[R-sig-eco] Survival model selection

[Diego Salazar]

Hi all,

I am trying to fit a survival analysis to determine effect of two 
factors on survival, factors have 3 and 5 levels respectively. N is 1050 
and experiment is balanced. I have used "survival" package for analyses.

I have problems in select the best survival model for my data, I fit 
several parametric models (weibull, expontential, gaussian, lognormal, 
loglogistic, logistic) and a Cox model. Model selection was made by AICc 
being Cox model the best option by this criteria, but I am not sure that 
AIC is valid to compare parametric and non-parametric survival models. 
Moreover, my data violates the proportional hazards assumption of Cox 
models.

For all of this, I don't know what model select.

Thanks in advanced

Diego Salazar

-- 
Diego Francisco Salazar Tortosa
Phd. student
Departamento de Ecología
Facultad de Ciencias
Universidad de Granada
Av. Fuente Nueva s/n
18071 Granada
Telefono: +34 958241000 ext 20007
Movil: +34 634851132
email:dsalazar at ugr.es
        dftortosa at gmail.com