Dataframe type for a Multistate model?

Posted:
Thu May 30, 2019 2:33 pm
by eveconnection
Hello,
I am having a 7 years dataset from birds to be used with a multi-state model (4 different states, 1 of them unobservable) but I am having trouble with finding an example of the right type of dataframe I need to create in order to run the MS-model. Does anyone have an example of this and also a MS-model script they can share so I can develop mine?
Many thanks in advance,
Eve
Re: Dataframe type for a Multistate model?

Posted:
Thu May 30, 2019 2:45 pm
by jlaake
- Code: Select all
library(RMark)
?mstrata
will give you an example of a multistate model. With an unobservable state you need to specify strata.labels argument in process.data step.
Have you read Appendix C in Cooch and White and the Workshop Notes for RMark? The link for the latter is shown when you type library(RMark)
This is RMark 2.2.7
Documentation available at
http://www.phidot.org/software/mark/rma ... tation.zipAll dataframe structures are the same for RMark. For most models, all that is required is a field named ch which is a character string. Everything else in the dataframe is optional. Models like nest success and others do have a different structure. But a MS model only differs from models like CJS etc in the values of ch.
Re: Dataframe type for a Multistate model?

Posted:
Thu May 30, 2019 3:10 pm
by eveconnection
Thank you so much! I was thinking the same, that the df needed to be the same for most of these models, but just needed somebody to confirm this.
Now I have another question related to the dataframe. Do I need to specify my states in the df? Like instead of having only 0 and 1, do I need to have 0,1,2,3,4 depending in which state my birds are?
And yes, I read C.17 in the Appendix C. Thanks for the Workshop notes, I wasn't having that one.
Many thanks.