Managing Service Quality: An International Journal Service quality in automated teller machines: an empirical investigation Bedman Narteh Article information: Downloaded by QATAR UNIVERSITY At 22:38 04 May 2015 (PT) To cite this document: Bedman Narteh, (2013),"Service quality in automated teller machines: an empirical investigation", Managing Service Quality: An International Journal, Vol. 23 Iss 1 pp. 62 - 89 Permanent link to this document: http://dx.doi.org/10.1108/09604521311287669 Downloaded on: 04 May 2015, At: 22:38 (PT) References: this document contains references to 75 other documents. To copy this document: permissions@emeraldinsight.com The fulltext of this document has been downloaded 1721 times since 2013* Users who downloaded this article also downloaded: Koushiki Choudhury, (2013),"Service quality and customers' purchase intentions: an empirical study of the Indian banking sector", International Journal of Bank Marketing, Vol. 31 Iss 7 pp. 529-543 http:// dx.doi.org/10.1108/IJBM-02-2013-0009 Shirshendu Ganguli, Sanjit Kumar Roy, (2010),"Service quality dimensions of hybrid services", Managing Service Quality: An International Journal, Vol. 20 Iss 5 pp. 404-424 http:// dx.doi.org/10.1108/09604521011073713 Houn-Gee Chen, Julie Yu-Chih Liu, Tsong Shin Sheu, Ming-Hsien Yang, (2012),"The impact of financial services quality and fairness on customer satisfaction", Managing Service Quality: An International Journal, Vol. 22 Iss 4 pp. 399-421 http://dx.doi.org/10.1108/09604521211253496 Access to this document was granted through an Emerald subscription provided by 545896 For Authors If you would like to write for this, or any other Emerald publication, then please use our Emerald for Authors service information about how to choose which publication to write for and submission guidelines are available for all. Please visit www.emeraldinsight.com/authors for more information. About Emerald www.emeraldinsight.com Emerald is a global publisher linking research and practice to the benefit of society. The company manages a portfolio of more than 290 journals and over 2,350 books and book series volumes, as well as providing an extensive range of online products and additional customer resources and services. Emerald is both COUNTER 4 and TRANSFER compliant. The organization is a partner of the Committee on Publication Ethics (COPE) and also works with Portico and the LOCKSS initiative for digital archive preservation. *Related content and download information correct at time of download. The current issue and full text archive of this journal is available at www.emeraldinsight.com/0960-4529.htm MSQ 23,1 Service quality in automated teller machines: an empirical investigation 62 Bedman Narteh Department of Marketing, University of Ghana Business School, Accra, Ghana Abstract Downloaded by QATAR UNIVERSITY At 22:38 04 May 2015 (PT) Purpose - The aim of this paper is to identify the dimensions of Automated Teller Machine (ATM) service quality and to evaluate customers' perceptions of the relative importance of these dimensions. Design/methods/approach - A structured questionnaire gleaned from the literature and focused group studies was used to collect data from 530 ATM customers of 15 banks in Ghana. Descriptive statistics, exploratory and confirmatory factor analysis, as well as multiple regression, were used to identify the relative importance of the dimensions of ATM service quality. Findings - The paper identified five dimensions of the \"ATMqual\" model. In order of importance, these dimensions are reliability, convenience, responsiveness, ease of use and fulfillment. Practical implications - The variables of the ATMqual scale provide practical levers for bank managers to improve customer experience with ATMs. The relative importance of the factors identified in the study also provide managers with a guide as to which issues to focus on in order to improve the efficiency and effectiveness of the ATMs. Originality/value - The paper provides a theoretical basis for conceptualising ATM service quality. The resulting dimensions, referred to as the ATMqual, thus address the paucity of a robust research in conceptualising and testing the dimensions of ATM service quality. Apart from the improved theoretical insight, the dimensions identified also provide bank managers with better understanding of and means to better manage customers' ATM experiences. Keywords Cashpoints, Customer services quality, Banking, Service quality, Electronic banking, Automated teller machines, Ghana Paper type Research paper Managing Service Quality Vol. 23 No. 1, 2013 pp. 62-89 r Emerald Group Publishing Limited 0960-4529 DOI 10.1108/09604521311287669 Introduction As a result of advancement in communication and information technology, service providers have found alternative means of delivering services to their customers. As such, the use of electronic channels to communicate, sell and deliver goods and services to customers has been widely adopted in business and even public services. In the area of retail banking, service providers have integrated electronic service delivery channels to bring banking services to customers and this is said to be influencing how banks interact with their customers in the market space (Parasuraman and Zinkhan, 2002; Bauer et al., 2005). Scholars have categorised electronic banking (e-banking) mostly into internet banking, telephone banking and automated teller machines (hence ATMs) ( Joseph and Stone, 2003; Ibrahim et al., 2006). The introduction of these electronic delivery channels in the banking industry has received wide academic and practitioner interest ( Joseph et al., 1999; Zeithaml et al., 2002; Lee and Lin, 2005; Ibrahim et al., 2006; Sayar and Wolfe, 2007; Ganguli and Roy, 2010). Prior research identified low-cost strategy as the key success factor in these electronic channels (Reibstein, 2002). Recently, quality dimensions have also emerged as a strategic issue which contributes to the success of e-banking channels (Loonam and O'Loughlin, 2008; Khan, 2010; Chong et al., 2010; Carlson and O'Cass, 2011). In addition, other researchers also argue that the Downloaded by QATAR UNIVERSITY At 22:38 04 May 2015 (PT) identification and management of quality dimensions in electronic channels could lead to competitive advantage and long-term business success (Parasuraman et al., 2005; Loonam and O'Loughlin, 2008). The extant literature postulates that research into how customers measure the quality of these electronic portals is still in the early stages (Zeithaml et al., 2002; Bressolles and Nantel, 2004; Parasuraman et al., 2005; Boshoff, 2007; Loonam and O'Loughlin, 2008; Chong et al., 2010). Joseph et al. (1999) argued that the different technologies might influence customer's perception of service quality in different ways and called for a study on the determinants of service quality based on the different technologies. Similarly, Curran and Meuter (2005) noted that consumer perceptions of service quality vary depending on the type of self-service technology used. In addition, Eriksson and Nilsson (2007) noted clear differences in e-banking technologies and urged banks to consider these real differences in designing customer contact strategies. In response, scholars have mostly focused on either determining the dimensions of service quality in internet banking (Loonam and O'Loughlin, 2008; Herington and Weaven 2009; Ho and Lin, 2010; Chong et al., 2010), and hybrid services (Ganguli and Roy, 2010) or compared the quality perceptions of providers and customers (Stiakakis and Georgiadis, 2009). A search through the extant literature also indicates that researchers have delineated the domain of ATM service quality and the measurement of its dimensions (Khan, 2010; Katono, 2011; Kumbhar, 2010). However, the studies have lacked sound theoretical bases for the conceptualisation of ATM service quality. Also, the convenience sampling methodology adopted in prior studies and the limited sample size utilised could affect the generalisability of the results which requires the need for a more comprehensive sampling methodology