You are on page 1of 14

Measuring Consumer Involvement Profiles

Author(s): Gilles Laurent and Jean-Nol Kapferer


Source: Journal of Marketing Research, Vol. 22, No. 1 (Feb., 1985), pp. 41-53
Published by: American Marketing Association
Stable URL: http://www.jstor.org/stable/3151549
Accessed: 18-03-2015 04:34 UTC
REFERENCES
Linked references are available on JSTOR for this article:
http://www.jstor.org/stable/3151549?seq=1&cid=pdf-reference#references_tab_contents
You may need to log in to JSTOR to access the linked references.

Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at
http://www.jstor.org/page/info/about/policies/terms.jsp
JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide range of content
in a trusted digital archive. We use information technology and tools to increase productivity and facilitate new forms of scholarship.
For more information about JSTOR, please contact support@jstor.org.

American Marketing Association is collaborating with JSTOR to digitize, preserve and extend access to Journal of Marketing
Research.

http://www.jstor.org

This content downloaded from 168.176.5.118 on Wed, 18 Mar 2015 04:34:04 UTC
All use subject to JSTOR Terms and Conditions

LAURENT
and JEAN-NOELKAPFERER*
GILLES
Thereis more than one kind of consumerinvolvement.Dependingon the antecedents of involvement(e.g., the product'spleasurevalue, the product'ssign or
symbolicvalue, risk importance,and prolability of purchaseerror), consequences
on consumerbehaviordiffer. The authorstlhereforerecommendmeasuringan involvementprofile,ratherthana singleinvolvement
level. Theseconclusionsare based
on an empiricalanalysisof 14 productcategories.

Measuring

Consumer

Involvement

The degreeof consumerinvolvementin a productcategory is now widely recognizedas a majorvariablerelevantto advertisingstrategy(Ray 1982;Rothschild1979;
Vaughn1980). Dependingon theirlevel of involvement,
individualconsumersdiffer in the extent of their decision processandtheirsearchfor information.Depending
on theirlevel of involvement,consumersmay be passive
or activewhen they receive advertisingcommunication,
and limit or extend their processingof this communication. To adaptto these differences, advertisersmay
considera numberof operationalvariablessuch as the
type of media, the degreeof repetition,the lengthof the
message, the tone of the message, and the quantityof
information(Tyebjee 1979). In practice,however, one
questionarisesfrequently:how can we know whethera
specific groupof consumersis indeed highly involved
in some productcategory?
Today,this questiongenerallyreceivesqualitativeassessmentfrom advertisingandproductmanagers.When
quantitativeindicatorsof involvementare used, the instruments
oftenboil downto a singlescale (Vaughn1980)
or to a single-item measure of perceived importance
(Agostini 1978; Hupferand Gardner1971; Lastovicka
and Bonfield 1982; Traylor1981). Shouldinvolvement
be reducedto a single dimension?Does "perceivedimportance"alone capturesall the richnessof the involvementconcept?Is it sufficientto classify people in terms
of a single involvementindicatoror shouldinvolvement
be analyzedin terms of multiplefacets, which need to
be measuredsimultaneouslyif one wants to provide

Profiles

managerswith a full pictureof the type of involvement


of a specific targetgroup?
Fifteenyears ago, in theirextensivereview of the involvementconcept, Kiesler, Collins, and Miller (1969)
calledit a pot-pourriconceptwhichmay encompassseveral independentelements. More recently Rothschild
(1979) concludedthatno single indicatorof involvement
couldsatisfactorilydescribe,explain,or predictinvolvement. In line with these remarks,we suggest that marketing researchersstop thinkingin terms of single indicatorsof the involvementlevel and instead use an
"involvementprofile"to specify more fully the nature
of the relationshipbetween a consumerand a product
category.
Ourobjectiveis to providemarketingand advertising
managerswith a scale specifyingthe natureandlevel of
consumerinvolvementthatis reliableand valid but also
convenient.Satisfyingthe conveniencecriterionimplies
thatthe itemsshouldmakesense for any productclassfromyogurtto bras, fromcolor TV sets to detergentsandthatthe totalnumberof items allows the scale to be
insertedat little extracost in a usage andattitudesurvey.
In the next sectionwe review the uses of the involvement concept, as revealed by the literatureand managers'interviewswith the authors.This review suggests
thatconsumersdiffer not only in level of involvement,
but also in type of involvement.Then we describe a
methodby which indicatorscan be developedfor each
type of involvement.Finally,dataanalysisprovidesevidence aboutthe reliabilityand validityof the indicators
as well as the usefulnessof thinkingin termsof involvement profile to predict selected aspects of consumers'
decisionprocessesand receptivityto advertising.

*Gilles Laurentand Jean-NoelKapfererare AssociateProfessors,


Ecoledes HautesEtudesCommerciales
(H.E.C.)andInstitutSuperieur
des Affaires (I.S.A.), Jouy-en-Josas,France. The orderof the authors'namesresultsfrom a randomdrawing.
Theauthorsthankthe anonymousJMRreviewersandthe Editorfor
theirhelpfulsuggestions.

INVOLVEMENT
ORINVOLVEMENTS?
Researchon consumerinvolvementgoes backto Sherif
andCantril's(1947) early work. Many authorshave re41
Journal of Marketing Research

Vol. XXII (February1985), 41-53


This content downloaded from 168.176.5.118 on Wed, 18 Mar 2015 04:34:04 UTC
All use subject to JSTOR Terms and Conditions

