Professional Documents
Culture Documents
Jurnal Information Systems Continuance
Jurnal Information Systems Continuance
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.
Management Information Systems Research Center, University of Minnesota is collaborating with JSTOR to
digitize, preserve and extend access to MIS Quarterly.
http://www.jstor.org
MIS
r ?i
Qarterjy
UNDERSTANDING
INFORMATION
SYSTEMS
CONTINUANCE:
AN EXPECTATION-
CONFIRMATION
MODEL1
MISQuarterly
Vol.25 No.3, pp. 351-370/September
2001 351
and its eventual success depend on its continued extension of acceptance behaviors (i.e., they
use ratherthan first-timeuse. This is so because, employthe same set of pre-acceptance variables
given its critical role in today's business pro- to explainbothacceptance and continuancedeci-
cesses, infrequent,inappropriate,and ineffective sions), implicitlyassume that continuance co-
long-termuse of IS often contributesto corporate varies with acceptance (e.g., Davis et al. 1989;
failures (Lyytinenand Hirschheim1987). Under- Karahannaet al. 1999), and are, therefore,unable
standing continued use or "continuance"(in to explain why some users discontinue IS use
contrastto initialuse or "acceptance")is the goal afteraccepting it initially(the "acceptance-discon-
of this study. tinuance anomaly").Further,priorresearch does
not elaborate upon users' psychological motiva-
IS continuance at the individualuser level is also tions "emerging"after their initialacceptance-
central to the survival of many business-to- whichpotentiallyinfluenceusers' subsequent con-
consumer electronic commerce firms, such as tinuance decisions but not their prioracceptance
Internetservice providers(ISPs), online retailers, decisions. In sum, currentacceptance models
online banks, online brokerages, online travel providea limitedexplanationof, and may some-
agencies, and the like. The effective subscriber times contradict,observedcontinuancebehaviors.
base, marketshare, and revenues of these firms
depend on both the number of initialadopters This paper is one of the earliest to conceptualize
(new subscriptions)and the numberof continued and test a theoreticalmodelof IS continuancethat
users (subscriptionrenewals). The importanceof takes intoaccountthe above distinctionsbetween
continuance,vis-a-visacceptance, is evidentfrom acceptance and continuancebehaviors. The pro-
the fact thatacquiringnew customers may cost as posed modelis based on expectation-confirmation
much as five times more than retainingexisting theory(ECT)(Oliver1980), whichis furtherrefined
ones, given the costs of searchingfor new custo- using auxiliarytheories and empirical findings
mers, setting up new accounts, and initiatingnew from prior IS use research. The hypothesized
customers to the IS (Parthasarathyand Bhat- model is then validated empiricallyusing data
tacherjee 1998). For example, a 5% increase in from a field survey of online banking users.
customer retentionin the insuranceindustrytypi- Research questions addressed in this paper are:
cally translates into 18% savings in operating (1) what are the salient motivationsunderlyingIS
costs (Crego and Schiffrin1995). Such trends users' intentionto continue using an IS after its
underscorethe importance,relevance, and time- initialacceptance, and (2) how do these motiva-
liness of studying IS continuance as a topic of tions influencecontinuanceintention?This study
organizationalinterest. is similarin spiritto Davis et al.'s formulationof
the technology acceptance model (TAM)in that it
Continuanceis not entirelyan alien concept in IS adapts ECTfromthe consumerbehaviorliterature
research. It has been examined variously as to propose a model of IS continuance, just as
"implementation"(Zmud 1982), "incorporation" TAMadapted the theoryof reasoned action from
(Kwon and Zmud 1987), and "routinization" the social psychology literatureto postulate a
(Cooperand Zmud1990) inthe IS implementation model of IS acceptance.
literature. These studies acknowledge the exis-
tence of a post-acceptance stage when IS use The remainderof the paper proceeds as follows.
transcends conscious behaviorand becomes part The next section describes ECTand integratesit
of normal routine activity. Likewise, innovation withpriorIS usage researchto theorizea modelof
diffusiontheory,in itsfive-stage adoptiondecision IS continuance behavior. The thirdsection des-
process (consisting of knowledge, persuasion, cribes the research methodologyused to empiri-
decision, implementation, and confirmation cally test the research model. The fourthsection
phases), suggests that adopters reevaluate their presents the results of data analysis. The fifth
earlieracceptance decision duringa final"confir- section discusses research implicationsfor the
mation"stage and decide whetherto continue or study's key findingsand its limitations.The final
discontinue using an innovation(Rogers 1995). section summarizesthe study's core findingsand
However, these studies view continuance as an its contributions.
