9.33%». an; . 4 . , . , .
. , . .34.

y ‘- ugmﬁ‘lfﬁ.v ‘
: . V ‘ , ‘ B: ..
,, , .. r

.x$~..1v..:#
( . . . u. .3
1 ,

. ..
.Itﬁ : :5:

t. 5.31; .3...

i“ I): o. :1. .3:

 

 

 

 

: ‘1...\:I:.X.
... In ... \l

 

‘mumm1Mimilimiiuniiniiiu1111mm "

3 1293 01810 3998

This is to certify that the

dissertation entitled

Cohort Analysis: An Application to Canadian Tourism

presented by

Lori Pennington-Gray

has been accepted towards fulﬁllment
of the requirements for

Ph.D. Park, Recreation and

degree in

 

 

Tourism Resources

Major professor

 

Date August 3 , 1999
yyﬂ

MS U is an Affirmative Action/Equal Opportunity Institution 0-12771

 

 

LIBRARY
Michigan State
University

 

 

 

PLACE IN RETURN BOX to remove this checkout from your record.
TO AVOID FINES return on or before date due.
MAY BE RECAU£D with earlier due date if requested.

 

DATE DUE | DATE DUE DATE our:
MAX 3169ZQP4
cm ,3 g iggg; g

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

1/98 Mass-p.14

COHORT ANALYSIS: AN APPLICATION TO CANADIAN TOURISM

By

Lori Pennington-Gray

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

1 999

ABSTRACT
COHORT ANALYSIS: AN APPLICATION To CANADIAN TOURISM
By

Lori Pennington-Gray

The primary purpose of this study was the application of cohort analysis to travel
philosophies, beneﬁts sought and travel interests of Canadians. Five research questions
guided the course of this study: (1) IS there a signiﬁcant relationship between period and
travel philosophies, beneﬁts sought and interests of Canadians? (2) Is there a signiﬁcant
relationship between age and travel philosophies, beneﬁts sought and interests of
Canadians over time? (3) Is there a signiﬁcant relationship between generation and travel
philosophies, beneﬁts sought and interests of Canadians? (4) Does generation, age or
time explain the greatest variation in travel philosophies, beneﬁts sought and travel
interests within Canadians? (5) Which structural model best measures the effects of age,
period and generation on three constructs: travel philosophy, beneﬁts sought and travel
interests?

Three-way analysis of variance was used to examine the effects of age, period and
generation on the travel-related variables. This method was used because it revealed the
most information and violated the fewest assumptions. The total variation explained and
interaction effects were presented for each of twelve travel-related variables. The cohort

analysis revealed that interaction effects among age, period and generation explained the

most between-group variation for all of the twelve variables. Interaction effects were
more signiﬁcant than any of the main effects.

Several major discoveries arose from the study. They were: (1) age, period and
generation explained very little of the total variation in the responses to the twelve travel-
related variables, (2) the interaction of the three independent variables (age, period and
generation) explained the greatest percentage of variation for all twelve variables, (3) the
combination of the triad method and three-way analysis of variance was an effective and
informative method of performing cohort analysis, and (4) future researchers need to
consider (a) obtaining panel data which measures a number of cohorts over the life span,
(b) keeping variables the same from one survey to the next, (c) including variables which
might be more “generationa ” in nature, and ((1) including variables which permit the

assessment of period effects.

Copyright by
Lori Pennington-Gray
1 999

I dedicate this dissertation to my husband, Doug Gray, for all his support and
encouragement along the way.

ACKNOWLEDGMENTS

Sincere appreciation is extended to my Ph.D. advisor, Dr. Joseph D. Fridgen, for
his support and advice. He was instrumental in the success of my Ph.D. program. His
patience, thoughtfulness and intellectual stimulation are examples I will try to follow.

I wish to express my gratitude to Dr. Daniel J. Stynes, for his ongoing help in the
completion of this manuscript. His command of statistics and research methods helped to
strengthen this dissertation.

I am truly grateful to Dr. Richard Spreng (Department of Marketing and
Logistics) and Dr. Bonnie Knutson (School of Hospitality Business), members of my
Ph.D. committee, for their continued guidance in the Shaping of this manuscript and their
continued support of me personally.

Special thanks are extended to Deborah L. Kersetter (The Pennsylvania State
University), my Master’s thesis advisor and member of my dissertation committee, for
her friendship and support. I know I couldn’t have made it without her. She was always
there when I needed it.

Sincere appreciation to my ﬁ'iend and mentor, Ronald McCarville (The
University of Waterloo), for his encouragement from the very start. Without his
suggestion to continue my education, I wouldn’t be where I am today.

Many thanks are extended to Dr. Stephen Smith, Scott Meis and Gerald Ballie,
members of the Canadian Tourism Commission. I am extremely grateful to them for
access to the data from 1983 and 1995. Without access to the Canadian data. this study

would not have been possible.

vi

D01.

SUP;
me l

than

I want to thank and acknowledge the National Tour Association and Michigan
State University. The ﬁnancial support provided by each helped me to stay on track and
ﬁnish in three years. For this, both my husband and l are truly grateful.

With that, my deepest and warmest gratitude must be extended to my husband,
Doug Gray. His support through this long process never waned. I am utterly grateful.

To my family, I extend my deepest gratitude for your understanding, patience and
support throughout my graduate program. Special thanks to Kirsten and Steve for helping
me through those hard times (comprehensive exams, interviews and the big book report),

thank goodness you were around.

vii

CH:-
ME'

TABLE OF CONTENTS

LIST OF TABLES ............................................................................................................ x
LIST OF FIGURES ...................................................................................................... xiv
ABBREVIATIONS ........................................................................................................ xv
CHAPTER 1
INTRODUCTION AND STATEMENT OF THE PROBLEM ....................................... 1
Purpose of the Study .............................................................................................. 10
Research Questions ................................................................................................ lO
Delimitations .......................................................................................................... 11
Deﬁnitions .............................................................................................................. 11
Organization of the Study ...................................................................................... 13
CHAPTER 11
REVIEW OF THE LITERATURE ................................................................................ l4
Cohort Studies in Research .................................................................................... l4
Cohort Studies in Marketing ......................................................................... 15
Cohort Studies in Travel Research ............................................................... 20
Demographics as a Component of Travel Research .............................................. 23
Life Cycle as a Variable in Travel Research ................................................ 23
Age as a Variable in Travel Research ........................................................... 28
Psychographics as a Component of Travel Research ............................................ 33
Beneﬁts Sought as a Component of Travel Research ............................................ 45
Change Over Time as a Component of Leisure Research ..................................... 53
Strauss and Howe’s Theory Applied to the Canadian Population ......................... 62
Cohort Analysis as a Method of Statistical Analysis ............................................. 64
Triad Method ................................................................................................ 65
Regression Analysis with Dummy Variables ............................................... 65
Structural Equations Modeling ..................................................................... 66
Summary .................................................. 69
CHAPTER III
METHODOLODGY ....................................................................................................... 70
Source of Data ........................................................................................................ 70
Canadian Travel Attitude and Motivation Study .......................................... 71
Domestic Tourism Market Research Study .................................................. 72
Conversion of the Data .......................................................................................... 73
Treatment of the Data ............................................................................................ 76

viii

CHAPTER IV

DATA ANALYSIS AND INTERPRETATION ............................................................ 78
Sample Proﬁle and Sample Bias ............................................................................ 78
Age ................................................................................................................ 80
Income ........................................................................................................... 80
Marital Status ................................................................................................ 81
Education Level ............................................................................................ 81
Gender ........................................................................................................... 82
Number of Income Eamers ........................................................................... 83
Accounting for Large Sample Sizes ...................................................................... 84
Period and Travel-Related Variables ..................................................................... 85
Age and Travel-Related Variables ......................................................................... 88
Generation and Travel-Related Variables .............................................................. 91
Using Different Methods of Cohort Analysis ........................................................ 93
Multicollinearity as a Complexity of Cohort Analysis ................................. 95
Using Cohort Tables as a Method of Cohort Analysis ................................. 96
Results of the Analysis ............................................................................ 97
Using Analysis of Variance as a Method of Cohort Analysis .................... 122
Results of the Analysis .......................................................................... 124
Using Structural Equations Modeling as a Method of Cohort Analysis ..... 130
Study Limitations .............................................................................................. 133
CHAPTER V
CONCLUSIONS, DISCUSSION AND RECOMMENDTIONS ................................. 135
Conclusions and Discussion ............................................................................. 135
Recommendations for Future Research ............................................................ 146
APPENDICES
Appendix A: Canadian Attitude and Motivation Study .................................... 151
Appendix B: Domestic Tourism Market Research Study ................................. 160
Appendix C: Three-way Analysis of Variance for travel philosophies,
beneﬁts sought and travel interest variables ............................................... 166
BIBLIOGRAPHY ......................................................................................................... l 7]

ix

Table
Table
Table
Table
Table
Table
Table

iabl

Tab

Te

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.

Table 18a.

LIST OF TABLES

Strauss & Howe’s Generations .................................................................. 7
Example of a Cohort Analysis Table ....................................................... 16
Comparison of Three Theories of Generations ........................................ 21
Alternative Views of the Life Cycle Model ............................................. 24

Example of Perreault, Darden and Darden’s (1977) A10 Statements ..... 37

1983’s Study Number of Respondents by Province ................................ 71
1995’s Study Number of Interviews by Province .................................... 72
Common Variables in Both Studies ......................................................... 73
Age of Each Generation at Each Period .................................................. 76
Sample Sizes of Each Generation at Each Period .................................... 77
Demographic Characteristics of Respondents in Both Studies ................ 79

Observed Differences on Standardized Scale related to Original Scale .. 85
Means of Travel Philosophies, Beneﬁts Sought and Travel Interests for
1983 Study, 1995 Study and Pooled Dataset .......................................... 86
Standard Deviations of Travel Philosophies, Beneﬁts Sought and Travel
Interests for 1983 Study, 1995 Study and Pooled Dataset ....................... 88
Means for Signiﬁcant Relationships between Generation and Travel
Philosophies, Beneﬁts Sought and Travel Interests ................................. 90

Means for Signiﬁcant Relationships between Generation and Travel

Philosophy, Beneﬁt Sought and Travel Interest Statements .................... 92
Multicollinearity of Age, Generation and Period .................................... 96
Age by Period for “Money Spent on Travel is Well Spent” ................... 98

T3‘
T31

T31

Tab

Tabl

Table

Table

Table

Table

Table 3

Table 33

Table 33

Table 18b.
Table 180.
Table 19a.
Table 19b.
Table 19c.
Table 20a.

Table 20b.

Table 20c.

Table 2 la.

Table 21b.

Table 21c.

Table 22a.

Table 22b.

Table 22c.

Table 23a.

Table 23b.

Generation by Period for “Money Spent on travel is Well Spent” .......... 98
Age by Generation for “Money Spent on Travel is Well Spent” ............ 98
Age by Period for “Important that People Speak my Language” .......... 100
Generation by Period for “Important that People Speak my Language”101
Age by Generation for “Important that People Speak my Language”... 101
Age by Period for Importance of “Having Fun and Being Entertained”102
Generation by Period for Importance of “Having Fun and Being
Entertain ” ........................................................................................... 102
Age by Generation for Importance of “Having Fun and Being
Entertained” ........................................................................................... 103
Age by Period for Importance of “National and Provincial Parks” ....... 105
Generation by Period for Importance of “National and Provincial

Parks” ..................................................................................................... 105
Age by Generation for Importance of “National and Provincial

Parks” ..................................................................................................... 105
Age by Period for Importance of “Beaches for Swimming and

Sunning” ................................................................................................ 106
Generation by Period for Importance of “Beaches for Swimming and
Sunning” ........................... ' ..................................................................... 107
Age by Generation for Importance of “Beaches for Swimming and
Sunning” ................................................................................................ 107
Age by Period for Importance of “High Quality Restaurants” .............. 109

Generation by Period for Importance of “High Quality Restaurants” 109

xi

Table 23c.
Table 24a.
Table 24b.
Table 24c.
Table 25a.

Table 25b.

Table 25c.

Table 26a.

Table 26b.

Table 26c.

Table 27a.

Table 27b.

Table 27c.

Table 28a.

Table 28b.

Table 280.

Table 29a.

Table 29b.

Table 29c.

Age by Generation for Importance of “High Quality Restaurants” ....... 1 10
Age by Period for Importance of “Budget Accommodations” .............. 111
Generation by Period for Importance of “Budget Accommodations”... ll 1
Age by Generation for Importance of “Budget Accommodations” ....... l 1 1
Age by Period for Importance of “First Class Accommodations” ........ 113
Generation by Period for Importance of “First Class

Accommodations” .................................................................................. l 13

Age by Generation for Importance of “First Class Accommodations”. 113

Age by Period for Importance of “Shopping” ....................................... 115
Generation by Period for Importance of “Shopping” ............................ 115
Age by Generation for Importance of “Shopping” ................................ 115
Age by Period for Importance of “Nightlife and Entertainment” .......... l 17

Generation by Period for Importance of “Nightlife and Entertainment”l 17
Age by Generation for Importance of “Nightlife and Entertainment” .. 117
Age by Period for Importance of “Museums and Art Galleries” ........... 119
Generation by Period for Importance of “Museums and Art Galleries” 1 19
Age by Generation for Importance of “Museums and Art Galleries” 120
Age by Period for Importance of “Amusement and Theme Parks” ....... 121
Generation by Period for Importance of “Amusement and Theme

Parks” ..................................................................................................... 121
Age by Generation for Importance of “Amusement and Theme

Parks” ..................................................................................................... 1 22

xii

Ta

Table 30.

Table 31.

Table 32.

Table 33.

Table 34.

Three-Way ANOVA Examining Total Interaction of Age, Generation and
Period on the Importance of “Nightlife and Entertainment” ................. 125
Three-Way Analysis of Variance Examining Total Interaction of Age,

Generation and Period on the Importance of “Having Fun and Being

Three-Way Analysis of Variance Examining Total Interaction of Age,

Generation and Period on the Importance of “Beaches for Swimming and

Sunning” ................................................................................................ 128
F Values of Three-way ANOVA for all Variables ................................ 129
Percentage of Total Percent of Explained Variance .............................. 130

xiii

Figure 1.

Figure 2.

Figure 3.

Figure 4.

Figure 5.

Figure 6.

Figure 7.

Figure 8.

Figure 9.

Figure 10.

LIST OF FIGURES

Lifestyle Expectation Model ...... ' ............................................................... 2

A Conceptualization of The Role & Relationship Of Respondents

Motives ................................................................................................... 46
Transformation from 5-Point Scale to 4-Point Scale .............................. 74
Transformation from a 4-Point Scale to a Standardized Scale ............... 75
Mapping Comparisons on a Standardized Scale ..................................... 76

Graphical Representation of No Interaction Effects between Generation
and Period for “Money Spent on Travel is Well Spent” ......................... 99
Graphical Representation of Interaction Effects of Generation and Period
for “Having Fun and Being Entertained” .............................................. 103
Graphical Representation of Interaction Effects of Age and Period for
“Beaches for Swimming and Sunning” ................................................ 108
Graphical Representation of Interaction Effects of Age and Period for
“Nightlife and Entertainment” .............................................................. 118

The Model ............................................................................................. 131

xiv

LIST OF ABBREVIATIONS

ANOVA ............................................................................................. Analysis of Variance
CTAMS ................................................... Canadian Travel Attitude and Motivation Study
DTS .................................................................. Domestic Tourism Market Research Study
SEM .................................................................................... Structural Equations Modeling

XV

 

Chapter I
INTRODUCTION AND STATEMENT OF THE PROBLEM

While mass marketing is a necessary part of many marketing plans, such
strategies are no longer as efﬁcient or cost effective as they once were. The marketplace
is continually changing. The marketplace has moved from a manufacturing driven to a
consumer driven economy (Kotler, 1996). In 1956, Wendell Smith wrote in the Journal
of Marketing of the need to target homogeneous components of a heterogeneous market
rather than the market as a whole. He was the ﬁrst to call this strategy “segmentation.”
In “The Third Wave,” Alvin Toffler warned that the "mass market has split into ever-
multiplying, ever-changing sets of mini-markets that demand a continually expanding
range of options. . .” (1980, p. 248). To remain competitive today, companies must be
able to develop and reﬁne their products and services to meet segments’ needs and
preferences.

Segmentation bases (or variables) include the characteristics of individuals,
groups or organizations. All are used to divide the mass market into more homogeneous
groups. These variables may include for example, age, sex, race, income, household size,
education, personality traits, emotional needs, attitudes, interests and opinions, activities
and organizational memberships. Authors have classiﬁed segmentation variables into
four main areas: (1) demographics, (2) psychographics, (3) behavior and (4) lifestyle (see

Figure 1).

 

 

 

 

Generations

  

Demographics ‘ Psychographics

’§

Figure 1: Lifestyle Expectation Model, Knutson, 1982

 
    

Lifestyle

 

 

 

Segmentation has been applied to a wide range of situations in a variety of
different settings. Early studies tended to focus on demographic segmentation techniques.
This was especially true within the context of tourism. According to Gladwell (1990) “the
most ﬁequently collected data for identifying the characteristics of tourists are
demographic data” (p.15). Anderson and Langmeyer, (1980) among others, believe that
age has a signiﬁcant interpretive power on travel behavior. According to Wells (1975),
“demographic proﬁles, essential though they may be, have not been deemed sufﬁcient”
(p. 196). Thus, research has moved towards using life cycle characteristics to segment
and proﬁle travelers. Zuzanek and Smale (1992) stated that life cycle explains greater
variance in consumer behavior than single item variables, such as age. This is because
life cycle theory combines biological characteristic, psychological orientations and social
roles.

Anderson and Langmeyer (1980) building off the lifecycle concept, found that
family obligations such as time and ﬁnances were less of a constraint for older travelers

than younger travelers and that incomes appeared to be somewhat comparable between

the two segments. In addition, they found that younger travelers tended to be more active
while on vacation than older travelers.

Zimmerman (1982) documented that patterns of daily trip frequency varied over
the life cycle and that households were more mobile at some stages than others. He found
that families with children tended to increase their frequency of travel as the children
aged. Lawson (1991) studying patterns of expenditures and types of vacations over the
life cycle, found that individuals with small children tended to travel less often and spend
less money than individuals with older children or no children at all.

An additional approach to life cycle segmentation is the use of psychographic
proﬁles to provide an understanding of tourists. One of the ﬁrst studies to use Attitude,
Interest, and Opinion Scales or psychographics in recreation and tourism was conducted
by Perreault, Darden and Darden (1977). Their study served as the foundation for many
future studies which used AIO scales to study travelers (c.f., Gladwell, 1990; Mayo,
1975; Schewe & Calantone, 1978). For example, Gladwell (1990) used psychographics
to examine state park travelers and found that three identiﬁable groups of travelers varied
in relation to their vacation behavior. She found that budget conscious travelers took
fewer vacations and shorter trips than two other groups. Mayo (1975) looked at auto
vacationers to national parks. His study, which used 85 psychographic statements to
proﬁle users, documented that the tourist who is attracted to national parks enjoys
adventure, does not plan in detail, is impulsive, likes the outdoors, likes to escape crowds
and has power over other individuals.

Many researchers have moved beyond psychographics to the use of “beneﬁts

sought” to address differences between various segments of a market. Such beneﬁts as

escape from reality (Bello & Etzel, 1985; Wahlers & Etzel, 1985), status and prestige
(McIntosh & Goeldner, 1986), social interaction (Pearce & Caltabiano, 1983), novelty
(Snepenger, 1987), relaxation (Crompton, 1979; Woodside & Jacobs, 1985), and climate
(Loker & Perdue, 1992) for example, appear to be important “beneﬁts” of traveling.
Hence, these beneﬁts and others have been used to segment travelers into distinct groups.
Many researchers also have focused on the relationship between sociodemographic
variables and the types of beneﬁts sought through travel.

Gitelson and Kcrstetter (1990) highlighted beneﬁts sought through travel in
relation to such variables as age, family income, education and gender. Hawes (1988)
originally found that age was an important predictor of beneﬁts sought by travel.
Contrary to what previous literature had suggested, however, he found that as individuals
age, they are less likely to seek rest and relaxation while on vacation. Woodside and
Jacobs (1985) discovered that national origin impacts the types of beneﬁts sought when
travelling to Hawaii. They suggested that Canadian visitors, American visitors and
Japanese visitors to Hawaii differ widely in the types of beneﬁts sought when
vacationing.

Given the challenges with traditional demographic segmentation and the
incompatible results of psychographic segmentation, several researchers have suggested
that a combination of psychographics and demographics is the best way to delineate
segments (Abbey, 1979; Perreault, Darden and Darden, 1977; Shewe and Calantone,

1978). Cohort segmentation is one way to combine these variables (Bonnici, 1992).

the
syr
the
111'.”
an:
pre

COT

will
H01
Spei
SOlli
hen:

unlit

Thes
"inte
”int:
tooth
When
COHSii

QCCOw

A cohort is deﬁned as individuals within a delineated population who experienced
the same signiﬁcant life event within a given period of time. Birth cohort is used
synonymously with generation. Cohort analysis typically refers to any study in which
there are measures of some characteristic of one or more cohorts at two or more points in
time. The majority of studies have examined consumption patterns by use of cohort
analysis. Coffee purchases, cola purchases, and automobiles are just a few of the
products which have been addressed using cohort analysis to examine changing
consumption patterns. Rarer is the examination of interests or philosophies held by
speciﬁc generations. According to several authors, a generation has a distinct personality
which binds them together (Foot & Stoffrnan, 1996; Smith & Clurman, 1997; Strauss &
Howe, 1991). It is this personality which is said to affect attitudes and interests toward
speciﬁc experiences and products. To date, cohort research has not focused on beneﬁts
sought. Considering studies have found that there is a strong relationship between
beneﬁts sought and other sociodemographic variables, this dearth of research is
unfortunate.

Typically, a cohort analysis involves simultaneous evaluation of three variables.
These variables include time period, age and generation. The process involves
“intercohort tren ” which examines the change in different cohorts as they age and
“intracohort trends” which examines the change in the cohort from one point in time to
another. For example, an intracohort trend would follow travelers born in 1943-1964
when they were 36 years old and again when they were 50 years old. Three effects are
considered: age effects, period effects and cohort effects. Age effects are those effects

accounted for by biological, psychological and social aging. Period effects are those

effects accounted for by environmental changes such as marketing or changes in
measurement and cohort effects are those differences in socialization or cohort
composition.

To date, cohort studies have rarely been applied to the area of travel and tourism.
Warnick (1993) conducted one of the few studies that has used cohort analysis. He
examined the US domestic travel market. His study suggested that the Baby Boomer
generation’s participation rate in domestic travel has declined at the same or a lower rate
than the overall population change. In addition, domestic travel is experiencing grth
due, in part, to the youngest generation (13th Generation). He suggested that
understanding generational personalities would be helpful in future travel trend
prediction.

Opperman (1995) used a systematic examination of past behavior to explore
cohort effects on the travel patterns of German residents. He used a “self-report”
methodology to analyze lifelong travel patterns of the respondents. His results suggested
that destination region preferences were dependent on cohort membership. To truly
understand these results, however, Opperman suggested that more longitudinal studies
are needed.

According to Strauss and Howe, a generation is shaped by its "age location," or
by participation in epochal events that occur during their course of life (e. g., Depression,
World War II, the Korean War, the Vietnam War, and the Energy Crisis). During each
stage of the life cycle, a set of collective behavioral traits and attitudes is produced. They
call this a "peer personality." This peer personality, in turn, gives history a dynamic

quality. Attitudes of one generation affect how their children are brought up and later

 

how those children raise their offspring. As a generation ages, its inner beliefs retain a
certain consistency over its life cycle, like that of an aging individual. Although the
beliefs and behavior of a generation rarely are uniform across all members, those who
differ from the peer norm usually are aware of their lack of conformation. A variety of
data such as work habits, ambition, and even attitudes towards travel can offer evidence
about peer personalities.

Strauss and Howe deﬁned a generation as "a cohort-group whose length
approximates the span of a phase of life and whose boundaries are ﬁxed by peer
personalities" (p. 60). This deﬁnition includes two elements that are of importance: the
length of a generational cohort and its peer personality. Unlike other social philosophers,
they based the length of a generational cohort on the length of a phase of life.
Accordingly, they determined that each cohort-group is up to 22 years long and possesses
a special uniﬁed personality that will later distinguish it from other age-brackets.

Strauss and Howe (1991) concluded that there are seven generations currently
living in the United States. Two of the generations are in the 90+ age group. The
remaining ﬁve generations fall between one year and eighty-nine years of age (Table 1).
These generations are identiﬁed by the year that they were born and by their
personalities. First, there is the "GI Generation" born between 1901 and 1924. The next
generation is the "Silent Generation," followed by the "Baby Boomers." After the
Boomers, comes the "Thirteenth Generation" or "Generation X." The last, or youngest

generation, is the "Millennial Generation."

Table 1. Strauss & Howe’s Generations

 

 

Generation Year Born
GI Generation 1901-1924
Silent Generation 1925-1943
Baby Boomers 1944-1964
13th Generation 1965-1981
Millennial Generation 1982-present

 

Members of the GI Generation have been characterized as America's conﬁdent

 

r
and rational problem-solvers. Members of the earlier years of the GI generation, which '-
produced Walt Disney, Arthur Godfrey, and Ronald Reagan, are characterized by “their
jaunty optimism,” whereas the last half of the generation (e.g., Lee Iacocca, George 9;

Bush) is characterized by “a more clean-cut rationalism.” World War I had a signiﬁcant
impact on the personalities of this generation and created a generation of enormous
economic and political power. Its personality carries a strong government ﬂavor and is
willing to put trust in government and authority. As well, the GI Generation came of age
preferring clearly deﬁned sex-roles and is considered a generation that is "heavily male-
ﬁxated" (p. 264).

The next generation is the Silent Generation born between 1925 and 1943. This
generation has been described as the "transitional" generation or "middle" generation.
The Silent Generation is one of caution, indifference, lack of adventure and imagination,
and basically just "silent." The ﬁrst half of the generation is one born during the
depression and, as a result, gives freely to charity, have a tender social conscience and
believe in a fair process more than ﬁnal results. The last half of the generation can
remember World War II from their childhood, and many joined the Peace Corps to show

their generational bond.

The next generation is the well-recognized Baby Boom Generation. Although
most demographers deﬁned the Boomers as being born between 1943 and 1960, Strauss
and Howe describe this generation as being born between 1944 and 1964. This
generation was born to an era of optimism and growth. The Boomers are characterized
by a "quest for self" attitude. This ﬁxation on self resulted in a strong individual
conscience rather than duty to community. With this generation, "made-in-the-USA"
became passe and consumer loyalty weakened; however, the Boomers continued to desire
the best in products and services. In all areas of life, the Baby Boomers have exhibited
an attitude of self-absorption, perfectionism and high self-esteem.

After the Baby Boomers comes a generation that has been recognized by many
labels: the 13th Generation, the Baby Busters, and Generation X. This generation was
born between 1965 and 1981. The 13th Generation, or Generation X, is a generation
characterized by divorce, poverty, latchkey syndrome and non-traditional families. This
generation is skeptical about its economic future and realizes its members will have to
work harder to achieve the same standard of living as their parents. Forced to grow up
fast, Generation X exhibits a matter of fact approach to sexuality. In addition, this
generation lacks faith in government and believes its members are responsible for their
own future.

While a theoretical foundation has been laid out by Strauss and Howe (1991 ),
this study will provide the ﬁrst look at generations in travel and tourism context. It was
also the ﬁrst study to assess travel philosophies, beneﬁts sought and interests held by

Canadians using a cohort analysis.

Purpose of the Study
The primary purpose of this study was to assess the effects of a person’s
generation on speciﬁc travel philosophies, beneﬁts sought and interests held by
Canadians using a cohort analysis. The study used several approaches to assess the
effects of age, period and generation on travel philosophies, beneﬁts sought and interests
held by Canadians, including analysis of variance, regression analysis and structural

equations modeling. The strengths and limitations were discussed for each.

Research Questions

This study addressed the following research questions:

(a) Is there a signiﬁcant relationship between time period and travel philosophies,
beneﬁts sought and interests of Canadians?

(b) Is there a signiﬁcant relationship between age and travel philosophies, beneﬁts sought
and interests of Canadians?

(c) Is there a signiﬁcant relationship between generation and travel philosophies, beneﬁts
sought and interests of Canadians?

((1) Which effect (generation, age or period) explains the greatest variation in travel
philosophies, beneﬁts sought and travel interests within Canadians?

(e) Is there a structural model which best measures the effects of age, period and

generation on three constructs: travel philosophy, beneﬁts sought and travel interests?

10

Delimitations
The study was delimited to the sample of Canadians who participated in either the
Canadian Tourism Attitude Motivation Study (1983) or the Domestic Tourism Market
Research Study (1995). It was further delimited to residents over the age of 15 residing
in Canada who had traveled in the past 12 months. Only the areas of travel philosophies,

beneﬁts sought, interests and demographics were addressed.

Deﬁnitions

The following terms are deﬁned for the purpose of this study. For many of the
terms, a more detailed operational explanation is included in Chapter III.

Generational cohort: individuals grouped together in 22+ year conﬁgurations
(Strauss & Howe, 1991).

Baby Boom Generation: individuals born between 1944 & 1964 (Strauss &‘
Howe, 1991).

Beneﬁts Sought: reasons for participation; beneﬁts a person seeks from a product
or service.

Cohort: people within a geographically or otherwise delineated population who
experienced the same signiﬁcant life event within a given period of time (Glenn, 1977).

Cohort Analysis: typically refers to any study in which there are measures of
some characteristic of one or more cohorts at two or more points in time. The term does
not apply to cohorts compared at one point in time, in a cross sectional fashion (Glenn,

1977).

ll

Generation: a "special cohort group whose length approximately matches that of
a basic phase of life or about 22 years over the last three centuries" (Strauss & Howe,
1991, p. 34).

Generational proﬁle: "A peer personality is a generational persona recognized
and deterrrrined by (1) common age location; (2) common beliefs and behavior; and (3)
perceived membership in a common generation" (Strauss & Howe, 1991, p. 64).

G1. Generation: individuals born between 1904 & 1924 (Strauss & Howe, 1991 ).

Life Cycle: The family life cycle (F LC), is the typical process of life through a
series of stages, usually within a familial context (Stampﬂ, 1978).

Pleasure travel: travel for the purpose of pleasure or amusement, not business.

Silent Generation: individuals born between 1925 & 1943 (Strauss & Howe,
1991)

Sociodemographic variables: variables used to describe the population, including:
age, income, education, gender and occupation.

Structural Equations Modeling: the method of looking at causal relationships
between variables and testing for signiﬁcance.

Thirteenth Generation: individuals born between 1965 & 1981 (Strauss & Howe,
1991)

Triad Method: emphasizes three levels of analysis in order to compute observable
differences infers which effects produced the differences, and imputes causes (Glenn,

1977)

12

Organization of the Study

The document is organized into ﬁve chapters. Following this introductory chapter
is a review of the relevant literature. Relative to the research questions, the following
topics are discussed: cohort studies in marketing and travel research; demographics as a
component of travel research; psychographics and life style as a component of travel
research; beneﬁts sought as a component of travel research; change over time as a
component of leisure research; Strauss and Howe’s theory applied to the Canadian
population; and cohort analysis as a method of statistical analysis. The third chapter

presents the sources of data, how the data were transformed for this study and the

 

treatment or methods used for the cohort analysis. Chapter IV provides proﬁles of the
study population compared to the Canadian population and possible biases that may exist,
the results of three bivariate analyses, the two-way interaction results, the three-way
interaction results, the structural equation model results and limitations of the study.
Finally, Chapter V contains the conclusions and discussion, and recommendations for

further research.

13

85 11
W35
C0311
I636

COTT

P9P

DUI

eat

Chapter 11
REVIEW OF THE LITERATURE
The purpose of this chapter is to review the literature on the theory of generations

as well as attitudes and motivations of pleasure travel. The survey of related literature
was divided into six major areas: 1) cohort studies in research; 2) demographics as a
component of travel research; 3) psychographics and life style as a component of travel
research; 4) beneﬁts sought as a component of travel research; 5) change over time as a
component of leisure research; 6) Strauss and Howe’s theory applied to the Canadian

population; and 7) cohort analysis as a method of statistical analysis.

Cohort Studies in Research

The term cohort originally referred to a Roman group of waniors or soldiers
(Glenn, 1978). Today, however, a cohort is typically deﬁned as “those people within a
geographically or otherwise delineated population who experienced the same signiﬁcant
life event within a given period of time” (Glenn, p. 8). Cohort analysis deals with any
type of cohort: birth, age, marriage, education, etc. The term generation is used
synonymously with birth cohort. According to Glenn (1978), a birth cohort may have
natural or arbitrary boundaries. In the former, boundaries are set by speciﬁc years and
events in time which affect a group of individuals, thereby creating a cohort or
“generation.” A cohort with arbitrary boundaries consists of individuals “internally
homogenous in some important respect and distinctively different from persons born

earlier or later” (p. 9).

