View metadata, citation and similar papers at core.ac.uk brought to you by CORE T he research reg ister for th is journ al is available at T h e cu rren t issue and fu ll tex t arch ive of this jou rn al is aprovidedv ailab le byat OAR@UM http://www.emeraldinsight.com/researchregisters http://www.emeraldinsight.com/0309-0566.htm

Service loyalty loyalty The effectsof service and the mediatingrole of satisfaction AlbertCaruana 811 Centre forCommunication Technology,University of Malta, ReceivedOctober 1999 Msida,Malta RevisedMay 2000; October2000 Keywords Loyalty,Service quality, , Banking Abstract Serviceloyalty, with its final effect on repurchasing bycustomers, appears tohave receivedrelatively little attention.This study starts by first delineating the concept ofservice loyaltyand proceeds to distinguish between service quality and customer satisfaction. A mediationalmodel that links servicequality to service loyalty via customer satisfaction is proposed.Appropriate measuresare identifiedand a postalsurvey is undertaken among1,000 banking .A response rate of20.5 per cent isobtained. Results indicate that customersatisfaction does play amediatingrole in theeffect ofservice quality on serviceloyalty. Theeffects of a number ofdemographic indicators on serviceloyalty are alsoreported. Implicationsare discussed,limitations of the study are notedand possible areas for further research are indicated.

Introduction Service loyalty,with its final effect onrepurchasing by customers, is perhaps oneof the most importantconstructs in services .Indeed, loyal customers thatindulge in repeat purchases are the bedrock of any business. Oneofthemore obvious questions relates tothe demographic characteristics of loyal customers,whether any such variables are more salient thanothers and howthese canbe usedfor segmentation purposes (e.g. Frank, 1967). However, workthat integrates the role ofservice loyaltywithin the context of other service marketingvariables like service qualityand customer satisfaction has received less attention. Service qualityhas been the subject ofconsiderable interest byboth practitionersand researchers inrecent years,spurred on bythe original work byParasuraman et al. (1985). Animportant reason for the interest inservice qualityby practitionersresults fromthe belief thatthis hasa beneficial effect onbottom-line performancefor the firm. However, practitioners often tend to use theterms service qualityand customer satisfaction interchangeably. Amongacademics thesatisfaction constructis recognisedas beingdistinct and hasdeveloped along fairly independent lines fromservice quality(e.g. Oliver, 1980).The concepts ofservice quality,customer satisfaction andservice loyaltyare related toeach other.Theoretically, the expectancy/ disconfirmation paradigmin process theorycan provide the grounding for this study,with

service qualityas anantecedent construct and service loyaltyas anoutcome EuropeanJournal of Marketing, variableof customer satisfaction. Abetterunderstanding of the effects of Vol.36 No. 7/ 8,2002, pp. 811-828. # MCBUPLimited,0309-0566 service qualityand customer satisfaction onservice loyaltycan help academics DOI10.1108/03090560210430818 European inthe development of a model ofservice marketing.It can also provide Journal of practitionerswith indications as towhere best todevote marketing attention Marketing andscarce corporateresources. Thisstudy seeks tocontribute to the development of a conceptual 36,7/8 frameworkthat integrates service loyalty,service qualityand customer satisfaction. Itreviews theliterature on these threeconstructs and outlines the 812 expectedrelationships ina researchmodel. Appropriate measures are identified andresearch is carriedout among retail bankingcustomers totest thehypothesised relationships. The demographic characteristics ofloyal customers arealso investigated.Implications fortheory development and managementare discussed.

Service loyalty Theconceptualisation oftheloyalty construct has evolved over the years. In theearly days the focus ofloyaltywas brandloyalty with respect totangible goods(Cunningham, 1956; Day, 1969; Kostecki, 1994; Tucker, 1964). Cunningham(1956) definedbrand loyalty simply as ``theproportion of purchasesof a householddevoted to the it purchasedmost often’’. Cunningham(1961) was tobroaden the spectrum of analysis byfocusing on store as opposedto brand loyalty using the same measures hehadused earlier forbrands. Over time thefoci havecontinued to expand,reflecting thewider perspective ofmarketing to include othertypes of loyalty such as vendor loyalty.However, few studies havelooked at customer loyalty of services (Oliver,1997). The intention of this section is toshow theevolution of the loyaltyconstruct over time, mappingout the construct’s domain and its specific componentsto provide a clear definitionof the service qualityconstruct used in this study. Areview ofthe literature indicates thatmuch of the initial research emphasised thebehavioural dimension of loyalty. This is epitomised by Tucker(1964, p. 32) whoholds that: Noconsideration should be givento whatthe subject thinks nor what goes on in hiscentral nervoussystem, his behaviour is the full statementof whatbrand loyalty is. Areview byJacoby (1971) confirmsthat prior studies havefocused entirely on behaviouraloutcomes andignored consideration of whatwent onin customers’ minds.Brand loyalty was simplymeasured in terms ofits outcome characteristics (Jacobyand Chestnut, 1978). This involved determining the sequence ofpurchase (Brown, 1952, 1953; Lawrence, 1969;McConnell, 1968; Tucker,1964), proportion of purchasedevoted to a givenbrand (Cunningham, 1956) andprobability of purchase (Frank, 1962; Maffei, 1960). Day(1969) arguedthat ` `thereis moreto brandloyalty than just consistent buyingof the same brand.Attitudes forinstance’ ’.Buildingon this work, Jacoby(1969, 1971) provideda conceptualisation ofbrand loyalty that incorporatedboth a behaviouraland an attitudinal component. The behaviouralaspect ofloyalty focuses ona measureof proportion of purchase of aspecific brand,while attitudeis measuredby a single scale (Day,1969) or Service muti-scale items (Selin et al.,1988).Day obtained a valuefor loyalty by dividing loyalty theratio of purchase of a brandby themean scores obtainedfor attitude. The behaviouraland attitudinal aspects ofloyalty are reflected inthe conceptual definitionof brand loyalty offered by Jacoby and Chestnut (1978). These authorshold that: 813 Brandloyalty is (1) biased(i.e. nonrandom), (2) behaviouralresponse (i.e. purchase),(3) expressedover time, (4) bysome decision making unit, (5) with respectto oneor more outof asetof suchbrands, and is a functionof psychologicalprocesses. Muchof the work on loyalty in the 1970s andearly 1980s hasused this conceptualisation (cf. Goldberg,1981; Lutz andWinn, 1974; Snyder, 1986). Morerecently, Dick and Basu (1994) suggestan attitudinal theoretical frameworkthat also envisages theloyalty construct as beingcomposed of ``relativeattitude’ ’and` `patronagebehavior’’. Afurtheraspect ofloyalty identified byother researchers inmore recent yearsis cognitiveloyalty. This is seen as ahigherorder dimension and involvesthe consumer’s conscious decision-makingprocess inthe evaluation of alternativebrands before a purchaseis effected. Gremlerand Brown (1996) extendthe concept of loyalty to intangible products, and their definition of service loyaltyincorporates the three specific componentsof loyalty considered,namely: the purchase, attitude and cognition. Service loyalty is defined as: Thedegree to whicha customerexhibits repeat purchasing behavior from a serviceprovider, possessesa positiveattitudinal dispositiontoward the provider, and considers using only this providerwhen a needfor this serviceexists (Gremler and Brown, 1996).

