Single Season, Multi-Method

[jackieguzy]

Hello,

I am wondering about setting up the detection design matrix for a multi-method model. We have 12 surveys at 21 sites, using two methods of detection for frogs (6 active searches and 6 frog call surveys). One model I want to run is one that allows detection to vary between the two methods, and also with a sample-specific covariate (rainfall). Example below. Is the intercept column (b1) necessary for this situation? I believe I have set up the array below so that active searches (b2) are treated differently than frog call surveys (b3).

b1 b2 b3 b4
1-1 1 1 0 z-score rain
1-2 1 0 1 z-score rain
2-1 1 1 0 z-score rain
2-2 1 0 1 z-score rain
3-1 1 1 0 z-score rain
3-2 1 0 1 z-score rain
4-1 1 1 0 z-score rain
4-2 1 0 1 z-score rain
5-1 1 1 0 z-score rain
5-2 1 0 1 z-score rain
6-1 1 1 0 z-score rain
6-2 1 0 1 z-score rain

[darryl]

Hi Jackie,
A couple of things.

First, no you don't need have all of the first 3 columns so could drop the intercept, or your could drop one of the other columns which would mean you could interpret the beta parameter as an estimate of how different the detection rate is for the 2 methods.

Second, by having the rain covariate in a single column, you're saying the magnitude of the effect of rain on detection is the same for the 2 detection methods. If you want it to be different then you're going to want to add another column and have the rain covariate in every second row (and 0 otherwise).

Third, you should have a space in a covariate name. It works the first time but PRESENCE will get confused when you reopen the project.

Cheers
Darryl

[jackieguzy]

Wow, thanks so much, Darryl! I really appreciate it!