[ADMB Users] crappy software

[Ben Bolker]

On 11-10-30 01:24 PM, dave fournier wrote:
> 
> This appears to be a good example of how crappy some of the R software is
> compared to SAS NLMIXED.
> It would be interesting to try this test on ADMB-RE.  I'll try this out.
> 
> http://onlinelibrary.wiley.com/doi/10.1002/sim.4265/abstract

  I tried this out (not the simulations, which would take a little while
to get together, but the data analysis: it appears (?) that the
same data set as is used by these guys is included in the 'geepack'
package).

  I can't get the results to agree (yet) with Zhang et al's stated
results, but all the R packages I have tried so far (glmmML, various
versions of lme4) agree very closely with each other and with the
results of glmmADMB with and without importance sampling ... the results
are posted at the bottom of:

http://glmm.wikidot.com/examples

  I'm sure Dave could code a much more efficient pure-ADMB version than
the current glmmADMB implementation, but I don't think the estimates
would change much?

  cheers
    Ben


  Feedback welcome.
> _______________________________________________
> Users mailing list
> Users at admb-project.org
> http://lists.admb-project.org/mailman/listinfo/users