changes in survival estimates with different input file

[Beatriz Martín]

Dear all,

I am using known fate model from MARK to estimate monthly survival rates in my birds. I am using four different groups: 1000: males-population 1; 0100 females-population 1; 0010 males-population 2; 0001 females-population 2 and I have used an individual covariate. My best model includes population, sex, month and the individual covariate.

I run also new models using two different input files separating each population. Now my best model includes only sex, month and covariate.
And now the problem: when I compare the parameter estimates for survival from the model that considers both populations together with the one with populations separated, these survival estimates are quite different. What is the reason? Have I done anything wrong in my runnings?

Thank you in advance,

Beatriz

[gwhite]

Beatriz:
You are definitely doing something wrong. Better re-read the Gentle Introduction on how to enter data.
Gary

[Beatriz Martín]

Beatriz:
You are definitely doing something wrong. Better re-read the Gentle Introduction on how to enter data.
Gary

I think I am doing it well...
This is an example of my input data format file when I consider sex and population in the same file (four groups, for sex and population, and one individual covariate):

/*1*/ 101010101011000000000000 1 0 0 0 1230;
/*2*/ 110000000000000000000000 0 0 0 1 1800;
/*3*/ 101010110000000000000000 0 1 0 0 1540;

And this one when I separate two populations in two different files and then I only consider the factor sex:

/*1*/ 101010101011000000000000 1 0 1230;

I can't find anything wrong in that!

Thanks again,

Beatriz

[gwhite]

Beatriz:
You may be getting the data into MARK correctly, but then did you run models that are completely comparable. For example, if you assume an additive effect of the individual covariate on the 4 group data set, you would not be able to duplicate this model when you break the data into 2 groups.

To compare, you should run a g*t model in all three data sets, and the answers will be the same if you have done everything correctly.
Gary

[Beatriz Martín]

Beatriz:
You may be getting the data into MARK correctly, but then did you run models that are completely comparable. For example, if you assume an additive effect of the individual covariate on the 4 group data set, you would not be able to duplicate this model when you break the data into 2 groups.

To compare, you should run a g*t model in all three data sets, and the answers will be the same if you have done everything correctly.
Gary

Yes, I was comparing additive models, that must be the reason of the difference between the survival estimates. Thanks a lot for the answer. I wanted to be sure I was doing things properly!

Beatriz.