changing time intervals affects model weights

[storrence]

Hi.

I ran the exact same models on the same datasets with the only difference being how I entered the time intervals in the input file.

For robust design, I did it 2 ways

as "period" where I had:

0 1 0 1 0...

and as "week": where I had:

0 (# days betw primary/7) 0 (# days betw primary/7) 0 ...

I am not concerned that the estimates themselves are different...I understand why. What I do not understand is why the likelihood and weights of different models changes. In one instance, a top model as "week" didn't even converge as "period." Why is this?

I calculated "period" because it would be a lot easier to calculate end of season survival and the new SE's doing it this way than by trying to restandardize everything to get this new estimate from weekly estimates.

Thanks.