1. Individual Differences in the Influence of Confidence: The Effects of Need for Closure and Need for Cognition Caroline Wesson & Briony Pulford University of Wolverhampton University of Leicester Participants 110 undergraduates (86 women and 24 men) volunteered to take part in the study, and ranged in age from 18 to 46 years, with a mean age of 21.20 years (S.D. = 4.81). Materials Three different questionnaire based tasks were used to measure confidence heuristic use: a general knowledge task (Task 1) an opinion based task (Task 2) an evaluative task (Task 3) Each task consisted of 12 questions followed by three alternative answers, each ‘spoken’ by a different speaker. Where relevant, accuracy was kept constant across all three speakers. In the experimental condition, one speaker accompanied all their answers with high confidence cues, one with medium confidence cues and one with low confidence cues. In the control condition no confidence cues were used. The confidence cues were developed in an earlier pilot study, and consisted of phrases such as: “I’m certain it’s…” (high confidence) “I think it’s…” (medium confidence) “I’m not sure but it could be…” (low confidence) Two personality questionnaires were also used: Need for Closure (Webster & Kruglanski, 1994) Need for Cognition (Cacioppo & Petty, 1982) Design & Procedure A 3 (‘Speaker’ confidence: high, medium, and low) x 2 (Need for Closure: High vs. Low) x 2 (Need for Cognition: High vs. Low) x 2 (condition: confidence cues vs. no confidence cues) mixed design, with repeated measures on the first variable was used. For each question on the three tasks participants were asked to circle the ‘name’ (A, B or C) of the speaker corresponding to the correct/most likely answer, as well as indicating how confident they were in their chosen answer (0% – 100%). Participants also completed the two personality questionnaires. D.V.s were amount of time each speaker’s answers chosen and participants confidence in their chosen answers. Results (1) Confidence Heuristic Use Influence of Confidence on Choice The addition of confidence cues to a speaker’s answer led to shifts in participants’ choice of answer compared to when no confidence cues were used, F(2, 216) = 54.00, p <.001. The extent of these shifts was dependent on the task, F(4, 432) = 13.52, p <.001. Figure 1 illustrates this three-way interaction, showing the differences in the mean scores between the experimental and control conditions. On all three tasks the addition of confidence cues resulted in a shift towards answers given by the high confidence speaker, largely to the detriment of the answers given by the low confidence speaker. This was most apparent on Task 3, although substantial shifts in choice towards the high confidence speaker’s answer were seen on Task 1, where both the medium and low confidence speakers lost out equally. The smallest shifts in choice were seen on Task 2. (All p .001 with the exception of Medium confidence on Task 2, t(108) =.57, p =.57). Hence, the three-way interaction appears to arise from Task 2 being very different to the other tasks. Influence of Confidence on Choice Confidence The addition of confidence cues led to an increase in participants’ confidence in their answers, regardless of the level of confidence being expressed, by 17.98% on Task 1, F(1, 75) = 13.67, p <.001, by 7.52% on Task 2, F(1, 90) = 6.90, p =.01, and by 27.51% on Task 3, F(1, 52) = 8.25, p =.006. However, as Figure 2 shows, control group confidence is much higher on Task 2 than on the other tasks, leading to much smaller increases in confidence, reinforcing the suggestion that this task is different to the other two tasks. Results (2) Individual Differences in Confidence Heuristic Use Influence of Confidence on Choice Significant Need for Closure x Speaker Confidence x Condition interactions were seen on Task 1, F(2, 162) = 3.29, p =.04, and Task 3, F(2, 162) = 3.63, p =.03, but not on Task 2, F(2, 162) =.22, p =.80. From Figure 3 it can be seen that on Tasks 1 and 3, High (vs. Low) Need for Closure participants showed greater shifts in choice towards answers expressed with high confidence and away from those expressed with medium confidence (all p <.05). Comparable shifts in choice for High (vs. Low) Need for Closure participants were seen on Task 2, and on the low confidence speakers answers on Tasks 1 and 3 (all at p >.05). Need for Cognition had no effect on participants’ choice of answers on any of the three tasks (all p >.05). Influence of Confidence on Choice Confidence Need for Closure made no difference to participants’ confidence in answers overall or on any of the three tasks, F(1, 87) =.11, p =.74 and F(2, 174) =.60, p =.55. However, a significant Need for Cognition x Condition interaction indicated that High (vs. Low) Need for Cognition participants did not differ in their overall confidence in the control group (M = 41.44 vs. M = 43.14) but when confidence cues were used High Need for Cognition participants were more confident in their chosen answers than Low Need for Cognition participants were (M = 71.27 vs. M = 56.63), F(1, 86) = 5.99, p =.02. The lack of a significant interaction with task type indicated that this pattern was the same across all three tasks, F(2, 172) = 1.00, p =.37. Introduction Decision-making is often social in nature - we turn to others for their opinions or advice when we are uncertain. How we evaluate and interpret this information can determine whether or not we use it. A potential method of evaluating the quality of another person’s information is by attending to the level of confidence with which they express themselves, where confidence is defined as the strength of a person’s belief that a specific statement represents their best or most accurate response (Peterson & Pitz, 1988). Previous research has found information expressed confidently to be used more than that which is expressed tentatively or with some element of doubt in many situations including: eyewitness testimony (Leippe, Manion, & Romanczyk, 1992) knowledge based tasks (Sniezek & Van Swol, 2001) group-decision making (Zarnoth & Sniezek, 1997). However, the amount of influence that confidence has is variable, leading to the suggestion that the influence of confidence may be mediated by task type (Zarnoth & Sniezek, 1997). The Confidence Heuristic Thomas and McFadyen (1995) suggest that the most confident individual’s answers may be selected as a default option, using confidence as a heuristic by which to simplify the decision- making process. In utilising the confidence heuristic more weight is placed on a persons expressed confidence than on the actual content of what the are conveying. Expressed confidence is taken as a cue to accuracy, knowledge and competency, and the expectation is that someone is more likely to possess reliable information if it is expressed confidently rather than tentatively. However, the confidence heuristic may not be a general cognitive heuristic, in that it may not be used by everyone, at least not all of the time, or even in the same way (Price & Stone, 2004; Thomas & McFadyen, 1995). This experiment investigates whether there are individual differences in this by considering the role of two measures of personality that are of direct relevance to this research. The first, Need for Closure, refers to the general tendency to prefer certain to uncertain knowledge, the desire for a firm answers and an aversion to ambiguity (Webster & Kruglanski, 1994). People who are high in Need for Closure are motivated to produce quick and confident judgements (Mayseless & Kruglanski, 1987). The second, Need for Cognition, is the tendency for an individual to engage in and enjoy thinking, an individual’s need to organise, abstract and evaluate information (Cacioppo & Petty, 1982). It has been suggested that low Need for Cognition individuals rely more on the use of heuristic cues when evaluating information than those who are high in Need for Cognition, who actively think about the content of the information (e.g. Chaiken, Liberman, & Eagly, 1989). Both Need for Closure and Need for Cognition may affect how people use the confidence heuristic. Aims The aim of this study is to find out whether listeners faced with different levels of expressed confidence utilise the confidence heuristic when making choices, and the effect that this has on their subsequent confidence in those choices. Whether or not this is a general cognitive heuristic is investigated in relation to the effect of Need for Closure and Need for Cognition on the two dependent variables. Objectives: The influence of expressed confidence was investigated in relation to the choices people make and the confidence they have in those choices, to ascertain whether individuals use the confidence with which a person expresses their answers as a heuristic. To determine whether the confidence heuristic is a general cognitive heuristic or is mediated by individual differences, the influence of Need for Closure and Need for Cognition was also considered. Design: The experiment used a 3 (‘speaker’ confidence: high, medium, and low) x 2 (Need for Closure: High vs. Low) x 2 (Need for Cognition: High vs. Low) x 2 (condition: confidence cues vs. no confidence cues) mixed design, with repeated measures on the first variable. Methods: 110 undergraduates took part in the experiment in which they were required to choose the correct/most likely answer to a series of questions belonging to three different task types, and to indicate their confidence in their chosen answer. Participants