Nest survival Delta-method very low CI's

[Miina Kovanen]

Hi!

I have calculated the seasonal survival probabilities in my data by multiplying the weekly survival estimates obtained by the best supported model. I think I managed to estimate the variances of the seasonal survival estimates using the delta-method, but the variances and 95% confidence intervals get very small - and I got sceptical. Maybe I've misunderstood something? Below is some introduction to my study, and explanation how I'm estimating the CI's.

Sorry if the question is very simple - I'd appriciate any help on this, only few people around here have any idea on these methods.

MY STUDY
I'm analysing 3-year radio telemetry data on survival of adult Black Grouse and self-learning to use program MARK Nest Survival Model. The data includes the histories of 253 individuals, marked in 3 cohorts, followed on monthly basis troughout the year. The data is "ragged" and includes censored cases as well. Thus, I'm applying Nest Survival model, and have coded the data in weekly intervals, to be able to analyse survival probabilities of interesting but uneven ecologically determined seasons.

We want to find out if the effects of age class (young or older), season, and year explain the survial rate. The model Age Class+Season+Age Class*Season got rather clear support over the other relevant models. This also makes sense in ecological terms.

ESTIMATING THE SEASONAL SURVIVAL AND 95%CI's
I obtained the seasonal survival probabilities of the age classes by multiplying the weekly survival estimates of the weeks included in each season. (Since the model assumes that the survival may vary between the seasons, but stay constant within each season, I simply raised the estimate for the first week of each season to the power determined by the number of weeks included in the season). Using the delta-method I got the estimates of the seasonal SD, which - to my understanding of statistical english - also average the seasonal standard errors. I hope this is correct? Then I calculated the CI's by the seasonal survival estimate +- estimated SE*1,96.

Many thanks in advance, this forum has been a great help to me!
Miina