While subjective segmentation bases [611641]

While subjective segmentation bases
are very popular among consumerresearchers, many practitioners still usebasic demographic variables such as ageand income. For example, the recenttrend towards generational marketinginvolves segmentation by birth groups.
5–7
In a similar vein, others have advocateda cohort segmentation.
8,9The basic
premise on which these approaches arebased is that people in differentgenerations and cohorts have collectivelyexperienced different external events andcircumstances (eg war, economicchanges) that have shaped theirbehaviours as consumers, making themdifferent from others who haveexperienced different types of such eventsover their life course.
10Others haveINTRODUCTION
Market segmentation is one of the mostimportant strategic marketing decisions.
1
Numerous ways for segmenting themarket have been suggested in themarketing literature, ranging from simpledemographics to behavioural, attitudinal,and lifestyle variables such as benefits andV alues and Life Styles (VALS).
2–4
Generally speaking, objective bases forsegmentation such as geo-demographicsenable marketers to measure and locatetheir segmented customers precisely butoffer little explanation for marketbehaviour. On the other hand, subjectivesegmentation bases such as values andbenefits seek to help us better understandmarket behaviour but present problemsin measuring and locating segments.
/H17015Palgrave Macmillan Ltd 1479-1862/06 $30.00 Vol. 14, 2, 115–128 Journal of Targeting, Measurement and Analysis for Marketing 115Life-changing events and
marketing opportunities
Received (in revised form): 28th September, 2005
Anil Mathur
is Associate Dean and Professor of Marketing at the Frank G. Zarb School of Business, Hofstra University, Hempstead, New
York, USA. He has published over 60 articles in journals and conference proceedings. He has also consulted for manycorporations on a variety of marketing problems.
Euehun Lee
is Associate Professor of Marketing at the School of IT Business, Information and Communications University, Korea. He hasworked as a consultant and as a marketing researcher for several firms and government agencies in Korea and the USA.
George P. Moschis
is Alfred Bernhardt Research Professor of Marketing and Director, Center for Mature Consumer Studies at J. Mack RobinsonCollege of Business at Georgia State University, Atlanta, Georgia, USA. He is also a member of the university’s gerontologyprogramme. He has consulted for leading corporations and government agencies worldwide, and lectures frequently. Amonghis more than 100 publications are four books published by Quorum Books.
Abstract The study reported in this paper shows that specific life-changing events
may represent marketing opportunities for advertisers and marketers to woo customers.A life events–based segmentation model is developed, tested and compared withcompeting cohort- and age-based models in predicting consumer behaviour. The resultssuggest that the inclusion of life events in segmentation models may be a significantimprovement over age- or cohort-based segmentation models.
Anil Mathur
Professor of Marketing,Frank G. Zarb School ofBusiness, 306 Weller Hall,134 Hofstra University,Hempstead, NY 11549,USA.
Tel:/H110011 516 463 5346;
Fax:/H110011 516 463-5268;
e-mail: [anonimizat]

The present paper advocates the use of
life events as an approach to marketsegmentation. Recent developments inthe information technologies andmarketing intelligence make it possiblefor researchers to gather information onvarious types of events that peopleexperience over their life course (egmarriage, graduation, retirement andeven the onset of health problems).
19
Although some previous researchersreport attempts to use life events assegmentation variables,
20,21systematic
approaches to the use of life events basedon theory and practical considerations (egcomparison of life events-based segmentsvs age-based segments) are largely sparse.Speci fically, the paper presents theoretical
bases for the proposed approach and theresults of a survey designed to test theefficacy of the life-events-based
segmentation model. Finally, the derivedmodel is compared with other populardemographics-based models (age andcohort) using the same approach as theone used by previous researchers.
22,23
THEORETICAL PERSPECTIVES
The proposed segmentation model isbased upon life-course research whichpostulates that behaviour at a given pointin time is the product of responses tochanging life conditions (such as events,changes or transitions) and the way theindividuals adapt to social andenvironmental circumstances.
24Research
on life course has been guided by threemajor theoretical orientations: humancapital, stress and normativeperspectives.
25Thehuman capital
perspective contends that a person ’s
behaviours are determined by their‘personal resources ’, which include
abilities, skills and knowledge. As peopleare exposed to different types of eventsand circumstances over their life course,they are likely to differ with respect toadvocated a combination of demographic
variables such as family life cycle that arenot context- and time-dependent.
11
One objective approach that has been
suggested and even cited in marketingtextbooks
12is based on life events or life
status changes that are unique to theperson experiencing them (and notnecessarily relevant to all people in thesame cohort). For example, divorce orrelocation can cause alteration inconsumption habits.
13,14In his landmark
study of life status changes and changesin consumer behaviour, Andreasen
15calls
for placing greater emphasis on life statusvariables and suggests that ‘measures of
status change should be seriouslyconsidered as predictor variables in futureconsumer studies in marketing,particularly those concerned withdeveloping market segments ’(p. 794).
Similarly, Kotler suggests that,
‘A company can consider occasions of
critical life events or transitions —marriage,
childbirth, illness, relocation, divorce, careerchange —as giving rise to new needs.
These should alert service providers —
banks, lawyers, and marriage, employment,and bereavement counselors —to ways they
can help ’.
16
Recent research on household
expenditures17also points to the
desirability of studying consumptionpatterns in the context of life events thatdefine life stages and transitions.
These suggestions beg the question: Is
life event-based segmentation a viableapproach? How does it compare withother popular objective segmentationmodels such as those based ondemographics and cohorts? Whilecohort-based segmentation has recentlyreceived support as a segmentationmodel,
18there is virtually no empirical
work that reveals the ef ficacy of cohort-
and event-based segmentation models inpredicting consumer behaviour.