MISQuarterly
Vol.25 No.3/September
2001 353
cal state resultingwhen the emotion surrounding to be less useful. These changes are explained
disconfirmed expectations is coupled with the by self-perception theory (Bem 1972), which
consumer's priorfeelings about the consumption posits that individuals continually adjust their
experience."Bothdefinitionsunderscorea psych- perceptions (e.g., expectation) as they acquire
ological or affective state relatedto and resulting new informationabout the focal behavior (by
from a cognitive appraisal of the expectation- observing others' and theirown behaviors). The
performancediscrepancy (confirmation). Lower adjusted perceptions then provide the basis for
expectation and/or higher performance lead to subsequent behaviors. Hence, once updated,
greater confirmation,which in turn positively post-consumption(modified)expectationreplaces
influence customer satisfaction and continuance pre-consumption (initial) expectation in con-
intention. The reverse causes disconfirmation, sumers'cognitivememoryas the basis forguiding
dissatisfaction, and discontinuance intention. subsequent decision processes. As elaborated
Hence, as shown in Figure 1, confirmationis upon later,this study postulates satisfactionas an
inversely related to expectation and directly additive function of modified (ratherthan initial)
relatedto perceived performance. expectationand confirmation.
weighed withan evaluationof outcomes (e,) in an tation. Third,(ex post) expectationis represented
expectancy theoreticsense (i.e., B,e,i),similarto in the proposed model by (ex post) perceived
that in many attitude theories (e.g., Ajzen and usefulness. This representationis consistentwith
Fishbein 1977). However, the high correlation ECT's definition of expectation as individual
observed between belief (EB,) and belief- beliefs (B,) or sum of beliefs (.B1), because
evaluation (;Bie,) operationalizations(Swan and perceived usefulness is a cognitive belief salient
Trawick 1981) suggests that the belief repre- to IS use (Davis et al. 1989). Althoughexpec-
sentationis notsubstantiallydifferentfrom(though tation theoreticallymay be a broaderconstruct,
conceptually simpler than) the belief-evaluation encompassing manyadditionalbeliefs (e.g., ease
representation. The next section examines how of use), based on TAM-basedstudies, perceived
ECTcan serve as a useful theoreticalframework usefulness is an adequate expectation in the IS
for explainingIS continuancebehaviors. continuancecontext because it is the only belief
thatis demonstratedto consistentlyinfluenceuser
intentionacross temporalstages of IS use (e.g.,
Model
An Expectation-Confirmation Davis et al. 1989; Karahanna et al. 1999).
of IS Continuance Figure 2 shows proposed associations among
these constructs.
IS users' continuancedecision is similarto con-
sumers' repurchasedecision because both deci- Per ECT, users' IS continuance intention is
sions (1) followan initial(acceptance or purchase) determinedprimarily bytheirsatisfactionwithprior
decision, (2) are influencedby the initialuse (of IS IS use. Several industrystudies provide anec-
or product)experience, and (3) can potentially dotal evidence for this association. Forinstance,
lead to ex post reversalof the initialdecision. IS an Inteco(1998) studycites negativeexperiences
continuance often imposes monetary and non- and dissatisfactionresultingfromslow access or
monetarycosts on IS users. Hence, rationalusers engaged lines, poorhelp lines, and othertechnical
most likely go though a non-trivialdecision problems as ISP users' primary reasons for
process, similarto that in ECT,priorto makingan service termination.Recall that satisfactionis an
informeddecision choice. However, in order to affect, captured as a positive (satisfied), indif-
adapt ECTto a differentcontext (i.e., IS contin- ferent,or negative (dissatisfied)feeling. Affect(as
uance), several theoretical extensions are re- attitude)has been theorizedandvalidatedinTAM-
quired. Such extensions provide unique oppor- based studies as an importantpredictorof inten-
tunitiesfortheoryrefinement.Potentiallythey can tion concerning IS use (e.g., Davis et al. 1989;
explainIS continuancedecisions betterthan ECT Karahannaet al. 1999; Taylorand Todd 1995).
alone. These studies provide indirect support for the
satisfaction-continuance intention association
First,while ECTexamines both pre-consumption derived from ECT. This leads to the first
and post-consumptionvariables (indicatedby t, hypothesis:
and t2 respectively in Figure 1), the proposed
continuance model focuses only on post-accep- H1. Users' level of satisfaction with
tance variables. This is so because the effects of initialIS use is positivelyassociated
any pre-acceptance variables are already cap- withtheirIS continuanceintention.
tured within the confirmationand satisfaction
constructs. Second, ECTonlyexamines the effect ECTposits thatuser satisfactionis determinedby
of pre-consumption(ex ante) expectation,butnot two constructs:expectationof the IS and confir-
post-consumption(ex post) expectation. As des- mation of expectation following actual use.
cribed before, ex post expectation is especially Expectationprovides the baseline level, against
importantfor productsor services where expec- whichconfirmationis assessed by users to deter-
tationmay change withtime, as is often the case mine their evaluative response or satisfaction.
with IS use. Hence, the proposed continuance Confirmationis positively related to satisfaction
model amends ECT to include ex post expec- with IS use because it implies realizationof the
MISQuarterly
Vol.25 No.3/September
2001 355
S I AgT Ai ~ *g.
