Statalist


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

st: RE: SV: Comparing ZINB Models with clustering by subject


From   "Anderson, Bradley" <[email protected]>
To   "[email protected]" <[email protected]>
Subject   st: RE: SV: Comparing ZINB Models with clustering by subject
Date   Mon, 26 Jan 2009 16:11:49 -0500

Responses to my original question didn't offer complete resolution, however, they did steer me in the right direction.  The solution I'm using is based on a scaled difference in chi-square test.  The source is:

Satorra Albert., & Peter. M. Bentler(1999)"A Scaled Difference Chi-square Test Statistic for Moment Structure Analysis." Psychometrika, 66, pp. 507-514.  It is also available is availabe on Dr. Satorra's web site.

After doing a bit of digging I found that the scalled difference in chi-square test could easily be calculated using Mplus output for the ZINB model.

Thanks,

Brad

-----Original Message-----
From: [email protected] [mailto:[email protected]] On Behalf Of Mads Meier Jæger
Sent: Monday, January 26, 2009 2:20 AM
To: [email protected]
Subject: st: SV: Comparing ZINB Models with clustering by subject

Bradley,

The zero-inflated Poisson with random effects (the latter bit to take into account the repeated measures design) might be useful. I haven't found any Stata program that does this model (maybe GLLAMM can do it with some tweaking?), but I've programmed the mixed ZIP using SAS' proc NL MIXED. Drop me a mail if you're interested in this code.

Mads

***

Mads Meier Jæger
Senior Researcher
The Danish National Centre for Social Research [email protected] http://www.sfi.dk/sw234.asp


-----Oprindelig meddelelse-----
Fra: [email protected] [mailto:[email protected]] På vegne af Anderson, Bradley
Sendt: 23. januar 2009 17:25
Til: [email protected]
Emne: st: Comparing ZINB Models with clustering by subject

I'm modelling a count outcome with an excess of 0 counts.  Conceptually and empirically a zero-inflated model makes sense.  It's a repeated measures design.  The number of observations per subject is relatively large (typically over 200) but varies across subjects.  Variance estimates clearly need to be adjusted for the within subject clustering.  I'd like to compare two models:  In model 1 the X is an ordered categorical variable that takes on the values 0-3.  In model 1 the effect of X is assumed linear.  In model 2, X is represented by three dummy indicators.  I want to compare the models to determine if the linear effect assumption is reasonable.  Normally I'd just do a likelihood-ratio test but that's not appropriate when using robust standard errors adjusted for clustering.  How can I compare the two models?


This e-mail and any files transmitted with it are confidential and intended solely for the use of the individual or entity to whom they are addressed.  If you are not the intended recipient, you are hereby notified that any disclosure, copying, distribution or taking of any action in reliance on the information contained in this e-mail is prohibited.  If you have received this e-mail in error, please notify sender by reply e-mail and delete this message and any attachment(s) immediately.  Thank you for your consideration in this matter.

*
*   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/


*
*   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/

This e-mail and any files transmitted with it are confidential and intended solely for the use of the individual
or entity to whom they are addressed.  If you are not the intended recipient, you are hereby notified
that any disclosure, copying, distribution or taking of any action in reliance on the information contained in
this e-mail is prohibited.  If you have received this e-mail in error, please notify sender by reply e-mail and
delete this message and any attachment(s) immediately.  Thank you for your consideration in this matter.

*
*   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 1996–2024 StataCorp LLC   |   Terms of use   |   Privacy   |   Contact us   |   What's new   |   Site index