How to estimate the goodness of fit of a sample containing censored values?
There is some older work on the matter (here), but i'd like to know if there is anything more modern.
I think a reasonable approach would be, to perform a monte carlo and generate samples from the target distribution and then censor them in the same way as my data. The test statistic can then be compared to the randomly generated distribution under the null.
This is the approach used for the KS test when the parameters of the distribution are estimated from the data (lilliefors 1969). Since the parameters of my fitted distribution were not estimated from the data, would this affect my power?
