Dismiss Notice
Join Physics Forums Today!
The friendliest, high quality science and math community on the planet! Everyone who loves science is here!

Mann-Whitney p = 1.0!

  1. Mar 14, 2009 #1
    Hey everyone,
    I got z = 0, and therefore p = 1.0 in a Mann-Whitney U test. Considering this is impossible, should I instead report z = 0.0001 and p = 0.99?
    Also, I have a different number of participants in my two conditions, I ran a parametric independent samples t-test, does anyone know if SPSS (the program I used) corrects for this?
    Thanks in advance.
  2. jcsd
  3. Sep 13, 2009 #2
    I don't think SPSS sorts for that.

    How have you entered your data in SPSS, since that might sometimes do the trick. You might just have entered it the wrong way. Try to transpose your data and see what SPSS then comes up with if you do the same...
  4. Sep 13, 2009 #3
    Thanks for the reply sander.
    I can’t remember how I entered it into SPSS as it was awhile ago now. I 'remember' (for what episodic memory is worth!) making several attempts. I normally do. I also did the calculations by hand (as was standard unless there was a lot of participants, in which case its far to laborious and more prone to errors). If I recall, it came close to z=0, but wasn't quite so; would this make sense? I'm pretty sure SPSS simply took it to be z=0, when it clearly couldn't have been.
    I think I resolved to put P=0.99 on the assignment.
  5. Sep 13, 2009 #4
    Hmmm now thats odd. SPSS is some weird stuff, it works just totally against your instincts.

    I think resolving the problem by saying p=0.99 doesnt solve the problem...

    It might be that your data is rounded, which might just make the difference between 0.99 and 1.00. SPSS usually uses 2.2 significance. (ie 12.34 -> 12.3456 will become 1.36).
Share this great discussion with others via Reddit, Google+, Twitter, or Facebook