Internet shoppng and Internet bankng n sequence Athanasos G. Patsots a, Tm Hughes b and Don J. Webber c a Department of Marketng, Deree College, Amercan College of Greece, Athens, Greece b Department of Busness and Management, Unversty of the West of England, Brstol, UK c Department of Accountng, Economcs and Fnance, Unversty of the West of England, Brstol, UK Abstract Ths paper presents an emprcal nvestgaton nto the factors that shape the propensty to use the Internet for shoppng and bankng through applcaton of bvarate probt regresson technques to data sourced from a survey of 259 respondents n Athens, Greece. Based on the observaton that Internet bankng usage typcally requres famlarty wth Internet shoppng, we estmate the margnal effects of the determnants of Internet bankng use condtoned on the determnants of Internet shoppng use. Our results suggest that not controllng for ths condtonng wll bas estmates and could lead to ncorrect polcy-recommendatons. For nstance, personal capacty s found to be an mportant determnant of the propensty to use Internet bankng n a non-sequental approach but t s found to have no sgnfcant effect after condtonng. In partcular, our results suggest that polcymakers should emphasse usefulness attrbutes of computer-based nnovatons when attemptng to ncrease the use of the Internet for bankng by people who already use the Internet for shoppng. Keywords: Internet bankng; Internet shoppng; adopton rate; bvarate probt regresson; condtonal margnal effects Address for correspondence: Dr Athanasos Patsots, Department of Marketng, Deree College, Amercan College of Greece, Athens, Greece. Emal: agpatsots@acg.edu 1
1. Introducton The Internet represents a huge source of nformaton that can be organzed and retreved n many dfferent ways based on ndvdual users needs (Mahajan et al., 2000). It facltates communcaton and shoppng through computer-medated envronments and t s a market where a large varety of new technologes and nterdependent products are ntroduced (Mahajan et al., 2000; Varadarajan and Yadav, 2002). It can also be a dscontnuous nnovaton process and lead to new product developments. Followng the ntroducton and acceptance of Internet shoppng, new technologcal nterfaces developed by banks, such as Internet bankng, are nnovatve delvery and communcaton channels where new products and servces are ntroduced. These nnovatons have facltated nteracton and the buldng of relatonshps between banks and ther customers (Tapp and Hughes, 2004). New technologes and especally the developments of self-servce technologes present several challenges for banks n terms of ther customer relatonshps. Banks that offer Internet bankng servces can beneft from lower costs due to the utlzaton of less human and physcal resources and the potental of economes of scale n bank operatons (Sh et al., 2008). Consumers transferrng ther decson makng processes from tradtonal offlne to onlne can engender cost and tme savngs benefts (Sh et al., 2008) at the expense of varous rsks (Durkn, 2007). Consequently banks need to alter ther operatons and nternal and external communcaton meda, and such major changes can encounter resstance. 1 A number of artcles present nvestgatons of the determnants of Internet use for a varety of servces. Such studes typcally examne ether one Internet servce n solaton or assume away structure or order between Internet servces. Ths paper purports that there s a sequence of Internet usage choces, wth consumers frst becomng famlar wth the Internet for ther shoppng experence and, once profcency n ths area has been acheved, consumers wll then consder usng the Internet for bankng servces. Based on the dea of a condtonal and sequental lnk between Internet shoppng and Internet bankng we proceed to examne emprcally the factors that nfluence the rate of Internet bankng adopton. Our results strongly support the assumpton of assocaton between Internet bankng and Internet shoppng, and once sequencng has been ntegrated nto the modellng approach we dentfy potental conflctng results and mportant polcy levers. The paper s structured as follows. The next secton presents a revew of extant lterature and our conceptual model. Secton 3 detals the data and the modellng approach. Secton 4 s a dscusson of fndngs and mplcatons, and Secton 5 concludes. 2. Theoretcal background Innovatons can be defned n terms of the amount of behavoural change necessary to use the nnovaton effectvely; they can be classfed along a contnuum from the least to the most dsruptve whch s related to the extent to whch the nnovaton s functonally new (Robertson, 1971). Technologcal nnovatons can be dscontnuous (Moore, 1991), are typcally vewed as beng rooted n new nformaton and computer-based technologes, and can dsrupt exstng patterns of behavour (Ftzsmmons and Ftzsmmons, 2011; Lttler, 1 For example, customers do not always welcome technology or they may ncreasngly use a combnaton of bankng methods. The theoretcal lterature on dffuson research of such technologcal nnovatons s welldeveloped but t lacks emprcal evdence on non-adopter behavour and focuses manly on mental behavour (Hernandez et al., 2009). There s lmted evdence on the possble dfferences between pre-adopton and usage behavour (Hernandez and Mazzon, 2007). 2
