Open Collections

UBC Theses and Dissertations

UBC Theses Logo

UBC Theses and Dissertations

How can I help you? : the role of recommendation agents in collaborative online shopping Huang, Shan 2012

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

Item Metadata

Download

Media
24-ubc_2012_fall_huang_shan.pdf [ 3.42MB ]
Metadata
JSON: 24-1.0307453.json
JSON-LD: 24-1.0307453-ld.json
RDF/XML (Pretty): 24-1.0307453-rdf.xml
RDF/JSON: 24-1.0307453-rdf.json
Turtle: 24-1.0307453-turtle.txt
N-Triples: 24-1.0307453-rdf-ntriples.txt
Original Record: 24-1.0307453-source.json
Full Text
24-1.0307453-fulltext.txt
Citation
24-1.0307453.ris

Full Text

How Can I Help You? The Role of Recommendation Agents in Collaborative Online Shopping by Shan Huang B.M., Tsinghua University, China, 2010 A THESIS SUBMITTED IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF MASTER OF SCIENCE in The Faculty of Graduate Studies (Business Administration) THE UNIVERSITY OF BRITISH COLUMBIA (Vancouver) October 2012 © Shan Huang, 2012ii Abstract This research examines two important, understudied questions: whether or not recommendation agents (RAs) assist collaborative online shopping (COS), in which two or more shoppers communicate directly and synchronously, and whether or not this depends on how restrictively the RA provides its service. Results from a laboratory experiment show that compared to the use of no RA or a restrictive RA, the use of a flexible RA leads to higher perceived shopping enjoyment (hedonic value).  Compared to the use of no RA, the use of a flexible RA or a restrictive RA is associated with higher perceptions of usefulness. However, no difference was found among the three conditions (no RA, flexible RA, and restrictive RA) in terms of perceived purchase decision quality (utilitarian value), nor did restrictiveness significantly affect perceived usefulness.  To look beneath the surface of these results, I examined protocols of the social interactions that took place during the task and I analyzed them with Interaction Process Analysis (IPA).  The results suggest that flexible RAs were preferred over no RAs or restrictive RAs because they strike a balance between leadership and empowerment during the social encounter. iii Preface The research described in this dissertation received approval from the Behavioural Research Ethics Board of the University of British Columbia (Certificate number: H11-00645). iv Table of Contents Abstract ........................................................................................................................ iiPreface ......................................................................................................................... iiiTable of Contents ..................................................................................................... ivList of Tables ............................................................................................................. viList of Figures .......................................................................................................... viiAcknowledgements ............................................................................................... viii1. Introduction .......................................................................................................... 12. Literature Review and Background ............................................................... 32.1. Interactive Decision Aids for Collaborative Online Shopping ............................ 3 2.2. Group Decision Support Systems and Collaborative Online Shopping .............. 5 2.3. Restrictiveness and Collaborative Online Shopping ............................................ 7 3. Research Model and Hypothesis Development .......................................... 93.1. Overview .............................................................................................................. 9 3.2. Effects on Perceived Shopping Enjoyment (Hedonic Value) ............................ 10 3.3. Effects on Perceived Purchase Decision Quality (Utilitarian Value) ................ 12 3.4. Effects on Perceived Usefulness and Intention to Reuse ................................... 14 4. Research Methodology .................................................................................... 174.1. Experimental Design .......................................................................................... 17 4.2. Sample and Incentive ......................................................................................... 19 4.3. Procedures .......................................................................................................... 20 5. Data Analysis ..................................................................................................... 215.1. Subject Demographics ....................................................................................... 21 5.2. Unit of Analysis ................................................................................................. 21 5.3. Measurement ...................................................................................................... 22 v 5.4. Manipulation Checks ......................................................................................... 25 5.5. MANCOVA Results .......................................................................................... 25 5.6. ANCOVA Results .............................................................................................. 26 5.7. Regression Results ............................................................................................. 27 5.8. Discussion of Results ......................................................................................... 28 6. Supplementary Analysis ................................................................................. 326.1. Coding of Shopping Transcripts ........................................................................ 32 6.2. Amounts and Patterns of Communications ........................................................ 34 6.3. Results of Interaction Process Analysis and Discussion .................................... 35 7. Contribution ....................................................................................................... 408. Limitations and Future Research .................................................................. 43References .................................................................................................................. 44Appendices ................................................................................................................ 52Appendix A: Design of Experimental Systems ................................................ 52Appendix B: Measurement Items ........................................................................ 57vi List of Tables Table 1: Factor Analysis ................................................................................................. 23 Table 2: Descriptive Statistics for Outcome and Control Variables ............................... 24 Table 3: Reliabilities and Correlation Between Constructs ............................................ 24 Table 4: MANCOVA Results ......................................................................................... 25 Table 5: ANCOVA Results ............................................................................................ 26 Table 6: Pairwise Comparisons ...................................................................................... 27 Table 7: Coefficients ....................................................................................................... 27 Table 8: Summary of Hypotheses Testing Results ......................................................... 28 Table 10: Descriptive Statistics for IPA ......................................................................... 35 Table 11: ANOVA Results ............................................................................................. 35 Table 12: Pairwise Comparisons .................................................................................... 36 Table A-1: Measurement Items of Outcome and Control Variables .............................. 57 vii List of Figures Figure 1: Research Model ................................................................................................. 9 Figure 2: Experimental Treatments ................................................................................ 19 Figure 3: Research Model ............................................................................................... 29 Figure 5: Expanded Research Models ............................................................................ 36 Figure A-1: Control Condition – Home Page ................................................................. 52 Figure A-2: Control Condition – Information Page of Products .................................... 52 Figure A-3: Control Condition – Shopping Cart ............................................................ 53 Figure A-4: Flexible RA – Homepage ............................................................................ 53 Figure A-5: Flexible RA – Sample Question of Attributes ............................................ 54 Figure A-6 Flexible RA – Recommendation Page ......................................................... 54 Figure A-7 Flexible RA – Helpful Tips .......................................................................... 55 Figure A-8 Restrictive RA – Home Page ....................................................................... 55 Figure A-9: Restrictive RA – Sample Question of Attributes ........................................ 56 Figure A-10 Restrictive RA – Recommendation Page ................................................... 56 viii Acknowledgements I would like to thank and express my heartfelt gratitude to my advisor, Prof. Izak Benbasat. His guidance, support, and positive attitude throughout the study encouraged me to keep up with the work all the way through. I am also very much grateful to my co-supervisor, Prof. Andrew Burton-Jones, for offering his feedback to my endless questions and inspiring me to continue my work and fulfill my potential.  Special thanks are owed to my parents, who have supported me throughout my years of education, both morally and financially. 1 1. IntroductionWe begin with an occasion in real life. Consider a couple that goes to a store to purchase a digital camera as a birthday gift for their son. There are three ways in which they might make their purchase decision. First, they may look around, evaluate the cameras together, and make the purchase decision unaided by anyone else.  Second, they may get a salesperson as soon as they step into the store and follow his or her guidance to select a camera. Third, they may choose an option in between, getting help from a salesperson if needed and following their own path to make the decision.  Two questions emerge from this scenario. First, does a salesperson indeed help in collaborative shopping? Second, how much flexibility (or restrictiveness) should a salesperson provide for the collaborative shoppers to make decisions?  Given the commonality of such a shopping scenario, we might expect that there would be substantial research addressing these two questions.  Surprisingly, there is not. There is even less research that has examined this in an IT context.  Given that recommendation agents are the equivalent of salespersons in online environments (Wang and Benbasat 2009), these two questions closely correspond with two important and understudied issues for the design of a collaborative online shopping system (COS): (1) whether an RA should be adopted for COS; (2) how flexible or restrictive should the design of the RA be. The objective of this study is to explore these two questions theoretically and empirically. I define COS as an activity, in which one shopper shops at an online store together and concurrently with one or more co-located shopping partners for a shared shopping goal (Zhu et al. 2010).  This will then provide a baseline for studying COS in situations in which shoppers are connected only virtually, or when they have different shopping goals.  