as —99 (“don’t know”) and —55 (“don’t want to
answer”) were recoded as missing data and excluded in the exploratory factor analysis
(n=5,485). A total of 5,485 complete cases were available for conducting factor analysis
of image items. Since the use of as many cases as possible is recommended for stable
results in factor analysis (Hair et al. 1998), all of these cases were included in the factor
analysis of this study. Only about 25% of respondents actually rated the ﬁrll set of image
items. The proﬁle of these respondents (image respondents) is provided next to the
proﬁle of all respondents in the sample proﬁles section of Chapter 4.

A similar dropout in the number of qualiﬁed respondents was also observed in the
multiple regression analysis conducted since multiple regression analysis also requires
listwise deletion. Some respondents terminated the call or hung up before they answered
the questions or could not or would not answer speciﬁc questions included in the
regression models developed for this study, including past travel behavior and
demographic questions. The general population model was applied to both visitor and
non-visitor respondents to identify the determinants of image for the general public. The
recent visitor segment model was applied to recent visitor respondents to identify the
determinants of image for the recent visitors to the destination. The proﬁle of these
respondents (model respondents) is provided next to the proﬁle of all respondents and

image respondents in the sample proﬁles section of Chapter 4.

75

The ﬁrst model, the general population model, includes 15 variables. These
variables are respondents’ image of Michigan (image factors extracted by Exploratory
Factor Analysis and the Overall image extracted by averaging the ratings of the 15 image
items), age, gender, race, income, state of residence (Illinois, Michigan, Ohio, Wisconsin,
Ontario), overall travel experience within the past 12 months, prior visitation to Michigan
and the season of the survey (winter, spring and summer). See Appendix A and B for the
speciﬁc questions operationalizing these variables.

Respondents’ image of Michigan is the dependent variable in the multiple
regression analyses throughout this study. The image battery consists of 15 selected
image attributes as can be seen in Table 5. Respondents were asked to state the extent of
their agreement with these image items on a 10-point Likert-type scale where 1 means
“do not agree at all” and 10 means “agree completely”. Although the image variable
(dependent variable) comprises interval data, it is arbitrarily assumed to be continuous for
the regression analysis purposes.

Table 5. Michigan image measures.

Michigan...

Is close enough for a weekend getaway

Has many interesting museums

Is great for summer outdoor recreation activities
Is an exciting place to visit

Has a lot of high quality lodging

Offers much scenic appeal

Is great for winter outdoor recreation activities
Is a good place to meet friendly people

Is a place everyone should visit at least once in their lifetime
Is a safe place to visit

Offers exciting nightlife and entertainment

Is a great place for a family vacation

Is a popular destination with vacationers

Has many interesting historic sites

Offers an excellent vacation value for the money

 

 

76

The independent variables of the general population model are age, gender, race,

income, state of residence, overall travel experience within the past 12 months, prior

visitation to Michigan, and the season of the survey. Most of these independent variables

were measured at the nominal (categorical) level as can be seen in Table 6.

Table 6. The original measurement levels and codes of the variables included in the

general population model.

Variables

Respondents’ image of MI

Respondents’ age

Respondents’ race

Respondents’ gender

Respondents’ total annual household
income (the median income is deﬁned
as $31K for 1996-1997 and $42K for
2001-2002)

Respondents’ state of residence

Overall travel experience within the
past 12 months

Prior visitation to M1

The season of the survey

Level of
Measurement
Interval-l tolO

Interval

Nominal

Nominal

Ordinal

Nominal

Ratio

Nominal

Nominal

77

Response Categories

l=“do not agree at all”
lO=“agree completely.”
-99=Don’t know/No response
Open-ended question
Responses range between 18-99
Open-ended question grouped into 7
major groups.

l=American Indian/Native
2=Asian/Paciﬁc

3=Black

4=Hispanic

5=White

6=Mix/Multiracial

7=Others

1=Male

2=Female

-99=Cannot determine
l=Above the median
2=Below the median
~55=Refused

-99=DK/NR

l=Illinois

2=Indiana

3=Michigan

4 =Minnesota (Excluded)
5=Ohio

6=Wisoonsin

7=Ontario

- 9 9=DK/ NR

0-100

Measured in 2 variables, see
explanation below.
Measured in 3 variables. see
explanation below.
I=Spﬁng

2=Summer

3=Fall

4=Winter

To be included in the multiple regression analysis, nominal and ordinal variables
were recoded as multiple dummy variables, with the values of 0 and l. The modiﬁed
response category codes of these variables and resulting multiple dummy variables used
in the regression analysis are provided in Table 7.

Table 7. Modiﬁed codes for the nominal and ordinal variables and resulting multiple
dummy variables included in the general population model.

 

 

Original Nominal Variables New Dummy Variables New Codes
Respondents’ race White l=White
O=Other races
Respondents’ gender Female l=Female
0=Male
Respondents’ total annual Above the median l=Above the median
household income 0=Below the median
Respondents’ state of residence“ Illinois l=Illinois residents
O=Other states’ residents
Michigan l=Michigan residents
O=Other states’ residents
Ohio l=Ohio residents
O=Other states’ residents
. Wisconsin 1=Wisconsin residents
O=Other states’ residents
Ontario 1=Ontario residents
O=Other states’ residents
Prior visitation to MI Visitation I=Visited Michigan
0=Did not visit Michigan
The season of the surveyb Winter 1=Measured in the Winter
0=Measured in other seasons
Spring 1=Measured in the Spring
0=Measured in other seasons
Summer 1=Measured in the Summer

0=Measured in other seasons

a , . . . . . . .
. A dummy variable was not created for Indiana to avord the srngular matrix problem in the regressron analysrs.

 

b I A dummy variable was not created for the Fall season to avoid the singular matrix problem in the regression analysis.
No modiﬁcation was necessary for continuous variables.

As can be seen in Table 7, each category in each variable was recoded into
individual dummy variables with the values of 0 and 1. The number of dummy variables
for each categorical variable is one less than the number of categories it has. For
example, the season of the survey variable has four categories, but only three dummy
variables with the values of 0 and l were created; similarly, the gender variable has two

categories, but only one dummy variable was created. Creation of dummy variables, one

78

less than the number of categories a variable has, is performed to avoid the singular
matrix problem in the multiple regression analysis (Hair et al. 1998). Destination image,
age and general travel experience variables were not recoded since they are continuous
variables in the regression analysis.

For each of these dummy variables, each case in the dataset was recoded as 0 and
1, 1 meaning the case represents a condition and 0 meaning it doesn’t represent that
condition. Let’s take the spring dummy variable to explain this ﬁrrther. On the spring
dummy variable, those cases that were measured in the spring were coded as 1 (measured
in the Spring), and all others were recoded as 0 (not measured in the Spring). The same
procedure was applied to other season dummy variables as well as all other dummy
variables.

Some variation was lost for the respondents’ race variable since all race categories
were recoded into one dummy variable categorizing them as white and nonwhite. This
was performed since the overwhelming majority of the respondents are white as will be
discussed in Chapter 4.

Prior visitation to Michigan was established from responses to three different
questions. As can be seen in Figures 7 and 8, the trip destination question appears at the
end of the most recent pleasure travel block. If the destination of this trip was not
Michigan, the respondents were asked if they took a pleasure trip to Michigan within the
past 12 months. If they said no, then they were asked if they have ever taken a pleasure
trip to Michigan. Thus, there were three types of visitor respondents: (1) visitors whose
most recent trip within the past 12 months was to Michigan, (2) visitors who took a trip to

Michigan within the past 12 months but not during their most recent trip, and (3) visitors

79

who visited Michigan but not within the past 12 months. The original visitation variables

and their response categories are presented in Table 8.

Table 8. Original visitation variables and their response catﬂegories.
Response Categories

Visitation Variables

 

l-What was the main destination of this trip? Open ended, but recoded during the
City/Place interview as:
State/Province/Country l=Michigan destination

2=Non-Michigan destination
2-Was a place in Michigan the main destination of any of l=Yes

 

the pleasure trips you've taken in the past 12 months? 2=No
-99=DK/NR

3-Have you ever taken a pleasure trip to a place in l=Yes

Michigan? 2=No
-99=DK/NR

 

Note: Question numbers are not actual sequence numbers on the questionnaire.

These questions measuring prior visitation were also recoded into three individual

dummy variables, and ﬁnally, all three dummy variables were recoded into one single

dummy variable. The original variable values, recoded new dummy variables and

ultimate visitation dummy variable are presented in Table 9.

Table 9. Original visitation variables, new dummy variables and the

dummy variable.

ultimate visitation

 

 

 

 

Original Visitation Variables Response New Dummy Ultimate Visitation
Categories Variables Dummy Variable
l-What was the main destination of this l=Michigan I=Visited
trip? destination Michigan
City/Place =Non-Michigan 0=Did not visit
State/Province/Country destination Michigan I=Visited Michigan
2-Was a place in Michigan the main l=Yes I=Visited at some point in their
destination of any of the pleasure trips 2=No Michigan lives
you‘ve taken in the past 12 months? -99=DK/NR 0=Did not visit 0=Did not ever visit
' Michigan Michigan
3-Have you ever taken a pleasure trip to a l=Yes I=Visited
place in Michigan? 2=No Michigan
-99=DK/NR 0=Did not visit
Michigan

 

 

Note: Question numbers are not actual sequence numbers on the questionnaire.

As can be seen in Table 9, each visitation variable was recoded into three

individual dummy variables with the values of 0 and 1. Thus, those cases that had a

80

Michigan destination on the ﬁrst visitation variable were coded as l and all others were
coded as 0. Similarly, those cases that had yes on the second variable were coded as 1
and all others as 0. The same procedure was applied to the third visitation variable. Cases
responding as —99 and —55 were recoded as system missing since they could not be
included in the multiple regression analysis.

Finally, these three variables were recoded into one visitation dummy variable by
using the count ﬁrnction of SPSS 10.0. With this function, the values of l in each of these
three variables are counted, resulting in another dummy variable with the values of 0 and
1, 1 meaning visited Michigan at some point and 0 meaning did not visit Michigan at all.
Since each respondent was supposed to be asked only one of these 3 questions, the
resulting count should not be greater than 1. However, some cases ended up having the
values of 2 and even 3, which meant that they were erroneously asked questions they
were not supposed to be asked. This was assumed to be due to the interviewer error, and
these values were replaced with 1.

The overall travel experience within the past 12 months variable was created by
combining the responses to two questions: (I) Have you taken a pleasure trip to any
destination in the past 12 months? and (2) How many pleasure trips? Since there is a skip
pattern involved in this questionnaire, those respondents who said “no” to the ﬁrst
question were not asked the second question. To include these non-travelers into the
overall travel experience within the past 12 months variable, these respondents were
coded as 0 on the second question. Thus, both travelers and non-travelers were included
in the overall travel experience within the past 12 months variable and related analyses;

inclusion of non-travelers is relevant to the general population model but not to the recent

81

visitor segment model since the recent visitor segment model encompasses only recent
visitors of Michigan. Since the general population model pertains to travelers and non-
travelers, all respondents giving valid responses were included in this analysis (n=3,554).

On the other hand, the recent visitor segment model was applied to only
respondents who took a pleasure trip in Michigan within the past 12 months. In this
model, the dependent variable is the same, respondents’ image of Michigan and the
independent variables are the same as in the general population model with the exception
of the prior visitation to Michigan. The prior visitation variable is excluded in the recent
visitor segment model since only visitor respondents are included in the test of this
model. However, additional visitation-related variables are included in this model. These
additional variables are: the frequency of visitation to Michigan within the past 12
months, whether or not the last visit to Michigan is the most recent trip, the season of the
last visit to Michigan (winter, spring & summer), the length of stay in Michigan during
the last visit (# of nights spent), and the number of activities participated in during the
last visit to Michigan.

Some of these additional variables were recoded as dummy variables by using the
same procedure that was used for the general population model discussed previously. The
original visitation-related variables, their levels of measurement and response categories
are presented in Table 10, and applied modiﬁcations are presented in Table 11. The
continuous variables, which include: the frequency of visitation to Michigan within the
past 12 months, the length of stay during the last visit in Michigan (# of nights), and the

number of activities participated in during the last visit in Michigan were not modiﬁed.

82

Table 10. Original visitation-related variables, their levels of measurement and response

categories.

 

 

 

 

 

 

Original Variables Level of Original Response Categories
Measurement
The frequency of visitation to Michigan within Interval Open-ended question
the past 12 months Responses range between 1-12
Whether or not the last visit to Michigan is the Nominal Open-ended, but recoded during
most recent trip' the inteniew as:
I-What was the main destination of this trip? l=Michigan destination
City/Place 2=Non-Michigan destination
State/Province/Country
2-Was a place in Michigan the main Nominal l=Yes
destination of any of the pleasure trips you've 2=No
taken in the past 12 months? -99=DK/NR
The season of the last visit to Michigan Nominal Open-ended question with a
numerical answer (1-12 months)
Months recoded into 4 seasons
12 and l-2=1=Winter,3-
5=2=Spring; 6-8=3=Summer; 9-
11=4=Fall
The length of stay during the last visit in Ratio Responses range between 0-199
Michigan (# of nights) Measured in two variables, see
explanation below.
The number of activities participated in during Ratio 0-11

the last visit in Michigan

Recreated variable. see explanation
below.

a : Those respondents who visited Michigan sometime but not within the past 12 months were not included since they didn’t have

responses for the trip-related variables.

Table 11. The original and recoded categories of the visitation-related variables.

Original Variables

Whether or not the last visit to
Michigan is the most recent trip
l-What was the main destination

Original Response
Catgories
Open ended, but recoded
during the interview as:
l=Michigan destination

New Dummy Variables

These two variables were

 

 

of this trip? 2=Non-Michigan recoded into one dummy
City/Place destination variable
State/Province/Country I=Visited MI most recently
2-Was a place in Michigan the l=Yes O=Visited MI less recently
main destination of any of the 2=No
pleasure trips you‘ve taken in the -99=DK/NR
past 12 months?
The season of the last visit to Winter I=Visited MI in the Winter
Michigan“ O=Visited MI in other seasons
Spring I=Visited MI in the Spring
O=Visited MI in other seasons
Summer I=Visited MI in the Summer

O=Visited MI in other seasons

' I A dummy variable was not created for the Fall season to avoid the singular matrix problem in the regression analysis.
No modification was performed on continuous variables.

83

As was mentioned before, the MRTMS contained two pleasure trip blocks: most
recent trip block and less recent trip block. If the respondents’ most recent trip was to
Michigan, they were not asked the less recent trip block questions. If their most recent
trip was not to Michigan, then they were asked if they took a trip to Michigan within the
past 12 month. Thus, Michigan visitors’ trip data pertaining to their trip to Michigan
within the past 12 months were stored in two data blocks in the same data set.

To apply the recent visitor segment model analysis, the data set was split into two
sets of respondents: (1) those who visited Michigan on their most recent trip (1,723
cases) and (2) those who visited Michigan sometime within the past 12 months (1,415
cases). Then, two new data ﬁles were created: (1) a ﬁle containing the most recent
Michigan travelers with their responses to the most recent trip block and (2) a ﬁle
containing the less recent Michigan travelers with their responses to the less recent trip
block. Since both blocks included the same questions regarding pleasure trip variables
and both ﬁles had these blocks along with image battery and sociodemographic variables,
these two ﬁles were merged into one ﬁle with a recent visitor sample including both most
recent and the less recent Michigan visitors. Thereby, both segments could be included in
the regression analysis, which was not possible with the original data format.

Once the most recent and the less recent blocks were combined, a few more
modiﬁcations were needed. First, since there was a skip pattern involved in the
questionnaire, those who did not travel to Michigan on their most recent trip seemed to
have missing data on the variable of whether or not the last visit to Michigan is the most
recent trip. To execute the regression analysis on all respondents, those less recent trip

respondents were coded as zero on this trip recency variable. A similar modiﬁcation was

84

done to the length of stay variable. If respondents said that their trip was a day trip, they
were not asked how many nights they spent in Michigan. To include these day-tripper
respondents in the regression analysis, they were coded as zero on the length of stay
variable (# of nights). Thus, the number of cases included in the recent visitor segment
model analysis is 1,269.

The number of activities is not an original variable; rather it was recreated from
another set of variables containing activities pursued on respondents’ trips. On the
survey, respondents are asked to give yes/no responses on 11 types of activities to
indicate whether or not they engaged in each of these activities during their last trips (see
survey instrument in Appendix A and B for these activities). These activity variables
were recoded into the number of activities variable by using the SPSS’s count ﬁinction,
which counted the yes responses (1) in each of these variables, thereby revealing the total
number of activities engaged in during the last trip. Similar to the procedure followed in
the general population model data preparation, cases with —99 and —55 were recoded as
system missing since they could not be included in the multiple regression analysis.

Data were weighted to adjust for the uneven sampling and participation rates
across the states and Ontario that were included in the study region. This was necessary
to achieve a sample proportionate to the actual population sizes across the study region.
Household populations in these states were obtained from the 2000 Census data of the
US. Census Bureau (2001). The weighting ratios for each state and Ontario are provided
in Table 12. For example, cases from Ontario were multiplied by 1.06095056689478 to
adjust the sample of households obtained to the population distribution of households in

Ontario. When data were weighted, those cases that did not have a valid response for the

85

state variable were dropped from the further statistical procedures. The sample sizes for
weighted and un-weighted datasets were also different as can be seen in Table 13.

Table 12. Weighting ratios used for each study region.

State Weighting Ratio
Illinois 1.47893712427784
Indiana 0.823092081 148211
Michigan 0.839421243061205

 

Ohio 1.183447641793
Wisconsin 0.657691714622163
Ontario 1.06095056689478

 

Table 13. Sample sizes for weighted and un-weighted datasets.

 

 

Un-weighted Weighted
All respondents = 21,111 All respondents = 20,704
All visitors = 3,138 All visitor = 2,903

 

Data Analyses

The Michigan Regional Travel Market Survey is a longitudinal ongoing project;
however, only data gathered in 1996, 1997, 1998, 2001 (November & December only)
and 2002 were analyzed because image measures were included in the instrument only
during these periods. On average, 400 interviews were completed each month during this
period. A total of 21,111 completed interviews were available from this period; however,
the number of interviews suitable for analysis was considerably less. As was explained
before, some cases were not used in some analyses because of the fact that respondents
either terminated the interview or did not answer all questions pertinent to these analyses.
For example, in the all-years’ data (1996, 1997, 1998, 2001-November & December-,
and 2002), only 5,485 cases were included in the exploratory factor analysis, since only

these cases contained responses for all of the 15 image measurement items.

86

As was mentioned in the literature review of factor analysis, it is recommended to
use the largest sample possible for developing a parsimonious factor model (Hair et al.
1998). For this reason, all usable cases in the all-years’ dataset (1996, 1997, 1998, 2001-
November & December-, and 2002) were utilized in exploratory factor analysis in this
study. Similarly, all usable cases in the all-years’ dataset were also used in testing both
the general population model and the recent visitor segment model. The purpose was to
be as inclusive as possible while testing the ﬁt of these models to the data. Subsequently,
both exploratory factor analysis and multiple regression analysis are also applied to the
data from two different time periods to validate the results obtained from the all-years’
data. These two data periods are the beginning years of the survey (including 1996 and
1997) and later years of the survey (including 2001 and 2002). In other words, the data
from 1998 are not included in the validation analyses. This was performed to differentiate
between the data of initial tests and data of validation tests.

The data were analyzed in a four-step process. In Table 14, a summary of each of
these steps, the variables included, variable types, and analysis tools used (including
frequencies, descriptives, exploratory factor analysis, and multiple regression analysis) is
provided. Analysis tools in the Statistical Package for Social Sciences (SPSS, Version
10.0) soffware were used.

In the ﬁrst step, both all respondents and the recent visitor subgroup were
examined in terms of demographic characteristics to gain a general understanding of the
characteristics of the sample. Frequency distributions, and descriptives, such as valid case
numbers, minimum and maximum values, means, and standard deviations were included

as statistical descriptions where necessary. Since a considerable amount of respondents

87

l

Table 14. Steps of analyses used in the study.

Step 1 - Sample Description & Comparisons

 

 

Variables Type Analysis Tools
Respondent sociodemographics (general population & recent Continuous Descriptives, t-test & One-way ANOVA
visitor segment) Categorical Frequencies & Chi-square

Comparisons of different groups of respondents &
different race groups

 

Step 2 — Image Description, Comparisons & Data Reduction

 

Destination image (15 items)
Comparisons of image for different groups and
different periods

Continuous

Descriptives, t-test & Exploratory Factor
Analysis

 

Step 3 - General Population Model Test

 

Dependent Variable: Destination image
Independent Variables

Respondents’ age

Respondents’ race-White

Respondents’ gender-Female

Respondents’ total annual household income-Above the
median

Respondents’ state of residence-Illinois
Respondents’ state of residence-Michigan
Respondents’ state of residence-Ohio

Respondents’ state of residence-Wisconsin
Respondents’ state of residence-Ontario

Overall travel experience within the past 12 months
Prior visitation to Michigan

The season of the survey-Spring

The season of the survey-Sununer

The season of the survey-Fall

Continuous

Continuous

Dummy
Dummy
Dummy

Dummy
Dummy
Dummy
Dummy
Dummy

Continuous

Dummy
Dummy
Dummy
Dummy

(OLS)
Multiple Regression Analysis

 

Step 4 - Recent Visitor Segment Model Test

 

Dependent Variable: Destination image

Independent Variables

Respondents’ age

Respondents’ race-White

Respondents’ gender-Female

Respondents’ total annual household income-Above the
median

Respondents’ state of residence-Illinois

Respondents’ state of residence-Michigan

Respondents’ state of residence-Ohio

Respondents’ state of residence-Wisconsin
Respondents’ state of residence-Ontario

Overall travel experience within the past 12 months
Prior visitation to Michigan

The season of the survey-Winter

The season of the survey-Spring

The season of the survey-Summer

Frequency of visitation to M]

Whether or not the last visit to MI is the most recent trip
The season of the last trip-Winter

The season of the last trip -Spring

The season of the last trip -Summer

The length of stay in MI during the last trip (it of nights)
The # of activities participated in during the last trip in M1

Continuous

Continuous
Dummy
Dummy
Dummy

Dummy
Dummy
Dummy
Dummy
Dummy
Continuous
Dummy
Dummy
Dummy
Dummy
Continuous
Dummy
Dummy
Dummy
Dummy
Continuous
Continuous

( OLS)
Multiple Regression Analysis

 

88

were eliminated in factor and regression analyses, those respondents who were included
in these analyses were compared to a group of respondents who provided partial
responses to the 15 image items and model-related variables. This was performed to see
if the respondents included in these analyses are different from those who were excluded
and to check for the possibility of bias in results regarding the general population. Chi-
square tests and t-tests were used to check for the signiﬁcance of the differences among
different groups of respondents.

Also, the white race is over-represented in some of the analyses regarding the
general population; therefore, different race groups were compared on selected key
variables to see if under-representation of non-white races might have caused a possible
bias in the results of this study. Chi-square tests and One-way ANOVA tests were used to
check for the signiﬁcance of the differences among different races.

The Chi-square test provides the distribution of a categorical variable on another
categorical variable and computes Chi-square statistics based on the differences between
the observed and the expected frequencies, assuming these observed and expected
frequencies to be equal for the null hypothesis to be true. The t-test provides the mean
values of a continuous variable on the categories of a dichotomous variable and computes
a t-statistic based on the differences between the mean values and variances for the two
groups, assuming the mean values to be equal for the null hypothesis to be true. The One-
way ANOVA provides the mean values of a continuous variable on the categories (more
than two) of a nominal variable and computes an f-ratio based on the differences between
the mean values and variances for the different categories, assuming the mean values to

be equal for the null hypothesis to be true.

89

In the second step, the 15 image items are examined using descriptive statistics.
The t-test was used to compare mean ratings of each image item between earlier (1996-
1997) and later (2001-2002) years of the MRTMS to assess the change in the perception
of Michigan. A similar comparison was conducted between ratings of visitors and non-
visitors. Also, the raw image items were subjected to factor analysis to reduce the image
data to few manageable and meaningful image factors. Exploratory factor analysis of
SPSS 10.0 was employed as a data reduction tool to derive meaningﬁrl and uncorrelated
factors to be used in subsequent analyses.

The items were factor analyzed using the principal component analysis as the
initial factor extraction method. Initially, the Eigenvalue-exceeding-one criterion was
used to decide on the number of factors to be extracted. Subsequently, these factors were
rotated using the Varimax rotation method to obtain uncorrelated factors to be used in
subsequent analyses. Image items with loadings 0.5 and higher were considered as
substantial loadings and used to represent the factors. Eventually, the internal stability of
the factors was tested by using Cronbach’s alpha. For additional analysis, factor scores
were computed for each case using the regression factor scores function of SPSS 10.0.
With this procedure, the regression scores can be saved automatically as soon as the
factors are extracted.

In the third step, the general population model was tested. The calculated scores
for the two image factors along with the overall image (mean image score), and
independent variables, namely, age, race, gender, income, the state of residence (Illinois,
Michigan, Ohio, Wisconsin and Ontario), overall travel experience within the past 12

months, prior visitation to Michigan, and the season of the survey (winter, spring and

90

summer) were used as inputs for the multiple regression analysis. Ordinary least square
regression was employed to investigate the relationship between the dependent variable
and the set of independent variables and to determine statistical signiﬁcance and relative
inﬂuence of each independent variable as an determinant to destination image held by the
general public. Independent variable coefficients were estimated and tested for
signiﬁcance and relative inﬂuence as delineated in the hypotheses discussed in Chapter 1.

In the fourth step, the recent visitor segment model was tested. The calculated
scores for the two image factors along with the overall image (mean image score), and
independent variables, namely, age, race, gender, income, state of residence (Illinois,
Michigan, Ohio, Wisconsin and Ontario), overall travel experience within the past 12
months, the season of the survey (winter, spring and summer), the frequency of visitation
to Michigan within the past 12 months, whether or not the last visit to Michigan is the
most recent trip, the season of the last visit to Michigan (winter, spring and summer), the
length of stay (# of nights) in Michigan during the last visit and the number of activities
participated in during the last visit to Michigan were used as inputs for the multiple
regression analysis. Ordinary least square regression procedures were applied.

The basic model of the study is:

Y = [3X + e
where, Y is the column of Y values (random or stochastic), X is a matrix of X values
(ﬁxed or nonstochastic), ﬂ is a column of coefficients, and c is a column of random error
terms (homoscedastic).

Multiple regression was used to test the multiple hypotheses that the coefficients

of the selected independent variables are not zero in explaining Michigan’s image. The

91

multiple regression model deﬁned in the ﬁrst chapter was estimated by using the ordinary
least square (OLS) regression model, assuming the dependent variable is continuous and
normally distributed, and there is constant variance and lack of multicollinearity and
singularity. SPSS version 10.0 was used to run the OLS regression. Coefficients, standard
errors, t-statistics, p-values and means of the independent variables were used to test the
hypotheses. Standardized or beta coefficients were also calculated to compare the
magnitude of the impact of each independent variable on Michigan’s image. Model ﬁt
statistics, namely, R2, the f-statistic, and a (p-value) were calculated and used to evaluate
the signiﬁcance of linear relationships between the dependent variable and the
independent variables in the population.

Because of the progressive dropout in the sample throughout different steps of
analyses, the sample size gets smaller in each step of analyses. However, since the
MRTMS database is rather extensive, the number of cases involved in each analysis of
this study is above 300. Within the frame of Central Limit Theorem, which was provided
in Chapter 2, the data will be assumed to represent the study population even if the
normality assumption was violated. Therefore, no test of normality was performed.
Multicollinearity and singularity were evaluated by the use of tolerance values and

correlation matrices calculated during the regression analyses.

92

CHAPTER IV

RESULTS AND DISCUSSION

The primary objective of this dissertation was to expand knowledge and
understanding of the relationships between destination image and selected variables
thought to inﬂuence destination image formation including: age, race, gender, income,
the state of residence, overall travel experience within the past 12 months, prior visitation
to the study destination, the frequency of visitation to the study destination within the
past 12 months, whether or not the last visit to the study destination is the most recent
trip, the season of the last visit to the study destination, the length of stay in the study
destination during the last visit, the number of activities participated in during the last
visit to the study destination and the season of the survey. Secondary data gathered by
computer-assisted telephone-interviewing laboratory (CATI-Lab) were used in the
analyses conducted. The variables used to test the study hypotheses can be categorized
into four groups: Destination image measures, respondents’ demographics, respondents’
past travel behavior, and a methodological variable.

Before executing the main analysis techniques of this study, which are
exploratory factor analysis and multiple regression analysis, numerical summary
measures of the data, descriptive statistics and frequency distributions, were developed
for all related variables to assess the central tendency, variability, and frequency of these
variables. The results of the analyses are presented in three sections.

The ﬁrst section contains the demographic proﬁles of all respondents and recent

visitor respondents for different survey periods. Throughout this chapter, all respondents ‘

93

(including visitors and non-visitors) will be dubbed as the ‘mixed sample’. The proﬁles
of the mixed sample and the recent visitor group are provided in three segments: (1) all
respondents, all cases with demographic information whether or not they have complete
information on image or model related variables (2) image respondents, cases with
demographic information and complete information on all 15 image items, and (3) model
respondents, cases with demographic information and complete information on all 15
image items and model related variables. This proﬁle segmentation was performed to see
if these groups grossly differ from one another. The same proﬁling was performed
separately on the all-years’ data (1996,1997, 1998, 2001, and 2002) as well as the 1996-
1997 and 2001-2002 subsamples since factor analysis and regression model test analyses
were applied on all-years’ data as well as these subsamples for validation of the results.
Given that the data were found to be robust, it was expected that analyses on these
different subsamples would lead to very similar overall ﬁndings. This section also
contains comparisons of different groups of respondents, including: (1) comparison of
respondents who completed all 15 image items with those who responded image items
partially (at least one item but not all), (2) comparison of respondents who completed all
general population model-related variables, including the 15 image items, with those who
responded image items partially (at least one item but not all), (3) comparison of
respondents who completed all general population model-related variables, including the
15 image items, with those who completed all 15 image items but not all general model-
related variables, (4) comparison of different races. These comparisons were performed

on few selected key variables to check for possible bias due to systematic exclusion of

94

respondents with partial non-response and/or due to overrepresentation of the white race
group in the results of the image analysis and the general population model test.

The second section deals with image measurement item analyses. First,
descriptive statistics including means and standard deviations of each image
measurement item are provided for different time periods and different groups, namely,
visitors (all three types of visitors) and non-visitors along with the t-tests on differences
between different periods and groups. Also provided is the correlation matrix of the
image measurement items. Then, results of the exploratory factor analysis on 15 image
items are provided for different subsamples, along with the schematic depiction of the
resulting two-factor model.

The third section contains the analyses pertaining to the regression models that
were developed. First, all model variables, both dependent and independent, that were
used to test study hypotheses are described. Sample means are presented for all variables,
and frequency distributions are also provided for dummy variables. The results of the
hypotheses tests for both the general population model and the recent visitor segment
model are presented in the remainder of this section. General population model results
include the relationships between destination image and proposed determinant variables,
including respondents’ sociodemographics, overall travel experience within the past 12
months, prior visitation to Michigan, and the season of the survey. Recent visitor segment
model results contain the relationships between destination image and proposed
determinant variables, including respondents’ sociodemographics, overall travel
experience within the past 12 months, the frequency of visitation to Michigan within the

past 12 months, selected last visit related variables and the season of the survey.

95

Sample Profiles

Since both factor analysis and model test analyses were applied to data from
different time periods for validation purposes, sample proﬁles were also analyzed for
these different time periods to check for discrepancies between these different
subsamples. The three different time periods are: (1) all-years (1996, 1997, 1998, 2001
and 2002), (2) the initial-years (1996-1997), and (3) the later-years (2001-2002). Within
each time period, different groups were analyzed separately in terms of descriptives and
frequencies: (1) all respondents, (2) image-respondents, those who answered all image
items with valid answers, and (3) model-respondents, those who provided valid responses
for all variables that are included in the regression models.

Cases with missing values or invalid responses for the image measurement items
were deleted listwise in the exploratory factor analyses; that is, only those respondents
giving a valid answer for all image items were included in the analysis. In addition, those
cases that did not have a valid data point for the variables included in the models
(highlighted variables in Table 15 and Table 21) were deleted in the regression model test
analyses. The proﬁles of image respondents and model respondents for each data period
are also provided for the purpose of comparisons.

The same procedure was applied for both the mixed sample and the recent visitor
segment. See Figure 9 for a visual depiction of these samples and subgroups. The model
respondents in the mixed sample refer to the general population model respondents; the
model respondents in the recent visitor segment refer to the recent visitor segment model

respondents.

96

FULL SAMPLE-1996, I997, 1998, 2001, 2002 (N=21,lll)III

1996-1997 Sample” 1998 Samplec
(n=9,890) (n=5,217)

'.O O O O O O O O O O O O O O

(“$995) '

.,WFEEET‘

 

.nssaeuusnsa.uss.issaisw

 

Image Respondentsf
(n=l,308)

oo-opocoxo~0199999q9!9392-s93
.TT.,,H_T,y H_H.,E,
‘. .'.;‘ (XI. ‘ '. ‘ .‘ H .>.'.'.‘ ' ’ . ' ‘.‘
V.w.-.

Image Respondentsf '
(n=2,580)

 

 

ti.

o
a
O
t
(a
9%
ta
4.2:.
g
(Q
W’
V
Q3
(1
ti
c}
QR
ﬁ'
17
t?
11
.L
1*
t“
o
f
t

lam-M“
{New
L Recent Visitor i;
Segment
Model

Respondents|I
(n=729)

"66:“ 6‘3 6*“; 23:66?“ '31; H Sid-'1? i 3"".‘5

o.......ooo...osooaoseaaeedaeasseieb.gyyegp_

 

 

 

Recent Visitor
Segment
Model
Respondents.
(n=346)

Recent Visitor
Segnent
Model _ 4,.
Respondents. E g
(n=l93) " ’

 

 

 

.,’.’.‘
i-. x0

 

 

 

aux-L ALA

ooooooooooooooooooogﬁoooqgoqeoeagopgoaiﬂ5_,

 

 

 

 

 

.\ ‘i ‘
._. V . 't .' . i . ‘ . vi ‘3‘ r
ooooooooooooooooooooooooo 60“

 

 

VISITORS”
~ ~ '(n=1,756)

.ooduodenaooooooooooooooooo

 

 

 

 

 

 

VISITORS“ : _.
(n=3,402) ,. F

_ VISITORSe
‘5 ‘(n=3,7'14)

________ummuwm

OOOOOOOOOOOOOOOOOOOOOOOOOOOO.,O-O_,O,.
- - - - - - - -

 

: All usable cases in this sample are used to test the study hypotheses.

: All usable cases in these samples are used for validation of the hypothesis test results.

: This sample is excluded in the validation tests for the differentiation purpose.

: Those who never visited Michigan.

: All three types of Michigan visitors (most recent within the past 12 months. less recent within the past 12 months. and visited
sometime before the past 12 months).

: Those respondents who provided valid responses for all 15 image items.

8: Those respondents who provided valid responses for all 15 image items and variables included in the general population model.

: Recent visitor respondents who provided valid responses for all 15 image items and variables included in the recent visitor segment
model.

Figure 9. Samples, subsamples and subgroups utilized in different analyses.

In comparing these subgroups, a subjective assessment was used instead of
inferential statistics since these groups are not mutually exclusive. So, if the percentages
in image and model respondents subgroups are higher, it means this demographic group

(e.g., black race group) is more likely to complete the survey with valid responses; if it is

97

lower, it means this demographic group is less likely to complete the survey with valid

responses.

The Mixed Sample Proﬁle

The frequency distributions and descriptive statistics for the mixed sample across
different time periods and different subsamples. are presented in Table 15. The total
sample size for all respondents for the all-years’ data is 21,111; however, if a state code
was not assigned for a case, it was discarded since it could not be used in the weighting
scheme used to balance the sample by respondent origin. Deleting these, reduced the full
sample size to 20,704. Some cases have missing values on some demographic variables;
therefore, sample sizes for different demographic variables vary in Table 15.

As can be seen in Table 15, in the all-years’ sample, Illinois and Ohio have the
highest percentages (21.3% and 20.9%, respectively) and Indiana and Wisconsin have the
lowest percentages (11.1% and 10.1%, respectively). With slight differences in
percentages, the same pattern exists in the 1996-1997 data. However, in the 2001-2002
data, the highest percentage belongs to Ontario (20.8%) while the second highest is still
Ohio (20.3%). The lowest percentages again belong to Indiana and Wisconsin (10.5%
and 9.8%, respectively).

