Open tedescolor opened 3 years ago
The confidence intervals from kmqte is from the the classical nonparametric bootstrap using the percentile method. It is not uniform. Also, the paper "Nonparametric Tests for Conditional Treatment Effects with Duration Outcomes" does not rely on this function and on this type of bootstrap. There, I used the multiplier bootstrap.
You can take a look at the zcdte, zcate, and the other z... functions. There, I used sup-t based confidence bands.
Hope this helps.
Thank you very much Pedro, I would like to use your paper as comparison. So, in order to recreate the table I will try to have a look to the z-functions. Sincerely thanks for your kindness, hoping I will not need any other help.
Best regards Lorenzo
Good morning, I would like to know if, in the function kmqte, it is used the Kolmogorov-Smirnov(KS) statistics or the Cramer-von Mises (CvM) for the confidence intervals. I am trying to reproduce the simulation table of the paper "Nonparametric Tests for Conditional Treatment Effects with Duration Outcomes" and I am not really sure how to obtain the value for the two statistics.
Really thanks, Best regards