[数量经济学研讨会]Spike and Slab Prior Distributions for Simultaneous Bayesian Hypothesis Testing, Model Selection, and Prediction, of Nonlinear Outcomes
发文时间:2015-04-23
数量经济学研讨会 [201503]

报告题目:Spike and Slab Prior Distributions for Simultaneous Bayesian Hypothesis Testing, Model Selection, and Prediction, of Nonlinear Outcomes
报告人:庞珣 教授
报告时间:2015年4月24日下午12:30-13:30
报告地点:明德主楼515B
内容简介:Most readers of empirical political science are not really interested in the low-level mechanics of building regression-style models, counting significance stars on tables, or the routine details on fit and robustness. Consumers of quantitative analysis in the discipline generally want to know from authors is what are the reliable determinants of some outcome of interest so that the discussion can return to the political science questions at hand. Over time, however, we have done readers a great disservice by increasingly focusing on the low-level mechanical aspects of model specification. This includes hiding the full model development process, a failed hypothesis testing paradigm, and fixation with artificial significance thresholds. In this paper we present an alternative modeling paradigm that simultaneously performs hypothesis testing, model comparison and variable selection, as well as a principled method for posterior prediction using Bayesian model averaging. We extend the linear spike and slab model to qualitative responses and design a hierarchy of forms over both the parameter and model spaces to achieve these goals. The performance of the models and methods are assessed with an empirical application in Political Science. Our strategy increases the information provided to readers in an efficient and overt manner. 


报告人简介:庞珣,清华大学国际关系学系教授、博士生导师,清华-卡内基全球政策中心中方主任,中国民主同盟中央经济委员会委员。美国华盛顿大学(圣路易斯)政治学博士,北京大学国际关系学院法学学士和硕士、北京大学经济研究中心经济学双学士。2010年取得政治学博士学位后,加入美国普林斯顿大学政治系,任tenure-track助理教授。2012年入职清华大学任国际关系学副教授,2014年晋升为教授。研究和教学兴趣主要集中在国际/比较政治经济学和计算社会科学,尤其是贝叶斯统计方法和马科夫链-蒙特卡洛模拟方法,发展复杂统计模型以分析政治学动态过程和政治网络互动。研究成果发表于政治学和国际关系学国际顶级期刊《政治分析》(Political Analysis)和《国际组织》(International Organization)等,并获得美国政治学会政治方法研究领域的多项奖励。


数量经济教研室 运筹学与数量经济研究所 中国人民大学经济学院 2015年04月

为了加强与国内外高水平数量经济学学者的交流和合作,更好地促进数量经济学与院内其他学科之间的合作交流,数量经济学教研室将举办数量经济学研讨会。研讨会侧重于理论计量经济学、应用计量经济学及数理经济学的最新研究成果,欢迎各位学者参加或报告。如果您有研究成果想要报告,请与数量经济学教研室章永辉博士(yonghui.zhang@hotmail.com)联系。