Hi everyone,
I've been away a bit. Thanks to Kit Baum, version 2.0.0 of nehurdle is now available in the Boston College repository. Type ssc install nehurdle [, replace] or search nehurdle, all at the command prompt. If you use the latter, for now make sure you get the package at the Boston College repository, and not the one published in The Stata Journal, because it is the one with the latest version. I have submitted a new article to The Stata Journal, so they may update the package, but SSC always has the latest version. To the original estimators I have added the following three:
1. Poisson truncated hurdle,
2. NB1 (negative binomial variance linear in mean) truncated hurdle, and
3. NB2 (negative binomial variance quadratic in mean) truncated hurdle.
The equations in these models can be estimated as two separate estimations, obviously, but the issue becomes with the post-estimation to do predictions of the observed variable. You will find that I have added a lot of options to use with predict, both for these models and for those that were already existing in the previous version of nehurdle that make post-estimation predictions very easy.
I hope this is useful to all of you researchers out there. Let me know if you have any questions.
I've been away a bit. Thanks to Kit Baum, version 2.0.0 of nehurdle is now available in the Boston College repository. Type ssc install nehurdle [, replace] or search nehurdle, all at the command prompt. If you use the latter, for now make sure you get the package at the Boston College repository, and not the one published in The Stata Journal, because it is the one with the latest version. I have submitted a new article to The Stata Journal, so they may update the package, but SSC always has the latest version. To the original estimators I have added the following three:
1. Poisson truncated hurdle,
2. NB1 (negative binomial variance linear in mean) truncated hurdle, and
3. NB2 (negative binomial variance quadratic in mean) truncated hurdle.
The equations in these models can be estimated as two separate estimations, obviously, but the issue becomes with the post-estimation to do predictions of the observed variable. You will find that I have added a lot of options to use with predict, both for these models and for those that were already existing in the previous version of nehurdle that make post-estimation predictions very easy.
I hope this is useful to all of you researchers out there. Let me know if you have any questions.