Service quality Definitions ofservice qualityhold that this is theresult ofthe comparison that customers makebetween theirexpectations abouta service andtheir perceptionof theway the service hasbeen performed (Lewis andBooms, 1983; Lehtinenand Lehtinen, 1982; Gro ¨nroos,1984; Parasuraman et al.,1985,1988, 1994).Lehtinen and Lehtinen (1982) givea three-dimensional view ofservice quality.They see it as consisting ofwhat they term ``interaction’ ’,``physical’’ and` `corporate’’quality.At ahigherlevel, andessentially froma customer’s perspective,they see qualityas beingtwo-dimensional, consisting of``output’’ and` `process’’quality.The model proposedby Gro¨nroos(1984, 1990) highlights therole oftechnical (oroutput) quality and functional (or process) qualityas occurringprior to, and resulting in, outcome quality. In this model technical qualityrefers towhatis deliveredto the customer, be it themeal ina restaurant, thesolution provided by a consultant,or the home identified bythe estate agent.Functional quality is concernedwith howthe end result ofthe process was transferredto the customer. This concerns both psychological and behavioralaspects thatinclude the accessibility tothe provider, how service employees performtheir task, what they say and how the service is done.Thus European while technical qualitycan often be quite readily evaluated objectively, this is Journal of moredifficult todo with functionalquality. The model also recognises that Marketing customers also havesome typeof image ofthe firm, which has a qualityimpact initself andfunctions as afilter. Thecustomers’ perceived quality is theresult 36,7/8 oftheevaluation they make of what was expectedand what was experienced, takinginto account the influence ofthe organisation’s image. 814 Inoperationalising the service qualityconstruct, Parasuraman et al. (1985, 1988,1994) havemade use ofqualitativeand quantitative research following generallyaccepted psychometricprocedures. This resulted in the development oftheoriginal 22-item SERVQUALinstrument that represents one of themost widely usedoperationalisations ofservice quality.It has provided researchers with thepossibility ofmeasuring the performance-expectations gap (gap 5) ostensibly composedof five determinants.In further developing the expectations side oftheir gap model, Berry and Parasuraman (1991) and Zeithaml et al. (1993) arguethat expectations canbe conceptualised toexist at two levels: thedesired; and the adequate. In between thereexists azone of tolerance reflecting thedegrees of heterogeneity individual customers are willing toaccept. Interestingly,the original service qualitygap (gap 5) now splits intotwo (Zeithmal et al.,1993).Gap 5A results fromthe contrast between perceivedservice anddesired service andis termedthe measure of service superiority(MSS). Gap 5B contrasts perceived service with adequateservice andis termedthe measure of service adequacy(MSA). The authors argue that companies providinga service abovethe adequate level havea competitive advantage.However, such companies needto strive so thatperceived service exceeds theservice level desiredby customers. This will ensure` `customer franchise’’whichresults inunwavering customer loyalty. Thecontentionby thedevelopers of SERVQUALthat the instrument can be appliedto determine the service qualityoffering of anyservice firmhas led to its extensive adoption(cf. Dabholkar et al.,1996).The various replications undertakenhave highlighted a numberof areas ofboth theoretical and psychometricconcern. First, theconceptualisation andusefulness ofthe expectations side ofthe instrument has been questioned (cf. Boulding et al., 1993;Cronin and Taylor, 1992, 1994; Forbes et al.,1986;Tse andWilton, 1988). Second,the problems expectation scores pose interms ofvariancerestriction havebeen highlighted (cf. Babakusand Boller, 1992; Brown et al.,1993).Third, thereare problems associated with difference scores includingfindings showingthat the performance items ontheir own explain more variance in service qualitythan difference scores (Babakusand Boller, 1992;Cronin and Taylor,1992, 1994). Cronin and Taylor (1992, 1994) show empirically thatthe perceptionitems inSERVQUAL exhibit a strongercorrelation with service qualitythan the difference score computationssuggested by SERVQUAL. Theytherefore suggest the use ofSERVPERF that consists solely ofthe 22 performanceitems ofSERVQUAL.Finally, the number of factors extractedis notstable (cf. Boumanand van der Wiele, 1992;Carman, 1990; Cronin and Taylor,1992, 1994; Gagliano and Hathcote,1994). Inresponse to the empirical findingsthat have emerged, Parasuraman et al. Service (1994) haveundertaken significant changes.First, therehas been a loyalty reconceptualisation andextension of the expectations side distinguishing between desiredand minimum expectations. Second, they have suggested the use ofa three-columnformat SERVQUAL that eliminates theneed to re-administeritems. Theauthors have also suggesteda reductionin the numberof items to21, the use ofnine-point instead of seven-point scales, and 815 recognise thepossibility ofthe existence ofthree rather than five dimensions, where` `responsiveness,assurance and empathy meld intoa single factor’’. The Gro¨nroosand the gap model ofservice qualityprovide parallel conceptualisation ofthe construct. The contribution made by Parasuraman et al. hasbeen in developing the widely usedSERVQUAL. Cronin and Taylor (1992,1994) haveshown that SERVPERF does abetterjob in measuring service quality.This paper takes theview thatthe conceptualisation ofservice qualityas agapis correct,but adopts the position by Rust et al. (1996,p. 249) whohold that service qualityis simplyconfirmation/ disconfirmationin satisfaction theory.Operationally this meansthat the gap is measureddirectly byaskingrespondents to providea score foreach ofthe performance items in SERVQUALin relation to their expectations ratherthan ask these separately andthen calculating thegap. This preserves the conceptualisation ofservice qualitybut has the advantage of being more statistically reliable andcutting thelength of the questionnaire.

