تاثیر خطر و اثر آن بر بهره وری جستجوی اطلاعات

We develop and test a theoretical framework of the joint influence of risk and affect on information search efficiency. Our framework proposes that information search is less efficient (i.e., less strategic) when risk is high, versus low. It further proposes that the influences of positive and negative affect on search efficiency are asymmetric and depend on the level of risk. Negative affect improves search efficiency when risk is high, but not when it is low. Positive affect degrades search efficiency when risk is low, but not when it is high. We find results consistent with our framework in two experiments. We discuss implications for affect research and for decision making in risky contexts, including financial statement auditing.

Information search is a critical aspect of many professional and personal decision tasks, including the auditor’s task of determining the acceptability of a client’s financial reports and the individual’s tasks of selecting an appropriate investment and deciding which school to attend or job to accept (e.g., Barrick and Spilker, 2003, Cloyd and Spilker, 1999, Einhorn and Hogarth, 1981 and Simon, 1977). The amount and relevance of information a decision-maker selects to examine in diagnostic tasks such as these affect the quality of subsequent decisions. In particular, information search efficiency, or the amount of relevant information gathered relative to the amount of information available, is an important determinant of the quality of subsequent decisions for at least two reasons. First, decision quality appears to be maximized when decision-makers use selective sets of highly relevant information (Keller and Staelin, 1987 and Rosman et al., 1999). The cognitive costs of using larger information sets can reduce decision quality through information overload, even if the quality of the information is held constant (Axt-Adam et al., 1993 and Keller and Staelin, 1987). If the quality of the information set is not held constant as the size increases, additional problems can arise from inappropriate reliance on irrelevant items (Bastardi and Shafir, 1998 and Chinander and Schweitzer, 2004) or from over-weighting of redundant information (Hall et al., 2007 and Joe, 2003). Second, in applied settings, gathering information is typically costly. For example, independent financial statement auditors are faced with the task of reaching critical decisions about the acceptability of a client’s reported financial numbers within limited time budgets. Professional investors face similar time and cost constraints in evaluating information and making investment choices. If external variables interfere with decision-makers’ ability to efficiently search for information, they may either extend information search, thereby increasing search costs and costs of missed deadlines, or make a decision with insufficient evidence, thereby increasing expected costs of decision error. Risk and affect are external variables common in professional decision settings that have the potential to influence search efficiency. We define risk as the variance of potential outcomes (e.g., Kahneman and Tversky, 1979, Libby and Fishburn, 1977 and Sharpe, 1964). Risk is an important aspect of natural decision settings, and is likely to influence information search behavior because decision-makers tend to prefer certainty to expected values and estimates (e.g., Sitkin & Pablo, 1992). In particular, auditors, investors, and other decision-makers faced with high variance in potential outcomes can reduce the variance by collecting more or more relevant information. For example, auditors extend evidence search when perceived risk is high (Bedard and Johnstone, 2004 and Blay et al., 2007). Affect is a general term used to describe a variety of feeling states, including moods, which are relatively diffuse, low-intensity, and enduring states of feeling generally good or bad, as well as emotions, which are relatively specific, intense, and short-lived (Forgas, Wyland, & Laham, 2006). Like risk, affect is pervasive in natural decision settings. Moreover, affect influences information processing in a variety of ways. Most research in this area focuses on the impact of ambient mood, as opposed to emotion (Forgas et al., 2006). Negative affect has been shown to focus information search (Fiedler et al., 1986, Isen, 1984, Keinan, 1987, Lewinsohn and Mano, 1993 and Mano, 1992). On the other hand, studies of the influence of positive affect on information processing tasks indicate that positive affect may reduce focus of attention (George & Brief, 1996) and increase risk aversion (Mittal & Ross, 1998), which would lead to less efficient information search (Mittal & Ross, 1998). We develop a theoretical framework that predicts when and how the increased focus from negative affect and the reduced focus from positive affect will moderate the effects of risk on the efficiency of information search. We conduct two experiments using tasks that are similar to an auditor’s verification task to test our predictions. We manipulate ambient affect and risk in both experiments. Information is costly.

