PSYCH OpenIR  > 社会与工程心理学研究室
中国人非线性概率加权的特征及机制探索
其他题名Exploring the characteristics and mechanisms of nonlinear probability weighting of Chinese people
隋晓阳
导师饶俪琳 ; 李纾
2023-06
摘要概率加权(probability weighting)指个体在决策过程中会对客观概率赋予一个主观的概率加权,然后根据主观的概率加权进行决策。概率加权是风险决策的重要组成部分,影响个体的风险决策行为。前人研究发现在风险决策研究领域,概率加权函数是非线性的,个体存在高估小概率、低估中高概率的模式。范围一频次理论和抽样决策理论认为个体对概率的主观判断受到现实世界概率分布的影响,个体记忆中的概率分布反映了现实世界的概率分布。前人研究对英国国家语料库中心的英文文字概率使用频次进行分析,发现当把描述风险的文字概率从小到大进行排序后,文字概率的相对位次表现出了倒S型模式。我们课题组前期研究发现将英文文字概率翻译成中文文字概率后,两种文字概率对应的数字概率值并不匹配(许洁虹&李纾,2007)。那么,中国人与西方人是否表现出相同的概率加权模式?为解决以上问题,本论文开展了以下研究。 研究一通过对微博和北京语言大学语料库中心的文字概率使用频次进行分析探索各文字概率的主观价值,并与西方人进行比较。子研究1.1首先构建了一个包含343个文字概率的文字概率词汇表,子研究1.2 (N=162)招募被试对文字概率表达的数字概率进行评定,子研究1.3和子研究1.4分别对微博和北京语言大学语料库中心各文字概率的使用频次进行分析,并选取西方人的数据进行重新分析((Stewart et al., 2006)并进行比较。结果发现无论是在西方人还是在中国人中,复合不变性函数对文字概率主观价值(即,概率主观加权)的拟合结果是最好的。中国人的弯曲度参数/大于西方人,表明中国人的概率加权函数更接近线性;中国人的高度参数δ大于西方人,表明中国人低估事件发生概率。这提示我们中国人高估小概率的倾向可能低于西方人。 研究二通过元分析验证中西方人群在高估小概率倾向上是否存在显著差异。高估小概率倾向由被试在确定选项(safe option)和单结果风险选项(C single-outcome risky option)中选择确定选项或风险选项的比例来测量。本研究选取了在确定选项和单结果风险选项中进行二择一的风险选择任务,其中小概率单结果风险选项的概率值在0%-34%之间。通过文献检索与筛选后,共纳入24个研究的137个数据集进行元分析。结果发现发现西方人比中国人更加高估小概率。 研究一和研究二的结果支持了范围一频次理论和抽样决策理论的预测,即个体对概率的主观判断受到现实世界文字概率分布的影响。抽样决策理论还指出现实世界的文字概率分布和个体记忆中的文字概率分布是对应的。考虑到个体对文字概率进行记忆提取时会受到文字概率熟悉度的影响,因此本研究进一步探索文字概率熟悉度是否对个体的概率加权产生影响,进而对个体的风险决策产生影响。研究三采用问卷调查和行为实验的方法探讨文字概率熟悉度对以文字概率表征的二择一风险选择任务的影响。子研究3.1和子研究3 .2分别探讨获益和损失情境下文字概率熟悉度对个体风险决策的影响,接着子研究3.3探讨了是否存在“文字一数字”框架效应,以及文字概率熟悉度是否影响“文字一数字”框架效应。子研究3.1 (N=110)和子研究3.2 (N= 83)通过以文字概率表征的二择一风险决策任务,发现在获益情境下,文字概率熟悉度可以正向预测被试对风险选项的选择;在损失情境下,文字概率熟悉度可以负向预测被试对风险选项的选择。子研究3.3 (N= 83)通过比较被试在文字概率和数字概率两种框架下的选择,发现存在“文字一数字”框架效应,被试在数字概率框架下更加偏好风险选项;同时被试对文字概率越熟悉,越可能发生在文字概率框架下选择风险选项,在数字概率框架下选择确定选项的框架效应。本研究揭示了文字概率熟悉度对个体的风险决策存在影响。 综上,本研究发现中国人高估小概率的倾向低于西方人,且文字概率熟悉度对个体的风险选择存在影响。在获益情境下,个体对文字概率越熟悉越可能选择风险选项;在损失情境下,个体对文字概率越熟悉越可能选择确定选项。同时本研究还发现存在“文字一数字”框架效应,且文字概率熟悉度对“文字一数字”框架效应存在影响。研究结果为文字概率相关研究提供了新的思路,未来研究可进一步探讨文字概率决策影响机制及利用“文字一数字”框架效应进行助推研究。
其他摘要Probability weighting refers to the assumption that the value of an outcome is multiplied by a decision weight that captures the subjective impact of that outcome on choice and that can deviate from the outcome's objective probability. Probability weighting is an important component of risky decision making and affects individuals' choice. Previous studies have found that in the field of risky decision making, the probability weighting function is nonlinear, and individuals tend to overweight small probabilities and underweight moderate and large probabilities. According to range frequency theory and decision by sampling, probability's subjective value is constructed from a series of binary, ordinal comparisons to a sample of probability values drawn from memory and its rank within the sample, and the sample reflects both the immediate distribution of probability values from the current decision's context and also the background, real-world distribution of probability values. Previous research analyzed the frequency with which English verbal probability occurred in natural language (i.e., the British National Corpus (BNC) World Edition), and found that the function has an inverse S-shape. Our previous study found that after translating English verbal probability into Chinese verbal probability, the corresponding numerical probabilities were not the same. Do Chinese and Western people exhibit the same probability weighting function? To address these issues, we conducted the following studies. In Study 1,we analyzed the frequency with which Chinese verbal probability occurred in Weibo and Beijing Language and Culture University Corpus Center (BLCU Corpus Center, BCC), and compared the results with those of Western people. In Substudy 1 .1,we constructed a vocabulary of Chinese verbal probability, including 343 Chinese verbal probability. In Substudy 1.2, we recruited 162 participants to estimate the numerical probability equivalent of each verbal probability. In Substudies 1.3 and 1.4, we analyzed the frequency with which Chinese verbal probability occurred in Weibo and BCC, and reanalyzed the frequency with which English verbal probability occurred in BNC. The results showed that the compound invariance model fit the data best, either in Western or Chinese people. The curvature parameter y was larger in Chinese people than in Western people, indicating that the function in Chinese people was more linear than Western people. And