Trouble running robust models with eh including "."

[dkwalkup]

Hello,

I am trying to run a robust model for a data set with multiple primary sessions, some with 7 secondary sessions, and some that only have 5 secondary sessions. To account for the difference, I have included two "." at the end of each short session. I can load the .inp file and set up the .dbf file fine, but when I try to run a model, I get the error: -- Encounter history must consist of only these characters: '10'. And in the notepad output it says: The character '.' was found in the encounter history.

Which makes sense, because I put them in there. However, In May this year I was at a short MARK course in VA and was able to run multiple models using a previous version of this .inp file with no '.' errors. The one I am trying to run now is a similar version, with this summer's trapping data added to it.

After double checking my current file and not finding any errors, I started a new analysis and set it up using my .inp file from May - which has not been modified at all since I set up that analysis - and now that file is giving me the same error as my new file. But, if I open the .dbf file from my May analysis with the different models I ran, and try to run another model, it runs the model with no errors, even through there are still "." in the encounter histories.

I double checked and reloaded MARK to make sure I was working from the most current version and went through the whole process again and I am still getting the same error. Does anyone have any advice on how to resolve this error?

Thanks! - Danielle

[cooch]

Hello,

I am trying to run a robust model for a data set with multiple primary sessions, some with 7 secondary sessions, and some that only have 5 secondary sessions. To account for the difference, I have included two "." at the end of each short session. I can load the .inp file and set up the .dbf file fine, but when I try to run a model, I get the error: -- Encounter history must consist of only these characters: '10'. And in the notepad output it says: The character '.' was found in the encounter history.

Which makes sense, because I put them in there. However, In May this year I was at a short MARK course in VA and was able to run multiple models using a previous version of this .inp file with no '.' errors. The one I am trying to run now is a similar version, with this summer's trapping data added to it.

After double checking my current file and not finding any errors, I started a new analysis and set it up using my .inp file from May - which has not been modified at all since I set up that analysis - and now that file is giving me the same error as my new file. But, if I open the .dbf file from my May analysis with the different models I ran, and try to run another model, it runs the model with no errors, even through there are still "." in the encounter histories.

I double checked and reloaded MARK to make sure I was working from the most current version and went through the whole process again and I am still getting the same error. Does anyone have any advice on how to resolve this error?

Thanks! - Danielle

There is no requirement for RD models that you have the same number of secondary sessions over primary sessions. And thus, no need to use '.' in the EH to 'balance things out. Examples in the book generally have the same number of secondary samples per primary sample, but that is an artifact of making the simulations used to generate the example data as simple as possible. Further, since you're using secondary samples within the primary sample for 'closed population' modeling, the notion of using a '.' doesn't make much sense (which is why it isn't supported).

So, I surmise you've made the problem more complicated than it needs to be.

[dkwalkup]

There is no requirement for RD models that you have the same number of secondary sessions over primary sessions. And thus, no need to use '.' in the EH to 'balance things out. Examples in the book generally have the same number of secondary samples per primary sample, but that is an artifact of making the simulations used to generate the example data as simple as possible. Further, since you're using secondary samples within the primary sample for 'closed population' modeling, the notion of using a '.' doesn't make much sense (which is why it isn't supported).

So, I surmise you've made the problem more complicated than it needs to be.

Thanks for the quick reply. I'll go back through my input file, I appreciate the help!

[gwhite]

Which exact RD data type were you running, because you should be able to have dots in the encounter histories? But as Evan says, you can specify unequal lengths for the primary sessions.

[dkwalkup]

Which exact RD data type were you running, because you should be able to have dots in the encounter histories? But as Evan says, you can specify unequal lengths for the primary sessions.

I realized yesterday afternoon that I was trying to run them in the 'Full Likelihood p and c' data type, whereas in May I was running them in the 'Huggins' p and c' data type. When I changed the model type to use Huggins' I was able to run a model using my updated .inp file with the dots.

[bjimenez]

Hello:

At this moment I am receiving the very same error message, even though my first attempts (some weeks ago), using the same .inp file worked well. I re-wrote the encounter history to make sure there were no " " in it; I also re-installed Mark earlier today and then I also switched from the full likelihood to the Huggins model but I still get the same error message. My current dataset includes four primary sessions with unequal secondary sessions and a different time interval between primary sessions 2 and 3.

As a final note, the error message in the notepad file includes:

...The character ' ' was found in the encounter history.

ERROR -- Encounter history must consist of only these characters: '10'

[bjimenez]

Hello:

At this moment, I am receiving the same error message when I try to run the RD. My current dataset includes four primary sessions with unequal secondary sessions and a different time interval between sessions 2-3. The error message includes the following statement:

The character ' ' was found in the encounter history.

ERROR -- Encounter history must consist of only these characters: '10'

My first attempts (some weeks ago) worked well and I am still using the same .inp file. I re-wrote the encounter history to make sure there were only 1s and 0s in it; I also re-installed MARK and then I also switched from the full likelihood to the Huggins model but I still get the error message. Your help will be much appreciated.

[bjimenez]

Hello:

I have just found the reason for the error message. It is just that I was using a file with pooled encounter histories, instead of the file with all of the secondary and primary sessions.