By evaluating alternatives according to consumers’ personal preferences and offering them personalized advice, RAs aim to make the online shopping task more effective and efficient (Xiao and Benbasat 2007). However, prior researchers and practitioners have hardly 2 considered RA use in a collaborative shopping context. Theoretically, little research has been conducted on the role of RAs in COS, although a wealth of studies have explored the effects of RAs on individual purchase-decision making and generally suggest that RAs can help individuals make more effective and efficient purchase decisions (Huabl and Murray 200). Practically, few systems are designed to support collaborative shopping online. Instead, most e-commerce systems, like RAs, are designed for the solitary user who buys items for him/herself, despite the fact that collaborative shopping is undertaken frequently in real life and the collaborative use of technology has long been an important part of practice (Brown et al. 2010). How should systems be designed to facilitate collaborative shopping? Alternatively, could they, in fact, hinder the collaborative process and outcomes?  In particular, given the positive effects of RAs on individuals’ shopping outcomes in past research, could RAs help in this context, and if so, how?  To facilitate collaborative shopping online, we first need to know how online shopping software such as RAs can influence collaborative shopping behaviors, and learn what aspects of them need to be improved for COS.  3 2. Literature Review and Background2.1. Interactive Decision Aids for Collaborative Online Shopping Interactive decision aids are designed to overcome information overload, reduce the complexity of information search, and facilitate purchase decision-making (Xiao and Benbasat 2007). Such capabilities of interactive decision aids are particularly valuable in online shopping environments. Electronic information can easily overwhelm online consumers with large volumes of data, straining their attention, memory and motivation (Häubl and Murray 2006). The absence of sales persons and physical contact with products makes online purchase decision making even more difficult. Interactive decisions aids such as RAs have therefore attracted significant attention in academia and industry (e.g., Haubl and Trifts 2000; Haubl and Murray 2003).  Almost all the prior studies on RA effects follow an information-processing model that regards the consumer as a logical thinker who solves problems to make purchase decisions (Bettman 1979), and they have emphasized the utilitarian dimensions of RA use (Xiao and Benbasat 2007). Ample evidence has been produced to support the positive utilitarian consequences of RA use, such as its potential to reduce individual users’ decision effort (e.g., Dellaert and Haubl 2005; Haubl and Murray 2006; Hostler et al. 2005), increase their decision quality (e.g. Haubl and Trifs 2000; Olson and Widing 2002; Dellaert and Haubl 2005; Hostler et al. 2005), and contribute to more favorable evaluations of shopping systems (e.g., Xiao and Benbasat 2007).   However, the experiential view, which highlights consumers’ hedonic responses to the shopping process, long recognized as an important part of consumer research (Holbrook and Hirschman 1982), has been neglected by conventional RA research.  The experiential view reminds us that shopping is not just an exercise in utility maximization; rather, it can also involve a “steady flow of feelings and fun” (Holbrook and Hirschman 1982, p.132).  Only a handful of consumer studies have examined the “fun side” of shopping and acknowledged 4 that shopping experiences can produce both utilitarian and hedonic value (e.g., Bloch and Bruce 1984, Hirschman 1984, Babin et al. 1994, Mathwick and Rigdon 2004). Because past research has overwhelmingly adopted a “shopping as work” assumption, it is currently unable to give a full account of how best to support consumers’ hedonic and utilitarian goals, especially when online.      One possible reason why so few studies have examined the influence of RA use on hedonic value is that RAs are typically intended to improve the utilitarian value of shopping (e.g., improving information processing ability, increasing decision quality and saving effort).  However, understanding their effect on hedonic value is also necessary.  First, IT can have both intended and unintended effects.  Thus, it is important to study both the outcomes that RAs were designed for (utilitarian value) and the outcomes they were not designed for, but may still affect, such as hedonic value.  Second, we know that shopping value includes both utilitarian and hedonic dimensions (Childers et al. 2001; Babin et al. 1994) and that the hedonic dimension is very important for shoppers, especially social shoppers (Bloch and Bruce 1984). People are particularly motivated to derive enjoyment from shopping with friends and/or family members (Arnold et al. 2003). Thus, if RA use increases the utilitarian value of COS but reduces the hedonic value, they may still not be appropriate technologies to use for COS. Third, system features, such as system restrictiveness, may potentially affect shoppers’ emotional feelings during shopping, which indicates that the design of RAs may indeed affect hedonic aspects of COS experience. And therefore, it is necessary to consider and examine the hedonic consequences of RA use. 5 2.2.  Group Decision Support Systems and Collaborative Online Shopping When the RA is considered in the COS context, it becomes a form of Group Decision Support Systems (GDSSs) for online shopping.  Although the influence of RA on COS has not received much attention, the effects of GDSSs on group process and outcomes have been extensively studied. Two major schools can be identified in past GDSS research: the decision theorist school and the institutionalist school (Dennis et al. 2001; DeSanctis and Poole 1994). The decision theorist school highlights the productivity, efficiency and satisfaction brought by information technologies to individuals and organizations (e.g., Dennis and Valacich 1999; Zigurs and Buckland 1998).  The institutionalist school regards information technology an opportunity for change of social interactions, rather than a causal agent of change (Barley and Tolbert 1988).  As the first attempt to study collaborative RA use, it seemed appropriate to follow the more traditional decision theorist school, and focus on the causal effect of the RA.  Nonetheless, rather than ignore the institutionalist perspective, I take on some of that perspective in a post-hoc analysis where I study how the effect of the RA on users’ perceptions is mediated by the complex and partially unintended effects of RA design on users’ social interactions.  Decision theorists argue that GDSSs can help overcome human’s bounded rationality and process losses associated with group interactions (DeSantics and Poole 1993), by facilitating and directing group communications (Dickson et al. 1993), providing information and computational infrastructure (Dennis et al.1988), and analyzing task-related information (Watson et al. 1988). These features and functions of GDSSs are expected to improve both the process (e.g., idea generation, conflict management, and equality of participation) and outcomes (e.g., decision quality, consensus, and satisfaction) of group decision-making (Dennis et al. 1988; Zigurs et al. 1991; DeSanctis and Poole 1994). Similarly, in a shopping context, RAs provide information, offer a computational infrastructure for collecting and storing users’ preferences, and they analyze users’ 6 information to make recommendations that improve purchase decision-making (Xiao and Benbasat 2007). However, while RAs for COS share some attributes of GDSSs, they also differ in at least three ways.  First, one of the important functions of conventional GDSSs is facilitating communications among co-located or distantly located group members through technologies such as synchronous video systems and teleconferences. Such a context differs substantially from the context that I examine, in which a pair of collocated shoppers collaborate together on a single computer.  Second, conventional GDSSs are used for organizational tasks, in which technologies such as brainstorming, idea generation, and voting and ranking techniques are used to facilitate tasks such as stakeholder analysis (e.g., Sambamurthy and DeSanctis 1989), allocation tasks (e.g., Dickson et al. 1989), or generate-choose tasks (e.g., Easton et al. 1990). The hedonic consequences of organizational decision-making are different from and apparently not as meaningful as those for shopping (Babin et al. 1994). Third, the criterion of purchase decision quality is more subjective than in many organizational decisions where the decisions taken can be clearly ranked in terms of quality (Mitra and Golder 2006).  For instance, different consumers will often evaluate the same product very differently because of different subjective preferences. A standard assumption in past research is that GDSSs have adequate information processing capabilities, leading to better utilitarian decision-making outcomes (DeSanctis and Poole 1994). But in the context of shopping, the effect on decision quality might be weakened due to the subjective nature of purchase decisions. Therefore, although past research on GDSSs and RAs can serve as theoretical base for predicting the influence of RA use on COS, it cannot be directly applied to the COS context. Additional research is needed.  7 2.3.  Restrictiveness and Collaborative Online Shopping Although there are large literatures on the use of GDSSs and RAs, there is not yet a consensus on the design and effects of such systems (Xiao and Benbasat 2001; Dennis et al. 2001). One reason is that any system can have many features and each one can have different effects (Wheeler and Valacich 1996; Anson et al. 1995, DeSanctis and Poole 1994).  Thus, we cannot conclude whether an RA is beneficial or not without understanding their features (Griffith and Northcraft 1994).   As indicated in my Introduction, RA restrictiveness is the feature I examine in this study.  System restrictiveness plays a central role in both GDSS and RA research (DeSanctis and Poole 1994; Wheeler and Valacich 1996), making it especially relevant and important to the COS context. Researchers have studied the effects of system restrictiveness in the context of individual shopping (Wang and Benbasat 2009) and organizational decision-making (Anson et al. 1995, Wheeler and Valacich 1996, Sliver 1988), and found that it leads to negative emotions (Wang and Benbasat 2009 and, Wheeler et al. 1993, Silver 1988) but positive decision-making outcomes (Wheeler et al. 1993, Anson et al. 1995). In particular, the restrictiveness of RAs used for individual online shopping has been examined in two levels: decision strategy restrictiveness and user-system interaction restrictiveness, because RAs are mainly designed to help elicit and construct attribute preferences through user-system interactions and to apply certain decision strategies to generate recommendations (Wang and Benbasat 2009).   The present study primarily focuses on user-system interaction restrictiveness. Unlike individual shoppers, collaborative shoppers have to establish their mutual preferences via interactions with the RA.  As a result, the outcomes of COS will likely depend heavily on how effectively the RA can facilitate collaborations and communications between them. Thus, the process of user-system interaction likely plays an even more important role in COS than in individual shopping. Moreover, the restrictiveness of user-system interaction could have competing (positive and negative) effects in a COS context, so it is not obvious, a 8 priori, whether it would help, or hinder.  For instance, collaborative shoppers may not require and may react negatively to restrictive guidance from an RA because when people do things together, they may prefer to discuss and collaborate freely with each other and may perceive a restrictive RA as interfering in their communications (Lenard et al. 1999).  On the other hand, because teams often need leaders, collaborative shoppers may benefit from third party’s guidance that can smooth disagreements and lead to a decision that reflects the preferences of both shoppers (Jago and Vroom 1980). As a result, learning what level of user-system interaction restrictiveness would be best in a COS context is both important and interesting to determine.  