Customersatisfaction Theexpectancy/ disconfirmationparadigm in process theory(Mohr, 1982) providesthe grounding for the vast majority of satisfaction studies and encompasses fourconstructs: (1) expectations; (2) performance; (3) disconfirmation;and (4) satisfaction. Dis/confirmationarises fromdiscrepancies between priorexpectations and actual performance.This conceptualisation is reflected inthe definition of satisfaction byTse andWilton (1988, p. 204) as: Theconsumer’s response to theevaluation of theperceived discrepancy between prior expectations(or some norm of performance)and the actual performance of theproduct as perceivedafter its consumption. At face valuethis definitionis verysimilar tothat put forward for service quality.However, a numberof distinctions areoften made between customer satisfaction andservice quality.These include thatsatisfaction is apost- decision customerexperience while qualityis not(Bolton and Drew, 1991; Boulding et al.,1993;Cronin and Taylor, 1994; Oliver, 1980, 1993; Parasuraman et al.,1988).A furtherpoint concerns expectations thatare defined differently European inthe satisfaction andquality literature. In the satisfaction literature Journal of ``expectationsreflect anticipatedperformance’ ’(Churchill andSuprenant, 1982, Marketing p.492) madeby the customer about the levels ofperformance during a transaction.On the other hand, in the service qualityliterature, expectations 36,7/8 areconceptualised as anormativestandard of future wants (Boulding et al., 1993,p. 8).These normativeor ideal standardsrepresent enduring wants and 816 needs thatremain unaffected by the full rangeof marketingand competitive factors.Normative expectations aretherefore more stable andcan be thought of asrepresentingthe service themarket oriented provider must constantly strive tooffer (Zeithaml et al., 1993). Oneof thehurdles in looking at antecedents andconsequences ofcustomer satisfaction is theabsence ofa consensus as towhat constitutes satisfaction. Withouta clear andbroadly accepted conceptualand operational definition the developmentof satisfaction measurementinstruments is somewhat arbitrary, andany conclusions aboutinteractions with otherconstructs are problematic. Toidentifythe conceptual domain of the customer satisfaction construct,Giese andCote (2000) conductresearch that involves a review ofthe satisfaction literaturetogether with groupand personal interviews. Theydefine the customeras theultimate userof a product.Their research suggests three generalcomponents that constitute thecustomer satisfaction construct.First, customersatisfaction is asummaryaffective responsethat varies inintensity. Second,the response pertains to a particularfocus, be it aproductchoice, purchaseor consumption.Finally, the response occurs ata particulartime that varies bysituation,but is generallylimited induration. The authors hold that these threeaspects providea frameworkfor a contextspecific operational definition.They describe customersatisfaction as: Asummaryaffective responseof varyingintensity, with atime-specific point of determinationand limited duration,directed toward focal aspects of productacquisition and/ orconsumption. Thisdefinition can be used to develop context relevant definitions. For the purposeof this studythe definition of customersatisfaction with retail banking services: Involvesa postpurchase, global affective summaryresponse, that maybe of different intensities, occurringwhen customers are questioned and undertaken relative to theretail bankingservices offered by competitors.

Researchmodel Therehas been significant effortin the past to look at the area of service quality,customer satisfaction and,to a lesser extent,service loyalty.However, thereis considerable confusionin the demarcation between service quality andcustomer satisfaction. Gro¨nroos(1984, 1990) andParasuraman et al. (1985,1988, 1994), both argue that perceived service qualityresults fromthe comparisonthat customers makebetween expectedquality and experienced or outcomequality. The expectancy/ disconfirmationparadigm that ultimately results insatisfaction ordissatisfaction makes asimilar point.Use is madeof Service this paradigmin process theoryto accommodate both the Gro ¨nroosand the loyalty gap model. Itis clear fromGro ¨nroos(1984) thatthe most importantaspect toperceived service qualityis thefunctional rather than the technical side ofquality.The gapmodel and its resultantSERVQUAL measure primarily focus onwhat Gro¨nroos(1984, 1990) terms thefunctional aspect ofquality. It is suggested 817 thatthese two models representparallel concepts thatcan both be viewed as onetype of confirmation/ disconfirmationin satisfaction theory.It is forthis reasonthat we adoptthe suggestion by Rust et al. (1996,p. 249) whoargue that service qualityis simplyconfirmation/ disconfirmationand who advocate the direct measurementof the perception items inSERVQUAL in relation to respondents’expectations. On its ownthe gap model hasno theoretical groundingand the use ofdifference score measures relative toideal expectations is questionable.The approachbeing suggested has the advantage ofproviding a clearer theoretical underpinningto the constructs, data that are morestatistically reliable while cuttingthe length of thequestionnaire. As aprocess intime, service qualitytakes place before,and leads to,overall customersatisfaction. AlthoughCronin and Taylor originally hypothesised thatsatisfaction is anantecedentof service quality,their research with amulti- industrysample showed,in a LISRELanalysis, an opposite relationship. Service qualityappears to be only one of the service factors contributingto customers’satisfaction judgements(Cronin and Taylor, 1992; Ruyter et al., 1997;Spreng and Mackoy, 1996). There are clearly otherantecedents. Overallsatisfaction with anexperience does lead tocustomer loyalty. Beardenand Teel (1983) arguethat customer satisfaction is importantto the marketerbecause ``itis generallyassumed to be asignificant determinantof repeatsales, positive wordof mouth and consumer loyalty’ ’.Similarly,Bloemer andPoiesz (1989) havealso arguedthat ``satisfaction canbe thoughtof as an importantdeterminant of brand loyalty’ ’,while Selnes (1993) arguesthat it is satisfaction with abrandthat leads tocustomer loyalty. This view is also supportedby Dick and Basu (1994). LaBarberaand Mazursky (1983) show empirically thatbrand loyal customers hada lower probabilityto switch brandsdue to higher levels ofsatisfaction. Onthebasis ofthe above, customer satisfaction is indicatedas actingas amediatorin the link between service qualityand service loyaltyas perFigure 1. Baronand Kenny (1986, p. 1177) providethe procedure that can be usedto investigate themediating effect depictedin Figure 1. This involves the computationof three regression equations: first, the regression of themediator (customer satisfaction) onthe independent variable (service quality),second, theregression of the dependent variable (service loyalty)on the independent variable(service quality);and third, the regression of the dependant variable (service loyalty) onboth the independent variable (service quality)and on the mediator(customer satisfaction). Formediation to hold:in the first regression equationthe independent variable must affect themediator; in the second European Journal of Marketing 36,7/8

818

Figure 1. Researchmodel

equationthe independent variable must be shown to affect thedependent variable;and in the final equationthe mediator must affect thedependent variableto the exclusion oftheindependent variable.