116 Journal of Targeting, Measurement and Analysis for Marketing Vol. 14, 2, 115 –128/H17015Palgrave Macmillan Ltd 1479-1862/06 $30.00Mathur, Lee and Moschis

new products that help them de fine the
new role and dispose of productsrelevant to the enactment of a previousrole.
324,25
Because the three life-course
perspectives suggest that the experience
andanticipation of certain life events, as
well as their timing , may affect people in
a similar way, it is expected that theseexperiences will manifest in differentconsumption patterns. These life-eventexperiences are hypothesised to be betterpredictors of market behaviour at a givenpoint in time than competingsegmentation models based on age andcohorts.
THE STUDY
Sample
Data for the study were collectedthrough mail questionnaire as a part of alarge national study. The sample wasrandomly drawn from the database of amajor mailing list vendor, which containsapproximately 87 million householdnames and addresses. Questionnaires weremailed to ten thousand adults chosen inproportion to the population of each ofthe 50 US states and speci fic age groups.
A total of 1,534 adults responded,reflecting a response rate of 15.34 per
cent. Although the response rate isrelatively low, it is consistent with theresponse rates for other national studiesof general populations. The surveyquestionnaire contained many questionsrelating to various issues; however, thepresent study utilises questions relating tolife events and consumption-relatinglifestyles. As many respondents had notexperienced any event included in thestudy or had experienced very fewevents, and because it is a commonprocedure in psychological research tosurvey or include only individuals orsamples of people who have experiencedpersonal resources and, consequently, the
types of behaviours they enact.Explanations for consumer behaviourover the life course from a human capitalperspective focus on observed differencesin consumer behaviour (primarilypatterns of information processing)among age groups.
26–28
The second perspective is based on
stress theory and research. Major life
events, changes and transitions (bothdesirable and undesirable) are oftentreated as ‘stressors ’that create a
generalised demand for readjustment bythe individual. People attempt to restorebalance and relieve frustrations andtensions accompanying disequilibrium byinitiating or modifying behaviours, whichare viewed as coping strategies.
29,30
Support for the stress perspective isfound in previous consumer studiesshowing that initiation, intensi fication or
changes in consumption habits re flect
efforts to handle stressful life events.
31,32
Finally, the normative perspective holds
that different behaviours are the outcomeof various roles people acquire and enactat different stages in life. As peopleacquire (or anticipate transitions into)new roles and relinquish old ones, theychange their behaviours accordingly, asthey rede fine their self-concepts and
attempt to enact socially prescribedroles.
33Certain events serve as markers of
transitions into roles that are normativelygoverned and predictable in bothoccurrence and timing. An individual isgradually socialised into a role eitherbefore the occurrence of a normativeevent (eg birth of first child into
‘parenthood ’) or upon the occurrence of
an unexpected life event (eg death ofspouse into ‘widowhood ’). According to
this perspective, changes in consumerbehaviour re flect rede finition of one ’s
self-concept as a result of the assumptionof a new role and the relinquishment ofold ones, as people attempt to acquire
/H17015Palgrave Macmillan Ltd 1479-1862/06 $30.00 Vol. 14, 2, 115 –128Journal of Targeting, Measurement and Analysis for Marketing 117Life-changing events and marketing opportunities

Life events
Respondents were also asked to indicate
whether or not they most recentlyexperienced any of 25 life events (seeAppendix) ‘in the past 6 months ’,‘in the
past 6 –12 months ’,a n d ‘more than 12
months ago ’. These life events were
selected based on previous research.
37–41
Several of the life events included in the
study are markers of role transitions (egmarriage, birth of a child) and can bemeasured relatively easily using objectivemeasures. Some other events included inthe study are merely stressors (eg seriousinjury); nevertheless, they have thepotential to trigger important lifestyle andconsumption behaviour changes ofimportance to marketers. Each life eventexperienced by the individual was codedas one (1); otherwise it was coded aszero (0). The timing of each life eventexperienced was also included as aseparate variable, measured on athree-point scale, re flecting the length of
time since the experience. In addition,subjects responded to a list of 14 lifeevents (see Appendix) they anticipated‘in the next 6 months ’and‘in the next
few years ’. Each event anticipated was
coded as one (1); otherwise it was codedas zero (0). The timing of eachanticipated event was also included as aseparate variable, measured on atwo-point scale, re flecting the time
before the individual expects toexperience each event.
ANALYSIS AND RESULTS
Hierarchical cluster analysis (using SPSS)was used to group respondents based ontheir past and anticipated experiences oflife events, as well as the timing of theseevents. A total of 78 variables wereincluded in the cluster analysis,comprising 25 life events experienced, 25variables representing the timing of suchexperiences, 14 anticipated events and 14certain events and compare them with
those who have not experienced them,
36
a judgment sample was drawn from thereturned questionnaires. First, allindividuals who had experienced two ormore events in the previous six monthswere included in the study sample(n/H11005340). Next, a random sample of
those who had experienced only oneevent in the previous six months(n/H11005203) and that of those who had not
experienced any event included in thestudy ( n/H11005322) was also included in the
study sample. The final sample used
consisted of 866 questionnaires. Althoughthe initial mailing list was randomlyselected, the final sample used in this
study was a judgment sample and cannot be considered to be a representativeof the total US population. The mainpurpose of the study, however, was toshow the advantage of segmentingmarkets based on variables not includedor studied previously, and not to estimatesegment sizes or population parameters.T h ea g er a n g eo ft h i ss a m p l ew a s2 1 –84
years, with a mean of 49.95 years and astandard deviation of 13.92 years —
figures that compare favourably with
Census data for the adult population.
