内容来源：OpenScience
通过分层校准解决多重比较问题
可重复性(Reproducibility)问题引起了多个研究领域的广泛关注，与之相关的统计概念也逐渐引发重视，比如假阳性(False positive)这个概念。2009年，fMRI领域著名的“死三文鱼的报告”(Bennett et al, 2009)让研究者对假阳性闻之色变。但严格地进行多重比较校正在减小假阳性的同时，也增加了假阴性，降低了研究的统计检验力。近年来，分层模型(hierarchical models，也称为层级模型、混合线性模型、混合效应模型、多层模型等)逐渐在各个领域内兴起，Neuron (Yu et al., 2022)、Annual Review of Psychology (Hoffman & Walters 2022)均开始倡导使用层级模型。本次OpenTalk中，我们非常有幸地邀请到美国精神卫生中心（NIMH）著名神经成像分析软件AFNI开发团体成员——陈刚博士，为我们介绍fMRI研究中层级模型的应用及其优势。
报告信息
题目：Addressing the issue of multiple comparisons through hierarchical calibration
报告语言：英文
分享嘉宾：Gang Chen, Scientific and Statistical Computing Core DIRP/NIMH, National Institutes of Health U.S.A.
My current research mainly focuses on hierarchical modeling using Bayesian inference to address issues such as multiplicity, regularization, nonlinearity, sample sizes and individual differences in neuroimaging, psychometrics, and behavior genetics.
摘要
In typical neuroimaging studies, the same model is applied simultaneously to hundreds of thousands of spatial units (e.g., voxels) across the brain. To ensure a reasonable error rate under the traditional statistical framework (i.e., multiple comparisons), results need to be adjusted for the number of statistical inferences carried out. Generally, this is done by means of a “cluster” methodology that leverages overall error rate against spatial extent (neighboring voxels). We suggest that the amount of information lost with this widespread method makes the cluster approach a major contributor to the reproducibility problem in the field. Specifically, we propose a hierarchical approach capable of considering shared information—not only immediately adjacent to individual spatial units, but across the whole brain. In addition, we recommend four steps researchers can take to alleviate information loss and improve reproducibility:
- Accurately model data hierarchies.
- Instead of focusing solely on statistical evidence, quantify effects.
- Abandon strict dichotomization.
- Report full results.
时间
北京时间[GMT+8] 4月09日(周六) 20:00~22:00
欧洲中部时间[CET] 4月09日(周六) 14:00~16:00
美国东部时间[EST] 4月09日(周六) 08:00~10:00
zoom信息
Meeting ID: 9139 4010 836
报告流程：报告50分钟，提问15~30分钟
主持人：胡传鹏（博士）南京师范大学心理学院
其它：本次线上报告录屏会在结束后上传于B站: COSN_live https://space.bilibili.com/252509184
组织团队(按名字首字母倒序排列)
OpenScience学术策划小组
张晗(博士), A*STAR, Singapore
张磊 (博士), University of Vienna, Austria
楊毓芳(博士), Freie Universität Berlin, Germany
杨金骉, MPI Psycholinguistics, the Netherlands
肖钦予, University of Vienna, Austria
王鑫迪(博士), 北京慧脑云
王庆(博士), Montreal Neurological Institute, Canada
金淑娴, Vrije Universiteit Amsterdam, the Netherlands
金海洋(博士), New York University Abu Dhabi, UAE
胡传鹏(博士), 南京师范大学
葛鉴桥 (博士), 北京大学
高梦宇(博士), 北京师范大学
耿海洋(博士), 香港大学
陈妍秀(博士), 中科院心理所
陈骥(博士), 浙江大学
NeuroChat团队
张文昊(UT Southwestern Medical Center, USA)
张洳源(上海交通大学)
张磊(University of Vienna, AUT)
应浩江(苏州大学)
徐婷(Child Mind Institute, USA)
王鑫迪(北京慧脑云)
滕相斌(MPI for Human Development, DEU)
鲁彬（中国科学院心理研究所）
孔祥祯(浙江大学)
胡传鹏(南京师范大学)
邸新(New Jersey Institute of Technology, USA)