UBC Theses and Dissertations

UBC Theses Logo

UBC Theses and Dissertations

Understanding e-service failures : formation, impact and recovery Tan, Chee-Wee 2011

Your browser doesn't seem to have a PDF viewer, please download the PDF to view this item.

Item Metadata

Download

Media
24-ubc_2011_fall_tan_cheewee.pdf [ 1.94MB ]
Metadata
JSON: 24-1.0102435.json
JSON-LD: 24-1.0102435-ld.json
RDF/XML (Pretty): 24-1.0102435-rdf.xml
RDF/JSON: 24-1.0102435-rdf.json
Turtle: 24-1.0102435-turtle.txt
N-Triples: 24-1.0102435-rdf-ntriples.txt
Original Record: 24-1.0102435-source.json
Full Text
24-1.0102435-fulltext.txt
Citation
24-1.0102435.ris

Full Text

UNDERSTANDING E-SERVICE FAILURES: FORMATION, IMPACT AND RECOVERY by Chee-Wee Tan B.Sc. National University of Singapore, Singapore, 2001 M.Sc. National University of Singapore, Singapore, 2004  A THESIS SUBMITTED IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF DOCTOR OF PHILOSOPHY in The Faculty of Graduate Studies (Business Administration)  THE UNIVERSITY OF BRITISH COLUMBIA (Vancouver)  July 2011 © Chee-Wee Tan, 2011  A BSTRACT E-service failure has been the bane of e-commerce by compelling consumers to either abandon transactions entirely or to switch to traditional brick-and-mortar establishments. More often than not, it is not the manifestation of e-service failure that drives away consumers, but rather, the absence or inadequacies of service recovery solutions that led to undeserved anger and frustration. Yet, despite the ‘dangers’ posed by e-service failures, there has not been a study to-date that systematically investigates how perceptions of failure emerge within an online transactional environment and what can be done to address these sources of potential consumer disappointments. Drawing on the Expectation Disconfirmation Theory (EDT) and the Counterfactual Thinking Perspective, this study synthesizes contemporary literature to arrive at separate typologies of e-service failure and recovery. Then, an integrated theory of e-service failure and recovery is constructed together with testable hypotheses. To empirically validate the model, two studies have been conducted and their designs elaborated. Essentially, findings from the two studies serve to inform both academics and practitioners on: (1) how consumer perceptions of different types of e-service failure manifest on e-commerce websites; (2) the impact of these perceptual failures on consumers’ expectations about transactional outcome, process and cost, as well as; (3) what kind of e-service recovery technology would be beneficial in alleviating negative failure consequences.  ii     P REFACE This research was approved by the University of British Columbia Behavioral Research Ethics Board (certificate number H10-00051).  iii     T ABLE OF C ONTENTS Abstract ................................................................................................................................................. ii  Preface .................................................................................................................................................. iii  Table of Contents................................................................................................................................. iv  List of Tables ....................................................................................................................................... vii  List of Figures ...................................................................................................................................... ix  Acknowledgements ............................................................................................................................... x  Chapter 1 – Introduction ..................................................................................................................... 1  1.1   Motivation and Research Objectives ...................................................................................... 1   1.2   Guiding Framework ................................................................................................................ 5   1.3   Dissertation Structure ............................................................................................................ 8   Chapter 2 – An Expectancy Perspective of E-Service Failure ....................................................... 10  2.1   E‐Service Failure: A Review of Alternate Frameworks ......................................................... 11   2.2   A System‐Oriented Typology of E‐Service Failures ............................................................... 15   2.2.1   Informational Failures .................................................................................................. 19   2.2.2   Functional Failures ....................................................................................................... 20   2.2.3   System Failures ............................................................................................................ 23   2.2.4   A System-Oriented Typology of E-Service Failures .................................................... 25   2.3   Summary ............................................................................................................................... 26   Chapter 3 – An Exploratory Study of E-Service Failure Causes (1st Study) ................................ 27  3.1   Questionnaire Development ................................................................................................ 28   3.2   Data Collection ..................................................................................................................... 30   3.3   Data Analysis ........................................................................................................................ 32   3.3.1   Analytical Procedures ................................................................................................... 33   3.3.2   Findings from Framework Comparison ....................................................................... 35  iv     3.4   Discussion ............................................................................................................................. 37   3.4.1   Implications for Research ............................................................................................. 37   3.4.2   Implications for Practice ............................................................................................... 38   3.4.3   Limitations .................................................................................................................... 39   3.4.4   Summary ....................................................................................................................... 40   Chapter 4 – A Counterfactual Thinking Perspective of E-Service Recovery ............................... 41  4.1   A Proposed Typology of E‐Service Recovery ........................................................................ 42   4.2   Summary ............................................................................................................................... 48   Chapter 5 – An Integrated Theory of E-Service Failure and Recovery ........................................ 49  5.1   An Expectation Disconfirmation Perspective of E‐Service Failure Consequences ............... 50   5.1.1   Consequences of Informational Failures ...................................................................... 53   5.1.2   Consequences of Functional Failures ........................................................................... 53   5.1.3   Consequences of System Failures ................................................................................ 54   5.2   A Counterfactual Thinking Perspective of E‐Service Recovery Effectiveness ...................... 55   5.2.1   Moderating Effect of Compensatory Recovery Technology ........................................ 57   5.2.2   Moderating Effect of Response Sensitivity Recovery Technology .............................. 58   5.2.3   Moderating Effect of Affinity Recovery Technology .................................................. 59   5.3   Summary ............................................................................................................................... 60   Chapter 6 – An Experimental Study of E-Service Failure and Recovery (2ND Study) ................. 61  6.1   Experimental Design ............................................................................................................. 61   6.1.1   A General Overview of Experimental Procedures ....................................................... 63   6.1.2   Manipulations of E‐Service Failures ............................................................................. 69   6.1.3   Manipulations of E‐Service Recoveries ........................................................................ 74   6.1.4   Measurement and Survey Questionnaires ................................................................... 80   6.2   Data Analysis ........................................................................................................................ 84  v     6.2.1   Manipulations of E‐Service Failures ............................................................................. 88   6.2.2   Manipulations of E‐Service Recoveries ........................................................................ 91   6.2.3   Hypotheses Testing ...................................................................................................... 97   6.3   Discussion ........................................................................................................................... 109   6.3.1   Implications for Research ........................................................................................... 112   6.3.2   Implications for Practice ............................................................................................. 113   6.3.3   Limitations .................................................................................................................. 114   6.3.4   Summary ..................................................................................................................... 115   Chapter 7 – Conclusion and Discussion ......................................................................................... 116  7.1   Implications for Research and Practice .............................................................................. 118   7.2   Future Research .................................................................................................................. 120   7.3   Conclusion .......................................................................................................................... 121   References ......................................................................................................................................... 123  Appendix A – Categorization of Extant E-Service Literature ..................................................... 133  Appendix B – Detailed Breakdown of Classifications of E-Service Failure Incidents ............... 135  Appendix C – Classification of Exemplary E-Service Failure Incidents ..................................... 142  Appendix D – Illustrative Examples of E-Service Recovery Technology in Practice................. 146  Appendix E – Dunnett T3 Test for E-Service Failure Manipulations [Failure Treatment Comparisons] .................................................................................................................................... 148  Appendix F – Dunnett T3 Test for E-Service Recovery Manipulations [Recovery Treatment Comparisons] .................................................................................................................................... 150  Appendix G – Dunnett T3 Test for Impact of E-Service Failures on Disconfirmed Expectancies [Failure Treatment Comparisons] .................................................................................................. 155  Appendix H – Dunnett T3 Test for Impact of E-Service Recoveries on Disconfirmed Expectancies [Recovery Treatment Comparisons]........................................................................ 157  Appendix I – Graphical Plots of Impact of E-Service Recoveries on Disconfirmed Expectancies ............................................................................................................................................................ 161  vi     L IST OF T ABLES Table 2.1: Typology of Service Encounter Failures [as adapted from Bitner et al. (1990) and/or Bitner et al. (1994)] ......................................................................................................................................... 11 Table 2.2: Typology of Retail Failures [as adapted from Kelley et al. (1993)].................................... 12 Table 2.3: Typology of Online Service Failures [as adapted from Holloway and Beatty (2003)] ....... 13 Table 2.4: Proposed E-Service Failure Typology ................................................................................ 19 Table 3.1: Descriptive Statistics for Online Survey Respondents [Sample N = 211] .......................... 31 Table 3.2: Intra- and Inter-Judge Reliabilities of E-Service Failure Classifications ............................ 35 Table 4.1: Comparison of Contemporary Frameworks of Service Recovery....................................... 45 Table 6.1 Hypothesis to be Tested........................................................................................................ 62 Table 6.2: Between-Subjects Experimental Design ............................................................................. 62 Table 6.3: Sample E-Service Failure Incidents .................................................................................... 70 Table 6.4: Example of Apology for Informational Failure................................................................... 77 Table 6.5 List of Measurement Items [All items were measured using a 7-point Likert scale ranging from ‘Strongly Agree’ to ‘Strongly Disagree’] .................................................................................... 82 Table 6.6 Descriptive Statistics for Experimental Participants [Sample N = 575] ............................... 84 Table 6.7 Inter-Construct Correlation Matrix....................................................................................... 85 Table 6.8 Loadings and Cross-Loadings of Measurement Items ......................................................... 86 Table 6.9 Descriptive Statistics for E-Service Failure Constructs ....................................................... 88 Table 6.10 ANOVA Results for E-Service Failure Constructs [Manipulation Checks] ...................... 88 Table 6.11 Dunnett t-Test (2-sided)a for E-Service Failure Constructs................................................ 89 Table 6.12 Descriptive Statistics for Controllability ............................................................................ 90 Table 6.13 One-Sample t-Test for Controllability................................................................................ 91 Table 6.14 ANOVA Results for Controllability ................................................................................... 91 Table 6.15 Dunnett T3 Test for Controllability .................................................................................... 91 Table 6.16 Descriptive Statistics for E-Service Recovery Constructs ................................................. 92 vii     Table 6.18 Dunnett t-Test (2-sided)a for E-Service Recovery Constructs............................................ 94 Table 6.19 Descriptive Statistics for Realism ...................................................................................... 95 Table 6.20 One-Sample t-Test for Realism .......................................................................................... 96 Table 6.21 ANOVA Results for Realism ............................................................................................. 96 Table 6.22 Dunnett T3 Test for Realism .............................................................................................. 96 Table 6.23 Descriptive Statistics for Disconfirmed Expectancy Constructs [by Failure Treatments] . 98 Table 6.24 ANOVA Test of Between-Subjects Effects [Failure Treatments  Disconfirmed Expectancies]........................................................................................................................................ 99 Table 6.25 Dunnett t-Test (2-sided)a for Disconfirmed Expectancy Constructs [by Failure Treatments] .............................................................................................................................................................. 99 Table 6.26 Descriptive Statistics for Disconfirmed Expectancy Constructs [by Recovery Treatments] ............................................................................................................................................................ 100 Table 6.27 Tests of Within-Subjects Contrasts for Disconfirmed Expectancy Constucts [Recovery Treatments  Disconfirmed Expectancies]....................................................................................... 103 Table 6.28 One-Sample t-Test for E-Service Recovery Treatment .................................................... 104 Table 6.29 One-Sample t-Test for Disconfirmed Expectancy Constructs [Test Value = 0.000] ....... 105 Table 6.30 Summary of Dunnett T3 Test for Disconfirmed Outcome Expectancy [Recovery Treatment Comparisons] .................................................................................................................... 106 Table 6.31 Summary of Dunnett T3 Test for Disconfirmed Process Expectancy [Recovery Treatment Comparisons]...................................................................................................................................... 107 Table 6.32 Summary of Dunnett T3 Test for Disconfirmed Cost Expectancy [Recovery Treatment Comparisons]...................................................................................................................................... 108 Table 6.33 Summary of Hypotheses Testing...................................................................................... 109 Table B-1: Typology of Service Encounter Failures [as adapted from Bitner et al. (1990) and/or Bitner et al. (1994)] [Sample N = 374] ............................................................................................... 135 Table B-2: Typology of Retail Failures [as adapted from Kelley et al. (1993)] [Sample N = 374] ... 137 Table B-3: Typology of Online Service Failures [as adapted from Holloway and Beatty (2003)] [Sample N = 374] ............................................................................................................................... 138 Table B-4: Proposed E-Service Failure Typology [Sample N = 374] ................................................ 140 viii     L IST OF F IGURES Figure 1.1: Overarching Theoretical Framework ................................................................................... 7 Figure 3.1: Diagrammatic Flow of Online Survey Questionnaire ........................................................ 30 Figure 3.2: Diagrammatic Flow of Content Analytical Procedures for E-Service Failure Incidents ... 33 Figure 5.1: Theory of E-Service Failure and Recovery ........................................................................ 50 Figure 6.1: Experimental Website from Control Group ....................................................................... 63 Figure 6.2: Introductory Page and Electronic Consent Form of Experimental Websites..................... 65 Figure 6.3: Description of Experimental Task ..................................................................................... 67 Figure 6.4: Diagrammatic Flow of Experimental Procedures .............................................................. 69 Figure 6.5: Illustration of Control versus Informational Failure Manipulation .................................... 71 Figure 6.6: Illustration of Control versus Functional Failure Manipulation......................................... 73 Figure 6.7: Illustration of Compensatory E-Service Recovery Technologies ...................................... 76 Figure 6.8: Illustration of Affinity E-Service Recovery Technologies ................................................ 78 Figure 6.9: Illustration of Response Sensitivity E-Service Recovery Technologies ............................ 79  ix     A CKNOWLEDGEMENTS The completion of this dissertation commemorates the end of my doctoral education in the University of British Columbia (UBC). Having spent the last six and a half years in UBC, I am extremely grateful to the team of dedicated faculty members and professional administrators, who have made my stay both intellectually stimulating and academically rewarding. Particularly, I will like to express my gratitude to a special group of people without whom this achievement would not be possible. First and foremost, I would like to express my gratitude to my supervisors, Dr. Izak Benbasat and Dr. Ron Cenfetelli, for their advice and guidance during the course of my study. Izak and Ron, thank you for helping me to grow both professionally as well as personally. I have benefited tremendously from your experience and insights and they have inspired me to aim for greater heights academically. Also, I would like to thank my dissertation committee member, Dr. Tim Silk for taking the time to share his knowledge and expertise on several occasions, which led to significant improvements to this dissertation. I am also grateful for the support and encouragement from friends and fellow colleagues, who have went through the same journey as I have. Among others, I would like to particularly express my appreciation to Bo Xiao, Ali Dashti, Sameh Al-Natour and Dr. Hasan Cavusoglu for the many good times we have shared during the past few years. Special Thanks Finally, I wish to dedicate this dissertation to my family members and especially my brother for their words of encouragement whenever I am down. Thank you for always being there for me. It is only because of your unwavering supporting and strong vote of confidence that I can continue to strive against all odds and accomplish this deed.  x     C HAPTER 1 – I NTRODUCTION E-service failures are common occurrences in e-commerce transactions. In a comprehensive review of modern websites spanning multiple industries, Oneupweb (2010), a digital marketing agency, reported that e-commerce transactions continue to exhibit an alarming 45% failure rate. Similar findings were echoed in Harris Interactive’s (2006) survey of 2,790 online consumers in the United States, revealing that 88% of consumers experienced problems when transacting online. The study by Harris Interactive (2006) further uncovered that e-service failures negatively impact emerchants by forcing 40% of online consumers to abandon transactions entirely (8%) or to switch to a physical competitor (32%). These results are corroborated in Forrester Consulting’s (2009) survey of 1,048 online shoppers: 79% of online shoppers who encountered any form of e-service failure will no longer purchase from the faulty website, 46% will develop a negative impression of the e-merchant, and 44% will notify friends and family about the negative experience. Additionally, 91% of consumers who experienced any form of e-service failure also stated that they are more likely to question e-merchants’ ability to safeguard confidential personal information disclosed during online transactions (Harris Interactive 2006). This implies that failure in one aspect of an e-commerce transaction will produce a negative spillover effect, causing customers to lose faith in other facets of the transactional process. This spillover could be attributed to heightened emotions during service failures that obstruct cognitive reasoning (Andreassen 2001; McColl-Kennedy and Sparks 2003). Because of the spillover, e-service failures may also adversely affect e-businesses in general: consumers may be reluctant to engage in e-commerce transactions as a consequence of earlier bad experiences.  1.1  Motivation and Research Objectives When service failures occur, consumers expect vendors to be competent and caring in  offering commensurable recovery measures (Bitner et al. 1990). Empirically, Smith et al. (1999) affirmed that it is possible to recover from almost any kind of service failure, regardless of its form 1     and magnitude, so long as the recovery measure is commensurate with the failure experienced by consumers. While service failures may be unwelcome occurrences, the effectiveness of corresponding service recovery measures determines whether consumers would be appeased and retained (Holloway and Beatty, 2003). As noted by Spreng et al. (1996), service recovery offsets consumers’ negativism towards failure events in three ways: (1) providing assurance of the fairness and sincerity of the offending vendor (i.e., admits to mistakes and makes restitution); (2) lessening the magnitude of negative consequences arising from the failures, and; (3) persuading victims to cast the blame elsewhere. Since e-commerce is reliant on the IT-enabled web interface as the focal point of contact between consumers and e-merchants (Benbasat and DeSanctis, 2001; Cenfetelli et al., 2008), it is not only susceptible to conventional forms of offline retail failures, such as wrong product delivery or slow customer service, it can also succumb to technological malfunctions. For instance, Forrester Consulting’s (2009) survey discovered that webpage loading delays and site crashes are the leading causes of failures for e-commerce websites, accounting for 23% and 17% of dissatisfactions in consumers respectively. If such e-service failures were to be met with incommensurable service recovery efforts, it may erode what little confidence customers may have with e-commerce transactions. Indeed, the empirical findings of Holloway and Beatty (2003) concluded that e-service recoveries are generally deemed to be inadequate or inequitable relative to the failures experienced. An in-depth appreciation of e-service failure and recovery hence makes a significant and timely contribution to extant literature for four reasons. First, although service failure and recovery is gaining momentum within marketing literature as an important investigative topic (e.g., de Matos et al., 2007; Rinberg et al., 2007, Smith et al., 1999; Tax et al., 1998), there is general consensus that we still have a somewhat limited understanding of the phenomenon, especially with regards to the e-commerce transactional environment (Holloway and Beatty, 2003). This trend is even more prevalent in the domain of Management Information Systems (MIS) where a review of prominent journals from the basket of eight (i.e., European Journal 2     of Information Systems, Information Systems Journal, Information Systems Research, Journal of Information Technology, Journal of Management Information Systems, Journal of Strategic Information Systems, Journal of the Association for Information Systems and MIS Quarterly) from 2001 to 2010 indicates that research into e-service failure and recovery is practically non-existent. Second, e-commerce is distinct from offline retail in that the entire transaction is accomplished via web-enabled services (Cenfetelli et al., 2008; Homburg et al., 2002). As the contact points increase between consumers and web technologies, opportunities for e-service failures grow exponentially as well (Holloway and Beatty, 2003). Specifically, e-commerce websites, due to their reliance on web technologies, are extremely vulnerable to the aftermath of failure occurrences as there is little room for physical intervention (Holloway and Beatty, 2003). A deeper understanding of e-service failures is therefore necessary to stem the loss of consumers that is prevalent even among sophisticated e-commerce websites (Forrester Consulting, 2009). Third, previous studies have hinted that e-service failures are not necessarily reflections of their physical counterparts. For example, Bitner et al.’s (1990, 1994) and Kelley et al.’s (1993) typologies of service failures, while they can account for failure incidents within physical retail and offline service channels, appear out of place when contrasted with that of Holloway and Beatty’s (2003), which uncovered failures exclusive to e-commerce transactions. Comparatively, service failures identified in Bitner et al.’s (1990, 1994) and Kelley et al.’s (1993) typologies generally revolve around interactional conflicts between consumers and store employees (e.g., ‘wrongful accusation of customers or failure in dealing with uncooperative customers’), whereas failure dimensions advocated by Holloway and Beatty (2003) include several technologically-induced problems (e.g., ‘navigational or payment difficulties’). Further, contemporary frameworks lack sufficient theoretical grounding for formulating explanations and predictions on how consumers respond to service failures.  3     Fourth, it is well-accepted that the majority of consumers, when confronted with service failures, will choose to simply forsake the transaction and terminate their relationship with the vendor (Hart et al., 1990; Tax and Brown, 1998). This trend may be even more pronounced for e-commerce transactions. Because consumers tend to participate in pseudo-relationships with multiple e-merchants and can readily switch among e-commerce websites with the mere click of a mouse button, the majority of them, when confronted with e-service failures, will choose to simply forsake the transaction and terminate their relationship with the offending e-merchant (Harris Interactive, 2006). The provision of suitable e-service recovery technologies on e-commerce website can be construed as the only chance for e-merchants to redeem themselves in the unfortunate event of a failure. Yet, as admitted by Holloway and Beatty (2003), current e-commerce websites are not only lagging in the provision of e-service recovery technologies to alleviate probable e-service failures, but even when such technologies are made available, recovery measures are usually incommensurate with the damages suffered by consumers. Whenever consumers feel betrayed as in the case of unrecovered service failures, Ward and Ostrom (2006) warned that consumers may exact revenge on the offending vendor through spreading negative word of mouth or engaging in sabotaging behaviors. To bridge the aforementioned knowledge gaps, this thesis develops a theory that explains and predicts consumers’ reactions to e-service failures and recoveries. Particularly, by drawing on the eservice and system success literatures to derive a novel typology of e-service failure that captures failure events unique to e-commerce settings, this thesis is the first of its kind to undertake a deductive approach in systematically categorizing e-service failures. Further, we subscribe to Smith et al.’s (1999) typology of four service recovery modes in prescribing actionable design principles that may inform e-merchants in the development of a holistic e-service recovery solution for coping with e-service failures. In so doing, this thesis endeavors to answer the following research questions: 1. How do e-service failures manifest on e-commerce websites and what is their impact on online consumer behavior?  4     2. How can information technology be leveraged to design effective e-service recovery mechanisms for addressing various forms of e-service failure?  1.2  Guiding Framework E-service failures are damaging to e-commerce transactions by decreasing consumers’  likelihood of attaining predetermined objectives (Bitner et al., 2000; McCollough et al., 2000) and must be countered through the provision of commensurable service recovery technologies (Smith et al., 1999; Tax et al., 1998). Depending on the probability of service failures and the existence of commensurable recoveries, the service encounter presents itself as a window of opportunity through which existing customers can be retained or lost and prospective ones may be attracted or deterred (Bitner, 1990; Maxham III and Netemeyer, 2003; Folkes, 1984). An integrated theory of e-service failure and recovery is therefore necessary for two reasons. First, such a theory is desirable as a step towards unraveling the interactional effect between failure events and recovery technologies in influencing online consumer behaviors (Holloway and Beatty, 2003; Kelley et al., 1993). By treating service failures and recoveries as distinct phenomena within extant literature, Smith et al. (1999) noted that scholars essentially rob their studies of any realism because such a distinction does not reflect pragmatic business circumstances. More importantly, a theory of e-service failure and recovery endows researchers with an explanatory framework by which to examine “specific determinants of an effective recovery and the relative importance of individual recovery attributes in restoring customer satisfaction across a variety of service failure conditions” (Smith et al., 1999, p. 357; see also Holloway and Beatty, 2003; Kelley et al., 1993; Maxham III and Netemeyer, 2003). Yet, with the notable exception of Smith et al. (1999), there is a paucity of studies that consider service failure and recovery in tandem within a single nomological network. Even then, Smith et al.’s (1999) model lacks sufficient explanatory and predictive power in disentangling the formation, impact and recovery of e-service failures due to two reasons. One, in treating service failure as a singular construct, Smith et al.’s (1999) study sacrifices the richness inherent within multidimensional failure frameworks (e.g., Bitner et al., 1990, 1994; Holloway and Beatty, 2003; 5     Kelley et al., 1993) by assuming homogeneity in consequences across failure events. Two, due to its emphasis on offline retail, Smith et al.’s (1999) work does not cater to the contextual uniqueness of ecommerce transactions in its conceptualization of service failures and recoveries. This limits its applicability to e-commerce websites in terms of prescribing actionable design principles that could be harnessed by e-merchants to improve online transactional experiences. A review of extant literature on service failure and recovery uncovers two predominant research streams. Studies belonging to the first research stream can be construed as preventive in that they seek to comprehend the causes of service failures and their impact on consumer behaviors. Core contributions of this line of work reside in the advancement of descriptive typologies of service failures (i.e., Bitner et al., 1990, 1994; Holloway and Beatty, 2003; Kelley et al., 1993) and in-depth appreciation of consumers’ reactions towards failure events, which range from attributional inclinations (e.g., Bitner, 1990; Folkes, 1984; Hess et al., 2007; Leong et al., 1997; Maxham III and Netemeyer, 2002a; Taylor, 1994) to behavioral responses such as complaints (e.g., Bove and Robertson, 2005; DeWitt and Brady, 2003), vendor switching (e.g., Keaveney, 1995) and negative word-of-mouth (e.g., DeWitt and Brady, 2003; Maxham III and Netemeyer, 2002a; Weun et al., 2004). Conversely, the second research stream focuses on prescribing feasible corrective actions to be undertaken in addressing service failures. Apart from introducing typologies of viable service recoveries to accommodate the myriad of offline failures (Kelley et al., 1993; Smith et al., 1999; Tax et al., 1998), primary contributions of this research stream also encompass detailed inspection into the effectiveness of recovery measures, such as compensation (e.g., Mattila and Patterson, 2004a, 2004b; McColl-Kennedy et al., 2003), explanation (e.g., Mattila and Patterson, 2004a, 2004b), rapport (e.g., DeWitt and Brady, 2003; Rosenbaum and Massiah, 2007) and voice (e.g., McColl-Kennedy et al., 2003; Karande et al., 2007), in rectifying service failures. This thesis therefore endeavors to merge the aforementioned research streams by proposing a theory that not only accounts for consumers’  6     reactions towards e-service failures, but also predicts the effectiveness of various recovery technologies in coping with these failures when they manifest. To construct the theory and develop testable hypotheses for empirical inquiries, we draw extensively on the Expectation Disconfirmation Theory (EDT) to explain the impact of e-service failures on online consumers and Counterfactual Thinking to postulate the effectiveness of various e-service recovery technologies in moderating different failure consequence. Figure 1.1 depicts the overarching theoretical framework underlying this thesis. Figure 1.1: Overarching Theoretical Framework  Counterfactual Thinking   E-Service Recoveries  E‐Service  Failures   Disconfirmed  Expectancies            Expectation Disconfirmation Theory  Our theory is constructed in two stages. In the first stage, we draw on the EDT to provide an underlying conceptual foundation for theorizing e-service failures and their immediate consequences. We then synthesized the e-service and system success research streams to advance a novel typology of e-service failure that classifies failure events into informational, functional and system categories. Under each of these categories is a collection of constituent failure dimensions that are synonymous with technological deficiencies which could emerge on e-commerce websites. These dimensions translate to actionable design principles that can be exploited by e-merchants to improve consumers’ e-commerce transactional experiences. The Critical Incident Technique (CIT) methodology was then employed to solicit events of e-service failures from reality. These failure events were scrutinized, via  7     content analytical techniques, to ascertain whether our proposed typology is more suited to the appreciation and classification of e-service failures in comparison to contemporary frameworks. In the second stage, we subscribe to consumers’ counterfactual thinking process in arguing for the inclusion of e-service recovery as an integral part of service delivery for e-commerce websites. Then, building on our proposed e-service failure typology and Smith et al.’s (1999) typology of service recovery, we advance an integrated theory of e-service failure and recovery by drawing on: (1) the EDT to postulate negative consequences of information, functional and system failures, and; (2) Counterfactual Thinking to predict the effectiveness of various e-service recovery technologies in moderating these failure consequences. An experimental study was subsequently conducted to validate the causal relationships presented in this theory. Together, the two studies will lay the foundation for unlocking a new line of research in the area of e-service failure and recovery.  1.3  Dissertation Structure This thesis comprises a total of 7 chapters, inclusive of the introduction. In Chapter 2, an  expectancy perspective of e-services is presented. The EDT is introduced as the conceptual foundation for deriving a working definition of e-service failure. Contemporary frameworks of service failures are then reviewed to clarify the reasoning behind our advancement of a novel typology of e-service failure that assimilates knowledge from e-service and system success literatures. Next, Chapter 3 outlines the data collection and analysis strategy for the first study by providing detailed explanations on the CIT and content analytical procedures. Based on the empirical findings, we compare and contrast our proposed typology against contemporary frameworks in classifying pragmatic incidences of e-service failures. The chapter ends with a summary of the theoretical contributions and pragmatic implications of the first study as well as its potential limitations. Subscribing to the notion of counterfactual thinking, Chapter 4 provides a working definition for eservice recovery. In addition, the chapter showcases three contemporary frameworks of service recovery to explain the rationale for choosing Smith et al.’s (1999) typology as the guiding theoretical 8     framework for modeling e-service recovery. Chapter 5 builds on preceding chapters by advancing an integrated theory of e-service failure and recovery together with testable hypotheses. Specifically, the chapter touches on the conceptualization of e-service failure consequences, elaborating on the delineation of the disconfirmation construct in the EDT to better reflect the multiplicity of negative consequences that may befall consumers who encounter e-service failures. It is also contended in the chapter that different type of e-service recovery technology may be commensurable with different failure consequence. Based on the theory, Chapter 6 outlines the design of a repeated measures experiment that manipulates different configurations of e-service failure and recovery treatment to examine the impact of their interactions on negative failure consequences. Chapter 6 closes by summarizing the theoretical contributions and pragmatic implications of the second study together with its potential limitations. The last chapter, Chapter 7 concludes with a general overview of the theoretical contributions and pragmatic implications of this thesis, the insights to be gleaned towards informing the design of e-commerce websites, and propositions for further research.  9     C HAPTER 2 – A N E XPECTANCY P ERSPECTIVE OF ES ERVICE F AILURE An e-service encounter involves the entire transactional process that begins when a consumer visits a website to query products and/or services to the moment when a product or service, which matches the consumer’s needs, has been delivered to his/her satisfaction (Boyer et al. 2002). Service failures in general can be conceived as consumers’ evaluations of service delivery falling below their expectations or ‘zone of tolerance’ (Zeithaml et al. 1993). An e-service failure therefore arises whenever an e-commerce website lacks the technological capabilities essential for a consumer to accomplish his/her intended transactional activities. The Expectation Disconfirmation Theory (EDT) was championed by Oliver (1980) as a theoretical framework for deciphering consumers’ reactions to the performance of products and/or services in relation to their pre-consumption expectations. The EDT posits that expectations, coupled with product/service performance, determine consumers’ postconsumption attitudes. This effect, in turn, is mediated by the disconfirmation of consumers’ expectations through product/service performance. That is, if a product/service satisfies or outperforms expectations (i.e., positive disconfirmation), positive post-consumption attitudes will develop, whereas if the product/service falls short of expectations (i.e., negative disconfirmation), negative attitudinal responses will develop (Oliver 1980). Since the applicability of the disconfirmation paradigm in the investigation of service failures has been reinforced in extant literature (e.g., Andreassen 2001; Bearden and Teel 1983; Bitner 1990; Smith et al. 1999), we subscribe to the EDT in defining e-service failure as an event whereby the performance of an eservice on an e-commerce websites falls short of consumers’ expectations (see Hess et al. 2007). Next, we review existing frameworks of service failures to justify our stance for advocating a novel typology of e-service failures that better captures failure events representative of and unique to e-commerce transactional environments.  10     2.1  E-Service Failure: A Review of Alternate Frameworks The earliest attempt at deriving a typology of service failures was undertaken by Bitner et al.  (1990). Relying on the CIT, Bitner et al. (1990) solicited 352 incidents of dissatisfied service encounters from the airline, hotel and restaurant industries. They then employed inductive classification techniques to arrive at three core categories of service failures (i.e., failure of service delivery system, failure to meet customer needs and requests as well as unprompted and unsolicited service behaviors), each with its own constituent dimensions. This typology was later expanded by Bitner et al. (1994) who, in further classifying 774 critical incidents on service failures reported in the exact same three industries, uncovered a fourth category of service failure that relates to problematic customer behaviors (see Table 2.1). Table 2.1: Typology of Service Encounter Failures [as adapted from Bitner et al. (1990) and/or Bitner et al. (1994)] Construct   Definition (Event in which...)   