Interval Estimation Based on Censored Samples from the Weibull Distribution

The focus of this paper is on interval estimation for percentiles of the Weibull and extreme value distributions based on failure censored samples. Standard large sample methods based on a t-like statistic are generally considered inadequate unless sample sizes are larger than 100. In this paper, we suggest a simple method for computing confidence limits which asymptotically give the correct confidence level. It is shown in a simulation study that these new confidence limits are accurate enough for practical purposes even for sample sizes as small as 10.