Hi all,
I try to compute goodness of fit tests for a multistate model in a JS model contexte. All the other tests (3G.SM,3G.SR, M.ITEC and M.lTEC) were not significative. But U-care fail to compute the WBWA test statistics. Looking to the detail, it seems that the individual histories are too 'original' or too 'sparse' to allow the completion of any contingency table, most of them were full of 0!
However, memory is suspected because my data is composed of individual history of birds yearly recorded in 2 wintering areas. The birds are faithfull to their wintering areas and the aim of this model is to estimate recruitment in the 2 wintering strategies.
There are 6 states in this model, but I reduced it to 4 (2 age classes x 2 sites) since the first state is before recruitment and the last state, dead, are never observe. Even in this case it is not possible to compute WBWA test.
Any suggestions are wellcome to explore memory in this data.
Many thanks in advance.
Fred