Time trend in certain groups only (multistate model)

posts related to the RMark library, which may not be of general interest to users of 'classic' MARK

Time trend in certain groups only (multistate model)

Postby monicaarso » Tue Mar 27, 2018 10:20 am

Hi,
I am fitting multistate models to estimate sex-specific survival. The three defined states (stratum): M, F, and U.
I have successfully fitted a series of models to estimate sex-specific survival with and without a time trend. I would like though to fit a model with a time trend on female survival (F), a constant male survival (M) and either constant or a time trend for the unknown state animals (U).

Process data and design data:
Code: Select all
#Process data
sexMS.processed=process.data(markdataMS,model="Multistrata",begin.time=1989)

#Create the design data
sexMS.ddl=make.design.data(sexMS.processed,parameters=list(Psi=list(pim.type="time")))


#add design data to allow survival probabilities to be M=F<>U, M=U<>F,F=U<>M
Code: Select all
sexMS.ddl$S$SFM="FM"
sexMS.ddl$S$SFM[sexMS.ddl$S$stratum=="U"]="SU"
sexMS.ddl$S$SFU="FU"
sexMS.ddl$S$SFU[sexMS.ddl$S$stratum=="M"]="SM"
sexMS.ddl$S$SMU="MU"
sexMS.ddl$S$SMU[sexMS.ddl$S$stratum=="F"]="SF"


Define survival parameters to run models
Code: Select all
#Time trend for each state (sex)
S.sexTime=list(formula=~Time:stratum)
#Time trend for F and M+U (together as one parameter)
S.MUTimeb=list(formula=~Time:SMU)


I wonder whether I need to modify the design matrix for S before I can define the model parameters? Any tips on how to specify the survival parameter would be much appreciated. I just haven't been able to see any examples.
Many thanks,
Monica
monicaarso
 
Posts: 33
Joined: Wed Feb 22, 2012 2:58 pm

Re: Time trend in certain groups only (multistate model)

Postby monicaarso » Tue Mar 27, 2018 11:30 am

Actually, I figured it out by looking at another post with a similar issue.
In case it is of anyone's interest the trick is to add a dummy variable for each of the groups, so that you can call them separately when defining survival. In my case:

Code: Select all
#define dummy variable for each sex
sexMS.ddl$S$Female=0
sexMS.ddl$S$Female[sexMS.ddl$S$stratum=="F"]=1
sexMS.ddl$S$Male=0
sexMS.ddl$S$Male[sexMS.ddl$S$stratum=="M"]=1
sexMS.ddl$S$Unknown=0
sexMS.ddl$S$Unknown[sexMS.ddl$S$stratum=="U"]=1


and then the models can be run, for example for a Time trend in Female survival but keeping male and unknown animals constant this would be:

Code: Select all
S.TF.Mdot.Udot=list(formula=~Time:Female+Male+Unknown)
p.time=list(formula=~time)
#fix some transitions in years I know there were no unknown animals being sexed as males or females.
Psi.time=list(formula=~-1+time,fixed=list(index=c(110,112,113,114,115,116,117,131,132,133,134,135,136,137,138,139,140,141,142,143),value=0))

model=mark(sexMS.processed,sexMS.ddl,model="Multistrata",model.parameters = list(S=S.TF.Mdot.Udot,p=p.time,Psi=Psi.time))


Hopefully I did not make anyone lose time with this.
Have fun Rmarkcoding.
monicaarso
 
Posts: 33
Joined: Wed Feb 22, 2012 2:58 pm


Return to RMark

Who is online

Users browsing this forum: No registered users and 14 guests