[R] Why does Bootstrap work for one of similar models but not for the other?

Thu Aug 19 18:29:45 CEST 2010

In future, when posting data in your message use dput() or
textConnection() so that helpeRs can more easily load them.

I was not able to replicate your results. Here's what I got:

Bootstrap Statistics :
       original        bias    std. error
t1*  0.99975370  0.0044205644  0.04110232
t2* -0.06091574 -0.0078646847  0.05778133
t3*  0.27506204  0.0006121326  0.05296862
t4* -0.03040424  0.0002096330  0.02826951

> sessionInfo()
R version 2.11.0 (2010-04-22)
i386-pc-mingw32

locale:
[1] LC_COLLATE=English_United States.1252
[2] LC_CTYPE=English_United States.1252
[3] LC_MONETARY=English_United States.1252
[4] LC_NUMERIC=C
[5] LC_TIME=English_United States.1252

attached base packages:
 [1] datasets  splines   grid      tcltk     stats     graphics  grDevices
 [8] utils     methods   base

other attached packages:
[1] boot_1.2-42     TinnR_1.0.3     R2HTML_1.59-1   Hmisc_3.7-0
[5] survival_2.35-8 svSocket_0.9-48 lattice_0.18-3  MASS_7.3-5

What does

getOption("digits")

give you just before you run each bootstrap?

