Wolney, I am not sure, but my guess is the second run is not using the design information, then you are getting artificially narrower CI and diverse point estimates due to the lack of weighting. check if your desing info (svyset) is the same before both runs.
Diego
Diego G. Bassani, M.Sc., Ph.D.
Centre for Addiction and Mental Health
Health Systems Research & Consulting Unit
33 Russell Street, Tower - T308
Toronto, ON - M5S 2S1
[email protected]
http://www.camh.net/hsrcu
Phone: +01 (416) 535 8501 # 6465
Fax: +01 (416) 979 4703
-----Original Message-----
From: [email protected]
[mailto:[email protected]]On Behalf Of Wolney Conde
Sent: April 21, 2006 9:01 AM
To: statalist
Subject: st: Stata behavior
Dear listers,
I'm facing a very strange Stata behavior. I've a long routine to execute
over a dataset in Stata (version 9, updated to 14 april), so I put it in
a do file. Two information: the data set is large (17,000 observations)
and has complex sample (the analysis use the survey family commands).
When I run the do file and write the command for estimation in command
window I get the first output below; when I execute the routine from the
do file (including the command for estimation) I have the second out
below, which is very different from the first one. Anyone has
experienced that sometime? Any suggestion to dealing with it? Any ideas
about the problem source?
Thanks in advanced
Wolney.
. run c:\wolney\chamada\chamada
_analise_old.do
. xi:svy:logistic haz77 i.bfam bens anoestm anoestc if samplea==1&idade4==3
i.bfam _Ibfam_0-1 (naturally coded; _Ibfam_0 omitted)
(running logistic on estimation sample)
Survey: Logistic regression
Number of strata = 9 Number of obs
= 2980
Number of PSUs = 479 Population size =
216878.4
Design df
= 470
F( 4, 467)
= 1.90
Prob > F =
0.1100
------------------------------------------------------------------------------
| Linearized
haz77 | Odds Ratio Std. Err. t P>|t| [95% Conf.
Interval]
-------------+----------------------------------------------------------------
_Ibfam_1 | .9830186 .372281 -0.05 0.964 .4670563
2.06897
bens | 1.02921 .1386758 0.21 0.831 .7897996
1.341192
anoestm | .9571184 .0734288 -0.57 0.568 .8231785
1.112852
anoestc | .8983084 .0532568 -1.81 0.071 .7995233
1.009299
------------------------------------------------------------------------------
. do "C:\DOCUME~1\Wolney\CONFIG~1\Temp\STD02000000.tmp"
. run c:\wolney\chamada\chamada_analise_old.do
.
. xi:svy:logistic haz77 i.bfam bens anoestm anoestc if samplea==1&idade4==3
i.bfam _Ibfam_0-1 (naturally coded; _Ibfam_0 omitted)
(running logistic on estimation sample)
Survey: Logistic regression
Number of strata = 9 Number of obs
= 2980
Number of PSUs = 479 Population size =
221958.46
Design df
= 470
F( 4, 467)
= 2.02
Prob > F =
0.0899
------------------------------------------------------------------------------
| Linearized
haz77 | Odds Ratio Std. Err. t P>|t| [95% Conf.
Interval]
-------------+----------------------------------------------------------------
_Ibfam_1 | .8659906 .3355343 -0.37 0.711 .4044431
1.854253
bens | 1.042337 .1354768 0.32 0.750 .8073995
1.345637
anoestm | .9503177 .070909 -0.68 0.495 .8207134
1.100389
anoestc | .903281 .0523593 -1.75 0.080 .8060371
1.012257
------------------------------------------------------------------------------
.
end of do-file
--
Wolney L Conde
/Department of Nutrition/
/School// of Public Health - University of Sao Paulo/
Av. Dr. Arnaldo 715
01246-904 Sao Paulo, Brazil
phone: +55-11-30667705 extension 232
fax: +55-11-30667705
*
* For searches and help try:
* http://www.stata.com/support/faqs/res/findit.html
* http://www.stata.com/support/statalist/faq
* http://www.ats.ucla.edu/stat/stata/
*
* For searches and help try:
* http://www.stata.com/support/faqs/res/findit.html
* http://www.stata.com/support/statalist/faq
* http://www.ats.ucla.edu/stat/stata/
