Discriminating Between the Logistic and the Normal Distributions based on Likelihood Ratio

by Franšois Aucoin and Fahim Ashkar.

Abstract: The problem of discriminating between the normal and the logistic distributions with unknown parameters and based on the likelihood ratio statistic is considered. The main interest is in determining the probability of correct selection (PCS) between these two distributions. Asymptotic approximations to these PCSĺs are derived and then assessed using Monte Carlo (MC) simulations. The MC simulations suggest the asymptotic approximations to be unreliable for sample sizes below 100. Results also reveal a clear difficulty in discriminating between the two distributions for sample sizes smaller than 50.

Key Words: Likelihood Ratio, Discrimination, Normal and Logistic Distributions, Asymptotic Approximation, Monte Carlo Simulations

Authors:
Franšois Aucoin, frank.aucoin@gmail.com
Fahim Ashkar, ashkarf@umoncton.ca

Editor: Subrata Chakraborty, subrata_arya@yahoo.co.in

READING THE ARTICLE: You can read the article in portable document (.pdf) format (188421 bytes.)

NOTE: The content of this article is the intellectual property of the authors, who retains all rights to future publication.

This page has been accessed 1933 times since AUGUST 17, 2010.


Return to the InterStat Home Page.