Variables
Consumption-related behaviours
Respondents were asked to indicate
whether they most recently initiated orchanged 24 consumption-relatedbehaviours ‘in the previous 6 months ’,
‘6–12 months ago ’,‘more than 12
months ago ’,o r ‘never had experienced
or done the activity ’.A ffirmative
responses to the any of the first three
categories were coded as one (1) foreach type of consumption-relatedresponse. A negative response was codedas zero (0) for these consumption-relatedbehaviours.
118 Journal of Targeting, Measurement and Analysis for Marketing Vol. 14, 2, 115 –128/H17015Palgrave Macmillan Ltd 1479-1862/06 $30.00Mathur, Lee and Moschis

segments. In order to name the derived
segments, each cluster was cross-classi fied
by the events experienced andanticipated that were used to createthem. The first cluster, which is the
largest, containing 42 per cent of thetotal sample, consists of people who hadexperienced or expected to experiencethe smallest number of life events.Because of the small number oflife-changing events experienced bypeople in this segment, the largest clusterwas named ‘The Unruf fled’. The second
segment consists of primarily older adults,with more than three-quarters of thosein the second segment born before 1940.Many individuals in this segment hadrecently experienced retirement andempty nest; they moved to a differentplace and became grandparents. Thissegment, which comprises 16 per cent ofthe total sample, was called ‘Free Birds ’.
The third segment was the smallest (9per cent) and the one that hadexperienced the largest number of lifeevents. The respondents in this segmentwere more likely to have experienced(or expected to experience) almost allthe events (except those experienced byFree Birds) than people in othersegments. This segment was called the‘Chronic Strugglers ’. Finally, the last
segment consists of 33 per cent of thesample, with three-quarters of thembeing baby boomers. Most people in thisgroup had experienced relatively fewerevents than the preceding two segments,with most of the experienced eventsrelated to family, such as marriage andbirth of a child. Because this group had adisproportionately high number of peopleliving with their spouse and child(ren),this group was called ‘Full Nesters ’.
A ss h o w ni nT a b l e1 ,t h em e a na g e
of Free Birds is the highest(mean/H1100563.1) and that of Full Nesters
is the lowest (mean /H1100542.0). The
Unruf fled and Chronic Strugglers arevariables representing anticipated timing
of such events. Theory and researchsuggested that the actual experience ofan event may have a different effect onthe person ’s behaviour than the
anticipated experience of the sameevent.
42Similarly, the timing of an event
might have a different effect on theperson ’s behaviour.
43
In the first stage of cluster analysis,
clusters were allowed to form freely andthe resulting agglomeration schedule(containing the coef fic i e n ta te a c hs t a g e )
was examined to determine theappropriate number of clusters for thispopulation. It has been well recognisedthat in any segmentation study, thedecision on the number of segments is asmuch an art as it is a science. Onerecent study of 2,000 adults conductedby Roper Starch Worldwide Inc. forModern Maturity segmented the adult
market based on life events.
44,45This
study, which identi fied seven segments,
was initially used as a guideline indeciding on the number of clusters toretain. Based on improvement in thecoefficient (squared Euclidean distance),
it was decided to retain four clusters. Asecond round of cluster analysis wasdone, with the number of clustersspeci fied as four. Resulting cluster
membership was saved and subsequentlyused to compare this segmentationapproach with other segmentationmethods (age-based and cohort-based)across 24 consumption-related variables.
SEGMENTS BASED ON
LIFE EVENTS
Demographic and other characteristics of
the derived segments based on life eventsare given in Table 1. Due to theover-representation of male heads ofhousehold in the mailing list, theproportion of male respondents is greaterthan that of females across all four
/H17015Palgrave Macmillan Ltd 1479-1862/06 $30.00 Vol. 14, 2, 115 –128Journal of Targeting, Measurement and Analysis for Marketing 119Life-changing events and marketing opportunities

(59.6 per cent) of Free Birds who are
retired or not working. Table 2 showsthe mean values of input variables forthe four life-event-based segments.
The study aimed to find out how
these event-based clusters differ in termsof various consumer behaviours. Table 3shows the differences in consumerbehaviours across the four lifeevent-based clusters. As shown in theTable, out of 24 consumer behavioursexamined in the study, there weresignificant differences for 18 behaviours.
For two of the investigated behaviours(moved into retirement home andreceived healthcare at home), thepercentage of respondents in the overallsample giving a positive response wasvery low.
One way to assess the value of lifeapproximately of the same age. The
four segments are mostly similar interms of their gender make-up. OnlyFree Birds comprise a relatively higherproportion of females. While FullNesters have the highest income level,with more than half (55.5 per cent) ofthem reporting an income of $50,000or more, the Free Birds and theUnruf fled have the lowest incomes.
