Statalist The Stata Listserver


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

RE: st: lincom post-estimation using svr commands in Stata8


From   "David Thomas" <David.Thomas@menzies.edu.au>
To   <statalist@hsphsun2.harvard.edu>
Subject   RE: st: lincom post-estimation using svr commands in Stata8
Date   Fri, 9 Jun 2006 16:35:44 +0930

Thanks Maarten.  I tried -nlcom- and it works.  But it is still a bit
weird.  I am not sure but I think it is OK.  The Manuals explain
problems with confidence testing using -nlcom-, and so with the p-values
and confidence intervals.

I tested this by calculating ORs, CIs, P-values for a single
coefficient.  I found that when I wrote:

nlcom exp(_b[var1])

I got the correct answer (same as in svrmodel) for the OR, but not for
the t, p-value and CIs.

And when I just asked for the coefficient:

nlcom _b[var1]

, I got the correct t and p-value.  I could then manually calculate the
correct CIs, by inserting the given value in:

display exp(x.xxx)

This is a bit tedious but seems to work.  So it must be taking into
account the survey design, replicate weights etc in svrmodel.

So I guess I can assume it works when I do:

nlcom exp(_b[var1] + _b[var2])
nlcom _b[var1] + _b[var2]

So far the results seem to look sensible.

Cheers

David

David P Thomas

Postdoctoral Research Fellow (Mondays - Thursdays)

Menzies School of Health Research, PO Box 41096, Casuarina NT 0811

ph: 08 89227610

fax: 08 89275187


-----Original Message-----
From: owner-statalist@hsphsun2.harvard.edu
[mailto:owner-statalist@hsphsun2.harvard.edu] On Behalf Of Maarten buis
Sent: Thursday, June 08, 2006 3:35 PM
To: statalist@hsphsun2.harvard.edu
Subject: Re: st: lincom post-estimation using svr commands in Stata8

--- David Thomas <David.Thomas@menzies.edu.au> wrote:
> I am using data from the Australian Bureau of Statistics with
> replicate weights.  I have built a logistic regression model which 
> includes a couple of interaction terms <snip>. I would like to
> calculate ORs with confidence intervals for the different age groups
> vs reference category for the variables with significant
interactions. 
> I tried using -lincom- <snip> But was left with the error message: 
> unrecognized command: svysvrmodel <snip> I could calculate the ORs 
> manually, but would have trouble with calculating the CIs that way.


You could try -nlcom- like the example below:

*------------begin example-------------
sysuse auto, clear
recode rep78 1=3 2=3
xi: logit foreign i.rep78*mpg, or

nlcom (mpg4: exp(_b[mpg] + _b[_IrepXmpg_4])) /*
   */ (mpg5: exp(_b[mpg] + _b[_IrepXmpg_5]))
*------------end example----------------

HTH,
Maarten



-----------------------------------------
Maarten L. Buis
Department of Social Research Methodology
Vrije Universiteit Amsterdam
Boelelaan 1081
1081 HV Amsterdam
The Netherlands

visiting adress:
Buitenveldertselaan 3 (Metropolitan), room Z214

+31 20 5986715

http://home.fsw.vu.nl/m.buis/
-----------------------------------------

Send instant messages to your online friends
http://uk.messenger.yahoo.com 
*
*   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/



© Copyright 1996–2014 StataCorp LP   |   Terms of use   |   Privacy   |   Contact us   |   What's new   |   Site index