Failure of Service Delivery System [as adapted from Bitner et al. (1990) and Bitner et al. (1994)]  Unavailable Service   Vendor fails to provide services  that are normally available or expected   Unreasonably Slow Service   Vendor is slow in servicing customers   Other Core Service Failure   Vendor  fails  to  meet  basic  performance  standards  for  other  aspects  of  the  core  service  (apart  from its absence or slowness)   Failure to Meet Customer Needs and Requests [as adapted from Bitner et al. (1990) and Bitner et al. (1994)]  Failure to Meet ‘Special Needs’  Customers    Vendor fails to recognize and accommodate customers’ special  demographical, physical and/or  sociological needs (e.g., disabilities)   Failure to Meet Customer  Preferences   Vendor  fails  to  recognize  and  accommodate  customers’  preferences  that  run  contrary  to  standard practices   Failure to Address Admitted  Customer Error   Vendor fails to resolve problems that arise from customers’ admitted errors   Failure to Manage Disruptive  Others    Vendor fails to deal appropriately with disruptive customers   Unprompted and Unsolicited Service Behaviors [as adapted from Bitner et al. (1990) and Bitner et al. (1994)]  Failure to Pay Attention to  Customer   Vendor fails to pay sufficient attention to customers during service encounters   Failure due to Out‐of‐the  Ordinary Service Behavior   Vendor  fails  to  perform  in  an  expected  manner  and  culminates  in  adverse  consequences  for  customers   Failure to be Sensitive to  Cultural Norms   Vendor fails to observe cultural norms during service encounters   Gestalt Evaluation Failure   Vendor fails to prevent isolated failures from affecting other related services   Failure to Perform Under  Adverse Circumstances   Vendor fails to perform efficaciously under unfavorable circumstances    Failure to Address Problematic Customer Behavior [as adapted from Bitner et al. (1994)]   11     Construct   Definition (Event in which...)   Failure to Address Drunken  Customers   Vendor fails to deal with  intoxicated customers who are causing troubles   Failure to Address Verbal and  Physical Abuse   Vendor fails to deal with customers who engage in physical and/or verbal abuses   Failure to Address Customers  Breaking Company Laws or  Policies   Vendor fails to deal with customers who refuse to comply with company rules and regulations   Failure to Address  Uncooperative Customers   Vendor fails to deal with customers who are generally rude, uncooperative and/or unreasonably  demanding   Bitner et al.’s (1990, 1994) typology of service failures has been applied by other scholars in investigating the phenomenon (e.g., McColl-Kennedy and Sparks 2003; Leong et al. 1997). However, because Bitner et al.’s (1990, 1994) typology is derived from pure service industries, it was subsequently refined by Kelley et al. (1993) for merchandise retailing. Like Bitner et al. (1990, 1994), Kelley et al. (1993) gathered 661 critical incidents on service failures involving general merchandise retailing (i.e., department stores, discount stores, variety stores and mail order retailers). Based on these critical incidents, Kelley et al. (1993) adapted Bitner et al.’s (1990, 1994) typology to the retail sector by collapsing overlapping dimensions, delineating ambiguous ones to improve clarity and eliminating the remainder that were deemed to be redundant (see Table 2.2). Table 2.2: Typology of Retail Failures [as adapted from Kelley et al. (1993)] Construct   Definition (Event in which...)   Failure of Service Delivery System and/or Product  Policy Failure   Vendor fails to enact service policies that are deemed to be just among customers   Slow/Unavailable Service   Vendor  fails  to  provide  services  that  are  normally  available  or  expected  and/or  is  slow  in  servicing customers   System Pricing Failure   Vendor erroneously price listed products   Packaging Errors   Vendor fails to properly package purchased products and/or label packages correctly   Product Defects   Purchased products fail to function as they are supposed to   Out‐of‐Stock   Vendor fails to supply accurate information on the inventory levels of listed products   Hold Disasters    Vendor fails to guarantee that products waiting to be claimed by customers do not become  lost or damaged   Alteration and Repairs Failure   Vendor fails ensure that product alterations or repairs are performed in a precise and speedy  fashion   Bad Information   Vendor misinforms customers in making transactional decisions   Failure to Meet Customer Needs and Requests   12     Construct   Definition (Event in which...)   Special Order/Request Failure   Vendor fails to fulfill special or unique requests that were promised to customers   Failure to Address Admitted  Customer Error   Vendor fails to resolve problems that arise from customers’ admitted errors   Unprompted and Unsolicited Service Behaviors  Mischarging   Vendor charges customers more than necessary for product purchases   Wrongful Accusation of Customers   Vendor  wrongfully  accuses  customers  of  inappropriate  actions  and/or  places  them  under  excessive surveillance during service encounters   Failure due to Service‐Induced  Embarrassment   Vendor embarrasses customers due to insensitivity or mistakes during service encounters   Attention Failures   Vendor fails to pay sufficient attention to customers during service encounters   Both Bitner et al.’s (1990, 1994) and Kelley et al.’s (1993) typologies of service failures suffer from the drawback of being derived from offline failure incidents and may not adequately denote issues pertinent to e-commerce transactions. Through in-depth interviews conducted with 30 individuals with prior experiences of e-service failures before surveying another 295 online shoppers, Holloway and Beatty (2003) proposed an alternate typology that comprises six categories of e-service failures (i.e., delivery problems, website design problems, customer service problems, payment problems, security problems, and miscellaneous). Each of these e-service failure categories in turn contains constituent dimensions that reflect failure events prevalent within e-commerce transactions (see Table 2.3). Table 2.3: Typology of Online Service Failures [as adapted from Holloway and Beatty (2003)] Construct   Definition (Event in which...)   Delivery Problems  Purchase Arrived Later than  Promised   E‐merchant is late in delivering purchased products to customers   Purchase Never Delivered   E‐merchant fails to deliver purchased products to customers   Wrong Item Delivered   E‐merchant delivers products that are different from what were purchased   Wrong Size Product Delivered   E‐merchant delivers products with different specifications from what were purchased   Purchase Damaged During  Delivery   E‐merchant fails to properly package purchased products to avoid damage during delivery   Website Design Problems  Navigational Problems at Site   E‐merchant fails to offer easy accessibility to service content offered   Product Poorly Presented at Site   E‐merchant fails to supply relevant information on product specifications   Insufficient Information  Provided at Site   E‐merchant fails to supply sufficient information on transactional activities   13     Construct   Definition (Event in which...)   Products Incorrectly Listed at  Site as in Stock   E‐merchant fails to supply accurate information on the inventory levels of listed products   Incorrect Information Provided  at Site   E‐merchant  fails  to  supply  correct  information  that  aid  customers  in  making  transactional  decisions   Customer Service Problems  Poor Customer Service Support   E‐merchant fails to meet customers’ service expectations when performing online transactions   Poor Communication with the  Company   E‐merchant fails to provide communication channels for customers to seek assistance   Unfair Return Policies   E‐merchant compels customers to return purchased products under unjust terms   Unclear Return Policies   E‐merchant fails to supply unambiguous information for returning purchased products   Payment Problems  Credit Card Overcharged   E‐merchant charges customers more than necessary for product purchases   Website Purchasing Process  Confusing   E‐merchant fails to offer a straightforward product purchasing process for customers   Difficulties Experienced While  Paying    E‐merchant fails to provide payment options desired by customers   Problems with Product Quality   Purchased products fail to function as they are supposed to   Consumer Dissatisfied with  Product Quality   Customers are disappointed with the way purchased products function   Security Problems  Credit Card Fraud   E‐merchant charges customers for unauthorized purchases   Misrepresented Merchandise   E‐merchant misinforms customers into purchasing products with unlisted specifications   Email Address Released to E‐ Marketers   E‐merchant  releases  customers’  disclosed  email  addresses  to  e‐marketers  without  proper  authorization   Miscellaneous  Failure to Address Unintentional  Customer Mistakes   E‐merchant  fails  to  resolve  problems  that  arise  out  of  unintentional  mistakes  on  the  part  of  customers   Retailer Charged Some  Customers More than Others   E‐merchant charges certain customers more than others for purchasing exact same products   Lack of Personalized  Information at Site   E‐merchant fails to tailor transactional information to meet customers’ requirements   Contemporary frameworks of service failures (i.e., Bitner et al. 1990, 1994; Holloway and Beatty 2003; Kelley et al. 1993) are inappropriate in capturing the spectrum of failure events unique to e-commerce transactions for three reasons. First, Bitner et al.’s (1990, 1994) and Kelley et al.’s (1993) typologies emphasize physical retail and encompass events for which the probability of occurrence is almost negligible in e-commerce settings (e.g., ‘failure in dealing with drunken customers’ and ‘wrongful accusation of customers’). Second, even for Holloway and Beatty (2003) 14     whose typology is tailored to e-commerce transactions, the blend of both service and non-service failure dimensions have made it exceedingly difficult to pinpoint actionable technological levers that can be exploited by e-merchants in improving the design of e-commerce websites (Benbasat and Zmud 2003). Though several failure dimensions in Holloway and Beatty’s (2003) typology pertain to technological flaws that compromise the service standards of e-commerce websites (e.g., ‘insufficient information and navigational problems’), others relate to troubles with business practices (e.g., ‘email address released to e-marketers’), product quality (e.g., ‘problems with product quality’) or purchase delivery (e.g., ‘purchase never delivered’), which go beyond web design technicalities. The same can be said for Bitner et al. (1990, 1994) and Kelley et al. (1993) in that many failure dimensions in both typologies are unsuited as technological prescriptions. Finally, as the three frameworks originate from a grounded methodological approach that relies on the inductive categorization of failure events, the resultant failure dimensions lack sufficient theoretical grounding such that one cannot offer explanations accounting for their manifestation. To overcome the limitations of existing frameworks, e-service and system success literatures are synthesized to advance a typology of e-service failure that delineates failure events for e-commerce transactions into three main categories, namely informational failures, functional failures and system failures.  2.2  A System-Oriented Typology of E-Service Failures Within service literature, scholars have noted that the consumption of an e-service must go  beyond outcome consumption to include process consumption because the functional processes leading to service fulfillment are transparent to the consumer (Grönroos 1998; Grönroos et al. 2000). For this reason, services have often been conceptualized as a mix of content and delivery elements (see Baker and Lamb 1993; Grönroos et al. 2000; Rust and Oliver 1994). Whereas service content is concerned with what consumers actually receive from the service encounter, service delivery relates to the manner by which customers interact with the service. Swartz and Brown (1989) maintained that  15     any theorization of services must include considerations for what services are offered as well as how these services are being offered. There is ample conceptual and empirical justification for drawing an identical distinction within e-commerce transactional environments. Without direct interaction with human service providers, Grönroos et al. (2000) argued that e-commerce websites must be “functionally advanced enough [i.e., effective service content] and technically easy to operate [i.e., efficient service delivery] by the customer so that he or she can get access to the service package” (p. 248). While an ecommerce website may offer effective content functions that assist consumers in purchasing desired products or services, accessibility to and interactivity with these functions are reliant on the website’s ability to harness the web medium for efficient delivery. This is a crucial distinction, yet one that has been infrequently made in theorizing e-services (Tan et al. 2010). Moreover, service content deficiencies result in negative consequences that are independent of those emerging from service delivery inadequacies. Whereas ineffective service content reduces consumers’ likelihood of obtaining favorable outcomes from transactional activities, inefficient service delivery amplifies the difficulty of performing such activities (Van Riel et al. 2001). Empirically, the significance of distinguishing between content and delivery in conceptualizing eservices has also been verified for both e-commerce (Cenfetelli et al. 2008) and e-government (Tan et al. 2010) contexts. Findings from these studies claim that regardless of how accessible or interactive service content may be, it serves little purpose if it does not satisfy consumers’ transactional requirements. Conversely, offering superior service content is pointless if it is not accessible to consumers through efficient delivery (Cenfetelli et al. 2008; Tan et al. 2010). From a design perspective, the distinction between content and delivery is paramount to the identification of actionable design principles that are informative and purposeful. Cenfetelli et al. (2008) and Tan et al. (2010) demonstrated that differentiating between content and delivery for e-services is vital for identifying design guidelines that entail both web features which assist consumers in obtaining 16     desirable transactional outcomes and medium characteristics which determine consumers’ accessibility to and interactivity with such features. Service content and delivery resonate with the notions of informational and system attributes within system success literature (see DeLone and McLean 1992, 2003; Wixom and Todd 2005). Since the seminal work of DeLone and McLean (1992) in which they positioned informational and system attributes as the cornerstone of system success, numerous empirical studies have attested to their significance in influencing users’ appraisal of technological systems (see DeLone and McLean 2003 for a comprehensive review). Whereas informational attributes characterize the value of information generated by a technological system (Wang and Strong 1996), system attributes are about the system’s technical performance (Hamilton and Chervany 1981). While system attributes are synonymous with the delivery aspects of services (Collier and Bienstock 2006; Fassnacht and Koese 2006), informational attributes constitute just one facet of service content because the functional capabilities of technological systems are largely neglected as distinct but complementary elements of service content (Janda et al. 2002). The concept of functionality has been employed to describe web-enabled functions that create value for consumers throughout the acquisition process of products and/or services (Cenfetelli et al. 2008; Lightner 2004). Functionalities are distinguishable from informational attributes in that they reflect service applications that generate and tailor transactional information to fit the requirements of individual consumers (Cenfetelli et al. 2008; Etezadi-Amoli and Farhoomand 1996; Janda et al. 2002; Lightner 2004). For instance, payment functionalities produce confirmation receipts verifying order information whereas tracking functionalities yield delivery details on the current whereabouts of purchased products. We therefore further differentiate between informational and functional attributes of e-services as constituent components making up the broader notion of service content and postulate that failures for e-commerce websites may be delineated into those associated with the informational,  17     functional, or system aspects of e-services 1 . Delineating e-service failures into informational, functional and system components echoes the work of Grover and Benbasat (2007) who, in inductively classifying 104 potentially unfavorable e-commerce events, suggested that consumer risk in online transactions is associated with information misuse, transactional functionality efficiency and system performance. The remainder of this chapter will examine informational, functional, and system failures in greater detail to arrive at viable working definitions and identify constituent dimensions belonging to each of these categories. Table 2.4 summarizes our proposed typology of e-service failure that distinguishes among informational, functional and system attributes; some of which overlap with those uncovered by Holloway and Beatty (2003) (e.g., inaccurate information, non-navigability and insecurity).  We will later outline an empirical study to substantiate our typology and validate its superiority over prior frameworks in the classification of e-service failure events.                                                               1  While the notion of functionality appears to overlap with the concept of service quality, it is not the case. As aptly surmised by Cenfetelli et al. (2008), functionalities represent the extent to which technological artifacts exist on e-commerce websites to fulfill consumers’ service expectations whereas service quality captures consumers’ overall evaluation of how well those functionalities are delivered. 18     Table 2.4: Proposed E-Service Failure Typology Construct   Definition (Event in which…)   Informational Failure  Inaccurate Information   Information  provided  on  an  e‐commerce  website  contains  errors  that  misinform  consumers  in  making  transactional decisions   Incomplete Information   Information provided on an e‐commerce website is insufficient to aid consumers in making transactional  decisions   Irrelevant Information   Information provided on an e‐commerce website cannot be utilized by consumers in making transactional  decisions   Untimely Information   Information  provided  on  an  e‐commerce  website  is  not  updated  to  support  consumers  in  making  transactional decisions   Functional Failure  Needs Recognition  Failure   Functionalities of an e‐commerce website are incapable of assisting consumers to formulate their needs  and preferences for products and/or services   Alternatives  Identification Failure   Functionalities of an e‐commerce website are incapable of assisting consumers to gather information on  and source for interested products and/or services   Alternatives Evaluation  Failure   Functionalities  of  an  e‐commerce  website  are  incapable  of  assisting  consumers  to  draw  comparisons  among interested products and/or services   Acquisition Failure   Functionalities of an e‐commerce website are incapable of assisting consumers to place orders for desired  products and/or services   Post‐Purchase Failure   Functionalities  of  an  e‐commerce  website  are  incapable  of  assisting  consumers  to:  (1)  obtain  purchased  products and/or services; (2) solicit advice on ways to maximize the utility of purchased products and/or  services, and; (3) dispose of unwanted products and/or services.   System Failure  Inaccessibility   Services of an e‐commerce website are not readily accessible   Non‐Adaptability   Services of an e‐commerce website are unable to accommodate diverse content and usage patterns   Non‐Navigability   Services of an e‐commerce website are difficult to navigate   Delay   Services of an e‐commerce website are inordinately slow in access   Insecurity   Services of an e‐commerce website are not safeguarded against access by unauthorized individuals   2.2.1  Informational Failures The saliency of informational attributes in determining system output is well documented  (DeLone and McLean 2003; Wixom and Todd 2005). Past studies have linked informational attributes to a host of positive benefits for task accomplishments such as workplace performance (Goodhue and Thompson 1995), productivity enhancements (Northrop et al. 1990), decisional quality (Wixom and Watson 2001), and system satisfaction (Etezadi-Amoli and Farhoomand 1996). There is an abundance of empirical evidence within e-service (see Appendix A) and system success (DeLone and McLean 2003) literatures that alludes to the importance of informational attributes in directing system outcomes. Similarly, Holloway and Beatty (2003) discovered that informational failures (e.g., 19     incorrect listing of out-of-stock items) capture a substantial fraction of technological problems that consumers associate with e-commerce websites. We hence postulate that informational failure constitutes a major deficiency of e-commerce websites and that it occurs whenever information provided on an e-commerce website is incapable of guiding consumers in the accomplishment of their transactional activities. Further, even though there exists disagreements among scholars over a representational list of preferred informational attributes for any given technological system (e.g., Wand and Wang 1996; Wang and Strong 1996), there is general consensus that accuracy, completeness, relevance, and timeliness are definitive of information quality (DeLone and McLean 2003; Wixom and Todd 2005). We hence posit that informational failures on e-commerce websites are caused by inaccurate, incomplete, irrelevant, and/or untimely transactional information.  2.2.2  Functional Failures Homburg et al. (2002) observed that the breadth and depth of service functionalities shape  consumers’ shopping experience by providing constant support throughout the entire transactional process. With e-commerce websites acting as the focal interface between consumers and e-merchants, the provision of timely assistance from pre- to post-transactional stages can only be realized through web-enabled functionalities, especially in nurturing a personalized customer service experience (Cenfetelli et al. 2008; Lightner 2004; Surjadjaja et al. 2003; Tan et al. 2010). Within service failure literature, Holloway and Beatty (2003) have identified ordering and payment difficulties as pervasive forms of functional failure experienced by consumers who transact via e-commerce websites. We hence define functional failure to have occurred whenever functionalities provided on an ecommerce website are incapable of supporting consumers in the accomplishment of their transactional activities. Further, in line with Jacoby’s (1998) delineation of consumer decision making into five sequential activities (i.e., needs recognition, alternatives identification, alternatives evaluation, product acquisition, and post-purchase), we contend that functional failure may occur for any of these activities. Also, while we have argued for the necessity of offering functionalities on e20     commerce websites to cater to the aforementioned core activities of consumer decision making, we acknowledge that these activities do not necessarily have to occur in sequence during online transactions. For instance, if a consumer has already decided on purchasing a specific product prior to visiting an e-commerce website, product acquisition functionalities would be rendered more relevant than those catering to needs recognition, alternatives identification and alternatives evaluation. Needs Recognition Failure: Needs recognition failure occurs whenever e-commerce websites fail to provide functionalities that assist consumers in making sense of their needs and preferences (Cenfetelli et al. 2008; Ives and Learmonth 1984; Lightner 2004). Because face-to-face communication and clarification are restricted for online transactions, the provision of functionalities supporting needs recognition serves three basic purposes: (1) to educate consumers about a product and/or service offered on an e-commerce website; (2) to get these consumers to realize how offerings from the website differ from that of its competitors, and; (3) to aid them in selecting the product and/or service best suited to their requirements (Piccoli et al. 2001). The absence of needs recognition functionalities would hence leave consumers in a state that is not much better off than when they first started, still lost as to how their transactional needs can be met. Alternatives Identification Failure: Once a consumer has narrowed down (e.g., via recommendation agents) to a smaller subset of products he/she is interested in, he/she may want to search for additional information on the performance of each product in the consideration set (Furse et al., 1984) or on possible locations for acquiring these products (Piccoli et al. 2001). Failure to provide functionalities that consolidate informational sources for easy referencing would compound onto consumers’ difficulty in identifying relevant alternatives. Alternatives Evaluation Failure: Individuals typically employ a two-stage cognitive evaluation process in making decisions with complex parameters (Jedetski et al. 2002; Xiao and Benbasat 2007). Whereas the first stage involves the refinement and transformation of consumers’ preferences into a subset of promising alternatives (i.e., needs recognition) (Xiao and Benbasat 2007), 21     it is only through the second stage of in-depth comparisons among generated alternatives that consumers eventually come to a purchase decision (Jedetski et al. 2002). For an in-depth comparison of alternatives to be meaningful, e-commerce websites must provide functionalities (e.g., Comparison Matrix) that organize evaluative criteria of product and/or service alternatives in an intuitive and easily comprehensible manner (Haubl and Trifts 2000; Jedetski et al. 2002). Otherwise, consumers’ decision making process would be impeded. Acquisition Failure: Acquisition pertains to functionalities that facilitate the completion of online transactions (Ives and Learmonth 1984; Ives and Mason 1990). Piccoli et al. (2001) stressed that technological functionalities can simplify the acquisition process by retaining consumer information (e.g., shipping and payment information), which can be reused for subsequent orders. Acquisition failures (e.g., missing ordering and payment functions) are hence fatal to e-commerce websites, with payment errors being identified by Holloway and Beatty (2003) as a key inhibitor of online transactions. Post-Purchase Failure: Post-purchase activities are those associated with the ownership and retirement of products and/or services. Whereas ownership is geared towards assisting consumers in obtaining and maximizing the utility of purchased goods, retirement is concerned with the clearance of products and/or services that have outlived their usefulness (Ives and Learmonth, 1984; Ives and Mason, 1990). Post-purchase functionalities include tracking services to monitor the status of purchase orders (Cenfetelli et al. 2008), FAQs to answer common enquiries regarding the usage of purchased products (Santos 2003; Singh 2002), return centers to facilitate the refund of defective products (Surjadjaja et al. 2003), and online auctions to support the disposal of unwanted products in a cost effective fashion (Piccoli et al. 2001). In a way, post-purchase functionalities are aimed at giving consumers ease of mind after purchases and their absence would compromise the entire transactional experience.  22     2.2.3  System Failures System quality has been demonstrated to streamline task performance through enhanced  adaptability to changing requirements (Vandenbosch and Huff 1997; Wixom and Watson 2001). Within the e-commerce context, the absence of key system attributes may undermine the delivery of service content for e-commerce websites and lead to unnecessary complications of the online transactional process for consumers (Cenfetelli et al. 2008). Holloway and Beatty’s (2003) categorization of e-service failures have similarly incorporated navigational difficulties as a core failure dimension. We hence define system failure to have occurred whenever service content (i.e., information and functionalities) offered by an e-commerce website is not delivered in a conducive manner that facilitates consumers in the accomplishment of their transactional activities. Adapting DeLone and McLean’s (2003) recommended metrics of system quality for successful e-commerce systems in conjunction with prevalent system attributes identified through our review of e-service literature (see Appendix A), we posit that system failures occur whenever e-service content is inaccessible, non-adaptable, non-navigable, delayed, and insecure in its delivery. Inaccessibility: As e-merchants struggle to maximize the potential of Internet technologies in overcoming physical limits during e-service delivery (Douglas et al. 2003; Janda et al. 2002; McKinney et al. 2002), diversity in the physiological capabilities of their target audience (e.g., dyslexia, visually impaired) and consumers’ adoption of divergent technologies (e.g., PCs versus Macintoshes, or Internet Explorer versus Firefox) are sometimes overlooked as fundamental elements affecting the accessibility of e-services (Shim et al. 2002). Within the domain of e-government, Tan and Benbasat (2009) reported that such inconsistencies in e-service accessibility are commonplace even among mature websites and reduce the efficiency of the Internet as an impartial delivery medium. Non-Adaptability: The strongest appeal of the Internet as an e-service delivery medium resides in its capacity to adapt and personalize transactions according to consumers’ stipulated requirements (Agarwal and Venkatesh 2002). Content personalization on an e-commerce website 23     however, is contingent on whether e-services are delivered in a manner that facilitates dynamic engagement and interaction with their target audience (Cagurati et al. 2005; Katz and Byrne 2003). Particularly, the capability of e-commerce websites to cope with diverse service content (e.g., multilingualism) and usage patterns (e.g., different conventions for data entry due to inter-country variations) plays a critical role in content personalization (Evanschitzky et al. 2004; Palmer 2002; Srinivasan 2002). Without personalizable content, non-adaptable e-services may impose unnecessary constraints on consumers’ transactional behaviors (Tan and Benbasat, 2009). Non-Navigable: The navigability of an e-commerce website governs the effort-performance expectancy of consumers (Childers et al. 2001; Tan and Benbasat 2009). The complexity of the navigational structure determines the ease by which an e-commerce website can be readily traversed by an inexperienced consumer and ultimately, affects the amount of cognitive effort that must be expended by the consumer to accomplish his/her transactional task (Korthauer and Koubek 1994). Non-navigability of e-commerce websites thus constitutes a form of e-service failure (Holloway and Beatty 2003). Delay: Response time has been shown to be a major deterrence against consumers’ adoption of e-commerce websites (Rose et al. 1999; Rose and Straub 2001; Torkzadeh and Dillon 2002). Past empirical studies have revealed an inverse relationship between response time and system users’ productivity (Barber and Lucas, 1983; Martin and Corl, 1986): long delays have been shown to lead to complaints of frustration (Doherty and Kelisky, 1979), feelings of dissatisfaction (Lee and MacGregor, 1985), sense of disorientation (Sears et al., 2000) and eventual abandonment (Nah, 2002). In light of the overwhelming evidence on response time being predictive of e-service quality (see Appendix A), delay should be recognized as a form of failure in e-commerce transactions. Insecurity: Security has received widespread attention in e-commerce literature (Wang 2002). Security in e-commerce websites pertains to protective measures to safeguard disclosed transactional information from unsanctioned or illegal intrusions by third parties (Collier and Bienstock 2003, 2006; 24     Etezadi-Amoli and Farhoomand 1996; Wang 2002). Holloway and Beatty (2003) also classified insecurity as a high priority failure for e-services. 2.2.4  A System-Oriented Typology of E-Service Failures Appendix A maps informational, functional, and system attributes in our typology (see Table  2.4) to previous studies that have advocated similar attributes as being desirable qualities of e-services. It can be deduced from Appendix A that the informational, functional, and system failures in our typology are representative of potentially problematic areas of concern for e-services. Clearly, our typology differs from contemporary frameworks in three ways. First, unlike Bitner et al. (1990, 1994) and Kelley et al. (1993), we concentrate primarily on the identification of a generic and representative collection of e-service failure causes that confront consumers in conducting e-commerce transactions. This leaves out service failures that rarely occur in e-commerce transactions such as the disastrous handling of bothersome consumers or the embarrassment and wrongful accusation of customers. Second, while we admit that problems related to business practices (e.g., ‘unfair return policies’), product quality (e.g., ‘consumer dissatisfied with product quality’), and purchase delivery (e.g., ‘purchase damaged during delivery’) are also part of e-commerce transactions, we opt to exclude such errors from our typology and focus exclusively on transactional failures that are within e-merchants’ abilities to rectify through improvements to web interface design. Further, in place of a wider variety of failure categories and dimensions depicted in past typologies, we choose to retain a precise but meaningful set of higher-order e-service failure categories (i.e., informational, functional, and system failures), each comprising a set of lower-order dimensions of e-service inadequacies. We believe that such a framework can yield targeted and purposeful design prescriptions for service enhancements on e-commerce websites. Finally, our typology circumvents the theoretical limitation of past frameworks by deductively deriving our e-service failure categories and dimensions from the synthesis of service and system success research streams. This forms a stronger conceptual foundation from which to rationalize about the causes of e-service failures. 25     2.3  Summary Grounded in an expectancy perspective, this chapter proposes a working definition of e-  service failure for e-commerce transactional environments. Further, it reviews existing typologies of service failures and discovers that the application of these frameworks to the classification of eservice failures is inherently challenging. Consequently, building on e-service and system success literatures, this chapter advances a typology of e-service failure that distinguishes among informational, functional and system aspects of an e-service encounter. The next chapter, Chapter 3, describes the first study that is designed to assess the comprehensiveness of our proposed typology in classifying e-service failure events in reality.  26     C HAPTER 3 – A N E XPLORATORY S TUDY OF E-S ERVICE F AILURE C AUSES (1 S T S TUDY ) Because my proposed e-service failure typology represents a loose adaption of web attributes highlighted within extant literature on e-service and system success, its validity is dependent on its ability to outperform contemporary frameworks of service failures (i.e., Bitner et al. 1990, 1994; Holloway and Beatty 2003; Kelley et al. 1993) in classifying failure events that transpired during actual e-commerce transactions. The first study therefore employs the Critical Incident Technique (CIT) for data collection to compare and contrast our proposed e-service failure typology against contemporary frameworks in order to ascertain its suitability in classifying incidences of e-commerce transactional failures. The CIT has been applied to the investigation of a variety of service and system-related phenomena including customer-vendor interactions (Nyquist and Booms 1987), service encounters (Bitner et al. 1990, 1994; Kelley et al. 1993), and virtual teams (Thomas and Bostrom 2010). The CIT comprises four sequential steps: (1) deciding the objectives of the activity; (2) formulating plans and agendas for the collection of critical incidents; (3) gathering and analyzing data, and; (4) interpreting empirical findings (Flanagan 1954). Reliability of the CIT has been verified in past information systems studies (e.g., Majchrzak et al. 2005; Thomas and Bostrom 2010). Keaveney (1995) noted that the CIT is especially appropriate when research objectives are targeted at both theory development and pragmatic relevance. In this study, a critical incident is conceived as any event, combination of events, or series of events between a consumer and an ecommerce website that caused the former to experience failure in the usage of e-services. Critical incidents were defined broadly to cast a wide net to ensure an adequate coverage of probable technological deficiencies across e-commerce websites. Respondents could report incidents along any stage of the online transactional process or on any aspect of the website.  27     3.1  Questionnaire Development Given the predominantly Internet-savvy target audience of respondents with previous e-  service failure experience, we opted to solicit critical incidents of e-service failures through digital means (Boyer et al. 2002; Stanton and Rogelberg 2001). A qualitative electronic survey questionnaire was carefully crafted for data collection. The questionnaire begins with a definition of e-service failure and some common examples to familiarize respondents with the phenomenon of interest. Then, respondents are requested to indicate their frequency of performing online transactions and whether they have experienced some form of e-service failure within the last six months. This single filtering question is devised to eliminate respondents with no recent experience of e-service failure. Next, respondents are instructed to either choose from a variety of e-merchants or provide a description of the type of e-commerce website on which they have encountered the e-service failure. In line with Keaveney’s (1995) advice, such a question offers a certain degree of structure to the type of ecommerce websites for which e-service failures may occur, without necessarily limiting respondents to the pre-specified list. Respondents are then asked to state their purpose for visiting the e-commerce website: Please  describe  in  detail your purpose for  visiting the website  for  which  you  have  experienced  the  online  service failure   This question on the purpose of the visit is to discern respondents’ transactional objectives because we do not presume that consumers transact online for the sole purpose of maximizing utility. These transactional objectives offer valuable background information on the situational context within which the e-service failure occurs. The next question touches on the actual phenomenon of interest by requesting respondents to elaborate on the e-service failure experienced, with additional probes for details. Because the development of our theory is confined to online transactional failures in order to generate  28     prescriptions for web interface design, the probes are deliberately phrased to emphasize the recollection of problems related to technological features on e-commerce websites: Please  describe  in  detail the  online  transaction  you  were  conducting  when  you  experienced  the  online  service failure as well as the events leading to this failure. You should elaborate on the following:    1.  What  you  have  managed  to  accomplished  on  the  website prior  to  the  occurrence  of  the  online  service failure     2.  Details  of  the  online  service  failure  experienced  [Please  be  specific  on  the  website  feature(s)  involved and why you perceive these feature(s) to have failed]   As respondents may have been exposed to multiple episodes of e-service failures, the same format of questioning was repeated twice to prompt each respondent to recall a minimum of one and a maximum of three critical incidents. A diagrammatic flow of the online survey questionnaire is depicted in Figure 3.1. In answering the questionnaire, respondents were never told to analyze why the failure incident(s) occurred. Rather, they were expected to merely narrate events that had transpired— something people do quite effortlessly (Bitner et al. 1990; Nyquist and Booms 1987). A pre-test was conducted with a convenient sample of faculty members and graduate students from a large North American university. No major issues surfaced during the pre-test.  29     Figure 3.1: Diagrammatic Flow of Online Survey Questionnaire Definition of e-service failures with common examples  Type of e-commerce website for e-service failure  Purpose for visiting ecommerce website Frequency of performing online transactions Description of e-service failure  prior failure experience? N  3.2  Y  another failure? (x 2)  Y  N  Data Collection Invitations were emailed to members belonging to a nationwide panel of e-business  consumers from a commercialized marketing research firm. In exchange for their participation, the marketing research firm awarded participating panel members points that can be redeemed for prizes. Due to the possibility of disabled e-mail accounts, spam filtering, or other forms of account blockages, no mechanism was available to gauge the diffusion rate of the invitation to all panelists. Following Cenfetelli et al. (2008), we reviewed the computer logs of the web server on which the electronic survey was hosted. The server logs recorded 991 visits to the online questionnaire, some of which may not be unique. Using the filtering question inserted at the start of the questionnaire to sift out respondents who have experienced e-service failure(s), 233 out of the 991 visitors to the survey satisfied our sampling criteria, thereby yielding a conservative estimate of 23.5% response rate. Flanagan (1954, p. 340) suggested that “if full and precise details are given, it can usually be assumed that this information is accurate. Vague reports suggest that the incident is not well remembered and that some of the data may be incorrect”. Accordingly, responses from 22 respondents were judged to be ambiguous and deleted, giving a final sample of 211 respondents for 30     data analysis. Table 3.1 summarizes the descriptive statistics for the sample together with a breakdown of the number of e-service failure incidents reported by various demographic groups. Paired t-tests between our sample and those documented in Cenfetelli et al.’s (2008) survey of 1,235 consumers on the service quality of e-commerce websites reveal no significant difference in demographic distribution (i.e., t(14) = -0.118, p = .907). Table 3.1: Descriptive Statistics for Online Survey Respondents [Sample N = 211] E‐Service Failure  Demographic Characteristic   No. of  Respondents [%]   Comparison    Frequency of E‐Commerce  Website Visitations   [Total Incidents = 316]  1   2   3   Total   Gender  Male   132 [62.56%]   34%   At least once per 2 weeks   90   20   22   196   Female   79 [37.44%]   66%   At least once per 2 weeks   53   11   15   120   Age 19‐29   32 [15.16%]   10%   At least once per 2 weeks   26   1   5   43   Age 30‐49   86 [40.76%]   60%   At least once per 2 weeks   62   10   14   124   Age 50‐64   71 [33.65%]   28%   At least once per 2 weeks   42   12   17   117   Age 65+   20 [9.48%]   2%   At least once per month   11   8   1   30   Unwilling to disclose   2 [0.01%]   0%   At least once per week   2   0   0   2   College education or higher   160 [75.83%]   87%   At least once per 2 weeks   103   25   32   249   Less than college education   49 [23.22%]   13%   At least once per 2 weeks   38   6   5   65   2 [0.01%]   0%   At least once per month   2   0   0   2   $0‐$29,999   68 [32.23%]   15%   At least once per month   48   10   10   98   $30,000‐$50,000   50 [23.70%]   24%   At least once per 2 weeks   34   8   8   74   $50,000‐$75,000   39 [18.48%]   28%   At least once per 2 weeks   28   6   5   55   $75,000+   44 [20.85%]   33%   At least once per 2 weeks   29   4   11   70   10 [0.05%]   0%   At least once per week   4   3   3   19   Age   Educational Level   Unwilling to disclose  Income   Unwilling to disclose   Cenfetelli et al. (2008)     A total of 316 e-service failure incidents were reported by the respondents (see Table 3.1). Of these 316 incidents, 58 (or 18%) contain descriptions of two distinct e-service failure episodes within a single recall and are therefore split into separate incidents to prevent confounds from surfacing during data analysis. An example of such recalls is this: 31     Incident:  “I  wanted  to  buy  a  plane  ticket  online.  