No abstract

SummaryIn the analysis of survival data wit& parametric models, i t is well known that the Weibull model is not suitable for modeling where the hazard rate ie non-monotonic. For moh c a m , loglogistic model is frequently used. However, due to the symmetric property of the log-logistic model, i t may be poor for the caaea where the hazard rate is ekewed or heavily tailed. In this paper, we suggest a generalization of the log-bghtic model by introducing 8 shape parameter. This generalized model is then applied to fit the lung cancer d a b of R~ITTICIE (1973). The resnlta seem to improve over those obtained by ueing the log-logistic model.

Various models have been developed for modeling the distribution of chromosome aberrations in the literature (e. g. CONSUL, 1989). Generalized Poisson distribution is among the popular ones. The parameters of this distribution provide meaningful interpretation of the induction mechanism of the chromosome aberrations. In this article, we apply several estimation methods to estimate the generalized Poisson parameters for fitting the number of chromosome aberrations under different dosages of radiations. The methods compared are moment, maximum likelihood, minimum chi-square, weighted discrepancy and empirically weighted rates of change. Our study suggests that the empirically weighted rates of change method results in smallest Mean Square Error and Mean Absolute Error for most dosages of radiations. The data used for this comparison are from JANARDAN and SCHAEFFER (1977).

No abstract

No abstract