Xem mẫu
Decision Factors for the Adoption of an Online Payment System by Customers
Table 4d. Hypotheses supported or rejected related to equation 1
Possible Determinant of Use Intention
Perceived Risk (PR) at low level 3HUFHLYHG%HQH¿WV3% Vendor’s Service Features (VSF)
Vendor’s Web site Features (VWF)
Client-side Technology (CST) Income Prospect (IP)
Gender Age Education
Internet Experience (IE)
Hypothesis
H1-1 H1-2 H2-1
H2-2
H3-1 H3-2 H3-3 H3-4 H3-5
H3-6
Test Result (correlation with UI)
6XSSRUWHG3RVLWLYHDQG6LJQL¿FDQW 6XSSRUWHG3RVLWLYHDQG6LJQL¿FDQW 6XSSRUWHG3RVLWLYHDQG6LJQL¿FDQW
6XSSRUWHG3RVLWLYHDQG6LJQL¿FDQW
8QVXSSRUWHGQHJDWLYHEXWLQVLJQL¿FDQW 8QVXSSRUWHGSRVLWLYHEXWLQVLJQL¿FDQW 6XSSRUWHG3RVLWLYHDQG6LJQL¿FDQW 6XSSRUWHG1HJDWLYHDQG6LJQL¿FDQW 8QVXSSRUWHGSRVLWLYHEXWLQVLJQL¿FDQW
8QVXSSRUWHGSRVLWLYHEXWLQVLJQL¿FDQW
2XU¿QGLQJVVXSSRUWVRPHRIWKHK\SRWKHVHV 6SHFL¿FDOO\RXUUHVSRQGHQWVKDYHVKRZQDVLJ-QL¿FDQWO\JUHDWHUWHQGHQF\WRDGRSWDQRQOLQHSD\-ment system if they: (a) perceive a low-level risk WRGRVRVXSSRUWLQJ+ESHUFHLYHEHQH¿WVRI WLPHHI¿FLHQF\RU¿QDQFLDOVDYLQJVERQXVWRGRVR VXSSRUWLQJ+F¿QGÀH[LEOHSURGXFWVHUYLFH features (supporting H2-1) or attractive Web site features (supporting H2-2) from online vendors. $OOWKHVHUHJUHVVLRQFRHI¿FLHQWVDUHVLJQL¿FDQW at the .01 level. In addition, after PR, PB, VSF, VWF, and IP are controlled for, the use intention of an online payment system is also positively associated with gender, while being negatively associated with age. Ilie et al. (2005) identify a VLJQL¿FDQWJHQGHUGLIIHUHQFHLQSHUFHLYHGLQQRYD-tion characteristics of communication technology adoptions, and suggest such a gender difference in perceptions can explain the gender difference in technology use intentions. Our evidence, on the other hand, supports the hypotheses that even
among those respondents who perceive the same ULVNEHQH¿WVDQGYHQGRU¶VWUDQVDFWLRQV\VWHP features, etc., a male is still more likely to adopt an online payment system than a female (sup-porting H3-3), while one’s intention to pay bills online decays with his/her age (supporting H3-4). We attribute this phenomenon to human nature (e.g., variety in risk tolerance) between different genders or ages. For example, even if younger and senior people understand equally well the VSHFL¿FULVNIRUWKHRQOLQHSD\PHQWV\VWHPLWVHOI younger persons are still more ready to accept the system than seniors, because the former is generally, by nature, more willing to risk trying new technology innovations and abandon the old methods for most of the technology innovations (Gilly et al., 1985).
2QWKHRWKHUKDQGRXU¿QGLQJVIDLOWRSURYLGH VXI¿FLHQWVWDWLVWLFDOVXSSRUWIRUWKHVLJQL¿FDQFH of the other determinants. Comprising multiple measurement items, neither the Client-side tech-
1204
Decision Factors for the Adoption of an Online Payment System by Customers
nology (CST) scale nor the Internet experience (IE) VFDOHVLJQL¿FDQWO\DIIHFWFXVWRPHUV¶LQWHQWLRQRI adopting online payment methods, therefore not supporting H3-1 or H3-6. Including only a single measurement item, neither the income prospect
Q1n =
1ȕ2 Q2nȕ3 Q3nȕ4 Q4n«ȕ22 Q22nİn ȕ1 + ȕm Qmnİn,
m 2
(IP) scale nor the education background variable where n = 1, 2, …, 148. (2) PDWHULDOO\LQÀXHQFHVFXVWRPHUV¶XVHLQWHQWLRQ
therefore not supporting H3-2 or H3-5. Inde-pendent of their individual differences in gender and age, customers are by far more concerned DERXWWKHSHUFHLYHGULVNDQGEHQH¿WVIRUXVLQJ an online payment system, as well as the service option features and Website design provided by vendors in the system.