2001; Veryzer, 1998b). Ther usage can nvolve a very hgh degree of technologcal uncertanty, a sequence of nnovatons based on the core applcaton, a longer development process, and a greater dstance from the end user n terms of customer famlarty wth the nnovaton and the tme t takes to evolve (Veryzer, 1998a, 1998b). Thus, technologcal nnovatons can requre a change n the behavour of potental adopters and the development of new sklls. The ntroducton of the core applcaton of the Internet (.e. lnkng computers n networks) has spawned a sequence of Internet based nnovatons that have facltated communcaton and shoppng. Such nnovatons nclude Internet shoppng and Internet bankng that requred major behavoural changes by potental adopters n ther busness and personal relatonshps. Internet shoppng s now wdespread and typcally ncludes books, cosmetcs, consumer durables and servce operatons, such as retalng, entertanment and travel. Internet bankng s also avalable and requres a more sophstcated applcaton of Internet technologes to satsfy consumers bankng needs, and t dffers from other selfservce nnovatons n fnancal servces snce t requres major changes n behavour (.e. a completely new way of consumer bankng). An order of successon There may be a logcal dependence of engagement wth nnovatons, and usage of Internet shoppng and Internet bankng may be a prme example. Ths paper purports a sequence of events whereby ndvduals make a seres of decsons: 1) Decde (conscously or otherwse) to use the Internet; 2) After some famlarty of use wth the Internet has been acheved, a next step n usng the Internet s for shoppng; 3) After some famlarty of use wth the Internet for shoppng has been acheved, a next step n usng the Internet s for bankng. Decson (1) s beyond the scope of ths paper. Ths paper examnes emprcally the factors that contrbute to decsons (2) and (3) for a sample of Internet users. The decsons above are each assocated wth dfferent, albet potentally sequental nnovatons that requre nteracton wth technologcal nterfaces and necesstate a degree of behavoural change by potental adopters. The conceptual framework for ths study, shown n Fgure 1, llustrates that someone must adopt the Internet frst, and then must adopt Internet shoppng as a prerequste to decdng whether to adopt Internet bankng. The orderng purported n Fgure 1 corresponds to both the sequence of ntroducton of the above technologcal nterfaces (shoppng and bankng) and the degree of the behavoural change requred by potental adopters to use these functonal nnovatons. As a result, there s a sequence of two dchotomous decsons addressng the respectve condtonal lnk: Factors nfluencng rates of adopton {Insert Fgure 1 about here} The rate of adopton refers to the frequency of new users of the nnovaton out of ts market potental (Rogers, 2003). Dffuson research has explctly consdered the communcaton process for the dffuson of a technologcal nnovaton usng mathematcal models (e.g. Bass, 1969; Fourt and Woodlock, 1960; Llen et al., 1992; Mansfeld, 1961). Followng these 3
classc works, a number of models have been developed to capture other dynamcs of the nnovaton dffuson process, such as the nfluence of the marketng mx on new product dffuson (Mahajan et al., 1990, 2000). The Bass model and ts revsed forms have been used n marketng for forecastng nnovaton dffuson (.e. the lfecycle dynamcs of a new product) n retal servce and consumer durable goods, among others (Mahajan et al., 1990). However, some of the assumptons underlyng the Bass model have been questoned, such as that market potental remans constant over tme and that adopton s an ndvdual decson (Llen et al., 1992). Moreover, to explan consumer acceptance dffuson research has focused on ) the perceved attrbutes of an nnovaton and ) the potental adopter s characterstcs. The emprcal lterature emphasses the mportance of gender, age and ncome dfferences (e.g. Gan et al., 2006) but offers lttle consstency on the mportance of other ndvdual characterstcs n explanng adopton of an nnovaton (e.g. Wang et al., 2008), and ths nconsstency may be based on the varety of research contexts, the nature of the nnovaton, the sample s representaton of the target populaton and the geographc context. The research also ndcates that the perceved attrbutes of an nnovaton are stronger predctors of adopton than the personal characterstcs of potental adopters (Gatgnon and Robertson, 1989; Lockett and Lttler, 1997; Moore and Benbasat, 1991; Rogers, 2003). Although the perceved attrbutes of an nnovaton can be used to explan adopton rates, researchers have devoted lttle effort n examnng how these factors affect adopton rates (Rogers, 2003). Wthn the context of Internet-based nnovatons, the lterature has examned emprcally the factors that nfluence consumer atttudes and ther effects on ntentons towards adoptng these servces. Many of these studes are based on the works of Rogers (2003) (nnovaton characterstcs), Davs (1989) and Davs et al., (1992) (technology acceptance model - TAM), Parasuraman (2000) (technology readness ndex - TRI), Dabholkar (1996) (servce qualty) and extensons and combnatons of these theores. In addton, the constructs of perceved rsk (Bobbtt and Dabholkar, 2001; Cunnngham et al., 2005; Curran and Meuter, 2005), nteractvty (to understand future buyer-seller actvtes n the electronc marketplace) (Sawhney et al., 2005; Varadarajan and Yadav, 2002; Yadav and Varadarajan, 2005a; Yadav and Varadarajan, 2005b), human nteracton (Glbert et al., 2004; Makarem et al., 2009; Smon and Usuner, 2007), perceved ablty/capacty (Btner et al., 2002; Ellen et al., 1991; Walker et al., 2002), tme and energy savngs (Walker and Johnson, 2006), aspects of servce convenence (Berry et al., 2002) and consumer demographcs and personal characterstcs (Dabholkar and Bagozz, 