System restrictiveness first appeared in the paper of Silver (1988) as an important structural feature of single-user Decision Support Systems (DSSs), and has been widely applied to the studies of GDSS, RA and other systems (Wang and Benbasat 2009, Poole et al. 1991). Silver (1988) defined system restrictiveness as the “degree to which and the manner in which a structure restricts its users’ decision-making processes to a particular subset of all possible processes” (p. 52). Following this original definition, in this study, we conceptualized system restrictiveness as a manner of limiting user-system interactions to the subset of possible activities and sequences. Consistent with Wang and Benbasat (2012), we employed user-guided and system-controlled RAs to operationalize the RAs with two different levels of user-system interaction restrictiveness: restrictive RAs (system-controlled) and flexible RAs (user-guided). In particular, restrictive RAs promote a well-defined and step-by-step approach to group decision-making. Users of restrictive RAs must stick closely to the order of eliciting and adjusting attributes preferences and reviewing product recommendations predetermined by the RAs. On the other hand, flexible RAs offer more user control of the process. Users can decide the types and the order of the preferences they wish to indicate, and easily adjust their preference and review recommendations at any time after a certain point during the process. 9 3. Research Model and Hypothesis Development3.1. Overview Figure 1 shows my theoretical model.  As the figure shows, I expect that use of an RA will influence online consumers’ evaluations of both their shopping experience and the system they used for the online shopping. To capture experiential as well as instrumental outcomes of COS, I examined consumers’ evaluations of utilitarian value and hedonic value (Babin at al. 1994).   Evaluations of the RA are similarly represented by two variables: perceived usefulness and intention to reuse the system.  Figure 1: Research Model 10 3.2. Effects on Perceived Shopping Enjoyment (Hedonic Value) Hedonic value stems from intrinsic enjoyment and is the subjective evaluation of shopping experience (Babin et al.1994). In this study, hedonic value is therefore conceptualized as perceived enjoyment of the COS process.    As noted before, RAs could have positive or negative effects on users’ enjoyment of COS.  On one hand, increases in enjoyment could stem from the RA simplifying user-system interactions and user-user interactions.  For instance, RAs could help consumers easily locate and focus on alternatives matching their preferences (Xiao and Benbasat 2007), reduce the amount of superfluous information to process, and free the online consumers from complex decision making tasks (Häubl and Trifts 2000). Also, by decreasing the complexity of collaborations and communications, RAs can help establish a common ground of mutual knowledge and assumptions, providing the collaborators with a shared referential base for discussion (Cramton 2002, Hanna et al. 2003, Zhu et al. 2010).  This can occur because RAs help the collaborators focus on one particular aspect of purchase decision-making at a time, such as a particular attribute or a particular recommendation.  Common ground can be accumulated after each collaborative decision on each aspect of the purchase decisions. Helpful tips on points to consider when making purchases can also help build a mutual knowledge base between collaborative shoppers.  This mutual base of assumptions and knowledge can then smooth and simplify the process of collaborations and communication. The emotional consequences of complexity are negative especially in intrinsically motivated tasks such as shopping (Deci1975). Greater complexity requires greater demands on cognitive capacity and decreases the pleasure of the experience.  On the other hand, excessive guidance and extremely systematic procedures will limit shoppers’ freedom, which, according to reactance theory, will negatively affect their enjoyment of COS. Reactance theory suggests that individuals are aroused when their freedom is threatened or restricted (Lessne and Venkatesan 1989). Restrictive RAs exert a highly structured procedure of eliciting and constructing preferences. Although this 11 simplifies the interactions that take place, it also limits the interactions that can take place between users and the system and between the users themselves.  Such RAs “force” shoppers to go through a programmed procedure. Shoppers may therefore be displeased with the limited flexibility and annoyed by their inability to perform certain behaviors (Clee and Wicklund 1980).  In short, restrictive RAs reduce the complexity of interactions (which can increase enjoyment) by reducing users’ freedom (which can decrease enjoyment).  Thus, I predict that the use of restrictive RAs will not lead to either significantly increased or decreased shopping enjoyment, compared to no use of RAs. In contrast, flexible RAs can decrease the complexity of interactions while retaining users’ freedom. Therefore, I predict that compared to no use of RAs and restrictive RAs, use of flexible RAs will be associated with increased perceived shopping enjoyment.  Hypothesis 1A: Compared to without use of RAs, use of flexible RAs will lead to greater perceived enjoyment of COS. Hypothesis 1B: Compared to without use of RAs, use of restrictive RAs will lead to similar level of perceived enjoyment of COS. Hypothesis 1C: Compared to use of restrictive RAs, use of flexible RAs will lead to greater perceived enjoyment of COS. 12 3.3. Effects on Perceived Purchase Decision Quality (Utilitarian Value) Utilitarian value stems from achieving the instrumental goal  of the task (Childers et al. 2001, Babin et al. 1994). Marketing research shows that it is not objective product quality but perceived quality that drive preferences, satisfaction, loyalty, sales, and profitability (Aaker and Jacobson 1994, Anderson and Sullivan 2003, Anderson et al. 1994, Bolton and Drew 1991, Rust et al. 1995).  In fact, prior researchers have not even been able to pin down a consistent or clear definition of objective purchase decision quality.  As a result, past research invariably examines perceived (rather than objective) decision quality as the dependent variable of interest.  Likewise, I use perceived decision quality to reflect the utilitarian value of shopping. Perceived decision quality is the overall subjective judgment of quality regarding the purchase decision (Mitra and Golder 2006, Haubl and Trifts 2000).   RAs can play valuable roles in both stages of decision-making: screening and in-depth comparison (Haubl and Trifts 2000). RAs enable users to screen all the alternatives by their needs and locate and focus on alternatives matching their preferences (Xiao and Benbasat 2007). RAs provide recommendations of products and sequence them based on the matching scores, which enhances the quality of the information processed by shoppers when they are performing comparisons across products (Singh and Ginzberg 1996). Therefore, RAs can improve users’ information-processing capability by increasing the amount and the relevance of the information processed. The greater the amount of relevant information an individual has processed, the greater is his or her confidence in his or her judgment (Paul and Sniezek 2004). As such, we hypothesize: Hypothesis 2A: Compared to without use of RAs, use of restrictive RAs will lead to greater perceived purchase decision quality. Hypothesis 2B: Compared to without use of RAs, use of flexible RAs will lead to greater perceived purchase decision quality. 13 Restrictive RAs require users to elicit their preferences to a certain number of product attributes, while users of flexible RAs enjoy the flexibility to choose the attributes they wish to consider and how they weight them. Therefore, users of restrictive RAs are likely to consider more attributes and thus have to process more information than users of flexible RAs. As we have argued, the amount of information that consumers consider can increase decision quality as well as increasing one’s confidence in those decisions (Paul and Sniezek 2004). I therefore predict that: Hypothesis 2C: Compared to use of flexible RAs, use of restrictive RAs will lead to greater perceived purchase decision quality. 14 3.4. Effects on Perceived Usefulness and Intention to Reuse Online shoppers are not only shoppers but users of IT artifacts (Koufaris 2002). Therefore, in addition to studying shopping value, I examine users’ perceptions of the usefulness of the shopping system (its perceived usefulness) and their attitude towards reusing it (their intention to use).  Perceived usefulness refers to “the degree to which a person believes that using a particular system would enhance his or her performance” (Davis 1989, p. 320).  Meanwhile, intention to reuse is a measure of the strength of the user’s intention to use a particular system he or she has just experienced and it is a strong predictor of actual behavior (Ajzen 1991, Ajzen and Fishbein, 1977).   RAs assist in screening large numbers of alternatives and provide recommendations according to consumers’ stated preferences. Quantitative decision models supported by RAs are designed to calculate how these alternatives match consumers’ preferences, thereby performing what consumer would otherwise have performed to make the purchase decision. Prior evidence suggests that although consumers want to save effort, they react positively to the effort exerted by others (Mohr and Bitner 1995), such as RAs (Xiao and Benbasat 2007).  Hypothesis 3A: Compared to without the use of RAs, use of flexible RAs will lead to greater perceived usefulness of system. Hypothesis 3B: Compared to without use of RAs, use of restrictive RAs will lead to greater perceived usefulness of system. Because restrictive RAs provide more guidance than flexible RAs, shoppers might react more positively to restrictive RAs than flexible RAs. However, restrictive RAs also restrain users from personalizing their interactions with RAs and expressing and adjusting their preferences in the process of preference construction, and prior research has found that the second effect is more dominant than the first (e.g., Wang and Benbasat 2009). Flexible 15 RAs are thus more likely to meet the needs of users than restrictive RAs. Moreover, a flexible environment provides users more freedom to actively use the system. The active engagement in interacting with the system can cause users to overestimate its value (Xiao and Benbasat 2007).  As a result, I hypothesize that: Hypothesis 3C: Compared to use of restrictive RAs, use of flexible RAs will lead to higher perceived usefulness of the system. Research in environmental psychology has argued that consumers’ attitude and behavioral responses are affected by their experiences while shopping (Ridgway et al. 1990).  For instance, positive shopping experiences can increase consumer loyalty (Kerin et al. 1992) and the utilitarian and hedonic aspects of online shopping experiences can affect their motivations to reuse a shopping system (Childers et al. 2002).  Pleasurable feelings can also increase shoppers’ intentions to return to a store (Dawson et al. 1990). Because of the importance of hedonic outcomes in collaborative shopping (Babin et al. 1994), shopping enjoyment is likely to be a significant predictor of collaborative shoppers’ intention to reuse a shopping system.  As a result, I postulate that: Hypothesis 4A: Perceived enjoyment of COS positively influences consumers’ intention to reuse the system Individuals adopt technology not only because they enjoy the experience associated with it but also because they derive value from it (Childers et al. 2001).  Users are more likely to adopt an RA if they perceive the RA is useful in performing the shopping goal and helps increase their decision quality (Häubl and Trifts 2000; Hostler et al. 2005, Wang and Benbasat 2009). If decision quality is perceived to be low, users will probably discontinue utilizing it.  Therefore, I posit that: 16 Hypothesis 4B: Perceived purchase decision quality will positively influence consumers’ intention to reuse the system.  Hypothesis 4C: Perceived usefulness of system will positively influence consumers’ intention to reuse the system. 