The research Thequestionnaire used consisted of37 items split between threeinstruments thateach measuredservice loyalty,service qualityand customer satisfaction respectively. Demographicvariables were also collected. Tomeasure service loyaltythe 12-item measuresuggested by Gremler and Brown (1996) that capturestheir conceptualisation ofservice loyaltyhas been used. Seven-point scales describedat either endby ``stronglyagree’ ’and``strongly disagree’ ’were used.To measure service qualitythe 21-item SERVQUALinstrument was used.However, in line with theconceptualisation envisaged,rather than collect expectationand perception items separately,service qualityis treatedas disconfirmationin satisfaction theory,and perceptions data relative to respondentexpectations arecollected directly.Therefore, for each perception item respondentswere askedto consider their views interms oftheir expectations onathree-pointscale. Was theperception on theparticular item worse thanexpected, about as expectedor betterthan expected? To measure customersatisfaction theinstrument provided by Bitner and Hubbert (1994) was used.This is inline with theconceptualisation ofcustomer satisfaction adopted.The instrument consists ofa four-itemscale thatlooks atpost- purchase,global affective summaryresponses measuredusing a five-point Likert-typescale appropriatelydescribed at either end.The wordings of the items, togetherwith descriptive statistics, appearin the Appendix. Postal questionnaireswere undertakenbecause ofconsideration of costs. It is knownthat almost all householdsin Malta have a bankaccount and use was thereforemade of the telephone directory as aconvenientsampling frame. The ``critical sample size’’forthe intended subsequent LISREL analysis is consideredto be 200 replies (Hair,1998, p. 605). Since thereply rate for anonymouspostal surveysin Malta is inthe region of 23 per cent, mailings were sent to1,000 households generated at randomfrom an electronic version Service ofthe telephone directory that excluded commercial subscribers.The covering loyalty letter soughtto generate interest, reassure potential respondents of the minimumeffort required in replying, and provided an undertaking of anonymity.By the closing date,three weeks later,194 replies were received. Giventhat 52 of theoriginal questionnaires were returnedfor various reasons, anacceptable responserate of 20.5 per cent was achieved.To test fornon- 819 responsebias use was madeof the technique suggested by Armstrong and Overton(1977). Thisassumes thatlate respondentsare similar tonon- respondents. t-tests were usedto compare the means for each oftheitems ofthe last 30replies received tothe mean for the same items fromthe rest ofthe respondents.Differences were notstatistically significant andthis was treated as sufficient assurancethat the data obtained were likely tobe a fair representationof the population of interest.

Results Respondentswere almost equallysplit between males (43.8per cent) and females, 73.2per cent were marriedand 12.4 per cent were single. Themean agewas 43(SD =15.6)and 73.2 per cent ofrespondentshad completed upto secondarylevel ofeducation. The check forreliability provided(Cronbach, 1951) alphasthat ranged from 0.79 to 0.95, which exceeded theacceptable cut- offpoint of 0.70 (Nunnally, 1978). The thirdcustomer satisfaction item (item 24 inthe Appendix) was deleted as this hasan item-to-total correlationof less than 0.5.The Appendix provides loadings resulting from a simultaneousfactor analysis followed byavarimaxrotation of all theitems ofthe three constructs thatmake up the questionnaire. Results show aclear factorstructure. The service loyaltyand customer satisfaction items loadon two separatefactors, while theservice qualityitems loadon threefactors. The loading of theservice qualityitems is inline with thelatest findingsby Parasuraman et al. (1994). Theresponsiveness,assurance and empathy items meldinginto one factor and distinct factors fortangibles and reliability arise. Also inline with recent findings,SERVQUAL item 6(item 22in the Appendix) that had previously beentreated as aresponsiveness item tendsto load with thereliability items. Theloading results inthe Appendix provide support for both convergent and discriminantvalidity, with items expectedto load together actually doing so. Theresults ofthe regression equations required to test themediation model areshown in Table I. Theconditions required for mediation to hold are present. Theeffect ofservice qualityon the service loyaltyis muchless inthe third equationthan in the second andthe R hasimproved. Although perfect mediationcannot be claimed as thebeta for service qualityin the third equation is still significant (at p <0.05),a mediationeffect canbe confirmedas thebeta valuehas declined veryconsiderably. Service quality acts onservice loyaltyvia customersatisfaction. As canbe expected, service qualityand customer satisfaction arecorrelated ( r = 0.45; p <0.00),resulting in multicollinearity, and this is compoundedby the possible presence ofmeasurement error in the European results.To overcomethese concernsstructural equation modelling can be used. Journal of Thishas the advantage of taking all therelevant paths being tested directly Marketing anddealing with complications ofmeasurementerror, correlated measurement errorand even feedback that are incorporated directly into the model (Baron 36,7/8 andKenny, 1986). To dothis themodel inFigure 1 was usedas thebasis ofa LISREL8.3 (Jo ¨reskog et al.,1999) analysis.The results provideda verygood fit 820 (À2(3) = 2.24; p =0.52;GFI 0.98)and standardised beta values thatare very close tothose obtainedfrom the multiple regression(Table II). The major difference is thatthe marginal decline inthe value of the beta for the link between service qualityand service loyaltyis enoughto render this link non-significant.The LISREL analysis canbe considered to be amorerobust test ofthe interrelationship among constructs, and the results obtainedprovide supportfor a completely mediatedeffect ofservice qualityon service loyalty viacustomer satisfaction. Meansfor each ofthe constructs were tested againstclassificatory variables. Statistically significant lower meanscores were obtainedfor the three constructsas theeducation level ofrespondentsincreases (TableIII). Results of afurtherANOVA also shows statistically significant differences between age groupsfor all constructs,with youngerrespondents tending to give lower scores (Table IV).Results of t-tests andan ANOVA forgender and marital status,respectively, indicatedno statistically significant differences. Toinvestigate whichelements ofage or education play the most salient effect, theservice loyaltyconstruct was investigatedfurther as this is ultimately whatdetermines defection orrepurchases. To do this analysis

y Customersatisfaction Service loyalty Serviceloyalty

R2 0.20 0.16 0.421 F 47.91*** 37.66*** 69.30*** Beta – servicequality 0.45*** 0.41*** 0.15* Table I. Beta – customersatisfaction 0.57*** Resultsof regression equationstesting Note: Betasreported are standardised values mediation *** p < 0.000; * = p < 0.05

