Bayesian Communication in the Social Sciences

社会科学中的贝叶斯传播

• 批准号: 9600040
• 负责人: John Geweke
• 金额: $ 15.42万
• 依托单位: University of Minnesota-Twin Cities
• 依托单位国家: 美国
• 项目类别: Standard Grant
• 财政年份: 1996
• 资助国家: 美国
• 起止时间: 1996-07-15 至 1999-06-30
• 项目状态: 已结题
• 来源: https://www.nsf.gov/awardsearch/showAward?AWD_ID=9600040&HistoricalAwards=false
• 关键词: Bayesian; Communication; Social; Sciences

This project will develop and implement procedures that will enhance communication between investigators who report results and their audience (i.e., other investigators, decision makers, and students). These user-friendly procedures will enable clients to alter investigators' priors, examine alternative posterior moments, and combine models. Software will be developed with standard file structures to facilitate control between it and user software. This software will enable clients to evaluate the accuracy of reweighted output, plot corresponding posterior densities, and average across models to obtain posterior moments unconditional on models. It is expected that this project will have two significant impacts in the social sciences. First, it will place within the reach of all quantitative social scientists many of the inherent advantages of Bayesian analysis by removing the unreasonable computational demands that have previously been required. Second, it will enable clients to rework investigators' analyses by modifying their priors, examining alternative posterior moments, and updating samples, in less time than is required to read a typical research report. Together these impacts should greatly encourage analysis of the robustness of policy recommendations with respect to assumptions and accelerate the application of new research to policy problems.

本项目将制定和实施程序，加强报告结果的调查人员与其受众（即，其他研究者、决策者和学生）。 这些用户友好的程序将使客户能够改变调查人员的先验，检查替代后验矩，并结合联合收割机模型。 将使用标准文件结构开发软件，以便于在软件和用户软件之间进行控制。 该软件将使客户能够评估重新加权输出的准确性，绘制相应的后验密度，并对模型进行平均，以获得模型上的无条件后验矩。 预计该项目将在社会科学领域产生两个重大影响。 首先，它将把贝叶斯分析的许多固有优势放在所有定量社会科学家的能力范围内，消除了以前所要求的不合理的计算要求。 其次，它将使客户能够修改他们的先验知识，检查替代的后验矩，并更新样本，从而在比阅读典型研究报告所需的时间更短的时间内修改研究人员的分析。 这些影响加在一起将极大地鼓励对政策建议相对于假设的稳健性进行分析，并加速新研究在政策问题上的应用。