and an expanded sample size to address the dimensions of ATM service quality. Moreover, a recent study on ATMs and service delivery in Ghana (Sanda and Arhin, 2011) found a high degree of customer complaints with ATM downtime, cash out, high charges and sometimes, poor service recovery efforts when customers have problems. It appears the bank managers have very little understanding of customer's expectations with the ATMs. The paper addresses these research gaps and makes two contributions to the e-banking service quality literature. The study proposes \"ATMqual\" as a scale for measuring the dimensions of ATM service quality. Second, the study evaluates the relative importance of these dimensions based on the perceptions of customers and provides insight into ATM service quality in a region often under-researched in academic discourse and inquiry. The study among others will enable bank managers to focus on and address issues critical to the increasing number of customers who perform their banking transactions through the ATM. The rest of the paper is structured as follows: the paper continues with a clarification of the concept of self-service technologies and e-banking. The second section discusses service quality in traditional channels. This is followed by a review of the literature on electronic service quality in order to provide much insight from which the current study could draw upon. Subsequently, the theoretical framework for the study and the proposed dimensions of the ATM service quality are discussed. In the fourth section, the methodology of the study is presented. The fifth section deals with the presentation and discussion of the results. The final section presents the conclusions and implications of the findings. Service quality in ATMs 63 MSQ 23,1 Downloaded by QATAR UNIVERSITY At 22:38 04 May 2015 (PT) 64 Literature review Since ATMs are part of the self-service technologies, this section begins with a discussion of self-service technologies. Self-service technology and e-banking Developments in information and communication technologies have given service providers alternative options outside the brick-and-mortar buildings, to deliver services to customers. Within the banking sector, what has emerged as the self-service technologies (SSTs) have increasingly been deployed to provide banking service to customers outside the traditional banking halls. Curran and Meuter (2005) argue that SSTs in banking enable customers to conduct their banking services without any direct interaction and assistance from the banking staff. Eriksson and Nilsson (2007) and Curran and Meuter (2005) noted that SSTs have gained popularity among banks and their customers due to their ability to standardise service delivery, provide expanded service delivery options and also reduce labour costs. SSTs in banking have been referred to as online banking or e-banking (Ibrahim et al., 2006; Casalo et al., 2007; Loonam and O'Loughlin, 2008; Chong et al., 2010). These words will, therefore, be used interchangeably. Various definitions abound in the literature for e-banking. Ibrahim et al. (2006) refer to it as the provision of information about a bank and its products via a page on the internet. Similarly, e-banking has been defined by Singh and Malhotra (2004) as the deployment of banking services and products over electronic and communication networks directly to customers. Service quality in traditional channels Service quality has received wide academic research over the past two decades (Parasuraman et al., 1988) because prior studies have shown that high levels of perceived customer service quality can exert positive influence on customer satisfaction (Parasuraman et al., 1988, 1991; Cronin and Taylor, 1992) and firm profitability (Lasser et al., 2000). Other researchers also postulate a direct relationship between service quality and relationship quality (Meng and Elliott, 2008; Ndubisi et al., 2011). Relationship quality has been explained as customers' perceptions of how well a relationship fulfils their expectations, predictions, goals and desires ( Jarvelin and Lehtinen, 1996). Ndubisi et al. (2011) postulate that firms that deliver consistent quality services are able to satisfy and retain their customers over a longer term. Consequently, the marketing literature has been bombarded with a stream of research purporting to unravel the antecedents, dimensions and consequences of relationship quality (Meng and Elliott, 2008; Ganguli and Roy, 2010; Ndubisi et al., 2011). Verhoef et al. (2007) found weak customer loyalty and by implication poor relationship quality in electronic channels due to the lack of personal contact between customers and service providers. The extant literature suggests that service quality is determined by the difference between customer expectations of service provider's performance and their evaluation of the service they receive (Parasuraman et al., 1985, 1988). Various models were developed in the literature to measure service quality. The SERVQUAL model, postulated by Parasuraman et al. (1988), and the alternative SERVPERF model by Cronin and Taylor (1992) have received wide research attention and application in the service quality literature. The SERVQUAL model, in spite of its criticisms (Cronin and Taylor, 1992), appears to have enjoyed canonical status with service quality research. Downloaded by QATAR UNIVERSITY At 22:38 04 May 2015 (PT) The model is predicated on the assumption that service quality is dependent on five major factors of reliability, tangibles, empathy, assurance and responsiveness (Parasuraman et al., 1988). Parasuraman et al. (2005) argued that the service quality research in the extant literature was dominated by people-delivered services. In situations where machines are used to replace employees in the service delivery process, new dimensions of service quality might be perceived as important by customers. A stream of research has recently emerged to address some of the quality issues with electronic services (Zeithaml et al., 2002; Parasuraman et al., 2005; Kim et al., 2006; Boshoff, 2007; Cristobal et al., 2007; Nusair and Kandampully, 2008; Shachaf et al., 2008; Herington and Weaven, 2009; Ho and Lin, 2010; Chong et al., 2010; Ganguli and Roy, 2010). The next section provides a review of the burgeoning literature in this field. Electronic service quality Researchers have postulated that electronic service quality is a key determinant in differentiating service offers and building competitive advantage because the cost of comparing alternatives are relatively low in online environments (Santos, 2003; Bauer et al., 2005). For a start, Zeithaml et al. (2002) defined electronic service quality as \"the extent to which a web site facilitates efficient and effective shopping, purchasing and delivery of products and services\". Bressolles and Nantel (2004) argued that the definition covers both pre- and post-service delivery experiences on the web. Bauer et al. (2005, p. 156) postulate that \"the most important step in providing a sophisticated level of service through e-banking portals is to identify and measure the dimensions of portal quality\". Research into the quality dimensions of e-banking portals did not start until early 2000 and is still ongoing ( Jayawardhena, 2004). In spite of the late start, there seems to be consensus that traditional service quality dimensions cannot be wholly imported to measure electronic service quality. Research into electronic service quality has progressed through two major routes. Some researchers have used the traditional service quality dimensions to advance their studies into electronic service quality (Gefen, 2002; Lee and Lin, 2005). Other researchers argue that electronic service environment is completely different, dominated by people-machine interface, and requires a new set of quality dimensions (Parasuraman et al., 2005; Nusair and Kandampully, 2008; Shachaf et al., 2008; Khan, 2010). Using the first approach, Gefen (2002) modified Parasuraman et al.'s SERVQUAL scale to measure online service quality and found a three-factor assessment scale