Mixed model based on NLME package

[alancai1974]

I'm wondering if the mixed model module could add support to NLME package. The current module is based on lme4 package (LMER function). The reason that I suggest NLME package (lme function) is that lme includes a number of function that is not available in lmer, for example, you can specify correlation types to fix autocorrelation.

[mcfanda@gmail.com]

Partially. The autocorrelation option, however, will be added soon

[alancai1974]

Are you also going to add variance/covariance related options? For example, lme allows modelling of heteroscedasticity using weight options and various variance specifications.

[mcfanda@gmail.com]

yes, that's my aim. GAMLj is going to be upgraded to version 3, in which a bunch of new possibilities is implemented (model comparisons, effect size, etc) for all models. Among them, I'm implementing also more options for structured errors and random effects covariances matrices in the mixed model. If you have suggestions along these lines, please send me an email. Help is always welcome.

[Philipp.B]

This is amazing news! I use your package for my master thesis and I am very thankful for your work. Is there a time frame when the update will come out?

[mcfanda@gmail.com]

not really. Being it a major update, it needs extensive testing. I think you'll be graduated already by the time it comes out.

[MAgojam]

I think you'll be graduated already by the time it comes out.

OT
Marcello, if it had happened 45 years ago, I would have waited for your module before graduating ...

Saluti,
Maurizio

[alancai1974]

This is a great news. Looking forward to seeing this happen.

[alancai1974]

just curious: why 45 years

I think you'll be graduated already by the time it comes out.

OT
Marcello, if it had happened 45 years ago, I would have waited for your module before graduating ...

Saluti,
Maurizio