[R-sig-ME] bug in confint with boot method and several random terms ?

Ben Bolker bbolker at gmail.com
Sun Jan 12 16:39:06 CET 2014


On 14-01-12 10:10 AM, espesser wrote:
> 
> Dear all,
> 
> Here is a reproducible example of a possible bug concerning
> confint.merMod with method="boot",
> when more than a single random term is specified.
> The problem does not concern the bootstrapping itself, but the external
> function  bootFun.
> Unhappily, I'm not skill enough to fix it.
> I did not found previous mention of this , but apologizes if I'm missing
> something.
> 
> Thank you for your help
> 
> R. espesser
> 

  This was fixed recently (11 days ago) in the development version.  I
will work on putting the latest version (just bumped to 1.1-4) on
http://lme4.r-forge.r-project.org/repos .  In the meantime, if you have
development tools installed (which has gotten to be relatively easy,
even on Windows ...) you can always install the latest version via
library("devtools"); install_github("lme4","lme4")

  We are hoping for another CRAN release fairly soon, but holding your
breath might be dangerous.

  Ben Bolker

> 
> 
> library(lme4)
> 
> data(Penicillin)
> 
> fm2 <- lmer(diameter ~ 1 + (1|plate) + (1|sample), Penicillin)
> summary(fm2)
> Linear mixed model fit by REML ['lmerMod']
> Formula: diameter ~ 1 + (1 | plate) + (1 | sample)
>    Data: Penicillin
> 
> REML criterion at convergence: 330.8606
> 
> Random effects:
>  Groups   Name        Variance Std.Dev.
>  plate    (Intercept) 0.7169   0.8467
>  sample   (Intercept) 3.7309   1.9316
>  Residual             0.3024   0.5499
> Number of obs: 144, groups: plate, 24; sample, 6
> 
> Fixed effects:
>             Estimate Std. Error t value
> (Intercept)  22.9722     0.8086   28.41
> 
>  confint(fm2, method="boot",nsim=100)
> Computing bootstrap confidence intervals ...
> Error in setNames(Cv_to_Sv(th, s = sigma(x)), c(tnames(x, old = FALSE,  :
>   'names' attribute [3] must be the same length as the vector [2]
> In addition: Warning message:
> In m0[lower.tri(m0, diag = TRUE)] <- x :
>   number of items to replace is not a multiple of replacement length
> 
> 
> 
> 
> # It is OK with only one random intercept:
> 
>  fm2_single <- lmer(diameter ~ 1 + (1|plate) , Penicillin)
> 
>  confint(fm2_single, method="boot",nsim=100)
> Computing bootstrap confidence intervals ...
>                          2.5 %     97.5 %
> sd_(Intercept)|plate  0.000000  0.7437895
> sigma                 1.751577  2.1949818
> (Intercept)          22.606299 23.3667589
> 
> 
> I
> ##########################
>  sessionInfo()
> R version 3.0.2 (2013-09-25)
> Platform: x86_64-w64-mingw32/x64 (64-bit)
> 
> locale:
> [1] LC_COLLATE=French_France.1252  LC_CTYPE=French_France.1252   
> LC_MONETARY=French_France.1252
> [4] LC_NUMERIC=C                   LC_TIME=French_France.1252
> 
> attached base packages:
> [1] stats     graphics  grDevices utils     datasets  methods   base
> 
> other attached packages:
> [1] boot_1.3-9      lme4_1.0-5      Matrix_1.1-0    lattice_0.20-23
> 
> loaded via a namespace (and not attached):
>  [1] colorspace_1.2-4   dichromat_2.0-0    digest_0.6.4      
> ggplot2_0.9.3.1    grid_3.0.2
>  [6] gtable_0.1.2       labeling_0.2       MASS_7.3-29       
> minqa_1.2.1        munsell_0.4.2
> [11] nlme_3.1-111       plyr_1.8           proto_0.3-10      
> RColorBrewer_1.0-5 reshape2_1.2.2
> [16] scales_0.2.3       splines_3.0.2      stringr_0.6.2      tools_3.0.2
>>
> 
>



More information about the R-sig-mixed-models mailing list