Documente Academic
Documente Profesional
Documente Cultură
REFERENCES
Linked references are available on JSTOR for this article:
http://www.jstor.org/stable/3150854?seq=1&cid=pdf-reference#references_tab_contents
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 14.139.242.60 on Tue, 12 Jan 2016 10:09:19 UTC
All use subject to JSTOR Terms and Conditions
RICHARDD. NORDSTROMand JOHN E. SWAN*
Journalof MarketingResearch
Vol. XIII (May 1976), 173-7
This content downloaded from 14.139.242.60 on Tue, 12 Jan 2016 10:09:19 UTC
All use subject to JSTOR Terms and Conditions
174 JOURNALOF MARKETINGRESEARCH,MAY 1976
May [13] showed that there was no difference in time. In addition, May [13], in a study of new car
loyalty between new car buyers on the basis of the purchases,found a direct relationshipbetween repur-
brandof car purchased.Therefore, it was determined chases of a brand and previous ownership of that
that use of a control dealer from the same city would brand. May's findings have been validated by two
prevent differences in economic variables, demo- recent doctoral dissertations [9, 15]. All three re-
graphic variables, and competitive variables which searchersfound that roughly 75%of the people who
would be likely if the control group were selected had made consecutive purchasesof a brandcontinued
from a different community. to purchase that brand on their next opportunityto
The cooperating Chevrolet agency was selected select a new car.
because of manysimilaroperatingcharacteristicssuch
as volume, numberof employees, and length of time METHOD
in the same community. Analysis of the data from
the questionnairesfurther supported the similarities Natural Experiment
between the control and experimentalgroups. Sample A natural experiment was used to obtain data
size variations reflected the sales ratio within the suitablefor examinationof consumerbrandand dealer
county. Response rates were nearly identical, 42.4% loyalty patterns [16]. Patternsof loyalty were estab-
for the control group and 45% for the experimental lished by respondents before any change had taken
group. Loyalty rates for the two groups were similar. place in the marketplacestructure. Identification of
Table I shows the distributionsof loyalty. A Kolmo- the potential respondentswas made possible by free
gorov-Smirnovtest showed that the two groups were access to records from two cooperating new car
similar as the maximum D was .089 for the brand agencies.
loyalty distribution and .068 for the store loyalty The Ford agency was sold in 1970. Therefore, only
distribution.The probability of difference was less buyers who had shown loyalty to one of the two
than .01 for either distribution. cooperatingfirms during the period 1960-1970 were
Automobiledealers are well suited for a study of included in the investigation. Loyalty was defined
the effects of a change in ownershipon loyalty. The operationallyas evidence of two or more purchases
retail channel throughwhich most new cars are sold from one agency as determined by an examination
in Americais somewhatunusual-at least in compari- of the part of the data base concerning sales. People
son with otherretailingarrangements.New car dealers who were qualifiedwere sent a questionnaireto obtain
normallyoffer one or two brands under a franchise their purchase history. Sales records only provided
grantedby one manufacturer.Thus, when two firms informationon purchasesfrom two sources.
are comparedthere need be little concern over varia- Questionnaires were color coded to signify how
tions in the size of assortment offered the public. many cars had been purchased from one of the
Further, the influence of a change in ownership on agencies. Thus, a yellow questionnairewas recognized
brand loyalty and store loyalty can be determined as coming from a person who had purchased three
much more clearly in a retailing arrangementwhere Chevrolets from the cooperating Chevrolet agency.
a limited numberof brandsare sold. If the history reported by the respondent disclosed
actual purchases of four Chevrolets, this response
HYPOTHESES was categorized as 100%brandloyal and 75% dealer
The first underlyinghypothesis tested was a null loyal. The purchasehistory also providedinformation
hypothesis that before the change in the Ford dealer- about the respondent'sactivity in the new car market
ship's ownership there was no difference between after the date of sale of the Ford agency.
membersof the control group (Chevroletbuyers) and When any business changes hands, the new owners
membersof the experimentalgroup (Ford buyers) in do not usually expect to retain all past customers.
termsof the frequency of occurrenceof variousbrand In the situation examined, the impact of the sale on
and dealer loyalty patterns. Because the two new car old customerswas believed to be about normal.There
agencies shared many common features, it was be- were no pressures on either party to the transaction.
lieved neither firm would be likely to attract a large As in many sell-out situations,every effort was made
groupof customerswho woulddisplayuniquepatterns to encourageold customersto supportthe new owners.
