Negative Intercept

questions concerning analysis/theory using program PRESENCE

Negative Intercept

Postby Angelica » Sun May 13, 2018 12:54 am

Hello,

I am running some single state models with binary (presence/absence) data from large mammals. I already chose my models and I have a huge doubt about the negative values of the intercepts. I have read there is not big deal about it, but I want to be clear why is this happening. I am using continuous and categorical covariates. I did the colineality analysis and standardized the continuous ones. Actually I used the normalized function of Presence, in case something is wrong with the covariates.

Could you please explain me why the negative intercepts?

I will appreciate so much your help.
Angelica
 
Posts: 6
Joined: Wed Jun 06, 2012 12:15 pm

Re: Negative Intercept

Postby jhines » Sun May 13, 2018 9:58 am

Negative values for the intercepts just means that the real parameter value is less than 0.5 when the covariate value(s) is(are) zero (covariate less than the mean in this case, since you've normalized). This has nothing to do with the effect of a covariate on the parameter. If you are expecting the parameter to always be > 0.5, regardless of any covariates, then you'd be expecting the intercept to be > 0.

The reason for this is because the real parameters are transformed using the logit-link function, which forces them to be between zero and one.
jhines
 
Posts: 599
Joined: Fri May 16, 2003 9:24 am
Location: Laurel, MD, USA


Return to analysis help

Who is online

Users browsing this forum: Google [Bot] and 2 guests