42

1985
JOURNAL
FEBRUARY
OF MARKETING
RESEARCH,

viewedthis field of consumerresearchandtheory(Arora


1982;Assael 1981;De Bruicker1979;Engel andBlackwell 1982;Ray 1973;Robertson1976). It is not our objective here to add anotherreview, but to focus on different facets or types of involvement.
In theory,involvementis consideredan individualdifferencevariable.It is a causalor motivatingvariablewith
a numberof consequenceson the consumer'spurchase
andcommunicationbehavior.Thus, dependingon their
level of involvement,consumerswill differgreatlyin the
extensivenessof their purchasedecision process (indiusedto comparebrands,
catedby the numberof attributes
the lengthof the choice process, and the willingnessto
reacha maximumor a thresholdlevel of satisfaction)or
in theirprocessingof communications(indicatedfor instanceby the extentof informationsearch,receptivityto
advertising,and the numberand type of cognitive responses generatedduring exposure) (Krugman1965,
1967).
The involvementliteratureandin-depthinterviewswith
advertisingmanagerssuggest that the hypotheticalconstruct"involvement"is not a unitaryone. Thereare differentviews of involvementas revealedby the uses of
the conceptand the conditionsimposedby differentresearchersto manipulateand measureit.
The Uses of the Concept
Researchersandpractitionerstendnot to use the word
"involvement"alone, but ratherimply a distinctionbetweentypes of involvement.For example, Houstonand
Rothschild(1977) make a distinctionbetween enduring
involvementand situationalinvolvement.The latterreflects concernwith a specific situationsuch as a purchase occasionor election. The former,stemmingfrom
the individual,reflectsa generalandpermanentconcern
with the productclass. The crucialdifferencebetween
these two types of involvementis suggestedby Rothschild(1979, p. 77): an individualmightusuallypurchase
variouslow-pricebrandsof liquorin a stochasticmanner
becauseof low enduringinvolvement;on the occasion
of a visit by the boss, however, a high involvementdecision would be madeto purchasea specific brand.Enduringinvolvementderivesfrom the perceptionthatthe
productis relatedto centrallyheld values (Arora1982),
those definingone's singularityand identity, one's ego
(OstromandBrock1968;Rokeach1968). Situationalinvolvementis heightenedwhen the consumerperceives
risk in a specific situation.
Anotherdifferentiationis subsumedby the practitioners' tendency to speak of "emotional involvement"
(Vaughn1980). Such a qualificationsupposesa contrario thattherecouldbe a non-emotional
involvement,such
as what Frenchsociologist Chombartde Lauwe (1979)
calls "rationalinvolvement,"devoid of any affect. For
instance,confrontinga choice of steam irons, the consumer would merely try to optimize a cost-benefit ratio,

withno emotionor interesttowardthe productcategory.

Pleasureis absent. This would not be the case for the


choice of a restaurant(Hirschmanand Holbrook1982).
A final differentiationis highlightedby authorswho
speakof "personalinvolvement"as thoughthere were
another,impersonal,kind of involvement.For instance
FrenchsemiologistBaudrillard(1968, 1970) posits that
"thereis involvementonly whenthereis sign." Looking
at some productalternatives,the consumerlooks for the
differencethatcorrespondsto his or herown identity,or
ego. When productchoice is perceivedas the sign of
oneself,involvementis present.In theirearlyworkSherif
andCantril(1947) madea similarrestriction.They spoke
of "ego involvement"to emphasizethe personal and
emotionalnatureof involvement.Greenwald(1965) proposes the term "solutioninvolvement"to denote the
commitmentof the consumerin the searchfor the right
solutionto a problem,and views this form of involvementas independentof ego involvementstemmingfrom
the individual'svery personaland centralvalues.
TheAntecedentsof Involvement
As a hypotheticalconstruct,involvementcannot be
measureddirectly. Looking at empiricalresearch,one
finds a great diversity in the operationalindicatorsof
involvement-furtherreflectingthe differencesin meaning of the constructfor differentresearchers.
Sherif and Hovland (1961) typically recruitedtheir
"highlyinvolved"subjectsamongWCTUwomen, emphasizingthree possible antecedentsof their involvement. Involvementcould stem from the "intrinsicimof an issue, its "personalmeaning"(Sherifand
portance"
Hovland1961, p. 197), a public standtaken, or strong
affect vis-d-vis an issue (Kiesler, Collins, and Miller
1969).
Workingin the cognitivedissonanceparadigm,Zimbardo(1960) experimentallymanipulatedinvolvement.
In choice or attitudechange experiments"highly involved"subjectswereled to believethatthey wouldhave
to make a public stand on their opinion in front of a
groupof spectators."Low involved" subjects, on the
contrary,perceivedtheir choice or opinions as incontypically manipsequential.Such an operationalization
ulateda perceivedriskantecedentof involvement(Chaffee andMcLeod1973)andspeciallya psychologicalrisk
relatedto the image one mightproject.
In marketing,price is probablythe most commonly
used indicatorof involvement.Because the risks of a
mispurchasearehigh when price is high, consumersare
likely to be involved(Rothschild1979). Durablegoods
also havebeenused to createconditionsof high involvementbecause,in case of mispurchase,one is stuckwith
a poor productfor a long time. Among those goods,
dresses are generallyconsideredas extremely ego-involvingbecauseof theirsymbolicmeaningvis-d-visrelevantothers,theircapacityto expressone's lifestyle or
personality(Levy 1959), or their hedonic character
(Hirschmanand Holbrook1982).

This content downloaded from 168.176.5.118 on Wed, 18 Mar 2015 04:34:04 UTC
All use subject to JSTOR Terms and Conditions