14

Cohort analysis typically refers to any study in which there is a measure of some
characteristic of one or more cohorts at two or more points in time. The term does not
apply to cohorts compared at one point in time, in a cross sectional fashion. However,
most cohort studies involve simultaneous cross sectional and period comparisons. This
study will apply cohort analysis which “measures some characteristics of one or more
cohorts at two or more points in time” (p. 9). For example, travelers born in 1943-1964
will be studied when they were 36 years old and again when they were 50 years old. The
study includes “intracohort trends” which examines the change in the cohort from one

point in time to another.

Cohort Studies in Marketing

One of the ﬁrst cohort studies in marketing was performed by Reynolds and
Rentz (1981). In the article, “Cohort Analysis: An Aid to Strategic Planning” (1981),
Reynolds and Rentz used cohort analysis to aid in strategic planning. The analysis was
used for the purpose of understanding certain environmental trends and aiding in
forecasting. The article presented the procedure using a nationwide survey conducted in
1969, 1975 and 1979. The researchers suggested that consumption data is often viewed
using cross sectional data and that more than one interpretation becomes possible by

using a “process viewpoint.” Their basic form of a cohort analysis is represented in Table

2.

15

Table 2. Example of a Cohort Analysis Table

 

 

Age Class 1930 1940 1950 1960 1970 1980

10-19 m 10 10 10 10 10

20-29 15 _1_§ 15 15 15 15 C12
30-39 20 20 29 20 20 20 Cl 1
40-49 25 25 25 g 25 25 C10
50-59 30 30 30 30 3Q 30 C9
60-69 35 35 35 35 35 35 C8
70-79 40 40 40 40 40 40 C7

 

C1 C2 C3 C4 C5 C6
1930 1940 1950 1960 1970 1980

 

“C” -represents ten-year cohorts measured at ten-year intervals. Cohort C7 is
underlined to follow movement through time.

Table 2 is an example of a cohort table. The table is a visual representation of
changes over time, within cohorts and among age groups. Variations can be interpreted
as either age effects (biological, psychological, and social aging), period effects
(environmental changes such as marketing or changes in measurement) or cohort effects
(historical differences in socialization, genetic change and cohort composition).

Reynolds and Rentz (1981) suggested an alternative model to the cohort table.
They suggested that a constrained regression model is more conceptually appealing.
Computing observable differences expands the analysis to include statistical differences.
Three possible types of observable differences exist (1) longitudinal differences (age +
period effects); (2) cross sectional differences (age + cohort effects) and (3) time lag
differences (period - cohort). Reynolds and Rentz (1981) looked at women’s agreement
with the statement “a woman’s place is in the home.” Their study computed observable
differences using t-tests and found that a combination of period and cohort effects were

responsible for differences to responses to this question over a 10-year period.

16

Rentz, Reynolds and Stout (1983) looked at changing consumption patterns with
cohort analysis. They looked at soft drink consumption in the US. and found that when
the data was examined using an aging interpretation, total soft drink consumption would
increase, however, a cohort interpretation would produce opposite results. Data was taken
from surveys of the US. population conducted in 1950, 1960, 1969 and 1979. Responses
were weighted to adjust for differences in demographic distributions between samples
and population. Data were examined using a constrained multiple regression model.
Cohorts, ages and periods were coded as dummy variables. The model with the best ﬁt
(the highest R2) was the cohort model (although the values were small for all three
models). The net impact of age, period and cohort effects shows an upward trend in cola
consumption, for each aging cohort due to the positive age effects. There were large
differences in cohort effects, which indicated that each succeeding cohort consumed more
soft drinks than the previous cohort. Rentz et a1. suggested that soft drink consumption is
inﬂuenced more strongly by consumer’s cohort membership than by their age and that
only by considering both age and cohort effects in succession can marketers make precise
forecasts of product consumption.

Rentz and Reynolds (1991) forecasted the effects of an aging population on coffee
consumption. Three measures of consumption were analyzed: (1) total coffee penetration
(percentage of the population who consumed any coffee the day before being surveyed);
(2) average cups consumed per drinker (among drinkers, average cups consumed the day
before being surveyed) and (3) cups consumed per person (total cups divided by total
population). Data for the study were taken from surveys of the US. population in 1950,

1960, 1970 and 1980. In each survey a stratiﬁed random probability design was used with

17

quotas for age and sex. Virtually identical consumption questions were asked on all
surveys. Three of the four surveys were personal interviews, while one was a telephone
survey. They found that the model that combined age, period and cohort effects ﬁt the
data the best. The regression model was then used to predict coffee consumption. They
found that forecasts based on cross sectional patterns of age produced results not only
different, but opposite to cohort analysis. The authors suggested that dummy variable
regression is an appropriate method for forecasting behaviors, attitudes and opinions.

Perhaps the most comprehensive book to date on the direct application of cohort
anlaysis is Smith and Clurrnan’s (1997) “Rocking the Ages, The Yankelovich Report of
Generational Marketing.” Smith and Clurman (1997) suggested that a generational
cohort deﬁnes who a person is, what s/he believes, and what s/he buys. The
Yankelovich report used 30 years of comprehensive data about consumers- their interests,
habits and lifestyles- to outline differences between the generations’ values and
motivations. The authors suggested that “each generation is driven by unique ideas about
the lifestyle to which it aspires” (1997, preface XIV).

Srrrith and Clurman (1997) grouped people into three generational cohorts-
Matures (born 1909-1945), Boomers (born 1946-1964) and Xers (born 1965-present).
These categories contrast with Strauss and Howe’s (1991) classiﬁcation system in that
they are broader and encompass more than one cohort group in both the Mature and Xer
generation. Based on Strauss and Howe’s theory of generational cycles, it would appear
that the combination of two or more cohort groups could alter “generational personality”
which is based on “shared life experiences in their formative years.” Much of the

information on values and motivations could possibly be broken out into more

18

comprehensive age groups in order to provide more precise target marketing. However,
the Monitor represents the ﬁrst publication devoted to documenting comprehensive
differences in interests, attitudes and values that affect purchasing behavior.

Smith and Clurman indicated that Matures are more adventuresome than previous
ﬁfty-plus generations. Boomers like experimentation and Xers like risk. However, they
suggest that marketers need to consider physical and psychological safety for Matures,
predictability for Boomers and a safety net for Xers (they like the risk but they don’t want
the fall to “kill” them). Research indicates that Boomers want to expand their cultural
experiences and “nostalgia travel is a big market” they think fondly of memories of their
youth. In contrast, Xers “make up another part of the growth market in mini-vacations,”
because they can’t afford to take a weeklong or two-week vacation. The authors
suggested that there is a greater quest for balance between work and leisure and that
“work will undergo the biggest metamorphosis” (p. 234). They suggested that just as
work shaped leisure in the eighties, leisure will shape work in the future.

The only Canadian study to date which looked at generations is called, “Boom,
Bust & Echo” (Foot & Stoffrnan, 1996). The authors looked at Canadian demographics
and indicated how Canada’s demographics are unique. Foot and Stoffrnan (1996)
suggested that different behaviors among unique demographic groups are useful
information for forecasting. One major difference between Foot and Stoffrnan’s theory
and Strauss and Howe’s (1991) theory is that they believe changes in social and
economic behavior result from the aging process and ﬂuctuations in birth rate as opposed

to changes in attitudes and behaviors which are homogeneous in each generation. They

19

suggested that the size of the generation drives consumption and that participation rates
are stable over time rather than dynamic.

According to Foot and Stoffrnan, there are eight cohorts that exist presently in
Canada. These generations fall between one and 100+ years in age. They are identiﬁed
by the year they were born and epochal events during their formative years. The ﬁrst one
is called “Pre-World War I,” born in 1914 and earlier. The next generation is the “World
War 1” generation, followed by “The Roaring Twenties,” “The Depression Babies,” t.
“World War II,” and the infamous “Baby Boom” generation. After the Baby Boom

comes GenX, followed by “The Baby Bust,” “The Baby-Boom Echo” and the “Future

 

Generation.” The generations outlined by Foot and Stoffrnan are shorter in year span
than those outlined by Strauss and Howe (Table 3). Also, Foot and Stoffrnan, indicated
that in Canada, GenXers are those born between 1960-1966. This is different from
Strauss and Howe’s deﬁnition which suggests that Generation Xers or the 13‘h

Generation (as they call it) are those born between 1961 and 1981.

Cohort Studies in Travel Research

The theory proposed by Strauss and Howe has been researched in relation to the
impacts of an aging population on domestic travel patterns. Warnick (1993) analyzed
whether generational characteristics and personalities impact U.S. domestic travel. He
used a database available through the Simmons Market Research Bureau to obtain a
national stratiﬁed random probability sample for each year from 1979 through 1991.
Warnick found that the Baby Boom Generation's participation rate in domestic travel

declined at the same or a lower rate than the overall population change.

20

Table 3. Comparison of Three Theories on Generations

 

 

 

 

 

 

Strauss & Howe Smith & Clurman Foot & Stoffman
GI Generation Matures Pre-World War I
1901-1924 1909-1945 1914 and earlier
World War I
1915-1919
The Roaring Twenties
1922-1929
Silent Generation The Depression Babies
1925-1943 1930-1939
World War II
1940-1946
Baby Boomers Boomer The Baby Boom
1944-1964 1946-1964 1947-1966
1?" Generation Trailing Boomer GenX
1965-1981 1960-1964 1960-1966
Xer The Baby Bust
1965-present 1 967-1 979
Millennial Generation The Baby Boom Echo
1982-present 1980-1994
Millenium Kids
1995-2010

 

The Silent Generation (those 35 to 44 in 1979 and then age 45 to 54 in 1989)
experienced a rise in participation rates in domestic travel. The rise in domestic travel
makes them a prime market for domestic travel. Each of the other generational segments
showed greater decline rates than the overall population. Warnick divided travelers into
light, moderate and heavy or frequent travelers. Results indicated that the US. domestic
travel market is experiencing growing participation rates among heavy travelers over the
age of 35. Also, in all market volume segments for those age 65 and older there has been
a decline in participation among travelers. Overall, domestic travel is experiencing a
generational decline for Baby Boomers and a portion of the Silent Generation, but a
generational growth for the 13"I Generation. He suggested that understanding

generational personalities will be helpful in future travel trend prediction.

21

Opperman (1995) looked at travel patterns of German residents. He used
longitudinal systematics to examine cohort effects. This type of research was used to
research lifelong travel patterns of respondents. Questionnaires were sent to a
convenience sample of 1,000 households in Germany. Respondents were asked to name,
for each year in their life, the purpose, length and travel party size for all trips which
lasted for more than three nights. Respondents could use diaries, travel logs or photo
albums for reference. A total of 1,975 trips were used for the analysis. Due to the small
sample size, only three cohort groups could be deﬁned for analysis. These grouping were
based on age cohorts, the age groups were 18-33 years, 34-48 years and anyone over 49
years. Opperman found that trips to Central Europe were decreasing from older to
younger generations and that destination region preferences were dependent on cohort
membership. Moreover, it was found that transportation usage differed by cohort groups.
Cohorts aged 34-48 and over 49 years appeared to have an increasing amount of air travel
compared to the youngest cohort group. Opperman did explain differences in
transportation by increasing discretionary income and the introduction of mass air
transportation. The overall concept of “period effects” was omitted from his analysis.
These effects help to explain variations in travel behavior in addition to age and cohort
effects. Oppennan’s results indicated that greater use of cohort analysis is needed to
examine changes in travel behavior. Longitudinal studies are needed to examine
differences in travel behavior and complement data gained from cross sectional inquires.

In the “Pop Culture” literature (literature published to attract a general audience),
a variety of theories of generations are presented. Interestingly, consistency among the

theories is not strong. The only consensus appears to be that monitoring generational

22

 

F.-

affects is a good tool for predicting the future. Lacking from the literature is an analysis
of the changing interests, philosophies and beneﬁts sought of cohort groups in the area of
travel and tourism. Strauss and Howe (1991) suggested that generational proﬁles affect
philosophies and interests, thus it is the researchers contention that an individual’s
generation should have an affect on philosophies, beneﬁts sought and interests for

pleasure travel.

Demographics as a Component of Travel Research

The following section of the literature, guided by the lifestyle model propopsed
by Knutson (1982); (see Figure l, Ch.1), highlights demographics as a predictor of travel
behavior. In fact, such explanatory variables as age, income, ethnicity and family life
cycle have been suggested as good predictors of travel behavior. Several studies have
investigated various age groups and their travel patterns (c.f., Anderson & Langmeyer,
1982; Blazey, 1987; Tongren, 1980; Rosenfeld, 1986; Vincent & de los Santos, 1990).
The demographic areas that are most pivotal to the concept of generations, however, are
lifecycle and age. Therefore, age and life cycle are the only areas that will be covered in

the literature review.

Life Cycle as a Variable in Travel Research

The literature has suggested that life cycle is a good segmentation technique for
travel (Cosenaz & Davis, 1981; Lawson, 1991). However, the examination of travel
behavior with respect to life cycle has not been extensively explored. All the research
studying the relationship between Family Life Cycle (FLC) and tourism is based on

previous models developed by researchers of F LC and consumer behavior.

23

Table 4. Alternative Views of the Life Cycle Model

 

 

Stage Lansing and Wells and Staples and Lawson
Morgan Gubar Murphy (1991)
(1955) (1966) (1979)

1 Bachelor- Bachelor- Young Single- Not classiﬁable
young, Single not young, single not young, single
at home living at home

2 Newly Married- Newly Married Young, Married Young singles
no children Couples- Without Children- under 25

no children no children

3 Full Nest 1- Full Nest 1- Other Young- Full Nest I-
youngest child youngest child young divorced preschool
under 6 under 6 with(out) children

children, young
married with
children

4 Full Nest 11- Full Nest II- Middle Aged- Full Nest 11-
youngest child 6 youngest child 6 married with(out) school aged
and over and over children children

divorced with(out)
children

married with (out)
dependent

5 Full Nest III- Full Nest 111- Older- Full Nest III-
older couple with older couples married older children,
dependent with dependent unmarried possible non
children children (divorced or dependent

widow)

6 Empty Nest- Empty Nest I- All Other Empty Nest I-
no children living no children still working, no
at home living at home, children

head in labor
force

7 Solitary Survivor- Empty Nest II- Empty Nest 11-
only one spouse head retired retired
living

8 Solitary Solitary

Survivor— Survivor-
in labor force retired

9 Solitary

Survivor-
retired

 

24

The more frequently cited research efforts (Table 4) are those of Lansing and
Morgan (1955), Wells and Gubar (1966), Staples and Murphy (1979) and Lawson
(1991).

Cosensaz and Davis (1981) examined the inﬂuence of life cycle on vacation
decision-making. A random sample of 252 families in the Dayton area was drawn to
exarrrine the decision making process over the life cycle. The study based on length of
marriage, average age of head of the house, and composition of the family. The study
focused on a new methodology for measuring dyadic dominance in the purchase
decision.

The authors found that in stage one of the life cycle (characterized by married less
than 14 years, average age about 34 years, children under 14 years), the decision-making
was syncratic (both partners made decisions together). In stage two (characterized by
married more than 14 years but less than 20 years, with children under 18 years and the
average age less than 44 years) the husband was the dominant decision-maker. Stage
three, (characterized by marriage for 20 to 30 years, children beginning to leave home,
average age approximately 54 years) was the stage where vacation purchases are at a
peak. This stage is predominantly wife—dominant. Stage four (characterized by marriage
more than 30 years, children have left home, and retirement) is largely wife dominant,
but moving back toward syncratic decision-making. Stage ﬁve shifts back to generally
syncratic, but is close to husband-dominant. This stage is characterized by couples
married more than 30 years with children still present and the dyad‘s age being less than
retirement (65 years of age). The last stage, the sixth stage is characterized by couples

married more than 40 years and retired without children at home. This stage is relatively

25

wife-dominant. The results signify the importance of recognizing the impact life stage
has on decision-making behavior.

Lawson (1991) studied patterns of tourist expenditures and types of vacation
across the family life cycle. His study addressed three separate areas: distribution of
tourists according to life cycle stage; expenditure on different elements of the vacation;
and differences across the life cycle in the type of holiday, activities pursued, amount of
travelling done, type of accommodation used and length of stay. The life cycle model L
used was based largely on Wells and Gubar's (1966) model. There were three alterations

to the model: (1) the break between Full Nest I and Full Nest II was at age ﬁve for the

 

youngest child (2) possible non-dependent children were included in the Full Nest III 9*
stage and (3) no allowance was made for solitary survivors still within the labor force
(see Table 4). The study classiﬁed expenditure data in nine different categories:
accommodations, meals, travel, arranged tours, entertainment, miscellaneous items, total
expenditures on vacation, total spent in destination, and other costs before arrival. The
average holiday expenditures were the greatest in the ﬁfth life cycle stage (Full Nest III,
older children, possibly independent) for all categories except entertainment, where the
third life cycle stage (Full Nest I, preschool children) spent the most money. Also, total
expenditures was greatest for the Full Nest III life cycle stage, with the seventh stage
signiﬁcantly behind. Life cycle theory (Wagner & Hanna, 1983) suggests discretionary
expenditures are high in stages one and two and fall at stage three before rising again in
stages ﬁve and six and then falling again. Theoretically, stages seven and eight would
see a decline in spending due to the ﬁnancial constraints of retirement. Lawson’s (1991)

results supported this theory.

26

Lawson (1991) presented a synopsis of the characteristics of each stage of the life
cycle. First, he found that young singles are highly active, but like to travel cheaply.
Visiting friends and relatives, being restricted to domestic travel, rafting, jet boating,
entertainment, and sporting events characterize their trips. Second, young couples reﬂect
the honeymoon market. Their trips are highly pre-arranged, tour-oriented; include visit
national parks; boating, sports, spas, and rafting. Third, the Full Nest 1 Stage is
characterized by visiting friends and relatives. Travel is limited due to the presence of
small children. Fourth, the Full Nest 11 Stage is mostly involved in visiting ﬁiends and
relatives and spas and sports. Fifth, the Full Nest III Stage is characterized by travel
through tour operators, use of rental cars and activities such as jet boating. Sixth, the
Empty Nest I is distinguished by travel through tour operators, use of rental cars, taxis,
and exploration of museums, demonstrating optimal purchasing power. Seventh, the
Empty Nest II Stage is identiﬁed by individuals who visit national parks, museums,
prefer traditional foods and enjoy cultural concerts. Eighth, Solitary Survivors' trips are
characterized by travelling to see friends and relatives, ﬁshing and enjoyment of cultural
concerts. Lawson (1991) concluded that life cycle is an effective segmentation tool for
marketers.

According to Zuzanek and Smale (1992), life cycle theory is suited to the study of
behavioral differences of speciﬁed population groups because of its functional approach.
They stated that life cycle, used as an independent variable, explains greater variance in
consumer behavior and leisure behavior than single item variables. This is because life
cycle theory combines biological characteristics with social role (marital status, income

level, social class, presence of children) and psychological orientation.

27

Age as a Variable in Travel Research

The travel literature has suggested that age is a good segmentation technique
(Anderson & Langmeyer, 1982; Tongren, 1980). Several studies have examined younger
travelers versus older travelers (Blazey, 1987; Shoemaker, 1990). As well, many travel-
related studies have looked at the theory of aging and its relationship to patterns of
behavior (Atchley, 1993; Javalgi, Thomas & Rao; 1992). This section examines age as a
variable for segmentation in the travel literature.

First, in 1980 Tongren studied the travel plans of the over 65 market pre- and
post- retirement. He found that (a) 52% of the respondents were planning to take 3-4 trips
per year when they reached retirement, (b) the older group engaged in a distinct pre-
retirement planning phase and a detailed post retirement search and implementation and
(c) over 50% of those over 65 still take a minimum of one to two trips per year.

Anderson and Langmeyer (1982) proﬁled travelers over and under the age of 50.
They sampled 826 households in the Dayton Standard Metropolitan Statistical Area. The
sample was stratiﬁed to represent populations proportionately according to their
demographic proﬁles in four counties. A response rate of 40% for the over-50 travelers
and 60% for younger travelers was achieved. The study examined family status, income,
education, occupation and health. Results indicated that the over-50 group preferred
simpler, pre-planned pleasure trips for the purpose of rest and relaxation or visiting
friends and relatives and were relatively inexpensive. While those under-50 traveled for
rest and relaxation, but were more likely to participate in outdoor recreation or visit
amusement facilities. Anderson and Langmeyer suggested that family obligations (time
and ﬁnancial) of the older segment are fewer yet their income appears to be somewhat

comparable to younger travelers.

28

The study also suggested that younger travelers tended to be more active while on
vacation and use vacation to escape from everyday routine. The authors suggested
marketers should pay more attention to the over 50 market because they are a lucrative
market.

Blazey (1987) examined the differences between participants and nonparticipants
in a senior travel program. He examined travel interests, constraints and demographic
characteristics of those over 55 years of age. The study population included a random
sample of 680 individuals who subscribed to a local newspaper that targeted seniors. A
mail-in questionnaire was sent out and 321 usable surveys were returned, yielding a
47.2% response rate. A screening question separated participants from nonparticipants
and a separate set of questions were asked of each group. Blazey suggested that
participants in a senior travel program are over-represented by females. Further, he found
that males to be signiﬁcantly more constrained by: lack of time, interruption of routine,
and other responsibilities, than females. Females were more likely to indicate they were
constrained by lack of a travel companion and safety. Interestingly, the nonparticipant
group reported higher incomes, was more likely to be pre-retirement age, and reported
poorer health conditions as a constraint to travel. Blazey suggested that health
considerations should be included in further research on constraints to travel.

Shoemaker (1989) studied members of the senior travel market and found that the
senior market could be segmented into several submarkets based on their reasons
(motivations) for pleasure travel. The study population consisted of residents’ age 55 and
older from the state of Pennsylvania. The sampling frame consisted of 5,000 randomly

selected residents. However, six counties in Pennsylvania were eliminated. The response

29

rate was 33%, yielding 407 usable questionnaires. Shoemaker used cluster analysis to
place respondents into groups. This procedure led Shoemaker to conclude that there were
three clusters: Family Travelers, Active Resters, and Older Set. These clusters were
distinctly different from each other on four variables: “to spend time with family,” “to
seek intellectual enrichment,” “to seek spiritual enrichment,” and “to meet people and
socialize.” Shoemaker concluded that the senior market can be segmented into smaller
homogeneous groups based on their motivations for travel.

Vincent and de los Santos (1990) focused on the senior winter traveler to Texas.
They examined winter Texans staying in mobile homes. A total of 1,757 questionnaires
were administered 70% of which were usable. The authors found that 70% of all winter
travelers to Texas staying longer than one month seek many incidental activities, plan
their travel and are from the midwestem states. The results also indicated that climate and
friendliness of the people were the two most important attractions when choosing a
winter vacation place.

Javalgi, Thomas and Rao (1992) studied senior and nonsenior groups in the US.
pleasure travel marketplace. They presented demographic information on both markets
and determined which types of pleasure trips they took and the differences in travel
related characteristics (mode of transportation, type of accommodations, etc.) The study
used data collected in a research project sponsored by Tourism Canada in 1985. The
results indicated that nonseniors are currently a better-educated group and more likely to
engaged in detailed information search process before making a purchase decision.
Seniors, on the otherhand, were more likely to buy trip packages and travel by bus or

airplane and use travel agents for making travel arrangements. The most common type of

30

trip for both groups was one taken for the purpose of visiting ﬁiends and relatives.
Nonseniors tend to take a close to home pleasure trip, a city trip, an outdoor vacation, a
resort vacation and a trip to a theme park or special event; whereas, seniors were more
likely to take a cruise or touring vacation. It is important for marketers to understand
these differences and aim effective marketing programs at both segments.

Zimmer, Brayley and Searle (1995) examined the differences between older
adults who travel and those who do not travel and the differences in their destination
choices. The data for the study were collected as part of the Canadian Aging Research
Network and the sample included 1,406 adults, 65 years of age and older, living in
Manitoba, Canada. No residents of long-term care institutions were included in the
study. Nontravelers were identiﬁed as those who had not taken a trip for the purpose of
pleasure within the last two years. A discriminant function analysis was performed to
determine how well a series of predictor variables were able to classify senior into those
who were travelers and those who were not travelers. The predictor which best-
distinguished travelers from nontravelers was age (r= -.68). That is, as age increased the
tendency to travel decreased. The two other best predictors were education and number of
mobility problems. Years of education was the best distinguishing factor (r=0.48).
Factors such as income, education, rural residency, willingness to spend money on
recreation, and health status inﬂuenced the choice of destination. One interesting ﬁnding
was that place of residence, be it nrral or urban, does not inﬂuence the choice of travel.
However, it does inﬂuence the chosen destination. This study showed that patterns of

leisure decision making tend to be consistent over the life span.

31

 

 

 

In Goeldner’s “1996 Travel Outlook” (1996), Peter Mason of Better Homes and
Gardens discussed the family leisure travel outlook. His report detailed family vacation
travel patterns prepared for Better Homes and Gardens by the US. Travel Data Center.
Generation X travelers (l 8 to 30 years) were found to spend the least on their family
vacation ($949), while Mature travelers (50 to 64 years) spent the most ($1,148). Baby
Boomers (31 to 49 year olds) averaged $1074 on their trips and the oldest age group,
“Older Matures” spent an average of $1038.

This section of the literature review has focused on literature relating to two main
demographics: life cycle and age and their relationship to travel behavior. The theory of
life cycle and aging are particularly interesting for understanding how behaviors change
as a result of movement through the aging process and different stages in life. However,
the literature suggests several constraints nright occur through using only demographics
as an explanation for travel behavior. For example, changing lifestyles have affected the
life cycle and no longer do people progress through the stages in a consecutive or
progressive method. In addition, people do not behave the way traditional aging theories
suggest. For example, leisure/travel behavior does not necessarily decrease with age.
Thus, it is important to look at other ways of predicting travel behavior, speciﬁcally,
analyzing the relationship between a person’s generation and travel behavior. This
approach may provide more insight into travel behavior than simply studying

demographics.

32

Psychographics and Life Style as a Component of Travel

Much of the life style and psychographic theory is based on the premise that
personality characteristics remain stable over time which leads to general patterns of
consistent behavior (Epstein & Teraspulsky, 1986; Swenson, 1990). For example, one of
the foundations for Plog’s allocentric/psychocentric model (one of the classiﬁcations of
travelers by personality types) is that although individuals change over time, their
personalities are enduring (Griffith & Albanese, 1996). Much of the literature which
examines psychographics and life style suggests that “although travel destination choices
will vary over time, the general types of destination decisions will remain relatively
stable” (Griffith & Albanese, 1996, p. 3). The following section of the literature review,
guided by the lifestyle model (Figure 1. Ch. 1) examines psychographics and lifestyle in
the travel literature.

Stanley Plog (1972) designed one of the most frequently cited models of traveler
proﬁles. He used behavioral characteristics to group people into “types” of travelers. He
found that the US. population fell along a continuum which be divided into ﬁve
segments: allocentric, near-allocentric, mid-centric, near psychocentric and
psychocentric. Psychocentrics are deﬁned as those who tend to visit well developed
tourist locations and prefer to travel with tour groups, while allocentrics are those who
prefer independent vacation experiences at destinations that have not yet developed a
mass tourism market. Further, he developed a predictive model using regression statistics
to determine which behaviors can be determined by each psychographic personality
characteristic and to what degree they are predictable. The estimates suggested that 4% of

the pOpulation are allocentric and 21% are psychocentric (Plog, 1985).

33

Cohen (1972) was one of the ﬁrst to identify tourists on a continuum based on
attitudes and interests towards the destination. Cohen’s tourist role typology was based
on the degree to which tourists seek novelty or familiarity in their travel. Four proposed
tourist roles were deﬁned as follows: the organized mass tourist, the individual mass
tourist, the explorer, and the drifter.

Mayo (1975) studied vacationers to National Parks and segmented them based on
their psychographic proﬁles. He sampled auto-vacationers at 24 sites across the US.
The sample population consisted of 670 auto-vacationers. Respondents were asked to
react to 85 psychographic statements that dealt with a variety of issues in their life and
vacation. Mayo used multiple regression to determine differences in the way individuals
rated the attractiveness of National Parks. Parks as a vacation destination were rated more
important among better-educated, higher income tourists. The tourist’s attitudes,
interests and opinions were correlated with the attractiveness ratings of the National
Parks. Eighteen of eighty-ﬁve were statistically signiﬁcant. Seven life-style traits were
the most signiﬁcantly correlated with the attractiveness ratings. These traits were: “The
Adventurer,” “The Planner,” “The Impulsive Decision-Maker,” “The Action-Oriented
Person,” “The Outdoorsman,” “The Escapist,” and “Tire Self-Designated Opinion
Leader.” More speciﬁcally, the results showed that the tourist who is attracted to
National Parks enjoys adventure and the unknown, does not plan the vacation in detail, is
an impulsive decision maker, action oriented, outdoors-like, likes to escape crowds and
people and feels he/she has persuasive power over other individuals.

Hawes (1977) stated that “Psychographics are meaningful... not merely

interesting” (p. 1). To validate this statement, he used 10 travel related activity, interest

34

and opin
'demogt
market 5
attractit'
educatic
when it
tale a Si
vacatior
categor
of cone
compo;
erplain
family

friends
m‘iel

female

be User

ClUSIer.

and opinion (AIOS) statements as a measure of psychographics. His thought was that
“demographics alone do not, in many cases, provide adequate discrimination between
market segments” (p. 3). Using data obtained from Market Facts, Inc., he found that (a)
attractiveness of a cabin on a quiet lake as a place to summer travel declined with age,
education and income, (b) females of higher education tended to be more spontaneous
when it came to vacation activities and (c) cities were considered an unattractive place to
take a summer vacation. A surprisingly large percentage of the population felt that
vacations should be planned for children. This was consistent across all life stage
categories. Hawes also looked at how well the 10 A10 statements interrelated. The use
of correlation analysis was employed to test the interrelatedness, followed by principal
components factor analysis. Four factors emerged with eigenvalues over 1.0 and
explaining 52% of the variance. The data showed that beneﬁts such as strengthening
family ties, recuperation from work, culturally enriching experiences, and strengthening
friendships were the most important variables associated with pleasure travel. Also, the
“travel now and play later” attitude was not thought of strongly by either males or
females, but particularly not by married, females aged 35-44 years.

Perreault, Darden and Darden (1977) suggested that psychographic scales could
be used to examine vacation life styles. Their study used AIO statements that were
clustered to show that vacation life styles differ according to sociologically relevant
variables. The purpose of the paper was to report “ a psychographic study designed to
produce a realistic portrait of vacation types” (p. 208). The data were collected with a
pre-tested mail questionnaire sent to 2,000 households in the Southeastern and Great

Lakes regions of the United States. The response rate was 33%, which provided 670

35

usable surveys. Measures of activities, interests and opinions provided the basis for the
study. Twenty-eight vacation-speciﬁc variables were used (Table 5). Households with
similar AIOs were grouped together.

Cluster analysis procedures produced ﬁve distinct groups of types of vacationers.
The groups were named: “The Budget Traveler,” “The Adventurer,” “The Homebodies,”
“The Vacationers,” “The Moderates.” The largest cluster, the “Budget Traveler,”
represented 28% of the total sample. Socioeconomic variables indicated that
“Homebodies” on average were the oldest; “Adventurers” had the greatest years of
formal education; “Vacationers” had the lowest paying jobs; and “Moderates” had
average scores across all the socioeconomic variables. The authors concluded that
socioeconomic characteristics and general behavioral characteristics were consistent with
the groups’ speciﬁc attitudes and interest in vacation activities.

Schewe and Calantone (197 8) proﬁled tourists to Massachusetts using
psychographic segmentation. In conjunction with the Massachusetts Department of
Commerce and Development, auto travelers were randomly sampled during the 1974-
1975 year. Individuals driving cars with non-Massachusetts license plates were
approached on the turnpike and given postage paid envelopes containing questionnaires.
A cash incentive was used to increase the response rate. A total of 1,498 useable surveys
were sent back. The questionnaire included 101 lifestyle statements, which were
measured on a six point likert type scale. The study indicated that the typical tourist was
family and socially oriented, satisﬁed and optimistic. Proﬁles were consistent over

seasons and over the year.

36

 

Table 5. Perreault, Darden and Darden (197” A10 Statements

 

Scale Name

Sample Statement

 

Vacation Travel Interest
Generalized Self Conﬁdence
Opinion Leadership

Plan Ahead Travel

I enjoy looking at vacation or travel magazines

I like to be considered a leader

My ﬁ'iends often ask my advise about vacation travel
I would never leave on a vacation without
reservations at my destination

 

Undecided Vacationer
Information Seeker
Impersonal Information Seeker

Leisure Time Venturesome

I often have difficulty in deciding where to visit on a
vacation

I often ask the advise of my friends regarding
vacation spots to visit

I often contact a travel agency for information about
vacation travel

I enjoy vacation travel to new places

 

Money Oriented Vacationer

Predisposition to Travel
Camper Traveler
Camping Impact

Given a windfall of money, I prefer to spend it on
vacation travel than something else.

I enjoy vacation travelling away from my hometown
My family enjoys camping.

