Model selection in causal inference

因果推理中的模型选择

• 批准号: 356000-2013
• 负责人: Lefebvre, Geneviève
• 金额: $ 1.09万
• 依托单位: Université du Québec à Montréal
• 依托单位国家: 加拿大
• 项目类别: Discovery Grants Program - Individual
• 财政年份: 2015
• 资助国家: 加拿大
• 起止时间: 2015-01-01 至 2016-12-31
• 项目状态: 已结题
• 来源: https://www.nserc-crsng.gc.ca/ase-oro/Details-Detailles_eng.asp?id=572714
• 关键词: Model; selection; causal; inference

Analysts use several statistical techniques, all requiring the specification of a model, in order to determine the true effect of an exposure (e.g., a drug) on a population. In this research program we argue for a change of practice, from the choice of a single model to one where an 'average' model is used. This 'average' model combines information from the most plausible models instead of using only the 'best' model. We will study the efficacy of the model averaging approach combined with various techniques currently used to determine the real effect of an exposure on a variable of interest.

分析师使用几种需要模型规范的统计技术，以确定暴露（例如药物）对人群的真实影响。在此研究计划中，我们主张改变实践，从选择单个模型到使用“平均”模型的模型。该“平均”模型结合了来自最合理的模型的信息，而不仅仅是使用“最佳”模型。我们将研究模型平均方法的功效，并结合当前用于确定暴露对感兴趣变量的实际影响的各种技术。