the elevation parameter 8 was larger in Chinese people than in Western people, implying that Chinese people underweighted the probability of event occurrence. The results suggested that the tendency of overweighting small probabilities in Chinese people may be lower than that in Western people. In Study 2, we conducted a meta-analysis to demonstrate whether there was a significant difference in the tendency of overweighting small probabilities between Chinese and Western people. This meta-analysis focusing on studies using the binary risky choice, in which participants were asked to choose between a safe option and a single-outcome risky option. The probabilities of risky options ranged from 00/o to 340/0. The tendency of overweighting small probabilities was measured by the proportion of risky option chosen in the gain domain, and the proportion of safe option chosen in the loss domain. After literature search and screening, a total of 137 datasets in 24 studies were included in the meta-analysis. The results showed that Western people showed a greater tendency of overweighting small probabilities than Chinese people. The results of Studies 1 and 2 supported range-frequency theory and decision by sampling, which suggested people make binary, ordinal comparisons between probability and probability values were compared with a decision sample comprising a sample of values from memory. Decision by sampling also assumed that the distribution of probability in memory reflects the distribution of probability values in the world. Given that the memory retrieval of verbal probability is influenced by familiarity of verbal probability, in Study 3, we further explored whether familiarity of verbal probability affects individuals' probability weighting and risky decision making. In Substudies 3 .1 and 3.2, we investigated how familiarity of verbal probability affects individuals' risky decision making in the gain and loss domains, respectively. In Substudy 3.3, we explored whether there is a "verbal-numerical" framing effect and whether familiarity of verbal probability affects the "verbal-numerical" framing effect. In Substudy 3 .1 (N=110), we found that in the gain domain, familiarity of verbal probability was a significant predictor predicting an increased likelihood of selecting the risky option. In Substudy 3.2 (N=83), we found that in the loss domain, familiarity of verbal probability was a significant predictor predicting an decreased likelihood of selecting the risky option. In Substudy 3.3 (N=83), we compared participants' choices in verbal and numerical framing, and found a "verbal-numerical" framing effect, with participants choosing the risky option more often in the numerical framing than in the verbal framing. In addition, familiarity of verbal probability was a significant predictor predicting the "verbal-numerical" framing effect. Specifically, the more familiar participants were with the verbal probability, the more likely they were to choose the risky option in the verbal framing and to choose the safe option in the numerical framing. In summary, we found that the tendency of overweighting small probabilities in Chinese samples was lower than that in Western samples, and familiarity of verbal probability was a significant predictor predicting an increased likelihood of selecting the risky option in the gain domain, while familiarity of verbal probability was a significant predictor predicting an decreased likelihood of selecting the risky option in the loss domain. We also found a "verbal-numerical" framing effect and that familiarity of verbal probability affects the "verbal-numerical" framing effect. The results provide new insights for future studies on verbal probability, future research could explore the mechanism of risky decision making under verbal probability, and conduct the nudging studies using the "verbal-numerical" framing effect.
关键词非线性概率加权 风险决策 熟悉度 文字概率 数字概率
学位类型博士
语种中文
学位名称理学博士
学位专业应用心理
学位授予单位中国科学院大学
学位授予地点中国科学院心理研究所
文献类型学位论文
条目标识符http://ir.psych.ac.cn/handle/311026/46183
专题社会与工程心理学研究室
推荐引用方式
GB/T 7714
隋晓阳. 中国人非线性概率加权的特征及机制探索[D]. 中国科学院心理研究所. 中国科学院大学,2023.
条目包含的文件
文件名称/大小 文献类型 版本类型 开放类型 使用许可
隋晓阳-博士学位论文.pdf(5386KB)学位论文 限制开放CC BY-NC-SA请求全文
个性服务
推荐该条目
保存到收藏夹
查看访问统计
导出为Endnote文件
谷歌学术
谷歌学术中相似的文章
[隋晓阳]的文章
百度学术
百度学术中相似的文章
[隋晓阳]的文章
必应学术
必应学术中相似的文章
[隋晓阳]的文章
相关权益政策
暂无数据
收藏/分享
所有评论 (0)
暂无评论
 

除非特别说明,本系统中所有内容都受版权保护,并保留所有权利。