Approaches to handling missing sampling occasions

questions concerning analysis/theory using program MARK

Approaches to handling missing sampling occasions

Postby VanessaSimons » Mon Jan 11, 2021 6:18 am

Hello,

This question concerns a Photo-id study with sampling 2003 to 2008 and 2012 to 2020. There was no funding for search effort 2009 to 2011.

For closed capture models (full likelihood p and c) built with the PIM chart (14.5)
The approach was to 'include a column of 0’s in the encounter histories for the missing occasion. In this case, you need to explicitly set p = 0 for the missing occasion'.

For CJS models
The approach was 'the input file you can simply not include an encounter column in the encounter history for the occasions where sampling did not occur. In this case, you then need to explicitly adjust the sampling interval to account for the missing occasion'

I need to use program CAPTURE (in the knowledge it is 'deprecated' etc)
For program CAPTURE is it possible to take the missing sampling occasions in to account?
Last edited by VanessaSimons on Mon Jan 11, 2021 3:33 pm, edited 1 time in total.
VanessaSimons
 
Posts: 1
Joined: Sun Jan 10, 2021 7:21 am

Re: Approaches to handling missing sampling occasions

Postby cooch » Mon Jan 11, 2021 12:34 pm

Unfortunately, your note has very few of the specifics anyone would need to offer assistance. It isn't clear to me at all what sort of analysis you're considering. You mention closed capture estimation, and open population estimation. So, which is it? One, or both?

In other words, less time summarizing comments about software (CAPTURE), and more on what you're trying to do. If you have missing 'years', it strikes me as entirely unlikely that you're dealing with a closed population. What I suspect you're working with are encounter data collected over a bunch of years, with some missing years. And I also suspect you've been tasked with saying something about abundance. At which point, you're talking about abundance estimation for an open population (I'm guessing). At which point you're simply taking your count statistic and dividing it by encounter probability in some fashion. Missing years are a problem, but can be handled for open populations by methods discussed in Chapter 4.

So, ignore what I'm guessing is a student thesis deadline, ignore the person who has 'ordered you' to use CAPTURE (which hasn't been supported in 10-15 years), and provide more details about what it is you're actually trying to do.

VanessaSimons wrote:Hello,

This question concerns a Photo-id study with sampling 2003 to 2008 and 2012 to 2020. There was no funding for search effort 2009 to 2011.

For closed capture models (full likelihood p and c) built with the PIM chart (14.5)
The approach was to 'include a column of 0’s in the encounter histories for the missing occasion. In this case, you need to explicitly set p = 0 for the missing occasion'.

For CJS models
The approach was 'the input file you can simply not include an encounter column in the encounter history for the occasions where sampling did not occur. In this case, you then need to explicitly adjust the sampling interval to account for the missing occasion'

I need (under command) to use program CAPTURE (in the knowledge it is 'deprecated' etc see below)
For program CAPTURE is it possible to take the missing sampling occasions in to account?
:?:

From searching earlier postings on the forum I appreciate the following about program CAPTURE: :!:
- program CAPTURE is largely deprecated, since closed population abundance estimation in MARK is far more flexible and 'current'.
The only reason I can see for running CAPTURE is if you really must generate estimates using a few models which are available in CAPTURE only.
- more faith in output from MARK than CAPTURE
- accept that MARK is more likely to be correct
- although program CAPTURE is available via MARK, you shouldn't use it (generally) for closed population abundance estimation
- all of ML-based models in CAPTURE can (and should) be built in MARK.
At this point, there is no compelling reason to use CAPTURE (sorry, ease of use does not constitute 'compelling'), and a number of reasons why you should use MARK.
- If you have a bit of time, you really should learn to use MARK, which is the de facto replacement for CAPTURE.
- model selection in CAPTURE was always a bit of a joke, and certainly leads to screwy results over time if you re-select the 'appropriate' model for each session (dataset).


Referring to the book
Chapter 2.3, 4.2.2 (p.4-24)
There are several different approaches to handling a missing sampling occasion – the ‘right’
approach is the one that you find most convenient, and which best suits your purposes...
cooch
 
Posts: 1628
Joined: Thu May 15, 2003 4:11 pm
Location: Cornell University


Return to analysis help

Who is online

Users browsing this forum: No registered users and 18 guests