of loyalty. Old managementsent letters, newspapercoveragewas
The second basic hypothesis was that people who adequate, and the new owners voiced their intention
had demonstratedhigh loyalty before the change in to retain all emiployees, all policies affecting cus-
ownership would be more likely to maintain their tomers, and a consistent level of customer service.
loyalty after the change in ownershipthanpeople who Thus, customers knew they would be able to deal
had shown less loyal behaviorbefore the change. This with the same people, could expect the same service,
hypothesis was based on basic learning theory plus and could expect the same procedures as before.
the work of other researchers [17] who found that Consequently, observed changes in loyalty patterns
loyalty patternstend to become more consistent over couldbe attributedlogicallyto the impactof the change
This content downloaded from 14.139.242.60 on Tue, 12 Jan 2016 10:09:19 UTC
All use subject to JSTOR Terms and Conditions
CHANGE WHEN A NEW CAR AGENCY IS SOLD?
DOES CUSTOMERLOYALTY 175
in dealership ownership rather than to changes in pinpoint any problems in obtaining desired informa-
personnel, policies, or procedures. tion. Two modest revisions were made as a result
of this procedure. After revision, a second pretest
GeographicSetting was conductedwith 10people who had been customers
The study was conducted in a Kansas community of a dealership 350 miles from the study area. No
of approximately16,000people. The city is in a county further alterations were deemed necessary after the
with 24,000 populationand an effective tradeterritory second pretest.
of roughly 40,000. Eight new car dealers operate in
the county. As is the case with national car sales, Non-ResponseBias
Ford and Chevrolet account for the bulk of new car Non-response bias was not considered a major
registrationsin the county. At the time of the study problem because respondents did not appear to be
only the Ford agency had been sold. The new owners differentupon comparisonwith non-respondents.Two
of the Ford agency were well known in the area. demographiccharacteristicswere compared, age and
education. They were selected because past research
Time had identified these two variables, more consistently
The sale of the Ford franchise and the collection thanany others,as beingcorrelatedwith brandloyalty.
of data were separated by three years. This period Because these two importantcharacteristicsseemed
of time was long enough to permit a majorityof the uniform for both respondents and non-respondents,
subjects to re-enterthe new car market. In fact, 67% the 44%response rate was considered adequate [20].
of all respondents actually had purchaseda new car
in the period 1970-1973. The tables summarize the RESULTS
data obtained only from those people who had made The first hypothesis was accepted. There was no
a purchaseafter the date of sale of the Ford agency. discernible difference between the groups in terms
Thus the analysis is relieved of a potential criticism of loyalty before the sale of the Ford agency. Table
of use of intentions to buy, which may or may not I shows the distributionof loyalty toward the brand
coincide with actual purchases. and toward the dealership. The figures are recorded
fromresponses of personswho actuallyhad purchased
Pretest a new car after the sale of the Ford agency. The
Pretest of the questionnaire was conducted with rate of reported purchase was consistent with the
a convenience sample of 38 adults who had purchased actual sales of the county. Acceptance of this
a new car. This test permittedan examinationof the hypothesis made it possible to conclude that both
informationobtained from the testing instrument to agencies were patronized by customers with similar
Table 1
RESPONDENTSCATEGORIZEDBY LOYALTY
BEFORE1970
Table 2
BRAND CHOICE BY PRIORLOYALTY
Ford Chevrolet
Less than Less than Less than Less than
1005% 100K%X 100%% 100%% 100%5 100 10O% 100%
Action brand brand store store brand brand store store
after 1970 loyal loyal loyal loyal loyal loyal loyal loyal
Bought same brand 29 9 22 7
Bought other brand 32 13 9 10
Total 61 22 31 17
Bought from same dealer 6 19 11 14
Bought from other dealer 23 35 5 18
Total 29 54 16 32
This content downloaded from 14.139.242.60 on Tue, 12 Jan 2016 10:09:19 UTC
All use subject to JSTOR Terms and Conditions
176 JOURNALOF MARKETINGRESEARCH,MAY 1976
loyalty dispositions. Therefore, it would be logical before 1970actually remainedloyal to the dealership
to assume that differences in behavior of these cus- after the sale at a higherrate than did the respondents
tomers after the sale of the Ford agency were not who had been 100%loyal before the sale. Thus, it
due to underlyingdifferencesin basic attitudestoward was concluded that the influence of a change in the
loyalty. Thus, an introduction of a change in the dealership's ownership adversely affected ability to
marketplaceenvironment,change of ownership, can predict continued patronageof store or brand. This
be imbuedwith causalityof any subsequentalterations conclusionaccountsfor problemswhicharisein efforts
in loyalty patternsin the experimentalgroup. to predict continuedpatronageof a store or purchase
The second hypothesis introducedthe proposition of a brandwhen the basis for predictionrelies heavily
that stronger past loyalty would be a predictor of on past loyalty behavior without regardfor environ-
a greaterprobabilityof future loyalty. Table 2 shows mentalfactors affecting loyalty.
