• 沒有找到結果。

Learning from the past: Measuring Internet banking service quality

N/A
N/A
Protected

Academic year: 2021

Share "Learning from the past: Measuring Internet banking service quality"

Copied!
6
0
0

加載中.... (立即查看全文)

全文

(1)

Learning from the past: Measuring Internet Banking

service quality

Yu-Lung Wu 1, Yu-Hui Tao 2, Pei-Chi Yang3 Ching-Pu Li 3

1 Department of Information Management, I-Shou University, Kaohsiung County, Taiwan 2 Department of Information Management, National University of Kaohsiung, Kaohsiung, Taiwan

3Department of Information Engineering, I-Shou University, Kaohsiung County, Taiwan

d9503006@stmail.isu.edu.tw

Abstract- Internet has played a pivotal role in transforming banking services into e-services. While several studies have examined the effective measurement of e-banking service quality, their lack of a holistic view has hindered accumulation of past knowledge. To address this issue, this study first reviews and summarizes the methodology, service quality dimensions, suggestions and limitations of seven e-banking service quality studies conducted in seven countries. An empirical study is then conducted to compensate for two shortcomings of a prior Taiwanese study. To improve understanding of e-banking service quality, a comprehensive scheme is proposed that has managerial implications.

Keywords: Internet Banking (e-banking), e-Service Quality (e-SQ), scale development, cultural factor

I. INTRODUCTION

Internet technology is widely applied in the service industry, and Internet banking (e-banking) is a prominent example. Theoretically, e-banking involves a customer connecting to bank computer systems via the Internet to access banking services without human contact. In this context, the majority of interactions between the bank and the customer are conducted digitally[1]

Internet technology has evolved considerably over the years, such that numerous new developed e-banking services differ considerably from older systems [2][3]. At least 65% of customers in Taiwan are satisfied with their e-banking services because key issues for users, such as security and privacy, have been addressed and considerably improved [4]. This high customer satisfaction contrasts with almost unanimously negative customer satisfaction from surveys conducted a few years ago[5][6]. Understanding Internet channel service quality thus is increasingly important[7]. Particularly, in the current competitive market, a good measurement scale that can provide a basis for future refinement and reassessment is crucial for sustainable development of banks.

After conducting initial research in 2007, this study reviewed the literature on Internet banking service quality (e-banking SQ) from Germany, Hong Kong, India, Taiwan, Turkey, the United Kingdom and the United States between 2001 and 2010[8] [1] [9] [10] [3] [11] [12]. On the one hand, this body of research contains too many versions of e-banking Web site service quality measurement to allow effective learning from the past, instead creating confusion. However,

these studies, conducted in seven countries provide an excellent basis for a holistic view of e-banking SQ, and a robust and general e-SQ measurement for practical applications. Consequently, further organization and suggestions regarding these e-banking studies is desirable for researchers and practitioners.

Based on the above analysis, this study thoroughly reviews the above studies to abstract a comprehensive framework of e-banking SQ measurement. While doing this, an additional empirical e-banking SQ measurement study is also conducted in Taiwan to address some of the issues raised in previous studies. The rest of this paper is organized as follows: SectionⅡ presents a literature review, and thus provides some background to the research design in Section Ⅲ, and the subsequent data analysis in Section Ⅳ. Conclusions are finally drawn in Section Ⅴ, along with a summary of its limitations and implications.

II. LITERATURE REVIEW

A. Service Quality Model

With regard to customer-based measurements of service quality, a review of the growing body of literature on service quality suggests that two schools of thought dominate extant thinking, namely the Nordic school of thought, based on the two-dimensional model of Grönroos [13] and the North American school of thought, based on the five-dimensional model of Parasuraman et al.[14]. For example, when developing a scale to measure traditional bank service quality, Bahia and Nantel [15] and Aldlaigan and Buttle[16] adopted the two-dimensional model while Karatepe et al. [17] adopted the five-dimensional model.

Early service quality models from both schools have tended to conceptualize factors related to service quality as components of service quality. The SERVQUAL instrument provides an example, and comprises a 22-item scale for measuring service quality along five dimensions proposed by Parasuraman et al. [18] [19]and later refined by Parasuraman et al. [20] [21] [22] However, the SERVQUAL instrument has also generated debate regarding the most appropriate ways to assess SQ [23] [24] [20] [21] [22] [25] Furthermore, a global measure of web site service quality is likely to suffer the same criticisms surrounding whether the SERVQUAL scale is industry or context dependent[26].

