error messages in POPAN

[treefrog]

Each time I attempt to run POPAN on any of my data sets I get an error message saying "Numerical convergence never reached, with a maximum G = number". The G is different for each dataset I run (e.g. G = 1735.55, 1160.992).

I would like to acquire estimates of probability of detection for each of 3 habitats sampled for 4-5 years. Number of surveys conducted at each site differs (e.g. 154, 83, & 64). The data can be partitioned into presence/absence per year (4 or 5 encounter occasions per habitat) or into our 3 main sampling periods for each habitat (unbalanced encounter occassions among sampling periods). However the number of species in the study is small (habitat 1= 10, habitat 2= 17, habitat 3= 15).

I'm very new to this program, but have worked through chapter 2 & 3 in the online book. Yet, I'm still confused as to where the estimates of p and phi are located.
Thank you for comments and suggestions.

[cooch]

Each time I attempt to run POPAN on any of my data sets I get an error message saying "Numerical convergence never reached, with a maximum G = number". The G is different for each dataset I run (e.g. G = 1735.55, 1160.992).

I would like to acquire estimates of probability of detection for each of 3 habitats sampled for 4-5 years. Number of surveys conducted at each site differs (e.g. 154, 83, & 64). The data can be partitioned into presence/absence per year (4 or 5 encounter occasions per habitat) or into our 3 main sampling periods for each habitat (unbalanced encounter occassions among sampling periods). However the number of species in the study is small (habitat 1= 10, habitat 2= 17, habitat 3= 15).

I'm very new to this program, but have worked through chapter 2 & 3 in the online book. Yet, I'm still confused as to where the estimates of p and phi are located.
Thank you for comments and suggestions.

Given the level of your question, I'm going to suggest strongly that you

(i) put away your own data
(ii) work through Chapters 1 -> 8 of the book in detail
(iii) then - and only then - look at Chapter 14 (The JS chapter).

If you're not even able to find out where the estimates are (making me wonder if in fact you've read chapter 3 at the least - see p. 10, for example), you're in no way ready to worry about things like convergence, etc.

Basically, chapters 1 -> 8 represent the minimum (and I do mean minimum) working knowledge needed to even begin to use MARK. If you haven't gone through that material, then it will be difficult (and not an efficient use of collective time) to try to answer your questions.

[jlaake]

I agree with the points that Evan made and strongly suggest that you work with one of the data sets that accompany the book.

Your post talked about detection probability and presence/absence and I'm wondering whether POPAN is what you want. What exactly are you trying to do? It sounds like occupancy modelling.

POPAN can be fairly tricky on convergence. You should look at the link functions that are being used. I've found that the defaults for those aren't always the best for convergence. One other trick I've used is to set initial values other than 0.

But go back and read some more, figure out what your goal is and whether POPAN is the correct model. Then work with one of the sample data sets that accompany the book so you can replicate the examples.

--jeff