17 4. Research Methodology4.1. Experimental Design A laboratory experiment with 3 (without RA, with flexible RA, with Restrictive RA) × 1 between-subject design was used to test my hypotheses. To simulate a real shopping situation, every participant was asked to invite one of his or her real friends to take part in the experiment. In the shopping task, the collaborative shoppers were asked to purchase a digital camera as a gift for one of their mutual friends, with the cost shared between the two shoppers. Just as in the vignette at the start of the thesis, the collaborative shoppers were collated; they were located in the same room and used one common computer for shopping online. The experiment compared three shopping environments (flexible RA, restrictive RA, and no RA). Participants could choose from 97 camera alternatives, all of them up-to-date models. The user-system dialogue that I used for the RAs in the experiment included 10 needs-based questions associated with 10 important attributes of digital cameras. For each question, users were provided with four options to indicate their preferences of attributes and were asked to elicit how important the attribute is. After the RA finished eliciting users’ preferences for attributes (questions), users would get access to the recommendation pages of cameras, in which all the cameras are presented in the order of recommendation scores, which describe how a camera matches users’ preferred camera attributes. Helpful tips were offered to explain the meaning of important attributes of digital cameras. To ensure the realism of the study, I obtained expert opinions on the design of the online camera store from the head photographer of a large university, a senior sales-person in a large digital camera store, and a photography instructor in a local college, and made adjustments on the design accordingly.  With the flexible RA, the user could start with any of the attribute questions.  The navigation bar located at the top of the webpage showed the 10 attributes, whether the user 18 had indicated his or her preferences for them, and whether the user was able to get access to the recommendations. They were able to get access to the recommendations after they provided their preferences regarding any three attributes. The user could go back and forth between different questions and recommendation pages to express and change their preferences, review recommended alternatives, and make decisions.   With the restrictive RA, the users began by answering a question about an attribute value and the importance of the attribute as specified by the RA. The navigation bar was absent. Users had to follow a specific procedure predetermined by the RA designer. After users answered one question, the next question was presented. Users could not go back and forth between questions.  After finishing all 10 questions, recommendation pages were presented automatically. After seeing the recommendations, if users wanted to adjust their preferences of attributes, they had to review all 10 attributes again to locate and change their preferences in the same sequence. As before, helpful tips were provided to clarify the meaning of important camera concepts. In the control condition, the screen listed all the alternatives in the database randomly. Users did not need to indicate their preferred cameras’ attributes and the system did not provide any recommendations. Users could sort the cameras by price and filter them by brand. No helpful tips were provided. Appendix 1 provides more detailed information on the design of experimental system. 19 4.2.  Sample and Incentive 84 (42 pairs) students from a large North American university participated in a pilot study aimed at testing the validity of the treatments and measures.  For the main study, 180 (90 pairs) students from the same university participated. Each participant received $10 for participation.  In addition, 30% of the pairs received an additional $10-$100 based on performance.  Participants were asked to provide their justifications for their shopping choices at the end of the task to increase the validity of the findings. Their performance was judged based on how convincing their justifications were and how seriously they performed the shopping task.  Control Condition Flexible RA Restrictive RA Figure 2: Experimental Treatments Images of RA screens removed for copyright reasons.20 4.3. Procedures Data were collected in the laboratory. Each pair was randomly assigned to one of the three experimental conditions. At the beginning of the experiment, a tutorial video was shown to describe the procedures and incentives, and the features of the shopping system they would use in the task.  Then, the two shopping partners were asked to complete a questionnaire about their mutual friend, for whom they were going to make a purchase.  Upon finishing the questionnaire, each pair was directed to a private room equipped with a personal computer, a monitor, a keyboard and a mouse. They were then asked to perform a shopping task and to make a purchase decision on a digital camera collaboratively. After finishing the shopping task, the participants were asked to respond to two questionnaires, with one finished individually in different rooms and the other finished together on the same computer. To explore the collaborative shopping process, we recorded the conversations between two shoppers (discussed later in Section 6).  21 5. Data Analysis5.1. Subject Demographics The average age of the participants was 22.7 years and the ages ranged from 19 to 55. 102 out of 180 participants were females. No significant differences were found between subjects randomly assigned to each of the three experimental conditions with respect to age, gender, product knowledge, task relevance, and friendship with their partners.  5.2. Unit of Analysis One of the key measurement issues in this study is to choose an appropriate unit of analysis (Gallivan and Benbunan-Fich, 2005). Most studies of groups obtain individual level data and average the scores in a group to build a group level data point (e.g. Chidambaram et al. 1990, G.K. Easton et al. 1990).  However, averaging individual scores to represent group scores without additional verification can cause problems (Gallupe 1990). In this paper, I measured the same variables at both the individual and group level. I first measured the variables individually and then asked the pair to get together to decide on a score on the same variables following the group discussion method (Gallupe and Mckeen 1990, Guzzo et al., 1993). Encouragingly, I found that the aggregated individual scores were highly correlated with the group scores (the intraclass correlation coefficients ranged from 0.86 to 0.95), in terms of all the seven variables involved (see 5.3).  I therefore felt assured in following common practice and using each group‘s average scores as the data for the following tests.  Thus, the following analyses are based on the aggregated individual scores. 22 5.3. Measurement The materials included measures for four dependent variables (perceived enjoyment, perceived purchase decision quality, perceived usefulness, and intention to reuse), two control variables (perceived involvement of shopping task and perceived friendship with shopping partners), and the manipulation check for perceived interaction restrictiveness.  Because this study examines user perceptions, I used a questionnaire to measure them and I adapted the items from prior studies (i.e., Widing and Talarzyk 1993, Koufaris et al. 2001, Zhu et al. 2010, and Wang and Benbasat 2009) (see Appendix 3).  Convergent and discriminant validity was assessed through an exploratory, principal components factor analysis (PCA) with direct oblimin rotation (see Table 1). The results showed that all items had loadings above the recommended value of 0.70 on the intended constructs and did not have cross-loadings above 0.40 on the other constructs.  Therefore, the convergent and discriminant validity of the seven constructs appeared satisfactory.  Table 2 shows the descriptive statistics.  The reliability of the measures also seemed acceptable because the Cronbach’s alphas were all greater than 0.70 (see Table 3). 23 Table 1: Factor Analysis Construct Item 1 2 3 4 5 6 7 Perceived Enjoyment (EN) EN1 .094 -.064 .077 .013 -.749 -.026 -.105 EN2 .024 .101 .004 .042 -.881 -.058 -.018 EN3 -.034 -.025 .021 .000 -.881 .093 .029 EN4 .031 -.002 .023 -.030 -.935 .087 .069 Perceived Purchase Decision Quality (DQ) DQ1 -.001 -.022 .101 .928 .118 .044 .087 DQ2 .027 .021 .027 .843 .033 -.011 -.098 DQ3 .167 -.024 .006 .855 -.038 -.019 .100 DQ4 -.214 .006 -.123 .720 -.214 .036 -.129 Perceived Usefulness of System (PU) PU1 .840 .035 -.043 .002 -.061 .025 -.090 PU2 .923 .026 .041 .004 -.036 -.006 -.019 PU3 .906 -.024 -.019 .023 -.013 .058 -.084  Intention to Use of System (IU)  IU1 .095 .064 .009 .017 .024 .044 -.858  IU2 .087 .013 .017 -.008 -.036 .037 -.886 IU3 .044 .048 .009 -.015 .004 .086 -.919 Perceived task involvement (TI) TI1 .127 .002 .073 .075 -.035 .765 -.050 TI2 -.105 .002 -.032 -.058 -.074 .878 .016 TI3 .089 -.011 .011 .051 .021 .851 -.108 Perceived Friendship with Partner (FR) FR1 .017 .057 .935 .035 .032 .064 .015 FR2 -.031 .034 .954 .030 .012 .053 -.008 FR3 -.006 -.051 .909 -.031 -.132 -.099 -.027 Perceived Restrictiveness of System Interaction (RSS) RSS1 -.043 .922 .027 -.017 .025 .005 -.031 RSS2 -.011 .941 .067 -.009 .070 .033 .017 RSS3 .075 .898 -.043 -.002 -.045 -.019 -.094 RSS4 -.002 .950 -.010 .011 -.063 -.026 .022 Note: Extraction Method: Principal Component Analysis. Rotation Method: Oblimin with Kaiser Normalization. 24  Table 2: Descriptive Statistics for Outcome and Control Variables  TREATMENT Perceived Enjoyment Perceived Purchase Decision Quality Perceived Usefulness of System Intention to Use of System Perceived Friendship with Partner Perceived Task Involvement Without RA Mean 5.67 6.43 4.14 3.82 6.22 4.91 N 30 30 30 30 30 30 Std. Deviation 0.77 0.53 1.26 1.61 0.74 1.14 With Flexible RA Mean 6.20 6.45 5.37 5.02 6.07 4.97 N 30 30 30 30 30 30 Std. Deviation 0.53 0.56 1.01 1.28 0.81 1.05 With Restrictive RA Mean 5.70 6.42 5.23 4.94 6.25 5.24 N 30 30 30 30 30 30 Std. Deviation 0.69 0.36 1.08 1.08 0.90 0.89 Total Mean 5.87 6.43 4.91 4.59 6.18 5.04 N 90 90 90 90 90 90 Std. Deviation 0.70 0.48 1.24 1.43 0.82 1.03  Table 3: Reliabilities and Correlation Between Constructs Construct Cronbach’s Alpha 1 2 3 4 5 6 7 1 Perceived Enjoyment  0.909 1       2 Perceived Purchase Decision Quality 0.858 .295** 1      3 Perceived Usefulness of System 0.950 .379** .193 1     4 Intention to Use of System 0.963 .441** .256* .806** 1    5 Perceived Task Involvement 0.830 .402** .137 .331** .384** 1   6 Perceived Friendship with Partner 0.936 .246* .135 .150 .143 .318** 1  7 Perceived Restrictiveness of System Interaction 0.951 -.201 -.091 .116 .177 .093 -.107 .1 Note: **Correlation is significant at the 0.01 level (2-tailed).       *Correlation is significant at the 0.05 level (2-tailed).   25 5.4. Manipulation Checks I expected that participants’ perception of system restrictiveness would differ between the group with restrictive RAs and the group with flexible RAs. As expected, the restrictive RAs was perceived to be significantly more restrictive than the flexible RAs in terms of interaction restrictiveness: restrictive RA (Mean = 4.96, SD = 0.69), flexible RA (Mean = 4.05, SD = 0.80), p<0.01).  However, while statistically significant, the absolute difference was not as large as I expected given the design of the two systems.  I return to this issue in the discussion.   5.5. MANCOVA Results I used MANCOVA to test the overall impact of the treatments on the dependent variables (perceived decision quality, perceived enjoyment, perceived usefulness and intention to use) (Hair et al. 1998).  The variable for ‘Experimental Treatments’ in this test had three levels: no RA, flexible RA, and restrictive RA.  Since the results in Table 4 showed that perceived task involvement was significantly correlated with some of the dependent variables, I included it as a control variable in the MANCOVA and ANCOVA thereafter. Table 4 also shows that there was a statistically significant difference among three conditions (no RA, flexible RA, and restrictive RA) on the combined dependent variables (F = 4.43, p<0.01; Wilks’ Lambda = 0.68; partial eta squared = 0.18). The significant result allows me to examine the impact of the experimental treatments on the individual dependent variables via ANCOVA. Table 4: MANCOVA Results Wilk’s Lambda DF F-Value Sig. Partial Eta Squared Experimental Treatments 0.68 8, 164 4.43 0.00 0.18 Perceived Task Involvement 0.80 4,82 5.20 0.00 0.20 Perceived Friendship with Partner 0.96 4,82 0.90 0.47 0.04 26 5.6. ANCOVA Results After adjusting for perceived task involvement and perceived friendship with shopping partners, there was a statistically significant difference on perceived enjoyment (F = 8.33, p<0.01; partial eta squared = 0.16) and perceived usefulness F = 11.00, p<0.01; partial eta squared = 0.21) among the three experimental treatments. However, the experimental treatments appeared to have no significant effect on perceived purchase decision quality (F = 0.079, p = 0.92; partial eta squared = 0.002).  