+ Reply to Thread
Page 3 of 3 FirstFirst 1 2 3
Results 31 to 37 of 37

Thread: Please help? Complete novice

  1. #31
    Fortran must die
    Points: 58,790, Level: 100
    Level completed: 0%, Points required for next Level: 0
    noetsi's Avatar
    Posts
    6,532
    Thanks
    692
    Thanked 915 Times in 874 Posts

    Re: Please help? Complete novice




    It is not generally a good idea to base your views on an issue on the writing of a single statistician. I have read books that argue that ANOVA, for example, is virtually useless given outliers should they occur. And that you essentially had to use bootstraping to utilize any GLM. To cite mearly one opinion. I think you have to look at the common usage in the field, not a single report.
    "Very few theories have been abandoned because they were found to be invalid on the basis of empirical evidence...." Spanos, 1995

  2. #32
    Human
    Points: 12,676, Level: 73
    Level completed: 57%, Points required for next Level: 174
    Awards:
    Master Tagger
    GretaGarbo's Avatar
    Posts
    1,362
    Thanks
    455
    Thanked 462 Times in 402 Posts

    Re: Please help? Complete novice

    Quote Originally Posted by noetsi View Post
    It is not generally a good idea to base your views on an issue on the writing of a single statistician.
    OK, now I have cited one paper – published in The American Statistician - in support or the statement I made.

    If we would accept your demand of several sources, then it up to Noetsi to show several papers that support the idea of post hoc power.

    I don’t agree that the majority decides what is allowed. Then no progress would be possible.

    But, Noetsi, I might quote you in the future about this demand.
    Last edited by GretaGarbo; 08-24-2012 at 04:00 PM.

  3. #33
    TS Contributor
    Points: 18,889, Level: 87
    Level completed: 8%, Points required for next Level: 461
    CowboyBear's Avatar
    Location
    New Zealand
    Posts
    2,062
    Thanks
    121
    Thanked 427 Times in 328 Posts

    Re: Please help? Complete novice

    Quote Originally Posted by noetsi View Post
    It is not generally a good idea to base your views on an issue on the writing of a single statistician. I have read books that argue that ANOVA, for example, is virtually useless given outliers should they occur. And that you essentially had to use bootstraping to utilize any GLM. To cite mearly one opinion. I think you have to look at the common usage in the field, not a single report.
    I agree that basing views on one person's writing is not always a great idea, but I think that "common usage" is probably an even more problematic criterion to judge something by. What techniques are in common use is determined by a lot of factors, several of them having nothing remotely to do with whether the techniques are actually sensible or useful. E.g. factors like: Is it easy to do in SPSS? Is it what my advisor does? Will it help give me the conclusions I want? Does it make my study look more "sophisticated"? Will reviewers understand it?

    Nil* hypothesis signifance testing is a good example of a set of analyses that are incredibly commonly used despite having massive problems associated with them, and a body of literature stretching back 60+ years pointing out those problems. But (nearly) everyone keeps using them because (nearly) everyone else keeps using them. (Edit - myself included, to be fair!)

    *I say "nil" not null intentionally - I mean null hypothesis significance testing in which the null hypothesis is a zero effect and the researcher wants to reject it in favour of some vague alternative hypothesis. The strong form of NHST where the researcher tests a specific null hypothesis implied by theory to see if there is evidence to falsify it is a bit of a different story.

  4. The Following User Says Thank You to CowboyBear For This Useful Post:

    victorxstc (08-14-2012)

  5. #34
    Pirate
    Points: 15,159, Level: 79
    Level completed: 62%, Points required for next Level: 191
    victorxstc's Avatar
    Posts
    875
    Thanks
    229
    Thanked 332 Times in 297 Posts

    Re: Please help? Complete novice

    Quote Originally Posted by CowboyBear View Post
    Nil* hypothesis signifance testing is a good example of a set of analyses that are incredibly commonly used despite having massive problems associated with them, and a body of literature stretching back 60+ years pointing out those problems. But (nearly) everyone keeps using them because (nearly) everyone else keeps using them. (Edit - myself included, to be fair!)
    Thanks, I agree that as stated above (and in our previous conversation) P values and relying only on P values being smaller or greater than 0.05 should be replaced with effect size measures.
    Last edited by victorxstc; 08-14-2012 at 06:54 PM. Reason: ficing the terrible grammar of an exploded multitasker brain at 4:24 am!! :p

  6. #35
    Fortran must die
    Points: 58,790, Level: 100
    Level completed: 0%, Points required for next Level: 0
    noetsi's Avatar
    Posts
    6,532
    Thanks
    692
    Thanked 915 Times in 874 Posts

    Re: Please help? Complete novice

    Personally I think that test of the null hypothesis are most troubled by the fact that they focus not on substantive effect size, but p test. So its commonly assumed if the p value is below alpha, to be simple, then the discovered effect size is important. Which is entirely wrong. I did a presentation to a federal administrator where one of the slides said "Ok its signficant, does it matter?"

    A variety of important methodologist such as Campbell have argued null hypotheisis test be abandoned for this reason

    The point about common usage is a good one. But the issue remains, for non-statisticians especially, what if you come across a writer that advocates radical change (such as essentially abandonding ANOVA without bootstraping) that few if any in the field appear to support? You can make a really good case for something that is wrong, especially to non experts
    "Very few theories have been abandoned because they were found to be invalid on the basis of empirical evidence...." Spanos, 1995

  7. #36
    Human
    Points: 12,676, Level: 73
    Level completed: 57%, Points required for next Level: 174
    Awards:
    Master Tagger
    GretaGarbo's Avatar
    Posts
    1,362
    Thanks
    455
    Thanked 462 Times in 402 Posts

    Re: Please help? Complete novice

    Quote Originally Posted by noetsi View Post
    It is not generally a good idea to base your views on an issue on the writing of a single statistician.
    Quote Originally Posted by CowboyBear View Post
    I agree that basing views on one person's writing is not always a great idea,
    In 1687 it was very good to just base views on one person: Isaac Newton.

    In 1859 it was very good to just base views on one person: Charles Darwin.

    In 1905 it was very good to just base views on one person: Albert Einstein.

    In 1922 it was very good to just base views on one person: R A Fisher.

    It is not the size of a crowd that make an idea important; it is the strength of the argument.

    And I think that is the way science advances. Come on! Question authority!

    (Besides I am still waiting for Noetsi to put forward several papers supporting the idea of post hoc power.)

  8. The Following User Says Thank You to GretaGarbo For This Useful Post:

    CowboyBear (08-14-2012)

  9. #37
    Pirate
    Points: 15,159, Level: 79
    Level completed: 62%, Points required for next Level: 191
    victorxstc's Avatar
    Posts
    875
    Thanks
    229
    Thanked 332 Times in 297 Posts

    Re: Please help? Complete novice


    Dear Greta, I just wanted to acknowledge the very graphic and neat way of expressing the details of your thoughts.

    In 2012, it is very good to take the insightful views of one person, really seriously: Greta Garbo.

+ Reply to Thread
Page 3 of 3 FirstFirst 1 2 3

           




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