2002; Ellott and Hall, 2005; Lee et al., 2010) have all been found to be mportant factors n explanng adopton. The above studes and related theores could be examned n order to dentfy what nfluences the ntal adopton of nnovatons (Mahajan et al., 2000) but equally there s space to examne post-adopton or usage behavour (Hernandez et al., 2009; Mahajan et al., 2000), whch s partcularly mportant for servce nnovatons that are used frequently after adopton. The construct of an e-servce qualty s an mportant factor contrbutng to repeat purchases from webstes (Zethaml et al., 2002) and the measurement of ths construct must rest on percepton data whch s based on the use (adopters) or non-use (non-adopters) of the Internet for shoppng. The most frequently found e-servce qualty percepton factors are: relablty, prvacy/securty, ste desgn, ease of use, responsveness, control, accessblty, speed of delvery, enjoyment and accuracy (Bobbtt and Dalholkar, 2001; Jauda et al., 2002; Jun and Ca, 2001; Shamdasan et al., 2008; Zethaml et al., 2002). There are some overlaps between aspects and constructs wth relablty, prvacy and securty concerns beng aspects of trust (Yousafza et al., 2009) and wth relablty also beng an aspect of perceved rsk (Curran and Meuter, 2005). Davs s constructs of usefulness and 4
ease of use are smlar to Rogers constructs of perceved relatve advantage and complexty, respectvely, whle aspects of relatve advantage and convenence are related. Rogers compatblty construct s broadly defned and captures knowledge aspects, such as personal ablty/capacty and cultural values. 2 There s lmted emprcal evdence on non-adopter behavour, whch s partcularly mportant for stuatons where adopters of the core applcaton, such as the Internet, have not adopted or have adopted only some of ts applcatons. The next secton presents an emprcal development of the adopton model to test for and explan a sequental lnk between adopton rates of Internet shoppng and adopton rates of Internet bankng, and hence to fll ths gap n the lterature. 3. Data and method Ths study draws on a data set collected va questonnares dstrbuted to organzatons n Athens, Greece, and has been descrbed n detal n Patsots et al. (2012). That study employed categorcal data to capture respondent demographcs and 7-pont Lkert scales to measure respondent perceptons on characterstcs of nnovaton. Descrptons of the varables used n ths study are presented n Table 1. Modellng approach {Insert Table 1 about here} Our proposton s that there s a clear sequence of decsons relatng to the adopton of computer-based nnovaton technologes: frst the Internet user must make the decson to use the Internet for shoppng (Yes = 1; No = 0) and then make a smlar decson to use the Internet for bankng (Yes = 1; No = 0). These two dchotomous choces are tradtonally modelled separately as dependent varables n regresson; we contnue to do ths but we model them smultaneously ntally and then sequentally. An alternatve way to thnk about these questons s to consder t as a (potental) sample selecton ssue: f these are sequental decsons then ndvduals who are not Internet shoppers are much less lkely to consder the Internet for bankng. Hence any efforts to dentfy the contrbutory effect of explanatory varables on the decson to use the Internet for bankng servces may produce based results, as part of the sample are not consderng usng the Internet for bankng, as they have not frst engaged enough n shoppng on the Internet. Therefore, estmates of the effect of explanatory varables on Internet bankng should be condtonal on the factors that nfluence adopton of the Internet for shoppng. An approprate method to employ n ths nstance s the bvarate probt regresson approach and condtonal margnal effects can be obtaned where P(Internet Banker = 1 Internet Shopper = 1). Gven margnal effect estmates of ths condtonal probablty, t 2 Some of these constructs may not be relevant to usage behavour. For nstance, technology readness factors focus on consumer trats and personal orentaton, such as nnovatveness, but do not necessarly predct technology usage (Meuter et al., 2003). Human nteracton may be an nhbtor to Internet shoppng and Internet bankng and s related more to pre-adopton behavour. As people use a combnaton of shoppng modes, Internet facltes may only serve as a substtute channel for users and hence t s less lkely for Rogers observablty to be an mportant factor (Black et al., 2001; Lassar et al., 2005). Conversely, Rogers tralablty construct (the lack of tral opportuntes to start usng Internet bankng) may be an mportant nhbtor to adopton. 5
would be possble to dentfy whether respondent demographcs and perceved attrbutes contrbute ether to the decson to partcpate n Internet bankng or to the decson to partcpate n Internet shoppng, or to both decsons. We adopt the formal model for estmatng the probabltes accordng to Greene (2003). Let y 1 be a latent varable that denotes the probablty that an ndvdual s an Internet banker, whch s dependent on a range of contrbutory factors, X 1. Also let X 2 be a latent varable that denotes the probablty that the ndvdual s an Internet shopper, where ths s also dependent upon a range of factors, X 2. The model s represented as follows: y1 1X1 1 y2 2X 2 2 where the values for y 1 are observable and related to the followng bnary dependent varables, on the bass of the followng condtons: Internet banker 1, f y 1 0 Internet banker 0, f y 1 0 and Internet shopper 1, f y 2 0 Internet shopper 0, f y 2 0 where Internet shopper 1 denotes that the ndvdual s an Internet shopper, and Internet banker 1 denotes that the ndvdual s an Internet banker. The errors ( 1, 2 ) are assumed to have the standard bvarate normal dstrbuton, wth V ) 1 V( ) and ( 1 2 Cov ( 1, 2 ). Thus the ndvdual s probablty of beng an Internet banker can be wrtten as: P( Internet banker) P ( Internet banker 1, Internet shopper 1) P X x, X x ) ( 1 1 2 2 x2 x1 2( z, z ; ) dz 1 2 1 ( 1X1, 2X 2; ) F dz 2 where F denotes the bvarate standard normal dstrbuton functon wth correlaton coeffcent. 