Follow up on overall psi estimate Question

questions concerning analysis/theory using program PRESENCE

Follow up on overall psi estimate Question

Postby bgerber » Fri Jan 11, 2008 6:51 pm

Without covariates is it presumptuous to make inferences using the overall psi estimate to areas that were not sampled? In my situation, I do not have a large enough sampling size (n=23 sites, 7 surveys) to tease apart covariate data that was taken at each site. Covariate estimates basically come back with 95% CL of 0.1 to 1 for each site.

If each site is considered a random point, can I regard the overall psi estimate as a representation of a larger area? How large of an area?

Sincerely, Brian Gerber
bgerber
 
Posts: 24
Joined: Wed Dec 12, 2007 1:01 pm

Random?

Postby jlaake » Fri Jan 11, 2008 7:17 pm

You don't need covariates to compute an overall Psi from a selection of sites. But your query "How big of an area?" makes me think that you didn't select sites with any particular sampling plan. Had you done that you would have had to define the region/area at the start for the sampling frame.

If you work through the example I gave and plug in a constant Psi then you'll get the variance for Psi that you started with because each variance and covariance is the variance of Psi so var(psi)= n^2*var(Psi)/n^2. Thus you would use the estimate of Psi and its variance (std err) from the output for a constant Psi. But your inference is only as useful as your sampling design.

--jeff
jlaake
 
Posts: 1417
Joined: Fri May 12, 2006 12:50 pm
Location: Escondido, CA

Postby bgerber » Sat Jan 12, 2008 12:37 pm

What I am asking of this data is possibly too much, as it was pilot study data that was given to me. The significance was to test the feasibility of carrying out such a study in a remote rainforest of Madagascar. The sampling design is truly the crux of the issue, thank you for bringing that to my attention. I was not thinking about the question correctly.

Sincerely, Brian
bgerber
 
Posts: 24
Joined: Wed Dec 12, 2007 1:01 pm

usefulness of pilot survey

Postby jlaake » Sat Jan 12, 2008 2:10 pm

As a pilot study it is still useful to help with designing a future survey to address questions like how many sites and how many visits per site. Also, it can potentially identify covariates that may be predictors of occupancy and even though they are uncertain in the data that you have that may help you design a future survey. If for instance they are habitat features you could consider stratifying sites to provide maximum contrast if your goal is to relate occupancy to habitat.. Hope this helps.
jlaake
 
Posts: 1417
Joined: Fri May 12, 2006 12:50 pm
Location: Escondido, CA

Postby bgerber » Sun Jan 13, 2008 3:54 pm

Thank you Dr. Laake. You have been most helpful.

Sincerely, Brian Gerber
bgerber
 
Posts: 24
Joined: Wed Dec 12, 2007 1:01 pm


Return to analysis help

Who is online

Users browsing this forum: No registered users and 12 guests

cron