When it comes to the image-respondent and model-respondent subgroups, the
percentages of all states decrease except for Michigan, which is more than double as
much as that of the all-respondents group across all time periods. This results from the
tendency of Michigan resident respondents to fully complete the survey while other

respondents were more likely not to complete the survey and provide valid responses to

98

£5: «was... 2 =a 5.. mom—«canoe 23> @0335 9.3 owe—E. u u

 

 

 

.3005 Gauge ”Emu—3E 05 5 032.3.) houmnvea a as 00620.: £ 036.55, 2E. u o dos—Er Miami.— 55» mom-«0 0» 05¢ m05dtd> 23.—«Ev he .5de mos—5’ Z n a
3.2: 8:238 seem as s 322: 8325 =a E 8888. 39, Eases 2.3 32¢ N u .«8« 2: _8« .33 .32 .82 9:32: "as: =a 2: n .
3.» «.E 3“ 2L «.2. 2e 3.; a: 2... .33 5522 «an;
E «a 2 o... «.v S 3. «.n as 3. at 22.38.. 2: 5 Sea Bases:
a: «._« «3 2.. «a. o.«« v.2 w: ”.3 3. at 223.5: 2: s .520 .25
I: n: 0.: 3; can 2; 3m 2.» :a s... at 223.5: 2: s 2 as 5.5 2% uniooﬁm
of 3: a: «.2 v.2 v.2 E. «.2 _ .1 3.. at 2333; 2c s :20 32.8-5
v3 «3 a: «3 an S.« a« «2 e: 9 52.5 22.88; 2: s mamas 25:2
”2 «Z 8.. ﬁ.. «2 n: «2 as «2 9 sass 5.53m“? 25 .3»
as. 93. 3... 0.3 ”.9. «d. 3.. a. s. 2.. vase.» 5:6 .3
3.. «.8 an m. _ n Is 3% 3n 3e .3 .5 8. -8. :55; 3. ea. stew» 55 2c:
Q: ”.2 92 S: v.2 «.2 «.2 e.«« «.2 «8. .3. SEN; no. so. size 5.3:. 220m
3: «.2 one a; 0.8 «so «.8 at as «8. -8. eta: .3. -09 355s 53.»... :2?
use 2:8... 3.5:.
«.2 3n «.8 2n .3 98 an an is east
as. 3... in 03. 3.. 3m «.3. 3... 3.” us:
.3; 5.5
a... E o.« 2 3 v.« I 3 «.« .35
m.« o.« 3. 3 2 «.v 3 3 an .535
3 «a. 3 2 an 3 I 3 2 5.25.5:
2 v.« 3 2 I «.« .2 2 «.m 3.6352;
mi 92 2: E 3 3. «a n: :2 Beam
3 3 f 2: 2: a; «a no «.9 25:5
2... v.3 “.8 $3 08 m2 is 0.8 S.“ 2:53
in EoEkOEEM
2 2 «a m.« 2 «a 3 an a... sec
“.3 «.3 «.«w 9% 23 I» g m: .3 was:
3 2 I S 2 a. E S f 25%:
”a «a 2 I E 3 3 i 3 seem.
3 mo 3 2 E 2 2 2 2 222a 53.
2 3 MS I. to to to so 3 $2.05. 232
o3; «.520 9.5—
«a 3 «.2 9a «.2 no. to Z: 2: 3.8.6
as «e M; Z: «a Z: «a E I: 5282:.
«.2 «.2 «.2 3: 5: o._« S: «.2 2a 3.6
3.. .5. «.2 «.«m a? a: 92 com 2: gases:
2 3 2: we «a n: 3.. as Z. «5%.;
«.2 «.2 E: “.2 .2 o.«« a: _.: «._« .353
as; 85:8: 3 35m
mAmvewE sﬁem. .w5 £8341: 333.95 3.3.95 953.95
u. we Eco—z gamed owné =< can: E52 9%»: owns: =< vamp: Eco—2 0&8 omen: =<
nasceaooana 83-82 anaeooqvus 52-3.: ._€o«.o«-w§.2u5 .2332 asst; 2.. s. 25:

 

 

.mqsoawnam as“ 35mm 2:: ESobE .6,“ momzmﬁonou .mﬁouconmﬂ =a mo mﬂocoswot Ea muzatomon 2C. .2 2an

99

all survey items. Michigan residents’ tendency to complete the survey might come from
their intimate familiarity with Michigan, their own state, or from the felt responsibility to
comply due to patriotic feelings.

The white race group comprises the majority of the sample across all data periods;
compared to the percentage in the all-years’ data (84.1%), the percentage is slightly
higher for the 1996-1997 and slightly lower for the 2001-2002 data (85.4% and 82.2%,
respectively). The percentages of white respondents in the image-respondent and model-
respondent subgroups are higher than that of the all-respondent groups across all data
periods, which means white respondents were more likely to complete the interview with
valid responses to all questionnaire items. This complies with ﬁndings in the
methodology literature about white peoples’ higher tendency to provide information
about themselves over the phone as was discussed in Chapter 2.

With only slight differences across periods, the second highest race represented in
the samples is black (about 6% for all periods). In general, percentages of Asian paciﬁc,
black, Hispanic and ‘other’ race groups decreased slightly in the image-respondent and
model-respondent subgroups while the percentages of native American race group had
slight increases in the image-respondent and model-respondent subgroups. This means
that Asian Paciﬁc, black, Hispanic and ‘other’ race groups were less likely to complete
the survey with valid answers.

In terms of the employment status, the biggest group is the employed full-time
group (about 60%), followed by the retired group (about 18%) across all periods. In the
image-respondent and model-respondent subgroups, the percentages of full-time

employed respondents have considerable increases while the percentages of retired '

100

respondents have noticeable decreases. In other words, full-time employed respondents
were more likely to complete the survey with valid responses while retired respondents
were more likely to either quit or not give valid responses for all survey questions.
Similarly, slight increases are observed in part-time employed respondents in the image-
respondent and model-respondent subgroups across all data periods with some anomalies
in the 2001-2002 data. In general, slight decreases are observed in the percentages of all
other employment groups, except for the student group, across all data periods.

The percentages of female and male respondents are almost identical across all
data periods (about 60% and 40%, respectively). Also, across all data periods, similar
increases (about 6%) in male groups and similar decreases (about 6%) in female groups
are observed in the image-respondent and model-respondent subgroups. This means that
male respondents were more likely to complete the survey with valid responses for all
survey questions.

Respondents who had above the median income ($31K for the initial years, $42K
for the later years) comprise the majority of the sample across all data periods (about
66%). Their percentages in the image-respondent and model-respondent subgroups have
considerable increases (between 5%-13%) across all data periods, meaning, they were
more likely to complete the survey with valid responses to all questions. In contrast,
comprising about 34% of the sample across all data periods, respondents who had income
below the median were less likely to complete the questionnaire with valid responses to
all questions since their percentages decrease about 10% in the image-respondent and
model-respondent subgroups. Respondents who earned more than $50K/$65K ($50K for

the initial years, $65K for the later years) annually comprise approximately 59% of the

101

sample across all data periods. Their percentage increases 3%-5% in the image-
respondent group while it decreases 7%-13% in the model-respondent group across all
data periods. These respondents were a little more likely to complete all 15 image items
with valid responses but far less likely to stay until the end of the survey and provide
valid responses for the variables included in the general population model.

On average, respondents are in their early 40s across all data periods, and the
mean ages for the image-respondent and model-respondent subgroups are a little younger
than that of the all-respondent group in each period except for the 2001—2002 period.

The average number of ﬁill-time wage earners in respondents’ households is
about 1.35 across all data periods; the number is slightly higher in the image-respondent
and model-respondent subgroups across all periods. This coincides with the pattern in the
employment and income variables. The more full-time wage earners and the higher the
income (except for those who earn more than $50K during the initial years of the study
and $65K during the later years of the study), the more likely they were to complete the
survey with valid responses to all questions. This also complies with ﬁndings in the
methodology literature about full-time employed and higher income people’s tendency to
cooperate in telephone surveys as was discussed in Chapter 2.

The average number of persons in respondents’ households is about 2.86 across
all data periods; it is a little higher in the image-respondent and model-respondent
subgroups. In line with this is the pattern observed in respondents having pre-school
children and/or school age children under age 18 in their households. Respondents having
preschool children in their households are about 14% of the samples, and those having

school age children under age 18 in their households is about 34% across all data periods.

102

The percentages of both of these types of respondents increase slightly in the image-
respondent and model-respondent subgroups. Thus, the more people in the household
and/or the more children in the household, the more they were likely to complete the
survey with valid responses to all questions. This is also in line with the likelihood of the
image-respondent and model-respondent subgroups’ being younger than the all-
respondent group across all data periods except for 2001-2002.

Those respondents who had a senior citizen in the household comprise about 25%
of the sample across all periods with moderate decreases (3%-9%) in the percentages in
the image-respondent and model-respondent subgroups. This is commensurate with the
lower average age for the image-respondent and model-respondent subgroups across all
data periods. It is commonly assumed that senior citizens are more likely to comply with
surveyors due to their loneliness. This might be true in this study too since the average
age of all respondents is a little higher than that of the model-respondents across all data
periods. However, older respondents were more likely to comprise the cases with invalid
responses, which were deleted listwise in the image-respondent and model-respondent
subgroups. Similar percentage decreases in the image-respondent and model-respondent
subgroups exist in the group of respondents who had a handicapped person in the
household, which is about 7%.

Although it is not a demographic variable, the visitation variable is also included
in this table since it is a key variable for the purposes of this study. This variable
comprises those respondents who visited Michigan at some point in their life, for the
most recent or less recent trip within the past 12 months, or before the past 12 months.

Visitors comprised 61% of the respondents in the all-years’ data, while this percentage

103

was 2% higher for the 1996-1997 data and 2.5% lower for the 2001-2002 data. The
percentage of visitors is 12%-20% higher in the image-respondent group and 15%-25%
higher in the model-respondent group across all data periods, meaning, respondents who
have visited Michigan were far more likely to complete the survey with valid responses
to all questions. As was proposed as an explanation for Michigan resident respondents’
higher tendency to complete the survey with valid responses, visitors’ tendency to
complete the survey with valid responses could be due to their familiarity with Michigan.
Non-visitors comprise about 25% of the image-respondent subgroup and about 20% of
the model-respondent subgroup across all data periods.

As can be seen in the proﬁles of different respondent groups in Table 15, although
it was necessary to perform the image and model analyses, the progressive elimination
process in the image—respondent and general population model-respondent groups might
have rendered these groups substantially different from the originally targeted random
population of the study region. As was mentioned before, the all-respondents, image-
respondent and model-respondent groups in Table 15 are not mutually exclusive;
therefore, statistical tests could not be run on the differences between these groups. To
gain a general understanding of the differences between these groups and potential bias in
the image and model test results, the image-respondent and general population model-
respondent groups were compared with two different groups of respondents: (l)
respondents who provided partial response to the 15 image items (at least 1 but not all),
thus, were deleted from the image analyses, and (2) respondents who provided full
response to the 15 image items but partial response to the general population model-

related variables, thus, were deleted from the model test analyses.

104

Comparison of Partial Image-Respondents with Full Image-Respondents and

General Population Model-Respondents

About 5,500 respondents were excluded from the image analyses, thus, from the
subsequent model test analyses because they did not provide a valid response to one or
more of the 15 image items. To see if these respondents were different from the retained
full image-respondent and general population model-respondent groups, comparative
tests were applied for these groups in terms of selected sociodemographic characteristics
and past travel behavior. Sociodemographics include: the state of residence, race, gender,
age, employment status, household income, the type and number of people living in the
household, and the number of full-time wage earners in the household. Past travel
behavior variables include: the number of trips taken to any destination within the past 12
months, prior visitation to Michigan, and the number of trips to MI within the past 12
months. Their image of Michigan could not be compared since the partial image-
respondent group does not have complete information on this item. The Chi-square test
was used to compare different groups on nominal variables, and the t-test was used to
compare different groups on continuous variables. The results of these comparisons are
presented in Tables 16 and 17.

As can be seen in Tables 16 and 17, the partial image-respondent group is
signiﬁcantly different from both the ﬁill image-respondent and general population model-
respondent groups for almost all variables. The differences between the partial image-
respondent and ﬁill image-respondent groups are almost identical to those between the
partial image-respondent and general population model-respondent groups, except for the

differences in the presence of a handicapped person in the household. The partial image-

105

Table 16. Comparison of full image-respondents (n=5,485) and partial image-respondents

     

 

 

Full image Partial ima e 1:
Selected Variables respondents‘I respondents Test statistics 2-Tailed
State of Residence (%) n=5,485 n=5,502 131050.688 0.000c
Illinois 17.1% 23.0%
Indiana 9.0% 12.5%
Michigan 36.6% 10.8%
Ohio 18.2% 22.3%
Wisconsin 8.7% 12.1%
Ontario 10.4% 19.2%
Race (%) n=4,628 n=4,600 36:45.638 0.000c
Native American 0.6% 0.5%
Asian or Paciﬁc 1.3% 1.2%
Black 5.6% 4.1%
Hispanic 1.2% 0.9%
White 87.4% 86.6%
Other 3.9% 6.7%
Gender (%) n=5,473 n=5,492 12:67.479 0.000c
Male 44.0% 36.5%
Female 55.9% 63.5%
n=4,375 n=4,003 =-8.7l4 0000"
Age 41.78 44.53
Employment (%) n=5, 167 n=5, 120 x’=121.106 0.000c
Employed full-time 66.0% 58.9%
Employed part-time 9.4% 8.9%
Retired 1 1.5% 18.7%
Not employed 2. 1% 2.3%
Homemaker 4.9% 5.5%
Student 4.3% 3.4%
In some other employment situation 1.6% 2.2%
n=3,685 n=3,297 12=31.776 0.000c
Total household income above the median income - $31K 77.4% 72.3%
for ‘96-‘97, $42K for ‘01-‘02 (Yes%)
n=2,270 n=1 893 12=8.925 0.012‘I
Total household income above $50K for ‘96-‘97, $65K for 64.6% 61.2%
‘01-‘02 (Yes%)
n=789 n=639 x2=17.578 0.000c
Pre-school child in the household (Yes%) 15.2% 12.4%
n=1,969 n=1,649 12:41.89] 0.000c
School-age child under age 18 in the household (Yes%) 37.9% 32.0%
n=923 n=1,301 12:84.855 0.000c
Senior citizen in the household (Yes%) 17.8% 25.2%
n=273 n=320 x2=5.300 0.071
Handicapped person in the household (Yes%) 5.3% 6.2%
' n=5,174 n=5,137 t=4.473 0.000c
Number of people in the household (Mean #) 2.97 2.83
n=5, 152 n=5,076 t=5.086 0.000c
Number of full-time wage earners in the household (Mean #) 1.50 1.39
n=5,406 n=5,346 t=2. 171 0.030d
Number of trips taken in the past 12 months (Mean #) 4.49 4.25
n=3,968 n=3,244 12:219149 0.000c
Visited Michigan (Yes%) 75.9% 62.5%
n=1,578 n=706 t=7.401 0.000c
Number of trips taken in MI in the past 12 months (Mean #) 3.26 2.52

 

: Responded to all 15 image items.
:Respondedtoatleastone image item butnot all.
: Statistically signiﬁcant at 99% level (p<0.01).

: Statistically signiﬁcant at 95% level (p<0.05).

O.

106

Table 17. Comparison of general population model-respondents (n=3,554) and partial
image-respondents (n=5,502).

 

 

Model Partial ima e 11
Selected Variables respondents. respondents Test statistics 2-Tailed
State ofResidence (%) n=3,554 n=5 502 17:861493 0.000“
Illinois 17.9% 23.0%
Indiana 9.4% 12.5%
Michigan 35.6% 10.8%
Ohio 18.3% 22.3%
Wisconsin 9.3% 12.1%
Ontario 9.4% 19.2%
Race (%) n=3,554 =4,600 78:43.843 0.000“
Native American 0.7% 0.5%
Asian or Paciﬁc 1.3% 1.2%
Black 4.8% 4.1%
Hispanic 1.0% 0.9%
White 88.8% 86.6%
Other 3.4% 6.7%
Gender (%) n=3,554 n=5,492 36:54.241 0.000“
Male 44.2% 36.5%
Female 55.8% 63.5%
n=3,554 n=4,003 t=-9.524 0.000c
Age 41.40 44.53
Employment (%) n=3,544 n=5, 120 x2=166. 189 0.000“
Employed full-time 68.6% 58.9%
Employed part-time 9.8% 8.9%
Retired 9.4% 18.7%
Not employed 1.8% 2.3%
Homemaker 5.2% 5.5%
Student 4.0% 3.4%
In some other employment situation 1.4% 2.2%
n=2 868 n=3,297 12:77.131 0.000“
Total household income above the median income - $31K for 80.7% 72.3%
‘96-‘97, $42K for ‘01-‘02 (Yes%)
n=1,805 n=1,893 12:15.683 0.000“
Total household income above $50K for ‘96-‘97. $65K for 66.0% 61.2%
‘01-‘02 (Yes%)
n=559 n=639 x“=20.130 0.000“
Pre-school child in the household (Yes%) 15.7% 12.4%
n=1,367 n=1,649 12:40.937 0.000“
School-age child under age 18 in the household (Yes%) 38.5% 32.0%
n=546 n=1,301 12:122479 0.000“
Senior citizen in the household (Yes%) 15.4% 25.2%
n=165 n=320 12:11.294 0.004“
Handicapped person in the household (Yes%) 4.6% 6.2%
n=3,552 n=5,137 t=4.504 0.000c
Nmnber of people in the household (Mean it) 2.99 2.83
n=3,543 n=5, 76 t=6.255 0.000c
Number of full-time wage earners in the household (Mean #) 1.53 1.39
n=3,554 n=5,346 t=7.931 0.000c
Number of trips taken in the past 12 months (Mean #) 5.24 4.25
n=2,808 n=3,244 x2=269.706 0.000“
Visited Michigan (Yes%) 79.0% 62.5%
n=1,302 n=706 t=6.969 0.000c
Number of trips taken in MI in the past 12 months (Mean #) 3.23 2.52

 

 

 

 

’ : Responded to all general population variables including 15 image items.

b . .
: Responded to at least one image ltem but not all.

c : Statistically signiﬁcant at 99% level (p<0.01).

107

respondents were less likely to be Michigan residents, but more likely to be Ontario
residents. They were more likely to identify themselves as ‘other’ race groups, which is
usually how Ontario residents reply in this study. Also, partial image-respondents were
more likely to be female, retired, older on average, with fewer number of people living in
the household, and less household income. In addition, partial image-respondents had
fewer trips on average not only to any destination but also to Michigan within the past 12
months, being less likely to have visited Michigan before. In other words, both ﬁill
image-respondent and general population model-respondent groups were more likely to
be younger, residents of Michigan, with above-the-median incomes, and more
experienced and more familiar with Michigan due to previous visitation to Michigan.
However, these differences in the proﬁles of excluded and retained groups of respondents
do not assure bias in the results of image analyses and model test analyses of this study
since it is not known if the excluded respondents’ image of Michigan was signiﬁcantly

different from that of the retained respondents.

Comparison of Full Image but not Model-Respondents with General

Population Model-Respondents

To search for evidence of differences in Michigan’s image between the excluded
and the retained respondents, another comparison was conducted between the general
population model-respondent group and respondents who provided full response to the 15
image items but not to the general population model-related variables (n=1,930). The

results of these comparisons are presented in Table 18.

108

Table 18. Comparison of general population model-respondents (n=3,554) and full image
but not model-respondent: (n=1,930)‘.lL

 

 

 

_ 3
Selected Variables Model Full image but not Test a
‘ respondents” model resp.b statistics 2-Tailed
State of Residence (%) n=3,554 n=1,930 xz=21,091 0001c
Illinois 17.9% 15.5%
Indiana 94% 8.3%
Michigan 35.6% 38.4%
Ohio 18.3% 18.0%
Wisconsin 9.3% 7.6%
Ontario 9.4% 12.1%
Race (%) n=3,554 n=1,073 76:32.414 ogooc
Native American 0.7% 0.6%
Asian or Paciﬁc 1.3% 1.3%
Black 4.8% 8.0%
Hispanic 1.0% 1.9%
White 88.8% 82.9%
Other 3.4% 5.4%
Gender (%) n=3,554 n=1,920 g“=3.810 0.149
Male 44. Zo/o 43 . 70/0
Female 55.8% 56.2%
n=3,5 54 n=821 t=~3.808 o_000°
Age 41.40 43.45
Employment (%) n=3,544 n=1,621 12:71.145 0000“
Employed full-time 68.6% 60.6%
Employed part-time 9.8% 8.8%
Retired 9.4% 16.2%
Not employed 1.8% 2.8%
Homemaker 5.2% 4.4%
Student 4.0% 5.2%
In some other employment situation 1.4% 2.0%
n=2,868 n==817 x2=88.700 0_000°
Total household income above the median income - S31K for ‘96-‘97. 80.7% 67.6%
S42K for ‘01-‘02 (Yes%)
n=1,805 n=465 xi=11.258 0,004“
Total household income above $50K for ‘96—‘97. $65K for ‘01-‘02 66.0% 59.6%
(Yes%)
n=559 n=230 g“=2.384 0.123
Pre-school child in the household (Y es%) 5.7% 14.1%
n=1 ,3 7 n=601 x’=1.823 0.402
School-age child under age 18 in the household (Yes‘%) 38.5% 36.8%
n=546 n=378 11:45.905 0,0006
Senior citizen in the household (Y es%) 15.4% 23.1%
n=165 n=108 x2=9.l66 (1010c
Handicapped person in the household (Yes%) 4.6% 6.6%
n=3,552 n=1,622 t=1.131 0.258
Number of people in the household (Mean #) 2.99 2.93
=3,543 n=1,609 t=2.552 0.0118
Number of fulletime wage earners in the household (Mean #) 1.53 1.45
n=3,554 n=1 852 t=13.923 0000“
Number oftrips taken in the past 12 months (Mean 4) 5.24 3.05
n=2 808 n=1 16o x“=59.422 0.000“
Visited Michigan (Yes%) 79.0% 69.2%
n=1,302 n=277 t=-0.953 0.341
Number oftrips taken in MI in the past 12 months (Mean 91) 3.23 3.37
n=3,554 n=1,930 t=1.405 0.160
Mean image score 7.50 7.43

 

 

 

b

" : Statistically signiﬁcant at 99% level (p<o.01).
d : Statistically significant at 95% level (p<0.05).

109

: Responded to all general population variables including 15 image rterns.
: Responded to all 15 image items but not all of the general population-related variables.

The same sociodemographic variables, past travel behavior variables as well as
the Michigan image were used in comparing these groups. The Chi-square test was used
to compare these groups on nominal variables, and t-test was used to compare these
groups on continuous variables.

As can be seen in Table 18, full image but not model-respondents are signiﬁcantly
different from the general population model-respondents in most of the selected criteria
(11 out of 17 variables). Similar to the partial image-respondents, these respondents were
older on average, more likely to be retired with fewer people living in the household and
less likely to have above the median or above $50Kl$65K household income. However,
surprisingly, these residents were more likely to be Michigan and Ontario residents. Since
the difference between the images of these two groups of respondents is not statistically
signiﬁcant, there isn’t enough evidence for the hypothesis of whether or not the excluded
and retained respondents had different images of Michigan. The results are inconclusive

and this hypothesis cannot be resolved with the available information.

Comparison of Different Races

As can be seen in Table 15, the white race group comprises the majority of the
respondents in the image-respondents and general population model-respondents groups.
The Census Bureau’s (2001) data for the Census 2000 was checked for the distribution of
different races in the US states in the study region (Ontario was excluded due to
unavailable census data). Racial distributions in the US Census data along with diiferent
groups utilized in the analyses of this study are provided in Table 19. As can be seen

from the table, Asian Paciﬁc and black race groups are slightly under-represented and the

110

.0328: 8: a. c2523.“. .203 Mogco no 53 mam—.00 u

.769: cote—smog Rheum 2: 2 26202 one—pats; =a 3.. «3:38. 2.? 32.6... on? 29:3 :3 05 :2 8522.8. 39:. H

 

 

v

.35: owns: 2 =a 8.. mom—5&8 2.: 32.6.5 2.3 29:8 :8 05 2 neon—5&2 0.8:... n 6

e282 3 8858a .628 as 5.? 5.; 32 e5 .88 .232 .32 .82 messes case :3 2F n a

9.8» 82 a me 298 25%.: Sub remoov Sac mam—BU 8:7. 33 beam 2: 2 .556. 65 2 votes 2 953 88 2:32: “ -

9v ﬁn ow .mm 93 few m... ed on m._ m._ m._ Bo ed 0o .32.

3; 6.: e12 . .2. N2. N: . e o no .2 . 3. an an - 2 2 Z - .3325

EN Wm _.v o.~ .8 ”Na $3 ado x._ «a Z 10 m o Nd ed 9N m._ 0. ad n.— 5283.?

vN ma Ev Z .3 ”cm v.5 13 Na n6 2. —.~— 9o 9o 9o m.— we Nd No E: e20

oé we o6 ON .3 new ohm «.5 we vs a.» ad.— o._ o._ 9. Nu he go 9o m..— aqua—9.2

h." a...” we o.~ No ado mdo 0.8 9m me me ad 00 m6 .5 MA 9o n6 5o ed 33:3:

9m Gal .6 we 0: ohm v.3 ﬁns «2n we 0m 92 elm v.~ _.N ad . - _.o ed £25:
‘33: cows... e=< 2.an e182 deuce: e=< 32.5 63.62 cone...— e=< 32.6 c...e_.ez eons... .5 .350 Love: coma: e=< Beau

ago—.5 8.55 sea—m 95.3.— ..e .53. 5822.3. 0332

 

 

 

 

 

 

.356 05 mo 89.98

of 2 23:3 330% 865.22% 2 :25“:me 308 28 8% ooom 3200 m: 2: 2 =2on 35% 2: 2 5:32me .223— .3 0381

111

white race

 

respondent
‘other’ rag
modiﬁcatio
caxegory. 1
study exce;

The

sociodemo g

 

 

MEL whic
responses n
different ra
travel beha
different fa
diﬁerem r:
presented '1
As
the “other
compared
hOUSehold
number 01
signiﬁc‘am

charaqeﬁE

white race group is over-represented, especially in the image-respondent and model-
respondent subgroups compared to their proportion in the Census data in each state. The
‘other’ race group in the study data also includes the Hispanic race group; this
modiﬁcation was conducted since Census data did not treat Hispanic origin as a racial
category. The ‘other’ race group in each state is also slightly over-represented in the
study except for Illinois.

These other race groups might be different from the white group in terms of other
sociodemographic characteristics, such as employment status, education and income
level, which can inﬂuence their travel behavior and ultimately the nature of their
responses to the Michigan image questions. Therefore, comparative tests were applied for
different race groups in terms of other selected sociodemographic characteristics, past
travel behavior, and their image of Michigan. The Chi-square test was used to compare
different race groups on nominal variables. One-way ANOVA test was used to compare
different race groups on continuous variables. The results of these comparisons are
presented in Table 20.

As can be seen in Table 20, respondents were grouped into ﬁve major races and
the “other” category, which embodied minority race groups. These race groups were
compared on nine selected sociodemographic variables, namely employment status,
household income, the type and number of people living in the household, and the
number of ﬁill-time wage earners in the household. Different races were found to be
signiﬁcantly different from one another in terms of all of these sociodemographic

characteristics at the 99% conﬁdence level.

112

no |
Table- ‘

 

|
39291).
Employmcr.
Employee
Employed
Retired
Not emplr
Homemal.
Student _
In some or
situation

  
 

Total housel.‘
income abox~
median incor

Total housch
income abou-
“CS” 0)

 

Preschool ch
household (Y

3.96 18 in the
(Yes%)

Senior citizer
. ho“Si‘hold (Y

Handicapped
the houSCholc

Number Of pg
[he householc

Table 20. Cormaarison of different races on selected variables.

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Native Asian or Test (1
Selected Variables American Pacific Black Hispanic White Other statistics 2-Tai1ed
Employment 93s? n=241 n=940 n=256 11:13.583 n=t.034 13:406325 0000*
Employed full-time 60.7% 55.6% 61.4% 67.2% 58.8% 48.6%
Employed part-time 13.5% 10.4% 7.6% 9.0% 9.3% 9.1%
Retired 7.9% 2.5% . 12.7% 2.3% 17.8% 26.0%
Not employed 2.2% 4.1% 6.0% 6.6% 2.9% 3.5%
Homemaker 4.5% 3.7% 3.2% 4.3% 5.7% 5.8%
Student 6.7% 20.3% 6.8% 7.0% 3.5% 4.2%
In some other emp. 4.5% 3.3% 2.4% 3.5% 20% 2.8%
situation
ﬂ n=146 n=478 n=132 n=8,490 n=48l x2=118.720 0000*
Total household 51.2% 69.9% 56.8% 59.7% 69.2% 57.1%
income above the
median income (Y es%)
n=23 n=86 n=257 n=64 n=4,773 n=261 x2=26.210 0003*
Total household 57.5% 61.9% 56.2% 49.6% 59.4% 59.5%
income above $50000
(Yes%)
n—17 n=48 n=180 n=62 n=1,882 n=l44 x2=47532 0000*
Pre-school child in the 19.1% 19.9% 18.9% 24.2% 13.8% 13.8%
household (Y es%)
n=33 n=110 n=445 n=124 n=4,573 n=354 12:119025 0000*
School-age child under 37.1% 45.6% 46.7% 48.2% 33.5% 34.1%
age 18 in the household
(Yes%)
n=18 n=46 n=231 n=33 n=3.245 n=352 x2=76.488 0000*
Senior citizen in the 20.2% 19.1% 24.3% 12.8% 23.8% 33.9%
household (Y es%)
° n=15 n_=_1_6_ n=107 n=19 n=860 n=96 x2=77.682 0000*
Handicapped person in 17.0% 4.2% 11.2% 7.5% 6.3% 9.3%
the household (Y es%)
n=89 n=242 n=952 n=257 n=13,609 n=1,032 f=24.786 0000*
Number of people in 3.07 3.64 3.07 3.56 2.85 2.87
the household (Mean #)
ﬁg n=237 n=929 n=255 n=13,512 n=1,016 f=l3.501 0000*
Number of full-time 1.55 1.68 1.48 1.62 1.38 1.21
wage earners in the
household (Mean #)
n=70 n=173 n=650 n=152 n=10.625 n=736 1:14.402 0000*
Number oftrips taken 4.21 3.40 2.70 2.88 4.21 3.29
in the past 12 months
(Mean #)
n_=ﬂ w n=339 n=65 n=6.817 n=351 x2=173.935 0000*
Visited Michigan 52.9% 38.7% 52.2% 42.8% 64.0% 47.8%
(Yes%)
n=13 n=14 n=95 n=l6 n=2,070 n=87 f=2.203 0.052
Number of trips taken 3.07 3.24 2.28 2.91 3.03 2.90
in MI in the past 12
months (Mean #)
n=30 9:59 n=257 n=55 n=4,047 n=180 [=11.224 0000*
Mean image score 7.40 6.62 7.10 6.91 7.57 7.13

 

* : The difference is signiﬁcant at 99% level (p-»' .01).

113

Th

terms of t

 

race grou‘
Michigan.
Different 1

different 0

 

and wheth
have \isite
compared
images of

race grour

The Rece
As
"1051 rece
dUﬁng an!
IhUs. the E
Was not a
Was uSed
“Weary
cases Wit]

induded s

Three past travel behavior variables, which were assumed to be key variables in
terms of their inﬂuence on destination image were also included in comparing different
race groups: the number of pleasure trips in the past 12 months, prior visitation to
Michigan, and the number of pleasure tn'ps taken in Michigan in the past 12 months.
Different race groups were compared on these variables and found to be signiﬁcantly
different on two of these variables: the number of pleasure trips in the past 12 months,
and whether or not they visited Michigan at all. These race groups were equally likely to
have visited Michigan sometime in their lifetime. Different race groups were also
compared in terms of their images of Michigan and found to have signiﬁcantly different
images of Michigan. These ﬁndings introduce the possibility of bias towards the white

race group in the results of the general population model of this study.

The Recent Visitor Group Proﬁle

As was explained in Chapter 3, respondents who visited Michigan during their
most recent trip within the past 12 months (1,723) and respondents who visited Michigan
during another trip within the past 12 months ( 1,415) were combined in a different ﬁle;
thus, the size of this sample is 3,138. However, as was mentioned before, if a state code
was not assigned for a case, it was discarded due to the weighting by state process that
was used. This resulted in a sample size of 2,903 for all recent visitor respondents in the
all-years’ data. In Table 21, sample sizes for different demographic variables vary due to
cases with missing values for some demographic variables. The visitation variable is not

included since all respondents in this group are recent Michigan visitors.

114

 

 

. r . @413... 2‘ .::.1 .: .~A~FN.§ 1:» iAChNt~ ..-V
I . .
may. 7:52 . .73 9.15:— :< P. Er: .352 .. .re. 3:2:— =< b .73 3.52 9.13.. .3255 Ex
- HR: 9-2}. ., 5 93-3....“ .ll 33.. _ 42.5 2.3-25. ..:.§.N-a2,.._ tit .233? -i 33%»: 9...}.acweammwﬁ al.

 

,mazoemnEm “ecu moan—wt 2:: gangsta 50.. mufamcmo—zov .mztuﬁcoumoc e085? 2500.53 marzruzcot ﬁzz m0>t5gbaot 9:. gm 93.:

080000.52 0.0.20.8 05 E 03.0.55 08060.5 0 0.0 vows—05 0. 030.50.» 00E. 0 o

.0025» 9:00.08 55 000.8 2 03. 0030500 0.00-00.00:. 8.. 00am“. 0032, Z n

.0500 0020 0. =0 80 8080.2 2.00 0000,80 2.3 32F u o

 

a
0002: .8808 8003 .082 20 .0 000220 8.00.50 =0 80 38008. 0:: 8006.0 2.3 08.: H P .0000 05 800 .002 .002 .002 0500.20 0.0.50 :0 2:- 0 .
0.0 0.0 _.0 0.0 0.0 0.0 0.0 0... _.0 0.0 82 2000002. 20 .0 850 00000205:
0.2 20 0.00 0.: 0.0. _.0_ 0.0. 0.0. 0.00 30 at 32.380 20 .0 5020 30.00
0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 30 at 200302. 05 5 0. 000 000.5 0.20 "00.39.00
0.: 0.2 0.0. 0.: _.0_ 2; 0.: 0.: 0.2 0.0 at 2003.000 20 .0 0.0.0 8000-20
8.0 00.0 00.0 00.0 00.0 _0.0 00.0 00.0 00.0 00 5020 20083.. 05 .0 88.0000 000003.0
3.0 3: 3... 00.. 00.. 000 _0._ 00.. 0E 00 03:0 950.0003 0:00 =00
0.00 0.00 0.0.. 0.00 0.00 0.:. 0.00- 000 .00 .280 0820 00<
2.0 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 00. -8. 8.5000 .00. 00. 0050000 55 202
0.00 0.00 0.00 0.2 _.00 0.00 0.2 0.00 0.00 000. -00. 80.0000. .. 00. .00. 05:02 00000:. 30.000
0.00 0.00 3.0 0.00 0.00 0.00 0.00 0.00 2.0 000. -00. 0205.0 :0. -00. 05:02 .3000... 0.80.0
.0000 2:8... 3.5:.
_.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 000.50
0.00 0.00 0.00 0.00 0.0.. 0.:- ...00 0. 3. 0.00 0:2
.700 00.3.00
0._ 0.0 0.0 0.. 0._ S 0._ 0._ 0.. 55o
0.. 0.0 0.0 0... 0.0 0.0 0.0 0.0 0.0 .8030
0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 00006050:
0.. 0.. _.0 0.. 0.. S 0.0 0.. 0.0 00.00.0800:
_.0_ 0.0_ 0.0. 0.0 0.0 0.0. 0.0. 0.: 0.: 02:00
0.0 0.0 0.0 0.2 0.0 0.0 0.0 0.0 0.0 2000-:00
0.00 3.0 20 _ .00 0.00 0.00 0.00 0.00 0.00 n0.00-=00
. A000 50:50.06”.—
00 0.0 0.0 0.. 0.0 0.0 0.0 .0 0.0 .200
0. .0 0.00 0.00 0.00 0.00 0.00 0.; 0.00 0.00 00.0:
0.. 0._ 0.0 0.0 0.0 0.0 0.0 0.0 0.0 2500:
0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0... 000.0.
0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 00080 503
0.. 0.. 0.0 0.0 0.0 0.0 0.0 0.0 0.0 50:05 30.2
00.00 0.85 83.
0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.50.5
0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 50:80:.
0.0 0.0 3.. 0.0_ 0.: 2.. 0.0. 0.: 0.: 3.6
0. :- 000 0.00 0.00 0.00 .00 0. .0 0.00 0.00 .3030.
E 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0500.0
0.0 0.0 0.2 0.0 0.0 0.2 0.0 _.0_ 0.2 0.8.3:
93.00 85280. 00 0.80
02.00.05 000005 0800.05 0800.05 0:00:05 082.. _ 0.0
0.90.. .032 0.M000 owns: =< 0.90.. .032 0.0—8.. owns: =< 0.280 .032 0.9.00 090:: =<

 

0032330 0:0 00800-0 08: 2000-Eu 00,0 mgr—aﬁonou Eco—0:802 8:200 0:002 MO 03053—0000 0:0 00202000020 0.:- ._N 0302.

00000-00000 0000-0000

00003-00000 002-002

 

000000000400 2000-5.

0033.50; 03900380:

115

As
make up 3
19964997
percentage
groups. ha
respectivelj
higher pen
(about 9%
respondent
decrease in

As
subgroups
except for
reSpondent
anSWers IO
likely 10 C0

As i
in the reCen
across a“ d
the percent
Which mEa
pﬁcentages

respond em

As can be seen in Table 21, in the recent visitor group, Michigan respondents
make up about half of the sample across all data periods (48.0% for all-years, 47.1% for
1996-1997, and 50.0% for 2001-2002). Ohio and Illinois, which had the highest
percentages in the mixed sample, come in as the second and third biggest recent visitor
groups, having almost identical percentages across all data periods (14.2% and 13.8%,
respectively). Indiana, which has one of the lowest percentages in the mixed sample, has
higher percentage of visitors than both Wisconsin and Ontario across all data periods
(about 9%). Having similar percentages across all data periods, Wisconsin and Ontario
respondents comprise the smallest groups in the recent visitor group with a moderate
decrease in the percentage of Wisconsin respondents in the 2001-2002 data.

As in the mixed sample, in the image-respondent and model-respondent
subgroups of the recent visitor group, the percentages of all states decrease about 2-5%
except for Michigan’s, which increases over 10% across all data periods. Recent visitor
respondents from other states were more likely not to complete the survey with valid
answers to all questions, while recent visitor respondents from Michigan were more
likely to complete the questionnaire with valid responses to all questions.