Therefore, H2A, H2B and H2C were not supported. Table 5: ANCOVA Results Source Dependent Variable DF Mean Square F Sig. Partial Eta Squared Experimental TreatmentsPerceived Enjoyment of Shopping 2  2.97 8.33 .000 .164 Perceived Purchase Decision Quality 2   0.02 0.08 .924 .002 Perceived Usefulness of System 2 12.49 11.00 .000 .206  Intention to Reuse the System 2 11.67 7.48 .001 .150 Fisher’s LCD test was performed for multiple comparisons between three experimental treatments (see Table 6). Participants with flexible RA perceived significantly higher level of enjoyment than the participants without RAs (p < 0.01) and with restrictive RAs (p < 0.01); participants with restrictive RAs and without RAs perceived similar level of enjoyment of online shopping (p > 0.1).  Therefore, H1A, H1B and H1C were supported.  Participants also perceived a significantly higher level of system usefulness when they are using RAs: both flexible RAs (p < 0.01) and restrictive RAs (p < 0.01) than when they are not using RAs, though the difference between restrictive RAs and flexible RAs was not significant (p > 0.1). Therefore H3A and H3B were supported, while H3C was not. Similarly, participants perceived a significantly higher level of intention to reuse when they used the RAs: both flexible RAs (p < 0.01) and restrictive RAs (p < 0.01), but, the differences between restrictive RAs and flexible RAs was not significant (p > 0.1).  27 5.7. Regression Results I used multiple regression to assess the extent to which perceived enjoyment of shopping, perceived purchase decision quality and perceived usefulness predicted intention to use. The model explained 66% of the variance in intention to reuse the system (see Table 7). Perceived usefulness and perceived enjoyment were both significant predictors of intention to reuse the system. Perceived usefulness had the strongest unique contribution: beta = 0.74, p < 0.01; perceived enjoyment shopping made a smaller contribution: beta = 0.14, p< 0.05. However, perceived decision quality was not a significant predictor of the dependent variable. Therefore, H4A and H4C were supported, but H4B was not. Table 7: Coefficients Coefficients t Sig. Adjusted R square Unstandardized Standardized B Std. Error Beta Perceived Purchase Decision Quality .215 .191 .072 1.123 .265 0.66 Perceived Enjoyment of Shopping .285 .140 .140 2.041 .044 Perceived Usefulness of System .855 .077 .739 11.086 .000 Table 6: Pairwise Comparisons Dependent Variable Treatment (I) vs. Treatment (J) Mean Difference (I-J) Std. Error Sig. Perceived Enjoyment of Shopping Without RA vs. Flexible RA -0.54 .155 .001 Without RA vs. With Restrictive RA 0.02 .156 .908 With Flexible RA vs. With Restrictive RA 0.56 .155 .001 Perceived Purchase Decision Quality Without RA vs. Flexible RA -0.02 .126 .886 Without RA vs. With Restrictive RA 0.03 .127 .804 With Flexible RA vs. With Restrictive RA 0.05 .127 .695 Perceived Usefulness of System Without RA vs. Flexible RA -1.23 .276 .000 Without RA vs. With Restrictive RA -0.98 .278 .001 With Flexible RA vs. With Restrictive RA 0.25 .277 .377 Intention to Reuse the System Without RA vs. Flexible RA -1.18 .324 .000 Without RA vs. With Restrictive RA -0.96 .326 .004 With Flexible RA vs. With Restrictive RA 0.22 .325 .504 28 5.8. Discussion of Results This study examined the effects of decision aids, such as RAs, on collaborative shopping behaviors and the importance of design features, such as system restrictiveness. The results highlight the differences between shopping without RAs, with flexible RAs, and with restrictive RAs.  As Table 9 shows, the preceding analyses found support for some, though not all, of the hypotheses.  Figure 3 shows the results of the structural model.    Table 8: Summary of Hypotheses Testing Results H1A Compared to without the use of RAs, use of flexible RAs will lead to greater perceived enjoyment of COS. Supported H1B Compared to without use of RAs, use of restrictive RAs will lead to similar level of perceived enjoyment of COS. Supported H1C Compared to use of restrictive RAs, use of flexible RAs will lead to greater perceived enjoyment of COS. Supported H2A Compared to without use of RAs, use of flexible RAs will lead to greater perceived purchase decision quality. Not Supported H2B Compared to without use of RAs, use of restrictive RAs will lead to greater perceived purchase decision quality. Not Supported H2C Compared to use of flexible RAs, use of restrictive RAs will lead to greater perceived purchase decision quality. Not Supported H3A Compared to without use of RAs, use of flexible RAs will lead to greater perceived usefulness of system. Supported H3B Compared to without use of RAs, use of restrictive RAs will lead to greater perceived usefulness of system. Supported H3C Compared to use of flexible RAs, use of restrictive RAs will lead to greater perceived usefulness of system. Not Supported H4A Perceived enjoyment of COS positively influences consumers’ intention to reuse the system. Supported H4B Perceived purchase decision quality positively influences consumers’ intention to reuse the system. Not Supported H4C Perceived usefulness of system positively influences consumers’ intention to reuse the system. Supported 29 Figure 3: Research Model In summary, the use of flexible RAs led to the highest level of perceived enjoyment, perceived usefulness, and intention to reuse. Intention to reuse the system was jointly explained by perceived enjoyment of shopping and perceived usefulness of system. Therefore, in general, flexible RAs have the most potential to improve the collaborative shopping experience and users’ evaluations of the COS system. These findings confirmed our hypotheses and supported the claims in prior research regarding the positive effects of RAs on system evaluations (Xiao and Benbasat 2007) and the negative effects of system restrictiveness on users emotions (e.g. Silver 1988; Wheeler et al. 1993). One remarkable result is that participants perceived higher enjoyment when they were using RAs than when they were not. Huang et al. (2011) predicted that RAs use would negatively influence shopping enjoyment because RAs use would force users to follow programmed procedures, making shopping more like work and thus harming the shopping enjoyment. However, we found that use of flexible RAs was associated with significantly greater shopping enjoyment than when using no RA. This result is encouraging and it is the first time to my knowledge that this result has been shown.    30 However, some of the results of this study contradicted my predictions and differ from the results of prior studies on individual use of RAs and on GDSSs. For instance, prior findings indicate that RAs improve individuals’ purchase decision quality and that GDSSs improve group-decision making outcomes (Xiao and Benbasat (2007); Dennis et al. 2001), but I found no significant differences on perceived purchase decision quality; participants in all three treatments reported very high perceived decision quality.  I suspect that the insignificant results might be for two reasons: the subjective nature of purchase decision quality and the presence of shopping partners. Consumers’ perceived quality actually does not reflect objective quality of products especially in short term (Bolton and Drew 1991).The perceptions of quality are influenced by many other sources, such as prior experience, brand reputation, store reputation and shopping involvement (Zeithaml 1988, Dodds et al. 1991, Boulding et al. 1993, Johnson et al. 1995).  RAs are designed to objectively improve consumers’ information processing ability and the accuracy of purchase decisions but may not always affect perceived quality. Also, in an individual online shopping context, a shopper has no shopping partner, and so the RA, as the sole advisor influencing purchase decisions, could have a stronger influence.  In COS, shopping partners play a very important role in establishing confidence toward purchase decisions because they can bring more information and knowledge for decision making (Shaw 1932), which may increase shoppers’ acceptance of their decisions. Stimulation and engagement from partners may also encourage individual shoppers to get involved in the shopping task. Active involvement can cause users to think highly of their purchase decisions (Xiao and Benbasat 2007). Moreover, cognitive dissonance theory suggests that people have an inner drive to hold their attitudes and beliefs in harmony and avoid inconsistencies (Brehm and Cohen 1962).  This need to maintain consistency may be especially strong when working with someone else, e.g., an individual may reconsider his/her own individual decision but might be less likely toreconsider a decision after just agreeing to it with another partner.      My investigation also shows that RA restrictiveness did not exert significant influence on users’ subjective evaluations of shopping systems, contrary to prior findings that system 31 restrictiveness negatively affected perceived system quality and intention to reuse (e.g. Silver 1988, Wang and Benbasat 2012). This difference might be caused by the social nature of COS. Dictionary.com defines flexibility as able to be easily modified to respond to altered circumstances or conditions. However, collaborative shoppers cannot utilize the flexibility as freely as they can when shopping on their own, because they have to get agreement on what actions to take and how to take them, before proceeding to the next steps. Therefore, collaborative shoppers might not feel as big a difference as individual shoppers do, reducing the difference between flexible RAs and restrictive RAs. The results of the manipulation check partially supports this intuition, showing that the absolute difference between flexible and restrictive RAs on perceived restrictiveness was not as large as initially expected. Moreover, when individuals are working in groups, conflicts and conformity can emerge in group interactions (Yukl 1989). Guidance from a third party, such as shopping system, might work as a “facilitator” who can smooth the process of group interactions. Therefore, collaborative shoppers might appreciate restrictiveness more than individual shoppers do.  Both of these arguments would suggest that the negative effect of RA restrictiveness might be weaker in the case of COS. 32 6. Supplementary Analysis: Interaction Process Analysisof Shopping Transcripts 6.1. Coding of Shopping Transcripts Although the results based on the measures of participants’ perceptions have shown interesting findings and related theoretical explanations were provided, more evidence is needed to further explain and support the results, especially several unexpected ones, such as that:  (1) use of flexible RAs led to the highest perceived enjoyment;  (2) no difference was found on perceived decision quality;  (3) system restrictiveness did not influence perceived usefulness of systems.  The process of collaborations and communications between the shopping partners makes collaborative shopping significantly different than individual shopping. Using Bales Interaction Process Analysis (IPA), I conducted an analysis of the conversations that took place between the two shopping partners to open the “black box” of the process of COS and to find objective evidences to explain the current results with the supplementary data from shopping transcripts. In particular, I examined the influence of RAs use on amounts and patterns of communications and explored the correlations between the process data, based on Bales IPA, and the outcome variables. The Bales IPA (Bales 1951) coding identifies and records the nature of each separate act in ongoing communications between collaborative shoppers. I compared the amount and pattern of communication in the three situations: without RA, with flexible RA, and with restrictive RA. The coding focused solely on spoken conversation (excluding nonverbal communications, such as facial expressions and gestures, which we found did not add any meaningful results), was done by two coding teams. Each team consisted of two well-trained coders and coded 50 transcripts, in which 40 transcripts are unique and the other 10 are the 33 overlap between the two teams. The coders in each coding team were asked to accomplish the following steps during the coding process: (1) identify each communication unit; (2) group the communication units into three categories: task-oriented communications (TOCs), positive social emotional communications (PSECs), and negative social emotional communications (NSECs), according to Bales Communication Categories (as shown in Figure 4); (3) meet to review the entire first 20 transcripts and resolve any inconsistencies until they agreed on one version of the results, and then code another 30 transcripts independently.  