Anchoring and Knowledge on the Numerical Value and Accuracy - Lab Report Example

The Effect of Anchoring and Knowledge on the Numerical value and Accuracy of an Estimated Date Abstract This study was designed to explore the possible effect of the value of the anchor and the presence of the anchor on the numerical value of the participants’ responses and on the accuracy of those responses. The study also tried to evaluate the effect that knowledge had on the dependent variables. It was expected that both the value of the anchor and the presence of the anchor would influence the value and the accuracy of the responses. Similarly, the study expected to find that knowledge enhanced these effects. The results demonstrated that while anchoring and knowledge did influence the numerical value of the responses, no effect was found on accuracy. The reasons for this are also discussed. The Effect of Anchoring and Knowledge on the Numerical Value and Accuracy of an Estimated Date Introduction When asked to respond to a question, people like to provide an answer that is as accurate as possible. When the correct response to a question is known or when the correct response is easy to calculate, the individual is not affected by external issues and can promptly provide this response (Mussweiler, Englich, & Strack, 2005). But when the person does not know the correct response to a question or when there is some ambiguity about which response is correct, the individual in question is likely to use any cue that will help him or her estimate a seemingly accurate response to the question at hand (Furnham & Boo, 2011). Often in such cases the correct answer is not available to the responder or the information necessary to calculate it may not be available. At such times, people use mental shortcuts called heuristics (Mussweiler, Englich, & Strack, 2005). One kind of heuristics that people often unwittingly use is that of anchoring. Anchoring involves the use of available information, whether relevant or arbitrary, to estimate the correct response to a question (Furnham & Boo, 2011). Anchoring effects are widespread and occur even when the person is aware that the initial value presented to him/her is arbitrary. Tversky and Kahneman (1974) have presented some of the seminal works in this area and have established that the anchoring effect works under a number of different conditions. In one of their experiments, they asked participants to estimate answer to different questions in terms of percentages. For example, they asked the participants to estimate the percentage of African countries in the UN. A wheel of fortune that indicated numbers between 1 and 100 was spun before the participants gave their answers, and it was found that the number indicated by the wheel (which was known to the participants to be very arbitrary) was strongly associated with the estimate provided by the participant. Participants who saw the wheel of fortune show the number 10 estimated the percentage as being significantly lower (25%) as compared to participants who saw the number 65 (average estimate in this case being 45%). Tversky and Kahneman (1974) also demonstrated that people use incomplete knowledge to draw estimates about a final response. For example, they found that when a sum was represented as 1 × 2 × 3 × 4 × 5 × 6 × 7 × 8, the estimated value was much lower than when the sum was represented as 8 × 7 × 6 × 5 × 4 × 3 × 2 × 1. It seems that the participants managed to compute the first few units and corrected for the final answer based on the value they were able to arrive at. When participants were given adequate time to solve the sum, anchoring did not occur (Tversky & Kahneman, 1974). A number of explanations of the anchoring effect have been proposed. Strack and Mussweiler (1997) found that participants would respond to an anchor pertaining to an event like the birth year of Einstein both when the anchor provided was plausible (for example, 1920) as well as when it was implausible (for example, 1245). They suggested that the anchor that is provided primes relevant information in the participants’ minds, and thus makes information in the appropriate direction (higher/lower than anchoring date) more accessible. Mussweiler, Englich and Strack (2005) have discussed the pervasiveness and the robustness of this phenomenon of anchoring as well as it’s relevance to a number of fields of research. According to them, a number of effects have been seen with respect to anchoring. The best one described is that of insufficient adjustments as described by Tversky and Kahneman (1974). This effect describes how people use the anchor to correct towards the right answer. Usually, these corrections are inadequate and are heavily influenced by the starting point introduced by the anchor. This occurs when the person knows the anchor to be unrelated to the response and when the person believes that the values are related (Northcraft & Neale, 1987). The second effect is that of conversational inferences. This means that people use the anchor to infer the possible range within which the true response may fall. This is best demonstrated by Northcraft and Neale’s (1987) study where people used the provided information to estimate the price of houses. The third effect seen is that of numerical priming, which suggests that the anchoring value simply primes the respondent to a certain set of numbers close to the anchoring value (Mussweiler, Englich, & Strack, 2005). The final mechanism suggested for anchoring is that it creates a selective accessibility of information which suggests that the anchor makes items similar to it more accessible to the respondent than items that share fewer similarities with the anchor (Mussweiler, Englich, & Strack, 2005). Bergman, Ellingsen, Johannesson and Svensson (2010) have found that while anchoring effects are pervasive, their effect reduces with increase in knowledge and cognitive ability. Based on the information reviewed, it may be said that an individual who receives an anchor is more likely to provide a response that is similar to the anchor (Furnham & Boo, 2011), but the knowledge that an individual has about the event will also influence this response. Since anchors help define a range for responses, the people who receive anchors should be more accurate in responding to questions, and, again, their knowledge will mediate these responses. Thus, the hypotheses for this study have been developed on the basis of this understanding of how anchoring interacts with knowledge to influence responses of participants. Figures Figure 1 represents the data in table 1. Mean scores and the standard error of responses are presented. Figure 1: Numerical value of responses and the standard estimate of error Figure 2 represents the interaction between these values of the anchor and