Document Actions
Boehning et al. (2004). Comparison of Estimators
Boehning, D., Sarol, J., Tattanasiri, S., Viwatwongkasem, C., and Biggeri, A. (2004). Comparison of non-iterative and iterative estimators of heterogeneity variance for the standardized mortality ratio. Biostatistics, 5(1):61–74.
This article deals with population heterogeneity where the parameter of interest varies in the population. The authors study the estimation of the heterogeneity variance and provide alternative estimators. Estimators are illustrated using data from hepatitis B cases in 23 regions of Berlin for 1995 as well as data on perinatal mortality in the Northwest Thames health region.
Four non-iterative estimators (Boehning, 2000; Marshal, 1991; Bautista, 1997; DerSimonian and Laird, 1986) and two iterative estimators (method of moments and pseudo-maximum likelihood) are applied to the hepatitis B data of Berlin. Furthermore, the behaviour of the estimators is analyzed using simulations using a mixture of two Poisson distributions to obtain the standardized moratility ratios. Among the iterative methods the pseudo-maximum likelihood methods are preferable. Among the non-iterative methods the Marshall (1991) estimator performs best. From the simulation study, the authors conclude that there is not a big loss in efficiency when using non-iterative compared to iterative methods.
Last updated May 1, 2006