with tangibles appearing paramount in the measurement. Lee and Lin (2005) also modified the SERVQUAL scale to identify the main factors that influence customer's perception of electronic service quality of online sites. The study found design, reliability, responsiveness and trust as dimensions of online store quality. Among the researchers who developed new models were Loiacono et al. (2002) who developed the Webqual for rating the quality of web sites. The 12 dimensions identified in their scale included informational-fit-to-task, interaction, trust, response time, design, intuitiveness, visual appeal, integrated communication, innovativeness, flow, emotional appeal, business processes and viable substitute. Criticising the scale, Parasuraman et al. (2005) argued that the purpose of the scale was not to measure service quality but to generate information for web site developers. Received wisdom from electronic quality research indicates that the perceptions of customer's quality Service quality in ATMs 65 MSQ 23,1 Downloaded by QATAR UNIVERSITY At 22:38 04 May 2015 (PT) 66 and provider's or designer's quality are both necessary and essential for the understanding of online site quality (Hillaris et al., 2007). In addition, Parasuraman et al. (2005) developed what they called a comprehensive model known as E-S-QUAL to measure service quality of web sites. The E-S-QUAL consisted of two parts, the first part which measured service quality and the other part which measured service recovery. The four dimensions that measured service quality are efficiency, fulfillment, system availability and privacy. The model also recognises the importance of service recovery in online transactions and developed the E-RecS-QUAL consisting of responsiveness, compensation and contact. The study concluded that efficiency and fulfillment dimensions were the most critical and equally important facets of web site quality (p. 230). In an attempt to measure the construct validity of the four-dimensional E-S-QUAL scale, Boshoff (2007) tested the framework with new data and concluded that a six-factor dimension fitted the data better. The six factors proposed were value, efficiency, delivery, privacy, speed and reliability. Only efficiency and privacy maintained their original positions. Boshoff (2007) further urged researchers to explore the concept of reliability as a construct in measuring electronic service quality. The model was exclusively developed for online shoppers and thus, could not apply to e-banking, thereby creating the need for further studies to design context-specific models that address the needs of the various electronic channels. Previous e-service quality research has attracted criticism for adopting traditional SERVQUAL dimensions in online environments, developed scales without sufficient empirical validation, focused on the evaluation of web site quality rather than entire service quality dimensions and excluded the assessment of the consumer-buying process (Kim et al., 2006; Parasuraman et al., 2005). The differences in frameworks and findings on online service quality research above points to one apparent conclusion: that there is no immediate closure to the issues of electronic service quality research. It is also significant to note that most of the e-service scales were developed for online sites. For e-banking, due to the different channels available, the study supports the view of Joseph et al. (1999) that studies on e-banking quality should focus on the different technologies available. Consistent with this view, the current study focuses on the service quality of ATMs. Prior conceptualisation of ATM service quality A number of marketing scholars identify ATM, internet and telephone banking as the principal e-banking channels ( Joseph et al., 1999; Joseph and Stone, 2003; Loonam and O'Loughlin, 2008 ). In the banking sector, customers tend to use these three different automated service delivery channels in a complementary way (Dabholkar, 1996). Santos (2003) defined e-banking service quality as the consumers' overall evaluation and judgement of excellence and quality of electronic service offering in the virtual market place. This definition encompasses all the broad groups of e-banking service delivery channels such as internet banking, telephone banking and ATM. The current study, however, focuses on ATM because of its distinctiveness, continued popularity and usage in retail banking. The current paper defines ATM as an electronic device which allows customers to deposit, withdraw, transfer money, pay bills and perform other financial transactions without the aid of a branch representative or a teller. From the foregoing, it is argued that ATM is the electronic versions of the brick-and-mortar banking halls and customers visit the ATM to perform financial transactions, be it withdrawals, deposits Downloaded by QATAR UNIVERSITY At 22:38 04 May 2015 (PT) or check balances as they would have done in the normal banking halls. The ability of the ATM to perform these functions to the satisfaction of customers will measure customer perceptions of ATM service quality. The study, therefore, defines ATM service quality as the customers' overall evaluation and judgement of the excellence of services provided through ATM channels. Research has shown that service quality, whether through traditional or electronic channels, have different dimensions (Parasuraman et al., 1988, 2005; Shachaf et al., 2008; Nusair and Kandampully, 2008; Stiakakis and Georgiadis, 2009). Some previous studies on e-banking service quality have followed the view that customers use e-banking channels in a complementary way and therefore, their service quality dimensions were jointly determined in a single study ( Joseph and Stone, 2003; Al-Hawari and Ward, 2006). Even though such studies identified the dimensions of e-banking channels such as internet banking, ATM and PC banking in a single study, not much attention has been paid to the quality dimensions of the individual technologies. Recently, researchers have started to address ATM service quality dimensions. Table I provides a summary of the identified studies on the dimensions of ATM service quality. From Table I, it is clear from the various studies that ATM quality is based on multiple dimensions. Moreover, some dimensions (ease of use, convenient location and reliability) appear to be common among the researchers. Researchers, however, disagree on the conceptualisation of the dimensions. For instance, Kumbhar (2010) combined security and responsiveness into one dimension even though these have been proven in the e-quality literature to belong to different conceptual domains (Parasuraman et al., 2005). The lack of consensus on the relevant dimensions requires the need for further systematic investigation into the quality dimensions of ATMs. Moreover, no clear-cut theoretical basis could be found in previous conceptualisations. Recent studies have shown that the service quality construct is hierarchical and multi-dimensional (Collier and Bienstock, 2006; Carlson and O'Cass, 2011), yet most of the variables (ease of use, reliability, responsiveness) were handpicked from the e-quality literature and applied to the domain of ATM service quality. To the best of our knowledge, a theoretical framework of ATM service quality is yet to emerge in the literature. Finally, another major problem with prior conceptualisation of ATM service quality is the type of samples used. Convenience samples through self-selection, (Mobarek, 2007; Khan, 2010; Kumbhar, 2010) were the major sampling methodology adopted in previous studies. Dilijonas et al. (2009) even failed to disclose the sample size of the cases used for their study. As noted by Fassnacht and Koese (2006), self-selection occurs when respondents are selected based on their response to a screening question rather than through random sampling. Khan (2010), for instance, acknowledged the weakness of the convenience sample used for the study and called for further studies on the dimensions of ATM service quality using random sampling. The current study is a response to