Trouble with GOF testing

[Luc te Marvelde]

I got trouble in getting the results of the RELEASEtest within Mark. When i start Release within MARK, i get a black window and then it wants to show me the results. 1st i get a message that the file is to big to be shown in notepad, when i want to open it in word, i get this message:

c:\...\mrk5289y.tmp
Cannot find this file
Please verify that the correct path and file name are given



What can i do to see the results? i already tried making an empty file with this name, but than the requested filename simply changes (the 4 numbers change)...


Who knows what to do?



and a second question:

when i have groups as well as covariates, should my .inp look like this:


/* name */ encounterhistory group1 group2 group3 cov1 cov2 cov3;


or do i have to put the '1' in after the encounterhistory?



Thnx in advance,

Luc te Marvelde
Masterstudent Biology
RUG, Groningen, The Netherlands

[Luc te Marvelde]

my inp-file looks like this:

/*AG.26001*/ 1000111000110000000111 1 0 0 0 0 0 0 0 ;
/*AG.26002*/ 1000111000000000000100 1 0 0 0 0 0 0 0 ;
/*AG.26003*/ 1001111100101000000000 1 0 0 0 0 0 0 0 ;


As you can see, i got 22 occasions and 8 groups.


i tried running RELEASE as standalone. i couldnt get my data in, but i saw that max 20 occasions can be used and max 4 groups..maybe this has something to do with my error?
just wanted to provide you with a little bit mroe info that you might need to answer my 1st question.

[darryl]

Hi Luc
How many marked individuals do you have? I would guess that you're going to need at least several hundred to get a reliable GOF result....
Cheers
Darryl MacKenzie

[Luc te Marvelde]

atm i got 390 marked individuals.. but this is just a test dataset... the final set will have about 600-700 animals probably!

[jhines]

Hi Luc,

I suspect that your problem is due to too many groups and occasions for RELEASE. I recompiled my version of RELEASE to allow up to 40 occasions and 8 groups. You can run RELEASE on your data by going to:

www.mbr-pwrc.usgs.gov/software/release.html

and pasting your input data into the input box. You'll need to add a couple of lines to the beginning like the example.

Jim Hines

[Luc te Marvelde]

wow, thank you very much! This site is very usefull! I bet loads of studies suffer the same 'problem' of exceding 20 occasions and 4 groups.

Maybe this site should be noted somewhere permanent on this site!

[cooch]

wow, thank you very much! This site is very usefull! I bet loads of studies suffer the same 'problem' of exceding 20 occasions and 4 groups.

Maybe this site should be noted somewhere permanent on this site!

It is - the Patuxent Software Archive is listed on the home page for this forum. This forum is not exclusivelyabout program MARK - although clearly the MARK segment(s) of the forum generates the most traffic. The forum is intended to facilitate discussion about analysis of data from marked individuals in general - and, the Patuxent group (Jim Hines) has created a lot of software - in some cases, applications which are also available (in intent) in MARK, but there are also a large number of programs in the Patuxent archive that have very specialized purposed, which aren't available in MARK, and several more which have been heavily customized by Jim - for example, the recompilation of RELEASE for more groups.

[Luc te Marvelde]

I would like to do this RELEASE-GOF analysis. Since i got more then 4 groups i'll do it using the internet site above. In my encounterhistory i use different time intervals... Question: can i also use different time intervals in RELEASE (using the site)? What do i have to type above my encountrehistory to make it work? (the 11th and 12th encounters have a timeinterval three times as big as the others)



Here is what my encounterhistory file looks like:
ringnumber LLLLLL 8 groups 3 covariates

/*AG.26001*/ 0110100011001110011 1 0 0 0 0 0 0 0 1540 302 191 ;
/*AG.26002*/ 0111000000001000000 1 0 0 0 0 0 0 0 1690 295 199 ;
/*AG.26003*/ 0111100010100000000 1 0 0 0 0 0 0 0 1650 290 195 ;
/*AG.26004*/ 0111000001001100000 0 0 0 0 0 1 0 0 1630 265 194 ;

[jhines]

Since you are trying to fit the most general model, the interval lengths are irrelevant. If you were interested in survival rates for a model where survival was assumed constant per unit time, then the intervals would be important. In the case where the interval lengths are three times as big as the others, the survival rate for that interval will be lower than if it were the same as the others (since animals would have to survive 3 times as long). GOF, however, is not affected.