Statalist


[Date Prev][Date Next][Thread Prev][Thread Next][Date index][Thread index]

st: propensity scores with multiple imputation


From   marisa sklar <skoister@yahoo.com>
To   statalist@hsphsun2.harvard.edu
Subject   st: propensity scores with multiple imputation
Date   Tue, 5 Aug 2008 13:54:24 -0700 (PDT)

Hello Everybody-

I am trying to evaluate the effectiveness of a community mental health
intervention for adolescents using a non-equivalent treatment group
design, with pre- and post-treatment assessments. There is quite a
bit of missing data, so I have used the ice command to create multiply
imputed datasets. I then used the pscore command, with these multiply
imputed datasets, to establish propensity scores and strata.

I now would like to estimate the average treatment effect across strata, with the combined multiply imputed datasets. My syntax is as follows...

mim, cat(fit): atts post_activities_t activities_t group, pscore(pscore2)
blockid(myblock2)

wherein the "activities_t" and "post_activities_t" are pre- and 
post-treatment assessment scores of overall engagement in activities,
and "group" is the assignment to either treatment group or 
control/comparison group.

I have also tried the above syntax using attnd, for nearest neighbor
matching, rather than stratification.

The error message I receive when running this syntax is "last estimates
not found."

I can successfully use the mim command with a regression statement, 
and I can successfully use the atts command without using the 
multiply imputed data.

Does anybody have any advice on how I can use the atts command with the
mim command to get an estimated treatment effect?  I'd greatly appreciate
it!

Much thanks!

Marisa


      
*
*   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   |   What's new   |   Site index