3 The bvarate probt model has full observablty f Internet banker and Internet shopper are both observed n terms of all ther four possble combnatons [) Internet banker = 1, Internet shopper = 1, ) Internet banker = 1, Internet shopper = 0, ) Internet banker = 0, Internet shopper = 1, and v) Internet banker = 0, Internet shopper = 0 ]. Category () s always equal to zero n our sample. If there s a clear sequence 3 2 2 2 (1/ 2)( x1 x 2 1 2 )/(1 ) 2 1/ 2 Greene (2003) shows that the densty functon s gven by: 2 2 x x e / 2 (1 ). 6
of decsons n adoptng Internet shoppng and then Internet bankng then t s approprate to deem ths to be a naturally constraned complete set of observatons and effectvely provdes us wth full observablty n our data. It s known that full observablty leads to the most effcent estmates (Ashford and Sowden, 1970; Zellner and Lee, 1965). The followng secton presents the results from applcaton of the bvarate probt method to the case explaned above, and specfcally wll dscuss the condtonal margnal effects obtaned from P(Internet Banker = 1 Internet Shopper = 1). 4. Results A sequental structure Fgure 2 shows the structure of our purported sequental structure. It shows that approxmately one-thrd of Internet users do not use the Internet for shoppng. More mportantly for our research, t llustrates that no one uses the Internet for bankng f they do not use the Internet for shoppng. In our sample, less than 30 percent of respondent use the Internet for bankng, and ths represents more than 40 percent of the respondents who use the Internet for shoppng. Fgure 2 corroborates the perspectve that the decsons to use the Internet for shoppng and for bankng may be sequental as t s n lne wth the proposton that usng the Internet for shoppng s a prerequste for usng the Internet for bankng. {Insert Fgure 2 about here} We draw our data from Patsots et al. s (2012) study whch captures the multdmensonal nature of each construct, but ncluson of all dmensons n our regresson approach would produce excessve multcollnearty and potentally ncorrect coeffcent estmates due to ncluded varable bas. 4 To crcumvent these problems we apply prncpal component analyss to each construct as a data reducton technque. We retan each construct s prncpal component and use these as regressors nto our bvarate probt analyss. 5 Table 2 presents descrptons of these prncpal components. Bvarate regresson results {Insert Table 2 about here} Our regresson results are presented n Table 3. They ndcate there s a very strong assocaton between the choces to use the Internet for shoppng and to use the Internet for bankng; ths s llustrated by the hghly statstcally sgnfcant result of the log-lkelhood of rho (ch 2 = 43.2469, p < 0.000). {Insert Table 3 about here} Our ntal regresson s the choce to use the Internet for shoppng. The results corroborate extant lterature by llustratng that males and the hgher educated are more lkely 4 5 The emprcal fndngs of ths study have several lmtatons. These refer to the chosen geographc context, the chosen predctors for modellng, the cross-sectonal nature of the survey, and the possble non-response bas consderng that sample respondents may not represent the workng populaton. See Churchll and Iacobucc (2002) for a descrpton of the prncpal components approach. 7
to use the Internet for shoppng. Moreover, the results show that greater enjoyment (sgnfcant at the 10 percent level) and greater personal capacty both enhance the lkelhood that an ndvdual wll use the Internet for shoppng, whle greater perceved rsks negatvely nfluence the lkelhood that a person wll use the Internet for shoppng, all gven that they are already Internet users. The postve nfluences observed, and especally on enjoyment, are n lne wth emprcal fndngs that emotons nfluence postvely usage behavour (Martn et al., 2008; Shamdasan et al., 2008; Watson and Spence, 2007; Wood and Moreau, 2006). Dfferent factors are mportant n enhancng the lkelhood that an Internet user wll use the Internet for bankng. Greater enjoyment and greater ease of use both enhance the lkelhood that an Internet user wll also use the Internet for bankng; conversely, lower perceptons of usefulness and a lack of tral wll both dmnsh ths lkelhood. Although the core applcaton s the same, usage of subsequent nnovatons may be nfluenced by dfferent factors and therefore proceed to estmate a sequental model. Condtonal margnal effects Set wthn a bvarate probt regresson approach, t s possble to examne our proposton that the equatons should be estmated n sequence. Accordngly, Table 3 also presents two sets of margnal effects estmates correspondng to those whch are not based on a sequental approach (.e. uncondtonal) and those that are based on a sequental approach (.e. condtonal). Several ponts are worthy of emphass here. Frst, there s a slght dsagreement on whch varables enhance the lkelhood that an Internet user wll use Internet bankng as the sgnfcant coeffcents correspondng to educaton and tme-energy are not sgnfcant at tradton