More than half of the Full Nestersbelong to the full-nest stage of theirlife cycle, with 56.4 per cent of themliving with spouse and child(ren). Onthe other hand, more than half of FreeB i r d sl i v ei na ne m p t yn e s t ,w i t h5 4 . 4per cent reporting living with spouseonly. A vast majority of ChronicStrugglers (75.3 per cent) areemployed, compared with almost half
120 Journal of Targeting, Measurement and Analysis for Marketing Vol. 14, 2, 115 –128/H17015Palgrave Macmillan Ltd 1479-1862/06 $30.00Mathur, Lee and Moschis
Table 1: Demographic profiles of segments based on life events
Chronic
The Unruffled Free Birds Strugglers Full Nesters(42%) (16%) (9%) (33%) Sig. level
Age (mean) 51.6 63.1 48.4 42.0 0.000
Sex 0.049
Male (%) 56.4 68.7 57.3 54.5Female (%) 43.6 31.3 42.7 45.5Income 0.000
Less than $20,000 (%) 12.4 15.0 8.0 2.8$20,000 –$34,999 (%) 26.7 22.6 25.3 17.7
$35,000 –$49,999 (%) 23.0 24.1 25.3 24.0
$50,000 –$74,999 (%) 22.5 17.3 21.3 32.5
$75,000 and above (%) 15.4 21.1 20.0 23.0Living status 0.000
Live alone (%) 31.5 22.1 16.0 9.1Live with spouse (%) 26.0 54.4 29.6 12.9Live with spouse and child(ren) (%) 19.3 14.7 27.2 56.4Live with child(ren) only (%) 23.2 8.8 27.2 21.6Employment status 0.000
Retired or not employed (%) 28.8 59.6 17.3 11.0Retired and employed (%) 6.1 10.3 7.4 2.5Employed (%) 65.1 30.1 75.3 86.5Education 0.241
High school or less (%) 20.2 25.2 13.6 16.4Some college (%) 33.2 34.8 32.1 35.0College graduate or more (%) 46.5 40.4 54.3 48.6Health status 0.000
No. of medical problems (mean) 1.42 2.39 1.60 0.92No. of prescription drugs (mean) 1.32 1.83 1.30 0.72Life events (mean #) 0.000
Events experienced 5.34 11.87 14.55 8.36Events expected 1.57 1.68 2.78 1.79n= 866 362 136 81 287
Sig. = significance

/H17015Palgrave Macmillan Ltd 1479-1862/06 $30.00 Vol. 14, 2, 115 –128Journal of Targeting, Measurement and Analysis for Marketing 121Life-changing events and marketing opportunitiesTable 2 : Mean values of input variables by segments based on life events
Experienced events Anticipated events
Chronic Chronic Full
The Unruffled Free Birds Strugglers Full Nesters The Unruffled Free Birds Strugglers Nesters(42%) (16%) (9%) (33%) (42%) (16%) (9%) (33%)
Moved to a different place* 0.414 0.963 0.926 0.965 0.332 0.279 0.494 0.307
(1.124) (2.846) (2.741) (2.822) (0.561) (0.515) (0.926) (0.540)
Marriage* 0.149 0.985 0.951 0.962 0.061 0.015 0.074 0.031
(0.412) (2.934) (2.852) (2.878) (0.111) (0.022) (0.136) (0.059)
Birth or adoption of a child* 0.055 0.890 0.704 0.836 0.028 0.007 0.025 0.070
(0.144) (2.632) (2.074) (2.429) (0.053) (0.015) (0.037) (0.115)
Divorce or separation* 0.130 0.206 0.605 0.268 0.030 0.022 0.037 0.031
(0.312) (0.603) (1.753) (0.781) (0.047) (0.029) (0.074) (0.056)
Last child moved out of 0.163 0.868 0.309 0.125 0.083 0.088 0.259 0.202
household* (0.417) (2.522) (0.827) (0.289) (0.155) (0.162) (0.482) (0.376)
Death of spouse 0.086 0.125 0.124 0.017
(0.238) (0.324) (0.346) (0.045)
Death of a parent or close 0.456 0.934 0.840 0.585 0.260 0.331 0.333 0.279
family member* (1.135) (2.581) (2.296) (1.627) (0.492) (0.654) (0.642) (0.509)
Birth of first grandchild* 0.127 0.882 0.247 0.108 0.033 0.037 0.099 0.105
(0.365) (2.559) (0.741) (0.289) (0.061) (0.052) (0.198) (0.185)
Major conflict with family member 0.356 0.478 0.691 0.509
(0.616) (1.154) (1.482) (0.979)
Retirement (at own will)* 0.177 0.581 0.198 0.056 0.083 0.154 0.148 0.063
(0.489) (1.654) (0.593) (0.153) (0.149) (0.279) (0.296) (0.125)
Lost job/business or forced 0.196 0.250 0.741 0.160 0.055 0.066 0.185 0.042
to retire* (0.489) (0.713) (2.124) (0.394) (0.091) (0.132) (0.309) (0.073)
Started work for the first time or 0.122 0.213 0.778 0.230 0.050 0.029 0.049 0.056
after not working for a long time* (0.287) (0.618) (2.309) (0.627) (0.066) (0.037) (0.074) (0.073)
Reduction in hours of 0.102 0.132 0.457 0.119 0.022 0.096 0.148 0.038
employment or giving up (0.218) (0.353) (1.235) (0.272) (0.033) (0.169) (0.247) (0.056)employment (at own will)*
Significant success at work or 0.301 0.485 0.790 0.537
personal life (0.483) (1.096) (1.765) (1.070)
Change jobs, same or different 0.249 0.397 0.926 0.460 0.160 0.081 0.284 0.185
type* (0.492) (1.140) (2.617) (1.143) (0.251) (0.125) (0.457) (0.303)
Major improvement in financial 0.185 0.309 0.654 0.362
status (0.387) (0.860) (1.728) (0.861)
Financial status a lot worse 0.304 0.309 0.716 0.324 0.160 0.191 0.185 0.129
than usual* (0.550) (0.640) (1.877) (0.606) (0.229) (0.324) (0.259) (0.157)
Family member's health a 0.279 0.471 0.593 0.265
lot worse (0.497) (1.140) (1.272) (0.443)
More responsibility for aged 0.271 0.427 0.494 0.174 0.227 0.265 0.333 0.251
relative* (0.500) (1.037) (1.124) (0.331) (0.384) (0.427) (0.568) (0.443)

122 Journal of Targeting, Measurement and Analysis for Marketing Vol. 14, 2, 115 –128/H17015Palgrave Macmillan Ltd 1479-1862/06 $30.00Mathur, Lee and Moschis
Table 2 : Continued
Experienced events Anticipated events
Chronic Chronic Full
The Unruffled Free Birds Strugglers Full Nesters The Unruffled Free Birds Strugglers Nesters(42%) (16%) (9%) (33%) (42%) (16%) (9%) (33%)
Gained a lot of weight 0.271 0.162 0.543 0.338
(0.586) (0.419) (1.444) (0.829)
Chronic illness or condition 0.155 0.338 0.407 0.174
diagnosed (0.356) (0.919) (1.012) (0.394)
Serious injury, illness or major 0.232 0.463 0.580 0.181
surgery (0.506) (1.191) (1.482) (0.387)
Community crisis or disaster 0.122 0.110 0.296 0.160
(hurricane crime, fire, flood, (0.287) (0.265) (0.803) (0.408)earthquake, etc)
Death or loss of a pet 0.235 0.331 0.630 0.279
(dog or cat) (0.575) (0.949) (1.704) (0.634)
Stopped smoking 0.163 0.441 0.420 0.178
(0.434) (1.302) (1.222) (0.502)
Table entries are mean values. Numbers in parentheses represent mean values for corresponding timing variable. Asterisks identi fy events included in the list of anticipated
events.

