[Basta-users] plotting survival curves manually using the posterior means

[Fernando Colchero]

Hi Jelle,

  Well, in that case I believe that the mistake might be in the BaSTA plotting function. Let me check and see if that's the case. Thanks for pointing this out!

   Best,

   Fernando



Fernando Colchero
Assistant Professor
Department of Mathematics and Computer Sciences
Max Planck Odense Center on the Biodemography of Aging

Tlf.               +45 65 50 46 35
Email           colchero at imada.sdu.dk
Web             www.sdu.dk/staff/colchero
Pers. web   www.colchero.com
Adr.              Campusvej 55, 5230, Odense, Dk

University of Southern Denmark





On Feb 20, 2013, at 11:51 AM, Jelle Boonekamp <jjboonekamp at gmail.com> wrote:

> Hi Fernando,
> 
> That is exactly what I am doing. I get the following parameter estimates from the BaSTA function.
> 
> 
> b0		-2.389 
> b1    	        0.478 
> gamma	-0.159
> 
> 
> Then I used these like this: mu(x | z) = exp(b0 + b1x + gamma) for plotting mortality rate. Upon visual comparison, this gave me a slightly higher mortality rate than what BaSTA is actually plotting. Now that I am thinking about this, I am probably making a mistake, because BaSTA is plotting an average mortality curve for both groups. What would be the equation (concerning the gamma parameter) for the average curve? Is it like this: mu(x | z) = exp(b0 + b1x + gamma/2)? Probably not, because that still seems to be off.
> 
> Best, Jelle
> 
> 
> -- 
> Jelle Boonekamp
> Behavioural Biology
> University of Groningen
> P.O. Box 11103
> 9700 CC Groningen
> The Netherlands
> 
> tel: +31.50.363 7853
> Sent with Sparrow
> 
> On Wednesday, 20 February 2013 at 10:48, Fernando Colchero wrote:
> 
>> Hi Jelle,
>> 
>>    I think I know what the problem was. The proportional hazards model would fix b0 and b1 and add an additional parameter for one of the two groups. So the equation becomes: mu(x | z) = exp(b0 + b1x + b2z), where z will assign 1 to group 1 and 0 otherwise. In this case, you can interpret b2 as a deviation from b0. Is that what you're trying to do? Now, the reason why the plots are different is a bit puzzling, unless we are missing something. We'll double check the plotting function to make sure that we haven't made a mistake there.
>> 
>>    Best,
>> 
>>    Fernando
>> 
>> 
>> 
>> Fernando Colchero
>> Assistant Professor
>> Department of Mathematics and Computer Sciences
>> Max Planck Odense Center on the Biodemography of Aging
>> 
>> Tlf.               +45 65 50 46 35
>> Email           colchero at imada.sdu.dk
>> Web             www.sdu.dk/staff/colchero
>> Pers. web   www.colchero.com
>> Adr.              Campusvej 55, 5230, Odense, Dk
>> 
>> University of Southern Denmark
>> 
>> 
>> 
>> 
>> 
>> On Feb 20, 2013, at 10:36 AM, Jelle Boonekamp <jjboonekamp at gmail.com> wrote:
>> 
>>> Hi BaSTA users,
>>> 
>>> I ran into something peculiar. When I manually plot the survivor function (2par Gompertz in my case) using the posterior mean parameter estimations, I get a slightly different survival curve/shape compared with the curve that BaSTA draws. Shouldn't these be exactly similar, or am I missing something? For the plotting I used the equation from table 3 on p16 of the cran manual, with proportional hazard implementation following equation 7b on page 19. The aim of this exercise was to estimate the effects of the Gompertz parameters on mean lifespan separately.
>>> 
>>> Best, Jelle
>>> 
>>> 
>>> -- 
>>> Jelle Boonekamp
>>> Behavioural Biology
>>> University of Groningen
>>> P.O. Box 11103
>>> 9700 CC Groningen
>>> The Netherlands
>>> 
>>> tel: +31.50.363 7853
>>> Sent with Sparrow
>>> 
>> 
> 

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.r-forge.r-project.org/pipermail/basta-users/attachments/20130220/2fae6b57/attachment-0001.html>