Mann-Whitney U Test: p=1.0 - Should I Report z = 0.0001?

  • Context: Graduate 
  • Thread starter Thread starter nobahar
  • Start date Start date
Click For Summary
SUMMARY

The discussion centers on the interpretation of results from a Mann-Whitney U test, where a participant reported a z-value of 0, resulting in a p-value of 1.0. The consensus suggests that reporting a z-value of 0.0001 and a p-value of 0.99 may be more appropriate, as a z-value of 0 is statistically implausible. Additionally, participants discussed the potential issues with data entry in SPSS, particularly regarding the handling of unequal sample sizes and rounding errors that could affect the results.

PREREQUISITES
  • Understanding of the Mann-Whitney U test and its statistical implications
  • Familiarity with SPSS software for statistical analysis
  • Knowledge of parametric vs. non-parametric tests
  • Basic concepts of statistical significance and p-values
NEXT STEPS
  • Research the correct interpretation of z-values in non-parametric tests
  • Learn about data entry best practices in SPSS to avoid common errors
  • Explore the implications of unequal sample sizes in statistical testing
  • Study the effects of rounding on statistical results and reporting
USEFUL FOR

Statisticians, researchers conducting non-parametric tests, students learning about statistical analysis, and anyone using SPSS for data analysis.

nobahar
Messages
482
Reaction score
2
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.
 
Physics news on Phys.org
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...
 
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.
 
nobahar said:
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.

Hmmm now that's odd. SPSS is some weird stuff, it works just totally against your instincts.

I think resolving the problem by saying p=0.99 doesn't 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).
 

Similar threads

Undergrad Mann-whitney U test? math n00b?
  • · Replies 1 ·
Replies
1
Views
3K
How to take into consideration errors in the Mann-Whitney test?
  • · Replies 13 ·
Replies
13
Views
3K
Undergrad Counting U Test for Mann-Whitney Test & Finding Z Value
  • · Replies 9 ·
Replies
9
Views
3K
Graduate MOND Theory Primer: Stacy McGaugh's Research Program Explored
  • · Replies 11 ·
Replies
11
Views
4K
  • Sticky
Physics Forums Global Guidelines
  • · Replies 2 ·
Replies
2
Views
504K