basic questions - GEMACO

questions concerning analysis/theory using programs M-SURGE, E-SURGE and U-CARE

basic questions - GEMACO

Postby Laurent B. » Mon Mar 26, 2007 10:46 am

Hello,
I am a beginner using M-Surge to study cost of reproduction in an iteroparous vertebrate. I am using 4 states, based on the “reproductive intensity” and am looking for difference in probability to breed and survive the following season (year t+1) according to breeding status year t.
I have some very basic questions regarding the GEMACO language:
1. does it make sense to use “different” states for the previous (from) and the current state (to)? For example, is it OK to define a model like: Psi [from(1, 2:4).to(1, 2, 3, 4]?
2. If yes, do I have to change the “Transition patterns” matrix for this particular model?
3. Then, I am using some covariates to model yearly variations in survival. For example, I would like to see whether variations in survival can be explained by variation in spring temperature (data normalized before analyses). The most parsimonious model I found for survival was : Phi[f(1, 2:4).t + a]. To test for covariate effect, is this OK to write: Phi[f(1, 2:4).t*x + f(1, 2:4) + a] ?

Thanks a lot,
Laurent
Laurent B.
 
Posts: 1
Joined: Mon Mar 26, 2007 10:07 am

Postby CHOQUET » Thu Mar 29, 2007 5:53 am

Hello Laurent,

Concerning :
1. It makes sense to do such a model however equalities of type
f(2).to(1)=f(3).to(1) can be destroy by the multinomial logit if either one
of the following equalities(in the same row) f(2).to(2)=f(3).to(2) or f(2).to(3)=f(3).to(3) or f(2).to(4)=f(3).to(4) is not satisfied.
2. With respect to 1., you have to set complementaries(parameters computed as one minus the others in a same row) at the same place from rows 2 to 4. For example, setting all the complementaries to the same column(one for example) allows the model described in 1.
3. It seems to be OK. Another way to write the same model could be
f(1,2:4).[i+t*x]+a
Sincerely,
Rémi
CHOQUET
 
Posts: 211
Joined: Thu Nov 24, 2005 4:58 am
Location: CEFE, Montpellier, FRANCE.


Return to analysis help

Who is online

Users browsing this forum: No registered users and 12 guests