Nest Success: **WARNING** message/beta estimates

[gilu]

Hi all!

I have set up a multi year nest success analysis which seemingly works. But I have two issues.

First, when I run a model, there appears as normal a little window that asks whether I want to append the model's output to the database. At the end of the window, however, are two warning messages that read as follows:

** WARNING ** Maximum time exceeds number of occasions.
** WARNING ** Invalid observation -- all times same

I presume this is something to worry about, and probably relates to the enconterhistory in the .INP-file. But I could not figure out what it means. Anybody knows about that?

Despite this warnings I will get model rankings that make very much sense, especially in the light of preliminary investigations (logistic regression, graphical visualization).

However, for at least two covariates that matter, the sign of the beta estimate is opposite of what is to be expected on preliminary investigation.

For instance, escape distance matters, and graphical investigation shows that successful hens remain longer on the nest when approached, and hence have a shorter escape distance. The beta estimate for escape distance, however, is positive. The same thing is with bodymass. Heavy hens (=older) tend generally to be better (considering the rather well known biology of the species and when plotted graphically), and the bodymass model indeed is a strong candidate model, but its beta estimate is negative. So light hens flushing off early should be best hens which does not make sense. Young hens are usuallly lighter and do flush off earlier, but results also say there is a considerable difference between young and old hen DSR (daily nest survival rate), young hens having lower DSR.

My question is, do I misinterprete the beta estimate, or do I have a problem with the input file? May be this also relates to the warning message.

Any input is highly appreciated!

Cheers!

Gilbert

[jbruggin]

Regarding the warnings. The first may indicate that you have a value of k that is greater than the number of time intervals you indicated when beginning your analysis. As for the second warning, I would check your data for cases where i=j=k.

John

[gilu]

Regarding the warnings. The first may indicate that you have a value of k that is greater than the number of time intervals you indicated when beginning your analysis. As for the second warning, I would check your data for cases where i=j=k.

John

Thanks John, you are right! I sorted that out and it works fine now.

cheers

Gilu