I  was  able  to  choose  the  destination,  date,  and  started  placing the order, then to realize later that: (1) the price changed during the time I was completing the order,  and (2) the website doesn't accept my credit card.”   To avoid the loss of valuable contextual information, we segregate the aforementioned description into two separate incidents in the following manner: Incident A: “I wanted to buy a plane ticket online. I was able to choose the destination, date, and started placing  the order, then to realize later that the price changed during the time I was completing the order.”  Incident B: “I wanted to buy a plane ticket online. I was able to choose the destination, date, and started placing  the order, then to realize later that the website doesn't accept my credit card.”   In splitting the 58 incidents, we arrive at a final sample of 374 data points for analysis.  3.3  Data Analysis Content analysis was carried out to sort the sample of 374 incidents into each of the four  typologies outlined in Chapter 2. Intra- and inter-judge reliabilities were compared to ascertain the validity of our proposed e-service failure typology in characterizing e-commerce transactional failures relative to the other frameworks. Noting past recommendations for CIT studies (see Boyatzis 1998; Butterfield et al. 2005; Keaveney 1995), we adhered to a set of content analytical procedures that have been developed exclusively for framework comparisons. A diagrammatic flow of the entire content analysis process for e-service failure incidents is depicted in Figure 3.2.  32     Figure 3.2: Diagrammatic Flow of Content Analytical Procedures for E-Service Failure Incidents Step 4: Retain 2 judges with  intra‐judge reliabilities > 0.85   374 E‐Service Failure  Incidents   Step 5: Sort on 100% of  sample for each of the four  typologies and create extra  failure dimension(s) if  necessary; 2 judges  Step 1: Sort on 20% of sample  for each of the four typologies  and create extra failure  dimension(s) if necessary; 5  judges   Step 6: Consult judges to  consolidate sorting results for  each of the four typologies;  same 2 judges   Step 2: Consult judges on  wording of original and newly  created failure dimension(s) if  any and revise accordingly;  same 5 judges   Step 7: Compare and contrast  the four typologies based on  intra‐ and inter‐judge  reliabilities   Step 3: Sort on same 20% of  sample; same 5 judges   N  intra‐judge  reliability> .70?   Y  Step 8: Follow‐up interviews  with judges to attain 100%  consensus on placement for  proposed typology; same 2  judges    3.3.1  Analytical Procedures To begin, five judges were recruited from students enrolled in an MBA program in a large  North American university to refine the wording of failure dimensions in each of the four typologies. Because Bitner et al.’s (1990, 1994) and Kelley et al.’s (1993) typologies originate from offline service failures, this step is essential in ensuring that the definitions of failure dimensions in these frameworks are amenable to e-commerce transactions. Further, as part of the program, these judges have taken a course on e-businesses and such familiarity with the e-commerce context would be advantageous in assessing the appropriateness of failure definitions for the various frameworks. 77 (or ~20%) incidents were randomly extracted from the sample and assigned to the five judges to be sorted into each of the four typologies. The entire sorting exercise is semi-structured. For each  33     typology, judges were told to place each incident into one of the pre-existing failure dimensions or to create an extra dimension if they were unsure of its placement. Upon the completion of the sorting exercise, the judges were consulted on the phrasing of the failure dimensions and modifications were made whenever necessary. Then, the judges were again presented with the same 77 incidents to be sorted into the refined failure dimensions. Intra-judge reliabilities (i.e., extent to which a single judge assigns an identical incident to the same failure dimension in both classification exercises) were computed. The second sorting exercise yielded intrajudge reliability values that exceed the recommended threshold of 0.70 (Boyatzis 1998) for each typology. This signifies consistency in the judges’ interpretation of failure dimensions and testifies to the clarity in the phrasing of the dimensions. Of the five judges, two with intra-judge reliability scores above 0.85 were retained for the remainder of the content analysis process. Inter-judge reliabilities were not factored into the selection of the two judges because we do not wish to bias subsequent sorting activities towards a particular framework. Next, sorting was performed for the entire sample of 374 incidents, which included the 77 incidents from before. The sorting exercise abided by the same protocol described previously. Once the sorting was completed, the two judges were questioned on the way they classified the incidents, with particular attention on the newly created failure dimensions for consolidation purposes. At this time, the judges were permitted to reassign incidents that they believed were wrongly placed, but they were not obliged to do so. The reason for not forcing a greater level of agreement among judges is to ensure that the judges are most comfortable with their current classification of the incidents. This provides a basis for comparing the viability of each typology in catering to the diversity of e-service failures that may arise during e-commerce transactions. Both intra- and inter-judge (i.e., extent to which different judges assign an identical incident to the same failure dimension) reliabilities were calculated for each typology based on the eventual classification of the incidents.  34     3.3.2  Findings from Framework Comparison Table 3.2 summarizes intra- and inter-judge reliabilities for each typology. Appendix B  contains a detailed breakdown of the classification of incidents for failure dimensions within each typology. Appendix C employs our proposed typology as a point of reference to: (1) showcase examples of failure incidents that have been unanimously sorted into each of our failure dimensions, and; (2) illustrate how these incidents have been classified with respect to other frameworks. Table 3.2: Intra- and Inter-Judge Reliabilities of E-Service Failure Classifications Intra‐Judge Reliability  [Sample N = 77]   Inter‐Judge Reliability  [Sample N = 374]   Bitner et al.’s (1990, 1994) Typology of Service Encounter Failures   0.74   0.50   Holloway and Beatty’s (2003) Typology of Online Service Failures   0.76   0.59   Kelley et al.’s (1993) Typology of Retail Failures   0.81   0.70   Proposed E‐Service Failure Typology    0.89   0.88   Framework   As can be seen from Tables B-1 and B-4 in Appendix B, the sorting exercises generated one additional dimension (i.e., ‘informational failure’) for Bitner et al.’s (1990, 1994) typology to cater to informational problems and three other dimensions (i.e., ‘mischarging’, ‘product delivery problems’ and ‘unresponsive to customer enquiries’) for our proposed typology to accommodate nontransaction-oriented errors. We grouped the three newly created failure dimensions under the higherorder category of ‘Non-Transaction-Oriented Failures’. Due to the interpretive nature of the content analytical procedures, we cannot claim that the results in Table 3.2, Appendixes B and C are definitive in proving the suitability of each of the four typologies in representing incidences of e-service failures. Nevertheless, several inferences can be drawn with regards to the pros and cons of applying each typology to the appreciation and classification of e-service failures. First, as can be concluded from the intra-reliability values in Table 3.2, our proposed typology fares better than other frameworks at distinguishing the underlying cause of one failure  35     incident from another. The same individual is able to match a failure incident with its cause more consistently based on our typology than with others. Second, the inter-reliability values in Table 3.2 demonstrate that our proposed typology enables greater consensus among individuals on the causes of seemingly dissimilar failure incidents. This is also substantiated through Table B-4 in Appendix B; the inter-judge reliabilities of all deductively-derived failure dimensions in our proposed typology exceed the recommended threshold of 0.70 (Boyatzis 1998) and are generally much higher than those of failure dimensions belonging to other typologies. Appendix C further exemplifies how disagreements may surface among individuals when typologies, besides the one we proposed, are applied to the classification of e-service failures. Third, of the four typologies, only ours can boast of a relatively even spread of failure incidents across different dimensions. In contrast, Bitner et al.’s (1990, 1994), Kelley et al.’s (1993), and Holloway and Beatty’s (2003) typologies tend to have all-inclusive dimension(s) with high concentrations of failure incidents (see Tables B-1, B-2 and B-3 in Appendix B). For instance, the dimension of ‘other core service failure’ in Bitner et al’s (1990, 1994) typology absorbs close to 41% of the total number of failure incidents in the dataset due to its blanket definition. The same can be said for the ‘slow/unavailable service’ dimension in Kelley et al.’s (1993) typology and the ‘poor customer service support’ dimension in Holloway and Beatty’s (2003) typology. This in turn lends credibility to the conciseness and relative orthogonality of the failure dimensions in our proposed typology. Finally, our proposed typology is the only one with no ‘empty’ dimensions, i.e. no dimension is left without instances of e-service failure. The other three typologies have at least one ‘empty’ dimension each, as seen from Tables B-1, B-2, and B-3 in Appendix B. As a whole, our proposed typology can therefore be deemed to be reasonably parsimonious; it neither carries failure dimensions which do not coincide with actual manifestations of e-service failures (e.g., ‘failure to manage disruptive others’ in Bitner et al.’s (1990, 1994) typology), nor does it include all-encompassing but 36     equivocal dimensions for which the prescription of actionable design principles is practically impossible (e.g., ‘poor customer service support’ in Holloway and Beatty’s (2003) typology).  3.4  Discussion While research into service failure has had a long tradition in the marketing discipline, there  are comparatively few studies which delve into this phenomenon in e-commerce transactions. Building on the EDT, the first study sets out to achieve two primary objectives. First, we synthesize e-service and system success literatures to construct a novel typology that delineates e-service failures into informational, functional, and system categories, each with its own collection of constituent dimensions. Then, leveraging on the CIT to solicit descriptive accounts of failure events that transpired during actual e-commerce transactions, our typology is compared and contrasted against contemporary frameworks of service failures to establish its suitability in classifying such events. Findings from our empirical investigation indicate that our proposed e-service failure typology can be deemed to be more comprehensive than contemporary frameworks in capturing the contextual uniqueness of failure events intrinsic to e-commerce transactions. The remainder of this chapter will summarize the theoretical contributions, pragmatic implications and potential limitations of this study. 3.4.1  Implications for Research The first study makes novel contributions to extant literature on e-service failure on two  fronts. First, we construct a typology of e-service failures that caters exclusively to e-commerce transactional environments. Through the deductive identification of generic and representational failure categories common to e-commerce websites (i.e., informational, functional, and system failures), our typology is the first to offer theoretically-grounded explanations for the manifestation of different forms of e-service failures. Further, under these higher-order failure categories, we have identified lower-order constituent dimensions that accentuate website design flaws which are within e-merchants’ ability to correct. Indeed, the pertinence of these failure categories and dimensions are corroborated with empirical evidence from our investigation. Based on the content analysis of e37     service failure incidents, our typology appears to be robust in classifying and characterizing ecommerce transactional failures in comparison to contemporary frameworks. Findings indicate that all the failure dimensions of our typology (as compared to other frameworks) conform to reality and are unambiguous in meaning. Second, despite a long tradition of research into the determinants of successful information systems, there has been no scholarly attempt to leverage on the knowledge accumulated in this area when conceptualizing e-service failures. Instead, scholars like Holloway and Beatty (2003) tend to emphasize the service aspect of e-service failures without giving enough notice to the technological side of things. This study is hence ground-breaking in that it subscribes to an assimilative strategy in constructing a typology of e-service failures by giving equal prominence to both service and system success research streams. Empirical evidence testifies to the importance of synthesizing e-service and system success literatures in the construction of an e-service failure typology that is both parsimonious and representative. 3.4.2  Implications for Practice This study should be of keen interest to e-merchants for two reasons. First, our e-service  failure typology can serve as an analytical toolkit for them to conduct benchmark studies on their ecommerce websites to assess whether visitors to the websites face transactional difficulties. Because the validity of the failure categories and dimensions in our typology are ascertained from critical incidents of e-service failures that transpired recently (i.e., past six months), faulty e-commerce websites may be more pronounced than what e-merchants imagined. This may also explain the 45% rate of failure in e-commerce transactions reported by Oneupweb (2010). Second, our e-service failure typology offers actionable design prescriptions for e-merchants to improve the quality of their e-commerce websites. Even though the failure dimensions do not delve into the technicalities of e-service implementation, they do offer guidelines on the areas to watch out for in the maintenance of e-commerce websites. E-commerce websites are never static and their 38     designs evolve over time to accommodate changing customer preferences (Wind 2001). Whenever the design of an e-commerce website is having a facelift, our typology could come in handy as a checklist to pinpoint any design flaws that may deter consumers from visiting. 3.4.3  Limitations Three caveats exist with regards to this study. First, while the CIT is suitable for eliciting  practical instances of events that have transpired, the retrospective nature of the recollection implies that the events may not be remembered accurately and there is no way of verifying whether recounts are distorted to fit respondents’ mental version of events. However, we have minimized such deviations in our empirical investigation. In getting respondents to recall e-service failure incidents they have encountered only in the past six months, we aim for recent experiences in order to minimize inaccurate or incomplete descriptions due to memory loss. Further, there is an inherent advantage in employing the CIT for data collection on e-service failures; failure events are more likely to have a lasting impression on respondents due to heightened emotions (Andreassen 2001; McColl-Kennedy and Sparks 2003; Smith et al. 1999). Second, while we have taken every effort to ensure that all data points (i.e., e-service failure incidents and consequences) are self-contained and content analytical procedures are rigorous, the interpretive nature of our research may impose a certain degree of subjectivity to our findings. Despite the care taken in the selection of judges for our content analysis, we acknowledge that judges’ interpretation of e-service failure events relies on personal judgments that may vary among individuals depending on past experiences. Therefore, findings from this empirical investigation should be validated through further research, especially across other e-service contexts such as egovernment. Third, because the judges for the content analysis were recruited from MBA students who have taken a course on e-businesses, there is a possibility that these judges may have acquired a certain level of familiarity with the e-service failure dimensions in our proposed typology. While such 39     familiarity may positively skew the intra- and inter-reliability values for our typology in comparison to Bitner et al.’s (1990, 1994) and Kelley et al.’s (1993) frameworks, we have reasons to believe that this may not be the case in our study. As can be inferred from Table 3.2, the intra- and inter-reliability values for Kelley et al.’s (1993) framework is higher than those for Holloway and Beatty’s (2003) framework, which contains a larger number of technological failure dimensions. If knowledge bias— arising from judges’ familiarity with the e-business context—were to have an upward biasing effect on our empirical results, then contrary to our findings, Holloway and Beatty’s (2003) framework would have higher intra- and inter-reliability values as compared to that of Kelley et al.’s (1993). 3.4.4  Summary This chapter outlines the design and execution of an exploratory study that employs the CIT  to solicit e-service failure events from practice. These e-service failure events in turn are analyzed, via content analytical techniques, to validate our proposed typology of e-service failure (see Table 2.4). Data gathered via the CIT attests to the robustness of the typology. The next chapter, Chapter 4, defines and introduces e-service recoveries as technological features that could be implemented on ecommerce websites to moderate failure consequences.  40     C HAPTER 4 – A C OUNTERFACTUAL P ERSPECTIVE OF E-S ERVICE R ECOVERY  T HINKING  E-service failures manifest whenever consumers detect service deviations from a priori expectations. This deviation may be due to one of two reasons: (1) when customers’ expectations are untenable (e.g., trying to acquire a product with non-existent attributes), or; (2) when an e-commerce website is ill-equipped with essential e-services to fulfill consumers’ valid expectations (Holloway and Beatty, 2003). Counterfactual thinking is contrasting what is perceived to be with what might have been, which Roese (1997) termed as contrastive thinking. When an individual is in a counterfactual frame of mind, he/she may (cognitively) alter parts of an event in assessing its consequence or outcome (Roese and Olson 1995). Folger and Cropanzano (2001) argued that counterfactual thinking is often subconsciously embraced by individuals to assess the seriousness of an offense. When applied to situations of service failures, counterfactual thinking tells us that a consumer will construe a sequence of events that vary from what actually took place (i.e. events which run contrary to reality) (McColl-Kennedy and Sparks, 2003). That is, the consumer is engaged in a contrastive evaluation process that gauges how things might have been if events had transpired differently. For instance, a consumer who experienced payment difficulties on an e-commerce website may reflect: “If only the payment functions had worked properly, I would have completed my transaction and obtained the product of my choice”. Thus, in evaluating any service failure event, Folger and Cropanzano (2001) claimed that a consumer engages in three contrastive frames of mind: what could have happened (e.g., the e-commerce website could have ensured that payment functions work properly), what should have happened (e.g., the e-commerce website should have provided alternative payment methods), and how it would have felt had alternative actions been taken (e.g., I would have been satisfied with the e-commerce website if either of the two measures had been implemented) (see also Teas, 1993). 41     Because e-service failures are typically accompanied by unwanted consequences (e.g., money spent, time or effort wasted) that leave the consumer feeling worse off than when he/she first started (Bitner et al., 1990), we define e-service recovery as the extent to which recovery technologies offered by an e-commerce website are able to moderate negative consequence(s) experienced by consumers in the event of an e-service failure. In this sense, e-service recoveries cater to the ‘should’ frame of mind in counterfactual thinking. The next section reviews contemporary frameworks of service recovery to clarify our choice of Smith et al.’s (1999) typology of service recovery modes as the theoretical framework for classifying recovery technologies on e-commerce websites.  4.1  A Proposed Typology of E-Service Recovery The first typology of service recovery was proposed by Kelley et al. (1993) who, in gathering  661 critical incidents of service failures for which 335 led to ‘good’ recoveries and 326 resulted in ‘poor’ recoveries, uncovered 12 recovery strategies that are commonly employed by vendors in addressing failure occurrences (see Table 4.1). Of the 12 recovery strategies, Kelley et al. (1993) observed that certain strategies may further exacerbate the situation. While extensive, Kelley et al.’s (1993) typology of service recovery suffers from the same fate as contemporary frameworks of service failures in that the CIT does not generate theoretical justification for the existence of inductively derived dimensions. Indeed, Kelley et al. (1993) do not present sufficient reasoning to justify why the recovery strategies prescribed in their typology are practiced in reality nor are they able to gauge the effectiveness of each recovery strategy in pacifying consumers in the event of service failures. Departing from the inductive technique to creating typologies of service recovery, Tax et al. (1998) synthesized past studies in marketing and social psychology to create a classification scheme of service recovery tactics that are founded on theories of justice. Consistent with previous theorizations of justice as a multi-dimensional concept (e.g., Colquitt et al., 2001; Masterson et al., 42     2000), Tax et al. (1998) distinguished among distributive, procedural and interactional justice in constructing their typology of service recovery. From their review of extant literature on complaint handling, Tax et al. (1998) arrived at 13 dimensions of service recovery that correspond to the three forms of justice. This was later expanded, via data gathered from the qualitative survey of 257 respondents, to 20 dimensions (see Table 4.1). Yet, the service recovery dimensions advocated in Tax et al.’s (1998) typology are inappropriate as actionable design principles for e-commerce websites despite the deductive manner of their derivation. First, it is difficult to transform the huge quantity of service recovery dimensions in Tax et al.’s (1998) typology into design guidelines that can be acted upon by e-merchants. Further, several dimensions deal with recovery measures that necessitate human touch (e.g., politeness and honesty) and would be challenging to replicate on e-commerce websites. Subscribing to the Social Exchange Theory (SET), Smith et al. (1999) built on the work of Tax et al. (1998) in advancing a typology of service recovery, which they empirically verified, through a mixed-design experiment, to be relatively inclusive in addressing a representative sample of service failures. Closely related to notions of justice, the SET has been applied pervasively by marketing scholars in articulating service recovery measures (e.g., Homans 1961; Walster and Berscheid, 1978; Walster et al., 1973). The SET accentuates mutual reciprocity among participants as the underlying mechanism of governance in any relational network, i.e. it involves the dynamic exchange of “diffuse, ill-defined obligations in terms of the nature, value, and timing of the benefits rendered and received by the parties” (Organ, 1990, p. 63). Individuals participating in social exchanges must therefore have “faith in the cooperative intentions of the other [parties] with whom they are engaging due to the lack of a mechanism that could enforce an equal exchange” (Gefen and Ridings, 2002, p. 51; Rosenbaum and Massiah, 2007). Conceptually, the SET acts as an appropriate theoretical lens from which to explicate the costs and benefits borne by participants within transactional relationships (Gefen and Ridings, 2002). 43     Smith et al. (1999) likened a service encounter to a social exchange whereby a consumer engages in a market transaction on the belief that he/she will receive a desirable outcome after expending a certain amount of resources to carry out stipulated transactional activities. But because of the presence of dysfunctional services, expended resources may be forfeited without producing the desired transactional outcome for the consumer. If such a loss is not reimbursed in kind through service recovery efforts on the part of the vendor, Smith et al. (1999) contended that the social exchange cannot be equalized, which may erode the willingness of the consumer to further participate in the exchange relationship (see Bitner et al., 1990). Applying the SET, Smith et al. (1999) proposed four modes of service recovery that align with the concepts of distributive, procedural and interactional justice similar to that of Tax et al. (1998), namely compensation, response sensitivity, affinity and initiative. Together, these four service recovery modes symbolize a targeted approach to the resolution of service failures by reimbursing consumers with equitable resources relative to incurred losses. That is, remuneration could be either utilitarian in that it involves the exchange of economic resources (e.g., money, goods or time) or symbolic in that it entails the exchange of socio-psychological resources (e.g., status, esteem, or empathy) (Smith et al., 1999). This thesis espouses Smith et al.’s (1999) typology as the most comprehensive yet parsimonious framework from which to isolate leverage points for optimizing eservice recovery efforts on e-commerce websites. Three reasons justify our claim. First, as illustrated in Table 4.1, both the retail recovery strategies advanced by Kelley et al. (1993) and the justice-oriented service recovery tactics introduced by Tax et al. (1998) can be readily subsumed under Smith et al.’s (1999) typology of service recovery modes. Second, as compared to Smith et al.’s (1999) taxonomy, Kelley et al.’s (1993) and Tax et al.’s (1998) typologies of service recovery are much too abstract and sophisticated to generate prescriptive guidelines for informing the design of e-service recovery technologies on e-commerce websites (Benbasat and Zmud, 2003; Voorhees and Brady, 2005). Moreover, as stated before, several dimensions in both typologies 44     demand a physical presence (e.g., honesty and manager/employee intervention) and would not be feasible as a recovery technology for e-commerce websites. Third, we also reject Kelley et al.’s (1993) and Tax et al.’s (1998) typologies due to their inclusion of both workable and non-workable (shown in red in Table 4.1) service recovery measures. E-commerce websites are extremely vulnerable to the repercussions from e-service failures as consumers face minimal switching costs in alternating among e-merchants (Harris Interactive, 2006). Therefore, it is not feasible for e-merchants to consider nonworkable e-service recovery solutions because there are no second chances for a failed recovery. Table 4.1: Comparison of Contemporary Frameworks of Service Recovery Service Recovery  Modes  [as adapted from  Smith et al. 1999]   Definition [Ability of  recovery measures to  appease customers  through…]   Justice‐Oriented  Service Recovery  Tactics  [Tax et al. 1998]   Compensation   Reimbursement of  tangible economic  resources such as  money, discount, free  merchandise and  coupons.   Reimbursement /  Reimbursement of the  Refund  full cost paid by  customers for products  or services prior to the  failure.   Refund   Reimbursement of the  full cost paid by  customers for products  or services prior to the  failure.   Replacement   Replacement of  defective products  purchased.   Replacement   Replacement of  defective products  purchased.   Store Credit   Offering credit for  Store Credit  customers’ next  transaction at the same  location from which  they have experienced  the failure.   No Comparison  Response  Sensitivity   Availability of  Repair  measures that  anticipate the  possibility of common  customer’s experiences  and are competent in  dealing with them  Assuming  efficiently.  Responsibility   Definition [Ability of  recovery tactics to  appease or anger  customers through…]   Service Recovery  Strategies  [as adapted from  Kelley et al. 1993]   Offering credit for  customers’ next  transaction at the  same location from  which they have  experienced the  failure.   Discount   Providing discounts on  desired products or  free merchandise.   Correction   Prompt rectification of  mistakes that were  made by the firm with  no physical  intervention from  employees.   No Comparison  Repairing defective  products purchased or  rectifying service  mistakes.   Definition [Ability of  recovery strategies to  appease or anger  customers through…]   Taking responsibility for  Manager/Employee  controllable service  Intervention  failures.   Intervention of service  personnel who are  empowered to solve  the problems at hand.   No Resolution   Ignoring customers  totally despite knowing  the failure occurrence.   Ignoring customers  totally despite knowing  the failure occurrence.   Process Control   Empowering customers  to take control over how  they wish the service  recovery to proceed.   Nothing   No Comparison   No Comparison   45     Service Recovery  Modes  [as adapted from  Smith et al. 1999]   Definition [Ability of  recovery measures to  appease customers  through…]   Justice‐Oriented  Service Recovery  Tactics  [Tax et al. 1998]   Definition [Ability of  recovery tactics to  appease or anger  customers through…]   Knowledge of  Process   Advising customers of  the procedural steps  involved in service  recovery.   Service Recovery  Strategies  [as adapted from  Kelley et al. 1993]   Failure Escalation  No Comparison   Affinity   Building of rapport  Apology  using socio‐ psychological resources  such as apologies, the  accordance of esteem,  politeness/courtesy,  concern and empathy.  Convenience   No Comparison   Definition [Ability of  recovery strategies to  appease or anger  customers through…]   Delays in providing  correction measures  such that the service  failure scenario is  allowed to escalate in  magnitude.   Offering apologies for  Apology  service failures and the  resulting inconvenience  to customers.   Offering apologies for  service failures and the  resulting  inconvenience to  customers.   Resolving service  failures with minimal  hassle and  inconvenience to  customers.    Undertaking corrective  actions that were  made with much  hassle and  inconvenience to  customers who have  already suffered from  the initial failure.   Timing / Speed   Resolving service  failures in as little time  as possible.    Flexibility   Allowing customers to  make special requests  during service recovery  that are not commonly  practiced.   Politeness   Assisting customers  through the service  recovery process in an  affable manner.   Empathy   Aiding customers  through the service  recovery process in a  personalized fashion.   Effort   Going beyond the call of  duty during service  recovery.   Explanation /  Information   Offering explanation  and/or information to  customers on the cause  behind the occurrence  of service failures   Honesty   Being truthful to  customers about the  occurrence of service  failures and/or recovery  measures   Attitude   Ensuring that service   Unsatisfactory  Correction   No Comparison   No Comparison   46     Service Recovery  Modes  [as adapted from  Smith et al. 1999]   Definition [Ability of  recovery measures to  appease customers  through…]   Justice‐Oriented  Service Recovery  Tactics  [Tax et al. 1998]   Definition [Ability of  recovery tactics to  appease or anger  customers through…]   Service Recovery  Strategies  [as adapted from  Kelley et al. 1993]   Definition [Ability of  recovery strategies to  appease or anger  customers through…]   recovery is to the  satisfaction of  customers.  Initiative   Demonstration of the  Correction Plus  capability of the  vendor to proactively  engage customers in  the solicitation of  unreported failures and  the provision of  innovative solutions  that are not expected  Follow‐Up  of the firm.   Providing services that  Correction Plus  goes beyond mere  correction of the failure  and puts customers in a  better position from  where they started  from.  Keeping customers  informed of the  progress of service  recovery efforts.   No Comparison   Customer‐Initiated  Correction  No Comparison   No Comparison   Providing services that  goes beyond mere  correction of the  failure and puts  customers in a better  position from where  they started from.   No Comparison   Offering service  recovery only upon  contact from  customers in scenarios  whereby the firm is  expected to have  known about the  failure beforehand.   Adapting Smith et al.’s (1999) typology to e-commerce websites, we postulate that e-service recovery technologies can be designed to appease consumers who have experienced failure occurrences through: (1) compensation where tangible economic resources are reimbursed; (2) response sensitivity where common errors are anticipated and guidance on their resolution is provided; (3) affinity where rapport is built using socio-psychological resources, and; (4) initiative where consumers are proactively engaged in the identification of unreported e-service failures. Further, as can be seen from Appendix D, recovery technologies, which are predominantly accessible from modern e-commerce websites, can be readily classified under one of the four recovery modes, thereby lending credibility to our choice of Smith et al.’s (1999) typology as the guiding theoretical framework for conceptualizing e-service recovery. While we have demonstrated the relevance of each the four service recovery modes in Smith et al.’s (1999) typology to e-commerce transactional environments, we have opted to drop the concept of initiative in this thesis. Because initiative on e-commerce websites is typically implemented as a 47     pop-up survey questionnaire that synchronizes with consumers’ transactional activities to solicit feedback on various web aspects (see Appendix D), it may be construed by consumers as an inhibitor (Cenfetelli, 2004). Citing pop-ups as a prime example, Cenfetelli (2004) defined inhibitors as “perceptions held by a user about a system’s attributes with consequent effects…that act solely to discourage use” (p. 475). Inhibitors capture the intuition that the presence of certain functionalities on e-commerce websites may serve to discourage usage among consumers even though their absence may not have any positive impact as well. In this sense, initiative in the form of pop-up survey questionnaires may not be a desirable mode of e-service recovery because it could induce failure perceptions among consumers instead: unlike pop-up advertisements, such questionnaires are likely to be highly disruptive to the transactional process by compelling consumers to switch between browser windows (see Appendix D).  4.2  Summary Reviewing contemporary frameworks of service recovery, this chapter adapts and positions  Smith et al.’s (1999) typology as an intuitive framework from which to taxonomize the myriad of eservice recovery technologies accessible from e-commerce websites and prescribe actionable design principles for e-merchants. Having derived guiding theoretical frameworks of e-service failure and recovery in the preceding chapters, Chapter 5 will construct an integrated theory together with testable hypotheses for empirical investigation.  48     C HAPTER 5 – A N I NTEGRATED T HEORY OF E-S ERVICE F AILURE AND R ECOVERY In this chapter, we focus on the construction of an integrated theory of e-service failure and recovery. Building on our proposed typology of e-service failure and Smith et al.’s (1999) typology of service recovery, we draw on: (1) the Expectation Disconfirmation Theory (EDT) to postulate negative consequences of information, functional and system failures, and; (2) Counterfactual Thinking to predict the effectiveness of compensatory, affinity and response sensitivity e-service recovery technologies in moderating these failure consequences. Figure 5.1 depicts our theory of e-service failure and recovery.  49     Figure 5.1: Theory of E-Service Failure and Recovery    Inaccurate Information  Incomplete Information   P1   Irrelevant Information  Untimely Information    Disconfirmed  Outcome  Expectancy     Needs Recognition Failure  Alternatives Identification Failure  P4, P5  P2  Alternatives Evaluation Failure  P4, P6  Disconfirmed  Process  Expectancy   Acquisition Failure  Disconfirmed  Cost Expectancy   Post‐Purchase Failure   Inaccessibility      Non‐Adaptability  P4, P7 Non‐Navigability  P3  Delay  Insecurity   Affinity   5.1    Response  Sensitivity  Compensation   An Expectation Disconfirmation Perspective of E-Service Failure Consequences Fundamental to service failure is the idea of expectation disconfirmation (Hess et al., 2007;  Hoffman and Bateson, 1997). It is well-accepted within service literature that consumers generally possess preconceived notions of service performance and that service failures manifest whenever those preconceptions have been violated (e.g., Andreassen 2001; Bearden and Teel 1983; Bitner 1990; 50     Hess et al. 2007; Smith et al. 1999). The same reasoning applies to e-service failures. Due to the existence of multiple frames of reference (e.g., previous transactional experience from both online and offline retail) from which online consumers may draw upon to base their evaluations of ecommerce websites, it is natural that the disconfirmation of these predefined expectations would be indicative of e-service failures. However, the EDT, in its current form, lacks sufficient explanatory and predictive power in modeling e-service failures and their consequences. Despite the extensive application of the EDT in investigating service failures, none has gone beyond theorizing disconfirmation as a monolithic construct. Implicitly, the manifestation of a service failure implies that a consumer’s expectations have not been fulfilled in some manner (Hess et al., 2007; Hoffman and Bateson, 1997). Yet to-date, there have been no scholarly attempts to further delineate the disconfirmation construct to yield insights into the type of consumer expectation that has been compromised whenever a particular form of service fails. Expectations are principal determinants of consumers’ positive attitudes towards e-commerce websites because they are the baseline from which evaluative judgments about focal e-services are formulated (Bhattacherjee, 2001). The disconfirmation of customer expectations is driven by the value to be gained from service utilization—the utility accorded to consumers due to perceptual differences between what is to be expected and what is actually given (Cronin et al. 2000; Parasuraman and Grewal 2000). Embodied within the concept of value is an inference to cost-benefit analysis (Cronin et al. 2000; Parasuraman and Grewal 2000) and, as reasoned by Davis et al. (1992), cost-benefits associated with technology usage are rooted in: (1) the capacity of the technology to produce desired task outcomes, as well as; (2) the tangible and intangible costs that must be expended by individuals in utilizing the technology. This distinction between outcome and cost associated with technology usage has been reflected in the well-established Technology Acceptance Model (TAM) as users’ perceptions of ‘usefulness’ and ‘ease of use’ towards technological systems (Davis, 1989). 51     Likewise, in a comprehensive review of technology acceptance models and theories being applied in system usage studies, Venkatesh et al. (2003) advanced a Unified Theory of Acceptance and Use of Technology (UTAUT) that delineates performance and effort expectancy as distinct influences affecting users’ receptivity towards technological systems. Yet, going beyond the cost and outcome associated with service utilization, there is ample evidence within service literature to suggest that the servicing process should not be ignored (e.g., Collier and Bienstock 2003, 2006). Berry et al. (1985) differentiated between process and outcome in conceptualizing services (see also Collier and Bienstock 2003, 2006; Fassnacht and Koese 2006). They argued that service process depicts consumers’ interactive exchange with a service, whereas service outcome is the consequence culminating from the execution of the service (Berry et al. 1985). The importance of incorporating process as a distinctive facet of services is also echoed by Mentzer et al. (2001). Mentzer et al.’s (2001) study on logistics service quality revealed that consumers’ service expectations can be segregated into those pertaining to order placement (i.e., process) and those regarding order receipt (i.e., outcome). Similarly, Jacoby (1998) divided consumers’ product acquisition process into five sequential stages (i.e., needs recognition, alternatives identification, alternatives evaluation, product acquisition, and post-purchase) and maintained that the provision of services to move transactional activities seamlessly along these stages is the key to fulfilling customers’ expectations. Arguably, consumers are likely to possess expectations about how transactional processes should flow on e-commerce websites and these expectations are disconfirmed whenever they encounter disruptions to their transactions due to the presence of e-service failures. We hence distinguish among outcome, process, and cost as distinct expectations that consumers harbor towards service utilization. That is, e-service failures may lead to the disconfirmation of consumers’ outcome, process and cost expectancies: 1.  Disconfirmed outcome expectancy manifest whenever the transactional outcome(s) obtained from the e-commerce website is not what is desired by the consumer, 52     2.  Disconfirmed process expectancy manifest whenever the transactional process on the e-commerce website does not proceed in a manner expected by the consumer, and;  3.  Disconfirmed cost expectancy manifest whenever a consumer expends more resources than anticipated in transacting via an e-commerce website.  5.1.1  Consequences of Informational Failures A basic tenet of consumer behavioral theory holds that when customers make purchase  decisions, the type of information is instrumental to the formation of decisional outcomes (Furse et al., 1984; Keaveney and Parthasarathy, 2001; Muthukrishnan and Chattopadhyay, 2007). As affirmed through existing studies of consumer satisfaction and service quality, the information employed by customers in making choice decisions impacts outcome predictability (e.g., Boulding et al., 1993; Oliver, 1997; Yi, 1990; Zeithaml et al., 1993, 1996). That is, if misinformation were to be supplied by an e-commerce website, whether intentionally or unintentionally, consumers may be misled into acquiring products and/or services that do not fit their requirements. Similarly, Collier and Bienstock (2003, 2006) have attested to informational attributes (e.g., accuracy and timeliness) as crucial antecedents of service outcome quality. Because the saliency of informational attributes in influencing task outcomes is well documented within system success (e.g., Bailey and Pearson, 1983; Ives et al, 1983; Wixom and Todd, 2005) and service failure (e.g., Gershoff et al., 2001; Holloway and Beatty, 2003) literatures, we propose that: Hypothesis 1: Informational failure on an e-commerce website will result in the disconfirmation of consumers’ outcome expectancy.  