Analysis of continuous data

questions concerning analysis/theory using program PRESENCE

Analysis of continuous data

Postby dmh1454 » Sun Dec 29, 2019 6:57 pm

After running my single season, single species models, my top model came out as psi(Channel), p(Day) but I am having trouble interpreting the psi(Channel) portion. This variable is the width of channels surveyed. After normalizing my data, the raw data comes out anywhere between -0.65 and 1.69. Looking at the parameters of my model, I see the a1 estimate is 3.15 with a SE of 1.31 while the a2 (column I put channel in) estimate is 2.36 with a SE of 2.15. Looking at this data, I believe a1 has a higher occupancy rate than a2. However, I do not know if a1 shows the occupancy of wider or narrower streams. I looked through user manuals and examples but most of them show constant (1 and 0) data. Are there any resources I can look at that go into detail about analyzing continuous data or can someone help walk me through it?
Posts: 7
Joined: Fri Aug 02, 2019 12:35 pm

Re: Analysis of continuous data

Postby jhines » Sun Dec 29, 2019 11:50 pm

From your description, I'm guessing that channel is a continuous covariate and your design matrix for psi looks like this:

Code: Select all
     a1     a2
psi  1    channel

So, a1 is the intercept, and a2 is the effect of channel. The value of occupancy for a site will be:

psi(i) = exp(a1+a2*channel) / (1+exp(a1+a2*channel)).

A1 is the logit of occupancy when channel=0. Any sites in your data with channel=0 will have occupancy =

psi = exp(a1) / (1+exp(a1)) = 0.959. You can look at the 'real' estimates of psi for those sites to verify this.

Since A2 is positive (2.36), sites with a value of channel > 0 will have a higher occupancy estimate than sites with channel=0. Sites with channel < 0 will have a lower occupancy estimate than sites with channel=0. A plot of channel vs psi would be an line which increases as channel increases.

So, you don't interpret a1 and a2 as occupancy estimates. They are the estimated coefficients of the formula to compute occupancy at each site.
Posts: 473
Joined: Fri May 16, 2003 9:24 am
Location: Laurel, MD, USA

Return to analysis help

Who is online

Users browsing this forum: No registered users and 1 guest