(2)

Hence, given the phenomenal growth of e-services, a stream of research exists that aims to understand the dimensions of e-SQ and their relationship with overall performance. For example, Loiacono et al. [27] proposed web quality (WebQual), a scale for rating websites on 12 dimensions. Yoo and Donthu [28] then immediately adapted WebQual into a four dimensional SITE-QUAL measurement scale. However, Parasuraman et al.[29] noted that neither WebQual nor SITE-QUAL capture all aspects of the purchasing process, and therefore cannot comprehensively assess site service quality.

In contrast with the general web quality structure of WebQual, Wolfinbarger and Gilly [30] [31]proposed .comQ and eTailQ as measures of web retail store quality. Based on the literature, Parasuraman et al. [29] combined various concepts of online service quality[27] [28] [30] [31] in the most comprehensive work on e-SQ to date. Parasuraman et al [29]. used an empirical test and a multiple item scale to assess online shopping provider service quality. The final analytical

results included 22 measurement items belonging to the four constructs for core service scale (E-S-QUAL), and 11 items belonging to the three constructs for recovery service scale (E-RecS-QUAL). Parasuraman et al.[29]applied both E-S-QUAL and E-RecS-QUAL to empirically test two online retailers, Amazon and Walmart.

Previous studies provide important theoretical framework and research instruments. Nevertheless, when considering the natural differences between e-banking and e-retailers, E-S-QUAL and E-RecS-E-S-QUAL may be inappropriate for direct application in measuring e-banking SQ. Therefore, careful investigation of e-banking service quality is desirable.

B. Internet Banking Service Quality

The literature on e-banking services has established a basic understanding of this new Internet application in terms of its service categories, satisfaction level, critical features, and service quality dimensions. Table I summarizes the above

seven e-banking SQ studies conducted from 2001 to 2010, and compares their methodologies, findings, suggestions . TABLE I Summary of the e-banking SQ literature

Reference/

country Methodology Findings Suggestions Limitations

Jun and Cai (2001) / USA

1. Dimensions identified using critical incident technique .

2. Content analysis.

Seventeen dimensions and 45 items.

1. Collecting data directly from e-banking customers.

2. Using factor analysis.

3. Developing overall e-banking SQ and investigating customer satisfaction.

1. Sampled customer comments from bulletin board system. 2. May miss key dimensions such as customer compliments and complaints.

Jayawardhena (2004) / UK

1. Initial items based on Parasuraman et al. (1988a, 1988b) and identified using focus group discussion. 2. Development method based on Parasuraman et al. (1988a, 1988b)

3. EFA, CFA and regression analyses.

Six dimensions and 26 items.

Continuing to use qualitative and quantitative techniques in different

contexts. N/A

Bauer et al. (2005) / Germany

1. Initial items based on Cronin and Taylor (1992) and Huizingh (2002), and identified using focus group discussion.

2. Development method based on Cronin and Taylor (1992)

3. EFA and CFA.

Six dimensions and 61 items.

1. Analyzing whether the identified measurement model can be generalized and applied to other portal types.

2. Investigating the relationship between the extracted quality dimensions and customer satisfaction or loyalty, respectively.

Only sampled from one bank.

Siu and Mou (2005) / Hong Kong

1. Initial items identified via focus group discussion (Internet users only).

2. Development method based on Zeithaml et al. (2000, 2002).

3. t-test, one-way ANOVA and multiple regression analyses.

Four dimensions and 19 items.

1. Using a larger sample.

2. Validating the model and extending the results to other industries and countries.

1. Only tested in Hong Kong. 2. Small respondent sample size (195).

Khan and Mahapatra (2009) / India

1. Initial items from focus group discussion (Internet users only).

2. Factor analysis, correlation analysis and regression analysis.

Five dimensions and 26 items.

1. Identifying the relative importance of each dimension.

2. Including the provider (banker) perspective.

1. Small respondent sample size (1,143).

2. Testing in urban India. Akinci et al.

(2010) / Turkey

1. Initial items based on Parasuraman et al. (2005) and identified via focus group discussion.

2. Development method based on Parasuraman et al. (2005).

3. CFA.

Seven dimensions and 28items.

Applying the E-S-QUAL and E-RecS-QUAL scales to different industries and culturally different countries. N/A Ho and Lin

(2010) /Taiwan

1. Initial items based on Cristobal et al. (2007). 2. Development method based on Parasuraman et al. (1988a, 1988b).

3. Factor analysis, principal components analysis.

Five dimensions and 17 items.

