Design matrix for repeated count data in PRESENCE

[pcutter]

I am using count data with one sample covariate and 1 site covariate. I have many missing values and several sites with no surveys whatsoever that I have included thinking that I might be able to do some predictive extrapolation based on the site covariate. The input data looks ok in the data browser. I originally tried this in MARK but the program did not seem to like double digit counts and lots of missing values.

When I open the Setup Numerical Estimation Run window, I am not getting the "custom" option as in the workbook example at:

http://www.uvm.edu/envnr/vtcfwru/spread ... Royle_.pdf

In fact, I am not getting white "window" within this window at all.

Is this a signal that my input data are just too sketchy? Other problems? Any troubleshooting advice is greatly appreciated.

[jhines]

Peter,

I removed the 'custom' option for this model since there are no variations possible with it. The option did nothing for this model anyway.

When the 'setup numerical estimation run' window appears, you should also get a design matrix window off to the right side of the screen. Sometimes, other windows may hide the design matrix window, but you should be able to bring it to the front by clicking on one of the PRESENCE icons in the taskbar at the bottom of the screen. The appearance of these windows is not affected by the amount of data.

A fairly common mistake is not clicking 'OK' after saving the data in the input form. After saving the data, the PRESENCE input filename should appear in a window, along with the results filename. If you don't click 'OK', the results file will not be created.

Cheers,

Jim

[pcutter]

Hi Jim,

I got things running and built a few models to compare.

I guess the only thing I'm stuck on is how to allow for estimates of a sample covariate to vary over different survey occasions. I'm used to having MARK provide a row for each estimate but in the design matrix here I can't seem to adjust the one row for 'p'.

Thanks again for your help working through this.

Pete

[jhines]

Peter,

You should have one row for each survey in the design matrix. If it's the 2-species model, you should have a second set of p's (pA, pB).

It seems like something isn't set up properly. Perhaps you should send me the input file (off list) and I'll check.

Jim

[darryl]

Here's a link to a PRESENCE walk-through that might help get you up and running with design matrices.

http://www.proteus.co.nz/PRESENCE_Examples.zip