TT

[bear16]

I am interested in including TT in my models since it has worked well in my abundance models in MARK. I noted that the overview for secr indicates that a "timecov" argument may be used. However, I am unfamiliar with how to set that up and then code for it. Would someone be willing to share code where they have included TT so that I can see how to add it to my own models?

Thank you!

[murray.efford]

Sorry, but can you please tell me what TT means save me the trouble of looking up MARK?
Murray

[bear16]

Of course. TT is a quadratic model of time. When I code for it in MARK, in the design matrix, I need to include two columns. This would be for five sessions:

T TT
1 1
2 2
3 4
4 8
5 16

Would something like this work?

secrsexandTT <- secr.fit(myCH, model = (g0 ~ g+T), groups = 'sex', buffer=6000, timecov=c(1,2,4,8,16))

Thank you again!

[murray.efford]

I see. Strictly, your code won't work as you intend because in 'secr' T refers to the 'built-in' covariate 1,2,3,4,5, and your model formula doesn't actually use the quadratic variable in timecov. For a true quadratic model you could use

Code: Select all

secrsexandTT <- secr.fit(myCH, model = (g0 ~ g+t1+t2), groups = 'sex', buffer = 6000,
    timecov = data.frame(t1 = 1:5, t2 = (1:5)^2))


(note the need to include both the linear and squared values as in the MARK design matrix).
Murray

[bear16]

Thank you! That is great!