Ezzati T M, Hoffman K, Judkins D R, Massey J T, Moore T F
National Center for Health Statistics, Hyattsville, MD 20782, USA.
Stat Med. 1995;14(5-7):571-83. doi: 10.1002/sim.4780140515.
Multiple data sources are sometimes available as potential sampling frames for population surveys, and in some situations the use of a multiple frame sample design is more advantageous than using a single sampling frame. The use of multiple sampling frames, however, has variance and bias implications, as well as sampling, data collection, and logistical considerations. These issues are addressed for a proposed dual frame sampling approach in the National Health Interview Survey (NHIS). The results of an investigation of the sampling efficiencies and operational issues in supplementing the NHIS area frame sample with a sample of elderly African and Hispanic Americans and persons with disabilities selected from Social Security Administration files are presented.
在人口调查中，有时会有多个数据源可作为潜在的抽样框，并且在某些情况下，使用多框抽样设计比使用单一抽样框更具优势。然而，使用多个抽样框会涉及方差、偏差问题，以及抽样、数据收集和后勤方面的考虑。针对美国国家健康访谈调查（NHIS）中拟议的双框抽样方法，对这些问题进行了探讨。本文展示了一项调查结果，该调查研究了用从社会保障管理局档案中选取的老年非裔和西班牙裔美国人以及残疾人样本补充NHIS区域框样本时的抽样效率和操作问题。
