Shannon Harry S, Hutson Royce, Kolbe Athena, Stringer Bernadette, Haines Ted
Department of Clinical Epidemiology and Biostatistics, McMaster University, CRL-221, 1280 Main St West, Hamilton, ON, Canada, L8S 4K1.
Emerg Themes Epidemiol. 2012 Sep 11;9(1):5. doi: 10.1186/1742-7622-9-5.
Various methods have been proposed for sampling when data on the population are limited. However, these methods are often biased. We propose a new method to draw a population sample using Global Positioning Systems and aerial or satellite photographs.
We randomly sampled Global Positioning System locations in designated areas. A circle was drawn around each location with radius representing 20 m. Buildings in the circle were identified from satellite photographs; one was randomly chosen. Interviewers selected one household from the building, and interviews were conducted with eligible household members.
Participants had known selection probabilities, allowing proper estimation of parameters of interest and their variances. The approach was made possible by recent technological developments and access to satellite photographs.
当总体数据有限时，已提出了各种抽样方法。然而，这些方法往往存在偏差。我们提出一种利用全球定位系统以及航空或卫星照片来抽取总体样本的新方法。
我们在指定区域随机抽取全球定位系统位置。在每个位置周围画一个半径为20米的圆。从卫星照片中识别该圆内的建筑物；随机选择其中一座。访谈员从该建筑物中选择一户家庭，并对符合条件的家庭成员进行访谈。
参与者具有已知的选择概率，从而能够对感兴趣的参数及其方差进行恰当估计。近期的技术发展以及获取卫星照片使得该方法成为可能。