As in the mixed sample, the white race group comprises the majority (about 90%)
in the recent visitor group across all data periods. The percentages of white recent visitors
across all data periods are a little over 5% higher than those of the mixed sample while
the percentages of all other race groups are smaller than those of the mixed sample,
which means white people were more likely to be recent visitors Michigan. The
percentages of white recent visitor respondents in the image-respondent and model-

respondent subgroups are about the same, which means that almost all white recent

116

 

 

recent VlSiT

h-Iidwestert

 

Transporta1
American:
As

gTOUp in tl
than those
resPOrtdent
is retired r
18 50.0 for
“me smart-
is abOUt I}:
vlsitors Ur
percentage
“1038 all

emplo.V'ed

resPODden.

visitor respondents provided valid responses for all image and model variables. As in the
mixed sample, the second highest race group represented in the recent visitor group
across all periods is black (between 3%-5%). Percentages of other race groups more or
less remained the same in the image-respondent and model-respondent subgroups except
for the ‘other’ race group, which incurred slight decreases. The distribution of races in the
recent visitor group is more or less similar to that of the Michigan visitors from the
Midwestem states reported by the US. Department of Transportation, Bureau of
Transportation Statistics (1995), which is: White=95.5%, Black=3.7%, Native
American=0.3%, Asian or Paciﬁc=0.3% and OtheF0.2%.

As in the mixed sample, full-time employed respondents comprise the biggest
group in the recent visitor group (a little over 60%); its percentages are slightly higher
than those of the mixed sample across all data periods. Thus, ﬁill-time employed
respondents were more likely to be recent visitors to Michigan. The second biggest group
is retired respondents (14.3% for the all-years’ data, 12.3% for the 1996-1997 data, and
18.5% for the 2001-2002 data). However, the percentages of retired recent visitors are a
little smaller than those of the mixed sample except for that of the 2001-2002 data, which
is about the same. This means retired respondents were less likely to be recent Michigan
visitors until 2001-2002. As in the mixed sample, retired recent visitor respondents’
percentages drop about 2%-3% in the image-respondent and model-respondent subgroups
across all data periods. Slight increases are observed in the percentages of part-time
employed and student recent visitor respondents in the image-respondent and model-
respondent subgroups across all data periods except for a little drop in student

respondents in the 2001-2002 data. Slight decreases in other employment groups are

117

observed in the image-respondent and model-respondent subgroups, except for the
homemaker recent visitor respondents in the 2001-2002 data, which remains about the
same in the image-respondent subgroup and increases a little in the model-respondent
subgroup.

The pattern in the gender variable for the recent visitor group is the same as that
of the mixed sample: a 40/60 male to female ratio, with male respondents’ being more
likely to complete the survey with valid responses to all questions across all data periods.
Similar to the mixed sample, recent visitor respondents who had above the median
income ($31K for the initial years, $42K for the later years) comprise the majority of the
sample across all data periods (77.7% for the all—years’ data, 78.7% for the 1996-1997
data, and 74.1% for the 2001-2002 data). These percentages are about 10% more than
those of the mixed sample across all data periods. Similarly, the percentages of recent
visitor respondents who had more than $50K/$651( annual income are about 3% higher
than those of the mixed sample while the percentages of recent visitor respondents who
had below the median income are about 10% less than those of the mixed sample. Not
surprisingly, respondents with above the median income were more likely to be recent
visitors to Michigan. As was observed in the mixed sample, there are slight decreases in
the percentages of the lower income recent visitor respondents and slight increases in the
percentages of higher income recent visitor respondents in the image-respondent and
model-respondent subgroups, with the exception of those who earn more than
$50K/$65K which remain the same in the image-respondent subgroup but decrease about
15% in the model-respondent subgroup. Except for those recent visitors with more than

$50Kl$65K annual income, recent visitors were more likely to complete the survey with '

118

valid TBSP
different 5
Tl
across all
responder
there is a
visitor res
younger
questions.
Tl
householc
145), “it
the avera;
290)» As
Visitor re-
SUbgTOUps
earTlers 1r.
peﬁod‘ W}
Th
household

The perCe

 

their housl

data peﬁg

valid responses to all questions regardless of their income level since differences between
different subgroups can be ignored.

The average age of recent visitor respondents is close to that of the mixed sample
across all time periods: early 405. There is a slight decrease in the mean age of the image-
respondent and model-respondent subgroups in the all-years’ and 1996-1997 data, but
there is a slight increase in the 2001-2002 data. This means that until 2001-2002, recent
visitor respondents who completed the survey with valid responses tended to be slightly
younger than those who terminated the survey or gave invalid responses to some
questions, while the opposite case existed in 2001-2002.

The average number of full-time wage earners in the recent visitor respondents’
households is a little higher than those of the mixed sample across all data periods (about
1.45), with the difference being smaller in the 2001-2002 data. A similar pattern exists in
the average number of persons in the households of recent visitor respondents (about
2.90). As in the mixed sample, the average number of persons in the households of recent
visitor respondents is slightly higher in the image-respondent and model-respondent
subgroups (about 2.97); the same pattern exists in the average number of full-time wage
earners in the households of recent visitor respondents except for the 2001-2002 data
period, which is identical across all groups.

The percentages of recent visitor respondents having pre-school children in their
households are similar to those of the mixed sample across all data periods (about 13%).
The percentage of recent visitor respondents having school age children under age 18 in
their households is almost identical to those of the mixed sample (about 34%) across all

data periods. There are slight increases in the percentages of recent visitor respondents

119

who had pre-school children in their households in the image-respondent and model-
respondent subgroups; however, increases in the percentage of recent visitor respondents
having school age children under age 18 in their households in the image-respondent and
model-respondent subgroups are moderately high, meaning, these types of recent visitor
respondents were more likely to complete the survey with valid responses.

The percentage of recent visitor respondents who had a senior citizen in their
households in 2001-2002 is almost identical to that of the mixed sample (about 24%), but
the percentages for other periods are about 4% smaller than those of the mixed sample.
The percentages of recent visitor respondents who had a handicapped person in their
households are about 1.5% smaller than those of the mixed sample across all data
periods. In general, these types of respondents were less likely to be recent visitors to
Michigan and less likely to complete the survey with valid responses since their
percentages are slightly smaller in the image-respondent and model-respondent

subgroups.

EFA Two-Factor Model
In this section, image items are described for different groups and sampling
periods, and possible explanations for observed differences and similarities are presented.
Finally, image item correlations are provided followed by the results of exploratory factor

analysis.

120

Image It
5'
single w
are prese
standard
for the i
den'ations
Table 22
Q

\
15 close en
Has many
IS great for
IS an excm
Has a lot 0
Offers mu:
15 great for

Image Item Descriptions

Since each image item is a relatively long sentence, each was converted to a
single word or a couple of words as a convenience. The conversions used for each item
are presented in Table 22. The 15 image items along with their corresponding means and
standard deviations are presented in Table 23. Since validation analyses were conducted
for the initial (1996-1997) and later (2001-2002) study periods, means and standard
deviations for each data period are also provided.

Table 22. Image items and correspondigg notations.

 

 

Image Item Notation
Is close enough for a weekend getaway getaway
Has many interesting museums museums
Is great for summer outdoor recreation activities summer rec.
Is an exciting place to visit exciting
Has a lot of high quality lodging lodging
Offers much scenic appeal scenic
Is great for winter outdoor recreation activities winter rec.
Is a good place to meet friendly people friendly
Is a place everyone should visit at least once in their lifetime must visit
Is a safe place to visit safe
Offers exciting nightlife and entertainment nightlife
Is a great place for a family vacation family place
Is a popular destination with vacationers popular place
Has many interesting historic sites historic
Oﬂ'ers an excellent vacation value for the money value

 

Table 23. Image item descriptive statistics for different samplin eriods.
All-years* (n=5,485) 19945-1997 (F2579) 2001-2002 (n=1,597)

 

 

 

 

 

 

 

Item Mean 1 Std. Dev. Mean I Std. Dev. Mean I Std. Dev.j
getaway 8.12 2.63 8.24 2.62 7.89 2.52
museums 6.49 2.62 6.14 2.63 7.32 2.46
summer rec. 8.09 2.16 8.13 2.17 8.04 2.14
exciting 7.15 2.33 7.09 2.39 7.34 2.20
lodging 7.36 2.21 7.22 2.26 7.76 2.06
scenic 8.19 2.08 8.17 2.10 8.27 2.04
winter rec. 7.89 2.27 7.91 2.31 7.99 2.15
friendly 7.58 2.25 7.42 2.28 8.02 2.13
must visit 8.06 2.40 8.06 2.46 8.10 2.23
safe 7.34 2.26 7.27 2.30 7.38 2.18
nightlife 6.62 2.33 6.38 2.31 7.17 2.27
family place 7.82 2.22 7.88 2.23 7.73 2.17
popular place 7.03 2.36 6.99 2.40 7.15 2.24
historic 7.05 2.27 6.96 2.32 7.29 2.13
value 7.32 2.20 7.32 2.20 7.38 2.13

 

* : The full sample including 1996. 1997. 1998, 2001 and 2002, with which the study hypotheses are tested.

121

The independent samples t-test was conducted to test for the differences between
the Michigan images for the initial (1996-1997) and later (2001-2002) study periods. The
results of these analyses are presented in Table 24. The cases involved in these analyses
comprise the image-respondent group, which includes those respondents who provided
valid responses for all the 15 image items.

Table 24. Results of the independent samples t-test on the differences between the initial
(1996-1997) and later (2001-2002) years of the study.

Levene's Test for

 

 

 

Equality of Variances t-test for Equality of Means

Sig. Mean Std. Error

F Si& t (11' (Z-tailed) Difference Difference

getaway .084 .772 4.217 4174 .000‘I -0.35 0.0822
museums 8.522 .004’I -l4.437 4174 .000’I 1.18 0.0817
summer rec. .004 .950 1.294 4174 .196 -0.089 0.0688
exciting 17.349 .000“ -3.289 4174 .001‘I 0.24 0.0738
lodging 25.206 .0003 -7.821 4174 .000'I 0.54 0.0695
scenic 2.784 .095 - l .468 4174 .142 0.0972 0.0662
winter rec. 17.646 .000“ -1.217 4174 .224 0.0872 0.0716
friendly 24.664 .0003 -8.428 4174 .000'I 0.60 0.0709
must visit 25.497 .000“ -.578 4174 .563 0.0437 0.0757
safe 2.909 .088 -1.590 4174 .112 0.11 0.0718
nightlife 2.417 .120 -10.776 4174 _000. 0.79 0.0730
family place .597 .440 2.027 4174 .043b -0. 14 0.0703
popular place 8.450 .0048 -2. 191 4174 .029b 0.16 0.0746
historic 17.703 .000‘I -4.615 4174 .000‘I 0.33 0.0716
value 2.377 .123 -.884 4174 .377 0.0611 0.0691

 

' : Statistically signiﬁcant at 99% level (p-*- 0.01).
b : Statistically signiﬁcant at 95% level (p<0.05).

As can be seen from Table 23, Michigan holds a positive image across all data
periods, having mean scores above six for all image attributes on a 10-point scale. For the
all-years’ data (1996, 1997, 1998, 2001, 2002), the lowest mean score is on the
‘museums’ attribute (6.49), and the highest mean score is on the ‘scenic’ attribute (8.19).
The attribute with the highest variation in scores is the ‘getaway’ attribute with a standard
deviation of 2.63, closely followed by the ‘museums’ attribute (2.62). The lowest

standard deviation (2.08) is associated with the ‘scenic’ attribute.

122

E
the initia
differenc'
getaway
deviation
attribute
standard

T.

of Michig

 

attribute l
item is th
the highe
highest st
the ‘muse

attnbme (

1n
hOWEVer,
\mUSeum S
during 1hr
cOmllared
Michigan.
ObSew e d

Examining mean ratings and standard deviations in ratings of each attribute for
the initial (1996-1997) and later data periods (2001-2002) reveals some similarities and
differences as can be seen in Table 23. In the 1996-1997 data, the strongest attribute is
‘getaway’ (8.24) and the weakest attribute is ‘museums’ (6.14). The highest standard
deviation belongs to the ‘museums’ attribute (263), closely followed by the ‘getaway’
attribute (2.62). The lowest variation is still in the ‘scenic’ attribute with the lowest
standard deviation (2.10).

There is a slight change in the most popular and the least popular image attributes
of Michigan in the 2001-2002 data. The most popular image attribute is still the ‘scenic’
attribute with the highest mean score of them all (8.27); however, the least popular image
item is the ‘popular place’ attribute with a mean score of 7.15. The image attributes with
the highest and lowest score variation are the same as those for the all-year data: the
highest standard deviation belongs to the ‘getaway’ attribute (2.52) closely followed by
the ‘museums’ attribute (2.46), and the lowest standard deviation belongs to the ‘scenic’
attribute (2.04).

In general, Michigan has been perceived strong in terms of scenic appeal;
however, perceptions of its weaknesses have changed. The weakest image attribute was
‘museums’ in the initial period of the study, while the ‘popular place’ attribute ranks last
during the later period. As can be seen in Table 24, when the mean image scores are
compared between the initial (1996-1997) and later (2001-2002) periods, it seems that
Michigan’s image has generally improved over the 5-year period. An increase is
observed in the majority of the image attributes in 2001-2002 data except for the

‘getaway’, ‘summer rec.’, and ‘family place’ attributes. The difference between initial

123

and l:
‘geta\
‘popu
stands
‘frienc
hﬁchu

respor

Comp

increas
year pe
SUbgrol
data. “I
represe;
COmpar.
Statisric‘.
and Stan

indepenc

I
{-

Michlgar

715 fOr

and later study periods is statistically signiﬁcant for nine out of 15 attributes; namely,
‘getaway’(-), ‘museums’, ‘exciting’, ‘lodging’, ‘ﬁiendly’, ‘nightlife’, ‘family p1ace’(-),
‘popular place’, and ‘historic’. Also, statistically signiﬁcant decreases are observed in the
standard deviation of eight attributes: ‘museums’, ‘exciting’, ‘lodging’, ‘winter rec.’,
‘ﬁiendly’, ‘must visit’, ‘popular place’, and ‘historic’. This is an indicator that
Michigan’s image of these attributes is stabilizing and crystallizing in the minds of

respondents.

Comparison of Visitors" and Non-visitors’2 Images

It was suspected that the improvement in Michigan’s image could be due to the
increasing portion of visitors in the image-respondent subgroup of the sample over the 5-
year period. As was presented in Table 15, the portion of visitors in the image-respondent
subgroup is 75.3% in the 1996-1997 data, while it increased to 79.3% in the 2001-2002
data. To see if the improvement in image could be due to the increase in the visitor
representation in the image-respondent group, visitors and non-visitors were also
compared by using the independent samples t-test. This was performed to see if there is a
statistically signiﬁcant difference between their perceptions of Michigan. Mean scores
and standard deviations for these two groups are presented in Table 25. The results of the
independent samples t-test are presented in Table 26.

As can be seen in Table 25, the most popular and least popular image attributes of
Michigan are the same for both groups, ‘scenic’ being the most popular (8.54 for visitors,

7.15 for non-visitors) and ‘museums’ being the least popular (6.82 for visitors, 5.55 for

 

l I All three types of visitors who provided valid responses for all 15 image items.
.,
“2 Those who never visited Michigan before and provided valid responses for all 15 image items.

124

  
  
  

non-visit
and stan-

Table .5

 

Item
getatx‘ay
museums
armmcr r.
excrUng
lodging
sceruc
“inter rec
friendly
must \isir
sale
nightlife
family pla
pepular pl'
historic
\alue

 

Table 2:”;

 

non-visitors). However, there are noticeable differences in terms of mean image scores
and standard deviations in scores between these groups.

Table 25. Image item descriptive statistics for visitors and non-visitors.
Visitors (n=3,968) Non-visitors (n=1,262)

 

 

 

 

 

Item Mean 1 Std. Dev. Mean I Std. Dev.
getaway 8.50 2.35 6.92 3.04
museums 6.82 2.55 5.55 2.56
summer rec. 8.45 1.95 6.98 2.40
exciting 7.51 2.16 6.04 2.44
lodging 7.69 2.05 6.38 2.39
scenic 8.54 1.83 7.15 2.41
winter rec. 8.19 2.10 7.01 2.50
friendly 7.92 2.05 6.63 2.50
must visit 8.43 2.16 7.01 2.71
safe 7.66 2.07 6.38 2.49
nightlife 6.83 2.28 6.00 2.37
family place 8.21 1.97 6.59 2.47
popular place 7.39 2.20 5.92 2.50
historic 7.37 2.15 6.10 2.33
value 7.67 ' 2.02 6.26 2.37

 

Table 26. The results of the independent samples t-test on the differences between
visitors and non-visitors.

 

 

 

Levene's Test for t-test for Equality of Means
Equality of Variances

F Sig. t df Sig. Mean Std. Error

(2-tailed) Difference Difference

getaway 273.671 .000" 19.332 5228 .000‘1 1.58 0.0819
museums 5.866 .015b 15.332 5228 .000‘ 1.27 0.0826
summer rec. 92.159 .000' 22.028 5228 .000‘I 1.47 0.0667
exciting 25.067 .000“ 20.507 5228 .000" 1.48 0.0722
lodging 64.414 .000‘I 19.051 5228 .000“ 1.31 0.0689
scenic 154.039 .000" 21.616 5228 .000‘I 1.39 0.0642
winter rec. 72.624 .000“ 16.483 5228 .000" 1.18 0.0713
friendly 1 14.077 .000‘I 18.446 5228 .000“ 1.29 0.0700
must visit 158.495 .0008 18.983 5228 .000“ 1.41 0.0745
safe 95.840 .000“ 18.135 5228 .000“ 1.28 0.0705
nightlife 1.132 .287 11.057 5228 .0008 0.82 0.0742
family place 131.221 .000" 23.786 5228 .000“ 1.62 0.0680
popular place 39.023 .000‘’ 20.103 5228 .000" 1.48 0.0734
historic 7.216 .007‘I 17.895 5228 .000" 1.27 0.0710
value 69.148 .0008 20.724 5228 .0008 1.41 0.0682

 

a : Statistically signiﬁcant at 99% level (p< 0.01).
b : Statistically significant at 95% level (p< 0.05).

As can be seen in Table 26, visitors’ mean image scores for all attributes are over

one point higher than those of non-visitors except for the ‘nightlife’ attribute and

125

diﬁerenc
and non-
(1) the
unfamilia
better im
A

visitors‘ i

 

attribute.
rather th;
biggest l
Signiﬁcan
‘scenic‘ ;
the ‘nigl
Statistical
Ir
i“Cluding
Protjded
thOSe Of
C1053,- 10
TCV‘eals
‘mghtlife

differences in all attributes are statistically signiﬁcant. The differences between visitors’
and non-visitors’ ratings of the set of image attributes could be interpreted in two ways:
(1) the perception of these attributes by non-visitors is inaccurate because they are
unfamiliar with Michigan (2) visitors come to Michigan because they already have a
better image of it as a desirable pleasure trip destination.

Also, visitors’ image score standard deviations are signiﬁcantly smaller than non-
visitors’ image score standard deviations for all image attributes except for the ‘nightlife’
attribute, possibly because their image of Michigan is derived from personal experience
rather than secondary information sources. The differences in standard deviations are
biggest for the ‘getaway’ and ‘scenic’ attributes and smallest and not statistically
signiﬁcant for the ‘nightlife’ attribute; meaning, visitation crystallized the ‘getaway’ and
‘scenic’ attributes more than the other attributes while this stabilization did not occur in
the ‘nightlife’ attribute since the difference between the variances is not big or
statistically signiﬁcant.

In Table 27, the differences between the images of all visitors and all non-visitors,
including those who gave invalid responses for some image measurement items are
provided. These all visitors’ and all non-visitors’ responses are not too different from
those of visitors and non-visitors who provided valid responses for all image items. A
closer look at the number of respondents who provided valid responses for each item
reveals that the number of respondents drops dramatically on the ‘museums’ and
‘nightlife’ attributes in both groups. This shows that there is a lack of knowledge and

perception on these aspects of Michigan among non-visitors as well as visitors.

126

 

 

 

getaxx :13
1111155111115

summer
exciting
lodging
scenic
Winter 1
friendl}
must \i
safe
nighm
{8.11111}
POW):
histori

in a
due

0V1

In

t}

it

Table 27. Image item descriptives for all visitors and all non-visitors.
Visitors (n=5,550.6.971) Non-visitors (o=l,936-2,95g

 

 

 

Item I n I Mean I Std. Dev. n Mean Std. Dev.
getaway 6946 8.33 2.51 2951 6.79 3.16
museums 5613 6.96 2.52 1936 5.80 2.66
summer rec. 6872 8.39 1.96 2645 7.13 2.37
exciting 6787 7.42 2.20 2448 6.08 2.47
lodging 6035 7.76 2.02 2027 6.69 2.35
scenic 6884 8.46 1.86 2579 7.37 2.29
winter rec. 6313 8.17 2.12 2426 7.22 2.44
friendly 6678 7.91 2.07 2359 6.94 2.44
must visit 6971 8.35 2.22 2732 7.22 2.67
safe 6596 7.65 2.11 2371 6.60 2.45
nightlife 5550 6.88 2.30 1976 6.19 2.38
family place 6848 8.15 2.02 2531 6.82 2.39
popular place 6315 7.37 2.23 2310 5.99 2.53
historic 6086 7.37 2.16 2062 6.27 2.34
value 6303 7.65 2.04 2078 6.41 2.37

 

Since visitors’ image of Michigan is signiﬁcantly better than that of non-visitors
in all attributes, it can be concluded that the improvement in Michigan’s image could be
due to the increasing portion of visitors in the image-respondent subgroup of the sample

over the 5-year period.

Image Item Correlations

The correlation matrix for image items is provided in Table 28. As can be seen in
the table, there are moderately high correlations (between 0.20-0.70) between image
items. High correlations are desired since the few factors extracted through factor

analysis should be providing maximum explanation of the original variables.

127

 

 

 

1.151%} 5:

Results

meanin
(XﬁChlé
Was us:
in 0rde
Since tr.
Variable
Ek'tracte
fOTCing
300d Ct

loading):

factor e

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Table 28. Ima :e item correlation matrix.
Pearson Corr. 1

Lgetaway 1.000 2

2-museums 2871.000 3

3-summer rec. .475 3881.000 4

4-exciting .364 .508 6491.000 5
-lodging .323 .534 .560 .664 1.000 6

[Io-scenic .388 .395 .661 .609 6061000 7

II7-winter rec. .342 .337 .525 .480 .470 5601.000 8

[IS-friendly .303 .436 .516 .579 .580 .554 .526 1.000 9
-must visit .325 .369 .520 .580 .496 .537 .456 .549 1.000 10

lO-safe .342 .307 .522 .548 .477 .528 .451 .574 .551 1.000 11

ll-nightlife .214 .516 .404 .568 .583 .416 .391 .509 .460 4021.000 12

lZ-familyplace .403 .400 .679 .672 .591 .674 .557 .610 .618 .641 4731.000 13
l3-popularplace .336 .442 .568 .656 .590 .583 .493 .551 .548 .533 .493 7051.000 14
l4-historic .305 .588 .510 .624 .601 .559 .473 .534 .529 .466 .548 .601 .623 1.000 15
lS-value .383 .435 .601 .640 .581 .596 .495 .582 .546 .587 .500 .666 .621 .641 1.000

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Listwise deleted n=5,485

Results of Exploratory Factor Analyses

Exploratory factor analysis was applied on 15 image items to derive fewer
meaningful and uncorrelated image factors to be used as the dependent variables
(Michigan’s image) in the multiple regression analyses. Principal component analysis
was used as the initial factor extraction method to extract uncorrelated factors organized
in order of decreasing variances. Factors with Eigenvalues-exceeding-one were kept
since those factors represent the variance equal or greater than that of the average original
variable. Then, these factors were rotated using the Varimax rotation method. Factors
extracted with this method are more meaningful since items are rotated orthogonally,
forcing them to approach the limits of 0 and :1; variables with loadings closer to 1 have
good correlation to the factor on which they loaded. Eventually, variables with substantial
loadings, equal to or greater than 0.5, were used to represent the factors.

The procedure explained above was applied to the all-years’ data ﬁrst. Initial

factor extraction with the Eigenvalues—exceeding-one criterion resulted in the extraction

128

of two factors. These two factors were rotated using the Varimax method. To validate the
results of the all-years’ data factor analysis, the same procedures were applied to the two
different data periods, 1996-1997 and 2001-2002. The results of these factor analyses are
presented in Table 29.

Table 29. Factor analysis results for different data periods.

 

 

 

 

 

All-years“ (n=5,485) 1996-1997 (n=2,580) 2001-2002 (n=1,597)
Dimensions Factor Loadings Dimensions Factor Loadings Dimensions Factor Loadings
F1 F2 F1 F1 F2
summer rec. 0.764 0.306 family place 0.870 summer rec. 0.767 0.258
family place 0.761 0.415 exciting 0.841 getaway 0.756 -0.129
scenic 0.716 0.376 value 0.807 family place 0.723 0.384
safe 0.696 0.294 scenic 0.804 safe 0.715 0.233
getaway 0.657 0.011 lodging 0.801 scenic 0.607 0.387
winter rec. 0.636 0.304 popular place 0.794 winter rec. 0.601 0.317
value 0.629 0504 summer rec. 0.789 popular place 0.554 0.543
must visit 0.603 0,421 historic 0.788 must visit 0.551 0.437
popular place 0.586 0.537 friendly 0.784 value 0.596 0.475
friendly 0.551 0.525 must visit 0.740
safe 0.706 nightlife 0.100 0.803
nightlife 0.196 0.793 winter rec. 0.689 museums 0.065 0.791
museums 0.124 0.791 nightlife 0.687 lodging 0.313 0.722
historic 0.402 0.722 museums 0.629 historic 0.383 0.673
lodging 0.439 0.687 getaway 0.519 exciting 0.537 0.579
exciting 0.567 0.613 friendly 0.402 0.547
Rotation converged in 3 iterations Only 1 component was extracted. Rotation converged in 3 iterations.
Cumulative ”A; of Variance Explained: The solution cannot be rotated. Cumulative 9'0 of Variance Explained:
62.23 Cumulative "/o of Variance Explained: 57.00 I5>_8.15

 

 

* : The full sample including 1996. 1997. 1998. 2001 and 2002. with which the study hypotheses are tested.
Extraction Method: Principal Component Analysis.

Rotation Method: Varimax with Kaiser Normalization.

Items ordered by the size of factor loadings.

As can be seen in Table 29, the factor analysis results of the all-years’ data and
the 2001-2002 data revealed two factors while factor analysis of the 1996-1997 data
revealed only one factor with all dimensions loading onto one factor with substantial
loadings (greater than 0.50). The cumulative variance explained by the factors extracted
is the smallest for the 1996-1997 data period (57.00%) while that of the all-years’ data is

the highest (62.23%).

129

The results of the all-years’ data are similar to those of the 2001-2002 data with
slight differences in dimensions’ factor loading sizes and directions. In the all-years’ data,
there are 4 ﬂoater (cross-loader) dimensions, loading highly, with loadings greater than
0.50, onto both factors: ‘value’, ‘popular place’, ‘friendly’, and ‘exciting’ dimensions.
However, factor loadings of ‘value’, ‘popular place’, and ‘ﬁiendly’ dimensions are higher
on Factor I while the factor loading of ‘exciting’ dimension is higher on Factor 11. Two of
these dimensions, ‘popular place’ and ‘exciting’, are also ﬂoaters in the 2001-2002 data.
The ‘popular place’ dimension loads higher onto Factor I while ‘ﬁiendly’ dimension
loads higher onto Factor 11 in this period.

The differences in the number of factors can be explained by the fact that the
percentage of visitors in the sample was higher in the later-years of the study. Since
visitors were found to have a differentiated image of the destination, their increasing
representation in the image measurement during later-years might have caused the results
of the all-years’ data to be differentiated too. To check if this explanation holds true, the
same factor analysis procedures were applied to visitors (all three types of visitors who
provided valid responses for all 15 image items) and non-visitors (who never visited
Michigan and provided valid responses for all 15 image items) separately. The results of
these analyses are presented in Table 30.

As can be seen in Table 30, factor analysis results of the non-visitor group are
similar to those of the 1996-1997 data, one factor with all dimensions loading onto one
factor with substantial loadings (greater than 0.50) except for the ‘getaway’ dimension
(0.40). Likewise, factor analysis results of the visitor group are similar to those of the all-

years’ data and especially to the 2001-2002 data. two factors with substantial dimension

130

 

loading

 

for the
respect
loaders
while 1
above e

Table 31

&

 

Dimen

 

family p
Summer

value

getaway
Popular 1
musr V151
“inter re

nightlife
thseurng
h1510nc
lOdgtng
exciting
fnendh‘
ROQI‘C‘D cor
CUmULun e

Gillan XI;
ham order:

S
image an
“Weary
ﬁlrthEr a1

Visual rep

loadings (bigger than 0.50). The cumulative variance explained by the factors extracted
for the visitor group is larger than that of the non—visitor group (58.02 and 56.78%,
respectively). The ‘popular place’, ‘exciting’, and ‘ﬁiendly’ dimensions are also cross-
loaders in the visitor group factor analysis results, the ﬁrst loading higher onto Factor I,
while the latter two loading higher onto Factor 11. Thus, it can be concluded that the
above explanation is highly likely to be true.

Table 30. Factor analysis results for visitor and non-visitor groups.

 

 

 

 

Visitors (n=3,968) Non-visitors (n=l,262)
Dimensions I Factor Loadings Dimensions I Factor Loadings;
F1 F2 F1
family place 0.771 0.360 family place 0.842
summer rec. 0.757 0.242 exciting 0.830
safe 0.700 0.207 value 0.812
scenic 0.697 0.333 historic 0.806
value 0.630 0.447 lodging 0.805
getaway 0.622 -0.018 scenic 0.788
popular place 0.590 0.501 popular place 0.787
must visit 0.589 0.382 summer rec. 0.782
winter rec. 0.582 0.287 friendly 0.771
nightlife 0.753
nightlife 0.150 0.792 safe 0.745
museums 0.081 0.789 must visit 0.725
historic 0.381 0.700 winter rec. 0.697
lodging 0.395 0.684 museums 0.646
exciting 0.567 0.580 getaway 0.408
friendly 0.51 1 0.512

 

Rotation converged in 3 iterations.
Cumulative 96 of Variance Explained: 58.02

Only 1 component was extracted.
The solution cannot be rotated
Cumulative % of Variance Explained: 56.78

 

 

Extraction Method: Principal Component Analysis.
Rotation Method: Varimax with Kaiser Normalization
Items ordered by the size of loadings.

Since the later-years’ data represent the current situation in terms of Michigan’s
image and since the factor analysis results for the later-years’ data resemble those of the
all-years’ data, the factor analysis results for the all-years’ data were adapted for the
further analyses in this study. The .ﬁnal factor results are presented in Table 31 and a

visual representation of the resulting 2-Factor Model is presented in Figure 10.

131

As ca'
of all 15 ima;
onto Factor
the internal
substantially
Table 31 S

Image I

Factor 1: TI
surnrne
fami 1}
scenic
safe
getawa
uinter
value
must \
p0pulz
friend

Factor [1:
mghtl
muscr
histor
lodgil
eXClll

imam .\1.

Rankin Met

Cullen 00m
hum Ordered

they are
dimenSiOI

Sc

attribmes

the phB-‘si.

As can be seen in Table 31, two factors were extracted with substantial loadings
of all 15 image dimensions, 10 dimensions loaded onto Factor I, and 5 dimensions loaded
onto Factor 11. The factors explain 62.23% of the original variables. The computation of
the internal stability Cronbach alpha coefficient indicates that both factors are stable with
substantially high internal consistencies, a=0.91 for Factor I and (1:087 for Factor 11.

Table 31. Summary of factor analysis results.

 

 

 

 

 

 

 

Factor % of Cumulative 0/o Factor Factor

Image Factors & Dimensions Loadings Variance of Variance Grand Alpha

Explained Explained Mean Value

Factor I: The Setting/Sense of Place 34.19 34.19 7.75 0.91
summer rec. 0.764
family place 0.761
scenic 0.716
safe 0.696
getaway 0.657
winter rec. 0.636
value 0.629
must visit 0.603
popular place 0.586
friendly 0.551
Factor I]: Activities/Things to do 28.04 62.23 6.93 0.87

nightlife 0.793
museums 0.791
historic 0.722
lodging 0.687
excitig 0.613

 

Extraction Method: Principal Component Analysis.
Rotation Method: Varimax with Kaiser Normalization.
Rotation converged in 3 iterations.

[tents ordered by the size of loadings.

Although individual image dimensions have good correlation with these factors,
they are not readily interpretable, especially Factor I. A close examination of the
dimensions. included in Factor 1 reveals that this factor is a collection of conceptually
different image attributes.

Some dimensions, namely, ‘summer rec.’, ‘winter rec.’, and ‘scenic’ are cognitive
attributes of image that deal with the factual knowledge of a destination, usually about

the physical appearance. On the other hand, other dimensions, such as ‘family place’,

132

 

 

.0662 LeuowkiN {EN

EIIQ‘QIIE

.. .Q—
$425.3 £>

BEER

  

 

ow 9 mwcik
\8E>co<

:3 u 2%..

EH!
.7

     
 

Snag:
.3 «we...—

BE

  

A39: owes: :meoaz cowown—éBH d— oSwE

EH

  

08E mo omcom
\mszom 2:

36 122.4

.352 heaven—1N «4mm

   
   
 

2...:
2.8me

one... 2223..
uua... £55
e__.._w_=

33

an? «2::
2:52.:

.99.. SE?
25%

Mia—5.
“5:38

.8; SEE—a
2552::
.3359“

5553

um—>
“e;
um?»
HNS,
H:>
5:,

ne>
nw>
nb>
ue>
nm>
35
un>
u~>
"Sr

133

‘safe’, ‘must visit’, ‘popular place’, and ‘friendly’ are affective attributes of image that
deal with the feelings and attitudes towards a destination.

The remaining two dimensions, ‘value’ and ‘getaway’ can go in either category.
Michigan is a close by destination for the respondents in this study who are residents of
the surrounding states and a province of Canada. Thus, these two dimensions would
normally be counted as a cognitive component. However, the responses would be highly
dependent on a particular respondent’s ability to afford a vacation. For a low-income
respondent who cannot afford a vacation, no matter how close Michigan actually is, the
response to this dimension conveys one’s ﬁnancial circumstances rather than the actual
knowledge of the physical distance or monetary cost of a vacation while considering
Michigan.

A holistic look at all of these dimensions reveals that there is a relationship
between these seemingly unrelated dimensions. They reﬂect the dimensions of a vacation
destination that is available, affordable, comfortable and functional; the respondents
know it, they visit it every now and then, for whatever reason, and they feel good about
it. Thus, this factor was named as ‘The setting/Sense of Place’. The grand mean of this
factor is 7.75, a rather high score, which corresponds to a high level of agreement on the
10-point scale, where 10 means “agree completely”.

The same mixed pattern exists in Factor 11; nevertheless, its interpretation is
easier than it is for Factor 1. Four dimensions, ‘nightlife’, ‘museums’, ‘historic’, and
‘lodging’ are cognitive components of image that deal with the physical attributes of a
destination while the ‘exciting’ dimension is an affective component of image comprising

the feelings or attitudes towards a destination. Again, a holistic perspective on these

134

 

dimensions
Therefore.
factor is 69
agreement.
“agree con
travel purpl
good as the
The
tourist dest
both in sun
Ohio or the
on Michiga

MiChigan.

Since
Wage of Mi
are USed as
For this pm];
regression fa
enraged, 1111
for the COYTe

ofthe ratings

dimensions reveals that these dimensions relate to what people do on pleasure trips.
Therefore, this factor was dubbed as ‘Activities/Things to do’. The grand mean of this
factor is 6.93, an above the average. score, which corresponds to a relatively high level of
agreement, but not as high as that of Factor I, on the 10-point scale, where 10 means
“agree completely”. Michigan is perceived as a close by destination good for different
travel purposes; however, its touristic activity amenities are perceived as good, but not as
good as the setting itself.

These ﬁndings appear to be commensurate with the reality of Michigan as a
tourist destination; it is a destination with ample natural resources for outdoor recreation
both in summer and winter, but it lacks big touristic attractions, such as Cedar Point of
Ohio or the Amish Community in Pennsylvania. Thus, it can be concluded that the data
on Michigan’s image are robust and represent a realistic measure of the image of

Michigan.

The Regression Models

Since the two image factors extracted by exploratory factor analysis represent the
image of Michigan, these two factors as well as the Overall Image (mean image score)
are used as the dependent variables in the two multiple regression models of this study.
For this purpose, factor scores of each factor were computed for respondents by using the
regression factor scoring function of SPSS 10.0. With this function, as soon as factors are
extracted, the scores for the respondents are also calculated and saved as new variables
for the corresponding factors. The Overall Image was extracted by computing the mean

of the ratings of 15 image items. Thus, these new variables contain the calculated factor

135

 

scores and
responses 1
variables it

population

General P1
The
the general
age, overal
included in
frequencies
With slight
$41116 as th.
between 0 E
Value of 1 .
The
respondemg
fur the gem
reSpondems
of the mode
of with e res
569/6 of [ht

rep resenIatio

scores and averaged image scores for the image-respondent subgroup. The calculated
responses of the two factors as well as the averaged image scores and the independent
variables were used as input in the multiple regression analyses for testing the general

population model and the recent visitor segment model of this study.

General Population Model Variables

The descriptive statistics and the correlation matrix of all the variables included in
the general respondents model are presented in Tables 32 and 33, respectively. Except for
age, overall travel experience, two image factors, and the Overall Image, all variables
included in the model are dummy variables with the values of O and 1. In Table 32,
frequencies are also provided for these dummy variables. As can be seen in the table,
with slight decimal differences (due to rounding in mean scores), frequencies are the
same as the mean values for these dummy variables since the mean is the middle point
between 0 and l, which equates to the weight of the value of 1, thus the percentage of the
value of 1.

