CHAPTER 2 LITERATURE REVIEW
2.4 A NCHORING E FFECT
國
立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
19
current situation or the expected result. “The location of the reference point and the consequent coding of outcomes as gains or losses can be affected by the expectations of the decision maker (Kahneman et al., 1979).” It can also be affected by the past experiences or the similar situations. Moreover, we can say reference point represents the attitude when considering a situation.
When an individual is going to consider using a new service, s/he sets a reference point to judge that s/he will feel gains or losses if s/he uses the new service. It is the threshold of taking a new service or not. So the reference point of the service can be used to represent the attitudes or thoughts of service because it is the criteria that makes the user think it is indifferent to him/her whether using the service or not. This can be the guide that how the service providers should improve their services.
For our research, we argue that reference point can be used to distinguish motivation for promotion and prevention since reference point is the threshold of accepting things and promotion-concerned people have a lower threshold and prevention-concerned people have a higher threshold.
2.4 Anchoring Effect
2.4.1 Anchoring Effect
Anchoring effect is a cognitive bias that describes that people are usually affected by during decision making process (Kahneman et al., 1974). Mussweiler (2002) said anchoring effect is the assimilation of a numeric estimate to a previously considered standard. Anchoring effect occurs when individuals use an initial piece of information to make subsequent judgments. Once the first piece
‧ 國
立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
20
of information provided is set as an anchor, judgments will be made by adjusting away from the anchor. And there is a bias toward interpreting other information around the anchor.
For example, when we are going to buy a used car, the initial price offered to us becomes the standard for the rest of the negotiations, so that prices lower than the initial price seem more reasonable even if they are still higher than what the car is really worth. Anchoring effect occurs in our daily life.
Kahneman and Taversky (1974) claimed that anchoring effect occurs because of the adjustments from the anchor are usually insufficient, giving the initial anchor a great influence over future judgments. Chapman and Johnson (2002) explained that the insufficient adjustments are caused by the lacking of cognitive effort.
Kahneman and Taversky (1974) implied that anchoring effects depend on the degree of uncertainty about the judgment. Supporting this notion, some related researches suggest that the size of the anchoring effect increases with uncertainty (Jacowitz et al., 1995). Jacowitz and Kahneman used some experiments to demonstrate that the more uncertain judges were about the judgment, the more were numeric estimates assimilated to the provided anchor values. For example, when a subject who lives in Asia is asked about the length of Mississippi River with a random number of 1700, s/he may use 1700 as an anchor and give the estimation close to the 1700. But a subject who lives in Minnesota may give a more accurate estimation than the former subject because s/he lives nearby Mississippi River and knows more about the river.
However, the research of Strack (1992) has mentioned that although the cognitive bias caused by the anchoring effect exists most of the time, sometimes people can still conquer the bias and make a more accurate estimation.
‧ 國
立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
21
Sometimes, people may not use the first provided information as an anchor or they may correct the information.
Anchoring effect happens in many of domains, like general knowledge, probability estimates, legal judgment, pricing decisions, and negotiation. We believe that it can be applied to understand the reasons behind people’s decision making process.
2.4.2 Selective Accessibility Model
However, Mussweiler (2002) gave a different explanation for anchoring effects and demonstrated that judgmental anchoring effects are semantic in nature. He said anchoring effect occurs because the semantic knowledge about the target is activated during the comparison with the anchor. In order to generate the final estimate, people judge primarily rely on easily accessible knowledge so that their estimate is influenced by the anchor-consistent knowledge that was generated before. For example, when we are going to buy a house but we are not sure the offered price is reasonable, the knowledge we have such as the price of the other houses in the neighborhood or the houses that have the similar specifications will be activated as the basis of the comparison. This leads to an assimilation of the final estimation to the anchor value.
The selective accessibility model has two parts (see Figure 2.2):
hypothesis-consistent testing (the selectivity hypothesis) and semantic priming (the accessibility hypothesis) (Mussweiler et al., 1999). The hypothesis-consistent testing is about after receiving an anchor value and a target object, the individual will test the hypothesis by examining the possibility that the target object’s value along the judgmental dimension is equal to the anchor
‧ 國
立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
22
value, which Mussweiler called “standard” (Mussweiler et al., 1999;
Mussweiler , 2002; Mussweiler , 2007). For the second part, semantic priming, if the consequence of test is equal, then it will increase the accessibility of standard-consistent knowledge and lead to an assimilation of the final estimation to the anchor value. If the consequence of test is unequal, then it will increases the accessibility of standard-inconsistent knowledge and lead to a contrast of the final estimation to the anchor value.
Figure 2.4.2 The Selective Accessibility Mechanism (Mussweiler, 2007)
The selective accessibility model points out that during the anchoring process, the anchor-related knowledge will be extracted and used to give the estimated value. We argue that can be used to urge people to judge the value of a specific target.2.4.3 Provided Anchors & Self-generated anchors
Previous researches (Epley et al., 2001, 2004, 2005, 2006) mentioned that anchors can be categorized into external provided anchors or self-generated
‧ 國
立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
23
anchors. When facing a decision making situation, people do not always have external information as provided anchors. But anchors still be generated by their past experience or similar situations they had, these anchors called the self-generated anchors. The self-generated anchors may activate more anchor-consistent knowledge when anchoring effect occurs than the provided anchors because people usually have strong belief in what they know. In the house buying example mentioned in the section 2.4.2, the offered price is the provided anchor and the knowledge we have such as the price of the other houses in the neighborhood or the houses that have the similar specifications is the self-generated anchor. We usually use the knowledge we have as the basis to start the negotiation because we prefer to rely on our information and the experience we know, which are the self-generated anchors. And we argue that these self-generated anchors represent the attitude and thoughts toward a specific target and these might affect the value of reference point.
2.4.4 Summarization
In this section, we have known that anchoring effect often occurs in our daily life.
Any information came into our mind first when we are going to make decision can be an anchor. The more uncertain the problem, the bigger size of anchoring effect. Anchoring effect activates the anchor-consistent knowledge. It seems that anchoring effect has a great power that keep the estimate close to the anchor during the decision making process. The effective ways to eliminate this kind of bias has not been proposed right now. According to the subsection 2.4.1, the size of the anchoring effect increases with uncertainty. Conversely, we can infer that
‧ 國
立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
24
the size of the anchoring effect decreases with certainty. So, there is a chance that we may think conversely to use the bias to retrieve some information.
The subsection 2.4.1 mentioned that there are still some situations that anchor does not have an obvious influence on the final estimate. We may explain the situations that anchor effect has only a small influence on the final estimate are caused by that the individual is certain about the target of estimation s/he made or the individual has his/her particular thoughts of the problem. If so, maybe we can use the size of anchoring effect to estimate whether an individual is interesting in or insisting in something and it can help us to find the attractive features for the users. For example, if an internet service company provides a solution with downloading speed 50Mb/s and uploading speed 50 Mb/s. We can consider the network speed as the provided anchors. A customer might said if the solution has a downloading speed 1000 Mb/s and uploading speed 100 Mb/s, s/he would consider using it. We can see that the adjustment of downloading speed is bigger than the uploading speed, which means the size of anchoring effect of downloading speed is relatively smaller and s/he is more interested in downloading speed.
‧ 國
立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
25