Seeking a larger sample size to increase credibility.

1. Only sampled from one bank. 2. Small respondent sample size (135).

III. RESEARCH DESIGN

Based on the methods used in seven e-Banking SQ studies, a comprehensive methodology for developing SQ measurement can be summarized into a six-step process consisting of generating the scale items, refining the scale items, conducting the field test and collecting the data, confirming the sample representation, performing exploratory factor analysis and the reliability test, and performing confirmatory factor analysis, a process which is also adopted in this study, as follows:

Step 1: Generate the scale items.

A comprehensive initial item set may be based on traditional SQ measurement scale studies such as Jayawardhena [1] and Bauer et al.[9], e-SQ measurement scale studies such as Akinci et al.[11]and Ho and Lin [32],

focus group discussion such as Khan and Mahapatra[3], or SQ related studies such as Parasuraman et al.[29]. In this study,

unlike the investigations listed in Table 1, the scale items for the service content were extracted from the relevant literature on domestic and foreign Internet banks([9] [33] [34] [35] [36] [37] [8] [38] [10] [39] [40]), SQ measurement methods [23] [24] [21] [22] and Web site SQ [27] [41] [28] [42] [43] [30] [31] These factors provided the basis for designing the appropriate assessment questions.

Step 2: Refine the scale items.

The initial scale items must be refined to better characterize the target domain based on the opinions of the domain experts and users. This study interviewed administrators of Internet banks to clarify their e-service situations and perspectives, as suggested by Jayawardhena [1], Khan and Mahapatra [3] and Siu and Mou [10]. These items are then adjusted based on interviewee perceptions of the importance of each SQ attribute of e-banking.

(3)

The composed questionnaire undergoes a field test involving a small sample. Based on the field test feedback, the questions are readjusted to finalize the formal questionnaire. This study randomly distributed the draft questionnaires to financial institutions, and the analyzed data set was free of invalid questionnaires from respondents with limited or no e-banking experience.

Step 4: Confirm the sample representation.

The sample data used in a survey study must be assessed to determine their representativeness of the population in the complete sample profile analysis. In this study, after presenting the sample profile, One-Way ANOVA was performed to assess significant differences relative to three larger national samples conducted in 2007.

Step 5: Perform exploratory factor analysis (EFA) and the reliability test.

With a representative sample data set, exploratory factor analysis is performed to identify the dimensions of the scale items, followed by a reliability test of the identified dimensions. This study uses the Principal Component Analysis and Varimax with Kaiser Normalization rotation in SPSS to perform orthogonal rotation. Furthermore, to increase convergent and discriminate validity, the following three criteria are applied in the EFA to obtain the final assessment dimension of the e-banking SQ[44]

1.Retaining items with eigenvalues exceeding 1.

2.Deleting items with factor loadings smaller than 0.5 or with large factor loadings on two factors.

3.Excluding factors that contain only one question.

According to Chang[45],the Cronbach’s α coefficient of each dimension, and each question, should be calculated. The questions with low relevant coefficients are then deleted to increase the relevant coefficient of the dimension.

Step 6: Perform confirmatory factor analysis (CFA). To validate the factor structure, CFA is performed following the EFA. Most EFA was conducted to assume a certain relationship, for example the two-level relationship between the SQ dimension and service quality[1] [9] [3] [32] or to simply apply regression analysis instead of CFA [1] [10] [3]. This study adopted a more rigorous approach for identifying the parsimonious model among the competing models of the n-factor model, as implemented by Noar [46]. The best-fit parsimonious model is also examined via analyses of reliability, convergent validity, and discriminant validity to assess its legitimacy.

IV. ESTABLISHMENT OF E-BANKING SQ ASSESSMENT DIMENSION

A. Design of Scale

According to the first three steps in the above procedure, relevant service attributes were first identified using the service items provided by the Internet banks, relevant scales addressed in the literature, and interviews with 23 administrators from 12 Internet banks in Taiwan. These attributes were then used as a basis for creating 76 appropriate survey questions following repeated reorganization and revision. Service quality was measured using the “perceptions-only” approach as adopted in Jayawardhena [1] The 76 service quality items were transformed into

Likert-scales, and the respondents were asked to indicate their perceptions of their Internet bank with regard to each item using a five-point scale ranging from “5 =strongly agree” to “1= strongly disagree.”

