Unusual encounter history problem

[HenSH]

Hi,
So I am trying to setup a new CJS model. The encounter history is rather large. It has 84 individuals and 262 encounters. When I go to input the encounter history .inp file I get a question

"Encounter history problem: ....contains only 84 semicolons. is this a proper mark encounter history file?"

If I click ok it opens the results browser. However when I go to generate a model I get another error

"Error - encounter history was too short for the number of occasions specified."

Here is the interesting part. If I reduce the number of encounter histories in the .inp file to the last 83, then it works fine. As soon as there are more encounters than individuals (than semicolons) I get the error.

Any ideas? Thank you.

[darryl]

Have you triple checked your input file? Sounds like it could be an error in the first line.

[HenSH]

I checked. It took me awhile to figure out that as soon as there were less encounter histories than individuals it would work. So I have produced many input files. All the ones over fail, all under work.

[cooch]

Have you triple checked your input file? Sounds like it could be an error in the first line.

The other 'usual suspect' is what you used to 'edit' the .inp file in the first place. When there are 'strange character' errors/warnings from MARK, much of the time it turns out the .inp file was created using Word or some such. may not be the case here, but worth reminding everyone...from the MARK FAQ:

9(b) 'garbage in, garbage out' - the input file - many of the problems people have getting started with MARK can ultimately be traced back to problems with the input (.inp) file. You are strongly encouraged to spend a lot of quality time reading the documentation about data formatting before you do anything else. Also, as a general suggestion, avoid - as in 'like the plague' - using Windows Notepad (or, even worse, Word) to do much of anything related to building/editing .inp files. Do yourself a favor and get yourself a real ASCII editor - there are a number of very good 'free' applications you can (and should) use instead of Notepad (e.g., Notepad++, EditPad Lite, jEdit, and so on...)

Also, unless the encounter history consists of '0' and '1', for a CJS model, MARK might choke, and truncate the string before it gets to the end. It is annoyingly easy to mistakenly use a capital O, rather than a 0.

[cooch]

I checked. It took me awhile to figure out that as soon as there were less encounter histories than individuals it would work. So I have produced many input files. All the ones over fail, all under work.

Ah -- now I get it. This 'error message' has been reported before (if you search for 'semicolon'), you'll find several discussions. For example, see my reply to the following:

viewtopic.php?f=1&t=1799&p=5339&hilit=semicolon#p5337

(although I'm not sure if it occurs when number of lines in .inp file is < number of occasions, of when number of individuals < number of occasions. I suspect the former. I'll work with Gary to find out.

[HenSH]

Thanks for directing me towards the other post. Still no dice though. Mark is working fine, the input files are done in notepad and are correct. Proof of this is that the input files with more semicolons than the number of occasions working perfectly. Just seems like a bug.

[cooch]

Thanks for directing me towards the other post. Still no dice though. Mark is working fine, the input files are done in notepad and are correct. Proof of this is that the input files with more semicolons than the number of occasions working perfectly. Just seems like a bug.

Gary confirms that the warning is just that, a 'warning'.

This warning is just to alert the user that there is a possible problem -- data seem pretty sparse. But, if the input file is correct, not a problem. Just having the number of individuals less than the number of occasions is not necessarily a problem, but is a crude check.

In other words, you can simply hit the 'OK' button, and proceed - at your own risk (i.e., with data so sparse, relative to the number of occasions, you're going to have significant challenges fitting any number of models).

[HenSH]

Thank you for that last reply. I guess what I don't still understand is why when I shorten the number of encounters to less than the number of individuals MARK lets me fit models? Why would creating an even sparser dataset enable model fitting?

[cooch]

Thank you for that last reply. I guess what I don't still understand is why when I shorten the number of encounters to less than the number of individuals MARK lets me fit models? Why would creating an even sparser dataset enable model fitting?

When you have fewer encounters than individuals, MARK guesses you might have a reasonable data set to proceed - no warning. When you have fewer individuals than occasions, then MARK suspects (probably correctly) that you don't have enough data to do much of anything, and issues a warning, which you're free to ignore. In either case, you can fit models. MARK doesn't stop you from being silly, only warns you that you might be heading down a 'dangerous' path.

[HenSH]

My point is it appears that when there are more encounters than individuals it is stopping me fit models. When I try to fit the model is when I get the

"Error - encounter history was too short for the number of occasions specified."

And MARK crashes.