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: How to Compare the Coefficients of Specific Variable where Each Group is Estimated Separately


From   Maarten Buis <maartenlbuis@gmail.com>
To   statalist@hsphsun2.harvard.edu
Subject   Re: st: How to Compare the Coefficients of Specific Variable where Each Group is Estimated Separately
Date   Wed, 19 Dec 2012 11:14:33 +0100

On Wed, Dec 19, 2012 at 7:16 AM, Yuval Arbel wrote:
> I am running a -stcox- model on two groups, where my particular
> interest is to compare the coefficients of the variable red_runmax1
>
> The difference between the two coefficients is -0.0246686 and I would
> like to show that this difference is significant
>
> Finally I tried to run the test on a model with interactions with the
> dummy. The problem is that I don't get the estimated -0.0246686
> difference of coefficients.

In order to reproduce the results from two separate -stcox- models,
you also need to make sure that the baseline hazard rate is also
different between the two groups. You can do so by adding the
-strata()- option. See the example below:

*----------------------------- begin example ------------------------------
use http://www.stata-press.com/data/r12/stan3.dta, clear
// estimate both model separately
bys surgery: stcox age year, nohr

// create interaction terms
// use tricks from <http://www.maartenbuis.nl/publications/ref_cat.html>
// to directly display the coefficients from the separate models, you
// can see they are exactly the same
gen surgXage = surgery*age
gen surgXyear = surgery*year
gen nonsurgXage = (surgery==0)*age if surgery < .
gen nonsurgXyear = (surgery==0)*year if surgery < .

stcox nonsurgXage nonsurgXyear surgXage surgXyear, strata(surgery) nohr

// test whether the effect of year is the same in two groups
test nonsurgXyear = surgXyear
// see the chi^2 statistic and p-value with more digits for later comparison
return list

// the same model, but now you can read the test statistic
// directly from the output.
// Notice, I did not include the main effect of surgery, this
// is implicitly added by the -strate()- option.
stcox age year surgXage surgXyear, strata(surgery) nohr
// see the chi^2 statistic and p-value with more digits
// they are exactly the same as in the model above
di "the chi^2 value is: " (_b[surgXyear]/_se[surgXyear])^2
di "the p-value is: " 2*normal(-abs(_b[surgXyear]/_se[surgXyear]))
*----------------------------- begin example ------------------------------
(For more on examples I sent to the Statalist see:
http://www.maartenbuis.nl/example_faq )

Hope this helps,
Maarten

---------------------------------
Maarten L. Buis
WZB
Reichpietschufer 50
10785 Berlin
Germany

http://www.maartenbuis.nl
---------------------------------
*
*   For searches and help try:
*   http://www.stata.com/help.cgi?search
*   http://www.stata.com/support/faqs/resources/statalist-faq/
*   http://www.ats.ucla.edu/stat/stata/


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