Date:         Fri, 3 Nov 2006 11:06:24 -0500
Reply-To:     Michael Ni <Michael.Ni@COGNIGENCORP.COM>
Sender:       "SAS(r) Discussion" <SAS-L@LISTSERV.UGA.EDU>
From:         Michael Ni <Michael.Ni@COGNIGENCORP.COM>
Subject:      Re: Proportional Hazard Model
Comments: To: Haris <Karovaldas@GMAIL.COM>
In-Reply-To:  <1162568662.484217.286520@m73g2000cwd.googlegroups.com>
Content-Type: text/plain; charset=us-ascii

Haris,

Thanks for your help! If there are 10 variables exist in the final
model, I need to check 10 interaction terms separately? In that case,
can I still check the assumption by using graphical techinuqe? If yes, how?

Thanks,
Michael


Haris wrote:

>Statistically, you use time-varying covariates that capture a
>systematic pattern of deviation for a constant HR.  The simplest and
>most often use is a simple interaction between a categorical predictor
>and Time.  A step above is an interaction with log of Time.  From there
>sky is the limit if you have a non-monotonic departutre from
>proportional hazards.  The key is to look at the change in hazard
>across time graphically!
>
>On Nov 3, 9:29 am, "Vadim Pliner" <Vadim.Pli...@VerizonWireless.com>
>wrote:
>
>
>>Michael Ni wrote:
>>
>>
>>>Hi,
>>>
>>>
>>>Could anyone tell me how to verify the assumption for proportional
>>>hazard model in SAS/PROC PHREG?
>>>
>>>
>>>Thanks a lot!
>>>
>>>
>>>MichaelThe SAS 9 documentation can tell you "how to verify the assumption for
>>>
>>>
>>proportional
>>hazard model in SAS/PROC PHREG": SAS/STAT -> PROC PHREG -> Assessment
>>of the Proportional Hazards Model - Experimental -> Checking the
>>Proportional Hazards Assumption. Or, you can read a section titled
>>"Checking the Proportional Hazards Assumption" in Alan Cantor's book
>>"Extending SAS Survival Analysis Techniques for Medical Research."
>>
>>HTH,
>>Vadim Pliner
>>
>>