Camping sites are beginning to overcrowd some of
the popular vacation areas

 

Relaxing Traveler

Weekend Traveler

If I can’t completely relax, I don’t feel that I’ve had a
vacation.

I prefer to take several short vacations rather than a
few longer ones

 

Cosmopolitan Traveler I would not vacation in an area where ﬁrst class
accommodations were not available.

First Class Traveler I spare no expense in making vacation travel
arrangements

One Up-Manship Traveler Having my friends over to see my vacation slides is

Educational Traveler
Sports Participation
Jet Setter or Vagabond Traveler

Economic Traveler

the best part of my vacation.

Educational vacations are the most fun and the most
rewarding

My vacation is normally planned so that 1 can
participate in my favorite sport

On vacation, I would not hesitate to hop on a plane
and travel to a more fun location

On vacation, I often eat in more economical
restaurants

 

Historical Vacation

Town Vacationer
Sportsman-Spectator

Functional Gregariousness

Visiting historical locations is an important
consideration in planning my vacation

The atmosphere of guided tours is too stiﬂing

I normally plan my vacation around watching my
favorite sporting event

The most important part of any vacation is meeting
new people

 

Familial Traveler
Vacation Gregariousness

Families that vacation together are happier
When I have a choice, I would rather travel to a new
place and meet new people

 

37

way tout
Canton

recomme

’L‘.

this by e
tdentitic
Commu
statemet
consiste
demogr;
WT e 2111
demogr
asked I(
dfmogr
1011i tra‘
designe
PTOftles
U10re q
20,1. A

ShOttld

Psychographics were used to provide a way of viewing the tourist and reveal the
way tourists felt about a variety of subjects. Similar to Hawes (1977), Schewe and
Calantone suggested that “psychographics cannot do the job alone” (p. 20). They
recommended the use of demographics to proﬁle groups.

Abbey (1979) asked the question “Does life-style proﬁling work?” He answered
this by examining 12 vacation life-style dimensions and 8 demographic variables. The
identiﬁcation of the life-style dimensions was based on Bauer’s Transactional
Communication Theory. Respondents indicated on a seven point scale which of the
statements they strongly agreed with or strongly disagreed with. The ﬁnal questionnaire
consisted of two parts: 48 statements which measured the life-style dimensions and 8
demographic questions. A total of 800 tour travelers were surveyed in Las Vegas. Half
were air travelers and half were motorcoach travelers. Eight proﬁles were drawn using a
demographic and life-style dimensions. Based on the proﬁles, two tour operators were
asked to design four tour packages; two budget and two ﬁrst class. Tour 1 was based on
demographic data and Tour 2 was based on life-style data. The results suggested that
tour travelers prefer tours designed with vacation life-style information rather than
designed with demographic data. The preferences were maintained across all eight
proﬁles. Abbey suggested that “life-style allows the designer to create a package that is
more compatible with the motivations, attitudes, and opinions of the tour travelers” (p.
20). Again, similar to the studies introduced earlier, results indicated that life-style data
should not replace demographic data but rather work in conjunction with it.

Bryant and Morrison (1980) studied visitors to Michigan by segmenting the target

market group in Michigan by vacation activity preferences using activity and recreational

38

interest 5
their dert
anions :1

effective

socioden
vacation
'mited h
predict 11
Kingdom
Were use
P351 12 r
The Slat:
1978).

eigettt'al
Mum:

chal’attte

interest scales. They developed six vacation activity preference types and determined
their demographic proﬁles using cross tabulations. The results indicated that marketing
actions and promotional activities could be focused at each of the groups to increase
effectiveness of the promotion and management priorities.

Schul and Crompton (1983) used travel speciﬁc psychographic and
sociodemographic variables to predict and explain search behavior of international
vacationers. The objectives of their study were to determine the relative ability of a
limited number of travel psychographics statements and sociodemographic variables to
predict travel search behavior. The data were obtained from 560 residents of the United
Kingdom who completed questionnaires concerning travel. Of the returned surveys, 544
were useable. Qualiﬁed respondents were over 25 years and had traveled by air in the
past 12 months. Sixteen psychographic statements were included in the questionnaire.
The statements were derived from a previous study done in the UK (Hay Associates,
1978). Factor analysis was performed on the 16 statements, resulting in 6 factors with
eigenvalues greater than one. Then a two step multiple regression analysis was
performed to predict search behavior using the psychographics and sociodemographic
characteristics. The results suggested that search behavior is better explained by travel
speciﬁc psychographics than by demographics. Travel speciﬁc psychographics were
found to correlate with length of time of the search. However, psychographics did not
explain the tendency to use multiple sources prior to the trip.

Adding to the literature on psychographics, Gladwell (1990) studied Indiana State
Park inn users and found identiﬁable groups of users with signiﬁcant differences on a

variety of characteristics. She surveyed 1,200 users of Indiana State Park Inns in 1983.

39

No. fol?
“line“
eluSter
racatit
and Sit

on the?

ditiere

allocet
"Actit
“ester
partici
(1987.;
relatin
Energ}
AilOCe
was In
includ.
the Us:
intern;
"133511:
test the

SUPPOr

No follow-up mailings took place. Three groups of users were identiﬁed:
“Knowledgeable Travelers,” “Budget Conscious Travelers, “ and ‘Travel Planners” using
cluster analysis on 26 A10 variables. The three groups had differences in relation to
vacation travel behavior measures. “Budget Conscious Travelers” took fewer vacations
and shorter trips than the other two groups and “Travel Planners” tended to travel further
on their vacation than the others. This study indicated that lifestyle measures could
differentiate travelers into types of travelers.

Nickerson and Ellis (1991) examined the extent to which Plog’s model of
allocentrism/psychocentrism and energy traveler types could be explained by the
“Activation Theory.” LISREL was used to analyze 171 randomly selected alumni of a
western state university. All subjects were selected by telephone in order to solicit
participation in the mail in questionnaire. The study examined the two theories. Plog’s
(1987) theory on psychocentrism and allocentrism and Fiske and Maddi’s (1961) theory
relating energy and allocentrism/psychocentrism. In activation theory, activation or
energy is assumed to be correlated with variety, intemality, and passivity.
Allocentrism/psychocentrism were measured using Plog’s ﬁve-item scale (1972). Energy
was measured using Thayer’s (1967) activation-deactivation checklist. Activation theory
included three other variables which were measured as follows: (1) variety- measured by
the use of the arousal seeking tendency scale (Mehrabian & Russell, 1974); (2)
intemality- measured using Eysenck’s Personality Inventory (1968); and (3) passivity-
measured through Lurnpkin’s brief locus of control scale (1985). LISREL was used to
test the ﬁt of the data to the model. All path coefficients were signiﬁcant. Thus, results

supported the use of the activation theory for explaining Plog’s model. The model also

40

sugges
orhog
locus (
anew

prefer:

lookirt
and de
maintz
Was be
cousin
behati

attribu

10llllSlc

35 "the

suggested that Plog’s model of allocentrisrn/pychocentrism and energy are not
orthogonal. The study concluded that four travel personalities exist: introvert-extemal
locus of control, extravert-extemal locus of control, introvert-intemal locus of control and
extravert-internal locus of control. Each personality type varied in terms of destination
preferences, travel companions, interaction with local cultures, and degree of activity.

Weaver, McCIeary, Lepisto and Darnonte (1994) expanded the literature by
looking at the relationship of destination selection attributes to psychological, behavioral
and demographic variables. A survey conducted for the Adult Longitudinal Panel (ALP),
maintained by the center for Adult Longitudinal Studies at Central Michigan University
was based on a life span model and involved data on psychology, demographics and
consumer behavior. The purpose was to examine whether psychographic, consumer
behavior and demographics can successfully predict groups of individuals clustered by
attributes sought. Weaver et al. found four cluster solutions that represented different
proﬁles in terms of destination attributes sought. The study revealed two behavioral
characteristics (“seek the new” and “fashion conscious”) and one demographic variable
(age) that explained statistically signiﬁcant differences among groups. The study
indicated that no psychological variables or psychographic characteristics showed
signiﬁcant differences among the groups. This study added to the literature by exarrrining
attributes sought in a destination and their relationship to demographic, psychographic
and consumer behavior variables.

Waryszak and Kim (1994) utilized psychographic segmentation of Korean
tourists visiting Australia to predict their travel behavior. They deﬁned psychographics

as “the study of personality, life styles and attitudes of individuals and groups, especially

41

in quam
chaser
(p.66).

iBum
0181 K
duenni
‘Ophnc
"Knowl
hart]
leaden
trgnuz
Leader
dearab
ﬁbdhi
humus

ﬁancn

Hegg
Ha\‘el:
Rokea,
“Win
Ilium

mea$11:

in quantitative terms, based on the belief that such readily measurable, descriptive
characteristics serve as better predictors of behavior than demographic characteristics”
(p.66). They used previous works by Johnston et al. (1986), Van Minder (1987), Evans
& Berman (1988), and Michman (1991) to shape their deﬁnition. The sample consisted
of 84 Koreans who had just returned from their vacation to Australia. A factor analysis
determined the underlying motivations of Korean tourists. Five factors were identiﬁed:
“Opinion Leaders,” “Pleasure Seekers,” “Experience Chasers,” “Adventurers,” and
“Knowledge Seekers.” Cross tabulations on demographics for each group indicated a
higher percentage of males in “Experience Chasers,” “Pleasure Seekers” and “Opinion
Leaders.” Older males with slightly higher average incomes, many of whom traveled in
organized groups dominated the “Experience Chasers” group; whereas, “Opinion
Leaders” were dominated by younger, less affulent males. The only differences among
desirable attributes were among “Opinion Leaders” and “Adventurers” with regards to
food in Australia. The article indicates that “demographic data is still a more reliable
dimension in measurement when a comparative demographic proﬁle can be compiled
than other approaches” (p.8).

Madrigal (1995) examined the relationship between List of Values (LOV) and
Plog’s traveler personality type scale. His study addressed the ability of each to predict
travel style. A convenience sample of 514 visitors to Arizona were asked to respond to
Rokeach’s Value Survey (1973), which consisted of 18 instrumental values and 18
terminal values. The LOV scale used in the study consisted of nine personal values
measured on a ﬁve-point scale. Plog’s allocentric/psychocentric model was used to

measure personality type. Discriminant anlaysis was used to determine whether the nine

42

personal values could differentiate group travelers ﬁ'om independent travelers. Of the
nine values in the LOV scale, four values differentiated between the groups: being well
respected, warm relationships with others, self-fulﬁllment and accomplishment. A
signiﬁcant predictor of group travelers was being well respected, whereas excitement and
self-respect were predictors of independent travelers. The study indicated that personal
values were a better predictor of independent vs. group travel than Plog’s
allocentrism/psychocentrism scale. Values of self fulﬁllment and accomplishment were
signiﬁcantly related to allocentrism, whereas, security was signiﬁcantly related to
psychocentrism. The results also suggested that Plog’s personality types can be
conceptualized more broadly into locus of control and that internal and external
motivators are consistent with Plog’s description of allocentrics and psychocentrics.
Grifﬁth and Albanese (1996) examined the correlation between Plog’s instrument,
underlying theoretical constructs and actual travel behavior. A sample of 145
undergraduates at a large rrridwestem university was given a questionnaire that examined
Plog’s 28 personality characteristics as well as a modiﬁed Nickerson scale. Nickerson’s
scale used two single-items to measure allocentrism and psychocentrism. Two types of
vacations were described: one which used adjectives such as novel, different, nontouristy
and the other which used adjectives such as national hotel chain, tourist shops and group
tour. The dependent variable in this study was travel behavior. The respondents were
asked to explain their travel behavior on their last trip. Correlations between travel
behavior and responses to Plog’s instrument were executed. Signiﬁcant correlations
between Nickerson’s scale, Plog’s 28 personality characteristics and actual travel

behavior appeared, suggesting that there is support for Plog’s model of travel behavior.

43

Thus. Plog.S
segmentation
Silver
the southeast
to determine
characteristic
commerciall;
states. A mo
response rat:
socioeconon
Dunlap and
measure of i
horizon. inf
anall’sis on
educationfh
infUrination
Sought. F0
Milly of.
and Whose
anall‘SiS w;

emerged. '1

Thus, Plog’s theoretical model is said to “provide a ﬁrm foundation for psychographic
segmentation in the ﬁeld of travel research” (p. 8).

Silverberg, Backrnan and Backrnan (1996) investigated nature-based travelers to
the southeastern United States using psychographic segmentation. Their study attempted
to determine if different segments in terms of environmental attitudes, demographic
characteristics and travel behavior. A sample was randorrrly selected from a
commercially purchased mailing list. A total of 1,200 individuals were contacted in three
states. A modiﬁed Dillman total design method was used to collect the data and yielded a
response rate of 36%. The questionnaire had ﬁve sections: trip purpose, trip behaviors,
socioeconomic and demographic information, environmental attitudes, and A105.
Dunlap and Van Liere’s (1978) NEP (new environmental paradigm) was used as a
measure of environmental attitude. Then, travelers behavior was assessed via planning
horizon, information source, accommodations and travel party type variables. Factor
analysis on the 46 A10 statements was performed. Six dimensions were identiﬁed:
education/history, camping/tenting, socializing, relaxation, viewing nature and
information. Then, cluster analysis was performed according to the nature-based beneﬁts
sought. Four clusters emerged. The only signiﬁcant difference existed in the primary
activity of viewing nature between individual’s whose purpose of trip was nature-based
and whose purpose was education. No other signiﬁcant differences were found. Factor
analysis was performed on the dimensions of environmental attitudes. Two factors
emerged. The factors were labeled conservationists and consumptives. Cluster analysis
was performed on the factors to proﬁle the groups by demographic and travel behavior

characteristics. Two groups clustered, they were named campers and noncarnpers. The

results suggesmd 1
sitlillllng appeal
Wed that nat1

mcbenelll.

Haley (19
consumCTS may 5
researchers have
lenehts sought.

Crompto
identiﬁed nine n
conceptual fratnl
psychological at
relatation, prest
The last two mo
Sociopsycholog

cultural motives

results suggested there are differences among nature-based tourists. For example,
socializing appeared to be more important to one group than another. In addition, it
appeared that nature-based travelers are looking for a package of beneﬁts rather than only

one beneﬁt.

Beneﬁts Sought as a Component of Travel Research

Haley (1968) was one of the ﬁrst researchers to suggest that different segments of
consumers may seek unique beneﬁts ﬁom using a product. Since his initial effort, many
researchers have attempted to discover motivations for travel and what might impact
beneﬁts sought.

Crompton (1979), for example, looked at the pleasure vacation travel market and
identiﬁed nine motives for travel. These motives for travel were developed into a
conceptual framework (see Figure 2). Eight were classiﬁed as sociological or
psychological and included: escape from normal life, exploration, evaluation of self,
relaxation, prestige, regression, enhancement of family relations, and social interaction.
The last two motives were more cultural: novelty and education. Crompton found that
socio-psychological motives were not related to the destination attributes; whereas

cultural motives and novelty were at least partially related to the destination traits.

45

 

 

Gr Disequilibrium

Break from
Routine
Stay in Non-pleasure
Home Locale Vacation Travel
Go on a Pleasure
Vacation

Socio-Psychological Cultural
Disequilibrium Disequilibrium

 
    

 

 

 

 

Figure 2. A Conceptualization of the Role and Relationship of Respondent’s Motives
(Crompton, 1979, p. 414)

Young, Ott and Feign (1978), using a national probability sample, divided the
US. travel market into six segments using beneﬁts as a variable. The sample was
collected by the Canadian government. Of the six segments identiﬁed, some included
visit friends and family, doing nothing, visiting a place that they have never been before.

Woodside, Moore and Etzel (1980) studied the vacation travel behavior and
perceived beneﬁts of South Carlonia’s residents. A questionnaire on travel and tourism
behavior was mailed to 834 members of the University of South Carolina's consumer
panel. Factor analysis was performed on 50 psychographic statements. Eleven factors
emerged: educational, affection, recommended place, physical activities, refreshing,
unique experience, rewarding, family togetherness, self-conﬁdence, unplanned, and

unwinding. Woodside et al. (1980) found that an important element in promoting

46

products is stressing the educational beneﬁts and unique experiences, and that residents
were more interested in personal beneﬁts than out of state travelers.

Crask (1981) explored vacation attribute statements and uncovered ﬁve
vacationer groups. The ﬁve groups were found to differ on most demographic variables.
Cluster one was a group that likes to travel for rest and relaxation or to get away from it
all and was characterized by middle age, middle income and the presence of school-aged
children. Cluster two depicted individuals who indicate that sightseeing was the most
important reason for travelling. High education and high income characterized this
cluster. Cluster three was cost conscious and included attraction-oriented vacationers.
This group was interested in shopping, amusement parks, and the cost of their trips and
was mostly young adults, middle to low income, and at least some post-secondary
education. Cluster four was characterized as sports enthusiasts and was looking for
sporting facilities when travelling. The group was characterized by low income and was
found to be either very young or over the age of retirement. Members of the ﬁfth group
were looking for camping experiences and opportunities to sightsee; they were mostly
middle age, with young children and were in the lower income brackets. Crask found
that members of each group were looking for different experiences when they vacationed
and that the demographic proﬁles differed by group.

Etzel and Woodside (1982) offered an approach to studying near-home
vacationers and distant travelers. A questionnaire addressing what people seek on
vacation was sent out to 900 Dayton area households. Fifty psychographic statements
taken from the literature, relating to the beneﬁts perceived by the respondents, were

included in the questionnaire. The researchers found that distant travelers have vacation

47

experiences and expectations quite different ﬁ'om those of near-home travelers. They
found that the longer the trip was, the more it was viewed as an adventure, exciting, a
change of pace and a once in a lifetime experience. Near-home travelers placed more
importance on relaxation and slower paced experiences than did distant travelers. Etzel
and Woodside (1982) suggested that when beneﬁt-oriented items are "combined with the
degree of satisfaction, the tourism manager is provided with a useful picture of the
consumer's desires" (p. 13).

Pearce and Caltabiano (1983) examined travel motivations by having respondents
describe positive and negative experiences from their last pleasure trip. Maslow’s (1943)
hierarchy of needs was used as a descriptive tool to code the responses. They found that
positive experiences were associated with fulﬁllment of love, physiological needs and
self-actualization. Negative experiences were associated with lack of physiological and
safety needs. Pearce and Caltabiano suggested that promoting how attributes address
particular needs is more effective than merely promoting the attributes themselves.

Adding to the literature on beneﬁts sought, Mazanec (1984) surveyed 250
Austrian citizens who indicated that they would take a pleasure vacation. Three segments
were identiﬁed. Segment I (45% of travelers) demonstrated interest in almost any beneﬁt
offered and appeared to be versatile and extravagant. Segment II (27% of travelers)
showed a higher interest in swimming and evening entertainment. Segment III (28% of
travelers) revealed an interest in nature travel, not comfort of accommodations. Mazanec
also broke each segment down by age and found that Segment I was predominantly
middle aged, Segment II was mostly young, and Segment III fell in the 50 and older

category. This study also indicated that some beneﬁts might be incompatible, depending

48

on the Segmen
interest in sigl
oriented and 1
Wood
MWMd
departure in a
nationalities.
and past expe
nationality. 13
Canada the l
501th by the
most impona
included son
bcnefrt of be
logemﬁmess.
Suategies for
Pitts
liaielleisure
and needs in
individuals V
Values Date
Ollhe Uane

SCale” lROlig

on the segment. Segment 11 members were highly motivated by swimming and had a low
interest in sightseeing or climbing or walking; whereas Segment III was highly nature-
oriented and less concerned with entertainment and shopping.

Woodside and Jacobs (1985) looked at different national samples and the beneﬁts
they sought through travel. Sixty travel groups were approached while waiting for
departure in airport waiting areas. A questionnaire was developed for each of three
nationalities. The questionnaire consisted of 26 likert type questions on beneﬁts realized
and past experiences. Woodside and Jacobs reported three different segments based on
nationality. Beneﬁts sought through travel were examined for three countries, including
Canada, the United States, and Japan. They found signiﬁcant differences in the beneﬁts
sought by the different nationalities. Canadian visitors reported rest and relaxation as the
most important beneﬁt of travelling to Hawaii. Americans sought similar beneﬁts, but
included some unique features. Mainland Americans were more likely to identify the
beneﬁt of being safe whereas, Japanese visitors were more likely to seek family
togetherness. The authors recognized that marketers might need to create separate
strategies for Canadians, Americans, and Japanese visitors.

Pitts and Woodside (1986) developed beneﬁt segments using cluster analyis of
travel/leisure criteria. The study examined values as a means of understanding motives
and needs in making leisure and travel decisions. The purpose of the study was to group
individuals with similar recreation choice criteria on the basis of common personal
values. Data for the study was obtained from a mail survey administered to 250 students
of the University of South Carolina. Values were measured using the “Rokeach Values

Scale” (Rokeach, 1973). The scale consisted of 18 items or values, which were to be

49

ranked in order of importance. Such attributes as cost, accessibility, excitement,
activities for group, good food, relaxation, physical exercise, educational beneﬁts,
comfortable accommodations and good for the whole family were included in the study.
The study used a two step analysis: cluster analysis of the beneﬁt statements, and then
discriminant analysis to create beneﬁt segments. Four segments emerged from the
cluster analysis. The clusters were then analyzed in relation to actual travel behaviors.

Gitelson and Kcrstetter (1990) reported an empirical test of the relationship
between sociodemographic variables, beneﬁts sought and subsequent vacation behavior.
In a questionnaire sent out to 2,700 individuals in North Carolina, they found that a
relationship existed between sociodemographic variables and beneﬁts sought by
vacationers. Their study supported Hawes' (1988) frnding that as individuals age they do
not seek rest and relaxation as much as those under the age of 30 years. In addition,
women tended to value the social dimensions more and rate the relaxation dimension as
being more important than men. Hawes study also suggested that individuals in the
rrriddle income bracket felt that the explorer dimension was more important than other
income brackets. ’

Loker and Perdue (1992) examined the feasibility of segmenting the tourist
market on the basis of beneﬁt segmentation. The North Carolina Office of Park and
Tourism Research conducted a year-long survey of non-resident travelers to North
Carolina. A survey questionnaire was sent out to 6,418 and resulted in a response rate of
56%. The factor-cluster model was used to distinguish six beneﬁt-based segments that
were then compared on the basis of ﬁve dependent variables. Selection of target markets

was based on proﬁtability, reachability, and accessibility. The study found that there were

50

six markets: naturalists, nondifferentiators, family and friends-oriented,
excitement/escape, escapists, and pure excitement seekers. It was determined that the
naturalists were highly educated males aged 30 to 59 years who reported middle incomes,
and vacationed in lodges or cottages. The nondifferentiators were professional males
living with middle to upper incomes who usually had children, stayed in hotels and
visited museums when on vacation, and enjoyed amusement parks and ﬁshing. The
family-oriented segment traveled for the purpose of interaction with family or friends.
This segment was represented by lower to middle income families. This group enjoyed
visiting beaches and historical sites. The next segment was the pure excitement seekers;
individuals who were self-employed or retired earned a middle to upper income,
generally traveled in small groups and enjoyed going to the beach area and/or sporting
events. Single or married individuals without children, who used travel brochures
frequently, and visited museums and beaches when on vacation, primarily characterized
the segment excitement/escape. The last segment was the escapists who were aged 30 to
49 years, earned a middle to upper income, planned trips one to three months in advance,
lodged in hotels and motels, and enjoyed visiting historical sites or scenic areas. The
study suggested that beneﬁts segmentation is a practical and useful method of segmenting
the vacation market.

Shoemaker (1994) segmented the US. travel market according to beneﬁts
realized. The segmentation study was based on the beneﬁts derived from travel that was
not destination speciﬁc. Major market segments were identiﬁed through the study of past
travel experience. Shoemaker used a telephone to survey qualiﬁed, random US.

households in six separate speciﬁc geographic regions. Qualiﬁed respondents were

51

actively involved in the decision-making process and had taken at least one overnight trip
in the past 12 months. The study found that there were four market segments, but only
three were considered an accurate size for analysis. These three markets were: “Get
Away/Farnily Travelers,” “Adventurous/Educational Travelers” and “Gamblers/Fun
Oriented Travelers.” It was determined that members of the “Get Away/Family
Travelers” were more likely to travel to a place where friends or relatives live, to visit
destinations with scenery, visit places which provide rest and relaxation, have the ability
to drive to the destination. In contrast, “Adventurous/Educational Travelers” visited a
place that had cultural activities, went to a place they had never been before, someone
else they knew had been there, offers a number of activities, has ﬁne dining/elegant
restaurants and offers hotel or accommodations in all price ranges. This group is less
concerned with a vacation destination that is one of rest and relaxation. Last, the
“Gamblers/Fun Oriented” market places high importance on activities associated with
gambling and fun. They look for opportunities for recreation or sport, a popular
destination, good night life and entertainment, good value for money and ﬁne
dining/elegant restaurants. This study suggested that beneﬁt segmentation is a viable
way to segment the market and that past trip experience explains beneﬁts realized which
in turn provide more meaningful beneﬁts segments.

Wight (1996) examined ecotourists in terms of beneﬁts sought. More speciﬁcally,
activities, preferences and motivations were examined. This study found that travelers in
ecotourism seek a wide variety of activities that are passive and active, general and
specialized, and land or water oriented. The essence of the article is that traveler

preferences are dynamic. Certain activities increase or decrease depending on the

52

experience of the ecotourist. Travelers in all markets are looking for products that
respect the environment. In addition, motivations and beneﬁts sought vary by destination
and are also dynamic. Both ecotourists who are interested in the outdoors and consumers
interested in cultural experiences rank natural settings as the most important feature.
Beneﬁts with high discriminating power for ecotourists and traditional travelers tended to
be on the primitive nature-end of the continuum.

This section of the literature review has focused on literature relating to beneﬁts
sought through pleasure travel. The literature suggests that there are several beneﬁts
sought when travelling for pleasure. Such beneﬁts might range from seeking rest and
relaxation to looking for various forms of entertainment. As well, the literature has
focused on the characteristics of the members of each beneﬁt segment. Unfortunately,
generational cohorts have not received any attention in the literature and little is known
about the types of beneﬁts each generation seeks and what impacts the types of beneﬁts

sought might have on the volume and type of pleasure travel in Canada.

Change Over Time as a Component of Leisure Research
To date there have been only a few national studies with large representative
samples of older adults across two or three generations. Rather, most studies of tourism
and leisure over time have been small local or regional studies of the leisure needs,
activities, preferences and patterns of older adults (McPherson 1983). In addition, almost
all of the research has been cross sectional in nature and compared different age groups
on their frequency of leisure participation. These studies typically relate types and rates

of participation across different segments of society where age is just one criterion used

53

to classify people. According to Mannell and Kleiber (1997) age group differences make
it tempting to assume that getting older leads to differences or changes in behavior.
However, simple examinations of linear progression do not provide enough information
to drawn conclusions about intergenerational differences.

The central concepts of behavioral change are stability and intraindividual
variability. There are four concepts of stability that have been covered in the literature;
structural invariance, normative stability, level stability, and ipsative stability. Structural
invariance refers to consistency in the multivariate patteming of relationships between
latent variables and the manifest variables that are used to represent those indicators in
empirical research (Glenn, 1977). Invariance of factor loadings is the best known and
most systematically studied methodological approach. Normative stability refers to the
extent to which individuals or other entities retain their relative ranks or statuses on
attributes of interest across times of measurement. Level stability refers essentially to the
extent to which scores remain invariant across times of measurement and ipsative
stability refers to the extent to which the organization of multiple attributes within the
individual remains invariant across times of measurement. Personality research illustrates
the context for this kind of stability. Stability is a class of concepts which has
contributed to the understanding of a variety of phenomena.

Intraindividual variability also encompasses a broad set of concepts: (1) within
person change that may or may not be reversible and that may or may not be synchronous
across individuals (i.e., learning); (2) within person changes that are more or less
reversible and that may or may not be synchronous over individuals (i.e., states, moods,

emotions); (3) patterns of organizations of intraindividual variability within the

54

mdir’idt
of an in
rariabil
There 2

cohort

variab
hhola
recen'
cohor
leisu
1996
said

prefi

near
inte
Ps0
Cl€\

COL

individual; (4) patterns of intraindividual variation reﬂecting the proﬁle or organization
of an individual’s attributes (hierarchy of attributes). Stability and intraindividual
variability summarize time (period) or age-related changes in attitude and behavior.
There appears to be a lack of literature which examines this variability or stability as a
cohort difference.

From the life course perspective, stability is uncommon. One explanation is that
variability aids in the adaptive capacity, especially in the transition to adulthood (Iso-
Ahola, 1980). The concept of stability dominated earlier research; however, more
recently researchers have focused on variability across the life course. Finally, in the
cohort literature, the concept of stability of preferences is lacking in the tourism and
leisure literature although it is quite common in the popular press (c.f., Foot & Stoffrnan,
1996; Smith & Clurman, 1997). The following section will examine the concept of
stability and variability of the tourism and leisure markets. Literature related to change in
preferences and attitudes will also be examined in the ﬁeld of tourism and leisure.

Development is a process that occurs over the entire life span and applies to
nearly all areas of human behavior. Development is not simply the accumulation of
interests, abilities and experiences; it is systematic and predictable change by which
people become qualitatively different in some way from what they were before. The
developmental perspective is facilitated by two models, the life span model and life
course model. Life span models are generally “psychological” in nature and tend to
emphasize those biological and psychological issues or challenges faced by individuals at
different stages of life (Erickson, 1963; Havighust, 1972). Life course models, in

contrast, are “sociological” in scope and examine how individuals behave in light of age-

55

related expe
examine ho
during their
trajectories
school con
experience
sports and
sedentary
Di
inﬂuence:
(Baltes, (

Species 3

Well be

hare St]

and ch;

related expectations and norms (Neugarten, 197 7). Life course models also tend to
examine how individuals act as a result of life events or unique turning points that occur
during their lives (Brim, 1980). Life course theory focuses on the interweave of life cycle
trajectories (i.e., family and work careers) and short-term transitions such as marriage or
school completion. Life course is structurally and experientially tied to family
experiences and work experiences. For example, young people tend to participate more in
sports and activities that require physical prowess; while older people participate in more
sedentary leisure activities (Lock, 1993; McPherson & Kozlik, 1987).

Development involves three types of inﬂuences, (l) normative age graded
inﬂuences; (2) normative history graded inﬂuences; and (3) non-normative life events
(Baltes, Cornelius, & Nesselroade, 1980). Normative age inﬂuences are changes that are
species speciﬁc and biological in origin, for example, the graying of hair with age. Such
changes are found across all cultures and historical periods. Normative history graded
development involves a form of evolutionary change associated with signiﬁcant historical
events. For example, the depression in the 19203, the Vietnam War in the 19603 and
secular trends. Finally, development is inﬂuenced by non-normative life events, such as
divorce where the changes experienced by people are precipitated more by unexpected
life circumstances than by local, social or large-scale cultural changes.

The notion of change often seems to be associated with growth, development and
well being. Several authors (Iso-Ahola, 1980a; Jackson & Dunn, 1994; Kleiber, 1985)
have suggested that a person’s development relies heavily on experiencing both stability

and change in one’s life; these states are in balance and work to stimulate each other. Too

56

much c

dei’elO}

Sllgge

1997')

llVi
Chris
lcism
actitr-j

titan

much change may be unsettling, whereas too much stability may not lead to successful
development. Novelty and change are important for growth and development.

Erickson’s model (1963) provides a useful perspective for thinking about leisure
behavior and its role in development over the life course. Erickson argued that
predictable changes do not end with the transition from adolescence to adulthood.
Erickson’s model has been thought of as too general a model; however, Leveinson’s life
span model is more speciﬁc and detailed. This model was ﬁrst applied to thinking about
changes in leisure behavior by Iso-Ahola (1980a). Levinson’s model suggested that there
are “seasons” of life during which certain developmental issues are predictably faced.
The model suggests that the seasons are involved in alternating periods of structure
building and structure changing throughout the adult life course. This research suggested
that for many, midlife brings an unwillingness to keep up appearances and associate with
people out of obligation, thus resulting in freedom for experimentation. Leisure provides
an opportunity for this experimentation.

Although there are a variety of what seems like competing theories, research
suggests that there is some continuity in leisure behavior over time (Mannell & Kleiber,
1997). What is less well known is the extent to which leisure behaviors people learn
during youth persist into adulthood. Several studies have concluded that youthful
activities are reﬂected in adult leisure choices (Burch, 1969; Buse & Enoch, 1977;
Christenson & Yoesting, 1973). Most of these studies have dealt with only a few types of
leisure activities and have used adults’ recall to determine the nature of their childhood
activities. It is estimated that 40 to 80 percent of adult leisure activities have a close

equivalent in childhood activities. These ﬁgures suggest that new patterns of

57

participation are indeed established throughout adulthood; however they tend to be
similar to activities pursued in early adulthood, rather than radically different activities.

