[R-sig-ME] strange interaction lme4a::lmer and lme4a::profile

[Douglas Bates]

On Thu, Jun 16, 2011 at 5:01 AM,  <joris.dewolf at cropdesign.com> wrote:
> Hi,
>
> I've got a strange behaviour with lme4a::lmer in combination with
> lme4a::profile()
>
> profile() seems to change the merMod object it has been calculating the
> profile for. The variance components and standard errors on fixed effects
> as returned by summary() are different before and after running profile()
> as you can see in the example below.
> >From the profile itself, the first run seems to be the correct one.
>
> Output and sessionInfo below. I can provide the data set on request.
>
> Anybody an idea what is going wrong?

Thanks for the very thorough report.  It's a bug that I forgot to fix.
 I'll do so now.

>
>
>
>> modL1.2 <- lmer(data=RPD15L1, value ~ transgenity +
> (transgenity|event),REML=FALSE)
>> summary(modL1.2)
> Linear mixed model fit by maximum likelihood ['summary.mer']
> Formula: value ~ transgenity + (transgenity | event)
>   Data: RPD15L1
>      AIC       BIC    logLik  deviance
>  500.6290  520.7116 -244.3145  488.6290
>
> Random effects:
>  Groups   Name                Variance Std.Dev. Corr
>  event    (Intercept)         0.4475   0.6690
>          transgenityPositive 0.9615   0.9806   0.295
>  Residual                     0.2971   0.5450
> Number of obs: 210, groups: event, 36
>
> Fixed effects:
>                    Estimate Std. Error t value
> (Intercept)          23.2766     0.1235  188.50
> transgenityPositive   1.0607     0.1801    5.89
>
> Correlation of Fixed Effects:
>            (Intr)
> trnsgntyPst 0.115
>> pr.modL1.2 <- profile(modL1.2)
>> summary(modL1.2)
> Linear mixed model fit by maximum likelihood ['summary.mer']
> Formula: value ~ transgenity + (transgenity | event)
>   Data: RPD15L1
>      AIC       BIC    logLik  deviance
>  500.6290  520.7116 -244.3145  488.6290
>
> Random effects:
>  Groups   Name                Variance Std.Dev. Corr
>  event    (Intercept)         0.1964   0.4432
>          transgenityPositive 0.4376   0.6615   0.523
>  Residual                     0.2971   0.5450
> Number of obs: 210, groups: event, 36
>
> Fixed effects:
>                    Estimate Std. Error t value
> (Intercept)         23.27657    0.09092  256.01
> transgenityPositive  1.06065 0.13364    7.94
>
> Correlation of Fixed Effects:
>            (Intr)
> trnsgntyPst 0.119
>
>
>> sessionInfo()
> R version 2.12.1 (2010-12-16)
> Platform: i386-pc-mingw32/i386 (32-bit)
>
> locale:
> [1] LC_COLLATE=Dutch_Belgium.1252  LC_CTYPE=Dutch_Belgium.1252
> [3] LC_MONETARY=Dutch_Belgium.1252 LC_NUMERIC=C
> [5] LC_TIME=Dutch_Belgium.1252
>
> attached base packages:
> [1] stats     graphics  grDevices utils     datasets  methods   base
>
> other attached packages:
> [1] boot_1.2-43        lme4a_0.999375-63  MatrixModels_0.2-1 minqa_1.1.14
>
> [5] Rcpp_0.9.0.2       Matrix_0.999375-46 lattice_0.19-13    rj_0.5.0-5
>
> loaded via a namespace (and not attached):
> [1] codetools_0.2-6 grid_2.12.1     nlme_3.1-98     rJava_0.8-8
> [5] splines_2.12.1  stats4_2.12.1   tools_2.12.1
>        [[alternative HTML version deleted]]
>
> _______________________________________________
> R-sig-mixed-models at r-project.org mailing list
> https://stat.ethz.ch/mailman/listinfo/r-sig-mixed-models
>