Intercoder and interteam reliabilities were calculated. In particular, the intercoder reliability was conducted on the 60 transcripts (30 for each coding team) that were coded independently by the individual coders and the interteam reliability was conducted on the 10 transcripts that overlap between the 2 coding teams. Following the common practice, I used the ratio of coding agreement to the total number of coding decisions to measure the intercoder reliability (Kassarjian 1977). Intercoder reliabilities range from 0.876 to 0.952, with an average of 0.901. The interteam reliability test was conducted on each the categories (i.e., TOC, PSEC, and NSEC) of the 10 transcripts, among the 4 coding teams, and was calculated using Intraclass correlation coefficient (ICC) (Shrout and Fleiss 1979).  The interteam reliabilities range from 0.937 to 0.969, with an average of 0.966. 6.2. Amounts and Patterns of Communications I conceptualized the amount of communications as the number of Bales units that each group conducts during the shopping. The Bales unit is the single communication “act”, which expresses a thought, sufficiently complete to permit the other person to interpret and so to react it. The units are also the smallest classifiable communication acts (Bales 1951). Not only the complete sentences but also the fragments of sentences, words, or phrases can be scored as Bales units.  Figure 4 shows the different categories and subcategories of Bale’s framework.   34 Figure 4 Bales Communication Categories As a social activity, collaborative shopping is a process of both making purchase decisions and maintaining social relations among shopping partners. In other words, making high quality purchase decisions and enjoying social interaction process are both important for collaborative shoppers. Bales (1951) believed that decision-making groups try to achieve two main goals—maintain the group and perform the task—and that communicative statements are the actions that groups use to achieve these goals. In particular, TOCs are related to the group’s goal of performing the task, while SECs include positive social and negative social communications and are associated with group maintenance (Bales 1951). Using Bales IPA, we can learn how and how well collaborative shopping teams achieve these shopping goals based on the analysis of TOCs and SECs. In my data, the most frequent exchanges were TOCs (77.51%), while SECs (22.49%) comprised a minor part of the exchanges. In particular, PSECs accounted for 17.34%, while NSECs accounted for 5.15%.  Table 10 shows the TOCs, PSECs, and NSECs for each condition in the experiment.   35  Table 10: Descriptive Statistics for IPA Treatment    Task Oriented Communications (TOCs) Positive Social Emotional Communications (PSECs) Negative Social Emotional Communications (NSECs) TOTAL Communications   Without RA Mean 182.43 38.68 21.85 242.97   N 30 30 30 30   Std. Dev 99.091 28.073 18.271 114.673 With Flexible RA Mean 265.77 69.43 7.91 343.10   N 30 30 30 30   Std. Dev 152.554 35.686 8.257 185.969 With Restrictive RA Mean 278.68 53.55 6.12 338.35   N 30 30 30 30   Std. Dev 124.603 33.122 7.672 148.861 Total Mean 242.29 53.89 11.96 308.14 N 90 90 90 90 Std. Dev 132.972 34.475 14.149 157.879 6.3. Results of Interaction Process Analysis and Discussion In Tables 11 and 12, I show the results for the Interaction Process Analysis.  Figure 5 shows the results of the structural model.  In the following subsections, I discuss the results for each question that I noted earlier.   Table 11: ANOVA Results   Sum of Squares df Mean Square F Sig. Task Oriented Communications  (TOCs) Between Groups 163753.47 2 81876.74 5.05 .008 Within Groups 1409911.98 87 16205.89     Total 1573665.45 89       Positive Social Emotional Communications (PSECs) Between Groups 14180.84 2 7090.42 6.73 .002 Within Groups 91600.31 87 1052.88     Total 105781.15 89       Negative Social Emotional Communications (NSECs) Between Groups 4451.18 2 2225.59 14.49 .000 Within Groups 13364.98 87 153.62     Total 17816.16 89       Total Amount of Communications Between Groups 191472.27 2 95736.14 4.11 .020 Within Groups 2026919.74 87 23297.93     Total 2218392.01 89        36  Table 12: Pairwise Comparisons Dependent Variable Mean Difference (I-J) Std. Error Task Oriented Communications  (TOCs) Without RA vs. Flexible RA -83.33* 32.87 Without RA vs. With Restrictive RA -96.25* 32.87 With Flexible RA vs. With Restrictive RA -12.92 32.87 Positive Social Emotional Communications (PSECs) Without RA vs. Flexible RA -30.74* 8.38 Without RA vs. With Restrictive RA -14.87 8.38 With Flexible RA vs. With Restrictive RA 15.875 8.38 Negative Social Emotional Communications (NSECs) Without RA vs. Flexible RA 13.94* 3.20 Without RA vs. With Restrictive RA 15.73* 3.20 With Flexible RA vs. With Restrictive RA 1.79 3.20 Total Amount of Communications Without RA vs. Flexible RA -100.13* 39.41 Without RA vs. With Restrictive RA -95.38* 39.41 With Flexible RA vs. With Restrictive RA 4.75 39.41 *. The mean difference is significant at the 0.05 level.                                           Figure 5: Expanded Research Model   37 Why flexible RAs leads to highest perceived enjoyment of shopping The results of the Bales IPA show that use of flexible RAs (Mean = 7.91, SD = 8.26) and restrictive RAs (Mean = 6.12, SD = 7.67) are associated with significantly lower NSECs than no use of RAs (Mean = 21.85, SD = 18.27, p < 0.01). Use of flexible RAs (Mean = 69.43, SD = 35.68) leads to significantly more PSECs than no use of RAs (Mean = 38.68, SD = 28.07, p< 0.01).  These results shed light on why use of flexible RAs lead to higher perceived shopping enjoyment than no use of RAs, from the perspective of RAs influence on the content and amount of between-shopper communications. The lower amount of NSECs and higher amount of PSECs associated with the use of flexible RAs indicate that, compared to no use of RAs, use of flexible RAs played a more important role in smoothing the conflicts and disagreements, and at the same time, contributed more to establishing positive communications between collaborative shoppers. Compared to use of flexible RAs, use of restrictive RAs did not lead to lower NSECs. I expected that restrictiveness could have reduced NSECs by simplifying the decision-making process, constraining the attention of collaborators on specific topics, and establishing a common ground. However, this result suggests that flexible RAs decreased NSECs just as much as restrictive RAs did. Although this finding contradicts the conventional view, it is encouraging because it suggests that designers might no longer need to make tradeoffs between decreasing the complexity of the shopping process and providing flexibility.  Flexible RAs seems to accomplish both goals. Despite these insights, the IPA results still do not completely explain why flexible RAs were associated with higher perceived enjoyment of shopping than restrictive RAs. This is because, as Figure 5 shows, PSEC increased perceived shopping enjoyment, but NSECs did not reduce perceived shopping enjoyment.  Thus, a full explanation still requires more research.     38  Why no effects of systems occur on perceived purchase decision quality Use of RAs does not exert significant influence on the perceptions of purchase decision quality (subjective), but it significantly increases the amount of TOCs (objective). The IPA shows that use of flexible (Mean = 265.77, SD = 152.55) and restrictive RAs (Mean = 278.68 SD = 124.603) led to significantly more TOCs than no use of RAs (Mean = 182.43, SD = 99.09, p<0.05).   This is not surprising. People often construct their shopping preferences dynamically and only elaborate on items that they are attending to (Tam and Ho 2005).  Because the RAs drew shoppers’ attention to more attributes, they forced them to elaborate (even if only briefly) on each one.  This is true even in the flexible RA condition, because they needed to agree on which features to consider. Interestingly, however, increased TOC is not associated with higher perceived decision quality. To explore why, I examined and compared the effects of perceived partner’s influence and perceived system’s influence on perceived purchase decision quality. I found that partner’s influence (Mean = 4.52, SD = 0.98) turned out to be as important as the influence of the system (Mean = 4.32, SD = 1.10, p > 0.1).  This result shows the existence of the competing effects of shopping partners with shopping systems on purchase decision making in the COS context and confirms my prior suspicion that the influence of the system on perceived purchase decision quality might be intervened or weakened by the presence of shopping partners. Moreover, the amount of TOCs in all three conditions is high. This finding together with the high perceived purchase decision quality in all three conditions imply that the presence of shopping partners may lead to both high task participation and high confidence in purchase decisions (i.e., perceived decision quality). In short, the results from the IPA analysis seem to confirm the arguments I had given earlier for the lack of results for perceived purchase decision quality.   39  Why restrictiveness of systems does not affect perceived usefulness of systems  The level of NSECs was found to negatively affect users’ perception of the usefulness of the system (as shown in Figure 5). One possible reason might be that shoppers tend to blame the unpleasant shopping experience on the shopping environment, and hence the shopping system. Therefore, users appear to have rated the no RA condition as less useful presumably because this condition was associated with higher NSECs. For the same reason, the lack of a difference in the level of NSECs between the flexible RAs and restrictive RAs can also partially explain why participants did not give significantly different ratings of perceived usefulness for these two conditions.      40 7. ContributionThis study offers both theoretical and practical contributions. First, it sheds light on an important but seldom studied research topic – COS – and to the best of my knowledge, this is the first study to examine the collaborative use of RAs. The growing phenomenon of social shopping, which involves shopping with one’s friends or others and the use of technology to mimic social interactions found in physical malls and stores, has been recognized by researchers (Schindler and Bickart 2005, Villanueva et al. 2008, Granitz and Ward 1996) and practitioners (e.g., group shopping sites, shopping communities, social shopping marketplace). Nevertheless, few systems are designed explicitly for COS and limited research has been devoted to facilitate collaborative shopping online. Although online consumers are able to get access to verbal information shared by others, so far, most of the shopping activities performed on the Web still can be characterized as solitary: users logged-in on a shopping site must browse pages, view and buy items essentially by themselves.  Of course, many offline consumption behaviors are decidedly not an individual affair; rather, individuals often undertake consumption activities to bond and share leisure time together with others (Evans et al. 1996). The rapid development of online shopping decision aids, such as recommendation agents, and collaborative technologies, such as group decision support systems, enjoy a big potential to fulfill the promise of collaborative shopping online. As a result, it is timely for researchers and practitioners to exert effort on understanding the issues related to the design of collaborative shopping system.  