The general population model respondents are considerably different from all
respondents in the all-years’ data in terms of some demographic variables. The mean age
for the general population model group is 41.40; this is slightly younger than that of all
respondents as was discussed in the sample proﬁles section of this chapter. The majority
of the model respondents are'white (88.8%); this is about 5% higher than the percentage
of white respondents in the all respondents group. Female respondents represent about
56% of the general population model respondents, which is about a 5% lower

representation than that of all respondents. Respondents with above-the-median income

136

 

make up ll
representati

Table 32. G:

(n
==

1

Inde
Age
Race-“ﬂute“
Gender-Fem:
Income-Abe

State-Illinois
State-h-Irchig
State-Ohio ‘

State- Wisco-
State-Oman
Overall [raw

V’isilalion-\

Survey $635
5““‘61‘ Sean

Del
The Settin;
Am‘iiliesr
‘ — The Vang
I 1 COded uh,
: Coded fen
.» C Ned m
’ Coded 1111
1 cm Xi:

uihrin'f

follow-e
Page“.
r"T'SIDEC
Summ
thus‘

”he.

make up the majority of the model respondents (80.7%), which is about a 14% higher

representation than that of all respondents.

Table 32. General population model variables’ descriptives and frequencies for the all-years’ data

 

 

 

(n=3,554).
Name Description Mean Std. Dev. Frequency
(% of code Q
Independent Variables
Age Age of the respondents 41.40 13.52 -
Race-White‘l White respondents 0.89 0.31 88.8
Gender-Female" Female respondents 0.56 0.50 55.8
Income-Above-the-medianc Respondents with above-the-median 0.81 0.39 80.7
income
State-Illinoisd Respondents from Illinois 0.18 0.38 17.9
State-Michigan‘ Respondents from Michigan 0.36 0.48 35.6
State-Ohio ‘ Respondents from Ohio 0.18 0.39 18.3
State-Wisconsing Respondents from Wisconsin 0.09 0.29 9.3
State-Ontarioh Respondents from Ontario 0.09 0.29 9.4
Overall travel experience Number of pleasure trips to any 5.24 5.70 -
destination in the past 12 Ms

Visitation-Visitor' Respondents who visited Michigan at 0.79 0.41 79.0

. any time -
Survey season-Winter’ Survey conducted in winter 0.23 0.42 23.4
Survey season-Spring" Survey conducted in spring 0.22 0.42 22.4
Survey season-SummerJ Survey conducted in sununer 0.24 0.43 23.7

Dependent Variables .

The Setting/Sense of Place Factor score for Factor 1 0.038 0.977 -
Activities/Things to do Factor score for Factor 2 -0.021 0.985 -
Overall Image Mean image score 7.498 1.615 -
- : The variable is continuous, fi'equency of each value is not given. 3 : Coded Wisconsin respondents = I; all others = 0.
“: Coded white respondents = 1; all others = 0. h : Coded Ontario respondents = 1; all others = 0.
b : Coded female respondents = 1; all others = 0. I : Coded visitor respondents = 1; all others = 0.
c : Coded above-median income respondents = 1; all others = 0. J : Coded winter surveys = I; all others = 0.
d : Coded Illinois respondents = 1; all others = 0. k : Coded spring surveys = 1'. all others = 0.
: : Coded Michigan respondents = I; all others = 0. l : Coded sununer surveys = I; all others = 0.

:CodedOhiorespondents =1:allothers= 0.

Michigan respondents represent the biggest percentage of all states (35.6%),
followed by Ohio and Illinois respondents (18.3% and 17.9%, respectively). The
percentages of Wisconsin and Ontario respondents are almost equal (9.3% and 9.4%,
respectively). These percentages are quite different ﬁom those of all respondents as was
summarized in Table 15, Michigan being about double as much as that of all respondents,
thus, reducing the representation of other states’ residents in the general population

model group.

137

The mean for the overall travel experience variable, meaning the mean number of
trips taken within the past 12 moths, is 5.24. Respondents who visited Michigan at some
point in their lives constitute the majority of the general model respondents, with a
percentage of 79.0, which is about 30% higher than that of all respondents described in
Table 15. As was mentioned before, visitors are more likely to ﬁnish the survey with
valid responses for all questions. Finally, the percentage of respondents surveyed in
winter, spring, and summer (included in the model as dummy variables) are 23, 22, and
24, respectively, an almost equal distribution between the four seasons.

As can be seen in the correlation matrix presented in Table 33, there is no
substantial correlation among the independent variables, which means that
multicollinearity is not likely a problem in the multiple regression analysis.

Table 33. General population model variables’ correlations.

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

[Pearson Correlation
dependent Variables l
l-Age 1.000 2
Ib-Race-White 0921.000 3
Ib-Gendcr-Female .042 .003 1.000 4
II4-Income-Above-the-median -035 .031 -.0681.000 5
IIS-State-lllinois -009 -010 .011 .038 1.000 6
[b-Smm-Miclogg .046 -021 .083 .050 -.3471.000 7
IP-State-Ohio .012 .040 -.018 .007 -221 -353 1.000 8
IE-State-Wisconsin -016 .049 -034 ..019 -.150 -.238 -.1521.000 9
9-State-Ontario -.063 -093 -054 .039 ..151 -240 ..153 -.1031.000 10
IO-Overall travel experience -010 .034 -.038 .034 -013 -039 -.058 .103 .010 1.000 11
ll-Visitation-Visitor .104 .085 -001 .033 -092 .294 -.062 -.058 -.182 .075 1.000 12
12-Surveyseason-Winter .053 .014 .007 -.006 -004 .047 -.026 .033 ..014 .029 .025 1.000 13
131-Survey season-Spring -023 -.016 .010 -002 -035 .024 .001 .019 -032 -.016 .022 -.2971.000 14
l4-SurveySeason-Summer .007 -012 .026 .027 .000 -020 .033 .009 .019 -.006 -039 -.308 -.3001.000
“Dependent Variables 15
IES-TheSetting’SenseofPlace .049 .135 .055 .023 -027 .248 -.098-.029-.212 .053 .336 .041 .059 0661.000 16
Ih6-Actlvitiesx'fhings to do .184 .006 .058 -070 -.107 .130 -032 -037 -004 .004 .158 -027 -050 -073 -.0561.000 17
“17- Overall Image .164 .104 .082 .031-.102 .288 .095-053 -.165 .043 .369 .013 .012 .000 .729 .641 1.000

 

 

 

 

 

 

 

 

 

 

 

 

Listwise deleted n=3,554.

Since the same multiple regression analysis was applied to the 1996-1997 and
2001-2002 data, the descriptives of all variables for these periods are also provided in

Table 34.

138

Table 34. General population model variables’ descriptive statistics for different data

 

 

 

Jeriods.
All-years“ Q1=3,554n996-1997 (n=1,990) 2001-2002 Q1=942)
Notation Variable Mean Std. Dev. Mean Std. Dev'. Mean Std. Dev.
Independent Variables
A Age 41.40 13.52 40.60 13.09 43.94 13.94
R Race-White 0.89 0.31 0.90 0.31 0.87 0.33
G Gender-Female 0.56 0.50 0.55 0.50 0.56 0.50
1 Income-Above-the-median 0.81 0.39 0.82 0.38 0.76 0.43
S-IL State-Illinois 0.18 0.38 0.19 0.40 0.13 0.34
S-MI State-Michigan 0.36 0.48 0.32 0.47 0.45 0.50
S-OH State-Ohio 0.18 0.39 0.19 0.39 0.16 0.37
S-WI State-Wisconsin 0.09 0.29 0.10 0.30 0.07 0.25
S-ON State-Ontario 0.09 0.29 0.10 0.30 0.10 0.30
OTE Overall travel experience 5.24 5.70 5.31 5.51 5.28 5.45
V Visitation-Visitor 0.79 0.41 0.78 0.42 0.85 0.36
SS-WI Survey season-Winter 0.23 0.42 0.27 0.44 0.19 0.39
SS-SP Survey season-Spring 0.22 0.42 0.25 0.43 0.15 0.36
SS-SU Survey Season-Summer 0.24 0.43 0.22 0.41 0.18 0.39
Dependent Variables

F1 The Setting/Sense of Place 0.038 0.977 0.106 0.963 -0. 169 1.004
F 11 Activities/Things to do -0.021 0.985 -0. 146 0.947 0.367 0.982
01 Overall Image 7.498 1.615 7.446 1.677 7.679 1.440

 

* : The full sample including 1996, 1997. 1998. 2001 and 2002, with which the study hypotheses are tested.

Means (frequencies for dummy variables) for the 1996-1997 data are more or less
similar to those of the all-years’ data. In the 2001-2002 data, however, there are some
noticeable differences, the biggest of which is in the respondents’ residence states. The
percentage of Michigan residents is 9% higher in the 2001-2002 data than that in the all-
years’ data. Consequently, the percentages of other states are 1% to 5% lower than those
of the all-years’ data. Also, the percentage of visitors is 6% higher than that of the all-
years’ data; and the percentages of survey seasons are 4% to 7% lower than those of the
all-years’ data. Finally, another difference in the 2001-2002 data is the means of the
scores of two image factors. In the all-years’ and the 1996-1997 data, mean of Factor I is
positive while that of Factor 11 is negative; the exact opposite case exists in the 2001-

2002 data. The Overall Image (mean image score) is highest for the 2001-2002 data and

139

lowest for the 1996-1997 data while the opposite is true for the standard deviation of the

Overall Image.

General Population Model Test Results
The multiple regression model deﬁned to investigate the relationship between
destination image, selected demographic variables, overall travel experience, prior

visitation and the season of the survey is as follows:

Image,- = Bo + 81A, + 82K + B361 + B41; + [358-114 + BsS-ML‘ + B7S-OH1 + BsS-WI;
+ BgS-ON, + BloOTE, + 311V, + BuSS-WL+ 313SS_SP,+ 314SS_SU;+ 8;

where,

Image,- = the image held by individual ,,

81.14 = coefﬁcients of independent variables l-14, and

s, = the error term for individual ,~

Since two image factors were extracted with the Exploratory Factor Analysis,
these two factors were used as the dependent variables in the multiple regression models

as well as the Overall Image (mean image score). Thus, the three regression models are:

FL '2 30 + [31A,- + BzRi + [336, + B41, + BsS-IL, + BsS-Nﬂ,‘ + B7S-OH1 + ﬁgs-WI; +
BgS-ON; + BmOTE, + 811V, + Buss-WIH' BI3SS_SP, + B]4SS_SU,-+ 8,-

F111: Bo + [31A,- + 32R, + 33G,“ + B41; + BsS-HJI + BaS-Nﬂ,‘ + B7S-OH,’ + BsS-WI, +
B9S-ON1'1' BloOTE1+ 811V, + Blsz-WIH' B]3SS_SP; + 314SS_SU1+ 8;

01; = Bo + 31A,- + 82R, + B36, + 341, + BsS-IL, + BsS-Nﬂ; + B7S-OH, + ﬁgs-WI, +
ﬁgs-ON; + 13100113,“? 311V; + Blsz-WIH‘ B]3SS_SP1+ B]4SS_SU,- + 8,

Ordinary least square regression was employed to determine the inﬂuence of

these selected variables on destination image. These multiple regression models were

estimated for both image factors and the Overall Image using the survey data of all-years

140

(1996,1997,1998,2001,2002); then, to validate the results, the same models were
estimated using the survey data for two different periods: 1996-1997 (initial two years of
the study) and 2001-2002 (later two years of the study). The data of 1998 are excluded
from the validation tests to differentiate the results from those of the all-years’ data as
much as possible. Estimation results for different data periods for Factor I (The Setting/
Sense of Place), Factor 11 (Activities/Things to do) and the Overall Image are provided in
Tables 35, 36 and 37, respectively.

In these tables, the “b” column provides the unstandardized coefﬁcients,
otherwise called estimates of the parameters for independent variables. A positive
estimate indicates a positive relationship between the Michigan image factor and each
independent variable. The “0’ column provides the standardized (beta) coefﬁcients,
which are used to make comparisons among independent variables in terms of the
magnitude of the effect on the dependent variable. Since the results of the all-years’ data
are inclusive, B values of the all-years’ data will be discussed in the text unless the sign of
the [3 values is different for different periods or the size of the [3 values are markedly
different for different data periods.

The “t” column provides the t-statistic, which is used to determine if a coefﬁcient
is signiﬁcantly different from zero in the statistical sense. The “a” column provides the
signiﬁcance level (p-value) for the two-way test of a coefficient being different from
zero. Finally, the “T” column presents the tolerance statistic for each variable, a statistic
that is used to check the extent of linear relationships among the independent variables

(multicollinearity).

141

.88.»... .26. .38 .. E885... 3.8.3.3» U.
.. adv... .63. :8 a 3885.. 2.8.5.3... . .
30. 2.. 0359.... .8.. 2.. if. 5.3 .88 v5 .....N .82 .82 .08. 8.8.2.. 6.95. .3. 2F u _.

 

 

 

 

   

 

 

8... .8... .8.... w... 8... 8... .8. 8.. .R. 8... on... 2... E. .8... 8...... 8.. .8. 8m. .2m-mm..oe=sm-=83w6:8
8... .8... c8... 8.. 8... S... 8.. 2m. .8.- 2...- 48. 2.... ..Q. .8... EN... 8.. .8. 8m. .88. m......m-8m§...pc=m
8... .8... 2.... 8m. 8... 8m. .8. 8». S...- 8... mm... 8...- 8... .8... 8.... 8.. .4... .2. 22.-mm. 5.5%-=83. 5.8m
8... .8... 83 8.. 3.... 8m. 8... .8... .8... EN. 8... 4S. 8... .8... 8?... SN. 8.... 8... .>.§...>-8.§.m.>
88. :4. 8... 8... m8. 8... .8. P... a... .8. 8... 8... .8. .2... 32 8... 8... 8... .mho.88:85.6.§.=ed.6
.2. 8m. 8....- 3... 4:. 2..- a. .8... 24.9 2...- 8... 8..- ..m. .8... 83- 3..- 8... .8.- 20.88.25-688
8.. 88. m8.- .....- 9... .8.- ..R. .3. .3..- 8...- a... 8.... .mm. 8.. m8..- 8...- 8... 8...- ..3-w.=.m=8m.3-aem
24. 8”. 9.....- 8...- o... 2..- .3. .8... 8m...- 8..- 8... 42.- .3. .8... 8.....- 8...- a... 8..- 50.92.523
8.. .2. 8... S... 8.. m8. 2.. .8... 2.... 2.. 8... E. m... .8... :2 m8. m8. 8.. ..2-m.=..w...6.2-2sm
m8. ”8. 8... .8. 3.. 8... 8... 8.. z...- .2..- 8... 8..- ....4. 8... 2m.- 8... 8... o...- ...-m.m.o=....-2sm
.8. E... 3...- 8...- 8... 2...- 33. N2. 3... m8. 8... 8... +8. 8.. .R. 2... 8... 8... ...Ends-2.3.8538...
.8. .N... 2.... m8. 8... S... .8. .8... .8.. 8... 8,... .8. 8... 18...... m8. 8... 8... 6.6.8.6..-.880
m8. ..8. 8... 8... 8... a... 88. .8... .84 8.. 8... o... .8. .8... 8.... 8.. 8... 8.. 2.223-665.
88. SM. 8.. R... 8... 8... 8... 8m. 8.... 2... .8. .8. .8. .8. 8... 8... .8. 8... 2...»...

8... 8....- o... 8.... 8... 8.9 8.. E.- 8... 8......- .8. .8..- 85.80.
H a . .. new .. H a . .. A; .. H a . .. aw .. 8.8:;

8.....1... 8....NL $2.1”... 8.....1... 88.8”. :2“... 8.....16 ~88”. 8.81.. 8.68:.

8.8.28 8.2.82 .28. -_.< .8: 25h

 

 

 

.. $518.... + 518.... + 55.8.... + ..>=.. + ..89.... + .zc-m... + .53... + .8.-m... + .58... + 3.8... + .5. + 6... + i... + ..<_.. + ... 1.. .—
.32.... mo omcomhccﬁm 0.:-V _ .80....— hom 8.58. 88 ESE coca—smog .2250 .mm 033.

 

142

.89... .96. $8 .. ...8.-.8.. 2.8.2.3: ”

p

.. . ...cv... .96. 8o: 2. ...8.-.2»... 2.8.2.3: n.
2.8. a. 8652.... .8.. 2.. 6...: 5.3 .8..” ...a .8. .82 .82 .82 8.8.2.. 6.8.: ..a 2F . ..

 

 

 

 

8... ...8. 8....- ...m.- 2.... :8.- ..8. c... .8. ..8. 8... :8. 8.. .8... 8....- ..2- 2.... .8.- Sm-mm..6EE=m-=88m .95.:
8... ...8. 2...:- m8- ..8. :8.- ..8. 2:. 8m.- 2... 8... 88.- 82-. .8... .2...- R..- 3.... 8? ...m-mm. 8.8-888 .95.:
8:. .8... .8.... .2.- 8... 8m.- .8. 8... 2n.- 2... :8. 8...- ..m... ...8. 48.n- 8..- .4... 8..- ..B.::. 3.5%-8.82.6.5:
n8. .8. :2... 8... 8... .2. 8... ...8. 8: .2. 8... 8m. 2.... ...8. 83 .2. .8. :8. .>.§....>-8.§...>
:8. 48. E.- 2... :8. 2....- 88. 2.8. 42.. .8. 48. 8... m8. .8. 2... n8. 2.... 8... .Eo.8=ocd..x6.9§..ep5
.8. 8m. 2....- ..2..- 8.. 2..- a. ...N. 3.... ..8. .8. 2.. 2H. 8:. E. 2... 48. 8... 2022.85.22:
:8. 8m. 8m.- .8.- .2. 8...- 1:2... 8.. 8...- 8... 8... 8..- .2. 8... up..- :2..- 8... 2...- 22.-2588.368:
.8. 8m. :8.- 2....- ..2. 2..- ..... 8m. ...8..- ..2..- .8. 2....- ..... 2... 8....- .2..- 8... 8...- 80.22568:
.8. 8.. 8....- .E..- o... 42.- 5. 2.... :8. 8... 8... 8... 2m. :8. 8... 8... 8... 8... 22.282.22.25
8... 8.... .8.- ..8. 8.. ..N..- 8... ...8. m8...- 8..- .8. 8....- 24. .8... .8..- ..2..- 8... m2. 37228....68:
.8. 3.... .8.- .8.- ..8. 2....- ..8. ...8. 2....- S... 8... 2..- ..8. .8... 82m. 8...- 8... 9..- 2.885-656.8588...
m8. :8. ...N. 8... .8. ..8. 88. ..8. ...8.. E... .8. 8.. 88. .8... 2.... .8. 8... .2. .o.6..§o-,.-s..8o
m2... .8... 82 8... 8... :2. 8... .8. 8:- 8...- :8. 8..- .8. ..mm. 28.. 2... .8. 2....- 2.6.2368...
8.... .8... 2...... 8.. ...8. .8. m8. .8... N8: ..2. N8. 88. .8. .8... :8... 8.. .8. 2... 2.6....

.8. :2.- 8.. :8.- 8... 2.....- ..2. ...8.- 8... 8....- 8... 8m.- .8880.
H a . : 2.: .. H a . .. a: .. H a . .. 2.: .. 3.8.8.»

88..."... 28...". 22..."... 88.81.. 2.8.21. 88..."... 88..."... :82". 88..."... 2.268.).

"8388 82-82 23...? was... 2:?

 

 

 

.4 + 451%.... + 5:182: + 55.32: + ..>:: + .8.-Os: + .2612: + 23%.: + 25.4: + £22.: + 3%.: + .1: + 6...: + W5. + 3:: + e: u .2 WI

 

.8.. o. mm=E§22>uu<v = .82.... 8.. 33:2 .8. .358 scum—2...: 3.0.60 .0: 2......-

143

.89... .26. :8 a .555: 2.8.2.2: n ..
.59... .25. .38 a .5025: 2.3.2.5: . .
.33. a. «32.3... .2... 2.. 5...... a... .8..: ...a .8: .82 .82 ...8. 8.8.2.. 28.3. ..e 2.: u ..

 

 

 

 

 

8:. E. 8:. :2.. :2. ..2. :2... 2;. 2... 2... :2.. :8. :2. ..2. :8. :8. 8... .8. S:-::..:E.....:-..o::o:.25:
3:. .8... ..:..: 2... :2. 2:. :8. 8:. 8:.- :8.- :2.. 8.... SH 2:. .8. .8 :2.. 3... .m:-::. 88:28:02.95:
:2.. .8... :8: :8. .2. 2:. .8. ..8. 8:.- :8.- .8... 8.... .5. 8:. 8:. 2... 8... 8... ..3-::. 3.5%-8:32.92:
:2.. .8... :8... 3:. :2. 8:. ..8. .8... 8:2 :::. :8. :2.. :8. ...8. :82 2:. 8... S... .>..o..:.>-..o......:.>
:2.. ..::. 8:... .8. :8. 8... 2.8. :2. :8... :8. 2.... :o... :8. 8:. 2:. :8. 2.... :o... .Eo.85..85.8§..296
.::. :2. £2- :2..- 8:. 8:- :::. .8... :8:- 2..,- :2. 8.....- 8:. .8... :3- .o..- 2.. :::.- .zo-:.o_.§=o-2:.:
:8... 8:. :2.- :8.- :::. 8..- :::. .82.. ..8.:- :8.- ..:.. ..:.- .::. .2... ::.<:- :2... :... 8::- 23-252.823.23:
:8. .2. 8:..- :8.- ::.. E.- ..q. .8... :2.:- 8...- :2. ...T ..... ...8. 8:..- ::...- :8. o::.- Eo-:.o...o-2:.:
8:. 8:. :::.- 2...- :2. .8.- 8:. .8... .2... 3.. 2. 2:. 2:. ...8. 2:: 8.... 8... :::. ..E-:...::._.o.z-2:.:
:3, 88. ..::.- 8:...- 82. 8..- 8:. ...8. :8:- :...- :2. 8:.- 2... .8... 2...:- o:.... :8. :2.- ...-:.:.2.....-2s:
.8. :8. 2..- :8- 8.. .8.- 2.... 8:. 2..- z...- 8... 8...- 2.8. 8... .8..- :8- 8... 8..- ...=a....=_-2..-BB<-2..8...
:2.. 8.... 8:. .8. 8... :2. 88. .8... :8: .8. 8... :::. :8. .8... .8.: :2.. 3... :2. 6.25:0...2800
:8. .8... 2.: :3. :2. :8. 2.... 8.. z... :8. E. :2. .8. .8... 23 8... :8. :2. 2.22385.
:8. ...8. 2:: ..2. 8... 2... :2.. ...8. 8:.» .8. 8... 2o. .8. .8... S: 2.. 8... z... 2.02:

8... 8...: 8::. 8...: 2.... ::::: 2:. ...8.: 2.... 22:: 8:. 8...: 25.2.8.
:- .. . : 2.: .. ..F .. . : .2: .. 2. .. . .. 2.: .. 832......

88..."... 82...". 8...“... 88...”... :33": :::...u.... 2......"5 5.3% 8:8"... 5.2.2..

8..:-.....: 82-82 ...8.-.2 v3...— 2.5

 

 

 

a +._::::::: + .mmummn: 33::2: + ._>:: + ﬂees: + ..z0.:.: + 55:: + ...8.-:5 + 32-2: + 5:... + .1: + 6.: + .2... + .<.: + .: ".5

E In

.380: owns: ...8.... owns.— =.2o>O o... 5.. 33:9. .8. .25.... cog—3...... .8250 s: 03....-

144

Also, model ﬁt statistics, namely R2, f statistic, and a (p-value), are provided in
these tables as measures of the model ﬁt for the data for each data period. These values
are used to test the null hypothesis that there is no linear relationship in the population
between the dependent variable and the independent variables. In other words, these
values are used to test multiple equivalent null hypotheses that all of the population
partial regression coefﬁcients are 0, and the population value for the multiple R2 is O. The
R2 reﬂects the percentage of the variance in the dependent variable explained by the
independent variables. The f-statistic, which is the ratio of the regression mean square to
the residual mean square, provides the basis for testing the null hypotheses that all of the
population partial regression coefﬁcients are 0, and the population value for the multiple
R2 is 0. The (1 serves as the signiﬁcance indicator of the one-way test for the null
hypothesis that the R2 is equal to zero.

Tolerance values of independent variables provided in Tables 35, 36, and 37 are
investigated to detect if there is a multicollinearity and/or singularity problem in the
regression analyses of this study. As was mentioned before, multicollinearity happens
when two or more variables in the regression model are highly correlated; and singularity
occurs when one score in the regression model is a linear combination of others. Both of
these conditions jeopardize the results of the multiple regression models by causing
instability in the matrix inversion. A variable with tolerance values below 0.19 is
assumed to be a linear combination of other independent variables (Hair et al.1998), thus,
dependent highly on other independent variables.

As can be seen in Tables 35, 36 and 37, all tolerance values are above the cutoff

point. The tolerance values of Illinois, Michigan, and Ohio are smaller than 0.5 for the

145

general population model across all data periods. Michigan has the lowest tolerance
values across all data periods (about 0.3). This might be the result of overrepresentation
of Michigan respondents in the sample.

As can be seen in Table 35, when estimated for the all-years’ survey data using
the two-tailed t-test, most of the independent variables, ten out of 14 independent
variables were found to be signiﬁcantly inﬂuential on the ‘Setting/Sense of Place’ Factor.
These variables are race-white, gender-female, state-Michigan, state-Ohio, state-Ontario,
overall travel experience, visitation-visitor, survey season-winter, survey season—spring,
and survey season-summer. However, when the same estimation was applied for the
1996-1997 data, four of these ten inﬂuential variables, namely, overall travel experience,
survey season-winter, survey season-spring, and survey season-summer were found to be
insigniﬁcant. Yet, when applied on the 2001-2002 data, ﬁve of these ten variables
namely, gender-female, state-Michigan, state-Ohio, state-Ontario, and overall travel
experience were found to be insigniﬁcant. Only two independent variables were found to
be signiﬁcantly inﬂuential across all data periods: race-white (whether or not respondents
are white) and visitation-visitor (whether or not respondents visited Michigan before).

This means that only whether or not respondents are white and whether or not
they visited Michigan are signiﬁcant in explaining the Setting/Sense of Place Factor. The
sign and size of [3 values are similar across different data periods, so the results of the all-
years’ data will be discussed. The coefﬁcients ([3 values) are positive for the race-white,
and visitation variables, white respondents (B=O.105) have a better perception of
Michigan than respondents from other race groups, and those who visited Michigan

(B=O.262) have a better perception of Michigan than the non-visitors on the Setting/Sense

146

of Place Factor. Having a greater coefﬁcient, however, the prior visitation variable is
more inﬂuential than the race variable in explaining this factor; in fact, its coefﬁcient is
the greatest among all variables that were signiﬁcant in the all-years’ data.

A similar dropout pattern is found in the estimation results for the
Activities/Things to do Factor with some differences in the numbers and types of the
inﬂuential variables. As can be seen in Table 36, when estimated for the all-years’ data,
and using the two-tailed t-test, the majority of the independent variables, eight out of 14
independent variables were found to be signiﬁcantly inﬂuential on the Activities/Things
to do Factor. These variables are age, gender-female, income-above-the-median, state-
Illinois, visitation-visitor, survey season-winter, survey season-spring, and survey season-
summer. However, when the same estimation was applied for the 1996-1997 data, six
variables were found signiﬁcant; survey season-winter, survey season-spring, and survey
season-summer dropped out and race-white added. Yet, when applied on the 2001-2002
data, the gender-female, income-above-the-median, state-Illinois, and visitation-visitor
variables dropped out and race-white variable was added, as in the 1996-1997 data.
Although the visitation-visitor variable was signiﬁcant across all data periods for the
Setting/Sense of Place Factor, it is not signiﬁcant for the Activities/Things to do Factor.
Only one independent variable is signiﬁcantly inﬂuential across all data periods:
respondents’ age.

This means that only respondents’ age is signiﬁcant in explaining the
Activities/Things to do Factor. The sign and size of [3 values are similar across different

data periods, so the results of the all-years’ data will be discussed. Since the coefﬁcient is

147

positive (6:0.159), older respondents have better perception of the Activities/Things to
do Factor than do younger respondents.

One would expect that the combination of those variables inﬂuential in explaining
each image factor would be inﬂuential in explaining the Overall Image of Michigan,
which is the mean image score of the 15 image items. As can be seen in Table 37, when
estimated for the all-years’ survey data using the two-tailed t-test, nine out of 14
independent variables were found to be signiﬁcantly inﬂuential on the Overall Image of
Michigan. These variables are age, race-white, gender-female, state-Illinois, state-
Michigan, state—Ohio, state, Wisconsin, state-Ontario, and visitation-visitor. When the
same estimation was applied for the 1996-1997 data, the same variables were signiﬁcant
except for the race-white variable. Yet, when applied for the 2001-2002 data, ﬁve of
these variables namely, age, race-white, visitation-visitor, survey season-winter and
survey-season-spring were found to be insigniﬁcant. Only two independent variables
were found to be signiﬁcantly inﬂuential across all data periods: age and visitation-visitor
(whether or not respondents visited Michigan before).

This is not commensurate with what was expected initially. Race-white and
visitation-visitor were found to be consistently inﬂuential in explaining the Setting/Sense
of Place Factor and age was found to be consistently inﬂuential in explaining the
Activities/ Things to do Factor, across all data periods. Only two of these variables were
found to be consistently inﬂuential in explaining the Overall Image of Michigan, across
all data periods: age and visitation-visitor.

This means that only respondents’ age and whether or not they visited Michigan

are signiﬁcant in explaining the Overall Image of Michigan. The sign and size of B values

148

are similar across different data periods, so the results of the all-years’ data will be
discussed. The coefﬁcients ([3 values) are positive for both variables, older respondents
(B=O.ll3) have better perception of the Overall Image of Michigan than do younger
respondents, and those who visited Michigan (B=O.288) have better perception of the
Overall Image of Michigan than do non-visitors. Having a greater coefﬁcient, prior
visitation variable is more inﬂuential than the age variable in explaining the Overall
Image of Michigan; indeed, its coefﬁcient is the greatest among all other variables that

were signiﬁcant in the all-years’ data.

Recent Visitor Segment Model Variables

Descriptive statistics and the correlation matrix of all the variables included in the
recent visitor segment model are presented in Tables 38 and 39, respectively. As was
discussed in the demographics section, the demographic proﬁle of the recent visitor
segment model-respondentsl is different from not only all respondentsz, but also the all
recent visitors group3 .

The recent visitor segment model respondents are slightly younger on average
(41.95 years of age) than both all respondents (43.1 years of age) and all recent visitors
(43.1 years of age). The differences in the percentages of white respondents are quite
noticeable. Over 90% of the recent visitor segment model respondents are white; this
percentage is about the same as that of the all recent visitors group, but about 7% higher

than that of all respondents.

 

1 Those who took a trip to Michigan within the past 12 months and provided valid responses for all model variables, including
a demographics.

‘3 All respondents with demographic information.

32 Those who took a trip to Michigan within the past 12 months and provided demographic information.

149

Table 38. Recent visitor segment model variables’ descriptives and frequencies

 

 

(n= 1,269).
Name Description Mean Std. Dev. Frequency
(% of code 1)
Independent Variables
Age Age of the respondents 41.95 13.22 -
Race-Whitea White respondents 0.91 0.28 91.2
Gender-Femaleb Female respondents 0.58 0.49 57.6
Income-Above-the-median° Respondents with above-the-median 0.81 0.40 80.5
income
State-Illinoisd Respondents from Illinois 0.10 0.30 9.9
State-Michigan° Respondents from Michigan 0.61 0.49 61.4
State-Ohiof Respondents from Ohio 0.11 0.31 10.6
State-Wisconsin“ Respondents from Wisconsin 0.05 0.23 5.4
State-Ontarioh Respondents from Ontario 0.05 0.21 4.8
Overall travel experience Number of pleasure trips to any 5.83 5.51 -
destination in the past 12 Ms
No visitor variable since they are all recent visitors
Visitation frequency Number of pleasure trips taken in 3.26 2.29 -
Michigan in the past 12 months
How recent visit-Most recent' Visited Michigan during the most 0.58 0.49 58.2
_ recent trip in the past 12 months
Visit season-WinterJ Last visit to Michigan was in winter 0.15 0.36 15.0
Visit season-Springk Last visit to Michigan was in spring 0.15 0.36 15.3
Visit season-Summe Last visit to Michigan was in summer 0.46 0.50 45.8
Nights Length of stay in Michigan during the 2.88 3.65 -
last visit
Activities Number of activities during the last 4.25 2.41 -
vrsrt
Survey season-Winter“ Survey conducted in winter 0.26 0.44 26.2
Survey season-Spring" Survey conducted in spring 0.22 0.42 22.4
Survey season-Summer° Survey conducted in summer 0.22 0.42 22.4
Dependent Variables
The Setting/Sense of Place Factor score for Factor 1 0.384 0.830 -
Activities/Things to do Factor score for Factor 2 0.176 0.917 -
Overall Image Mean image score 8.165 1.272 -

 

W an 9.0 0'9

:The variable is continuous, frequency of each value is not given.
:Codedwhite respondents = 1; all others = 0.

: Coded female respondents = 1; all others = 0.

: Coded above-median income respondents = 1; all others = 0.
: Coded Illinois respondents = 1: all others = 0.

: Coded Michigan respondents = 1; all others = 0.

: Coded Ohio respondents = 1; all others = 0.

: Coded Wisconsin respondents = 1: all others = 0.

:CodedOntarioreepondents=1;allothers=0.
:Codedmostrecentvisitors= l;lessrecentvisitors=0.
:Codedwintertn'ps= l;ailothers=0.
:Codedspringtrips= 1;all others=0.
:Codedsummertrips=1;allothers=0.

:Coded wintersurveys= l;allothers=0.

:Coded spring surveys= I;allothers=0.

o :Coded summersurveys= l;all others=0.

Female respondents constitute 58% of the recent visitor segment model

respondents, which is about 2% lower than those of all respondents and the all recent

visitors. The percentage of the above-the-median income respondents for the recent

150

 

acorn—etc". .mozutg $on EoEmom .553 880M .3 053.

151

visitor segment model respondents (81%) is a little higher than that of the all recent
visitors group (77.7%), but noticeably higher than that of all respondents (66.8%).

The representation of different states in the recent visitor segment model
respondents group is rather different from those of both all respondents and all recent
visitors. The majority of the recent visitor segment model respondents are Michigan
residents (61%), while they constitute 48.1% of the all recent visitors group and only
18.5% of all respondents. Wisconsin and Ontario residents each represent only 5% of the
recent visitor segment model respondents; these percentages are about 2% lower than
those of the all recent visitors group while the decrease is 5% for Wisconsin and 13% for
Ontario when compared to all respondents.

When compared with the general population model respondents, the only
noticeable difference in the recent visitor segment model respondents’ demographics is
the distribution of the residence states, Michigan’s being 25% higher and other states’
being 4% to 8% lower in the recent visitor segment model respondents group.

The mean for the overall travel experience variable (average number of trips taken
in any destination in the past 12 months) is 5.83, while mean visitation frequency to
Michigan (average number of trips taken in Michigan in the past 12 months) is 3.26.
Since the recent visitor segment model respondents are all recent visitors, there is no
visitation-visitor variable in this model; however, there are additional last visit related
variables: (1) whether or not the last visit to Michigan is the most recent trip, (2) the
season of the last visit to Michigan, (3) the length of stay in Michigan during the last
visit, measured as the number of nights spent, and (4) the number of activities

participated in Michigan during the last visit.

152

The majority of the recent visitor segment model respondents (58%) visited
Michigan in their most recent trip within the past 12 moths. As for the season of the last
visit, summer was the most popular season (46%), while winter and spring were a lot less
popular seasons and equal in distribution (15%) for the recent visitor segment model
respondents. On average, recent visitors spent 2.88 nights and participated in 4.25
activities during their last trip to Michigan.

In terms of the survey season, the recent visitor segment model respondents are
similar to the general population model respondents. However, there are differences in
the mean image factor scores and mean Overall Image scores between these two groups;
mean image factor scores for the recent visitor segment model respondents are all
positive and both mean factor scores and the mean Overall Image score are higher than
those of the general population model respondents.

As can be seen in the correlation matrix presented in Table 39, there is no
substantial correlation among the independent variables; this is desirable since it is
evidence that multicollinearity is not likely a problem in the multiple regression analysis.

Since the same multiple regression analysis was applied to the 1996-1997 and
2001-2002 data periods, the descriptives for all variables for these periods are also
provided in Table 40. As was observed in the general population model respondents,
there are some discrepancies in the recent visitor segment model respondents’
characteristics in the 2001-2002 data period. Means (frequencies for dummy variables)
for the 1996-1997 data are more or less similar to those of the all-years’ data. In the
2001-2002 data, however, there are some noticeable differences in respondents’ age,

income, residence states, and how recent was the last visit to Michigan.