percentage of respondents in the four
clusters who engaged in various types ofconsumption-related activities during theprevious 12 months.
As it can be seen in Table 3, a larger
percentage of Chronic Strugglers thanthe Unruf fled engaged in all 24
consumer activities during the 12 monthspreceding the survey. The differenceswere particularly noticeable for productsand services that people use to cope withevents in marketing strategy is to assess
the consumption activity of the twoextreme life event-based segments —the
Unruf fled and the Chronic Strugglers. If
new consumer needs stem from personaltransitions because people buy productsand services that ease transition andaccommodate change, then life-eventchanges provide an importantopportunity for advertisers and marketersto woo customers. Table 2 shows the
/H17015Palgrave Macmillan Ltd 1479-1862/06 $30.00 Vol. 14, 2, 115 –128Journal of Targeting, Measurement and Analysis for Marketing 123Life-changing events and marketing opportunities
Table 3 : Consumption-related differences across event-based segments
The Unruffled Free Birds Chronic Full
(42%) (16%) Strugglers (9%) Nesters (33%)% % % % Prob.
Financial servicesSet new investment goals (retirement, home, etc) 61.3 66.2 74.1 72.5 0.012Made more changes than usual in key investments 55.0 61.8 55.6 60.6 0.367
(CDs, mutual funds, stocks and bonds)
Received professional legal or financial advice 50.6 52.9 76.5 55.7 0.000
for the first time or after not receiving for a long time
HousingHome purchase or sale 64.4 86.8 82.7 86.1 0.000Moved into a retirement or nursing home 0.6 1.5 2.5 0.3 0.205Home remodelling or refurnishing 72.7 88.2 90.1 85.0 0.000Recreational/cultural activitiesWent on a vacation abroad for the first time 47.8 49.3 55.6 44.3 0.320
or after not going for a long time
Took on a new hobby or recreational activity 69.9 76.5 86.4 81.2 0.001Change in attendance of cultural events 52.2 51.5 59.3 50.5 0.579Change in the amount or type of television viewing 68.2 66.2 86.4 72.5 0.006Social activitiesChange in attendance of religious activities 53.0 60.3 75.3 63.8 0.001Change in social relations 67.4 71.3 88.9 74.2 0.001Food, beverages, smokingIncreased consumption of alcoholic beverages 22.4 18.4 39.5 24.4 0.004Ate out a lot more times than usual 67.1 61.0 77.8 71.4 0.042Started smoking for the first time or after not 14.9 28.7 27.2 27.2 0.000
smoking for a long time
ShoppingBought more gifts than usual 53.6 47.8 69.1 55.7 0.021Spent more than usual on clothes 55.2 47.1 70.4 58.2 0.008Made more buying decisions than usual 39.0 57.4 64.2 62.7 0.000
together with spouse
Health-relatedReceived professional counselling for the 26.2 19.1 50.6 32.1 0.000
first time or after not receiving for a long time
Used more anti-depressants or tranquilisers 11.3 10.3 23.5 12.2 0.019
than usual
Started diet/weight control or exercise programme 66.0 61.0 82.7 71.8 0.004Received healthcare or personal-care servicesAt home for the first time 7.2 8.1 7.4 9.4 0.772AltruismGave more money or time than usual to charities 59.9 61.8 72.8 59.6 0.158InsuranceChange in amount or type of insurance coverage 62.4 68.4 81.5 76.7 0.000
Table entries are percentage of individuals in each cluster that have experienced or engaged in that behaviour.Prob = probability

segments have a less skewed
distribution in terms of age and cohortmembership.
Following the approach used by
previous researchers to comparesegmentation models,
48,49regression
analysis was used to compare the threetypes of segmentation schemes. Threeseparate regression models were tested foreach consumer behaviour-related activityusing the three segmentation variables asindependent variables. Regression modelsfor event-based segmentation had fourparameters, corresponding to an interceptterm plus three dummy variables for fourmutually exclusive event-based segments.Regression models for cohort-basedsegments had five parameters
(corresponding to an intercept term plusfour dummy variables) for the five
mutually exclusive cohort-basedsegments. Finally, regression models forage-based segments had five parameters
(corresponding to an intercept term plusfour dummy variables) for the five
mutually exclusive age groups. As theseregression models are not nested theycould not be directly compared.Therefore, two additional regressionmodels combining event-based segmentswith age and cohort, respectively, werestress and anxiety such as alcohol,
cigarettes, mood-altering drugs andprofessional counselling. Collectively,these findings suggest that changes in
customers ’lives create new consumption
needs and corresponding opportunitiesfor marketers and advertisers to appeal tothese needs.
COMPARISON TO AGE- AND
COHORT-BASED SEGMENTS
Age-based segments were created by
dividing the sample into five
traditionally used adult age groups(21–34, 35 –44, 45 –54, 55 –64, 65 years
or older). Similarly, cohort-based groupswere created by dividing the sampleinto five traditionally recognised cohorts
based on the year of birth: GenerationX (those born in or after 1965), BabyBoomers (those born between 1946and 1964), War Babies (those bornbetween 1940 and 1945), DepressionGeneration (those born between 1930and 1939) and GI Generation (thoseborn before 1930).
