confidence interval for p (0-1)??

[Mauricio]

Hi
I am having some problems with one of four site covariates. when I run it alone or in company of another site or sampling covariates I got standard errors higher than 1 and the program is unable to estimate the confidence interval for p. (my PAO estimate is 1)
The covariate is categorical (1=presence- 0=absence). I have another site covariate with very similar values but I haven't had any problems with that.
My site covariates are
bromelids (1 or 0)
Gambas(huge tropical roots/ 1or 0)
Ph (continue)
Cs(leaf litter coverage %)
Any Idea of the problem?

Best regards
Mauricio

[Bird Counter]

Hi Mauricio,

I'm a biologist, not statistician, but I have had the same issue when I have no variability in occupancy associated with the covariate (that is, the species is always or never present at some level of the covariate). Also, if p is extremely low at some level of your covariate (e.g., when bromeliad = 0), this can cause the models to become non-estimable. When you have a combination of low occupany and/or low detectability at one level of the covariate, all kinds of problems can occur. Look at the distribution of your species (1,0) at different levels of the covariate that is creating the problem as a first step.

[Mauricio]

Hi and thanks Bird counter for your response
Maybe I am in a worst condition , I am nor Biologist neither Statistician I am Psychologist, anyway. I was checking the data and I could not find any pattern. PAO is 1 in all my models and the lowest value of p is 0.3161 in pao (pg), p(.). and PAO 1 and the lowest p = 0.2738 in pao(pg), p(fullID).
PG sometimes is present when the specie is detected and sometimes not so I am just wondering if I should eliminate that variable and the models where it is involved.
any idea?
Best regards
mauricio