Second, different from most prior studies that only focus on generating design guidelines based on the utilitarian consequences of RAs, this study examined the influence of RAs on both utilitarian and hedonic value of COS. RAs’ appearance online is mostly driven by the utilitarian motivations, as I have clarified. It is still very important to know whether the utilitarian motivations behind RAs design would harm the hedonic dimensions of shopping experience or not. This research made the first attempt to answer this question and show that RA use can increase the hedonic shopping value perceived by shoppers. This 41 result is encouraging, for it theoretically confirms the effects of RAs on both the two dimensions of shopping value, and practically supports the use of RAs for collaborative shopping online.  Third, this study did not examine RAs effects in general, but described and studied RAs in terms of a specific system feature: system restrictiveness.  Numerous other features could have been studied (DeSanctis and Poole 1994).  RAs are essentially combinations of different features. A parsimonious and effective approach is to focus on a particular system feature of RA while controlling for other features, as this enables researchers to show and explain the mechanism of influence more clearly and reasonably. The results of this research show the negative influence of system restrictiveness on hedonic shopping value (i.e. perceived shopping enjoyment), which encourages system designers to provide enough flexibility in designing collaborative shopping systems. Fourth, the insignificant differences between the three treatments in terms of perceived purchase decision quality reveal the potential difference between collaborative shopping and individual shopping. As I have discussed, the social nature of COS and subjective nature of purchase decision quality can reduce the effects of RAs on purchase decision quality. However, it is arbitrary to conclude that RAs do not help users make better purchasing decisions. Instead, this insignificant result may imply that perceived decision quality, which has been a standard construct in consumer research, may not be the most appropriate measure in the COS context. Therefore, this finding should alert future researchers to the need to find or create more appropriate and sensitive constructs to measure the quality of collaborative purchase decisions.  Fifth, this research shows important results that the connections between outcome variables and process variables are not quite as simple as previously assumed. It would be usually assumed that utilitarian process variables (i.e., TOC) would correlate to utilitarian outcome variables (i.e., perceived decision quality and perceived usefulness of systems), and hedonic process variables (i.e., PSECs and NSECs) may relate to hedonic outcome variable (i.e., perceived enjoyment of shopping). However, TOCs did not correlate with perceived 42 decision quality or perceived usefulness of systems, while the amount of NSECs was found to affect perceived usefulness of system. Little prior research has explored the connections between process and outcome in online shopping. However, process data has great potential to further explain the mechanisms of influences exerted by the systems, especially in collaborative shopping context. This research makes one of the first steps in using the process data to explain the influence of RAs.  Sixth, the present study also went a step further than past studies in validating the use of measures collected at the individual level for analysis at the level of a pair. Prior researchers often worried about the validity of constructing the group level data by aggregating the individual level data, although most of previous studies simply adopted this method without further validating this measure. In this study, I adopted two methods to construct group-level measures of the same variables: by the discussion method (Guzzo et al., 1993) and by averaging individual level data, and examined consistency between these two measures. The high intraclass correlation between the two measures provides reassurance regarding the validity of constructing group-level measures by using the average of individual level data. Future research can therefore adopt this approach as well.   43 8. Limitations and Future ResearchAs a first step in this area, my study’s scope is limited in several ways and could be extended. First, I only studied the simplest COS situation, in which two co-located shoppers shop together for a shared shopping goal.  Future research can focus on other possible instances of COS, such as two remotely located shoppers shopping together, more than two shoppers shopping together, or collaborative shoppers shopping together for different shopping goals. Collaborative technologies such as electronic meeting systems, computer-supported cooperative work, video conferencing system and navigation support can be applied to facilitate the potential instances of COS. Second, only one type of products, digital cameras, was examined. However, the effects of RAs may vary among different types of products. The product-technology fit has been found to be a significant factor in yielding the best consumer experience (Jahng et al. 2000). For example, RAs may play a more important role in the shopping tasks for search products than experience products. Future researchers should pay more attention to the moderating role of product type on RA effects. Third, this research only adopted subjective measures to evaluate the effects of RA use on COS outcomes. The coexistence of the interactions between collaborative shoppers and between shoppers and system make the COS context both complex and interesting. Future researchers can further explore RA effects on COS through describing and studying these two types of interactions using the objective measures, such as the amount and type of communications between shoppers and searching behavior while shopping. Fifth, this study only emphasizes the restrictiveness of user-system interactions. However, system restrictiveness can also be manipulated in other ways, such as strategy restrictiveness (e.g. Wang and Benbasat 2009). Future researchers can also examine the RAs effects with focusing on other type of restrictiveness, such as strategy restrictiveness.  44 References Aaker, David A., and Robert Jacobson. 1994. The financial information content of perceived quality. Journal of Marketing Research 31 (2, Special Issue on Brand Management) (May): pp. 191-201. Adams, Dennis A., R. Ryan Nelson, and Peter A. Todd. 1992. Perceived usefulness, ease of use, and usage of information technology: A replication. MIS Quarterly 16 (2) (Jun.): pp. 227-247. Ajzen, Icek. 1991. The theory of planned behavior. Organizational Behavior and Human Decision Processes 50 (2) (12): 179-211. Ajzen, Icek, and Martin Fishbein. 1977. Attitude-behavior relations: A theoretical analysis and review of empirical research. Psychological Bulletin 84 (5) (09): 888-918. Anderson, Eugene W., Claes Fornell, and Donald R. Lehmann. 1994. Customer satisfaction, market share, and profitability: Findings from sweden. Journal of Marketing 58 (3) (Jul.): pp. 53-66. Anderson, Eugene W, and Linda Court Salisbury. 2003. The formation of Market‐Level expectations and its covariates. Journal of Consumer Research 30 (1) (June): pp. 115-124. Anson, Robert, Robert Bostrom, and Bayard Wynne. 1995. An experiment assessing group support system and facilitator effects on meeting outcomes. Management Science 41 (2) (Feb.): pp. 189-208. Arnold, M. J., and K. E. Reynolds. 2003. Hedonic shopping motivations. Journal of Retailing 79 (2): 77-95. Babin, B. J., W. R. Darden, and M. Griffin. 1994. Work and/or fun: Measuring hedonic and utilitarian shopping value. The Journal of Consumer Research 20 (4): 644-56. Bales, Robert F., and Fred L. Strodtbeck. 1951. Phases in group problem-solving. The Journal of Abnormal and Social Psychology 46 (4): 485-95. Barley, Stephen R., and Pamela S. Tolbert. 1997. Institutionalization and structuration: Studying the links between action and institution. Organization Studies (Walter De Gruyter GmbH & Co.KG.) 18 (1) (01): 93. 45 Bettman, James R. 1979. Memory factors in consumer choice: A review. Journal of Marketing 43 (2) (Spring): pp. 37-53. Bloch, P. H., and G. D. Bruce. 1984. Product involvement as leisure behavior. Advances in Consumer Research 11 (1): 197-202. Bloch, P. H., N. M. Ridgway, and D. L. Sherrell. 1989. Extending the concept of shopping: An investigation of browsing activity. Journal of the Academy of Marketing Science 17 (1): 13-21. Bolton, Ruth N., and James H. Drew. 1991. A multistage model of customers' assessments of service quality and value. Journal of Consumer Research 17 (4) (Mar.): pp. 375-384. Brown, S. A., A. R. Dennis, and V. Venkatesh. 2010. Predicting collaboration technology use: Integrating technology adoption and collaboration research. Journal of Management Information Systems 27 (2): 9-54. Chidambaram, Laku, Robert P. Bostrom, and Bayard E. Wynne. 1990. A longitudinal study of the impact of group decision support systems on group development. Journal of Management Information Systems7 (3, Management Support Systems) (Winter): pp. 7-25. Childers, T. L., C. L. Carr, J. Peck, and S. Carson. 2001. Hedonic and utilitarian motivations for online retail shopping behavior. Journal of Retailing 77 (4): 511-35. Clee, Mona A., and Robert A. Wicklund. 1980. Consumer behavior and psychological reactance. Journal of Consumer Research 6 (4) (Mar.): pp. 389-405. ———. 1980. Consumer behavior and psychological reactance. Journal of Consumer Research 6 (4) (Mar.): pp. 389-405. Cramton, Catherine Durnell. 2002. Finding common ground in dispersed collaboration. Organizational Dynamics 30 (4) (0): 356-67. Davis, Fred D. 1989. Perceived usefulness, perceived ease of use, and user acceptance of information technology. MIS Quarterly 13 (3) (Sep.): pp. 319-340. Dawson, Scott, Peter H. Bloch, and Nancy M. Ridgway. 1990. Shopping motives, emotional states, and retail outcomes. Journal of Retailing 66 (4) (Winter90): 408. Deci, Edward L. 1975. Intrinsic motivation. New York, NY, US: Plenum Press. 46  Dellaert, B. G. C., and G. Häubl. 2005. Consumer product search with personalized recommendations. Unpublished Working Paper, Department of Marketing, Business Economics and Law, University of Alberta. Dennis, Alan R., Joey F. George, Len M. Jessup, Jay F. Nunamaker Jr., and Douglas R. Vogel. 1988. Information technology to support electronic meetings. MIS Quarterly 12 (4) (Dec.): pp. 591-624. Dennis, Alan R., and Joseph S. Valacich. 1999. Research note. electronic brainstorming: Illusions and patterns of productivity. Information Systems Research 10 (4) (December): pp. 375-377. Dennis, Alan R., Barbara H. Wixom, and Robert J. Vandenberg. 2001. Understanding fit and appropriation effects in group support systems via meta-analysis. MIS Quarterly 25 (2) (Jun.): pp. 167-193. DeSanctis, Gerardine, and Marshall Scott Poole. 1994. Capturing the complexity in advanced technology use: Adaptive structuration theory. Organization Science 5 (2) (May): pp. 121-147. Dickson, G. W., J. E. Lee, L. Robinson, and R. Heath. 1989. Observations on GDSS interaction: Chauffeured, facilitated, and user-driven systems. Paper presented at System Sciences, 1989. Vol.III: Decision Support and Knowledge Based Systems Track, Proceedings of the Twenty-Second Annual Hawaii International Conference on, . Dickson, Gary W., Joo-Eng Lee Partridge, and Lora H. Robinson. 1993. Exploring modes of facilitative support for GDSS technology. MIS Quarterly 17 (2) (Jun.): pp. 173-194. Dodds, William B., Kent B. Monroe, and Dhruv Grewal. 1991. Effects of price, brand, and store information on buyers' product evaluations. Journal of Marketing Research 28 (3) (Aug.): pp. 307-319. Easton, George K., Joey F. George, Jay F. Nunamaker Jr., and Mark O. Pendergast. 1990. Using two different electronic meeting system tools for the same task: An experimental comparison. Journal of Management Information Systems 7 (1) (Summer): pp. 85-100. Evans, K. R., T. Christiansen, and J. D. Gill. 1996. The impact of social influence and role expectations on shopping center patronage intentions. Journal of the Academy of Marketing Science 24 (3): 208-18. 47  Gallivan, M. J., and R. Benunan-Fich. 2005. A framework for analyzing levels of analysis issues in studies of e-collaboration. Professional Communication, IEEE Transactions on 48 (1): 87-104. Gallupe, R. Brent, and James D. McKeen. 1990. Enhancing computer-mediated communication: An experimental investigation into the use of a group decision support system for face-to-face versus remote meetings. Information & Management 18 (1) (1): 1-13. Granitz, N. A., and J. C. Ward. 1996. Virtual community: A sociocognitive analysis. Advances in Consumer Research 23 : 161-6. Griffith, Terri L., and Gregory B. Northcraft. 