CJS modelling ; searched fledglings different time period

[Lutfor_Rahman]

Dear members,
I have a situation where trying to build cjs model using weekly survival data.
A total of over 200 fledglings were marked initially and were searched mostly every week for 5 weeks (recorded as re-sighted or not in the subsequent visit). However, 100 of those fledglings were also search further 2-3 weeks more. I am trying to do postfledging survival analyses for the period of 8 weeks.

Is there any way, I can incorporate in my model building those fledglings which were not search upto 8th week or should I only include those fledglings which were searched only upto 8th week.

If I take only later cases, I have to through away about my 50% of data. Should I do any other survival analysis which will be best suitable for my data?

Any suggestions will be much appreciated.

Kind regards
Lutfor

[jlaake]

One approach would be to assign the fledglings to groups and for the group that was not followed for the last 3 weeks set p=0.

--jeff

[Lutfor_Rahman]

Thanks Jeff,
But I am not sure how I can set p(0) in model building if I want to modelling talking all those data;

If I have a set of data like below (for example),
Group 1:
11010
10111
11000
Group 2:
11001000
10111110
11000011

I should do format data like below (both groups together and during model building re-sighting probability P(0)-even not sure how to command that option):
11010000 01
10111000 01
11000000 01
11001000 10
10111110 10
11000011 10

Your suggestions will be much appreciated.

Kind regards
Lutfor

[jlaake]

Yes you have to pad the capture histories as you have shown. I assume here that you are using the MARK gui interface and not RMark. How you set p=0 for those 3 weeks depends on the PIM structure you choose for your model. If you have the g*t PIM structure then you just need to fix the real parameters matching group 1 for times 6-8 to zero with the run window. See Cooch and White book for examples.

--jeff

[Lutfor_Rahman]

Dear Jeff,

Thank you so much for your helpful suggestions. However, after reading Cooch and White book, I still wondering how i would relate them with co-variates ( I have about 6 co-variables) on each data point. Now, I do have two groups of data (One groups of data which search for 4 weeks and another groups of data search made for 7 weeks.

11010000 01
10111000 01
11000000 01
11001000 10
10111110 10
11000011 10

Is there any way, I can do analysis taking into co-variates.

Kind regards
Lutfor

[jlaake]

Chapter 11 in Cooch and White shows how to use covariates.

Your freq column 01 etc should have spaces 0 1

Also, to set some p values to 0 when you run a model you click on the Fix Parameters button and assign 0 to the appropriate parameters.

I suggest that you look through some of the examples that come with MARK and read through ch 11. You may also want to attend one of the classes that are offered.

regards --jeff