153

Table 40. Recent visitor segment model variables’ descriptive statistics across different
data periods.
All-years" (n=1,269) 1996-1997 (n=730) 2001-2002 (n=347)

 

 

Notation Variable Mean Std. Dev. Mean Std. Dev. Mean Std. Dev.
Independent Variables
A Age 41.95 13.22 40.49 12.76 45.82 13.69
R Race-White 0.91 0.28 0.91 0.29 0.91 0.28
G Gender-Female 0.58 0.49 0.56 0.50 0.59 0.49
1 Income-Above-the-median 0.81 0.40 0.81 0.39 0.76 0.43
S-IL State-Illinois 0.10 0.30 0.10 0.30 0.06 0.24
S-MI State-Michigan 0.61 0.49 0.58 0.49 0.71 0.45
S-Ol-I State-Ohio 0.11 0.31 0.12 0.33 0.08 0.27
S-WI State-Wisconsin 0.05 0.23 0.06 0.24 0.03 0.18
S-ON State-Ontario 0.05 0.21 0.05 0.22 0.05 0.21
OTE Overall travel experience 5.83 5.51 6.04 5.40 5.32 4.31
VF Visitation frequency 3.26 2.29 3.15 2.28 3.49 2.21
VR How recent visit-Most 0.58 0.49 0.52 0.50 0.70 0.46
recent

VS-WI Visit season-Winter 0.15 0.36 0.16 0.37 0.11 0.32
VS-SP Visit season-Spring 0.15 0.36 0.14 0.34 0.16 0.37
VS-SU Visit season-Summer 0.46 0.50 0.46 0.50 0.44 0.50
N Nights 2.88 3.65 2.94 3.87 2.43 1.97
AC Activities 4.25 2.41 3.93 2.37 4.70 2.39
SS-WI Survey season-Winter 0.26 0.44 0.29 0.45 0.21 0.41
SS-SP Survey season-Spring 0.22 0.42 0.24 0.43 0.17 0.38
SS-SU Survey season-Summer 0.22 0.42 0.20 0.40 0.18 0.39

Dependent Variables

F 1 The Setting/Sense of Place 0.384 0.830 0.490 0.772 0.074 0.942
F 11 Activities/Things to do 0.176 0.917 0.077 0.879 0.414 0.960
01 Overall Image 8.165 1.272 8.187 1.277 8.050 1.266

 

* : The ﬁill sample including 1996. 1997. 1998. 2001 and 2002, with which the study hypotheses are tested.

The recent visitor segment model respondents in the 2001-2002 data are about 5
years younger on average. Above the median income respondents are about 76% of the
recent visitor segment model respondents in the 2001-2002 data, while this percentage is
about 5% higher for other periods. The percentage of Michigan residents is about 10%
higher, and those of other states (except for Ontario) are 2% to 4% lower in the 2001-
2002 data than in the all-years’ and the 1996-1997 data. Finally, 70% of the recent visitor
segment model respondents visited Michigan in their most recent trip within the past 12
months. This percentage was 12% lower for the all-years’ data and 18% less for the

1996-1997 data. The Overall Image is above 8 for all data periods; however, it is lower ‘

154

for the 2001-2002 data than it is for the 1996-1997 data, while the opposite is true for the

standard deviation of the Overall Image.

Recent Visitor Segment Model Test Results

The multiple regression model speciﬁed to investigate the relationship between
recent visitors’ destination image and selected demographic variables, selected past travel
behavior variables, and the season of the survey is as follows:

Image, = Bo + 31A, + [32K + 33G, + B41," + 858-114, + B6S-MI, + B7S-OH,’ + ﬁgs-WI,

+ ﬁgs-ON, + BloOTE, + 311W, + BuVR, 'l' BuVS-WI, + 314VS-SP, +
BuVS-SU, + BmN, + BnAC, + BmSS-WIH‘ BwSS-SP, + BzoSS-SU, + 8,
where,
Image,- = the image held by individual ,,
[31.20 = coefﬁcients of independent variables 1-20, and
a, = the error term for individual ,
Since two image factors were extracted with the Exploratory Factor Analysis,
these two factors and the Overall Image were used as the dependent variables in the
multiple regression models. Thus, the two regression models are:
F I, = Bo + 31A, + 32R, + 83G, + B41, + BjS-IL, + BsS-MI, + B7S-OI'I, + ﬁgs-WI, +
ﬁgs-ON,- + BloOTE, + BuVF, + BuVR, + B]3VS-WI, + BuVS-SP, + B,5VS-
SU, + 316N,+ B]7AC, + BmSS-WIH‘ 319SS-SP,+ BzoSS-SU, + 8,

F H, = Bo + 31A, + 32R,- + 83G, + B41, + BsS-IL, + B6S-NH, + B7S-OH, + ﬁgs-WI, +
BgS-ON, + BIOOTE,‘ + BuVF, + BuVR, + BuVS-WI, + 314VS-SP,+ BISVS-
SU, + BléNj + B]7AC, + BlgsS-WI,+ 019SS-SP, + BzoSS-SU, + 8,

01,: Bo + 81A, + 82R, + B36, + B41, + BsS-IL, + ﬁgs-MI, + B7S-OH, + ﬁgs-WI, +
ﬁgs-ON, + BloOTE, + 311W,- ‘l' BuVR, + BuVS-WI, + BMVS-SP, + BISVS-
SU, + 816N, + BI7AC, + B,gSS-WI,-+ B,gSS-SP,- + BzoSS-SU, + 8,-

These OLS multiple regression models were estimated for both image factors and

the Overall Image using the all-years’ data; then, to validate the results, the same models

155

were estimated using the survey data for two different periods: 1996-1997 (the initial two
years of the study) and 2001-2002 (later two years of the study). Estimation results for
Factor I (The Setting/Sense of Place), Factor 11 (Activities/Things to do) and the Overall
Image are provided in Tables 41, 42 and 43, respectively.

As can be seen in Tables 41, 42 and 43, all tolerance values are above the cutoff
point of 0.19. As in the general population model, the tolerance values of Illinois,
Michigan, and Ohio are the smallest across all data periods. Michigan has the lowest
tolerance values across all data periods (about 0.29), which might be the result of
overrepresentation of Michigan respondents in the sample.

As can be seen in Table 41, when estimated for the all-years’ survey data using
the two-tailed t-test, nine out of 20 independent variables were found to be signiﬁcantly
inﬂuential on visitors’ perception of the Setting/Sense of Place Factor. These variables
are race-white, income-above-the median, state-Illinois, state-Ontario, visitation
frequency in Michigan, nights spent in Michigan, survey season-winter, survey season-
spring, and survey season-summer. However, for the 1996-1997 data, only ﬁve of these
variables remained signiﬁcant; state-Illinois, nights spent in Michigan, survey season-
winter, survey season-spring, and survey season-summer were not signiﬁcant, and state-
Ohio was added as signiﬁcant this time. When applied to the 2001-2002 data, only race-
white and survey seasons were found to be signiﬁcantly inﬂuential on the Setting/Sense
of Place Factor. Only one independent variable is signiﬁcantly inﬂuential across all data
periods: recent visitor respondents’ race.

This means that only respondents’ race is signiﬁcant in explaining the

Setting/Sense of Place Factor for recent visitors. This is the same as that of the general

156

Amodvaa so. as: 9.2 8.. .95. $3 a... Eon :5... «5595 53:85.95 9: “56508 a 35 8.5%.: n p
Sodvs .8. .33 22 8.. .25. {can 5 Son Eat €96.95 3.56555 5 Eat—beau u E: 8:86.: ” a
.83 a... 885er .53m 2.. 523 .2... .32 a: :5" .52 .32 .82 5:522: 29:3 =5 25 u ..

 

 

 

 

 

5a.. .25. 5.5 m2. 3. 5.5. :5. 3.... 52..- NS- 55. «5.. «S. .25. mm: 5.. $5. 5mm. 5395555588.3:5
2b .25. 23 SM. 3. New. 25. Now. 5%.- 25.- as. 25.- 55. .25. a: 2:. m5. 5%. €38msameoaoiutzm
:5. .25. 5:. :5. E. :5. ms. 55. .5..- 25.- :5. 25.- $5. .25. 83. 52. .5. Sm. 2335253583.FEW
:5. :5. an- 55.. :5. N5. 35. 2:. $2.55. :5. 25.- 2.». SN. 82.58.. 55. 25.- 6332.62
43. E. 5: 58. no. 58. 8.... 2.. no: “no. :5. :5. :5. .NS. 82 a5. 55. 25. 25532
:5. :5. Se. 58. 2:. 48. 5%. 25. 5.. 58. 58. 58. :5. 25. SN. 25. 58. 25. 53>:555853855
:5. NS. NS. $5. 5. E. as. :5. 34.- 25.- :5. 58- 53. Now. 5. 25. ...8. So. EwmzmccaweasmES
o:.. :5. :5. :5. E. 25.- 53. MS. 5..- 25.- :5. 58- 85. E. 52.- 25.- 58. NS- :B-m>:2=§-=882§>
5%. SN. :2 5.5. 2:. 5:. a» :5. 22 55 52. 25. an. 53. oz: 58. 53. 58. 95558..82-:23:82..5=
25. mm. 52 :5. ms. 58. 59 .25. 55:: :5. 55. 52. .:5. 332:. :o. 58. AgvecuscécouaeS
E. 5:. 8m. :5. :5. 25. MS 5. on so 25. :5 :5. NS. 2.». 48. :5. :5. Show85:855.;2535
25. :5. 95:55.- E. 5.. $5 .:5. 32.3... E. 52.- .45. .25. ”2.3:. as. 53..- Aszvoesaoésm
5%. om. 5%. m8. 5:. 3:. 2.... 5%. NS. 58. N3. 58. 55. N2. :5. N8. E. 5:. 23952823055
5%. an. Sn- 53.- SN. 53.- E. .25. 85 3:- 52. 2o. 5:. N8. 53.38.- 2:. 25.. 50-825095
E. 95. ms. 25.- 52. :o. 43 52. 5:8 2:. 2: EN. .2. N3. 35. :5. :5. 22.955.55.23
5mm. 8“. 58.35. 52. NZ. :5 E. 8:55 5:. 5:. :5. .55. 58.22 2:. :5. A.=-m:_o===-2sm
5.8. 95. 8M. ms, 5:. m8. :5 .35. $2 58 58. 52. N3. .52. $2 5.5. 58. 52. e=a_BE-05-BB<aESE
$5. 2.5. 5%. 58. 85. 58. N; NE. SN. 25- no. 25.- .5. SN. 52 58. $5. 58. Evangesecuo
5.8. .58. 22:: 5:. Em. NR .25. 35:: 55. :2. ME... .25. :23: as. 5:. €223-85:
:5. a... E. 55.. 25. 25.- 43 :5. 58 25 Sc. 25. 5.5 B. 22-58.- So. 25.- 25...»...

25. 83- 5:. 2? ~54. 87 2:. 3.. 25. SS- 2:. 52..- 35.258
H u . 5 :8 .. .._. a . a A; a H a . 5 am a 8525»
55.5"... 585i £25m 25.58 Eat 5:5".5 855“... 5:31 23".: 55.255

8.5.83 3.2.52 .eao>-=< wee... 2:;

 

 

 

 

e + 53%.“. + 5-5.5 + 53-525 + 9...... + is: + .3
-92: + _._m-m>:= + 55.91.: + 35:: + ”.55— + 3.....er + .293. + 55.3. + 56.3. + .425... + and“: + .5. + 6.: + #3. + {a + .5 n .— ..—

.AoouE mo omcomhﬁsom on: _ 552$ 5.: 8:52 58 3on EoEwom 585$ Booed .3 038.

157

€5.35 53 :53 o3. 8.. .2615? 5 28 62.. Ens-Eu 25359” a. .5568 a 55 8529: U a

Codvﬁ use .33 92 Su .96. $8 .5 83 59¢ «co-Ema. 3585:»: mm 285508 a 55 8.855 n a
583. 8a 88:89»: >35. 05 523 55 .83 93 Sam .32 Koa. doa— mﬁvaoﬁ 39:8 :3 2F ” _..

 

 

 

 

5e. .25. 83-mon- 52. E.- _$. 52. an: 5.5. 25. 5:. S. .25. 25545.- NE. SN. Sm-mm:2=§m-§§ 5:5
E. .25. 5:32.- 5:. 5.5.- 25. 55. 53. :5. N8. :5. 5. .25. 555.35.- E. E.- Em-mmvwaﬁmeaaa 5:5
:5. .25. 32.8:- 3. E. 58. .5... :5. NS. :5. $5. 5. .52. 58.28.- 55. 5.- 23553535858 52.5
:5. N5. 5:: 2:. NS. :5. 35. .55. 3.: 2:. :5. 58. S. .25. 2:... 5:. :5. 25. 6385352
55. 25. E. 25.- 58. :5.- 85. 5.3.. :5. 58. 25. .55. 5. 53. SN. 55. :5. :5. 2:552
:5. an. 58.- 55.. 52. 2.5.- 5%. E. E._-55.- :5. ~:.- 5. E. 8:73.. ..5. ET Sm-m>:oe§m-§§§>
:5. :m. 23.95.. 52. 5:. So. .5... E: 35. 5:. 2: 3. 55. 58. :5. 35. :5. awmzmaeaméaaaﬁS
5:. 5...... :5.- 58. 9:. 5:- 55. E... 58.- :5.- 5: :5.- Se. 5:. SM. 25. a5. 58. c35>§2=§é§3§>
5...». 2... 25. $5. 4:. :5. 52. ...8. :55 25. M55. :2. 52. .25. :2 95. 55. 52. Ame/5:82 8233.582 .6:
$5. 55. :5. 55. Re. 25. E. :5. :5. «8. £5. .25. .5: E. on... ...8. N5. N5. £>E§85=2§E>
MS. 5. 59.725.- ..5. 25.- MS. a. 58.- 35.- 55. :5.- R». 5:. 55:55.. 25. :5.- ABOVS=88§9§=296
55. $5. 53..- 58.- E. 2..- Eo. 8m. :5. 55. E. 8.. $5. a... 35. m8. 3:. 25. 2988559255
5%. E. 52.35.- 52. E..- 2.... 3.. 35.- 58.- a... 5:.- 25. E. 8...-:5.- 52. 2:- 53-958.885-23
53.. 52. an- 53.- E. 5N.- t... E. an. 55. 52. 2.5. 5:... £5. 25.- :5.- o:. 25.. 59925-255
E. E. 25.3.2.- SN. 3.. E. E. 35. 25. 2.. ...5. E. :5. 2:.- 25. 55. 55.- 52-553.522-23
52. .25. 52.52.. E. :5.- :m. :8. 52.2... 5:. E.- 5... .55. 255-5:- 5:. an- 335555-23
E... 52. ~37 MS.- 52. :8.- 85. Po. N335.- :5. 9.:- E. .25. $2.55.. 85. X..- SEES-05.885858:
as. E. 5: :5. 2:. E. E. .:5. 325:. 55. E. E. .55. $2 58. 53. oz. 652955-5200
mm... mm. 58. 55. 5:. 8: N3 45. 25.35.- E. 5753. 52. 54:5.- 55. 5.. $223.85:
:5. .:5. 2.35:. :5. NS. 55. .25. 55... 3... :5. 55. 55. .25. 23 5:. :5. :5. :50»...

E. 5:. SH. 5:. :5. :3- E. 2...- Na. 82- 2... 58.. 55988
H a . a ...3 a ..F a . 5 .55 a H a . 5 ...3 a 822...;

55.5"... Ea": 52.5"”: 55.5”: 58.1 SSH”: 55.5"... «Six Nedwm 5.252

25358 32.5.: Eng-=4. THE... 25

 

 

 

 

e + .35.... + ...m-mmaa + 53-5.... + 65.25 + .225 + 53>}.
+ page: + 3352:: + .452: + 55:5 + .952: + ..zo-m£ + 65.3. + 50.4: + sue-ms: + Jim... + .3. + .06 + 2.: + 3:: + a u ..= ..—

.Aow 8 mmcitmoﬁ>m8<v : 885m .8 3.3.2 58 .256 :58QO .353 Booed .Nv 635..-

158

Anodvnv «8.. >55 25. 8.. .93. exam-o .a Sou Ea... Eat-Ev rundown—EV 5. 5.32-:08 a 85 8.36:. n

a

Ccdvmv «no. .33 03. 8.. .26. 5.58 3 Son Spa Eat-Eu b.5055.» 5. 566508 a .55 8.8:»... n -
538. a. 33.53... :53. 2.. 5...: 5.... .855 55 .555 .555. .85. .555. 55.523. 25.5. =5 .5 H ..

 

 

 

 

 

E. 555. 2... 555. 2.. 3.. 555. 55... 58. 85. 55. 55.. $5. .8. 555.. 555. 555. 55.. 55-565.553-588.5:55
555. .555. 555.5 55. 3.. 5:. :5. 555. 555. 58 ..N. .8. $5. .58. 3.5.5 85. 555. 55.. €5-55.55555-58585255
55... 5:. 85.. 58. 55. 58. 555. 555. 52.. 5.5. 5: 5.... 555. .555. 555.5 58. 85. 58. 23-55.55.358....5255
:5. .85. 35.5 58. 58. $5. 555. 58. .55.. .55. 58. .55. N55. ..55. .52 85. 25. $5. 6.5.3.552
E. 555. .3. 5.5. .8. 555. 555. 555. 85.. 555. 5.5. 58. 555. 555. .55.. 15. 555. 5.5. 2.2.52
.8 555. 55.. 555. 55. 58 .55. 8... $5.- :5.- 55.. 58- 555. 555. 58.- 5.5. 58. 58.- 53>.55.555-588.55
555 55... .55.- :5.- 55.. 5:.- 555. 5.5. 5:..- 5.5.- 5:. $5.- 555. E. 5:.- 58- 555. E..- ...m-mtmccawaoaaES
85. 8... 555.- 58- a... 8..- 555. 2.5. 555.- 58.- 5:. :5.- 58. 555. $5.- 58.- 8.. 555.- ..3-w>.§=§-=8§:5>
:5. .85. 52.5 555. 3. 5.5. 555. 18.8.5 58. .55 55. $5. .555. 55: 555. ..55. 5:. $5.582.8...-55:58.55:
5:. 555. 555.. 555 58 58 58. .555. 555.5 .2. N8. 555 555 .555. 511 5... 5.5 555. C>..§.=.§.=o.§.m.>
5:. 55... E.- 58.- :5. 555- .8. 555. 5:.- 555- 555. .55.- 35. 555. 5.5.- 5.5.- 5.5 :5- .Eo..u=..s%o_p§.55.80
5%. 8... 85.- 35.- 555. 555- 8.5. .58. 555.5- 555.- :5. 55... 555 .555. 25.5- E..- 5: 555,- 205.25.505.55
555. 5.5. 52.- :5.- 555. 555- 555. 58. E..- ~8.- 555. $5- 58. 5:. 555..- 85.- 52. 58.- 235.525.53-255
55... 55.. 5.2..- 58. 555. 555.- 555. .25. 85.5- 5:.- E. 53.- 2.... .555. $5.5- :5- E. 555.. $9925.23
555. .55. 58..- 55... 55.. 5.5.- 555. 5.5. 85. .8. 8.. 555. 555, 5.... .55.- 5.5.- 55.. 58.- 52-955.522-255
85. 5.. 5m..- 58.- 555. 555- 5.5 $5. 53.- :5.- 55.. K..- 55.. 3.5. 555.- 5.5.- 55.. 555.- 3.-m:_o=_...-2..m
555. 555. 8..- :5.- 55.. 35.- R5. 85 58.- 85.- 52. 555- $5. .55. 2. .- 58.- :5. 85.- c.=2...E-£.-38<-5=_8=.
58. .25. 85.5 8.. 55.. 555 :5 .555. 555.5 555. 555. 555. 555. .555. 5:5 555. 555 85. 6.2.2...-5550
555. 555. 58.. 555. 5:. 555. .55. .555. 58.5 555. 5:. 555 58 ..55. 55.... 58. 8. 555. 2.223-855
:5. .85. 5:5 55.. 85. :5. $5. .555. 555.5 2.. 85. 5.5. 555. .555. 555... 2.. 85, :5. 2.05...

555. 58.5. :5. 58.5 555. $5.55 555. 58.5 555. 555.55 85 555.5 .5228.
.5 5 . 5 .55 5 .5 5 . 5 5.5 .5 5 . 5 5.5 5 8.5.5....»
555.5": 552". .555"... 555.5".5 5...5u\ 5:5"... 555.5"... 555.1 85.5””: 55.5.52

855.855 35.555. 2.55-=5. may... 25.-5

 

 

 

.. + 55-5525 + 5.5-5555 +...3-mm._5 + .325 + .225
+ ...m-m>._5 + 5.5-9.5 + £35.25 + .5525 + 55:5 + .5525 + 2.5-5.5 + 55-5.5 + 55.55 + 55-5.5 + 3-5.5 + ...5 + 6.5 + £5 + {5 + .5 n ...o

.5285 owns... :52... 0mg... =So>o o... 8.. 5.358 .58 .25.: .553... 585$ .503— .9. 035,—-

159

population model results for the Setting/Sense of Place Factor; visitation is embedded in
the results since these respondents are recent visitors already. The sign and size of [3
values are similar across different data periods, so the results of the all-years’ data will be
discussed. Since the coefﬁcient is positive (B=O.143), white recent visitor respondents
have better perception of the Setting/Sense of Place Factor than do recent visitor
respondents ﬁ'om other race groups.

As can be seen in Table 42, for the Activities/Things to do Factor, nine variables
were found to be signiﬁcantly inﬂuential for the all-years’ data: age, gender-female,
income-above-the-median, state-Illinois, whether or not the last visit to Michigan was the
most recent trip, the number of activities participated in during the last visit in Michigan,
and survey seasons. For the 1996-1997 data, only the ﬁve following variables remained
signiﬁcant: age, gender-female, state-Illinois, whether or not the last visit to Michigan
was the most recent trip, and the number of activities participated in during the last visit
in Michigan. For the 2001-2002 data, age, state-Illinois and survey seasons were found to
be signiﬁcantly inﬂuential. Thus, for the Activities/Things to do Factor, two independent
variables are signiﬁcantly inﬂuential across all data periods: recent visitor respondents’
age and whether or not they are ﬁ'om Illinois.

This means that respondents’ age and whether or not they are from Illinois are
signiﬁcant in explaining the Activities/Things to do image for recent visitors. Age was
also signiﬁcant for the Activities/Things to do Factor in the general population model.
The coefﬁcient of age variable is bigger than that of state-Illinois variable, meaning,
recent visitor respondents’ age is more inﬂuential than whether or not they are from

'Illinois in explaining their images of Michigan. The sign and size of [3 values are similar

160

across different data periods, so the results of the all-years’ data will be discussed. Since
the coefﬁcient for age variable is positive (B=O.l78), older recent visitor respondents
have better perception on the Activities/Things to do Factor than do younger recent
visitor respondents. The coeﬁicient for the state-Illinois variable, however, is negative
([3: —O.108). Thus, recent visitor respondents from Illinois are more likely to have a
worse perception of Michigan on the Activities/1‘ hings to do Factor than do recent visitor
respondents from other states in the study region.

According to the assumption that the variables consistently inﬂuential in
explaining the two image factors across all data periods must be consistently inﬂuential
in explaining the Overall Image of Michigan, recent visitors’ race, age and Illinois
residence must be inﬂuential in explaining the Overall Image of Michigan. However, as
in the results of the general population model test, this assumption did not hold true for
the recent visitor segment model test either.

As can be seen in Table 43, for the Overall Image of Michigan, ten variables were
found to be signiﬁcantly inﬂuential for the all-years’ data: age, race-white, gender-
female, state-Ohio, state-Ontario, visitation frequency to Michigan, whether or not the
last visit to Michigan was the most recent trip, activities participated in during the last
visit in Michigan, survey season-winter and survey season-spring. For the 1996-1997
data, only the seven following variables remained signiﬁcant: age, race-white, gender-
female, state-Ohio, state-Ontario, visitation frequency to Michigan and whether or not the
last visit to Michigan was the most recent trip. For the 2001-2002 data, age, gender-
female, whether or not the last visit to Michigan was the most recent trip, activities

participated in during the last visit in Michigan and survey season-spring were '

l6]

signiﬁcantly inﬂuential. Thus, for the Overall Image of Michigan, three independent
variables are signiﬁcantly inﬂuential across all data periods: recent visitor respondents’
age, gender-female and whether or not their last visit to Michigan was their most recent
trip.

This means that respondents’ age, gender and whether or not recent visitors’ last
visit to Michigan was their most recent trip are signiﬁcant in explaining the Overall
Image of Michigan for recent visitors. Recent visitor respondents’ race and Illinois
residence were not signiﬁcant as expected, yet two new variables were added. The
coefﬁcient of age variable is bigger than that of gender and whether or not their last visit
to Michigan was their most recent trip variables, meaning, recent visitor respondents’ age
is more inﬂuential than their gender and whether or not their last visit to Michigan was
their most recent trip in explaining their Overall Image of Michigan. The sign and size of
B values are similar across different data periods, so the results of the all-years’ data will
be discussed. Since the coefﬁcient for age variable is positive (B=0.ll7), older recent
visitor respondents have better perception of the Overall Image of Michigan than do
younger recent visitor respondents. The second variable in terms of the size of the
inﬂuence is the gender variable; having a positive coefﬁcient (B=0.095), female recent
visitors have better perception of the Overall Image of Michigan than do male recent
visitors. Finally, with a positive coefﬁcient (B=0.064), recent visitors who visited
Michigan in their most recent trip have a better perception of the Overall Image of
Michigan than do recent visitors who visited Michigan in a less recent trip.

As can be seen in Tables 35, 36, 37, 41, 42 and 43, both the general population

model and recent visitor segment model test results are signiﬁcant for all time periods,

162

with very small R2 values, however. The R2 ranges between 0.081 and 0.246. R2 values
pertaining to the Activities/Things to do Factor are smaller for both models. There are
different plausible arguments for these low R2 values. First, selected variables may not be
good explanatory variables in the regression equation as argued by Pindyck and
Rubinfeld (1981). There might be other more important factors that determine these
image factors of Michigan, especially the Activities/Things to do Factor, such as
promotional media used by destination marketing organizations in Michigan. Also, as
argued by Achen (1982) and Court and Lupton (1997), the sample is the deﬁning factor
of the variances, not the underlying relationship between the variables; the large variation

in the data might be reducing the size of R2 (Pindyck & Rubinfeld, 1981).

163

CHAPTER V

SUMMARY AND CONCLUSIONS

This last chapter is comprised of three sections. The ﬁrst section includes a
summary of ﬁndings and discussion. The second section contains the limitations of this

study followed by the future research suggestions in the last section.

Summary of Results

Destination image determinant variables were deﬁned based on the existing
destination image literature and within the limits of the secondary data used in this study.
Michigan’s image was measured over an extended period of time. The determinants
included: (1) selected sociodemographics, namely, race, gender, age, income, and the
state of residence (2) selected past travel behavior variables, namely, overall travel
experience, prior visitation to Michigan, the frequency of visits to Michigan, whether or
not the last visit to Michigan was the most recent trip, the season of the last visit to
Michigan, the length of stay in Michigan during the last visit, and the number of activities
participated in during the last visit to Michigan, and (3) a methodological variable,
namely, the season of the survey. The following section provides a summary of the

ﬁndings of this study along with a brief discussion of them.
Michigan’s Image

Over the ﬁve-year period covered in this study, Michigan has enjoyed a solid and

positive image; mean scores for all Michigan image attributes remained above six on a

164

lO-point scale across different data periods. Overall, Michigan’s ‘scenic appeal’ has been
its strongest attribute since it generally has the highest mean scores (ranging between
8.17-8.27). However, perceptions of its weaknesses have changed; its weakest image
attribute was ‘museums’ (6.14) for the beginning years of this survey, while the weakest
image dimension was the ‘popular place’ (7.15) attribute during the later years.

Overall, Michigan’s image has improved over the ﬁve-year period since an
increase is observed in ratings of the majority of the image attributes, except for the
‘getaway’, ‘summer rec.’, and ‘family place’, which incurred decreases in scores during
the later years of the study. The difference between initial and later study periods is
statistically signiﬁcant for nine out of 15 attributes; namely, ‘getaway’(-), ‘museums’,
‘exciting’, ‘lodging’, ‘ﬁiendly’, ‘nightlife’, ‘family place’(-), ‘popular place’, and
‘historic’. Over the ﬁve-year period, a decrease was observed in the standard deviation of
all image attributes, statistically signiﬁcant for eight attributes: ‘museums’, ‘exciting’,
‘lodging’, ‘winter rec.’, ‘ﬁiendly’, ‘must visit’, ‘popular place’, and ‘historic’. This
means Michigan’s image has been stabilizing and crystallizing in the minds of current
and potential tourists.

The variation in the scores of the strongest dimension, ‘scenic’, has always been
the lowest (ranging between 2.04-2.10), meaning the perception of Michigan’s scenic
appeal has been not only positive, but also uniform across different types of respondents.
The attribute with the highest variation in scores has been ‘getaway’ (ranging between
2.52-2.63), closely followed by the ‘museums’ attribute (ranging between 2.46-2.63).

This means that perceptions of these attributes have not crystallized as much as it did for

165

the ‘scenic’ attribute. The variation could be attributed to certain people’s lack of
knowledge or perception on these attributes.

The improvement in Michigan’s image could be due to the promotional efforts of
Travel Michigan over the years, or it could be attributed to the increasing portion of
Michigan residents and visitors6 in the image-respondents group during the later years of
the study. The percentage of Michigan residents in the image-respondents group is 33.1
in the 1996-1997 data, while it increased to 47.1 in the 2001-2002 data. The percentage
of visitors in the image-respondents group is 75.3 in the 1996-1997 data, while it
increased to 79.3 in the 2001-2002 data.

Visitors’ mean image scores for all attributes are over one point higher than those
of non-visitors7 except for the ‘nightlife’ attribute, which is 0.82 point higher, differences
between the two groups being statistically signiﬁcant for all attributes. This means that
Michigan is perceived as weak on the ‘nightlife’ attribute although visitation improves
the perception of this attribute. The difference between image scores of visitors and non-
visitors is the biggest for the ‘family place’ and ‘getaway’ attributes, which could mean
the perception of these attributes by non-visitors is unrealistically less positive than the
objective reality.

Also, standard deviations in image scores of visitors are signiﬁcantly lower for all
attributes except for the ‘nightlife’ attribute. This means that visitation induces a more
uniform and stable image of Michigan. The difference is biggest for the ‘getaway’ and
‘scenic’ attributes and smallest and not signiﬁcant for the ‘nightlife’ attribute, meaning,

visitation crystallizes the ‘getaway’ and ‘scenic’ attributes more than the other attributes

 

6: Those who visited Michigan at any time.
7: Those who never visited Michigan.

166

while such a stabilization did not occur in the perception of the ‘nightlife’ attribute. There
is a lack of knowledge and/or perception of the ‘museums’ and ‘nightlife’ attributes of
Michigan among not only non-visitors but also visitors, since the number of respondents
who provided valid ratings for these items drop dramatically in both groups.

The most popular and least popular image attributes of Michigan are the same for
both visitors and non-visitors, ‘scenic’ being the most popular or highest rated (8.54 for
visitors, 7.15 for non-visitors) and ‘museums’ being the least popular or lowest rated
(6.82 for visitors, 5.55 for non-visitors).

Exploratory factor analysis on the 15 image items yielded two factors for the all-
years’ data and the 2001-2002 data, while only one factor for the 1996-1997 data. The
results of the all-years’ data were similar to those of the 2001-2002 data with slight
differences in dimensions’ factor loading sizes and directions. Factor analysis was
applied to visitor and non-visitor groups separately; non-visitor group results were similar
to those of the 1996-1997 data, while visitor group results were similar to those of the all-
years’ data and especially the 2001-2002 data. The results of the all-years’ data were
adopted for the further analyses.

Two factors were extracted with substantial loadings of all 15 image dimensions,
ten dimensions loaded onto Factor I, and ﬁve dimensions loaded onto Factor II. The
factors explained 62.23% of the variation in the original variables. Factors were internally
stable, with substantially high Cronbach’s alphas (a=0.9l for Factor I and a=0.86 for
Factor 11).

Although both factors were combinations of conceptually different image

dimensions and components, Factor I was named ‘The Setting/Sense of Place’ since it

167

included ten cognitive and affective dimensions of a vacation destination that is available,
affordable, comfortable and functional; these attributes were ‘summer rec.’, ‘winter rec.’,
‘scenic’, ‘family place’, ‘safe’, ‘must visit’, ‘popular place’, ‘friendly’, ‘value’, and
‘getaway’. The grand mean of this factor was 7.75, a rather high score, on the 10-point
scale, where 10 means “agree completely”.

Factor 11 was named ‘Activities/Things to do’ since it included dimensions that
people would look for while considering a destination for pleasure trip purposes; these
attributes were ‘nightlife’, ‘museums’, ‘historic’, ‘lodging’, and ‘exciting’. The grand
mean of this factor is 6.93, an above average score, which corresponds to a relatively
high level of agreement on the lO-point scale, but not as high as that of Factor 1.

Michigan is perceived as a close by destination that is good for diﬂ’erent travel
purposes; however, its touristic amenities are perceived as good but not as good as they
might be. This reﬂects the reality of Michigan as a tourist destination; it is a destination
with ample natural resources for outdoor recreation for all seasons, but it lacks big

touristic attractions.

Significance of Demographics of the General Population

The inﬂuence of sociodemographics was estimated for both the general
population8 and the recent visitor group’. For the all-years’ data, the mean age for the
general population model respondents was 41 .40; 88.8% of these respondents were white,

55.8% were female, and 80.7% had above the median income. Michigan respondents

 

8: including both visitors and non-visitors.
9: Those who visited Michigan within the past 12 months.

168

comprised 35.6%, followed by Ohio and Illinois Wisconsin and Ontario, 18.3%, 17.9%,
9.3% and 9.4%, respectively.

When estimated for the Setting/Sense of Place Factor for the all-years’ data, ﬁve
sociodemographic image detemrinants included in this study were found to be
signiﬁcantly inﬂuential on destination image: race-white, gender-female, state-Michigan,
state-Ohio, and state-Ontario. The same sociodemographics were found to be signiﬁcant
when estimated for the 1996-1997 data; however, the only signiﬁcant one was race-white
variable when estimated for the 2001-2002 data. Thus, the sociodemographic variable
consistently found to be signiﬁcant, thus robust in explaining the Setting/Sense of Place
Factor is race-white, more speciﬁcally, whether or not respondents are white. Having a
positive coefﬁcient (B=0.lOS for the all-years’ data), white respondents have a better
perception of the Michigan’s Setting/Sense of Place Factor than do respondents from
other race groups.

A similar dropout pattern was found in the estimation results for the
Activities/Things to do Factor with some differences in the numbers and types of the
inﬂuential variables. When estimated for the all-years’ data, age, gender-female, income-
above-the-median, and state-Illinois were found to be signiﬁcant. Along with these
variables, race-white was also signiﬁcant when estimated for the 1996-1997 data.
However, for the 2001-2002 data, only age and race-white were found to be signiﬁcant.
Thus, the only sociodemographic variable consistently signiﬁcant, thus robust in
explaining Michigan’s Activities/Things to do Factor is respondents’ age. Having a
positive coefficient (B=0.159 for the all-years’ data), older respondents have a better

perception of Michigan’s Activities/Things to do Factor than do younger respondents.

169

It was expected that a combination of those variables found signiﬁcant in
explaining the individual image factors, namely age and race-white, would also be found
signiﬁcant in explaining the Overall Image of Michigan (the mean image score). When
estimated for the Overall Image of Michigan for the all-years’ data, all sociodemographic
variables were found to be signiﬁcant, except for the income-above the median variable.
When estimated for the 1996-1997 data, race-white was also dropped out. For the 2001-
2002 data, only age and race-white were found to be signiﬁcant. Thus, the only
sociodemographic variable consistently signiﬁcant, thus robust in explaining Michigan’s
overall image is age; race-white was not robust. Having a positive coefﬁcient (B=0.1 13
for the all-years’ data), older respondents have a better perception of Michigan’s Overall

Image than do younger respondents.

Significance of Demographics of the Recent Visitor Segment

The demographic proﬁle of the recent visitor segment model respondents was
found to be different from, not only all respondents, but also from all recent visitors in the
all-years’ dataset. Recent visitor segment model respondents were younger on average
(41.95 years of age); 91.2% of them were white, 57.6% of them were female, and 80.5%
of them had above-the-median income. Michigan residents comprised 61.4% of the
recent visitor segment model respondents, followed by Ohio, Illinois, Wisconsin and
Ontario residents, 10.6%, 9.9%, 5.4% and 4.8%, respectively.

When estimated for the Setting/Sense of Place Factor for the all-years’ data, four
sociodemographic variables were signiﬁcant: race-white, income-above-the—median,

state-Illinois, and state-Ontario. For the 1996-1997 data, three of the same variables

170

(race-white, income-above-the-median and state-Ontario) as well as state-Ohio were
found to be signiﬁcant. When estimated for the 2001-2002 data, only one of these
variables was signiﬁcant: race-white. Thus, the only sociodemographic variable
consistently signiﬁcant, thus robust in explaining the Setting/Sense of Place Factor for
the recent visitors was race, more speciﬁcally, whether or not the recent visitor
respondents are white. Having a positive coefﬁcient (B=0.l43 for the all-years’ data),
white recent visitor respondents have a better perception on the Setting/Sense of Place
Factor than do recent visitor respondents from other race groups.

When estimated for the Activities/Things to do Factor for the all-years’ data, age,
gender-female, income-above-the-median, and state-Illinois were found signiﬁcant; for
the 1996-1997 data, age, gender-female and state-Illinois remained signiﬁcant, while for
the 2001-2002 data, only age and state-Illinois were found to be signiﬁcant. Thus, only
two sociodemographic variables, age and state-Illinois, were consistently signiﬁcant, thus
robust in explaining Michigan’s Activities/Things to do Factor for recent visitors. The
coefﬁcient for the age variable is positive (B=0.l78 for the all-years’ data), meaning,
older recent visitor respondents have a better perception of Michigan’s Activities/Things
to do Factor than do younger recent visitor respondents. The coefﬁcient for the state-
Illinois variable, however, was negative (B= —O.108 for the all-years’ data), meaning,
recent visitor respondents from Illinois are more likely to have a worse perception of
Michigan on the Activities/Things to do Factor than do recent visitor respondents from
other states in the study region. The coefﬁcient of the age variable was bigger than that of
the state-Illinois variable, meaning, age is a more inﬂuential determinant than Illinois

residence on explaining Michigan’s Activities/Things to do Factor.

