www.phidot.org

With simulated data I have explored a bit how bias estimated survival rate will be in CJS model in case capture-recapture events do not happen once during a period. Maybe there is some trick with modeling such data but I was not able to get reliable estimates though have tried many different ways to constrain models and design matrix.
My feeling is that at least with bird resighting data we almost never have short and intense sessions of resighting – the data are usually coming in small portions on a daily basis.
So the questions are:
1. Is there some trick to account for continuous recapture / resight?
2. Where in the literature this problem is discussed in details?
Thanks!

Eldar wrote:My feeling is that at least with bird resighting data we almost never have short and intense sessions of resighting – the data are usually coming in small portions on a daily basis.

Overly general - depends on the bird species, and the nature of the study.

So the questions are:
1. Is there some trick to account for continuous recapture / resight?

Usual approach is simple pooling. Everything is collapsed into 'seen at least once during a period, or not seen at all'. There has been some discussion given to the idea that using the number of encounters as a covariate is a way to accommodate heterogeneity in encounter probabilities, but I know of no formal studies.

The other approach has been to consider such things as a permutation of a Barker model, or RD-Barker, where you specify primary sampling periods, and everything in between is treated like opportunistic encounters (i.e., Barker model).

If you do have truly continuous detections, then the final approach is to pretend birds are fish, and delve into the fisheries literature on continuous-encounter models.

the final approach is to pretend birds are fish, and delve into the fisheries literature on continuous-encounter models

Had no success with searching for continuous time CJS (not JS) models, where the aim will be estimate survival. Could you please provide any examples?
thanks a lot!

Eldar wrote:

the final approach is to pretend birds are fish, and delve into the fisheries literature on continuous-encounter models

Had no success with searching for continuous time CJS (not JS) models, where the aim will be estimate survival. Could you please provide any examples?
thanks a lot!

Thats because 'CJS' can't be 'continuous', by definition. If its 'continuou time', its not 'CJS'.

That's right. But were there attempts to estimate survival under nonperfect observations (p<1) with continuous time (not in CJS framework)?

Have you seen this Eldar?

http://onlinelibrary.wiley.com/doi/10.1111/2041-210X.12059/abstract