address some of these weaknesses in previous ATM service quality research. Theoretical framework for the study Even though we applaud the efforts of prior researchers in conceptualising ATM service quality, the current paper argues that the process could benefit from further development. The point of reference in this paper is the theoretical framework proposed by Collier and Bienstock (2006) in conceptualising online service quality. Service quality in ATMs 67 MSQ 23,1 Researcher/ author and year Contextual area Joseph and Stone (2003) USA Al-Hawari et al. (2005) Australia Mobarek (2007) Botswana Dilijonas et al. (2009) Baltic States Khan (2010) Pakistan Kumbhar (2010) India Katono (2011) Uganda Downloaded by QATAR UNIVERSITY At 22:38 04 May 2015 (PT) 68 Table I. Studies on ATM service quality Findings Found that among the customers convenient location; secure services; special services for the disabled; user friendliness; and personalisation (acknowledge me by name) were the dimensions of ATM service quality. The study lacked a theoretical foundation Found in the study that secured places; user friendliness; convenient location and proper functioning as the key dimensions of ATM service quality. The study used a convenient sample and also lacked a theoretical foundation Found speed of operations and waiting time as the important predictors of ATM service quality. This study appears not comprehensive enough, as it views the ATM as a mechanism for ensuring only fast service and as such it ignores all the other important dimensions such as ease of use, privacy and protection of customer's transactions. For instance, one may wonder the true value of an ATM which is fast but not reliable and cannot protect or compensate customers against fraud in Botswana Categorised the factors that determine ATM service quality into essential resources (adequate number of ATMs, convenient and secure locations and user friendly system); important dimensions of operation of ATM (maximum speed, minimum errors, high uptime and cash back); and value-based aspects (reasonable cost, offering to cover maximum needs of customers) as the major dimensions of ATM service quality. The major limitation of this study, however, is its lack of methodological rigour through which the variables for the study were developed The study found reliability; ease of use; privacy; convenience; efficiency and responsiveness as the dimensions of ATM service quality. The study, due to difficulties in obtaining customer information relied on a convenience sample which limits the generalisability of the findings. The researcher called for further studies on the dimensions of ATM service quality using a random sample The study found 11 dimensions for assessing ATM service quality system availability; e-fulfillment; efficiency; security; responsiveness; easiness; convenience; cost effectiveness; problem handling; compensation and contact. A cursory look at the dimensions indicates overlapping among the factors. For instance, problem handling, compensation and contact could all be addressed within the assurance construct. Similarly, efficiency and cost effectiveness seems to be related to each other The study established that tangibles, card issues, reliability and location are the most important student service quality evaluation dimensions of ATMs. The study used a convenience sample and lacked any sound theoretical foundation Downloaded by QATAR UNIVERSITY At 22:38 04 May 2015 (PT) They argued that the touch points in online service delivery are the process, the outcome and the recovery phases, respectively. The three-phases approach suggested above is an improvement over the two-stage approach of process and outcome phases proposed by Mentzer et al. (2001) to investigate logistics service quality because it also considers the difficulties that might have occurred during the service delivery phase and how they must be addressed. The addition of a recovery phase in the current study thus, provides managerial pointers to remedy ATM service failure. It also extends the process and outcome dimensions of quality in traditional service delivery proposed by Gronroos (1984). This conceptualisation also concurs with Bauer et al. (2005), who view e-banking quality dimensions to include issues that occur during the transactional phase as well as the post-transaction phase. It is also similar to the quality and the recovery dimensions proposed in the study of Parasuraman et al. (2005) in online shopping sites. Finally, this conceptualisation also resonates with recent studies that have demonstrated that the service quality construct is hierarchical and multi-dimensional (Collier and Bienstock, 2006; Carlson and O'Cass, 2011). The process dimension deals with the quality of interaction between the customer and the ATM as well as how customers transact business through the ATM. Fassnacht and Koese (2006) argued that in self-service technologies, customers merely interact with a technical interface, leading to a pure person-to-technology service encounter. The people-machine encounters raise issues for customer quality concerns. The current study proposes that process issues in ATMs relate to ease of use of the ATMs, convenient locations, privacy and security as well as reliability or functionality of the machines. The process dimensions of quality are explained in the next section. Reliability: reliability has often emerged as a major dimension in most service quality research. Within electronic quality research in particular, reliability has also emerged as one of the major determinants of service quality (Shachaf et al., 2008). Wolfinbarger and Gilly (2003) found that reliability is the strongest predictor of customer satisfaction and the second predictor of customer loyalty. The reliability dimension is critical because it embeds the dynamic capability to perform the promised service dependably and accurately. In the electronic environment, Zeithaml et al. (2002) used the concept to denote both technical reliability (the quality of the technology itself) and functional reliability (the ability of the technology to perform as expected). Thus, within the ATM environment, technical reliability may be equated to the design and installation of the ATM while functional reliability may also denote the performance of the ATM (e.g. ability of the ATM to provide error-free services). Shachaf et al. (2008) found reliability as one of the major determinants of electronic service quality in academic and public libraries. Similarly, in online transactions, Stiakakis and Georgiadis (2009) found reliability as the fundamental criterion of superior electronic service quality. Within the ATM environment, Khan (2010) and Katono (2011) found reliability an essential ATM quality dimension. Convenience: most early studies found locational convenience as an important influence on bank selection (Kaynak and Kucukemiroglu, 1992; Gray, 1977). As the ATM offers an alternative way of conducting banking transactions outside the banking halls, locational convenience of the ATM sites is also an important quality determinant. Locational convenience refers to the site at which the ATMs are located. Most ATMs are located in bank branches, or off sites such as in shopping malls and college campuses. Sometimes, the bank's ATM card is compatible with other banks ATM platforms and this makes it possible for customers to withdraw money from other ATMs at a small fee. If the ATMs are conveniently located, it minimises the Service quality in ATMs 69 MSQ 23,1 Downloaded by QATAR UNIVERSITY At 22:38 04 May 2015 (PT) 70 inconvenience of moving long distances in order to carry out transactions. So far, from Table I, convenient location appears to be the most widely used dimension of ATM service quality as it appeared in all the scales with the exception of that of Mobarek (2007). Ease of use: technology could sometimes be intimidating to some customers and therefore, one expects that an ATM should be able to simplify this to the barest minimum so as to motivate customers to use the facility. Ease of use is one concept that has been