*? S?~Y~1~~1)I~I _~~???YI
expected benefits of IS use, whiledisconfirmation they can also influence subsequent continuance
(perceived performance lagging expectation) decisions. However,empiricalstudies comparing
denotes failureto achieve expectation. Although the relative effects of perceived usefulness and
the confirmation-satisfaction association is yet to ease of use during pre-acceptance and post-
be examined empirically in IS use research, acceptance stages of IS use reportthat (1) useful-
industrystudies provideanecdotalsupportforthis ness impacts attitudesubstantivelyand consis-
association. Forinstance, online brokerageusers tently duringboth stages of IS use, and (2) ease
attributetheir service dissatisfaction to brokers' of use has an inconsistenteffect on attitudein the
inabilityto maintainserver uptime,execute timely initialstages, whichseems to furthersubside and
orders, and provide reasonable margin rates become non-significantinlaterstages (Daviset al.
(Selwyn 1999). Similarly,online shoppers are 1989; Karahannaet al. 1999). This finding led
disillusionedand dissatisfied with late deliveries, Karahannaet al. to observe,
inaccurate billing, and non-availabilityof items
listedon e-retailingsites (Sliwaand Collett2000). users gain experience with the system,
Hence: ease of use concerns seem to be
resolved and displaced by more instru-
H2. Users'extent of confirmationis posi- mental considerationsinvolvingthe effi-
tivelyassociated withtheirsatisfac- ciency of the innovation to increase
tion withIS use. one's job performance(i.e., perceived
usefulness). (p. 200)
As described before, drawing from TAM (e.g.,
Davis et al. 1989), post-consumptionexpectation In keeping with these observations, perceived
is representedas ex post perceived usefulness in usefulness is expected to be the most salient ex
the proposed IS continuance model. TAMfound post expectation influencingusers' post-accep-
perceived usefulness and perceived ease of use tance affect (satisfaction).3Hence:
as salient beliefs influencing IS acceptance
behaviorsacross a broadrange of end-user com- H3. Users' perceived usefulness of IS
putingtechnologies and user populations (e.g., use is positively associated with
Davis et al. 1989; Mathieson 1991; Taylor and theirsatisfactionwithIS use.
Todd 1995). Perceived usefulness captures the
instrumentalityof IS use, while ease of use taps
into the self-efficacy dimension. Because per- 3Aninformaltest validatedthe expected non-significant
effect of perceived ease of use on satisfaction in this
ceived usefulness and ease of use are the primary study,althoughthis associationwas not stated or tested
motivatorsof IS acceptance, it is plausible that as a formalhypothesis.
TAM hypothesizes perceived usefulness as a low. Theoretical support for this association
direct predictorof acceptance intention(in addi- comes fromcognitivedissonance theory (Festin-
tionto its indirecteffect via attitude)to accountfor ger 1957), which suggests that users may
circumstances where high instrumentalityconsi- experience cognitivedissonance or psychological
derations may override low affect in motivating tension iftheirpre-acceptanceusefulness percep-
usage intentions. As Davis et al. note, tions (which earlier led to acceptance) are
disconfirmedduringactual use. Rationalusers
people formintentionstowardbehaviors may tryto remedythis dissonance by distortingor
they believe willincrease theirjob perfor- modifyingtheirusefulness perceptionsin orderto
mance, over and above whatever posi- be more consistent with reality. In other words,
tive or negative feelings may be evoked confirmationwilltend to elevate users' perceived
towardthe behavior.(p. 986) usefulness and disconfirmationwill reduce such
perceptions. Hence:
Enhanced performance is instrumental in
achievingvariousrewardsthatare extrinsicto the H5. Users' extent of confirmation is
task context, such as promotionsor monetary positively associated with their
gains (Vroom1964). IS use is oftenviewedas the perceived usefulness of IS use.
means to that end. Such means-end behavioris
largely based on cognitive decision rules or In order to highlight the value-added by the
heuristics that are invoked without conscious proposed IS continuancemodel, it is instructiveto
thought whenever faced with similar behavioral compare it withTAMas an IS acceptance model.
