TY - JOUR
T1 - A comparative study of tests for the difference of two Poisson means
AU - Ng, H. K.T.
AU - Gu, K.
AU - Tang, M. L.
N1 - Funding Information:
We would like to express our sincere thanks to the editor, the associate editor and the two anonymous reviewers for their comments which greatly improved this article. The author would also like to thank Prof. W.R. Schucany for his useful suggestions. The work of the first and second authors is supported by the SMU University Research Council Grant and work of the third author is supported by Hong Kong Baptist University Grant FRG/04-05/II-01 and a grant from the Research Grant Council of the Hong Kong Special Administration Region (Project number CUHK4371/04M).
PY - 2007/3/1
Y1 - 2007/3/1
N2 - We investigate different test procedures for testing the difference of two Poisson means. Asymptotic tests, tests based on an approximate p-value method, and a likelihood ratio test are considered. Size and power performance of these tests are studied by means of Monte Carlo simulation under different settings. If one wants to control the actual significance level at or below the pre-chosen nominal level, tests based on approximate p-value method are the desirable candidates. If one allows tests whose actual significance levels may occasionally exceed the pre-chosen nominal level by an acceptable margin, asymptotic tests based on an unbiased estimate and constrained maximum likelihood estimate are reasonable alternatives. We illustrate these testing procedures with a breast cancer example.
AB - We investigate different test procedures for testing the difference of two Poisson means. Asymptotic tests, tests based on an approximate p-value method, and a likelihood ratio test are considered. Size and power performance of these tests are studied by means of Monte Carlo simulation under different settings. If one wants to control the actual significance level at or below the pre-chosen nominal level, tests based on approximate p-value method are the desirable candidates. If one allows tests whose actual significance levels may occasionally exceed the pre-chosen nominal level by an acceptable margin, asymptotic tests based on an unbiased estimate and constrained maximum likelihood estimate are reasonable alternatives. We illustrate these testing procedures with a breast cancer example.
KW - Asymptotic tests
KW - Constrained maximum likelihood estimation
KW - Level of significance
KW - Likelihood ratio test
KW - Monte Carlo simulation
KW - Power
UR - http://www.scopus.com/inward/record.url?scp=33846626507&partnerID=8YFLogxK
U2 - 10.1016/j.csda.2006.02.004
DO - 10.1016/j.csda.2006.02.004
M3 - Journal article
AN - SCOPUS:33846626507
SN - 0167-9473
VL - 51
SP - 3085
EP - 3099
JO - Computational Statistics and Data Analysis
JF - Computational Statistics and Data Analysis
IS - 6
ER -