Abstract: In the DSS literature, most studies have concentrated on the direct effects of DSS use and DSS design on decision outcomes and user performance in the workplace. Fewer DSS studies have integrated decision process variables, such as user beliefs and attitudes, in their models. In this paper, we examine the mediating role of decision process variables in the use of an online customer DSS. We do so through an experimental study of an alternative-based and an attribute-based DSS for product customization by online customers. Using Cognitive Fit and Flow theories, we develop a theoretical model with four mediating decision process variables (perceived usefulness, perceived ease of use, perceived enjoyment, and perceived control) and two of their antecedents: interface design (attribute-based vs. alternative-based) and task complexity (choice set size). Our results show that the impact of DSS interface design on behavioral intentions is fully mediated by perceived usefulness and perceived enjoyment, though not by perceived control. Specifically, we find that users of an attribute-based DSS express higher perceived usefulness and perceived enjoyment than users of an alternative-based one. In addition, we verify that task complexity has an interesting relationship with usefulness and enjoyment, both of which follow an inverted U-shaped curve as choice set size increases. Finally, we find that for users of the alternative-based DSS, perceived ease of use and perceived control decrease as task complexity increases. However, the attribute-based DSS alleviates that diminishment for both variables. Among other contributions, our results indicate the importance of including decision process variables when studying DSS as well as the complex effect of task complexity on those variables. Our study also provides some important guidelines for online companies that provide customer DSS on their websites, especially the danger of providing too many product choice options which can overwhelm customers and harm their shopping experience.
Abstract: Shopping engines of different designs have been researched fruitfully as a means of convenient access to the goods and services on the Web. Some shopping engines are designed to function autonomously in one stage, searching to quickly maximize decision accuracy as a function of several parameters. Others are designed to strongly involve the user, searching in multiple stages to satisfy decision accuracy requirements. This study tested single-stage and multiple-stage shopping engines, designed with two approaches, termed QuickSearch and AdaptiveSearch, on 205 users who were trying to maximize accuracy while minimizing effort. The results show that the best performing shopping engine used two stages, first QuickSearch, and then AdaptiveSearch. The results imply that QuickSearch and AdaptiveSearch, although logically equivalent, have different impacts on shopping for differentiated, multi-attribute goods and services. In practice, the results of this study suggest that shopping engines should be designed to first save the shopper effort, and then provide attribute-focused support for examining the resulting set of items.
Abstract: Many online retailers try to assist consumers via parametric search engines xD;(PSEs), i.e., attribute-based search engines. The objective of this study is to model xD;the effectiveness of four PSEs in using search effort and domain knowledge to xD;increase decision quality, decision confidence, perceived ease of use and xD;perceived usefulness. Our model comprises a set of behavioral decision theory xD;antecedents to the technology acceptance model. We tested users with four PSEs xD;in a laboratory experiment and modeled the results with Partial Least Squares xD;(PLS) analysis. The main result of this paper is a PLS model showing that the xD;effects of search effort and domain knowledge are mediated through decision xD;quality and decision confidence to impact perceived ease of use and perceived xD;usefulness. The model explains the variance in decision quality (17.3%), decision xD;confidence (28.3%) and perceived usefulness (27.0%). Overall, this study shows xD;that input, process and outcome variables are important for predicting the xD;effectiveness of PSEs. Implications for research and practice are discussed.
Abstract: We offer a review of three software packages that estimate directed acyclic graphs (DAGs) from data. The three packages, MIM, Tetrad and WinMine, can help researchers discover underlying causal structure. Although each package uses a different algorithm, the results are similar. All three packages are free and easy to use. They are likely to be of interest to researchers who do not have strong theory regarding the causal structure in their data. DAG modeling is a powerful analytic tool to consider in conjunction with, or in place of, path analysis, structural equation modeling, and other second-generation statistical techniques.
Abstract: This study examines the impact of handheld computers on students in a first year Information Technology course (IT101), measuring IT attitudes, skills, and performance. Students in standard sections used their laptop computers and traditional textbooks. Students in intensive sections used both handheld computers and laptop computers, that is, multiple platforms. Both standard and intensive sections covered the same topics. Although students in both standard and intensive sections became more skilled during the semester, students in the intensive sections improved more in programming and Microsoft Excel skills. Students in intensive sections also scored significantly better than students in the standard sections on the common portion of the final exam. Finally, according to anecdotal evidence, students in the intensive sections gained significantly more skill-based confidence in IT unrelated to handheld computers compared to students in the standard sections. Seven of our ten expectations were supported. While students benefited from the intensive version of IT101, several factors could account for the improvements: a self-selection factor, an active learning vs. textbook learning factor, and possibly an instructor factor. Future research should shed light on the relative weights of these factors.