UnstandardisedStandardised Parameter value value t-values

Gamma Servicequality customersatisfaction 0.67 0.53 7.04*** ! Beta Table II. Customersatisfaction serviceloyalty 0.59 0.59 8.44*** MLestimatesfor Servicequality service! loyalty 0.17 0.14 1.91 structuralmodel ! parameters Note: *** p < 0.000 CHAID(chi-squared automated interaction detector) was appliedto the data. Service Theresults intree diagramform are depicted in Figure 2. These show that loyalty 62.5per cent ofrespondents are less loyal.Education is theprimary variable thatexplains the presence ofservice loyalty,with those thathave completed educationat post-secondary level (codes 3-5) andabove being proportionally (72.8per cent) theleast loyal.Moreover, the analysis indicates thatwithin the categoryof those thathave completed educationat post-secondary level and 821 above,90 per cent ofthose inthe 15-39 yearsage bracket (code 1-5) areeven less loyal.

Discussion Thefindings indicate thatthe questionnaires identified tomeasure service loyalty,customer satisfaction andservice qualityexhibit acceptable psychometricproperties in terms ofboth reliability andvalidity. The results confirmthe hypothesised relationships inthe research model. Service quality is foundto act onservice loyaltyvia customer satisfaction. Theresults also show thatwhile genderand marital status providedno basis fordifferentiation amongconstructs, education and age play a majorrole indetermining the differentperceptions of customers aboutthe constructs investigated. Analysis usingCHAID for the service loyaltyconstruct shows thatit is education followed byagethat is thesalient segmentationvariable. Theresearch contributes to our knowledge by providing support for the contentionthat customer satisfaction performsa mediatingrole inthe link between service qualityand service loyalty.Service qualityhas been found to beanimportantinput to customer satisfaction andexplains 53 percent ofits variance.However, the main focus ofmanagement attention should be on customersatisfaction, ofwhich service qualityis animportant antecedent. Identification ofthevarious elements, besides service quality,that contribute to overall customersatisfaction becomes critical. Otherelements thatcould be contributingto customer satisfaction couldinclude otherconstructs such as transactionsatisfaction, valueand corporate reputation or image.Similarly, it is just as critical toidentifyother elements, inaddition to customer satisfaction, thathave a direct impacton service loyalty.A clearer understandingas tothe sequence ofrelationshipbetween service quality,customer satisfaction, service loyaltyand ultimately theperformance of the firm can help ensure better targetingof limited marketingresources. It is interestingthat service loyaltyis primarilyaffected byeducation and only to a secondaryextent by age.

Post- Post- PrimarySecondary secondary Graduate graduate F Sig Table III. Servicequality 53.91 54.61 52.44 45.6145.06 6.42 0.000 ANOVAfor means of Customersatisfaction 12.81 13.14 11.95 10.9410.43 4.01 0.004 constructson the basis Serviceloyalty 73.08 73.84 67.16 64.8359.21 3.72 0.006 ofeducation European Customer Journal of Servicequality satisfaction Serviceloyalty Marketing Age group 36,7/8 15-19 48.30 12.33 68.86 20-24 51.01 11.50 60.61 25-29 48.80 10.07 57.78 822 35-34 49.82 12.91 73.90 40-44 56.00 12.90 72.22 45-49 54.27 12.82 70.00 50-54 51.82 11.68 69.45 55-59 58.61 15.00 80.20 60-64 55.43 12.94 74.14 65-69 54.28 14.08 74.74 Table IV. 70-74 56.02 14.17 81.13 ANOVAfor means of 75-80 55.00 14.00 84.00 constructson the basis F 1.882 2.407 2.504 of agebracket Sig. 0.039 0.007 0.005

Figure 2. Treediagram of results fromCHAID analysisof serviceloyalty

Education,occupation and income areconsidered to be critical elements in social class. Itis likely thatthe variable of influence onservice loyaltycould quitewell besocial class. Knowingwhether the different social class groupings exhibitdifferent levels ofservice loyaltycould have considerable marketing implications. Thistype of analysis canhelp identify segments thathave the Service highestpotential of defection andwhere best totarget limited marketing loyalty resources.It is crucial forservice firms toestablish benchmarksfor the constructsunder consideration and to regularly and systematically monitorits performanceamong its customers inthe same mannerthat the firm monitors its managementaccounts. Different benchmarks may be required for the differentsegments thatthe firm targets. Moreover, such monitoring can be 823 extendedto include acomparisonof the firm’s performance on the three constructsto those ofits competitors. All researchhas its limitations andthis studyis noexception. In a strict sense theresults pertainonly to the respondents and generalisations toa wider populationor industry should be done with caution.The sample size is not largebut adequate for the type of analysis undertaken.While thereis an argumentfor the relationship investigated to be undertaken with morethan onesector tosupport the findings’ generalisability, this needs tobe balanced with ``chameleoneffects’’ thatare likely tocreep inand cause themeaning of customersatisfaction tovary as aresult ofthe different research contexts (Giese andCote, 2000). Thisstudy provides a numberof directions forfuture research. Work can focus onidentifying and possibly developing a uniquemeasure of customer satisfaction fora bankingcontext. It is also possible tolook at developing a richermodel thatincorporates other constructs beyond the three used in this studyand to consider their interactive effects. Therole ofvalue and its exact relationshipto customersatisfaction andservice loyaltycould prove to be an interestingarea of study. Similarly, the role ofcorporatereputation or image is worthyof further elaboration. Does corporatereputation have an effect on service quality,on customer satisfaction oron both? Is therea reverse relationshipfrom these two constructsto corporate reputation? How important is therole ofcustomerinvolvement? What is theeffect ofswitching cost? How importantare emotions inservice loyalty?