However, we note that the regression model as Equation 1 is largely based on using scales as explanatory variables, and Table 4b shows that
the regression intercept, Ι0LVVLJQL¿FDQWO\GLI-IHUHQWIURP]HURFRHI¿FLHQW W
Table 5a indicates that the R-Square and adjusted R-Square of the model in Equation (2) are respectively .640 and .580, both showing an improvement over Equation (1) with .525 and .505. Cohen and Cohen’s (1983) test result (with F-value of 10.651, p < .01) also indicates a considerable increase in explanatory power when comparing the item-based Equation (2) with the scale-based Equation (1). The 29 variables (Q2-Q30) which serve as proxies for customers’ perceived risk, SHUFHLYHGEHQH¿WVRQOLQHSD\PHQWVHUYLFHIHD-tures, vendors’ Web site features and customers’
p 7KHH[LVWHQFHRIVXFKDVLJQL¿FDQW characteristics, jointly explain approximately 64%
non-zero intercept suggests that there are other IDFWRUVPLVVLQJIURPRXUPRGHOVSHFL¿FDWLRQVDQG the explanatory power of this regression model can be substantially improved by including ad-ditional explanatory variables (e.g., Brav, Lehavy,
of the variation in customers’ intention to adopt online payment methods.
Table 5b estimates the possible impact that each of the explanatory variables may have on customers’ payment-method preferences.
& Michaely, 2005). Additional analyses to provide extra explanatory powers are not uncommon; see Suh and Lee (2005) and Wasko and Faraj (2005) as examples.
Using All Measurement Items as Explanatory Variables
To further explore the possible underlying factors WKDWPD\LQÀXHQFHDFXVWRPHU¶VLQWHQWLRQWRDGRSW online payment methods, we extended our regres-sion analysis by using respondents’ use intention (Q1) as the dependent variable, the other 21 percep-tion items (Q2-Q22) as independent variables, and customer individual difference factors (Q23-Q30) as covariates. The extended model and regression results are presented as follows:
• $PRQJ WKH ³SHUFHLYHG ULVN´ LWHPV 4 4DQG4DUHVLJQL¿FDQWO\DQGSRVLWLYHO\ associated with the dependent variable Q1,
DVȕ3,ȕ5DQGȕ6DUHVLJQL¿FDQWO\SRVLWLYHDW the .05-.10 level. A customer would be more
willing to adopt online payments provided that he or she feels safe to provide personal in-formation online, considers legal regulations DUHVXI¿FLHQWWRGLVFLSOLQHWKRVHHQJDJHG in online payment fraud, and considers the vendor/creditor’s online transaction network is secure (p = .068).
• $PRQJWKH³SHUFHLYHGEHQH¿WV´LWHPV4 44DQG4DUHVLJQL¿FDQWO\DQGSRVL-
WLYHO\DVVRFLDWHGZLWK4DVȕ7,ȕ8ȕ11, and ȕ12DUHVLJQL¿FDQWO\SRVLWLYHDWWKH level. A customer will be more likely to adopt
1205
Decision Factors for the Adoption of an Online Payment System by Customers
Table 5a. The OLS model summary & ANOVA analysis related to equation 2
R Square
.640
Adjusted R2
.580
Std. Error
.827
Durbin-Watson
2.103
F Sig.