One of the most widely cited studies of variations in leisure activity across the life
span was done with a survey in the 19705, of 1,441 people between the ages of 20 and 94
(Gordon, Gaitz & Scott, 1976). This study showed that overall leisure activity level
decreased with age and that this pattern was consistent across a variety of activities (e. g.,
cooking and solitary activities). In other words, older adults placed less importance on for
example, skiing than did younger adults. Activities done outside the home, requiring
physical exertion and high involvement indicated the strongest negative correlation with
age. The negative correlation between activity and age has also been found in studies
conducted by Kelly (1987a) and Unkel (1981). The limitation of this study was that it
was cross sectional in nature and as a result, cohort and period effects were not able to be
isolated as indicators of change.

Iso-Ahola (1980a) suggested a curvilinear relationship, with the variety of leisure
activities reaching its peak in middle adulthood and declining towards old age. His study
suggested that for most of people’s adult lives there is a decline in the number of personal
leisure resources available corresponding to the apparent decline in number of leisure
activities engaged in and the intensity of involvement.

Crawford, Godbey and Crouter (1986) investigated the stability of leisure
preferences across a two-year time span. Leisure preference data were collected from 126
married couples shortly after their marriage and then two years later. They used rank
order stability (assessed by correlational methods), and mean stability (assessed by a t-

test) to assess the stability of preferences over time. The study focused on single item

58

measures rather than latent variables (assessed by factor stability). They found that
individual’s preferences for speciﬁc leisure activities are signiﬁcantly and positively
correlated across a two-year period. Results indicated gender differences in stability
coefﬁcients, the average number of preference declines over time, and that husbands and
wives show no pattern of increased leisure preferences over the ﬁrst two years. Crawford
et al. concluded that psychological and environmental inﬂuences co-predict the stability
of leisure preferences.

McGuire, Dottavio and O’Leary (1987) used the 1982-1983 Nationwide
Recreation Survey (N RS) to determine whether late life differences in outdoor leisure
activity participation could be explained by early life participation. The study involved
6,000 individuals 12 years of age and older. They found that those age 65 and over could
be categorized as “expanders” or “contractors” based on their patterns of leisure
involvement. Expanders altered their leisure patterns by the addition of new activities
throughout the life course, whereas contractors showed evidence of continuity and
maintained the same pattern of leisure involvement over the life course.

McGuire and Dottavio (1987) looked at the data from the NRS to examine why
people in different ages quit, replaced, added or continued leisure activities. This study
examined three constructs of abandonment, continuity and liberation in outdoor
recreation participation among different age groups over the life course. The
investigators found that each of the three patterns existed at each life course stage. Thus,
abandonment, continuity and liberation all occurred at all different stages of the life

course.

59

Stability of leisure participation and motivational factors among 139 community
residents were assessed by Lounsbury and Hoopes in 1988. Their study used a list of 113
leisure activities based on the Leisure Activities Blank and ﬁve participation factors to
assess stability over a ﬁve-year period. Unlike, Crawford et a1. (1986), these researchers
examined the internal consistency of latent variables of participation and motivation and
assessed the reliabilities so that estimates of “true stability” could be generated. They
used a MANOVA to examine change over the ﬁve-year period. The results indicated a
high level of stability over a ﬁve-year period for leisure activity participation as well as
leisure motivation. In addition, results revealed large proportions of variability attributed
to between subject differences and subject by time interactions. They suggested that such
stability may be attributable to a person’s lifestyle or the constancy of environmental
opportunities.

Searle, McTavish and Brayley (1993) supported these patterns in a study of 1,209
Canadians in Manitoba. The sample ranged from 16 to 65 years of age. They found that
about 20% of the sample were classiﬁed as replacers, 25% were quitters, 16% were
adders, and 40% were continuers. They also found that the number of people who were
continuers increased with the age of the cohort group. The researchers suggested that
perhaps other factors besides age need to be examined to help explain patterns of
participation.

Iso-Ahola, Jackson and Dunn (1994) conducted a study of 3,927 households in
Alberta, Canada and found that the number of people indicating they started new
activities within the last year decreased steadily over three phases of adulthood until

retirement. The number of replacing activities, activities dropped or replaced by another,

60

also declined over the life course. This study found that over the four stages of adulthood
studied, those who chose to continue with the same activities increased in number while
those who started new activities decreased. The authors concluded that the “tendency to
seek novelty through new leisure activities declines with advancing life stages, whereas
the tendency to maintain stability through old and familiar activities increases with life
stages” (p. 243). The study does not support the theory that older adults disengage from
life and become inactive, but rather they become more selective and discriminating.

In 1989, Scott and Willits published a more comprehensive and long-range report
of the linkages between various adolescent recreational activities and corresponding adult
leisure choices using data from a long-term panel study of rural Pennsylvania residents.
In that analysis, panel members involvement in ﬁve types of leisure activities in 1947
when they were in high school was positively related to the frequency of participation in
these same activities 37 years later (in 1984) when the subjects were in their ﬁfties. The
strength of these relationships varied somewhat by speciﬁc type of activity. The strengths
of the relationships for sports and creative/artistic activities were signiﬁcantly greater for
women then men. Not only were there positive correlation between adolescent and adult
leisure patterns, but adolescent participation was more predictive of midlife involvement
in socializing, formal organizational participation, and for women in creative/artistic
activities and sports than respondents, gender, health rating, education or income.

In 1993, these same panel members were re-surveyed, and data on the frequencies
of their participation in these same ﬁve types of leisure activities were assessed again. In
1992, the subjects were in their early sixties- a period of life during which adults in

American society are likely to be experiencing various life changes. Although there is

61

 

(m

clearly diversity among adults in their early sixties, occupational achievements have
generally peaked and retirement looms on the horizon or may already have occurred. The
impending or actual loss of long-standing occupational identities and generally reduced
incomes can threaten customary lifestyles, while at the same time provide freedom from
earlier responsibilities.

One of the problems with studying leisure patterns and preferences over the life
course is the confounding problem of age and historical events (period effects). It is
important to be critical when studying ﬁndings from cross sectional studies because it
may be difﬁcult to isolate age effects ﬁ'om cohort effects. Therefore, the use of
longitudinal data on the same persons across time avoids the possibility of age effect bias
(LaPage & Reiyan, 1974; Smale & Dupuis, 1995; Freysinger & Ray, 1994; Schmitz-

Scherzer & Thomas, 1983; Crawford, Godbey & Crouter, 1986).

Strauss and Howe’s Theory Applied to the Canadian Population

According to Foot and Stoffrnan (1996), much of the thinking related to
attitudinal and social changes in America can be applied to Canada because they have
followed a similar pattern. Canada’s population pyramid contains a massive bulge,
representing the huge generation of the Baby Boomers born in a 20-year period from
1946 to 1966. By comparison, the GI Generation and the Silent Generation that preceded
the Baby Boomer Generation and Generation X are small. The most recent generation,
the Millennial Generation, (children of the Baby Boomers) is also a comparatively large
generation. This pattern is the same as that of the United States, although the precise

dates may vary by a year or two.

62

 

 

 

.

,......
.~

 

r... ... . .1... _ ..
2.1.... , .

, r . .

   

. .5.

 

 

According to Foot and Stoffrnan (1996), Canada’s baby boom was the loudest in
the industrialized world. In fact, only three other Western countries — the United States,
Australia and New Zealand had baby booms. Part of the reason was that these four
countries were immigrant receivers, and immigrants tend to be in their 205, the prime
childbearing years. Foot and Stoffrnan (1996) have suggested that the generational
proﬁles includes “those born elsewhere during those same years but now living in
Canada” (p. 13).

The US. boom started earlier, in 1946, and it also ended earlier, in 1964. At its
peak in 1957, the US. boom hit 3.7 children per family, nearly half a baby fewer than
Canadian women were producing at the peak of the Canadian boom. The Americans
started their boom earlier because more of their war effort was in the Paciﬁc. The US.
troops were brought home in 1945 and kids started appearing in 1946. Canadian troops
came home later, so Canadian births did not leap upwards until 1947. One third of
Canadians today are boomers and for that reason, according to Foot and Stoffrnan, they
are a slightly more important factor in Canadian life than American boomers are in
American life.

Additional support for the application of the proﬁles by Strauss and Howe to the
Canadian population comes from Matthew Elrnslie (1998). From 1997 to 1999, he has
been active in developing support for the application of Strauss and Howe’s theory to the
Canadian population. He has suggested that the “Canadian saeculum matches the
American saeculum exactly all the way down the line” (source: www.fourthtuming.com). He
has also assumed that all of British North America (what is now known as Canada)

operates on a single saeculum and not a bunch of differing local ones. His argument is

63

that “due to the overwhelming nature of modern American media, the two countries
cycles were pretty much in sync” (source: www.fourthturning.com). Matthew Elmslie
has gone through history and found support for the phases that are outlined in Strauss and
Howe’s theory. He has suggested there are ﬁve phases: (1) Pre—European, (2) Pre-
French, (3) French; (4) British and (5) Canada. He suggested there is no essential
difference between a British possession in North America (Canada) and a non-British
possession in North America (United States). Based on the research of these three
Canadian authors, the application of Strauss and Howe’s generational theory to the

Canadian population was deemed reasonable.

Cohort Analysis as a Method of Statistical Analysis

There are a number of ways to study change in individuals over time. They range
from cross sectional to longitudinal in nature. However, the best way to study change
over time is by use of the cohort sequential research design. This allows the researcher to
control for year of birth (which tells period effects) and age (life stage and age). Repeated
measures on multiple cohorts allow the researcher to partial out the sources of group
differences and changes associated with aging. This method of longitudinal research
gives the most complete picture of the all three effects over time. There are three
statistical ways to implement a cohort sequential research design. They include, the triad
method, regression analysis with dummy variables and structural equations modeling.

Discussions on each of the three statistical techniques follow.

lri

Triad Method

The least rigorous technique to analyze change over time is called the Triad
Method. This method uses t-tests or AN OVAs to examine differences in cohort’s
attitudes or preferences over time. Palmore’s (1978) triad method “peels off” the data into
three layers or levels and makes sense out of them in a dependent fashion. The ﬁrst layer
consists of measuring (1) longitudinal differences (difference between two measurements
of the same cohort over time); (2) cross sectional (difference between younger and older
cohorts at a speciﬁc time; (3) time lag (difference between two cohorts measured at two
different points in time such that their age is equal at the respective times of
measurement). Palmore suggests that period effects may be caused by changing physical
environments, group composition, measurement techniques, etc. and cohort effects may
be caused by historical differences in social or physical environments during critical

earlier years.

Regression Analysis with Dummy Variables

An additional method for examining differences over time, which takes into
account all three effects, is regression analysis with dummy variables. The ﬁrst
researchers to use this technique were Rentz and Reynolds (1991), who examined
differences in coffee consumption over time. Rentz and Reynolds noted that many of the
previous techniques used to perform cohort analysis were not adequate in separating the
effects of age, cohort membership and period. They noted that multiple cross sections can
be used but, in their original form, are linearly dependent. Therefore, they suggested the

following model:

65

Yir = P+Bi+7j+5k+ eij
Where the effect of the it}. age group is given by 131, the effect of the jth period by yj and
the effect of the km cohort by 8., where u is the grand mean of the dependent variable and
e is the random error. The i age groups are coded into i-l dummy variables, the j periods
into j-l dummy variables, and the k cohorts into k-l dummy variables. Because of the
linear dependency, the linear effects of age, period and cohort are confounded but the
nonlinear effects are not. Therefore, in data that are linear this method should not be used.
Because the model is underidentiﬁed, one of the parameters must be constrained in order
to estimate the other parameters. Typically, two or more age, period or cohort effects are
constrained to be equal. Because there are three variables, three combinations of
constraints exist. In order to test the relative importance of the three dimensions, all three
combinations are performed and one model is then chosen as the appropriate model to
interpret. Tests of signiﬁcance are used to test the ﬁt of the models and determine which

is the most appropriate.

Structural Equations Modeling

Another technique used with longitudinal data to measure cohort effects is
structural equations modeling. The most powerful statistical analysis of cohorts involves
structural equations modeling. According to Bollen (1989), longitudinal data often
require complex multivariate statistical models and techniques for their analyses.
Structural equations modeling involves observed variables, which are regarded as
indicators of latent variables (Bollen, 1989). Each indicator has a relationship with the

latent variable but one indicator alone may be a poor measure of the latent variable. By

66

using several indicators of a latent variable we get a better measurement of it. Using
latent variables incorporates the concept of measurement error (observational error) in the
observed variables. Measurement errors occur because of imperfection in the various
measurement instruments (tests, questionnaires, scales, etc.) used to measure such
concepts. Even if one could construct a valid measurement of these concepts it would
still be impossible to obtain perfectly reliable measurements. One assumption is that the
measurement model is assumed to be invariant over time as well as over cohorts. This
assumption enables us to use the same scale in examining changes in the latent variable
over time and differences in the latent variable between cohorts. To explain the values of
the latent variable we build a structural model that contains a mean structure and a
covariance structure. The mean structure is interpreted in terms of cohort (or age) effects
and time effects. The covariance structure is an autoregressive model, which explains the
latent variable at time t by the value of the latent variable at time H and by other
observed explanatory variables (Bollen, 1989).

According to Glenn (1977), the general model is one that has observations on a
number of individuals over t periods of time from each of G cohorts. At each occasion we
have measurements on the same p variables. The observed score of an individual from a
cohort on variables i at time period t, where the sample size is adequate. Initially, we
examine one cohort. We assume that the p variables are fallible measures of a single
latent variable. The measurement model speciﬁes how the p response variables are used
to measure the latent variable at occasion t. The measurement model can be used to study
measurement properties such as validity and reliability of the responses variables and

how these vary over time and over cohorts. It is assumed that the error terms are

67

uncorrelated with the observed variables for every cohort. Furthermore, it is assumed that
error terms are uncorrelated with each other. Both origin and the unit of measurement are
arbitrary and, therefore, to deﬁne a scale which is common to all occasions, we ﬁx the
origin of one lambda to one for one cohort or set the variance to one for all cohorts. The
covariance matrix and mean vector will be different for different cohorts. This may in
principle hold for the phi (relationship between latent constructs), but it may be of
interest to see to what extent phi is invariant over cohorts. The structural model is used to
describe how the latent variable depends on other variables whose effects are to be
estimated and studied. Explanatory variables are included in the structural model. These
explanatory measures include sex, education, experience and income. The changes in
coefﬁcient alpha represent effects of cohort and time each confounded by age. The
coefficient alphas reﬂect changes in level (means) over time. The Beta coefﬁcient is
usually called a stability coefﬁcient, because it reﬂects the extent to which the later latent
variable can be predicted from the earlier latent variable. High values of beta are
associated with high stability. The error term in the equation is a random disturbance
term representing the combined effects of all variables inﬂuencing the latent term and not
included in the indicator variable (Bollen, 1989; Glenn, 1977).

To test the measurement model, the phi matrix is free to vary, one of the lambda’s
is set to one, and the ﬁrst element of k is set to zero. The test of the measurement model
is essentially the test of the hypothesis that v and lambda are invariant across time. Note
also that the hypothesis states that there is a correlation of error terms across time, such
that they are correlated with the same variable over time but not with other variables. The

test of the measurement model is then measured using the chi-square statistic to evaluate

68

how well the data ﬁt the model. To test the structural model, the lambda’s and v’s must
be constrained to be equal across time. In this analysis, all the means of the latent
variables in all cohorts can be estimated except for a single additive constant.

To test the structural model, one assumes that the measurement model holds and
is invariant over occasions and cohorts and imposes additional structures on the mean
vector k and covariance matrix or phi. If the vector is unconstrained and free to vary over
cohorts the structure on the means does not impose any restrictions. When the time
periods are greater than three, the hypothesis of ﬁrst order autoregression imposes
restrictions on the phi covariance matrix. When the larnbdas are constrained to be equal
over time, the means are then estimable and signiﬁcant differences in the size of the
means can be measured. We are then able to determine the differences in attitudes or

behaviors of the different cohorts over time.

Summary

The beginning of this literature review proposed a theoretical foundation for the
study of cohorts or generations. The review then focused on research relating to
attitudes, motivations and beneﬁts sought through travel. Until recently, the idea of
segmenting the market based on generations was unheard of. However, researchers have
begun to recognize the importance of cohorts. Though, there has been a lack of empirical
evidence regarding the relationship between a person’s generation and the impact on
attitudes and motivations towards travel. The purpose of this study was to see if there

was empirical evidence to support generational theory in a tourism context.

69

Chapter III
METHODOLOGY
Cohort analysis was used to investigate the impact of generational variables on
twelve travel-related variables. The preparation of the data for the cohort analysis will be
addressed in the following sections: (a) source of data; (b) conversion of the data; and (c)

treatment of the data.

Source of Data

The data used in this study was secondary in nature. Cohort analysis requires data
from two or more periods in time, usually which cover a period of 10-15 years or more.
The availability of data from more than two time periods which measure the same
variables on the same population using the same method is limited. Similar to the
approach referred to in the Yankelovich report, this study used two time periods as the
bases for cohort analysis. Studies providing good travel data were obtained through the
Canadian Tourism Commission. Speciﬁcally, the survey data were obtained through two
national surveys ftmded by Tourism Canada (later called the Canadian Tourism
Commission). Permission was obtained from the Canadian Tourism Commission to use
both the data sets. These studies were the “Canadian Travel and Tourism Motivation
Study” (CTAMS) and the “Domestic Tourism Market Research Study” (DTS).

The selection of the subjects in both surveys included Canadians, aged 15 and
older. The data was obtained through a combination of both telephone surveys and
personal interviews. The selection of respondents for CTAMS was obtained through a

telephone survey using a random digit dial sampling technique. The selection of

70

 

respondents for the DTS used two different data collection methods. A telephone survey
was conducted nation-wide on a random basis, while the remaining respondents were
selected in shopping malls in large cities to meet regional quotas. Then, a more in-depth
survey consisting of in-person interviews with 1,457 randomly selected Canadians was

conducted. This survey was conducted in 48 urban centers with a population over 30,000.

Canadian Tourism Attitude and Motivation Study

The Canadian Tourism Attitude and Motivation Study was conducted by the
Department of Regional Industrial Expansion in the autumn of 1983. The objective of the
study was to “develop an understanding of the Canadian travel market through the
deﬁnition and analysis of population sub-groups based on: vacation needs; attitudes to
vacations, interests with respect to vacations and their actual behavior” (p. 1). The sample
size was 11,500 and equally distributed throughout the 10 provinces. The 1983 sample

was 11,500 and distributed fairly equally among the 10 Provinces in Canada (Table 6).

Table 6. Number of Respondents by Province in 1983 CTAMS Study

 

 

 

Province Number of Interviews

Newfoundland 1500
Prince Edward Island 948
Nova Scotia 1346
New Brunswick 1430
Quebec 1437
Ontario 1808
Manitoba 1602
Saskatchewan ' 1 896
Alberta 1158
British Columbia 1055
Total 11,500

 

71

Domestic Tourism Market Research Study

The Domestic Tourism Market Research Study was conducted by Coopers &
Lybrand Consulting in September and October of 1995. The objective of the study was
to examine in detail the motivations, attitudes and perceptions of Canadians about
tourism, travel opportunities and destinations. The overall purpose of the research was “to
determine how best to market Canadian destinations and products to Canadians and
examine possibilities for import substitution” (p.i). The results were based on a
combination of telephone and in-person interviews with over 3,356 Canadians in all
regions of the country. The Domestic Tourism Travel’s survey sample size was 1,899
telephone surveys and 1,457 in-person surveys. The sample size varied in the 10
Provinces and was weighted accordingly (Table 7). The data had a weight variable so that

age, gender and province were representative of the entire population.

Table 7. Number of Interviews by Province in 1995 DTS Study

 

 

Province Number of Interviews
Newfoundland 119
Prince Edward Island 27
Nova Scotia 185
New Brunswick 154
Quebec 429
Ontario 438
Manitoba 235
Saskatchewan 210
Alberta 440
British Columbia 447
Total 3356

 

 

72

Conversion of the Data

Due to the nature of the study, it was necessary to have scalar equivalence among

the variables. Thus, all variables needed to be identical in both measurement, meaning

and wording. A total of 12 variables across both studies were worded exactly the same.

These variables were categorized as three main constructs (by Tourism Canada): (1)

Travel Philosophies; (2) Beneﬁts Sought and (3) Travel Interests. Under the “Travel

Philosophy” section, there were two variables which had scalar equivalence; under the

“Beneﬁts Sought” section there was one variable that had scalar equivalence and under

the “Travel Interest” section there were nine variables that had scalar equivalence.

Therefore, there are a total of 12 variables that were used to examine the effects of age,

generation and period on travel-related variables (Table 8).

Table 8. Common Variables in Both Studies

 

 

 

 

CTAMS- 1983 Domestic Travel- 1995
Travel Philosophy
V1 Money spent on travel is well Money spent on travel is well
spent spent
V2 Important people speak my Important people speak my
language language
Benefits Sought
V3 Having fun, being entertained Having fun, being entertained
Travel Interests
V4 National/provincial parks National/provincial parks
V5 Beaches for swimming and Beaches for sunning and
sunning swimming
V6 High quality restaurants High quality restaurants
V7 First class hotels First class hotels and resorts
V8 Budget accommodations Budget accommodations
V9 Nightlife and entertainment Nightlife and entertainment
V10 Museums, art galleries Museums and art galleries
V11 Amusement/Theme parks Theme parks and amusement
parks
V 1 2 Shopping Shopping

 

73

The travel philosophy variables at period one (CTAMS) were measured on a 5-
point likert scale, ranging from “strongly agree” to “strongly disagree.” The travel
philosophy variables at period two (DTS) were measured on a 4-point scale, ranging from
“strongly agree” to “strongly disagree”. Therefore, the variables were standardized on a
0-1 scale, where 0 was “strongly agree” and 1 was “strongly disagree.” Z-scores were not
used. Not using Z-scores prevented the standard deviation from being normalized on a 0
to 1 scale. It was felt that maintaining the original range of responses provided more
information which in turn helped in analyzing patterns. Figure 3 maps the conversion

from the ﬁve and four point scales to the standardized scale.

 

 

 

 

 

 

 

 

 

Figure 3. Transformation from S-Point Scale to 4-Point
Scale

The scale was interpreted as follows: 0 = “strongly agree,”.25 = “somewhat
agree,”.50 = “neutral,”.75 = “somewhat disagree” and 1 = “strongly disagree.”

The beneﬁts sought variable was measured in both time periods on a 4-point
scale, which ranged from “very important” to “not at all important.” However, in time
period one (CTAMS) a score of one was “very important” and in time period two a score
of one was “not at all important;” therefore, time two was reverse coded so the scales

were the same.

74

The ﬁnal set of variables, the travel interest variables, were also measured on a 4-
point likert type scale from “very important” to “not at all important.” The same variables
were measured on identical scales. Similar to the beneﬁt sought variable, however, the
scores needed to be reverse coded for one period into to make them comparable.

For the purpose of comparison, all variables were standardized. Therefore, the
beneﬁts sought and travel interest scales were standardized on a zero to one scale. Figure

4 maps the transformation from the four-point scale to the standardized scale. I

 

 

JS
'l

3
.l l

<3<—~
{new

L»)

 

 

 

Figure 4. Transformation from 4-Point Scale to
Standardized Scale

The way the scales were interpreted were that a score of 0 represented
“important,” a score of .33 on the standardized scale represented “somewhat important,”
a score of .66 represented “not very important” and a score of 1 represented “not at all
important.” Figure 5 indicates how the two scales could be used to map comparisons

among the 12 variables on a standardized scale.

75

 

 

2 3
l \l/
.25--.33 --------- .50 .66

 

 

 

 

 

9
we
cl>_/A\)

 

 

 

Figure 5. Mapping Comparisons on a Standardized Scale

Treatment of the Data

The data were analyzed using the statistical package, SPSS (1995). Initially, the
FREQUENCIES procedure was applied to obtain a “picture” of the data in the form of
raw frequencies and percent of occurrence for the sample.

The “age” variable in each dataset was recoded to represent the generations
outlined by Strauss and Howe (1991). When coding the generation variable it was
necessary to leave out speciﬁc age categories so that they would not overlap. This

allowed each age category to represent only one generation.

Table 9. Age of Each Generation at Time of Two Studies

 

Generation Year Born Age in 1983 Age in 1995
G1 Generation 1901-1924 82-59 94-71
Silent Generation 1925-1943 58-40 70-52
Baby Boomer 1944-1964 39-19 51-31
Generation X 1965-1981 18-2 30-14

 

76

Table 10. Sample Sizes of Each Generation in Two Studies

 

 

Sample Size Sample Size
Generation Year Born in 1983 in 1995
G1 Generation 1901-1924 1953 419
Silent Generation 1925-1943 1 394 590
Baby Boomer 1944-1964 4133 1485
Generation X 1965-1981 360 862
Total 7840 3356

 

Data were initially analyzed using t-tests and analysis of variance. T-tests were
preformed to measure bivariate relationships and AN OVAs were used to measure
multivariate relationships.

The cohort analysis was performed by means of a three-way analysis of variance.
The analysis of variance results were reported in a narrative form and in a table showing
the main effects and interaction effects, the R2 for each model was also reported. The
AN OVA procedure tested for signiﬁcance at the .05 level.

The ﬁnal procedure was to analyze the data by means of structural equations
modeling using the EQS program (Bentler, 1985). This method permits the analyst to
specify the measurement of latent variables and to hypothesize the relations that occur
among those latent variables. The adequacy of the measurement model and the tenability
of the hypothesized causal relations among the latent variables can be tested
simultaneously. A poor ﬁt of the measurement model or causal model indicates that the
hypothesized relations are inconsistent with the data. A good ﬁt supports the legitimacy

of the model but does not rule out plausible alternatives that were never tested.

77

Chapter IV
DATA ANALYSIS AND INTERPRETATION
The purpose of this study was to assess speciﬁc travel philosophies, beneﬁts
sought and preferences held by Canadians using a cohort analysis. This chapter has been
divided into the following sections: (a) sample proﬁle and biases; (b) accounting for large
sample sizes (c) period and travel-related variables; ((1) age and travel related variables;
(e) generation and travel-related variables; (0 using different methods of cohort analysis;

(g) using structural equations modeling as a method of cohort analysis.

Sample Proﬁle and Biases

Both the Canadian Tourism Attitude and Motivation Study (CTAMS) and the
Domestic Tourism Market Research Study (DTMRS) provided demographic variables
that helped to proﬁle the samples. The following demographic variables were included in
both studies: age, income, marital status, education, gender, and number of income
earners in household. The purpose of proﬁling the respondents was to address the
representativeness of the sample to the Canadian population. P0pulation statistics were
obtained from the 1984 and 1996 Census collected by Statistics Canada. Biases were

addressed in the narrative. The results are presented in Table 11.

78

Table 11. Demographic Characteristics of Respondents in Both Periods

 

 

Temographrc 1983 1983 1995 1995
Characteristic CTAMS CTAMS - . . DTS ° DTS
Sample Population* Sample Population’
% % % %
Age
15-19 3.3 9.6 8.4 8.4
20-24 11.3 11.3 8.6 8.7
25-34 28.2 22.7 23.1 20.8
35-44 20.0 18.3 22.9 20.8
45-54 11.8 12.8 16.2 15.6
55-64 11.9 11.7 11.5 10.7
65+ 13.5 13.6 9.4 15.1
Income
Under median 44.1 42.8 50.0 40.9
Median l 1.7 6.4 6.5 6.5
Over Median 44.3 50.8 43.4 52.6
Marital Status
Married 68.0 42.7 59.8 42.9
Single 21.0 49.6 27.2 47.8
Other 11.0 7.7 12.9 9.3
Education Level
University/College 32.0 3 l .5 48.6 39.7
High School 48.0 20.3 48.7 23.1
Grade School 20.0 47.8 2.6 36.8
Gender
Male 49.0 49.3 38.2 49.6
Female 51.0 50.7 61.8 50.4
Number of income
earners
1 38.0 31.8 40.0 29.7
2 36.0 41.9 52.3 43.4
3 or more 25.0 26.3 7.7 26.9

 

 

 

0 Percentages may not add up to 100% due to rounding.
o (*) Source: Statistics Canada 1986 Census
0 (’) Source: Statistics Canada 1994 Census

79

Age
The age groups were not coded the same in both studies. Therefore, the categories

in the DTS were recoded to match the age categories in the CTAMS study. The DTS was
recoded to match CTAMS because the groupings were more evenly distributed in the
CTAMS than the groupings in the DTS study. Most categories in the 1983 study were in
10 year increments, while age categories in the DTS varied considerably.

Approximately, 28% of the CTAMS study were 25 to 34 years, while slightly less
(23.1%) were the same age in the 1995 study. The youngest and the oldest categories had
the least representation in both studies. In the CTAMS study, the youngest category (15-
19) represented 3.3% of the entire sample, while the same category in the DTS
represented 8.4%. Similarly, in the CTAMS study, the oldest group (65+) represented
13.5% while in 1995 this category represented 9.4% of the sample.

Examining the Canadian population parameters revealed that the youngest
category in the 1983 study and the oldest category in the 1995 study were under
represented. The 25-34 year old category and the 35-44 year old category in both the
1983 and 1995 study over represented the numbers of Canadians in that category.
However, the proportions in each age group were relatively the same as the proportions
in the population. Thus, based on age categories, the samples appeared to be fairly

representative of the Canadian population.

Income
The median income category in the 1983 study was $20,000 to $24,999 per
annum. In 1995, the median income category was $40,001 to $45,000. This was fairly

representative of the Canadian population, where the median income in 1983 was

80

$20,332 and the median income in 1995 was $47,124. The 1995 study was slightly
biased in its estimate of the number of Canadians who fell below the median. However,
this bias does not appear to be that large. In the 1983 study, 44.1% of the sample reported
an income under the median, while in 1995, 50% of the sample indicated an annual
income under the median. Whereas, in 1983, 44.3% reported having incomes over the
median and in 1995, 43.4% reported having incomes above the median. The relative
proportions of those over and under the median were consistent with the population in

both periods.

Marital Status

The largest segment of both samples were those who were married. In the 1983
study, 68% reported being married, while in 1995, slightly less (59.8%) reported being
married. A slightly larger group reported being single in 1995 (27.2%) compared to
21.0% in 1983. Approximately 13% and 11% fell into the “other” category in 1995 and
1983 respectively. The “other” category was made up of divorcees and widows.
Comparison of sample estimates to population parameters revealed that in both studies
the number of single Canadians were under represented. The result of this bias may be

that the importance of certain travel-related variables may be affected.

Education Level

The levels of education of the two samples were very different. Education

statistics ﬁom Statistics Canada were reported for those over the age of 21, this

81

represented only those who have had an ‘opportunity’ to obtain an education. Therefore,
the sample data only looked at the results of those over the age of 21.

In 1983, a much larger proportion of the sample reported that their highest level
of education was grade school. In 1995, only 2.6% of the sample reported having only a
grade school education. The number of respondents who had a university or college
education in 1983 was less than in 1995; 32% reported being university or college
educated in 1983 while 48% reported being university or college educated in 1995. The
percentage of respondents who reported having a high school degree was essentially the
same for both samples (48%).

Examining the population parameters suggested that the 1983 sample over
represented Canadian’s whose highest level of education was high school and under
represented Canadian’s whose highest level of education was grade school. In 1995, a
severe bias was imposed by under representing the number of people with grade school
education. This may result in biases associated with the importance of travel-related
variables which are educational in nature, such as the importance of art galleries and

museums.

Gender

The percentage of males to females was relatively proportionate in the 1983 study
(49.0% males and 51.0% females). However, in the 1995 study, females were over
represented. They represented almost 62% of the total sample whereas men only
represented 38% of the sample. Examining the population parameters suggest that this

represents an over sampling of approximately 15%. This bias could affect scores on

82

certain types of travel-related variables. For example, shopping is traditionally viewed as
a more “female” activity and therefore responses on this variable may be biased due to an

over representation of females in the 1995 sample.

Number of Income Earners

The number of dual income earning families went up from 1983 to 1995. In 1983
only 36% reported having two people working whereas in 1995, slightly over half
reported having two people working in the home. The number of homes who reported
having more than three people in the home working was greater (15%) in 1983 than in
1995 (7.7%). However, this number was underrepresented in the sample. The number of
households in the population in 1995 with 3 or more income earners was approximately
27%. The sample did not capture enough of this group and therefore, there could be
sample biases. These biases should be taken into account when analyzing the results.

Overall, the sample estimates are fairly representative of the population, except on
a few variables. The age variable under represented the youngest category in 1983 and
the oldest category in 1995 and the two middle-aged categories (age 25-34) were over
represented. Gender, marital status, and education are the three variables which have the
strongest biases associated with them. The number of women sampled in 1995 was over
represented considerably and should be kept under consideration when evaluating the
results. Likewise, the percentage of married people in the sample was over represented
and the number of single people was under represented. This also needs to be taken into
consideration when evaluating the results. Finally, the number of individuals in the

sample whose highest level of education was “grade school” were under represented in

83

both studies. These variables may present problems when analyzing and interpreting the
relationships among age, generation and period on the travel-related variables. In
variables where there are anticipated relationships between these demographic variables
and the importance of the travel-related variable, period effects have a greater effect. The
problem associated with these biases is that the researcher is unable to isolate the effects
of these biases and the other independent variables, making the analyses even more

complicated.

