+ Reply to Thread
Results 1 to 4 of 4

Thread: Post-hoc Bonferroni alternatives

  1. #1
    Points: 256, Level: 5
    Level completed: 12%, Points required for next Level: 44

    Posts
    2
    Thanks
    0
    Thanked 0 Times in 0 Posts

    Post-hoc Bonferroni alternatives




    I have conducted a number of Mann-Whitney U-tests on my data, and now need to adjust for multiple comparisons. Unfortunately, using a Bonferroni adjustment seems extremely conservative. Are there any alternatives that are more sensible? Perhaps tests that are more suitable for non-parametric comparisons?

    Many thanks

  2. #2
    TS Contributor
    Points: 10,570, Level: 68
    Level completed: 30%, Points required for next Level: 280
    Karabiner's Avatar
    Location
    Schalke 04, Germany
    Posts
    1,620
    Thanks
    28
    Thanked 397 Times in 379 Posts

    Re: Post-hoc Bonferroni alternatives

    Why do you call it post-hoc tests? Have there been global tests, with many groups,
    before you conducted the 2-group Mann-Whitney tests?

    Kind regards

    K.

  3. #3
    Points: 256, Level: 5
    Level completed: 12%, Points required for next Level: 44

    Posts
    2
    Thanks
    0
    Thanked 0 Times in 0 Posts

    Re: Post-hoc Bonferroni alternatives

    No, I have just conducted a number of Mann-Whitney tests looking at the difference between two groups on a number of measures.

    Sorry for the confusion, and thank you in advance of any help that you can offer.

  4. #4
    TS Contributor
    Points: 10,570, Level: 68
    Level completed: 30%, Points required for next Level: 280
    Karabiner's Avatar
    Location
    Schalke 04, Germany
    Posts
    1,620
    Thanks
    28
    Thanked 397 Times in 379 Posts

    Re: Post-hoc Bonferroni alternatives


    Well, you give no background information (research question(s), number of tests,
    sample size, relationship of the tested hypotheses, consequences of false-positive or
    false-negative test results), so answers are difficult. In general, I am aware of: no
    correction, a more conservative level of significance than usual (e.g. 1%), Bonferroni,
    Bonferroni-Holm correction, Hochbergs False Discovery Rate (FDR).

    Kind regards

    K.

+ Reply to 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