level of acceptance wthn the uncondtonal set up. These fndngs suggest that tme-energy and hgher educaton affect the lkelhood of usng the Internet for bankng and not only the decson to use the Internet for shoppng, as would be nferred under the uncondtonal approach. Perhaps these results are reflectng the possblty that havng hgher educaton ncreases the speed of learnng new nnovatons and therefore ncreases the lkelhood that an Internet user wll more quckly adopt Internet bankng facltes. The greater coeffcent estmates for the tme-energy varable when the structure s condtonal also emphasses the ncreasng mportance of speed when usng the Internet for bankng transactons. Second, when the varables have been dentfed as beng hghly statstcally sgnfcant, the coeffcents estmates correspondng to the condton approach are typcally much larger than for the uncondtonal approach. Ths suggests the standard uncondtonal approach leads to coeffcent estmates that are based towards zero (.e. havng no effect), and therefore underestmates ther mpact on the lkelhood of usng the Internet for bankng. The clearest example of ths bas corresponds to havng post-graduate educaton, where the margnal effect s more than four-tmes greater under the condtonal approach relatve to the uncondtonal approach. A lack of tral has a much greater hnderng effect on usng the Internet for bankng under the condton approach, beng almost 50 percent greater than expected under the uncondtonal approach. Thrd, the emprcal observaton that there s a negatve nfluence of usefulness on Internet bankng use offers new evdence on Internet bankng usage behavour; ndeed our emprcal estmates suggest that usefulness s the most mportant lever for polcy formaton, and emphassng the usefulness of Internet bankng could lead to the greatest amount of new Internet bankers. Moreover, the negatve effects of usefulness on Internet bankng adopton rates contrasts wth exstng research on the nfluence of usefulness on customer ntentons and usage toward technology-based servce offerngs (TAM model and extensons: Glbert et 8
al., 2004; Hernandez et al., 2009; McKechne et al., 2006; Ozdemr and Trott, 2009). In a study comparng Internet bankng acceptance to older self-servce technologes for bankng, such as the ATM and phone, Curran and Meuter (2005) found that ease of use and usefulness are not mportant predctors of Internet bankng, although there are mportant for the adopton of ATM. They concluded that for an nnovaton to be successful,.e. reach the majorty of potental customers, t must be both useful and easy to use. Thus, the low adopton rates of Internet bankng may be assocated wth the negatve nfluence of usefulness. Consderng that usefulness s smlar to relatve advantage and that relatve advantage has been found to be one of the strongest predctors of an nnovaton s adopton rate (Rogers, 2003), the negatve effect of usefulness on Internet bankng propensty may further explan low adopton rates. Another explanaton for the negatve nfluence of usefulness on Internet bankng s that Internet shoppers may perceve Internet bankng to be less useful when compared to alternatve channels, such as the ATM, and therefore not nterestng for further consderaton. Fourth, a lack of tral of Internet bankng s found to be strongly assocated wth nonusage behavour, whch could be dffcult to crcumvent as tralablty of Internet bankng cannot readly be experenced before adopton. In our study, although Internet bankng users are postvely nfluenced by ease of use, enjoyment, and to some extent value tme-energy savngs, usefulness and lack of tral explan non-use. Thus, Internet shoppers may not be fully aware of the usefulness aspects of Internet bankng use. Ths s supported by the negatve nfluence of lack of tral,.e. less or no opportuntes to understand how t works do not add knowledge on the usefulness aspects. It may also be the case that customers use a combnaton of bankng methods other than the onlne. The lterature supports a mult-channel ntegraton (Coelho and Easngwood, 2003; Zethaml et al., 2009). Earler results ndcate that consumers have a preference for a mx of delvery channels rather than exclusve relance upon anyone sngle channel (Howcroft et al., 2002), and new delvery channels tend to complement rather than replace the exstng ones (Hughes, 2006). Fnally, t s nterestng n the results of ths study that nteractvty, and any perceved rsks, securty and prvacy concerns do not nfluence adopton rates. Ths also contrasts emprcal research on the precedng nfluences. It may be the focus of extant work on pre-adopton behavour, compared to the method adopted n ths study examnng usage behavour, that explans ther non-mportance. 5. Concluson and mplcatons Ths study examned emprcally the condtonal lnk between Internet shoppng and Internet bankng n order to dentfy key factors nfluencng Internet bankng adopton rate. The focus was on the behavour of Internet users that adopt and use Internet shoppng frst n order to decde whether to use the Internet for ther bankng needs. The applcaton of bvarate probt regresson analyses revealed that for those wth Internet shoppng experence, the probablty of usng Internet bankng servces s postvely affected by customer perceptons on enjoyment, ease of use, and tme-energy savngs, and negatvely nfluenced by usefulness and lack of tral. The results of ths study support the presence of a condtonal lnk, and suggest Internet shoppng and Internet bankng should be examned sequentally. Emprcal research usually examnes dffuson theores wthn the context of one technologcal nnovaton. The results of ths study strongly support the proposton that dffuson research should examne new technologcal nnovatons sequentally when they are based on a core applcaton and the adopton of the frst would probably lead or not to the adopton and use of the second. 9
