Sequential Quantal Analysis Applications to Simulations

Date of Award

3-1991

Document Type

Thesis

Degree Name

Master of Science

Department

Department of Mathematics and Statistics

Abstract

Simulations on computers are being used more often to model the workings of various systems. These programs can be very costly and time consuming to run. Often multiple runs are required to obtain the necessary amount of information to do analysis. When the outcome of a simulation run is a quantal or binary response, then techniques originally created for medical drug testing can be applied to create a model for the response. In the medical field, it is often desirable to limit the number of test subjects required to determine the toxicity level of a drug and sequential quantal analysis is a technique used to limit this number and still obtain a reasonably good fit for a subjects' response model. A question arose as to whether sequential quantal analysis could be applied to computer simulations in a similar manner to limit the number of computer runs required. This thesis attempts to discover an answer to that question using a simple simulation model with a quantal response, two sequential methods of obtaining data, and four models. The two sequential methods are the Robbins Monro Method and the Up and Down Method. The four models include the probit, logit, pareto, and linear model approximations. All models were fitted and compared to similar models obtained using a standard fixed sample size data set.

AFIT Designator

AFIT-GOR-ENC-91M-2

DTIC Accession Number

ADA239066

Comments

There is no digital version of this thesis available at this time.

This document is currently not available here.

Share

COinS