that the Chevrolet (control) group reacted as would It was not possible to conclude that the impact of
be predicted from this hypothesis. Of the Chevrolet the change of a market structure variable had been
respondentswho had displayed 100%loyalty before greater on either type of loyalty. Very clearly,
1970, 71% remained loyal. This finding conforms to however, a significant change in loyalty behavior
rates of new car loyalty providedby other researchers emerged after the shift in ownership. It should be
[9, 13, 15]. However, the Ford(experimental)group's noted, perhapsparenthetically,thatrespondentscould
loyalty was significantly lower after the change in give evidence of 100%brandloyalty and not be 100%
the ownership of the Ford dealership. Only 47% of store loyal. However, it was impossible to be store
the previously 100% brand loyal Ford respondents loyal and not also brand loyal. This restriction was
remained loyal after the agency was sold. Ford re- inherentin the franchise system of new car distribu-
spondents who faced a change in their environment tion.
did not react as would be expected from observations
based on purchasebehaviorassociated with historical DISCUSSION
patterns of buying behavior. The general conclusion As was noted, the situation studied was not a
drawnfrom this unexpected reduction in loyalty was situationof extremes. No bankruptcy,conflict, or any
that a change in ownership had caused the purchase other abnormal feature surroundedthe sale of the
patterns of previously loyal Ford buyers to undergo agency. Therefore, it was assumed that any changes
some alteration. Statistical support was given this in the behavioralpatternscould be associated directly
conclusionby applicationof a Fisherexact probability with the change in ownershipand were not a reaction
showingthatthe two groups(controland experimental) to some other element present in the transaction.The
were not similar [19]. effects of the sale cannot be rated as above, equal
Dealer loyalty also was determinedfrom data ob- to, or below normal,as no "normal"has been estab-
tainedin the study(see Table3). As with brandloyalty, lished. It was definitely apparentthat the change in
the Chevrolet (control) group reacted as would be ownership resulted in altered patterns of loyalty.
expected. The higher dealer loyalty subjects before Membersof the experimentalgroup displayed much
1970were the higherdealerloyalty subjectsafter 1970. less propensityto remainloyal than did members of
The fact that the experimentalgroup did not react the control group. A summaryof data is presented
as would be expected reflects the fact that intensity in Table 4.
of priorloyalty is not an absolutepredictorof behavior. Perhaps the primaryfinding of this research con-
Respondentsin the experimentalFord groupwho had cerns the learning constructs which have been the
displayed less than 100% loyalty to the dealership basis for prediction of a consumer's current choice
of a product given his past purchases. Predictionof
Table 3 behavior on the basis of past satisfaction was found
SELECTIONSMADE BY 100% LOYALRESPONDENTS to be reliable only for the control group and not for
AFTER1970 the experimentalgroup. One conclusion drawn from
this discovery was that currentbrandchoice behavior
Bought Bought Bought
can be related to prior brandloyalty if, and only if,
same same other marketplacefactors congruent to the choice process
brand brand brand remain constant over time. Alterationof any market
from from from variableis likely to upset the probabilityof continued
same other other
Loyalty group dealer dealer dealer loyalty. Dealerownershipwas shownto be a significant
influenceon loyalty. In particular,a change in owner-
100% brand loyal before 1970
Ford (61) 21 8 32 ship had an impact on brandand source selections.
Chevrolet (31) 20 2 9 Findings developed from this research show that
100% dealer loyal before 1970 future studies of loyalty toward a brand or a store
Ford (29) 6 8 15 should include an identification of market structure
Chevrolet (16) 11 0 5 variableswhich may influencethe consumer'sactions.
This content downloaded from 14.139.242.60 on Tue, 12 Jan 2016 10:09:19 UTC
All use subject to JSTOR Terms and Conditions
CHANGE WHEN A NEW CAR AGENCY IS SOLD?
DOES CUSTOMERLOYALTY 177
Table 4
SUMMARYOF OBSERVATIONS
Controlgroup Experimental
Observations (Chevrolet) group (Ford)
1. Of the respondents:the people who had purchaseda new car of any makeafter the sale of the
Ford dealershipwere 42.4% 45%
(48) (84)
2. Of the people in 1:
those who had been 100%obrandloyal before the sale were 64.6% 73.5%
those who had been 100%store loyal before the sale were 33.3% 35%
The data above indicate that there was no difference between the two groups surveyed in terms of loyalty before
the sale of the Ford agency.