After completing the draft questionnaire, a small-sample pretest was conducted on e-banking users who were able to freely express their opinions. A descriptive analysis was conducted to delete unnecessary questions identified by the majority of the respondents, and to revise questions with unclear or inappropriate meanings. Following several modifications, the formal questionnaire was finalized and included 44 questions.

B. Sample Statistics

The formal questionnaire was randomly distributed in front of the offices of major financial institutions in Taiwan. Since the sample respondents come from different banks, the collected data was rich in content compared with that of Ho and Lin [32]. A total of 432 questionnaires were completed, of which 24 were either incomplete or completed with repetitive or predictable answers, while a further 96 respondents had no e-banking experience. Thus the final samples comprised a total of 312 (72.22%) valid questionnaires.

According to Step 4 of the scale development process, a One-Way ANOVA was used to determine the differences in gender, age, years of using Internet banking, and occupation between the mean of the present sample and the samples from the Pollster Online Survey (2007, 3,621 usable respondents), the 104 Survey (2007, 3,013 usable respondents), and the InsightXplorer, Ltd. survey (2007, 2,160 usable respondents). The P-values all exceeded 0.05, demonstrating no significant difference between the present sample and the other three large-survey samples. The lack of any significant difference indicates that the collected information has certain reference value.

C. Exploratory Factor Analysis

The total number of 312 valid respondents is too small to be halved as originally planned for conducting EFA and CRA with over 200 respondents in each case as suggested by Gefen et al.[47]. Therefore, this study used the entire sample for both the EFA and the CFA, an approach deemed acceptable in some of the literature, including Jamal et al. [48], Başol [49] and Byrne[50].

Before implementing the EFA in accordance with Step 5, the values of the Bartlett test of sphericity (9067.077) and the corresponding P-value (0.000) from the sample data were calculated. Since both values reached significance, the data set was suitable for performing EFA. Additionally, the Kaiser-Meyer-Olkin value was calculated as 0.858, very close to 1, indicating that the data sampling was appropriate.

For the EFA, this study deleted question items with factor loadings below 0.5 or that contained multiple factors with high factor loadings [51]. This study used 44 items to repeatedly perform EFA until all the items met the factor-loading criteria. Following four runs, 21 questions and five factor dimensions were obtained. When the largest common features occurred within the same dimension, these five-factor

(4)

dimensions were respectively named and defined as follows: efficiency, privacy/security, reliability, responsiveness, contact.

These five factors explained 49.033% of the variance with a high combined reliability of 0.987. For the individual factor reliability, Cronbach’s α ranged from 0.943 to 0.987 for these five factors, indicating good internal consistency among the items within each dimension. The findings indicated that the functions of a bank web site in relation to the five dimensions are strongly emphasized by customers and regarded as the essence of the supplied services. Summarizes the quality attributes of each factor. The extremely high value of each factor loading indicates that these five dimensions have good construct validity.

To confirm the consistency of the items within the same factor dimension, the Cronbach’s α values of the 21 items were calculated to assess whether any item might have too low a value causing inconsistency corresponding to its factor dimension. The analytical results showed that the factor dimensions have very high internal consistencies exceeding 0.6. Meanwhile, deleting certain questions could not increase the scale reliability, with total scale reliability exceeding the generally recommended level, 0.7 [45]. These results confirmed that the 21 questions and their corresponding five dimensions representing the e-banking SQ assessment scale were highly reliable.

D. Assumption Test for the Confirmatory Factor Analysis

Before applying the CFA, this study conducted a multivariate normality test on the skewness and kurtosis of each measurement item to confirm the assumed normal distribution of the data for using CFA. Extreme skewness

exists if the absolute skewness exceeds 3.0 and is problematic if the absolute value of kurtosis exceeds 10.0 [52]. Following the calculation, the skewness fell within the range -2.81 to 0.697 and the kurtosis fell within the range -0.74 to 7.45 in all cases, well within the limits of acceptability. Therefore, the 21 observed variables have near normal distribution, and it is reasonable to apply the maximum likelihood-based SEM method for parameter estimation.

