Abstract: Anchoring effect was refer to the phenomena of estimation bias that the judgment and decision making result or target value approached excessively to the initial information or value. It was a heuristic bias found in exploring decision making behavior and process in the descriptive paradigm of decision making research. The study of anchoring effects was of great significant to understand the negative influence of the heuristic strategy in decision making and seeking effective ways to reduce the decision making bias. The present study examined how anchor value affects the decision making process and target value in external-anchor decision making tasks with eye tracker. Simple random design was used in the experiment with material of general knowledge backed by positive and negative information. The independent variable was anchor value with high and low levels. Dependent variables were judgment ratio of anchor value comparing questions, index of anchoring effect, frequencies of choices and eye movement index (IA_FSA_COUNT and IA_FSA_DURATION etc.). Participants were 64 university students, with 31 males and 33 females. Subject’s task was making anchor question judgments after reading the background information, and then estimating the target value. The background information and the decision making tasks were presented in four steps overlaying. The statistics indicated that Anchoring Effect Index was above 0.76, and there were no differences between anchor groups. IA-FSA-COUNT and IA-FSA-DURATION measures indicated: (1) At the stage of answering anchor comparing questions, IA-FSA counts from “anchor value IA” to “negative information IA” were more than that to “positive information IA”, and IA-FSA duration from “anchor value IA” to “negative information IA” was longer than that to “positive information IA”. (2) At the stage of answering target value estimate questions, IA-FSA counts from “estimate question IA” to “anchor value IA” were more than that to “positive and negative information IA”, and IA-FSA duration from “estimate question IA” to “anchor value IA” was longer than that to “positive and negative information IA”. The results suggested that external-anchor of dual background information aroused high level anchoring effect, and there were no differences between anchor value types. Participants paid more attention to the negative information. The negative information weighed more heavily in understanding the context. Anchoring effect was aroused at the stage of answering anchor comparing questions. The anchoring effect was generated by the primary processing to the anchor value. The results supported the explaining of selective-accessibility mechanism and the negativity bias in information processing.

Key words: external-anchor question, dual information, anchor value, eye-movement characteristics

摘要： 采用正反双向背景信息日常生活外部锚问题，选取64名大学生，运用眼动技术，考察锚值及双向信息对外部锚问题决策的影响。结果发现：高、低锚组锚定效应指数均在0.76以上，且不存在显著差异。回答锚值比较问题时，从锚值向反向信息的眼跳次数和注视时间显著多于向正向信息的眼跳次数和注视时间。回答估计值问题时，从估计值问题向锚值的眼跳次数和注视时间显著多于向双向信息的眼跳次数和注视时间。研究表明，外部锚引起程度较高的锚定效应。反向信息在问题理解及决策中得到更多关注。锚值在回答锚值比较及估计值问题阶段均对决策产生显著影响。支持了选择通达和信息加工负性偏向观点。

关键词: 外部锚问题, 双向信息, 锚值, 眼动特征