46,47Cross-tabulations
of event-based segments by age groupsand cohorts are given in Table 4.While Full Nesters contain the vastmajority of Baby Boomers, other
124 Journal of Targeting, Measurement and Analysis for Marketing Vol. 14, 2, 115 –128/H17015Palgrave Macmillan Ltd 1479-1862/06 $30.00Mathur, Lee and Moschis
Table 4: Cross-tabulation of event-based segments by age groups and cohort membership
The Unruffled Free Birds Chronic Full Nesters
(41%) (16%) Strugglers (9%) (33%)% % % % Sig. level
Age (years) 0.000
21–34 11.1 0.0 3.9 21.5
35–44 26.5 2.3 37.7 43.0
45–54 23.9 20.5 35.1 26.8
55–64 16.5 30.3 10.4 7.0
65 and above 21.9 47.0 13.0 1.8Cohorts 0.000
Generation X 4.6 0.0 0.0 6.0Baby Boomers 43.6 9.1 57.1 75.4War Babies 13.4 13.6 19.5 9.9Depression Generation 16.5 30.3 10.4 7.0GI Generation 21.9 47.0 13.0 1.8n= 866 362 136 81 287
Sig. = significant

17 out of 24 lifestyle variables
examined was signi ficant ( p< 0.05)
when event-based segmentationvariables were added to regressionmodels with cohort variables (mean R
2
increased to 0.036).
DISCUSSION AND IMPLICATIONS
One could argue that all these
segmentation methods produce low R2
values. However, this does not reduce
the value of segmentation models. As hasbeen pointed out in previous research,
52
segmentation approaches using generalconsumer characteristics generallyproduce similarly low R
2values.
Moreover, as discussed by Novak andMacEvoy,
53even low R2could re flect
significant differences across segments,
which could have major impact onmarketing strategies. In addition,segments based on age and cohorts werenot pro filed in terms of their speci fic
consumption-related behaviours becausethe purpose of the study was todemonstrate the bene fit of using life
events as a segmentation base rather thanto estimate sizes of segments based onsuch characteristics. Moreover, the dataused in the study were based on ajudgment sample and, therefore, thederived segment pro files might not be of
much value to marketing practitioners.
A couple of other caveats are in order.
First, the present study did not address allpossible aspects of the person ’s consumer
behaviour in the marketplace. Therefore,the results may not be generalisable tosituations (behaviours) other than thosestudied. Secondly, while age- andcohort-based segments can be formed onanap r i o r i basis (and can be fixed across
studies and time), the derived segmentsbased on life events may differ innumber and size across studies and time,depending on the sample and life-eventlists used.developed. One model had eight
parameters, corresponding to aninteraction term, three dummy variablesfor the four mutually exclusiveevent-based segments, and four dummyvariables for the five mutually exclusive
age groups. Similarly, the other modelhad eight parameters, corresponding toan interaction term, three dummyvariables corresponding to the fourmutually exclusive event-based segments,and four dummy variables for the five
mutually exclusive cohorts.
R
2values for the three alternative
segmentation methods for eachconsumption-related item, as well asthose for the two models combiningevent-based segments with age andcohort membership, respectively, wereexamined. Generally, the sizes of R
2
values were very close to those
obtained using other segmentationmodels.
50,51The mean R2for
event-based segmentation across the 25consumer behaviours was 0.018, whilethe values for cohort-basedsegmentation and for age-basedsegmentation were 0.02 and 0.02,respectively. As these models are notnested, a statistical comparison couldnot be done. However, regressionmodels combining event-based segmentswith age and cohort membershipprovided support for the value ofevent-based segmentation. Whenevent-based segmentation variables wereincluded in the regression models withage-group variables, R
2values increased
for all equations (mean R2increased to
0.036). Incremental R2for event-based
segmentation variables ( R2
events|age groups )
was signi ficant ( p< 0.05) for 18 out of
24 lifestyle variables studied. Thisshows that event-based segments addadditional explanatory power to thesegmentation models based on age.Similarly, incremental R
2for
event-based segments ( R2
events|cohort ) for
/H17015Palgrave Macmillan Ltd 1479-1862/06 $30.00 Vol. 14, 2, 115 –128Journal of Targeting, Measurement and Analysis for Marketing 125Life-changing events and marketing opportunities

gathering on life events experienced by a
person easier than ever before. Forexample, information regarding eventssuch as marriage and birth of a child isobtainable from public records.Moreover, these events representmarketing opportunities, as people buyproducts and services to accommodatechange and ease transitions. As peopleexperience major life-changing events,they re-evaluate their priorities, productneeds, brand and store preferences, andthe criteria by which they selectproducts. Segments based on suchlife-changing events re flect such
differences in consumer behaviour,making certain segments more receptiveto marketing offerings than othersegments. Thus, there is an opportunityfor targeting different segments withdifferent products.
In sum, the results of the present
research demonstrate the viability of thelife-events segmentation basis along withdemographic variables such as age andcohort. Future research could examine alarger number and types of life events asbases for segmentation. Measures of lifeevents experienced could also be furtherrefined by examining more detailed
measures of timing of events and perhapssequence of life events, especially eventson which information can easily beobtained by marketers. There is also aneed for studying a wider range ofconsumption-related behaviours,especially consumer responses tomarketing-mix variables.
References
1 Kotler, P . (2003) ‘Marketing Management ’, 11th
edn. Prentice Hall, Upper Saddle River, NJ.
2Ibid.
3 Bernstein, P . (1978) ‘Psychographic still an issue on
Madison Avenue ’.Fortune , 16th January, pp. 78 –84.