The above fndngs suggest further research nqury and practcal mplcatons for servce provders. Future research could further apply the sequental modellng approach n smlar servce contexts (e.g., smart phone, personal assstance shoppng) and n other geographc areas exhbtng dfferent adopton rates. Ths s partcularly applcable when technologcal nnovatons are ntroduced, or have been ntroduced, sequentally n the market, and there s a lack of an accurate sample frame. Usage behavour could be examned further to dentfy the most relevant predctors and confrm the key nfluences revealed by ths study. As the chosen nnovatons are n the market for some tme now, dfferent factors mght be more mportant nfluences to ther adopton rates. Ths would facltate a better understandng of non-usage behavour too. In addton, longtudnal studes based on panel data could provde better ndcaton of the future drecton of the condtonal lnk and the most mportant nfluencng factors. Fnally, the evdence provded n ths research study on the condtonal lnk between Internet shoppng and Internet bankng suggests that even though the bass of a technologcal nterface s the computer, new servces developed through ths can have dfferent perceved characterstcs. Thus, banks should be aware that Internet bankng use presents unque characterstcs compared wth the other alternatve channels as well as wth related servces. Bank customers that use the Internet for shoppng but not for bankng should be nformed for the usefulness aspects of onlne bankng, as well as be gven the opportunty to experence how t works, gven the perceved lack of tral. In partcular, the usefulness aspects should be examned further through a comparson of the alternatve methods for bankng. It may be that some customers fnd Internet bankng to be less useful when compared to other method(s), or they may not be fully aware of ts benefts. Exstng users of Internet bankng facltes value the enjoyment aspects, so servce provders should make the experence enjoyable so as to sustan postve emotons. As a result, good knowledge of the behavour of users would facltate ther effort to understand those lkely to be new users and ncorporate ths understandng n ther marketng strategy to develop a more targeted communcatons polcy that would generate useful customer feedback. References Bass, F.M. (1969). A new product growth model for consumer durables. Marketng Scence, 15, 215-227 Berry, L.L., Seders, K. and Grewal, D. (2002). Understandng servce convenence. Journal of Marketng, 66 (3), 1-17. Btner,M.J., Ostrom, A.L. and Meuter, M.L. (2002). Implementng successful self-servce technologes. Academy of Management Executve, 16 (4), 96-108. Black, N., Lockett, A., Wnklhofer, H. and Ennew, C. (2001). The adopton of Internet fnancal servces: a qualtatve study. Internatonal Journal of Retal and Dstrbuton Management 29(8), 390-398 Bobbtt, M. and Dabholkar, P.A. (2001). Integratng atttudnal theores to understand and predct use of technology-based self-servce: the nternet as an llustraton. Internatonal Journal of Servce Industry Management, 12 (5), 423-450. Churchll, G.A. and Iacobucc, D. (2002). Marketng Research: Methodologcal Foundatons, 8th ed. Orlando, FL: Harcourt College Publshers. Coelho, F. and Easngwood, C. (2003). Multple channel structures n fnancal servces: A framework. Journal of Fnancal Servces Marketng, 8 (1), 22-34. Cunnngham, L., Gerlach, J., Harper, M. and Young, C. (2005). Perceved rsk and the consumer buyng process: Internet arlne reservatons. Internatonal Journal of Servce Industry Management, 16 (4), 357-372 Curran, J.M. and Meuter, M.L. (2005). Self-servce technology adopton: comparng three technologes. Journal of Servces Marketng, 19 (2), 103-113. Dabholkar, P.A. (1996). Consumer evaluatons of new technology-based self-servce optons: An nvestgaton of alternatve models of servce qualty. Internatonal Journal of Research n Marketng, 13 (1), 29-51. 10
Dabholkar, P. and Bagozz, R. (2002). An atttudnal model of technology-based self-servce: Moderatng effects of consumer trats and stuatonal factors. Journal of the Academy of Marketng Scence 30(3), 184-201 Davs, F.D. (1989). Perceved usefulness, perceved ease of use, and user acceptance of nformaton technology. MIS Quarterly, 13 (3), 319-340. Davs, F.D., Bagozz, R.P. and Warshaw, P.R. (1992). Extrnsc and ntrnsc motvaton to use computers n the workplace. Journal of Appled Socal Psychology, 22 (14), 1109-1130. Durkn, M. (2007). Understandng regstraton nfluences for electronc bankng. Internatonal Revew of Retal, Dstrbuton and Consumer Research. 17(3), 219-231 Ellen, P., Bearden, W. and Sharma, S. (1991). Resstance to technologcal nnovatons: an examnaton of the role of self-effcacy and performance satsfacton. Journal of the Academy of Marketng Scence 19(4), 297-307 Ellott, K.M. and Hall, M.C. (2005). Assessng consumers propensty to embrace self-servce technologes: Are there gender dfferences? The Marketng Management Journal, 15 (2), 98-107. Ftzsmmons, J.A. and Ftzsmmons, M.J. (2011). Servce Management: Operatons, Strategy, Informaton Technology. 7 th ed. New York, NY: McGraw-Hll/Irwn. Fourt, L.A. and Woodlock, J.W. (1960). Early predcton of market success for grocery products. Journal of Marketng. 