investigated in the technology adoption studies especially in workplace environments (Davis, 1989). In this regard, Davis (1989, p. 985) defines ease of use as \"the degree to which the prospective user expects the target system to be free of effort\". Chong et al. (2010) argues that if users feel that online banking is easy to use and hassle-free, then the chances of them using the system will be greater. This study used the concept \"ease of use\" to denote the degree to which the ATM ensures a hassle-free transaction for the customers. In his study, Cooper (1997) found ease of use as one of the three important characteristics customers consider in the adoption of innovative service. It has been strongly asserted that ease of use is one of the fundamental factors in determining the adoption and use of various corporate information technologies (Davis, 1989). Gounaris and Koritos (2008) also found that perceived ease of use predicted customer adoption of online banking. Even though researchers have used different names such as user-friendliness ( Joseph and Stone, 2003; Al-Hawari et al., 2005) user-friendly system (Mobarek, 2007) and easiness (Kumbhar, 2010) to describe this factor, it was found to be a major dimension of ATM service quality. Security and privacy: an ATM should also provide customers with security and privacy. Zeithaml et al. (2002) contend that privacy is the protection of personal information while security involves the protection of customers from fraud and financial loss. Clarke (1999) defined privacy as the individual's right to be alone and extends it to include communications and personal data. Casalo et al. (2007, p. 585) state that \"privacy refers to a set of legal requirements and good practices with regard to the handling of personal data whereas security refers to the technical guarantees that ensure that the legal requirement and practices with regard to privacy will be met effectively\". Kim et al. (2006) found privacy to be of critical value for customers in online transactions. Jahangir and Begum (2008) as well as Hernandez and Mazzon (2007) found in their respective studies in Bangladesh and Brazil that security is an important enabler of online banking. Similarly, Chong et al. (2010) found security and privacy important in the adoption of internet banking in Vietnam. Thus, this study argues that the way ATM custodians handle the privacy and security of customer information and data could be crucial in gaining customer trust in ATMs. It is important to note that security and privacy were not included in the dimensions found on the two African studies (Mobarek, 2007; Katono, 2011). However, they were important quality dimensions in the USA ( Joseph and Stone, 2003), Australia, (Al-Hawari et al., 2005), India (Kumbhar, 2010) and Pakistan (Khan, 2010). As a result, the current study could speculate that security and privacy issues become more pronounced in developed countries as well as emerging markets, due to the relatively developed financial systems than in developing countries. The outcome phase measures the results customers expect during the transactional phase. The paper identifies accuracy and fulfillment as the two outcome dimensions. Accuracy: as ATMs are mechanised, they are expected to eliminate many common errors that might occur with in-person interactions at the banking halls. ATMs should provide accurate transactions such as correct identification of name of customer, recording of routine transactions and printing of receipts. Collier and Bienstock (2006) found information accuracy and order accuracy to be critical dimensions of online retailer's service quality. Accuracy has not featured in any of the scales identified in the extant literature. However, it is represented in Table II as an ATM quality dimension and will therefore, be explored in this study. Fulfillment: Parasuraman et al. (2005) used fulfillment of online shopping sites to denote the extent to which the site's promises about order delivery and item availability are met. Similarly, Wolfinbarger and Gilly (2003) found that fulfillment of web sites Downloaded by QATAR UNIVERSITY At 22:38 04 May 2015 (PT) Critical dimension Reliability Responsiveness Ease of use Convenience Fulfillment Security Accuracy Related items Literature evidence ATM functions all the time ATM provides consistent services The ATM system compensates for wrongful deductions ATM problems are settled to my satisfaction ATM contact person is available for redress of problems Breakdown ATMs are fixed promptly ATM cards are promptly replaced ATM banking settles complaints in a reasonable time ATM provides clear instructions on usage ATMs are easy to use for transactions ATM language is easy to understand ATM provides graphics and adverts of bank services ATM provides education and direction of usage ATM provides different range of services ATMs are conveniently located ATM waiting time is acceptable ATM cards are compatible on other platforms ATM gives quality bank notes ATM provides enough money during transactions ATM satisfies most of my banking needs ATM charge are reasonable ATM provides fast services ATM gives instant money all the time I feel safe during ATM transactions I have trust and security in the ATM I fear for security of my personal information ATM provides accurate account records ATM provides precise receipt to confirm transactions Al-Hawari et al. (2005), Lee and Lin (2005), Khan (2010), Katono (2011) Joseph and Stone (2003), Lee and Lin (2005), Khan (2010), Kumbhar (2010) Service quality in ATMs 71 Joseph and Stone (2003), Mobarek (2007), Chong et al. (2010), Kumbhar (2010) Joseph and Stone (2003), Al-Hawari et al. (2005), Khan (2010), Katono (2010) Wolfinbarger and Gilly (2003), Parasuraman et al. (2005), Kumbhar (2010) Joseph and Stone (2003), Al-Hawari et al. (2005), Parasuraman et al. (2005), Khan (2010) Collier and Bienstock (2006) Table II. Critical ATMqual dimensional scale MSQ 23,1 Downloaded by QATAR UNIVERSITY At 22:38 04 May 2015 (PT) 72 had a significant effect on overall quality, satisfaction and loyalty intentions. Within the ATM environment, the study used fulfillment quality to measure the extent to which the ATM performs outcomes which meet the customer's expectation. Here, one could mention the genuineness of notes delivered by the ATM (eliminate counterfeits), the amount provided to customers per transaction and the ATM's transactional charges. Apart from the work of Kumbhar (2010), the other scales did not discuss the issues of fulfillment as an ATM quality dimension. Finally, the recovery phase also measures the quality of a bank's efforts to remedy possible ATM service failures. The recovery phase is important because most ATM transactions take place outside the banking halls without the active involvement of staff. As ATM technologies are sometimes prone to failures, such as inability to provide cash during transactions even though the account holder has been debited, as well as long periods of machine downtime, one expects an efficient recovery strategy to be implemented to reduce customer's frustrations. This dimension is represented by responsiveness. Responsiveness: it is not always possible to expect flawless ATM service delivery at all times to customers as technologies can, sometimes, fail to deliver as expected. Responsiveness measures the extent to which the banks put in place measures to recover services when ATM services are negatively confirmed. Effective service recovery has been shown to have a positive impact on customer satisfaction (Magnini et al., 2007). Most electronic service quality scales have responsiveness as a major dimension (Parasuraman et al., 2005; Kim et al., 2006; Nusair and Kandampully, 2008; Shachaf et al., 2008). With ATMs, responsiveness quality deals with the banks ability to handle customer complaints arising as a result of transactional failures. Acceptable recovery practices may vary from a simple apology to compensating customers against losses incurred such as monies illegally withdrawn from their accounts. Responsiveness was found as an important ATM quality by both Khan (2010) and Kumbhar (2010). ATM quality dimensions and overall service quality Prior research indicates differential effects of the dimensions of ATM quality on customers overall perception of ATM service quality (Khan, 2010; Al-Hawari and Ward, 2006). As a result of the extensive literature reviews and the theoretical conceptualisation proposed for this study, the current study proposes reliability, convenience, ease of use, security and privacy, accuracy, fulfillment and responsiveness as the relevant dimensions of the ATMqual scale, hence they will form the bases for the empirical investigation. Methodology The aim of the paper is to determine the dimensions of ATM service quality and the relative importance of these dimensions in predicting customers' overall ATM service quality. In order to achieve the aim of the research, a multi-step process which demands systematic adherence to best practices was followed at each stage of the research process (Churchill, 1979; Blankson et al., 2007). Development of the scale The scale for the study was developed through a two-stage process. The first stage involved a review of the main measurement scales used in previous studies of electronic services notably the E-S-QUAL (Parasuraman et al., 2005), Webqual Downloaded by QATAR UNIVERSITY At 22:38 04 May 2015 (PT) (Loiacono et al., 2002) and the ATM quality dimensions suggested by Joseph and Stone (2003); Mobarek (2007); Dilijonas et al. (2009) and Khan (2010). The reviews produced a list of 40 items considered critical for measuring e-banking service quality. As the aim was to develop a specific scale for measuring ATM service quality in Ghana, the second stage of the process involved detailed focus group discussions with actual users of ATMs. Subsequently, three focus group discussions, each made up of ten convenient samples of ATM users, were held. The first group was made up of undergraduate students, the second group was made up of graduate students and the third group was made up of workers of the University of Ghana, respectively. University of Ghana was selected because it is the oldest University in Ghana with the highest student and worker population. Moreover, there is a proliferation of banks and ATM terminals on the University campus. Staff and students of the university are therefore familiar with the ATM environment for a long time and could, subsequently, participate in the focus group discussions. Following good practices (Kirk and Miller, 1986), the participants were asked to freely talk and then list the words and statements that described their expectations and experiences with regard to ATM usage. Words such as accurate recording of transactions, proper functioning of the ATM, protection of personal data and information, to quick repair if the ATM breaks down were openly mentioned and discussed in the groups. Similarly, issues of ATM cards being \"trapped in\" as well as transaction recovery through an identified contact person, to \"compensation\" when monies are illegally withdrawn from customer's accounts also featured prominently in the discussions. Insights from the interviews were subsequently utilised to further refine the survey instrument. To enhance the content validity of the survey instrument, further consultations were held with three ATM custodians from three banks in order to gain more insight into the perceptions of customer's ATM experience. Based on the literature review, focus group discussions and consultations with the ATM custodians, a 28-item survey instrument was developed that was further discussed with two distinguished professors each with a long history of research in electronic services. Based on their recommendations, some of the statements were re-worded to make them clearer to the respondents. Finally, the revised 28-item survey instrument, anchored on a Likert type scale ranging from 5 (strongly agree) to 1 (strongly disagree) were pilot tested with 20 ATM users, each with a considerable number of years of ATM experience. The result indicated that the wording in English was clear to the customers and also the variables appeared comprehensive enough to capture customer's ATM expectations. The measures undertaken above to enhance the validity of the survey instrument is consistent with best practice in the literature (Churchill, 1979; Blankson et al., 2007). Table II presents the final 28 variables and their sources which were used for the empirical investigation. Data collection A copy of the questionnaire and a permission letter were sent to all the 26 banks in Ghana asking to use their ATM sites for the data collection. Only 15 banks agreed to participate in the study. Ten trained assistants were then deployed to 45 randomly selected ATM sites of the 15 banks for the data collection. Two ATM sites were randomly selected for each bank in Accra, the capital city, while one site was selected outside the capital in order to capture the ATM experiences Service quality in ATMs 73 MSQ 23,1 Downloaded by QATAR UNIVERSITY At 22:38 04 May 2015 (PT) 74 across a wider spectrum of customers located at different places in Ghana. A random and systematic sampling was adopted where every fifth customer who completed a transaction was intercepted, the purpose of the study explained to them, and those who agreed to participate in the study were given the forms and persuaded to fill it on the spot. Those who requested to take the questionnaire home were given self-addressed envelopes through which the completed questionnaires were mailed back to the researcher. At the end of three weeks, 540 questionnaires were received out of which 530 were found complete and processed for the analysis. Descriptive statistics, factor analysis and multiple regression were used for the data analysis. Data analysis, interpretation and results Demographic profile Within the context of the financial services industry, some of the most commonly investigated demographic variables are age, gender, education and occupation. An analysis of the gender composition of the respondents indicated that 67.9 per cent of the respondents (n 360) were male while 32.1 per cent (n 170) of the respondents were female. Only 5.7 per cent of the respondents were o20 years while 83.8 per cent were between 21 and 50 years, with 10.1 per cent of the respondents being above 50 years. The respondents were also well educated as 79.7 per cent possessed diploma certificates, bachelor or postgraduate degrees. In terms of occupational distribution, 41.5 per cent were salaried workers, 27.9 per cent were students mostly of tertiary institutions, while 24.9 per cent were self-employed. Furthermore, an analysis of the ATM usage pattern of the respondents revealed that about 80.3 per cent of the respondents used their cards for transactions once or twice in a week, 14 per cent of the respondents used their ATM cards thrice in a week while 5.7 per cent claimed they used their ATMs four times per week. The result appears in Table III. Descriptive statistics Table IV displays the means and standard deviations of the various variables used. The results indicate moderate to high mean values. The highest means were 4.4743 (ATM gives instant money) and 4.4371 (ATM provides receipt to confirm transactions) whilst the lowest was 1.7943 (ATM gives quality bank notes). Hence, it is evident from Table IV that a lot of the respondents agreed that the ATMs of their banks provide them with instant cash. Exploratory factor analysis (EFA) The variables were also factor analysed. Prior to the extraction of factors, the Bartlett test of sphericity (approximately: w2 1,165.162, df 162, significance 0.000) and the KMO measure of sampling adequacy (value of 0.808) confirmed that there was significant correlation among the variables to warrant the application of EFA. Only factors whose eigenvalues were equal to or 41 were selected (Malhotra and Birks, 2007). Moreover, variables with loadings of at least 0.5 (Hair et al., 2006) and factors with a reliability threshold of 0.7 (Hair et al., 2006) were included in the analysis. In the initial exploration, eight variables were deleted because they