Dynamic Employer-Household Data and the Social Data Infrastructure

信息时代的成熟带来了新的挑战。面对迅速变化的经济和社会，各国政府被迫作出影响深远的经济和社会政策决定。然而，这些决定是基于有限的静态和昂贵的调查数据。与此同时，从家庭、商业机构和政府实体收集的数据中获得了大量行政数据集。这些数据可以描述工人、企业、政府和社会之间的动态互动，但在美国并没有得到充分利用，不仅是因为这些数据的独特优势还不清楚，还因为保密和访问的关键问题还没有得到充分解决。这一结果付出了巨大的代价。如果使用得当，关联数据将具有重要的科学意义，并能以多种方式增强我们的基本社会数据基础设施。它们还可用于减轻被调查者的负担，提高数据质量，并为依赖人口普查局数据进行决策的联邦、州和地方机构提供更多可用信息。此外，这些数据可以为研究界提供有价值的工具。这项研究计划将根据人口普查局的人口和经济产品，并使用链接信息，创建三个原型数据集，使数据集在家庭/个人和公司/机构两方面都是纵向的。主要调查人员、人口普查局和其他外部研究人员将使用这些数据来检验上述主张的价值。这些数据集虽然本身具有巨大的用途，但也将以另外两种方式促进知识的发展。首先，通过创建数据和开发访问模式，研究团队将明确解决一系列机密性和访问/可用性问题，利用内部人口普查局的专业知识以及Comell大学、城市研究所、马里兰大学和NCSA的外部研究人员的专业知识。其次，该项目将推进链接技术和链接数据的统计特性的知识，以便所有学科的研究人员都可以使用这些技术。该建议规定与人口普查局进行广泛合作，人口普查局实际上是该项目的主要赞助者，此外还有三个欧洲统计机构。该基础设施项目预计将涉及更多的政府和私营部门合作伙伴，特别是在保密问题得到更充分解决的情况下。将开发一套创新的访问支持工具，将完整的模拟环境（在Comell大学支持站点）与最新的基于网络的协作工具（来自伊利诺伊大学的NCSA）相结合。这些工具将与计算机视频电话会议集成到人口普查局和康奈尔大学的网站上。根据详细规定的访问规则，包括已建立的保密和披露审查程序，人口普查局将支持外部研究人员使用基于网络的支持工具和视频电话会议设施访问相关数据。这种外部研究机会将超越“最先进的技术”，其加强是提案中涉及的另一项主要活动。康奈尔大学的组成部分还包括支持一个限制访问的数据站点，该站点将存放来自其他国家国家统计机构的机密关联数据。INSEE链接的法国数据已经被批准用于该网站。瑞典统计局表示愿意完成一项协议，允许将瑞典的数据联系起来。如果得到资助，德国的DIW将在网站上发布GSOEP的链接版本。继续努力同其他国家机构谈判限制进入协定。限制访问的数据站点提供链接的数据分析工具和超级计算机设施，以使用这些数据。从这些数据中产生的新知识可能影响深远。原型美国数据集提供了解决社会和经济行为基本问题的能力。来自其他国家的限制获取数据提供了一个实验室，可以在其中检验某一特定国家发现的结果的普遍性。提案中的所有数据进展都为发现保密保护方面的技术进步提供了机会，同时使跨学科的新伙伴关系得以形成，重点是了解社会和经济系统、组织和机构。有关该提案的更多信息可在http://old-instructl.cit.cornell.edu:8000/nsf-infrastructure/上获得

The maturation of the information age has created new challenges. Governments, faced with rapidly changing economies and societies are forced to make far-reaching economic and social policy decisions. These decisions, however, are based on limited static, and expensive survey data. At the same time, large administrative data sets are available that are derived from data collected from households, business establishments and governmental entities. These data, which could describe the dynamic interactions of workers, businesses, government and society, are not fully used in the United States, not simply because the unique advantages of these data have not been clear, but also because key issues of conf dentiality and access have not been fully addressed. This outcome has come at substantial cost. Linked data, if used wisely, have a great deal of scientific importance and can enhance our basic social data infrastructure in a number of ways. They can also be used to reduce respondent burden, increase data quality, and enhance the information available to the federal, state and local agencies which rely on Census Bureau data for decision making. In addition, these data can provide a valuable tool to the research community. This research proposal will create three prototype data sets based upon the Census Bureau's demographic and economic products and using link information that permits the data sets to be longitudinal in both the household/individual and firm/establishment dimensions. The Principal Investigators, the Census Bureau, and other external researchers will use these data to examine the value of the claims made above. These data sets, while of immense use in their own right, will also advance knowledge in two additional ways. First, by creating the data and developing the access modality, the research team will explicitly address a series of confidentiality and access/availability issues, using internal Census Bureau expertise and that of external researchers at Comell University, the Urban Institute, the University of Maryland and NCSA. Second, the project will advance the knowledge of both linkage technology and the statistical properties of linked data so that researchers in all disciplines can use these techniques. The proposal delineates an extensive collaboration with the Census Bureau, which is indeed a major sponsor of the project, in addition to three European statistical agencies. The infrastructure project expects to involve additional government and private sector partners, particularly as the confidentiality issues are more fully addressed. An innovative set of access support tools that combine a complete simulated environment (at the Comell University support site) with the latest in web-based collaboration tools (from the NCSA at the University of Illinois) will be developed. These tools will be integrated with videoJcomputer teleconferencing access to the Census Bureau and Cornell sites. Under carefully specified access rules that encompass established confidentiality and disclosure review procedures, the Census Bureau will support external research access to the linked data, using the web-based support tools and the video teleconferencing facilities. This external research access will go beyond "state of the art" and its enhancement is another major activity covered in the proposal. The Cornell University component also includes the support of a restricted-access data site that will house confidential linked data from the national statistical agencies of other countries. Linked French data from INSEE have already been approved for this site. Statistics Sweden has expressed a willingness to complete an agreement that would allow linked Swedish data. The DIW in Germany will release a linked version of the GSOEP to the site, if it is funded. Efforts continue to negotiate restricted access agreements with other national agencies. The restricted access data site provides linked data analysis tools and a supercomputer facility for use with these data. The new knowledge that can be generated from these data is potentially far reaching. The prototype American data sets provide the capacity to address fundamental questions in social and economic behavior. The restricted-access data from other countries provides a laboratory in which to test the generality of results found for a particular country. All of the data advances in the proposal provide the opportunity to discover technological advances in confidentiality protection while enabling new partnerships to be formed across disciplines that focus on understanding social and economic systems, organizations and institutions. Additional information concerning this proposal is available at http://old-instructl.cit.cornell.edu:8000/nsf-infrastructure/