25 (October), 31-38. Gan, C., Clemes, M., Lmsombuncha, V. and Weng, A. (2006). A logt analyss of electronc bankng n New Zealand. Internatonal Journal of Bank Marketng. 24 (6), 360-383. Gatgnon, H. and Robertson, T.S. (1989). Technology dffuson: an emprcal test of compettve effects. Journal of Marketng, 53 (1), 35-49. Glbert, D., Balestrn, P. and Lttelboy, D. (2004). Barrers and benefts n the adopton of e-government. Internatonal Journal of Publc Sector Management 17(4) 286-301 Greene, W.H. (2008). Econometrc Analyss, Upper Saddle Rver, NJ: Prentce-Hall. Hernandez, B., Jmenez, J. and Martn, M.J. (2009). Adopton vs acceptance of e-commerce: two dfferent decsons. European Journal of Marketng, 43 (9/10), 1232-1245. Hernandez, J.M.C. and Mazzon, J.A. (2007). Adopton of nternet bankng: proposton and mplementaton of an ntegrated methodology approach. Internatonal Journal of Bank Marketng, 25(2), 72-88. Howcroft, B., Hamlton, R. and Hewer, P. (2002). Consumer atttude and the usage and adopton of home-based bankng n the UK. Internatonal Journal of Bank Marketng. 20 (3), 111-121. Hughes, T. (2006). New channels/old channels Customer management and mult-channels. European Journal of Marketng, 40 (1/2), 113-129. Jauda, S., Troccha, P.J. and Gwnner, K.P. (2002). Consumer perceptons of Internet retal servce qualty. Internatonal Journal of Servce Industry Management 13 (5), 412-431. Jun, M. and Ca, S. (2001). The key determnants of Internet bankng servce qualty: A content analyss. Internatonal Journal of Bank Marketng, 19 (7), 276-291. Lassar, W.M., Manols, C. and Lassar, S. (2005). The relatonshp between consumer nnovatveness, personal characterstcs, and onlne bankng adopton. Internatonal Journal of Bank Marketng, 23 (2), 176-199. Lee, H-J, Cho, H.J., Xu, W. and Farhurst, A. (2010). The nfluence of consumer trats and demographcs on ntenton to use retal self-servce checkouts. Marketng Intellgence and Plannng, 28 (1), 46-58. Llen, G.L., Kotler, P. and Moorthy, K.S. (1992). Marketng Models. Englewood Clffs, NJ: Prentce-Hall. Lttler, D. (2001). The dosyncrases of new product adopton for technologcally nnovatve offerngs. The Marketng Revew. 2, 3-19. Lockett, A. and Lttler, D. (1997). The adopton of drect bankng servces. Journal of Marketng Management, 13 (8), 791-811. Mahajan, V., Muller, E. and Bass, F.M. (1990). New product dffuson models n marketng: A revew and drectons for research. Journal of Marketng. 54 (January), 1-26. Mahajan, V., Muller, E. and Wnd, Y. (2000). New Product Dffuson Models, Boston, MA: Kluwer Academc. Makarem, S.C., Mudamb, S.M. and Podoshen, J.S. (2009). Satsfacton n technology-enabled servce encounters. Journal of Servces Marketng, 23 (3), 134-144. Mansfeld, E. (1961). Techncal change and rate of mtaton. Econometrcs, 29 (October), 741-766. Martn, D., O Nell, M., Hubbard, S. and Palmer, A. (2008). The role of emoton n explanng consumer satsfacton and future behavoural ntenton. Journal of Servces Marketng, 22 (3), 224-236. McKechne, S., Wnklhofer, H. and Ennew, C. (2006). Applyng the Technology Acceptance Model to onlne retalng of fnancal servces, Internatonal Journal of Retal & Dstrbuton Management 34(4/5), 388-410 Meuter, M.L., Ostrom, A.L., Btner, M.J. and Roundtree, R. (2003). The nfluence of technology anxety on consumer use and experences wth self-servce technologes. Journal of Busness Research, 56, 899-906. Moore, G.C. (1991). Crossng the Chasm, New York, NY: Harper Busness. 11
Moore, G.C. and Benbasat, I. (1991). Development of an nstrument to measure the perceptons of adoptng an nformaton technology nnovaton. Informaton Systems Research, 2 (3), 192-222. Ozdemr, S. and Trott, P. (2009). Explorng the adopton of a servce nnovaton: A study of Internet bankng adopters and non-adopters. Journal of Fnancal Servces Marketng, 13 (4), 284-299. Parasuraman, A. (2000). Technology readness ndex (TRI): a multple-tem scale to measure readness to embrace new technology. Journal of Servce Research, 2 (4), 307-320. Patsots, A.G., Hughes, T. and Webber, D.J. (2012). Adopters and non-adopters of nternet bankng: a segmentaton study. Internatonal Journal of Bank Marketng, 30 (1), 20-42. Robertson, T. (1971). Innovatve Behavour and Communcaton, New York, NY: Holt, Renhart and Wnston. Rogers, E.M. (2003). Dffuson of Innovatons, 5th ed. New York, NY: Free Press. Sawhney, M., Verona, G. and Prandell, E. (2005). Collaboratng to create: the Internet as a platform for customer engagement n product nnovaton. Journal of Interactve Marketng, 19 (4), 4-17. Shamdasan, P., Mukherjee, A. and Malhotra, N. (2008). Antecedents and consequences of servce qualty n consumer evaluaton of self-servce Internet technologes. The Servce Industres Journal, 28 (1), 117-138. Sh, W., Shambare, N. and Wang, J. (2008). The adopton of nternet bankng: An nsttutonal theory perspectve. Journal of Fnancal Servces Marketng. 12 (4), 272-286. Smon, F. and Usuner, J. (2007). Cogntve, demographc, and stuatonal determnants of servce customer preference for personnel-n-contact over self-servce technology. Internatonal Journal of Research n Marketng, 24, 163-173. Tapp, A. and Hughes, T. (2004). New technology and the changng role of marketng. Marketng Intellgence and Plannng. 