E-SURGE - no StDev

[Erlend]

Hi!

I have a set of capture histories for observations of guillemot nest sites (they breed in cliffs and don't bulid nests), and I have made the following observations:
0 = adult (may have a chick)
1 = no chick (confirmed)
2 = chick (confirmed)
If a chick dies/dissapears the site may later become occupied by adults, and it's often impossible to see whether a chick is there or not. In order to get unbiased estimates of survival I want to estimate the probability that there is a chick on the nest site even if I can only see an adult. Examples of 4 capture histories:
00020; 22002; 02011; 00000

I've tried to understand and use E-SURGE, but so far my estimates are without standard deviations, which obviously is a bad sign.

Question 1: Is E-SURGE the right application for this?
Q 2: Why are the SDs not estimated?

Thanks for any help and comments!

Erlend

[CHOQUET]

Hello Erlend,

Reply to Q1 The multievent framework is clearly adapted to deal with your problem, because you have to deal with uncertainty of the two states 'with/without chick' if there is a 'adult'. And E-SURGE is to my knowledge the only program which do that.
If I don't make any mistake, I think that you need two states :
'chick, no chick'.

However you should take care that the empty history '00000' will not be modelized (each probability is calculated conditional to the first capture). If you want to analyse each history conditional to the first date then you will have to recode the observations as:
0 = no observation(perhaps, there will be no 0 in our data set)
1 = adult (may have a chick)
2 = no chick (confirmed)
3 = chick (confirmed)

Reply to Q2 To get SE, just tick the box for Hessian before fitting the model.

Rémi

[Erlend]

Thank you very much!

[Erlend]

But when I do that, the "E" for Event in the output file is changed into a "C". Why is that?

[CHOQUET]

a very minor bug. C stand for capture here.