When updating the R version requirements for lmerTest, I stumbled upon this issue (for which we happen to have a test in lmerTest). It appears that KRmodcomp gives different ddf estimates using R 3.3.2 - I think the 3.4.4 results are correct:
> library(pbkrtest)
Loading required package: lme4
Loading required package: Matrix
> m <- lme4::lmer(Reaction ~ Days + I(Days^2) + (Days | Subject), sleepstudy)
> KRmodcomp(m, matrix(c(0, 1, 0), nrow=1))
Note: method with signature ‘sparseMatrix#ANY’ chosen for function ‘kronecker’,
target signature ‘dgCMatrix#ngCMatrix’.
"ANY#sparseMatrix" would also be valid
F-test with Kenward-Roger approximation; computing time: 0.23 sec.
large : Reaction ~ Days + I(Days^2) + (Days | Subject)
small : L beta = L betaH
L=
1 x 3 sparse Matrix of class "dgCMatrix"
[1,] . 1 .
betaH=
[1] 0
stat ndf ddf F.scaling p.value
Ftest 6.9551 1.0000 150.3885 1 0.009236 **
---
Signif. codes: 0 ‘***’ 0.001 ‘**’ 0.01 ‘*’ 0.05 ‘.’ 0.1 ‘ ’ 1
> sessionInfo()
R version 3.3.2 RC (2016-10-26 r71594)
Platform: x86_64-apple-darwin13.4.0 (64-bit)
Running under: OS X El Capitan 10.11.6
locale:
[1] en_US.UTF-8/en_US.UTF-8/en_US.UTF-8/C/en_US.UTF-8/en_US.UTF-8
attached base packages:
[1] stats graphics grDevices utils datasets methods base
other attached packages:
[1] pbkrtest_0.4-6 lme4_1.1-13 Matrix_1.2-7.1 fortunes_1.5-3
loaded via a namespace (and not attached):
[1] Rcpp_0.12.6 splines_3.3.2 MASS_7.3-45 munsell_0.4.3
[5] colorspace_1.3-2 lattice_0.20-34 minqa_1.2.4 car_2.1-4
[9] plyr_1.8.4 tools_3.3.2 nnet_7.3-12 parallel_3.3.2
[13] grid_3.3.2 gtable_0.2.0 nlme_3.1-128 mgcv_1.8-15
[17] quantreg_5.29 MatrixModels_0.4-1 lazyeval_0.2.0 assertthat_0.1
[21] tibble_1.2 numDeriv_2016.8-1 nloptr_1.0.4 ggplot2_2.2.1
[25] scales_0.4.1 lmerTest_3.0-0 SparseM_1.74
Restarting R session...
> library(pbkrtest)
Loading required package: lme4
Loading required package: Matrix
> m <- lme4::lmer(Reaction ~ Days + I(Days^2) + (Days | Subject), sleepstudy)
> KRmodcomp(m, matrix(c(0, 1, 0), nrow=1))
F-test with Kenward-Roger approximation; computing time: 0.24 sec.
large : Reaction ~ Days + I(Days^2) + (Days | Subject)
small : L beta = L betaH
L=
1 x 3 sparse Matrix of class "dgCMatrix"
[1,] . 1 .
betaH=
[1] 0
stat ndf ddf F.scaling p.value
Ftest 6.9551 1.0000 114.4314 1 0.00952 **
---
Signif. codes: 0 ‘***’ 0.001 ‘**’ 0.01 ‘*’ 0.05 ‘.’ 0.1 ‘ ’ 1
> sessionInfo()
R version 3.4.4 (2018-03-15)
Platform: x86_64-apple-darwin15.6.0 (64-bit)
Running under: OS X El Capitan 10.11.6
Matrix products: default
BLAS: /System/Library/Frameworks/Accelerate.framework/Versions/A/Frameworks/vecLib.framework/Versions/A/libBLAS.dylib
LAPACK: /Library/Frameworks/R.framework/Versions/3.4/Resources/lib/libRlapack.dylib
locale:
[1] en_US.UTF-8/en_US.UTF-8/en_US.UTF-8/C/en_US.UTF-8/en_US.UTF-8
attached base packages:
[1] stats graphics grDevices utils datasets methods base
other attached packages:
[1] pbkrtest_0.4-7 lme4_1.1-17 Matrix_1.2-12 fortunes_1.5-4
loaded via a namespace (and not attached):
[1] minqa_1.2.4 MASS_7.3-49 compiler_3.4.4 parallel_3.4.4 tools_3.4.4
[6] Rcpp_0.12.16 splines_3.4.4 nlme_3.1-131.1 grid_3.4.4 nloptr_1.0.4
[11] lattice_0.20-35
Hi,
When updating the R version requirements for lmerTest, I stumbled upon this issue (for which we happen to have a test in lmerTest). It appears that KRmodcomp gives different ddf estimates using R 3.3.2 - I think the 3.4.4 results are correct:
Cheers,
Rune