[R-sig-ME] confint.merMod(): method = "boot" throws error

Bertolt Meyer b.meyer at psychologie.uzh.ch
Wed Feb 26 18:24:39 CET 2014


Dear all,

I am trying to obtain confidence intervals on the parameters of a merMod with bootstrapping. However, I always get the following error:

> confint.merMod(mmodel.7, method = "boot", nsim = 100)

Computing bootstrap confidence intervals ...
Error in if (const(t, min(1e-08, mean(t, na.rm = TRUE)/1e+06))) { : 
 Missing value where TRUE/FALSE is required

However, using confint.merMod() with method = "Wald" on the same model works just fine:

> confint.merMod(mmodel.7, method = "Wald")
                      2.5 %       97.5 %
(Intercept)     4.1936561326  4.768229069
sex2            0.0569452437  0.393527752
age            -0.0122385301  0.004655781
tenure          0.0238700983  0.048076967
network_size   -0.0022468908  0.008291748
dage           -0.0085075176  0.006415557
dtenure         0.0005230516  0.019461232
dsex1          -0.3336595531 -0.022736774
time_l         -8.2010266502 -6.205144078
time_q          0.9684946890  2.249428767
time_c         -2.7975435909 -1.524616092
dtenure:time_l -0.4159371669 -0.109693667
dtenure:time_q  0.0291175752  0.249497811
dtenure:time_c -0.1558022397  0.055678434

I tried this with different models, it's always the same. Does anybody have an idea why it fails with method = "boot"?

Thank you and best greetings,
Bertolt

--
Dr. Bertolt Meyer, Dipl.-Psych.
Senior Research and Teaching Associate
Social and Business Psychology
Institute of Psychology, University of Zurich 
Binzmuehlestr. 14/13
CH-8050 Zurich
Switzerland

b.meyer at psychologie.uzh.ch



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