www.phidot.org

I am modeling a different first week capture probability compared with subsequent weeks (1,0,0,0,0,0). How do I obtain estimates of g0 for week 1 and weeks 2-6? Thanks!
Joe

Using the 'captdata' dataset that has 5 occasions (times):

Code: Select all

 timecovars <- data.frame(firstweek = factor(c(1,2,2,2,2)))
 secr.fit(captdata, model = g0~firstweek, timecov = timecovars)


The levels of the firstweek factor can be anything you like - c('one','two','two','two','two') would work just as well. The g0 estimate displayed when the model fit is printed will be for the first week, by default.
Murray

Sorry Murray; I should have made myself clearer. I already fit the model with covariates (1,0,0,0,0,0) but I want to get occasion-specific estimates of g0 (1 estimate for week 1 and 1 estimate for week 2-6). I should probably just do the linear algebra but was hoping there was an easier way to go about it.

The reason I am doing this is I suspect a positive site-specific behavioral response and am worried that our subsampling protocol results in some animals whose 1st capture (according to the CH) is actually a recapture of a previous sample that was not genotyped, leading to a potential bias in the estimate (Augustine et al. 2014, Methods in Ecology and Evolution). I think the first week should be relatively unaffected because we genotyped all the samples that week and there should be no site-specific behavioral response. However, we subsampled weeks 2-6 so I want to estimate g0 for the first week and use that estimate for the other occasions to see if my estimates are biased. Can g0 be a fixed term? Thanks.
Joe

I meant to include this in my earlier reply.

Code: Select all

timecovars <- data.frame(firstweek = factor(c(1,2,2,2,2)))
fit <- secr.fit(captdata, model = g0~firstweek, timecov = timecovars)
nd <- data.frame(firstweek = factor(c(1,2)))
predict(fit, newdata = nd)

If your model includes other covariates then you'll need to include them as columns in nd. e.g. bk = c(0,0) for the base level of a site-specific learned response.
Murray