Abstract:
The objective of this study is to compare parameter estimations from interval-censored data. The parameter estimation methods are maximum likelihood estimation (MLE) and graphical estimation (GE). Moreover, we also study bias correction method for graphical estimation. In case of fixed observed time, MLE is more efficient than GE for lognormal distribution and loglogistic distribution. GE is more efficient than MLE for weibull distribution, when the shape parameter is equal to 1 or the sample size is small with the shape parameter or . In case of fixed censoring rate, MLE is more efficient than GE for lognormal distribution, loglogistic distribution and weibull distribution, when the shape parameter . GE is more efficient than MLE for weibull distribution when the shape parameter or with small sample size. With bias correction of graphical estimation for lognormal distribution, loglogistic distribution and weibull distribution, the bias correction graphical estimation (BCGE) is more efficient than GE, especially, for small sample size, however, the efficiency decreases as sample size increases.