had loadings of o0.5 (I have trust and security in the ATMs; ATM cards are promptly Downloaded by QATAR UNIVERSITY At 22:38 04 May 2015 (PT) Profile of respondents Age Under 20 21-30 31-40 41-50 51-60 461 Total Gender Male Female Total Education SSSCE Diploma Degree Postgraduate Others Total Occupation Student Salaried worker Self-employed Pensioner Total ATM usage per week Once Twice Thrice Four times Total Number % 30 194 182 70 28 26 530 5.7 36.6 34.3 13.2 5.3 4.9 100 360 170 530 67.9 32.1 100 35 58 235 130 72 530 6.6 10.9 44.4 24.5 13.6 100 148 220 132 30 530 27.9 41.5 24.9 5.7 100 256 170 74 30 530 48.3 32.0 14.0 5.7 100 replaced; ATM satisfies most of my banking needs; ATM banking settles complaints in a reasonable time; ATM cards are compatible on other ATM platforms; ATM charges are reasonable; I fear for security of my personal information; ATM gives instant money all the time). The rest of the 20 variables were factor analysed, which yielded eight factors as illustrated on Table V. The eight factors explain satisfactory 65.907 per cent of the variance. Varimax rotated principal component loadings The rest of the 20 variables were then rotated using the Varimax rotation with Kaiser normalisation. The results indicated that the rest of the 20 variables loaded onto six factors and there was no cross-loading on any factor. Two variables were, however, re-specified. \"ATM provides enough money during transactions\" which originally loaded onto reliability was added to fulfillment variables due to conceptual fitness. Similarly, \"Machine breakdowns are fixed promptly\" was re-specified from reliability to responsiveness. In both cases, the re-specification led to higher reliability for the Service quality in ATMs 75 Table III. Profile of respondents MSQ 23,1 Downloaded by QATAR UNIVERSITY At 22:38 04 May 2015 (PT) 76 Table IV. Descriptive statistics Variables ATM provides different range of services ATM provides enough money during transactions ATM Functions all the time ATM provides consistent services Breakdown ATMs are fixed promptly I have trust and security in the ATM ATM cards are promptly replaced ATM language is easy to understand ATM is easy to use for transactions ATM provides graphics and adverts of bank services ATM provides education and direction of usage ATM satisfies most of my banking needs ATM provides accurate account records ATM banking settles complaints in a reasonable time ATM machines are conveniently located ATM cards are compatible on other ATM platforms ATM provides fast services ATM waiting time is acceptable ATM charges are reasonable ATM provides clear instructions on usage The ATM system compensates for wrongful deductions ATM contact person is available for redress of problems ATM problems are settled to my satisfaction I fear for security of my personal information I feel safe during ATM transactions ATM gives instant money all the time ATM gives quality bank notes ATM provides precise receipt to confirm transactions Mean SD 2.85 3.16 2.17 3.21 2.38 3.57 3.38 4.20 4.10 3.03 3.51 2.67 3.79 2.63 3.40 3.10 4.06 4.17 2.76 2.90 3.23 3.33 3.18 3.96 4.00 4.47 1.79 4.44 1.157 1.319 1.114 1.103 1.171 1.189 1.251 0.899 0.893 1.120 1.096 1.215 1.067 1.218 1.218 1.401 0.864 0.869 1.287 1.044 1.027 1.017 1.031 0.785 0.790 0.653 0.649 0.664 latent factors as suggested by Hair et al. (2006). Four variables loaded highly on factor 1 and were all related to \"responsiveness\". This factor included items such as ATM system compensates for wrongful deductions, ATM contact person is available for redress of problems, ATM problems are settled to my satisfaction and broken down ATMs are fixed promptly. The second factor had five variables which were all related to \"ease of use of ATMs\". Variables that loaded on this factor included ATMs provide clear instructions for usage, ATMs are easy to use for transactions, ATM language is easy to understand, the ATM provides graphics and adverts of bank services and ATM provides education and direction of usage. Factor 3 also had three variables which related to \"reliability\" of the ATM with items in the form of ATM provides consistent services, ATM functions all the time and ATM provides accurate services. Factor four had items dealing with the \"convenience\" in using the ATMs. Items that loaded onto this factor include ATMs provide a different range of services, ATM waiting time is acceptable and ATMs are conveniently located. The fifth factor contained results related to \"fulfillment\" with items such as ATMs provide receipts to confirm transactions, ATMs give quality bank notes, ATM provides enough money during transactions and ATMs provide fast services. Finally only one variable (I feel safe during ATM transactions) loaded onto factor 6 and as such, was not used for further analysis. Downloaded by QATAR UNIVERSITY At 22:38 04 May 2015 (PT) Variables ATM provides different range of services ATM provides enough money during transactions ATM Functions all the time Breakdown ATMs are fixed promptly ATM language is easy to understand ATMs are easy to use for transactions The ATM provides graphics and adverts of bank services ATM provides education and direction of usage ATMs are conveniently located ATM provides accurate account records ATM provides fast services ATM waiting time is acceptable ATM provides clear instructions on usage The ATM system compensates for wrongful deductions ATM contact person is available for redress of problems ATM problems are settled to my satisfaction I feel safe during ATM transactions ATM provides precise receipt to confirm transactions ATM gives quality bank notes % of Cumulative Communality Factor Eigenvalue variance % 0.678 1 3.299 11.644 11.644 0.692 0.708 0.659 0.829 0.848 2 3 4 5 6 2.243 1.491 1.172 1.128 1.097 10.990 8.600 7.982 7.946 7.034 22.635 31.234 39.216 47.162 54.197 0.749 7 1.081 5.900 60.097 0.746 0.771 0.831 0.765 0.830 0.766 8 1.011 5.810 65.907 Service quality in ATMs 77 0.822 0.732 0.766 0.873 0.794 0.607 Reliability of the factors The internal reliability of the six factors was analysed through Cronbach's a coefficient. Only factors that met the minimum value of 0.7 as postulated by (Hair et al., 2006) were accepted. Also, item-to-total correlation was set above 0.3 (Parasuraman et al., 1988). Out of the six factors, five were maintained because they had Cronbach's a values of between 0.712 and 0.884. The security factor was dropped from further analysis because only one variable loaded onto it and as such, its reliability could not be ascertained. The average variance extracted (AVE) of the different factors, defined as the average variance shared between a construct and its measures was above 0.50 for each factor (Fornell and Larcker, 1981; Hair et al., 2006). This indicates that majority of the variance was explained by the five factors. The result is illustrated on Table VI. Final factors The final factors that explain ATM service quality within the Ghanaian banking industry are now presented. Following analysis of the initial EFA, reliability analysis, conceptual coherency and validity of the indicated factors, it is suggested that the five factors identified in Table V (reliability, convenience, responsiveness, ease of use and fulfillment) represent, tentatively a robust matrix of dimensions underpinning the ATMqual model proposed in this study, which predicts ATM service quality in the Ghanaian banking sector. Table V. Principal component factor loadings MSQ 23,1 Downloaded by QATAR UNIVERSITY At 22:38 04 May 2015 (PT) 78 Table VI. Internal consistency and final revised structure Factor and items Factor 1: responsiveness The ATM system compensates for wrongful deductions ATM problems are settled to my satisfaction ATM contact person is available for redress of problems Breakdown ATMs are fixed promptly Factor 2: ease of use ATM provides clear instructions on usage ATMs are easy to use for transactions ATM language is easy to understand ATM provides graphics and adverts of bank services ATM provides education and direction of usage Factor 3: reliability ATM Functions all the time ATM provides consistent services