Bert Gunter
Genentech nonclinical Statistics

On Thu, Aug 19, 2010 at 12:50 AM, Reiko Akiyama <reiko.akiyama at ebc.uu.se> wrote:
> Dear all,
>
> Could anyone help me figure out why bootstrap works for one of similar
> models but not for the other and how I can solve it?
>
> I am using R 2.11.1 in Windows and would like to get confidence intervals
> for my models A and B by bootstrapping. However, bootstrap gives expected
> output for the model A but not for B, which I found was puzzling because the
> structure of the models is similar as I describe below. I had another person
> running the models in another computer and the same thing happens so this
> does not seem to be computer-specific. I could not find a clue for a
> solution in the R archive or in the R book (at least to the extent I
> understood).
>
> Here are the properties of the models A and B and what happens when I run
> bootstrap.
>
> modelA: rA~stA1+ stA2+stA3
> model B: rB~stB1+stB2+stB3
> The variables for the models A and B are in the same dataset called ?data?.
> The sample size is 32 for both models and the value range and distribution
> of the variables in the two models are similar. (Variables from both models
> are at the end of this enquiry.)
>
> [bootstrap of the model A]
>>
>> A.fun<-function(data,indices)coefficients(lm(rA~stA1+
>> stA2+stA3,data=data[indices,]))
>> bootA<-boot(data,A.fun,1000);bootA
>
> ORDINARY NONPARAMETRIC BOOTSTRAP
> Call:
> boot(data = data, statistic = A.fun, R = 1000)
> Bootstrap Statistics :
>       original       bias    std. error
> t1*  1.00016501 -0.004350842  0.05309877
> t2*  0.02343475  0.008501989  0.07638795
> t3* -0.01602954 -0.004980400  0.07806805
> t4*  0.03601194 -0.005417404  0.08510128
>
> [bootstrap of the model B]
>>
>>
>> B.fun<-function(data,indices)coefficients(lm(rB~stB1+stB2+stB3),data=data[indices,])
>> bootB<-boot(data,B.fun,1000);bootB
>
> ORDINARY NONPARAMETRIC BOOTSTRAP
> Call:
> boot(data = data, statistic = B.fun, R = 1000)
> Bootstrap Statistics :
>       original  bias    std. error
> t1*  0.99975370       0           0
> t2* -0.06091574       0           0
> t3*  0.27506203       0           0
> t4* -0.03040424       0           0
>
> What am I missing here?
> I highly appreciate any comments and suggestions.
>
> Best Wishes,
> Reiko Akiyama
> Uppsala University
> Sweden
>
> [Variables from the model A]
>>
>> rA
>
>  [1] 0.7100881 1.0406464 1.1100229 0.6182664 0.7345739 1.0577865 0.6856024
>  [8] 0.5264447 1.5793340 1.1793993 0.6488737 1.0214076 1.3589618 1.0528893
> [15] 1.5242409 1.3761019 0.9427032 0.6794809 1.4752693 0.7737512 1.0120797
> [22] 0.8692458 1.2079660 1.0610513 0.8570029 0.9794319 1.0957395 0.8243552
> [29] 0.4162586 1.4079334 1.0692132 1.1059419
>>
>> stA1
>
>  [1] -0.9126354 -0.8331680 -1.0239203 -0.3721959 -0.5311308  0.7564474
>  [7] -1.1828933 -1.2146727 -0.8172593 -0.9921410 -0.5152602 -0.9285442
> [13] -0.4198840 -0.9444529 -0.4198840 -0.8331680  1.2810163  1.4081718
> [19]  1.7102091  2.3460247  1.3806653  1.3127957  1.2333282  1.4240806
> [25] -0.1337555 -0.1973142  0.2954372 -0.1337555 -0.4039753 -0.3880665
> [31]  0.2795666 -0.2291317
>>
>> stA2
>
>  [1] -0.2292617 -0.4917962 -0.6437899 -1.2241293 -0.3398026 -2.0946384
>  [7] -1.0721356 -1.2655821 -1.3484877 -1.8873744 -0.7543307 -0.9615948
> [13] -0.3674378  0.4483537  0.8761467 -0.8786892  0.5312593  1.1524988
> [19]  0.3234425 -0.4088906  0.5102565  1.1945044  1.7748438  0.6827002
> [25]  0.6418001  1.1801340  0.4207184  0.8076114  0.9181522  0.6827002
> [31]  0.9037819  0.9181522
>>
>> stA3
>
>  [1]  0.86459627 -0.23416149 -2.00372671  0.04161491  0.78881988 -2.50869565
>  [7] -0.02608696 -0.84161491 -0.95465839 -0.28012422  0.47080745  0.07577640
> [13]  0.84223602  0.24472050  2.83975155  0.43043478 -0.75652174 -0.92795031
> [19]  0.29192547 -0.78633540 -0.78385093 -0.51242236  0.59627329  0.19068323
> [25]  0.02919255  1.17018634 -0.19440994  0.68385093  1.08881988 -0.28385093
> [31] -0.71118012  1.06583851
>
> [Variables from the model B]
>>
>> rB
>
>  [1] 1.5385568 1.5885100 1.3587255 0.8991566 1.4086787 0.3097095 0.9191378
>  [8] 0.3996252 0.7393065 0.6993440 1.2488286 1.4186693 1.4586318 1.8282851
> [15] 0.8991566 0.9790816 1.0889785 1.0090535 0.7792690 0.8991566 0.8791753
> [22] 0.7892597 0.6294096 0.9690910 1.0689973 0.5994377 0.6793628 0.7293159
> [29] 0.9690910 0.7393065 0.7193253 1.7583507
>>
>> stB1
>
>  [1] -0.67898627 -0.94275552 -1.32045796  0.03417996 -1.18276552  2.01872951
>  [7] -1.75937865 -1.85016395 -0.70319013 -0.89159673 -0.35055299 -0.38890124
> [13] -0.81445562 -0.98941255 -0.95548269 -0.63066192  0.52759406  1.27063302
> [19]  1.19746568  1.34424498  0.62679931  1.15103096  1.24195520  0.94395043
> [25]  0.20232868  0.71085978  0.53654199  0.67470683  0.41377202  0.38428833
> [31]  0.58178180 -0.40626910
>>
>> stB2
>
>  [1]  2.18599646  1.64030436  0.29913150 -0.30874645  2.29052340 -2.13238029
>  [7] -0.78386750 -0.53233418 -0.96552917 -1.00046883 -0.02227346 -0.71399585
> [13]  0.42490201  1.27034048  0.09650296 -0.32970563 -0.23188840 -0.24586119
> [19]  0.04061179 -0.07118592 -0.49040812 -0.04323265 -0.06419952 -0.18297594
> [25]  0.40393513 -0.73495504 -0.53233418 -0.23188840  0.13144263 -0.21092921
> [31] -1.24501576  2.29052340
>>
>> stB3
>
>  [1] -0.3683333  0.3416667 -1.7883333 -1.8133333 -0.6166667  0.8783333
>  [7] -1.4433333 -0.5150000  0.3066667 -0.5016667 -0.1850000 -0.2116667
> [13]  1.2116667 -0.5783333  0.4533333 -0.3300000 -0.1733333 -0.7183333
> [19] -0.5000000 -0.3983333 -1.2733333 -0.2333333 -0.2333333 -0.9150000
> [25]  0.3366667  2.4200000  1.6016667  0.5116667  0.9283333  1.8750000
> [31]  1.0866667  0.5950000
>
> ______________________________________________
> R-help at r-project.org mailing list
> https://stat.ethz.ch/mailman/listinfo/r-help
> PLEASE do read the posting guide http://www.R-project.org/posting-guide.html
> and provide commented, minimal, self-contained, reproducible code.
>

-- 
Bert Gunter
Genentech Nonclinical Biostatistics
467-7374
http://devo.gene.com/groups/devo/depts/ncb/home.shtml