I have a question about the GEMACO coding for ages when individuals of different ages are present at the first marking occasion. The ESURGE manual gives an example of how to do this for two “marking groups” (e.g. when there are juveniles (1yo) and adults (2+yo) present at the first occasion). In this case the GEMACO shortcut for Juveniles is [a(1).g(1)] and for the Adults for [a(2 : last age).g(1)&g(2)]. And then the shortcut for “RealAge” would be [Juveniles + Adults]
When I do this and set Transition ~ RealAge I get two parameters for the two age groups – as expected. All good.
F To T A G S
Beta #1 1 1 1 1 2 1
Beta #2 1 1 1 1 1 1
But the parameterisation with 3 ages (and 3 marking groups) seems to be different.
Three marking groups: g(1) = marked as Adults; g(2) = marked as Juveniles; g(3) = marked as Older (they’re in alphabetical order )
Maximum number of age classes = 3.
GEMACO code is:
Juvs = [a(1).g(2)]
Ads = [a(2).g(2)& a(1).g(1)]
Olds = [a(3).g(2)& a(2:3).g(2)&g(3)]
RealAge= [Juvs+Ads+Olds]
The difference is that when I set Transition ~ RealAge I get three parameters but two of these are different groups (the Juvs and Ads) and the third is an “age” parameter at the second time period rather than a third age group.
F To T A G S
Beta #1 1 1 1 1 2 1
Beta #2 1 1 1 1 1 1
Beta #3 1 1 2 2 1 1
And when I look at the output from the model with Transition ~ [RealAge + t] for my dataset (which has 5 time intervals) I get 9 parameters associated with group 1 (five of which are for age 1, and four of which are age 2 from time 2) then five parameters for group 2, then a single parameter for group 3.
Params From To Time Age Group Step Estimates
Surv 1 1 1 1 1 1 0.729
Surv 1 1 2 1 1 1 0.921
Surv 1 1 3 1 1 1 0.417
Surv 1 1 4 1 1 1 0.467
Surv 1 1 5 1 1 1 0.268
Surv 1 1 2 2 1 1 0.859
Surv 1 1 3 2 1 1 0.273
Surv 1 1 4 2 1 1 0.315
Surv 1 1 5 2 1 1 0.161
Surv 1 1 1 1 2 1 0.709
Surv 1 1 2 1 2 1 0.913
Surv 1 1 3 1 2 1 0.393
Surv 1 1 4 1 2 1 0.443
Surv 1 1 5 1 2 1 0.248
Surv 1 1 1 1 3 1 0.586
Is this ok? Or have I done something wrong?
Which group is which? (e.g. Is Group 1 made of Ads & Olds? )
Can I assume that the single parameter for group 3 at time 1 is the ‘missing’ estimate at the first time period from Age 2/Group 1?