SILPAKORN UNIVERSITY SCIENCE AND TECHNOLOGY JOURNAL, Vol 4, No 2 (2010)

A Simulation Comparison of New Confidence Intervals for the Coefficient of Variation of a Poisson Distribution

Wararit Panichkitkosolkul

Department of Mathematics and Statistics, Faculty of Science and Technology, Thammasat University, Phathumthani


Abstract


This paper proposes four new confidence intervals for the coefficient of variation of a Poisson distribution based on obtaining confidence intervals for the Poisson mean. The following confidence intervals are considered: confidence intervals for the coefficient of variation of a Poisson distribution based on Wald (W), Wald with continuity correction (WCC), Scores (S) and Variance stabilizing (VS) confidence interval. Using Monte Carlo simulations, the coverage probabilities and lengths of these confidence intervals are compared. Simulation results have shown that the confidence interval based on WCC has desired closeness coverage probabilities of 0.95 and 0.90. Additionally, the lengths of newly proposed confidence intervals are slightly different. Therefore, the confidence interval based on WCC is more suitable than the other three confidence intervals in terms of the coverage probability.

Key Words: Coefficient of variation; Confidence interval; Coverage probability; Length; Poisson distribution

Full Text: PDF