www.phidot.org

Hi all
I am needing some help with conducting a closed pop assessment using IELogitNormalMR.

The Setup:

Three primary surveys, each of varying secondary survey length. And each primary survey also varies in terms of Known Marked Animals.
The population between the primary surveys are open, and within the secondary surveys closed (so, in essence we have three closed pop surveys (5 days, 4 days and 4 days) across 14 months.

I have been using the IELogitNormalMR model for the assessment with the following code:

Closed_Proc=process.data(Capensis_Closed,model="IELogitNormalMR", counts=list("Marked Superpopulation"=c(67,67,67,67,67,36,36,36,36,12,12,12,12),"Unmarked seen"=c(67,27,20,28,27,33,36,30,26,20,26,12,8)),time.intervals = c(0,0,0,0,1,0,0,0,1,0,0,0))
Closed.ddl=make.design.data(Closed_Proc)
Capensis_Closed_Mod1=mark(Closed_Proc, Closed.ddl)

I continue getting the following error:
Warning messages:
1: In data.frame(estimate = as.numeric(x$estimate), se = as.numeric(x$se), :
NAs introduced by coercion
2: In data.frame(estimate = as.numeric(x$estimate), se = as.numeric(x$se), :
NAs introduced by coercion
3: In data.frame(estimate = as.numeric(x$estimate), se = as.numeric(x$se), :
NAs introduced by coercion

An example of what my data looks like:

ch
10100........
10011........
10110........
10010........
100000100....
100000101....
.....1100....
.....1111....
.....1010....
11100........
101110011....
100011011....
111011111....
100100100....
10010....1001
100100100....
101110101....
1010010000100
100000011....
100111011....
110000011....
10010........
10111....1000
11000........

Any help or suggestions would be much appreciated.

Thanks!

Hi Louw,

This isn't really an RMark problem, it's a model specification and/or optimization problem. Try starting with simpler models, and, if they still don't converge, setting initial values for the parameters (N and sigma can be particularly squirrelly). For example, try starting with a simple model with no individual heterogeneity:

Code: Select all

Capensis_Closed_Mod2=mark(Closed_Proc2, Closed.ddl2,
                          model.parameters=list(p=list(formula=~1),
                                                sigma=list(formula=~1,fixed=0),
                                                alpha=list(formula=~session*time)))


If that converges, then starting building up from there. If not, try setting initial values.

Cheers,
Brett