EITM: Developing the Tools to Understand Human Performance: An Empirical Infrastructure to Foster Research Collaboration

从雇主和雇员的角度理解工作场所对于理解人的表现至关重要。只有将雇主和雇员的微观数据整合起来，纵向联系起来，并使研究界能够访问，才能理解工作场所。开发集成数据的数据基础设施是一项艰巨的任务，因为传统的数据开发方法是分别收集家庭和企业的数据。幸运的是，这些单独收集的数据可以通过美国联邦统计系统中提供的包含雇主和雇员信息的丰富的管理数据集进行集成。为这些数据开发一个访问系统也是一项艰巨的任务，因为有关企业和家庭的基础数据受到法律保密限制的保护。在联邦统计系统内，可以建立综合的微观数据，挑战在于如何使用户社区能够为经批准的统计目的访问这些数据，同时保护数据的机密性。对这些数据的现有访问是通过NSF/人口普查研究数据中心网络。虽然这一系统非常成功，但有一些限制，因此，相对于联邦统计系统中微观数据的潜在使用，目前的使用非常有限。该项目概述了一个多层访问结构，该结构建立在最近的数据基础设施发展和当前存在的访问模式之上。这种多层访问结构的关键组成部分是开发推理有效的公共使用合成微数据，在虚拟研究数据中心访问更丰富的合成微数据，进而限制对人口普查/国家科学基金会研究数据中心网络中黄金标准微数据的访问。开发推理有效的综合数据是统计理论和应用前沿的一项重大工作。多层访问系统的开发是处理必须面对的保密问题的前沿。企业和家庭的微观数据（特别是综合数据）对社会科学具有基础性的重要性，必须向研究界开放，但也必须保护这些数据的机密性。这笔拨款支持一个人口普查数据产品的原型合成数据系统- LEHD基础设施文件（个人，雇主，工作），以测试构建合成数据的可行性和实用性。拟议活动的更广泛影响拟议的活动有可能大大增加社会科学研究界对微观数据的访问。这种获取途径的增加将产生广泛的影响，但在本项目下制定的方法和协议将对所有科学学科产生更广泛的影响。要解决社会科学、卫生科学和环境科学中的一系列广泛问题，就需要关于家庭和企业的丰富综合微观数据。开发如此丰富的数据、推理有效的合成数据和多层访问系统是科学界许多不同部门面临的问题。来自各个学科的许多社会科学家将访问本提案中开发的数据系统

Understanding of the workplace from the perspective of both employers and employees is vital for understanding human performance. Understanding the workplace can only occur if micro data on employers and employees are integrated, linked longitudinally and made accessible to the research community. Developing the data infrastructure for integrated data is a monumental task as the traditional approach towards data development is to collect data on households and businesses separately. Fortunately, such data collected separately can be integrated via the rich administrative data sets that contain information on both employers and employees that are available in the U.S. federal statistical system. Developing an access system for such data is also a monumental task because the underlying data on businesses and households are protected by legal confidentiality restrictions. Within the federal statistical system, integrated micro data can be created and the challenge is to make such data accessible to the user community for approved statistical purposes while protecting the confidentiality of the data. Existing access to such data is via an NSF/Census Research Data Center network. While this system has been very successful, there are a number of limitations so that, relative to the potential use of the micro data in the federal statistical system, the current use is very limited. This project outlines a multi-layered access structure that builds on recent data infrastructure developments and the access modalities as they currently exist. Key components of this multi-layered access structure are the development of inference-valid public use synthetic micro data, access to richer synthetic micro data at a virtual Research Data Center, and in turn limited access to the gold standard micro data in the Census/NSF Research Data Center network. The development of inference-valid synthetic data is a major undertaking at the frontier of statistical theory and applications. The development of the multi-layered access system is at the frontier of dealing with the confidentiality protection issues that must be confronted. The micro data on businesses and households (and especially the integrated data) are of fundamental importance for the social sciences and must be accessible to the research community but the confidentiality of these data must also be protected. This grant supports a prototype synthetic data system for one Census data product - the LEHD infrastructure files (individual, employer, job) to test the feasibility and usefulness of constructing synthetic data. Broader Impacts of the Proposed ActivityThe proposed activity has the potential for dramatically increasing access to micro data for the social science research community. This increased access will have broad impacts but even broader impacts arise for all scientific disciplines from the methodologies and protocols developed under this project. Rich integrated micro data on households and businesses are required to address a wide range of issues in the social sciences, health sciences, and environmental sciences. Developing such rich data, inference-valid synthetic data, and a multi-layered access system are issues confronting many different parts of the scientific community. Many social scientists from a wide range of disciplineswill access the data system developed in this proposal