ATM provides accurate services Factor 4: convenience ATM provides different range of services ATMs are conveniently located ATM waiting time is acceptable Factor 5: fulfillment ATM provides receipt to confirm transactions ATM gives quality bank notes ATM provides fast services ATM provides enough money during transactions Factor 6: security I feel safe during ATM transactions No. of items Loadings Item-total correlation 4 0.822 0.766 0.632 0.563 0.732 0.659 0.589 0.617 0.766 0.848 0.829 0.641 0.596 0.605 0.749 0.746 0.529 0.693 0.708 0.716 0.765 0.627 0.689 0.560 0.678 0.771 0.830 0.586 0.519 0.643 0.794 0.607 0.831 0.587 0.624 0.553 0.692 0.635 5 3 3 4 Cronbach's a AVE 0.785 0.56 0.765 0.62 0.885 0.53 0.712 0.58 0.796 0.54 1 0.873 The dependent variable The three variables measuring overall ATM service quality (I am satisfied with the performance of the ATM; I will recommend the ATM to other customers; Overall, I am happy with the ATM service quality) were also checked for their loadings and Cronbach's a. Results showed that all the variables had high loadings between 0.784 and 0.901 with a Cronbach's a coefficient of 0.739. Thus, the three variables also formed a single matrix that measured consumer's perception of overall ATM service quality in the Ghanaian banking sector. Multiple regression analysis In order to test the relationship between the dimensions of ATM service quality and overall service quality, multiple regression analysis was used. Overall ATM service quality was used as the dependent variable whilst the independent variables were represented by recovery, ease of use, convenience, reliability and fulfillment. The compute command in SPSS version 20 was used to compute the value loadings associated with the variables measuring each of the factors based on numeric transformations in order to obtain values for both the dependent and independent Downloaded by QATAR UNIVERSITY At 22:38 04 May 2015 (PT) variables. Table VII presents a summary of the multiple regression least square results for the dependent and independent variables. The results indicate that there is a strong and significant relationship between the dimensions of ATM service quality and overall ATM service quality (F 20.087, probability F-statistico0.00). The R2-value 0.574 indicated that the independent variables explained 57.4 per cent of the variance in ATM service quality with an adjusted R2 of 56.5 per cent. On the individual determinants, reliability of ATM services was found to be the most important determinant of overall ATM service quality (b 0.222, t 3.986, p 0.000o0.05). This was followed by convenience of ATMs (b 0.217, t 3.688, p 0.000o0.05). The third predictor was responsiveness (b 0.060, t 3.582, p 0.22740.05) while the forth was ease of use (b 0.115, t 2.1195, p 0.023o0.05). The final determinant of overall ATM service quality was fulfillment (b 0.059, t 1.196, p 0.23340.05) but this latent factor was not a statistically significant predictor of overall ATM service quality. Service quality in ATMs 79 Confirmatory factor analysis After identifying the five factors through EFA, a confirmatory factor analysis was used to enable cross-validation of the model. The 20 variables found in Table V were put into another questionnaire which was used to collect data from another 230 ATM users randomly selected from 10 ATM sites in Accra, the capital city of Ghana. Structural equation modelling (SEM) using AMOS 17.0 was used to perform the confirmatory factor analysis. The analysis indicated that the variables loaded in accordance with the pattern revealed in the EFA. In line with Hair et al. (2006) the ATMqual model fit was evaluated. Specific fit indices used were the Chi-square, comparative fit index (CFI), normed fit index (NFI), non normed fit index (NNFI), goodness-of-fit index (GFI) and root mean square error of approximation (RMSEA). Scholars (Byrne, 2001; Holmes-Smith et al., 2004; Hair et al., 2006) have argued that a value of 0.90 is required to accept a model as good while a value of at least 0.95 is needed to judge the model fit as good. For the RMSEA, a value o0.05 is required to judge the model fit as good while an upper level of 0.08 is still acceptable (Hair et al., 2006). The following results were obtained for the fit indices: w2 4405, df 182, CFI 0.961, NNFI 0.958, NFI 0.98, GFI 0.982, RMSEA 0.045. Taken together, the fitness indices indicated that the measurement model fits the data very well. Variable Constant Responsiveness Ease of use Reliability Convenience Fulfillment SE of estimate R2 Adjusted R2 b SE t Probability 0.060 0.115 0.222 0.217 0.059 0.331 0.054 0.066 0.052 0.060 0.073 2.752 3.582 2.195 3.986 3.688 1.196 0.056 0.000** 0.003* 0.000** 0.000** 0.233 F-statistics Probability (F-statistic) 20.087 0.000 0.622 0.574 0.565 Notes: n 530. *p o0.05, **p o0.01 Table VII. Multiple regression analysis MSQ 23,1 Downloaded by QATAR UNIVERSITY At 22:38 04 May 2015 (PT) 80 Reliability and validity The dimensionality of the model was first tested after the EFA by reviewing the loadings of the factors. Each item had a significant loading on the intended factors and no significant loading on another factor was found. The scales were, therefore, found to be uni-dimensional (Hair et al., 2006). The reliability of the model was also explored through Cronbach's a coefficient and the item-to-total correlation (Hair et al., 2006; Parasuraman et al., 1988). The a coefficient values for the factors were as follows: reliability (0.885), convenience (0.712), responsiveness (0.785), fulfillment (0.796) and ease of use (0.765). All the reliability coefficients surpassed the 0.7 lower limit threshold of acceptability (Hair et al., 2006). Moreover, all the variables set up in Table V had item-total correlation value of at least 0.519, which are higher than the lower limit of 0.3 put forth by Parasuraman et al. (1988). Content validity was derived through an exhaustive step-by-step approach throughout the research process. A thorough literature review was first conducted to determine the dimensions of ATM service quality. Due to the perceived unique nature of the Ghanaian banking environment, three focus group discussions were conducted which were used to design the questionnaires. Further consultations were held with three ATM custodians from two multi-national banks and a local bank in order to gain further insight into their ATM experiences with customers. Furthermore, the resulting items were discussed with two professors of e-banking who helped in the refinement of the variables used in the questionnaire. Finally, the resulting instrument was pilot tested with 20 ATM users. These steps conform to best practices in the literature (Blankson et al., 2007). Convergent validity, which defines the degree to which items of a given construct measure the same latent construct (Fornell and Larcker, 1981; Hair et al., 2006) was assessed through the AVE, standard factor loadings and the fit indices (Fornell and Larcker, 1981; Blankson et al., 2007). As a rule of thumb, the AVE for each factor should exceed 0.5 to achieve convergent validity (Fornell and Larcker, 1981). The AVE scores in Table VI range between 0.52 and 0.62, indicating that the factors underpinning ATM service quality found in this study demonstrate convergent validity. In addition, convergent validity was also investigated through the EFA where the extracted latent factors can be seen as a set of common underlying dimensions of the research construct. Moreover, the fit indices obtained in the confirmatory factor analysis all met the acceptable criteria. Finally, most of the variables measuring the ATMqual model had high loadings in the region of 0.7 and above. Discriminant validity according to Fornell and Larcker (1981) is a measure of the extent to which latent factors are distinct, i.e. they should not correlate so highly to seem to measure the same underlying dimension (Siekpe, 2005). Fornell and Larcker (1981) and Hair et al. (2006) postulate that discriminant validity can be assessed using the AVE with the corresponding inter-construct squared correlation estimates. The AVE values