A Framework for Analyzing Decision Aid User Interactions with Decision Aids

ABSTRACT This paper presents a model that describes how decision aid users interact with decision aid recommendations. Past research assumes that users reject a decision aid recommendation if they disagree and rely upon a decision aid recommendation if they agree with the recommendation. In fact, user interaction with an aid may be more complex and is influenced by organizational policy regarding relying upon an aid recommendation and the perceived utility, self-perception, and impression management gained through relying upon an aid. Understanding user interaction with a decision aid recommendation presents opportunities to further decision aid research. The model may also be useful in setting an organization context of using decision aids. INTRODUCTION Whether a decision aid is accepted by its users is an important question in decision aid research. Systems acceptance in general is a crucial question in information systems research. This paper presents a framework for analyzing user acceptance of decision aid recommendations. A pair of frameworks already exist that focus on antecedents of decision aid acceptance or rejection. The framework presented in this paper focuses on the decision aid user's actual decision about whether to accept a decision aid recommendation. The framework is based on the Social Response Context Model (MacDonald et al., 2004), and is useful for predicting how decision aid users may react to different social contexts in using decision aids. This framework is useful in two ways. First, it provides guidance for future research on decision aid reliance. Most research on decision aid reliance focuses on factors that are associated with decision aid reliance without placing those factors within a larger context. This has led to a fragmented understanding of decision aid reliance. Development of a framework can guide the structure of future research. Second, a framework can guide implementers of decision aids both in recognizing pitfalls of an implementation and in devising strategies to increase decision aid reliance. The following sections will review the research literature regarding decision aids and present the framework. DECISION AIDS Decision aids may be used for a variety of purposes, such as improving group communications, promoting creativity, or assisting data gathering. For the context of this paper, a decision aid is a tool used to recommend a solution to a problem. Research on decisions aids has sought to answer two related questions. First, does using a decision aid improve the quality of a decision? Second, are decision aid users willing to use a decision aids recommendation? Decision aids regularly outperform decision makers' un-aided decisions (Dawes et al, 1989). Examples of decisions aids that have been shown to have superior accuracy to experts include bankruptcy prediction (Sun, 2007), management fraud assessment (Hansen et al., 1996; Eining & Jones, 1997; Bell & Carcello, 2000), and audit materiality judgment (DeZoort et al., 2006). In addition to improving accuracy, firms may adopt decision aids to guide or direct a decision process (Silver, 1990). Even though decision aids tend to improve the quality of a decision and promote firm goals, decision aid recommendations are often ignored by users. Investigating what encourages or discourages users to accept or reject decision aid recommendations has long been an area of interest to researchers. A firm implementing a decision aid would find the aid useless without users being willing to adopt the aid's recommendation. Indeed, may decision aids used by audit firms have fallen into disuse (Gill, 1995). Probably the most common finding in research is that more expert users are less likely to rely upon a decision aid than novice decision makers (Arkes et al., 1986; Whitecotton, 1996; Glover et al., 1997). This seems intuitive, as more experienced decision makers have more confidence in their decision ability than novices. …