Far East Journal of Theoretical Statistics
Volume 13, Issue 1, Pages 81 - 100
(May 2004)
|
|
STATISTICAL INFERENCE ON MIXTURE OF TWO BINOMIAL DISTRIBUTIONS WITH APPLICATIONS IN GENETIC LINKAGE
M. M. Shoukri (Canada) and M. H. Asyali (Saudi Arabia)
|
Abstract: The finite mixture of distributions has many applications in medicine, epidemiology, image segmentation, human and animal genetics, and many other fields. The massive development in the statistical techniques to investigate their properties is mainly due to the facts that these models constitute an effective tool to model unobserved population heterogeneity. This paper investigates the asymptotic properties of statistical tests applied to the mixture of two binomial distributions with a special parametric structure. Such tests are frequently used to detect genetic linkage under heterogeneity. We investigate the efficiency of the method of moments relative to the method of maximum likelihood, and through Monte-Carlo simulations we evaluate the asymptotic distribution of the likelihood ratio test to detect genetic linkage under population heterogeneity. |
Keywords and phrases: mixture distributions, likelihood inference, EM algorithm, efficiency contours, Monte-Carlo simulations. |
|
Number of Downloads: 244 | Number of Views: 767 |
|