score tests in secr R package

questions concerning anlysis/theory using program DENSITY and R package secr. Focus on spatially-explicit analysis.

score tests in secr R package

Postby howeer » Mon Apr 26, 2010 4:14 pm

Hi,
I wish to perform score tests comparing the fit of a null secr model model (hazard rate detection function) to models with session effects on each of g0 and sigma, and want to be sure I'm using the correct betaindex specification. I think the order of parameters is 1 = g0, 2 = sigma, 3 = z for hazard rate. Beta index numbers should then be as follows, for a data set with 9 sessions?

# SCORE TEST SESSION EFFECTS
st.g0SESS=score.test(g0.sig.fit, g0~session, betaindex = c(1, 10, 11))
st.sigSESS=score.test(g0.sig.fit, sigma~session, betaindex = (1, 2, 11))
#

Thanks,
Eric Howe
howeer
 
Posts: 39
Joined: Wed Jun 21, 2006 10:49 am

Re: score tests in secr R package

Postby murray.efford » Tue Apr 27, 2010 6:58 am

Eric
Yes, if you are using conditional likelihood (hence no density parameter) I think you have that right. I would be interested if this differs from the packaged automatic version. You could verify it by starting to fit the more complex model and counting off parameters from the first one or two likelihood evaluations. (Do note the caveat on score tests: they remain 'experimental' until we can resolve doubts about using the observed information, and information theoretic purists probably gag on them).
Murray
murray.efford
 
Posts: 686
Joined: Mon Sep 29, 2008 7:11 pm
Location: Dunedin, New Zealand


Return to analysis help

Who is online

Users browsing this forum: No registered users and 12 guests