References Armstrong,J.S. andOverton, T.S. (1977),` `Estimating non-responsebias in mail surveys’’, Journal ofMarketing Research ,Vol. XIV, August,pp. 396-402. Babakus,E. andBoller, G.W. (1992),` `An empiricalassessment of theSERVQUAL scale’ ’, Journal ofBusiness Research ,Vol. 24No.3, May, pp. 253-68. Baron,R.M. andKenny, D.A. (1986),``The moderator-mediator variable distinction in social psychologicalresearch: conceptual, strategic and statistical considerations’’, Journal of Personalityand Social Psychology ,Vol. 51No. 6, pp. 1173-82. Bearden,W.O. andTeel, J.E. (1983),` `Selecteddeterminants of consumersatisfaction and complaintbehavior’ ’, Journal ofMarketing Research ,Vol. 20,February, pp. 21-8. Berry,L.L. andParasuraman, A. (1991), Marketing Services – Competingthrough Quality , Free Press,Macmillan, NewYork, NY. Bitner,M.J. andHubbert, A.R. (1994),` `Encountersatisfaction versus overall satisfaction versus quality’’,in Rust, R.T. andOliver, R.L. (Eds), ServiceQuality: New Directions in Theoryand Practice,Sage,London, pp. 72-94. European Bloemer,J.M.M. andPoiesz, T.B.C. (1989),` `Theillusion of consumersatisfaction’ ’, Journal of Journal of ConsumerSatisfaction, Dissatisfaction and Complaining Behavior ,Vol. 2,pp. 43-8. Bolton, R.N. andDrew, J.H. (1991),` `A multistage modelof customers’assessments of service Marketing quality andvalue’ ’, Journal ofConsumerResearch ,Vol. 17,March, pp. 375-84. 36,7/8 Boulding,W., Kalra,A., Staelin, R.andZeithaml, V.(1993),` `A dynamicprocess model of service quality: fromexpectations to behaviouralintentions’’, Journal ofMarketing Research , 824 Vol. 30,February, pp. 7-27. Bouman,M. andvan der Wiele, T.(1992),` `Measuringservice quality in thecar service industry:building and testing aninstrument’ ’, International Journal ofService Industry Management,Vol. 3No.4, pp. 4-16. Brown,G.H. (1952),` `Brandloyalty – fact orfiction?’’, AdvertisingAge ,Vol. 23,9 June,pp. 53-5. Brown,G.H. (1953),` `Brandloyalty – fact orfiction?’ ’, AdvertisingAge ,Vol. 24,26 January, pp. 75-6. Brown,T.J., Churchill,G.A. Jr andPeter, J.P. (1993),` `Improvingthe measurement of service quality’’, Journal ofRetailing ,Vol. 68No.1, Spring, pp. 127-39. Carman,J.M. (1990),` `Consumerperceptions of servicequality: anassessment of theSERVQUAL dimensions’’, Journal ofRetailing ,Vol. 66No. 1, Spring,pp. 33-55. Churchill,G.A. Jrand Surprenant, C. (1982),` `An investigationinto thedeterminants of customer satisfaction’’, Journal ofMarketing Research ,No.XIX, November,pp. 491-504. Cronbach,L.J. (1951),` `Coefficient alphaand the internal structure of tests’’, Psychometrika , Vol. 16 No.3, September, pp. 297-333. Cronin,J.J. andTaylor, S.A. (1992),` `Measuringservice quality: are-examinationand extension’’, Journal ofMarketing ,Vol. 56,July, pp.55-68. Cronin,J.J. andTaylor, S.A. (1994),``SERVPERF versus SERVQUAL: reconciling performance basedand perception based – minus – expectationmeasurements of servicequality’’, Journal ofMarketing ,Vol. 58,January, pp. 125-31. Cunningham,R.M. (1956),` `Brandloyalty – what, where,how much’’, Harvard BusinessReview , Vol. 39,November-December, pp. 116-38. Cunningham,R.M. (1961),` `Customerloyalty to storeand brand’ ’, Harvard BusinessReview , Vol. 39,November-December, pp. 127-37. Dabholkar,P.A., Thorpe,D.I. andRentz, J.O. (1996),` `A measureof servicequality forretail stores:scale development and validation’ ’, Journal oftheAcademy of Marketing Science , Vol. 24No. 1, Winter, pp.3-16. Day,G.S. (1969),` `A two dimensionalconcept of brandloyalty’ ’, Journal ofAdvertising Research , Vol. 9,September, pp. 29-36. Dick, A.S. andBasu, K. (1994),` `Customerloyalty: towardan integratedconceptual framework’’, Journal ofthe Academy of Marketing Science ,Vol. 22No. 2, pp. 99-113. Frank,R.E. (1962),` `Brandchoice as a probabilityprocess’ ’, Journal ofBusiness ,Vol. 35,January, pp. 43-56. Frank,R.E. (1967),` `Correlatesof buyingbehavior for grocer products’’, Journal ofMarketing , Vol. 31,October, pp. 48-53. Forbes,J.D., Tse,D.K. andTaylor, S. (1986),` `Towarda modelof consumerpost-choice responsebehaviour’’, in Lutz,R.L. (Ed.), Advancesin ConsumerResearch , Vol. 13, Associationfor Consumer Research, Ann Arbor, MI, pp.658-61. Gagliano,K.B. andHathcote, J. (1994),` `Customerexpectations and perceptions of servicequality in apparelretailing’ ’, Journal ofServicesMarketing ,Vol. 8No.1, pp. 60-9. Giese,J. andCote, J. (2000),` `Defining customersatisfaction’’, Academyof Marketing Science Review,availableat: www.amsreview.org/amsrev/theory/giese00-01.html Goldberg,S.M. (1981),` `An empiricalstudy of lifestyles correlatesto brandloyalty behaviour’’, Service Advancesin ConsumerResearch ,Vol. 9,pp. 456-60. loyalty Gremler,D.D. andBrown, S.W. (1996),` `Serviceloyalty: its nature,importance and implications’ ’, in Edvardsson,B., Brown,S.W., Johnston,R. andScheuing, E.E. (Eds), Proceedings American Marketing Association ,pp.171-80. Gro¨nroos,C. (1984),` `A servicequality modeland its marketimplications’ ’, European Journal of Marketing,Vol. 18No. 4, pp.36-44. 825 Gro¨nroos,C. (1990), ServiceManagement and Marketing ,D.C. Heath& Co., Lexington,MA. Hair,J.F., Anderson,R.E., Tatham,R.L. andBlack, W.C., MultivariateData Analysis , 4th ed., Macmillan, NewYork, NY. Jo¨reskog,K.G., So¨rbon,D., duToit, S.anddu Toit, M.(1999), LISREL8: NewStatistical Features , Scientific SoftwareInternational, Chicago, IL. Jacoby,J. (1969), Towardsa Multi-brandModel of Brand Loyalty ,Paper105, Department of Psychology,Purdue University, West Lafayette,IN. Jacoby,J. (1971),` `A modelof multi-brandloyalty’ ’, Journal ofAdvertising Research , Vol. 11, pp. 25-31. Jacoby,J. andChestnut, R.W. (1978), Brand Loyalty:Measurement and Management ,JohnWiley, NewYork, NY. Kostecki,M.M. (1994), Marketing Strategiesfor Services:Globalisation, Client Orientation, Deregulation ,PergamonPress, Oxford. LaBarbera,P.A. andMazursky, D. (1983),` `A longitudinal assessmentof consumersatisfaction/ dissatisfaction:the dynamic aspect of thecognitive process’ ’, Journal ofMarketing Research,Vol. 20,November, pp. 393-404. Lawrence,R.J. (1969),` `Patternsof buyerbehavior: time fora newapproach?’ ’, Journal of ,Vol. VI, May,pp. 137-44. Lehtinen,U. andLehtinen, J.R. (1982),` `Servicequality – astudyof dimensions’’,unpublished workingpaper, Institute, Helsinki, pp.439-60. Lewis,R.C. andBooms, B.H. (1983),` `Themarketing aspects of servicequality’ ’,in Berry,L.L., Shostack,G. andUpah, G. (Eds), Emerging Perspectivesin ServiceMarketing , American MarketingAssociation, Chicago, IL, pp.99-107. Lutz,R.J. andWinn, P.R. (1974),` `Developinga Bayesianmeasure of brandloyalty: apreliminary report’’,in Curhan,R.C. (Ed.), Proceedingsof theAmerican Marketing Association , Vol. 36, Chicago,IL. McConnell,J.D. (1968),` `Thedevelopment of brandloyalty: anexperimental study’ ’, Journal of Marketing Research ,Vol. V,February,pp. 13-19. Maffei, R.B. (1960),` `Brandpreference and simple Markov processes’ ’, Operations Research , Vol. 8, March-April,pp. 210-18. Mohr,L.B. (1982), Explaining Organizational Behavior ,Jossey-Bass,San Francisco, CA. Nunnally,J.C. (1978), PsychometricTheory ,2nded., McGraw Hill, NewYork, NY. Oliver,R.L. (1980),` `A cognitivemodel of theantecedents and consequences of satisfaction decisions’’, Journal ofMarketing Research ,Vol. XVII, November,pp. 460-9. Oliver,R.L. (1993),` `A conceptualmodel of servicequality andservice satisfaction: compatible goals,different concepts’’,in Swartz,T.A., Bowen,D.E. andBrown, S.W. (Eds), Advances in ServicesMarketing andManagement: Researchand Practice ,Vol. 2, JAI Press, Greenwich,CT, pp.65-85. Oliver,R.L. (1997), Satisfaction.A BehaviouralPerspective on Consumers ,McGrawHill, Maidenhead. European Parasuraman,A., Zeithaml, V.A. andBerry, L.L. (1985),` `A conceptualmodel of servicequality Journal of andits implication forfuture research’ ’, Journal ofMarketing ,Vol. 49,April, pp.41-50. Parasuraman,A., Zeithaml, V.A. andBerry, L.L. (1988),` `SERVQUAL:a multiple-item scalefor Marketing measuringconsumer perceptions of servicequality’ ’, Journal ofRetailing ,Vol. 64No. 1, 36,7/8 Spring,pp. 12-40. Parasuraman,A., Zeithaml, V.A. andBerry, L.L. (1994),` `Alternativescales for measuring service quality: acomparativeassessment based on psychometricand diagnostic criteria’ ’, Journal 826 ofRetailing ,Vol. 70No.3, pp. 201-30. Rust, R.T., Zahorik,A.J. andKeiningham, T.L. (1996), ServiceMarketing ,HarperCollins, New York, NY. Ruyter,K., Bloemer,J. andPeeters, P. (1997),` `Mergingservice quality andservice satisfaction: anempirical test of anintegrative model’ ’, Journal ofEconomic Psychology , Vol. 18, pp.387-406. Selin, S., Howard,D.R., Udd,E. andCable, T. (1988),` `An analysisof consumerloyalty to municipalrecreation programs’ ’, LeisureSciences ,Vol. 10,pp. 210-23. Selnes,F. (1993),` `An examinationof theeffect of productperformance on brand reputation, satisfactionand loyalty’ ’, European Journal ofMarketing ,Vol. 27No. 9, pp. 19-35. Snyder,D.R. (1986),` `Serviceloyalty andits measurement:a preliminaryinvestigation’ ’,in Vankatesan,M., Schmalensee,D.M. andMarshall, C. (Eds), Proceedingsof the American Marketing Association ,Chicago,IL, pp.44-8. Spreng,R.A. andMackoy, R.D. (1996),` `An empiricalexamination of amodelof perceivedservice quality andsatisfaction’ ’, Journal ofRetailing ,Vol. 72No. 2, pp.201-14. Tucker,W.T. (1964),` `Thedevelopment of brandloyalty’ ’, Journal ofMarketing Research , Vol. 1, August,pp. 32-5. Tse,D.K. andWilton, P.C. (1988),` `Modelsof consumersatisfaction formation: anextension’’, Journal ofMarketing Research ,Vol. XVII, November,pp. 460-9. Zeithaml, V.A., Berry,L.L. andParasuraman, A. (1993),` `Thenature and determinants of customerexpectations of service’’, Journal ofthe Academy of Marketing Science , Vol. 21 No.1, pp. 1-12. Appendix Service )

