Interval estimates and their precision

3rd International Conference on Science & Engineering in Mathematics, Chemistry and Physics 2015

数学;化学;物理学

Marek, Lubo^1 ; Vrabec, Michal^1

University of Economics, Faculty of Informatics and Statistics, W. Churchill Sq. 4, Prague 3

CZ - 130 67, Czech Republic^1

关键词: Binomial approximation;    Confidence interval;    Interval estimate;    Random sample;    Relative frequencies;    Sampling without replacements;    Specific values;    Theoretical methods;   

Others  :  https://iopscience.iop.org/article/10.1088/1742-6596/622/1/012038/pdf
DOI  :  10.1088/1742-6596/622/1/012038

来源: IOP

【 摘 要 】

A task very often met in in practice is computation of confidence interval bounds for the relative frequency within sampling without replacement. A typical situation includes preelection estimates and similar tasks. In other words, we build the confidence interval for the parameter value M in the parent population of size N on the basis of a random sample of size n. There are many ways to build this interval. We can use a normal or binomial approximation. More accurate values can be looked up in tables. We consider one more method, based on MS Excel calculations. In our paper we compare these different methods for specific values of M and we discuss when the considered methods are suitable. The aim of the article is not a publication of new theoretical methods. This article aims to show that there is a very simple way how to compute the confidence interval bounds without approximations, without tables and without other software costs.