Delta method advice

[Jochen]

Hi,
from a MARK analysis I have obtained model averaged estimates of monthly survival over a 110 month study period with seasonal variation. I saved the V-C-matrix as dbf, too.
Now I am looking for SEs for the mean survival for the 12 months of the year, and for their product, annual survival.
So, 2 steps are needed, to get from 110 months to 12 months and then to a single year.

I guess I need the delta method and I saw a forum post recommending the deltamethod.special function in RMark, but I am not sure about the details how to apply this here. Maybe somebody has more advice on this, e.g. a worked example?

Any advice is appreciated
Jochen

[egc]

Hi,
from a MARK analysis I have obtained model averaged estimates of monthly survival over a 110 month study period with seasonal variation. I saved the V-C-matrix as dbf, too.
Now I am looking for SEs for the mean survival for the 12 months of the year, and for their product, annual survival.
So, 2 steps are needed, to get from 110 months to 12 months and then to a single year.

I guess I need the delta method and I saw a forum post recommending the deltamethod.special function in RMark, but I am not sure about the details how to apply this here. Maybe somebody has more advice on this, e.g. a worked example?

Any advice is appreciated
Jochen

Mean survival -- section 6.15, chapter 6 of the MARK book.

[Eurycea]

Since that section in the MARK book does not mention the delta method, is it because that method is completely inappropriate, or because the other approaches are better/easier? Say the OP had survival estimates for each of 110 months, but someone wanted the average survival and SE for months that end in a Tuesday or Wednesday. Would the delta method be appropriate here?

[cooch]

Since that section in the MARK book does not mention the delta method, is it because that method is completely inappropriate, or because the other approaches are better/easier? Say the OP had survival estimates for each of 110 months, but someone wanted the average survival and SE for months that end in a Tuesday or Wednesday. Would the delta method be appropriate here?

You can apply the Delta method (trivial to execute) to whatever you want (including the mean of a set of estimates), but, variance will be ~sum of (process+sampling), whereas the random effects approach (either using the method of moments approach - Appendix D, or MCMC - Appendix E) generates correct estimate of process variance only.