43

PROFILES
CONSUMER
INVOLVEMENT
MEASURING

EmpiricalData
Empiricaldataalso highlightthe necessityof thinking
in termsof differenttypesof involvement.For example,
Lastovickaand Gardner(1979) asked their subjectsto
evaluate14 productson a series of items measuringimportance,commitment,and affect. Their analysis revealed three types of products:low involvement,high
involvement,and special interestor enthusiastproducts
(productsexpressingone's hobby). The differencebetweenthe two last types lay in the presenceof affect and
hedoniccharacterin the lattercase.
To summarize,our review of uses and indicatorsof
involvementand of empiricalresearchsuggeststhatthe
(permanentor situational)state of "involvement"may
stem from differenttypes of antecedents.Frequentuse
of the word"involvement"with a qualifier(personalinvolvement, emotionalinvolvement,etc.) suggests that
the sourceof involvementis importantinformationand
thatresearchersor managersshouldnot be contentwith
knowing only that an individualis or is not involved.
Knowingthe level of involvementoffersa staticdescription. Understandingof the sourcesof involvementprovides a dynamicpictureof the consumer'ssubjectivesituationandgives clues as to whatappealsshouldbe used
in communicatingwith consumers.
Beyondcontroversiesover definitionsof involvement,
ourreviewof currentresearchandpracticesindicatesfive
antecedents,or facets, of involvement.
1. The perceivedimportanceof the product(its personal
meaning).
2. The perceivedrisk associatedwith the productpurchase,
which in turnhas two facets (Bauer 1967):
-the perceivedimportanceof negativeconsequencesin
case of poor choice and
-the perceivedprobabilityof makingsuch a mistake
3. The symbolic or sign value attributedby the consumer
to the product,its purchase,or its consumption.This differentiatesfunctionalrisk from psychosocialrisk (Bauer
1967)
4. The hedonicvalue of the product,its emotionalappeal,
its abilityto providepleasureand affect.

Insteadof developing a composite of items tapping


these differentsources to obtain a single index of involvementlevel, it seems essentialto keep the full pictureof the natureof consumerinvolvementby measuring the consumers'positionon eachof thesefive facetsthus providingtheir involvementprofile.
Beforeturningto the proposedmeasurementmethod,
we shouldnote thatthereis not a one-to-onecorrespondencebetweenthe facets of the involvementprofileand
Houstonand Rothschild'sdistinctionbetween enduring
and situationalinvolvement.Two facets of the involvementprofilecorrespondto enduring,nonsituational,aspects of the consumer'srelationshipto a product:the
of theproductandits hedonicvalue.
perceivedimportance
Two otherfacets, however,aremoredifficultto classify:

the perceivedrisk associatedwith the productand the


sign valueattributedto the product.Certainproductsentail a risk in all circumstances(e.g., a vacuumcleaner),
whereasfor otherproductsthe risk dependson the situation(e.g., a wine to be drunkalone or with the boss).
The formercase couldbe describedas enduringinvolvement, the latter as situationalinvolvement. Similarly,
certainproductsmay have an enduringsymbolicvalue,
whereasotherproductsmay have a symbolicvalue only
in the presenceof relevantothers.
METHOD
Ourobjectivewas to createa reliableand valid measure for each of the facets of involvement.Following
Churchill's(1979) suggestions, once the facets were
identified,we generateda pool of items for each facet.
Sourcesof these items were twofold, a literaturereview
andin-depthinterviewsof a sampleof housewives.Three
surveyswere necessaryto purifythe measuresand obtain five scales that would be satisfactorypsychometrically but also shortenough to be of practicaluse. Two
preliminarydatacollectionwaves wherecompletedwith
samplesof about100housewives,eachpersonbeingasked
aboutseveralproducts.The resultsreportedhereafterare
based on the third wave, for which a sample of 207
housewiveswas recruitedon the basis of age and socioeconomicquotas.Face-to-faceinterviewingwas done at
home. Eachhousewifewas interviewedon two product
categories,with a systematicrotationof productcategoriesby interviewee.Thus, the dataanalysiswas based
on 414 cases. Fourteenproductcategorieswere studied.
To be qualifiedfor the interview,in additionto meeting
the sociodemographiccriteriaof the quotas, the housewife had to be a consumerof the two products.
Each facet of involvementwas measuredby a multiitemscalewitha 5-pointLikert-type
responseformat(fully
disagreeto fully agree). Table 1 lists some items correspondingto each facet thatwere used in the thirdand
final datacollection.
In creatingconditionswherebythe facetscould appear
independentif such were the case, the selection of the
stimulusproductswas crucial. Fourteenproductswere
selectedto representcontrastingprofiles on the dimensions of perceivedsign value, perceivedhedonicvalue,
perceivedrisk, and perceivedimportance.These products were suggestedby qualitativein-depthinterviewing
of housewives;for each dimension,the housewiveswere
asked what typical productcame to mind among four
categories(food, durables,textile, and drugs). For instanceTV sets, washingmachines,dresses,andbraswere
mentionedas high risk products(the consequencesof a
mispurchaseare great). Low-pricefrequentlypurchased
itemswere at the otherextreme.Whenaskedwhatproducts were devoid of any hedonic character,the housewives mentioneddetergents,vacuumcleaners,and irons.
For the housewivesthese productsare ties to household
chores. At the other extreme(high hedonic character),

This content downloaded from 168.176.5.118 on Wed, 18 Mar 2015 04:34:04 UTC
All use subject to JSTOR Terms and Conditions

FEBRUARY
JOURNALOF MARKETING
1985
RESEARCH,

44

Table 1
FACETS
MEASURES
OF THEPRESUMED
(translatedfrom originalFrenchitems)
Number
Code
of
items
name
Facet
4
Productperceived
Importance
importance
Perceivedimportanceof negative
consequencesof a
mispurchase
Subjectiveprobability of a mispurchase

Hedonicvalue of
the productclass

Perceivedsign value
of the product
class

Examples
of items
is very importantto me.
For me
does not matter.
Whenyou get a
Risk
, it's not a
importance
big deal if you
makea mistake.
When you get a
Risk
, it's hard
probability
to makea bad
choice.
I can't say thatI
Pleasure
particularlylike
Sign

You can reallytell


abouta personby
the
she
picks out.