contexts, withoutnecessarily activatingthe posi- The continuancemodel is similarto TAMon two
tiveaffectassociated withperformance-contingent counts: (1) it employs individualcognitivefactors
rewards (Davis et al. 1989). Although the forpredicting(continued)IS use, and (2) it reflects
usefulness-intention association was originally the belief-affect-intentioncausality characteristic
derived in an acceptance context, it is likelyto of most IS use theories. However,the proposed
holdtrueincontinuancecontexts,because human model is differentfrom TAMon at least three
tendencies for subconsciously pursuing instru- counts. First,it explains continuance behaviors,
mental behaviors or striving for rewards are in contrastto TAM,whichfocuses on initialaccep-
independent of the timing or stage of such tance (but has been applied to both acceptance
behaviors. This leads to the fourthhypothesis: and continuance contexts). As stated before,
acceptance and continuance are two temporally
H4. Users' IS continuance intentionis andconceptuallydistinctand possiblyincongruent
positivelyassociated withtheirper- phases of IS use. Second, althoughno study has
ceived usefulness of IS use. yet compared TAM with ECT in continuance
contexts, ECT is a theoreticallyrichermodel by
Finally,just as the cognitive beliefs in IS accep- virtueof its inclusionof unique post-acceptance
tance contexts (i.e., ease of use and perceived variables (satisfaction, confirmation). Because
usefulness) are related(Davis et al. 1989), those these variablesare in greatertemporalproximity
in IS continuancecontexts (i.e., confirmationand to post-acceptance (continuance)behavior,they
perceived usefulness) may also be related. For are likelyto predictcontinuancebetterthanTAM's
instance, users may have low initialusefulness pre-acceptancevariables(i.e., usefulness, ease of
perceptionsof a new IS because they are unsure use, and attitude). Third,TAMcannot providea
what to expect from its use. Nonetheless, they reasonable explanation of acceptance-discon-
may stillwantto accept itwiththe intentof making tinuanceanomalybased on a common set of pre-
their usage experience a basis for formingmore acceptance variables. ECTexplainsthis anomaly
concrete perceptions. Althoughlow initialuseful- by suggesting that one's disconfirmationand
ness perceptions are easily confirmed, such dissatisfactionwithIS use may lead to its eventual
perceptionsmay be adjustedhigheras a resultof discontinuance, despite positive perceptions of
the confirmationexperience, when users realize pre-acceptancevariables. Empiricaltesting of the
that their initialperceptions were unrealistically proposed model follows in the next section.
MISQuarterly
Vol.25 No.3/September
2001 357
!
W1oUUiS1eIJtIfSJItIIWvr#[@13S6UMO
I0 II[-I
(2) they were too long to incorporatein studies score decreases with decreasing variance in
where multipleconstructs are being measured eitherthe pre-or post-consumptionratings,which
(e.g., Ives et al.'s short-forminstrumentcontains is likelyin inferredconfirmationbecause the two
39 items). This satisfactionscale is similarto atti- ratings tend to be highly correlated (Prakash
tude scales used in IS acceptance research (and 1984). Fourth,a model that includesexpectation,
reasonably so because both constructs connote perceived performance,and inferredconfirmation
affect). Nonetheless, this scale captured post- is likely to be over-specified because inferred
usage affect in contrast to pre-usage affect confirmationis determinedentirelyby the firsttwo
capturedin attitudescales. variables. Finally,perceivedconfirmationis empi-
ricallyfound to predict satisfaction better than
Confirmation is operationalized in the ECT inferred confirmation,even with a single-item
literaturein three ways: objective, inferred,and measure (R2= 0.73 versus 0.56) (Tse and Wilton
perceived (Yi 1990). Objective confirmation 1988), because of its temporal proximityto
employs an externaljudge to "objectively" assess satisfaction and because human intentions are
the expectation-performancediscrepancy under guided by perceptions (of confirmation),even if
the assumption that product or service perfor- such perceptionsare biased or inaccurate(Ajzen
mance can be judgeduniformlyacross consumers and Fishbein 1977). Given these limitations,Yi
using some predefinedand unambiguouscriteria recommends perceived confirmationas the most
(Olshavsky and Miller1972). Despite ease of appropriatemeasure of confirmation. Further,
measurement and manipulation,objectiveconfir- attribute-levelmeasurement of perceived confir-
mationis not an accurate predictorof satisfaction mationdoes not provideany significantimprove-
because it ignores variations in consumers' ment in explanation over overall product-level
expectation and performance perceptions (Yi measurement(Oliver1980). Hence, product-level
1990). Inferred confirmation calculates the confirmationis adequate for most studies.