Accounting for Large Sample Sizes

The data in this study contained 14,000 cases. Based on the size of the sample, it
is appropriate to conclude that the population estimates are the actual population
parameters. However, one of the problems associated with such large datasets is the
ability to easily reject the null hypothesis or show statistical differences with relatively
small differences. In large datasets minor differences are statistically signiﬁcant.
Therefore, it was appropriate to set a standard or “benchmark” to evaluate how big a
difference is a meaningful difference. Translation of the differences back to the scales,
allows us to determine what may be a meaningful difference. For example, a .05
difference in means represents a difference of .20 on the original 5-point scale. That
translates into a move of 20% from a score of “strongly agree” to “agree.” A difference
of .10 translates into a difference of .4 on the original ﬁve-point scale. That is less than
half a scale value and therefore probably not meaningful. Table 12 identiﬁes how

differences translate into the four and ﬁve-point scale. It was determined that less than a

84

.05 difference is probably not meaningful and therefore absolute differences of greater

than .05 will be considered “meaningful” differences.

Table 12. Observed Differences on the Standardized and Original Scales

 

 

Difference on 0-1 Scale Difference on the Difference on the
4 point scale 5 point scale
.05 .15 .20
.10 ' .30 .40
.15 .45 .60

 

Each of the single variables (age, generation and period) were analyzed both for
statistical differences and meaningful differences. Meaningful differences were computed
by subtracting the highest value from the lowest value. It is recognized that this is a fairly
liberal method of determining differences and that more conservative methods could have
been used. However, it was felt that this liberal method was appropriate for the purpose

of assessing where meaningful differences might be evident.

Period and Travel-Related Variables
The t-test procedure was preformed to test for statistical differences in means
across the two periods. Nine out of twelve were signiﬁcantly different at the 99%
conﬁdence interval (Table 13).
The three travel-related variables that were not statistically different between the
two periods were: “beaches for swimming and sunning,” “budget accommodations” and
“amusement and theme parks.” In all three cases, the ranking in both periods were

identical. “Beaches for swimming and sunning” was rated as “somewhat important.”

85

Both “budget accommodations” and “amusement and theme par ” ranked between

“somewhat important” and “not very important.”

Table 13. Means of Travel Philosophy, Beneﬁts Sought and Travel Interests
for 1983 Study, 1995 Study and Pooled Data

 

 

Philosophy, Beneﬁts and Interest Men Men Men Sig- Diff-

1983 1995 Pooled high-

Data low

*Money spent on travel is well spent .44 .37 .42 .00 .07
*Irnportant people speak my language .50 .53 .51 .00 .03
Having fun, being entertained .24 .17 .22 .00 .07
National/provincial parks .30 .31 .30 .02 .01
Beaches for swimming and sunning .31 .31 .31 .32 .00
High quality restaurants .41 .46 .43 .00 .05
Budget accommodations .48 .48 .48 .71 .00
First class accommodations .34 .30 .33 .00 .04
Shopping .36 .31 .35 .00 .05
Nightlife and entertainment .43 .42 .43 .02 .01
Museums and art galleries .42 .45 .42 .00 .03
Amusement and theme parks .43 .43 .43 .20 .00
Overall Mean .39 .38 .39 .00 .01

 

Scale 0-1 where 0= “very important,” .33 = “somewhat important,” .66 = “not very
important,” and 1= “not at all important.”

* Scale 0-1 where 0= “strongly agree,” .25 = “agree,” .50 = “neither agree nor disagree,”
.75 = “disagree,” and 1= “strongly disagree.”

As previously mentioned it is important to examine the differences between mean
scores on a practical level as well as a statistical level. Because the number of cases in the
dataset exceeded 14,000, evaluation of differences was based on “meaningful
differences” of greater than .05. Only four of the twelve travel-related variables showed a
meaningful difference between time periods. These variables included, “money spent on
travel is well spent,” “having fun and being entertained,” “ high quality restaurants,” and
“shopping.” The remaining variables revealed a practical difference of less than .05 and

were therefore deemed not signiﬁcantly different.

86

“Having ﬁrn and being entertained” was the most important travel-related
variable. The mean score on this variable was .22 which represented a score of better
than “somewhat important” but less than “important” on the standardized scale.
Canadians were more likely to ﬁnd this important in 1995 than in 1983 (.17 and .24
respectively). There was a difference of .07. This suggests that Canadians moved from
feeling that having ﬁm and being entertained was “somewhat important” closer to
“important.”

The least important travel-related variable was “people I encounter speak my
same language.” The overall mean was .51. Respondents did not feel strongly one way
or the other toward people speaking the same language. Canadians in 1995 ranked this
statistically lower in importance than in 1983.

Agreement with the statement “money spent on travel is well spent” indicated a
meaningful difference of .07 between the two time periods. Canadians tended to rate this
as more important in 1995 than they did in 1983.

“High quality restaurants” became less important over time. The importance of
high quality restaurants declined by .05 from 1983 to 1995. Canadians tended to rate this
less important in 1995 than they did in 1983. This difference translates into a difference
of less than half of the scale value between strongly agrees to agree.

Finally, “shopping” became more important over time. The difference between
1983 and 1995 was a difference of .05 again which represented a difference of .15 on the
original scale. This suggests that Canadians are moving closer to “important” from

“somewhat important.”

87

 

(lab
is we
they
“bea

imp

bet

5X3

The standard deviations for all twelve variables ranked between .19 and .27
(Table 14). The variable with the smallest standard deviation was “money spent on travel
is well spent.” This suggests that there was less variability in the responses to how much
they agreed with this statement. The variable with the largest standard deviation was
“beaches for swimming and sunning” indicating a greater range in responses to the
importance of this variable in pleasure travel.

Table 14. Standard Deviations of Travel Philosophy, Beneﬁts Sought and
Travel Interests for 1983 Study, 1995 Study and Pooled Data

 

 

Philosophy, Benefits and Interest Std. Std. Std. Dev
Dev. Dev. Pooled
1983 1995 Data
*Money spent on travel is well spent .19 .16 .19
*lmportant people speak my language .23 .27 .24
Having fun, being entertained .22 .21 .22
National/provincial parks .25 .24 .25
Beaches for swimming and sunning .27 .25 .27
High quality restaurants .25 .26 .26
Budget accommodations .24 .25 .24
First class accommodations .24 .25 .25
Shopping .25 .25 .25
Nightlife and entertainment .26' .25 .25
Museums and art galleries .26 .23 .25
Amusement and theme parks .25 .24 .24
Overall standard deviation .24 24 .24

 

Scale 0-1 where 0= “very important,” .33 = “somewhat important,” .66 = “not very
important,” and 1= “not at all important.”

* Scale 0-1 where 0= “strongly agree,” .25 = “agree,” .50 = “neither agree nor disagree,’
.75 = “disagree,” and 1= “strongly disagree.”

9

Age and Travel-Related Variables

The analysis of variance procedure was used to test for statistical differences
between means of the responses to the travel-related variables across subgroups. The

examination of age and the 12 travel-related variables was performed on the pooled data.

88

Post hoc tests were not performed because due to the large sample size small differences
resulted in signiﬁcant differences among groups (e.g., differenCe of .02). Therefore,
instead of post hoc tests practical differences were used to reveal differences. A practical
difference of .05 was deemed a meaningful difference. All twelve means were
signiﬁcantly different, meaning that there was a statistical difference between at least one
variable and the overall mean. On some of the variables, there was a negative linear
relationship. The importance of the travel interests, “national and provincial parks,” “first
class accommodations,” “nightlife and entertainment” and “amusement and theme
parks,” for example, was less important with age. One of the variables had a positive
linear relationship. The importance of “museums and art galleries” was more important
with age. Table 15 shows the mean scores based on age, the signiﬁcance level of the

AN OVA and the practical difference between the highest mean score and the lowest
mean score for each age category.

Again, because statistical differences are sensitive to sample size, analysis of
meaningful differences was important. Analysis using differences between the highest
mean score and the lowest mean score in each age category illustrated a more meaningful
difference in scores. Seven of the twelve travel-related variables indicated differences
greater than .05. These differences were meaningful for the following variables,
“important that people speak my language,” “having fun and being entertained,”
“national and provincial parks,” “beaches for swimming and sunning,” “ﬁrst class
accommodations,” “nightlife and entertainment” and “amusement and theme parks.” On
six of the seven variables, the largest differences were seen between the oldest age group

(65+) and either the youngest group (15-19) or the second youngest group (20-24).

89

Table 15. Means for Signiﬁcant Relationships between Age and Travel
Philosophy, Beneﬁts Sought and Interest Statements

 

Philosophy, Age Age Age Age Age Age Age All Sig. Diff.
19 24 34 44 54 64 low

Interest

Statements

 

*Money spent on .42 .42 .41“ .42 .42 .42 .433 .42 .010 .02
travel is well

 

 

 

 

spent

*lmportant .52 .53 .53 .54’l .51 .48 .45a .51 .000 .09
people speak my

language

Having fun, .153 .15 .21 .23 .25 .21 .26a .22 .000 .11
being entertained

National/provinc .27“ .27 .30 .31 .31 .31 .33a .30 .000 .06
ial parks

Beaches for .26 .24a .25 .29 .33 .29 .458| .31 .000 .21
swimming and

sunning

High quality .45a .42 .41ﬁl .41a .42 .44 .45al .43 .001 .04
restaurants

Budget .50'll .49 .49 .48 .46”l .48 .48 .48 .000 .04
accommodations

Firstclass .29a .31 .32 .33 .33 .35 .37a .33 .000 .08
accommodations

Shopping .34'I .34ll .35 .36 .34' .35 .37a .35 .008 .03
Nightlife and .31 .30T .38 .43 .47 .50 .55‘il .43 .000 .25
entertainment

Museums andart .44a .44a .44” .42 .42 .40a .41 .42 .000 .04
galleries

Amusement and .37a .39 .41 .43 .45 .48 .50al .43 .000 .13
theme parks

Overall Mean .36a .36al .38 39 .39 .39 .42a .39 .000 .06

 

Scale 0-1 where O= “very important,” .33 = “somewhat important,” .66 = “not very
important,” and l= “not at all important.”

"' Scale 0-1 where O= “strongly agree,” .25 = “agree,” .50 = “neither agree nor disagree,”
.75 = “disagree,” and 1= “strongly disagree.”

(a) indicates where largest differences between age groups exist

90

The two variables which indicated the greatest difference in scores between the oldest
age group and the youngest age group were “beaches for swimming and sunning” and

“nightlife and entertainment” (difference of .21 and .25 respectively).

Generation and Travel-Related Variables

The analysis of variance procedure was used to test for differences between
means of the responses to the travel-related variables across generational subgroups. The
examination of generation and the 12 travel-related variables was performed on the
pooled data. Post hoc tests were not performed because due to the large sample size small
differences resulted in signiﬁcant differences among groups (e.g., difference of .02).
Therefore, instead of post hoc tests, practical differences were used to reveal differences.
A practical difference of .05 was deemed a meaningﬁrl difference. Analysis of variance
revealed that the subgroup means differed from each other and from the grand mean.

For the travel-related variables “money spent on travel is well spent,” “having
fun, being entertained,” “national and provincial parks,” “beaches for swimming and

,9 6‘

sunning,” “frrst class accommodations, shopping,” “nightlife and entertainment,” and
“amusement and theme parks,” younger generations rated these interests more important
than older generations. The importance of budget accommodations in pleasure travel was
more important to older generations than younger generations. All other variables had no
pattern relative to generation. Table 16 displays the mean scores based on generation, the
signiﬁcance level of the AN OVA and the practical difference between the highest mean
score and the lowest mean score.

As with the previous two analyses, examination of meaningful differences was an

important step to analyzing the true differences based on generation. Nine of the twelve

91

travel-related variables indicated differences greater than .05. These differences were

,9 66'

meaningful for the following variables, “money spent on travel is well spent, important
that people speak my language,” “having fun and being entertained,” “beaches for
swimming and sunning,” “high quality restaurants,” “ﬁrst class accommodations,”

“shopping,” “nightlife and entertainment,” and “amusement and theme parks.”

Table 16. Means for Signiﬁcant Relationships between Generation and Travel
Philosophy, Beneﬁts Sought and Interest Statements

 

Philosophy, Benefits Gen X Baby Silent GI All Sig. Diff.
and Interest Statements 300m“ *1]?th

 

*Money spent on travel .39al .42 .43 .44a .42 .000 .05
is well spent
*Important people speak .52 .53 a .52 .46 a .51 .000 .07

 

my language

Having fun, being .14 a .19 .25 .27 a .22 .000 .13
entertained

National/provincial parks .29 a .29 .31 .32 a .30 .000 .03
Beaches for swimming .28 .25 a .30 .43 a .31 .000 .18
and sunning

High quality restaurants .46 a .42 .41 a .44 .42 .000 .05

 

Budget accommodations .49 .49 a .47 r .48 .48 .002 .02

 

First class .27 ' .32 .34 .36 a .33 .000 .09
accommodations

Shopping .32 a .35 .36 .37 a .35 .000 .05
Nightlife and .36 F .36 .45 .54 a .43 .000 .18
entertainment

Museums and art .44 a .44 .42 .40 a .42 .000 .04
galleries

Amusement and theme .39 a .41 .44 .49 a .43 .000 .10
parks

Overall Mean 36 a .37 .39 .42 ' .39 .000 .06

 

Scale 0-1 where 0= “very important,” .33 = “somewhat important,” .66 = “not very
important,” and 1= “not at all important.”

* Scale 0-1 where 0= “strongly agree,” .25 = “agree,” .50 = “neither agree nor disagree,”
.75 = “disagree,” and 1= “strongly disagree.”

(a) indicates where largest differences between age groups exist

92

On six of the nine variables, the largest differences were seen between the GI
generation and Generation X. In all cases, the GI generation ranked the following
variables as less important than Generation X, “money spent on travel is well spent,”
“having fun and being entertained,” “ﬁrst class accommodations,” “shopping,” “nightlife
and entertainment” and “amusement and theme parks.” For “it is important that people
speak my language” and “beaches for swimming and sunning,” the greatest difference
was between Baby Boomers and the GI Generation. In both cases, the GI generation felt
these activities were less important than the Baby Boomer generation. The difference
between the Silent Generation and the GI Generation was greatest for the importance of
“high quality restaurants.” The Silent Generation ranked “high quality restaurants” more

important than the GI Generation.

Using Different Methods of Cohort Analysis

Cohort analysis is used in this study to determine the amount of variance
explained by the variables age, period and generation. There are several methods to
determine the amount of explained variation. Methods, which are generally accepted, can
be both statistical and non-statistical in nature. The two most common methods include
the triad method using cohort tables or dummy variable regression. The triad method for
cohort analysis is generally accepted as a non-statistical method, this method “eyeballs”
the data to detect patterns. This study used a variation of the triad method in order to get a
“picture” of the data.

The second method that is widely accepted is regression. This technique is in the

form of dummy variable regression and provides estimates of the effects on a dependent

93

variable. To perform this analysis, the effects of either two age levels, two generations or
two periods are assumed to be equal. The assumption that the effects are equal is not a
problem; however, the technique assumes that age, generation and period are not additive
(do not interact). That means that age effects are the same for all generation and periods;
that generation effects are the same for all ages and periods and that period effects are the
same for all ages and generations. This assumption is not realistic; nor after using
analysis of variance was it found to be true.

In addition, the majority of research which uses dummy variable regression in
cohort analysis has used subgroup means to perform the regression analysis. One of the
problems with this is that subgroup means do not reveal how much variation there is
within subgroups. It is important to note that failure to report within group variation can
be misleading. In fact, individual variation, on average, accounts for 95% of the variation
in scores. Therefore, subgroup means isolate the ﬁve percent variation between groups
and ﬁts the regression line through the group means. Even though there are problems
with this technique it appears to be the most common method used in marketing research.
The cohort studies that are most widely accepted (Rentz & Reynolds, 1981, 1991; Rentz,
Reynolds, & Stout, 1983) have all used subgroup means. It is important to note, however,
that the R25 reported in these articles are much higher than one would expect if individual
scores were used. Thus, caution should be taken when this method is used to forecasting
future behavior. Considerable errors could be made if future efforts were made based on
between group variations of only ﬁve percent.

Therefore, in order to reveal the within group variation and account for the

relationship among the three independent variables, a three-way analysis of variance

94

procedure was used in this study. This method is accepted in the literature; however
rarely used (Glenn, 1977). This method was used as opposed to dummy variable
regression because it violated fewer assumptions and revealed more information.
Typically, analysis of variance is used in experimental design; however, in cohort

analysis the examination of effects is post hoc or after the fact.

Multicollinearity as a Complexity of Cohort Analysis

Many authors have suggested that “the inﬂuences of cohort membership, age and
period is appealing however, operationally complex” (Rentz & Reynolds, 1991, p.18).
The effects of generation, age and period are confounded in traditional research.
Longitudinal designs using repeated measurements or independent samples of the same
generations, confound the effects of age and the effects of period. Mason et a1. (1973)
presented a lucid discussion of the implications of confounding. They showed that the
confounding makes the parameter estimation in most statistical models impracticable.
Multicollinearity exists when there is a high correlation between independent variables.
The R2 is not affected by multicollinearity; however, parameter estimates are attenuated
(under estimated). The major problem with multicollinearity is that the effects of age,
period and generation can not be separated and therefore problems are caused when
interpreting the amount of variance explained by each of the effects. According to
Kennedy (1998), there are two options for researchers faced with multicollinearity: (1) do
nothing or (2) incorporate additional information. The second option includes any of the
following alternatives: obtaining more data, formalizing relationships among regressors

(independent variables), specifying a relationship among some parameters, dropping a

95

variable, incorporating estimates from other studies, forming a principal component, and
shrinking the OLS estimates.

As with all cohort studies, multicollinearity is a rather central problem. One
statistic that demonstrates the extent of multicollinearity is a correlation table. The
correlations between the three independent variables (age, generation and period) in this
study revealed a very high correlation between age and generation (r=.91). The
correlation between age and period was negative and not as strong; however, due to
sample size, it was still statistically signiﬁcant (r= -.50). The ﬁnal correlation between
generation and period was slightly less than age and period; however still signiﬁcantly

different (r=.32).

Table 17. Multicollinearity of Age, Generation and Period

 

 

Age Period Generation
Age 1 .00
Period -.50** 1.00
Generation .91 ** .32” 1.00

 

** signiﬁcant at .01 level of signiﬁcance

Using Cohort Tables as a Method of Cohort Analysis

Cohort analysis using cohort tables should be “viewed as a method of research,
not a statistical technique” (Rentz & Reynolds, 1981, p. 63). A cohort table involves
simultaneous process and cross-sectional comparisons. The ability to examine
developmental (age) and cohort differences simultaneously is regarded as a‘unique
property of cohort tables (Riley et al., 1972). In this study, however, all three effects are
presented together. The main reason is because the categories of age groups (seven

categories in increments of 10) are not clearly representative of the generations deﬁned

96

by Strauss and Howe. Because the original surveys had age categorized as a nominal
variable, separation into the exact dates outlined by Strauss and Howe was not possible.
Therefore, overlapping age categories were dropped and those that clearly fell into the
four generations outlined by Strauss and Howe were kept. The best way to communicate
the results in a cohort table was by examining the bivariate relationships separately.

The tables are constructed so that subgroup means are reported for each of the
age, period and generations on each of the travel-related variables (Tables 18-29). SPSS
for Windows was used to generate the subgroup means. The interpretation of the tables is
not straightforward. The analysis is performed by looking for patterns in the data. If the
patterns in both time periods are the same the variables can be treated independently and
explained independently; however, if the patterns are not the same, the independent

variables must be treated as an interaction effect.

Results of Analysis Using Cohort Tables
Money Smut on Travel is Well Smut

There was no particular pattern based on age or generation. In fact, upon
examination of all three tables, it would appear that period effects play the largest role in
agreement with the statement “money spent on travel is well spent.” “Money spent on
travel is well spent” had an overall mean of .44 in 1983 and a mean of .37 in 1995. This
represented an absolute difference of .06. Using the “standard” set forth in the previous
section this appears to be a meaningful difference. Canadians in 1995 were more likely to

“agree” with the statement.

97

Canadians between 25 and 34 were the most likely to agree with the statement
“money spent on travel is well spent” in 1983 (Table 18a). In 1995, Canadians between
35 and 44 were the most likely to agree that “money spent on travel is well spent.” In
1983, the Baby Boomer and Silent Generations agreed most with the statement that
“money spent on travel is well spent.” In 1995, the Baby Boomer Generation was most
likely to agree with the statement. Examination of Table 18c revealed that as each
generation aged, they were more likely to agree that “money spent on travel is well

spent.”

Table 18a. Age by Period on “Money spent on Travel is Well Spent”

 

 

Age Class 1983 1995 Pooled Data
15-19 0.44 0.39 0.42
20-24 0.44 0.37 0.42
25-34 0.43 0.36 0.41
35-44 0.43 0.35 0.42
45-54 0.44 0.38 0.42
55-64 0.44 0.36 0.42
65+ 0.45 0.37 0.43
Total 0.44 0.37 0.42

 

Scale 0-1 where 0= “strongly agree,” .25 = “agree,” .50 = “neutral,”
.75 = “disagree” and 1= “strongly disagree.”

Table 18b. Generations by Period on “Money Spent on Travel is Well

 

 

Spent”

Generation Class 1983 1995 Pooled Data
Generation X 0.44 0.37 0.39
Baby Boomer 0.43 0.36 0.41
Silent Generation 0.43 0.37 0.43

GI Generation 0.44 0.37 0.44
Total 0.44 0.37 0.40

 

Scale 0-1 where 0= “strongly agree,” .25 = “agree,” .50 = “neutral,”
.75 = “disagree” and l= “strongly disagree.”

98

Table 18c. Age by Generation on “Money Spent on Travel is Well Spent”

 

 

Age Class Generation Baby Silent GI Overall
X Boomer
15-19 0.42 0.42
20-24 0.37 0.42 0.42
25-34 0.36 0.42 0.41
35-44 0.38 0.43 0.42
45-54 0.37 0.41 0.42
55-64 0.36 0.44 0.42
65+ 0.43 0.43
Total 0.39 0.42 0.43 0.44 0.42

 

Scale 0-1 where 0= “strongly agree,” .25 = “agree,” .50 = “neutral,”
.75 = “disagree” and l= “strongly disagree.”

A graph illustrating the effects between generation and period illustrates “no
interaction” between these two variables (Figure 6). A chart of all three variables would

be three-dimensional and none of the lines would intersect.

 

.42 q

.40 I

......................... Period of time

.36 u _
1983

-——-

.34 1995
13th Generation Baby Goomer Silent ngeration GI Generation

 

 

 

Generation

Figure 6. Graphical Representation of Interaction Effects between Generation and
Period for “Money Spent on Travel is Well Spent.”

99

Immrtant that People Smale my Langu_age

Overall, agreement with the statement“ it is important that people speak my
language” was greater in 1983 than in 1995 (Table 19a). It appeared that in the 1983 data
Canadians were more likely to agree with the statement that “it is important that people I
encounter speak my same language” the older they got. The statement that it is important
that people speak my language was more agreed upon by the two oldest age groups in
1983 than other age groups. Whereas, in 1995, the opposite appeared to be true, that
younger Canadians agreed more with the same statement. Examination of the patterns
across both time periods for the seven age groups does not reveal a consistent pattern.
This may suggest there are interaction effects present.

In 1983, the GI Generation was the most likely to agree with the importance of
encountering people their same language, whereas, in 1995, the GI Generation was most
likely to agree that it is important that people speak my same language (Table 19b).
Examination of a cross section of age and generation revealed that there were no

consistent patterns within generations or age groups (Table 19c).

Table 19a. Age by Period for “Important That People Speak my
Language”

 

 

Age Class 1983 1995 Pooled Data
15-19 0.52 0.52 0.52
20-24 0.53 0.53 0.53
25-34 0.53 0.52 0.53
35-44 0.53 0.55 0.54
45-54 0.51 0.53 0.51
55-64 0.47 0.53 0.48
65+ 0.43 0.55 0.45
Total 0.50 0.54 0.51

 

Scale 0-1 where 0= “strongly agree,” .25 = “agree,” .50 = “neutral,”
.75 = “disagree” and 1= “strongly disagree.”

100

Table 19b. Generation by Period for “Important People Speak my

 

 

Language”

Generation Class 1983 1995 Pooled Data
Generation X 0.52 0.52 0.52
Baby Boomer 0.53 0.54 0.53
Silent Generation 0.52 0.53 0.52

GI Generation 0.45 0.55 0.46
Total 0.50 0.54 0.51

 

Scale 0-1 where 0= “strongly agree,” .25 = “agree,” .50 = “neutral,”
.75 = “disagree” and 1= “strongly disagree.”

Table 19c. Age by Generation for “People Speak my Same Language”

 

 

Age Class Gen X Baby Silent GI Overall
Boomer
15-19 0.52 0.52
20-24 0.53 0.53 0.53
25-34 0.51 0.53 0.53
35-44 0.55 0.53 0.54
45-54 0.53 0.51 0.51
55-64 0.53 0.47 0.48
65+ 0.45 0.45
Total 0.52 0.53 0.52 0.46 0.51

 

Scale 0-1 where 0= “strongly agree,” .25 = “agree,” .50 = “neutral,”
.75 = “disagree” and 1= “strongly disagree.”

Having Fun and Being Entertained

The importance of “having fun and being entertained” appeared to have two
effects, an age effect as well as a period effect. In the pooled data, “having fun and being
entertained” was overall more important for younger Canadians than it was for older
Canadians (Table 20a). Examination of the data for the two years revealed that similar
patterns failed to exist. Again this lack of consistency in patterns suggests an interaction
between the two variables may exist.

Examination of generations over time indicated there was no consistent pattern in

responses to “having fun and being entertained” (Table 20b). In 1983, each consecutive

101

generation rated “having fun and being entertained” as less important, however, in 1995,
this pattern was not as consistent.

Examination of the age and generation table revealed each generation was lower
than the next on most of the scores (Table 20c). No real pattern existed for each
generation as it aged.

A slight pattern in responses over time and as one ages may suggest a
generational effect. The combination of changes in the environment (time period) as one

ages might be interpreted as generational effects.

Table 20a. Age by Period for “Having Fun, Being Entertained”

 

 

Age Class 1983 1995 Pooled Data
15-19 0.13 0.17 0.15
20-24 0.15 0.14 0.15
25-34 0.22 0.17 0.21
35-44 0.25 0.16 0.23
45-54 0.27 0.19 0.25
55-64 0.29 0.18 0.21
65+ 0.28 0.18 0.26
Total 0.24 0.17 0.22

 

Scale 0-1 where 0= “very important,” .33 is “somewhat important,”
.66 is “not very important,” and 1= “not at all important.”

Table 20b. Generation by Period for “Having Fun, Being Entertained”

 

 

Generation Class 1983 1995 Pooled Data
Generation X 0.13 0.15 0.14
Baby Boomer 0.20 0.18 0.19
Silent Generation 0.26 0.19 0.25
GI Generation 0.28 0.18 0.27
Total 0.24 0.17 0.22

 

Scale O-l where 0= “very important,” .33 is “somewhat important,”
.66 is “not very important,” and 1= “not at all important.”

102

Table 20c. Age by Generation for “Having Fun, Being Entertained”

 

 

Age Class Gen Baby Silent GI Overall
X Boomer
15-19 0.15 0.15
20-24 0.14 0.16 0.15
25-34 0.15 0.21 0.21
35-44 0.16 0.25 0.23
45-54 0.18 0.26 0.25
55-64 0.18 0.29 0.27
65+ 0.26 0.26
Total 0.15 0.19 0.25 0.27 0.22

 

Scale 0-1 where 02 “very important,” .33 is “somewhat important,”
.66 is “not very important,” and 1= “not at all important.”

The graphical representation of the relationship between generation and period for
the variable “having ﬁm and being entertained” illustrates the large interaction effects
(Figure 7). The graph indicates a strong negative linear relationship between generation
and importance of “having fun and being entertained” in 1983 and a slightly weaker
negative linear relationship in 1995. This suggests that older generations are less likely to

agree that “having fun and being entertained” is an important part of pleasure travel.

 

.21

Period of time

1983

 

.1 1995
13th Generation Baby Rom er Silent ngoration GI Generation

 

 

 

Generation

Figure 7. Graphical Representation of Interaction Effects of Generation and Period
for “Having Fun and Being Entertained.”

103

National and Provincial Parks

Over time, the importance of “national and provincial parks” has declined from
1983 to 1995. The absolute change from 1983 to 1995 was a decline of .04 (overall
mean1995 — mean1933). Examination of the pooled data indicated that as Canadians age
they are less likely to agree that “national and provincial parks” are important to their
pleasure travel. This pattern is not consistent at either time period but when the data are
pooled reveals more of a linear pattern.

“National and provincial parks” were more important for Generation Xers and
Baby Boomers than for any other generation (Table 21b). In the 1983 data, Generation X
was more likely to agree that “national and provincial parks” are important whereas, in
1995, Baby Boomers and the Silent Generation were more likely to agree “national and
provincial parks” are important to pleasure travel.

Examination of Table 21c revealed that for three of the four generations
(Generation X , Baby Boomers, and GI Generation) that “national and provincial parks”
became less important as they aged, whereas, for the Silent Generation, “national and
provincial parks” became slightly more important. However, using the standard of an
absolute difference of .05 to measure meaningful differences would suggest that this is

probably not a meaningful difference.

104

Table 21a. Age by Period for National and Provincial Parks

 

 

Age Class 1983 1995 Pooled Data
15-19 0.24 0.32 0.27
20-24 0.26 0.32 0.27
25-34 0.30 0.31 0.30
35-44 0.26 0.32 0.31
45-54 0.28 0.32 0.31
55-64 0.32 0.30 0.31
65+ 0.33 0.33 0.33
Total 0.28 0.32 0.30

 

Scale 0-1 where 0= “very important,” .33 is “somewhat important,”

.66 is “not very important,” and l= “not at all important.”

Table 21b. Generation by Period for National and Provincial Parks

 

 

Generation Class 1983 1995 Pooled Data
Generation X 0.24 0.32 0.29
Baby Boomer 0.26 0.31 0.29
Silent Generation 0.31 0.31 0.31
GI Generation 0.32 0.33 0.32
Total 0.28 0.32 0.30

 

Scale 0-1 where 0= “very important,” .33 is “somewhat important,”

.66 is “not very important,” and 1= “not at all important.”

Table 21c. Age by Generation for National and Provincial Parks

 

 

Age Class Gen Baby Silent GI Overall
X Boomer
15-19 0.27 0.27
20-24 0.32 0.26 0.27
25-34 0.31 0.30 0.30
35-44 0.32 0.31 0.31
45-54 0.31 0.31 0.31
55-64 0.30 0.32 0.31
65+ 0.33 0.33
Total 0.29 0.29 0.31 0.32 0.30

 

Scale 0-1 where 0= “very important,” .33 is “somewhat important,
.66 is “not very important,” and 1= “not at all important.”

105

9,

Beaches for Swimming and Sunning

The largest effect on the importance of “beaches for swimming and sunning” in
pleasure travel appeared to be age effects (Table 22a). In 1983, younger Canadians were
more likely to rank “beaches for swimming and sunning” as important (Table 22a). In
1995, the pattern was not as clear. Those between 20-25 and 35-44 years of age ranked
“beaches for swimming and sunning” more important than any other age group.

The overall importance of “beaches for swimming and sunning” to Canadians
remained fairly consistent over the 12 year period; however closer examination revealed
that in 1983 Canadians responses ranged from close to very important to much less than
somewhat important; whereas in 1995, the scores all hovered around “somewhat
important.”

Table 22c indicates that as each generation ages, they are less likely to frnd
“beaches for swimming and sunning” as important to their pleasure travel. Table 22c
revealed a fairly consistent pattern in age effects for each generation. In addition, older
generations rated “beaches for swimming and sunning” slightly less important than the

younger generations.

lable 22a. Age by Period for Beaches for Swimming and Sunning

 

 

 

Age Class 1983 1995 Pooled Data
15-19 0.19 0.35 0.26
20-24 0.22 0.30 0.24
25-34 0.24 0.31 0.25
35-44- 0.28 0.30 0.29
45-54- 0.33 0.33 0.33
55-64 0.40 0.34 0.29
65+ 0.48 0.33 0.45
Total 0.31 0.32 0.31

 

Scale 0-1 where 0= “very important,” .33 is “somewhat important,”
.66 is “not very important,” and l= “not at all important.”

106

Table 22b. Generation by Period for Beaches for Swimming and Sunning

 

 

Generation Class 1983 1995 Pooled Data
Generation X 0.19 0.31 0.28
Baby Boomer 0.23 0.31 0.25
Silent Generation 0.30 0.33 0.30
GI Generation 0.44 0.33 0.43
Total 0.31 0.32 0.31

 

Scale 0-1 where 0= “very important,” .33 is “somewhat important,”
.66 is “not very important,” and 1= “not at all important.”

