Known Fate encounter histories do not match occasions

[kernicholson]

Known fate capture histories are to consist of two-digit pairs. So, if we capture 3 times, this would mean 6 digits correct? So, when setting up your initial input in MARK, and it asks for Encounter History - is it actually asking for Encounter Intervals? Or really the Encounter History?

I tried it assuming that encounter history was the same, that I put in 3. However, this now is estimating 3 survival parameters. I do not have 3 survival parameters do I? Everyone was captured at the same time in June, survival check in August (3 months later) and then in April (8 months after August check). So, the survival intervals = 2. Yet, if you indicate 3 encounter histories it asks for 3 separate time intervals and as I said, estimates 3 survival parameters.

In proceeding I attempted by indicating my time intervals = 1, 3, 8 (although now I think maybe it should be 0, 3, 8 if that is even at all possible). But, regardless, I get survival estimates of 0.98 for S1, 0.95 for S2 and 1 for S3...S3 also has 1 for the confidence intervals. So survival for S3 doesn't make sense to me.

What's going on? Do I have a 6 digit capture history but tell MARK my encounter history = 2 with the time interval of 3 months and 8 months? Or do I create a 4 digit capture history somehow?...which is impossible for an individual that was seen in June, not in August but seen again in April. As that history should look something like 110010 or would it be 100010?

[cooch]

Based on your question, I'd strongly suggest you read - and work through - chapter 16 (which covers known-fate models).

[kernicholson]

I have been doing that exact thing, thus the reason for me even trying to post the question. Myself and 2 other people actually. All of us are no farther in figuring out what the answer is. In the known fate chapter they do not discuss actually programming the input files in detail and, there is no discussion covering setting a difference in time between capture sessions (likely that was chapter 2 if I remember correctly). But again, when they tell you that you can account for difference in length between capture sessions, the example just explains the reason for the calculation and it informs us this can be done. It doesn't quite explain the answer to my question. In chapter 16 I do recognize that they indicate 8 encounters, and create a history of 16 two-digit pairs, but the information in that chapter is lacking particularly when it comes to interpretation of results and implications of the choices made to generate those results.

[cooch]

I have been doing that exact thing, thus the reason for me even trying to post the question. Myself and 2 other people actually. All of us are no farther in figuring out what the answer is. In the known fate chapter they do not discuss actually programming the input files in detail and, there is no discussion covering setting a difference in time between capture sessions (likely that was chapter 2 if I remember correctly). But again, when they tell you that you can account for difference in length between capture sessions, the example just explains the reason for the calculation and it informs us this can be done. It doesn't quite explain the answer to my question. In chapter 16 I do recognize that they indicate 8 encounters, and create a history of 16 two-digit pairs, but the information in that chapter is lacking particularly when it comes to interpretation of results and implications of the choices made to generate those results.

The assumption is that you've worked through the 'core' chapters first: 1 -> 7. If you had, you'd have worked through Chapter 2, which is about formatting input files. You'd know what an LDLD format is, amongst other things.

And, if you think the chapter is lacking, feel free to write your own.

[B.K. Sandercock]

You have three occasions, June, August and April.
Each occasion would have a pair LD values in the encounter history for a total of 6-digits: LDLDLD
n occasions gives n-1 intervals.
Three occasions yields two intervals and two estimates of survival.
The unequal duration of the two intervals can be specified when you bring the data into Mark.
If you specify 3 and 8, you will get monthly estimates.
Depending on the period survival rates, the monthly estimates might be close to the boundary of one
I suggest you might scale by the length of the shortest interval instead.
If you specify 3/3 = 1 and 8/3 = 2.7 you will get 3-month estimates of survival for both intervals.