expectation-performancediscrepancyin terms of
prespecified productor service attributes. It is Prior perceived confirmationscales in the ECT
assessed as a summation of difference scores literaturehave employed items such as "unsatis-
rated by consumers on pre-selected attributes factory," "extremely pleasing" (which reflect
before and after the consumption experience satisfaction),or "wouldcome here again"(which
(SwanandTrawick1981). Perceivedconfirmation overlap with intention) (e.g., Oliver and Linda
representsconsumers'subjectivepost-onlyrating 1981; Swan and Trawick1981). Giventhe poten-
of the same discrepency,eitherat overallproduct tial confounding of these items with other
or service level or at individual attributelevel constructsof interest(e.g., satisfaction,intention),
(Oliver1980; Tse and Wilton1988). three confirmation items were indigenously
created based on the conceptual definitionof this
Althoughinferredand perceived confirmationare construct (i.e., respondents' perception of the
both good predictors of satisfaction, inferred expectation-performancecongruence). Two of
confirmationsuffers from several limitations(Yi these items examined the perceived congruence
1990). First, post-consumptionratings may be intermsof user experience and service level. The
biased by pre-consumptionratings (consistency thirditem assessed respondents'overallextent of
bias) because respondents may provide similar confirmation.
ratingsin orderto be internallyconsistent, poten-
tiallyignoringtheirtrue post-consumptionbeliefs.
Second, inferred confirmation suffers from a
"ceilingor floor effect." If respondents give the Data Analysis and Results
highest (lowest) rating on the pre-consumption
scale and the productperformswell above (below) Scale Validation
expectations, they can still only give the highest
(lowest) rating on the post-consumption scale, Constructvalidityforthefourmeasurementscales
inappropriatelyimplyinga confirmationof magni- (IS continuance intention,satisfaction,perceived
tude zero. Third, the reliabilityof a difference usefulness, and confirmation)was assessed via
LI1?YYI?I?11L
OF19MEN
Chi-square
Variables Constrained Chi-square Degrees of Freedom Differencea
None 116.21 71
CI + S 167.98 72 51.77
CI + PU 204.29 72 88.08
CI + C 210.24 72 94.03
S + PU 279.70 72 163.49
S+C 180.43 72 64.22
PU + C 213.37 72 97.16
Legend: CI = IS continuanceintention,S = Satisfaction,
PU = Perceived usefulness, C = Confirmation.
aAllX2differenceswere significantat p = .001 level.
0.892
Perceived 0.294a
usefulness )
(0.20)
Sr~ - V..I
five hypothesized paths in the model were respectively). The effect of perceived usefulness
significantat p < 0.01. Intentionto continue IS on users' intention in both acceptance and
use was predictedby satisfaction (P = 0.57) and continuance contexts attests to the robustness
perceived usefulness (P = 0.29), which explained and salience of this association across temporal
32% and 9% of the intentionvariance respec- stages of IS use. However,the size of this effect,
tively. In additionto its direct effect, perceived relativeto that of affect, seems to decrease over
usefulness also had a small indirecteffect on IS time. ECT provides some intuitionfor under-
continuance intention(1 = 0.13) via the satisfac- standing this change. Users' pre-acceptance
tion construct,explaining1.7%of the variance in attitudeis based solely on cognitive beliefs (e.g.,
the dependent variable. Satisfaction,in turn,was usefulness, ease of use) formed potentiallyvia
predictedby confirmation(P= 0.53) and perceived second-hand informationfrom referent others,
usefulness (1 = 0.23), which explained 28% and popularmedia, or othersources. These influence
5% of the satisfaction variance respectively. sources may be biased. Hence, user attitude
Confirmationalso had a small indirecteffect (P = potentiallymay be inaccurate, unrealistic, and
0.10) on satisfaction,viathe perceivedusefulness uncertain. Incontrast, post-acceptance satisfac-
construct. Finally,confirmationwas a significant tion is grounded in users' first-handexperience
predictor of perceived usefulness (P = 0.45), with the IS. It is, therefore, more realistic,
explaining20%of the usefulness variance. Impli- unbiased, and less susceptible to change (Fazio
cations of these effects for IS continuance are and Zanna 1981). Users' may accommodatethis
discussed in the next section. uncertainty in affect by underweighting more
uncertainattitude in their acceptance decisions
and overweighting more certain satisfaction in
continuancedecisions.