knowledge. It is evident that there is no interaction between these lines. Figure 2: Interaction between anchor value and knowledge. Figure 3 provides a graphical representation of the data in table 2. The mean accuracy scores and standard error of responses are presented. Figure 3: Graph for accuracy of responses and the standard error in accuracy (smaller score is associated with more accuracy) Discussion This experiment was carried out to test the effect of anchoring based on arbitrary numbers and previous knowledge on the numeric value and the accuracy with which participants were able to guess the year in which a variety of events occurred. The events in question were posed as general knowledge questions about little known historical events of the first millennium. The results based on the present data have demonstrated that anchoring did indeed affect the estimate provided by the participant. Knowledge also affected the responses of the participants – the participants with more knowledge were more likely to provide a higher number as compared to participants with no knowledge. Thus, the first hypothesis – “for participants given an anchor, the higher their anchor is the higher their estimates will tend to be” – and the auxiliary hypothesis – “the anchoring effect in hypothesis 1 will be different for participants reporting more knowledge” – were both verified by the data. On the basis of these results, one can safely conclude that it is evident that arbitrary values do influence the estimated responses. These findings are in line with those of Tversky and Kahneman (1974). Mussweiler, Englich and Strack (2005) and Furnham and Boo (2011) have demonstrated the pervasive and robust nature of the anchoring effect and have explained how even an arbitrary number may provide a starting point from which the participants will try to correct for the answer. It is also possible that when an anchor is presented, even when the participant knows that it is arbitrary, similar numbers become more accessible to him or her (Strack & Mussweiler, 1997). This could lead to producing the anchoring effect. On the other hand, it was observed that the accuracy of responses as measured for this experiment was unaffected by either independent variable. Thus the second hypothesis – “participants given an anchor will produce estimates closer to the true year than those not given an anchor” – and the third hypothesis – “participants reporting more knowledge will make more accurate estimates and this effect will be greater for participants given an anchor” – were not supported statistically. Although the trends observed in the data were in line with the hypotheses, these trends were not statistically significant and thus may not be accepted empirically. Strack and Mussweiler (1997) believe that anchoring helps construct a response that is close to the true answer by priming relevant information. This has also been suggested by Mussweiler, Englich and Strack (2005) through their analysis of research conducted over 20 years. They have suggested that anchoring can enhance knowledge accessibility. The present results, however, do not support the mentioned theories, since neither anchoring nor knowledge seemed to improve the accuracy of the estimate. One possible reason for this could be the manner in which the variable of knowledge has been operationalised. For the purpose of this study, even the most minimal level of knowledge (some knowledge about only one out of 9 questions) was included under the condition of “Some Knowledge”, while only absolutely no knowledge about any of the questions was included under the condition of “No knowledge”. It has not been possible to demonstrate that these two conditions were truly different in the participants’ own experiences. Even if this difference is assumed, the condition of ‘Some Knowledge’ may be too wide as regards the type of responses it includes. This condition includes a vast heterogeneity of possible responses from some knowledge about only one event to a lot of knowledge about all events. Thus, the condition is likely to contain extreme scores associated with both guessing and knowledge. Since the initial data collect described knowledge under three possibilities – no knowledge, some knowledge and a lot of knowledge – it might help to use all three classifications of knowledgeableness in the analysis in order to improve the distinctions and reduce contamination. It is also possible that a particular participant has knowledge about a few events and guessed about some others. Guessing may contaminate the effect of knowledge when aggregating the responses for each subject. In order to avoid this, it may be possible to first evaluate the responses of all participants to each question and then aggregate the differences if necessary. This would help avoid the contamination of guessing with knowledge for each participant. Alternately, it is also possible to use the median as the measure of central tendency used to demonstrate the scores of each participant. Since the median is less sensitive to extreme values, it may provide a more accurate estimate of the experience of the participant. Conclusion In conclusion it may be said that for the present experiment, participants who did experience anchoring did demonstrate sensitivity to the value of the anchor. The results also demonstrated that knowledge affected the estimated value for participants who experienced anchoring. But when comparing participants who did not experience anchoring with those who did, no difference in accuracy was evident. Similarly, participants showed no difference in accuracy with respect to knowledge. The lack of significant differences for these groups may be associated with some methodological issues which have been discussed. These possible weaknesses should be considered during further analysis. References Bergman, O., Ellingsen, T., Johannesson, M., & Svensson, C. (2010). Anchoring and cognitive ability. Economics Letters, 107(1), 66-68. Furnham, A., & Boo, H. C. (2011). A literature review of the anchoring effect. Journal of Socio-Economics, 40(1), 35-42. Mussweiler, T., Englich, B., & Strack, F. (2005). Anchoring effect. In R. Phol (ed.), Cognitive illusions: A handbook on fallacies and biases in thinking, judgment and memory. New York, NY: Psychology Press. Northcraft, G. B., & Neale, M. A. (1987). Experts, amateurs, and real estate: An anchoring-and-adjustment perspective on property pricing decisions. Organizational Behavior and Human Decision Processes, 39, 84–97. Strack, F., & Mussweiler, T. (1997). Explaining the enigmatic anchoring effect: Mechanisms of selective accessibility. Journal of Personality and Social Psychology, 73, 437-446. Tversky, A., & Kahneman, D. (1974). Judgment under uncertainty: Heuristics and biases. Science, 185, 1124-1131. Read More