171

According to the previously stated expectation, both age and Illinois residence
should have been found inﬂuential in explaining the overall Image of Michigan. Age,
race-white, gender-female, state-Ohio, and state-Ontario were found to be signiﬁcant for
both the all-years’ and 1996-1997 data, while only age and gender-female remained
signiﬁcant for the 2001-2002 data. Thus, only two sociodemographic variables were
consistently signiﬁcant, thus robust in explaining Michigan’s Overall Image for recent
visitors; age is one of them as expected, but the other is gender-female instead of state-
Illinois. The coefﬁcient for the age variable is greater and positive (B=0.1 17 for the all-
years’ data), meaning, older recent visitor respondents have a better perception of
Michigan’s Overall Image than do younger recent visitor respondents. The coefﬁcient for
the gender-female variable is also positive (B=0.095 for the all-years’ data), meaning,
female recent visitor respondents have a better perception of Michigan’s Overall Image

than do male recent visitor respondents.

Signiﬁcance of Past Travel Behavior of the General Population

Only two past travel behavior variables in the dataset were usable for the general
population model: overall travel experience and prior visitation. Although overall travel
experience was signiﬁcant for the Setting/Sense of Place Factor in the all-years’ data, it
was not for the 1996-1997 or the 2001-2002 data. However, prior visitation was
consistently signiﬁcant across all data periods. Since its coefficient was positive (B=0.262
for the all-years’ data), respondents who have visited Michigan have a better perception
of the Setting/Sense of Place Factor than do non-visitors. For the Activities/Things to do

Factor, only prior visitation was signiﬁcant for the all-years’ and the 1996-1997 data, but '

172

not for the 2001-2002 data. Prior visitation was also consistently signiﬁcant across all
data periods for Michigan’s Overall image, as expected. Since its coefﬁcient was positive
(B=0.288 for the all-years’ data), respondents who have visited Michigan have a better

perception of the Overall Image of Michigan than do non-visitors.

Signiﬁcance of Past Travel Behavior of the Recent Visitor Segment

Eight past travel behavior variables were included in the recent visitor segment
model, most of them related to the respondents’ last visit to Michigan; these variables
were: overall travel experience (the number of trips within the past 12 months), the
frequency of visitation to Michigan (the number of Michigan trips within the past 12
months), whether or not the last visit to Michigan was the most recent trip, the season
(winter, spring and summer) of the last visit to Michigan, the length of stay in Michigan
during the last visit (the number of nights spent), and the number of activities participated
in during the last visit to Michigan.

When estimated for the Setting/Sense of Place Factor, the frequency of visitation
to Michigan and the length of stay during the last visit to Michigan (#of nights spent)
were found to be signiﬁcant for the all-years’ data. For the 1996-1997 data, only the
ﬁequency of visitation to Michigan was signiﬁcant, while, for the 2001-2002 data, none
of these past travel behavior variables was found to be signiﬁcant. For the
Activities/Things to do Factor, only whether or not the last visit to Michigan was the
most recent trip and the number of activities participated during the last visit to Michigan
was signiﬁcant for the all-years’ and the 1996-1997 data; however, none of these past

travel behavior variables was found to be signiﬁcant for the 2001-2002 data either. For

173

the Overall Image of Michigan, the frequency of visitation to Michigan, whether or not
the last visit to Michigan was the most recent trip and the number of activities
participated in during the last visit to Michigan were found to be signiﬁcant for the all-
years’ data. The activities variable dropped out in the 1996-1997 data, while the visitation
ﬁ'equency dropped out in the 2001-2002 data. Only whether or not the last visit to
Michigan was the most recent trip was found to be consistently signiﬁcant across all data
periods. Since its coefficient was positive (B=0.068 for the all-years’ data), respondents
who visited Michigan during their most recent trip have a better perception of the Overall
Image of Michigan than do visitors who visited Michigan during a less recent trip within

the past 12 months.

Significance of the Season of the Survey

For the general population model, all three seasons (winter, spring and summer)
included in the analysis, were found to be signiﬁcant for both of Michigan’s image
factors for the all-years’ and the 2001-2002 data but not for the1996-1997 data. The same
pattern also existed for the recent visitor segment model. For the Overall image of the
general population, survey season-winter and survey season-spring were found to be
signiﬁcant for only the 2001-2002 data. The same seasons were found to be signiﬁcant
for the recent visitor segment in all-years’ data while only survey season-spring was
found to be signiﬁcant for the 2001-2002 data.

As well as the season of the survey, other methodological factors need to be
investigated further. As was mentioned before, the image held by consumers is ﬁltered

through methodological factors while being measured and documented by researchers. In

174

the case of mail or telephone survey technique, maybe only those people who feel
strongly positive or negative about the destination respond to the survey, leaving the
images of neutral people out of the picture.

As was discussed in the results chapter, although the initial sample of this study
was random, the sample included in the ﬁnal model tests runs the risk of being
nonrandom due to data cleaning procedures used in this study. This shows that
researchers deﬁne the nature of the data not only through the data collection techniques
they use, but also through data preparation and analysis procedures they use; therefore,
methodological factors’ impact on destination image need further attention in ﬁxture

research.

Models

Although several determinants were found to be signiﬁcant for different data
periods, estimation results were robust or signiﬁcant across all data periods for only a few
proposed image determinants as summarized in Table 44. Had the data been from a
limited time period, such as only the initial years of the study or the later years of the
study, estimation results would have included more signiﬁcant variables but would not
have necessarily held for other periods. This is an important ﬁnding in terms of the time
validity of research results. Results of image studies might be speciﬁc to the study
periods rather than across an extended period of time. Researchers need to be cautious
while interpreting the results of their research and providing destination authorities with

managerial suggestions such as positioning and promotion.

175

Table 44. Summary of all model test results.

Model [General Population Recent Visitor

 

 

 

 

 

Model Segment Model

Image Factors] F1 F1] or F1 F1] or
Variables I Timanm123123123123123123
Age(A) ssssss ssssss
Race-White(R) 553 538 ssss ss
Gender-Female(G) s s s s s s s s s s s
Income-Above—the—median (I) s s s
State-Illinois (S-IL) s s s s s s s
State-Michigan (S-Ml) s s s s
State-Ohio (S-OH) s s s s s s s
State-Wisconsin (SW) S s
State—Ontario (S-ON) s s s s s s s 3
Overall travel experience (OTE) s
Visitation- Visitor (V) s s s s s s s - - - - - - - - -
Visitation frequency (VF) - - - - - - - - - s s s

Howrecentvisit-Most recent(VR) - - - - - - - - - s s s s s
Visit season-Winter (VS-WI) - - - - - - - - -
Visit season-Spring (VS-SP) - - - - - - - - -
Visit season-Summer (VS-SU) - - - - - - - - -

 

 

 

 

 

 

 

Nights(N) ---------s

Activities(AC) --------- ss 5 5
Survey season-Winter (SS-WI) s s s s s s s s s 3
Survey season-Spring (SS-SP) s s s s s s s s s s 5
Survey season-Summer (SS-SU) s s s s s s s 8

F1: The Setting/Sense of Place. 1: The ﬁlll sample including 1996. i997. 1998, 2001 and 2002.
PH: Activitiesrntingstodo. 2: Initial years ofthe study(l996-I997).

01: Overall Image. 3: Later years ofthe study (2001-2002).

S: Significant in explaining image. -I Not included in the model.

With few differences, the same variables were found to be consistently signiﬁcant
for both the general population model and the recent visitor segment model in explaining
each image factor of Michigan. This could be due to high representation of visitors in the
general population model respondents group. As was mentioned in the demographics
section, 79% of those general population model respondents were also visitors, who
visited Michigan at some point in their lives. Or, it could be due to the robustness of the
estimation results for Michigan’s image determinants, not only over different time
periods but also for both visitors and non-visitors.

Some of the same variables as well as few others were also found to be
consistently signiﬁcant for the Overall Image of Michigan. The inconsistencies between

the results of the Overall Image and the two factors could be due to the following

176

reasons: (1) these determinants may not be stable for different measures of image or (2)
the arithmetic mean of 15 image scores may not be a good measure of the Overall Image.
Maybe respondents should have been asked for their overall image of Michigan; this
needs further investigation in ﬁlture studies.

As depicted in Figure 11, model test results revealed that Michigan’s image has
two factors that can be predicted with a few identiﬁed determinants. In general, the
Setting/ Sense of Place Factor depends on respondents race and whether or not they
visited Michigan and the Activities/Things to do Factor depends on respondents’ age.
However, for the recent visitor segment, also Illinois residence is a determinant of
Michigan’s Activities/Things to do Factor. Michigan’s Overall Image also depends on
age, and prior visitation, as well as gender and the most recent visitation of the recent
visitors segment.

MICHIGAN ’S IMAGE
OVERALL IMAGE
The Setting/Sense of Place Activities/Things to do

if f.

T i
+ +
PRIOR
GENDER
(Female)

VISITATION
(Visitor)
FOR ALL FOR RECENT
VISITORS

 

 

 

 

 

 

 

 

 

 

  
 

ILLINOIS
RESIDENCE

       
   
 

 

 

 

 

 

 

 

 

Note: The signs and sizes ofthe determinants are based on the signs and sizes oftheir standardized betas.

Figure 11. A model of Michigan’s image and its detemiinants.

177

A model of destination image containing only six of the proposed determinants
would be under-speciﬁed; a lot of other possible determinants might be missing in this
model, especially information sources including both commercial advertising and others.
As was discussed in the results chapter, estimation of both models had very small R2
values, especially for the Activities/ Things to do Factor. Promotional information
sourcing from Michigan marketing organizations could be more explanatory in
explaining this factor than any other determinant. Therefore, more of these inﬂuential

factors need to be integrated into these models in future studies.

Implications of Results

Positioning Michigan

Positioning is creating and maintaining a positively distinctive place for the
destination in the minds of target markets, and destination image is an important part of
this endeavor (Goodrich 1978; Calantone, et a1. 1989; Ahmed 1991; Crompton, Fakeye,
& Lue 1992; Gartner 1993; Bramwell & Rawding 1996; Baloglu & Brinberg 1997).
Positioning is an important part of destination marketing due to: (1) increased
competition for the same tourist markets, (2) different tastes and preferences of tourist
markets, and (3) substitutability of destinations due to similarities in destination products.
Therefore, travel destinations must apply different positioning for different markets
taking the characteristics, perceptions and needs of these markets into consideration
(Crompton, Fakeye, & Lue 1992; JOppe, Martin & Waalen 2001).

The ﬁrst step in effective positioning for a destination is assessing its strengths,

weaknesses, similarities, distinctive competencies, and competitive advantages in target

178

markets’ minds in comparison with the competing destinations (Crompton, Fakeye, &
Lue 1992; Baloglu & McCleary 1999). Then, the destination is differentiated in target
markets’ minds by selecting its unique, strong, and important attributes, {and emphasizing
them while communicating to the target markets (Crompton, Fakeye, & Lue 1992;
Baloglu & McCleary 1999).

Michigan’s Setting/ Sense of Place Factor is stronger than its Activities/Things to
do Factor; therefore this factor as a whole can be used in positioning Michigan. However,
since this factor is an accumulation of several attributes, it can cause a blurred image for
Michigan (Aaker & Shansby 1982). Therefore, choosing the strongest attribute within
this factor might result in a better positioning and clearer image for Michigan. The
strongest item in this factor and among all attributes (for both visitors and non-visitors) is
the ‘scenic appeal’ attribute. This attribute should be used with an explicit strategy to
create a ‘mental ﬁx’ in target markets’ minds (Lovelock 1984). Michigan should be
positioned as a destination with scenic beauty and serenity while boosting its touristic
amenities on the side. This strategy could increase Michigan’s ‘popularity’ attribute,

which has been the weakest image attribute during the later years of the study.

Promotion of Michigan

Effective promotion is not only part of effective positioning but also necessary for
effective maintenance of positive destination image. (Hunt 1975; Goodrich 1978; Reilly
1990; Court & Lupton 1997; Murphy 1999). To implement the positioning strategy
deﬁned above, promotional materials should include messages that would create the

desired ‘mental ﬁx’ about the position of Michigan among other destinations. An '

179

example of such a message could be: “Experience Michigan’s scenic beauty and serenity
in action”

Michigan is notorious for its lack of museum and nightlife opportunities. There is
a lack of perception and/or knowledge about these attributes of Michigan among both
visitors and non-visitors since the number of respondents providing ratings for these
attributes drop dramatically in both groups. Promotional materials should apply more
coverage of these attributes showing and describing whatever Michigan has to eliminate
the uncertainty about these attributes. Also, ratings of the ‘getaway’, ‘summer rec.’ and
‘family place’ attributes were lower during the later years of the study; these attributes
need to be highlighted more in the promotional media. The differences between visitors’
and non-visitors’ perceptions are statistically signiﬁcant for all attributes, the difference
being the largest for the ‘family place’ and ‘getaway’ attributes. Testimonial advertising
could be used to improve non-visitors’ image of Michigan, especially ‘family place’ and
‘getaway’ qualities that are unknown to non-visitors.

Not only informative but also persuasive promotion should be targeted to non-
visitors. Since visitation improves Michigan’s image, promotional messages should be
persuasive and ﬁ’amed to induce visitation, using suggestive language along with coupons
for several tourism products in Michigan. Previous visitors should be persuaded to visit
again by directing reminder messages to them.

Since few sociodemographic variables were found to be signiﬁcant in explaining
Michigan’s image, segmentation should be applied to reach and convince target markets
effectively. Youth and males don’t perceive Michigan as exciting as others do; they

should be targeted and persuaded that Michigan is more exciting than they actually think.

180

There seems to be apathy towards Michigan among people from race groups other than
white. In the printed media, other races, including black, Asian, and Hispanic, should be
reﬂected more to gain the attraction of other races for Michigan and to induce more
visitation from these groups.

Visitors from Illinois are more likely to be disappointed with Michigan as a travel
destination. This is understandable since Illinois has a bigger touristic attraction than any
of those in Michigan: Chicago, the windy city that is ﬁlll of touristic amenities. Also
younger people, especially males, look for more action when they are on vacation. This
should be considered, especially when younger people ﬁom Illinois are targeted in
promotion. Promotional messages should be realistic and not inﬂated. Maybe, the idea of
a quick getaway from the hustle and bustle of big cities could be used in promoting to the

urbanized market segment.

Product Improvement in Michigan

It is obvious that Michigan needs to improve its tourism products, especially
touristic activity-related attributes. The ‘museums’ and ‘nightlife’ attributes deﬁnitely
need to get some attention from Michigan’s destination marketing organizations since not
only non-visitors but also visitors don’t rate these attributes highly. Current facilities need
to be improved, and new facilities need to be developed. Also, Michigan’s image of these
attributes could be consolidated by packaging related products and marketing to potential
markets at discounted rates (Tasci, Aziz, & Holecek 2003).

In addition, other tourist attractions need to be developed. The recent addition of

Michigan Adventure in Muskegon, as an alternative to Ohio’s Cedar Point, has not been

181

highly successﬁil possibly due to its high price for its not-so-exciting features (from
personal communication with previous visitors to thisfacility). This adventure park is not
the equal of Cedar Point in quality; so its price needs to be lowered to match its offerings
and to induce more visitation not only from residents but also people from neighboring

states, including the younger population of Illinois.

Limitations
There are several possible limitations that should be noted with regards to this
dissertation. Their source is mainly in the secondary data used in this study, the Michigan
Regional Travel Market Survey. Using this secondary data set posed limitations in terms
of: (1) the nature, consistency, and relevance of the data, (2) potential bias from

sampling, nonresponse and recall, and (4) the lack of generalizability.

The Nature, Consistency and Relevance of the Data

The data used in this study were collected through the Michigan Regional Travel
Market Survey by the Travel Tourism and Recreation Resource Center (TTRRC) at
Michigan State University. During the ﬁrst three years of the study, the main sponsor of
this study was Travel Michigan, Michigan’s official travel bureau. Therefore, the main
goal of the study was to describe Michigan’s prime market characteristics and behaviors
(including both visitors and non-visitors) along with measuring the effectiveness of the
state’s promotional efforts. Therefore, the questionnaire was designed to describe
respondents’ most recent trips, less recent Michigan trips, Michigan image, promotional

responsiveness, and demographic characteristics. Since Travel Michigan was no longer

182

the sponsor of this study after 1999, questions related to Michigan image and promotional
responsiveness variables were deleted from the instrument starting in 1999. However,
during the last two months of 2001 and for all of 2002, the same image items were put
back in the instrument with only slight modiﬁcations.

First, since the questionnaire was designed to serve the purposes of the sponsors
instead of the purposes of this study, the data include general information rather than
speciﬁc information that would best serve accomplishing the purposes of this study.
Second, the deletion of some questions across different time periods made inclusion of
some relevant variables in the regression model impossible. For example, promotion-
related variables could have been useful in the image model; however, they were not used
since they were included in the survey during the initial three years but not later years.

Also, some relevant sociodemographic variables are not included in this dataset.
For example, respondents’ education level might be a factor in their image of a speciﬁc
destination. People with higher levels of education might have been exposed to more
media about Michigan, thus heard or read more about Michigan. They could, therefore,
have a different image of Michigan than that of less educated and informed people. Also,
occupation is another missing sociodemographic variable. Farmers might have a different
image of Michigan than that of business people who might have more travel experience
due to their tendency to travel more, at least for business purposes.

Moreover, no psychographic variables are included in this secondary data. Some
psychographic variables, especially personality type of the respondents, might be
inﬂuential factors in explaining destination image. For example, those people who are

characterized as analytical might have a greater tendency to search for information,

183

gather factual data, and base their responses on what they know rather than just their
feelings as would a creative personality type.

Finally, the nature of the variables in the multiple regression model poses
analytical disadvantages. It is acknowledged that one assumption of the multiple
regression analysis is violated; namely, the use of the categorical rather than continuous
data on several variables in the multiple regression model. Most of the variables included
in the model of this study are categorical, thus had to be transformed into dummy
variables. Some of these variables, such as gender, race, and the state of residence have to
be categorical; however, the income variable could have been continuous which might
have yielded different results in the multiple regression analyses.

If the survey was designed speciﬁcally for the purposes of this study, more
focused, consistent, relevant and richer data could have been extracted. Instead, whatever
relevant data were available from this lengthy survey instrument were used to carry out

this study.

The Potential Bias from Sampling, N onresponse and Recall

First, in the 1996-1998 study period, respondents were not asked the Michigan
image questions if they did not take any trip within the past 12 months. However, in the
2001-2002 study period, Michigan image questions were asked of all respondents.
Although the majority of the respondents (80.8%) of respondents in the initial years’ data
reported that they traveled during the past 12 months, the remaining proportion (19.2%)
is a considerable amount of people considering the sampling scale of this study. A t-test

on the mean image values for these two samples might appear viable to see if there is any

184

bias due to the sampling differences between these two periods. However, the results
cannotbe conclusive since differences in Michigan’s image for these two different time
periods can be attributed to many other factors including the change in Michigan’s image
due to the promotional activities carried by the destination marketing organizations over
the course of this study.

Second, some race groups may have been underrepresented in the general
population model, which was originally designed to assess the determinants of
destination image for the general public. Different race groups were compared on
selected key variables, namely, sociodemographics, past travel behavior, and the image
of Michigan, by using Chi-square and One-way ANOVA tests. Differences between
races were signiﬁcant, which introduces the possibility of sampling bias into the results
of the general p0pulation model developed in this study.

Also, older respondents might have been underrepresented in this study. As was
discussed in the results chapter, the mean age for the image-respondents and model~
respondents subgroups were younger than that of the all respondents. Older respondents
were more likely to terminate the survey or not to respond to all questions asked as was
discussed before. It would be logical to assume that older respondents feel fatigue or
can’t remember; thus, can’t ﬁnish the survey or give invalid responses, such as “I don’t
know” or “I don’t remember”. Sudman and Dradbum (1973) found that older
respondents had trouble in remembering their past experiences. This could be the reason
for older respondents not providing valid responses for some image items, or it could be
due to their discomfort with giving information about themselves over the phone. The

same reason could be true for lower income respondents, who might have been

185

underrepresented in this study since the overwhelming majority (about 80%) of
respondents used in both models had above the median income.

Bias due to nonresponse is also a possibility in this study. As was mentioned in
the methods chapter, the refusal rate of the eligible respondents was about 29%. Although
the check for nonresponse bias revealed few differences between respondents and non-
respondents, nonresponse bias is still a concern in this study since it is impossible to
assure that the sample is representative of the intended population of this study.

Another potential source of bias could be from respondents’ inaccurate recall on
some quantitative variables: the number of trips they took in the past 12 months, the
number of trips they took in Michigan in the past 12 months, and the number of nights
they spent in Michigan. The responses to these variables might contain recall bias since it
is hard to remember the exact numbers of this type, especially for older respondents.
Those who provided valid responses for these variables might have provided approximate
responses, which could reduce the accuracy of the ﬁndings, and thus the nature and

accuracy of the model test results.

Generalizability

Another possible limitation is the lack of generalizability due to the limited nature
of the study region. The sampling ﬂame of this study included Michigan, four
neighboring states (Illinois, Indiana, Ohio, and Wisconsin), and the province of Ontario.
Also the recent visitors group, which represented both the majority of cases included in
the general population model test and all cases included in the recent visitor segment

model test, included only those who visited Michigan within the past 12 months rather '

186

than all visitors. Therefore, the results might be limited to not only these states and
province, but also to recent Michigan travelers who are familiar with Michigan. Also, the
majority of respondents in both models were Michigan residents; thus, the results might
be biased towards residents’ image of their own state.

People from distant states such as Florida might have a totally different image of
Michigan than those from these neighboring states. Also visitors who did not travel to
Michigan within the past 12 moths might be different in image and its determinant
variables. Thus, the results of this study may not be generalizable to all current and
potential tourist markets of Michigan. Since the distance to a destination is found to be an
important factor in destination image, people from distant states might have distinctively
different images of Michigan than those of residence of neighboring states. Furthermore,
the results of this study are probably limited to Michigan’s image rather than other states

or cities with various unique properties and unique market characteristics.

Suggestions for Future Research

This study has once again shown that destination image is a rather complex
human phenomenon with various aspects in need of ﬁirther research. Based on the above
discussion, several ﬁlture research suggestions are provided. First, the ﬁndings of this
study can be double-checked by utilizing a different survey mode (e. g., mail survey,
face-to-face interview). As was mentioned before, there is a possibility of bias due to the
low response rate, the high nonresponse rate as well as the high partial nonresponse from
older people, people from race groups other than white, lower income respondents, and

nonresident respondents. The reason could reside in methodological factors. Certain types

187

of people, such as older and Hispanic may still be uncomfortable with providing
information about themselves over the phone. In general, the effectiveness of the
telephone interviewing technique has been diminishing due to the increasing use of fax
machines, answering machines, caller IDs, unknown call zappers, and the increasing
telemarketing activities. People become more leery of talking to an unknown person on
the phone due to the fear of becoming a 'subject to a scam, which has been the feeding
force behind the efforts for creating a national no-call list (National Do Not Call Registry
2003) to avoid unknown callers.

For this reason, other survey modes could be used to check for the stability of the
ﬁndings of this study. Mail surveys could be used to check the general model ﬁndings;
Michigan visitors can be intercepted at various welcome centers in Michigan to conduct
face-face surveys in an effort to check the recent visitor segment model ﬁndings. Also,
the proliferation of computers and the Internet opens up more convenient, easier, cheaper
and faster data collection modes for the researchers (Tasci & Knutson 2003). E-mail or
Internet surveys can also be used to conduct similar surveys. Using different survey
modes would also enable more investigation of the impact of methodological factors on
destination image.

Second, it is recommended that a shorter and more concise survey instrument be
designed to alleviate the problems of low response rate, high nonresponse, and high
partial nonresponse rate. The original questionnaire utilized in the Michigan Regional
Travel Market Survey was rather long, comprised of 140 questions on average. Although
people rarely were asked all questions due to the skip pattern in the instrument, an

interview took about 12 minutes on average to complete, extending to 20 minutes in

188

cases where a respondent was asked all the questions. Respondents usually asked for the
amount of time the survey would take. The interviewers were informed to respond in a
non-deﬁnitive way due to the skip pattern; they were allowed to say 10-15 minutes
depending on answers if the respondent insisted on getting a time estimate. Interviewers’
hesitance in giving time estimate and also 10-15 minutes might have turned off some
respondents at the beginning of the survey. Also, some respondents might have
terminated the call or hung up during the interview as they realized the survey was taking
more time than they expected.

Also, since the survey instnlment was not speciﬁcally designed for the purposes
of this study, several questions in the survey were not directly related to this study.
Therefore, the survey instrument should be designed to contain only related variables to
provide more speciﬁc information for the purposes of this study: Michigan image,
relevant past travel behavior variables, and sociodemographics. In addition, for a more
comprehensive investigation of the determinants of destination image, other relevant
variables, such as respondents’ education level, occupation, and psychographic variables
could also be included in a new survey instrument. Psychographics, such as the type of
personality, the level of dogmatism and conservatism, lifestyle, value system, and status
concerns have been found to determine the image of products in relation to their country
of origin (Bilkey & Nes 1982). These factors could very well also inﬂuence destination
image formation. Moreover, the promotion responsiveness variables that were used
between 1996-1998 and deleted in later years could also be included in this new survey
instrument to enable the test of the inﬂuence of promotion on Michigan’s image for two

different data periods.

189

For better regression results, it is recommended to use continuous variables where
possible. In this study, the majority of the variables in the regression models were
categorical, posing analytical disadvantages by violating the assumption of the use of
continuous data in regression analysis. Some of those variables, namely, gender, race,
and the residence state have to be categorical; however, the income variable could be
continuous which might yield different results in the multiple regression analyses.

Third, the models tested in this study can also be tested for other sampling origins
as well as for other study destinations. In other words, similar studies could be conducted
to measure Michigan’s image and its determinants for regions other than Midwestem
states and also other destinations’ images and their determinants. As was mentioned in
the limitations section, the ﬁndings of this study could be speciﬁc to these states rather
than generalizable across different regions in the US. Therefore, conducting similar
studies in other states, especially distant states, such as Florida and California is
recommended to see if the ﬁndings for Michigan hold across different regions. Also
similar studies could be conducted for other states to determine if image determinants are
similar or different for different states. Image determinants for a state more popular than
Michigan could be different from those of Michigan.

Fourth, ﬁthher analyses can be conducted on the data utilized in this study. Both
the general population and the recent visitor segment models can be tested by using the
individual image dimensions as the dependent variables. Doing so could reveal different
determinants as signiﬁcantly inﬂuential for different dimensions. For example, income
could be signiﬁcant in explaining the ‘Michigan offers an excellent vacation value for the

money’ dimension. In addition, the same analyses can be run by weighting the data I

190

towards other important sociodemographic variables. The data was weighted by the state
variable in this study to adjust the sample to the actual population. The data could be
weighted by age, income, and race to adjust the sample to the actual population and test if
the same analyses would result in different ﬁndings. In addition, segmentation could be
applied to the data to check for the compatibility between the segments created and the
segments of PRIZM, lifestyle segmentation system used by Travel Michigan (Claritas
2003). In case of high compatibility, PRIZM segments can be used for effective image
management activities.

A ﬁnal recommendation for future research is qualitative investigation of the
underlying reasons of the signiﬁcant impact of visitation on destination image. Visitors
provided a more positive image of Michigan across all attributes assessed in this study.
One explanation could be that people have unrealistically less positive images of
Michigan until they visit; once they visit, their images improve. However, one can
logically expect that as well as unrealistically negative images, people could also develop
unrealistically positive images about places due to romantic feelings stimulated by
promotional materials. Therefore, there could be other underlying reasons to explain the
consistent positive inﬂuence of visitation in this study. These reasons could be residing in
psychological or socio-psychological mysteries of human beings, which could be
explained through different theories in related ﬁelds of research. Gaining a better
understanding of such human phenomena requires a scrutiny of the subject matter, which
is more feasible using qualitative methods, such as focus groups, in-dept interviewing,

observation, and photo interpretation. Therefore, qualitative methodologies should be

191

used in ﬁxture studies to investigate the pertinence and applicability of theories in

explaining visitation’s signiﬁcant impact on destination image.

192

APPENDIX A.
19964998 QUESTIONNAIRE

193

FINAL YEAR 1 CERTEC/SAPMINR PHONE SURVEY QUESTIONNAIRE 11/13/95

[ENTER DATE OF INTERVIEW]
Month>_ Day>__ Year> l9_
Hello, my name is . I'm calling from Michigan State University. We're conducting a study to learn how often people in the

Midwest and Canada take trips. Your household was randomly selected to represent your community. We'd greatly appreciate your
help in answering a few questions about trips you've made. May I speak to the adult over 17 years old who will have the next
birthday? [IF THIS PERSON IS NOT AT HOME, ASK TO SPEAK TO THE ADULT AT HOME WHO WILL HAVE THE NEXT
BIRTHDAY]

We're defining a "trip" as any overnight or day trip to a place at least 50 miles from your home. unless it was taken in commuting to
work or school.

[RECORD GENDER OF RESPONDENT] > __ M=Male F=Female -99=Can‘t determine

[DOUBLE ENTRY REQUIRED]
1. Have you taken any kind oftrip in the past 12 months? > _
l=Yes 2=No —> GO TO QUESTION 130 -99=DK/NR ——-- GO TO QUESTION 130

BEGIN INTRODUCTORY BLOCK
[READ OPTIONS 14'. IF NECESSARY, PROBE FOR PRIMARY PURPOSE OF TRIP]
2. Was your most recent trip primarily for the purpose of... > _
I=Visiting friends or relatives; 2=Recreation; 3=Business; or 4=Some other purpose? -—> ASK QUESTION 3 -99=DK/NR

3. And what would that purpose be? >

 

We're defining a "pleasure trip” as any overnight or day trip to a place at least 50 miles from your home that was made for your
enjoyment. including vacations. weekend getaways. shopping trips, and trips to visit friends or relatives.

4. Have you taken a pleasure trip to Illinois in the past 3 years?

[CONTINUE FOR EACH STATE/PROVINCE: ”How about ?] l=Yes 2=No -99=DK/NR
Illinois > _ Ohio > _ Indiana ,T- __ Wisconsin ’- _ Michigan > _ Ontario 3" __ Minnesota ‘> __
[DO NOT READ LIST]
5. During the next 12 months, do you expect to take more. fewer. or about the same number of pleasure trips as you did during the
previous 12 months? > __ 1=More 2=Fewer 3= Same -99=DK/NR

[DO NOT READ LIST; ACCEPT UP TO 3 RESPONSES]
6. Where do you turn most often when you need information to help plan a pleasure trip? 2-

ORGANIZATIONS OTHER
1=Chamber of commerce 10=Friendslrelativeefco-workers
2=Convention/visitors bureau l 1=CD~ROM
3=State travel office/ call state 800 number 12=Highway welcome centers
4=Travel agency l3=1ntemet/ on—line service

l4=Travel show

PUBLICATIONS 15=Other source
5=Magazine(s) l6=No source(s)
6=Travel section of newspaper -99=DK/N R

7=Mobil Travel Guide
8=AAA/CAA/auto club publications
9=Other travel guide

[READ OPTIONS 1-4]
7. Which one of the following media has been most helpful to you in selecting the destinations you have visited on pleasure trips?
1=Magazines; 2=Newspapers: 3=Television; or 4=Radio? -99=DK/NR

8. How would you rate the desirability of Illinois as a pleasure trip destination on a scale from 1 to 10. where 1 means "not at all
desirable" and 10 means "very desirable?"
[REPEAT FOR EACH REMAINING STATE/PROVINCE: ”How about ?"]
STATE! RATING STATE’ RATING -99=DK/NR
PROVINCE [l-lO] PROVINCE [1-10]
Illinois >_Ohio >_ Indiana > _Wisconsin S> __ Florida 1’ __ Ontario > _ Michigan > __Colorado> _Minnesota > __

 

[END INTRODUCTORY BLOCK]
[BEGIN PROMOTIONAL AWARENESS AND RESPONSE BLOCK]

9. In the past 12 months, have you seen or heard any advertisements promoting travel to any destinations? >- _
l=Yes 2=No -—> GO TO QUESTION 16 -99=DK/NR --> GO TO QUESTION 16

[ENTER UP TO 5 PLACES: PROBE FOR STATES ASSOCIATED WITH UNCOMMON PLACES; PROBE: Any other places?]
10. What places have you seen or heard ads for?

194

 

 

 

VVVV

 

V

 

[DON'T READ]
l=Michigan or a place in Michigan mentioned 2=Only non-Michigan places mentioned -——> GO TO QUESTION 16
-99=DK/NR —-> GO TO QUESTION 16 > _

11. On a scale from I to 10, where 1 means "poor" and 10 means "excellent," how would you rate the quality of the Michigan ads
you‘ve seen or heard? > ______ -99=DK/N R

[DO NOT READ LIST; PROBE TO FIT A CATEGORY]

12. Where did you see or hear the most recent ad promoting travel to Michigan? __ ~99=DK/N R
1=TV 9=Direct mail advertisement
2=Radio 10= Internet/'on-line service
3=Newspaper l l=CD-ROM
4=Magazine l2=Chamber of commerce
5=Billboard/outdoors 13=Convention and visitors bureau
6=Travel agent l4=Highway welcome center
7=Travel show 15=At the destination
8=Travel guide 16=Other

13. Did this ad promote travel to a speciﬁc destination in Michigan or travel to Michigan in general? ’2 _
l=Travel to a speciﬁc destination in Michigan 2=Travel to Michigan in general -99=DK/NR

14. Did the ad provide a toll-ﬁ'ee number that people could call to request further information? > _ l=Yes 2=No -99=DK/NR
15. Did you contact the organization that sponsored this ad to request additional travel information? > _1=Yes 2=No -99=DK/NR

16. Do you recall any of the slogans that are used to promote travel to any states or Canadian provinces? > _
l=Yes 2=No —-> GO TO QUESTION 19 -99=DK/NR —-> GO TO QUESTION 19

[DO NOT READ LIST: ACCEPT UP TO 3 RESPONSES]
17. Which slogans do you recall? > _ __ __

l=Illinois: "Illinois. Don‘t Miss It!"

2=Indiana: "You Could Use A Little Indiana"

3=Indiana: "Wander Indiana"D

4=Kentuckyz "Kentucky...What You've Been Looking For"

5=Minnesota: "Explore Minnesota"

6=Michiganz "Say Yes to Michigan" [REMEMBER IF THIS IS MENTIONED]

7=Michigan: ”Yes Michigan” [REMEMBER IF THIS IS MENTIONED]
=New York: "I Love New York"

9=Ohio: "Ohio...The Heart of It All"

10=Ontarioz "Discover Ontario"

ll=0ntario: ”Ontario: Yours to Discover"

l2=Wisconsin2 "Escape to Wisconsin"

13=Wisconsinz "You‘re Among Friends"

 

14=Other
18. Other >

[DON'T ASK IF MICHIGAN SLOGAN(S) WERE MENTIONED IN RESPONSE TO ABOVE QUESTION]
19. Have you ever heard the slogan, "Say Yes to Michigan" or "Yes Michigan"? > _ l=Yes 2=No -99=DK/'NR
20. Have you ever heard the slogan, "Michigan: A Destination for All Seasons"? T> _ l=Yes 2=No -99=DK/NR

21. During the past 12 months. have you called any state or province's toll-free number to request travel information? y. _
l=Yes 2=No --> GO TO QUESTION 23 -99=DK./N R --> GO TO QUESTION 23

[ENTER ALL STATES/PROVINCES MENTIONED: PROBE: Any others?]
22. What states' or provinces' toll-free numbers have you called?
[DON'T READ] l=Michigan mentioned .... GO TO QUESTION 24 2=Michigan not mentioned ‘2’ __

 

23. Do you know if the State of Michigan has a toll-free number you can call to obtain information on travel in Michigan? 2 __
l=Yes 2=No -99=DK/NR

END PROMOTIONAL AWARENESS AND RESPONSE BLOCK

BEGIN MICHIGAN IMAGE BLOCK
[PROBE: What others come to mind?; ACCEPT up To 3 RESPONSES]

195

24. When you think of Michigan as a pleasure trip destination, what positive impressions, if any. come to mind?

>

 

V

 

V

 

[PROBE: What others come to mind?; ACCEPT UP TO 3 RESPONSES]
25. And what negative impressions, if any, come to mind?
>
>
>
[ACCEPT UP TO 3 RESPONSES]
26. What. if any, recreation activities or facilities do you feel are missing in Michigan?
>
>
>
[ACCEPT UP TO 3 RESPONSES]
27. What types of winter recreation opportunities do you feel Michigan is known for?

>

 

 

 

 

 

 

 

‘
I)

 

\.
I)

 

We'd like to know how much you agree or disagree with some statements about Michigan. Please use a scale from 1 to 10. where 1
means you "do not agree at all" and 10 means you "agree completely." Michigan. . . -99=DK/NR

28.1s close enough for a weekend getaway .............
29. Has many interesting museums ...................... -. _
30. Is great for summer outdoor recreation activities

—n_

31 Is an exciting place to visit. .................... > _

32. Has a lot of high quality lodging ................. > __

33. Offers much scenic appeal .3-

34. Is greatforwinteroutdoorrccreation activities—. >_

35.15 a good place to meet ﬁiendly people ........... ,5

36.1s a place everyone should visit at least once in their lifetime ............... _
37. Isasafeplacetovisit ......................... >

38. Offers exciting nightlife and entertainment. ...... > _
39.15 a great place for a family vacation ............ __
40. Is a popular destination with vacationers.........>
41. Has many interesting historic sites ............... ">
42. Offers an excellent vacation value for the money. ‘ _

END MICHIGAN IMAGE BLOCK

Now we'd like to ask you about pleasure trips that you may have taken. Again. we're deﬁning "pleasure trips" as any overnight or day
trips to places at least 50 miles from your home that were made for your enjoyment. including vacations, weekend getaways. shopping
trips. and trips to visit frienck or relatives.