were given three alternative answers to each question to choose from. In the experimental group these were accompanied by high, medium or low confidence cues developed in a previous pilot study, whereas in the control group there were no cues as to the speaker’s confidence. Participants also completed two personality questionnaires measuring Need for Closure and Need for Cognition. Results: The addition of confidence cues to a speaker’s answer resulted in a shift towards choosing answers expressed with high confidence and away from those expressed with low confidence. Regardless of a speaker’s confidence level, the addition of confidence cues led to an increase in participants’ confidence in their answers. However, the extent of these effects was dependent on task type. In relation to the personality measures used, Need for Closure had an effect on participants’ choice of answer whereas Need for Cognition affected participants confidence in their chosen answers. High (vs. low) Need for Closure participants showed a greater shift towards answers expressed with high confidence and away from those expressed with medium confidence. High (vs. low) Need for Cognition participants were more confident in their chosen answers. Conclusions: People do appear to use a heuristic that uses the confidence of a person as an indicator of the validity of their information. People use the heuristic when they are uncertain as a means of making choices and having confidence in those choices. However, the extent to which the confidence heuristic is used, and the way in which it is used, is influenced by individual differences. Abstract Method Discussion The results of this experiment provide support for the suggestion that people use the confidence heuristic as a way of simplifying the decision-making process. When people feel uncertain they turn to the confidence with which an answer is expressed as a basis for reaching a ‘best’ answer. Specifically, the higher the level of confidence used by a speaker, the greater that speaker’s influence is on the choices made by the listener. The extent to which we are influenced by another person’s confidence is mediated by task type, supporting Zarnoth and Sniezek’s (1997) suggestion. Taking participants confidence in their answers into account, it would seem that the greater our own uncertainty, the more we will use and rely upon the confidence that another person expresses as a way of making a decision. However, just the very expression of confidence by someone in relation to their information, regardless of the level of that confidence, is enough to make us more confident ourselves.. Although the confidence heuristic does seem to be a general cognitive heuristic, being situationally induced when uncertainty is high, the extent to which another’s confidence is used as a heuristic appears to be mediated by individual differences, supporting previous suggestions (Price & Stone, 2004; Thomas & McFadyen, 1995). Specifically, Need for Closure appears to affect our use of another’s confidence when making choices, whereas Need for Cognition affects how confident we are in those choices. For high Need for Closure individuals, using the confidence heuristic may satiate their desire for confident knowledge, allowing them to make quick decisions and confident choices by employing this strategy. Low Need for Closure individuals are less motivated to produce quick and confident judgements and so rely less heavily on the confidence heuristic. Need for Cognition did not lead to any differences in the extent to which the confidence heuristic was used, possibly because of situational factors that reduced all individuals’, regardless of their Need for Cognition, to relying on heuristic processing. Need for Cognition did, however, have an effect on how confident people were in their chosen answer. It may be that high Need for Cognition individuals do still try and engage in more cognitive effort when choosing their answers than their low Need for Cognition counterparts, which is reflected in their higher levels of confidence Further research is warranted to identify other factors that mediate our use of the confidence heuristic, such as how we perceive speakers who use different levels of expressed confidence and whether there are gender differences, in terms of both speaker and listener, in confidence heuristic use. Conclusions: People do appear to use a heuristic that uses the confidence of a person as an indicator of the validity of their information. People use the heuristic when they are uncertain as a means of making choices and having confidence in those choices. However, the extent to which the confidence heuristic is used, and the way in which it is used, is influenced by individual differences. Email: c.j.wesson@wlv.ac.uk