variance components, odd results

[abreton]

I recently added two covariates to a mark/variance components analysis (Burnham's method) that I'd otherwise completed about a month ago. After running the first covariate through the procedure, I noticed that MARK suggested an adjusted (not naive) sigma2 of 0.0011688, more about this estimate in a moment -- my procedure was,

(1) retrieve time-dependent (unconstrained base) model
(2) go to output > specific model output > variance components > real parameters
(3) select parameters #24-25.
(4) select 'user specified'
(5) enter covariate values into (single column) design matrix ... click ok.

I tried the second covariate, again sigma2 was exactly 0.0011688 -- that seemed unbelievable ... so I re-ran the variance components procedure a few more times using other covariates (that I'd run in the past) and each time sigma2 = 0.0011688. Recall the variance procedure provides an option to graph some of the results, here's an example from one of the covariates that gave a sigma2 of 0.0011688,

https://www.box.com/s/f75d63a36c40874bbd34

Note the model estimates (red line), I thought this might be helpful/an important insight.

Subsequently, I've updated MARK to the latest version available on phidot. On the first run with the updated software MARK gave sigma = 0.0012586, on the second run 0.0011688 ... and all subsequent runs 0.0011688.

Prior to posting I searched this and the 'analysis' forum and no one seems to have encountered this before, With so many recent improvements to MARK I wonder if I might have stumbled on a bug...

Thanks for any assistance ... send me an email and I'll forward my MARK files and covariate values.

andre

[abreton]

The only odd part of the process was the user that was apparently so motivated to use the stupid part of his brain he failed to notice the obvious when he was pasting values over the intercept ... all bashing at this point will be accepted as it is well deserved.

Please disregard my original post ... there is no problem with the variance components procedure ... it's working as well as it did a few months ago when I was, apparently, using the common sense part of my brain.

andre