+ Reply to Thread
Results 1 to 6 of 6

Thread: Bootstrap strange estimate

  1. #1
    Points: 440, Level: 8
    Level completed: 80%, Points required for next Level: 10

    Posts
    8
    Thanks
    0
    Thanked 0 Times in 0 Posts

    Bootstrap strange estimate




    Hello everyone, I'm a statistical novice, I would like your help. I'm analyzing non-normally distributed variables. To develop the confidence interval I preferred to use non-parametric method: bootstrap.
    Unfortunately I get output biased estimates and standard errors high, even though I used a lot of replies. How can I do?

  2. #2
    Omega Contributor
    Points: 38,413, Level: 100
    Level completed: 0%, Points required for next Level: 0
    hlsmith's Avatar
    Location
    Not Ames, IA
    Posts
    7,004
    Thanks
    398
    Thanked 1,186 Times in 1,147 Posts

    Re: Bootstrap strange estimate

    Tell us more about what you are doing. Confidence interval for what? Also, perhaps post a histogram of your BS estimate.
    Stop cowardice, ban guns!

  3. #3
    Points: 440, Level: 8
    Level completed: 80%, Points required for next Level: 10

    Posts
    8
    Thanks
    0
    Thanked 0 Times in 0 Posts

    Re: Bootstrap strange estimate

    this is R code:

    x=read.csv2("D:bootstrap2.csv")
    attach(x)
    x

    library(boot)
    mean.func=function(x,ind)
    median(x[ind])

    bfer=boot(ferr, mean.func, R=5000)

    boot.ci(bfer)
    Attached Thumbnails Attached Thumbnails Click image for larger version

Name:	asb1.JPG‎
Views:	5
Size:	63.9 KB
ID:	5955   Click image for larger version

Name:	asb2.JPG‎
Views:	6
Size:	53.4 KB
ID:	5956   Click image for larger version

Name:	asb3.JPG‎
Views:	4
Size:	51.1 KB
ID:	5957  

  4. #4
    Omega Contributor
    Points: 38,413, Level: 100
    Level completed: 0%, Points required for next Level: 0
    hlsmith's Avatar
    Location
    Not Ames, IA
    Posts
    7,004
    Thanks
    398
    Thanked 1,186 Times in 1,147 Posts

    Re: Bootstrap strange estimate

    So you are just putting 95 percentile CI on your means? The histograms and q-q plots look fine. What is your concern?
    Stop cowardice, ban guns!

  5. #5
    Points: 440, Level: 8
    Level completed: 80%, Points required for next Level: 10

    Posts
    8
    Thanks
    0
    Thanked 0 Times in 0 Posts

    Re: Bootstrap strange estimate

    my doubt is the st error....too high. can I present these values? I am not shure

  6. #6
    Omega Contributor
    Points: 38,413, Level: 100
    Level completed: 0%, Points required for next Level: 0
    hlsmith's Avatar
    Location
    Not Ames, IA
    Posts
    7,004
    Thanks
    398
    Thanked 1,186 Times in 1,147 Posts

    Re: Bootstrap strange estimate


    Well you can see that it is symmetrical and that is the purpose of the bootstrap. The bootstrap cant just make your se smaller the more samples you use like increasing the sample size in traditional approaches. I actually remember hearing a Prof say that with more samples the SE will get wider with BS, not sure if that is case specific. A general rule of precision is your CI is < 30% of your estimate, but that is just an arbitrary rule.
    Stop cowardice, ban guns!

+ Reply to Thread

           




Tags for this Thread

Posting Permissions

  • You may not post new threads
  • You may not post replies
  • You may not post attachments
  • You may not edit your posts






Advertise on Talk Stats