Experiment 2 provides converging evidence that high risk degrades search efficiency in a neutral affect condition. Additionally, negative affect improves search efficiency in high-risk settings and positive affect degrades search efficiency in low-risk settings. We began this section by stating that our framework may not hold under higher demand for information because search may be less strategic overall or because positive affect participants might avoid information and search less. Neither of these explanations is consistent with the data. While some of the (untabulated) search measures indicate less strategic search in Experiment 2 versus 1 (e.g., search was generally less targeted in Experiment 2); others show the opposite pattern (e.g., sequential search scores are generally lower—indicating more strategic search—in Experiment 2). Thus, the greater expected demand for information in Experiment 2 did not appear to reduce strategic search. Also, on average, positive affect participants collected more cues rather than fewer in Experiment 2 as compared with Experiment 1 (untabulated). Indeed, Fig. 1 and Fig. 2 show very similar patterns of results across the two experiments, indicating that our theoretical framework extends to situations with varying levels of demand for information. General discussion Despite the importance of risk in decision-making (e.g. Johnson & Tversky, 1983), prior research has not addressed the influence of risk on information search behavior. Further, although there is some evidence regarding the role of negative affect on information search (Isen, 1984, Lewinsohn and Mano, 1993 and Mano, 1992), far less work has investigated the role of positive affect on information search (Jonas et al., 2006). We develop a general framework to predict the joint influence of risk and affect on information search efficiency. Our framework results in three main predictions. First, we posit that under neutral affect, information search is less efficient when risk is higher. Second, we propose that negative affect is associated with more efficient search than neutral affect when risk is high. We expected that this focusing effect of negative affect would occur when risk is high and individuals are willing to bear the cognitive costs of determining a high quality strategy, but not when risk is low and motivation to engage in the additional effort is lower. Third, we propose that positive affect is associated with less efficient search than neutral affect when risk is low. We expected that this distraction effect of positive affect would occur when risk is low, but not when risk is high and individuals are more motivated to avoid it. Results of our experiments support this framework. Theoretical implications and future directions In many contexts, prior research has posited and found that decision-makers gather more information in high-risk situations (Bedard and Johnstone, 2004 and Cho and Lee, 2006; Kadous et al., 2008). Our findings indicate that the additional information search does not focus on the most diagnostic (relevant) cues. Because less relevant information can impair decision making, our finding has implications for the quality of decisions made in risky situations. Recent research indicates that this finding likely extends to more complex professional contexts such as financial statement auditing, in that auditors exposed to increased risk collect more information in an audit, but that additional information decreases audit quality because it is not diagnostic (Hammersley et al., 2011). Future research can further investigate whether our framework extends to other professional contexts, such as those with incomplete information and those with affect-laden risk. For example, we manipulated risk as variance in expected outcomes (Experiment 1) and variance in payouts (Experiment 2). Neither of these manipulations is likely to influence affect directly; however; manipulations of risk that influence affect (Slovic, 2002) may lead to different conclusions. Prior literature provides considerable evidence that negative affect focuses information search (Isen, 1984, Keinan, 1987 and Lewinsohn and Mano, 1993) in tasks where all information is equally relevant. The results of our experiments indicate that this focusing effect is context dependent. That is, more efficient search is unlikely to occur in low-risk settings where the need for information is low and the importance of searching well is minimal. On the other hand, gains from the focusing effect of negative affect are obtained when risk is higher. In contrast, little evidence existed about the influence of positive affect on information search, and evidence on the influence of positive affect on other decision-making tasks is mixed. Although positive affect can reduce focus of attention (George & Brief, 1996), it also increases creative problem solving (Isen et al., 1987; Davis, 2009). The results of our experiments indicate that positive affect distracts participants, making their search less efficient, but also shows that this effect is, too, context dependent. In high-risk settings, neutral affect decision makers appear to gather information relatively indiscriminately, and the distraction of positive affect does not increase this inefficiency. On the other hand, in low-risk settings neutral affect participants tend to gather relatively cues in a more efficient manner, and the distraction effect results in less efficient search for positive affect decision makers. The results discussed above provide evidence of the asymmetric effects of positive and negative affect. That is, positive and negative affect do not necessarily have opposing influences on cognitive tasks (Forgas, 1995). Our findings provide insight into the conditions under which positive and negative affect will influence decision behavior. Future research can examine variables in addition to risk that cause asymmetric impacts of affect on important decision tasks. Future research can also examine whether our framework is generalizable to variations in task features. For example, our task requires strategic thinking, but not a great deal of creativity. Tasks incorporating more complexity, such as those with multiple variables and a changing relationship among the variables (e.g., Gonzalez, 2005) demand more innovation for successful performance. Because positive affect facilitates innovative problem solving (Isen et al., 1987; Davis, 2009), it is possible that positive affect may have beneficial impacts in information search tasks with this feature. This possibility merits further investigation. In addition, the cues in our information search task involved no uncertainty. Choosing a cue resulted in perfect information about that cue. However, information is not always perfectly reliable. Future research could use our framework to investigate information search efficiency when the reliability of individual cues is in question. For example, if unreliable information increases assessed risk, our framework predicts decreased search efficiency with less reliable information. Further, if information from multiple sources is inconsistent, the conflict may increase task difficulty, therefore increasing negative affect (Sawers, 2005). Here our framework predicts that the negative affect would mitigate the influence of higher perceived risk to moderate any increase in search inefficiency.