22 (3), 284-296. Varadarajan, P.R. and Yadav, M.S. (2002). Marketng strategy and the Internet: An organzng framework. Journal of the Academy of Marketng Scence 30(4) 296-312. Veryzer, R.W., Jr. (1998a). Key factors affectng customer evaluaton of dscontnuous new products. Journal of Product Innovaton Management, 15, 136-150. Veryzer, R.W., Jr. (1998b). Dscontnuous nnovaton and the new product development process. Journal of Product Innovaton Management, 15, 304-321. Walker, R.H. and Johnson, L.W. (2006). Why consumers use and do not use technology-enabled servces. Journal of Servces Marketng, 20 (2), 125-135. Walker, R.H., Crag-Lees, M., Hecker, R. and Francs, H. (2002). Technology-enabled servce delvery: An nvestgaton of reasons affectng customer adopton and rejecton. Internatonal Journal of Servce Industry Management, 13 (1), 91-106. Wang, G., Dou, W. and Zhou, N. (2008). Consumpton atttudes and adopton of new consumer products: A contngency approach. European Journal of Marketng. 42 (1/2), 238-254. Watson, L. and Spence, M.T. (2007). Causes and consequences of emotons on consumer behavour A revew and ntegratve cogntve apprasal theory. European Journal of Marketng, 41 (5/6), 487-511. Wood, S.L. and Moreau, C.P. (2006). From fear to loathng? How emoton nfluences the evaluaton and early use of nnovatons. Journal of Marketng, 70 (3), 44-57. Yadav, M.S. and Varadarajan, R.R. (2005a). Understandng product mgraton to the electronc marketplace: A conceptual framework. Journal of Retalng, 81 (2), 125-140. Yadav, M.S. and Varadarajan, R.R. (2005b). Interactvty n the electronc marketplace: an exposton of the concept and mplcatons for research. Journal of the Academy of Marketng Scence, 33 (4), 585-603. Yousafza, S., Pallster, J. and Foxall, G. (2009). Mult-dmensonal role of trust n Internet bankng adopton. The Servce Industres Journal, 29 (5), 591-605. Zethaml, V.A., Btner, M.J. and Gremler, D.D. (2009). Servces Marketng, 5th ed. New York, NY: McGraw- Hll/Irwn. Zethaml, V.A., Parasuraman, A. and Malhotra, A. (2002). Servce qualty delvery through webstes: A crtcal revew of extant knowledge. Journal of the Academy of Marketng Scence, 30 (4), 362-375. 12
Internet Use No Internet Shoppng Use No Internet Bankng Use Internet Shoppng Use Internet Bankng Use Predctors - Perceved Innovaton Attrbutes - Demographc Characterstcs Fgure 1: Conceptual framework 13
I use the Internet for shoppng I use the Internet for Bankng Yes = 42.13 % Fnal probabltes 27.78 % Yes = 65.93 % No = 57.87 % 38.15 % No = 34.07 % Yes = 0.00 % 0.00 % No = 100.00 % 34.07 % 1 Fgure 2: Tree dagram 14
Table 1. Descrpton of non-prncpal component varables Varables Descrpton Dependents: - Internet shoppng use: - Internet bankng use: Bnary = 1 f someone uses; 0 non-use Bnary = 1 f someone uses; 0 non-use Independent Indvdual characterstcs: Gender (bnary) Age (years) Educaton 1 f male (47%); 0 f female (53%) 18-25 (26.3%); 26-35 (41.5%); 36-45 (20%); 46-55 (7.8%); 56-65 (4.4%) School (3.7%); College (20%); Unversty (43%); Postgraduate (33.3%) 15
Table 2: Results of prncpal components analyss Enjoyment Tme and energy Ease of use Usefulness Interactvty Personal capacty Perceved rsks Prvacy Securty concerns concerns Lack of tral Loadngs 7a, b, c 8g1, 2, 3, 4 8h,, j 8g5, e, f 9 (all 33) 10b, c, d, e 11a, b, c 11d1, 2, 3 11d4, 5, 6 11e, f, g, h,, j Egenvalue 2.263 2.594 1.483 1.753 13.343 2.474 2.265 2.421 2.348 4.022 Proporton (%) 0.754 0.649 0.494 0.584 0.404 0.619 0.755 0.807 0.783 0.670 1
Table 3: Estmates of probt regressons Unrelated bvarate probt Uncondtonal margnal effects Condtonal margnal effects Coef. Std. error p dy/dx Std. error p dy/dx Std. error p Internet shopper Male 0.627 0.193 0.001 Age: 18-25 0.110 0.228 0.630 Age: 26-35 Control varable Age: 36-45 0.1600 0.269 0.552 Age: 46-55 -0.057 0.401 0.887 Age: 56-65 0.106 0.452 0.815 Educaton School 0.4500 0.460 0.328 Educaton College 0.189 0.234 0.420 Educaton Degree Control varable Educaton Post-grad 0.969 0.262 0.000 Enjoyment 0.128 0.074 0.082 Tme and energy -0.035 0.079 0.659 Ease of use 0.124 0.096 0.199 Usefulness -0.134 0.095 0.160 Interactvty 0.035 0.032 0.273 Personal capacty 0.272 0.075 0.000 Perceve rsks -0.309 0.100 0.002 Prvacy 0.190 0.116 0.102 Securty -0.137 0.102 0.177 Lack of tral 0.040 0.065 0.538 constant -0.165 0.203 0.417 Internet banker Male 0.589 0.226 0.009 0.168 0.065 0.009 0.148 0.083 0.075 Age: 18-25 -0.156 0.291 0.591-0.043 0.077 0.577-0.077 0.109 0.480 Age: 26-35 Control varable Control varable Control varable Age: 36-45 0.776 0.292 0.008 0.256 0.106 0.016 0.285 0.099 0.004 Age: 46-55 0.144 0.475 0.761 0.043 0.148 0.772 0.065 0.187 0.729 Age: 56-65 1.002 0.494 0.042 0.360 0.192 0.060 0.382 0.170 0.025 Educaton School -0.653 0.678 0.335-0.338 0.098 0.160-0.319 0.263 0.224 Educaton College -0.135 0.295 0.647-0.037 0.077 0.634-0.079 0.108 0.464 Educaton Degree Control varable Control varable Control varable Educaton Post-grad -0.164 0.252 0.514-0.045 0.068 0.504-0.196 0.090 0.029 Enjoyment 0.291 0.093 0.002 0.082 0.026 0.002 0.098 0.036 0.006 Tme Energy 0.137 0.093 0.140 0.039 0.026 0.138 0.059 0.033 0.078 Ease Of Use 0.225 0.110 0.041 0.064 0.031 0.038 0.072 0.039 0.066 Usefulness -0.316 0.101 0.002-0.089 0.029 0.002-0.107 0.036 0.003 Interactvty 0.004 0.037 0.909-0.001 0.010 0.909-0.003 0.013 0.817 Personal Capacty 0.127 0.095 0.180 0.036 0.027 0.177 0.013 0.035 0.699 Perceved Rsks -0.109 0.103 0.292-0.031 0.029 0.291-0.001 0.040 0.975 Prvacy 0.051 0.110 0.640 0.014 0.031 0.640-0.005 0.042 0.898 Securty -0.025 0.099 0.798-0.007 0.028 0.798 0.009 0.037 0.816 Lack Of Tral -0.231 0.073 0.002-0.065 0.022 0.003-0.097 0.030 0.001 constant -1.154 0.234 0.000 Rho 1 5.21e-09 Log Lkelhood -191.55 Notes: N=259; bold font hghlghts statstcal sgnfcance at the 10% confdence level. Lkelhood-rato test of rho=0, ch 2 (1) = 43.2469, Prob > ch 2 = 0.0000. 1