manyfood itemscameto mind(wine, chocolate,yogurt,


jams, etc.) as well as perfumesand dresses. In termsof
sign value, dresses, bras, jeans, wines, perfumes,and
cars were mentionedspontaneously.
A choice had to be made among all the mentioned
products.In this selectionprocesswe droppedproducts
lackingsubstantialpenetration(e.g., only a minorityof
housewiveswearjeans) and productswhich presumably
were high (or low) on all facets (cars, perfume,paper
towels);suchproductswouldpreventthe facets fromappearingdistinct.Finally, we did not retainproductspurchasedessentiallyby the husband(cars). The final list
of products consists of washing machines, vacuum
cleaners, irons, TV sets, dresses, bras, detergents,
shampoo,facial soaps, toothpaste,oil, yogurt, chocolate, and champagne.As we show subsequently(Table
3), meanscoresof each productcategoryfor each facet
showed high correspondencewith the a priori judgments.
EMPIRICALANALYSISOF THE FACETS

We first examinethe qualityof each scale. Discriminantvalidityalso is assessed. Then special emphasisis
givento the analysisof the relationshipsbetweenfacets.
Finally,we look at predictivevalidityby focusingon the
relationshipsof the facets to consumers'decision processes and communicationreceptivity.
Evaluationof the Qualityof Each Scale
Two criteriawere used to evaluateeach scale, multiproductfit andreliability.The firstcriterionis very importantif one wants to build a tool appropriatefor any
product.Unfortunatelymany items that would fit well

in the case of, say, washingmachinesarefoundsilly by


the intervieweewhen appliedto yogurtand vice-versa.
two surveyswerenecessaryto prunea large
Consequently
initialitem base. Items were rejectedif they had a significantnumberof nonresponsesor don't-knowanswers
(and afternegativefeedbackfrom the field team about
the interviewees'actualreactionto them). At the third
data collection phase, all items met the first criterion.
The secondcriterionis Cronbach'salphameasureof internalconsistencyof a scale (CarminesandZeller 1979).
To makethe full instrumenteasy to use in commercial
marketstudies, we limited each scale to no more than
five items;forreliabilitypurposes,each scale hadno less
thanthreeitems. Despitethe small numberof items per
scale, the Cronbach'salphavaluesprovedsatisfactoryimportance.80, sign .90, pleasure.88, risk importance
.82, and risk probability.72.
Traitand DiscriminantValidity
Campbell(1960)andNunnally(1978) suggestthateach
scale shouldmeasurea single dimensionif it is consideredto have "traitvalidity."Discriminant
validityof each
scale representsthe distinctivenessof each scale vis-dvis others.It mightbe possible-despite differentnames
anditems and good alphavalues-for two scales to be
so correlatedthatthey cannotbe consideredas measuring differentconcepts.They wouldlack discriminantvalidity (Campbell1960). To test simultaneouslytraitand
discriminantvalidity, we undertooka factoranalysisof
the items using all 414 observations.To have traitvalidity, a scale should load on one and only one factor.
To have discriminantvalidity, a scale should not load
on the same factoras anotherscale.
Because the scales tap differentfacets of the same
concept,the factorsshouldnot be expecteda priori to
be orthogonal.With this in mind, we used an oblique
factor analysis. The eigenvalue criterionleads to four
significantfactors, reproducing66% of the total variance. Table 2 reportsthe loadingsof the items.
The loading patternsshow that each scale is singlefactored(traitvalidity). As predictedby theory (Bauer
1967), each dimensionof perceivedrisk loads on a factor. However,"perceivedimportanceof the product"and
the first dimensionof risk do not display discriminant
validity,but insteadload on the samefactor.Therefore,
in furtheranalyses, these items are merged to form a
single scale of seven items (resultingin a Cronbach's
alphaof .87). Its code name is "imporisk"-denoting
thatfor consumersto deem a productimportantis akin
to feeling that a mispurchasewould have high negative
consequences.
Fromthis dataanalysis,we concludethatthe involvementprofileshouldhave four distinctfacets.
1. Imporisk(the perceivedimportanceof the productand
the perceivedimportanceof the consequencesof a mispurchase).
2. The subjectiveprobabilityof a mispurchase.

This content downloaded from 168.176.5.118 on Wed, 18 Mar 2015 04:34:04 UTC
All use subject to JSTOR Terms and Conditions

45

PROFILES
INVOLVEMENT
CONSUMER
MEASURING

Table 2
OF THEITEMS
OF THE
ANALYSIS
FACTOR
OBLIQUE
FACETSa
INVOLVEMENT

Importance1
Importance2
Importance3
Importance4

Factor 1
.59
.56
.62
.74

Pleasure 1
Pleasure 2
Pleasure 3
Pleasure 4
Pleasure 5

Factor 2

Factor 3

Factor 4

-.73
-.68
-.82
-.67
-.58

Sign 1
Sign 2
Sign 3
Sign 4
.62
Risk importance1
.74
Risk importance2
.74
Risk importance3
Risk probability1
Risk probability2
Risk probability3
'Omittedloadingsare inferiorto .25.

.78
.94
.73
.77

.76
.64
.50

3. The hedonicvalue of the productclass.


4. The perceivedsign value of the productclass.

It appearsthatinvolvementcannotsimply be equated
withperceivedrisk. Ourresultsprovidea directandpositive responseto Chaffeeand McLeod's (1973) conclusion after their literaturereview: "Althoughperceived
risk appearsclearly a sufficient conditionfor involvement, it is problematicwhetherit is a necessaryone.
Therewouldseemto be a numberof morepositivesources
of involvement,such as rewardsinherentin the product
afterpurchase"(p. 389).
WhatAre the RelationshipsBetweenFacets?
Obliquefactorsarenot expectedto be independentbecause the facets belong to the same construct.The following table is the matrixof the correlationsbetween
facets, computedover all 414 observations(each facet
score is measuredby the scale describedbefore).

