Qualitative and quantitative assessment of geographic clustering of population samples selected using different methods of random digit dialing.
Qualitative and quantitative assessment of geographic clustering of population samples selected using different methods of random digit dialing.
复制标题
对使用不同随机数字拨号方法选择的人口样本的地理聚类进行定性和定量评估。
DOI：
10.1093/oxfordjournals.aje.a115777
复制
发表时间：
1990
影响因子：
5
通讯作者：
Heuser,L
中科院分区：
文献类型：
--
作者：
Davis,S;Ross,A;Voigt,LF;Heuser,L
随机数字拨号法是流行病学研究中常用的随机抽样方法。虽然随机数字拨号通常被认为是提供代表性的样本，覆盖率和无应答错误可能会影响样本的代表性程度。本调查是为了确定使用基本随机数字拨号技术的变化选择的样本的地理分布是否准确地反映了潜在的人口分布，如果不是，这些样本是否倾向于或多或少比他们从中选择的人口分散。利用1983-1986年进行的三项病例对照研究的对照组数据。在华盛顿州西北部的三个县的地理区域内，对998名对照的居住地址进行了定位，并分配了X-Y坐标和人口普查区名称。最初，对照的空间分布进行了检查，以图形的相关的年龄-性别结构的基础人口。没有明显的差异，地理格局进行了更正式的统计评价的基础上，利用两种措施来描述一组点的空间分布的质心技术。结果表明，所选择的样本既不多也不少分散比基本人群。然而，使用原始号码选择的样本的地理中心往往偏离其各自种群的中心。几种可能的解释，这种转变被认为是，并建议在人口抽样的进一步评价和调查疾病的时空聚集的分析方法的扩展。
Random digit dialing is a method commonly used to select random population samples in epidemiologic research. Although random digit dialing is generally presumed to provide representative samples, coverage and nonresponse errors may affect the degree to which the sample is representative. The present investigation was undertaken to determine whether the geographic distributions of samples selected using variations of the basic random digit dialing technique accurately reflect the underlying population distribution, and if not, whether such samples tend to be either more or less dispersed than the populations from which they were selected. Data regarding control groups from three case-control studies conducted from 1983–1986 were utilized. The residence addresses of 998 controls were located and assigned anX-Ycoordinate and census tract designation within a three-county geographic area in northwest Washington State. Initially, the spatial distributions of controls were examined graphically in relation to the age-sex structure of the underlying population. No differences in geographic pattern were apparent A more formal statistical evaluation was conducted based on centrographic techniques utilizing two measures to describe the spatial distribution of a set of points. Results indicate that the samples chosen were neither more nor less dispersed than the underlying populations. However, the geographic centers of samples selected using primary numbers tended to be shifted from the centers of their respective populations. Several possible explanations for such shifts are considered, and extensions of the analytical approach are suggested in relation to the further evaluation of population sampling and the investigation of space-time aggregations of disease.