Bookmark and Share

Notice: On April 23, 2014, Statalist moved from an email list to a forum, based at statalist.org.


[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

st: Zero truncation in Multilevel regression


From   [email protected]
To   [email protected]
Subject   st: Zero truncation in Multilevel regression
Date   Tue, 22 Mar 2011 10:28:56 +0100

Dear Statalist members,
I'm implementing a Multilevel regression analysis, using the "gllamm" command (since levels are 3: individual, district, municipality). The outcome is a count variable, whose I consider only positive values. I'd like to replicate, in the Multilevel analysis, the same approach I can adopt in the one-level regression, in particular with respect to overdispersion and zero-truncation.
As for the former issue, I can't use a Negative Binomial distribution (since it is not among the ones performed by gllamm), but I can include a Random effect at the individual level (obviously, choosing the Poisson distribution).
On the contrary, with respect to zero-truncation, I don't know what to do. As regards censoring and truncation, I see it is implemented, in Stata, in the case of Gaussian distribution, Survival analysis and Panel data. With zero-truncated count variables, in the one-level case, I can use the command "ztp" for Poisson regressions (in case of the Negative Binomial, "ztnb"). But, in the case of Multilevel regression, I didn't find any way to "inform" Stata that my dependent variable is zero-truncated.
Since my outcome assumes values ranging from 1 to 586 and it is highly asymmetric (its median value is 2) I can't use either the Multinomial distribution, nor the Gaussian one.
Do you have any suggestions on how to solve this issue?
Thank you,
Federico
*
*   For searches and help try:
*   http://www.stata.com/help.cgi?search
*   http://www.stata.com/support/statalist/faq
*   http://www.ats.ucla.edu/stat/stata/


© Copyright 1996–2018 StataCorp LLC   |   Terms of use   |   Privacy   |   Contact us   |   Site index