Imporisk
Risk probability
Sign
Pleasure

Imporisk

Risk
probability

Sign

.47
.40
.46

.16
.15

.53

A relationshipdoes exist betweenfacets. However, the


correlationsindicatethat one facet cannotbe fully predictedby another.It is not possible to pick up a single
index, for no single facet alone catches the richnessof
therelationshipbetweena consumeranda productclass.
The extent of correlationswarns that a consumermay
be high on one facet but low on another.Two scatterplots illustratethe relationshipsbetween facets of in-

volvement.Figure1 showshow the perceivedsign value


varies with the perceived importanceof consequences
(imporisk).Each point correspondsto the averagescale
scoresof a productcategory,computedover all responon thatproductcategory.We see that,
dentsinterrogated
their
visible
covariation,one facetcannotbe fully
despite
predictedby the other. For example, thoughthey have
similarsign values, chocolate and irons are perceived
differentlyin termsof the importanceof consequences.
Figure 2 illustratesthe relationshipbetween perceived
sign value and perceivedpleasurevalue, using average
scale scores of each productcategory.
These scatterplots suggestthe desirabilityof measuring the full involvement
profile of a consumerin a product categorybecause no facet alone summarizesconsumers'relationshipsto products.Table 3 describesthe
averageprofiles of the productcategorieson the four
facets. Thereis a good correspondencebetweenempirical data and the a priori judgmentsthat led to the selectionof this productsample.However,the profilesof
some productswarranta comment. The involvement
profileof washingmachinesdiffersfromthatof vacuum
cleanersor irons. Thereis pleasurevalue in purchasing
a washing machine for it liberatesthe housewives by
giving them free time, whereasthe latterproductsnecessitatethe housewives'presenceandevoke "bondage"
to householdchores.Devoidof anyhedonicor sign value,
vacuum cleaners create risk involvement. A vacuum
cleaneris an expensivedurableproductand in case of a
poor choice one is stuck with it for many years. Furthermore,there are many differentfeatureson the various brandsandthe consumermay not feel at all assured
of makinga good choice. Her subjectiveprobabilityof
mispurchaseis high. Facial soap position mirrorssoap
advertisingappeals.Thereis no mentionanymoreof the
washingpower,butratherthe fragranceandthe physical
andpsychologicalsensations.Facial soap advertisingis

Table3
INVOLVEMENT
PROFILESa
Importance Subjective
of
probability
Pleasure Sign
negative
of
consequences mispurchase value value
181
121
112
147
Dresses
Bras
117
115
106
130
111
118
106
109
Washingmachines
122
95
TV sets
112
100
112
70
78
Vacuumcleaners
110
72
103
95
76
Irons
120
125
125
109
Champagne
65
Oil
89
97
92
86
106
78
83
Yogurt
Chocolate
80,
89
123
75
90
96
103
81
Shampoo
94
95
95
105
Toothpaste
Facialsoap
82
90
114
118
79
82
56
63
Detergents
'Averageproductscore = 100.

This content downloaded from 168.176.5.118 on Wed, 18 Mar 2015 04:34:04 UTC
All use subject to JSTOR Terms and Conditions

46

JOURNAL
OF MARKETING
FEBRUARY
1985
RESEARCH,

Figure1
RELATIONSHIP
BETWEEN
FACETS
OF INVOLVEMENT:
RISKIMPORTANCE/SIGN

A
Risk
Import an4e

waAie
U

0 nad

.j^AA

0 T.V.

m Ai9
* c, kom

, -euv.I4J_V 4t?oumut

*4kampoo

d'uAd

* footkcpaJLe

*yogkwtt
*octcgejt
v

0* Loap

t c~oPtP

now verysimilarto perfumeadvertising.Marketershave


preventedsoap frombecominga low involvementproduct by playingon two conditionsof involvement,pleasureand sign value.
Naturally,the figures are averagesfor each product
category,based on small samples. Consumersdo vary
in theirperceptions.Intraproduct
typologies are recommendedto isolate the groupsof consumerswith homogeneousinvolvementprofiles.

SomeConsequencesof the InvolvementProfile


Theorypredictsthat involvementexerts a stronginfluenceon consumers'decision processesand information search.Becauseinvolvementis capturedbetterwhen
all its antecedentconditions(facets) are taken into account, it is useful to investigatethe influence of these
facetson consumerbehavior.
Thebehavioralconsequencesof involvementhavebeen

This content downloaded from 168.176.5.118 on Wed, 18 Mar 2015 04:34:04 UTC
All use subject to JSTOR Terms and Conditions

47

MEASURING
CONSUMER
INVOLVEMENT
PROFILES

Figure 2
BETWEEN
FACETSOF INVOLVEMENT:
RELATIONSHIP
PLEASURE/SIGN

PIQasure

dtae60

.T.v.

* crtoet *?

* A

* yogftkA

*wuaz
?

* Aik.m.

(h,A).?*

1
*0

ampaghe

* b4
P9t&dCI~4AL-

tootkp acte

aCLmVaaoV

e*oi

* dttveht
S,'g9

reviewed often (Assael 1981; Engel and Blackwell 1982;


Finn 1982; Robertson 1976). Traditional views hold that
highly involved consumers (Assael 1981, p. 84)
-seek to maximizeexpectedsatisfactionfrom their brand
choicethroughanextensivechoiceprocess(Chaiken1980),
e.g., comparingmanybrands,spendingtime, using multiple attributes,
-are informationseekers, activelylookingfor information
from alternativesources,

-are more likely to be influencedby referencegroups,


-are more likely to expresstheir lifestyle and personality
characteristics
in theirbrandchoice, and
-process communication
cognitivelyby goingthroughstages
of awareness, comprehension,attitude, and behavior
(Krugman1965; Rothschild1969).
In the context of our study, we selected the first two
behavioral consequences to assess the contribution of
measuringthe full involvement profile instead of a single

This content downloaded from 168.176.5.118 on Wed, 18 Mar 2015 04:34:04 UTC
All use subject to JSTOR Terms and Conditions