5.1.2  Consequences of Functional Failures Functional failures cause dissonance to manifest in e-commerce transactional processes.  Empirical findings suggest that service functionalities, no matter how well designed they may be, are rendered meaningless if they cannot satisfy consumers’ transactional needs (Cenfetelli et al. 2008). The same opinion was expressed by Piccoli (2001), who claimed that one must “think creatively about how technology can be integrated into your products and into your customer’s experience 53     [because] the most innovative ideas are often not the most costly or resource-intensive, but simply those based on an understanding of how customer needs can effectively be satisfied” (p. 45). Ecommerce websites in this sense, should not only mirror physical retailers in the range and sophistication of services being offered to consumers, but must also made available transactional functionalities, which are otherwise infeasible via conventional media (Barnes and Vidgen 2003; Homburg et al., 2002). Studies conducted in both e-commerce (Cenfetelli et al., 2008) and e-government (Tan et al., 2010) domains have claimed that consumers’ service expectations for online transactions are not only distinguishable from those for their offline counterparts, but that these expectations also vary depending on which stage of the transactional process consumers are currently engaged in. The availability of complementary web-enabled functionalities to cater to the spectrum of service expectations throughout the online transactional process is therefore deterministic of an e-commerce website’s eventual acceptance by its target audience (Cenfetelli et al., 2008; Tan et al., 2010). For instance, while recommendations agents are probably needed in the beginning of an e-commerce transaction to assist consumers in product selection, ordering and payment functions become salient in the later stages for product acquisition purposes. Olsen (2003) has blamed functional failures along various stages of the online transactional process for systematically ejecting consumers from ecommerce websites and contributing to a low conversion rate of 34% among purchase-ready customers (see also Holloway and Beatty, 2003). Given the growing evidence that alludes to the decisive role of service functionalities in sustaining a fluid e-commerce transactional process (e.g., Cenfetelli et al., 2008; Lightner, 2004; Tan et al., 2010), we propose that: Hypothesis 2: Functional failure on an e-commerce website will result in the disconfirmation of consumers’ process expectancy.  5.1.3 Consequences of System Failures The minute a consumer visits an e-commerce website, he/she already begins to incur a cost for the transaction, be it effort expended or time spent. Because system attributes affect the efficiency 54     with which consumers can access service content on an e-commerce website (DeLone and McLean, 2003; Wixom and Todd, 2005), it is inevitable that the presence of system failures lowers consumers’ effort-performance expectancy as a much higher transactional cost must now be incurred to attain satisfactory service performance (Holloway and Beatty, 2003). Empirical justification for such a relationship is abundant. Prior research has testified to an inverse relationship between response time and the amount of resources invested by system users (Barber and Lucas, 1983; Martin and Corl, 1986). Studies have shown that delays on e-commerce websites induce a sense of loss in consumers because they are forced to spend way more time than projected in accomplishing online transactions (Lee and MacGregor, 1985; Sears et al., 2000). Nah (2002) noted that, in the worst case scenario, consumers would rather terminate the transaction than waste time on unbearably slow e-commerce websites. Besides response time, there are yet other system attributes that have obtained support in extant service literature as having an impact on consumers’ transactional costs such as accessibility (e.g., Agarwal and Venkatesh, 2002; Douglas et al., 2003; Shim et al., 2002), adaptability (e.g., e.g., Agarwal and Venkatesh 2002; Kim et al. 2006; Palmer 2002; Ribbink et al. 2004; Semeijn et al. 2005; Srinivasan et al. 2002; Surjadjaja et al. 2003) and navigability (e.g., Meliàn-Alzola and PadronRobaina 2006; Semeijn et al. 2005; Surjadjaja et al. 200). Since the impact of system attributes on transactional costs has received broad consensus among researchers, we propose that: Hypothesis 3: System failure on an e-commerce website will result in the disconfirmation of consumers’ cost expectancy.  5.2  A Counterfactual Thinking Perspective of E-Service Recovery Effectiveness When e-service failures occur, counterfactual thinking would compel consumers to question  if e-commerce websites have taken steps to improve the situation, the absence of which would indicate a misguided recovery process (McColl-Kennedy and Sparks, 2003). Counterfactual thinking plays a critical role in consumers’ assessment of e-service recovery because the suitability of recovery  55     technologies would depend on whether they conform to measures that customers anticipate to be present on e-commerce websites to alleviate negative failure consequences. From a counterfactual thinking perspective, we differentiate between the mere presence of a recovery technology and its commensurability in determining consumers’ reactions towards e-service failures. Whereas the presence of e-service recoveries signals an e-merchant’s willingness to make amends for e-service failures and lessen to an extent the immediate fallout from failure incidents, it is the commensurability of these recoveries that would ultimately decide consumers’ eventual behavioral responses towards the offending e-commerce website. Previous research lends supports to this distinction as well. McColl-Kennedy et al. (2003) empirically certified that the presence of any manner of service recovery is better than inaction. The existence of e-service recoveries, regardless of their compatibility, would moderate the negative consequences arising from e-service failures. Further, as observed by Tam and Ho (2006), ecommerce websites emits cues (or signals) that consumers exploit to augment transactional activities such as discerning the risk of the transaction (Schlosser et al., 2006) or detecting deception (Xiao and Benbasat, 2011). The presence of e-service recoveries thus fulfills a secondary purpose of instilling a calming effect on consumers that increases their confidence in salvaging failed transactions. For example, just by having self-service resolution centers alone may convince consumers that they can most certainly recover from any unfavorable transactional outcomes (e.g., defective products, wrong purchases and misplaced deliveries). We therefore propose that: Hypothesis 4: The presence of any e-service recovery technology (compensation, response sensitivity or affinity) will negatively moderate the positive relationship between an e-service failure and consumers’ disconfirmed expectancy. Yet, equal emphasis must be placed on the commensurability of e-service recovery technologies. Smith et al (1999) observed an interaction effect between failure and recovery by illustrating that consumers prefer recoveries that are commensurate with the form and magnitude of failure consequence experienced. A mismatch between failure consequence and recovery is limited in 56     easing consumers’ displeasure. For example, if unusually long delay in the loading of webpages is the problem, compensating consumers with product discounts may be a disproportionate response. Rather, having an alternate site with minimal graphics to accelerate the loading process would be the preferred solution. Separating the presence of e-service recovery technologies from their commensurability would thus enrich our theory. In subsequent sections, we revisit extant literature on service recovery to propose that although any type of recovery technology (i.e., compensation, response sensitivity and affinity) is better than inaction at moderating consumers’ disconfirmed expectancies that arise from eservice failures, certain recovery may be more effective than others depending on the nature of the expectation being disconfirmed.  5.2.1  Moderating Effect of Compensatory Recovery Technology Compensation is a standard recovery procedure in which consumers are reimbursed (in the  form of coupons, discounts, free merchandise and refunds) for any losses they may have suffered as a consequence of service failures (Smith et al., 1999). Though there are considerable disagreements over the level of compensation to be awarded for a given failure incident, existing studies are unanimous in attesting to the importance of compensation in counterbalancing negative service outcomes (e.g., McCollough et al., 2003; Lovelock and Wirtz, 2004; Wirtz and Mattila, 2004). Through content analysis of qualitative complaints, Tax et al. (1998) claimed that compensation is particularly advantageous in assisting consumers to recover from undesirable service outcomes. Likewise, Mattila and Patterson (2004a, 2004b) uncovered that compensation drives customers’ perceptions of outcome fairness whereas McColl-Kennedy et al. (2003) revealed that it is especially pertinent in alleviating service failures that plague outcome-driven consumers. Conceivably, while there are no signs to indicate that compensation may not be relevant to other forms of service failures, the bulk of empirical evidence appears to suggest that it is better suited as a recovery for disconfirmed outcomes. Since e-commerce transactions take place virtually with little room for physical intervention, compensation measures must not only guarantee that consumers are sufficiently 57     reimbursed for damages suffered, they should also entail digital means for customers to arrange for reimbursements without having to engage in human contact. For example, if a consumer is misinformed (due to informational failures) into acquiring a product that does not meet his/her requirements, self-service resolution centers such as those offered by major e-commerce players (e.g., Amazon.com, eBay.com or Expedia.com) should be provided to assist the consumer in seeking amends for the misguided purchase. We therefore propose that: Hypothesis 5: Compensatory e-service recovery technology will have a stronger negative moderating effect on the positive relationship between an e-service failure and consumers’ disconfirmed outcome expectancy as compared to response sensitivity and affinity recovery technologies.  5.2.2  Moderating Effect of Response Sensitivity Recovery Technology Response sensitivity has been an integral part of service quality and measures vendors’  propensity to be helpful and prompt in responding to consumers (Cenfetelli et al., 2008; Clemmer and Schneider, 1993, 1996; Parasuraman et al., 1985, 1988). A well-timed and fitting response to service failures has been observed to improve consumers’ assessment of service encounters (Berry et al., 1994; Bitner et al., 1990; Clark et al., 1992; Hart et al., 1990; Kelley et al., 1993; Smart and Martin, 1992; Taylor, 1994). In the context of e-commerce websites, the criticality of response sensitivity has been acknowledged by Gefen (2002), who stated that while it may be “doubtful if automated systems today can provide the kind of responsive service that salespeople can, but there are some responsiveness aspects that also relate to websites: providing prompt service, providing helpful guidance when problems occur, and telling customers accurately when the ordered services will be performed or the products delivered” (p. 32). This view is also borne out in previous studies of recovery voice whereby scholars noted that it is imperative for vendors to demonstrate their receptivity and sensitivity towards customer feedback during service failures by giving dissatisfied consumers a chance to voice their opinions about how existing services may be enhanced to avert similar failures in the future (e.g., Karande et al., 2007; McColl-Kennedy et al., 2003; Sparks and McColl-Kennedy, 2001). Arguably, response sensitivity is the most appropriate mode of recovery 58     whenever transactional processes are abruptly disrupted because swift and targeted responses should be imminent to prevent customer exits. E-service recoveries exhibiting response sensitivity are also evident from modern e-commerce websites whereby one is accustomed to finding comprehensive guides on Frequently Asked Questions (FAQs) (e.g., Amazon.com, eBay.com and Expedia.com), live help (e.g., Dell.com and Futureshop.ca) and/or customer feedback forms (e.g., Amazon.com, Dell.com, eBay.com, Expedia.com and Futureshop.ca). Evidently, the provision of such recoveries moderates the negative consequences of functional failures by offering: (1) ready answers to common transactional queries (e.g., step-by-step tutorials on how to order and pay for a product), or; (2) communication channels for consumers to report transactional problem(s) and seek assurance that measures are being undertaken to prevent a repeat of such problems (e.g., automated response to feedback). We therefore propose that: Hypothesis 6: Response sensitivity e-service recovery technology will have a stronger negative moderating effect on the positive relationship between an e-service failure and consumers’ disconfirmed process expectancy as compared to compensatory and affinity recovery technologies.  5.2.3  Moderating Effect of Affinity Recovery Technology Affinity (with the most common manifestation being an apology) is a valuable reward that  redistributes esteem (a social resource) in an exchange relationship (Smith et al., 1999). Apologies from vendors communicate respect and empathy to consumers in the event of service failures, which in turn lowers the latter’s condemnation of the disappointing service encounters (Hart et al. 1990; Kelley et al. 1993). Studies have shown that apologies project a sense of care and transparency on the part of vendors by being upfront with consumers on the causes of service failures (Hart et al. 1990; Houston et al. 1998; Kelley et al. 1993; Taylor 1994). Costs incurred by consumers for e-service failures vary considerably on an individual basis (Mattila 2001). For instance, whereas one consumer might view a delay in loading time for an e-commerce website to be exceedingly wasteful, another may find the same delay to be reasonably acceptable. Without sufficient knowledge on the value each individual consumer attaches to the amount of resources (tangible or otherwise) he/she invested on an 59     offending e-commerce website, an apology could be a more universal remedy in that it goes a long way towards “[acknowledging] the costs that were imposed upon the consumer” (Houston et al. 1998, p. 742). This is also evident from existing studies of service delays. Service delays not only impose economic overheads (e.g., time wasted) as well as social and emotional tolls on consumers (e.g., anger and frustration) (Larson 1987), they also involve opportunity costs in that customers could have seek out other alternatives (Leclerc et al. 1995). Without the ability to accurately assess the loss suffered by consumers in service delays, Houston et al. (1998) admitted that vendors may be better off giving an apology. In apologizing, vendors not only express their recognition of the delay, but they also convey sympathy towards consumers’ unfortunate predicament (Taylor 1994). We therefore propose that: Hypothesis 7: Affinity e-service recovery technology will have a stronger negative moderating effect on the positive relationship between an e-service failure and consumers’ disconfirmed cost expectancy as compared to compensatory and response sensitivity recovery technologies.  5.3  Summary Building on our proposed e-service failure typology and Smith et al.’s (1999) typology of  service recovery, we construct a theory of e-service failure and recovery by assimilating the preventive and corrective research streams within extant literature. Specifically, we draw on the EDT to account for disconfirmed expectancies that arise from informational, functional and system failures and counterfactual thinking to hypothesize the effectiveness of compensatory, response sensitivity and affinity recovery technologies in alleviating these disconfirmed expectancies. Based on this theory, Chapter 6 outlines the design and execution of a repeated measures experimental study to validate the relationships embedded within the theory.  60     C HAPTER 6 – A N E XPERIMENTAL S TUDY OF E-S ERVICE F AILURE AND R ECOVERY (2 ND S TUDY ) As noted by Holloway and Beatty (2003), existing e-commerce websites are lagging in the provision of e-service recovery technologies to alleviate e-service failures. Therefore, to validate our theory of e-service failure and recovery, an online experiment was conducted to achieve two research objectives: (1) to ascertain the impact of the three predominant forms of e-service failure (i.e., informational failure, functional failure and system failure) on consumers’ disconfirmed expectancies, and; (2) to explore the effectiveness of the three e-service recovery technologies (i.e., compensation, response sensitivity and affinity) in alleviating negative consequences arising from these failures.  6.1  Experimental Design While prior research has testified to the existence of an interaction effect between failure and  recovery (e.g., Holloway and Beatty, 2003; Kelley et al., 1993; Smith et al, 1999), it is not yet apparent what type(s) of recovery technologies would be effective in moderating negative consequences that arise from a particular form of failure. To this end, the validation of our theory of e-service failure and recovery closes the knowledge gap pertaining to the commensurability of recovery technologies relative to the failure experienced. The experiment was designed as a matchup between probable e-service recovery technologies with specific instantiations of e-service failures. Given the constraints of an experimental study, it is not feasible to manipulate every form of e-service failure in our theory. Rather, we selected three pervasive types of e-service failure, one each from the categories of informational, functional and system failure, to serve as treatments in our experiment. Hypotheses being tested in this experiment are summarized in Table 6.1.  61     Table 6.1 Hypothesis to be Tested H1: Informational failure on an e‐commerce website will result in the disconfirmation of consumers’ outcome expectancy.  H2: Functional failure on an e‐commerce website will result in the disconfirmation of consumers’ process expectancy.  H3: System failure on an e‐commerce website will result in the disconfirmation of consumers’ cost expectancy.  H4:  The  presence  of  any  e‐service  recovery  technology  (compensation,  response  sensitivity  or  affinity)  will  negatively  moderate  the  positive relationship between an e‐service failure and consumers’ disconfirmed expectancy  H5: Compensatory e‐service recovery technology will have a stronger negative moderating effect on the positive relationship between  an  e‐service  failure  and  consumers’  disconfirmed  outcome  expectancy  as  compared  to  response  sensitivity  and  affinity  recovery  technologies.  H6:  Response  sensitivity  e‐service  recovery  technology  will  have  a  stronger  negative  moderating  effect  on  the  positive  relationship  between  an  e‐service  failure  and  consumers’  disconfirmed  process  expectancy  as  compared  to  compensatory  and  affinity  recovery  technologies.  H7: Affinity e‐service recovery technology will have a stronger negative moderating effect on the positive relationship between an e‐ service  failure  and  consumers’  disconfirmed  cost  expectancy  as  compared  to  compensatory  and  response  sensitivity  recovery  technologies.   The experiment employs a 3 [Type of E-Service Failure: Informational Failure, Functional Failure and System Failure] x 2 [Compensation: Present and Absent] x 2 [Affinity: Present and Absent] x 2 [Response Sensitivity: Present and Absent] between-subjects factorial design with 619 participants (see Table 6.2). Apart from the twenty-four treatment groups, a control group [N = 25] with no eservice failure was also included as part of the experimental design to contrast the presence of an eservice failure against its absence on participants’ disconfirmed expectancies. Together, this yields a total of 644 participants in the experiment. Table 6.2: Between-Subjects Experimental Design E‐Service Recovery     COM []  AFF []   COM []  AFF []   AFF []   AFF []     E‐Service Failure   RES []   RES []   RES []   RES []   RES []   RES []   RES []   RES []   Informational Failure (IF)   Group 1   Group 2   Group 3   Group 4   Group 5   Group 6   Group 7   Group 8   [N = 25]   [N = 26]   [N = 26]   [N = 25]   [N = 27]   [N = 26]   [N = 25]   [N = 26]   Group 9   Group 10  Group 11  Group 12  Group 13  Group 14  Group 15  Group 16   [N = 26]   [N = 25]   [N = 25]   Functional Failure (FF)   System Failure (SF)   [N = 25]   [N = 25]   [N = 25]   [N = 27]   [N = 25]   Group 17  Group 18  Group 19  Group 20  Group 21  Group 22  Group 23  Group 24   [N = 25]   [N = 28]   [N = 27]   [N = 25]   [N = 26]   [N = 25]   [N = 27]   [N = 27]   COM – Compensation  AFF – Affinity  RES – Responsiveness   62     6.1.1  A General Overview of Experimental Procedures Experimental Setting: An experimental website featuring an artificial e-merchant was  created for each of the twenty-four treatment groups as well as for the control group. Experimental websites for the twenty-five cells are entirely identical with the exception of an e-service failure and the inclusion of an e-service recovery technology corresponding to the specific manipulation for each of the twenty-four treatment groups (see Table 6.2). Each website features the same lineup of 45 laptop computers with product attributes that are consistent with models, which are freely available from online marketplaces at the time of the experiment. This preserves the realism of the experimental setting. Figure 6.1 depicts the experimental website from the control group that is void of any e-service failure and recovery manipulations. Figure 6.1: Experimental Website from Control Group  Like the first study, experimental participants were recruited with the aid of a commercialized marketing firm. Sourcing for external participants increases the generalizability of our empirical findings at the expense of lesser control over experimental procedures. Nevertheless, with ample pre63     testing and fine-tuning of the experimental design, we are convinced that this loss of control does not pose a threat to the internal validity of the experiment. An email invitation was broadcasted to the entire panel of potential subject pool accessible from the commercialized marketing firm. The email contains a synopsis of the experimental procedures together with a link to one of the twenty-five experimental websites if the recipient is willing to participate in the study. Each participant is rewarded with participation-based incentives from the marketing firm. Arrangements were made with the marketing firm to randomly assign participants to one of the twenty-five treatment groups. The first page of the experimental websites is a preamble that contains detailed descriptions of the experimental procedures as well as an electronic consent form for participants to ‘sign’ (see Figure 6.2). Only upon giving consent will participants be permitted to continue with the experiment. Participants were also reminded that their participation is entirely voluntary and they can choose to withdraw from the experiment at any point in time by closing the browser window.  64     Figure 6.2: Introductory Page and Electronic Consent Form of Experimental Websites  Electronic consent form for experimental participants to agree to proceed with the experiment  For participants who chose to proceed with the experiment, they were given an unconstrained experimental task that requires them to browse through the various laptops being offered on the experimental websites and select THREE that best match their current needs (see Figure 6.3). Participants were also told that they are automatically entered into a raffle context for which TEN lucky winners stand a chance of winning a 50% discount towards the laptop they have chosen as their 65     top choice: laptop prices range from USD $299 to USD $1,999 with the mean being USD $819.67 and the median being USD $692 due to realistic pricing of laptop models (i.e., laptop models are priced according to market rates). It was also made known to participants that even if they were to emerge as winners of the raffle contest, arrangements will be made with an e-merchant to award them with a voucher towards the exact laptop they have picked as their top choice and at the price specified on the website. The voucher is non-transferable and cannot be used as store credit or be exchanged for cash, i.e. the voucher can ONLY be used cover the cost of the laptop they have chosen. The aforementioned conditions are imposed to substantially incentivize participants without prompting them to aim for the most expensive laptop right away. Because winners of the raffle contest were still expected to foot the other 50% of the laptop cost, participants would be motivated in their search for laptops that fit their requirements and at a price which is affordable to them. The design of the experimental task was finalized after three rounds of pilot testing in which we rejected a predefined task with performance incentives as well as one that is identical to the preceding task but with a smaller discount of 20% for winners of the raffle contest: the former being rejected due to its inability to motivate participants for an unrelated task and the latter being rejected for the insufficiency of the discount to generate adequate motivation among participants (since they still expect to pay for the other 80% of the laptop cost).  66     Figure 6.3: Description of Experimental Task  Once ready, participants were directed to the artificial e-merchant for their treatment group to begin shopping for the laptops. There was no time limit imposed on participants to make the purchase decision. To complete the experimental task, participants must: (1) select three laptops; (2) place them in the shopping cart; (3) rank them according to how close they think each comes to meeting their needs, and; (4) click the submit button. Upon submitting their choices, participants were directed to an online survey questionnaire to provide evaluations of the recently completed transaction. This questionnaire is designed both as a manipulation check to determine whether failure manipulations have been successful and as data collection for the impact of e-service failure manipulations on consumers’ disconfirmed expectancies (i.e., consequences of e-service failure treatments must be measured before the introduction of recovery technologies). Answering the survey questionnaire completes the experiment for the control group as well as for treatment groups 8, 16 and 24, which are not exposed to any type of e-service recovery technology. Conversely, participants belonging to Groups 1 – 7, 9 – 15 and 17 – 23 were forwarded to another page containing the e-service recovery technology that corresponds to the manipulation for 67     their treatment group. From the experimental procedures stated in the preamble of the experimental websites, participants were aware that they can take as long as they want to explore the ‘e-services’ (i.e., recovery technologies) that have been provided in relation to their earlier shopping experience. Once participants were satisfied that they have done all they can with the e-service recovery technology, they can proceed onto the final stage of the experiment where they are presented with a second survey questionnaire, which contains both manipulation checks for the e-service recovery technologies being provided and measures evaluating the effectiveness of these technologies in alleviating failure consequences. Figure 6.4 depicts a simplified diagrammatic flow of the experimental procedures.  68     Figure 6.4: Diagrammatic Flow of Experimental Procedures Randomly assign to one of  twenty‐five treatment  groups    Given experimental task on  purchasing three laptops  which matches needs    Redirect to artificial e‐ merchant and present with  e‐service recovery  technology [Treatment  Groups 1‐ 7, 9 – 15 & 17 –  23]  Direct to artificial e‐ merchant for completing  experimental task   Utilize e‐services to select  three laptops which  matches needs   Explore recovery technology  as a means of alleviating the  failure encountered   Place the three laptops in  shopping cart   Rank choice of three laptops  according to how close they  come to matching needs   N  Satisfy with  outcome?   nd  2 survey measuring  effectiveness and impact of  recovery technology on  failure consequence   Y st  1  survey measuring  consequence of e‐service  failure [Control + Treatment  Groups]  Control + Treatment Groups 8, 16 & 24  6.1.2  Manipulations of E­Service Failures  While there are comparatively many studies in the marketing discipline that embraces experimental designs in the investigations of offline service failures, there is no such tradition for eservice failures. Consequently, due to the paucity of experimental studies that are contextualized to eservice failures, manipulations for failure treatments were based primarily on the triangulation  69     between actual failure incidents solicited from the first study (see Table 6.3) and past empirical studies that experiment with web features, albeit from a system success angle. Table 6.3: Sample E-Service Failure Incidents Construct   Sample E‐Service Failure Incident [First Study]   Informational Failure   Out‐of‐Stock  Products: “I was ordering clothes and the color and sizes  of the  clothing I was ordering was  supposed to be in stock, but I received an error message saying no longer available or out of stock.”   Functional Failure   Product  Comparison  Error:  “I  recently  tried  to  order  several  items  from  a  retail  store  via  their  website,  www.kohls.com.  After  choosing  several  products  and  entering  the  desired  quantities,  I  decided  to  visit  Overstock.com to compare prices for similar items before placing the order with Kohl’s.  Before switching  websites, I created a username and password  on the Kohl’s website, assuming that my "basket" contents  would be saved. However, after navigating to the Overstock website and then returning to Kohls.com, my  basket contents had been cleared. Other shopping sites that I've used tend to be very sticky with my basket  contents  even  when  I  am  not  logged  in  as  a  user.  As  long  as  I'm  entering  from  the  same  IP  address,  my  shopping basket contents are usually retained. But this was not the case on the Kohl’s site. I did not recreate  my online order with them.”   System Failure   Long  Delays:  “I  was  taking  a  look  at  my  online  account  to  see  the  recent  transactions.  It  was  slow  and  it  delays my time spent on the Internet and viewing my personal information online.”   Informational Failure: Consistent with the qualitative data collected in the first study, one recurring form of informational failure is the inclusion of out-of-stock items in the product catalogues and/or consideration sets generated by recommendation agents on e-commerce websites. Out-of-stock items being listed on e-commerce websites was also highlighted as a constituent dimension of eservice failure in Holloway and Beatty’s (2003) typology. Our manipulation of informational failure therefore took the form of the first three products placed into the shopping cart being tagged as out of stock (see Figure 6.5). To further amplify the effects of our informational failure treatment, the experimental website was designed to bolster participants’ expectations of all listed products being in stock by: (1) including an option for participants to exclude out-of-stock items from the product catalogue, and; (2) having an inventory status display for every individual item in the product catalogue (see Figure 6.5). This gives participants the impression that any product listed on the ecommerce website must be in stock. Hereafter, we refer to our manipulation of informational failure as ‘out-of-stock’ for clarity.  70     Figure 6.5: Illustration of Control versus Informational Failure Manipulation Functioning shopping cart  Functioning Experimental Website [Control Group]  Option to remove out-of-stock products from catalogue  Inventory status display for products in catalogue  Out-of-Stock Items [Treatment Groups 01 – 08]  71     Functional Failure: Consumers rely on in-depth comparisons among product alternatives to arrive at purchase decisions (Jedetski et al., 2002; Payne et al., 1993). Therefore, e-commerce websites must provide functionalities that organize attribute information in an intuitive and easily comprehensible manner to facilitate inter-product comparison. The experimental studies of Haubl and Trifts (2000) as well as Jedetski et al. (2002) have alluded to the significance of the Comparison Matrix as an alternatives evaluation tool for such purposes. Although comparison matrixes are becoming a permanent fixture of recommendation agents as can be found at Procompare.com [http://www.procompare.com/], it is not mandatory for these two functionalities to always co-exist. Futureshop [http://www.futureshop.ca] for one, implements the comparison matrix as a feature of its product catalogue due to the lack of recommendation agents. Because each laptop comprises a total of 16 product attributes with real values that were extracted from existing e-commerce websites, comparison matrixes were incorporated as a feature of product catalogues on experimental websites to enable participants to compare and contrast viable alternatives. In the absence of recommendation agents, the comparison matrix would be an essential feature of the experimental websites to assist participants in laptop selection. But for those websites with functional failure manipulation (i.e., treatment groups 9 – 16), an error message was displayed instead whenever participants tried to make use of the comparison matrix (see Figure 6.6). The provision of a dysfunctional comparison matrix thus amplifies our functional failure manipulation. Hereafter, we refer to our manipulation of functional failure as ‘missing comparison’ for clarity.  72     Figure 6.6: Illustration of Control versus Functional Failure Manipulation Functioning Comparison Matrix  Functioning Experimental Website [Control Group]  Error page  Product Comparison Error [Treatment Groups 09 – 16]  System Failure: Response time is indicative of system quality (see Bailey and Pearson, 1983; Ives et al., 1983). Evidence from past experimental studies claims that consumers undertake psychological discounts at 8 seconds (Kuhmann, 1989), lose interest at 10 seconds (Ramsay et al., 1998), become impatient at 12 seconds (Hoxmeier and DiCesare, 2000), engage in disruptive actions 73     at 15 seconds (Shneiderman, 1998) and abandon the task at 38 seconds (Nah, 2002). Because the purpose of our system failure manipulation is to induce sufficient disappointment among participants in the system aspects of the experimental websites without encouraging thoughts of task abandonment, it was implemented as a 12 seconds time delay being forcefully imposed on the loading of every product-related webpage (i.e., product catalogue and comparison matrix) for websites with such treatments, but not on the other pages (e.g., homepage and shopping cart). By creating a sharp contrast in loading speed between product-related and other webpages, we amplified the effects of our system failure treatment. Further, it can be calculated from our system failure manipulation that the total time delay experienced by participants is equivalent to the number of viewed products multiplied by 12 seconds, i.e. the greater the number of products being viewed by a participant, the longer the time delay he/she experienced. This creates frustration among participants who, despite having an incentive to go through each viable laptop alternative, experienced an extra 12 seconds of time delay for viewing every other interested alternative. Hereafter, we refer to our manipulation of system failure as ‘delay’ for clarity. 6.1.3  Manipulations of E­Service Recoveries  For manipulations of e-service recovery technologies, a review of existing e-commerce websites was conducted to generate ideas on manipulations of e-service re recovery technology that are realistic. Appendix D exemplifies e-service recovery technologies that have implemented by ecommerce websites in practice. Aligning our manipulations of e-service recovery technologies with Appendix D thus enhances the generalizability of our empirical findings. Further, as part of our manipulation checks, participants are asked to evaluate the realism of our recovery treatments in order to verify whether the treatments comply with actual practices on contemporary e-commerce websites. Compensation: Compensation has been a standard practice of service recovery in which consumers are reimbursed for any damages they may have suffered as a consequence of failure incidents (Smith et al., 1999). Lovelock and Wirtz (2004) claimed that a stingy compensation may be 74     conceived by consumers as being worse off (or potentially offending) than when no compensation is offered. Past studies on compensation of service failures have manipulated the recovery measure as a discount on the eventual purchase, which ranges anywhere from 20% (Wirtz and Mattila, 2004) to 50% (McCollough et al., 2003). As it is not the intention of this research to quibble over the level of compensation for a given instance of e-service failure, the manipulation will take the form of a selfserving help center (see Figure 6.7) that walks participants through a series of options to diagnose the problem and eventually offers a discount of 15% on the prices of laptops chosen by participants. Given that laptop computers are high-priced products and realistic pricing is adhered to in this experiment, a discount of 15% already translates to a substantial amount of monetary value for participants. Moreover, this discount is over and above the incentive of 50% being awarded in the raffle contest. Because participants stand a chance of being rewarded with a further discount on the laptop price if they were to emerge as winners of the raffle contest, our compensation treatment would be effective as a form e-service recovery as verified in the pretests. Hereafter, we refer to our manipulation of compensation as ‘discount’ for clarity.  75     Figure 6.7: Illustration of Compensatory E-Service Recovery Technologies  Order Details  Compensation  Affinity: Apology communicates respect and empathy to the consumer (Smith et al., 1999). Linguistic researchers have broken down an apology into four primary components, namely remorse, responsibility admission, promise of forbearance and offer of repair (Scher and Darley, 1997). 76     Among the four components, responsibility admission and remorse contains information that is deemed indispensable to an apology as they convey “admission of blameworthiness and regret for an undesirable event” (Darby and Schlenker, 1982, p. 352). Conversely, promises of forbearance boost the effectiveness of apologies through reassuring the victim that the transgression will not be repeated whereas offer of repair relates to the remedial function of an apology by proposing to repair the situation such that it is as if the transgression had not occurred in the first place (Scher and Darley, 1997). It should be noted that an offer of repair is distinct from compensation in that it does not specify the terms of the remedial action. Rather, it solicits the victim’s input on what a viable remedy might be. Table 6.4 illustrates an example of an apology for e-service failures according to the four components. Table 6.4: Example of Apology for Informational Failure Component   Purpose   Expression   Responsibility  Admission   Acknowledge of the violation of normative or ethical  standards of conduct   “An  error  occurred  when  we  tried  to  process  your  order”   Remorse   Expression of remorse or regret over one’s actions   “We apologize for the inconvenience”   Promise of Forbearance   Promise to not repeat one’s transgression   “Rest  assured,  we  are  already  working  on  the  problem to ensure that it does not happen again”   Offer of Repair   Offer to make recompense for one’s actions   “If there is any way we can make it up to you, please  let us know”   Since it has empirically shown that the number of apology components is deterministic of its effectiveness (Schlenker and Darby, 1981), we manipulated affinity through the creation of generic webpages (see Figure 6.8), each comprising all four apology components and catering to one of the three types of e-service failures. Further, we ensure that while the apologies were designed to admit to the existence of e-service failures on the part of the experimental websites, they do not specify the cause of these failures and instead, emphasize a desire for reconciliation. Otherwise, it may draw unnecessary attention to the cause of the failure and prompt failure attributions among participants, thereby confounding the empirical results. Hereafter, we refer to our manipulation of affinity as ‘apology’ for clarity.  77     Figure 6.8: Illustration of Affinity E-Service Recovery Technologies  Apology  Response Sensitivity: Response sensitivity relates to the procedural aspects of recovery in that it is concerned with whether actions have been taken to address the failure incident (Smith et al., 1999). Within the e-commerce domain, response sensitivity can take the form of digitized evaluation forms with several predefined categories for participants to provide feedback on the e-service failure encountered (see Figure 6.9). After which, an automatic response email is generated to acknowledge receipt of participants’ feedback and assure them that actions are being taken to address their feedback. While the response email is instantaneously generated, it is consistent with the empirical results of Smith et al. (1999), who indicated that an immediate response fares much better than a delayed response in alleviating consumers’ dissatisfaction. Giving dissatisfied consumers a chance to voice their concerns with regards to the service failure encountered has been empirically verified to be a salient recovery measure by numerous researchers (e.g., Karande et al., 2007; McColl-Kennedy et al., 2003; Sparks and McColl-Kennedy, 2001). Hereafter, we refer to our manipulation of response sensitivity as ‘feedback’ for clarity. 78     Figure 6.9: Illustration of Response Sensitivity E-Service Recovery Technologies  Problem Selection  HTML Editor for Feedback  79     Automated Response to Feedback  6.1.4  Measurement and Survey Questionnaires  The list of measurement items being employed in this study is summarized in Table 6.1.4.1. Apart from the standard e-service failure, e-service recovery and disconfirmed expectancy constructs, we included controllability as a manipulation check for our failure treatments and realism as a manipulation check for our recovery treatments. Controllability: Controllability is the degree to which a consumer deems the cause of a service failure to be volitional or not (Hess et al., 2003). Attributions of controllability determine consumers’ evaluations regarding the competency of vendors to prevent service failures from happening (Weiner, 2000). Bitner (1990) and Taylor (1994) discovered that negative reactions are more salient in consumers who attribute service failures to vendors’ oversight. Likewise, Leong et al. (1997) found that consumers were more dissatisfied whenever vendors are perceived to exercise substantive control over the cause of service failures. As part of our failure manipulation, it is vital to ensure that participants deem our treatments to be controllable because otherwise, they may not regard it as an e-service failure and discount its impact. 