Case selection and encounter history file

[JonL]

Can I use covariate codes to provide case slection capabilities? E.g., can I run an analysis only on males in a file and then on females? OR would I have to provide two different files? In that case, would I have to code females as "0 1" and males as "1 0" (as seemed to be done in "A gentle introduction", page 2.2.2. Or , can I code females as "1" and males as "2" (in the same column). . And possibly juvenilse as "3" in thesame column. Can I enter individual weight (physical weight, gramems, that is) in one column with the idea to only run the analysis on animals above a certain weight? I guess this can be found somewhere in the documentation but it is quicker to ask, hoping for a quick and simple answer. These are the kind of things you need to know befor preparing a big data file.

Jon Loman

[cooch]

Can I use covariate codes to provide case slection capabilities? E.g., can I run an analysis only on males in a file and then on females? OR would I have to provide two different files? In that case, would I have to code females as "0 1" and males as "1 0" (as seemed to be done in "A gentle introduction", page 2.2.2. Or , can I code females as "1" and males as "2" (in the same column). . And possibly juvenilse as "3" in thesame column.

No - if you want to run an analysis on males and females separately, you'd need different files. Same for 'juveniles'. Your idea of simply using '1', '2', or '3' in a single column won't work - I'd suggest a thorough reading of chapter 7 (linear models) to understand why.

Can I enter individual weight (physical weight, gramems, that is) in one column with the idea to only run the analysis on animals above a certain weight?

Not exactly - you can have MARK create a dummy variable structure comparing individuals of weight above a certain value with those below that value (e.g., heavy versus light) using design matrix functions (described in Chapter 12). But, you can't tell MARk to 'do the analysis only on the heavy ones - selecting them from all individuals in the file'.

I guess this can be found somewhere in the documentation but it is quicker to ask, hoping for a quick and simple answer. These are the kind of things you need to know before preparing a big data file.

MARK does not have data manipulation tools - at least, not in the sense you're describing them. This is for a very good reason - because whatever query language-data manipulation features Gary put into MARK would undoubtedly work well for some folks, and not for others.

So, the basic paradigm is simple: use your favourite data manipulation program to create encounter histories, create your input files, and feed them into MARK. Depending on your facility with your data handling software, this is usually a fairly trivial task. Sure, you may end up with multiple files, each representing a subset of the data, but that is usually not a problem.

Moreoever, pulling out pieces of the data set - which I assume you want to do because you want an estimate for a particular subset (say, females) isn't necessary - if you have both males and females in the data set, you'll get estimates for both sexes. Again, if it isn't immediately obvious why this is the case, then I suggest a thorough reading of Chapter 7.