Analysis of type II censoring data related to the weibull distribution

The main purpose of this thesis is the consideration of regression models for type II censoring time to failure data in which the under lying distribution is Weibull and where the effect of the explanatory variables on the scale parameter of the distribution is multiplicative. The moment properties of maximum likelihood method (ML), weighted least squares (WLS) and modified weighted least squares (MWLS) estimates of the regression coefficients with approximations to biases of such estimators are examined theoretically and by simulation. Moreover, properties of confidence interval and testing procedures for the regres3ion coefficients and goodness of fit tests for the exponential regression model are examined .