Discussion of Results
The above findinghas importantimplicationsfor
IS practice. While ignoringpre-acceptance user
ExplainingIS Continuance attitude may not severely impact IS productor
Intention
service acceptance among new users, ignoring
Results of the studysupportECT'scontentionthat post-acceptanceuser satisfactioncan have disas-
satisfactionwith IS use is the strongest predictor trous consequences for user retention (contin-
of users' continuance intention (R2 = 0.32), uance). Because perceived usefulness is more
followed by perceived usefulness as a significant crucialforacceptance intentionand satisfactionis
but weaker predictor (R2 = 0.09). Coupled with a more dominantfor continuanceintention,IS firms
and other supply-side institutionsresponsiblefor
strong intention-behaviorassociation theorized
and validatedin priorIS use research (e.g., Davis enhancingIS use should adopta two-foldstrategy
et al. 1989; Taylor and Todd 1995), the above for maximizingtheir return on investments in
associations suggest that satisfaction and per- customertraining:informnew (potential)users of
ceived usefulness are important(indirect)predic- the potentialbenefits of IS use and educate old
tors of actual continuancebehaviors. (continued)users on how to use IS effectivelyso
as to maximizetheirconfirmationand satisfaction
with IS use.
Comparing the above results with prior TAM-
based studies of IS acceptance, some interesting
patterns emerge. Perceived usefulness was a Further, satisfaction also may be the key to
strongerpredictorof acceptance intentionin TAM explaining the IS acceptance-discontinuance
than attitude(Davis et al. 1989; Taylorand Todd anomaly(user discontinuanceof IS afterits initial
1995), while satisfaction was the stronger phenomenonin IS
acceptance),a little-understood
predictorof continuance intention in this study use research. TAM,which predictsuser intention
than perceived usefulness. Perceived usefulness based on perceived usefulness and attitude,
is a cognitivebelief,whileattitudeand satisfaction cannot explainthis anomalysatisfactorily,unless
both reflectuser affect (pre-and post-acceptance eitheror both determinantschange frompositive
to negative from pre-acceptance to post-accep- Noteworthyin this context are the similaritiesand
tance phases. Because satisfaction was the differences between confirmationand service
stronger predictor of continuance intention quality(SERVQUAL5) constructsin the marketing
(explaining32% of intentionvariance) relativeto literature(Parasuramanet al. 1988). Bothconfir-
perceived usefulness (the direct and indirect mation and SERVQUALrelate to a customer's
effects of perceived usefulness jointlyexplained evaluationof the expectation-performance discre-
10%of the intentionvariance)in this study, users pancy. Nonetheless, they differ on several
dissatisfiedwithIS use (due to disconfirmationof counts. First, SERVQUALapplies only to ser-
expectation) may discontinue IS use, despite vices, whileconfirmationapplies to both products
having positive perceptionsof its usefulness. In andservices. Second, expectationin SERVQUAL
other words, dissatisfaction, and not perceived is anchored to a "desired"level based on a
usefulness, is the necessary condition for IS general understanding of the overall class of
discontinuance. Futurestudies that extend TAM service, ratherthan a reasonable service-specific
into continuance contexts should, therefore, level forconfirmation.Desires congruencyis sub-
integrate satisfaction and its antecedents (e.g., stantiallydifferentfrom expectation congruency
confirmation)with existing TAM constructs to (confirmation) intheirmagnitudesand satisfaction
providea betterunderstandingof this anomaly. effects (Spreng and Olshavsky 1993). Third,
SERVQUALis an inferred confirmationscale.
Hence, it suffers from the limitationsof inferred
confirmation(consistency bias, ceiling or floor
ExplainingSatisfactionwith IS Use effect, and low reliability)discussed earlier. In
contrast,the perceivedconfirmationscale (typical
SatisfactionwithIS use was predictedprimarily by of confirmation)is not subjectto these limitations.
users' confirmationof expectationfromIS use and Fourth,adaptationsof SERVQUALto IS contexts
secondarilybytheirperceivedusefulness of initial (e.g., Pitt et al. 1995) have been employed in
IS use (R2= 28%and 5%respectively). However, circumstanceswhere users are well aware of the
because these determinantsjointlyexplain only service provider'shardware, physical facilities,
33% of the satisfaction variance, IS satisfaction and employees. Thus, they can make an in-
formedjudgmenton items such as whetherIS has
may have additionalsalient predictorsthan those
identifiedusing the ECTlens. up-to-datehardwareand software, IS employees
are well-dressed and neat in appearance, IS
insists on error-freerecords,and so forth. These
As expected from ECT, confirmationwas a
items are not applicable to most e-commerce
stronger predictorof satisfactionthan perceived
services, given their remote and anonymous
usefulness in this study. Confirmationis a new
nature.The above issues motivatedthe construc-
constructin IS use research. Conceptualizingthis
tion of a new confirmationscale for this study.