12.290** .000
7DEOHE7KHUHJUHVVLRQFRHI¿FLHQWVUHODWHGWRHTXDWLRQ
Test Statistic Collinearity Statistics
&RHI¿FLHQW Value
ȕ1 -1.207 ȕ2 .047 ȕ3 .092 ȕ4 -.098 ȕ5 .422 ȕ6 .129 ȕ7 .282 ȕ8 .220 ȕ9 .138 ȕ10 -.115 ȕ11 .227 ȕ12 .353 ȕ13 .199 ȕ14 .069 ȕ15 .233 ȕ16 .010 ȕ17 .017 ȕ18 -.189 ȕ19 .043 ȕ20 .023 ȕ21 .251 ȕ22 -.087
Std. Error .912 .093 .044 .104 .112 .069 .115 .107 .071 .129 .086 .124 .072 .124 .103 .119 .094 .096 .084 .090 .136
.088
t-value -1.323
.508 2.091* -.947 3.772** 1.870 2.452* 2.056* 1.944 -.890
2.651** 2.840** 2.758** .556 2.262* .085 .184
-1.956 .508 .259
1.838
-.984
p-value .188 .612 .047 .345 .000 .068 .022 .041 .053 .375 .009 .005 .007 .579 .041 .933 .855 .053 .613 .796 .089
.327
Tolerance VIF
.545 1.835 .511 1.956 .462 2.164 .354 2.821 .721 1.388 .514 1.944 .470 2.127 .537 1.862 .213 4.695 .235 4.247 .295 3.395 .565 1.769 .439 2.276 .338 2.959 .450 2.222 .558 1.791 .474 2.108 .741 1.350 .593 1.687 .504 1.982
.572 1.750
Notes: (a) The t-statistics are derived from testing the null hypothesis that each of the regression
FRHI¿FLHQWVȕ1±ȕ30HTXDOV]HUR³QRLQÀXHQFHRQUHVSRQGHQWV¶SUHIHUHQFHV´ELQGLFDWHV VLJQL¿FDQFHDWWKHDQGOHYHOUHVSHFWLYHO
1206
Decision Factors for the Adoption of an Online Payment System by Customers
Table 5c. The covariate effects related to equation 2
5HJUHVVLRQ&RHI¿FLHQWZLWK4
Covariate Q23 Q24 Q25 Q26 Q27 Q28 Q29
Q30
Parameter .506 -.322 .252 .149 .187 .336 -.095
.002
t-value 2.945** -1.999* 2.933** 1.720 1.462 3.201** -.812
.017
p-value .004 .049 .004 .088 .146 .002 .418
.987
1RWHLQGLFDWHVVLJQL¿FDQFHDWWKHDQGOHYHOUHVSHFWLYHO
online payment methods provided that he or she considers meeting payment deadlines and avoiding late penalties as particularly important, considers the online payment system is easy to use and fast, and consid-ers the access to computers and Internet is easy to obtain. However, our respondents do not consider saving postage costs will be particularly important for them to choose
³SD\RQOLQH´DVȕ10LVQRWRQO\LQVLJQL¿FDQW but also negative. The discount/bonus (Q9)
provided by creditors/vendors for placing and paying for orders online is marginally
LQÀXHQWLDODVȕ9 is marginally positive (p = .053).
• $PRQJ³YHQGRUVHUYLFHIHDWXUHV´4LV
VLJQL¿FDQWO\DQGSRVLWLYHO\DVVRFLDWHGZLWK
4DVȕ13LVVLJQL¿FDQWO\SRVLWLYHDWWKH level. A customer will be more likely to
adopt online payment methods provided that the vendor’s online payment system offers customers the option feature of recurring DXWRPDWLFGHGXFWLRQV7KLV¿QGLQJLVFRQ-sistent with the fact that customers highly regard the importance of meeting payment deadlines and avoiding late penalties, since
monthly automatic deduction with the minimum amount due is the most time- and cost-effective way to avoid late penalties. $PRQJ³YHQGRU:HEVLWHIHDWXUHV´4LV
DOVRSRVLWLYHO\DVVRFLDWHGZLWK4DVȕ15 is SRVLWLYHO\VLJQL¿FDQWDWWKHOHYHO
• $PRQJWKH³FOLHQWVLGHWHFKQRORJ\´LWHPV
(Q17-Q21), all but Q18 are positively associ-ated with Q1, but none of these regression FRHI¿FLHQWVDUHVLJQL¿FDQWDWWKHOHYHO From our observations, it appears that a customer’s preference to pay bills online does not strongly depend on the hardware or software that he or she is equipped with, including anti-virus/spyware programs, op-erating system, or even high-speed Internet VHUYLFHVXFKDV`6/WKHUHIRUHUHDI¿UPLQJ our earlier result that the scale of client-side technology (CST) does not materially affect use intention (UI). When deciding whether WR³SD\RQOLQH´DFXVWRPHULVFRQFHUQHG more about the vendor’s technology level than about his/her own technology level.
