Missing data for some individuals

questions concerning analysis/theory using program MARK

Missing data for some individuals

Postby alipichun » Wed Jun 27, 2018 1:14 pm

I am analyzing data from marked adult gannets with re-sighting to obtain the adult survival rate.
Data is available over the past 8 years. During those 8 years new individuals have been tagged. I have therefore data for 8 years for some individuals but for others I have data for less years. Is it reasonable to include individuals without data available for the 8 years?, and If it is, how do I insert the data in the program? Should I count the missing years as 0 or as 1?
alipichun
 
Posts: 3
Joined: Wed Jun 27, 2018 11:55 am

Re: Missing data for some individuals

Postby jlaake » Wed Jun 27, 2018 1:24 pm

I'll try to be gentle here but you need to spend sometime reading Cooch and White and learning about capture-recapture before you think about entering data. It doesn't sound like you understand the basic concepts of capture-recapture and capture histories. But to answer your question your data might look something like

ch
10001001
00100111

For the first capture history the bird was released in year 1 and then resighted in years 5 and 8. The second bird was released in year 3 and then resighted in years 6,7,&8.

Now go read Cooch and White book on MARK.

--jeff
jlaake
 
Posts: 1417
Joined: Fri May 12, 2006 12:50 pm
Location: Escondido, CA

Re: Missing data for some individuals

Postby cooch » Wed Jun 27, 2018 1:31 pm

jlaake wrote:Now go read Cooch and White book on MARK.

--jeff


Yup -- all 1200 pages. By tomorrow. ;-)

In all seriousness, I echo jeff's comment. You need to learn some of the basics, first. I'd suggest reading chapters 1 -> 4 in the MARK book, to start. That will get you pointed in the right direction. Then, depending on how sophisticated you want to get, Chapter 6, and then, beyond that, that nature of your data (and the questions you're interested in) will determine what you'll need to read.

But, absolutely -- go through 1 -> 4 (and, by go through, that means sit down, with those chaptrs, read them, and work through the exercises). Then, and only then, start fussing with your own data.
cooch
 
Posts: 1628
Joined: Thu May 15, 2003 4:11 pm
Location: Cornell University

Re: Missing data for some individuals

Postby alipichun » Wed Jun 27, 2018 1:33 pm

Ups! I meant If I should write it as 0 or as "dot"
alipichun
 
Posts: 3
Joined: Wed Jun 27, 2018 11:55 am

Re: Missing data for some individuals

Postby jlaake » Wed Jun 27, 2018 1:37 pm

If you are talking about years prior to release, you can use . or 0 because for a CJS model, nothing before the first 1 in the capture history matters. Go read ch 1-4 as Evan suggested.

--jeff
jlaake
 
Posts: 1417
Joined: Fri May 12, 2006 12:50 pm
Location: Escondido, CA

Re: Missing data for some individuals

Postby cooch » Thu Jun 28, 2018 9:36 am

jlaake wrote:If you are talking about years prior to release, you can use . or 0 because for a CJS model, nothing before the first 1 in the capture history matters. Go read ch 1-4 as Evan suggested.

--jeff


Jeff is, of course, entirely correct. However, I generally suggest that

1\ if you sampled, but didn't detect, use a zero.

2\ if you didn't sample, then you can use a 'dot'. I say *can*, because (a) not all models in MARK support the 'dot' notation, and (b) you need to be sure you know what you're doing. Sometimes using a '0', and then explicitly fixing p to 0 forces you to really think - carefully - about what a missing sample is/does to your analysis. This is covered in various places in the book, but the key 'treatment' is in the -sidebar- starting on p. 24 of chapter 4

For simple CJS analysis, point is somewhat moot. The likelihood is conditioned on events following the first capture, so whether you use a 0 or dot won't much matter. This being Jeff's point.
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: Google [Bot] and 11 guests

cron