LIFE TESTING
成果类型：
Article
署名作者：
EPSTEIN, B; SOBEL, M
刊物名称：
JOURNAL OF THE AMERICAN STATISTICAL ASSOCIATION
ISSN/ISSBN：
0162-1459
DOI：
10.2307/2281004
发表日期：
1953
页码：
486-502
关键词：
摘要：
In biological work life testing may refer to expts. in which a group of animals is exposed to some stimulus (irradiation, drugs, etc.) and the time required for each animal to attain a certain response (death, a physiological threshold, etc.) is recorded. The authors discuss methods for estimating mean response times in which response times for only r out of n animals in a group are required. They show that for certain distributions, estimates based on r out of n response times are just as accurate, with respect to both Type I and II errors, as estimates based on r out of r animals. The principal advantage of the former method is a considerable saving in time, although (n - r) more animals must be employed. If n = 20 and r = 10, the 1st method requires only 23%, on the average, of the time required for the 2d method.