d loyalty * e 3 u n Q i S t n o c ( * 2 4 4 4 6 2 2 5 5 6 7 5 6 ...... Q 0 0 0 0 0 0 S 827 * 2 3 1 4 6 . . Q 0 0 S * 1 2 8 T 6 7 6 . . . A 0 0 0 S * 9 8 0 0 2 3 7 3 9 8 0 6 Y 6 7 7 6 7 8 8 8 8 8 9 8 ...... O 0 0 0 0 0 0 0 0 0 0 0 0 L v e 7 2 7 4 1 3 7 4 2 3 3 1 4 1 4 5 1 7 7 8 9 8 7 5 5 3 5 8 6 4 4 2 4 5 4 5 5 D ...... 0 1 0 8 5 5 5 5 5 6 5 ...... 1 1 1 1 1 1 1 1 1 1 1 1 4 d 1 1 1 2 0 0 0 0 0 0 0 1 t S d e t e l e n 1 0 9 3 7 0 8 2 9 9 2 8 9 4 2 8 2 2 7 6 1 0 5 3 D a 7 1 6 4 8 9 8 2 9 8 9 8 4 0 2 4 4 5 3 6 ...... 2 4 4 5 e . . . . 5 6 5 5 5 5 5 6 5 5 5 5 0 4 4 2 2 2 2 2 4 2 2 2 7 1 M s e c i v u r o e y d s e e Z d Z r , i Y a h Y v t X l o i X l s r a h w p e m r t h i t s e e s i e l u v s w b b w e o o o l s e n t r l s l i s i d s p s e t p e Z e i ) w o u n n f Z e i s n d e Y a b s c i e s e e Y i i p s c c t X s u w e i v i X u r e a r r b n v I v e s h e b s e r o r t m v e o s h i e m i e n d o t e c t Z s d o w s e i r r w o d r e ( e o Y h e v f d o n d v r h s o t o e m e X e w m t s t e a i , y m s k e h t s i e s o e h a r a r t h t n d n e Z n c t t e e s w s s m i e a i I w Z v u s e I n r Y d s s i o u b r a i , m o e i d e t k w b Y r i c u i e s e X o f e y a n r p t y b g X c m l h m s e n t e a g r m t s e m c s o n s n o o h i e u i t a r b p n r r t g d h u e t t e d c k i i h i o e s e o c h x e l p n g n t l n r m f I i i i d i d e c b e w y u y i d f e a a r t l e o s s t n o r n s r c a n i s e c n i r r p b e a a h o i d u a p r s s a a I t t u s x w d p v g s b e e r e l s s l h e a s s o e g e r l y c e h n o n e n e e s u i u d e a y c t n g e r o e h c c o n i l h y o m v Z w i n i i n i t i n e f h i s m s t a r e t h i e m i o v v i t w o d o m t Y i c e c c e o c o Z r r i y t w a i t r n c s s a a l l t I o e e s X f d Z e Z f s l l i Y f v o a t I s e s s l n l r t s s v a a c u y i g e e Y y s e Y c i i X l i y e o t a i p c t t t d t s b n e l g g t k g a i s c i o X i l X n n o h i r e l a a a u r a n t n n a l s t y r g a o o e , s s

