st: re: group based trajectory analysis

["Ariel Linden, DrPH" <ariel.linden@gmail.com>]

Steve,

You can contact Bobby Jones directly via the website I listed below for
"traj". I have been trouble-shooting a Stata-based version of the program
that Bobby is working on. Unfortunately, I am having problems with the
program due to the fact that it uses a plug-in. It ran fine for me in
Windows XP on a 32 bit machine, but now I have Windows 7 and a 64 bit
machine, and the code just doesn't work. Maybe it will work for you.

In addition, there is a user-written (Partha Deb) program in Stata called
-fmm- (finite mixture modeling). This modeling approach is the underpinning
of "traj", so it could possibly serve as an alternative. Partha told me that
fmm can be used for longitudinal analysis, but I am not clear how to do
that. I've given it a lot of thought, and just can't seem to wrap my head
around the issue of determining "group assignment" when there are multiple
time periods involved.

I'd be curious as to the findings of your exploration...

Ariel

________________________________________

Dear all
 
I found this thread on stata list and noticed that Ariel thought a traj
program is forthcoming for Stata. Can anyone update on this or any other
means to conduct this type of analysis with Stata? 
 
I have longitudinal cohort data on BMI for n=childre7000 children and trying
to make sense of ther weight trajectories by identifying and grouping
differing growth paths. Links to alternate approaches in Stata also greatly
appreciated.
 
Help much appreciated.
 
BW
 
Steve
 
Dr Steven Allender

Associate Professor/ Deputy Director
World Health Organization Collaborating Center for Obesity Prevention Deakin
University

Senior Researcher
Coronary heart disease statistics
Department of Public Health
University of Oxford
Re: st: Time series and psmatch2
________________________________________
From	Ariel Linden <alinden@lindenconsulting.org>

To	statalist@hsphsun2.harvard.edu

Subject	Re: st: Time series and psmatch2
Date	Wed, 9 Dec 2009 10:12:22 -0800
________________________________________
Following up on Austin's comments: yes, it is perfectly reasonable to use
each monthly or yearly observation as covariates in the propensity score
model. 

There are a couple alternative options available for this problem:

(1) roll up the periods into one aggregate value and propensity score match,
then check the balance on each period separately. Rosenbaum and Rubin have
suggested in more than one paper that the propensity score will tend to
achieve balance on these covariates, and I find this to be the case almost
100% of the time.

(2) there is a user-defined program in SAS called "traj"
(http://www.andrew.cmu.edu/user/bjones/index.htm) which takes multiple
observations per person and assigns them to a trajectory group (through a
clustering model). Once the subjects are assigned to the given trajectory
groups, you can further propensity score match them within the group. BTW,
the traj program is currently being written for Stata (yeah)!

A marvelous paper on this concept is:

Amelia Haviland, Paul R. Rosenbaum, Daniel S. Nagin. Combining Propensity
Score Matching and Group-Based Trajectory Analysis in an Observational
Study. Psychological Methods. 2007, Vol. 12, No. 3, 247-267


I hope this helps.

Ariel


*
*   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/