Table 22c. Age by Generation for Beaches for Swimming and Sunning

 

 

Age Class Gen Baby Silent GI Overall
X Boomer
15-19 0.26 0.26
20-24 0.29 0.22 0.24
25-34 0.29 0.25 0.27
35-44 0.30 0.28 0.29
45-54 0.32 0.33 0.33
55-64 0.33 0.40 0.39
65+ 0.45 0.45
Total 0.28 0.25 0.30 0.43 0.43

 

Scale 0-1 where 0= “very important,” .33 is “somewhat important,”
.66 is “not very important,” and 1= “not at all important.”

When examining a graph of the relationship between age and period for the
variable “beaches for swimming and sunning,” it was apparent that large interaction
effects were present. The graph indicated a strong linear relationship between age and
importance of “beaches for swimming and sunning” in 1983 and a less distinct

relationship in 1995 (Figure 8).

107

 

.4-

.3.

 

'2‘ Period of time
1983
_1 1995

 

 

 

young-19 20'24 25:34 3534 45754 5564 65+

Age category

Figure 8. Graphical Representation of Interaction Effects of Age and Period for
“Beaches for Swimming and Sunning”

Hi uali Restaurants

The importance of “high quality restaurants” was more important in 1983 than in
1995. An overall change in score from .41 to a score of .46 suggests that Canadians found
“high quality restaurants” more important in 1983. The scores suggest that Canadians rate
“high quality restaurants” somewhat important to their pleasure travel. The absolute

difference between 1983 and 1995 was .03. Based on the previous standard, this does not

indicate a meaningful difference.

108

Canadians between 45-54 in 1983 and 35-44 year old Canadians in 1995 (Table
23a) rated “high quality restaurants” more important. Members of these age categories
were also members of the Silent Generation. The Silent Generation tended to rate “high
quality restaurants” more important than any other generation, although as they age it
becomes less important to them (Table 23b and 230). There was no consistent pattern in
responses from one generation to the next over time, however one consistent pattern was
that as each generation aged, it rated “high quality restaurants” less important (Table

23c)

Table 23a. Age by Period for High Quality Restaurants

 

 

Age Class 1983 1995 Pooled Data
15-19 0.43 0.49 0.45
20-24 0.41 0.47 0.42
25-34 0.40 0.46 0.42
35-44 0.40 0.45 0.41
45-54 0.39 0.47 0.41
55-64 0.43 0.46 0.44
65+ 0.45 0.46 0.45
Total 0.41 0.46 0.42

 

Scale 0-1 where 0= “very important,” .33 is “somewhat important,”
.66 is “not very important,” and 1: “not at all important.”

Table 23b. Age by Generation for High Quality Restaurants

 

 

Generation Class 1983 1995 Pooled Data
Generation X 0.43 0.43 0.46
Baby Boomer 0.40 0.46 0.42
Silent Generation 0.40 0.44 0.41
GI Generation 0.44 0.46 0.44
Total 0.41 0.46 0.42

 

Scale 0-1 where 0= “very important,” .33 is “somewhat important,”
.66 is “not very important,” and 1= “not at all important.”

109

Table 23c. Age by Generation for High Quality Restaurants

 

 

Age Class Gen Baby Silent GI Overall
X Boomer
15-19 0.45 0.45
20-24 0.47 0.41 0.42
25-34 0.47 0.41 0.41
35-44 0.45 0.40 0.41
45-54 0.48 0.40 0.42
55-64 0.46 0.43 0.44
65+ 0.45 0.45
Total 0.46 0.42 0.41 0.44 0.42

 

Scale 0-1 where 0= “very important,” .33 is “somewhat important,”
.66 is “not very important,” and 1= “not at all important.”

Budget Accommodations

The importance of “budget accommodations” in pleasure travel was relatively
consistent over time. In fact, examination of the overall mean scores in each time period
revealed no difference in the rating of the importance of “budget accommodations”
(Table 24a).

Scrutiny of the responses to the importance of “budget accommodations” by each
age group in the two time periods revealed no consistent patterns. All responses tended to
hover around a neutral response; however, responses varied by age category (Table 24a).

“Budget accommodations” were most important to the Silent Generation in 1983
and both the Silent Generation and the GI Generation in 1995 (Table 24b). No clear
pattern existed when the combination of age and generation were examined (Table 24c).

This suggests a possible interaction.

110

Table 24a. Age by Period for Budget Accommodations

 

 

Age Class 1983 1995 Pooled Data
15-19 0.50 0.50 0.50
20-24 0.49 0.48 0.49
25-34 0.49 0.50 0.49
35-44 0.47 0.49 0.48
45-54 0.46 0.46 0.46
55-64 0.48 0.47 0.48
65+ 0.49 0.47 0.48
Total 0.48 0.48 0.48

 

Scale 0-1 where 0= “very important,” .33 is “somewhat important,”
.66 is “not very important,” and 1= “not at all important.”

Table 24b. Generation by Period for Budget Accommodations

 

 

Generation Class 1983 1995 Pooled Data
Generation X 0.50 0.49 0.49
Baby Boomer 0.49 0.48 0.49
Silent Generation 0.47 0.47 0.47
GI Generation 0.48 0.47 0.48
Total 0.48 0.48 0.48

 

9,

Scale 0-1 where 0= “very important,” .33 is “somewhat important,
.66 is “not very important,” and 1= “not at all important.”

Table 24c. Age by Generation for Budget Accommodations

 

 

Age Class Gen Baby Silent GI Overall
X Boomer
15-19 0.50 0.50
20-24 0.48 0.49 0.49
25-34 0.49 0.49 0.49
35-44 0.47 0.48 0.48
45-54 0.47 0.46 0.46
55-64 0.47 0.48 0.48
65+ 0.48 0.48
Total 0.49 0.48 0.47 0.48 0.48

 

Scale 0-1 where 0= “very important,” .33 is “somewhat important,”
.66 is “not very important,” and 1= “not at all important.”

lll

First Class Accommodations

The importance of “ﬁrst class accommodations” in pleasure travel gained
importance over time (Table 25a). Overall, Canadians rated “ﬁrst class accommodations”
more important in 1995 than they did in 1983 (Table 25a). In fact, scores on this variable
indicate that “ﬁrst class accommodations” are fairly important to pleasure travel
(mean1995=.29).

When examining the change in importance within the generations over time, it is
interesting to note that Generation X had the most drastic change in their rating of the
importance of “ﬁrst class accommodations.” In fact, the percentage change got smaller
for each older generation. This suggests a period effect that is affecting younger .
Canadians at a greater rate than older Canadians (Table 25b).

Table 25c indicates that “ﬁrst class accommodations” is more important for
Generation X in their middle years of the generation and for the remaining three
generations “ﬁrst class accommodations” becomes less important as they age.
Examination of the total scores for each generation reveals that each consecutive
generation rates “ﬁrst class accommodations” less important than the previous generation

and that this also becomes less important as one ages (Table 25c).

112

Table 25a. Age by Period for First Class Accommodations

 

 

Age Class 1983 1995 Pooled Data
15-19 0.31 0.26 0.29
20-24 0.32 0.23 0.31
25-34 0.33 0.27 0.32
35-44 0.34 0.31 0.33
45-54 0.34 0.31 0.33
55-64 0.35 0.33 0.35
65+ 0.37 0.35 0.37
Total 0.34 0.29 0.33

 

Scale 0-1 where 0: “very important,” .33 = “somewhat important,”
.66 = “not very important,” and l= “not at all important.”

Table 25b. Generation by Period for First Class Accommodations

 

 

Generation Class 1983 1995 Pooled Data
Generation X 0.31 0.25 0.27
Baby Boomer 0.33 0.30 0.32
Silent Generation 0.34 0.33 0.34
GI Generation 0.36 0.35 0.36
Total 0.34 0.29 0.33

 

,9

Scale 0-1 where 0= “very important,” .33 is “somewhat important,
.66 is “not very important,” and 1= “not at all important.”

Table 25c. Age by Generation for First Class Accommodations

 

 

Age Class Gen Baby Silent GI Overall
X Boomer
15-19 0.29 0.29
20-24 0.24 0.32 0.31
25-34 0.25 0.33 0.32
35-44 0.31 0.34 0.33
45-54 0.30 0.34 0.33
55-64 0.33 0.35 0.35
65+ 0.37 0.37
Total 0.27 0.32 0.34 0.36 0.33

 

Scale 0-1 where 0= “very important,” .33 is “somewhat important,”
.66 is “not very important,” and l= “not at all important.”

113

Shopping

Overall, “shopping” gained in importance from 1983 to 1995. In 1983, Canadians
rated “shopping” as a component of pleasure travel as a little less than “somewhat
important.” In 1995, Canadians rated “shopping” as slightly better than “somewhat
important.” The importance of “shopping” to Canadians increased from 1983 to 1995. An
absolute difference in mean scores from 1983 to 1995 suggests this difference is a
meaningful difference (greater than .05).

Examination of Table 26a revealed that neither time period showed consistent
patterns across age groups. In 1983, the youngest age group (15-19) indicated “shopping”
was the most important. In 1995, the age groups from 35-65+ indicated that “shopping”
was more important than the younger age groups.

Table 26b indicates that in 1983, Generation X rated shopping most important and
in 1995 all generations rated shopping about the same level of importance. The pooled
data revealed a increase in importance from the youngest generation to the oldest
generation; however, closer examination indicates that each period is distinctly different
from the other.

Table 26c suggests that for the three youngest generations (Generation X, Baby
Boomers and Silent) as they aged, they rated shopping more important. For the oldest

generation, the GI Generation, shopping became slightly less important with age.

114

Table 26a. Age by Period for Shopping

 

 

Age Class 1983 1995 Pooled Data
15-19 0.32 0.36 0.34
20-24 0.34 0.30 0.34
25-34 0.36 0.32 0.35
35-44 0.37 0.31 0.36
45-54 0.35 0.31 0.34
55-64 0.36 0.31 0.35
65+ 0.39 0.31 0.37
Total 0.36 0.31 0.35

 

Scale 0-1 where 0= “very important,” .33 is “somewhat important,”
.66 is “not very important,” and l= “not at all important.”

Table 26b. Generation by Period for Shopping

 

 

Generation Class 1983 1995 Pooled Data
Generation X 0.32 0.32 0.32
Baby Boomer 0.36 0.31 0.35
Silent Generation 0.36 0.31 0.36
GI Generation 0.37 0.31 0.37
Total 0.36 0.31 0.35

 

Scale 0—1 where 0= “very important,” .33 is “somewhat important,”
.66 is “not very important,” and l= “not at all important.”

Table 26c. Age by Generation for Shopping

 

 

Age Class Gen Baby Silent GI Overall
X Boomer
15-19 0.34 0.34
20-24 0.30 0.34 0.34
25-34 0.30 0.36 0.35
35-44 0.31 0.37 0.36
45-54 0.31 0.35 0.34
55-64 0.31 0.36 0.35
65+ 0.37 0.37
Total 0.32 0.35 0.36 0.37 0.35

 

Scale 0-1 where 0= “very important,” .33 is “somewhat important,

.66 is “not very important,” and 1= “not at all important.”

115

’9

Nightlife and Entertainment

Overall, “nightlife and entertainment” in pleasure travel was slightly more
important in 1995 than it was in 1983 (Table 27a). “Nightlife and entertainment” had an
overall mean of .43 in 1983 and a mean of .42 in 1995. The absolute difference of .01
suggests that this difference is probably not a meaningful difference. Canadians rated
“nightlife and entertainment” as a fairly neutral component to their pleasure travel.

Upon examination of all three tables, it would appear that age effects play the
largest role in the importance of “nightlife and entertainment.” In 1983, there is a clearly
negative linear relationship between age and importance of “nightlife and entertainment.”
Thus, as one ages, “nightlife and entertainment” becomes less important. In 1995, the
relationship is not as clear. Canadians who are between the ages of 20-24 and 55-64 rated
“nightlife and entertainment” most important. The pooled data would suggest a negative
linear relationship between age and importance of “nightlife and entertainment.”

Examination of Table 27b indicates a relationship between generation and
importance of “nightlife and entertainment.” This relationship is more clear and greater in
1983 than in 1995. In 1995, the ratings are more clustered around the mean and less
variable. Thus, the differences in the two periods would suggest the presence of an
interaction effect between generation and period.

Investigation of the age by generation table (Table 27c) illustrated that for each
generation, “nightlife and entertainment” becomes less important as they age. This is

consistent within each generation and within time periods.

116

Table 27a. Age by Period for Nightlife and Entertainment

 

 

Age Class 1983 1995 Pooled Data
15-19 0.23 0.42 0.31
20-24 0.27 0.40 0.30
25-34 0.37 0.42 0.38
35-44 0.44 0.41 0.43
45-54 0.48 0.44 0.47
55-64 0.53 0.40 0.50
65+ 0.58 0.42 0.55
Total 0.43 0.42 0.43

 

Scale 0-1 where 0= “very important,” .33 is “somewhat important,”
.66 is “not very important,” and 1= “not at all important.”

Table 27b. Generation by Period for Nightlife and Entertainment

 

 

Generation Class 1983 1995 Pooled Data
Generation X 0.23 0.41 0.36
Baby Boomer 0.35 0.42 0.36
Silent Generation 0.45 0.41 0.45
GI Generation 0.56 0.42 0.54
Total 0.43 0.42 0.43

 

9’

Scale 0-1 where 0= “very important,” .33 is “somewhat important,
.66 is “not very important,” and 1= “not at all important.”

Table 27c. Age by Generation for Nightlife and Entertainment

 

 

Age Class Gen Baby Silent GI Overall
. X Boomer
15-19 0.31 0.31
20-24 0.40 0.27 0.30
25-34 0.41 0.38 0.38
35-44 0.41 0.43
45-54 0.44 0.44 0.47
55-64 0.48 0.52 0.50
65+ 0.55 0.55
Total 0.36 0.36 0.45 0.54 0.43

 

”

Scale 0-1 where 0== “very important,” .33 is “somewhat important,
.66 is “not very important,” and 1= “not at all important.”

117

A graph of the relationship between age and period for the variable “nightlife and
entertainment” illustrates the large interaction effects (Figure 9). The interaction effects
are represented by the intersection of the lines. The graph indicated a strong linear
relationship between age and importance of “nightlife and entertainment” in 1983 and a

less distinct relationship in 1995.

 

.41

Period of time

1983

.3.

 

.2 1995
young-19 20724 25734 3534 45754 ass-'64 6: +

 

 

 

Age category

Figure 9. Graphical Representation of Interaction Effects of Age and Period for
“Nightlife and Entertainment.”

Museums and Art Galleries

The largest effect for the change in the importance of “museums and art galleries”
appears to be change in period (Table 28a). The mean score in 1983 was .40 while the
mean score in 1995 was .45. This represents an absolute change of .05, a meaningful
difference. “Museums and art galleries” were rated as more important in 1983 than in
1995.

Table 28a revealed the oldest two age groups rated “museums and art galleries” as

most important (.33 and .39 respectively) in 1983. In 1995, those between the ages of 34-

118

44 rated “museums and art galleries” as most important. No clear linear relationship is
apparent in either of the time periods. However, the pooled data suggests that “museums
and art galleries” are more important to older Canadians than to younger Canadians
Examination of Table 28b indicates minimal changes in responses across the
generations and across time. The GI Generation had the greatest change between 1983
and 1995 in their rating of the importance of “museums and art galleries.” Baby Boomers
were the most likely to rate “museums and art galleries” consistently over time.
“Museums and art galleries” became less important to all generations other than the Baby

Boomers as they aged (Table 28c).

Table 28a. Age by Period for Museums and Art Galleries

 

 

Age Class 1983 1995 Pooled Data
15-19 0.43 0.47 0.44
20-24 0.44 0.45 0.44
25-34 0.44 0.47 0.44
35-44 0.41 0.43 0.42
45-54 0.33 0.45 0.42
55-64 0.39 0.44 0.40
65+ 0.40 0.45 0.41
Total 0.40 0.45 0.42

 

Scale 0-1 where 0= “very important,” .33 is “somewhat important,”
66 is “not very important,” and 1= “not at all important.”

Table 28b. Generation by Period for Museums and Art Galleries

 

 

Generation Class 1983 1995 Pooled Data
Generation X 0.43 0.47 0.44
Baby Boomer 0.43 0.44 0.44
Silent Generation 0.41 0.45 0.42
GI Generation 0.39 0.45 0.40
Total 0.40 0.45 0.42

 

Scale 0-1 where 0= “very important,” .33 is “somewhat important,”
.66 is “not very important,” and 1= “not at all important.”

119

Table 28c. Age by Generation for Museums and Art Galleries

 

 

Age Class Gen Baby Silent GI Overall
X Boomer
15-19 0.44 0.44
20-24 0.45 0.44 0.44
25-34 0.49 0.44 0.44
35-44 0.43 0.42 0.42
45-54 0.43 0.41 0.41
55-64 0.44 0.39 0.40
65+ 0.41 0.41
Total 0.46 0.44 0.42 0.40 0.42

4%

Scale 0-1 where 0= “very important,” .33 is “somewhat important,”
.66 is “not very important,” and l= “not at all important.”
Amusement and Theme Parks

Overall, the period appeared to have little effect on the importance of “amusement
and theme parks” in pleasure travel (Table 29a). The means scores in both 1983 and 1995
were relatively close at .42 in 1983 and .43 in 1995.

Age appeared to have the greatest effect on the importance of “amusement and
theme par ” (Table 29a). Overall, younger Canadians rated “amusement and theme
parks” more important than older Canadians. In 1983, a clear pattern existed in the
responses to the importance of “amusement and theme parks.” Older Canadians rated
“amusement and theme parks” less important than younger Canadians. In 1995, this
pattern was not as clear. Those between the ages of 20-24 rated “amusement and theme
parks” as most important. The variation in scores in 1995 was much less than the
variation in scores in 1983. The scores in 1995 were all fairly neutral compared to 1983
where they ranged from somewhat important to closer to not very important.

The generation table (Table 29b) indicated that overall older generations rated

“amusement and theme par ” as less important. However, this pattern was not as

120

consistent in both time periods. In 1983, the pattern was fairly clear. Generation X rated
amusement and theme parks as most important. In 1995, the range of scores was much
less and a linear pattern did not exist. This may indicate an interaction between the two
variables.

Examination of the age by generation table illustrated that as each generation
aged, they were less likely to rate “amusement and theme parks” as important (Table
29c). Each consecutive generation tended to rate “amusement and theme parks” less
important than the next and as each generation aged they were less likely than the next to

ﬁnd “amusement and theme parks” important to their pleasure travel.

Table 29a. Age by Period for Amusement and Theme Parks

 

 

Age Class 1983 1995 Pooled Data
15-19 0.32 0.44 0.37
20-24 0.39 0.39 0.39
25-34 0.40 0.43 0.41
35-44 0.42 0.43 0.43
45-54 0.45 0.44 0.45
55-64 0.49 0.44 0.48
65+ 0.51 0.44 0.50
Total 0.42 0.43 0.43

 

Scale 0-1 where 0= “very important,” .33 is “somewhat important,”
.66 is “not very important,” and 1= “not at all important.”

Table 29b. Generation by Period for Amusement and Theme Parks

 

 

Generation Class 1983 1995 Pooled Data
Generation X 0.32 0.42 0.39
Baby Boomer 0.39 0.43 0.41
Silent Generation 0.45 0.43 0.44
GI Generation 0.50 0.44 0.49
Total 0.42 0.43 0.43

 

Scale 0-1 where 0= “very important,” .33 is “somewhat important,”
.66 is “not very important,” and 1= “not at all important.”

121

Table 29c. Age by Generation for Amusement and Theme Parks

 

 

Age Class Gen Baby Silent GI Overall
X Boomer
15-19 0.37 0.37
20-24 0.39 0.39 0.39
25-34 0.43 0.41 0.41
35-44 0.43 0.43 0.43
45-54 0.44 0.46 0.45
55-64 0.44 0.49 0.48
65+ 0.50 0.50
Total 0.39 0.41 0.44 0.49 0.43

 

Scale 0-1 where 0= “very important,” .33 is “somewhat important,”
.66 is “not very important,” and 1= “not at all important.”

Analysis of Variance as a Method of Cohort Analysis

Analysis of variance as a cohort method is an appropriate method because it
reveals the main effects of the three variables (age, period and generation), the tvvo-way
interactions (age-period, age-generation, and period-generation) and the three-way
interactions (age-period-generation). The R2 indicated how much variation in the
responses is accounted for by the interaction of the three variables.

Analysis of variance is a technique which partitions the total sum of squares into
two components, the variation between the subgroups (variation in sample means) and
the variation within each subgroup divided by generation, period and age (variation
among the treatment group means). The total deviation of a score from the grand mean is
represented by (Xij — X3). The between group variation due to an effect of the
independent variable is represented by (Xu —Xg), this is the difference of the subgroup
mean from the grand mean. The measure of within group or error variation (sum of
square of error) is calculated by xij-x. and represents the variation of scores within the

group from the group mean If the independent variable has no effect, then the variation

122

among the treatment group means will be due only to sampling error. The F statistic is
computed as a ratio of group variation to the between group variation. When the mean
squares between groups is the same as the mean square of errors, the F ratio is about 1.00.

One assumption of analysis of variance is that the F statistic will follow the F
distribution. If this assumption is violated, the accuracy of the approximations will be
affected. The analysis of variance test also depends on the assumption that the variances
of the different populations are equal. However, experience has shown that F-tests are
robust and not adversely affected by deviations from normality in the distribution
(Blalock, 1972 ).

Examination of the correlations between the independent variables previously
revealed an extremely high correlation between age and generation. In a cohort analysis
with a greater number of periods this correlation would be reduced; however, because
this study only had two years, the correlation was higher than normal. In spite of this
correlation, it was felt that the analysis of variance would provide support for the
relationship between the three independent variables and demonstrate the percentage of
variation accounted for by the combination of variables (age, generation and period).

The main effects, two-way interaction and the three-way interaction among age,
period and generation for each of the travel-related variables were presented. When the
signiﬁcance of an interaction between variables is greater than the main effects, it
suggests there is multicollinearity or that change in one variable is dependent on change
in another variable. When two or more explanatory variables are related, it is not possible

to disentangle their separate effects on the dependent variable. In past cohort analysis

123

these effects are treated as minimal or ignored. In this study, it was felt that it was

important to address the amount of interaction among age, period and generation.

Results of the Three- Way Analysis of Variance

The results of the tlrree-way analysis of variance for three variables (“nightlife
and entertainment,” “having fun and being entertained” and “beaches for swimming and
sunning”) are presented below (Table 30-32). The percentage variance explained (R2)
was reported as part of the output from the three-way analysis of variance procedure in
the Statistical Package for the Social Science 8.0 (SPSS). Three variables were chosen to
report because they had the largest R2 (variance explained by the combination of age,
period and generation). The results for the remaining variables are available in Appendix
C.

The greatest amount of variation in “nightlife and entertainment” was explained
by the interaction of age, period and generation (Table 30). A total of 11% of the
variation was accounted for by the interaction of the three independent variables, age,
period and generation. This was the largest amount of variation explained across all the
12 variables. The interaction among age, period and generation was highly signiﬁcant
and represented 95% of the total percentage variance explained (see Table 34). In
addition, the interaction between period and generation was statistically signiﬁcant. Large
interaction effects indicate that it is not possible to partial out the main effects of age,
period or generation. However, as with past cohort studies, holding the interaction

effects constant will allow the strength of the main effects to be addressed.

124

In the variable, “nightlife and entertainment,” the main effects of age were fairly
large (ot=.000). Thus, one could argue that of the three independent variables (age, period
and generation) age effects had the largest effects on nightlife and entertainment.
Examination of the mean scores from Table 27a show that nightlife and entertainment in
pleasure travel was more important for younger Canadians than it was for older
Canadians (note: lower scores indicate greater importance). The results support what was
illustrated in the “eyeballing” method. With the “eyeballing” method, age effects
appeared to be the strongest effects.

Further investigation of the analysis of variance indicated that the error (within
group variation) was larger than the sum of squares for the main effects or the interaction
effects. This suggested that there was more within group variation than between group
variation. This ﬁnding was consistent among all 12 travel related variables. Therefore,
the amount of variation between individuals was much greater than the amount of

variation within groups of Canadians.

Table 30. Three way AN OVA Examining Total Interaction of Age, Generation
and Period on the Importance of “Nightlife and Entertainment” in Pleasure
Travel

 

 

Independent Variables Sum of df Mean Square F Sig.
and Interaction Squares

Age 3.40 6 0.57 9.65 .000
Period 0.09 1 0.09 1.51 .219
Generation 0.17 2 0.08 1.43 .240
Age "‘ Period 0.16 2 0.08 1.34 .253
Age * Generation 0.17 1 0.17 2.84 .092
Period * Generation 0.25 1 0.25 4.23 .040
Age * Period * Generation 89.36 17 5.26 89.43 .000
Error 699.96 1 1909 0.06

‘

 

R9 .113 (Adjusted Rial 12)

125

For the travel-related variable “having ftm and being entertained,” the greatest
amount of variation was again explained by the interaction of age, period and generation
(Table 31). A total of 5% of the variation in the response to the importance of “having
ﬁm and being entertained,” was accounted for by age, period and generation. The
interaction between age, period and generation was highly signiﬁcant. None of the two-
way interactions were statistically signiﬁcant. Although the main effects of age and
generation were statistically signiﬁcant, due to large interaction effects it was not
possible to partial out the main effects. Therefore, drawing conclusions about any of the
main effects is impossible. However in order to better understand which of the main
effects may have the greatest effect on the importance of “having fun and being
entertained,” the interaction effects were held constant.

Examination of the mean scores indicate that “having fun and being entertained”
was more important for younger generations of Canadians than it was for older
generations of Canadians and that it was gaining importance over time (note: lower
scores indicate greater importance). The largest main effect on “having fun and being
entertained” was generation (F value = 4.12, or=.02). The ﬁndings from the analysis of
variance are consistent with the “eyeballing” technique. In the “eyeballing” technique it
was apparent through examination of age and generation that each generation ranked
“having fun and being entertained” more important than the last and that over time this
was also becoming more important.

Similar to “nightlife and entertainment,” the error (within group variation) was

larger than the sum of squares for the main effects or the interaction effects. This

126

suggested that there was more within group variation than between group (generation)
variation.
Table 31. Three Way AN OVA Examining Total Interaction of Age, Generation

and Period on the Importance of “Having Fun and Being Entertained” in
Pleasure Travel

 

 

Independent Variables Sum of df Mean Square F Sig.
and Interaction Squares

Age .753 6 .126 2.77 .011
Period .166 1 .166 3.65 .056
Generation .376 3 .188 4.12 .016
Age "' Period .019 3 .009 .211 .810
Age * Generation .005 l .005 .103 .748
Period * Generation .041 1 .041 .888 .346
Age * Period * Generation 28.34 17 1.67 36.57 .000
Error 542.64 1 1905 .045

 

R‘= .050 (Adjusted R2=.048)

Finally, just like with the other two variables, the greatest amount of variation in
“beaches for swimming and sunning” was explained by the interaction of age, period and
generation (Table 32).. A total of 8.6% of the total variation was accounted for by the
combination of age, period and generation. Again, it is not possible to partial out the main
effects. However, for the purpose of isolating the largest main effect, the interaction
effects were held constant and the F scores for the main effects were examined. In this
case, the largest main effect was age. Examination of the mean scores indicated that
“beaches for swimming and sunning” were more important for younger generations of
Canadians than for older generations of Canadians (note: lower scores indicate greater
importance). This is the same conclusion that was come to with the “eyeballing”
technique.

Similar to “nightlife and entertainment” and “having fun and being entertained,”

the error (within group variation) was larger than the sum of squares for the main effects

127

or the interaction effects, suggesting that there was more within group variation than
between group variation. Thus, the amount of variation among individuals was much
greater than the amount of variation among different groups (generations) of Canadians.
Table 32. Three Way AN OVA Examining Total Interaction of Age, Generation

and Period on the Importance of “Beaches for Swimming and Sunning” in
Pleasure Travel

 

 

Independent Variables Sum of df Mean Square F Sig.
and Interaction Squares

Age 4.45 6 .742 1 1.24 .000
Period .000 1 .000 .000 .986
Generation .109 3 .054 .826 .43 8
Age * Period .155 3 .078 1.18 .308
Age * Generation .001 1 .012 .019 .893
Period * Generation .005 1 .005 .084 .772
Age * Period * Generation 73.68 17 4.33 65.61 .000
Error 786.27 1 1902 .067

 

R2: .086 (Adjusted R2=.084)

Table 33 summarizes the F values for the main effects and the interaction among
age, period and generation for each of the travel-related variables. For all the variables,
there was a signiﬁcant interaction of age, period and generation. This result indicates that
the relationship among the three variables explains the most variation. To reiterate, when
two or more explanatory variables are related, it is not possible to disentangle their
separate effects on the dependent variable. In fact, the combination of age, generation and
period is so great that it accounts for between 74% and 96% of the total percentage of

explained variance (Table 34).

128

Table 33. F Values from Three-Way AN OVA for All Travel Variables

 

 

Main Main Main R’l
Effect Effect Effect Interaction Adj R2
ATge Period Generation “A
Money well spent 0.85 9.52W 0.06 16.96"“ .021/.020

Speak same language 4.51:“ 17.49.” 3.66:: 19.77: .025/024
Having fun, being 2.77 3.65 4.12 36.57 050/.043

entertained

 

A
r

 

 

National/provincial 0.48 5.1 7 0.50 4.48m .006/.005
parks

Beaches for swim & 11.24‘” .000 0.33 65.61‘” 036/.034
sun

High quality 0.86 5.33' 0.17 7.42‘” .010/.009
restaurants *

First class 1.13 0.62 1.12 130' 003/001
accommodations

Budget 0.33 1.36 1.69 903‘” 013/01 1
accommodations

Shopping 2.35: 4.59: 0.24 6.90"; .010/.003
Nightlife and 9.65“" 6.34'" 1.43 39.44‘" .113/.112
entertainment

Museums, art 0.35 633’” 2.03 503‘" 003/007
galleries

Amusement, theme 216‘ 0.70 0.13 2032‘” 029/027
parks

 

* signiﬁcant at 0t= .05
" signiﬁcant at or= .01
"* signiﬁcant at ot= .001

There were two variables where the F value for a main effect was greater than the
F value for the interaction effects. For these two variables, period explained the most
variation. The two variables were “museums and art galleries” and “national and
provincial parks.” According to Blalock (1972), “if the interaction effect is relatively
large, this simple distinction between main and interaction effects will be difficult to
translate into substantive or theoretical terms” (p. 346). He also suggests that when the
main effect is signiﬁcant but the interaction effect is large (F value is close to the main

effect) it does not make theoretical sense to obscure real differences by talking about the

129

average effects of age, period or generation. Therefore, although the main effects for
national and provincial parks and museums and art galleries were larger than the
interaction effects, the large size of the interaction effects rules out any meaningful

interpretation of period effects alone.

Table 34. Percentage of Main Effects and Interaction Effects for Total Percent of

 

 

Explained Variance (%)
Travel-Related Variable Main Effect Main Effect Main Effect Interaction
Age Period Generation A‘G‘P
(%) (%) (%) (%)
Money well spent 2 3 0 95
Speak same ianguage 6 4 3 83
Having fun, being entertained 3 1 1 95
National/provincial parks 3 6 1 86
Beaches for swim & sun 6 0 0 94
High quality restaurants 4 4 0 92
First class accommodations 17 1 5 74
Budget accommodations 3 1 2 92
Shopping 10 3 0 84
Nightlife and entertainment 5 0 0 95
Museums, art galleries 4 6 3 85
Amusement, theme parks 4 0 0 96

 

*May not total to 100% because 2-way interaction percentages not presented

Using Structural Equations Modeling as a Method of Cohort Analysis

The structural equation modeling technique was used to examine whether there
was a structural model that existed which measured the relationship between age, period
and generation on three travel-related constructs. The EQS program by Bentler (1985)
was used to perform the cohort analysis with structural equations modeling. The initial
step was to see if the 12 variables represented the three constructs outlined by the
Canadian Tourism Commission (Figure 10). As mentioned in the methods section of this
manuscript, the two variables “money spent on travel is well spent” and “it is important

that people speak my language” represented the “travel philosophy” construct. The

130

“beneﬁt-sought” construct was a single item indicator represented by “having fun and
being entertained.” The “travel interest” construct was hypothesized as a reﬂective
variable made up of “national and provincial parks,” “beaches for swimming and
sunning,” “high quality restaurants,” “budget accommodations,” “ﬁrst class
accommodations,” “shopping,” “nightlife and entertainment,” “art galleries and

museums” and “amusement and theme par .”