construct and validatingits effect on IS contin-
uance intention(viasatisfaction)are twocontribu-
tions of this study. The larger effect size of
confirmation,relative to perceived usefulness,
suggests that users view realizingtheir expecta- 5SERVQUAL is a 45-itemmeasureof servicequalityin
tion as being more salient than instrumentality
of whichconsumersassess theirexpectationpriorto
IS in forming affect and intention about IS service(via22 items),perceivedperformance afterthe
continuance. Confirmationalso had a significant service (via another22 items), and a single-item
measureofoverallservicequality.The22 itemstapinto
positive effect on perceived usefulness, sug- five dimensionsof service quality:tangibles(e.g.,
gesting that user perceptionof IS instrumentality appearanceof personnel),
physicalfacility,equipment,
toperform
(ability
reliability servicedependably
promised
may also be adjustedby theirextent of confirma- and accurately),responsiveness(willingnessto help
tion. While prior TAM-based studies suggest customers),assurance (employees'knowledgeand
temporal changes in perceived usefulness, this abilityto inspireconfidence),and empathy(caring,
study provides some insighton possible causes individualizedattention).SERVQUAL is computedin
termsof the differencescores between the two 22-item
underlyingsuch changes. pairs,plus the overallqualityitem.
Confirmationis a cognitive belief (the extent to shows that individualratingson each of the three
which users' expectation of IS use is realized continuance intentionitems ranged from 1 to 7.
duringactual use) derived from priorIS use. In Thus, some respondents demonstratedan intent
addition,it influences subsequent IS use via the to discontinue OBD use. Given that online
satisfaction (affect) and intention constructs. bankingis stillin its infancy,it may be reasonably
Such belief-affect-intention-behavior-belief caus- expected that some of these intended discon-
ality suggests the presence of feedback loops, tinuers would eventually discontinue OBD use,
whereby users' beliefs, affect (satisfaction),and whichwould alleviatethe above bias.
intentionare continuallyrefinedand modifiedfrom
their initialacceptance to long-termcontinuance Fourth, because survey respondents had used
or discontinuance. Such feedback is also evident OBD between two and 36 months at the time of
in other contexts like the linkbetween user parti- the survey,one mayquestionwhetherperceptions
cipation (behavior) and involvement (affect) in of recent IS acceptors were systematically
system development (Hartwickand Barki1994). different from those of earlier acceptors (as
Otherwise,it has received littletheoreticalatten- suggested by Rogers 1995). As an approximate
tion thus far. Understandingsuch complex and test of this bias, the sample was dividedintotwo
dynamicaspects of IS use motivations,although (early and later) groups via a median split. The
beyondthe scope of the currentstudy, represents research model was then runseparatelyfor each
an interestingand potentiallyfruitfularea forfuture group. Despite minordifferences in model fit and
research. pathcoefficients,the overallpatternof resultswas
similaracross bothgroups. Even if earlieraccep-
tors differedfromlateradopterson certaindimen-
Limitationsof the Study sions (e.g., innovativeness),such differences do
not appear to influencetheircontinuancepercep-
This study suffers fromseveral limitations. First, tions significantly.
given the low response rate, the results may be
influenced by non-response bias. Operational Finally,an ideal empiricaldesign for testing ECT
limitationspreventedthe researcher from polling wouldbe a longitudinalcomparisonof customers'
a section of non-respondentsto test for this bias. pre-acceptance and post-acceptanceperceptions,
However,sample demographics(age and income in orderto faithfullycapturethe complex,dynamic
level) were comparedwith"average"OBDcusto- interrelationshipsbetweenacceptance and contin-
mers, as providedby this online bank. Difference uance decisions. However, the cross-sectional
of means tests found that respondents did not nature of this study restricted such temporal
differsignificantlyin age or income level fromthe comparisons. Interestingissues such as changes
target population. Therefore,bias is unlikely. in perceived usefulness and attitude across
acceptance and continuance phases were not
Second, the novelty associated with an online examined. These represent potential ways of
mode of data collection may have biased the extendingthe currentresearch.