E. Confirmatory Factor Analysis

This research adopted five nested models null model suggested by Noar [46] namely one factor model, uncorrelated factor model, correlated factor model and hierarchical model, and achieved best fit using four absolute indices of chi-square difference (χ2./d.f), goodness of fit index (GFI), adjusted GFI (AGFI), and root mean square error of approximation (RMSEA), and two relative indices of nonnormed fit index (NNFI) and comparative fit index (CFI). The recommended values are χ 2/d.f. less than 3.0 as suggested by Bagozzi and Yi [53], GIF exceeding 0.9 as suggested by Etezadi-Amolo and Farhoomand [54], RMSEA less than 0.08 as suggested by Hair et al.[44] and NFI, NNFI, and CFI all exceeding 0.9 as suggested by Bentler and Bonett [55], respectively. Table II shows that the null model, one-factor model and uncorrelated one-factor model do not meet the recommended values of all six fit indices. The correlated factor model is the only one satisfying all six recommended fit indices, while the hierarchical model only meets the recommended values of the χ2/.d.f and GFI.

TABLE II. Evaluation outcomes of competing models

Fit index Recommended value model Null One factor model Uncorrelated factor model Correlated factor model Hierarchical model

Absolute χ 2/d.f.() <3 20.03 15.58 8.25 2.41 2.73 GFI >0.9 0.68 0.73 0.92 0.90 RMSEA <0.08 0.17 0.15 0.06 0.07 AGFI >0.8 0.57 0.66 0.83 0.79 Realtive NNFI >0.9 0.56 0.82 0.95 0.88 CFI >0.9 0.63 0.84 0.97 0.89

The correlated model suggests that these five SQ dimensions can be appropriately discriminated from each other while at the same time also interrelated with one another. Meanwhile, a correlated model also suggests the possibility of a higher-order model[46]. Even though the preferred hierarchical model failed to meet three fit indices, it is only 0.01 or 0.02 short from meeting the recommended values of AGFI, NFI, and CF as seen in Table II.. In other words, the hierarchical model is still very promising and thus implies a potential second-order factor to account for the relationships among the five SQ dimensions in this study.

With the correlated model as the best fit model, the five SQ dimensions can only be examined individually, which may reflect a mixture of outcomes in e-banking SQ studies. For instance, all five dimensions are significantly related to the overall service quality in the work of Jayawardhena [1], whereas only three of the four dimensions are in the study of Siu and Mou[10]. The loss from not being able to sum these

five SQ dimensions into one scale may constitute further empirical work in developing a shorter version of the scale (such as one item per subscale), which may allow the researchers additional space to assess other important constructs in the survey, as well as reduce the response burden on participation[46] For the objective of this review and comparative study, these benefits of the hierarchical model are not as critical as in a pure scale development study.

As a result, the correlated factor model is identified as the parsimonious model depicted in Figure 1, with the path diagram and standardized parameter estimations.

To sustain the legitimacy of the best-fit parsimonious model, the correlated factor model is examined to assess its reliability, convergent validity, and discriminant validity. Regarding the reliability of the correlated-factor model, all the individual item reliabilities fall within the range 0.50 to 0.68, and thus meet the value of greater than 0.5 suggested by Hair [44]while the composite reliability (Efficiency: 0.89;

(5)

Privacy/security: 0.82; Reliability: 0.83; Responsiveness: 0.81; Contact: 0.85) all significantly exceed the value of 0.6 suggested by Bagozzi and Yi [53]. That is, the five factors have good reliability.

Figure 1- Correlated-factor model for Internet banking service quality Regarding the convergent validity, all the standardized parameter values (λ) fall within the range 0.53 to 0.83, higher than the minimum value of 0.45 suggested by Jöreskog and Sörbom [56] This phenomenon indicates that all the measurement variables adequately reflect the latent variables constructed by this study. Furthermore, the average variance extracted for the five latent variables exceeds the ideal value of 0.5[57], indicating the contribution of the constructed latent variables exceeds that of the biases].

To further test the discriminant validity of the correlated-factor model, this study adopted a complementary assessment proposed by Anderson & Gerbing[58] to “determine whether the confidence interval (+two standard errors) around the correlation estimate between the two factors includes 1.0.” The latent variables all exhibit good discriminant validity because the pairwise reliability intervals do not include 1.00.

Internal structural testing shows that all the measurement items have adequate reliability and every factor has good constructive reliability. The convergent validity test and the two discriminant validity tests demonstrate the need for the five factors in this correlated-factor model.

