SNARC效应量大小及置信区间的可信问题的实例分析  

Case Study on SNARC Effect Size and the Credibility About Estimation of Confidence Interval

作　　者：何华[1] He Hua(School of Education,Soochow University,Suzhou 215123)

机构地区：[1]苏州大学教育学院,苏州215123

出　　处：《心理学探新》2023年第1期77-83,共7页Psychological Exploration

基　　金：江苏省教育学会“十三五”教育科研规划重点课题(19B1N2SZ18)。

摘　　要：通过实例分析SNARC效应量和置信区间可信度问题。(1)效应量是心理学实验结果报告中一个非常重要的部分,依据SNARC效应量的特殊性提出一种新的统计方法(在线性回归模型中引入混合虚拟变量)处理Aleottia等(2020)的开放实验数据,尽管该方法比一般方法更复杂。(2)参数的区间估计是一种基本统计推断形式。根据枢轴量分布,置信区间在一定置信度下可估计总体参数所在的可能范围。文章通过构建一个特殊实例分析了置信区间的估计过程,并和假设检验、贝叶斯统计进行对比分析,结果显示,虽然置信区间可以用来估计参数,但是存在依据某置信区间无法作出正确估计的情况,通过贝叶斯统计能分析出其中原因。Two problems(SNARC effect size and confidence interval)are analyzed by some cases.(1)Effect size is a very important part in the report of the psychological experiment results,the research inquires into the statistical methods of comparing the size on SNARC effect in the field of number cognition,which is different from the common definitions and methods on effect size.The research raises a new statistical method(Mixed dummy variable being introduced into linear regression model)to process the open data in Aleottia et.al.(2020)study,although more complicated than general methods.(2)Interval estimation of parameters is a basic form of statistical inference.According to the distribution of the pivot quantity,confidence intervals can estimate the possible range of the population parameter undera certain confidence.The research analyzes the estimation process of a certain confidence interval by constructing a special example,and makes a comparative analysis with Null Hypothesis Significance Testing and Bayesian statistics.The results show that although the confidence interval is reliable in population,there is a situation that the accurate judgment cannot be made according to a certain confidence interval,and the reason can be analyzed through Bayesian statistics.

关 键 词：SNARC效应 虚拟变量 假设检验 置信区间 先验分布 后验分布 贝叶斯统计 

分 类 号：B841.2[哲学宗教—基础心理学]