Indicator variables for categorical covariates

questions concerning analysis/theory using program PRESENCE

Indicator variables for categorical covariates

Postby JoshWeaver » Thu Apr 14, 2022 8:37 am

I have one site covariate which is categorical (habitat feature = clearing, stream etc.), I have converted this covariate into four (5-1) indicator variables to be included in single-species, single-season occupancy models.

My question is whether these indicator variables must always be modelled together, or can they be modelled independently? I have noticed that one of the habitat features appears to have a significant effect on occupancy, but when modelled alongside the other indicator variables the model is not ranked highly.

As one habitat feature offers a good explanation for site occupancy but the other habitat features do not, can this habitat feature be modelled independently?

Any guidance would be appreciated as I'm new to occupancy modelling.

Thank you.
JoshWeaver
 
Posts: 2
Joined: Thu Apr 14, 2022 8:06 am

Re: Indicator variables for categorical covariates

Postby jhines » Thu Apr 14, 2022 8:54 am

No, they don't all have to be included. When all 4 indicator covariates are included, you get 5 independent estimates of occupancy. If you only included one of the covariates in a model, let's say stream, then you would get an estimate of occupancy which applies to the other 4 categories, and an estimate of occupancy for stream sites. If you included clearing and stream in the model, you'd get an estimate for the other 3 site types, one for clearing and one for stream for a total of 3 different estimates of occupancy. In general, any covariates of the 5 that you don't include in the design matrix will have the same estimate of occupancy, and the covariates which are included in the design matrix will have different estimates.

If you wanted a model with one category for both clearing and stream and one for the other 3 categories, you will need to create a new indicator covariate which contains a '1' if the site is either clearing or stream, and '0' otherwise. Then, use only that new covariate in the design matrix.
jhines
 
Posts: 598
Joined: Fri May 16, 2003 9:24 am
Location: Laurel, MD, USA

Re: Indicator variables for categorical covariates

Postby JoshWeaver » Thu Apr 14, 2022 9:08 am

That’s great, thank you very much for the explanation and the quick response!
JoshWeaver
 
Posts: 2
Joined: Thu Apr 14, 2022 8:06 am


Return to analysis help

Who is online

Users browsing this forum: No registered users and 1 guest

cron