Ⅴ. CONCLUSIONS AND LIMITATIONS Through a review of seven e-banking SQ measurement studies conducted in seven countries during the past decade, this investigation holistically reviewed the methodology, suggestions and limitations associated with the development of e-banking SQ measurement. Furthermore, through an empirical investigation conducted in Taiwan to address some shortcomings of a previous study, this study also presented a holistic view of the dimensions and categories of SQ measurement. Based on this diverse literature, including diversity in terms of both time and country, two broad conclusions can be obtained as follows:

First, a comprehensive method for developing SQ measurement for e-banking has been consolidated into a six-step process, comprising scale item generation, scale item refinement, field testing and data collection, confirmation of

sample representativeness, exploratory factor analysis and reliability testing, and confirmatory factor analysis. Particularly, the combined methods of EFA and CFA and the competing-model approach for the parsimonious model[46] are critical to achieving a robust research outcome. Only three out of seven studies listed in Table 1 applied the combined methods of EFA and CFA, and none used the competing-model approach for the parsimonious competing-model. The managerial implication is that any researcher or practitioner wishing to develop a robust SQ measurement can easily do so by referring to section Ⅲof this study for a concise description of the necessary methodology and section Ⅳ of this study for a description of its application.

Second, limitations remain even after reviewing eight studies on e-banking SQ measurement scale development. As noted in section Ⅱ.(B), among the suggestions and limitations proposed by the past seven studies, only one suggestion and one limitation remain unaddressed. The first unaddressed limitation was the inclusion of customer compliments and complaints in the key dimension, which may be related to the satisfactory and dissatisfactory experiences of customers identified by Jun and Cai [8]. Second, the unaddressed suggestion, made by Akinci et al. [11], was to apply the E-S-QUAL and E-RecS-E-S-QUAL scales to a culturally different country and a different industry. Although this study addressed culturally different countries through reviewing the literature comprising studies conducted in seven countries, none of the other six studies has addressed the initial application of E-S-QUAL and E-RecS-UQLA. Finally, all the results were based on data collected from a user population with actual experience of e-banking, creating a potential sample bias through the exclusion of individuals who browsed the site but did not use the e-banking services. Failure to use e-banking services could result from various reasons, ranging from safety concerns to operational failure. A true population of Internet users, including those not currently using e-banking services, is desirable to confirm the generalizability of this study. Although the research on measuring e-banking SQ is extensive, further research addressing these three limitations is highly desirable.

REFERENCES

[1] C. Jayawardhena, "Measurement of Service Quality in Internet banking: The Development of an Instrument," Journal of Marketing Management, vol. 20, 2004.

[2] J. Z. Lu, " The current development situation of Internet banking,"

Financial and Monetary Bimonthly Magazine, vol. 41, 2005.

[3] M. S. Khan and S. S. Mahapatra, "Service quality evaluation in internet banking: an empirical study in India.," Indian Culture and Business

Management, vol. 2, pp. 30-46, 2009.

[4] Pollster Online Survey., 2007.

[5] (2004,2005 104 Survey. Available:

http://www.104survey.com/104Survey/portal/index.jsf.

[6] (2005,2007, Pollster Online Survey. Available:

http://www.pollster.com.tw/(2007/12/28)

[7] N. P. Mols, "The role of structural equation modeling in scale development," Structural Equation Modeling, vol. 10, pp. 622-647, 2000. [8] M. Jun and S. Cai, "The key determinants of Internet banking service

quality: A content analysis," International Journal of Bank Marketing, vol. 19, pp. 276-291, 2001.

[9] H. H. Bauer, et al., "Measuring the quality of e-banking portals,"

(6)

[10] N. Y. M. Siu and J. C. W. Mou, "Measuring service quality in internet banking: The case of Hong Kong," Journal of International Consumer

Marketing,, vol. 17, pp. 99-116, 2005.

[11] S. Akinci, et al., "Re-assessment of E-S-Qual and E-RecS-Qual in a pure service setting," Journal of Business Research, vol. 63, pp. 232-240, 2010. [12] C. I. Ho and Y. L. Lee, "The development of an e-travel service quality

scale," Tourism Management, vol. 28, pp. 1434-1449, 2007.

[13] C. Grönroos, "A service quality model and its marketing implications,"

European Journal of Marketing, vol. 18, pp. 36-44, 1984.

[14] A. Parasuraman, et al., "A Conceptual Model of Service Quality and Its Implications for Future Research," Journal of Marketing Management, vol. 49, pp. 41-50, 1985.

[15] K. Bahia and J. Nantel, "A reliable and valid measurement scale for the perceived service quality of banks. ," International Journal of Bank

Marketing, vol. 18, pp. 84-91, 2000.