1985
JOURNAL
OF MARKETING
FEBRUARY
RESEARCH,

48

indicatorof involvementlevel. A scale of extensiveness


of choice process was built with three self-perception
Likert-typeitems (numberof attributesused in comparing brands,amountof time spent, degree of attention
exertedduringchoice). This 3-item scale resultedin a
highreliabilitycoefficient(Cronbach'salpha= .80). Informationseekingwas operationalized
by threeself-perthe
items
measuring tendency to keep permaception
informed
about the productclass, interest for
nently
articlesand TV programsabout the product,and pro-

pensityto look at advertisingin the productclass. The


reliabilitycoefficient(.60) indicatedthatthesethreeitems
did not actuallyconstitutea scale. Thussubsequentanalyses were done on each item separately.
To hypothesizethat involvementhas an influenceon
certainaspectsof consumerbehaviordoes not imply that
theseaspectsdependfrominvolvementalone. Otherexplanatoryvariablesmay be at work. For this reason, in
this sectionon predictivevalidity, we take into account
variablesotherthan involvementthat may be expected

Figure 3
A FACETOF INVOLVEMENT
BETWEEN
RELATIONSHIP
VALUE)AND A CONSEQUENCE
(PLEASURE
OF
THE
DECISION
(EXTENSIVENESS
PROCESS)

EXtervs iveneSS o

ihe. decision procass

. dxci

S u)AC-tIAA.m/

* Lv6udU~

eL

))ua&ICAtVL

* r.v.
0

* 4rWA9

* 4kokpDyo e
*0ot

k
foDtpaJ)

aa.,ftpa it

oap

*yoqkvtd
Pleature-

This content downloaded from 168.176.5.118 on Wed, 18 Mar 2015 04:34:04 UTC
All use subject to JSTOR Terms and Conditions

value

49

MEASURING
PROFILES
CONSUMER
INVOLVEMENT

to influence the two consequencesof interest:extensivenessof the choice process and informationseeking.
Omittingthese variableswouldresultin specificationerror. A review of previousempiricalresearchsuggested
the inclusion of two other variables,perceiveddifferences between alternativesand price. Perceiveddifferences act as a majorstimulusof choice and searchbehavior(Assael 1981; Claxton,Fry, and Portis 1974; De
Bruicker1979;Ray 1973;Rothschild1979). Whenprice
is high, the expectationof obtaininga betterprice jus-

tifies spendingmore time in the choice process and acandCuntively searchingfor information(Dommermuth
diff 1967; Kiel and Layton 1981; Newmanand Staelin
1972). A 3-item scale measuredthe perceived differentiationvariable(Cronbach'salpha = .71). Price was
measuredby the index of average retail prices of the
productcategory.We used the logarithmof the price to
reduce the skewness of the variable. Because we had
four separatedependentvariables,we ran four separate
regressionson all 414 observations.

Figure4
AND A CONSEQUENCE
A FACETOF INVOLVEMENT
BETWEEN
RELATIONSHIP
(RISKIMPORTANCE)
OF THEDECISIONPROCESS)
(EXTENSIVENESS

Extensiveness
of the decision
process

dju^UL uWu
?vacuum rar~.
* T..
rNmachJLA
9

JtiPP

agL .

*skajayoo
* tootkpa.t6r

* doa

* vedst e

* yogIkv

kisk

This content downloaded from 168.176.5.118 on Wed, 18 Mar 2015 04:34:04 UTC
All use subject to JSTOR Terms and Conditions

importance

JOURNAL
OF MARKETING
FEBRUARY
1985
RESEARCH,

50
Table 4 reports the results of the regressions. A major
conclusion of the analysis is that the facets of the involvement profile have different influences on the dependent variables. Sometimes one facet is determinant
and sometimes another facet exerts the major influence.
Analysis of the standardized regression weights shows
that the extensiveness of the decision process is influenced above all by the perceived importance of the product and by the negative consequences of a mispurchase.
The second variable influencing extensiveness is the degree of perceived difference between alternatives. Consumers have a tendency to keep permanently informed

when they perceive the product category as important,


or when it has sign value or pleasure value. Consumers
take an interest in articles and programs when the product has pleasure value and sign value. Propensity to expose oneself to advertising is dependent on the pleasure
value of the product class.
Interestingly, the importance facet does not affect all
aspects of communication behavior. The pleasure facet
influences communication behavior but has no influence
on the extensiveness of the choice process. The perceived probabilityof making a mispurchaseexerts a small
positive influence on the extensiveness of the decision

Figure 5
RELATIONSHIP
BETWEEN
A FACETOF INVOLVEMENT
AND A CONSEQUENCE
(RISKIMPORTANCE)
(LIKING
ADVERTISING)
i

L;king

cadvertising

* dMu6

* tun
CJA"1)pa9I.e

*co ecoto&te

* r.v.

*tootpate

Ojaca

* wa.bAng

*4karknpoo
* vc4teA XLJ'ItI

*yo9hwdi
0

dAekgtpd

dW
A
O~UI(IZ

0* 4uWL

Rtsk

This content downloaded from 168.176.5.118 on Wed, 18 Mar 2015 04:34:04 UTC
All use subject to JSTOR Terms and Conditions

importance

INVOLVEMENT
MEASURING
CONSUMER
PROFILES

51

Table 4
INFLUENCE
OF THEINVOLVEMENT
FACETS
regression
(sta.Jordized
weights)
Risk
(importance)
Extensive decision process

Keepingpermanentlyinformed
Interestin articlesandTV programs

Lookingat advertising
?p< 0.001.
bp< 0.01.

Sign
value

Pleasure
value

Risk
(probability)

.61'

.lOb

.00

.06c

.27a

.18'

.15b

.08

.13

.14b

.28'

.05

.06

.37a

Perceived
differentiation

Price
.ob

-.08

R2

.17

.71

.05

.28

.01

.03

.00

.20

-.04

.01

.00

.17

cp < 0.05.

