C.I. for derived parameters of model averaged set

questions concerning analysis/theory using program PRESENCE

C.I. for derived parameters of model averaged set

Postby kyle.artym » Tue Sep 22, 2015 4:18 pm

Hi there,

I am trying to derive confidence intervals for specific covariates. I calculated the model-averaged values from the back transformed values, as recommended in Darryl's workshop.

I used the Delta method to calculate SEs and CIs for each unique model, where the CIs are bound between [0,1]. However, I cannot find a formula to produce CIs for the model averaged parameter estimates.

Since values for psi are near 1 for some model average estimates, Y +/- (1.96*modelaveragedSE) are not bounded between [0,1].

Can you recommend a methodology to derive CIs for derived parameters of model averaged set that are bound between [0,1]. If this is not possible, could you recommend a satisfactory way of reporting these limits in the literature?

Thank you,

Kyle
kyle.artym
 
Posts: 8
Joined: Wed Aug 26, 2015 11:55 am

Re: C.I. for derived parameters of model averaged set

Postby darryl » Tue Sep 22, 2015 4:31 pm

No claims it's the perfect solution, but what I do is take the model-averaged estimate and SE, transform them onto the logit scale (good practice of the delta method), calculate the limits with +/- 1.96 logit(SE), then back transform the limits to the 0-1 scale. This will keep them within the boundaries.
Darryl
darryl
 
Posts: 495
Joined: Thu Jun 12, 2003 3:04 pm
Location: Dunedin, New Zealand

Re: C.I. for derived parameters of model averaged set

Postby kyle.artym » Tue Sep 22, 2015 4:49 pm

Awesome, thanks Darryl!
kyle.artym
 
Posts: 8
Joined: Wed Aug 26, 2015 11:55 am

Re: C.I. for derived parameters of model averaged set

Postby cooch » Tue Sep 22, 2015 5:03 pm

darryl wrote:No claims it's the perfect solution, but what I do is take the model-averaged estimate and SE, transform them onto the logit scale (good practice of the delta method), calculate the limits with +/- 1.96 logit(SE), then back transform the limits to the 0-1 scale. This will keep them within the boundaries.
Darryl


Although not specifically considering an occupancy analysis, much of this thread, and Darryl's approach, are laid out in *much* detail in the MARK book (http://www.phidot.org/software/mark/docs/book/). See in particular Chapter 4, section 4.5.1, Chapter 6, section 6.16, and Chapter 11, section 11.8.
cooch
 
Posts: 1628
Joined: Thu May 15, 2003 4:11 pm
Location: Cornell University

Re: C.I. for derived parameters of model averaged set

Postby Alan Dextrase » Wed Sep 23, 2015 12:38 pm

Hello Kyle,

I created a simple spread sheet to automate the procedure that Darryl describes. if you send me an e-mail (alandextrase(at)trentu.ca), I can send it along.

Alan
Alan Dextrase
 
Posts: 17
Joined: Tue Dec 07, 2010 10:01 pm
Location: Peterborough ON, Canada


Return to analysis help

Who is online

Users browsing this forum: Google [Bot] and 9 guests

cron