1994. Distinguishing between the forest and the trees: Media, features, and methodology in electronic communication research. Organization Science 5 (2) (May): pp. 272-285. Guzzo, R. A., P. R. Yost, R. J. Campbell, and G. P. Shea. 1993. Potency in groups: Articulating a construct. British Journal of Social Psychology. Hanna, Joy E., Michael K. Tanenhaus, and John C. Trueswell. 2003. The effects of common ground and perspective on domains of referential interpretation. Journal of Memory and Language 49 (1) (7): 43-61. Häubl, G., and K. B. Murray. 2006. Double agents: Assessing the role of electronic productrecommendation systems. Sloan Management Review 47 (3): 8–12. Häubl, G., and V. Trifts. 2000. Consumer decision making in online shopping environments: The effects of interactive decision aids. Marketing Science: 4-21. Hirschman, E. C., and M. B. Holbrook. 1982. Hedonic consumption: Emerging concepts, methods and propositions. The Journal of Marketing: 92-101. Hirschman, Elizabeth C. 1984. Experience seeking: A subjectivist perspective of consumption. Journal of Business Research 12 (1) (3): 115-36. Holbrook, Morris B., and Elizabeth C. Hirschman. 1982. The experiential aspects of consumption: Consumer fantasies, feelings, and fun. Journal of Consumer Research 9 (2) (Sep.): pp. 132-140. Hostler, R. E., V. Y. Yoon, and T. Guimaraes. 2005. Assessing the impact of internet agent on end users' performance. Decision Support Systems 41 (1): 313-23. 48  Jago, Arthur G., and Victor H. Vroom. 1980. An evaluation of two alternatives to the Vroom/Yetton normative model. The Academy of Management Journal 23 (2) (Jun.): pp. 347-355. Jahng, J., H. Jain, and K. Ramamurthy. 2000. Effective design of electronic commerce environments: A proposed theory of congruence and an illustration. Systems, Man and Cybernetics, Part A: Systems and Humans, IEEE Transactions on 30 (4): 456-71. Jungjoo Jahng, H. K. Jain, and K. Ramamurthy. 2006. An empirical study of the impact of product characteristics and electronic commerce interface richness on consumer attitude and purchase intentions.Systems, Man and Cybernetics, Part A: Systems and Humans, IEEE Transactions on 36 (6): 1185-201. Kassarjian, Harold H. 1977. Content analysis in consumer research. Journal of Consumer Research 4 (1) (Jun.): pp. 8-18. Kerin, Roger A., Ambuj Jain, and Daniel J. Howard. 1992. Store shopping experience and consumer price-quality-value perceptions. Journal of Retailing 68 (4) (Winter92): 376. Kling, Rob, and Suzanne Iacono. 1988. The mobilization of support for computerization: The role of computerization movements. Social Problems 35 (3, Special Issue: The Sociology of Science and Technology) (Jun.): pp. 226-243. Koufaris, M., and W. Hampton-Sosa. 2002. Customer trust online: Examining the role of the experience with the web-site. Department of Statistics and Computer Information Systems Working Paper Series, Zicklin School of Business, Baruch College, New York. Koufaris, M., A. Kambil, and P. A. LaBarbera. 2001. Consumer behavior in web-based commerce: An empirical study. International Journal of Electronic Commerce 6 (2): 115-38. Lessne, Greg, and M. Venkatesan. 1989. Reactance theory in consumer research: The past, present and future. Advances in Consumer Research 16 (1) (01): 76-8. Mathwick, Charla, and Edward Rigdon. 2004. Play, flow, and the online search experience. Journal of Consumer Research 31 (2) (September): pp. 324-332. Mitra, Debanjan, and Peter N. Golder. 2006. How does objective quality affect perceived quality? short-term effects, long-term effects, and asymmetries. Marketing Science 25 (3) (May - Jun.): pp. 230-247. 49  Mohr, Lois A., and Mary Jo Bitner. 1995. The role of employee effort in satisfaction with service transactions. Journal of Business Research 32 (3) (3): 239-52. Moore, Robert, and Girish Punj. 2001. An investigation of agent assisted consumer information search: Are consumers better off?. Vol. 28Association for Consumer Research. Olson, E. L., II Widing, and E. Robert. 2002. Are interactive decision aids better than passive decision aids? A comparison with implications for information providers on the internet. Journal of Interactive Marketing 16 (2): 22-33. Orlikowski, Wanda J. 1992. The duality of technology: Rethinking the concept of technology in organizations. Organization Science 3 (3, Focused Issue: Management of Technology) (Aug.): pp. 398-427. Paese, Paul W., and Janet A. Sniezek. 1991. Influences on the appropriateness of confidence in judgment: Practice, effort, information, and decision-making. Organizational Behavior and Human Decision Processes 48 (1) (2): 100-30. Poole, Marshall Scott, Michael Holmes, and Gerardine Desanctis. 1991. Conflict management in a computer-supported meeting environment. Management Science 37 (8) (08): 926-53. Ridgway, Nancy M., Scott A. Dawson, and Peter H. Bloch. 1990. Pleasure and arousal in the marketplace: Interpersonal differences in approach-avoidance responses. Marketing Letters 1 (2) (Jun.): pp. 139-147. Roland T. Rust, Anthony J. Zahorik, and Timothy L. Keiningham. 1995. Return on quality (ROQ): Making service quality financially accountable. Journal of Marketing 59 (2) (Apr.): pp. 58-70. Sambamurthy, V., and G. Desanctis. 1990. An experimental evaluation of GDSS effects on group performance during stakeholder analysis. Paper presented at System Sciences, 1990., Proceedings of the Twenty-Third Annual Hawaii International Conference on, . Schindler, R. M., and B. Bickart. 2005. Published word of mouth: Referable, consumer-generated information on the internet. Online Consumer Psychology; Understanding and Influencing Consumer Behavior in the Virtual World: 35-62. Shaw, M. E. 1932. A comparison of individuals and small groups in the rational solution of complex problems. The American Journal of Psychology 44 (3): 491-504. 50  Shrout, Patrick E., and Joseph L. Fleiss. 1979. Intraclass correlations: Uses in assessing rater reliability. Psychological Bulletin 86 (2) (03): 420-8. Silver, M. S. 1988. User perceptions of decision support system restrictiveness: An experiment. Journal of Management Information Systems 5 (1): 51-65. Singh, Danièle Thomassin, and Michael J. Ginzberg. 1996. An empirical investigation of the impact of process monitoring on computer-mediated decision-making performance. Organizational Behavior and Human Decision Processes 67 (2) (8): 156-69. Springer, Leonard, Mary Elizabeth Stanne, and Samuel S. Donovan. 1999. Effects of small-group learning on undergraduates in science, mathematics, engineering, and technology: A meta-analysis. Review of Educational Research 69 (1) (Spring): pp. 21-51. Villanueva, J., S. Yoo, and D. M. Hanssens. 2008. The impact of marketing-induced versus word-of-mouth customer acquisition on customer equity growth. Journal of Marketing Research 45 (1): 48-59. Wang, Weiquan, and Izak Benbasat. 2009. Interactive decision aids for consumer decision making in E-commerce: The influence of perceived strategy restrictiveness. MIS Quarterly 33 (2) (06): 293-320. Wang, Weiquan, and Izak Benbasat. 2007. Recommendation agents for electronic commerce: Effects of explanation facilities on trusting beliefs. Journal of Management Information Systems 23 (4) (Spring2007): 217-46. Wang, Weiquan, and Izak Benbasat. 2012. A Contingency Approach to Investigating the Effects of User-System Interaction Modes of Online Decision Aids. Forthcoming in Journal of Information Systems Research.  Watson, R. T., G. DeSanctis, and M. S. Poole. 1988. Using a GDSS to facilitate group consensus: Some intended and unintended consequences. MIS Quarterly: 463-78. Wheeler, B. C., and J. S. Valacich. 1996. Facilitation, GSS, and training as sources of process restrictiveness and guidance for structured group decision making: An empirical assessment. Information Systems Research 7 : 429-50. Widing, R. E., and W. W. Talarzyk. 1993. Electronic information systems for consumers: An evaluation of computer-assisted formats in multiple decision environments. Journal of Marketing Research 30 (2): 125-41. 51 Xiao, B., and I. Benbasat. 2007. E-commerce product recommendation agents: Use, characteristics, and impact. Management Information Systems Quarterly 31 (1): 9. Yukl, Gary. 1989. Managerial leadership: A review of theory and research. Journal of Management 15 (2) (06): 251. Zeithaml, Valarie A. 1988. Consumer perceptions of price, quality, and value: A means-end model and synthesis of evidence. Journal of Marketing 52 (3) (Jul.): pp. 2-22. Zhu, L., I. Benbasat, and Z. Jiang. 2010. Let's shop online together: An empirical investigation of collaborative online shopping support. Information Systems Research 21 (4): 872-91. Zigurs, I., G. DeSanctis, and J. Billingsley. 1991. Adoption patterns and attitudinal development in computer-supported meetings: An exploratory study with SAMM. Journal of Management Information Systems7 (4): 51-70. Zigurs, Ilze, and Bonnie K. Buckland. 1998. A theory of Task/Technology fit and group support systems effectiveness. MIS Quarterly 22 (3) (Sep.): pp. 313-334. 52 Appendices Appendix A: Design of Experimental Systems Figure A-1: Control Condition – Home Page Figure A-2: Control Condition – Information Page of ProductsImages of RA screens removed for copyright reasons.53 Figure A-3: Control Condition – Shopping Cart Figure A-4: Flexible RA – Homepage Images of RA screens removed for copyright reasons.54 Figure A-5: Flexible RA – Sample Question of Attributes Figure A-6 Flexible RA – Recommendation Page Images of RA screens removed for copyright reasons.55 Figure A-7 Flexible RA – Helpful TipsFigure A-8 Restrictive RA – Home Page Images of RA screens removed for copyright reasons.56 Figure A-9: Restrictive RA – Sample Question of Attributes Figure A-10 Restrictive RA – Recommendation Page Images of RA screens removed for copyright reasons.57 Appendix B: Measurement Items Table A-1: Measurement Items for the Outcome and Control Variables Construct Names Measurements Items  (7 point scale) Sources Perceived Enjoyment of Shopping (EN) 1. I/We found this collaborative online shopping experience interesting.2. I/We found this collaborative online shopping experience enjoyable.3. I/We found this collaborative online shopping experience exciting.4. I found this collaborative online shopping experience fun.Koufaris et al. (2001) Perceived Purchase Decision Quality (DQ) 1. I/We am confident that our choice is the best for this shopping task.2. I/We am confident that we made the best purchase decision for thisshopping task. 3. I/We believe that we have made the best choice for this shoppingtask. 4. I/We believe that we selected the best digital camera for thisshopping task. Widing and Talarzyk (1993) Perceived Usefulness of System (PU) 1. Using this shopping website can improve our online shoppingperformance. 2. Using this shopping website can improve our online shoppingefficiency. 3. Using this shopping website can improve our online shoppingeffectiveness. Xu (2011) Perceived Intention to Use of System (IU) 1. Assuming we want to buy a digital camera, we intend to use thisshopping website. 2. Assuming we want to buy a digital camera, we would use thisshopping website. 3. Assuming we want to buy a digital camera, we plan to use thisshopping website. Wang and Benbasat (2009) Perceived Task Involvement (TI) 1. This shopping task that we have experienced is relevant to me.2. This shopping task that we have experienced is of concern to me.3. This shopping task that we have experienced matters to me.McQuarrie and Munson (1992) Perceived Friendship with Partner (FR) 1. How close would you characterize your friendship with yourshopping partner? 2. How strong would you characterize your friendship with yourshopping partner? 3. How important would you characterize your friendship with yourshopping partner? Zhu et al. (2010) Perceived Restrictiveness of System Interaction (RSS) 1. We had limited control over how we could change our desiredattributes for the camera. 2. This shopping website did not allow us to be flexible in sequencingour answers to questions. 3. The process of specifying attributes for camera on the shoppingwebsite was highly structured. 4. The shopping website constrained the process of changing thedesired attributes for the camera Wang and Benbasat (2012) 

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.24.1-0307453/manifest

Comment

Related Items