Multi-season, multi-state model in PRESENCE error

[bross]

I’m currently trying to run the multi-state, multi-season model (with the Nichols et al. 2007 parameterization) with 2 known states and 53 seasons (35 surveys each season). Each time I try to run this particular model, PRESENCE returns a “confirm” window with an AIC = 0 and the DOS window gives the following error: “This application has requested the Runtime to terminate it in an unusual way. Please contact the application’s support team for more information.” Admittedly, I am more familiar with MARK and RMark than PRESENCE, but I have been unable to find documentation of this particular error anywhere in the literature (as it seems different than other errors stated in the FAQ of the forum), so I am unsure what steps need to be taken. Any help would be greatly appreciated.
Thanks,
Beth

[darryl]

My first suggestion is that you double (and triple) check your data to make sure you only have 0's, 1's and 2's or "-" or "." for missing values.

Then check that you're model structure is ok in the design matrix. What sort of model are you trying to fit?

Cheers
Darryl

[jhines]

Hi Beth,

There's an error in the input (pao) file. There should be 387 lines for each covariate, but the 7th covariate only has 386. I didn't check to see if it would work by fixing this as I think you might want to consider pooling the surveys within each season. There isn't nearly enough data to estimate 1,855 detection probabilities, and only a few sites have more than about 15 surveys per season. This would greatly reduce the number of estimated parameters and simplify the models. Since you don't have survey covariates, pooling surveys should be pretty straightforward.

By the way, the error message you described in your original post is a generic Windows 'something went wrong' error message, probably due to a subscript out of range. When you get this running PRESENCE, it's usually a problem with the input file.

Cheers,

Jim

[BDWilliams]

I'm having similar issues with the multi-season, multistate models but it's not the input file. I've even tried entering in fake data by hand. Any suggestions?

-- Branson