[DOUBLE ENTRY REQUIRED]
43.1n the past 12 months, have you taken any pleasure trips to any destination?
1 -Yes 2= No --> GO TO QUESTION 79 -99= DK/NR ---- GO TO QUESTION 79

(ACCEPT 1-999)
44. About how many pleasure trips have you taken in the past 12 months? > _

[IF RESPONDENT IS UNABLE TO GIVE A SPECIFIC NUMBER. PROBE:]

In the past 12 months. would you say you've taken. . .
2=l to 3 pleasure trips? S=4 to 6 pleasure trips? 8=7 to 9 pleasure trips? 15=10 to 20 pleasure trips?
25=More than 20 pleasure trips? -99=DK/NR

[NOTE1 USE CODES ONLY IF RESPONDENT DOESN'T GIVE A SPECIFIC RESPONSE]

BEGIN CULTURAL HERITAGE TOURISM BLOCK

45. Did you visit any museums, halls of fame. or historic sites on any of the pleasure trips you took in the past 12 months? > _
l=Yes 2=No -—-> GO TO QUESTION 52 ~99=DK/NR --2 GO TO QUESTION 52

46. Were any of these located in Michigan? “‘
l= Yes 2= No --> GO TO QUESTION 50 -99=DK/N R --> GO TO QUESTION 52

[PROBE TO FIT CATEGORIES: ACCEPT UP TO 5 RESPONSES]
47. What types of museums. halls of fame. or historic sites did you visit in Michigan? >
MUSEUMS/HALLS OF FAME HISTORIC SITES
l=Art museum 8=Battleﬁeld 15=Home
2=Children's museum 9=Bridge l6=Lighthouse

196

3=Hall of Fame 10=Cemetery l7=Ship

4=Historical museum 1 1=Church 18=Town
5=Maritime museum 12=F arm l9=Underwater preserve
6=Natural history museum l3=Fishery 20=Other
7=Science museum 14=Fort ~99=DK/NR
48. Other ‘2

 

49. On these pleasure trips, did you visit any museums. halls of fame, or historic sites in any other states or countries? > _
l=Yes 2=No --> GO TO QUESTION 52 -99=DK/NR -—2 GO TO QUESTION 52

[PROBE TO FIT CATEGORIES; ACCEPT UP TO 5 RESPONSES]
50. What types of museums. halls of fame, or historic sites did you visit in other states or countries? 7* __ __ _ __ _
MUSEUMS/HALLS OF FAME HISTORIC SITES

l=Art museum 8=Battlefleld 15=Home
2=Children's museum 9=Bridge 16=Lighthouse
3=Hall of Fame 10=Cemetery l7=Ship
4=Historical museum 11=Church 18=Town
5=Maritime museum 12=Farm l9=Underwater preserve
6=Natural history museum 13=Fishery 20=Other
7=Science museum l4= Fort -99= DK/N R
51. Other >

 

END CULTURAL HERITAGE TOURISM BLOCK
BEGIN BASIC PLEASURE TRIP PROFILE BLOCK

52. Now I‘d like to ask you about your most recent pleasure trip.

[PROBE FOR MONTH AND DAY: ENTER NUMERICAL VALUES FOR MONTH AND DAY: IF NECESSARY. PROBE

FOR BEST GUESS OF DAY]
Approximately when did this trip begin - the month and day?
MONTH CODES
l=January 4=April 7=July 10=October

2=February 5=May 8=August I 1=November
3=March 6=June 9=September 12=December
MONTH > __ DAY > _ -99=DK/NR

[ACCEPT UP TO 3 RESPONSES: PROBE FOR SPECIFIC PURPOSE(S), ESPECIALLY 1F RESPONDENT SAYS
"VACATION"]
53. What was the purpose or purposes ofthis trip?

)
)

 

 

 

[ASK 1F MORE THAN 1 PURPOSE MENTIONED; PROBE FOR SPECIFIC PURPOSE. ESPECIALLY 1F RESPONDENT
SAYS "VACATION"]

54. What would you say was the primary purpose of this trip? >
[IF RESPONDENT WAS ON A GROUP TOUR. PROBE FOR SIZE OF IMMEDIATE TRAVEL
PARTY AS OPPOSED TO SIZE OF ENTIRE GROUP] (ACCEPT 1-99)

 

55. How many persons. including yourself. were in your immediate travel party? __

[IF NECESSARY. PROBE FOR BEST GUESS OF AGE]
56. Beginning with yourself. please give me the gender and age of each person who went on this trip:
M=MALE F=FEMALE -55=REFUSED -99=DK/NR

GENDER AGE GENDER AGE
RESPONDENT s. _ __ PERSON #2 __ __
PERSON #3 _ > __ PERSON #4 > __ __
PERSON #5 ;- __ r» _ PERSON #6 _ _
PERSON #7 > __ ,- __ PERSON #8 _ _
PERSON #9 2 __ r- __ PERSON #10 2 __ __

57. Did your immediate travel party consist of family members only? ;.- _ l=Yes 2=No -99=DK/N R

58. Was this an overnight or day trip? i> _
l=Overnight 2=Day trip --> GO TO QUESTION 63 -99=Dl\’/NR --.-‘- GO TO QUESTION 63

(ACCEPT 1.999)

59. How many nights were you away from home? __ -99=DK/NR
(ACCEPT 0999; IF 0, SKIP NEXT QUESTION)

197

60. How many nights were spent in the state containing the main destination of this trip? ?- __ -99=DK/N R
(ACCEPT 0-999)

61. While you were in the state mntaining the main destination of this trip, about how much. if anything. did you spend per night on
lodging in hotels, motels, Bed & Breakfasts. or rental cabins? > $_ -99=DK/NR

[DO NOT READ LIST UNLESS NECESSARY TO STIMULATE RESPONSES]
62. What was the main type of lodging you used? ‘2 __ -99=DK/N R
1=Friend or relative's home 2=Hotel, motel. or lodge 3=Bed & Breakfast 4=Rented cabin. cottage. or condominium
5=Owned cabin, cottage, or condominium 6=County, state. or federal campground 7=Commercial campground (e.g.. KOA)
8=Boat/ship 9=Other
[READ LIST]

63. Which. If any, of the following recreation activities did you participate in? l=Yes 2=No -99=DK/NR
Attend a festival or event? ......................... , _
Shopping? ........................................... _
Casino gaming? ......................................
Nightlife? . ......................................... : _
Visit a museum or hall of fame? ..................... I> _
Visit an historic site? ............................. > _
Visit some other type of attraction? ................ >
Fall color touring outside of traveling to and from your destination? .................. :-
General touring or driving for pleasure? ............. -
Outdoor recreation? >

.—

——_.—

 

[ACCEPT UP TO 5 RESPONSES] (ASK IF OUTDOOR RECREATION AFFIRMED ABOVE)
64. What outdoor recreation activities did you participate in?

 

VV

 

 

v‘vv

 

 

END BASIC PLEASURE TRIP PROFILE BLOCK

[IF NECESSARY, PROBE FOR CITY/PLACE FARTHEST FROM HOME]
65. What was the main destination ofthis trip?
City/Place: 3'
State/Province/Country: »
[DON'T READ: DOUBLE ENTRY REQUIRED]
l=Michigan destination 2=Non-Michigan destination - > GO TO QUESTION 78

 

 

BEGIN SUPPLEMENTAL MICHIGAN PLEASURE TRIP PROFILE BLOCK

[USE NAME OF DESTINATION FROM ABOVE QUESTION IN BLANK] (ACCEPT 50-9999)
66. About how many miles did you travel to get to ? > miles -99=DK/NR

 

[ASK IF AT LEAST 1 RESPONSE WAS GIVEN TO QUESTION 64'. USE IST SUCH RESPONSE IN BLANK]

67. How would you rate the quality of Michigan's opportunities on a scale from 1 to 10. where 1 means "poor" and 10
means "excellent"? > _ -99=DK/NR
(ACCEPT 0-999999)

68. What would be your best estimate of how much your immediate travel party spent altogether on this trip while in Michigan? >
S -99=DK/NR

69. Was this a vacation trip? _ l=Yes 2=No -99=DK/NR

[ENTER RESPONSE. E.G.. 90 DAYS. 2 WEEKS. 3 MONTHS]
70. About how far in advance of this trip did you make a ﬁnal decision about where to go?

 

71. Were any of the travel arrangements for this trip made by a travel agent? > __ l=Yes 2=No -99=DK/NR
72. For this trip, did you purchase a package. for which you paid one price, that included at least one night of lodging? > _
l=Yes 2=No -99=DK/NR
[DO NOT READ LIST; ACCEPT UP TO 3 RESPONSES]
73. What types of transportation did you use? ‘ 1=Car/truck without camping equipment 2=Car/truck with camping
equipment 3= Self-contained recreation vehicle 4= Rental car 5= -Airplane 6=Train 7=Ship/boat 8= —Motorcycle

9= ~Bicycle 10= Motorcoacthus -— ASK QUESTION 75 11=0ther --- ENTER UNDER QUESTION 74 -99=DK/N R

198

74. Other >
75. Was this a motorcoach tour? > _ l=Yes 2=No -99=DK/NR
76. What did you most enjoy about this trip? :-
77. And what did you least enjoy about this trip?

 

 

 

GO TO QUESTION 107
END SUPPLEMENTAL MICHIGAN PLEASURE TRIP PROFILE BLOCK
[DOUBLE ENTRY REQUIRED]
78. Was a place in Michigan the main destination of any of the pleasure trips you‘ve taken in the past 12 months? > ____
l=Yes --> GO TO QUESTION 81 2=No -99=DK/NR

79. Have you ever taken a pleasure trip to a place in Michigan? > __ l=Yes 2=No —->- GO TO QUESTION 12]

[PROBE FOR YEAR: ENTER LAST TWO DIGITS OF YEAR]
80. When was the last time you took a pleasure trip to a place in Michigan? 3* l9_ -99=DK/N R
00 TO QUESTION 121

BEGIN FULL MICHIGAN PLEASURE TRIP PROFILE BLOCK

[IF NECESSARY. EXPLAIN THAT WE NEED A PROFILE OF THEIR MOST RECENT PLEASURE TRIP IN MICHIGAN AS
WELL AS THEIR MOST RECENT PLEASURE TRIP IN GENERAL]

81. Now I'd like to ask you about your most recent pleasure trip in Michigan.
[PROBE FOR MONTH AND DAY: ENTER NUMERICAL VALUES FOR MONTH AND DAY: IF NECESSARY. PROBE FOR
BEST GUESS OF DAY]

Approximately when did this trip begin - the month and day?

MONTH CODES

l=January 4=April 7=July 10=October

2=February 5=May 8=August l 1=November

3=March 6=June 9=September 12=December

MONTH > __ DAY > _ ~99=DK/NR

[ACCEPT UP TO 3 RESPONSES: PROBE FOR SPECIFIC PURPOSE(S), ESPECIALLY 1F RESPONDENT SAYS "VACATION"]
82. What was the purpose or purposes of this trip?

V

 

V

 

>
[ASK IF MORE THAN 1 PURPOSE MENTIONED: PROBE FOR SPECIFIC PURPOSE. ESPECIALLY IF RESPONDENT
SAYS "VACATION"]

 

83. What would you say was the primary purpose of this trip?

 

[IF RESPONDENT WAS ON A GROUP TOUR. PROBE FOR SIZE OF IMMEDIATE TRAVEL PARTY AS OPPOSED TO SIZE
OF ENTIRE GROUP] (ACCEPT 1-99)

84. How many persons, including yourself. were in your immediate travel party? > __
[IF NECESSARY, PROBE FOR BEST GUESS OF AGE]

85. Beginning with yourself. please give me the gender and age of each person who went on this trip:
M=MALE F=FEMALE -55=REFUSED ~99=DK/NR

GENDER AGE GENDER AGE
RESPONDENT 5 _ 1“» __ PERSON #2 >~ __ _
PERSON #3 > _ > _ PERSON #4 ‘2 _ > _
PERSON #5 > _ > _ PERSON #6 i‘r- _ > _
PERSON #7 :> _ > __ PERSON #8 _ _
PERSON #9 > _ > _ PERSON #10 >- _ _
86. Did your immediate travel party consist of family members only? 2 __ l=Yes 2=No ~99=DKINR

87. Was this an overnight or day trip? 2' __
l=0vemight 2=Day trip -—‘> GO TO QUESTION 92 -99=DK/NR -—-> GO TO QUESTION 92

(ACCEPT 1-999)

88. How many nights were you away from home? _ -99=DK/NR
(ACCEPT 0-999; IF 0. SKIP NEXT QUESTION)

89. How many nights were Spent in Michigan? ___ -99=DK/N R
(ACCEPT 0-999)

199

90. While in Michigan, about how much. if anything, did you spend per night on lodging in hotels. motels, Bed & Breakfasts, or rental
cabins? > $__ -99=DK/NR

[DO NOT READ LIST UNLESS NECESSARY TO STIMULATE RESPONSES]
91. What was the main type of lodging you used? 1“ __ 1=Friend's or relative's home 2=Hotel. motel. or lodge 3=Bed &
Breakfast 4=Rented cabin, cottage, or condominium 5=Owned cabin. cottage. or condominium 6=County, state, or federal
campground 7=Commercial campground (cg, KOA) 8=Boat/ship 9=Other -99=DKfN R

[READ LIST]
92. Which, if any, of the following recreation activities did you participate in while you were in Michigan?
l=Yes 2=No -99=DK/N R

Attend a festival or event? ......................... 2-
Shopping? ...... __
Casino gaming? ....... >-
Nightlife? L" __
Visit a museum or hall of fame? ..................... > _
V lsit an historic site? ............................. 2' _
Visit some other type of attraction? ................ ‘
Fall color touring outside of traveling to and from— your destination? ................. :-
General touring or driving for pleasure? ............ -> _
Outdoor recreation? .>

 

 

 

 

[ACCEPT UP TO 5 RESPONSES] (ASK IF OUTDOOR RECREATION AFFIRMED ABOVE)
93. What outdoor recreation activities did you participate in while you were in Michigan?

>

 

 

 

 

 

[USE lST RECREATION ACTIVITY MENTIONED IN RESPONSE TO QUESTION 93 ABOVE IN BLANK] (ASK ONLY IF
OUTDOOR RECREATION AFFIRMED IN QUESTION 92)
94. How would you rate the quality of Michigan's opportunities on a scale from 1 to 10. where 1 means "poor" and 10
means "excellent"? > _ -99=DK/NR

[IF NECESSARY, PROBE FOR CITY/PLACE FARTHEST FROM HOME]
95. What was the main destination of this trip? City/Place in Michigan: "

 

[USE NAME OF DESTINATION FROM ABOVE QUESTION IN BLANK] (ACCEPT 50-9999)
96. About how many miles did you travel to get to . miles -99=DK/NR

 

(ACCEPT 0-999999)
97. What would be your best estimate of how much your immediate travel party spent altogether on this trip while in Michigan? >
S -99=DK/NR

98. Was this a vacation trip? > __ l=Yes 2=No -99=DK/NR

[ENTER RESPONSE. E.G.. 90 DAYS. 2 WEEKS. 3 MONTHS]
99. About how far in advance of this trip did you make a final decision about where to go?

 

100. Were any of the travel arrangements for this trip made by a travel agent? I~ _ l=Yes 2=No -99=DK/NR

101. For this trip. did you purchase a package. for which you paid one price. that included at least one night of lodging? ‘1‘ _
l=Yes 2=No -99=DK/NR

[DO NOT READ LIST; ACCEPT UP TO 3 RESPONSES]
102. What types of transportation did you use? > __ l=Car/truck without camping equipment 2=Car/truck with
camping equipment 3= Self-contained recreation vehicle _4=Rental car 5=Airplane 6=Train 7=Ship or boat 8=Motorcycle
9= —Bicycle 10= Motorcoach/Bus ...- ASK QUESTION 104 ll=Other -—- ENTER UNDER QUESTION 103 -99=DK/NR

103. Other >
104. Was this a motorcoach tour? L- _ l=Yes 2=No -99=DK/NR
105. What did you most enjoy about this trip?

106. And what did you least enjoy about this trip?

 

 

 

END FULL MICHIGAN PLEASURE TRIP PROFILE BLOCK
BEGIN INFLUENCE BLOCK

107. Before you left home for this most recent pleasure trip in Michigan. did you see or hear any advertisements about travel in
Michigan? > __ l=Yes 2=No -->- GO TO QUESTION 120 -99=DK/NR ...::.~ 00 TO QUESTION 120

200

108. Did you see orhear I ad ormorethan 1 ad about travel In Michigan? >
1= 1 ad 2= More than I ad --—> [USE THE PHRASE "THESE ADS" RATHER THAN "THIS AD" IN QUESTIONS IN
THIS SECTION] -99= DK/NR

[DO NOT READ LIST: PROBE FOR ANSWERS]

109. Where did you see or hear this (these) ad(s) about travel in Michigan? > __. -99=DK/N R
I=TV 8=Direct mail advertisement
2=Radio 9=lnternet/on-line service
3=Newspaper lO=CD—ROM
4=Magazine 1 l=Chamber of commerce
5=Billboard/outdoors 12=Convention and visitors bureau
6=Travel agent 13=Highway welcome center
7=Travel show 14=Other

1 10. Did this (these) ad(s) have no inﬂuence. a partial inﬂuence, or a primary inﬂuence on your decision to travel in Michigan? '> _
I=No inﬂuence 2=Partial inﬂuence 3=Primary inﬂuence -99=DK/NR

1 11. Did this (these) ad(s) promote travel to a speciﬁc destination In Michigan or travel to Michigan In genera17>
I =Travel to a speciﬁc destination In Michigan 2= Travel to Michigan In general —-- GO TO QUESTION 113
-99= DK/NR --> GO TO QUESTION 113

112. Which destination in Michigan? >

 

I 13. Did the (these) ad(s) include the Michigan travel slogan?)
l =Yes 2= No --> GO TO QUESTION 115 -99= DK/N R -- GO TO QUESTION 115

1 14. What do you remember the slogan to be? > __ 1="Say Yes to Michigan" 2="Yes Michigan" 3=Other -99=DK/NR

1 15. Did you contact the organization that spomored this (these) ad(s) to request additional travel information? > _
l=Yes --> GO TO QUESTION 1 I8 2=No ~99=DKINR

116. Did you contact any other organization to obtain travel information about Michigan?
I— TYes 2= No --> GO TO QUESTION 120 -99= DK/NR -—> GO TO 01 IE—STION 120

I 17. What organization did you contact? >

 

l 18. Did you receive the information you requested before you left home for your trip?
1T Yes 2= No --> GO TO QUESTION 120 -99= DK/NR --- GO TO QUESTION 120

119. Did the information on Michigan you received have no inﬂuence. a partial inﬂuence. or a primary inﬂuence on your decision to
travel in Michigan? > __ 1=No inﬂuence 2=Partial inﬂuence 3=Primary inﬂuence -99=DK/N R

END INFLUENCE BLOCK

(ACCEPT 1-999)
120. About how many pleasure trips to places in Michigan have you taken in the past 12 months? __ pleasure trips

[IF RESPONDENT IS UNABLE TO GIVE A SPECIFIC NUMBER. PROBE:]
In the past 12 months, would you say that you've taken. . . 2=I to 3 pleasure trips? 5=4 to 6 pleasure trips? 8=7 to 9 pleasure
trips? 15=10 to 20 pleasure trips? 25=More than 20 pleasure trips? -99=DK/NR

[NOTE2 USE CODES ONLY IF RESPONDENT DOESN'T GIVE A SPECIFIC RESPONSE]

BEGIN MICHIGAN TRAVEL EXPECTATIONS BLOCK
121. During the next 12 months. do you plan to take any pleasure trips to places in Michigan? :- _
l=Yes 2=No -—-> GO TO QUESTION I24 -99=DK/NR
[DO NOT READ LIST]

122. Compared to the preceding 12 months. during the next 12 months do you expect to take more. fewer, or about the same number
of pleasure trips in Michigan? > _ 1=More 2=Fewer 3=Same -99=DK/NR

123. Do you plan to take any pleasure trips in Michigan. . . l=Yes 2=No -99=DK/N R
This fall? i) __ How about this Thanksgiving? ;. __ How about this Christmas or New Years? f:-- _

END MICHIGAN TRAVEL EXPECTATIONS BLOCK
BEGIN MICHIGAN TRIP VOLUME BLOCK

124. Now we'd like to ﬁnd out how many trips you may have recently taken in Michigan. Here we'd like to get information on any
kind of trips you may have taken in Michigan, including business trips.

20]

[RESPONSE SHOULD INCLUDE ANY TRIPS RESPONDENT MAY HAVE ALREADY TOLD YOU ABOUT] (ACCEPT O-
31; IF 0 OR DK/NR. GO TO QUESTION 130)

How many trips of any kind to places in Michigan have you taken that occurred wholly or partially during [MONTH PRECEDING
CURRENT MONTH]? ‘2 trips ~99=DK/NR

[IF MORE THAN 1 TRIP WAS TAKEN, SAY: I'd like to ask you about the most recent trip that occurred wholly or partially
during [MONTH PRECEDING CURRENT MONTH]

125. Was this trip primarily for the purpose of conducting business or attending a convention. seminar, or meeting? ‘5’ __
l=Yes —> GO TO QUESTION 127 2=No ~99=DKINR

126. Was this trip primarily for some purpose other than business or pleasure. such as moving a household, or going to a funeral or
wedding in another city? > _ l=Yes 2=No —- -~ GO TO QUESTION I30 -99—=DK/NR ---> GO TO QUESTION I30

127. Was this an overnight or day trip? :- _
I=Overnight 2=Day trip ....:~ GO TO QUESTION 130 -99=DK/NR ——- _~. GO TO QUESTION 130

(ACCEPT 0-999)
128. How many nights were spent in Michigan? L" _ -99=DK/NR

[DO NOT READ LIST UNLESS NECESSARY TO STIMULATE RESPONSES]
129. What was the main type of lodging you used? 2’ __ I=Friend or relative‘s home 2=Hotel, motel. or lodge 3=Bed &

Breakfast 4=Rented cabin, cottage. or condominium 5=Owned cabin, cottage. or condominium 6=County. state. or federal
campground 7=Commercial campground (e.g., KOA) 8"—Boat/ship 9=Other -99=DK/NR

END MICHIGAN TRIP VOLUME BLOCK
BEGIN PERSONAL’HOUSEHOLD CHARACTERISTICS BLOCK

130. To conclude. we'd like to ask just a few questions to help us classify your answers. In what city do you live? >
131. And your state or province? >
132. And your zip or postal code?
133. In what county do you live? 2

 

 

 

[READ LIST]
134. Do any of the following types of persons live in your household?
l=Yes 2=No -55=Refused -99=DK/NR
Pre-school child? > _ School-age child under age 18? . - __ Senior citizen? 1- __ Handicapped person? 4‘» __

(ACCEPT 1-99)
135. How many persons, including yourself. live in your household? ____

(ACCEPT 0.99)

136. How many full-time wage-earners live in your household? _ -55=Refused -99-T-DK/N R
[READ LIST; ACCEPT UP TO 2 RESPONSES]
137. Are you ...... ‘> __ _ 1=Employed full-time: 2=Employed part-time: 3=Retired; 4=Not employed; STA homemaker,
6=A student; or 7=In some other employment situation? -99=DKI’N R

138. What racial or ethnic group do you belong to? 3-

 

139. The median household income is $31,000. Would you say your total household income before taxes in 1994 was above or below
the median? > _ l=Above the median 2=Below the median 2 GO TO QUESTION I41
-55-T-Refused [DO NOT READ] ...2 GO TO QUESTION I41 -99=DK/NR —-> GO TO QUESTION 14]

 

I40. Was your total household income above 350.000? __ l=Yes 2=No -55=Reﬁrsed ~99=DKfN R

END PERSONAUHOUSEHOLD CHARACTERISTICS BLOCK

141. That's all the questions I have. Would you like to know the number to call for free information on travel in Michigan? > __
l=Yes —-— > The number is 1-800-5432YES. Thank you very much for your time!!! Have a good evening! [TERMINATE]
2=No -- Thank you very much for your time!!! Have a good evening! [TERMINATE] -99=DK/NR

INTERVIEWER CODE NUMBER 1‘

 

202

APPENDK B.

1996-1 998 QUESTIONNAIRE

©Copyright 2002 Michigan State University, 203
Travel. Tourism & Recreation Resource Center

YEAR 7 — TTRRC TELEPHONE SURVEY — QUESTIONNAIRE l (REVISED 03/29/02)

[ENTER INTERVIEWER CODENUMBER: DOUBLEENTRYREQUIRED] l‘ _
[ENTER CODE NUMBER] >
[ENTERAREA CODE: DOUBLEENTRYREQUIRED] 2* __

Hello. I'm calling from Michigan State University. My name is ......... We're conducting a study on travel and tourism. May I speak
to the adult, 18 or older, who will have the next birthday?
{IF THIS PERSON IS NOTATHOME. ASK TO SPEAK TO THE ADULTATHOME WHO WILL HA VE THE NEXT
BIRTHDA Y. ]

We'd greatly appreciate your help in answering a few questions about trips you've made.

{ENTER GENDER OF RESPONij > _ l=Male 2=Female -99=Cannot determine
1. Does anyone in your household own or lease a car. van. recreation vehicle. pick- -up truck or sport-utility vehicle??-
1= Y es 2= No -) GO TO QUESTION 4 -99= DK/NR -) GO TO QUESTION_ 4
[READ LIST. ]
2. Would you say the price of gasoline has affected the amount of driving you do on pleasure tripsK 7' __
l=a great deal. 2=a little or 3=not at all? -99=DK/NR
[READ LIST. ]
3. Would you say the price of gasoline has affected the amount of driving you PLAN to do on pleasure tripsK 7‘ __
l=a great deal, 2=a little or 3=not at all? -99=DK/NR

4. During the past twelve months. have you traveled on any highways In Michigan?
I= Yes 2= No 9 GO TO QUESTION 6 -99= DK/NR -) GO TO QUESTION 6

5. How would you rate the overall quality of Michigan’ s highways on a scale from I to 10, where 1 means “very poor” and 10
means “outstanding”? -99= DK/NR

BEGIN AIR TRAVEL BLOCK

6. Have you traveled by commercial airplane within the last 30 days? :1" _
l=Yes 2=No -) GO TO QUESTION 9 -99=DK/NR -) GO TO QUESTION 9

[AIR TRA VEL EXPERIENCE MEANS EXPERIENCES AT THE DEPARTURE AIRPORT. DURING THE FLIGHTAND AT THE
ARRIVAL AIRPORT. ]

7. On a scale from I to 10. where I means “not at all satisﬁed" and 10 means “extremely satisﬁed,” how satisfied were you with
your last air travel experience? i' __

[ASK THIS QUESTION ONLY IF SATISFACTION WITH AIR TRA VEL EXPERIENCE WAS RATED I . 2 0R 3. j
8. Why were you dissatisfied with your last air travel experience?

 

,>

 

\
)

 

9. Have you cancelled a commercial air trip within the last 30 days? i» _
l=Yes 2=No -) GO TO QUESTION 11 -99=DK/NR -) GO TO QUESTION 11

10. Why did you cancel it?

‘3
,

 

>

 

\)

 

END AIR TRAVEL BLOCK

BEGIN MICHIGAN IMAGE BLOCK
[ACCEPT UP T O 3 RESPONSES. ]
l 1. As a travel destination, what do you think Michigan is known for?

 

'4

 

 

[A C CEPT UP TO 3 RESPONSES. ]
12. What. if any, tourism-related facilities, services or opportunities do you feel are missing in Michigan?

 

‘\

1

 

 

[READ]
l3. We’d like to know how much you agree or disagree with some statements about Michigan. Please use a scale from 1 to 10,

©Copyright 2002 Michigan State University. 204
Travel, Tourism & Recreation Resource Center

where 1 means “do not agree at all" and 10 means “agree completely." -99=DK'NR

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Michigan...

[ACCEPT I - IO or —99.]

Is close enough for a weekend getaway ...................................................................................................................... > _
Has many interesting museums ................................................................................................... _____
Is great for sununer outdoor recreation activities ....................................................................................................... > _
Is an exciting place to visit ...... ...................................................... “ __
Has a lot of high quality lodging ............................................................................................. 1" __
Offers much scenic appeal .......................................................................................................... > _
ls great for winter outdoor recreation activities ................................... ................. ‘2 _
Is a good place to meet fiiendly people ....................................................................................................................... > _
Is a place everyone should visit at least cum in their lifetime .................................................... ‘2 __
Is a safe place to visit ........................................................... _
Offers exciting nightlife and entertainment .. ...................... 1‘» __
Is a great place for a family vacation .......................................................................................................................... 2 __
Is a popular destination with vacationers ........................................................ r- __
Has many interesting historic sites ........................................................ _
Offers an excellent vacation value for the money .................................................................... __
Has great shopping opportunities .......................................................................................... '3- __

What is the Internet address for the Michigan’s official tourism web site?
-99=DK/N R 3‘

 

END MICHIGAN IMAGE BLOCK
BEGIN PLEASURE TRIP BLOCK

We're defining a "pleasure trip" as any overnight or day trip to a place at least 50 miles from your home that was made for your
enjoyment, including vacations, weekend getaways, shopping trips. trips to a second home. and trips to visit friends or relatives.

14. On a scale from 1 to 10. where 1 means “not at all important" and 10 means “extremely important." how important is each of the

 

 

following factors during your pleasure trips? -99=DK/NR
Comfort _
Convenience _
Cost. ................................................................................ _
Safety and security ......................................................... l" _

[DO NOTREAD THESE OPTIONS UNLESS THE RESPONDENT DOESN 'T RECALL THEM. /
15. Which one of these four factors is the most important to you during pleasure trips? - __
I=Comfort 2=Convenience 3=Cost 4=Safety and security -99=DK/NR

[DOUBLE ENTRY REQUIRED. ]
16. In the past twelve months. have you taken any day or overnight pleasure trips to any destination? '-’- _
l=Yes 2=No -) GO TO QUESTION 55 -99=DK/NR -) GO TO QUESTION 55

[ACCEPT] — 999.]
17. About how many pleasure trips have you taken in the past twelve months? - __

[IF RESPONDENT IS UNABLE TO GIVE A SPECIFIC NUMBER. READ THE FOLLOWING OPTIONS. ]
In the past 12 months. would you say you've taken...

2=1 to 3 pleasure trips 5=4 to 6 pleasure trips 8=7 to 9 pleasure trips 15=10 to 20 pleasure trips
25=More than 20 pleasure trips -99=DK./NR
END PLEASURE TRIP BLOCK

BEGIN MOST RECENT PLEASURE TRIP PROFILE BLOCK

[ASK FOR MONTH AND DA Y : ENTER NUMERICAL VALUES FOR MONTH AND DA Y; IF NECESSAR Y. ASK FOR BEST
GUESS OF DA Y. j
18. Now I'd like to ask you about your most recent pleasure trip. Approximately when did this trip begin — the month and day?

MONTH > __ DAY .- _ -99=DK/NR
MONTH CODES
1=January 4=April 7=July 10=October
2=February 5=May 8=August I I=November
3=March 6=June 9=September l2=December

[ACCEPT I — 3 RESPONSES. ASK FOR SPECIFIC PURPOSEYS). ESPECIALLY IF RESPONDENT SAYS "VA CA TION. " ASK
IF MORE THAN ONE PURPOSE MENTIONED. ]
19. What was the purpose or purposes ofthis trip?

)

 

 

©Copyright 2002 Michigan State University, 205
Travel, Tourism & Recreation Resource Center

>

 

20. What would you say was the PRIMARY purpose of this trip?
>

 

[ACCEPT 1 — 3 RESPONSES.) [DO NOT READ THE LIST.)

21. Whattypesoftransportation did you use? > _ __
I=Car/truck without camping equipment 2=Car/truck with camping equipment 3=Self-contained recreation vehicle

4=Rental car 5=Airplane 6=Train 7=Ship/boat 8=Motorcycle 9=Bicycle
10=Motorcoach/bus l 1=Other 9 ENTER UNDER QUESTION 22 -99=DK/N R
22. Other >

 

[ACCEPT I — 99.) [IF RESPONDENT WAS ON A GROUP TOUR. ASK FOR SIZE OF III/MEDIA TE TRAVEL PARTY AS
OPPOSED TO SIZE OF ENTIRE GROUP.)
23. How many persons, including you. were in your immediate travel party? _ -99==DK/NR

[ACCEPT I - I 30 FOR AGE VARIABLES.) [IF NECESSARY. ASK FOR RESPONDENT'S BEST GUESS OF AGE(S). )
24. Beginning with you, please give me the gender and age of each person who went on this trip:
1=Male 2=Female ~55=Refused -99=DK/NR

GENDER AGE GENDER AGE
RESPONDHVT :- _ '> __ PERSON #2 >- __ v" _
PERSON #3 > _ > __ PERSON #4 __ 2- _
PERSON #5 > _ > __ PERSON #6 _ :. _
PERSON # 7 ‘2 _ ‘2 _ PERSON #8 _ ‘ __
PERSON #9 > _ > PERSON #10

25. Was this an overnight or day trip? 1" _
I=Overnight 2=Day trip 9 GO TO QUESTION 33 -99=DK./N R 9 GO TO QUESTION 33

[ACCEPT I - 999.]
26. How many nights were you away from home? 1- _ -99=DKIN R

[ACCEPT 0 - 999: IF ZERO, SKIP NEXT QUESTION.)
27. How many of those nights did you spend in the state or province that was the main destination of this trip? > __ -
99=DKINR

/A C CEPT I — 5 LOCATIONS.)
28. In which locations did you spend these nights?

,/

 

 

 

 

>
>
,>

 

[ACCEPT 0-999.)
29. While you were in the state or province containing the main destination of this trip, about how much. if anything, did you spend
per night on lodging in hotels. motels. bed & breakfasts or rental cabins? 5- $___ -55-T-REFUSED -99=DK/NR
[DO NOT READ LIST UNLESS NECESSARY TO ST [MUM T E RESPONSES. ]
30. What was the main type oflodging you used? > _
1=Friend's or relative's home 2=Hotel. motel, or lodge 3=Bed & Breakfast 4=Rented cabin. cottage. or condominium
5=Owned cabin. cottage, or condominium 6=County, state. or federal campground 7=Commercial campground (e.g., KOA)
8=Boat/ship 9=Other -99=DK/N R

31. Did you spend the night at any casino hotels? > _
l=Yes 2=No 9 GO TO QUESTION 33 ~99=DK/NR 9 GO TO QUESTION 33

[ACCEPT I -— 5 CASINO LOCATIONS.)
32. Which ones?

 

>

 

)-

 

 

>

 

[READ THE LIST.) [IF RESPONDENT SA YS "YES" TO "OUT DOOR REC REA TION ", ASK THE NEXT QUESTION;
OT HERWTSE SKIP THE NEXT QUESTION.)

 

 

 

33. Which. if any. of the following activities did you participate in? l=Yes 2=No -99=DK/NR

Shopping ............. > __
Nightlife . .......... > _
©Copyright 2002 Michigan State University, 206

Travel, Tourism 8!. Recreation Resource Center

 

Visit a state or national park .................................................... 7»

 

 

 

 

 

 

 

 

Visit a museum or ball of fame .. ......................................................................... ~. __
Visit an historic site ....................................................................... ;. _
Visit some other type of attraction ............................................................................... ‘2 __
Explore a small city or town ............................................ > _
Dine at a unique restaurant ......................................................................... > __
Fall color touring outside of traveling to and from your destination . ............................................. > _
General touring or driving for p' We ......................................................................... ‘2 __
Outdoor recreation ............................. J)

 

[ACCEPT I - 5 RESPONSES.) [ASK ONLYIFOUTDOOR RECREA TION AFFIRMED ABOVE.)
34. What outdoor recreation activities did you participate in?

 

VV

 

\v'

 

 

J' V

 

35. Did you attend any festivals or events on this trip? --
l=Yes 2=No 9 GO TO QUESTION 38 -99=DK/NR 9 GO TO QUESTION 38

[ACCEPT] — 5 RESPONSES.)
36. What festivals or events did you attend?

 

 

 

 

 

[ACCEPT 0 - 999999. )
37. How much did you spend on-site at that/those festival(s) or event(s)? 1“ S __ ~55=REFUSED -99=DK/N R

38. Did you visit any farm markets. roadside produce stands or u-pick farms or orchards on this trip? “s _
l=Yes 2=No 9 GO TO QUESTION 42 -99=DK/NR 9 GO TO QUESTION 42

39. Did you purchase anything there? > __
l=Yes 2=No 9 GO TO QUESTION 42 ~99:DK/NR 9 GO TO QUESTION 42

[ACCEPT I — 5 RESPONSES.)
40. What did you purchase?

/

 

 

 

 

 

[ACCEPT o - 999999. )

41. How much did you spend in total? ‘f‘ S _____ -55=REFUSED -99=DK/NR
42. Did you visit any wineries on this trip? .> __
l=Yes 2=No 9 GO TO QUESTION 46 -99=DK/NR 9 GO TO QUESTION 46

43. Did you purchase anything there? > __
l=Yes 2=No 9 GO TO QUESTION 46 -99=DK/NR 9 GO TO QUESTION 46

[ACCEPT] - 5 RESPONSES.)
44. What did you purchase?

/

 

 

\g

 

,>

 

>

 

[ACCEPT 0 - 999999. )
45. How much did you spend in total? S __ o55=REFUSED -99=DK/NR

46. Did you do any casino gaming on this trip? __
l=Yes 2=No 9 GO TO QUESTION 51 -55=Refused 9 GO TO QUESTION 51 -99=DK/NR 9 GO TO QUESTION 51

[ACCEPT I — 5 RESPONSES.)
47. Which casinos did you visit?

©Copyright 2002 Michigan State University. 207
Travel. Tourism & Recreation Resource Center

 

.7

 

V

 

 

\' V

 

48. How satisfied were you with your visit to [NAME OF FIRST CASINO MENTIONED ABOVE ) on a scale from I to 10, where 1
means "extremely dissatisfied" and 10 means "extremely satisfied"? 2 __ -99=DKIN R

49. Did you plan to participate in casino gaming before you left home on this trip? ,-,. __
l=Yes 2=No 9 GO TO QUESTION 51 -99=DK/N R 9 GO TO QUESTION 51

50. Was casino gaming the only reason. the primary reason or a secondary reason for this trip?>- _
I=Only 2=Primary 3=Secondary -99=DK’N R

[ACCEPT 0 - 999999. ) .
51. In US. dollars. what would be your best estimate of how much your immediate travel party spent altogether while in the state
or province containing the main destination of this trip? 7- S -55=REFUSED -99=DK/NR

[ENTER RESPONSE. E.G.. 90 DA YS, 2 WEEKS, 3 MONTHS.)
52. About how far in advance of this trip did you begin to make plans for it?

