Problem with random effects models for lambda

[dtempel]

I'm trying to run some random-effects models for lambda using the Pradel phi, p, lambda model structure. I'm getting some unusual results for my two user-specified models (a quadratic time trend, and a log-linear time trend). In the MARK notepad output, I'm getting integer values for the G trace (e.g. 3.00000) and sigma of 0.00000 for these two models.

I ran these exact same models a few months ago, and at that time, I got results that actually made sense. Interestingly, the QAIC values for each model are similar now to what I got before, even though the output (G trace, sigma) has changed. I've tried uninstalling MARK and installing the most recent version, but that didn't work. It also doesn't matter which link function is specified (sin, log, logit); I still get a zero estimate for sigma.

Does anyone have any suggestions? Anyone experience this before?