Sampling errors of systematic and random surveys of cover-type areas
成果类型：
Article
署名作者：
Osborne, JG
刊物名称：
JOURNAL OF THE AMERICAN STATISTICAL ASSOCIATION
ISSN/ISSBN：
0162-1459
DOI：
10.2307/2279222
发表日期：
1942
页码：
256-264
关键词：
摘要：
To estimate the composition of an area by cover-type classes, sampling is usually systematic, the national forest-survey procedure being one line in 10 miles of width. By superimposing on a map 20 lines per mile, randomly placed, 3 types of sampling were tried[long dash]completely random, stratified random and a systematic type consisting of lines 1 mile apart. The estimates of the mean area in cultivated land were practically the same in the 3 types of sampling, but the stratified random surveys were only 1/2 - 1/4 as efficient as systematic surveys of the same intensity. The nature of the variation found in populations in place is such that the variate measured may be considered as a continuous function of position and the problem of sampling reduces, sensibly, to one of curve fitting. A polynomial of 6th degree was fitted to the variate, number of chains of cultivated land, and formulas developed for estimating the error of predicting the mean of another survey. If data taken in the systematic manner investigated are used with random sampling formulas, biased estimates of the sampling errors result; but from estimates of the correlation of measured and unmeasured lines dependable estimates of the sampling errors of this kind of systematic samples are obtained.