Figure6
A FACETOF INVOLVEMENT
BETWEEN
RELATIONSHIP
AND A CONSEQUENCE
(PLEASURE
VALUE)
(LIKING
ADVERTISING)
i

Liking
advert;sing

*dfUd

edunmpagq

T.V.*
e tootUL1Od4e

*jOa--r

chocoa&t

*4lwipoa
*yoqY0kL
Y~O9IUVd

e~p4l&
* WLOL,

* oi

Pleasure value

This content downloaded from 168.176.5.118 on Wed, 18 Mar 2015 04:34:04 UTC
All use subject to JSTOR Terms and Conditions

JOURNAL
OF MARKETING
FEBRUARY
1985
RESEARCH,

52
process, but has no influence on the other dependent
variables.
Figures 3 through 6 are graphic illustrations of these
results. Each of them shows, on the basis of average
productscores, the influence of one facet of involvement
(abscissa) on a possible consequence (ordinate). They
suggest that the extensiveness of the decision process is
weakly influenced by a product's pleasure value (Figure
3), but strongly influenced by risk importance (Figure
4). In contrast,propensityto exposure to advertisingdoes
not depend much on risk importance (Figure 5), but derives mainly from the product's pleasure value (Figure
6).
DISCUSSION AND CONCLUSIONS
Looking at consumer behavior textbooks (Assael 1981;
Engel and Blackwell 1982), one sees that involvement
theory makes rather simple predictions on the effects of
involvement on consumer behavior. Typically, when
consumers are involved, they should engage in a number
of behaviors (active search, extensive choice process,
active informationprocessing, etc.); when consumers are
not involved, they should not engage in these behaviors.
Knowing the conditions that gave rise to involvement
has no role in the theory. Prediction of behaviors entails
knowing only the consumer's level of involvement. In
contrast, our research was prompted by the fact that
managers and researchers use the word "involvement"
with a qualifier, implying that the term used alone is too
imprecise unless one specifies what kind of involvement
is concerned. Here, we propose that the nuances in
meanings of involvement derive from differences in the
antecedent conditions producing involvement. The literaturereview suggested five such antecedent conditions
of involvement-perceived importance of the product or
the situation, perceived sign value, perceived pleasure
value, and perceived risk (itself divided in two subcomponents). A factor analysis indicated that though they
were correlated, each facet of involvement brought some
specific information. One could not capture the consumer's involvement through a single index; all facets of the
involvement profile must be taken into account simultaneously.
Regression analyses showed that all facets contributed
to the prediction of behavior. Also, some facets influence specific behaviors but not other behaviors. Therefore no precise prediction on the consequences of involvement could be made unless the antecedentconditions
were specified. Knowing the involvement level on one
facet (e.g., perceived importance, the classical indicator
of involvement) is not sufficient. The full profile must
be known because different facets have different influences on selected aspects of consumer behavior.
On practical grounds, the involvement profile can be
used to segment the market. Rather than merely indicating high-low involvement divisions of the market, the
profile allows identification of consumers high on some
facets but low on others. Moreover, the involvement

profile affords a better understandingof the dynamics of


consumer involvement. Looking at the facets, one may
understandbetter where involvement originates, which
provides clues as to what types of appeals should be used
in communication for each segment.
How does the involvement profile compare with FCB's
advertising planning matrix (Vaughn 1980)? FCB's approach is based on the plotting of products or people or
situations in two dimensions, involvement and think-feel
(whether the decision is based more on facts or more
on feeling). Because the items used to build the involvement index have not been published, it is difficult to assess whether FCB's involvement represents one of our
facets or a mix of them. The think-feel dimension is related to the weight of the sign and pleasure value facets
within the involvement profile. When one of these two
facets is strongly present, people should engage more in
"feel" decisions than in "think"decisions. However, the
think-feel dimension confounds these two determinants,
thus providing fewer clues to advertising managers looking for a clear picture of the involvement dynamics.
On theoretical grounds, Rothschild (1979) spoke of
involvement as "a vague concept" (p. 78) and Kiesler,
Collins, and Miller (1969) called it "a pot-pourri concept" (p. 279). Our results strongly suggest that, as it
stands now, involvement theory may be oversimplified.
Involvement does not systematically lead to the expected
differences in behavior. They depend on the antecedents
of involvement, as measured here by the involvement
profile. Some consequences depend on certain facets but
not on others. Therefore at a metatheoretical level, if
prediction of specific behavioral outcomes depends on
knowledge of the specific facets (risk, sign, pleasure,
importance), one may question the utility of thinking of
the gross concept called "involvement" and instead substitute an analytical distinction between the facets.
REFERENCES
Agostini, J. M. (1978), "Communication
publicitaireet implicationdu consommateur.Cons6quencespratiquespourla
conceptiondes messageset le choix des media,"IREP,Proceedingsof the 18th AnnualConference,79-86.
Arora,R. (1982), "Validationof an S-O-R Model for Situation, Enduring, and Response Componentsof InvolveResearch,19 (November),505ment,"Journalof Marketing
16.
Assael, H. (1981), ConsumerBehavior. New York: Wadsworth.
Baudrillard,Jean (1968), Le systemedes objets. Paris:Gallimard,CollectionTel.
(1970), La societedde consommation.Paris:SGPP.
Bauer,R. A. (1967), "ConsumerBehavioras Risk Taking,"
in Risk Takingand InformationHandlingin ConsumerBehavior, D. F. Cox, ed. Boston: HarvardUniversityPress.
Divisionof Research,GraduateSchool of BusinessAdministration,HarvardUniversity,23-33.
for APA Test
Campbell,D. T. (1960), "Recommendations
StandardsRegardingConstruct,Traitand DiscriminantValidity,"AmericanPsychologist, 15, 546-53.
Carmines,E. G. and R. A. Zeller (1979), "Reliabilityand

This content downloaded from 168.176.5.118 on Wed, 18 Mar 2015 04:34:04 UTC
All use subject to JSTOR Terms and Conditions