• :H¿QGQRVLJQL¿FDQWUHVXOWVEHWZHHQD customer’s intention to pay bills online and his/her family income growth prospect
1207
Decision Factors for the Adoption of an Online Payment System by Customers
4DVȕ22LVQRWVLJQL¿FDQWO\GLIIHUHQW from zero.
To further account for customers’ charac-teristics, we once again employed a covariate regression analysis using Q1 (use intention) as the dependent variable, Q20-Q22 (individual items for user perceptions) as between-subjects factors, Q23-Q30 (gender, age, education, and Internet experience items), respectively, as covariates in the model. The covariate effect estimates are VKRZQLQ7DEOHF7KHFXVWRPHU¶V³SD\RQOLQH´ use intention has positive associations with male gender and with education (p = .004 in both cases), and a negative association with age (p = .049). The between-subjects covariate effects of customer gender, age, education background, and Internet experience are in line with those reported in the scale-variable-based result presented in Table 4c.
7KHGDWDLQ7DEOHFUHDI¿UPVRXUSULRU¿QG-ings that males, younger customers and those with higher education levels are more willing to use an online payment system than their counterparts. In addition, it is interesting to observe that those most willing to pay bills online are those customers who frequently trade securities online (between 4DQG4WKHFRHI¿FLHQW W p = .002), rather than those who frequently shop, bid or sell goods online. One of the possible explanations is that online security trading typically involves larger amounts of electronic funding. Online brokerage accounts require a certain amount of cash deposit to open, and the trader must use bank deposits rather than credit cards to pay for the trades. Compared with online shoppers and bidders who typically use credit cards (with credit card companies allowing customers to dispute unauthorized payments) and pay relatively smaller amounts for their deals, online security traders have experienced considerably greater risk within their online payment/funding process; therefore, they will be more inclined to accept online pay-
ment methods and less likely to overestimate the risk related to making online payments.
When comparing results in Tables 5a, 5b, and FZLWKWKRVHLQ7DEOHVDEDQGFZH¿QG that instead of using scales, using measurement items as explanatory variables and/or as covari-ates can improve the statistical performance of regression analysis for online-payment-adoption determinants. (1) The R-square and adjusted R-Square improved, implying a greater explanatory SRZHUIRUWKHPRGHO7KHFRQVWDQWFRHI¿FLHQW EHFRPHVLQVLJQL¿FDQWFRQVLGHUDEO\UHGXFLQJWKH ³XQH[SODLQHG´SRUWLRQIRUWKHPRGHODQGZH ¿QGVRPHQHZVLJQL¿FDQWHYLGHQFHVXSSRUWLQJ the positive impact of a customer’s education and Internet experience on his or her pay-online use intention.
However, we were concerned that multicol-linearity might arise for a regression model like Equation 2 that incorporates all twenty-nine measurement items as explanatory variables. If multicollinearity does exist, it would cause a severe problem of biased and unreliable regres-VLRQHVWLPDWHVZKLFKE\IDURXWZHLJKVWKH³LP-provement in statistical performance.” Therefore, we performed a collinearity analysis, and the resulting statistics are documented in the last two FROXPQVLQ7DEOHE:H¿QGWKDWDOOLQGLYLGXDO VIF statistics are below 10, the average VIF value is below 6, and none of the tolerance values is below 0.1. Our regression estimates appear not to be materially affected by the multicollinear-ity problems. As a further proof, our regression estimates remain stable even after we drop from the model some of the explanatory variables that might seem highly correlated, such as Q4 vs. Q5, Q18 vs. Q21, etc. Thus, we feel reasonably FRQ¿GHQWZLWKWKHXQELDVHGQHVVRIFRHI¿FLHQW estimates obtained from the regression analysis WKDWXVHVWKH³XVHUSHUFHSWLRQ´PHDVXUHPHQW items (Q2-Q22) as explanatory variables and the HLJKW³XVHULQGLYLGXDOGLIIHUHQFH´LWHPV4 Q30) as covariates.
1208
...
- tailieumienphi.vn
nguon tai.lieu . vn