3. Of those who had been 100l%brandloyal before the sale:
those who remainedbrandloyal after the sale were 71% 47%
4. Of those who had been 100%store loyal before the sale:
those who remainedstore loyal after the sale were 69% 20.7%
those who remainedbrandloyal after the sale were 69% 48%
those who did not retainloyalty to either the brandor the store were 0% 57%
Conclusion: A change in the marketingstructurevariable, ownership, had a significanteffect on behavioralpatterns
of customers. This effect was reflected in a shifting of loyalty patterns among members of the experimental
group.
At least, future researchers should be aware of the 7. Day, George S. "A Two Dimensional Concept of Brand
possible effects on loyalty which could be induced Loyalty," Journal of Advertising Research, 9 (1969),
by a change in any major market structure variable 29-35.
such as ownership. 8. Enis, Ben. "Store Loyalty as a Basis for Market
An extension of this work might be a study of the Segmentation," Journal of Retailing, 46 (Fall 1970),
42-56.
significance of market structure variables on retention 9. Gieselan, Robert Dale. "Automobile Brand Loyalty,"
of loyalty. The effect of other market structure varia-
unpublished doctoral dissertation, University of Illinois,
bles on loyalty also could be considered. A study 1970.
of the effects of changes in market structure variables 10. Guest, Lester. "Brand Loyalty Revisited: A Twenty
on industrial buyers might be worthwhile. Year Report," Journal of Applied Psychology, 48 (1964),
Practical implications were gleaned from the project. 93-7.
It may be very important, from the dealer's viewpoint, 11. Hansen, Flemming. Consumer Choice Behavior. New
to seek continuity of operation as a means of holding York: The Free Press, 1972.
loyal buyers. Manufacturers will observe that the study 12. Jacoby, Jacob, and David B. Kyner. "Brand Loyalty
vs. Repeat Purchase Behavior," Journal of Marketing
findings reinforce the value of emphasis on consistent
dealer representation. The findings also highlight a Research, 10 (February 1973), 1-9.
13. May, Frederick. "Adaptive Behavior in Automobile
need to recognize that loyalty is dependent on factors Brand Loyalty," Journal of Marketing Research, 6
outside the physical product. (February 1969), 62-5.
14. McConnell, J. Douglas. "Development of Brand Loyal-
REFERENCES
ty: An Experimental Study," Journal of Marketing
1. Alport, Mark, II. "Personality and the Determimants Research, 5 (January 1968), 13-9.
of Product Choice," Journal of Marketing Research, 15. McHale, Henry Patrick. "Correlates of Automobile
9 (February1972),89-92. Brand Loyalty," unpublished doctoral dissertation,
2. Bass, Frank, and W. Wayne Talarzyk. "An Attitude University of Illinois, 1971.
Model for the Study of Brand Preference," Journal 16. Nordstrom, Richard D. "Analysis of Store and Brand
of Marketing Research, 9 (February 1972), 93-6. Loyalty Behavior: An Ex Post Facto Experiment In-
3. Belk, RussellW. "An ExploratoryAssessmentof Situa- volving New Car Purchases," unpublished doctoral
tional Effects in Buyer Behavior," Journal of Marketing dissertation, University of Arkansas, 1974.
Research, 11 (May 1974), 156-63. 17. Rothbert, Robert. "Consumer-Retailer Loyalty," Jour-
4. Benson, PurnellH. "A Modelfor Analysisof Consumer nal of Retailing, 47 (Winter 1971), 72-82, 95.
Preference," Journal of Applied Psychology, 39 (1955), 18. Sheth, Jagdish N. "Factor Analytic Model of Brand
375-81. Loyalty," Journal of Marketing Research, 5 (November
5. Carman,JamesM. "Correlatesof BrandLoyalty: Some 1968), 395-404.
Positive Results," Journal of Marketing Research, 7 19. Siegel, Sidney. Non-Parametric Statistics. New York:
(February1970),67-76. McGraw-Hill, 1956.
6. Cunningham,Ross M. "CustomerLoyalty in Store and 20. Simon, Julian L. Basic Research Methods in Social
Brand," Harvard Business Review, 39 (January-Fe- Science. New York: Random House, 1969.
bruary1961), 127-38.
This content downloaded from 14.139.242.60 on Tue, 12 Jan 2016 10:09:19 UTC
All use subject to JSTOR Terms and Conditions