80     Realism: We also verify the realism of our recovery manipulations. For our empirical findings to be valid and generalizable to e-commerce websites, we have to guarantee that our recovery treatments are comparable to what has been offered externally. Because we are interested to determine the extent to which existing recovery technologies on e-commerce websites are applicable to the alleviation of prevalent forms of e-service failures, it is imperative for our recovery treatments to be viewed by participants as identical to that offered by e-merchants. With the exception of controllability, measurement items for the majority of latent variables are newly generated following standard psychometric procedures (Nunnally and Bernstein, 1994). This is not surprising given the novelty of the research topic. The phrasing of the measurement items were refined and validated from three rounds of pilot testing. In developing the measurement items, we distinguished among constructs that were measured in the first survey questionnaire versus those that were measured in the second. Because informational failure, functional failure, system failure and controllability were manipulation checks for failure treatments, they were measured in the first survey questionnaire, but not the second. Conversely, compensation, affinity, response sensitivity and realism were manipulation checks for recovery treatments. Therefore, they were measured in the second survey questionnaire, but not the first. Only disconfirmed expectancy constructs were measured across both survey questionnaires as repeated measures. Further, to reduce instrumentation biases from contaminating the results, measurement items for e-service failure constructs were reverse coded. Because the e-service failure constructs assess the extent to which our treatments for informational, functional and system failures were successful (i.e., primary effects), the reverse coding of the measurement items guards against response bias by preventing participants from speculating about the phenomenon being investigated in the experiment and reacting in a fashion that they deem to be desirable.  81     Table 6.5 List of Measurement Items [All items were measured using a 7-point Likert scale ranging from ‘Strongly Agree’ to ‘Strongly Disagree’] st  Construct   nd  Measures   1  Survey  Questionnaire   2  Survey  Questionnaire   Information provided on the website helps  me  in  obtaining  desired  outcomes  from  the  e‐commerce  transaction.  (Reverse‐ Coded)   Present   Absent   Information  provided  on  the  website  improves  the  outcomes  I  can  attain  from  the  e‐commerce  transaction.  (Reverse‐ Coded)   Present   Absent   Information  provided  on  the  website  is  useful to me in getting preferred outcomes  from  the  e‐commerce  transaction.  (Reverse‐Coded)   Present   Absent   Functionalities  provide  on  the  website  support me in conducting the e‐commerce  transaction. (Reverse‐Coded)   Present   Absent   Functions  provide  on  the  website  make  it  easy  for  me  to  conduct  the  e‐commerce  transaction. (Reverse‐Coded)   Present   Absent   Functions  provided  on  the  website  assist  me  in  conducting  the  e‐commerce  transaction. (Reverse‐Coded)   Present   Absent   Service  content  on  the  website  is  readily  accessible  to  me  when  conducting  the  e‐ commerce transaction. (Reverse‐Coded)   Present   Absent   Service  content  on  the  website  loads  properly when conducting the e‐commerce  transaction. (Reverse‐Coded)   Present   Absent   Service  content  on  the  e‐government  website is easy to access when conducting  the  e‐commerce  transaction.  (Reverse‐ Coded)   Present   Absent   Extent  to  which  an  e‐ commerce  website  offers  services  by  which  consumers  can  seek  reimbursement  for  an  e‐ service failure encountered   I can seek reimbursement for transactional  problems I encountered on the website.   Absent   Present   I  can  seek  to  redress  transactional  problems I encountered on the website.   Absent   Present   I  can  seek  compensation  for  transactional  problems I encountered on the website.   Absent   Present   Extent  to  which  an  e‐ commerce  website  offers  services  that  empathize  to  consumers  following  an  e‐ service failure encounter   Upon  encountering  transactional  problems, the website tries to appease me.   Absent   Present   Upon  encountering  transactional  problems,  the  website  displays  empathy  with my situation.   Absent   Present   Upon   Absent   Present   Definition   E‐Service Failure Constructs [Newly Created]  Informational Failure   Functional Failure   System Failure   Extent to which information  provided  on  an  e‐ commerce  website  is  incapable  of  adding  value  to  consumers  towards  the  achievement  of  their  transactional objectives   Extent  to  which  functionalities  provided  on  an e‐commerce website are  incapable  of  supporting  consumers  along  the  process  of  achieving  their  transactional objectives   Extent  to  which  service  content  (i.e.,  information  and  functionalities)  offered  by  an  e‐commerce  website  is  not  delivered  in  a  efficient  manner  that  facilitates consumers in the  achievement  of  their  transactional objectives   E‐Service Recovery Constructs [Newly Created]  Compensation   Affinity   encountering   transactional   82     st  Construct   nd  1  Survey  Questionnaire   2  Survey  Questionnaire   The website prepares for future problems  based on my feedback on the transactional  problems I encountered.   Absent   Present   The website anticipates the potential  transactional problems I could encounter  by allowing me to provide feedback.   Absent   Present   The website plans for potential  transactional problems I could encounter  by being responsive to my feedback.   Absent   Present   The  transactional  outcome  obtained  from  the  e‐commerce  website  is  worse  than  what I expected.   Present   Present   My  expectations  about  the  transactional  outcome  obtained  from  e‐commerce  website are not met.   Present   Present   The  transactional  outcome  obtained  from  the  e‐commerce  website  is  below  my  expectations.   Present   Present   The  transactional  process  on  the  e‐ commerce  website  is  worse  than  what  I  expected.   Present   Present   My  expectations  about  the  transactional  process  on  the  e‐commerce  website  are  not met.   Present   Present   The  transactional  process  on  the  e‐ commerce  website  is  below  my  expectations.   Present   Present   The amount of effort I spent on transacting  via the e‐commerce website is more than I  anticipated.    Present   Present   I expected the amount of effort I had spent  on transacting via the e‐commerce website  to have been less.   Present   Present   The effort I spent on transacting via the e‐ commerce  website  is  above  my  expectations.   Present   Present   The  transactional  problems  I  have  encountered  are  controllable  by  the  website.   Present   Absent   The  transactional  problems  I  have  encountered  are  preventable  by  the  website.   Present   Absent   The  transactional  problems  I  have  encountered are avoidable by the website.   Present   Absent   The   Absent   Present   Definition   Measures  problems,  the  website  is  sympathetic  to  my plight.   Responsive Sensitivity   Extent  to  which  an  e‐ commerce  website  offers  services  that  anticipate  common  concerns  and  provide  mechanisms  /  guidelines  to  permit  consumers to report and/or  resolve  an  e‐service  failure  encountered   Disconfirmed Expectancy Constructs [Newly Created]  Disconfirmed Outcome  Expectancy   Disconfirmed Process  Expectancy   Disconfirmed Cost  Expectancy   Extent  to  which  the  transactional  outcome  obtained  from  the  e‐ commerce  website  is  undesired by the consumer   Extent  to  which  the  transactional  process  on  the  e‐commerce  website  does  not  proceed  in  a  manner  expected  by  the  consumer   Extent to which a consumer  expends  more  resources  than  anticipated  in  transacting  via  an  e‐ commerce website   Attribution & Realism  Controllability   Realism   Extent  to  which  consumers  view  the  cause  of  an  e‐ service  failure  to  be  volitional  or  not  [as  adapted  from  Hess  et  al.  (2007)]   Extent  to  which  services   website’s   response   to   the   83     st  Construct   6.2  2  Survey  Questionnaire   The  website’s  response  to  the  transactional problems I faced is similar to  what  other  e‐commerce  websites  would  do.   Absent   Present   The  website’s  response  to  the  transactional  problems  I  faced  is  believable.   Absent   Present   Measures   provided by an e‐commerce  website  to  address  e‐ service  failure  are  realistic  and comparable to other e‐ commerce  websites  [Newly  Created]   nd  1  Survey  Questionnaire   Definition   transactional problems I faced is realistic.   Data Analysis Data was analyzed via a blend of ANOVA/MANOVA, one-sample t-tests and Dunnett’s t-  test.  Whereas  ANOVA/MANOVA provides a statistical test of mean differences across treatment groups, one-sample t-tests and Dunnett’s t-test offer a way of comparing means from multiple treatment groups against a reference group or value. Data Sample: A total of 644 participants were recruited for the experiment based on the aforementioned experimental procedures. Of these 644 participants, 69 responses were discarded due to data runs or for being outliners, resulting in an eventual sample of 575 data points for analysis. Table 6.6 summarizes the descriptive statistics for the sample pool of experimental participants. Paired t-tests between our sample and those documented in Cenfetelli et al.’s (2008) survey of 1,235 consumers on the service quality of e-commerce websites reveal no significant difference in demographic distribution (i.e., t(15) = 0.000, p = 1.000). Table 6.6 Descriptive Statistics for Experimental Participants [Sample N = 575] No. of  Respondents [%]   Comparison    E‐Commerce Transactions  Frequency   Male   209 [36.35%]   34%   About once per month   Female   365 [63.48%]   66%   About once per month   1 [0.17%]   0%   Less than once per year   Age 19‐29   78 [13.57%]   10%   About once per month   Age 30‐49   266 [46.26%]   60%   About once per month   Demographic Characteristic   Gender   Unwilling to disclose  Age   84     No. of  Respondents [%]   Comparison    E‐Commerce Transactions  Frequency   193 [33.57%]   28%   About once per month   Age 65+   36 [6.26%]   2%   About once per 3 months   Unwilling to disclose   2 [0.35%]   0%   About once per 3 months   College education or higher   439 [76.35%]   87%   About once per month   Less than college education   134 [23.30%]   13%   About once per 3 months    2 [0.35%]   0%   About once per 3 months   $0‐$29,999   196 [34.09%]   15%   About once per 3 months   $30,000‐$50,000   124 [21.57%]   24%   About once per month   $50,000‐$75,000   118 [20.52%]   28%   About once per month   $75,000+   118 [20.52%]   33%   About once per fortnight   19 [3.30%]   0%   About once per 3 months   Demographic Characteristic   Age 50‐64   Educational Level   Unwilling to disclose  Income   Unwilling to disclose  Cenfetelli et al. (2008)     Measurement Model: All constructs were modeled reflectively. The test of our measurement model involves the estimation of internal consistency as well as the convergent and discriminant validity of the measurement items included in our survey instrument. We assessed the measurement properties of the reflective items in the model using Cronbach’s alpha (Nunnally and Bernstein 1994), composite reliability, and the Average Variance Extracted (AVE) (Fornell and Larcker 1981). As illustrated in Table 6.7, all constructs far exceed recommended thresholds, thus supporting convergent validity. Table 6.7 Inter-Construct Correlation Matrix Composite  Cronbach’s  AVE  Reliability  Alpha  [> 0.50]  [> 0.70]   [> 0.70]       AFF   CP   CTR   FF   IF   PDCE  PDOE  PDPE   DCE   DOE   DPE   REA   RES   SF                                                                                                                                                                                                                                                                                                                   AFF   0.894   0.962   0.941   0.946   CP   0.808   0.927   0.885   0.188  0.899   CTR   0.859   0.948   0.924   0.079  0.016  0.927  FF   0.918   0.971   0.956   0.138  0.162  ‐0.083 0.958  IF   0.936   0.978   0.966   0.147  0.110  ‐0.018 0.516 0.967  PDCE   0.792   0.919   0.870   ‐0.171  ‐0.143 0.192 ‐0.318 ‐0.249 0.890   PDOE   0.910   0.968   0.950   ‐0.253  ‐0.353 0.096 ‐0.365 ‐0.327 0.467  0.954   PDPE   0.957   0.985   0.977   ‐0.290  ‐0.216 0.153 ‐0.413 ‐0.395 0.579  0.736  0.978   85     Composite  Cronbach’s  AVE  Reliability  Alpha  [> 0.50]  [> 0.70]   [> 0.70]       AFF   CP   CTR   FF   IF   PDCE  PDOE  PDPE   DCE   DOE   DPE   REA   RES   SF                                                           DCE   0.757   0.903   0.846   ‐0.142  ‐0.243 0.099 ‐0.284 ‐0.246 0.587  0.454  0.466  0.870   DOE   0.927   0.974   0.961   ‐0.201  ‐0.201 0.220 ‐0.504 ‐0.464 0.576  0.729  0.781  0.487  0.963   DPE   0.949   0.983   0.973   ‐0.200  ‐0.170 0.246 ‐0.466 ‐0.381 0.511  0.609  0.695  0.416  0.839  0.974  REA   0.761   0.904   0.843   0.279  0.183  ‐0.074 0.428 0.406 ‐0.332 ‐0.516 ‐0.568 ‐0.290  ‐0.557  ‐0.502 0.872  RES   0.838   0.939   0.904   0.234  0.165  0.095 0.182 0.187 ‐0.181 ‐0.278 ‐0.285 ‐0.192  ‐0.259  ‐0.150 0.252 0.915  SF   0.845   0.942   0.908   0.144  0.136  ‐0.092 0.507 0.431 ‐0.285 ‐0.344 ‐0.384 ‐0.289  ‐0.481  ‐0.400 0.444 0.180 0.919      AFF  –  Affinity;  CP  –  Compensation;  CTR  –  Controllability;  FF  –  Functional  Failure;  IF  –  Informational  Failure;  PDCE  –  Post  Disconfirmed  Cost  Expectancy;  PDOE  –  Post  Disconfirmed  Outcome  Expectancy;  PDPE  –  Post  Disconfirmed  Process  Expectancy;  DCE  –  Pre  Disconfirmed  Cost  Expectancy;  DOE  –  Pre  Disconfirmed Outcome Expectancy; DPE – Pre Disconfirmed Process Expectancy; REA – Realism; RES – Response Sensitivity; SF – System Failure   Barclay et al. (1995) put forward two criteria for determining discriminant validity. First, the square root of AVE for each construct should be greater than its correlations with any other construct. This indicates that the construct shares more variance with its own measures than it shares with other construct (Fornell and Larcker 1981). As can be seen from the inter-construct correlation matrix in Table 6.7, all construct display sufficient discriminant validity. Second, the factorial loading for an item should be higher for the construct it is supposed to measure than for any other construct. This criterion holds for the constructs in this study as verified by the factorial loadings and cross-loadings of measures shown in Table 6.8. A more stringent guideline was recommended by Gefen and Straub (2005), who maintained that a minimum difference of 0.10 between item loadings and cross-loadings is compulsory to state a claim of discriminant validity. As can be inferred from Table 6.8, all items satisfy this strict guideline for discriminant validity. Table 6.8 Loadings and Cross-Loadings of Measurement Items     AFF   CP   CTR   DCE   DOE   DPE   FF   IF   PDCE   PDOE   PDPE   REA   RES   SF   AFF1   0.911   0.170   0.088   ‐0.103   ‐0.148   ‐0.154   0.065   0.092   ‐0.165   ‐0.211   ‐0.234   0.219   0.168   0.102   AFF2   0.960   0.175   0.074   ‐0.142   ‐0.205   ‐0.195   0.152   0.157   ‐0.161   ‐0.236   ‐0.285   0.273   0.231   0.138   AFF3   0.964   0.188   0.066   ‐0.151   ‐0.207   ‐0.210   0.159   0.156   ‐0.162   ‐0.263   ‐0.295   0.288   0.252   0.161   CP1   0.180   0.893   0.019   ‐0.230   ‐0.221   ‐0.166   0.163   0.135   ‐0.144   ‐0.345   ‐0.235   0.181   0.178   0.171   CP2   0.179   0.911   0.014   ‐0.204   ‐0.143   ‐0.149   0.135   0.074   ‐0.108   ‐0.295   ‐0.163   0.132   0.119   0.086   CP3   0.144   0.893   0.007   ‐0.214   ‐0.158   ‐0.138   0.129   0.072   ‐0.124   ‐0.297   ‐0.166   0.171   0.133   0.086   CTR1   0.095   0.023   0.947   0.042   0.155   0.193   ‐0.055   0.002   0.157   0.054   0.123   ‐0.046   0.114   ‐0.065   CTR2   0.037   0.005   0.922   0.126   0.264   0.271   ‐0.092   ‐0.036   0.220   0.123   0.177   ‐0.097   0.078   ‐0.107   86         AFF   CP   CTR   DCE   DOE   DPE   FF   IF   PDCE   PDOE   PDPE   REA   RES   SF   CTR3   0.062   0.007   0.912   0.149   0.247   0.259   ‐0.102   ‐0.035   0.187   0.124   0.152   ‐0.086   0.055   ‐0.105   DCE1   ‐0.118   ‐0.193   0.085   0.881   0.468   0.389   ‐0.258   ‐0.206   0.526   0.443   0.448   ‐0.252   ‐0.126   ‐0.234   DCE2   ‐0.132   ‐0.262   0.105   0.914   0.461   0.400   ‐0.275   ‐0.244   0.522   0.425   0.441   ‐0.309   ‐0.226   ‐0.306   DCE3   ‐0.122   ‐0.149   0.052   0.812   0.298   0.262   ‐0.186   ‐0.179   0.491   0.280   0.283   ‐0.152   ‐0.126   ‐0.186   DOE1   ‐0.172   ‐0.198   0.194   0.456   0.955   0.781   ‐0.476   ‐0.446   0.548   0.703   0.741   ‐0.551   ‐0.279   ‐0.471   DOE2   ‐0.193   ‐0.192   0.231   0.478   0.972   0.813   ‐0.485   ‐0.448   0.554   0.698   0.756   ‐0.537   ‐0.236   ‐0.467   DOE3   ‐0.216   ‐0.190   0.211   0.472   0.961   0.829   ‐0.496   ‐0.445   0.561   0.703   0.759   ‐0.520   ‐0.232   ‐0.451   DPE1   ‐0.202   ‐0.181   0.210   0.396   0.807   0.969   ‐0.463   ‐0.376   0.483   0.591   0.674   ‐0.490   ‐0.147   ‐0.374   DPE2   ‐0.186   ‐0.144   0.247   0.413   0.822   0.977   ‐0.450   ‐0.369   0.513   0.586   0.679   ‐0.486   ‐0.152   ‐0.394   DPE3   ‐0.196   ‐0.171   0.263   0.409   0.823   0.977   ‐0.447   ‐0.369   0.499   0.602   0.679   ‐0.491   ‐0.140   ‐0.402   FF1   0.124   0.135   ‐0.073   ‐0.262   ‐0.477   ‐0.441   0.959   0.525   ‐0.278   ‐0.341   ‐0.369   0.401   0.157   0.501   FF2   0.146   0.154   ‐0.103   ‐0.290   ‐0.510   ‐0.461   0.962   0.487   ‐0.339   ‐0.366   ‐0.431   0.427   0.186   0.481   FF3   0.126   0.176   ‐0.061   ‐0.264   ‐0.461   ‐0.437   0.953   0.471   ‐0.297   ‐0.342   ‐0.387   0.403   0.179   0.475   IF1   0.131   0.087   ‐0.009   ‐0.237   ‐0.437   ‐0.357   0.497   0.965   ‐0.238   ‐0.303   ‐0.367   0.385   0.186   0.416   IF2   0.150   0.112   ‐0.013   ‐0.230   ‐0.455   ‐0.373   0.498   0.972   ‐0.238   ‐0.333   ‐0.380   0.411   0.191   0.424   IF3   0.145   0.121   ‐0.030   ‐0.248   ‐0.453   ‐0.375   0.502   0.965   ‐0.247   ‐0.312   ‐0.398   0.381   0.166   0.410   PDCE1   ‐0.129   ‐0.104   0.160   0.544   0.520   0.455   ‐0.266   ‐0.235   0.898   0.426   0.504   ‐0.293   ‐0.148   ‐0.288   PDCE2   ‐0.174   ‐0.141   0.198   0.507   0.568   0.490   ‐0.332   ‐0.244   0.908   0.463   0.577   ‐0.348   ‐0.185   ‐0.251   PDCE3   ‐0.149   ‐0.135   0.145   0.524   0.430   0.409   ‐0.237   ‐0.178   0.863   0.342   0.445   ‐0.226   ‐0.144   ‐0.222   PDOE1   ‐0.210   ‐0.325   0.068   0.419   0.679   0.543   ‐0.322   ‐0.276   0.408   0.932   0.648   ‐0.432   ‐0.257   ‐0.309   PDOE2   ‐0.243   ‐0.346   0.095   0.440   0.715   0.601   ‐0.363   ‐0.332   0.467   0.966   0.730   ‐0.518   ‐0.270   ‐0.328   PDOE3   ‐0.267   ‐0.338   0.110   0.441   0.690   0.595   ‐0.358   ‐0.324   0.459   0.963   0.723   ‐0.522   ‐0.269   ‐0.347   PDPE1   ‐0.258   ‐0.204   0.157   0.443   0.752   0.665   ‐0.411   ‐0.398   0.551   0.737   0.974   ‐0.562   ‐0.290   ‐0.378   PDPE2   ‐0.305   ‐0.219   0.148   0.460   0.762   0.684   ‐0.401   ‐0.373   0.568   0.706   0.980   ‐0.557   ‐0.287   ‐0.385   PDPE3   ‐0.288   ‐0.212   0.144   0.464   0.777   0.693   ‐0.400   ‐0.387   0.579   0.715   0.980   ‐0.546   ‐0.260   ‐0.362   REA1   0.293   0.239   ‐0.089   ‐0.296   ‐0.559   ‐0.485   0.426   0.395   ‐0.350   ‐0.555   ‐0.588   0.928   0.267   0.406   REA2   0.145   ‐0.002   ‐0.120   ‐0.125   ‐0.324   ‐0.329   0.277   0.267   ‐0.150   ‐0.208   ‐0.264   0.733   0.087   0.337   REA3   0.262   0.182   ‐0.008   ‐0.299   ‐0.531   ‐0.476   0.396   0.381   ‐0.324   ‐0.509   ‐0.561   0.940   0.261   0.417   RES1   0.230   0.184   0.063   ‐0.185   ‐0.272   ‐0.174   0.191   0.185   ‐0.205   ‐0.283   ‐0.302   0.291   0.909   0.191   RES2   0.224   0.135   0.108   ‐0.163   ‐0.204   ‐0.111   0.165   0.169   ‐0.130   ‐0.236   ‐0.232   0.182   0.912   0.143   RES3   0.183   0.128   0.095   ‐0.178   ‐0.227   ‐0.119   0.136   0.156   ‐0.155   ‐0.239   ‐0.241   0.206   0.924   0.156   SF1   0.173   0.141   ‐0.040   ‐0.291   ‐0.430   ‐0.352   0.506   0.443   ‐0.289   ‐0.323   ‐0.374   0.408   0.182   0.932   SF2   0.096   0.127   ‐0.153   ‐0.275   ‐0.468   ‐0.391   0.427   0.365   ‐0.281   ‐0.336   ‐0.340   0.400   0.155   0.911   SF3   0.131   0.105   ‐0.057   ‐0.230   ‐0.426   ‐0.359   0.465   0.380   ‐0.214   ‐0.289   ‐0.344   0.415   0.160   0.914   AFF  –  Affinity;  CP  –  Compensation;  CTR  –  Controllability;  FF  –  Functional  Failure;  IF  –  Informational  Failure;  PDCE  –  Post  Disconfirmed  Cost  Expectancy;  PDOE  –  Post  Disconfirmed  Outcome  Expectancy;  PDPE  –  Post  Disconfirmed  Process  Expectancy;  DCE  –  Pre  Disconfirmed  Cost  Expectancy;  DOE  –  Pre Disconfirmed Outcome Expectancy; DPE – Pre Disconfirmed Process Expectancy; REA – Realism; RES – Response Sensitivity; SF – System Failure   87     6.2.1  Manipulations of E­Service Failures  Manipulation checks were performed for our failure treatments. To begin, we spilt the entire sample into the four failure treatments and the descriptive statistics for each treatment condition are summarized in Table 6.9. Table 6.9 Descriptive Statistics for E-Service Failure Constructs Dependent Variable  Informational Failure   Functional Failure   System Failure   Failure Treatment   2  N   Mean    Std. Deviation   23  2.2174  1.04803  Out‐of‐Stock   184  4.2532  1.23079  Missing Comparison   184  2.5527  1.35368  Delay   184  2.4128  1.25157  Total   575  3.0387  1.51981  23  2.3483  1.05638  Out‐of‐Stock   184  2.8115  1.50415  Missing Comparison   184  4.2320  1.20897  Delay   184  2.5070  1.35145  Total   575  3.1501  1.54358  23  2.2752  1.00891  Out‐of‐Stock   184  2.8442  1.33520  Missing Comparison   184  2.8132  1.41123  Delay   184  4.2107  1.08945  Total   575  3.2488  1.43825  No Failure   No Failure   No Failure   ANOVA results indicate statistically significant differences in informational failure [F(3) = 82.96; p = .000], functional failure [F(3) = 59.89; p = .000] and system failure [F(3) = 52.59; p = .000] among participants assigned to different failure treatments (see Table 6.10). This implies that there are substantial differences in how participants perceive the presence of informational, functional and system failures across the four treatment conditions. Table 6.10 ANOVA Results for E-Service Failure Constructs [Manipulation Checks] Dependent Variable  Failure Treatment  Type III Sum of Squares  Informational Failure   df   Mean Square   Between Groups   402.462  3  134.154  Within Groups   923.371  571  1.617  1325.834  574  Total   F   Sig.   82.959  .000                                                                2  Because measurement items for e-service failure constructs were reverse coded, a higher mean would correspond to participants’ acknowledgement of the presence of an e-service failure. 88     Dependent Variable  Failure Treatment  Type III Sum of Squares  Functional Failure   System Failure   Between Groups   df   Mean Square   327.339  3  109.113  Within Groups   1040.287  571  1.822  Total   1367.626  574  Between Groups   257.058  3  85.686  Within Groups   930.296  571  1.629  1187.354  574  Total   F   Sig.   59.891  .000   52.593  .000   Although our ANOVA results detect significant variations among failure treatments with regards to participants’ perception of the presence of informational, functional and system failures, the analysis is unable to pinpoint the exact treatment(s) that gives rise to these perceptual differences. Dunnett’s t-test was thus conducted to compare perceptual differences between participants assigned to each of the three failure treatments and those allocated to the control group (see Table 6.11). Table 6.11 substantiates our failure manipulations through three observations: (1) participants assigned to the out-of-stock treatment condition reported a greater presence of informational failure as compared to the control group; (2) participants assigned to the missing comparison treatment condition reported a greater presence of functional failure as compared to the control group, and; (3) participants assigned to the delay treatment condition reported a greater presence of system failure as compared to the control group. It is also observable from Table 6.11 that there are traces of negative spillover effects from each treatment condition: each failure treatment generally results in a worse off evaluation for all aspects (i.e., informational, functional and system) of the service encounter as compared to the control group. This is especially true for the system aspects of the experimental websites whereby participants assigned to the out-of-stock and missing comparison treatment conditions reported a slightly greater presence of system failure as compared to the control group. Table 6.11 Dunnett t-Test (2-sided)a for E-Service Failure Constructs Dependent Variable  Informational Failure   (I) Failure Treatment   (J) Control   Mean Difference (I‐J)   Sig.   Out‐of‐Stock   No Failure   *  .000  Missing Comparison   No Failure   .33527  .351  Delay   No Failure   .19543  .671  2.03582  89     Dependent Variable   (I) Failure Treatment   (J) Control   Mean Difference (I‐J)   Sig.   Functional Failure   Out‐of‐Stock   No Failure   .46321  .194  Missing Comparison   No Failure   *  .000  Delay   No Failure   .15875  .793  No Failure   +  .076  +  .097  *  .000  System Failure   Out‐of‐Stock  Missing Comparison  Delay   No Failure  No Failure   1.88370  .56902 .53799 1.93543  a. Dunnett t‐tests treat one group as a control, and compare all other groups against it.  *. The mean difference is significant at the 0.05 level.  +  . The mean difference is significant at the 0.10 level.   The validity of our failure manipulations is further corroborated by conducting Dunnett’s T3 test, which compares and contrasts the four treatment conditions against one another (see Appendix E). As illustrated in Appendix E, each failure manipulation (i.e., out-of-stock, missing comparison and delay) has a statistically significant effect on participants’ evaluation of its corresponding failure construct as compared to other treatment conditions. Next, we performed a manipulation check on participants’ perception of the controllability of our failure manipulations. Table 6.12 summarizes the descriptive statistics for controllability according to the different failure treatments. Table 6.12 Descriptive Statistics for Controllability Failure Treatment   N   Mean   Std. Deviation   Out‐of‐Stock   184  3.2954   1.31900  Missing Comparison   184  3.3912   1.24015  Delay   184  3.5054   1.25163  Total   552  3.3974   1.27133  We conducted a one-sample t-test against the neutral pivot value of 4.0 in our measurement scale (see Table 6.13): we tested the null hypothesis that controllability is unrelated to e-service failures. As illustrated in Table 6.13, participants assigned to the out-of-stock [t(183) = -7.246; p = .000], missing comparison [t(183) = -6.659; p = .000] and delay [t(183) = -5.360; p = .000] treatment conditions reported a statistically significant mean difference in controllability, thereby attesting to participants’ belief that our failure manipulations are controllable and can be avoided. 90     Table 6.13 One-Sample t-Test for Controllability Test Value = 4.000  Failure Treatment  t   df   Sig. (2‐tailed)   Mean Difference   Out‐of‐Stock   ‐7.246   183   .000  ‐.70457  Missing Comparison   ‐6.659   183   .000  ‐.60880  Delay   ‐5.360   183   .000  ‐.49457  Conversely, ANOVA results indicate statistically insignificant differences among failure treatments with regards to participants’ perception of controllability [F(2) = 1.260; p = .285] (see Table 6.14). This implies that participants do not regard one particular failure manipulation as being more controllable than others. Table 6.14 ANOVA Results for Controllability Failure Treatment  Type III Sum of Squares  Between Groups   df   Mean Square   4.068   2  2.034  Within Groups   886.503   549  1.615  Total   890.570   551  F   Sig.   1.260  .285  Our ANOVA results are further corroborated based on Dunnett’s T3 test for controllability among failure treatments, which reveal insignificant mean differences across treatment conditions (see Table 6.15). Table 6.15 Dunnett T3 Test for Controllability (I) Failure Treatment   (J) Failure Treatment   Out‐of‐Stock   Missing Comparison   Mean Difference (I‐J)   Sig.   Missing Comparison   ‐.09576  .854  Delay   ‐.21000  .314  .09576  .854  ‐.11424  .761  Out‐of‐Stock   .21000  .314  Missing Comparison   .11424  .761  Out‐of‐Stock  Delay   Delay   6.2.2  Manipulations of E­Service Recoveries  Like e-service failures, manipulation checks were performed for our recovery treatments. Table 6.16 summarizes the descriptive statistics for each of the eight recovery treatment condition.  91     Table 6.16 Descriptive Statistics for E-Service Recovery Constructs Dependent Variable  Compensation   Recovery Treatment   N   Mean   Std. Deviation   No Recovery   69  4.9180  1.48686  Feedback   69  4.5703  1.37225  Apology   69  4.6335  1.36414  Apology x Feedback   69  4.6475  1.50609  Discount   69  2.3722  .93190  Discount x Feedback   69  2.4203  .94941  Discount x Apology   69  2.4730  .95557  Discount x Apology x Feedback   69  2.5413  1.00678  552  3.5720  1.65508  No Recovery   69  4.4539  1.70836  Feedback   69  4.0001  1.76222  Apology   69  2.7006  .94484  Apology x Feedback   69  2.2894  1.04928  Discount   69  3.4296  1.97331  Discount x Feedback   69  3.7346  2.17624  Discount x Apology   69  2.3720  .97909  Discount x Apology x Feedback   69  2.0628  1.03113  552  3.1304  1.73249  69  4.1065  1.83361  Feedback   69  2.7775  .97148  Apology   69  4.2758  1.52770  Apology x Feedback   69  2.4058  1.02557  Discount   69  3.6136  1.73377  Discount x Feedback   69  2.3671  .98234  Discount x Apology   69  3.6761  1.88099  Discount x Apology x Feedback   69  2.4541  1.04006  552  3.2096  1.60153  Total  Affinity   Total  Response Sensitivity  No Recovery   Total   MANOVA was performed to detect whether between-subject differences exist across different recovery treatment conditions (see Table 6.17). A series of observations can be made. First, there are statistically significant differences in compensation [F(1) = 501.32; p = .000], affinity [F(1) = 11.50; p = .001] and response sensitivity [F(1) = 7.97; p = .005] for the discount treatment condition. This indicates that participants assigned to the discount treatment condition  92     perceived the presence of compensation, affinity and response sensitivity differently from those who have not been exposed to the same recovery treatment. Second, there is a statistically significant difference in affinity [F(1) = 358.41; p = .000], but not in compensation [F(1) = 0.15; p = .695] and response sensitivity [F(1) = 0.00; p = .960] for the apology treatment condition. This indicates that participants assigned to the apology treatment condition perceived the presence of affinity differently from those who have not been exposed to the same recovery treatment. Third, there is a statistically significant difference in response sensitivity [F(1) = 156.19; p = .000], but not in compensation [F(1) = 0.00; p = .985] and affinity [F(1) = 1.81; p = .179] for the feedback treatment condition. This indicates that participants assigned to the feedback treatment condition perceived the presence of response sensitivity differently from those who have not been exposed to the same recovery treatment. Finally, none of the interaction effects are statistically significant. Combining the four observations, it would seem to substantiate the validity of our recovery treatments. Table 6.17 Tests of Between-Subjects Effects for E-Service Recovery Constructs [Recovery Treatments  E-Service Recovery Constructs] Type III Sum of  Squares   df   Mean Square   715.575  1  Affinity   25.655  Response Sensitivity   Recovery Treatment   Dependent Variable   Discount   Compensation   Apology   715.575  501.316  .000   .469   1.000  1  25.655  11.500  .001   .020   .923  15.489  1  15.489  7.965  .005   .014   .804  .220  1  .220  .154  .695   .000   .068  358.407  1  358.407  160.657  .000   .221   1.000  Response Sensitivity   .005  1  .005  .002  .960   .000   .050  Compensation   .001  1  .001  .001  .985   .000   .050  4.041  1  4.041  1.812  .179   .003   .269  303.725  1  303.725  156.191  .000   .216   1.000  .062  1  .062  .044  .835   .000   .055  4.650  1  4.650  2.085  .149   .004   .302  Response Sensitivity   .816  1  .816  .420  .517   .001   .099  Compensation   .263  1  .263  .184  .668   .000   .071  5.568  1  5.568  2.496  .115   .004   .351  Compensation   Affinity  Response Sensitivity  Discount x Apology   Compensation  Affinity   Discount x Feedback   Observed  Power   Sig.   Affinity   Feedback   Partial Eta  Squared   F   Affinity   93     Recovery Treatment   Apology x Feedback   Discount x Apology x  Feedback   Dependent Variable   Type III Sum of  Squares   df   Mean Square   Response Sensitivity   5.263  1  .012  Affinity  Response Sensitivity   Partial Eta  Squared   Observed  Power   F   Sig.   5.263  2.706  .100   .005   .376  1  .012  .008  .928   .000   .051  2.556  1  2.556  1.146  .285   .002   .188  3.039  1  3.039  1.563  .212   .003   .239  .354  1  .354  .248  .619   .000   .079  Affinity   2.140  1  2.140  .959  .328   .002   .165  Response Sensitivity   3.354  1  3.354  1.725  .190   .003   .259  Compensation   Compensation   Results from our Dunnett’s t-test (see Table 6.18) and T3 test (see Appendix F) further corroborate our recovery manipulations in that participants’ perception of a recovery construct coincides with its existence in a specific recovery treatment. As illustrated in Table 6.18, participants assigned to treatment conditions containing discount, apology and feedback reported a greater presence of compensation, affinity and response sensitivity respectively as compared to the control group. Although our Dunnett’s t-test results reveal that there are other recovery treatments (i.e., discount and discount x feedback) besides those involving apology that give rise to participants’ perception of the presence of affinity (see Table 6.18), our Dunnett’s T3 test (see Appendix F) indicate that such perceptions are still stronger for the apology treatment condition as compared to any others. Table 6.18 Dunnett t-Test (2-sided)a for E-Service Recovery Constructs Dependent Variable  (I) Type of Recovery   (J) Control   Mean Difference (I‐J)   Sig.   Compensation   Feedback   No Recovery   ‐.34768  .387  Apology   No Recovery   ‐.28449  .602  Apology x Feedback   No Recovery   ‐.27043  .653  No Recovery   *  .000  *  .000  *  .000  *  .000  Discount  Discount x Feedback  Discount x Apology   Affinity   No Recovery  No Recovery   ‐2.54580  ‐2.49768 ‐2.44493  Discount x Apology x Feedback   No Recovery   Feedback   No Recovery   ‐.45377  .344  No Recovery   *  .000  *  .000  *  .001  Apology  Apology x Feedback  Discount   No Recovery  No Recovery   ‐2.37667  ‐1.75333  ‐2.16449 ‐1.02435  94     Dependent Variable  (I) Type of Recovery   (J) Control   Mean Difference (I‐J)   Sig.   *  .034  *  .000  *  .000  *  .000  No Recovery   .16928  .974  Apology x Feedback   No Recovery   *  .000  Discount   No Recovery   ‐.49290  .201  Discount x Feedback   No Recovery   *  .000  Discount x Apology   No Recovery   ‐.43043  .328  No Recovery   *  .000  Discount x Feedback   No Recovery   ‐.71928  Discount x Apology   No Recovery   ‐2.08188  Discount x Apology x Feedback   No Recovery   ‐2.39116  Response Sensitivity  Feedback   No Recovery   Apology   Discount x Apology x Feedback   ‐1.32899  ‐1.70072  ‐1.73942  ‐1.65246  a. Dunnett t‐tests treat one group as a control, and compare all other groups against it.  *. The mean difference is significant at the 0.05 level.   Table 6.19 summarizes the descriptive statistics for realism according to the eight recovery treatments. Table 6.19 Descriptive Statistics for Realism  Recovery Treatment   N   Mean   Std. Deviation   Feedback   69   2.9810  1.22998  Apology   69   3.1741  1.17789  Apology x Feedback   69   2.8941  1.32922  Discount   69   3.1883  1.24265  Discount x Feedback   69   3.3141  1.33191  Discount x Apology   69   2.8307  1.09001  Discount x Apology x Feedback   69   3.0678  1.05848  483   3.0643  1.21574  Total   To investigate participants’ perception of the realism of our recovery manipulations, we conducted a one-sample t-test against the neutral pivot value of 4.0 in our measurement scale (see Table 6.20): we tested the null hypothesis that realism is unrelated to e-service recoveries. As illustrated in Table 6.20, participants assigned to the discount [t(275) = -12.530; p = .000], apology [t(275) = -14.315; p = .000] and feedback [t(275) = -12.482; p = .000] treatment conditions reported a statistically significant mean difference in realism, thereby suggesting that participants’ believe the recovery manipulations to be realistic and comparable to what has been offered in reality. 95     Table 6.20 One-Sample t-Test for Realism Test Value = 4.000  Recovery Treatment  t   df   Sig. (2‐tailed)   Mean Difference   Discount   ‐12.530   275   .000  ‐.89978  Apology   ‐14.315   275   .000  ‐1.00833  Feedback   ‐12.482   275   .000  ‐.93576  Conversely, ANOVA results indicate statistically insignificant differences among recovery treatments with regards to participants’ perception of realism [F(6) = 1.410; p = .209] (see Table 6.21). This implies that participants do not regard one particular recovery manipulation as being more realistic than others. Table 6.21 ANOVA Results for Realism Recovery Treatment  Type III Sum of Squares  Between Groups   df   Mean Square   12.439   6  2.073  Within Groups   699.974   476  1.471  Total   712.413   482  F   Sig.   1.410  .209  Our ANOVA results are also corroborated based on Dunnett’s T3 test for realism among recovery treatments, which reveal insignificant mean differences across treatment conditions (see Table 6.22). Table 6.22 Dunnett T3 Test for Realism (I) Type of Recovery   (J) Type of Recovery   Response Sensitivity   Apology   Sig.   ‐.19304  1.000   .08696  1.000   Discount   ‐.20725  1.000   Discount x Feedback   ‐.33304  .937   Discount x Apology   .15029  1.000   ‐.08681  1.000   Feedback   .19304  1.000   Apology x Feedback   .28000  .986   Discount   ‐.01420  1.000   Discount x Feedback   ‐.14000  1.000   Discount x Apology   .34333  .803   Discount x Apology x Feedback   .10623  1.000   Apology x Feedback   Discount x Apology x Feedback  Apology   Mean Difference (I‐J)   96     (I) Type of Recovery   (J) Type of Recovery   Apology x Response Sensitivity   Feedback   ‐.08696  1.000   Apology   ‐.28000  .986   Discount   ‐.29420  .982   Discount x Feedback   ‐.42000  .746   Discount x Apology   .06333  1.000   ‐.17377  1.000   Feedback   .20725  1.000   Apology   .01420  1.000   Apology x Feedback   .29420  .982   Discount x Feedback   ‐.12580  1.000   Discount x Apology   .35754  .789   Discount x Apology x Feedback   .12043  1.000   Feedback   .33304  .937   Apology   .14000  1.000   Apology x Feedback   .42000  .746   Discount   .12580  1.000   Discount x Apology   .48333  .353   Discount x Apology x Feedback   .24623  .995   Feedback   ‐.15029  1.000   Apology   ‐.34333  .803   Apology x Feedback   ‐.06333  1.000   Discount   ‐.35754  .789   Discount x Feedback   ‐.48333  .353   Discount x Apology x Feedback   ‐.23710  .988   Feedback   .08681  1.000   Apology   ‐.10623  1.000   .17377  1.000   Discount   ‐.12043  1.000   Discount x Feedback   ‐.24623  .995   Discount x Apology   .23710  .988   Discount x Apology x Feedback  Discount   Discount x Response Sensitivity   Discount x Apology   Discount x Apology x Response Sensitivity   Apology x Feedback   Mean Difference (I‐J)   Sig.   6.2.3  Hypotheses Testing  This section describes the results from our testing of hypotheses in the theoretical model. Impact of E-Service Failures: As discovered in our first study, e-service failures are detrimental to online transactions through disconfirming consumers’ expectations about transactional outcome, process and/or cost. Therefore, the first stage of our hypotheses testing is to validate the 97     causal relationships between e-service failures and their negative consequences as derived in our first study. Table 6.23 summarizes the descriptive statistics for the various disconfirmed expectancy constructs according to the four failure treatment conditions. Table 6.23 Descriptive Statistics for Disconfirmed Expectancy Constructs [by Failure Treatments] Dependent Variable  Disconfirmed Outcome Expectancy   Disconfirmed Process Expectancy   Disconfirmed Cost Expectancy   Failure Treatment  No Failure   N   Mean   Std. Deviation   23  5.7530  1.01194   Out‐of‐Stock   184  3.9510  1.43448   Missing Comparison   184  4.7318  1.50839   Delay   184  4.9584  1.51330   Total   575  4.5953  1.54460   23  5.8404  .91995   Out‐of‐Stock   184  4.5581  1.66356   Missing Comparison   184  4.1486  1.32778   Delay   184  4.8786  1.58636   Total   575  4.5809  1.55922   23  5.1304  1.17979   Out‐of‐Stock   184  4.4113  1.26906   Missing Comparison   184  4.5326  1.14213   Delay   184  4.0361  1.29688   Total   575  4.3588  1.25964   No Failure   No Failure   To investigate the impact of informational, functional and system failures on participants’ disconfirmed expectancies (i.e., hypotheses 1, 2 and 3), we performed ANOVA to compare betweensubject differences among the four treatment conditions (see Table 6.24). Results indicate that there exist substantial differences in participants’ evaluation of disconfirmed outcome [F(3) = 20.80; p = .000], process [F(3) = 12.70; p = .000] and cost [F(3) = 8.50; p = .000] expectancy across the four treatment conditions.  98     Table 6.24 ANOVA Test of Between-Subjects Effects [Failure Treatments  Disconfirmed Expectancies] Type III Sum of  Squares   df   Mean Square   134.893  3  Disconfirmed Process Expectancy   87.275  Disconfirmed Cost Expectancy   38.915  Source   Dependent Variable   Failure  Treatments   Disconfirmed Outcome Expectancy   Partial Eta  Squared   Observed  b Power    .000   .099  1.000  12.698  .000   .063  1.000  8.496  .000   .043  .994  F   Sig.   44.964  20.797  3  29.092  3  12.972  The next step is to conduct the Dunnett’s t-test to compare between-subject differences for each disconfirmed expectancy construct between participants assigned to each of the three failure treatments and those allocated to the control group. Results of the Dunnett’s t-test are summarized in Table 6.25. From Table 6.25, participants assigned to out-of-stock, missing comparison and delay treatment conditions reported higher levels of disconfirmed outcome, process and cost expectancies as compared to the control group. Yet, a closer inspection of participants’ evaluation of disconfirmed expectancies based on multi-group comparisons via Dunnett’s T3 test reveal that there is a dominant effect for each failure treatment (see Appendix G). Whereas all three failure treatments (i.e., out-ofstock, missing comparison and delay) impacts consumers’ disconfirmed outcome expectancy as compared to the control group, the effect is the strongest for the out-of-stock treatment condition. The same goes for the failure treatments of missing comparison and delay: missing comparison has the strongest impact on participants’ evaluation of disconfirmed process expectancy whereas delay has the strongest impact on participants’ evaluation of disconfirmed cost expectancy. These results support hypotheses 1, 2 and 3. Table 6.25 Dunnett t-Test (2-sided)a for Disconfirmed Expectancy Constructs [by Failure Treatments] Dependent Variable  Disconfirmed Outcome Expectancy   (I) Failure Treatment  Out‐of‐Stock  Missing Comparison  Delay   Disconfirmed Process Expectancy   Out‐of‐Stock  Missing Comparison  Delay   (J) Control   Mean Difference (I‐J)   No Failure   ‐1.8020  *  .000   ‐1.0213  *  .004   *  .027   ‐1.2823  *  .000   ‐1.6918  *  .000   *  .008   No Failure  No Failure  No Failure  No Failure  No Failure   ‐.7946  ‐.9618  Sig.   99     Dependent Variable  Disconfirmed Cost Expectancy   (I) Failure Treatment  Out‐of‐Stock   (J) Control   Mean Difference (I‐J)   No Failure   *  .017   +  .052   *  .000   ‐.7192  Missing Comparison   No Failure   ‐.5978  Delay   No Failure   ‐1.0943  Sig.   a. Dunnett t‐tests treat one group as a control, and compare all other groups against it.  *. The mean difference is significant at the 0.05 level.  +  . The mean difference is significant at the 0.10 level.   Impact of E-Service Recoveries: The remainder of this section will focus on the impact of eservice recoveries on participants’ disconfirmed outcome, process and cost expectancies. Table 6.26 summarizes the descriptive statistics for the various pre- and post- disconfirmed expectancy constructs according to the eight recovery treatments. Table 6.26 Descriptive Statistics for Disconfirmed Expectancy Constructs [by Recovery Treatments] Dependent Variable  Disconfirmed Outcome Expectancy   Type of Recovery   Std. Deviation   69  4.5848  1.53202   Apology   69  4.4539  1.58651   Apology x Feedback   69  4.7052  1.80556   Discount   69  4.9035  1.40525   Discount x Feedback   69  4.3190  1.62535   Discount x Apology   69  4.7491  1.63100   Discount x Apology x Feedback   69  4.7051  1.38555   483  4.6315  1.57369   Feedback   69  4.6088  1.52743   Apology   69  4.4348  1.51499   Apology x Feedback   69  4.8406  1.76722   Discount   69  5.4930  1.17097   Discount x Feedback   69  5.3765  1.52527   Discount x Apology   69  5.6670  1.12597   Discount x Apology x Feedback   69  5.5703  1.06361   483  5.1416  1.47315   Feedback   69  4.5800  1.64757   Apology   69  4.5268  1.57860   Apology x Feedback   69  4.3332  1.83901   Discount   69  4.9906  1.33440   Discount x Feedback   69  4.4054  1.70455   Discount x Apology   69  4.5943  1.47057   Total  Disconfirmed Process Expectancy   Mean   Feedback   Total  Post Disconfirmed Outcome Expectancy   N   100     Dependent Variable   Type of Recovery   Mean   Std. Deviation   69  4.4929  1.46039   483  4.5605  1.58653   Feedback   69  4.8935  1.58572   Apology   69  4.5409  1.55011   Apology x Feedback   69  5.1690  1.63756   Discount   69  5.0533  1.42770   Discount x Feedback   69  5.0435  1.69485   Discount x Apology   69  4.7970  1.41332   Discount x Apology x Feedback   69  5.2946  1.30703   483  4.9702  1.53026   Feedback   69  4.1935  1.34136   Apology   69  4.1299  1.40378   Apology x Feedback   69  4.1642  1.35495   Discount   69  4.5894  1.07116   Discount x Feedback   69  4.2706  1.37138   Discount x Apology   69  4.6330  1.03714   Discount x Apology x Feedback   69  4.4686  1.04424   483  4.3499  1.24964   Feedback   69  4.0532  1.31966   Apology   69  4.0870  1.29823   Apology x Feedback   69  4.2372  1.52221   Discount   69  4.2901  1.32220   Discount x Feedback   69  4.2078  1.53933   Discount x Apology   69  4.2951  1.39960   Discount x Apology x Feedback   69  4.2703  1.03428   483  4.2058  1.35160   Discount x Apology x Feedback  Total  Post Disconfirmed Process Expectancy   Total  Disconfirmed Cost Expectancy   Total  Post Disconfirmed Cost Expectancy   Total   N   To investigate the impact of e-service recoveries on participants’ evaluation of disconfirmed expectancies, we performed repeated measures MANOVA to compare within-subject differences between pre- and post- disconfirmed expectancies for the seven recovery treatment conditions (see Table 6.27). A series of observations can be made from Table 6.27. First, there are statistically significant differences in participants’ evaluation of disconfirmed outcome [F(1) = 103.66; p = .000], process [F(1) = 66.25; p = .000] and cost [F(3) = 7.58; p = .006]  101     expectancy across time. This indicates that the introduction of e-service recovery in general cause variations to participants’ evaluation of disconfirmed expectancies. Second, there are statistically significant differences in participants’ evaluation of disconfirmed outcome [F(1) = 88.00; p = .000] and cost [F(1) = 4.63; p = .032] expectancy, but not in their disconfirmed process [F(1) = 2.36; p = .125] expectancy for the discount treatment condition. This indicates that participants assigned to the discount treatment condition perceived the disconfirmation of their outcome and cost expectancy differently as compared to those who have not been exposed to the same recovery treatment. Third, there is a statistically significant difference in participants’ evaluation of disconfirmed process [F(1) = 5.68; p = .017] expectancy, but in their disconfirmed outcome [F(1) = 0.42; p = .516] and cost [F(1) = 0.00; p = .992] expectancy for the apology treatment condition. This indicates that participants assigned to the apology treatment condition perceived the disconfirmation of their process expectancy differently as compared to those who have not been exposed to the same recovery treatment. Fourth, there is a statistically significant difference in participants’ evaluation of disconfirmed process [F(1) = 42.95; p = .000] expectancy, a marginally significant difference in their disconfirmed outcome [F(1) = 2.87; p = .091] expectancy and no difference in disconfirmed cost [F(1) = 0.92; p = .337] expectancy for the feedback treatment condition. This indicates that participants assigned to the feedback treatment condition perceived the disconfirmation of their process and outcome expectancy differently as compared to those who have not been exposed to the same recovery treatment Finally, even though most interaction effects involving multiple e-service recoveries are statistically insignificant, there is a marginally significant difference in participants’ evaluation of disconfirmed outcome [F(1) = 3.45; p = .064] expectancy, but not in their disconfirmed process [F(1) = 1.89; p = .017] and cost [F(1) = 0.93; p = .335] expectancy for the discount x apology x feedback 102     treatment condition. This indicates that participants assigned to the treatment condition involving all three e-service recoveries perceived the disconfirmation of their outcome expectancy differently as compared to those who have not been exposed to the same recovery treatment. Table 6.27 Tests of Within-Subjects Contrasts for Disconfirmed Expectancy Constructs [Recovery Treatments  Disconfirmed Expectancies] Type III Sum of  Squares   df   Mean Square   Disconfirmed Outcome  Linear  Expectancy   56.749  1  Disconfirmed Process  Expectancy   Linear   36.630  Disconfirmed Cost  Expectancy   Linear   Source   Measure   Time   Time * Discount   Time * Apology   Time * Feedback   Time   Disconfirmed Outcome  Linear  Expectancy  Disconfirmed Process  Expectancy   Linear   Disconfirmed Cost  Expectancy   Linear   Disconfirmed Outcome  Linear  Expectancy  Disconfirmed Process  Expectancy   Linear   Disconfirmed Cost  Expectancy   Linear   Disconfirmed Outcome  Linear  Expectancy  Disconfirmed Process  Expectancy   Linear   Disconfirmed Cost  Expectancy   Linear   Time * Discount  *   Disconfirmed Outcome  Linear  Apology  Expectancy  Disconfirmed Process  Expectancy   Linear   Disconfirmed Cost  Expectancy   Linear   Time * Discount  *   Disconfirmed Outcome  Linear  Feedback  Expectancy   Time * Apology  *   Feedback   Disconfirmed Process  Expectancy   Linear   Disconfirmed Cost  Expectancy   Linear   Disconfirmed Outcome  Linear  Expectancy   Partial Eta  Squared   Observed  a Power    .000   .155  1.000  66.253   .000   .105  1.000  4.527  7.581   .006   .013  .785  1  48.180  88.004   .000   .134  1.000  1.307  1  1.307  2.364   .125   .004  .336  2.765  1  2.765  4.630   .032   .008  .575  .232  1  .232  .423   .516   .001  .100  3.140  1  3.140  5.679   .017   .010  .662  6.321E‐5  1  6.321E‐5  .000   .992   .000  .050  1.570  1  1.570  2.868   .091   .005  .394  23.748  1  23.748  42.953   .000   .070  1.000  .551  1  .551  .923   .337   .002  .160  .009  1  .009  .016   .901   .000  .052  .241  1  .241  .437   .509   .001  .101  .529  1  .529  .885   .347   .002  .156  .250  1  .250  .456   .500   .001  .103  .007  1  .007  .012   .911   .000  .051  .714  1  .714  1.196   .275   .002  .194  .678  1  .678  1.238   .266   .002  .199  F   Sig.   56.749  103.655   1  36.630  4.527  1  48.180  103     Source   Measure   Time   Disconfirmed Process  Expectancy   Linear   Disconfirmed Cost  Expectancy   Linear   Type III Sum of  Squares   df   Mean Square   1.260  1  .113  Time * Discount  *   Disconfirmed Outcome  Linear  Apology  *   Expectancy  Feedback  Disconfirmed Process  Linear  Expectancy  Disconfirmed Cost  Expectancy   Linear   Partial Eta  Squared   Observed  a Power    .132   .004  .326  .189   .664   .000  .072  1.887  3.446   .064   .006  .458  1  1.045  1.890   .170   .003  .279  1  .555  .929   .335   .002  .161  F   Sig.   1.260  2.279   1  .113  1.887  1  1.045 .555  Although our MANOVA results detect significant variations among recovery treatments with regards to participants’ evaluation of the disconfirmed expectancies, the analysis does not yield insights into whether these e-service recoveries alleviate or worsen failure consequences. To examine the impact of e-service recovery in general on participants’ evaluation of disconfirmed expectancies, we conducted a one-sample t-test that compares within-subject differences against the value of 0.0 (see Table 6.28): we test the null hypothesis that e-service recoveries in general neither alleviates nor worsens failure consequences. As illustrated in Table 6.28, participants exposed to e-service recoveries in general reported an improvement in their disconfirmed outcome [t(482) = 9.25; p = .000] and process [t(482) = 7.66; p = .000] expectancy, but experienced a further boost to their disconfirmed cost [t(482) = -2.72; p = .000] expectancy. This partially supports hypothesis 4. Table 6.28 One-Sample t-Test for E-Service Recovery Treatment Test Value = 0.000  Dependent Variable  t   df   Disconfirmed Outcome Expectancy Difference   9.251  482  .000  .50994  Disconfirmed Process Expectancy Difference   7.664  482  .000  .41551  ‐2.717  482  .007  ‐.15025  Disconfirmed Cost Expectancy Difference   Sig. (2‐tailed)  Mean Difference   The same analytical procedures were applied to each of the seven recovery treatments and the results are summarized in Table 6.29. From Table 6.29, we can make four deductions. One, disconfirmed outcome expectancy can be improved through the provision of e-service recoveries 104     involving discount [t(68) = 4.30; p = .000], discount and feedback [t(68) = 5.78; p = .000], discount and apology [t(68) = 5.40; p = .000] or all three (i.e., discount, apology and feedback) [t(68) = 5.86; p = .000]. Two, disconfirmed process expectancy can be improved through the provision of e-service recoveries involving feedback [t(68) = 2.26; p = .027], apology and feedback [t(68) = 4.43; p = .000], discount and feedback [t(68) = 4.52; p = .000] or all three (i.e., discount, apology and feedback) [t(68) = 4.71; p = .000]. Third, none of the e-service recoveries improves disconfirmed cost expectancy. Rather, the provision of e-service recoveries involving discount [t(68) = -2.11; p = .039] as well as discount and apology [t(68) = -2.05; p = .044] only serve to further worsen disconfirmed cost expectancy. Lastly, of the seven recovery treatments, the provision of a mere apology has no effect consumers’ disconfirmed outcome, process or cost expectancy. Table 6.29 One-Sample t-Test for Disconfirmed Expectancy Constructs [Test Value = 0.000] Disconfirmed Outcome  Recovery Treatment   Disconfirmed Process   Disconfirmed Cost   Mean  Difference   t‐Test   Mean  Difference  t‐Test   Mean  Difference   t‐Test   Feedback   0.024   t(68) = 0.195 (p = 0.846)   0.314   t(68) = 2.259 (p = 0.027)   ‐0.140   t(68) = ‐1.287 (p = 0.203)   Apology   ‐0.020   t(68) = ‐0.249 (p = 0.804)   0.014   t(68) = 0.167 (p = 0.868)   ‐0.043   t(68) = ‐0.311 (p = 0.757)   Apology x Feedback   0.135   t(68) = 1.534 (p = 0.130)   0.836   t(68) = 4.428 (p = 0.000)   0.072   t(68) = 0.455 (p = 0.651)   Discount   0.589   t(68) = 4.293 (p = 0.000)   0.063   t(68) = 0.673 (p = 0.503)   ‐0.299   t(68) = ‐2.109 (p = 0.039)   Discount x Feedback   1.058   t(68) = 5.778 (p = 0.000)   0.638   t(68) = 4.518 (p = 0.000)   ‐0.062   t(68) = ‐0.463 (p = 0.645)   Discount x Apology   0.918   t(68) = 5.398 (p = 0.000)   0.203   t(68) = 1.743 (p = 0.086)   ‐0.338   t(68) = ‐2.053 (p = 0.044)   Discount x Apology x  Feedback   0.865   t(68) = 5.861 (p = 0.000)   0.802   t(68) = 4.713 (p = 0.000)   ‐0.198   t(68) = ‐1.323 (p = 0.190)   Table 6.29 reveals the existence of multiple e-service recovery solutions for alleviating negative failure consequences. Dunnett’s T3 test was hence performed on within-subject differences for disconfirmed expectancies among the seven recovery treatments to determine which of these solution(s) would be most effective in alleviating a particular form of disconfirmed expectancy. Appendix H presents the results from our Dunnett’s T3 test and these results are summarized in Tables 6.2.3.8, 6.2.3.9 and 6.2.3.10 for disconfirmed outcome, process and cost expectancy respectively. As can be deduced from the Table 6.30, any e-service recovery involving discount 105     would mostly outperform if not comparable to any other recovery not involving discount in alleviating disconfirmed outcome expectancy. But concurrently, it is also clear from Table 6.30 that all four e-service recoveries that involve discount are comparable to one another in alleviating disconfirmed outcome expectancy. Combining these results with those from Table 6.29, we can conclude that no added value is gained from including apology and/or feedback over and above discount in alleviating disconfirmed outcome expectancy. This supports hypothesis 6. Table 6.30 Summary of Dunnett T3 Test for Disconfirmed Outcome Expectancy [Recovery Treatment Comparisons] Recovery Treatment (B) FED   APO   APO x  FED   DIS   DIS x FED   DIS x  APO   DIS x  APO x  FED   Feedback [FED]   –                     Apology [APO]   A = B   –                  Apology x Feedback [APO x FED]   A = B   A = B   –               Discount [DIS]   A > B   A > B   A = B   –            Discount x Feedback [DIS x FED]   A > B   A > B   A > B   A = B   –         Discount x Apology [DIS x APO]   A > B   A > B   A > B   A = B   A = B   –      Discount x Apology x Feedback [DIS x APO x FED]   A > B   A > B   A > B   A = B   A = B   A = B   –     Recovery Treatment (A)   A = B – Recovery treatment (A) is comparable to recovery treatment (B) in influencing disconfirmed outcome expectancy  A > B – Recovery treatment (A) is better than recovery treatment (B) in reducing disconfirmed outcome expectancy  A < B – Recovery treatment (A) is worse than recovery treatment (B) in reducing disconfirmed outcome expectancy   Table 6.31 summarizes the results from Dunnett’s T3 test among the seven recovery treatments with regards to disconfirmed process expectancy. As can be deduced from the Table 6.31, any e-service recovery involving feedback would mostly outperform if not comparable to any other recovery not involving feedback in alleviating disconfirmed process expectancy. But concurrently, it is also clear from Table 6.31 that all four e-service recoveries that involve feedback are comparable to one another in alleviating disconfirmed process expectancy. Combining these results with those from Table 6.29, we can conclude that no added value is gained from including discount and/or apology over and above feedback in alleviating consumers’ disconfirmed process expectancy. This supports hypothesis 7. 106     Table 6.31 Summary of Dunnett T3 Test for Disconfirmed Process Expectancy [Recovery Treatment Comparisons] Recovery Treatment (B) FED   APO   APO x  FED   DIS   DIS x FED   DIS x  APO   DIS x  APO x  FED   Feedback [FED]   –                     Apology [APO]   A = B   –                  Apology x Feedback [APO x FED]   A = B   A > B   –               Discount [DIS]   A = B   A = B   A < B   –            Discount x Feedback [DIS x FED]   A = B   A > B   A = B   A > B   –         Discount x Apology [DIS x APO]   A = B   A = B   A = B   A = B   A = B   –      Discount x Apology x Feedback [DIS x APO x FED]   A = B   A > B   A = B   A > B   A = B   A = B   –     Recovery Treatment (A)   A = B – Recovery treatment (A) is comparable to recovery treatment (B) in influencing Disconfirmed Process expectancy  A > B – Recovery treatment (A) is better than recovery treatment (B) in reducing Disconfirmed Process expectancy  A < B – Recovery treatment (A) is worse than recovery treatment (B) in reducing Disconfirmed Process expectancy   Table 6.32 summarizes the results from Dunnett’s T3 test among the seven recovery treatments with regards to disconfirmed cost expectancy. As can be deduced from the Table 6.32, there is no inherent advantage for any e-service recovery in alleviating disconfirmed cost expectancy. Combining these results with those from Table 6.29, we can conclude that no e-service recovery is effective in alleviating disconfirmed cost expectancy and in certain cases, recoveries may even exacerbate the situation (i.e., discount or discount and apology). Hypothesis 8 is not supported.  107     Table 6.32 Summary of Dunnett T3 Test for Disconfirmed Cost Expectancy [Recovery Treatment Comparisons] Recovery Treatment (B) FED   APO   APO x  FED   DIS   DIS x FED   DIS x  APO   DIS x  APO x  FED   Feedback [FED]   –                     Apology [APO]   A = B   –                  Apology x Feedback [APO x FED]   A = B   A = B   –               Discount [DIS]   A = B   A = B   A = B   –            Discount x Feedback [DIS x FED]   A = B   A = B   A = B   A = B   –         Discount x Apology [DIS x APO]   A = B   A = B   A = B   A = B   A = B   –      Discount x Apology x Feedback [DIS x APO x FED]   A = B   A = B   A = B   A = B   A = B   A = B   –     Recovery Treatment (A)   A = B – Recovery treatment (A) is comparable to recovery treatment (B) in influencing Disconfirmed Cost expectancy  A > B – Recovery treatment (A) is better than recovery treatment (B) in reducing Disconfirmed Cost expectancy  A < B – Recovery treatment (A) is worse than recovery treatment (B) in reducing Disconfirmed Cost expectancy   The aforementioned empirical results are further corroborated when the impact of e-service recoveries on disconfirmed expectancies were plotted graphically (see Appendix I). Three inferences can be drawn from Appendix I with regards to the impact of e-service recoveries in alleviating negative failure consequences. First, there is a statistically significant interaction effect between eservice recoveries and disconfirmed outcome expectancy [F(1, 6) = 10.936; p = .000], implying that differences can be detected in disconfirmed outcome expectancy for e-service recoveries over time. Further, upward slopes can be observed for the recovery treatments of discount, discount and apology, discount and feedback as well as discount, apology and feedback on disconfirmed outcome expectancy3. thereby indicating that these four types of e-service recovery are effective in alleviating disconfirmed outcome expectancy. Likewise, there is a statistically significant interaction effect between e-service recoveries and disconfirmed process expectancy [F(1,  6)  = 6.043; p = .000],  implying that differences can be detected in disconfirmed process expectancy for e-service recoveries over time. Also, upward slopes can be observed for the recovery treatments of feedback, discount and                                                              3  Because measurement items for disconfirmed expectancies are phrased negatively and the Likert scale ranges from ‘Strongly Agree’ to ‘Strongly Disagree’, an upward slope would imply an improvement in participants’ disconfirmed expectancies when e-service recoveries are introduced. 108     feedback, apology and feedback as well as discount, apology and feedback on disconfirmed process expectancy, thereby indicating that these four types of e-service recovery are effective in alleviating disconfirmed process expectancy. Conversely, the interaction effect between e-service recoveries and disconfirmed cost expectancy is not statistically significant [F(1, 6) = 0.912; p = .486], implying that differences cannot be detected in disconfirmed cost expectancy for e-service recoveries over time. Results of our hypotheses testing are summarized Table 6.33. Table 6.33 Summary of Hypotheses Testing Hypothesis   Support   Additional Insights   H1: Informational failure on an e‐commerce website will  result  in  the  disconfirmation  of  consumers’  outcome  expectancy.   Supported   While  informational  failure  has  the  strongest  impact  on  disconfirmed  outcome  expectancy,  it  also  negatively  affects  disconfirmed process and cost expectancy.    H2:  Functional  failure  on  an  e‐commerce  website  will  result  in  the  disconfirmation  of  consumers’  process  expectancy.   Supported   While  functional  failure  has  the  strongest  impact  on  disconfirmed  process  expectancy,  it  also  negatively  affects  disconfirmed outcome and cost expectancy.    Supported   While  functional  failure  has  the  strongest  impact  on  disconfirmed  cost  expectancy,  it  also  negatively  affects  disconfirmed outcome and process expectancy.    H3: System failure on an e‐commerce website will result  in the disconfirmation of consumers’ cost expectancy.  H4:  The  presence  of  any  e‐service  recovery  technology  (compensation,  response  sensitivity  or  affinity)  will  negatively  moderate  the  positive  relationship  between  an  e‐service  failure  and  consumers’  disconfirmed  expectancy  H5:  Compensatory  e‐service  recovery  technology  will  have  a  stronger  negative  moderating  effect  on  the  positive  relationship  between  an  e‐service  failure  and  consumers’  disconfirmed  outcome  expectancy  as  compared  to  response  sensitivity  and  affinity  recovery  technologies.  H6:  Response  sensitivity  e‐service  recovery  technology  will  have  a  stronger  negative  moderating  effect  on  the  positive  relationship  between  an  e‐service  failure  and  consumers’  disconfirmed  process  expectancy  as  compared  to  compensatory  and  affinity  recovery  technologies.  H7:  Affinity  e‐service  recovery  technology  will  have  a  stronger  negative  moderating  effect  on  the  positive  relationship between an e‐service failure and consumers’  disconfirmed  cost  expectancy  as  compared  to  compensatory  and  response  sensitivity  recovery  technologies.   6.3  Partially  Supported   Supported   Supported   Not  Supported   While  the  presence  of  an  e‐service  recovery  alleviates  disconfirmed outcome and process expectancy, it negatively  affects disconfirmed cost expectancy.   While  the  provision  of  any  e‐service  recovery  involving  compensatory  mechanisms  will  alleviate  disconfirmed  outcome  expectancy,  there  is  no  inherent  advantage  to  be  gained  from  including  affinity  and  response  sensitivity  recovery technologies over and above compensation.   While  the  provision  of  any  e‐service  recovery  involving  response  sensitivity  mechanisms  will  alleviate  disconfirmed  process  expectancy,  there  is  no  inherent  advantage  to  be  gained  from  including  compensatory  and  affinity  recovery  technologies over and above response sensitivity.   None  of  the  e‐service  recoveries  can  alleviate  disconfirmed  cost  expectancy  with  some  (i.e.,  compensation  or  compensation and affinity) even making it worse.   Discussion The absence of a cohesive theory of e-service failure has thwarted academic efforts to expand  research in the area and stymied practitioners’ attempts to roll out effective recovery technologies to alleviate negative consequences arising from failure occurrences. This second study therefore 109     represents a small but significant step in that direction. We begin by drawing on the Expectation Disconfirmation Theory (EDT) to delineate negative consequences of e-service failures into those affiliated with consumers’ outcome, process and cost expectations. Then, subscribing to consumers’ counterfactual thinking process in the event of e-service failures, we derived a working definition of e-service recovery. Next, we reviewed contemporary frameworks of service recovery and identified Smith et al’s (1999) typology as the most comprehensive and parsimonious theoretical framework from which to theorize the effectiveness of specific recovery technologies in moderating negative failure consequences. An integrated theory of e-service failure and recovery together with testable hypotheses was subsequently constructed through the assimilation of the preventive and corrective research streams within extant literature. The theory is tested via an online experiment that crossmatches different categories of e-service failures (i.e., informational, functional and system failures) with different types of recovery technologies (i.e., compensation, response sensitivity and affinity) to determine their interactional effects on consumers’ disconfirmed expectancies. Experimental findings raise several points of interest. First, our findings reaffirm that e-service failures have an adverse influence on consumers’ outcome, process and cost expectations with respect to an e-commerce transaction. As uncovered in our experimental study, different forms of e-service failure have a domineering effect on one of three expectancies harbored by consumers: (1) informational failure has the strongest impact on consumers’ disconfirmed outcome expectancy; (2) functional failure has the strongest impact on consumers’ disconfirmed process expectancy, and; (3) system failure has the strongest impact on consumers’ disconfirmed cost expectancy. Further, experimental findings also reveal the existence of negative spillover effects from every form of e-service failure. Second, our experimental study indicates that the mere presence of an e-service recovery has a heterogeneous impact on consumers’ disconfirmed expectancies. Whereas e-service recoveries in general tend to alleviate consumers’ disconfirmed outcome and process expectancy, their presence 110     may also contribute to a further increase in consumers’ disconfirmed cost expectancy. A probable reason behind such an observation may be due to the inherent nature of e-service recoveries. Though e-service recoveries—in the form of compensatory and response sensitivity mechanisms—can alleviate negative outcome and process consequences of e-service failures, consumers will still have to expend considerable resources to effectively utilize these mechanisms in order to extract their benefit, thereby contributing to a deterioration in disconfirmed cost expectancy. Third, our findings demonstrate that the effectiveness of different e-service recovery technologies vary depending on which consumers’ expectation is being disconfirmed. Whereas eservice recoveries involving compensation are suited to the alleviation of disconfirmed outcome expectancy, those involving response sensitivity are better at alleviating disconfirmed process expectancy. Further, experimental findings illustrate that having compensation and response sensitivity as standalone recovery solutions would be sufficient in alleviating consumers’ disconfirmed outcome and process expectancy and the inclusion of additional recovery technologies do not compound their effectiveness. Finally, the experiment study claims that none of the e-service recovery technologies are effective at alleviating consumers’ disconfirmed cost expectancy. In fact, the provision of compensation or compensation and affinity recoveries made it worse by boosting consumers’ disconfirmed cost expectancy. A probable reason for this observation has been raised by Bitner (1990), Wirtz and Mattila (2004), who insinuated that compensations may imply an admission of guilt on the part of vendors, thereby leading consumers to view service failures as being avoidable and unnecessary. Similar sentiments were expressed by Weiner (2000) for apologies in that if phrased improperly, apologies could be interpreted as vendors’ confession of responsibility. If consumers were to be led into thinking that e-service failures are preventable, then it could be that they are likely to view any extra cost expended during online transactions to be redundant, even if such costs are incurred from the utilization of recovery technologies to rectify the failure. 111     6.3.1  Implications for Research The second study contributes to extant literature on e-service failure and recovery in four  ways. First, we advance a theory with hypotheses that explain and predict consequences of e-service failures from consumers’ perspective. To the best of our knowledge, there is no prior study that explores consequences of e-service failures. Specifically, we draw a distinction among consumers’ expectations with regards to transactional outcome, process, and cost, and postulate that different categories of e-service failure will disconfirm these expectations in distinctive ways. These hypothesized relationships between e-service failures and disconfirmed expectancies were then scrutinized via an online experiment. Based on our experimental findings, we establish disconfirmed outcome, process and cost expectancy as prominent consequences of informational, functional and system failure respectively. Further, our empirical evidence attests to the existence of negative spillover effects from e-service failures. This is a significant development because to the best of our knowledge, this experiment is the first academic study to systematically investigate and corroborate the existence of negative spillover effects for service failures despite similar claims being made in practitioner literature (Forrester Consulting, 2009; Harris Interactive, 2006). Second, our empirical findings expand on the previous work of McColl-Kennedy et al. (2003), who contended that the presence of any manner of service recovery is better than inaction in the event of service failures. By delineating disconfirmed expectancies into those associated with transactional outcome, process and cost, our study reveals that the benefits of e-service recovery is not necessarily homogeneous as claimed by McColl-Kennedy et al. (2003): even though e-service recovery in general may decrease consumers’ disconfirmed outcome and process expectancy, it has the opposite effect on their disconfirmed cost expectancy. Third, by subscribing to Smith et al.’s (1999) typology in conceptualizing and investigating the effectiveness of e-service recoveries, we are able to identify the range of recovery solutions that may be appropriate for moderating a particular form of disconfirmed expectancy experienced by consumers in the event of an e-service failure. We discovered that having compensation and response 112     sensitivity alone would suffice in alleviating consumers’ disconfirmed outcome and process whereas no e-service recovery technology would be able to neutralize the disconfirmation of consumers’ cost expectation. This study is thus novel in that it not only furthers knowledge on the effectiveness of different combinations of e-service recovery technologies in alleviating negative failure consequences (see Table 6.29), it also yields insights into the most parsimonious recovery solution for a given consumer expectation, which has been disconfirmed by the occurrence of an e-commerce transactional failure. Finally, our study reveals that certain combinations of e-service recovery technologies (i.e., compensation, compensation and apology) can lead to a further deterioration in consumers’ disconfirmed cost expectancy. In a way, we build on empirical findings from previous studies— which allude to consumers’ tendency to view compensation and apology as an admission of guilt (e.g., Bitner, 1990; Weiner, 2000; Wirtz and Mattila, 2004)—by establishing compensatory and affinity mechanisms as having an added detrimental effect on consumers’ cost expectancy over and above the negative influence of e-service failures. Consequently, this study sheds light on the necessity of achieving equilibrium between the provision of commensurable e-service recoveries and its diverse impact on consumers’ disconfirmed expectancies. 6.3.2  Implications for Practice This study should be of interest to both e-merchants and online consumers for three reasons.  First, e-merchants can definitely benefit from an in-depth appreciation of the negative consequences that may arise from different categories of e-service failures. By delineating consequences into disconfirmed outcome, process, and cost expectancies, we provide clarity to the consequences of various categories of e-service failures. That is, while the occurrence of an e-service failure will always result in the disconfirmation of consumers’ expectations about the outcome, process and cost of e-commerce transaction, there are dominant effects for each category of e-service failure. This information could enable e-merchants to effectively channel resources to improve high priority e113     services. For instance, it can be deduced from our study that the transactional process may be compromised when an e-commerce website is lacking functionalities that cater to needs recognition, alternatives identification, alternatives evaluation, acquisition and/or post-purchase activities. Since Olsen (2003) noted that 66% of consumers are dissuaded from making purchases due to problems at various stages of the online transactional process, it may be wise for e-merchants to first focus on securing the delivery of functionalities that support the preceding activities. Second, depending on the form of consumer expectation which has been compromised by the occurrence of an e-service failure, our experimental study illustrates that different types of e-service recovery may be desirable. For this reason, there is urgency for e-merchants to better diagnose the cause of an e-service failure such that commensurable and prudent recovery solutions may be offered. This allows e-merchants to conserve resources by providing targeted e-service recoveries that are shown to be effective. Online consumers would also benefit from the provision of commensurable eservice recoveries since they can be assured of better recovery from e-commerce transactional failures. Finally, in aligning our manipulations of e-service recoveries with practical examples from actual e-commerce websites, we supply concrete evidence that attest to the effectiveness of these recovery technologies in alleviating negative failure consequences. Because e-commerce websites are still lacking in the provision of commensurable e-service recoveries to counter failure occurrences (Holloway and Beatty, 2003), our adaptation of Smith et al.’s (1999) typology, while simplistic, does yield actionable design principles that may be leveraged by e-merchants to implement recovery solutions that would be effective in alleviating negative failure consequences in accordance with our empirical findings. 6.3.3  Limitations Findings from this experimental study should be interpreted conservatively. As deducible  from Table 2.4, we have only extracted a small subset of e-service failures to serve as treatment conditions in our experiment. Although we have endeavored to be representational in our failure 114     treatments by manipulating one form of e-service failure from each of the three primary categories of informational, functional and system failure, we acknowledge that our conclusions on failure consequences should be interpreted with caution when dealing with other failure events beyond what has been investigated in the experiment. Conversely, our empirical findings on e-service recoveries are more credible in that as long as one is able to diagnose the failure consequence experienced by consumers, commensurable recovery technologies may be offered. The only caveat in our recovery treatments lies in the restrictive set of eservice recovery technologies being investigated. Despite incorporating the most pervasive types of eservice recovery technologies into our experimental design, we are aware that there are other webenabled recovery solutions (e.g., FAQs and Live Help), which have escaped notice in this thesis. 6.3.4  Summary This chapter outlines the design and execution of an online experiment that investigates: (1)  negative consequences of e-service failures, and; (2) the effectiveness of e-service recovery technologies to alleviate these consequences. Empirical findings from the experiment suggest that eservice failures should be avoided due to the presence of direct and indirect effects on consumers’ disconfirmed expectancies. This is especially problematic when such failures disconfirm consumers’ cost expectancy since there appears to be no recovery solution to moderate the situation based on our empirical investigation. The next chapter, Chapter 7, will summarize the key findings of the thesis and provide directions for future research.  115     C HAPTER 7 – C ONCLUSION AND D ISCUSSION While research into e-service failure and recovery is still in its infancy, appeals from both academic and practitioner communities for a better appreciation of e-service failure and recovery have attested to the urgency of advancing knowledge on: (1) the causes of e-service failures; (2) their impact on online consumers, and; (3) the range of recovery measures or technologies that can be exploited by e-merchants to facilitate consumers in overcoming these unpleasantries (e.g., Holloway and Beatty, 2003; Forrester Consulting, 2009). More often than not, it is not the occurrence of eservice failures that frustrate consumers, but rather, the absence or incommensurability of recourse channels, which led to unwarranted customer exits. This thesis therefore endeavors to contribute to an in-depth appreciation of the phenomenon by: (1) uncovering technological deficiencies responsible for the manifestation of e-service failures; (2) deciphering consumers’ reactions to these failure events, and; (3) prescribing matching e-service recovery technologies that may be harnessed by e-merchants to deal with each form of failure. Specifically, we endeavor to provide answers to the two research questions stated in Chapter 1: 1.  How do e-service failures manifest on e-commerce websites and what is their impact on online consumer behavior? To answer the aforementioned question, we begin by building on the EDT to derive a  working definition of e-service failure. We then synthesize e-service and system success research streams to advance a novel typology of e-service failure that captures failure events exclusive to ecommerce transactions. It is a primary contention of this thesis that when transacting online, consumers are exceedingly vulnerable to informational (e.g., out-of-stock products), functional (e.g., missing comparison matrix), and system (e.g., delays) failures. Utilizing the CIT, we solicited descriptive accounts of e-service failures to validate the suitability of our proposed typology in classifying failure instances for e-commerce transactions as compared to contemporary frameworks. Empirical findings suggest that our proposed typology of e-service failure is not only more comprehensive than contemporary frameworks in classifying e-service failures by delineating failure 116     causes into a series of informative dimensions, but it is also more parsimonious than others by precluding any dimensions which do not coincide with actual failure events that manifest within ecommerce transactions. Next, we distinguish consequences of e-service failures among those that relate to the disconfirmation of consumers’ outcome, process and cost expectations. We posit that informational, functional and system failure will result in consumers’ disconfirmed outcome, process and cost expectancy respectively. These causal relationships between e-service failures and negative consequences were tested via an online experiment in which we found that not only were our hypotheses validated, but the existence of negative spillover effects were detected as well. This implies that e-service failures are ‘contagious’ in adversely affecting consumers’ evaluations of service encounters. Interestingly, such negative spillover effects for e-service failures have not been empirically corroborated in any systematic academic investigation despite hints of their existence in practitioner surveys. 2.  How can information technology be leveraged to design effective e-service recovery mechanisms for addressing various forms of e-service failure? Subscribing to Smith et al.’s (1999) typology of service recovery modes, we introduce a  classification of e-service recovery technologies that, in the event of service failures for e-commerce websites, empower consumers to: (1) seek reimbursement for the trouble they faced (i.e., compensation); (2) regain social resources lost due to the failure occurrence (affinity); (3) provide feedback regarding their negative transactional experience (i.e., response sensitivity), and; (4) reflect on various aspects of their recent transaction to identify service concerns, which might have otherwise escape notice (i.e., initiative). A review of pragmatic e-commerce websites also testifies to the pertinence of our proposed typology of e-service recovery in that the spectrum of recovery technologies on existing sites can be matched to one of the four dimensions in the typology. We then construct an integrated theory of e-service failure and recovery together with testable hypotheses, which were investigated via an online experiment. Experimental findings offer insights into the 117     effectiveness and parsimony of e-service recovery solutions in alleviating negative failure consequences. We discovered that while compensation and response sensitivity can exist as standalone e-service recovery technologies in alleviating consumers’ disconfirmed outcome and process expectancy, there is no recovery solution in our investigation which may be effective in overcoming consumers’ disconfirmed cost expectancy.  7.1  Implications for Research and Practice A major contribution of this thesis resides in the advancement of a novel typology of e-  service failure that exemplifies the unique characteristics of e-commerce transactional environments. Forsaking contemporary frameworks of service failures for the reasons covered in Chapter 2, our proposed e-service failure typology represents a deductive adaptation of e-service and system success literatures to draw emphasis to actionable technological prescriptions for the design of e-commerce website. While the validity of several dimensions of our proposed e-service failure typology have received corroboration from past studies in end-user computing (i.e., informational and system failures), they have not been tested within an e-commerce setting. Coupled with the fact that certain constructs of the typology have not been subjected to empirical validation in the past (i.e., functional failures), a thorough examination of whether each dimension actually corresponds to the manifestation of an e-service failure in reality is warranted. Further, because this thesis contends that our proposed typology of e-service failure should be superior to contemporary frameworks of service failures in its explanatory and predictive powers with respect to e-commerce transactions, it should outperform other typologies (i.e., Bitner et al., 1990, 1994; Holloway and Beatty, 2003; Kelley et al., 1993) in the classification of failure events. From our exploratory study, we conclude that not only is our proposed typology of e-service failure representative of a parsimonious collection of theoretically-grounded failure dimensions which are useful for deriving conceptual explanations and predictions, but it also embodies valuable benchmarks for practitioners to be vigilant against possible technological deficiencies that may exist on e-commerce websites. 118     Another contribution of this thesis stems from its derivation of distinct e-service failure consequences. Drawing on the EDT, we identified three predominant consequences of e-service failures and these are: (1) the obtainment of transactional outcomes which are undesirable to consumers (i.e., disconfirmed outcome expectancy); (2) the disruption of transactional process which hinders the natural flow of e-commerce transactions (i.e., disconfirmed process expectancy), and; (3) the expenditure of additional resources beyond what is expected to complete e-commerce transactions (i.e., disconfirmed cost expectancy). We uncovered that each of these consequences is salient to a particular category of e-service failure even though spillover effects do exist as alluded to by practitioners. In this sense, we bear a word of caution for e-merchants in that ‘prevention may be better than cure’ in the case of e-service failures due to their contagious nature. Finally, by affirming the effectiveness of the three types of e-service recovery technologies (i.e., compensation, affinity and responsiveness) as well as their combinations on each of the three failure categories (i.e., informational, functional and system), this thesis not only sheds light on the exact e-service recovery solution to be deployed in conjunction with the negative consequence that originates from a specific category of failure, but it also illustrates how these recovery mechanisms can be implemented technologically on e-commerce websites. While Smith et al. (1999) have previously investigated the interactional effects between service failure and recovery for offline services, they have not gone beyond theorizing failure as a monolithic construct. Conceivably, this study is the first of its kind to examine different e-service failure categories and recovery technologies in tandem, thereby bridging existing gaps between preventive and corrective streams of literature. For this reason, this thesis holds promise for practitioners in three ways. First, it contributes to an optimal allocation of resources by enabling e-merchants to tailor their e-service recovery technologies to target failures that are most pronounced for e-commerce transactions. Second, it enlightens practitioners on the technological design of e-service recovery solutions. Third, it offers e-merchants  119     a glimpse into the adequacy of their existing e-service recovery technologies in catering to various forms of failures.  