Estimating Bias in Similarity Measures for the Pareto Populations

Quantitative measures, known as similarity measures are used by animal ecologists for comparative studies of diet, habitat preference, and seasonal patterns of abundance and by economists to compare income distributions of different segments of the population. The complements of similarity measures,...

Full description

Saved in:
Bibliographic Details
Published inJournal of combinatorics, information & system sciences Vol. 36; no. 1-4; p. 103
Main Authors Mulekar, Madhuri S, Fukasawa, Takeshi
Format Journal Article
LanguageEnglish
Published New Delhi Prints Publications Pvt. Ltd 01.01.2011
Subjects
Combinatorics
Estimation bias
Income inequality
Monte Carlo simulation
Pareto optimum
Online AccessGet full text

Cover

More Information
Summary:Quantitative measures, known as similarity measures are used by animal ecologists for comparative studies of diet, habitat preference, and seasonal patterns of abundance and by economists to compare income distributions of different segments of the population. The complements of similarity measures, known as measures of heterogeneity are also used to study disparities in populations. Here Matusita's and MacArthur-Levin's measures for similarity are developed for Pareto distributions which are commonly used to describe income distributions. Approximations for bias of estimates are derived and the behavior of bias and MSE is studied using Monte Carlo technique. Results show that for sample sizes used commonly in practice, the estimates are nearly unbiased, but for smaller sample sizes, negatively biased estimates are quite common. [PUBLICATION ABSTRACT]
ISSN:0250-9628
0976-3473