Groups or individual covariates ?

[marine]

Hello everybody,

I am a student of biology in France (sorry for my bad english !) working on a data set of capture-mark-recapture.
My aim is to test the effect of sex and age on apparent survival and recapture in lemmings.
I am a beginner in CMR anaysis and I would like to have some advices.
For my study, I decided to use a Cormack-Jolly-Seber (CJS) model.
I thought to consider age and sex as groups namely 4 groups : adult males (AM), juvenile males (JM), juvenile
females (JF) and adult females (AF) because the individuals marked as both juveniles and adults. So, my file .inp contains encounter histories
and 4 columns = 4 groups (AM, JM, JF, AF).
I also have to model uneven time-intervals between sampling occasions. Then, I thought to do the goodness-of-fit for the 4 groups and to test the different models
with additive and interactive effects.
Do you think that my study design is correct and how could I test the additive effect in the design matrix? or could I consider age and sex
as individual covariates ?


Thank you very much for your help
Marine

[cooch]

This (and related issues) are covered in various places in the 'individual covariates' chapter of the 'MARK book':

http://www.phidot.org/software/mark/doc ... chap11.pdf

Since your focus concerns GOF, the fact that there are no direct approaches for models with individual covariates (whereas there are for models where classification factors are coded as explicit 'groups') is probably a consideration.

[marine]

Hello everybody,

Thank you very much for your answer. I have a problem with my design matrix. I don't know how to test the additive effect of age, sex and time with 2 age classes (adult, juvenile) on survival. This is the 3 models that I would like tested:
Phi(A1 + sex) (A2 + time) p(time) with A1=juvenile and A2=adult
Phi (A1 + sex)(A1 * sex * time) p(time)
Phi (A1 * time)(A1 + sex*time) p(time)

I thought to make my design matrix so:
1 intercept columnn
1 column for sex (male=1 and female=0)
1 column for age (adult male = 1, juvenile male = 0, adult female = 1, juvenile female =0 )
5 columns for time
1 column for the sex*age interaction.
5 columns for sex*time interaction.
5 columns for age*time interaction.
5 columns for sex*age*time interaction.

I understand with combined age classes but how do I make my design matrix with 2 age classes separately in order to test the 3 models above-said ?

Thank you very much for your help.
Marine

[nelly]

I had the same problem. I would like to know the solution, me too !