7.2  Future Research Our study lays the groundwork for opening up an entirely new line of research into e-services.  Subsequent empirical investigations should be undertaken to further refine and validate our theory. While we have endeavored to be representative in our investigation of e-service failures for both studies, it is still cross-sectional in nature. There is still much to be explored about the frequency and longitudinal effects of e-service failures on online consumer behaviors. Previous studies of offline service failures show that consumers react much more unfavorably towards failure events that have a higher rate of recurrence (e.g., Folkes et al. 1987; Leong et al. 1997). Future research can therefore investigate whether consumers react differently to: (1) the frequency with which a particular form of e-service failure recurs, and; (2) the time duration between two consecutive recurrences of the same failure. Attribution Theory claims that individuals are rational information processors whose behaviors are directed by their causal inferences (Folkes, 1984). Whenever an e-service failure occurs, it is likely to trigger a cognitive attribution process that involves an assessment of the losses incurred (Bearden and Teel, 1983) and an attribution of blame for the ensuing problem (Bitner 1990; Folkes 1984). Because past studies have shown a strong correlation between consumers’ causal attributions of service failures and their evaluations of service encounters (e.g., Hess et al. 2007), it is worth investigating whether the different forms of e-service failures in our typology result in different causal attributions, and the impact of such attributions on online consumer behaviors and loyalty towards e-merchants. Future research into e-service recovery might also look into resolutions for the service recovery paradox. As explained by Andreassen (2001), the service recovery paradox is founded on the assumption that the potential benefits a firm reaps from the manifestation of service failures 120     complemented with excellent recovery mechanisms is greater than that in a scenario absolutely devoid of service failures throughout the entire transaction. Empirical evidence however, produces mixed conclusions (e.g., Andreassen, 2001; McCollough et al., 2000; Smith and Bolton, 1998; Tax et al., 1998). It is hence interesting to explore this paradox in future investigations to unravel the extent to which effective recovery technologies can induce positive evaluations among consumers in the event of e-service failures. Intuitively, there may be a marked difference in the conditions of the online transactional environment that could possibly challenge the legitimacy of this paradox. For example, due to the possibility of boosted expectations towards e-service recoveries as documented by Holloway and Beatty (2003), it can be the case that it is difficult if not impossible to fashion recovery technologies in ways which are acceptable to consumers. Consequently, the extent to which e-service recovery technologies drive consumers’ positive behavioral intentions may be negligible. Conversely, Voorhees and Brady (2005) illustrated that responsiveness towards service failures is positively correlated with future intentions, thereby suggesting that the strategic leveraging of technology in creating e-service recovery solutions may be critical in retaining consumers following failure occurrences.  7.3  Conclusion In summary, this thesis contributes to extant literature by developing typologies of the  various manifestations of e-service failures for e-commerce websites and the technological artifacts (i.e., recovery technologies) to be harnessed by e-merchants in aiding consumers to overcome transactional difficulties. Then, constructing an integrated theory of e-service failure and recovery, this thesis outlines the design and execution of two empirical studies to clarify the negative impact on online consumers that is brought about by the manifestations of varying failure occurrences and the exact recovery solution, which is effective in addressing negative failure consequence. In light of our empirical findings, it should be emphasized this thesis is but a preliminary step in understanding the causes of e-service failures and prescribing viable countermeasures that can be undertaken by e121     merchants to alleviate negative failure consequences. Concerted research efforts are still very much needed in the future to expand on this foundation in hope of creating a seamless e-commerce transactional experience for online consumers.  122     R EFERENCES Agarwal, R., and Venkatesh, V. “Assessing a Firm’s Web Presence: A Heuristic Evaluation Procedure for Measurement of Usability,” Information Systems Research (13:20), 2002, pp. 168-186. Andreassen, T. W. “From Disgust to Delight: Do Customers Hold a Grudge?” Journal of Service Research (4:1), 2001, pp. 39-49. Bailey, J. E., and Pearson, S. W. “Development of a Tool for Measuring and Analyzing Computer User Satisfaction,” Management Science (29:5), 1983, pp 530-45. Baker, J. A., and Lamb, C. W. Jr. “Measuring Architectural Design Service Quality,” Journal of Professional Services Marketing (10:1), 1993, pp. 89-106. Barber, R. E. and Lucas, H. C. “System Response Time, Operator Productivity, and Job Satisfaction,” Communication of the ACM (26:11), 1983, pp. 972-986. Barclay, D., Thompson, R., and Higgins, C. “The Partial Least Squares (PLS) Approach to Causal Modeling: Personal Computer Adoption and Use as an Illustration,” Technology Studies, (2:2), 1995, pp. 285-309. Barnes, S. J., and Vidgen, R. “An Evaluation of Cyber-Bookshops: The WebQual Method,” International Journal of Electronic Commerce (6:1), 2001, pp. 11-30. Bearden, W. O., and Tee1, J. E. “Some Determinants of Consumer Satisfaction and Complaint Report,” Journal of Marketing Research (20:1), 1983, pp. 21-28. Benbasat, I., and DeSanctis, G. “Communication Challenges: A Value Network Perspective,” in Information Technology and the Future Enterprise: New Models for Managers, G. W. Dickson and G. DeSanctis (eds.), Prentice Hall, Upper Saddle River, 2001, pp. 144-162. Benbasat, I., and Zmud, R. W. “The Identity Crisis Within the IS Discipline: Defining and Communicating the Discipline's Core Properties,” MIS Quarterly (27:2), 2003, pp. 183-194. Berry, L. L., Zeithaml, V. A., and Parasuraman, A. “Quality Comes in Services,” Business Horizon (28:1), 1985, pp. 44-52. Bhattacherjee, A. “Understanding Information Systems Continuance: An Expectation-Confirmation Model,” MIS Quarterly (25:3), 2001, pp. 351-370. Bitner, M. J. “Evaluating Service Encounters: The Effects of Physical Surroundings and Employee Responses,” Journal of Marketing (54:1), 1990, pp. 69-82. Bitner, M. J., Booms, B. H., and Mohr, L. A. “Critical Service Encounters: The Employee’s Viewpoint,” Journal of Marketing (58:1), 1994, pp. 95-106. Bitner, M. J., Booms, B. H., and Tetreault, M. S. “The Service Encounter: Diagnosing Favorable and Unfavorable Incidents,” Journal of Marketing (54:1), 1990, pp. 71-84. Bitner, M. J., Brown, S. W., and Meuter, M. L. “Technology Infusion in Service Encounters,” Journal of the Academy of Marketing Science (28:1), 2000, pp. 138-49. Boulding, W., Kalra, A., Staelin, R., and Zeithaml, V. A. “A Dynamic Process Model of Service Quality: From Expectations to Behavioral Intentions,” Journal of Marketing Research (30), 1993, pp. 7-27. Bove, L. L., and Robertson, N. L. “Exploring the Role of Relationship Variables in Predicting Customer Voice to a Service Worker,” Journal of Retailing and Consumer Services (12:2), 2005, pp. 83-97. Boyatzis, R. E. Transforming Qualitative Information: Thematic Analysis and Code Development, Sage Publications, California, 1998. Boyer, K. K., Hallowell, R., and Roth, A. V. “E-services: Operating Strategy – A Case Study and A Method for Analyzing Operational Benefits,” Journal of Operations Management (20:2), 2002, pp.175-188. Butterfield, L. D., Borgen, W. A., Amundson, N. E., and Maglio, A. S. T. “Fifty Years of the Critical Incident 123     Technique: 1954-2004 and Beyond,” Qualitative Research (5:4), 2005, pp. 475-497. Cai, S., and Jun, M. “Internet Users’ Perceptions of Online Service Quality: A Comparison of Online Buyers and Information Searchers,” Managing Service Quality (13:6), 2003, pp. 504-519. Carugati, A., Hadzilias, E., and Demoulin, N, “Setting The Framework for Developing E-Government Services on Cultural Heritage,” in Proceedings of the 13th European Conference on Information Systems, D. Bartmann (ed.), Resensburg, Germany, 2005. Cenfetelli, R. T. “Inhibitors and Enablers as Dual Factor Concepts in Technology Usage,” Journal of the Association for Information Systems (5:11-12), 2004, pp. 472-492. Cenfetelli, R.T., Benbasat, I., and Al-Natour, S. “Addressing the What and How of Online Services: Positioning Supporting-Services Functionality and Service Quality for Business to Consumer Success,” Information Systems Research (19:2), 2008, pp. 161-181. Childers, T. L., Carr, C. L., Peck, J., and Carson, S. “Hedonic and Utilitarian Motivations for Online Retail Shopping Behavior,” Journal of Retailing (77:4), 2001, pp. 511-535. Clark, G. L., Kaminski, P. F., and Rink, D. R. “Consumer Complaints: Advice on How Companies Should Respond Based on an Empiricai Study,” Journal of Services Marketing (6:1), 1992, pp. 41-50. Clemmer, E. C., and Schneider, B. “Fair Services,” in Advances in Services Marketing and Management, Vol 5, T. A. Swartz, D. E. Bowen, and S. W, Brown (eds.), JAI Press, Greenwich, CT, 1996, pp. 109-226. Clemmer, E. C., and Schneider, B. “Managing Customer Dissatisfaction with Waiting: Applying Social-Psychological Theory in a Service Setting,” in Advances in Services Marketing and Management, Vol 2, T. A. Swartz, D. E. Bowen, and S. W, Brown (eds.), JAI Press, Greenwich, CT, 1993, pp. 213-229. Collier, J. E., and Bienstock, C. C. “A Conceptual Framework for Measuring E-Service Quality,” in Developments in Marketing Science, Academy of Marketing Science Conference Proceedings, H. Spotts, (ed.), Academy of Marketing Science, Corel Gables, FL, 2003, pp. 158-162. Collier, J. E., and Bienstock, C. C. “Measuring Service Quality in E-Retailing,” Journal of Service Research (8:3), 2006, pp. 260-275. Colquitt, J. A., Conlon, D. E., Wesson, M. J., Porter, C. O. L. H., and Ng, Y. K. “Justice at the Millennium: A MetaAnalytic Review of 25 Years of Organizational Justice Research,” Journal of Applied Psychology (86:3), 2001, pp. 425-445. Cronin, J. J., Brady, M. K., Hult, G. T. M. “Assessing the Effects of Quality, Value, and Customer Satisfaction on Consumer Behavioral Intentions in Service Environments,” Journal of Retailing (76:2), 2000, pp. 193-218. Darby, B. W., and Schlenker, B. R. “Children's Reactions to Apologies,” Journal of Personality and Social Psychology (43:4), 1982, pp. 742-753. Davis, F. D. “Perceived Usefulness, Perceived Ease of Use, and User Acceptance of Information Technology,” MIS Quarterly (13:3), 1989, pp. 319-340. Davis, F. D., Bagozzi, R. P., & Warshaw, P. R. Extrinsic and intrinsic motivation to use computers in the workplace. Journal of Applied Social Psychology, (22:14), 1992, pp. 1111-1132. de Matos, C. A., Henrique, J. L., and Rossi, C. A. V. “Service Recovery Paradox: A Meta-Analysis,” Journal of Service Research (10:1), 2007, pp. 60-77. DeLone, W. H., and McLean, E. R. “Information Systems Success: The Quest for the Dependent Variable,” Information Systems Research (3:1), 1992, pp. 60–95. DeLone, W. H., and McLean, E.R. “The DeLone and McLean Model of Information Systems Success: A Ten-Year Update,” Journal of Management Information Systems (19:4), 2003, pp. 9-30. Devaraj, S., Fan, M. and Kohli, R. “Antecedents of B2C Channel Satisfaction and Preference: Validating E-Commerce Metrics,” Information Systems Research (13:3), 2002, pp. 316-333. DeWitt, T., and Brady, M. K. “Rethinking Service Recovery Strategies,” Journal of Service Research (6:2), 2003, pp. 124     193-207. Doherty, W. and Kelisky, R. “Managing VM/CMS Systems for User Effectiveness,” IBM Systems Journal (18:1), 1979, pp. 143-163. Douglas, A., Muir, L., and Meehan, K. “E-Quality in the E-Services Provision of Legal Practices,” Management Service Quality (13:6), 2003, pp. 483-491. Etezadi-Amoli, J., and Farhoomand, A. F. “A Structural Model of End User Computing Satisfaction and User Performance,” Information and Management (30:1), 1996, pp. 65-73. Evanschitzky, H., Iyer, G. R., Hesse, J., and Ahlert, D. “E-Satisfaction: A Re-Examination,” Journal of Retailing (80), 2004, pp. 239-247. Fassnacht, M., and Koese, I. “Quality of Electronic Services: Conceptualizing and Testing a Hierarchical Model,” Journal of Service Research (9:1), 2006, pp. 19-37. Flanagan, J. C. “The Critical Incident Technique,” Psychological Bulletin (51), 1954, pp. 327-58. Folger, R., and Cropanzano, R. “Fairness Theory: Justice as Accountability,” in Advances in Organizational Justice, J. Greenberg and R. Cropanzano (eds.) Stanford University Press, Stanford, CA, 2001, pp. 1-55. Folkes, V. S. “Consumer Reactions to Product Failure: An Attributional Approach,” Journal of Consumer Research (10:4), 1984, pp. 398-409. Folkes, V. S., Koletsky, S., and Graham, J. L., “A Field Study of Causal Inferences and Consumer Reaction: The View from the Airport,” Journal of Consumer Research (13:4), 1987, pp. 534-539 Fornell, C., and Larcker, D. F. “Evaluating Structural Equation Models with Unobservable Variables and Measurement Error,” Journal of Marketing Research (18:1), 1981, pp. 39-50. Forrester Consulting. eCommerce Web Site Performance Today: An Updated Look At Consumer Reaction To A Poor Online Shopping Experience, Forrester Research, Inc, Cambridge, MA, 2009. Furse, D. H., Punj, G. N. and Stewart, D. W. “A Typology of Individual Search Strategies Among Purchasers of New Automobiles,” Journal of Consumer Research (10), 1984, pp. 417-431. Gefen, D. “Customer Loyalty in e-Commerce,” Journal of the Association for Information Systems (3:1), 2002, pp. 2751. Gefen, D., and Ridings, C. M. “Implementation Team Responsiveness and User Evaluation of Customer Relationship Management: A Quasi-Experimental Design Study of Social Exchange Theory,” Journal of Management Information Systems (19:1), 2002, pp. 47-69. Gefen, D., and Straub, D. “A Practical Guide to Factorial Validity Using PLS-Graph: Tutorial and Annotated Example,” Communications of the Association for Information Systems, (16:5), 2005, pp. 91-109. Gershoff, A. D., Broniarczyk, S. M., and West, P. M. “Recommendation or Evaluation? Task Sensitivity in Information Source Selection,” Journal of Consumer Research (28:3), 2001, pp. 418-438. Goodhue, D. L., and Thompson, R. L. “Task-Technology Fit and Individual Performance,” MIS Quarterly (19:2), 1995, pp. 213-236. Gounaris, S., and Dimitriadis S. “Assessing Service Quality on the Web: Evidence from Business-to-Consumer Portals,” Journal of Services Marketing (17:5), 2003, pp. 529-548. Grönroos, C. “Marketing Services: The Case of a Missing Product,” Journal of Business and Industrial Marketing, (13:5), 1998, pp. 322-38. Grönroos, C., Helnomen, F., Isoniemi, K., and Lindholm, M. “The NetOffer Model: A Case Example from the Virtual Marketplace,” Management Decision (38:4), 2000, pp. 243-252. Grover, S., and Benbasat, I. “An Event-Driven Model of E-Commerce Transaction Perceived Risk: Identifying Unwanted Events to Inform IT Design,” Working Paper, Sauder School of Business, University of British Columbia, 2007. 125     Gummerus, J., Liljander, V., Pura, M., and van Riel, A., “Customer Loyalty to Content-Based Websites: The Case of an Online Health-Care Service,” Journal of Service Marketing (18:3), 2004, pp. 175-186. Hamilton, S., and Chervany, N. L. “Evaluating Information System Effectiveness - Part I: Comparing Evaluation Approaches,” MIS Quarterly (5:3), 1981, pp 55-69. Harris Interactive. New Survey Reveals Online Transaction Issues have Negative Impact on Customers and Businesses, 2006. (available online at: http://www.harrisinteractive.com/news/newsletters/clientnews/2006_tealeaf.pdf, accessed June 5, 2008) Hart, C. W. L., Heskett, J. L., and Sasser, W. E. “The Profitable Art of Service Recovery,” Harvard Business  Review (68), 1990, pp. 148–156. Haubl, G., and Trifts, V., “Consumer Decision Making in Online Shopping Environments: The Effects of Interactive Decision Aids,” Marketing Science (19:1), 2000, pp. 4–21. Hess Jr., R. L., Ganesan, S., and Klein, N. M. “Interactional Service Failures in a Pseudorelationship: The Role of Organizational Attributions,” Journal of Retailing (83:1), 2007, pp. 79-95. Hess, R. L. Jr., Ganesan, S., and Klein, N. M. “Service Failure and Recovery: The Impact of Relationship Factors on Customer Satisfaction,” Journal of the Academy of Marketing Science (31), 2003, pp. 127–145. Spring. Hoffman, K. D., and Bateson, J. E. J. Essentials of Services Marketing, The Dryden Press, Fort Worth, TX, 1997. Holloway, B. B., and Beatty, S. E. “Service Failure in Online Retailing: A Recovery Opportunity,” Journal of Service Research (6:1), 2003, pp. 92-105. Homans, G.C. Social Behavior: Its Elementary Forms, New York: Harcourt, Brace, 1961. Homburg, C., Hoyer, W. D., and Fassnacht, M. “Service Orientation of a Retailer’s Business Strategy: Dimensions, Antecedents, and Performance Outcomes,” Journal of Marketing (66:4), 2002, pp. 86-101. Houston, M. B., Bettencourt, L. A., and Wenger, S. “The Relationship Between Waiting in a Service Queue and Evaluations of Service Quality: A Field Theory Perspective,” Psychology & Marketing (15:8), 1998, pp. 735-753. Hoxmeier, J. A., and DiCesare, C. “System Response Time and User Satisfaction: An Experimental Study of Browser Based Applications,” in Proceedings of the Sixth Americas Conference on Information Systems, Long Beach, CA, 2000, pp. 140-145. Ives, B., and Learmonth, G. “The Information System as a Competitive Weapon,” Communications of the ACM (27:12), 1984, pp. 1193-1201. Ives, B., and Mason, R. O. “Can Information Technology Revitalize Your Customer Service?” Academy of Management Executive (4:4), 1990, pp. 52-69. Ives, B., Olson, M., and Baroudi, J. J. “The Measurement of  User Information Satisfaction,” Communications of the ACM, 1983, pp. 785-793. Jacoby, J. “Consumer Behavior: A Quadrennium,” Annual Review of Psychology (49:1), 1998, pp. 319-344. Janda, S., Trocchia, P. J., and Gwinner, K. P. “Consumer Perceptions of Internet Retail Service Quality,” International Journal of Service Industry Management (13:5), 2002, pp. 412-431. Jedetski, J., Adelman, L., and Yeo, C. “How Web Site Decision Technology Affects Consumers,” IEEE Internet Computing (6:2), 2002, pp. 72-79. Jiang, J. J., Klein, G., and Carr, C. L. “Measuring Information System Service Quality: SERVQUAL from the Other Side,” MIS Quarterly (26:2), 2002, pp. 145-166. Karande, K., Magnini, V. P., and Tam, L. “Recovery Voice and Satisfaction After Service Failure,” Journal of Service Research (10:2), 2007, pp. 187-203. Katz, M. A., and Byrne, M. D. “The Effect of Scent and Breadth on Use of Site-Specific Search on E-commerce Web Sites,” ACM Transactions on Human-Computer Interaction (10:3), 2003, pp. 198-220. Keaveney S. M., and Parthasarathy, M. “Customer Switching Behavior in Online Services: An Exploratory Study of 126     the Role of Selected Attitudinal, Behavioral, and Demographic Factors,” Journal of the Academy of Marketing Science (29:4), 2001, pp. 374-390. Keaveney, S. M. “Customer Switching Behavior in Service Industries: An Exploratory Study,” Journal of Marketing (59:2), 1995, pp. 71-82. Kelley, S. W., Hoffman, K. D., and Davis, M. A. “A Typology of Retail Failures and Recoveries,” Journal of Retailing (64:4), 1993, pp. 429-452. Kim, H-W., Xu, Y., and Koh, J. “A Comparison of Online Trust Building Factors between Potential Customers and Repeat Customers,” Journal of the Association for Information Systems (5:10), 2004, pp. 392-420. Kim, M., Kim J-H., and Lennon, S. J. “Online Service Attributes Available on Apparel Retail Websites: An E-SQUAL Approach,” Managing Service Quality (16:1), 2006, pp. 51-77. Kim, S. Y., and Lim, Y. J. “Consumers’ Perceived Importance of and Satisfaction with Internet Shopping,” Electronic Marketing (11:3), 2001, pp. 148-154. Kim, S., and Stoel, L. “Apparel Retailers: Website Quality Dimensions and Satisfaction,” Journal of Retailing and Consumer Services (11), 2004, pp. 109-117. Korthauer, R. D., and Koubek, R. J. “An Empirical Evaluation of Knowledge, Cognitive Style, and Structure upon the Performance of a Hypertext Task,” International Journal of Human-Computer Interaction (6:4), 1994, pp. 373-390. Kuhmann, W. “Experimental Investigation of Stress-inducing Properties of System Response Times,” Ergonomics (32:3), 1989, pp. 271-280. Larson, R. C. “Perspectives on Queues: Social Justice and the Psychology of Queuing,” Operations Research, (35:6), 1987, pp. 895-905. Leclerc, F., Schmitt, B. H., and Dube, L. “Waiting Time and Decision Marking: Is Time like Money?” Journal of Consumer Research, (22:1), 1995, pp. 110-119. Lee, E. and MacGregor, J. “Minimizing User Search Time in Menu Retrieval Systems,” Human Factors (27), 1985, pp. 157-162. Leong, S. M., Ang, S. H., and Low, L H. L. “Effects of Physical Environment and Locus of Control on Service Evaluation: A Replication and Extension,” Journal of Retailing and Consumer Services (4:4), 1997, pp. 231-237. Lightner, N. J. “Evaluating e-Commerce with a Focus on Customer,” Communications of the ACM (47:10), 2004, pp. 88-92. Loiacono, E. T., Watson, R. T., and Goodhue, D. L. “WEBQUAL: A Measure of Website Quality,” in Proceedings of 2002 Marketing Theory and Applications, Chicago, IL, 2002, pp. 432-438. Lovelock, C. H., and Wirtz, J. Services Marketing – People, Technology, Strategy, 5th ed., Prentice-Hall, Upper Saddle River, NJ, 2004. Majchrzak, A., Chellappa, R. K., and Cooper, L. P. “Personalizing Knowledge Delivery Services: A Conceptual Framework,” in New Frontiers in Knowledge Management, K. C. DeSouza (ed.), New York: Palgrave-Macmillan, 2005, pp. 51-71. Martin, G. L., and Corl, K. G. “System Response Time Effects on User Productivity,” Behavior and Information Technology (5:1), 1986, pp. 3-13. Masterson, S. S., Lewis, K., Goldman, B. M., and Taylor, M. S., “Integrating Justice and Social Exchange: The Differing Effects of Fair Procedures and Treatment on Work Relationships,” Academy of Management Journal (43), 2000, pp. 738-748. Matilla, A. S., “The Effectiveness of Service Recovery in a Multi-Industry Setting,” Journal of Services Marketing (15:7), 2001, pp. 583-596. Mattila, A. S., and Patterson, P. G. “Service Recovery and Fairness Perceptions in Collectivist and Individualist Contexts,” Journal of Service Research (6:4), 2004a, pp. 336-346. 127     Mattilla, A. S., and Patterson, P. G. “The Impact of Culture on Consumers’ Perceptions of Service Recovery Efforts,” Journal of Retailing (80:3), 2004b, pp. 196-206. Maxham III, J. G., and Netemeyer, R. G. “A Longitudinal Study of Complaining Customers’ Evaluations of Multiple Service Failures and Recovery Efforts,” Journal of Marketing (66:4), 2002a, pp. 57-71. Maxham III, J. G., and Netemeyer, R. G. “Firms Reap what they Sow: The Effects of Shared Values and Perceived Organizational Justice on Customers’ Evaluations of Complaint Handling,” Journal of Marketing (67:1), 2003, pp. 4662. McColl-Kennedy, J. R., and Sparks, B. A. “Application of Fairness Theory to Service Failures and Service Recovery,” Journal of Service Research (5:3), 2003, pp. 251-266. McColl-Kennedy, J. R., Daus, C. S., and Sparks, B. A. “The Role of Gender in Reactions to Service Failure and Recovery,” Journal of Service Research (6:1), 2003, pp. 66-82. McCollough, M., Berry, L. L., and Yadav, M. S. “An Empirical Investigation of Customer Satisfaction after Service Failure and Recovery,” Journal of Service Research (3:2), 2000, pp. 121-137. McCollough, M., Daus, C. S., and Sparks, B. A. “The Role of Gender in Reactions to Service Failure and Recovery,” Journal of Service Research (6:1), 2003, pp. 66-82. McKinney, V., Yoon, K., and Zahedi, F. “The Measurement of Web-Customer Satisfaction: An Expectation and Disconfirmation Approach,” Information Systems Research (13:3), 2002, pp. 296-315. Melian-Alzola, L., and Padron-Robaina, V. “Tangibility as a Quality Factor in Electronic Commerce B2C,” Managing Service Quality (16:3), 2006, pp. 320-338. Mentzer, J. T., Flint, D. J., and Hult, T. M. “Logistics Service Quality as a Segment-Customized Process,” Journal of Marketing (65:1), 2001, pp. 82-104. Muthukrishnan, A. V., and Chattopadhyay, A. “Just Give Me Another Chance: The Strategies for Brand Recovery from a Bad First Impression,” Journal of Marketing Research (44:2), 2007, pp. 334–345. Nah, F. “A Study of Web Users’ Waiting Time,” in Intelligent Support Systems Technology: Knowledge Management, V. Sugumaran (ed.), IRM Press, 2002, pp. 145-152. Northrop, A., Kraemer, K. L., Dunkie, D., and King, J. L. “Payoffs from Computerization: Lessons Over Time,” Public Administration Review (50:5), 1990, pp. 505-514. Nunnally, J., and Bernstein, I. Psychometric Theory, McGraw Hill, New York, 1994. Nyquist, J. D., and Booms, B. H. “Measuring Services Value From the Consumer Perspective,” in Add Value to Your Service, C. Surprenant, (ed.), American Marketing Association, Chicago, 1987, pp. 13-16. O’Neill, M., Wright, C., and Fitz, F. 2001. “Quality Evaluation in Online Service Environments: An Application of the Importance-Performance Measurement Technique,” Managing Service Quality (11:6), 2001, pp. 402-417. Oliver, R. L. “A Cognitive Model of the Antecedents and Consequences of Satisfaction Decisions,” Journal of Marketing Research (17:4), 1980, pp. 460-469. Oliver, R. L. Satisfaction: A Behavioral Perspective on the Consumer, McGraw-Hill, New York, 1997. Olsen, H. “Supporting Customers’ Decision-Making Process,” The Interaction Designer’s Coffee Break (6), 2003. (available online at: http://www.guuui.com/issues/02_03.php, accessed November 16, 2009) Oneupweb. Revolutionizing Website Design: The New Rules of Usability, Oneupweb, 2010. (available online at http://www.oneupweb.com/landing/10_usability_study/?source=pitchengine_2010usability_060910&guid=7F4760D73B72-DF11-9242-0010B5500080#, accessed January 27, 2011) Organ, D. W. “The motivational Basis of Organizational Citizenship Behavior,” Research in Organizational Behavior (12:1), 1990, pp. 43-72. Palmer, J. W. “Web Site Usability, Design and Performance Metrics,” Information Systems Research (13:2), 2002, pp. 151-167. 128     Parasuraman, A., and Grewal, D. “The Impact of Technology on the Quality-Value-Loyalty Chain: A Research Agenda,” Journal of the Academy of Marketing Science (28:1), 2000, pp. 168-174. Parasuraman, A., Berry, L. L., and Zeithaml, V. A. “A Conceptual Model of Service Quality and its Implications for Future Research,” Journal of Marketing (49:4), 1985, pp. 41-50. Parasuraman, A., Berry, L. L., and Zeithaml, V. A. “SERVQUAL: A Multiple-Item Scale for Measuring Customer Perceptions of Service Quality,” Journal of Retailing (64:1), 1988, pp. 12-40. Parasuraman, A., Zeithaml, V. A., and Malhotra, A. “E-S-Qual: A Multiple-Item Scale for Assessing Electronic Service Quality,” Journal of Service Research (7:3), 2005, pp. 213-233. Payne, J. W., Bettman, J. R., and Johnson, E. J. The Adaptive Decision Maker, Cambridge University Press, 1993. Piccoli, G., Spalding, B. R., and Ives, B. “The Customer-Service Life Cycle: A Framework for Improving Customer Service through Information Technology,” Cornell Hotel and Restaurant Administration Quarterly (42:3), 2001, pp. 38-45. Ramsay, J., Barbesi, A., and Preece, J. “A Psychological Investigation of Long Retrieval Times on the World Wide Web,” Interacting with Computers (10:1), 1998, pp. 77-86. Ribbink, D., van Riel, A. C. R., Liljander, V., and Streukens, S. “Comfort Your Online Customer: Quality, Trust, and Loyalty on the Internet,” Managing Service Quality (14:6), 2004, pp. 446-456. Ringberg, T., Odekerken-Schröder, G., and Christensen, G. L. “A Cultural Models Approach to Service Recovery,” Journal of Marketing (71:3), 2007, pp. 194-214. Roese, N. J. “Counterfactual Thinking,” Psychological Bulletin, (121:1), 1997, pp. 133-48. Roese, N. J., and Olson, J. M. “Counterfactual Thinking: A Critical Overview,” in What Might Have Been: The Social Psychology of Counterfactual Thinking, N. J. Roese and J. M. Olson (eds.), Lawence Erlbaurn, Hillsdale, NJ, 1995. Rose, G. M., and Straub, D. “The Effect of Download Time on Consumer Attitude Toward the E-Service Retailer,” eService Journal (1:1), 2001, pp. 55-76. Rose, G., Khoo, H., and Straub, D. W. “Current Technological Impediments to Business-To-Consumer Electronic Commerce,” Communications of the AIS (1:5), 1999, pp. Rosen, D. E., and Purinton, E. “Website Design: Viewing the Web as a Cognitive Landscape,” Journal of Business Research (57:7), 2004, pp. 787–794. Rosenbaum, M. S., and Massiah, C. A. “When Customers Receive Support from Other Customers: Exploring the Influence of Intercustomer Social Support on Customer Voluntary Performance,” Journal of Service Research (9:3), 2007, pp. 257-270. Rust, R. T., and Oliver, R. L. “Service Quality – Insights and Managerial Implications from the Frontier,” in Service Quality: New Directions in Theory and Practice, R. T. Rust and R. L. Oliver, (eds.), Sage, Thousand Oaks, CA, 1994, pp. 1-19. Santos, J. “E-Service Quality: A Model of Virtual Service Quality Dimensions,” Management Service Quality (13:3), 2003, pp. 233-246. Scher, S. J., and Darley, J. M. “How Effectives are the Things People Say to Apologize? Effects of the Realization of the Apology Speech Act,” Journal of Psycholinguistic Research (26:1), 1997, pp. 127-140. Schlenker, B. R., and Darby, B. W. “The Use of Apologies in Social Predicaments,” Social Psychology Quarterly (44:3), 1981, pp. 271-278. Schlosser, A. E., White, T. B., and Lloyd, S. M. “Converting Web Site Visitors into Buyers: How Web Site Investment Increases Consumer Trusting Beliefs and Online Purchase Intentions,” Journal of Marketing (70:April), 2006, pp. 133148. Schubert, P. “Extended Web Assessment Method (EWAM): Evaluation of Electronic Commerce Applications from the Customer’s Viewpoint,” International Journal of Electronic Commerce (7:2), 2002, pp. 51-80. 129     Sears, A., Jacko, J. A., and Dubach, E. M. “International Aspects of WWW Usability and the role of High-End Graphical Enhancements,” International Journal of Human-Computer Interaction (12:2), 2000, pp. 243-263. Semeijn, J., van Riel, A. C. R., van Birgelen, M. J. H., and Streukens, S. “E-Services and Offline Fulfillment: How ELoyalty is Created,” Managing Service Quality (15:2), 2005, pp. 182-194. Shchiglik, C., and Barnes. S. J. “Evaluating Website Quality in the Airline Industry,” Journal of Computer Information Systems (44:3), 2004, pp. 17-25. Shim, J. P., Shin, Y. B., and Nottingham, L. “Retailer Web Site Influence on Customer Shopping: An Exploratory Study on Key Factors of Customer Satisfaction,” Journal of the Association for Information Systems (3:1), 2002, pp. 53-76. Shneiderman, B. Designing the User Interface, Addison-Wesley, Boston, 1998. Singh, M. “E-Services and their Role in B2C E-Commerce,” Managing Service Quality (12:6), 2002, pp. 434-446. Smart, D. T., and Martin, C. L. “Manufacturer Responsiveness to Consumer Correspondence: An Empirical Investigation of Consumer Perceptions,” Journal of Consumer Affairs (26:1), 1992, pp. 104-28. Smith, A. K., and Bolton, R. N. “An Experimental Investigation of Service Failure and Recovery: Paradox or Peril?” Journal of Service Research (1:1), 1998, pp. 65-81. Smith, A. K., Bolton, R. N., and Wagner, J. “A Model of Customer Satisfaction with Service Encounters Involving Failure and Recovery,” Journal of Marketing Research (36:3), 1999, pp. 356-372. Sparks, B., and McColl-Kennedy, J. R. “Justice Strategy Options for Increased Customer Satisfaction in a Services Recovery Setting,” Journal of Business Research (54:3), 2001, pp. 209-218. Spreng, R. A., MacKenzie, S. B., and Olshavsky, R. W. “A Reexamination of Determinants of Consumer Satisfaction,” Journal of Marketing (60:1), 1996, pp. 15-32. Srinivasan, S. S., Anderson, R., and Ponnavolu, K., “Customer Loyalty in E-Commerce: An Exploration of its Antecedents and Consequences,” Journal of Retailing (78:1), 2002, pp. 41-50. Stanton, J. M., and Rogelberg, S. G. “Using Internet/Intranet Web pages to Collect Organizational Research Data,” Organizational Research Methods (4:3), 2001, pp. 200-217. Surjadjaja, H., Ghosh, S., and Antony, J. “Determining and Assessing the Determinants of E-Service Operations,” Managing Service Quality (13:1), 2003, pp. 39-53. Swartz, T. A., and Brown, S. W. “Consumer and Provider Expectations and Experiences Evaluating Professional Service Quality,” Journal of the Academy of Marketing Science (17:2), 1989, pp. 189-195. Tam, K. Y., and Ho, S. Y. “Understanding the Impact of Web Personalization on User Information Processing and Decision Outcomes,” MIS Quarterly (30:4), 2006, pp. 865-890. Tan, C. W., and Benbasat, I. “IT-Mediated Customer Services in E-Government: A Citizen’s Perspective,” Communications of the AIS, (24:12), 2009. Tan, C. W., Benbasat, I., and Cenfetelli, R., “IT-Mediated Customer Service Content and Delivery in Electronic Governments: An Empirical Investigation of the Antecedents of Service Quality,” Working Paper, Sauder School of Business, University of British Columbia, 2010. Tax, S. S., and Brown, S. W. “Recovering and Learning from Service Failure,” Sloan management Review (40:1), 1998, pp. 75-88. Tax, S. S., Brown, S. W., and Chandrashekaran, M. “Customer Evaluations of Service Complaint Experiences: Implications for Relationship Marketing,” Journal of Marketing (62:2), 1998, pp. 60-76. Taylor, S. “Waiting for Service: The Relationship between Delays and Evaluations of Service,” Journal of Marketing (58:2), 1994, pp. 56-69. Teas, R. K. “Expectations, Performance Evaluation, and Consumers’ Perceptions of Quality,” Journal of Marketing (57:4), 1993, pp. 18-34. 130     Thomas, D. M., and Bostrom, R. P. “Vital Signs for Virtual Teams: An Empirically Developed Trigger Model for Technology Adaptation Interventions,” MIS Quarterly (34:1), 2010, pp. 115-142. Torkzadeh, G. and Dhillon, G. “Measuring Factors that Influence the Success of Internet Commerce,” Information Systems Research, (13:2), 2002, pp. 187-204. Van Riel, A. C. R., Liljander, V., and Jurriens, P. “Exploring Consumer Evaluations of e-Services: A Portal Site,” International Journal of Service Industry (12:4), 2001, pp. 359-377. Vandenbosch, B., and Huff, S. L. “Searching and  Scanning: How Executives Obtain Information  from Executive Information Systems,” MIS Quarterly (21:1), 1997, pp. 81-107 Venkatesh, V., Morris, M. G., Davis, G. B., and Davis, F. D. “User Acceptance of Information Technology: Toward a Unified View,” MIS Quarterly (27:3), 2003, pp. 425-478. Voorhees, C. M., and Brady, M. K. “A Service Perspective on the Drivers of Complaint Intentions,” Journal of Service Research (8:2), 2005, pp. 192-204. Walster, E., Berscheid, E., and Walster, G. W. “New Directions in Equity Research,” Journal of Personality and Social Psychology (25:2), 1973, pp. 151-176. Walster, G. W., and Berscheid, E. Equity: Theory and Research, Allyn and Bacon, Boston, MA, 1978. Wand, Y., and Wang, R. Y. “Anchoring Data Quality Dimensions in Ontological Foundations,”  Communications of the ACM (39:11), 1996, pp. 86-95. Wang, R. Y., and Strong, D. M. “Beyond  Accuracy: What Data Quality Means to Data  Consumers,” Journal of Management Information Systems (12:4), 1996, pp. 5-34. Wang, Y-S. “The Adoption of Electronic Tax Filing Systems: An Empirical Study,” Government Information Quarterly (20), 2002, pp. 333-352. Ward, J. C., and Ostrom, A. L. “Complaining to the Masses: The Role of Protest Framing in Customer Created Complaint Web Sites,” Journal of Consumer Research (33:2), 2006, pp. 220-230. Weiner, B. “Attributional Thoughts about Consumer Behaviour,”  Journal of Consumer Research (27), 2000, pp. 382– 387. Weun, S., Beatty, S. E., and Jones, M. A. “The Impact of Service Failure Severity on Service Recovery Evaluations and Post-Recovery Relationships,” Journal of Services Marketing (18:2/3), 2004, pp. 133-46. Wind, Y. J. “The Challenge of ‘Customerization’ in Financial Services,” Communications of the ACM (44:6), 2001, pp. 39-44. Wirtz, J., and Mattila, A. S. “Consumer Responses to Compensation, Speed of Recovery and Apology after a Service Failure,” International Journal of Service Industry Management (15:2), 2004, pp. 150-166. Wixom, B. H., and Todd, P. A. “A Theoretical Integration of User Satisfaction and Technology Acceptance,” Information Systems Research (16:1), 2005, pp. 85-102. Wixom, B. H., and Watson, H. J. “An Empirical Investigation of the Factors Affecting Data Warehousing Success,” MIS Quarterly (25:1), 2001, pp. 17-41. Wolfinbarger, M., and Gilly, M. C. “eTailQ: Dimensionalizing, Measuring and Predicting etail Quality,” Journal of Retailing (79:3), 2003, pp. 183-198. Xiao, B. and Benbasat, I “Product-Related Deception in e-Commerce: A Theoretical Perspective,” MIS Quarterly, 2011, (Forthcoming). Xiao, B., and Benbasat, I. “E-Commerce Product Recommendation Agents: Use, Characteristics, and Impact,” MIS Quarterly (31:1), 2007, pp. 137-209. Zeithaml, V. A. “Service Excellence in Electronic Channels,” Managing Service Quality (12:3), 2002, pp. 135-138. Zeithaml, V. A., Berry, L. L., and Parasuraman, A. “The Behavioral Consequences of Service Quality,” Journal of 131     Marketing (60:2), 1996, pp. 31–46. Zeithaml, V. A., Berry, L. L., and Parasuraman, A. “The Nature and Determinants of Customer Expectations of Service,” Journal of the Academy of Marketing Science (21:1), 1993, pp. 1-12. Zeithaml, V. A., Parasuraman, A., and Malhotra, A. “Service Quality Delivery through Web Sites: A Critical Review of Extant Knowledge,” Journal of the Academy of Marketing Science (30:4), 2002, pp. 362-375. Zhang, P., von Dran, G. M., Blake, P., and Pipithsuksunt, V. “Important Design Features in Different Web Site Domains: An Empirical Study of User Perceptions,” e-Service Journal (1:1), 2001, pp. 77-91.  132     A P P E N D IX A – C A T E G O R I ZA T I O N  OF  E X T A N T E-S E R V I C E L I T E R A T U R E  Informational Attributes   Functional Attributes   System Attributes   Author(s)  ACC  COM  REL  TIM  NER  AID  ALE  ACQ  POP  AES  NAV  ADT  SPD  SEC  Agarwal and Venkatesh (2002)      X   X   X   X            X   X   X   X         Barnes and Vidgen (2001)   X         X   X            X   X   X   X   X   X   Cai and Jun (2003)   X   X      X               X   X   X            Cenfetelli et al. (2008)               X   X      X   X   X   X   X   X      Childers et al. (2001)               X               X   X      X      Collier and Bienstock (2003, 2006)   X   X      X            X   X   X   X         X   Devaraj et al. (2002)   X         X   X      X      X   X   X   X      X   Douglas et al. (2003)   X   X   X   X      X         X   X   X         X   Evanschitzky et al. (2004)   X   X            X   X         X   X      X   X   Fassnacht and Koese (2006)   X   X   X   X   X   X      X      X   X      X   X   Gefen (2002)                        X   X   X      X      X   Gounaris and Dimitriadis (2003)               X         X   X               X   Gummerus et al. (2004)               X               X   X         X   Janda et al. (2002)   X                     X      X            X   Jiang et al. (2002)                           X   X      X      X   Kim and Lim (2001)   X   X   X   X                  X         X      Kim and Stoel (2004)         X               X         X      X   X   Kim et al. (2004)   X            X   X      X   X   X   X   X      X   Kim et al. (2006)   X            X         X   X   X   X   X   X   X   Loiacono et al. (2002)      X   X      X   X   X         X   X      X   X   McKinney et al. (2002)   X   X   X   X            X      X   X            Meliàn‐Alzola and Padron‐Robaina (2006)               X   X      X         X            O’Neill et al. (2001)                           X   X   X   X      X   Palmer (2002)   X   X         X            X   X   X   X   X      Parasuraman et al. (2005)                        X      X         X   X   Ribbink et al. (2004)               X               X   X   X      X   Rosen and Purinton (2004)                     X         X   X            Santos (2003)      X         X            X   X   X      X   X   Schubert (2002)         X                     X            X   Semeijn et al. (2005)   X            X                  X   X      X   Shchiglik and Barnes (2004)   X   X   X         X      X   X   X   X            Shim et al. (2002)   X               X   X      X   X               Singh (2002)                  X      X   X   X            X   Srinivasan et al. (2002)               X      X      X   X      X      X   Surjadjaja et al. (2003)   X         X   X         X   X   X   X   X   X   X   133     Informational Attributes   Functional Attributes   System Attributes   Author(s)  ACC  COM  REL  TIM  NER  AID  ALE  ACQ  POP  AES  NAV  ADT  SPD  SEC  Wolfinbarger and Gilly (2003)                        X   X   X   X         X   Zeithaml (2002); Zeithaml et al. (2002)      X      X            X      X   X         X   Zhang and von Dran (2001)         X   X   X   X      X                     ACC  –  Accuracy;  COM  –  Completeness;  REL  –  Relevance;  TIM  –  Timely;  NER  –  Needs  Recognition;  AID  –  Alternatives  Identification;  ALE  – Alternatives  Evaluation;  ACQ  –  Acquisition;  POP  –  Post‐Purchase;  AES  –  Accessibility;  NAV  –  Navigability;  ADT  –  Adaptability;  SPD  –  Speed;  SEC – Security   134     A P P E N D IX B – D E T A I L E D B R E A K D O W N  OF  CLASSIFICATIONS  OF  E-S ER V I C E F A I L U R E I N C I D E N T S  Table B-1: Typology of Service Encounter Failures [as adapted from Bitner et al. (1990) and/or Bitner et al. (1994)] [Sample N = 374] Incident Coding  Construct   Definition (Event in which...)   No. Unique  Incidents* [%]   No. Common  + Incidents  [%]   Inter‐Judge  ‡ Reliability    Failure of Service Delivery System [as adapted from Bitner et al. (1990) and Bitner et al. (1994)]  Unavailable Service   E‐merchant fails to provide services  that are normally available or expected   19 [5.08%]   2 [0.53%]   0.11   Unreasonably Slow Service   E‐merchant is slow in servicing customers   25 [6.68%]   13 [3.48%]   0.52   Other Core Service Failure   E‐merchant fails to meet basic performance standards for other aspects of the core service (apart from  150 [40.11%]  its absence or slowness)   39 [10.43%]   0.26   Failure to Meet Customer Needs and Requests [as adapted from Bitner et al. (1990) and Bitner et al. (1994)]  Failure to Meet ‘Special Needs’  Customers    E‐merchant  fails  to  recognize  and  accommodate  customers’  special    demographical,  physical  and/or  sociological needs (e.g., disabilities)   24 [6.42%]   7 [1.87%]   0.29   Failure to Meet Customer  Preferences   E‐merchant fails to recognize and accommodate customers’ preferences that run contrary to standard  practices   83 [22.19%]   17 [4.55]   0.20   Failure to Address Admitted  Customer Error   E‐merchant fails to resolve problems that arise from customers’ admitted errors   6 [1.60%]   4 [1.07%]   0.67   Failure to Manage Disruptive  Others    E‐merchant fails to deal appropriately with disruptive customers   0 [0.00%]   0 [0.00%]   0.00   Unprompted and Unsolicited Service Behaviors [as adapted from Bitner et al. (1990) and Bitner et al. (1994)]  Failure to Pay Attention to  Customer   E‐merchant fails to pay sufficient attention to customers during service encounters   44 [11.76%]   14 [3.74%]   0.32   Failure due to Out‐of‐the Ordinary  Service Behavior   E‐merchant  fails  to  perform  in  an  expected  manner  and  culminates  in  adverse  consequences  for  177 [47.33%]  customers   82 [21.93%]   0.46   Failure to be Sensitive to Cultural  Norms   E‐merchant fails to observe cultural norms during service encounters   24 [6.42%]   7 [1.87%]   0.29   Gestalt Evaluation Failure   E‐merchant fails to prevent isolated failures from affecting other related services   17 [4.55%]   0 [0.00%]   0.00   Failure to Perform Under Adverse  Circumstances   E‐merchant fails to perform efficaciously under unfavorable circumstances    7 [1.87%]   4 [1.07%]   0.57   0 [0.00%]   0 [0.00%]   0.00   Failure to Address Problematic Customer Behavior [as adapted from Bitner et al. (1994)]  Failure to Address Drunk Customers  E‐merchant fails to deal with  intoxicated customers who are causing troubles   135     Incident Coding  Construct   Definition (Event in which...)   