Scheme A Capture Histories?

[Onoufrios]

Hello all.
I am trying to perform a goodness of fit test in a POPAN formulation analysis with 2 groups in MARK. I am only able to perform the Release GOF test, but when i try it and look at the output, after the tables for the observed recaptures for the 2 groups and the sum of groups there is the line 'Data type is Scheme A Capture Histories' and then the results for Test2 and Test1.

a) What does the 'Data type is Scheme A Capture Histories' refer to and why isn't Test3 performed?
(I was able to perform the GOF test using UCARE but i would like to know if there is something wrong with my data or maybe my .inp file.)

b) Because of low recapture rates i have rather sparse data, so i suspect that a full time dependent model won't fit the data. In order to assess the fit of simpler models, should i (i) try GOF using CJS or (ii) try the Link-Barker formulation as suggested in another post and then use the results (and c hat) in my POPAN analysis (or neither/ something completely different)?

The first lines of my input file, ten occasions, 2 groups, few recaptures..:

1000001100 0 1;
1000000000 0 1;
1000000000 0 1;
1000000000 0 1;


Thanks in advance and congrats for the really useful posts in this forum

[apinya]

Hello,
I either have this problem. Getting the line "Data type is Scheme A Capture Histories". My data is quite similar as you--- few recaptures. this may be the case.
hope someone give us any suggestion how to deal with this.
sorry for no answer.
best regards, apinya

[cooch]

Hello,
I either have this problem. Getting the line "Data type is Scheme A Capture Histories". My data is quite similar as you--- few recaptures. this may be the case.
hope someone give us any suggestion how to deal with this.
sorry for no answer.
best regards, apinya

No worries. Scheme A (and scheme B) are 'relics' (of a sort) of the 'blue fish monoigraph', and program RELEASE (which was described in depth for the first time in the book). The two schemes (A and B) reflect 'partial' encounter histories (as opposed to the full history that is generated if each individual has a unique mark). Scheme A and B refer to sampling protocols that apply to batch marked populations. The schemes are characterized by an initial mark, a second mark upon first recapture, and removal on next (3rd) encounter. Scheme A differs from Scheme B in that in A, the second mark is put on regardless of when the second encounter occurs (relative to the release occasion), while in scheme B, only individuals encountered on the second occasion after the release occasion are given a second mark. The relative merits of these different schemes are discussed at length in chapter 1 in 'the blue book'.

However, for your purposes, if RELEASE sees an encounter history that 'looks' like it could be generated under scheme A (which would occur if you have relatively few recaptures -- in scheme A and B an encounter history can have no more than 3 total 'encounters'), it reports that it thinks you are in faxt using that sampling scheme -- even if you're not. More importantly, this doesn't affect what you're trying to do (in other words...you're safe in ignoring it

[Onoufrios]

Thank you for the informative reply!
I will ignore it, but MARK won't.. i still can't perform a Test3 on my dataset!
Also, I would be grateful if you could provide some insight on the second question
Thanks again.

[cooch]

Thank you for the informative reply!
I will ignore it, but MARK won't.. i still can't perform a Test3 on my dataset!
Also, I would be grateful if you could provide some insight on the second question
Thanks again.

Short of looking at your data, I don't know, but a likely explanation is...data sparseness. In general, when RELEASE doesn't generate an expected test, that is fairly diagnostic of data sparseness, which often complicates (or renders impractical) the construction of the various contingency tables on which release is built.

[Onoufrios]

O.K. i got the message: as i said, few recaptures!
I would expect a notification of some sort in the Release output saying that Test# couldn't be performed. Oh, well.. this is the least of my worries right now!
Cheers!