Estimating p(site)

[stephanietodd]

Hello,
I have used camera traps to collect my data, so there shouldn't be variation in detection across 'surveys' (in this case a survey=3 camera trap days) but I think it is likely that there will be differences in detection between sites (i.e. cameras) simply because often at a site there was only one place we could possibly place a camera, and if the study species liked that particular spot then it would be more likely to detect it. This can't really be quantified as a covariate, but there should be different estimates. I have two ideas of how the model p(site) could be generated:

1) when entering data, select 10 sampling covariates (one for each site), which will result in 10 tabs each with a matrix in the same format as the detection data. Enter all zeros, except for the row that corresponds to the site, which is filled with all ones. Then when the model is run in the SNER window you have 10 beta values for detection in the matrix and fill each column with a covariate (site)- I think this makes the most sense

2) Don't add any detection covariates when entering the data. Instead add 10 beta values when creating the model, and enter all ones in the matrix.

Can someone let me know if either of these are correct? or if not the correct way of running the model with p(site)

Kind regards,
Stephanie