survey responses. A single item check for this
novelty revealed that all respondents were
comfortable with online surveys (in varying
degrees). In addition,94% of the respondents Conclusions
had earlierfilledout online forms, suggesting the
absence of noveltyeffect. The goal of this paper was to identify salient
determinantsof IS continuance intentionand to
Third, because respondents were current (and understand how they influence the dependent
continuing)users of OBD,they may be biased in variable. Toward that goal, ECT was adapted
their perceptions, in contrast to OBD discon- from the consumer behavior literatureand inte-
tinuers. Even though discontinued users could grated with priorIS use research to theorize a
not be contacted to test forthis bias, the rawdata model of IS continuance. Data collected from a
MISQuarterly
Vol.25 No.3/September
2001 367
Spreng, R. A.; MacKenzie,S. B.; and Olshavsky, Yi, Y. "A CriticalReview of Consumer Satis-
R. W. "AReexaminationof the Determinantsof faction," in V. A. Zeithmal (ed.), Review of
Consumer Satisfaction,"Journal of Marketing Marketing(4) AmericanMarketingAssociation,
(60), July 1996, pp. 15-32. Chicago, 1990, pp. 68-123.
Swan, J. E., and Trawick,I. F. "Disconfirmation
of Zmud, R. W. "Diffusionof Modern Software
Expectations and Satisfaction with a Retail Practices: Influenceof Centralizationand For-
Service,"Journal of Retailing(57), Fall 1981, malization," Management Science (25:10),
pp. 49-67. 1982, pp. 966-979.
Tan, M., and Teo, T. S. H. "FactorsInfluencing
the Adoptionof InternetBanking,IJournalof the
AIS (1:5), July 2000, http://jais.isworld.org/
articles/1-5/. Aboutthe Author
Taylor,S., and Todd, P. A. "Understanding Infor-
mationTechnologyUsage:A Test of Competing Anol Bhattacherjee is an associate professorof
Models,"InformationSystems Research (6:2),
computerinformationsystems at the Universityof
1995, pp. 144-176. South Florida. He received Ph.D. and M.B.A.
Tse, D. K., and Wilton,P. C. "Modelsof Consu-
mer Satisfaction: An Extension,"Journal of degrees fromthe Universityof Houstonand B.S.
and M.S. degrees from IndianInstituteof Tech-
MarketingResearch (25), May 1988, pp. 204-
212. nology (India). His research interests include
behavioralissues in electronic commerce, infor-
Vroom,V. H. Workand Motivation,Wiley, New
mationtechnology (IT)acceptance and use, and
York,1964
IT-enabledorganizationaltransformation.
Hisprior
Westbrook, R. A., and Reilly, M. D. "Value-
research has been published in Information
Percept Disparity:AnAlternativeto the Discon-
firmationof ExpectationTheory of Consumer Systems Research, Journal of MIS, Decision
Satisfaction," in Advances in Consumer Sciences, Decision Support Systems, IEEE
Research, K. B. Monroe(ed.), Association for Transactionson Systems, Man,and Cybernetics,
ConsumerResearch, AnnArbor,MI,1983, pp. DataBase, Information &Management,and other
94-99. refereedjournals.
MISQuarterly
Vol.25 No.3/September
2001 369
Appendix
Scale Items and Exploratory Factor Analysis
Factor Loadingsa
Scale Item 1 2 3 4
IS continuanceintention:
CI1 I intendto continue using OBD ratherthan discontinue .117 .027 -.030 .699
its use.
C12 Myintentionsare to continue using OBDthan use any .040 -.016 .020 .790
alternativemeans (traditionalbanking).
C13 If I could, I would liketo discontinue my use of OBD -.044 .055 .063 .764
(reverse coded).
Satisfaction:
Howdo you feel about youroverall experience of OBD .759 .071 .032 -.033
S1 use:
VerydissatisfiedNery satisfied.
S2 Very displeasedNery pleased. .626 -.046 .076 .159
S3 VeryfrustratedNerycontented. .829 -.007 -.034 .023
S4 delighted.
Absolutelyterrible/Absolutely .735 .019 .044 .010
Perceived usefulness:
PU1 Using OBDimprovesmy performancein managing .097 .834 -.021 -.063
personalfinances.
PU2 Using OBD increases my productivityin managing -.050 .809 .040 .027
personal finances.
PU3 Using OBDenhances my effectiveness in managing .058 .784 -.034 -.023
personalfinances.
PU4 Overall,OBD is useful in managing personalfinances. -.087 .721 .054 .132
Confirmation:
C1 Myexperience withusing OBDwas betterthan what I -.051 .019 .772 .051
expected.
C2 The service level providedby OBDwas betterthan what -.047 -.018 .754 .023
I expected.
C3 Overall,most of my expectations fromusing OBDwere .056 .011 .783 -.078
confirmed.
Eigenvalue 6.201 1.819 1.410 1.008
factoranalysis (principalcomponents withdirectobliminrotation).
aExploratory