iSixSigma

p value = 0 ?

Six Sigma – iSixSigma › Forums › Old Forums › General › p value = 0 ?

Viewing 15 posts - 1 through 15 (of 15 total)
  • Author
    Posts
  • September 23, 2006 at 2:47 pm #44683

    MC
    ★ 10 Years ★
    Participant

    I would like to know, what does exactly a p value of null mean. I´m wondering if the programs performed well! Is it perhaps rather an indication that my objects of study are more than highly significant, i.e. totally different? Should I better make further test in order to get some positive value, although it will be very small?Thanks!

    0
    September 23, 2006 at 4:17 pm #143743

    Picklyk
    ★ 10 Years ★
    Participant

    Hi MC,
    If the p is below 0.05 you can say that based on your test of a sample, looking at the population your subgroups are significantly different from one another.
    If you have a p value of 0 it just means that there is 0% chance of you making an alpha error; meaning there is a 0 per cent chance of you stating that the subgroups of your population are different when in fact they are not.
    Jay
     

    0
    September 23, 2006 at 6:46 pm #143745

    Quainoo
    ★ 10 Years ★
    Member

    I have a feeling that the p value cannot really be equal to zero.
    It can get very close to it (very small value) but not exactly zero.
    Vincent
     

    0
    September 23, 2006 at 10:14 pm #143755

    Dr Casey
    ★ 10 Years ★
    Participant

    Diagnosis – urethritis.

    0
    September 23, 2006 at 10:56 pm #143756

    MC
    ★ 10 Years ★
    Participant

    Thank you Jai and Vincent for your answer. Now I will sleep better, although I had also the same feeling as Vincent. As p is an area, perhaps you can say it tends to null when x tends to infinite. I guess, the statistical programs say p = 0 depending on how did you set your number of decimals.
    Any way, I think we three agree.Thanks :))

    0
    September 23, 2006 at 11:13 pm #143757

    Orang_Utan
    ★ 10 Years ★
    Participant

    How you get P value=0? No way!

    0
    September 24, 2006 at 1:41 pm #143766

    Szentannai
    ★ 10 Years ★
    Member

    Hi,
    AFAIK the statistical programs will display tge p value to a precision of 3 or decimals only. If p < 0.0001 all they will show is a zero. So what you see probably means only that the p is so small that the program did not bother to print the exact value and not that it is strictly 0.Regards
    Sandor

    0
    September 24, 2006 at 2:22 pm #143767

    MC
    ★ 10 Years ★
    Participant

    Hello! and thank you for the answers regarding the p = 0. Jai, I could sleep better, thinking that my populations were in fact different! Vincent, I agree with you also, perhaps it´s a matter of properly defining. As p is in fact an area, would it be propperlier to say that p tends to null if X tends to infinite? (also for Orang) Thank you Sandor, I was also looking for some explanation – experience like that. I also agree with you, but look, the precission of my programs is like this:p 0.00795
    p 0.00000 So the p value must be really very small, don´t you agree!?Have a nice sunday forum!

    0
    September 24, 2006 at 2:51 pm #143769

    Szentannai
    ★ 10 Years ★
    Member

    Hi Mc,
    I also think that your p value is very small but not zero.
    Have a nice weekend too!
    Regards
    Sandor

    0
    September 28, 2006 at 2:02 pm #143941

    Ron
    ★ 10 Years ★
    Member

    Hi,
    The p value is used to determine if enough evidence exists to reject the null hypothesis in favor of the alternate hypothesis.
    The p value is the probability of incorrectly rejecting the null hypothesis.
    Simply remember, “if p is low (often <0.05) Ho must go."
    Also, we never “accept” the null or alternate hypothesis… rather we “reject” or “fail to reject.”  This is a common error I have seen many new practitioners make.
    Finally, I have seen the discussion below of whether p can be equal to 0 or very close to zero… the answer is yes it can and I can send actual data sets to prove this.  Don’t fret over this… else you run the risk of entering statistical analysis paralysis.
    I hope this helps.

    0
    September 28, 2006 at 4:29 pm #143951

    Clowers
    ★ 10 Years ★
    Participant

    Use the P value as the percentage. If the P=0, subtract that from 100% and you are 100% confident that there is a statistical significance in the data you tested. Rejecting the NULL (that there is no difference) and ACCEPTING the alternative (that there is a difference) P=0.05, then you are 95% confident that the data is statistical. if the P=.5, then 50% really isn’t statistically significant.

    0
    October 25, 2006 at 2:18 pm #145677

    Bode
    ★ 10 Years ★
    Participant

    You sound like one of my instructors.  Have you ever been to Suffolk, VA?

    0
    October 25, 2006 at 2:23 pm #145679

    Ron
    ★ 10 Years ★
    Member

    No, never taught in the great state of VA.  But glad to hear your instructor and I agree!
    Ron

    0
    April 22, 2008 at 8:08 pm #171391

    arghh
    ★ 10 Years ★
    Participant

    The P Value should not be ‘Null’ it is zero.  There’s a difference.

    0
    April 23, 2008 at 7:50 pm #171419

    Outlier, MDSB
    ★ 10 Years ★
    Participant

    I have 2 questions:
    1. What does your response mean? It doesn’t make sense.
    2. Why are you randomly responding to a 2-year old post?

    0
  • Author
    Posts
Viewing 15 posts - 1 through 15 (of 15 total)

The forum ‘General’ is closed to new topics and replies.