[SciPy-user] help with scipy.stats.mannwhitneyu
josef.pktd@gmai...
josef.pktd@gmai...
Thu Feb 5 08:33:05 CST 2009
On Thu, Feb 5, 2009 at 6:37 AM, Wavy Davy <bloodearnest@gmail.com> wrote:
> Hi all
>
> I am using the mannwhitneyu in the stats module, and I was looking the
> code and I see this notice in the docstring.
>
> "Use only when the n in each condition is < 20 and you have 2
> independent samples of ranks. "
>
> Am I reading it correctly that this test should only be used with
> sample sizes less than 20?
>
> I am not a statistican, more a python coder. I have been pointed and
> this test as a more robust version of the t-test, so forgive my
> ignorance.
>
> Any help would be much appreciated.
>
> --
> Simon
I briefly looked at the test, the implementation of the test statistic
is mostly as described in
http://en.wikipedia.org/wiki/Mann-Whitney_U_test
It seems the test statistic is defined with the opposite sign from the
definition in wikipedia.
The doc string statement "Use only when the n in each condition is <
20", I think should be >20, since the pvalue is based on the
asymptotic distribution, which is only correct in larger samples.
I didn't see any unit tests for this test, but I will try to verify
the results later today.
wilcoxon is a similar test for paired instead of independent samples,
and there the recommendation in the docstring is for N>20.
Josef
More information about the SciPy-user
mailing list