Error in documentation of lmer?

According to the documentation of lmer, the way to get the fitting
algorithm steps is by setting the control = list(msVerbose=TRUE). For
me, this seems to be ignored.
(fm1 <- lmer(Reaction ~ Days + (Days|Subject), sleepstudy,
control=list(msVerbose=TRUE)))
Linear mixed model fit by REML 
Formula: Reaction ~ Days + (Days | Subject) 
   Data: sleepstudy 
  AIC  BIC logLik deviance REMLdev
 1756 1775 -871.8     1752    1744
Random effects:
 Groups   Name        Variance Std.Dev. Corr  
 Subject  (Intercept) 612.092  24.7405        
          Days         35.072   5.9221  0.066 
 Residual             654.941  25.5918        
Number of obs: 180, groups: Subject, 18

Fixed effects:
            Estimate Std. Error t value
(Intercept)  251.405      6.825   36.84
Days          10.467      1.546    6.77

Correlation of Fixed Effects:
     (Intr)
Days -0.138

However, after looking at the lmer function, it appears that by setting
verbose=TRUE, I can get the steps.
(fm1 <- lmer(Reaction ~ Days + (Days|Subject), sleepstudy,
verbose=TRUE))
  0:     1768.4124: 0.516398 0.0967302  0.00000
  1:     1761.9813: 0.571375 0.550876 0.167097
  2:     1755.4534: 0.790274 0.455893 0.215675
  3:     1745.0053: 0.812187 0.239613 0.105945
  4:     1744.2582: 0.817183 0.213879 0.0692586
  5:     1743.9764: 0.834405 0.230507 0.0310488
  6:     1743.8121: 0.877690 0.218240 0.0280506
  7:     1743.7121: 0.964311 0.243094 0.0246631
  8:     1743.6293: 0.970155 0.232066 0.0135558
  9:     1743.6284: 0.966654 0.230511 0.0163398
 10:     1743.6283: 0.966755 0.230912 0.0156969
 11:     1743.6283: 0.966734 0.230909 0.0156904
Linear mixed model fit by REML 
Formula: Reaction ~ Days + (Days | Subject) 
   Data: sleepstudy 
  AIC  BIC logLik deviance REMLdev
 1756 1775 -871.8     1752    1744
Random effects:
 Groups   Name        Variance Std.Dev. Corr  
 Subject  (Intercept) 612.092  24.7405        
          Days         35.072   5.9221  0.066 
 Residual             654.941  25.5918        
Number of obs: 180, groups: Subject, 18

Fixed effects:
            Estimate Std. Error t value
(Intercept)  251.405      6.825   36.84
Days          10.467      1.546    6.77

Correlation of Fixed Effects:
     (Intr)
Days -0.138

Is this a mistake in the documentation? Or have I missed something. I
have seen this on R 2.9.1 on Windows XP and on R 2.9.0 on Linux with
lme4 0.999375-31

Mark Lyman
Hmm, both control = list(msVerbose = TRUE) and verbose = TRUE should
result in verbose output from the optimization, although verbose =
TRUE is the preferred form now.

It looks as if at some point I got carried away with cleaning up the
code and removed the part of lmer that handles the control argument.
I have restored that and will issue a new version of the lme4 package.

Thanks for the report.
According to the documentation of lmer, the way to get the fitting
algorithm steps is by setting the control = list(msVerbose=TRUE). For
me, this seems to be ignored.

(fm1 <- lmer(Reaction ~ Days + (Days|Subject), sleepstudy,
control=list(msVerbose=TRUE)))
Linear mixed model fit by REML
Formula: Reaction ~ Days + (Days | Subject)
? Data: sleepstudy
?AIC ?BIC logLik deviance REMLdev
?1756 1775 -871.8 ? ? 1752 ? ?1744
Random effects:
?Groups ? Name ? ? ? ?Variance Std.Dev. Corr
?Subject ?(Intercept) 612.092 ?24.7405
? ? ? ? ?Days ? ? ? ? 35.072 ? 5.9221 ?0.066
?Residual ? ? ? ? ? ? 654.941 ?25.5918
Number of obs: 180, groups: Subject, 18

Fixed effects:
? ? ? ? ? ?Estimate Std. Error t value
(Intercept) ?251.405 ? ? ?6.825 ? 36.84
Days ? ? ? ? ?10.467 ? ? ?1.546 ? ?6.77

Correlation of Fixed Effects:
? ? (Intr)
Days -0.138

However, after looking at the lmer function, it appears that by setting
verbose=TRUE, I can get the steps.

(fm1 <- lmer(Reaction ~ Days + (Days|Subject), sleepstudy,
verbose=TRUE))
?0: ? ? 1768.4124: 0.516398 0.0967302 ?0.00000
?1: ? ? 1761.9813: 0.571375 0.550876 0.167097
?2: ? ? 1755.4534: 0.790274 0.455893 0.215675
?3: ? ? 1745.0053: 0.812187 0.239613 0.105945
?4: ? ? 1744.2582: 0.817183 0.213879 0.0692586
?5: ? ? 1743.9764: 0.834405 0.230507 0.0310488
?6: ? ? 1743.8121: 0.877690 0.218240 0.0280506
?7: ? ? 1743.7121: 0.964311 0.243094 0.0246631
?8: ? ? 1743.6293: 0.970155 0.232066 0.0135558
?9: ? ? 1743.6284: 0.966654 0.230511 0.0163398
?10: ? ? 1743.6283: 0.966755 0.230912 0.0156969
?11: ? ? 1743.6283: 0.966734 0.230909 0.0156904
Linear mixed model fit by REML
Formula: Reaction ~ Days + (Days | Subject)
? Data: sleepstudy
?AIC ?BIC logLik deviance REMLdev
?1756 1775 -871.8 ? ? 1752 ? ?1744
Random effects:
?Groups ? Name ? ? ? ?Variance Std.Dev. Corr
?Subject ?(Intercept) 612.092 ?24.7405
? ? ? ? ?Days ? ? ? ? 35.072 ? 5.9221 ?0.066
?Residual ? ? ? ? ? ? 654.941 ?25.5918
Number of obs: 180, groups: Subject, 18

Fixed effects:
? ? ? ? ? ?Estimate Std. Error t value
(Intercept) ?251.405 ? ? ?6.825 ? 36.84
Days ? ? ? ? ?10.467 ? ? ?1.546 ? ?6.77

Correlation of Fixed Effects:
? ? (Intr)
Days -0.138

Is this a mistake in the documentation? Or have I missed something. I
have seen this on R 2.9.1 on Windows XP and on R 2.9.0 on Linux with
lme4 0.999375-31

Mark Lyman

_______________________________________________
R-sig-mixed-models at r-project.org mailing list
https://stat.ethz.ch/mailman/listinfo/r-sig-mixed-models