Bookmark and Share

Notice: On March 31, it was announced that Statalist is moving from an email list to a forum. The old list will shut down on April 23, and its replacement, statalist.org is already up and running.


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

RE: st: RE: Non-linear multilevel modeling


From   Cameron McIntosh <cnm100@hotmail.com>
To   STATA LIST <statalist@hsphsun2.harvard.edu>
Subject   RE: st: RE: Non-linear multilevel modeling
Date   Wed, 7 Dec 2011 14:19:36 -0500

I would guess that most of the percents in this case are between 0 and 100... but if there are some 0s, perhaps some of the methods developed by our friends who sort and count cells might help:
Bagwell, C.B. (2005). Hyperlog-a flexible log-like transform for negative, zero, and positive valued data. Cytometry Part A, 64(1), 34-42.http://www.cores.utah.edu/labs/flowcytometry/pdf/HyperLog.pdf

Parks, D.R., Roederer, M., & Moore, W.A. (2006). A New "Logicle" Display Method Avoids Deceptive Effects of Logarithmic Scaling for Low Signals and Compensated Data. Cytometry Part A, 69, 541–551.http://offsite.treestar.com/downloads/Biexp_Parks.pdf
Frelinger, J., Kepler, T.B., & Chan, C. (2008). Flow: Statistics, visualization and informatics for flow cytometry. Source Code for Biology and Medicine 2008, 3(10).http://www.scfbm.org/content/pdf/1751-0473-3-10.pdf

although I've never seen these used in regression in social science contexts. Anyway, thought this might be of interest.

Cam
> From: n.j.cox@durham.ac.uk
> To: statalist@hsphsun2.harvard.edu
> Date: Wed, 7 Dec 2011 17:07:20 +0000
> Subject: st: RE: Non-linear multilevel modeling
> 
> The question with percents is always whether the limits bite in terms of functional forms that do, or do not, make sense. If your percents are a long way from 0 and 100, the fact that they are percents may be secondary. Also, you can model nonlinear changes with year using a polynomial or splines, so linear models are not restricted to straight-line fits. But whatever you do, watch for predictions that escape [0,100]. 
> 
> Or if your percents are never 0 or 100, you could model logit(depvar/100). 
> 
> Sooner or later I imagine -xtlogit- will permit continuous proportions in the response, but that doesn't seem to have happened yet. 
> 
> Nick 
> n.j.cox@durham.ac.uk 
> 
> Jennyfer Wolf
> 
> which options are there to do non-linear multilevel modelling in Stata?
> My dependent variable is a percentage and I have only one independent
> variable which is time.
> My different levels are regions and countries.
> 
> For the moment I just have a model for a linear relationship:
> 
> xtmixed depvar year || region: || country:
> 
> Thank you very much for your answer.
> Jennyfer
> *
> *   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/
 		 	   		  
*
*   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 1996–2014 StataCorp LP   |   Terms of use   |   Privacy   |   Contact us   |   Site index