[16] A. H. Aldlaigan and F. A. Buttle, "SYSTRA-SQ: a new measure of bank service quality. ," International Journal of Service Industry Management, vol. 13, pp. 362-381, 2002.

[17] O. M. Karatepe, et al., " Measuring service quality of banks: Scale development and validation," Journal of Retailing and Consumer Services, vol. 12, pp. 373-383, 2005.

[18] A. Parasuraman, et al., "SERVQUAL a multiple-item scale for measuring consumer perceptions of service quality," Journal of Retailing, vol. 64, pp. 35-48, 1988a.

[19] A. Parasuraman, et al., "Communication and control processes in the delivery of service quality," Journal of Marketing Management, vol. 52, pp. 35-48, 1988b.

[20] A. Parasuraman, et al., "Refinement and reassessment of the SERVQUAL scale," Journal of Retailing, vol. 67, pp. 420-450, 1991.

[21] A. Parasuraman, et al., "Reassessment of expectations as a comparison standard in measuring service quality: implications for further research,"

Journal of Marketing Management, vol. 58, pp. 111-124, 1994a.

[22] A. Parasuraman, et al., "Alternative scales for measuring service quality: A comparative assessment based on psychometric and diagnostic criteria,"

Journal of Retailing, vol. 70, pp. 201-230, 1994b.

[23] J. M. Carman, "Consumer perceptions of service quality: an assessment of the SERVQUAL dimensions," Journal of Retailing, vol. 66, pp. 33-55, 1990.

[24] J. J. J. Cronin and S. A. Taylor, "Measuring service quality: a reexamination and extension," Journal of Marketing Research, vol. 56, pp. 55-68, 1992. [25] R. K. Teas, "Expectations, performance evaluation, and consumers'

perceptions of quality," Journal of Marketing Management, vol. 57, pp. 18-34, 1993.

[26] Z. Yang, et al., " Development and validation of an instrument to measure user perceived service quality of information presenting Web portals,"

Information & Management, vol. 42, pp. 575-589, 2005.

[27] E. T. Loiacono, et al., WeBQualTM: A Web site quality instrument,

Worcester, Mass: Worcester Polytechnic institute. Working paper., 2000.

[28] B. Yoo and N. Donthu, "Developing a scale to measure the perceived quality of an Internet shopping site (SITEQUAL)," Quarterly Journal of

Electronic Commerce, vol. 2, pp. 31-46, 2001.

[29] A. Parasuraman, et al., "E-S-QUAL: a multiple-item scale for assessing electronic service quality," Journal of Service Research, vol. 7, pp. 213-233, 2005.

[30] M. Wolfinbarger and M. C. Gilly, comQ: Dimensionalizing, Measuring and

Predicting Quality of the E-tail Experience. Measuring and Predicting

Quality of the E-tail Experience, Marketing Science Institute, Cambridge, MA., 2002.

[31] M. Wolfinbarger and M. C. Gilly, "eTailQ: Dimensionalizing, Measuring and Predicting Etail Quality," Journal of Retailing, vol. 79, pp. 183-198, 2003.

[32] C. T. B. Ho and W. C. Lin, "Measuring the service quality of internet banking: scale development and validation," European Business Review, vol. 22, pp. 5-24, 2010.

[33] F. Calisir and C. A. Gumussoy, "Internet banking versus other banking channels: Young consumers’ view," International Journal of Information

Management, vol. 28, pp. 215- 221, 2008.

[34] W. X. Chou and X. Y. You, "The study about Internet banking service quality, relation quality, and customer loyalty.," Journal of Business

Administration, vol. 65, pp. 31-60, 2005.

[35] W. J. Deng, et al., "Confirmation of key service quality attributes of Internet banking - Three factor theory and the application of IPA," Journal of

Quality, vol. 14, pp. 351-365, 2007.

[36] K. Eriksson and D. Nilsson, "Determinants of the continued use of self-service technology: The case of Internet banking.," Technovation, vol. 27, pp. 159-167, 2007.

[37] C. Jayawardhena and P. Foley, "Changes in the banking sector-the case of Internet banking in the UK," Journal of Internet Research: Networking and

Policy, vol. 10, pp. 19-30, 2000.

[38] Y. B. Lin, et al., "The application of customer relationship management towards Internet banking services: The influence of website service quality

towards customer loyalty," Chiao Tung Management Review, vol. 27, pp. 57-85, 2007.