CONSUMER
INVOLVEMENT
MEASURING
PROFILES
ValidityAssessment,"Sage UniversityPaper.
Chaffee,S. H. and J. M. McLeod(1973), "ConsumerDecisions and InformationUse," in ConsumerBehavior:Theoretical Sources, S. Ward and T. S. Robertson,eds. EnglewoodCliffs, NJ: Prentice-Hall,Inc., 385-415.
Chaiken,S. (1980), "HeuristicVersusSystematicInformation
Processingand the Use of SourceVersusMessage Cues in
Persuasion,"Journalof Personalityand Social Psychology,
39 (5), 752-66.
Chombartde Lauwe, M. J. (1979), Un mondeautre: l'Enfance. Paris:Payot.
forDevelopingBetter
Churchill,G. A., Jr.(1979), "AParadigm
Measuresof MarketingConstructs,"Journal of Marketing
Research, 16 (February),64-73.
Claxton,J. D., J. N. Fry, andB. Portis(1974), "A Taxonomy
of Prepurchase
InformationGatheringPatterns,"Journalof
ConsumerResearch, 1 (December),35-42.
De Bruicker,F. S. (1979), "An Appraisalof Low-Involvement ConsumerInformationProcessing,"in AttitudeResearch Plays for High Stakes, J. C. Maloney and B. Silverman,eds. Chicago:AmericanMarketing
Association,11230.
Dommermuth,W. P. and E. W. Cundiff(1967), "Shopping
Goods, ShoppingCentersand Selling Strategies,"Journal
of Marketing,31 (October),32-6.
Engel, F. and R. D. Blackwell (1982), ConsumerBehavior,
4th ed. New York:The DrydenPress.
Finn, W. F. (1982), "Itis Time to Lay the Low-Involvement
Hierarchyto Rest," in Proceedingsof the Associationfor
ConsumerResearch, 13th Conference,99-102.
Greenwald,H. S. (1965), "TheInvolvementControversyin
PersuasionResearch,"unpublishedmanuscript,Columbia
University.
Hirschman,E. C. andM. B. Holbrook(1982), "HedonicConsumption:EmergingConcepts,Methodsand Propositions,"
Journalof Marketing,46 (Summer),92-101.
Houston,M. J. and M. L. Rothschild(1977), "A Paradigm
for Researchon ConsumerInvolvement,"WorkingPaper
11-77-46, Universityof Wisconsin-Madison.
Hupfer,N. T. and D. M. Gardner(1971), "DifferentialInvolvementwith ProductsandIssues:An Exploratory
Study,"
in Proceedingsof the Associationfor ConsumerResearch,
2nd Confterence,D. M. Gardner,ed. College Park, MD,
262-70.
Kiel, G. C. and R. A. Layton(1981), "Dimensionsof ConsumerInformationSeeking Behavior,"Journalof Marketing Research, 18 (May), 223-9.
Kiesler,C. A., B. E. Collins, and N. Miller (1969), Attitude
Change.New York:JohnWiley & Sons, Inc.
Krugman,H. E. (1965), "TheImpactof TelevisionAdvertising: earning Without Involvement," Public Opinion
Quarterly,29 (Fall), 349-56.
(1967), "The Measurementof AdvertisingInvolve-

53
ment,"Public OpinionQuarterly,30 (Winter),583-96.
Lastovicka,J. L. andE. H. Bonfield(1982), "Do Consumers
Have BrandAttitudes?"Journalof EconomicPsychology,
2, 57-75.
andD. M. Gardner(1979), "Componentsof Involvement," in AttitudeResearchPlays for High Stakes, J. C.
Maloneyand B. Silverman,eds. Chicago:AmericanMarketingAssociation,53-73.
Levy, Sidney J. (1959), "Symbolsfor Sale," HarvardBusiness Review, 37 (July-August), 117-19.
InforNewman,J. W. and R. Staelin (1972), "Prepurchase
mationSeekingfor New Carsand MajorHouseholdAppliances,"Journalof MarketingResearch, 9 (August), 24957.
Nunnally, J. C. (1978), PsychometricTheory. New York:
McGrawHill Book Company.
Ostrom,I. M. and T. C. Brock(1968), "A CognitiveModel
of AttitudinalInvolvement,"in Theoriesof CognitiveConsistency,R. P. Abelsonet al., eds. New York:RandMcNally.
Ray, M. L. (1973), "MarketingCommunicationand the Hierarchyof Effects," in Sage AnnualReviewsin CommunicationResearch,F. GeraldKlineandP. Clark,eds. Beverly
Hills, CA: Sage Publications.
(1982), Advertisingand Communication
Management,
EnglewoodCliffs, NJ: PrenticeHall, Inc.
ConsumerBeRobertson,T. S. (1976), "Low-Commitment
havior,"Journalof AdvertisingResearch, 16 (April), 1924.
Rokeach,M. (1968), Belief, Attitudesand Values. San Francisco: Jossey-Bass.
Rothschild,M. L. (1979), "AdvertisingStrategiesfor High
andLow InvolvementSituations,"in AttitudeResearchPlays
for HighStakes,J. C. MaloneyandB. Silverman,eds. Chicago: AmericanMarketingAssociation,74-93.
Sherif,M. and H. Cantril(1947), The Psychologyof Ego-lnvolvement.New York:JohnWiley & Sons, Inc.
and C. L. Hovland(1961), Social Judgment:Assimilationand ConstrastsEffects in Communicationand AttitudeChange. New Haven, CT: Yale UniversityPress.
Traylor, M. B. (1981), "ProductInvolvementand Brand
Commitment,"Journal of AdvertisingResearch, 21 (December),27-33.
Tyebjee,T. T. (1979), "Refinementof the InvolvementConcept: An AdvertisingPlanningPoint of View," in Attitude
ResearchPlaysfor High Stakes,J. C. Maloneyand B. Silverman,eds. Chicago:AmericanMarketingAssociation,94111.
Vaughn, R. (1980), "How AdvertisingWorks: A Planning
Model,"Journalof AdvertisingResearch,20 (October),2733.
Zimbardo,P. G. (1960), "Involvementand Communication
Discrepancy as Determinantsof Opinion Conformity,"
Journalof Abnormaland Social Psychology,60, 86-94.

This content downloaded from 168.176.5.118 on Wed, 18 Mar 2015 04:34:04 UTC
All use subject to JSTOR Terms and Conditions

You might also like