 

 
     
 

Travel
Philosophy

 

 

Age

Figure 10. The Model

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Conﬁrmatory factor analysis was performed using the statistical software package
EQS. In order for the constructs to be “conﬁrmed” three tests needed to indicate a good
ﬁt. The tests used were the CPI, RMR and the GFI. The results indicated that the “travel
philosophy” construct was a poor construct. The goodness of ﬁt scores were too poor to
conclude that these two indicators were measuring the same thing. Therefore, they
needed to be measured separately and called two separate items. The “beneﬁts sought”
construct was a single item indicator and therefore was not tested. The “travel interest”

construct initially had nine indicators; however, after evaluating the LMTest (indicated

131

cross loadings of indicators on other constructs and large errors) and the t value (which
indicated the signiﬁcance of each indicator), the ﬁnal construct had four indicators. These

’9 ‘6

four indicators were “shopping, beaches for swimming and sunning,” “nightlife and
entertainment” and “national and provincial parks.”

The next step was to measure the relationships among the three exogenous
variables (independent variables) and the three constructs. This is where the initial
problems occurred. The model was hypothesized to measure each of the effects on each
of the constructs and determine which of the effects was the strongest. This is the same as
the regression procedure however, SEM allows an overall “goodness of ﬁt” which
measures all variables at once.

Results revealed that the full model did not ﬁt the data. The goodness of ﬁt
measures were so weak that even manipulating the parameters as indicated by the LM
Test did not yield a good ﬁt. Several problems may have been underlying the lack of ﬁt.
Some of these may include problems due to multicollinearity and problems due to
unstable constructs. Due to the multicollinearity of the three exogenous variables, a clear
model was impossible to derive. Even when age, period and generation were allowed to
correlate, the relationship between the exogenous variables and the endogenous variables
was not direct. A consistent ﬁnding with each of the constructs was that each exogenous
variable wanted to correlate indirectly with the construct through the individual indicator.
This suggested that the constructs were perhaps not as reliable as the CF A suggested.

One suggestion that has arisen to deal with this problem is the creation of

“formative” constructs (c.f., Bollen & Lennox, 1991; Cohen, Cohen, Teresi, Marchi &

Velex, 1990; MacCallum & Browne, 1993). These constructs are measured “without

132

error.” The idea is that the construct is “formed” by the indicators. In this study it was not
appropriate to do this because to form the construct “travel philosophy” by the two
indicators “money is well spent” and “important that people speak my language” would
be illogical. In fact, it is counter intuitive. The construct travel philosophy is NOT simply
the addition of those two indicators, in fact it has several other indicators which have not
been accounted for. Therefore, this solution is an inappropriate resolution to the problem.
An additional solution to the problem is to perform “group” analyses. These
groups would be similar to dummy variables and a model would be run for each dummy
variable. However, this is merely a variation of dummy variable regression and one of the
assumptions of dummy variable regression is that the three effects do not interact. This
assumption is not appropriate given the results of the three-way AN OVA and therefore
the decision NOT to use dummy variable regression was made. Likewise, with structural
equations, this method would not capture the interaction among all three variables.
Therefore, it was concluded that structural equation modeling was not an effective way to

measure cohort analysis in this case.

Study Limitations
Before advancing to the conclusions of this study, there are several limitations
that should be acknowledged. It is important to address these limitations to fully
understand the conclusions and recommendations that have come about as a result of this
study.
First, the study only included two time periods. Although cohort analysis only

needs a minimmn of two time periods, three or more periods result in more robust

133

datasets. One of the problems that arose because having only two time periods was that it
was difﬁcult to conclude a linear relationship over time. Having only two points does not
indicate a linear trend one way or the other. Additional research efforts need to keep this
in mind when performing cohort analysis.

An additional problem that arises by not having enough time periods is that we
can not follow a generation over its life span. In this data, Generation X was never
measured at a later life stage and the GI generation was never measured in their youth.
Thus, it becomes more difficult to partial out the age effect from the generational effects
because older aged Canadians were also older generations of Canadians. Future studies
need to address this when performing a cohort analysis.

Second, another limitation of the study was that the age variable in the original
studies was measured at the ordinal level. Therefore, pre-existing categories made it
difficult to recategorize age into Strauss and Howe’s generations. When recoding the data
into generation, one age category was dropped because it contained two generations. If
age was measured at the interval level, the categorizing both the age and generation
variables for this study would have been more precise. This should be kept in mind for
future research.

Finally, because these studies were not designed to perform a cohort analysis
there were limited numbers of variables that had scalar equivalence. Both datasets were
very large and had over 100 variables each; however, only 12 were worded identically
and measured the same construct (term used loosely after SEM results). More variables
that were measured in the same manner would have perhaps resulted in a better structural

equation model.

134

Chapter V
CONCLUSIONS, DISCUSSION AND IMPLICATIONS

Conclusions and Discussion

Considerable research has relied on cohort analysis as a means to predict future
consumption behaviors. This study sought to extend this type of research by focusing
solely on travel-related philosophies, beneﬁts and interests.

The study set out to assess which effect (age, period and generation) explained the
most variation in travel philosophies, beneﬁts sought and interests of Canadians who
travel. Speciﬁcally, generational effects on travel philosophies, beneﬁts sought and
interests were of importance. However, no conclusive ﬁndings relating a person’s
generation and travel philosophies, beneﬁts sought and interests were made. Essentially,
attributing any changes in travel philosophies, beneﬁts sought or travel interests to
generation effects was less than conclusive.

Several methods were used to examine the effects of generation on travel
philosophies, beneﬁts sought, and interests. The use of these different methods helped to
reinforce the difficulty of isolating generational effects on the 12 travel-related variables.
Using these different methodologies yielded several important. These discoveries are
discussed below.

One of the most important discoveries that this study revealed was that the
combined effects of age, period and generation do not account for much variation in
individual responses to travel philosophies, beneﬁts sought and travel interests. The
amount of variation explained by the combination of age, period and generation ranged

from less than 1% to a high of 11.3%. Therefore, generation accounts for even less

135

 

variation in responses than the combination of the three independent variables. These
ﬁndings suggest that people ‘3 response to travel changes over time.

What does this mean for tourism marketers? Overall because generation does not
account for much variation in response to travel-related variables other variables must be
considered if a researcher is going to assess the role that generation plays in a person’s
travel philosophies, interests and beneﬁts sought. Marketers should not completely
dismiss the idea of generational marketing. However, what needs to be done is consider
generational variables in conjunction with other demographic and psychographic
variables.

Previous literature has suggested that sociodemographic variables such as family
life cycle, income, geographic location, and education can inﬂuence the beneﬁts sought,
philosophies and interests of travelers (c.f., Anderson & Langemeyer, 1982; F odness,
1992; Shoemaker, 1994; Zirnmer, Brayley & Searle, 1995). McPherson (1991) suggests
that “aging and leisure behavior are inﬂuenced by a number of sociodemographic factors
that must be included in studies addressing the question of leisure” (p. 425). He proposed
these factors could include such things as “gender, race, ethnic membership, education,
lifelong and present income and assets, marital status (past and present), and type and
place of residence (rural vs. urban, north vs. south, east vs. midwest vs. west, etc.)”(p.
425)

One variable that should be included in future cohort studies is family life cycle.
F odness (1992) found that as family structure changes (i.e., the family moves through the
life cycle) predictable changes occur in terms of family decision-making and interests.

His study found that in families with children the wife predominantly made the decision

136

 

of what to do on vacation. He also found that “roles are likely to change over time, both
in general and within the speciﬁc stages of the family life cycle (p. 12). In addition,
Lawson (1991) in his study of tourist expenditure and types of vacations across the
family life cycle found that those in the “full nest group” tended to stay with ﬁiends and
relatives more frequently than members of other life cycle stages. He suggested that
constraints imposed by small children were easy to discern in their lack of activities.
Several authors have suggested that family structure changes over time and changes in
travel patterns represent “unique parameters for each generation.” Therefore, it is
recommended that the family life cycle variable be included in future cohort studies.

Related to family life cycle, income might be an additional variable that might
help to explain the amount of variation in responses. One theory of income, dating back
to the 19503, introduces considerations based on family life cycle position. The
hypothesis recognizes predictable life cycle patterns in earnings, asset accumulation and
consumption (Blomqvist, 1981; Mincer, 1974). Poterba (1989) found that randomly
chosen individuals had less than 50% chance of being in the same income quintile in
1971 and 1978, indicating that the annual income distribution is unstable over time.
Searle, Mctavish, and Brayley (1993) suggest tourism spending is affected by income.
These authors suggest that a small change of income at the bottom end of the scale bring
a large proportionate change in tourist spending. Therefore, monitoring changes in
household income over time might help to explain changes in responses to the
importance of speciﬁc travel-related variables.

An additional variable that needs to be considered is gender. Women have been

found to be more interested in educational experiences and have rated the social

I37

 

dimension higher than males. Although there is some support for the theory that women’s
interests and preferences for travel have changed over time (Hawes, 1988; Kinnaird &
Hall, 1994; Pennington & Kcrstetter, 1997; Samuel, 1999) there is little to no research
that has examined changes in interests or preferences for travel among cohorts over time.
However, it would make sense that different generations of women would respond
differently to speciﬁc preferences for travel given that younger generations have
considerably more freedom than older generations had in their formative years.

Finally, research should consider a person’s ethnicity as a factor in explaining
variation in responses to travel philosophies and interests over time. Several studies have
documented observable differences in travel behavior, especially for outdoor activities
among ethnic or racial subgroups in the US population. Ethnicity theory maintains that
“lower levels of participation in outdoor recreation by minorities are primarily explained
by distinct subcultural values about leisure” (Bowker & Leeworthy, 1998, p. 64). Some
theorists (Drucker, 1997; Metz, 1999) believe that the inﬂuence of technology and the
media has lead to more assimilated cultures and less ethnic diversity. This has been called
“cultural pluralism.” Proponents of this theory argue that the world is becoming more
similar and true than changes over time based on ethnicity may be reduced. The opposing
theory is called a cultural mosaic (V illanueva, 1996). This theory argues that individuals
will continue to maintain their cultural identity regardless of inﬂuences such as media
and globalization. Larrner’s (1999) article in Newsweek suggests that in America, “a
country that constantly redeﬁnes itself, the rise of Latinos also raises questions about
race, identity and culture- and whether the United States will ever truly be one nation” (p.

49) or one culture. These conﬂicting views; however, need to be considered in cohort

138

 

 

studies. How does a changing population (i.e., more ethnically diverse or more
homogenous) inﬂuence generational proﬁles? Future research needs to be consider these
issues and examine changes in interests and preferences for travel over time in relation to
a person’s ethnicity.

These are just some of the variables which might have an inﬂuence on changing
philosophies and interests for travel over time. It is necessary for future researchers to be
aware that all of these factors, “alone or in combination, can inﬂuence past, present or E-
future leisure values, preferences and opportunities” (McPherson, 1991, p. 425). Thus, a

combination of sociodemographic variables might provide a better picture of what

 

variables effect changes in travel philosophies, beneﬁts sought and interests. One *'
recommendation is to examine a person’s generation in relation to other demographic

variables in terms of its predictive power. For example, the previously discussed

variables in addition to other variables could be included in a regression procedure which

measures the strength of each variable alone and in combination in determining future

travel philosophies, beneﬁts sought or travel interests.

The next signiﬁcant discovery that came out of this study was the extent of the
interaction of age, period and generation in explaining differences in travel philosophies,
beneﬁts sought, and travel interests. In fact, the percentage accounted for by the
combination of age, period and generation ranged ﬁ'om 74% to 96% of the total percent
variance explained. This is a substantive ﬁnding. Most other cohort studies fail to report
the extent of this interaction. The presence of such a large interaction makes it impossible

to partial out the main effects of age, period or generation. Therefore, conclusions can not

139

be drawn for any one main effect. This is an important methodological issue for
researchers.

Previous studies have acknowledged the problem of multicollinearity or
interaction effects; however, many of the techniques ignore the relationship between the
three variables and proceed to make conclusions anyway. In this particular study,
collinearity was more of a problem because there were only two time periods with a
range of only 12 years. Therefore, age and generation were more closely aligned. In
previous cohort studies, the number of time periods has been greater and therefore, age
and generation have not shown as strong a correlation. However, all cohort studies should
address the issue of collinearity, particularly those which have limited time periods over a
limited time frame. It is important for readers to understand that fewer time periods will
result in higher correlations among the independent variables, which in turn will result in
greater interactions.

In marketing, Rentz & Reynolds (1981) addressed the issue of “confounding.”
They suggested the three independent variables are inherently correlated as a result of the
formal linear dependency between the operations. However, the authors downplay the
confouding and argue that “cohort analysis can provide tentative evidence of the probable
inﬂuence of age, period and cohort membership (p. 65). No where in the article is the
correlation or percentage variance explained presented.

In their next article, Rentz and Reynolds (1983) again mentioned the linear
dependency but proceeded to use a constrained dummy variable regression model to
estimate cohort effects. They suggested that the “constrained multiple regression model is

the most appealing conceptually and statistically” (p. 14). However, again this article

140

 

fails to present the relationship between the confounding variables, or the R2 indicating
the amount of variance explained by the combination of the three variables. Upon
completion of the present study, the researcher feels that ignoring the interaction and the
amount of variation explained by the independent variables can lead to false conclusions.
In both Rentz and Reynolds studies the ﬁndings were used to forecast future behavior.
This author would argue that forecasting future behavior based on 5% of the explained
variation is problematic.

In addition, the presence of such large interaction effects has signiﬁcant
marketing implications. First, due to the presence of interaction effects, it is difﬁcult to
partial out how much change in travel philosophies, beneﬁts sought or interests is a result
of a person’s generation. Therefore, marketers should be cautioned about developing
marketing plans which strictly focus on a person’s generation. The relationship between a
person’s age, generation and time period need to be considered simultaneously. It is not
enough to focus on a person’s generation in isolation. The idea of generations is
important and should not be downplayed. However, measuring speciﬁc characteristics or
preferences of generations is more difﬁcult than the literature suggests. Marketers should
be aware that segmenting based on age is probably just as effective as generational
marketing. In fact, knowing how different ages behave and what is happening at that time
period may be more beneﬁcial than spending lots of money trying to proﬁle different
generation’s preferences and interests.

In addition, tourism marketers should be cautious of trendy literature that suggests
generational marketing is path to success. One book that gained considerable attention in

late 19903 is “Rocking the Ages: The Yankelovich Monitor on Generational Marketing”

141

 

(1997). The main focus of the book was to provide support for generational marketing.
The Yankelovich Monitor used longitudinal data to examine changes in attitudes and
consumption of three different generations over time. The authors illustrated changes in
attitudes and behavior both cross-sectionally and over time. At ﬁrst glance it would
appear that there are several differences among generations which provide a strong basis
for generational marketing. However, one of the main problems is the authors’ failure to
address or even measure the interaction among age, period and generation. Smith and
Clurman are not alone in their procedures, in fact, they do what most cohort analysis do-
they ignore the interaction of the three effects and focus strictly on the main effects (in
this case generation). Unfortunately, ignoring the confounding effects of age, period and
generation may lead to false conclusions. Therefore, without taking the generation’s age
and the time period into account, the changes in attitude and consumption are not as
meaningful. In addition, Smith and Clurman fail to recognize the amount of within
generation variation. They assume that all members of a particular generation will
respond the same way. Smith and Clurman use years of data to support the theory that
different generations will respond in different ways to different products. However, it is
important for marketers to recognize that much of the complexities of cohort analysis are
not addressed in their book. Although the book is based on surveys of the general
population, many of the limitations of their particular data have also not been presented
or addressed in their book. Therefore, tourism marketers should be cautious of adopting
strategies suggested in the popular literature without being fully informed of methods of

data collection and the statistics used for interpretation.

142

A third conclusion that has come out of this study is that the use of analysis of
variance is an effective and appropriate method to explain the effects of age, period and
generation. The two methods of cohort analysis which are most common are the triad
method and dummy variable regression. Although, both methods are useful in
understanding the complexity of the cohort analysis; they are do not provide information
about the interaction among the three independent variables and the interaction’s effect
on the dependent variable. In addition, dummy variable regression typically uses
subgroup means rather than individual scores. This can be misleading in that, generally
group differences account for less than 5% of the overall differences.

Various other models have been proposed which attempt to unravel the effects of
age, period and generation; however, all of these methods are all based on assumptions.
For example, Baltes and Reinert’s (1969) bifactorial model assumes that period effects
are zero. Because most cohort analysis involves change over a ten-year period, this
assumption would seem difﬁcult to support. In addition, Schaie and Strother (1968a)
assume that period effects are positive; however, it was demonstrated in this study that
period effects could be negative rather than positive. Therefore, it is sometimes a great
leap in the social sciences to assume that all period effects are positive. This document
has demonstrated that fatal ﬂaws can be made if the researcher makes the wrong
assumptions.

One of the only models which avoids the statistical problems of confounding
variables is the triad method. Bonnici et a1. (1992) suggested this method is “logically

appealing, simple and offers a methodologically adequate tool for the marketing

143

practitioner” (p. 3). Therefore, it is argued that this method should be used initially in all
cohort analyses.

The present study has demonstrated how the use of three-way analysis of variance
can be used in conjunction with the triad method. The use of the three-way analysis of
variance enabled complex patterns and relationships to be identiﬁed. The statistical
procedure enabled the researcher to illustrate empirically what was happening in the data.

Results of the analysis of variance revealed that when large interaction effects are L
present, isolating the main effect of generation would be inappropriate. Hypothetically, if

only generational effects were present, the three-way analysis of variance would indicate

 

signiﬁcant main effects for generation and no signiﬁcant effects for age, period and any
of the interactions. Therefore, using the three-way analysis of variance illustrates the
main effects and the interaction effects. This provides enough information for the
researcher to be able to draw conclusions about the relationships between age, period,
generation and behavior. Therefore, future cohort analysts should consider using the
combination of three-way analysis of variance and cohort tables.

An additional discovery that came out of the present study was that a structural
equation modeling is a difﬁcult procedure to use to address generational effects.
Although structural equations are an appropriate method to measure the overall “ﬁt” of a
model, several factors cause problems which affect the overall ﬁt of the model. The main
problem with structural equations is that there is too much individual variation in
responses and as a result the constructs were not stable over time. In addition, assessing
the three effects (age, period and generation) simultaneously is not possible given the

large numbers of categorical independent variables. A method available in structural

144

equations modeling which could be used is “group” analysis. However, it was determined
that “group” analysis is essentially the same as dummy variable regression. As with
dummy variable regression the groups’ analysis would make several assumptions that
may not be valid. Thus, it was felt that the use of structural equations did not provide an
effective means of cohort analysis in this study.

Finally, cohort analysis is much more complex than what researchers might think.
It is not simply assessing how a generation’s preference or behavior for a product or
service changes over two or more times. It includes pulling apart the intertwined
variables of age, period and generation, determining how much interaction exists between
the variables, ﬁnding measures at multiple points in time and ﬁnding a statistical
technique which best illustrates what is going on over time. This complexity effects the
practical nature of generational segmentation. Thorough examination of the segmentation
literature supports this complexity. Segmentation is caught between two extreme worlds,
one which argues we can segment an entire hemisphere into “nine nations” (Kahle, 1986)
and another which argues that each individual is it’s own market and that we can market
to a segment of one (Zieger, 1999). The “nine nations” argument proposes nine regions in
North America which can be uses as a basis for market segmentation under the
assumption that values and cultural inﬂuences on consumption differ across geographical
regions. The “individual market” argument suggests that each individual is it’s own
market and that we need to understand the individual to maintain and increase our
business. Undoubtedly, these two theories come at understanding markets from opposite

perspectives. Generational marketing as a whole is more similar to the “nine nations”

145

argument (geographical regions); however, the ﬁndings in this study have provided more

endorsement for the “individual market” argument.

Recommendations for Future Research

This study has demonstrated that there are several issues that future researchers
need to keep in mind if they are going to perform an effective cohort analysis. First,
although several cohort methodologists (c.f., Achlety, 1989; Glenn, 1977; McPherson,
1990) have supported the use of repeated cross-sectional design, there are several
limitations with this method. One of the main limitations of this method is ﬁnding
variables that are “scalar” equivalent. Typically, researchers design surveys in isolation
ﬁ'om previous studies. Thus, it is rare that variables are worded the same over time. If
marketers are to learn more about generations, researchers should attempt to keep the
wording of variables consistent among studies. A strong recommendation for the
Canadian Tourism Commission, Travel Industry Association of America, state travel
offices and other large tourism research organizations is to develop a set of “core”
questions which would not change over time. It would be valuable for the leaders in
public sector tourism to commission partners, who collect national or state-level tourism,
to include this small set of “core” questions on all surveys over a set number of years.
This would allow for more effective longitudinal studies and the examination of cohorts
over time.

In addition to the limited variables with scalar equivalence, repeated cross-section
designs play into individual variations. As demonstrated in this study, analysis of

variance isolates individual variations in relation to group variations. The use of repeated

146

 

 

cross section designs only inﬂate the amount of individual variation in responses. This
leads into the next recommendation.

It is recommended that tourism researchers strive to collect longitudinal panel
data. A representative national panel surveyed every 10 years for at least 50 years would
enable researchers to more readily determine which of the three effects (age, period and
generation) explains the most variation in travel philosophies and interests. Typically a
cohort study looks at changes in attitudes or behavior in 5-10 year increments over a 20-
30 year period. According to LaPage (1994), “sometimes it is desirable to periodically
recontact the sample of respondents” (p. 482). He argues “panels are superior for
providing in-depth answers to questions of why market shifts and trends are taking place”
(p. 482). In addition to providing in-depth answers, a longitudinal panel design enables
tracking of changes in attitudes and preferences of multiple cohorts over time. Following
several cohorts over their life span might help the researcher to isolate the effects of age,
period and generation. True differences in cohorts are better revealed when the researcher
can examine differences in several cohorts at different times in their life. For example,
measuring consumption and preference patterns for the GI Generation at a several
different ages, then for the Silent Generation at the same ages and ﬁnally for the Baby
Boomers might revealed more meaningful differences.

Panels are an extremely powerful means for understanding social change, for
assessing the impact of unforeseen events and for short-term forecasting. In fact, Lapage
(1994) argues that “the magnitude of change in people’s leisure-behavior patterns, and
their receptivity to new leisure opportunities, appear to be far greater than would be true

for household purchases” (p. 486) and therefore, panel research is a distinctly appropriate

147

 

 

method of travel and tourism research. Future cohort analysts are urged to consider both
collecting and obtaining panel data to assess generational effects on different aspects of
travel and tourism.

An additional recommendation is that researchers who wish to examine cohort
effects should design and use variables that are more “generational” in nature.
Researchers need to think creatively about questions that may appeal to one generation
over another. Strauss and Howe (1991) proﬁle generations based on birth year and a set
of epochal events which deﬁne the generation. Consideration should be given to the
events and the possible “nostalgia” that may surround it. For example, if the Silent
Generation is touted as “ﬁ'ugal” and the Baby Boomers as “excessive” then inclusion of
variables which measure these attitudes might be included in future travel studies. Future
researchers should attempt to design speciﬁc measures that a speciﬁc generation may
relate to. Proﬁling a generation based on these variables may provide a clearer idea of the
effects of a person’s generation on speciﬁc philosophies or preferences.

Fourth, researchers who perform longitudinal studies should consider including
variables which may indicate environmental changes. This is perhaps the most difﬁcult
variable to measure when examining changes over time. In order to truly account for
period effects, surveys need to include measures which focus on environmental changes,
such variables might include changes in the use of technology in tourism or changes in
marketing. Because it is important to measure this change over time, it is necessary that
the variables be measured in each time period and in the same manner. Additional
measures may be added to future studies; however, continuing to ask the same questions

using the same environmental variables in subsequent surveys is essential.

148

 

 

Fifth, perhaps cohort analysis travel and tourism needs to consider “travel”
epochal events rather than world epochal events to deﬁne the generations. What might
be interesting is deﬁning speciﬁc generations based on epochal “travel” events, such as
the rise in airline travel or growth in “cybertravel.” Deﬁning a generation based on an
event in travel that may have shaped their travel philosophies or interests might reveal
greater differences between the generations. Future cohort analysis should consider
developing proﬁles of different generations which are based on some of the larger travel [1
events. These generations may be named “the airline generation,” “the automobile

generation,” or “the computer assisted travel generation.” It is argued that generations

 

would differ based on differential exposure to different types of travel. J
In all, the results of this study have led to a more comprehensive understanding of

cohort analysis as it relates to the travel philosophies, beneﬁts sought and travel interests

of Canadians. The study has been especially useful in isolating several issues that require

attention when conducting a cohort analysis.

149

 

APPENDICES

150

 

 

 

 

APPENDIX A

Tourism Canada

Canadian Tourism Attitude and
Motivation Study

Segmentation Report

 

151

 

 

 

P1

 

 

 

CANADIAN TOURISM ATTITUDE AND
MOTIVATION STUDY

RESEARCH REPORT
SEGMENTATION ANALYSIS

"

TOURISM CANADA

Published under the authority of the Department of
Regional Industrial Expansion

August 1986

 

152

 

 

 

 

 

 

This
and 1
C an;
CTA
cont

Exp

The

datr

Car

att.

Va

be

 

 

Research Backgmund

This report presents the results of a segmentation analysis of the vacation
and pleasure travel market in Canada based on data provided by the
Canadian Tourism Attitude and Motivation Survey (CTAMS). The
CTAMS data base is the result of a national survey of Canadians
conducted in the fall of 1983 by the Department of Regional Industrial

Expansion (Tourism) through the Labour Force Survey.

The purpose of this analysis is to develop a framework for the CTAMS
data base that provides an understanding of the Canadian travel market and
that can be used as input to marketing development decisions in the
Canadian tourism industry. The framework is based on the deﬁnition and
analysis of population sub-groups which are derived from Canadian’s
attitudes with respect to taking vacations, planning and arranging for
vacations, their vacation needs and interests, and their actual vacation

behaviours.

 

 

153

 

 

 

 

 

The framework will provide insight to the Canadian vacation travel market

in terms of:

o What various segments or groups of the Canadian pOpulation want in a
vacation

0 What products or environments are necessary to deliver these wants

0 How to Canadian travellers perceive the alternate vacation destinations
available to them based on the destinations they have chosen to meet
the attributes desired by them

0 Whether there is a potential to alter the perceptions of various vacation
products/environments available in Canada in order to inﬂuence the

choice of vacation destinations by more Canadians.

The analysis was done by Burak Jacobson Research Partners Inc. on behalf

of the Department of Regional Industrial Expansion (Tourism).

 

154

 

 

 

 

 

Research Obiectives

The speciﬁc objectives of this analysis are:

1.

To identify the factors which inﬂuence the travel decisions of
Canadians

To develop an information base on current attitudes and motivations to
leisure travel among Canadians

To identify market groups, including any special markets, that offer
potential for increased travel within Canada by Canadians

To develop priorities for market development based on any market
groups for which the potential for increased travel in Canada can be
identiﬁed

To identify elements of the Canadian tourist product that could attract
more Canadian visitors if expanded, improved or better promoted

To provide a basis for positioning Canada and the provinces as travel
destinations relative to foreign destinations and, where feasible, to do

the same for major sub-provincial regions as well

 

155

 

 

 

In addition, tabulations were run across the derived groups to test the
following speciﬁc hypotheses:
a) The groups derived in this analysis do not differ on the basis of their
place of residence:
0 By region or province
0 By urban versus rural locations
0 By metro versus non-metro locations
b) The groups derived do not differ in terms of the:
0 Characteristics of trips actually taken and recorded in the
CTAMS questionnaire including destinations
o The duration of trips taken e.g., short trips (1 to 3 nights)
long trips (4 or more nights)
c) The groups derived do not differ on the basis of the demographic
characteristics of respondents belonging to each group.
d) The groups derived do not differ in terms of their frequency of travel

and the proportion of travellers versus non-travellers.

 

156

 

 

 

The Data Base

The CTAMS data base is the result of a national survey of Canadians
conducted by DRIE (Tourism) in the fall of 1983 and consists of 11,500

completed interviews.

The CTAMS data base covers ﬁve categories of travel information about
Canadians. Three categories comprise attitudinal information:
1. Attitudes concerning vacations in general, and the planning and
arrangements involved
2. Beneﬁts important in choosing a vacation
3. Activities and Interests desired in a vacation
Two categories provide behavioral information:
1. Actual travel behaviour data
2. Description of a speciﬁc vacation or pleasure trip taken in the
previous 12 months, in terms of the beneﬁts sought and the
activities and interests important in attracting them to that

choice

 

157

 

 

 

The research vehicle used to collect the CTAMS data was the Statistics
Canada Labour Force Survey. This survey consists of six “rotation” groups
each consisting of a national sample of just over nine thousand respondents
so that the total sample approaches 60,000 person. Each rotation group
remains in the sample for six months and the groups are staggered over the

period.

The CTAMS survey method was to give a supplementary questionnaire to

selected respondents in the Labour Force survey.

For the CTAMS research one rotation group provided the sample for
Ontario, Quebec, Alberta, and British Colmnbia. Two rotations groups
were needed to obtain a sufﬁcient sample for Nova Scotia, New
Brunswick, Saskatchewan, and Manitoba; three groups for Newfoundland;
and four rotations for Prince Edward Island sample. The territories were

not included.

 

158

 

 

 

The Sample

The sample was designed with a disproportionate distribution across the
provinces so that the results would have similar reliability for each

province. The distribution of attempted interviews was as follows:

Newfoundland 1,500 interviews
Prince Edward Island 948

Nova Scotia 1,346

New Brunswick 1,430

Quebec 1,437

Ontario 1,808

Manitoba 1,602
Saskatchewan 1,896

Alberta 1,158

British Columbia 1,055

Just over 80% of respondents completed the questionnaire for a total of

1 1,500 interviews.

 

 

159

 

 

 

 

APPENDIX B

Canadian Tourism Commission

Domestic Tourism Market
Research Study

Main Report

 

160

 

 

 

DOMESTIC TOURISM MARKET
RESEARCH STUDY

MAIN REPORT

CANADIAN TOURISM COMMISSION

Canadian Cataloguing in Publication Data

1 99'6

 

161

 

 

 

 

Introduction: Study Backggound and Objectives

The Domestic Tourism Market Research study is the ﬁrst major initiative
since the 19803 to examine in detail the motivations, attitudes, and
perceptions of Canadians about tourism, travel opportunities and
destinations. The overall purpose of the study is to determine how best to
market Canadian destinations and products to Canadians and examine

possibilities for import substitution.

This study is particularly timely, given the new federal support for
domestic tourism marketing with the recent creation of the Canadian
Tourism Commission (CTC). A good understanding of the marketplace is
obviously necessary to provide a sound basis for future market planning

and marketing efforts by the CTC and industry alike.

In addition, import substitution is a growing issue as a result of the rapid
rise of Canada’s international travel account deﬁcit over the last decade.
Curbing the ﬂow of Canadians who vacation outside the country and

convincing them to travel domestically is clearly one way to reduce this

deﬁcit.

 

I62

 

 

 

According to Statistics Canada, there was a 60% increase in the number of

overnight trips made by Canadians outside Canada between 1984 and

1993. In contrast, overnight domestic travel rose by only 26% in the same

time period. This suggests that the industry should look seriously at ways

in which to market its products and attractions to outbound travellers who

may be amenable to travelling within Canada instead.

The speciﬁc objectives of this study are as follows:

To determine the incidence, market size and characteristics of
Canadians who take overnight pleasure trips to intra-provincial
destinations, inter-provincial destinations, the US, and other
countries.

To assess Canadian’s awareness of and attitudes toward Canada
overall, major regions in Canada and the US. as pleasure travel
destinations.

To determine the impressions of product offerings associated with
different trip types in Canada and the US.

To evaluate the pleasure destination decision-making process of
Canadians.

To determine the perceived advantages and barriers to pleasure
travel in Canada and the US.

To segment the Canadian overnight pleasure travel market in terms
of travel motivations and travel philosophy

 

 

163

 

 

 

 

Methodology

Two different data collection methodologies were used in this study:

' A telephone survey of 1,899 Canadians, aged 15 or older, was
undertaken in September, 1995. About 91% of these interviews
(1.725) were conducted on a random basis, while the remaining
respondents were selected to meet regional quotas.

' A more in-depth survey consisting of in-person interviews with 1,457
Canadians was conducted in October 1995.

Combined the two surveys yielded interviews with 3,3 56 Canadians
including 2,694 travellers and 662 non-travellers. The results for the 2,694
travellers are accurate 19 times out of 20 to within plus or minus 1.9%.
Some questions were asked only in the in-person survey (n=1,233); for

these questions, the margin of error rises to about 2.8%.

For the purpose of this study, a pleasure trip was deﬁned to be any trip of
at least one night away from home, taken entirely or in part for vacation,
recreation or holiday purposes, including out-of-town visits to family or

ﬁiends or trips to a cottage.

 

164

 

 

 

 

Organization of the report

The remainder of this report is organized as follows:

Chapter 2 presents an overview of the domestic travel market including
market size, incidence rates, trip characteristics, trip decision-making,
travel motivations and attitudes and future travel intentions.

Chapter 3 focuses on the issue of brand loyalty and looks at the
potential for import substitution in the domestic market

Chapter 4 focuses on destination marketing and looks at the destination
preferences of Canadian travelers, as well as their perceptions of
Canada and the US.