[39] B. Vatanasombut, et al., "Information systems continuance intention of web-based applications customers: The case of online banking,"

Information and Management, vol. 45, pp. 419-428, 2008.

[40] C. S. Yiu, et al., "Factors affecting the adoption of Internet Banking in Hong Kong-implications for the banking sector," International Journal of

Information Management, vol. 27, pp. 336-351, 2007.

[41] V. Riel, et al., "Exploring consumer evaluations of e-services: a portal site,"

International Journal of Service Industry Management, vol. 12, pp. 359-377,

2001.

[42] V. A. Zeithaml, et al., "Service quality delivery through Web sites: A critical review of extant knowledge.," Journal of the Academy of Marketing

Science, vol. 30, pp. 362-375, 2002.

[43] V. A. Zeithaml, et al., E-Service Quality: Definition, Dimensions and

Conceptual Model. Cambridge. MA. Marketing Science Institute, 2000.

[44] J. F. Hair Jr., et al., Multivariate Data Analysis 6th. . Prentice-Hall, 2006. [45] S. X. Chang, Research Method. Taiwan, 2001.

[46] S. M. Noar, "The role of structural equation modeling in scale development," Structural Equation Modeling, vol. 10, pp. 622-647, 2003. [47] D. Gefen, et al., "Structural Equation Modeling Techniques and Regression:

Guidelines," Communications of AIS, vol. 4, 2000.

[48] A. Jamal, et al., "Profiling consumers: A study of Qatari consumers’ shopping motivations," Journal of Retailing and Consumer Services, vol. 13, pp. 67-80, 2006.

[49] G. Başol, "Validity and Reliability of the Multidimensional Scale of Perceived Social Support-Revised, with a Turkish Sample," Social behavior

and personality, vol. 36, pp. 1303-1314, 2008.

[50] B. M. Byrne, Structural Equation Modelling with Amos: Basic Concepts

Applications and Programming. New Jersey, 2001.

[51] J. Hair, et al., Multivariate data analysis. Upper Saddle River., 1998. [52] R. B. Kline, Principles and Practice of Structural Equation Modeling. New

York, 1998.

[53] R. P. Bagozzi and Y. Yi, "On the evaluation of structural equation mode,"

Journal of Academy of Marketing Science, vol. 16, pp. 74-94, 1988.

[54] Etezadi-Amolo and A. F. Farhoomand, "A structural model of end user computing satisfaction and user performance," Information and

Management, vol. 30, pp. 65-73, 1996.

[55] P. M. Bentler and D. G. Bonett, "Significance tests and goodness of fit in the analysis of covariance structures," Psychological Bulletin, vol. 88, pp. 588-606, 1989.

[56] K. G. Jöreskog and D. Sörbom, LISREL 8: Structural Equation Modeling.

Scientific Software International Cop. Chicago, 1996.

[57] C. R. Fornell and D. F. Larcker, "Structural equation models with unobservable variables and measurement error," Journal of Marking

Research, vol. 18, pp. 39-50, 1981.

[58] J. C. Anderson and D. W. Gerbing, "Structural equation modeling in practice: A review and recommended two-step approach.," Psychological

數據

TABLE II.    Evaluation outcomes of competing models
Figure 1- Correlated-factor model for Internet banking service quality  Regarding the convergent validity, all the standardized  parameter values (λ) fall within the range 0.53 to 0.83, higher  than the minimum value of 0.45 suggested by Jöreskog and  Sörb

參考文獻

相關文件

The temperature angular power spectrum of the primary CMB from Planck, showing a precise measurement of seven acoustic peaks, that are well fit by a simple six-parameter

In case of non UPnP AV scenario, any application (acting as a Control Point) can invoke the QosManager service for setting up the Quality of Service for a particular traffic..

To explore different e-learning resources and strategies that can be used to successfully develop the language skills of students with special educational needs in the

Explore different e-learning resources and strategies that can be used to successfully develop the language skills of students with special educational needs in the..

* All rights reserved, Tei-Wei Kuo, National Taiwan University, 2005..

The min-max and the max-min k-split problem are defined similarly except that the objectives are to minimize the maximum subgraph, and to maximize the minimum subgraph respectively..

Different from services provided by retail banks that we normally enjoy, private banks provide a variety of services other than banking. These services include suggestions

With λ selected by the universal rule, our stochastic volatility model (1)–(3) can be seen as a functional data generating process in the sense that it leads to an estimated