Abstract: Since technology adopters are sensitive to perceived ease of use, online shoppers who are novices to a product category may be particularly sensitive. We developed a predictive model of decision quality centered on perceived ease of use for online shoppers who are novices to a product category. The antecedents of perceived ease of use in this study are novice category knowledge, perceived time pressure and purchase involvement. We had undergraduate students interact with multiple-stage shopping engines (MSSEs) of varying effort-accuracy tradeoffs in our purpose-built Web store. Both novice category knowledge and perceived time pressure negatively impacted purchase involvement and perceived ease of use of the MSSEs. Perceived ease of use of the MSSEs positively impacted their perceived usefulness. Perceived usefulness positively impacted decision quality, which was measured objectively, with data envelopment analysis, and subjectively, with the weighted additive method. Six of the seven hypotheses were supported. Surprisingly, there were no significant differences among the MSSEs and purchase involvement did not positively impact perceived ease of use. The results, analyzed via partial least squares, demonstrate the potential value of giving multiple-stage shopping engines to online shoppers who are novices to a product or service category.
Abstract: Champions are critical to the success of an information system implementation. Research shows that success in implementing a large information system may hinge on there being more than one type of champion. This study investigates the types of champions used in federated inter-organizational systems (FIOS) in a state telemedicine context. Case studies were conducted in Georgia, Pennsylvania, Ohio, and Wisconsin to identify the network of champions in state telemedicine systems. We found that FIOS that relied on a network of champions, including a sponsorship champion at the state level, as well as a technical champion and user champion at site locations, were more successful than those that lacked such a network. We suggest that our model of champions in FIOS applies not only to state telemedicine, but also to any large-scale system implementation spanning a federation of loosely coupled organizations.
Abstract: How do we model and improve the learning of technical skills such as network subnetting? This paper reports an experimental study that tested three hypotheses, two derived from Kolb’s experiential learning cycle, which is concrete-based, and one derived from the advance organizer technique, which is abstraction-based. The two hypotheses following Kolb’s cycle were supported, and the one based on advance organizers was not supported. The results indicate that the learning of network subnetting starts concrete, becomes more abstract, and ends concrete. The lack of support for advance organizers indicates that they are not a good fit for technical problem solving, or that they need to be designed differently for it. We draw some implications for research and practice. We also call for additional research to extend Kolb’s cycle to other technical and scientific problem solving domains.
Abstract: Telemedicine networks bring together a range of medical facilities often governed by an external entity or task force. In order for a telemedicine network to be successful, it needs sufficient support from key individuals within the network. The collective skill set and logistical placement of these individuals, knows as champions, is critical to produce an effective telemedicine implementation. This study investigates the types of champions used in state telemedicine programs to determine the most effective champion mix. Research was conducted on Georgia, Pennsylvania, Ohio, and Wisconsin programs to identify the cadre of champions existing in state telemedicine systems. Data collected included interviews, as well as information related to each program reported in magazines, newspapers, medical journals, academic journals, web sites, state legislation records, minutes from state telemedicine task force meetings and internal evaluations. Triangulated qualitative analysis indicates that the telemedicine systems that relied on a network of champions that included a sponsorship champion at the state level, as well as a technical champion and user champion in each site were more successful than those that lacked such a network. Furthermore, the more efficacious implementations employed a greater overall number of technical and user champions at the participating sites.
Abstract: In this study, we address the question of the effect of domain knowledge on performance with a decision tool in a preferential choice task. Specifically we explore the relationship between knowledge, effort and performance (accuracy), drawing on theories from both behavioral decision theory and cognitive science. Using a purpose-built web shopping simulation we study 56 subjects interacting with two different decision tools varying in effort and knowledge requirements; a lexicographic tool (LEX) and a parametric search tool (PS. Using a novel measure for accuracy based on Data Envelopment Analysis we find that effort positively influences accuracy, with a greater effect observed for the more powerful decision tool, PS. In conflict with theory and intuition we find knowledge has a negative effect on accuracy, and that this effect is greater with the more powerful decision tool. We interpret our results in terms of knowledge-effort substitution, the content and representation of knowledge, and the conceptual characteristics of the decision tool. Our results emphasize the importance and complexity of knowledge considerations in understanding performance with decision support tools. Pragmatically, it highlights the importance of considering who a decision support system is designed for, and how the use of that decision support may both constrain and enable effective performance.
Abstract: Prior efforts in personalization have focused primarily on modeling individual consumer's preferences so that products for which they have a higher likelihood of purchasing are presented. In this study we explore the potential of an approach to personalization focusing on customizing shopping tools based on a consumer's Product Category Knowledge. The low Product Category Knowledge user may not be able to use the shopping tools as well as the high Product Category Knowledge user, because lower Product Category Knowledge users allocate cognitive power to learning the product attribute space at the expense of using the tool effectively. Alternatively, shopping tools may effectively guide the decision making of low Product Category Knowledge users, but be perceived as too restrictive by high Product Category Knowledge users, thus diminishing their Decision Confidence. To determine which of these scenarios holds true we had sixty-six subjects interact with different decision tools of varying effort-accuracy tradeoffs in our purpose-built web store for purchasing a computer. We examine the impact of Product Category Knowledge on perceptions of ease of use and usefulness of the decision tools, and ultimately Decision Confidence, a key predictor of purchase likelihood. The results evidence the potential value of adapting tools to the degree of user Product Category Knowledge. High Product Category Knowledge users may require less restrictive decision tools to promote Decision Confidence, whereas low Product Category Knowledge users may require simpler tools and more decisional guidance