4 Novak, T. P . and MacEvoy, B. (1990) ‘On
comparing alternative segmentation schemes: TheList of Values (LOV) and Values and Life Styles(VALS) ’,Journal of Consumer Research ,V o l .1 7 ,N o .
1, pp. 105 –109.It is understood that the timing of
occurrence of life events might have aneffect, and, as such, was included in theanalysis. The present research utilised anordinal measure of the length of timesince experience of an event, however,although it is possible that full impact ofthe timing of events was not captured inthis ordinal measure. Future researchersmight consider more precise measure ofevent timing for including insegmentation models based on lifeevents.
These limitations notwithstanding, the
present study demonstrated the value ofevents when they are added tocohort-based and age-based models.Although marketers and advertisers realisethe importance of the aging consumermarket, many of them continue to useage-based segmentation approaches.However, recent research has shown thatage does not directly affect one ’s
behaviour. Schiffman and Sherman
54
assert that age is no longer an indicatorof one ’s physical state but merely a state
of mind. Consequently, age may not beadequate in explaining the consumptionbehaviour of individuals. There may beother factors that have greater impact onone’s behaviour. Indeed, as stated by
Neugarten and Neugarten,
55‘age has
become a poor predictor of the timingof life events, as well as a poor predictorof a person ’s health, work status, and
therefore, also, of a person ’s interests,
preoccupations, and needs ’(p. 36).
Therefore, it seems more desirable to uselife events along with age to predictbehaviour.
Segmentation methods based on life
events provide an opportunity for onenot only to understand the basis forobserved differences, but also to useobjectively measured variables (events).Although collecting information on lifeevents is still dif ficult, the proliferation of
information technologies have made data
126 Journal of Targeting, Measurement and Analysis for Marketing Vol. 14, 2, 115 –128/H17015Palgrave Macmillan Ltd 1479-1862/06 $30.00Mathur, Lee and Moschis

Journal of Marketing Research ,V o l .2 7 ,N o .2 ,
pp. 175 –184.
27 Gaeth, G. J. and Heath, T. B. (1987) ‘Cognitive
processing of misleading advertising in young andold adults: Assessment and training ’,Journal of
Consumer Research ,V o l .1 4 ,N o .1 ,p p .4 3 –54.
28 John, D. R. and Cole, C. (1986) ‘Age differences in
information processing: Understanding de ficits in
young and elderly consumers ’,Journal of Consumer
Research ,V o l .1 3 ,N o .3 ,p p .2 9 7 –315.
29 Lazarus, R. and Folkman, S. (1984) ‘Stress,
Appraisal, and Coping ’, Springer, New York, NY .
30 Pearlin, L. I. (1982) ‘Discontinuities in the study of
aging ’,i nH a r e v e n ,T .K .a n dA d a m s ,K .J .( e d s )
‘Aging and Life Course Transitions: An
Interdisciplinary Perspective ’, The Guilford Press,
New York, NY , pp. 55 –74.
31 Andreasen (1984) op. cit .
32 O ’Guinn, T. C. and Faber, R. J. (1989)
‘Compulsive buying: A phenomenological
exploration ’,Journal of Consumer Research ,V o l .1 6 ,
N o .2 ,p p .1 4 7 –157.
33 Hagestad, G. O. and Neugarten, B. L. (1985) ‘Age
and the life course ’, in Binstock, R. and Shanas, E.
(eds) ‘Handbook of Aging and the Social Sciences ’,
2 n de d n ,V a nN o s t r a n dR e i n h o l d ,N e wY o r k ,N Y ,pp. 35 –61.
34 McAlexander (1991) op. cit .
35 Mehta, R. and Belk, R. W . (1991) ‘Artifacts,
identity, and transition: Favorite possessions ofIndians and Indian immigrants to the United States ’,
Journal of Consumer Research ,V o l .1 7 ,N o .4 ,p p .
398–411.
36 Norris, F . H. and Uhl, G. A. (1993) ‘Chronic stress
as a mediator of acute stress: The case of HurricaneHugo ’,J o u r n a lo fA p p l i e dS o c i a lP s y c h o l o g y ,V o l .2 3 ,
No. 16, pp. 1263 –1284.
37 Andreasen (1984) op. cit .
38 Norris and Uhl (1993) op. cit .
39 Silvers (1997) op. cit .
4 0S t o n e ,A .A . ,H e l d e r ,L .a n dS c h n e i d e r ,M .S .
(1988) ‘Coping with stressful events: Coping
dimensions and issues ’, in Cohen, L. H. (ed.) ‘Life
Events and Psychological Functioning ’,S a g e
Publications, Newbury Park, CA, pp. 182 –210.
41 Turner, J. R. and Avison, W . R. (1992) ‘Innovations
in the measurement of life stress: Crisis theory andthe signi ficance of event resolution ’,Journal of Health
and Social Behavior ,V o l .3 3 ,M a r c h ,p p .3 6 –50.
42 Pearlin (1982) op. cit .
43 Hagestad and Neugarten (1985) op. cit .
44 AARP (1995) op. cit .
45 Silvers (1997) op. cit .
46 Crispell, D. (1993) ‘Where generations divide: A
guide ’,American Demographics ,M a y ,p p .9 –10.
47 Noble, S. M. and Schewe, C. D. (2003) ‘Cohort
segmentation: An exploration of its validity ’,Journal
of Business Research , Vol. 56, pp. 979 –987.
48 Kamakura and Novak (1992) op. cit .
49 Novak and MacEvoy (1990) op. cit .
50 Kamakura and Novak (1992) op. cit .
51 Novak and MacEvoy (1990) op. cit .5 Higgins, K. T. (1998) ‘Generational marketing ’,
Marketing Management ,V o l .7 ,F a l l ,p p .6 –9.
