detection functions in secr 2.3.1

[howeer]

Hello,
Does "detectfn = 8" still yield the cumulative gamma detection function?
I fit a model with detectfn = 8 and used plot.secr to view the detection function. It looked more like an exponential distribution.

Also, I was unable to fit a model with "h2" affecting both h0 and sigma, which was unexpected based on previous work. When I used detectfn = 0 (half-normal), the shape of the detection function was as expected and I was able to fit the heterogeneity model.
Thanks,
Eric

[murray.efford]

Eric
Yes, both exponential (2) and cumulative gamma (8) detection functions still work as far as I know. Try this:

Code: Select all

fit.2 <- secr.fit(captdata, detectfn = 2)
fit.8 <- secr.fit(captdata, detectfn = 8)
plot(fit.2)
plot(fit.8, add = TRUE, col = 'red')


(and you should have noticed that detectfn = 8 causes an additional parameter to be fitted).

The shoulder on a fitted cumulative gamma function may shrink to near zero, giving the effect you describe.

Your other issue seems to be data-specific. Mixture models often give trouble, and adding the extra parameter may be the straw that breaks the camel's back.

Murray