a i i t Table AI. h e r d e y m g r o r y a e n c o d m t d d p u r d n u y o n i v r r i a n t d d i i o o p l p l o i b l e q o

t e Items with descriptive t n o e e e s l m e e p l v v m f i d s m l e c u g s s p e p a l y n m y e l r e i e o o o i t i r m m o o r e e n e e a a c o i r o r e r r c a e c statistics andfactor r n i o o i n i S I E S D R T B T c t t C F P B B C I P D P P K P W v v v s s s r ...... r . . . r ...... loadings e u e u e e 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1 2 3 D S 1 1 1 S C 1 1 1 1 C S 1 1 1 2 2 2 2 European

Journal of * 1 0 4 2 3 8 8 6 6 . . . . Q

Marketing 0 0 0 0 S 36,7/8 * 2 Q 828 S * 5 4 5 1 1 5 8 9 9 9 1 6 6 5 7 7 6 7 6 7 4 ...... Q 0 0 0 0 0 0 0 0 0 0 S * T A S * Y O L v e 6 4 8 8 5 6 7 7 7 0 8 3 9 0 3 D 5 5 5 5 5 5 5 5 5 7 6 6 5 6 1 ...... d 0 0 0 0 0 0 0 0 0 0 0 0 0 0 8 t S n 0 2 8 1 2 9 5 8 3 4 5 9 7 0 8 a 6 6 5 6 6 5 5 5 5 3 4 3 5 4 7 ...... e 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 5 M s n o i t s e u q r s e n e r c o e i i m e v h o m r c t s o e s t n a s f s u a r c u e s g a c h r n e n t e i r s o p i r i t r h t w e r p s a e t t c s a h a n c i s s t a n e m l e w s w u a a o h f a o u t n o o t n s d o n e q h a i t t a o e r s e u s n i t r t r c s d s e o t s a u i ’ r i t e r s g o t o i n s e c s e e i n c d e r n f e o n e e m e r h o s e l e t t y r o e e s c t r l t m t w t p a h a n a s o n n t o n i , e t i l u s t e 5 n s l d c s a t t d e a i 3 k a u s e f . f s s u e n i i i c r h a e 0 t r n e d a n s t i u s i t e i b o l o h n t o i n s v t c t a l ’ i w o c a r a h e s l c e e d a e d i e l r h f m s t f t d v n e v a n e i s s t a e n a r o s g g s e m n n a h d u h r s s p i n n n o e e r e i i i e s l t l l t l o o e o o o s e s a m a a a m r h h h h h u c u s m e e p o h s i c b t g t w w o w w w o p p t t s u n t p p e y s s s s s s s i u q t a a n s h e e e e e e i u d c e t e l s e e e e e e c i a y y e a y y y y y y g n l l o n g r l l n u g r o o o o o o n e i L l l l l n l l o a a i e q n i t v d i p p p p p p u u * k d v p a n e v s s i a o : i a i i e m m m m m m c o r i e c R E M E E G E H E C M V E V v t s ...... r

Table AI. o e e 4 5 6 7 8 9 0 1 2 3 4 5 6 7 D 2 2 2 2 2 2 3 3 3 3 3 3 3 3 S N