Organizational Behavior and Human Decision Processes, Volume 70, Issue 2

this influence is measured by the improvement in the Why do people take advice? To find out, we provided quality of the judgment. Here we shall be concerned a low, medium, or high level of training on a task in primarily with factors that determine the influence that which judgments varied in importance. Then, in a test advice has on a judge. session, we eliminated feedback but made advice availHogarth and Einhorn’s (1992) Belief Adjustment able. Different features of our results suggest that model provides some insight into how advice may be there are three distinct reasons for taking advice. First, integrated with an individual’s prior beliefs. Consider even experienced judges took some advice from nova doctor who has assessed a terminally ill patient’s life ices: people appear reluctant to reject completely the help offered to them. Second, all judges took more adexpectancy and then asked a colleague for an indepenvice from advisors more experienced than themselves, dent opinion. According to Hogarth and Einhorn (1992), and the amount that they took was related to the differthe doctor forms a final opinion by taking a weighted ence in level of experience: people appeared to be tryaverage of his or her prior opinion and the colleague’s ing to use advice to improve their judgments. Third, advice. The weight given to the advice depends on the experience enabled people to distinguish judgments person’s sensitivity to new information. This, in turn, on the basis of their importance. Experienced judges depends on situational variables such as the credibility took about twice as much advice for the most important of the source of the new information (e.g., the colleague’s judgments: they appeared to be sharing responsibility seniority). However, Hogarth and Einhorn (1992) also when the risk associated with error was high. Thus, we model advice-taking in terms of three components: argue that sensitivity to information varies with peraccepting help, improving judgment, and sharing resonal variables: their example of this is a scientist who sponsibility. q 1997 Academic Press is not sensitive to information about a favorite theory. Their model therefore implies that people receiving advice should take their beliefs about their own level of In many institutions, judges or executives take indiexpertise into account as well as their beliefs about the vidual decisions after conferring with one or more adviexpertise of their advisors. sors or consultants. With a few rare exceptions (e.g., At present, there appears to be no direct evidence Gardner & Berry, 1995; Sniezek & Buckley, 1993, 1995), that either of these factors is important in determining judge-advisor systems of this type have not been studthe influence of advice on judgments themselves. Howied systematically by psychologists. To investigate ever, recent work on people’s confidence in their judgthem, a person can be asked to make an initial judgment ments suggests that they take their perceptions of both on the basis of some evidence, provided with advice their own expertise (Arkes, Christensen, Lai, & Blumer, from some source, and then asked for a final judgment. 1987; Trafimow & Sniezek, 1994) and their advisors’ Influence of the advice is measured by the size of the expertise (Sniezek & Buckley, 1995) into account. Arkes shift between the initial and final judgment. Effect of et al. (1987) and Trafimow and Sniezek (1994) found that experimental manipulations designed to reduce Parts of this paper were presented at the Seventh Meeting of the participants’ perceptions of their expertise in answerEuropean Society for Cognitive Psychology, Lisbon, 1994, and at the Sixteenth International Symposium on Forecasting, Istanbul, 1996. ing a set of questions decreased their confidence in their The research was in part funded by Economic and Social Research answers, whereas those designed to elevate their perCouncil Grant R000236827. ception of their expertise increased their confidence in Address correspondence and reprint requests to N. Harvey, Departtheir answers to the same set of questions. Sniezek ment of Psychology, University College London, Gower Street, London WC1E 6BT, UK. Fax: 144 171 436 4276. E-mail: [email protected]. and Buckley (1995) examined whether two advisors’