# Accession Number:

## AD0614792

# Title:

## A TWO-STEP SAMPLE SIZE PROBLEM.

# Descriptive Note:

## Doctoral thesis,

# Corporate Author:

## COLORADO STATE UNIV FORT COLLINS

# Personal Author(s):

# Report Date:

## 1965-03-01

# Pagination or Media Count:

## 104.0

# Abstract:

A problem in sampling is given in which an estimator of the mean of a normal population is determined such that the estimator deviates from the true population parameter by less than a given percentage of the true parameter with at least a certain specified probability. The problem is solved in the multi-step framework in which a minimum total sample size is to be attained. Certain assumptions on the existence of an upper bound on the coefficient of variation or a known positive lower bound on the mean are made. One, two and three step procedures are developed, of which the two-step process emerges as the most practical. A feasibility study is conducted using a high speed computer to determine, for a given confidence level, the form of an expression for the sample size as a function of the sufficient statistics mean and standard deviation of the first sample.