Fixing covariate values for model-averaged additive models

questions concerning analysis/theory using program PRESENCE

Fixing covariate values for model-averaged additive models

Postby leroy » Thu Jun 21, 2018 2:44 am

Hi,

I have model-averaged across multiple single-season models with additive covariates for occupancy. I want to plot the effect of one covariate while holding the other covariates constant (by using the mean value for each covariate across all of my sites). However the model-averaged estimates don't come with betas that I can use for this purpose.

Does anyone have any suggestions on how to deal with this? Currently I plot my model-averaged estimates against a covariate of interest (one that is in my list of supported models) and acknowledge that the plot represents the association between that covariate and occupancy after accounting for the influence of the other covariates included in my list of supported models.

Any advice would be appreciated.

Thanks,

Leroy.
leroy
 
Posts: 10
Joined: Thu Aug 04, 2016 4:04 am

Re: Fixing covariate values for model-averaged additive mode

Postby darryl » Thu Jun 21, 2018 3:18 am

One trick to do this is in PRESENCE is to add some 'sites' to your data set that have all missing detection/nondetection values (so has no effect on parameter estimation) but with the set of covariate values that you're interested in. When you do model averaging, PRESENCE will give you an estimate for those extra sites.

If you're an R user, you can do this pretty easily with RPresence package.

Cheers
Darryl
darryl
 
Posts: 495
Joined: Thu Jun 12, 2003 3:04 pm
Location: Dunedin, New Zealand

Re: Fixing covariate values for model-averaged additive mode

Postby leroy » Tue Jun 26, 2018 9:59 am

Thanks a lot for the advice Darryl. Much appreciated. It seemed to work well.

I have a situation where an additive model is the most supported model for detection. One of my covariates in this model is temperature and its influence on detection is quite strong when the other covariate is held constant for all sites. However, the other categorical covariate (detector type) appears to not have any influence on detection when temperature is held at its mean. Just wondering why detector type comes out as being supported in an additive model when it doesn't appear to influence detection on its own?

Cheers,

Leroy.
leroy
 
Posts: 10
Joined: Thu Aug 04, 2016 4:04 am


Return to analysis help

Who is online

Users browsing this forum: No registered users and 12 guests