 

[IF NECESSARY. ASK FOR CITY/PLACE FARTHEST FROM HOME.)
53. What was the main destination ofthis trip?

City/Place >-

State/Province/Country 7'

 

 

[DO NOTREAD.) [DOUBLE ENTRY REQUIRED.)
‘2 l=Michigan destination 9 GO TO QUESTION 92 2=Non-Michigan destination

END MOST RECENT PLEASURE TRIP PROFILE BLOCK

[DOUBLE ENTRY REQUIRED.)
54. Was a place in Michigan the main destination of any of the pleasure trips youye taken in the past twelve months? ‘I __
l=Yes 9 GO TO QUESTION 56 2=No -99=DK/NR

55. Have you ever taken a pleasure trip to a place in Michigan? ‘I’ _
l=Yes 9 GO TO QUESTION 94 2=No 9 GO TO QUESTION 94 -99=DK/N R 9 GO TO QUESTION 94

BEGIN GENERAL MICHIGAN PLEASURE TRIP PROFILE BLOCK
56. Now I'd like to ask you about your most recent pleasure trip in Michigan.

[IFNECESSARK EATIAIN THAT WENEED A PROFILE OF THEIR MOST RECENT PLEASURE TRIP IN MICHIGAN AS
WELL AS THEIR MOST RECENT PLEASURE TRIP IN GENERAL.)

[ASK FOR MONTH AND DA Y. ENTER NUMERICAL VALUES FOR MONTH AND DA Y. IF NECESSARY. ASK FOR BEST

GUESS OF DA Y ]
Approximately when did this trip begin - the month and day?
MONTH > _ DAY > _ ~99=DKINR
MONTH CODES
I=January 4=April 7=July 10=October
2=February 5=May 8=August I 1=November
3 =March 6=June 9=September 12=December

[ACCEPT I — 3 RESPONSES. ASK FOR SPECIFIC PURPOSEfS). ESPECIALLY IF RESPONDENT SAYS "VACA TION. "]
57. What was the purpose or purposes of this trip?

 

vi

 

V

 

[ASK IF MORE THAN ONE PURPOSE MENTIONED: ASK FOR SPECIFIC PURPOSE. ESPECIALLY IF RESPONDENT SA YS
"VACA TION. ")
58. What would you say was the PRIMARY purpose of this trip? .1-

 

[ACCEPT I -— 3 RESPONSES.) [DO NOTREAD LIST.)
59. Whattypesoftransportationdidyou use?>_____
I=Car/truck without camping equipment 2=Car/truek with camping equipment 3=Self-contained recreation vehicle
4=Rental car 5=Airplane 6=Train 7=Ship or boat 8=Motorcycle 9=Bicycle 10=Motorcoach/bus
ll=Other 9 ENTER UNDER QUESTION 60 -99=DK/NR

©Copyright 2002 Michigan State University. 208
Travel. Tourism & Recreation Resource Center

60. Other >

 

[ACCEPT I - 99.) [IF RESPONDENT WAS ON A GROUP TOUR. ASK FOR SIZE OF IMMEDIA TE T RA VEL PARTYAS
OPPOSED T 0 SIZE OF ENTIRE GROUP.)
61. How many persons. including you, were in your immediate travel party? > __ -99=DK/N R

[ACCEPT I - 130 FOR AGE VARIABLES.) [IF NECESSARY. ASK FOR RESPONDENT'S BEST GUESS OF AGE. )
62. Beginning with you, please give me the gender and age of each person who went on this trip.

I=Male 2=Female ~55=Refused -99=DK/NR

GENDER AGE GENDER AGE
RESPONDENT 3» __ I“ _ PERSON #2 “ _ 1- __
PERSON #3 1" __ 9 __ PERSON #4 __ _
PERSON #5 > _ ~;. __ PERSON #6 _ _
PERSON #7 ‘2 _ ~;. __ PERSON #8 __ __
PERSON #9 __ :1 _ PERSON #10 _ __
63. Was this an overnight or day trip?
I=Overnight 2= —Day trip 9 GO TO— QUESTION 71 -99=DK/NR 9 GO TO QUESTION 71
[ACCEPT] — 999.)

64. How many nights were you away from home? __ ~99=DKINR

[ACCEPT 0 — 999: IF ZERO, SKIP NEXT QUESTION.)
65. How many nights were spent in Michigan? - __ -99=DK/NR

[ACCEPT I - 5 LOCATIONS.)
66. In which locations in Michigan did you spend these nights?

>

 

/‘

 

’)

 

 

 

[ACCEPT 0 — 999.]
67. While in Michigan. about how much. if anything. did you spend per night on lodging in hotels. motels. bed & breakfasts or
rental cabins? > S_ -55=REFUSED -99=DK/NR

[DO NOTREAD LIST UNLESS NECESSARY T O S TIM UIA TE RESPONSES.)
68. What was the main type of lodging you used? >
1=Friend‘s or relative's home 2= Hotel, motel. or lodge 3=Bed & Breakfast 4=Rented cabin. cottage. or condominium
5=Owned cabin. cottage. or condominium 6= —County, state or federal campground 7=Commercial campground (e.g., KOA)
8=Boat/ship 9=Other -99=DK/N R

69. Did you spend the night at any casino hotels In Michigan? “'
1— TYes 2= No 9 GO TO QUESTION 71 -55= Refused 9_ GO TO QUESTION 71 -99= DK/NR 9 GO TO QUESTION 71

[ACCEPT] — 5 CASINO LOCATIONS.)
70. Which ones?

 

 

 

 

/)

 

[RE/ID LIST.) [IF RESPONDENTSAYS "YES" TO "OUT DOOR RECREA TION". ASK THE NEXT QUESTION: OTHERWISE
SKIP THE NEXT QUESTION.)

 

 

 

 

 

 

 

 

 

 

71. Which. if any, of the following activities did you participate in? l=Yes 2=No -99=DK/NR
Shopping ......................................................................................................................... > _
Nightlife .......................................................................................................................... ,> _
Visitastateornationalpark ..................................................................................................................................................... >_
Visit a museum or hall of fame .................................................................................................................... '2' __
Visit an historic site .......................................................................................................................... > _
Visit some other type of attraction ............................................................................................................................................. _
Explore a small city or town ...................................................................................................................................................... 2 __
Dine at a unique restaurant ....................................................................................... ‘> __
Fall color touring outside of traveling to and ﬂour your destination ........................................................................................ 9' __
General touring or driving for pleasure .................. .. ...... :- __
Outdoor recreation ................. ‘> _
©Copyright 2002 Michigan State University. 209

Travel, Tourism & Recreation Resource Center

[ACCEPT I - 5 RESPONSES.) [ASK ONLY IF OUTDOOR RECREA TION AFFIRMED ABOVE.)
72. What outdoor recreation activities did you participate in while you were in Michigan?

 

 

 

 

VVVVV

 

73. Did you attend any festivals or events on this trip? >
l=Yes 2=No 9 GO TO QUESTION 76 -99=DK/NR 9 GO TO QUESTION 76

[ACCEPT I — 5 RESPONSES.)
74. What festivals or events did you attend?

 

 

 

v v v v

 

 

[ACCEPT 0 - 999999. )

75. How much did you spend on-site at that/those festival(s) or event(s)? S __ -55=REFUSED -99=DK/NR
76. Did you visit any farm markets. roadside produce stands or u-pick farms or orchards on this trip? > _

l=Yes 2=No 9 GO TO QUESTION 80 -99=DK/N R 9 GO TO QUESTION 80

77. Did you purchase anything there? “2 __
l=Yes 2=No 9 GO TO QUESTION 80 -99=DK/N R 9 GO TO QUESTION 80

[ACCEPT I - 5 RESPONSES.)
78. What did you purchase?

3
/

 

 

 

V V V V

 

 

[ACCEPT 0- 999999. )
79. How much did you spend in total? S _ -55=REFUSED ~99=DK’N R

80. Did you visit any wineries on this trip? ‘> __
l=Yes 2=No 9 GO TO QUESTION 84 -99=DK/N R 9 GO TO QUESTION 84

81 . Did you purchase anything there? "4' __
l=Yes 2=No 9 GO T O QUESTION 84 -99=DK/NR 9 GO TO QUESTION 84

[ACCEPT I — 5 RESPONSES. )
82. What did you purchase?

3

 

 

 

v v V -.

 

 

[ACCEPT 0 - 999999.]
83. How much did you spend in total? 3' S _ ~55=REFUSED -99=DK/N R

84. Did you do any casino gaming on this trip? __
l=Yes 2=No 9 GO TO QUESTION 89 -55=Refused 9 GO TO QUESTION 89 -99=DK/NR 9 GO TO QUESTION 89

[ACCEPT I — 5 RESPONSES.)
85. Which casinos did you visit?

 

 

‘J \J' V

 

/

 

/‘

 

[ACCEPT I — 10.)
86. How satisfied were you with your visit to [NAME OF FIRST CASINO MENTIONED ABO VE) on a scale from I to 10, where 1
means "extremely dissatisﬁed" and 10 means "extremely satisfied"? _ -99=DK/NR

©Copyright 2002 Michigan State University, 210
Travel, Tourism & Recreation Resource Center

87. Did you plan to participate in casino gaming before you left home on this trip? 1" _
l=Yes 2=No 9 GO TO QUESTION 89 -99=DKfN R 9 GO TO QUESTION 89

88. Was casino gaming the only reason, the primary reason or a secondary reason for this trip? ,> __
1=OnIy 2=Primary 3=Secondary ~99=DKINR

[ACCEPT 0- 999999. )
89. In US. dollars what would be your best estimate of how much your immediate travel party spent altogether on this trip while in
Michigan? > S -55=REFUSED -99=DK/NR

[ENTER RESPONSE. E.G.. 90 DAYS. 2 WEEKS. 3 MONTHS.)
90. About how far in advance of this trip did you begin to make plans for it?

 

[IF NECESSARY, ASK FOR CITY/PLACE FARTHEST FROM HOME.)
91. What was the main destination of this trip?
City/Place in Michigan: >

 

END GENERAL MICHIGAN PLEASURE TRIP PROFILE BLOCK
BEGIN GENERAL MICHIGAN PLEASURE TRIP HISTORY BLOCK

92. 'Was this most recent pleasure trip in Michigan the first pleasure trip you've ever taken in this state? 2» _
l=Yes 9 GO TO QUESTION 94 2=No -99=DK/NR

[ACCEPT] — 999.]
93. About how many pleasure trips to places in Michigan have you taken in the past twelve months?

[IF RESPONDENT IS UNABLE T 0 GI l/EA SPECIFIC NUMBER. READ THE FOLLOWING OPTIONS.)

In the pan [2 months. would you say you've taken...
2=I to 3 pleasure trips 5=4 to 6 pleasure trips 8=7 to 9 pleasure trips 15:10 to 20 pleasure trips
25=More than 20 pleasure trips -99=DK/NR

END GENERAL MICHIGAN PLEASURE TRIP HISTORY BLOCK
BEGIN MICHIGAN TRAVEL EXPECTATIONS BLOCK

94. During the next twelve months, do you expect to take more, fewer, or about the same number of pleasure trips to ANY
DESTINATION as you did during the previous twelve months? 9“_ 1=More 2=Fewer 3=Same -99=DK/NR

95. How about to MICHIGAN? > __
1=More 2=Fewer 9 GO TO QUESTION 97 3=Same 9 GO TO QUESTIONIOI
(starting on March I, 2002 9 GO TO QUESTION 98)
-99=DK/N R 9 GO TO QUESTIONIOI
(starting on March I, 2002 9 GO TO QUESTION 98)

96. Why do you think you will take more pleasure trips to Michigan in the next 12 months? ‘t-
9 GO TO QUESTION 101
(starting on March I. 2002 9 GO TO QUESTION 98)

 

97. Why do you think you will take fewer pleasure trips to Michigan in the next 12 months? I»
9 GO TO QUESTION 101
(starting on March I, 2002. no skipping)

 

98. Did you take a pleasure trip to Michigan during last year's Memorial Day? ‘- _
l=Yes 2=No -99=DK/NR

99. Do you plan to take a pleasure trip to Michigan during this year's Memorial Day? r- __
l=Yes 9 GO TO QUESTION 101 2=No -99=DK/NR 9 GO TO QUESTION 101

100. Why don't you plan to take a pleasure trip to Michigan this Memorial Day??-

 

101. Do you plan to take any pleasure trips to places in Michigan
l=Yes 2=No -99=DK/NR
...duringthis winter season? > _
How about during the next twelve months? > __

END MICHIGAN TRAVEL EXPECTATIONS BLOCK

BEGIN INTERNET BLOCK
102. Do you have access to the Internet? > ____

l=Yes 2=No 9 GO TO QUESTION 106 -99=DK/NR 9 GO TO QUESTION 106
©Copyright 2002 Michigan State University. 2] 1

Travel, Tourism & Recreation Resource Center

103. During the past twelve months have you used the Internet to obtain travel information?
l=Yes 2=No -99=DK/NR

104. Have you made a travel-related purchase over the Internet in the past twelve months? 1* __
l=Yes 2=No 9 GO TO QUESTION 106 -99=DK/NR 9 GO TO QUESTION 106

[ACCEPT] — 999.)
105. How many times? 3* __

END INTERNET BLOCK
BEGIN PERSONAL/HOUSEHOLD CHARACTERISTICS BLOCK

[DO NOTREAD LIST.) [DOUBLE ENTRY REQUIRED.)
106. To conclude. we'd like to ask just a few questions to help us classify your answers. In what state or province do you
permanently reside? > _
l=Illinois 2=Indian 3=Michigan 4=Minnesota 5=Ohio 6=Wisconsin 7=Ontario -99=DK/NR
107. In what county do you live? >
108. What is your zip or postal code? 9
109. In what city do you live? 2
110. On a scale from I to 10, where I means “not at all satisfied“ and 10 means “extremely satisfied," how satisfied are you with the
following options in YOUR HOME COMMUNITY... -99=DK/N R
The overall quality of life ...................................................... ..
Festivals and events ........................................................... . __
Sporting events ................................................................. -
Nightlife or entertainment ...................................................

Local parks and recreation opportunities (prog‘ams. open space)? ‘ :

 

 

 

I 1 1. Do any ofthe following types ofpersons live in your household? l=Yes 2=No ~55=Refused -99=DK/NR

Pre-school child .....................
School-age child under age 18....
Senior citizen ........................
Handicapped person ................

[ACCEPT] -— 99.]
I 12. How many persons. including yourself. live in your household? - _ -99=DKINR

[ACCEPT] - 99.)
l 13. How many adults over age 17. including yourself. live in your household? - _ -99=DK’NR

[ACCEPT 0 — 99.]
114. How many full-time wage earners live in your household? _ ~55=Refused -99'— DK/N R

[ACCEPT I — 2 RESPONSES.) [DO NOTREAD NUMBERS.)
115. Are you... > __ _ -55=Refused -99=DK/NR
I=employed fullotirne 2=employed part-time 3=retired 4=not employed 5=a homemaker 6=a student
7=in some other employment situation

I 16. What racial or ethnic group do you belong to? -55=Refused -99=DK/NR
BEGIN VOLUNTEERISM BLOCK

Now we would like to ask you just a couple of questions about volunteer work. By “volunteer work" we mean NOT just belonging to
an organization. but actually working in some way to help others or the environment for no monetary pay.

I 17. Have you done ANY volunteer work in the past 12 months? 2* __
l=Yes 2=No 9 GO TO QUESTION 134 -99=DKINR 9 GO TO QUESTION I34

[ACCEPT 0— 366.)
1 18. How many times did you volunteer your time in the past 12 months? >

[EXAMPLES OF VOL UNTEER WORK IN MANAGEMENT OR PLANNING ACTIVITIES REALTED TO OUTDOOR RE C REA TION.
NATURAL RESOURCES OR THE ENVIRONMENT:

- C AMPGRO UND A TTENDANT OR HOST ("UNPAID ")

- FUNDRAISLN'G

- GENERAL MAINTENANCE (FACILITY INSPECTIONS. LITTER PICK UP. PAIN TING, ETC.)

- INT ERPRET A TION/EDUCA TION (DELIVER NA TURE. HISTORICAL OR CULTURAL REL/I TED PROGRAMS. ETC.)

- MEMBERSHIP ON AN INFORMAL ADVISOR Y GROUP/T ASK FORCE

©Copyright 2002 Michigan State University. 212
Travel, Tourism & Recreation Resource Center

- MEMBER OF A PLANNING TEAM
- MEMBER OF A POLICY-MAKING GROUP (FORMAL CITIZEN BOARD OR COMMISSION)
- OFFICEASSISTANCE (ENIELOPE STUFFING. FEE COLLECTION. BOOKKEEPING, ETC.)
- RESOURCE MONITORING (BIRD C OUNTS. STREAM OR WETLANDS INSPECTIONS. IN l/ENTORIES. ETC.)
- RESOURCE STE WARDSH IP/RESTORA TION (NON -NA Til/E SPECIES REMOVAL, PRESCRIBED B URNS. ETC. )]
119. Excluding coaching, in the past 12 months. have you done any volunteer work that involved participating in MANAGEMENT
OR PLANNING ACTIVITIES related to outdoor recreation. natural resources or the environment?>_
l=Yes 2=No 9 GO TO QUESTION 134 ~99=DKINR 9 GO TO QUESTION 134

120. In the past 12 months. what was the PRIMARY way you volunteered your time in management or planning activities related to
outdoor recreation. natural resources or the environment?

 

[ACCEPTO— 999.)
121. How many times did you volunteer your time this way in the past 12 months?

122. For which group, agency or organization did you volunteer?

 

123. In the past 12 months. have you volunteered your time in any OTHER way that involved management or planning activities
related to outdoor recreation, natural resources or the environment? .. _
l=Yes 2=No 9 GO TO QUESTION 131 ~99=DKINR 9 GO TO QUESTION 131

124. What did you do? >

 

[ACCEPT 0 — 999. ]
125. How many times did you volunteer your time this way in the past 12 months?

126. For what group, agency or organization did you volunteer?

 

127. In the past 12 months, have you volunteered your time in any ADDITIONAL way that involved management or planning
activities related to outdoor recreation, natural resources or the environment? 1* __
l=Yes 2=No 9 GO TO QUESTION I31 -99=DKINR 9 GO TO QUESTION I31

128. What did you do? i“

 

[ACCEPT 0 — 999.)
129. How many times did you volunteer your time this way in the past 12 months?

130. What type of group. agency or organization did you volunteer with or for? ‘-

[READ THE LIST.)
131. When you volunteer your labor, time and/or services for NATURAL RESOURCES. THE ENVIRONMENT OR OUTDOOR
RECREATION do you expect to receive any of the following in return ?
l=Yes 2=No -99=DK/NR
Thanks ........................................................................ - __
Public recognition ........................................................... '- _
More access to managers than citizens who do not participate .......
More inﬂuence on agency/organization decisions and policies ....... "

Support (transportation. food/beverages, child care) ...................

 

132. Do you expect something else not mentioned above? _
l=Yes 2=No 9 GO TO QUESTION I34 ~99=DKfNR 9 GO TO QUESTION 134

133. What else do you expect in return for your volunteering?

 

END VOLUNTEERISM BLOCK

134. The median household income is $42,000. Would you say your total household income before taxes in 2000 was above or
below the median? > _

l=Above the median 2=Below the median 9 GO TO QUESTION 136
-55=Refused 9 GO TO QUESTION I36 -99=DK/NR 9 GO TO QUESTION I36
I35. Was your total household income above S65.000? " _ l=Yes 2=No -55=Refused -99=DK/NR

END PERSONAL/HOUSEHOLD CHARACTERISTICS BLOCK

[READ]
136. That's all the questions I have. Thank you very much for your time! Have a good evening!

 

[TO TERMINATE. HIT THE ENTER KEY ONCE.) -'__
Legend

MS:

~99=DKINR

-33=Did not travel in Michigan
-55=Reﬁrsed

©Copyright 2002 Michigan State University. 213
Travel. Tourism & Recreation Resource Center

 

 

 

REFERENCES

Aaker, D. (1981). Multivariate Analysis in Marketing. Palo Alto, CA: The Scientiﬁc
Press.

Aaker, D. A. & Shansby J. G. (1982). “Positioning Your Product.” Business Horizons,
May-June, 5662.

Achen, C. H. (1982). Interpreting and Using Regression. Newbury Park, CA: Sage
Publications.

Aﬁfi, A. (1984). Computer-Aided multivariate Analysis. Belmont, CA: Lifetime Learning
Publications.

Ahmed, Z. U. (1996). “The Need for the Identiﬁcation of the Constituents of a
Destination's Tourist Image: A Promotion Segmentation Perspective.” Journal of
Promotional Services Marketing, 14 (1), 37-60.

Alhemoud, A. & Armstrong, E. (1996). “Image of tourism attractions in Kuwait.”
Journal of Travel Research, 34 (Spring), 76-80.

Aquilino, W. S. (1994). “Interview mode effects in surveys of drug and alcohol use: A
ﬁeld experiment.” Public Opinion Quarterly, 58 (2), 210-240.

Ashworth, G. (1989). “Urban tourism: an imbalance in attention.” In C. Cooper
(Eds), Progress in tourism, recreation and hospitality management, London:
Belhaven.

Babbie, E. (1998). The Practice of Soda/Research. (8th ed). Belmont, CA: Wadsworth
Publishing Company.

Baloglu, S. (2001). “Image variations of Turkey by familiarity index: informational and
experiential dimensions.” Tourism Management, 22, 127-133.

Baloglu, S. & Brinberg, D. (1997). “Affective images of tourism destinations.” Journal of
Travel Research, 35 (4), 11-15.

Baloglu, S. & McCleary, K. W. (1999). “US. international travelers’ images of four
mediterranean destinations: A comparison of visitors and nonvisitors.” Journal of
Travel Research, 38 (November), 144-152.

Bass, F. M. (1975). “Unexplained variance in studies of consumer behavior.” In J. Farley
& J. Howard (Eds) Control of Error in Market Research Data. Lexington, MA:
D. C. Heath and Company.

Bernstein, I. (1988). Applied Multivariate Analysis. New York, NY: Springer Verlag.

214

Bilkey, W. & Nes, E. (1982). “Country-of-origin effects on product evaluations.” Journal
of International Business Studies, 8 (Spring/Summer), 89-99.

Bojanic, D. C. (1991). “The use of advertising in managing destination image.” Tourism
Management, 12 (Spring), 353-55.

---( 1992). “A look at a modernized family life cycle and overseas travel.” Journal of
Travel and Tourism Marketing, 1 (1), 61-79.

Bramwell, B. & Rawding, L. (1996). “Tourism marketing images of industrial cities.”
Annals of Tourism, 23, 201-221.

Britton, R. A. (1979). “The image of the third world in tourism marketing.” Annals of
Tourism Research, 6 (July/Sept), 318-327.

Calantone, R. J., Benedetto, A. D., Hakam, A., & Bojanic, D. C. (1989). “Multiple
multinational tourism positioning using correspondence analysis.” Journal of
Travel Research, 28 (Fall), 25-32.

Chen, J. S. & Hsu, C. H. C. (2000). “Measurement of Korean tourists’ perceived images
of overseas destinations.” Journal of Travel Research, 38 (May), 411-416.

Chen, P. J. & Kerstetter, D. L. (1999). “International students’ image of rural
Pennsylvania as a travel destination.” Journal of Travel Research, 37 (February),
256-266.

Choi, W. M., Chan, A., & Wu, J. (1999’). “A qualitative and quantitative assessment of
Hong Kong’s image as a tourist destination.” Tourism Management, 20, 361-365.

Churchill, Jr. G. A. & Peter, P. J. (1984). “Research design effects on the reliability of
rating scales: A Meta analysis.” Journal of Marketing Research, 21(November),
365-366.

Churchill, Jr.G. A. (1999). Marketing Research: Methodological Foundations, The
Dryden Press.

Claritas. (2003). Segmentation-Prim. Last viewed August 5, 2003 on the WWW:
http://www.claritas.com/index.html?section=search&content=search.htm

Coshall, J. T. (2000). “Measurement of tourists’ images: The Repertory Grid Approach.”
Journal of Travel Research, 39, (August), 85-89.

Court, B. & Lupton, R. A. (1997). “Customer portfolio development: Modeling

destination adapters, inactives, and rejecters.” Journal of Travel Research, 36 (1),
35-43.

215

Crompton, J. L. (1979). “An assessment of the image of Mexico as a vacation Destination
and the inﬂuence of geographical location upon that image.” Journal of Travel
Research, 17(1), 18-23.

Crompton, J. L., Fakeye, P. C., & Lue, C. C. (1992). Positioning: The Example of the
Lower Rio Grande Valley in the Winter Long Stay Destination Market. Journal of
Travel Research, 31 (Fall), 20-26.

Dadgostar, B. & Isotalo, R. (1992). “Factors affecting time spent by near-home tourists in
city destinations.” Journal of Travel Research, 30 (Fall), 34-3 9.

Dann, G. M. S. (1996). Tourists’ images of a destination-An alternative analysis.” Recent
Advances in Tourism Marketing Research. 5, (1/2), 41-55.

Dillman, D. A. (1978). Mail and Telephone Surveys. New York, NY: Wiley & Sons.

Echtner, C. M. & Ritchie J. R. B. (1991). “The meaning and measurement of destination
image.” Journal of Tourism Studies, 2 (2), 2-12.

Embacher, J. & Buttle F. (1989). “A Repertory Grid Analysis of Austria's image as a
summer vacation destination.” Journal of Travel Research, 27 (Winter), 3-7.

Ervin, K. S. & Gilmore, G. (1999). “Traveling the superinforrnation highway: Aﬁican
Americans' perceptions and use of cyberspace technology.” Journal of Black
Studies, 29 (3), 398-407.

Etzel, M. & Woodside, A. (1982). “Segmenting vacation markets: The case of the distant
and near-home traveler.” Journal of Travel Research, 20 (Spribg), 10-14.

Fakeye, P. C. & Crompton, J. L. (1991). Image differences between prospective, ﬁrst-
time, and repeat visitors to the Lower Rio Grande Valley.” Journal of Travel
Research, 30 (Fall), 10-16.

Fink, J. C. (1983). “CATI’s ﬁrst decade: The Chilton experience.” Sociological Methods
and Research, 12, 153-168.

Fraenkel, J. R. & Wallen, N. E. (1996). How to design and evaluate research in
education. (3rd ed.). McGraw-Hill, Inc.

Frey, J. (1989). Survey Research by Telephone. Newbury Park, CA: Sage Publications.

Fridgen, J. D. (1987). “Use of cognitive maps to detemiine perceived tourism regions.”
Leisure Sciences, 9, 101-117.

Gartner, W. C. (1986), "Temporal Inﬂuences on Image Change." Annals of Tourism
Research, 13, 635-644.

216

---(1989). “Tourism Image: Attribute Measurement of State Tourism Products Using
Multidimensional Scaling Techniques.” Journal of Travel Research, 28 (Fall), 16-
20.

---(1993). “Image formation process.” Journal of Travel and Tourism Marketing, 2 (2/3),
191-215.

Gartner, W. C., & Shen, J. (1992). “The Impact of Tiananmen Square on China's Tourism
Image.” Journal of Travel Research, 30 (Spring), 47-52.

Gazel, R. C., Schwer, R. K., & Daneshvary, R. (1998). “Interview mode choice by survey
respondents: A methodological analysis.” Social Science Computer Review, 16
(2), 185-191.

Gentry, J. & Doering, M. (1979). “Sex role orientation and leisure.” Journal of Leisure
Research, (Second Quarter), 102-1 1.

Goodrich, J. N. (l978)a. “The Relationship between preferences for and perceptions of
vacation destinations: Application of a choice model.” Journal of Travel
Research, 16 (2), 8-13.

Goodrich, J. N. (1978)b. “A new approach to image analysis through multidimensional
scaling.” Journal of Travel Research, 16 (Winter), 3-7.

Grube, J. W. (1997). “Monitoring youth behavior in response to structural changes:
Alternative approaches for measuring adolescent drinking.” Evaluation Review,
21 (2), 231-245.

Gyte, D. & Phelps, A. (1989). “Patterns of destination repeat business: British tourists in
Mallorca, Spain.” Journal of Travel Research, 27 (Summer), 24-28.

Gunn, C. (1972). Vacationscape: Designing Tourist Regions. Austin: Bureau of Business
Research, University of Texas.

Hair, Jr. J. F., Anderson, R. E., Tatham, R. L., & Black, W. C. (1998). Multivariate Data
Analysis. New York: Macmillan.

Hamilton, LC. (1992). Regression with Graphics-A second course in Applied Statistics.
Duxbury Press, Belmont, California.

Herzog, A. R., Rodgers, W. L., & Kulka, R. A. (1983). “Interviewing older adults: A
comparison of telephone and face-to-face modalities.” Public Opinion Quarterly,
47 (3), 405-418. -

Hu, Y. & Ritchie, J. R. B. (1993). “Measuring destination attractiveness: A contextual
approach.” Journal of Travel Research, 32 (2), 25-34.

217

Hunt, J. D. (1975). “Image as a factor in tourist development.” Journal of Travel
Research, 13 (Winter), 1-7.

Jain, A., Pinson, C., & Ratchford, B. (1982). Marketing Research: Applications and
Problems. New York: John Wiley and Sons.

Joppe, M., Martin, D. W., & Waalen, J. (2001). “Toronto’s image as a destination; A
comparative importance-satisfaction analysis by origin of visitors.” Journal of
Travel Research, 39 (3), 252-260.

Kaye, B. K. & Johnson, T. J. (1999). “Research methodology: Taming the cyber
frontier.” Social Science Computer Review, 17 (3), 323-337.

Krysan, M., Schuman, H., Scott, L. J ., & Beatty, P. (1994). “Response rates and response
content in mail versus face-to-face surveys.” Public Opinion Quarterly, 58 (3),
381-399.

Lavrakas, P. J. (1993). Telephone Survey Methods: Sampling, Selection and Supervision.
(2"d Ed). Newbury Park, CA: Sage Publications.

Lovelock, C. H. (1984). Services Marketing. Englewood Cliffs, NJ: Prentice Hall.

Lubbe, B. (1998). Primary image as a dimension of destination image: An empirical
assessment.” Journal of Travel and Tourism Marketing, 7 (4), 21-43.

MacKay, K., & Fesenmaier, D. (1997). “Pictorial element of destination in image
formation.” Annals of Tourism Research, 24, 537-65.

---(2000). “An exploration of cross-cultural destination image assessment.” Journal of
Travel Research, 38 (4), 417-423.

Mayo, E. (1973). “Regional images and regional travel behavior. Research for changing
travel patterns: Interpretation and utilization.” In Proceedings of the Travel
Research Association 4'” Annual Conference. Salt Lake City, UT: Travel and
Tourism Research Association, pp. 211-18.

Mayo, E. & Jarvis, L. (1981). The Psychology of Leisure Travel. BostonzCBI.
McAuliffe, W. E., Geller, S., LaBrie, R., Paletz, S., & Foumier, E. (1998). “Are
telephone surveys suitable for studying substance abuse? Cost, administration,

coverage, and response rate issues.” Journal of Drug Issues, 28 (2), 455-482.

McQueen, J. & Miller, K. (1985). “Target market selection of tourists: A comparison of
approaches.” Journal of Travel Research, 23 (Summer), 2-6.

218

Miller, L. & Smith, K. (1983). “Handling Nonresponse Issues.” Journal of Extension,
(September/October), 45-50.

Milman, A. & Pizam, A. (1995). “The role of awareness and familiarity with a
destination: The central Florida case.” Journal of Travel Research, 33 (Winter),
21-27.

Murphy, L. (1999). “Australia’s image as a holiday destination-perceptions of
backpacker visitors.” Journal of Travel and Tourism Marketing, 8 (3), 21-45.

National Do Not Call Registry. (2003). Last viewed August 6, 2003 on the WWW:
http://www.donotcall.gov

Pearce, P. L. (1982). “Perceived changes in holiday destinations.” Annals of Tourism
Research, 9, 145-164.

Phelps, A. (1986). "Holiday destination image: The problem of assessment." Tourism
Management, 7 (3), 168-80.

Pindiyck, R. S. & Rubinfeld, D. L. (1981). Econometric models and economic forecasts.
(2"d ed.). McGraw-Hill Book Company.

Rockwood, T. H., Sangster, R. L., & Dillman, D. A. (1997). “The effect of response
categories on questionnaire answers: Context and mode effects.” Sociological
Methods and Research, 26 (1), 118-140.

Ross, G. F. (1993). “Ideal and actual images of backpacker visitors to Northern
Australia.” Journal of Travel Research, 32 (3), 54-57.

Rossi, P., Wright, J ., & Anderson, A. (1983). Handbook of Survey Research. San Diego,
CA: Academic Press.

Schroeder, T. (1996). “Relationship of residents’ image of their state as a tourist
destination and their support for tourism.” Journal of Travel Research, 34 (4), 71-
73.

Schul, P. & Crompton, J. (1983). “Search behavior of international vacationers: Travel
speciﬁc lifestyle and sociodemographic variables.” Journal of Travel Research,
21 (Fall), 25-30.

Selby, M. & Morgan, N. J. (1996). “Reconstructing place image-A case study of its role
in destination market research.” Tourism management, 17 (4), 287-294.

Sirakaya, E., Petrick, J., & Choi, H. S. (2002). “Customer Satisfaction Scores Are Not
What They Seem?” Paper presented at the 33rd Annual Conference of Travel and
Tourism Research Association, June 23-26, 2002, Arlington, Virginia.

219

Smith, M. C. & MacKay, K. J. (2001). “The organization of information in memory for
pictures of tourist destinations: Are there age-related dilferences?” Journal of
Travel Research, 39 (3), 261-266.

Snepenger, D., Meged, K., Snelling, M., & Worrall, K. (1990). “Information search
strategies by destination-naive tourists.” Journal of Travel Research, 28
(Summer), 13-16.

Sniderman, P. M. & Grob, D. B. ( 1996). “Innovations in experimental design in attitude
surveys.” Annual Review of Sociology, 22, 377-399.

Sonmez, S., Apostolopoulos, Y., & Tarlow, P. (1999). “Tourism in crisis: Managing the
effects of terrorism.” Journal of Travel Research, 38 (1), 13-18.

StatPac Inc. (1995). StatPac Gold IV: Overview and Operation. Minneapolis, MN:
StatPac Inc.

Sudman, S. & Bradbum, N. M. (1973). “Effects of time and memory factors on response
in surveys.” Journal of American Statistical Association, 68 (344), 805-815.

Tapachai, N. & Waryszak, R. (2000). “An examination of the role of beneﬁcial image in
tourist destination selection.” Journal of Travel Research, 39 (1), 37-44.

Tasci, A. D. A., Aziz, A., & Holecek, D. F. (2003). “Characteristics and potential of
Midwestern package travelers.” The proceedings of 34’" Annual Conference of
Travel and Tourism Research Association.

Tasci, A. D. A. & Holecek, D. F. (2003). “Measurement of the impact of visitation on
destination image.” In the Proceedings of 8’” Annual Graduate Education and
Graduate Students Research Conference in Hospitality and Tourism. Las Vegas,
Nevada, p. 650-654.

Tasci, A. D. A. & Knutson, B. J. (2003). “Online research modes: Waiting for leisure,
hospitality and tourism researchers.” Journal of Hospitality & Leisure Marketing,
10 (3-4), 57-83.

Turner, C. F., Ku, L., & Rogers, S. M. (1998). “Adolescent sexual behavior, drug use,
and violence: Increased reporting with computer survey technology.” Science,
280 (5365), 867-873.

US. Census Bureau. (2001). http://www.census.gov/

US. Census Bureau. (2001). “Race Alone or in Combination, for States, Puerto Rico, and
Places of 100,000 or More Population: 2000.” United States Census 2000.

220

(Release date: 04/2/2001). Retrieved June 16, 2003 on the WWW:
http://www.census.gov/population/cen2000/phc-t6.html

US. Department of Transportation, Bureau of Transportation Statistics. (1995). 1995
American Travel Survey.

Vogt, C. A. & Andereck, K. L. (in press). “Destination perceptions across a vacation.”
Journal of Travel Research.

Vogt, C. A. & Stewart, S. I. (1998). “Affective and cognitive effects of information use
over the course of a vacation.” Journal of Leisure Research, 30(4), 498-520.

Walmsley, D. J. & Young, M. (1998). “Evaluative images and tourism: The use of
personal constructs to describe the structure of destination images.” Journal of
Travel Research, 36 (3), 65-69.

Woodside, A. (1980). “First time versus repeat visitors: Analyzing multiple travel market
segments.” In Proceedings of the Travel Research Association 11'” Annual Travel
and Tourism research Association Conference. Salt Lake City, UT: Travel and
Tourism Research Association, p. 15-18.

Woodside, A. & Lysonski, S. (1989). “A general model of traveler destination choice.”
Journal of Travel Research, 27 (Spring), 8-14.

Woodside, A. & Pitts, R. (1976). “Effects of consumer life styles, demographics, and
travel activities on foreign and domestic travel behavior.” Journal of Travel
Research, 14 (Winter), 13-15.

Wright, D. L., Aquilino, W. S., & Supple, A. J. (1998). “A comparison of computer-
assisted and paper-and-pencil self-administered questionnaires in a survey on

smoking, alcohol, and drug use.” Public Opinion Quarterly, 62 (3), 331-353.

Yong, K. S., & Gartner, W. C. (in press). “Perceptions in international urban tourism- An
analysis of travelers to Seoul, Korea.” Journal of Travel Research.

221