Chapter 5 summarizes the philosophy and motivational segments of the
domestic market and explores speciﬁc opportunities for import
substitution among these segments

Chapter 6 concludes the report

Other products resulting from this study include two special reports, one

on off and Shoulder season marketing and the other on regional marketing

opportunities. A set of computer volumes containing the detailed survey

data are also available. These Show the results of all questions in the

questionnaire broken down by major categories.

 

165

 

 

APPENDIX C

THREE-WAY ANALYSIS OF VARIANCE FOR TRAVEL PHILOSOPHY, BENEFIT
SOUGHT AND TRAVEL INTEREST VARIABLES

166

 

E...

THREE-WAY ANALYSIS OF VARIANCE FOR REMAINING TRAVEL
PHILOSOPHY, BENEFIT SOUGHT AND TRAVEL INTEREST VARIABLES

Three Way ANOVA Examining Total Interaction Of Age, Generation And Period
On Money Spent On Travel Is Well Spent

 

 

Independent Variables Sum of ' df Mean Square F Sig.
and Interaction Squares

Age .177 6 .003 .849 .532
Period 0 .330 1 .330 9.519 .002
Generation .007 3 .002 .064 .979
Age * Period .008 3 .003 .076 .973
Age * Generation .008 l .008 .234 .629
Period * Generation .009 1 .009 .266 .606
Age * Period * Generation 10.59 18 .589 16.956 .000
Error 486.1 18 14006 .003

 

E1: .021 (Adjusted R2=.020)

Three-Way AN OVA Examining Total Interaction Of Age, Generation And
Period On People I Meet Speak Same Language

 

 

Independent Variables Sum of df Mean Square F Sig.
and Interaction Squares

Age 1.47 6 .250 4.509 .000
Period .963 l .963 1 7.400 .000
Generation .608 3 .203 3.664 .012
Age * Period .676 3 .225 4.074 .007
Age * Generation .005 1 .005 .943 .331
Period "' Generation .186 1 .186 3.360 .067
Age "' Period * Generation 19.56 18 1.09 19.77 .000
Error 774. 1 1 9 l 3990

 

W: .025 (Adjusted RT=.024)

167

Three Way AN OVA Examining Total Interaction Of Age, Generation And Period
On Importance Of National And Provincial Parks

 

 

Independent Variables Sum of df Mean Square F Sig.
and Interaction Squares

Age .180 6 .030 .483 .821
Period .321 1 .321 5.167 .023
Generation .062 3 .03 l .499 .607
Age * Period .066 3 .033 .536 .585
Age * Generation .001 l .015 .025 .876
Period " Generation .005 1 .053 .086 .769
Age * Period " Generation 4.73 17 .278 4.48 .000
Error 737.38 1 1884 .062

 

Rf: .006 (Adjusted R2=.005)

Three Way AN OVA Examining Total Interaction Of Age, Generation And
Period On Importance Of High Quality Restaurants

 

 

Independent Variables Sum of df Mean Square F Sig.
and Interaction Squares

Age .337 6 .056 .866 .519
Period .350 1 .350 5.39 020‘
Generation .021 3 .01 1 . 166 .847
Age * Period .016 3 .008 .122 .885
Age "' Generation .017 1 .017 .267 .605
Period * Generation .015 l .015 .234 .629
Age * Period "' Generation 8.19 17 .482 7.42 .000
Error 772.07 11890 .065

 

R7: .010 (Adjusted R2=.009)

Three Way AN OVA Examining Total Interaction Of Age, Generation And
Period On Importance Of First Class Accommodations

 

 

Independent Variables Sum of df Mean Square F Sig.
and Interaction Squares

Age .408 6 .067 1.18 .313
Period .035 l .035 .616 .433
Generation . 129 3 .064 1 . 1 2 .327
Age * Period .026 3 .013 .229 .795
Age * Generation .019 I .019 .342 .559
Period "' Generation .008 1 .008 .140 .708
Age * Period * Generation 1.76 17 .103 1.80 .023
Error 683.38 1 1890 .058

 

R‘= .003 (Adjusted R2=.001)

168

Three Way AN OVA Examining Total Interaction Of Age, Generation And
Period 0n Importance Of Budget Accommodations

 

 

Independent Variables Sum of df Mean Square F Sig.
and Interaction Squares

Age .317 6 .053 .876 .511
Period .082 1 .082 1.36 .244
Generation .204 3 .102 1 .69 . 1 84
Age "' Period .097 3 .049 .807 .446
Age "' Generation .076 1 .076 1.26 .261
Period “ Generation .062 1 .062 1.04 .309
Age " Period " Generation 9.25 17 .544 9.03 .000
Error 715.81 11884 .060

 

RI: .013(Adjusted R2=.011)

Three Way AN OVA Examining Total Interaction Of Age, Generation And
Period On Importance Of Shopping

 

 

 

Independent Variables Sum of df Mean Square F Sig.
and Interaction Squares

Age .900 6 .150 2.35 .029
Period .293 1 .293 4.59 .032
Generation .031 3 .015 .240 .786
Age * Period .093 3 .047 .733 .481
Age "' Generation .045 1 .045 .701 .402
Period * Generation .017 1 .017 .271 .603
Age * Period * Generation 7.48 17 .440 6.90 .000
Error 759.94 1 191 I .064

 

if: .010 (Adjusted 112-003)

Three Way AN OVA Examining Total Interaction Of Age, Generation And
Period On Importance Of Museums And Art Galleries

 

 

Independent Variables Sum of df Mean Square F Sig.
and Interaction Squares

Age .321 6 .053 .847 .533
Period .435 1 .435 6.88 .009
Generation .257 3 .128 2.03 .131
Age * Period .062 3 .031 .490 .613
Age * Generation .032 l .032 .507 .476
Period * Generation .032 1 .032 .505 .477
Age "‘ Period * Generation 6.37 17 .375 5.93 .000
Error 752.92 1191 1 .063

 

Rf: .003 (Adjusted R2=.007)

I69

Three Way AN OVA Examining Total Interaction Of Age, Generation And
Period On Importance Of Amusement And Theme Parks

 

 

Independent Variables Sum of df Mean Square F Sig.
and Interaction Squares

Age .73 6 .123 2.16 .043
Period .041 1 .041 .697 .404
Generation .021 3 .010 .177 .838
Age * Period .004 3 .002 .036 .965
Age "' Generation .006 1 .006 .104 .747
Period * Generation .020 1 .002 .343 .558
Age * Period "‘ Generation 20.88 17 1.23 20.82 .000
Error 702.16 11902 .059

 

R2: .029 (Adjusted R2=.027)

 

170

BIBLIOGRAPHY

171

Bibliography

Anderson, B. B. & L. Langmeyer (1984). The Under 50 and Over 50 Travelers: A Proﬁle
of Similarities and Differences. Joumgl of Travel Research. 20 (1), 20-24.

Atchley, R. C. (1988). Social Forces and Aging (5th ed.). Belmont, CA: Wadsworth
Publishing Company.

Atchley, R. C. (1989). A Continuity Theory of Normal Aging. The Gerontologist, 29,
1 83-1 90.

Baltes, P. B.; Cornelius, S.W. & J. R. Nesselroade (1980). Cohort Effects in
Developmental Psychology. In J. R. Nesselroade and PB. Baltes (Eds.)

Longitudinal Research in the Study of Behavior and Development (pp. 61-87).
New York, NY: Academic Press.

Baltes, P. B. (1968). Longitudinal and Cross-Sectional Sequences in the Study of Age
and Generation Effects. Human Development, 11, 145-171.

Baltes, P. B. & G. Reinert (1969). Cohort Effects in Cognitive Development of Children
as Revealed by Cross-sectional Sequences. Developmental Psychology, 3 (1),
169-177.

Bello, D. C. & M. J. Etzel (1985). The Role of Novelty in the Pleasure Vacation. Annals
of Tourism Research 6, 408-24.

Bentler, P. M. (1985). Theog and Implementation of EQS: A Structural Equations
Program. Los Angeles: BMDP Statistical Software.

 

Blalock, H. M., Jr. (1966). The Identiﬁcation Problem and Theory Building: The Case of
Status Inconsistency. American Sociology Review, 25 (2), 52-61.

Blalock, H. M., Jr. (1972). Social Statistics (2"d Ed.) New York: McGraw Hill, Inc.

Blazey, M. A. (1987). The Differences Between Participants and Nonparticipants In A
Senior Travel Program. Journgl of Travel Research, 25 (4), 7-12.

 

Blomqvist, N. S. (1981). A Comparison of Distributions of Annual and Lifetime Income:
Sweden around 1979. Review of Income and Wealth, 27 (1), 243-264.

Bowker, J. M. & V. R. Leeworthy (1998). Accounting for Ethnicity in Recreation
Demand: A Flexible Count Data Approach. Joumgl of LeisLure Research, 30 (1),
64-78.

Bollen, K. A. & R Lennox (1991). Conventional Wisdom on Measurement: A Structural
Equation Perspective. Psychological Bulletin, 110, 305-314.

172

Bollen, K. A. (1989). Structural Euations with ngqnt Variables. New York: John Wiley
and Sons, Inc.

Bonnici, J. L. (1992). Cohort Analysis: Analyzing Data on Product Use. Journal of
Applied Business Research, 8 (4), 1-6.

Brim, O. G., Jr. (1980). Types of Life Events. Journ_al of Sociﬁsues. 36, 148-157.

Bryant, B. & A. Morrison (1980). Travel Market Segmentation and the Implementation
of Market Strategies. Journal of Tryavel Research, 18 (3), 2-8.

 

 

Burch, W. R., Jr. (1969). The Social Circles of Leisure: Competing Explanations. Journal
of Leisure Researc_h, 1 (2), 121-147.

 

Buse, R. C. & N. Enosh (1977). Youth Experience: Effect on Participation in
Recreational Activities. Land Economics, 53, 468-486.

Calantone, R. J. & J. S. Johar (1984). Seasonal Segmentation of the Tourism Market
Using Beneﬁt Segmentation Framework. Journal of Travel Resea_pc_:h, 23 (2), 14-
24.

Caltabiano, M. J. (1995). Main and Stress-Moderating Health Beneﬁts of Leisure. Loisir
et Societe/ Socieg and Leisure, 18, 33-52.

Christensen, J. E. & D. R. Yoesting (1973). Social and Attitudinal Variants In High and
Low Use of Outdoor Recreational Facilities. Journal of Leisure Research, 5 (1), 6-
15.

 

Cohen, E. (1972). Toward a Sociology of International Tourism. Social Research. 39,
164-82.

Cohen, R; Cohen, J .; Teresi, J.; Marchi, M, & C. N. Velez (1990). Problems in the
Measurement of Latent Variables in Structural Equations Causal Models. Applied

Psychological Measurement, 14, 183-196.

Coopers and Lybrand Consulting (1996). Domestic Tourism Market Research Study:
Final Rewrt. Report for Canadian Tourism Commission.

Cosenza, R. M. & D. L. Davis (1981). Family Vacation Decision Making over the Family
Life Cycle: A Decision and Inﬂuence Structure Analysis. Journal of Travel
Research, 20 (3), 17-23.

Crask, M. R. (1981). Segmenting the Vacation Market: Identifying the Vacation

Preferences, Demographics, And Magazine Readership of Each Group. Journal of
Travel Research, 20 (3), 29-34.

173

Crawford, D. W.; Godbey, G. & A. C. Crouter (1986). The Stability of Leisure
Preferences. Joumgl of Leisure Resegh, 18 (1), 96-115.

Crompton, J. L. (1979). Motivations for Pleasure Vacation Annals of Touris_m Research.
6 (4), 408-424.

Cutler, R. C. (1993). Older Women's Leisure Activity and Quality of Life. In J. R. Kelly
(Eds), Activig and Aging: Staying Involved in Lgter Life (pp. 86-98). Newbury
Park, CA: Sage Publications.

Dunlap, R. E. & K. D. Van Liere (1978). The New Environmental Paradigm. Journal of
Environmeng Education. 9(4), 10-19.

Elmslie, M. (1997). Beyond America: Canada. [Online] Available at
www.fourthturning.com/cgi.local/netform/beyondamerica/a.cgi/3---3, April, 1, 1999.

Epstein, S. & L. Teraspulsky (1986). Perception of Cross Situational Consistency.
Journgl of Personplitv and Socig Psychology, 50(6), 1152-1163.

Erickson, E. H. (1963). Childhood and Society. New York: Norton.

Etzel, M. J. & R.G. Wahlers (1985). The Use of Requested Promotional Material by
Pleasure Travelers. Journal of Travel Research, 23 (4), 2-7.

Etzel, M. J. & A. G. Woodside (1982). Segmenting Vacation Markets: The Case of the
Distant and Near Home Travelers. Journal of Tgvel Research, 22 (1), 10-14.

Evans, J. R. & B. Berman (1988). Marketing. N.Y.: McMillan.

Fiske, D. W. & S. R. Maddi (1961). Functions of Varied Experience. Homewood, IL:
Dorsey.

F odness, D. (1992). The Impact of Family Life Cycle on the Vacation Decision Making
Process. Journal of Travel Research, 31 (2), 8-14.

Foot, D. K. & D. Stoffrnan (1996). Boom Bust and Echo: How to Proﬁt from the
Coming Demogpaphic Shift. Toronto, Canada: Macfarlane Walter & Ross.

 

Freysinger, V. J. & R. 0. Ray (1994). The Activity Involvement of Women and Men in
Young and Middle Adulthood: A Panel Study. Liisure Scignces. 16, 193-217.

Gitelson, R. J. & D. L. Kcrstetter (1990). The Relationship between Sociodemographic

Variables, Beneﬁts Sought and Subsequent Vacation Behavior: A Case Study.
Journal of Travel Research, 26:24-29.

174

Gladwell, N. J. (1990). A Psychographic and Sociodemographic Analysis of State Park
Inn Users. Journal of Travel Research, 28 (4), 15-21.

Glenn, N. D. (1977). Cohort Anplvsis. Series: Quantitative Applications in the Social
Sciences, Sage University Paper: Newbury Park.

 

Gordon, C.; Gaitz, C. M. & J. Scott (1976). Leisure and Lives: Personal Expressively
Across the Life Span. In R. Binstock and E. Shanas (Eds). Handbook of Aging
and the Social Sciences (pp. 310-341). New York: Van Nostrand Reinhold Co.

Grifﬁth, D. A. & P. J. Albanese (1996). An Examination Of Plog’s Pyschographic Travel
Model Within A Student Population. Journal of Travel Research, 34 (4), 47-52.

Haley, R. (1968). Beneﬁt Segmentation: A Decision Oriented Research Tool. Journal of
Marketing, 23 (6), 30-35.

Havighurst, R. J. (1972). Development Tasks and Education. New York: Longman,
Green.

Havighurst, R. J. (1973). Social Roles, Work, Leisure and Education. In C. Eisdorfer &
M. P. Lawton (Eds). The Psychologv of Adult Development and Aging.
Washington, DC: APA Press.

Hawes, D. K. (1977). Psychographics are Meaningful... Not Merely Interesting. Journal
of Travel Research, 15 (1), 1-7.

Hirschman, E. C. (1984). Leisure Motives and Sex Roles. Journal of Leisure Research,
16, 209-223.

Iso-Ahola, S. E. (1980). Toward A Dialectic Social Psychology of Leisure and
Recreation. In S. E. lso-Ahola (Ed.), Social Psychological Perspectives on Leisure
and Recreation (pp. 19-3 7). Springﬁeld, IL: Charles C. Thomas.

Iso-Ahola, S. E.; Jackson, E. & E. Dunn (1994). Starting, Ceasing and Replacing Leisure
Activities over the Lifespan. Journal of Leisure Research, 26(3), 227-235.

Jackson, E. L., & E. Dunn (1991). Is Constrained Leisure An Internally Homogeneous
Concept? Leisure Sciences, 13, 161-84.

Javalgi, R. G.; E. G. Thomas, & S. R. Rao (1992). Consumer Behavior in the US
Pleasure Travel Marketplace: An Analysis of Senior and Nonsenior Travelers.
Journal of Travel Research, 31 (2), 14-20.

Johnson, E. M.; Scheuing, E. E. & K. A. Gaida (1986). Service Buyer Beh_2_1vior.
Proﬁtable Service Marketing. Homewool: Dow Jones-Irwin.

175

 

Joreskog, K. G. & D. Sorbom (1988). LISREL VI: Analysis of Linear Structural
Relationships by Maximum Lil_(elihood and Legst Squales Method. Mooresville,
IN: Scientiﬁc Software, Inc.

Kahle, L. R. (1986). The Nine Nations of North America and the Value Basis of
Geographic Segmentation. Journ_al of Marketing. 50 (2), 37-48.

Kelly, J. R. (1974). Socialization toward Leisure: A Developmental Approach. Journal of
Leisure Research, 6(2), 181-193.

Kelly, J. R. (1977). Leisure Socialization: Replication and Extension. Journ_al of Leisure
Research, 9 (2), 121-132.

Kelly, J. R. (1987). Freedom to be: A New Sociology of Leisure. New York: Macmillan.

Kinnaird, V. & D. Hall (1994). Tourism: A Gender Analysis. New York: John Wiley and
Sons, Inc.

Kleiber, D. A. (1985). Motivational Reorientation in Adulthood and the Resource of
Leisure. In D. Kleiber & M. Maehr (Eds), Motivation and Adulthood. (pp. 125-
150). Greenwich, CT: JAI Press.

Kotler, P. (1967). Marketing Management: Analysis, Planning, and Control. Englewood
Cliffs, NJ: Prentice Hall.

LaPage, W. F. (1994). Using Panels for Travel and Tourism Research. In Ritchie, J .R, &
C. R. Goeldner (2nd Eds). Travel, Tourism and Hospitality Research: A Handbook
for Mmgers and Researchers (pp. 481-491). Toronto, ON: John Wiley and Sons,
Inc.

LaPage, W. F. & D. P. Ragain (1974). Family Camping Trends: An Eight-Year Panel
Study. Journal of Leisure Research, 6(2), 101-112.

Larmer, B. (1999). Latino America. Newsweek, 134 (2), 48-51.

Lawson, R. (1991) Patterns of Tourist Expenditure and Types of Vacation across the
Family Life Cycle. Journal of Travel Research, 29 (4), 12-19.

Loker, L. E. & R. R. Perdue (1992). A Beneﬁt-Based Segmentation of a Non-Resident
Summer Travel Market. Journ_ﬂ of Travel Resemh, 29 (1), 12-18.

Lounsbury, J. W. & L. L. Hoopes (1988). Five-Year Stability of Leisure Activity and
Motivation Factors. Journal of Leisure Research, 20 (1), 31-45.

MacCallum, R.C. & M. W. Browne (1993). The Use of Causal Indicators in Covariance
Structure Models: Some Practical Issues. Psychological Bulletin, 114, 533-541.

176

 

Maclntosh, R. W.; Goeldner, C. R. & J. R. B. Ritchie (1995). Tourism, Principles,
Practices and Philosophies. (Ed. 7). New York: John Wiley and Sons, Inc.

Madrigal, R. & L. R. Kahle (1994). Predicting Vacation Activity Preferences on the Basis
of Value-System Segmentation. Journal of Travel Research, 32 (3), 22-29.

Mannell, R. C. & D. A. Kleiber (1997). A Social Psychology of Leisure. State College,
PA: Venture Publishing.

Mason, P. (1996). The Family Leisure Travel Outlook. In C. R. Goeldner (Ed.) 1996
Travel Outlook.

Mayo, E. F. (1975). Tourism and the National Parks: A Psychographic and Attitudinal
Study. Journal of Leisure Research, 14 (1), 14-18.

Mazanec, J. A. (1984). How to Detect Travel Market Segments: A Clustering Approach.
Journpl of Trgvel Research, 23 (1), 17-21.

McGuire, F .; O'Leary, J .; Yeh, C-K. & F. Dottavio (1989). Integrating Ceasing
Participation with other Aspects of Leisure Behavior: A Replication and
Extension. Journal of Leisure Research, 21 (4), 316-326.

McPherson, B. D. (1983). Aging as a Social Process. Toronto, ON: Butterworth.

McPherson, B. D. (1991). Aging and Leisure Beneﬁts: A Life Cycle Perspective. In
Driver, B.L., Brown, P. J. & G.L. Peterson (Eds). Beneﬁts of Leisure (pp. 423-
429). State College, PA: Venture Publishing Inc.

McQuire, R.; Dottavio, F. D. & J. O’Leary (1987). The Relationship of Early Life
Experiences to Later Life Leisure Involvement. Leisure Sciences 9 (2), 251-257.

 

Mehrabian, A. & J. A. Russell (1973). A Measure of Arousal Seeking Tendency.
Environmental Behavior, 5, 315-333.

Metz, J. B. (1999). In the Pluralism of Religious and Cultural Worlds. Cross Currents, 49
(2), 227-236.

Michman, R. D. (1991). Lifesgle Market Segmentation. N.Y.: Praeger Publishers.

Mieczkowski, Z. (1990). World Trends in Tourism and Recreation. N.Y.: Peter Lang
Publishing, Inc.

Miller, S. J. (1965). The Social Dilemma of the Aging Leisure Participant. In A. M. Rose

and W. Peterson (Eds.) Older People and Their SociyaJ Wong (pp. 77-92).
Philadelphia, PA: F. A. Davis Company.

177

 

Mincer, J. (1974). Schooling, Exmrience and Earnings. New York: Columbia University
Press.

Murphy, P. E. & W. A. Staples (1979). A Modernized Family Life Cycle. Journal of
Consumer Research, 6(1), 12-22.

Naisbitt, J. (1984). Megatrends. New York: Warner.

Neugarten, B. L. (1977). Personality and aging. In J. Birren & K W. Schaie (Ed.)
Handbook of the Psychology of Aging (pp. 626-649). New York: Van Nostrant
Reinhold.

Neugarten, B. L.; Moore, J. W. & J. C. Lowe (1968). Age Norms, Age Constraints and
Adult Socialization. In B. L. Neugarten (Ed.) Middle Age and Aging (pp. 22-28).
Chicago, IL: University of Chicago Press.

Nickerson, N. P. & G. D. Ellis (1991). Traveler Types and Activation Theory: A
Comparison of Two Models. Journal of Travel Research. 29 (3), 26-32.

Opperman, M. (1995). Family Life Cycle and Cohort Effects: A Study of Travel Patterns
of German Residents. Journgl of Ewe! & Tourism Marketing, 4 (1), 23-45.

Palmore, E. (1978). When Can Age, Period and Cohort be Separated? Socigl Forces, 57
(1), 282-295.

Pearce, P. L &. M. L. Caltabiano (1983). Inferring Travel Motivations from Traveler’s
Experiences. Journal of Travel Research, 22 (I), 16-20.

Pennington, L. & D. L. Kcrstetter (1994). An Exploratory Study of the Beneﬁts Sought
by College-Educated Women Traveling for Pleasure: 13 there a Generational
Effect? Abstracts 1994 Leime Research Symmsium. National Recreation and
Park Association. San Antonio, TX.

Perreault, W. D.; D. K. Darden & W. R. Darden (1977). A Psychographic Classiﬁcation
of Vacation Life-styles. Journal of Leisure Research, 9 (1), 208-23.

Peters, T. (1988). Thriving on Chaos, Handbook for a Management Revolution. New
York: Knopf.

Pitts, R. E. & A. G. Woodside (1986). Personal Values and Travel Decisions. Journal of
Travel Research, 25 (1), 20-25.

Plog, S. (1972). Developing the Family Travel Market. In the Values of Trgvel Research:

Planning, Techniques and Applications. Travel Research Association Third
Annual Conference Proceedings, (pp. 209-221), Quebec, August, 13-16.

178

 

Plog, S. (1974). Why Destination Areas Rise and Fall in Popularity. The Cornell Hotel
and Restaurant Administration Quarterly, 14 (1), 55-58.

Poterba, J. M. (1989). Lifetime Incidence and the Distributional Burden of Excise Taxes.
America_n Economic Review. 79 (2), 325-330.

Rentz, J. O. & F. D. Reynolds (1991). Forecasting the Effects of an Aging Population on

Product Consumption: An Age-Period-Cohort Framework. J ourngl of Marketing
Research, 28 (3), 355-360.

Rentz, J. O.; F. D. Reynolds & R. G. Stout (1983). Analyzing Changing Consumption
Patterns with Cohort Analysis. Journgl of Marketing Research, 20 (1). 12-20.

Rentz, J .O. (1980). The Impact of Aging, Period and Cohort Membership on
Consumption of a Product Class, unpublished Ph.D. dissertation, University of

Georgia.

Reynolds, F. D. & J. O. Rentz (1981). Cohort Analysis: An Aid to Strategic Planning.
Journ_al of Marketing, 45 (2), 62-70.

Robinson, J. & G. Godbey (1996). The Great American Slowdown. American
Demoggrphics, 18 (6), 42-48.

Rokeach, M. (1973). The Nature of Human Values. New York: The Free Press.

Rosenfeld, J. P. (1986). Demographics on Vacation. Americh Demoggrphics, 8 (1), 38-
43.

Rosma G. & M. Blenman (1989). Vacation Patterns of the Elderly German. Annals of
Touris_rn Resear_cl_1_. 16(2), 178-88.

Samuel, N. (1999). Women, Leisure and the Family in Contemppm Sociegy: A
Multinational Persmctive (Ed.) UK: CAB International.

Schewe, C. D. & R. Calantone (1978). Psychographic Segmentation of Tourists. Journal
of Travel Research, 16(1), 14-20.

Schmitz-Scherzer, R., & H. Thomae (1983). Constancy and Change of behavior in Old
Age: Findings from the Bonn Longitudinal Study of Aging. In W. K. Schaie (Ed.)

Longitudinal Studies Of Adult Psychological Development (pp. 191-221). New
York: Guilford Press.

Scott, D, & E. K. Willits (1989). Adolescent and adult leisure patterns: A 37-Year
Follow-Up Study. Leisure Sciences, 11, 323-335.

179

 

Scott, D, & E. K. Willits (1998). Adolescent and Adult Leisure Patterns: A
Reassessment. Journ_al of Leisure Research, 30(3), 319-330.

Searle, M. S.; Mactavish, J. B. & R. E. Brayley (1993). Integrating, Ceasing Participation
with other Aspects of Leisure Behavior: A Replication and Extension. Journal of
Leisure Research, 26 (3), 227-237.

Shaie, K. W. & C. R. Strother (1968). A Cross-Sectional Study of Age Changes in
Cognitive Behavior. Psychological Bulletin, 75 (12), 671-680.

Shih, D. (1986). VALS as a tool of Tourism Market Research: The Pennsylvania
Experience. Journal of Travel Research, 23 (1), 2-11.

Shoemaker, S. (1989). Segmentation of the Senior Pleasure Travel Market Journal of
Travel Research, 29 (2), 8-12.

Shoemaker, S. (1994). Segmenting the US Travel Market According To Beneﬁts
Realized. Journal of Travel Researph, 32 (3), 8-19.

Shuh, P. & J. Crompton (1983). Search Behavior of International Vacationers: Travel
Speciﬁc Lifestyle and Sociodemographic Variables. Journal of Travel Research.
22 (2), 25-31.

Silverberg, K; S. Backman & K. Backrnan (1996). A Preliminary Investigation into the
Psychographics of Nature-Based Travelers to the Southeastern United States.
Journal of Travel Research 35 (2), 19-28.

 

Smale, B. J. A. & S. L. Dupuis (1995). A Longitudinal Analysis of the Relationship
Between Leisure Participation and Psychological Well-Being Across the
Lifespan. Paper presented 1995 National Recreation and Parks Association.

Smith, J. W. & A. Clurman (1997). Rocking the Ages: The Yankelovich Remrt on
Generational Marketing. New York: HarperCollinS Publishers, Inc.

Smith, W. (1957). Product Differentiation and Market Segmentation of Alternative
Marketing Strategies. Journal of Marketing, 1 (1), 4.

Snepenger, D. J. (1987). Segmenting the Vacation Market by Novelty Seeking Role.
Journal of Travel Resear_ch, 25 (2), 8-12.

Sofranko, A. J. & M. F. Nolan (1972). Early Life Experiences and Adult Sports
Participation. Journal of Leisure Research, 4(1), 6-18.

180

 

Statistics Canada. (1988). Postcensal Annual Estimates of PomLation Bv Marital Status
Age. Sex And Components Of Growth By Canada Provinces AngTerritories,
June 1 1988.

 

 

Statistics Canada. (1996). Postcengrl Annual Estimates of Ponulgtion by Marital Statup,

Age, Sex and Commnents of Grth by Canada Provinces and Territories. June
1 1996.

 

Strauss, W. & N. Howe (1991). The Cycle of Generations. American Demographics,
13(4), 24-33, 53.

Strauss, W. & N. Howe (1991). Generations: The History of America’s Future. 1584 to
2069. Quill: New York, New York.

Swenson, C. A. (1990). Selling to a Segmented Market: The Lifesgle Approach. New

York: Quorum Books.

Thayer, R. E. (1967). Measurement of Activation through Self-Report. Psychological
Remus, 7, 86-94.

Toﬁler, A. (1980). The Third Wave. New York: William Morrow.

Tongren, H. N. (1980). Travel Plans of the Over 65 Market Pre and Post Retirement.
Journpl of Trgvel Research. 19(1), 7-11.

Tourism Canada (1986). Market Research and the Canadian Tourim Attitude a_r_1g ~

Motivation Study: Research Report Segmentation Analysis. Ottawa, Canada:
Statistics Canada.

Unkel, M. (1981). Physical Recreation Participation of Females and Males During The
Adult Life Cycle. Leisure Sciencg, 4, 1-27.

Van Minden, T. T. R. (1987). Dictiom of Marketing Research. Aron: The Bath Press.
Villanueva, C. (1996). A Mosaic Future. Futures, 28 (6-7), 692-695.

Vincint, V. C. and G. de los Santos (1990). Winter Texans: Two Segments of the Senior
Travel Market. Journal of Travel Research, 29 (1), 9-12.

Warnick, R. (1993). Back to the Future: US. Domestic Travel and Generational Trends
1979 to 1991. Paper presented at the Resort and Commercial Recreation
Association, Mohonk Mountain House, New Paltz, New York.

 

Warnick, R. (1993). US. Domestic Travel: Back to the Future. The Impacts of an Aging
US. Population on Domestic Travel Trends. The Annual Review of Travel, (pp.
75-89). New York: American Express Travel Related Services, Inc.

181

 

Waryszak, R. Z. & H. Kim (1994). Psychographic Segmentation of Tourists as a
Predictor of their Travel Behavior: An Exploratory Study of Korean Tourists to

Australia. Journal of Hospitalig and Leisure Marketing, 2 (4), 5-19.

Weaver, P. A.; McCleary, K. W.; Lepisto, L. & L. T. Damonte (1994). The Relationship
of Destination Selection Attributes to Psychological, Behavioral and

Demographic Variables. The Journal of Hospitalig & Leisure Marketing, 2 (2),
93-109.

Wells, W. D. (1975). Psychographics: A Critical Review. Joumg of Marketing Researih,
12(5), 196-213.

Wells, W. D., & G. Gubar (1966). Life Cycle Concept in Marketing Research. Journal of
Marketing Research, 3 (10), 355-63.

Wight, P. A. (1996). North American Ecotourists: Market Proﬁle and Trip
Characteristics. Journal of Travel Research, 34 (1), 2-10.

Woodside A.G.; Moore, E. M. & M. J. Etzel (1980). Vacation Travel Behavior and
Perceived Beneﬁts of Home State Residents. Journal of Travel Resea_r_ch, 18 (4),
28-35.

 

Woodside, A. G. & R. E. Pitts (1976). Effects Of Consumer Life Styles, Demographics
and Travel Activities on Foreign and Domestic Travel Behavior. Journal of Tame;
Research, 14(1), 13-15.

Woodside, A. G. & L. W. Jacobs (1985). Step Two in Beneﬁt Segmentation: Learning
the Beneﬁts Realized by Major Travel Markets. Journal of Travel Resear_cl_1, 24
(1), 7-13.

Yoesting, D. R. & D. L. Burkhead (1973). Signiﬁcance of Childhood Recreation
Experience on Adult Leisure Behavior: An Exploratory Analysis. Journal of
Leisure Research, 5, 25-36.

Young, S.; Ott, L. & B. Feigin (1978). Some Practical Considerations in Market
Segmentation. Journal of Marketing Research, 15 (3), 405-410.

Zeiger, A. (1999). Customization Nation. Incentive, 173 (5), 35-50.

Zimmer, Z.; Brayley, R. E. & M. S. Searle (1995). Whether to Go and Where to Go:
Identiﬁcation of Important Issues, Journal of Travel Research, 33 (3), 3-11.

Zimmerman, C. (1982). The Life Cycle Concept as a Tool for Travel Research.
Transpprtation, 11, 51-69.

182

 

Zuzanek, J. & B. J. A. Smale (1992). Life-Cycle Variations in Across-the-Week
Allocation of Time to Selected Daily Activities. Lpi_s_ir et Societe/ Society a_n_d
Leisure, 15, 559-586.

183

T NIV

u . LIBRARIES
ll 1| HIMIWIWINHI
1 3998

£3 03

SA

5K3

MICHIGRN
1111111111“
13132

11