6 Rice, F . (1995) ‘Making generational marketing
come of age ’,Fortune , June 26, pp. 110 –114.
7 Smith, W . J. and Clurman, A. (1997) ‘Rocking the
Ages ’, Harper Business, New York, NY .
8 Schewe, C. D. and Meredith, G. E. (1994) ‘Digging
deep to delight the mature adult consumer ’,
Marketing Management ,V o l .3 ,N o .3 ,p p .2 1 –35.
9 Schewe, C. D., Meredith, G. E. and Noble, S. M.
(2000) ‘Defining moments: Segmenting by cohorts ’,
Marketing Management ,V o l .9 ,F a l l ,
pp. 48 –54.
10 Noble, S. M. and Schewe, C. D. (2003) ‘Cohort
segmentation: An exploration of its validity ’,Journal
of Business Research , Vol. 56, pp. 979 –987.
11 W ells, W . D. and Gubar, G. (1966) ‘Life cycle
concept in marketing research ’,Journal of Marketing
Research ,V o l .3 ,N o v e m b e r ,p p .3 5 5 –363.
12 Kotler (2003) op. cit .
13 McAlexander, J. H. (1991) ‘Divorce, the disposition
of the relationship and everything ’, in Holman, R.
and Solomon, M. R. (eds) ‘Advances in Consumer
Research ’, Vol. 18, Association for Consumer
R e s e a r c h ,P r o v o ,U T ,p p .4 3 –48.
14 Schewe and Meredith (1994) op. cit .
15 Andreasen, A. R. (1984) ‘Life status changes and
changes in consumer preferences and satisfaction ’,
Journal of Consumer Research ,V o l .1 1 ,N o .3 ,
pp. 784 –94.
16 Kotler (2003) op. cit . p. 293.
17 Wilkes, R. E. (1995) ‘Household life-cycle stages,
transitions and product expenditures ’,Journal of
Consumer Research ,V o l .2 2 ,N o .1 ,p p .2 7 –42.
18 Noble and Schewe (2003) op. cit .
19 Andreasen (1984) op. cit .
20 American Association of Retired Persons (1995)
‘Modern maturity study findsfifties the most
turbulent decade in life ’,AARP News , 13th
December, AARP , W ashington, DC.
21 Silvers, C. (1997) ‘Smashing old stereotypes of
50-plus America ’,Journal of Consumer Marketing ,
Vol. 14, No. 4, pp. 303 –309.
22 Kamakura, W . A. and Novak, T. P . (1992)
‘Value-system segmentation: Exploring the meaning
of LOV ’,Journal of Consumer Research ,V o l .1 9 ,
N o .1 ,p .1 1 9 –132.
23 Novak and MacEvoy (1990) op. cit .
24 Mayer, Karl U. and Tuma, N. B. (1990). ‘Life
course research and event history analysis: Anoverview ’,i nM a y e r ,K .U .a n dT u m a ,N .B .( e d s )
‘Event History Analysis in Life Course Research ’,
The University of Wisconsin Press, Madison, WI.
25 Abeles, R. P ., Steel, L. and Wise, L. L. (1980)
‘Patterns and implications of life course organization:
Studies from project talent ’, in Baltes, P . B. and
B r i m ,O .G .( e d s ) ‘Life-Span Development and
Behavior ’,V o l .3 ,A c a d e m i cP r e s s ,N e wY o r k ,N Y ,
pp. 307 –337.
26 Cole, C. A. and Gaeth, G. J. (1990) ‘Cognitive and
age-related differences in the ability to usenutritional information in a complex environment ’,
/H17015Palgrave Macmillan Ltd 1479-1862/06 $30.00 Vol. 14, 2, 115 –128Journal of Targeting, Measurement and Analysis for Marketing 127Life-changing events and marketing opportunities

Birth of first grandchild*
Major con flict with family member
Retirement (at own will)*Lost job/business or forced to retire*Started work for the first time or after
not working for a long time*
Reduction in hours of employment or
giving up employment (at own will)*
Signi ficant success at work or personal
life
Change jobs, same or different type*Major improvement in financial status
Financial status a lot worse than usual*Family member ’s health a lot worse
More responsibility for aged relative*Gained a lot of weightChronic illness or condition diagnosedSerious injury, illness or major surgeryCommunity crisis or disaster (hurricane,
crime, fire,flood, earthquake, etc)
Death or loss of a pet (dog or cat)Stopped smoking
Asterisk (*) indicates item also used as
anticipated life event.52 Frank, R. E., Massy, W . F . and Wind, Y . (1972)
‘Market Segmentation ’, Prentice Hall, Englewood
Cliffs, NJ.
53 Novak and MacEvoy (1990) op. cit .
54 Schiffman, L. G. and Sherman, E. (1991) ‘Value
orientations of new-age elderly: The coming of anageless market ’,Journal of Business Research ,V o l .2 2 ,
N o .2 ,p p .1 8 7 –194.
55 Neugarten, B. L. and Neugarten, D. A. (1986)
‘Changing meanings of age in the aging society ’,I n
Pifer, A. and Bronte, L. (eds) ‘Our Aging Society
Paradox and Promise ’,W . W .N o r t o n&C o m p a n y ,
New York, NY , pp. 33 –51.
APPENDIX: LIFE EVENTS
EXPERIENCED ANDANTICIPATED USED FORCLUSTERING
Life events
Moved to a different place*Marriage*B i r t ho ra d o p t i o no fac h i l d *Divorce or separation*Last child moved out of household*D e a t ho fs p o u s eDeath of a parent or close family
member*
128 Journal of Targeting, Measurement and Analysis for Marketing Vol. 14, 2, 115 –128/H17015Palgrave Macmillan Ltd 1479-1862/06 $30.00Mathur, Lee and Moschis