No. Unique  Incidents* [%]   No. Common  + Incidents  [%]   Inter‐Judge  ‡ Reliability    0 [0.00%]   0 [0.00%]   0.00   Failure to Address Customers  E‐merchant fails to deal with customers who refuse to comply with company rules and regulations  Breaking Company Laws or Policies   0 [0.00%]   0 [0.00%]   0.00   Failure to Address Uncooperative  Customers   E‐merchant  fails  to  deal  with  customers  who  are  generally  rude,  uncooperative  and/or  unreasonably  demanding   0 [0.00%]   0 [0.00%]   0.00   E‐merchant fails to provide quality information to customers in making transactional decisions   10 [2.67%]   4 [1.07%]   0.40   Failure to Address Verbal and  Physical Abuse   E‐merchant fails to deal with customers who engage in physical and/or verbal abuses   Informational Failure  Informational Failure   *  Total number of unique incidents assigned to each category by both judges  + Total number of identical incidents assigned to each category by both judges  ‡ Number of identical incidents divided by number of unique incidents   136     Table B-2: Typology of Retail Failures [as adapted from Kelley et al. (1993)] [Sample N = 374] Incident Coding  Construct   Definition (Event in which...)   No. Unique  Incidents* [%]   No. Common  + Incidents  [%]   Inter‐Judge  ‡ Reliability    19 [5.08%]   10 [2.67%]   0.53   Failure of Service Delivery System and/or Product  Policy Failure   E‐merchant fails to enact service policies that are deemed to be just among customers   Slow / Unavailable Service   E‐merchant  fails  to  provide  services  that  are  normally  available  or  expected  and/or  is  slow  in  servicing  customers   221 [59.09%]   164 [43.85%]   0.74   System Pricing Failure   E‐merchant erroneously price listed products   45 [12.03%]   12 [3.21%]   0.27   Packaging Errors   E‐merchant fails to properly package purchased products and/or label packages correctly   32 [8.56%]   20 [5.53%]   0.63   Product Defects   Purchased products fail to function as they are supposed to   10 [2.67%]   3 [0.80%]   0.30   Out‐of‐Stock   E‐merchant fails to supply accurate information on the inventory levels of listed products   12 [3.21%]   6 [1.60%]   0.50   Hold Disasters    E‐merchant  fails  to  guarantee  that  products  waiting  to  be  claimed  by  customers  do  not  become  lost  or  damaged   1 [0.27%]   1 [0.27%]   1.00   Alteration and Repairs Failure   E‐merchant fails ensure that product alterations or repairs are performed in a precise and speedy fashion   1 [0.27%]   0 [0.00%]   0.00   Bad Information   E‐merchant misinforms customers in making transactional decisions   8 [2.14%]   2 [0.53%]   0.25   9 [2.41%]   0 [0.00%]   0.00   8 [2.14%]   3 [0.80%]   0.38   Failure to Meet Customer Needs and Requests  Special Order / Request Failure   E‐merchant fails to fulfill special or unique requests that were promised to customers   Failure to Address Admitted  Customer Error   E‐merchant fails to resolve problems that arise from customers’ admitted errors   Unprompted and Unsolicited Service Behaviors  Mischarging   E‐merchant charges customers more than necessary for product purchases   5 [1.34%]   0 [0.00%]   0.00   Wrongful Accusation of  Customers   E‐merchant  wrongfully  accuses  customers  of  inappropriate  actions  and/or  places  them  under  excessive  surveillance during service encounters   0 [0.00%]   0 [0.00%]   0.00   Failure due to Service‐Induced  Embarrassment   E‐merchant embarrasses customers due to insensitivity or mistakes during service encounters   10 [2.67%]   0 [0.00%]   0.00   Attention Failures   E‐merchant fails to pay sufficient attention to customers during service encounters   107 [28.61%]   39 [10.43%]   0.36   *  Total number of unique incidents assigned to each category by both judges  + Total number of identical incidents assigned to each category by both judges  ‡ Number of identical incidents divided by number of unique incidents     137     Table B-3: Typology of Online Service Failures [as adapted from Holloway and Beatty (2003)] [Sample N = 374] Incident Coding  Construct   Definition (Event in which...)   No. Unique  Incidents* [%]   No. Common  + Incidents  [%]   Inter‐Judge  ‡ Reliability    Delivery Problems  Purchase Arrived Later than Promised   E‐merchant is late in delivering purchased products to customers   3 [0.80%]   0 [0.00%]   0.00   Purchase Never Delivered   E‐merchant fails to deliver purchased products to customers   22 [5.88%]   15 [4.01%]   0.68   Wrong Item Delivered   E‐merchant delivers products that are different from what were purchased   10 [2.67%]   6 [1.60%]   0.60   Wrong Size Product Delivered   E‐merchant delivers products with different specifications from what were purchased   7 [1.87%]   4 [1.07%]   0.57   Purchase Damaged During Delivery   E‐merchant fails to properly package purchased products to avoid damage during delivery   2 [0.53%]   2 [0.53%]   1.00   90 [24.06%]   33 [8.82%]   0.37   8 [2.14%]   1 [0.27%]   0.13   Insufficient Information Provided at Site  E‐merchant fails to supply sufficient information on transactional activities   38 [10.16%]   17 [4.55%]   0.45   Products Incorrectly Listed at Site as in  Stock   E‐merchant fails to supply accurate information on the inventory levels of listed products   10 [2.67%]   7 [1.87%]   0.70   Incorrect Information Provided at Site   E‐merchant fails to supply correct information that aid customers in making transactional decisions  3 [0.80%]   2 [0.53%]   0.67   186 [49.73%]   83 [22.19%]   0.45   Poor Communication with the Company  E‐merchant fails to provide communication channels for customers to seek assistance   22 [5.88%]   13 [3.48%]   0.59   Unfair Return Policies   E‐merchant compels customers to return purchased products under unjust terms   6 [1.60%]   0 [0.00%]   0.00   Unclear Return Policies   E‐merchant fails to supply unambiguous information for returning purchased products   1 [0.27%]   0 [0.00%]   0.00   Credit Card Overcharged   E‐merchant charges customers more than necessary for product purchases   8 [2.14%]   5 [1.34%]   0.63   Website Purchasing Process Confusing   E‐merchant fails to offer a straightforward product purchasing process for customers   18 [4.81%]   2 [0.53%]    0.11   Difficulties Experienced While Paying    E‐merchant fails to provide payment options desired by customers   20 [5.35%]   5 [1.34%]   0.25   Problems with Product Quality   Purchased products fail to function as they are supposed to   5 [1.34%]   2 [0.53%]   0.40   Consumer Dissatisfied with Product  Quality   Customers are disappointed with the way purchased products function   0 [0.00%]   0 [0.00%]   0.00   Website Design Problems  Navigational Problems at Site   E‐merchant fails to offer easy accessibility to service content offered   Product Poorly Presented at Site   E‐merchant fails to supply relevant information on product specifications   Customer Service Problems  Poor Customer Service Support   E‐merchant fails to meet customers’ service expectations when performing online transactions   Payment Problems   138     Incident Coding  Construct   Definition (Event in which...)   No. Unique  Incidents* [%]   No. Common  + Incidents  [%]   Inter‐Judge  ‡ Reliability    Security Problems  Credit Card Fraud   E‐merchant charges customers for unauthorized purchases   4 [1.07%]   1 [0.27%]   0.25   Misrepresented Merchandise   E‐merchant misinforms customers into purchasing products with unlisted specifications   2 [0.53%]   0 [0.00%]   0.00   Email Address Released to E‐Marketers   E‐merchant  releases  customers’  disclosed  email  addresses  to  e‐marketers  without  proper  authorization   2 [0.53%]   2 [0.53%]   1.00   Failure to Address Unintentional  Customer Mistakes   E‐merchant  fails  to  resolve  problems  that  arise  out  of  unintentional  mistakes  on  the  part  of  customers   5 [1.34%]   3 [0.80%]   0.60   Retailer Charged Some Customers More  than Others   E‐merchant charges certain customers more than others for purchasing exact same products   0 [0.00%]   0 [0.00%]   0.00   Lack of Personalized Information at Site   E‐merchant fails to tailor transactional information to meet customers’ requirements   56 [14.79%]   17 [4.55%]   0.30   Miscellaneous   *  Total number of unique incidents assigned to each category by both judges  + Total number of identical incidents assigned to each category by both judges  ‡ Number of identical incidents divided by number of unique incidents   139     Table B-4: Proposed E-Service Failure Typology [Sample N = 374] Incident Coding  Construct   Definition (Event in which…)   No. Unique  Incidents* [%]   No. Common  + Incidents  [%]   Inter‐Judge  ‡ Reliability    Informational Failures  Inaccurate Information   Information  provided  on  an  e‐commerce  website  contains  errors  that  misinform  consumers  in  making transactional decisions   37 [9.89%]   28 [7.49%]   0.76   Incomplete Information   Information  provided  on  an  e‐commerce  website  is  insufficient  to  aid  consumers  in  making  transactional decisions   27 [7.22%]   20 [5.35%]   0.74   Irrelevant Information   Information  provided  on  an  e‐commerce  website  cannot  be  utilized  by  consumers  in  making  transactional decisions   11 [2.94%]   9 [2.41%]   0.82   Untimely Information   Information provided on an e‐commerce website is not updated to support consumers in making  transactional decisions   25 [6.68%]   20 [5.35%]   0.80   Needs Recognition Failure   Functionalities of an e‐commerce website are incapable of assisting consumers to formulate their  needs and preferences for products and/or services   3 [0.80%]   3 [0.80%]   1.00   Alternatives Identification Failure   Functionalities  of  an  e‐commerce  website  are  incapable  of  assisting  consumers  to  gather  information on and source for interested products and/or services   8 [2.14%]   8 [2.14%]   1.00   Alternatives Evaluation Failure   Functionalities  of  an  e‐commerce  website  are  incapable  of  assisting  consumers  to  draw  comparisons among interested products and/or services   1 [0.27%]   1 [0.27%]   1.00   Acquisition Failure   Functionalities of an e‐commerce website are incapable of assisting consumers to place orders for  desired products and/or services   63 [16.84%]   52 [13.90%]   0.83   Post‐Purchase Failure   Functionalities  of  an  e‐commerce  website  are  incapable  of  assisting  consumers  to:  (1)  obtain  purchased products and/or services; (2) solicit advice on ways to maximize the utility of purchased  products and/or services, and; (3) dispose of unwanted products and/or services.   26 [6.95%]   21 [5.61%]   0.81   Inaccessibility   Services of an e‐commerce website are not accessible   73 [19.52%]   64 [17.11%]   0.88   Non‐Adaptability   Services  of  an  e‐commerce  website  are  unable  to  accommodate  diverse  content  and  usage  patterns   18 [4.81%]   17 [4.55%]   0.94   Non‐Navigability   Services of an e‐commerce website are difficult to navigate   28 [7.49%]   21 [5.61%]   0.75   Delay   Services of an e‐commerce website are inordinately slow in access   33 [8.82%]   30 [8.02%]   0.91   Insecurity   Services  of  an  e‐commerce  website  are  not  safeguarded  against  unsanctioned  access  by  unauthorized individuals   7 [1.87%]   7 [1.87%]   1.00   Functional Failures   System Failures   140     Incident Coding  Construct   Definition (Event in which…)   No. Unique  Incidents* [%]   No. Common  + Incidents  [%]   Inter‐Judge  ‡ Reliability    Non‐Transaction‐Oriented Failures  Mischarging   E‐commerce website charges the consumer for unauthorized or unfulfilled purchases   9 [2.41%]   5 [1.34%]   0.56   Product Delivery Problems   Product(s) purchased on an e‐commerce website is not delivered or damaged during delivery   31 [8.29%]   15 [4.01%]   0.48   Unresponsive to Customer Enquiries   Responses to online customer enquiries are not forthcoming   18 [4.81%]   9 [2.41%]   0.50   *  Total number of unique incidents assigned to each category by both judges  + Total number of identical incidents assigned to each category by both judges  ‡ Number of identical incidents divided by number of unique incidents      141     A P P E N D IX C – C L A S S I F I C A T I O N  OF  E X EM P L A R Y E-S E R V I C E F A I L U R E I N C I D E N T S  Proposed E‐Service Failure Typology  Bitner’s (1990, 1994) Typology   Failure Incident   Holloway & Beatty’s (2003)  Typology   Kelley et al.’s (1993) Typology   1st Judge   2nd Judge   1st Judge   2nd Judge   1st Judge   2nd Judge   1st Judge   2nd Judge   Incident  1:  You  must  be  registered  to  be  a  member  and  pay  a  fee  and  usually  through  internet  processors  such  as  PayPal  or  clickbank...with  a  full  refund  policy  within  thirty  days  or  so,  but  they  never  pay  me  back.  So  the  information offered on the site is not accurate.   Inaccurate  Information   Inaccurate  Information   Incorrect  Information  Provided at Site   Incorrect  Information  Provided at Site   Failure to Meet  Customer  Preferences   Other Core  Service Failure   Bad Information   Special Order /  Request Failure   Incident 2: A few times, I was looking to buy some  hair  products  online.  After  spending  a  lot  of  time  adding  products  to  shopping  carts  and  entering  my  contact  information,  I  was  informed  that  the  companies did not mail orders to places outside of  US.  This  was  never  made  known  to  me  before  I  initiated the transaction.   Incomplete  Information   Incomplete  Information   Lack of  Personalized  Information at  Site   Insufficient  Information  Provided at Site   Failure to Meet  ‘Special Needs’  Customers   Informational  Failure   Policy Failure   Policy Failure   Incident  3:  In  the  past,  I  had  been  able  to  view  more  organized  information  about  products,  but  recently, the website began providing me with less  detailed/off‐center  image  information,  which  no  longer meets my needs.   Irrelevant  Information   Irrelevant  Information   Product Poorly  Presented at Site   Poor Customer  Service Support   Failure due to  Out‐of‐the  Ordinary Service  Behavior   Failure due to  Out‐of‐the  Ordinary Service  Behavior   Bad Information   Slow /  Unavailable  Service   Incident  4:  I  wanted  to  buy  a  plane  ticket.  I  was  able  to  choose  the  destination,  date,  and  started  placing  the  order,  then  to  realize  later  that  the  price  changed  during  the  time  i  was  completing  the order.   Untimely  Information   Untimely  Information   Website  Purchasing  Process  Confusing   Website  Purchasing  Process  Confusing   Failure due to  Out‐of‐the  Ordinary Service  Behavior   Failure due to  Out‐of‐the  Ordinary Service  Behavior   System Pricing  Failure   System Pricing  Failure   Incident  5:  Looking  to  buy  something  online  and  searching  for  the  item  I  wanted,  I  can't  find  it  because  the  website  cannot  help  me  to  pinpoint  the item I am looking for.   Needs  Recognition  Failure   Needs  Recognition  Failure   Lack of  Personalized  Information at  Site   Lack of  Personalized  Information at  Site   Failure to Meet  Customer  Preferences   Failure to Pay  Attention to  Customer   Slow /  Unavailable  Service   Attention  Failures   Incident  6:  I  visited  Amazon.com  to  search  for  a  DVD  I  wanted  to  purchase.  I  have  often  searched  for and found things on Amazon.com successfully,  but  because  this  DVD  turned  out  to  be  out  of  print,  it  made  it  harder  to  find  at  a  decent  price.  The only DVDs for sale I could find were over $50,  which  I  was  not  willing  to  spend.  I  couldn't  imagine  that  out  of  all  the  sellers  on  Amazon,  there  wasn't  a  used  DVD  for  cheaper.  After  shuffling  and  searching  around  for  a  very  long  time, I was able to dig deeper than the first search   Alternatives  Identification  Failure   Alternatives  Identification  Failure   Navigational  Problems at Site   Poor Customer  Service Support   Other Core  Service Failure   Failure to Meet  Customer  Preferences   Slow /  Unavailable  Service   System Pricing  Failure   142     Proposed E‐Service Failure Typology  Bitner’s (1990, 1994) Typology   Failure Incident   Holloway & Beatty’s (2003)  Typology   Kelley et al.’s (1993) Typology   1st Judge   2nd Judge   1st Judge   2nd Judge   1st Judge   2nd Judge   1st Judge   2nd Judge   Incident  7:  I  recently  tried  to  order  several  items  from  a  retail  store  via  their  website,  www.kohls.com.  After  choosing  several  products  and  entering  the  desired  quantities,  I  decided  to  visit  Overstock.com  to  compare  prices  for  similar  items before placing the order with Kohl’s.  Before  switching  websites,  I  created  a  username  and  password on the Kohl’s website, assuming that my  "basket" contents would be saved.  However, after  navigating  to  the  Overstock  website  and  then  returning  to  Kohls.com,  my  basket  contents  had  been  cleared.  Other  shopping  sites  that  I've  used  tend  to  be  very  sticky  with  my  basket  contents  even when I am not logged in as a user. As long as  I'm  entering  from  the  same  IP  address,  my  shopping basket contents are usually retained. But  this was not the case on the Kohl’s site. I did not  recreate my online order with them.   Alternatives  Evaluation  Failure   Alternatives  Evaluation  Failure   Lack of  Personalized  Information at  Site   Poor Customer  Service Support   Failure to Meet  Customer  Preferences   Failure to Meet  Customer  Preferences   Attention  Failures   Attention  Failures   Incident  8:  I  wanted  to  purchase  cinema  tickets  online.  I  could  find  the  movie,  theatre,  and  time.  However,  when  I  got  to  the  credit  card  payment,  the  (externally‐powered)  transaction  module  failed  to  validate  my  transactions. I  pay  with  that  card very often on other Websites so I don't think  it  was  due  to  my  card  or  me  entering  the  wrong  info. I tried 4 times to reprocess the payment but  it never managed to process it.   Acquisition  Failure   Acquisition  Failure   Poor Customer  Service Support   Difficulties  Experienced  While Paying   Other Core  Service Failure   Failure due to  Out‐of‐the  Ordinary Service  Behavior   Slow /  Unavailable  Service   Slow /  Unavailable  Service   Incident  9:  I  wanted  to  order  a  video  game  through Amazon.ca, which I had successfully done.  I  was  able  to  add  the  item  to  my  cart  and  successfully  check  out.  A  couple  hours  later,  I  realized  that  I  had  forgotten  to  order  another  item.  Amazon  had  the  option  to  amend  orders  before they were processed, but when I returned  to  my  account;  my  order  had  already  been  processed. My original order was over $39, which  qualified it for free shipping, but the second item  that I wanted to order was not. I did not want to  place another order and have to pay for shipping,   Post‐Purchase  Consultation  Failure   Post‐Purchase  Consultation  Failure   Failure to  Address  Unintentional  Customer  Mistakes   Insufficient  Information  Provided at Site   Failure to  Address  Admitted  Customer Error   Failure to  Address  Admitted  Customer Error   Failure to  Address  Admitted  Customer Error   Attention  Failures   results and find a DVD for $30. I think the search  function  is  poorly  designed.  I  should  have  been  able to find the cheaper DVD without taking such  a long time to search.   143     Proposed E‐Service Failure Typology  Bitner’s (1990, 1994) Typology   Failure Incident  1st Judge   Holloway & Beatty’s (2003)  Typology   Kelley et al.’s (1993) Typology   2nd Judge   1st Judge   2nd Judge   1st Judge   2nd Judge   1st Judge   2nd Judge   Inaccessibility   Poor Customer  Service Support   Poor Customer  Service Support   Failure due to  Out‐of‐the  Ordinary Service  Behavior   Gestalt  Evaluation  Failure   Slow /  Unavailable  Service   Slow /  Unavailable  Service   Non‐Adaptability  Lack of  Personalized  Information at  Site   Lack of  Personalized  Information at  Site   Failure to Meet  Customer  Preferences   Failure to Meet  Customer  Preferences   Slow /  Unavailable  Service   Attention  Failures   Incident  12:  I  had  accessed  the  main  page  and  navigated  through  it  to  the  product  I  was  interested  in.  At  that  point  I  tried  to  use  the  button  allowing  me  to  get  more  information  but  despite  continued  attempts  using  the  button  the  Non‐Navigability  Non‐Navigability  required  page  failed  to  load  and  I  got  an  error  message  stating  the  requested  page  was  unavailable. I attempted several times to go back  to the home page and re‐navigate to this spot but  the requested page failed to load   Poor Customer  Service Support   Product Poorly  Presented at Site   Other Core  Service Failure   Other Core  Service Failure   Slow /  Unavailable  Service   Slow /  Unavailable  Service   Incident  13: I choose the laptop I wanted to buy.  Then  I  was  redirected  on  the  site  for  the  credit  card  payment.  I  entered  my  credit  card  information, number and expiry date, and clicked  on PROCESS. Nothing happened. 5 minutes later I  clicked  again  on  PROCESS.  Nothing  happened  again.  I  clicked  again  5  minutes  later  and  it  worked.   Delay   Delay   Navigational  Problems at Site   Website  Purchasing  Process  Confusing   Unreasonably  Slow Service   Failure due to  Out‐of‐the  Ordinary Service  Behavior   Slow /  Unavailable  Service   Slow /  Unavailable  Service   Incident  14:  I  logged  on  to  my  account  and  was  hijacked  to  a  site  to  enter  a  sweepstakes  instead  that had the terms and conditions to participate in  several  levels  of  "reward  programs".  These  seem  to lead to endless and expensive participations.   Insecurity   Insecurity   Lack of  Personalized  Information at  Site   Lack of  Personalized  Information at  Site   Failure due to  Out‐of‐the  Ordinary Service  Behavior   Failure to Pay  Attention to  Customer   Attention  Failures   Slow /  Unavailable  Service   when  I  could have  just  added  the  second  item to  go  with  the  first,  and  get  free  shipping  for  both  items.  In  the  end,  I  decided  not  to  order  the  second item.  Incident 10: I went to Amazon.com to purchase a  present  for  my  husband.  I  got  almost  the  whole  way  through  the  checkout  process  before  the  website  malfunctioned  on  my  browser  and  I  lost  my order.   Inaccessibility   Incident  11:  When  I  went  to  send  an  email  to  inquiry  about  my  purchase  order,  the  website  asked for my name, address, account number, etc.  I could not proceed further because when it came  Non‐Adaptability time to enter my STATE I couldn't because it was  an American site and the STATE section could only  be  filled  out  from  a  pre‐installed  list.  I  am  from  Canada and I couldn't override it.   144     Proposed E‐Service Failure Typology  Bitner’s (1990, 1994) Typology   Failure Incident   Holloway & Beatty’s (2003)  Typology   Kelley et al.’s (1993) Typology   1st Judge   2nd Judge   1st Judge   2nd Judge   1st Judge   2nd Judge   1st Judge   2nd Judge   Mischarging   Mischarging   Purchase Never  Delivered   Purchase Never  Delivered   Failure to Pay  Attention to  Customer   Failure due to  Out‐of‐the  Ordinary Service  Behavior   Policy Failure   Policy Failure   Incident 16: I buy things from them several times  a  year  and  have  done  so  for  years.  One  of  the  items was meant to be a gift and according to the  shipping estimate would have arrived in plenty of  Product Delivery  Product Delivery  time.  The  week  the  item  was  scheduled  to  be  Problems  Problems  delivered  I  received  every  other  item  I  have  ordered  except  the  gift.  What  failed  was  Amazon  NOT  informing  that  an  item  is  being  shipped  or  not available on the date promised.   Purchase Never  Delivered   Purchase Never  Delivered   Other Core  Service Failure   Other Core  Service Failure   Packaging Errors   Packaging Errors   Incident  17:  I  was  able  to  easily  find  the  product  that I wanted to purchase. I saw that there was an  area on the website where I could ask a question  Unresponsive to  Customer  to  which  I  submitted  my  query.  I  submitted  my  Enquiries  query  and  after  two  days,  had  not  received  a  response.  I  submitted  another  query,  and  waited  an additional two days and still nothing.   Poor  Communication  with the  Company   Poor  Communication  with the  Company   Failure to Pay  Attention to  Customer   Unreasonably  Slow Service   Attention  Failures   Attention  Failures   Incident  15:  I  was  on  bearshare.com  wanting  to  join so i could download some music. I was to pay  $60 for the year after i put in the information and  my  card  was  charged,  the  page  would  not  finish  submitting  and  I  received  no  membership  to  download music and was out by $60 and have not  heard from the web site.  I have complained to the  website and requested my money back.   Unresponsive to  Customer  Enquiries      145     A P P E N D IX D TECHNOLOGY E‐Service Recovery  Compensation   – IN  ILLUSTRATIVE  EXAMPLES  OF  E-S E R V I C E  RECOVERY  PRACTICE Technological Implementation   Offer  Self‐Serving  Help  Centers  (see  example  below)  for  consumers  to  seek  compensation  for  negative  transactional  experiences   Amazon.com provides various options for consumers to state the problem encountered    Affinity   Offer Apology (see example below) to consumers regarding any negative transaction experience   Amazon.com apologizes to consumers for a transactional error        146     E‐Service Recovery  Response Sensitivity   Technological Implementation  Offer  Evaluation/Inquiry  Forms  (see  example  below)  for  consumers  to  provide  feedback  regarding  any  negative  transaction experience   Amazon.com provides a general template for consumers to give feedback that can range across multiple predefined topics    Initiative   Offer Proactive  Feedback  Mechanisms (see example below) to prompt consumers to reflect on various aspects of their  transactional experience in order to identify unreported areas of concern   Mazda.ca proactively enlists the assistance from visitors to complete an online survey questionnaire as they browse the website     147     A P P E N D IX E – D U N N E T T T3 T E S T F O R E-S E R V I C E F A I L U R E M A N I P U L A T I O N S [F A I L U R E T R E A T M E N T C O M P A R I S O N S ] Dependent Variable  Informational Failure   (I) Failure Treatment  No Failure   Out‐of‐Stock   (J) Failure Treatment   Mean Difference (I‐J)   Informational Failure   *  .000   Missing Comparison   ‐.33527  .659   Delay   ‐.19543  .954   2.03582  *  .000   1.70054  *  .000   1.84038  *  .000   .33527  .659   *  .000   Delay   .13984  .885   No Failure   .19543  .954   *  .000   Missing Comparison   ‐.13984  .885   Informational Failure   ‐.46321  .336   *  .000   ‐.15875  .985   .46321  .336   *  .000   .30446  .225   1.88370  *  .000   Informational Failure   1.42049  *  .000   Delay   1.72495  *  .000   .15875  .985   ‐.30446  .225   *  .000   Informational Failure   ‐.56902  .110   Missing Comparison   ‐.53799  .153   *  .000   No Failure   .56902  .110   Missing Comparison   .03103  1.000   *  .000   .53799  .153   ‐.03103  1.000   *  .000   1.93543  *  .000   1.36641  *  .000   No Failure  Missing Comparison  Delay   Missing Comparison   No Failure  Informational Failure   Delay   Informational Failure   Functional Failure   No Failure   Missing Comparison  Delay  Out‐of‐Stock   No Failure  Missing Comparison  Delay   Missing Comparison   Delay   No Failure   No Failure  Informational Failure  Missing Comparison   System Failure   No Failure   Delay  Out‐of‐Stock   Delay  Missing Comparison   No Failure  Informational Failure  Delay   Delay   No Failure  Informational Failure   ‐2.03582  ‐1.70054  ‐1.84038  ‐1.88370  ‐1.42049  ‐1.72495  ‐1.93543  ‐1.36641  ‐1.39745  Sig.   148     Dependent Variable   (I) Failure Treatment   (J) Failure Treatment  Missing Comparison   Mean Difference (I‐J)  1.39745  *  Sig.  .000   The mean difference is significant at the 0.05 level.   149     A P P E N D IX F – D U N N E T T T3 T E S T F O R E-S E R V I C E M A N I P U LA T I O N S [R E C O V E R Y T R E A T M E N T C O M P A R I S O N S ]  RECOVERY  Dependent Variable   (I) Type of Recovery   (J) Type of Recovery   Compensation   No Recovery   Feedback   .34768   .988   Apology   .28449   .999   Apology x Feedback   .27043   1.000  .000   *  .000   *  .000   *  2.37667    .000   No Recovery   ‐.34768   .988   Apology   ‐.06319   1.000   Apology x Feedback   ‐.07725   1.000   Discount x Feedback  Discount x Apology  Discount x Apology x Feedback   Discount  Discount x Feedback  Discount x Apology  Discount x Apology x Feedback  Apology   No Recovery  Feedback  Apology x Feedback  Discount  Discount x Feedback   Apology x Feedback   2.54580   2.49768   2.44493    *  .000   *  .000   *  .000   *  2.02899    .000   ‐.28449   .999   .06319   1.000   ‐.01406   1.000   2.19812   2.15000   2.09725    *  .000   *  .000   *  .000   *  2.26130   2.21319    Discount x Apology   2.16043    Discount x Apology x Feedback   2.09217    .000   ‐.27043   1.000   Feedback   .07725   1.000   Apology   .01406   1.000   No Recovery   Discount  Discount x Feedback  Discount x Apology  Discount x Apology x Feedback  Discount   Sig.   *  Discount   Feedback   Mean Difference (I‐J)   No Recovery  Feedback  Apology   *  .000   *  .000   *  .000   *  .000   *  .000   *  .000   *  .000   *  2.27536   2.22725   2.17449   2.10623   ‐2.54580   ‐2.19812   ‐2.26130    Apology x Feedback   ‐2.27536    .000   Discount x Feedback   ‐.04812   1.000   Discount x Apology   ‐.10087   1.000   Discount x Apology x Feedback   ‐.16913   1.000   150     Dependent Variable   (I) Type of Recovery  Discount x Feedback   (J) Type of Recovery  No Recovery  Feedback   .04812   1.000   Discount x Apology   ‐.05275   1.000   Discount x Apology x Feedback   ‐.12101   1.000   *  .000   *  .000   *  .000   *  ‐2.17449    .000   Discount   .10087   1.000   Discount x Feedback   .05275   1.000   ‐.06826   1.000   No Recovery   ‐2.44493   ‐2.09725   ‐2.16043    *  .000   *  .000   *  .000   *  ‐2.10623    .000   Discount   .16913   1.000   Discount x Feedback   .12101   1.000   Discount x Apology   .06826   1.000   Feedback   .45377   .972   No Recovery  Feedback  Apology  Apology x Feedback   Apology  Apology x Feedback  Discount  Discount x Feedback  Discount x Apology  Discount x Apology x Feedback  No Recovery   ‐2.37667   ‐2.02899   ‐2.09217    *  .000   *  .000   *  1.02435    .038   .71928   .587   1.75333   2.16449    *  .000   *  2.39116    .000   ‐.45377   .972   2.08188    *  .000   *  1.71072    .000   Discount   .57058   .874   Discount x Feedback   .26551   1.000   Apology  Apology x Feedback   Discount x Apology  Discount x Apology x Feedback  Apology   *  .000   Discount x Apology x Feedback   Feedback   .000   ‐2.22725    Apology x Feedback   No Recovery   .000   *  ‐2.15000    Apology x Feedback   Apology   Affinity   .000   *  ‐2.49768    ‐2.21319    Feedback   Discount x Apology x Feedback   Sig.   *  Apology   Discount   Discount x Apology   Mean Difference (I‐J)   No Recovery  Feedback  Apology x Feedback   1.29957    *  .000   *  .000   *  .000   *  ‐1.29957    .000   .41116   .369   1.62812   1.93739   ‐1.75333    151     Dependent Variable   (I) Type of Recovery   (J) Type of Recovery  Discount  Discount x Feedback  Discount x Apology  Discount x Apology x Feedback   Apology x Feedback   No Recovery  Feedback  Apology  Discount  Discount x Feedback  Discount x Apology  Discount x Apology x Feedback   Discount   No Recovery   .168   ‐1.03406    .013   .32855   .720   *  .006   *  .000   *  ‐1.71072    .000   ‐.41116   .369   .63783   ‐2.16449    *  .001   *  ‐1.44522    .000   ‐.08261   1.000   .22667   .997   ‐1.14014    *  Feedback   ‐.57058   .874   Apology   .72899   .168   *  Apology x Feedback   1.14014    .001   Discount x Feedback   ‐.30507   1.000   *  .004   *  1.36681    .000   No Recovery   ‐.71928   .587   Feedback   ‐.26551   1.000   Apology  Apology x Feedback  Discount  Discount x Apology  Discount x Apology x Feedback  No Recovery  Feedback  Apology  Apology x Feedback  Discount  Discount x Feedback  Discount x Apology x Feedback  Discount x Apology x Feedback   *  .038   Discount x Apology x Feedback   Discount x Apology   ‐.72899   Sig.   ‐1.02435    Discount x Apology   Discount x Feedback   Mean Difference (I‐J)   1.05754    *  .013   *  1.44522    .000   .30507   1.000   1.03406    *  .000   *  .000   *  .000   *  ‐1.62812    .000   ‐.32855   .720   .08261   1.000   1.36261   1.67188   ‐2.08188    *  .004   *  ‐1.36261    .000   .30928   .865   ‐1.05754    *  .000   *  .000   *  ‐.63783    .006   ‐.22667   .997   No Recovery   ‐2.39116    Feedback   ‐1.93739    Apology  Apology x Feedback  Discount  Discount x Feedback   *  .000   *  .000   ‐1.36681   ‐1.67188    152     Dependent Variable   (I) Type of Recovery   (J) Type of Recovery  Discount x Apology   Response Sensitivity   No Recovery   Apology   ‐.16928   1.000   Discount x Apology  Discount x Apology x Feedback  No Recovery  Apology  Apology x Feedback  Discount  Discount x Feedback  Discount x Apology   .000   .49290   .949   *  1.73942    .000   .43043   .994   *  .000   *  .000   *  ‐1.49826    .000   .37174   .564   1.65246   ‐1.32899    *  ‐.83609    .019   .41043   .333   *  .017   Discount x Apology x Feedback   .32348   .812   No Recovery   .16928   1.000   Apology x Feedback  Discount  Discount x Feedback  Discount x Apology  Discount x Apology x Feedback  No Recovery  Feedback  Apology  Discount  Discount x Feedback  Discount x Apology   *  .000   *  1.87000    .000   .66217   .399   1.49826    *  1.90870    .000   .59971   .678   *  .000   *  ‐1.70072    .000   ‐.37174   .564   1.82174    *  .000   *  ‐1.20783    .000   .03870   1.000   ‐1.87000    *  ‐1.27029    .000   Discount x Apology x Feedback   ‐.04826   1.000   No Recovery   ‐.49290   .949   *  Feedback   .83609    .019   Apology   ‐.66217   .399   *  .000   *  Apology x Feedback   1.20783    Discount x Feedback   1.24652    .000   Discount x Apology   ‐.06246   1.000   Discount x Apology x Feedback  Discount x Feedback   *  1.70072    ‐.89855    Feedback   Discount   .865  .000   Discount x Feedback   Apology x Feedback   *  1.32899    Discount   Apology   ‐.30928   Sig.   Feedback   Apology x Feedback   Feedback   Mean Difference (I‐J)   No Recovery  Feedback   *  .000   *  ‐1.73942    .000   ‐.41043   .333   1.15957    153     Dependent Variable   (I) Type of Recovery   (J) Type of Recovery  Apology  Apology x Feedback   Discount x Apology   *  Sig.   ‐1.90870    .000   ‐.03870   1.000   *  .000   *  Discount   ‐1.24652    Discount x Apology   ‐1.30899    .000   Discount x Apology x Feedback   ‐.08696   1.000   No Recovery   ‐.43043   .994   *  Feedback   .89855    .017   Apology   ‐.59971   .678   Apology x Feedback  Discount  Discount x Feedback  Discount x Apology x Feedback  Discount x Apology x Feedback   Mean Difference (I‐J)   No Recovery  Feedback  Apology  Apology x Feedback  Discount  Discount x Feedback  Discount x Apology   *  1.27029    .000   .06246   1.000   *  .000   *  .000   *  ‐1.65246    .000   ‐.32348   .812   1.30899   1.22203    *  ‐1.82174    .000   .04826   1.000   *  ‐1.15957    .000   .08696   1.000   *  ‐1.22203    .000   *. The mean difference is significant at the 0.05 level.   154     A P P E N D IX G – D U N N E T T T3 T E S T F O R I M P A C T O F E-S E R V I C E F A I L U R E S D I S C O N F I R M E D E X P E C T A N C I E S [F A I L U R E T R E A T M E N T C O M P A R I S O N S ] Dependent Variable  Disconfirmed Outcome Expectancy   (I) Failure Treatment  No Failure   (J) Failure Treatment  Out‐of‐Stock  Missing Comparison  Delay   Out‐of‐Stock   No Failure  Missing Comparison  Delay   Missing Comparison   Delay   No Failure   No Failure   Out‐of‐Stock   No Failure   No Failure  Out‐of‐Stock  Delay  No Failure  Out‐of‐Stock  Missing Comparison   No Failure   Missing Comparison   .000   *  .001   *  ‐1.0074   ‐1.0213    *  .012   *  1.0074    .000   .2266   .624   ‐.7946    *  .000   *  .000   *  .001   *  ‐1.2823    .000   .4095   .055   ‐.3205   .306   1.2823   1.6918   .9618    *  ‐1.6918    .000   ‐.4095   .055   *  .000   *  ‐.9618    .001   .3205   .306   ‐.7300    *  .000   Out‐of‐Stock   .7192   .060   Missing Comparison   .5978   .157   *  1.0943    .002   No Failure   ‐.7192   .060   Missing Comparison   ‐.1214   .913   *  Delay   .3751    .031   No Failure   ‐.5978   .157   .1214   .913   Out‐of‐Stock  Delay  Delay   .000   *  .7300    Delay  Out‐of‐Stock   *  ‐.7808    .624   Delay   Disconfirmed Cost Expectancy   .000   ‐1.8020    ‐.2266   Missing Comparison   Delay   .012   *  .7946    Delay   Delay   Missing Comparison   .001   *  1.0213    .000   Missing Comparison   Out‐of‐Stock   .000   *  1.8020    .7808    Missing Comparison  No Failure   Sig.   *  Out‐of‐Stock   Out‐of‐Stock   Disconfirmed Process Expectancy   Mean Difference (I‐J)   No Failure  Out‐of‐Stock   *  .001   *  .002   *  .031   .4965   ‐1.0943   ‐.3751    ON  155     Dependent Variable   (I) Failure Treatment   (J) Failure Treatment  Missing Comparison   Mean Difference (I‐J)  *  ‐.4965    Sig.  .001   156     A P P E N D IX H – D U N N E T T T3 T E S T F O R I M P A C T O F E-S E R V I C E R E C O V E R I E S O N D I S C O N F I R M E D E X P E C T A N C I E S [R EC O V E R Y T R E A T M E N T C O M P A R I S O N S ] Dependent Variable   (I) Type of Recovery   Disconfirmed Outcome Expectancy  Feedback  Difference   Mean Difference (I‐J)  Sig.  .0438  1.000  Apology x Feedback   ‐.1113  1.000  Discount   ‐.5654  .054  *  .000  *  .001  *  .001  Feedback   ‐.0438  1.000  Apology x Feedback   ‐.1551  .986  *  .004  *  .000  *  .000  *  .000  Feedback   .1113  1.000  Apology   .1551  .986  Discount   ‐.4541  .122  *  .000  *  .002  *  .001  .5654  .054  *  .004  Apology x Feedback   .4541  .122  Discount x Feedback   ‐.4686  .585  Discount x Apology   ‐.3283  .945  Discount x Apology x Feedback   ‐.2754  .978  *  .000  *  .000  *  .000  Discount   .4686  .585  Discount x Apology   .1403  1.000  Discount x Apology x Feedback   .1932  1.000  *  .001  *  .000  *  .002  .3283  .945  ‐.1403  1.000  (J) Type of Recovery  Apology   Discount x Feedback   ‐1.0339  Discount x Apology   ‐.8936  Discount x Apology x Feedback  Apology   Discount  Discount x Feedback  Discount x Apology  Discount x Apology x Feedback  Apology x Feedback   Discount x Feedback  Discount x Apology  Discount x Apology x Feedback  Discount   Feedback  Apology   Discount x Feedback   Feedback  Apology  Apology x Feedback   Discount x Apology   Feedback  Apology  Apology x Feedback  Discount  Discount x Feedback   ‐.8407  ‐.6091 ‐1.0777  ‐.9374  ‐.8845  ‐.9226 ‐.7823  ‐.7294  .6091  1.0339 1.0777 .9226  .8936 .9374  .7823  157     Dependent Variable   (I) Type of Recovery   Mean Difference (I‐J)  Sig.  .0529  1.000  *  .001  *  .000  *  .001  .2754  .978  Discount x Feedback   ‐.1932  1.000  Discount x Apology   ‐.0529  1.000  .2991  .765  ‐.5223  .433  .2507  .946  Discount x Feedback   ‐.3242  .889  Discount x Apology   .1106  1.000  Discount x Apology x Feedback   ‐.4884  .436  Feedback   ‐.2991  .765  *  .003  ‐.0484  1.000  *  .006  ‐.1886  .987  *  .002  .5223  .433  *  .003  *  .008  Discount x Feedback   .1981  1.000  Discount x Apology   .6329  .101  Discount x Apology x Feedback   .0339  1.000  Feedback   ‐.2507  .946  Apology   .0484  1.000  *  .008  *  .019  ‐.1401  1.000  *  .005  .3242  .889  (J) Type of Recovery  Discount x Apology x Feedback   Discount x Apology x Feedback   Feedback   .8407  Apology   .8845  Apology x Feedback  Discount   Disconfirmed Process Expectancy  Difference   Feedback   Apology  Apology x Feedback  Discount   Apology   Apology x Feedback  Discount  Discount x Feedback  Discount x Apology  Discount x Apology x Feedback  Apology x Feedback   Feedback  Apology  Discount   Discount   Apology x Feedback  Discount x Feedback  Discount x Apology  Discount x Apology x Feedback  Discount x Feedback   Feedback   ‐.8214  ‐.6233  ‐.7875  .8214  .7730  ‐.7730 ‐.5749  ‐.7391  Apology   .6233  *  .006  Apology x Feedback   ‐.1981  1.000  *  .019  .4348  .323  Discount x Apology x Feedback   ‐.1642  1.000  Feedback   ‐.1106  1.000  Discount  Discount x Apology   Discount x Apology   .7294  .5749  158     Dependent Variable   (I) Type of Recovery   Mean Difference (I‐J)  Sig.  .1886  .987  ‐.6329  .101  .1401  1.000  Discount x Feedback   ‐.4348  .323  Discount x Apology x Feedback   ‐.5990  .087  .4884  .436  Apology   *  .7875  .002  Apology x Feedback   ‐.0339  1.000  *  .005  Discount x Feedback   .1642  1.000  Discount x Apology   .5990  .087  Apology   ‐.0965  1.000  Apology x Feedback   ‐.2122  .998  .1594  1.000  Discount x Feedback   ‐.0775  1.000  Discount x Apology   .1978  1.000  Discount x Apology x Feedback   .0583  1.000  Feedback   .0965  1.000  ‐.1157  1.000  Discount   .2559  .989  Discount x Feedback   .0190  1.000  Discount x Apology   .2943  .978  Discount x Apology x Feedback   .1548  1.000  Feedback   .2122  .998  Apology   .1157  1.000  Discount   .3716  .826  Discount x Feedback   .1346  1.000  Discount x Apology   .4100  .793  Discount x Apology x Feedback   .2704  .993  Feedback   ‐.1594  1.000  Apology   ‐.2559  .989  Apology x Feedback   ‐.3716  .826  Discount x Feedback   ‐.2370  .994  Discount x Apology   .0384  1.000  ‐.1012  1.000  Feedback   .0775  1.000  Apology   ‐.0190  1.000  Apology x Feedback   ‐.1346  1.000  (J) Type of Recovery  Apology  Apology x Feedback  Discount   Discount x Apology x Feedback   Feedback   Discount   Disconfirmed Cost Expectancy  Difference   Feedback   Discount   Apology   Apology x Feedback   Apology x Feedback   Discount   Discount x Apology x Feedback  Discount x Feedback   .7391  159     Dependent Variable   (I) Type of Recovery   Discount x Apology   Discount x Apology x Feedback   Mean Difference (I‐J)  Sig.  Discount   .2370  .994  Discount x Apology   .2754  .988  Discount x Apology x Feedback   .1358  1.000  Feedback   ‐.1978  1.000  Apology   ‐.2943  .978  Apology x Feedback   ‐.4100  .793  Discount   ‐.0384  1.000  Discount x Feedback   ‐.2754  .988  Discount x Apology x Feedback   ‐.1396  1.000  Feedback   ‐.0583  1.000  Apology   ‐.1548  1.000  Apology x Feedback   ‐.2704  .993  .1012  1.000  Discount x Feedback   ‐.1358  1.000  Discount x Apology   .1396  1.000  (J) Type of Recovery   Discount   160     A P P E N D IX I – G R A P H I C A L P L O T S DISCONFIRMED EXPECTANCIES Dependent Variable   OF  IMPACT  OF  E-S E R V I C E R E C O V E R I E S  ON  Graphical Plot   Disconfirmed  Outcome Expectancy      161     Dependent Variable   Graphical Plot   Disconfirmed Process  Expectancy      162     Dependent Variable   Graphical Plot   Disconfirmed Cost  Expectancy      163     

Cite

Citation Scheme:

        

Citations by CSL (citeproc-js)

Usage Statistics

Share

Embed

Customize your widget with the following options, then copy and paste the code below into the HTML of your page to embed this item in your website.
                        
                            <div id="ubcOpenCollectionsWidgetDisplay">
                            <script id="ubcOpenCollectionsWidget"
                            src="{[{embed.src}]}"
                            data-item="{[{embed.item}]}"
                            data-collection="{[{embed.collection}]}"
                            data-metadata="{[{embed.showMetadata}]}"
                            data-width="{[{embed.width}]}"
                            async >
                            </script>
                            </div>
                        
                    
IIIF logo Our image viewer uses the IIIF 2.0 standard. To load this item in other compatible viewers, use this url:
http://iiif.library.ubc.ca/presentation/dsp.24.1-0102435/manifest

Comment

Related Items