Collapsing Occasions

[horchard]

Hi Murray,

I am using proximity detectors and testing behavioural and temporal covariates (homogeneous density, no spatial covariates yet), so I started using fastproximity = FALSE. However, the best fitting models are producing really high density estimates with wide confidence intervals. These same models when ran with collapsed occasions had much more reasonable estimates.
The data consists of 3, 1 week long sampling periods spaced approximately 1 month apart (1 occasion in January, 1 in February, and 1 in March). There is a relatively low number of spatial recaptures.
I am wondering if it would be better to let the occasions collapse, as it may just be too complicated for the dataset and if that's what could be causing these high density estimates. Would there be any issues with this if I still wanted to test k, b, T, etc.?

*update: I understand that these covariates simply won't run with collapsed occasions, which in hindsight, makes sense!

Thank you!

Hailey