Nest survival encounter occasions

[DTI]

Hello,

I am trying to use the nest survival module in MARK to calculate daily survival rate over the course of a study. There are 42 nests, 67 encounter occasions, and 15 individual covariates (though for ease of reading I have only included one in the .inp input below). As expected, I get the error message "The Encounter Histories file <path> only contains 43 semicolons." After clicking "Ok" and creating the MARK Database File for this analysis, I open up the Reduced Design Matrix with only 1 column to construct a Null Model ( {S(.)} ).

The problem is: there are only 65 rows in the matrix, rather than the 66 I would expect it should have, given that there are 67 encounter occasions! When I fill the first (and only) column in the matrix with 1's, the number of rows remains at 65, but the row labels all shift up one (e.g. the first row, previously labeled "1:S" becomes "2:S", the last row, previously labeled "65:S", becomes "66:S").

I double checked that my spacing was correct in the input file, and that there are indeed 67 encounter occasions (e.g. nest /*AMT09080109*/ is first encountered on day 1, and nest /*AMT09290110*/ was last alive and last checked on day 67). In addition, I double checked that the correct fates were assigned to nests (i.e. a successful nest was not given a "1" in the fate column, and that any failed nests had different "last checked" and "last active" dates). Also, this does not appear to be an issue of lack of data, as after quadrupling the size of the input file, the problem still exists (though the warning message about number of semicolons is no longer given). As a final check that it isn't specific to my computer, we've tried this analysis on 3 separate computers and all seem to have the same issue.

I'm not sure if this is being caused by a data issue, a bug in MARK, or (most likely) user error. Any help or suggestions you could offer would be much appreciated!


----INPUT FILE----

Nest survival group=1;
/*AMR09010109*/ 20 41 41 0 1 19.26;
/*AMR09030109*/ 21 40 42 1 1 35.92;
/*AMT09080109*/ 1 15 20 1 1 40.24;
/*AMT09100209*/ 51 65 65 0 1 37.94;
/*AMT09120109*/ 20 44 46 1 1 48.17;
/*AMT09150109*/ 26 43 43 0 1 26.78;
/*AMT09190109*/ 19 48 48 0 1 20.9;
/*AMT09200109*/ 9 12 19 1 1 23.09;
/*AMT09230109*/ 19 36 39 1 1 21.3;
/*AMT09250109*/ 28 41 41 0 1 28.88;
/*AMT09280109*/ 19 45 45 0 1 30.7;
/*AMT09300109*/ 19 28 32 1 1 30.98;
/*AMT09320109*/ 35 55 55 0 1 27.31;
/*AMT09350109*/ 9 35 35 0 1 37.44;
/*AMT09360109*/ 20 35 35 0 1 28.57;
/*AMR09370109*/ 1 39 39 0 1 16.23;
/*AMR09420109*/ 26 29 34 1 1 18.51;
/*AMR09430109*/ 20 43 43 0 1 27.09;
/*AMR09010110*/ 22 44 44 0 1 28.11;
/*AMR09020110*/ 19 41 41 0 1 33.87;
/*AMT09090110*/ 22 28 33 1 1 22.48;
/*AMT09120110*/ 32 40 43 1 1 27.71;
/*AMT09210110*/ 15 17 25 1 1 33.62;
/*AMT09230110*/ 40 40 41 1 1 22.11;
/*AMT09250110*/ 33 36 40 1 1 25.63;
/*AMT09270110*/ 15 41 41 0 1 27.35;
/*AMT09290110*/ 20 67 67 0 1 35.38;
/*AMT09330110*/ 22 42 42 0 1 19.81;
/*AMR09440110*/ 22 37 37 0 1 23.89;
/*AMR09450110*/ 29 41 41 0 1 28.58;
/*AMR09460110*/ 15 44 44 0 1 30.21;
/*AMR09470110*/ 22 44 44 0 1 25.62;
/*AMT10520110*/ 27 34 36 1 1 17.61;
/*AMT10590110*/ 20 41 41 0 1 22.31;
/*AMT10620110*/ 25 28 30 1 1 26.82;
/*AMT10630110*/ 29 48 48 0 1 29.42;
/*AMT10640110*/ 27 50 50 0 1 31.9;
/*AMT10650110*/ 27 49 49 0 1 46.91;
/*AMT10660110*/ 27 51 51 0 1 38.55;
/*AMT10680110*/ 40 48 53 1 1 30.87;
/*AMT10730110*/ 27 50 50 0 1 21.51;
/*AMR10810110*/ 32 59 59 0 1 32.53;

[Rotella]

I don't have any problems with your dataset when I run it in MARK 5.1 (Build 2600). MARK shows 66 possible survival parameters in the PIM and models like S(.) and S(t) can be run.

[DTI]

Thank you for your quick response! I am using MARK 6.0 (build 6002), so I will see if I can find a download for the older version.

[gwhite]

David:
This is a bug in the Visual Objects software that I've not found a work around for. Something about 66 rows in the design matrix causes a problem. What you can do is relabel a column, and I think the last row will then reappear, or at least I think that is the trick I remember to get it back on the screen. The row is present, just not displayed.
Gary

[DTI]

Gary,

Thank you for the clarification! Yes, when I open the real parameter estimates I see that there are 66 estimates given. It seems that once I use the "Copy Value Down" option in the Design Matrix, the first row is no longer displayed.

At first, I noticed that the first real parameter estimate would be different than the rest when I had the 1 covariate in the model (and thus two columns in my Design Matrix), which was confusing because there was no time effect in the model. After your clarification, I see that this is because after I used the "Copy Value Down" option to fill the first column with 1's, I was only able to fill the second column from the second row down (leaving a "0" in the first row of the second column).

As suggested by the manual, if I set up my design matrix in Excel and paste it into the design matrix (without using the "Copy Value Down" option, so as not to lose the first row) I can get around this problem. As long as I have 66 rows in my Excel spreadsheet, it seems to fill all of the rows, even if the 66th row isn't displayed in the Design Matrix.

Thanks again.

Dave