[SciPy-User] MannWhitneyU test critical values

[josef.pktd at gmail.com]

On Thu, Sep 9, 2010 at 5:14 AM, <kossnair at web.de> wrote:

>  Hi there,
> I just choke on a trivial problem (I think):
> I have 13 test data sets and two methods, that return a measure of
> performance. (higher value is better)
> Now I want to use scipy.stats.mannwhitneyu for significance testing.
> However, I did not find a possibility to retrieve a critical value.
> If I am right the method returns a critical value, that has to be compared
> with a tabulated value.
> Can anybody point me towards a way of reading this tabulated value in a
> python script?
> Thank you very much in advance,
> kossnair
>

All statistical tests (except anderson-darling) return p-values, which you
can use to decide whether to reject or not to reject the null hypothesis.
For example, a pvalue below 0.05 means, the null hypothesis can be rejected
with an alpha of 5 % (confidence level 1-alpha).

For the actual critical values, we have to look at the source to see which
distribution for the test-statistic is used. I don't remember what
mannwhitneyu uses, and I don't have time until later to look it up.


Josef




>
>
> Neu: WEB.DE De-Mail - Einfach wie E-Mail, sicher wie ein Brief!
> Jetzt De-Mail-Adresse reservieren: https://produkte.web.de/go/demail02
>
> _______________________________________________
> SciPy-User mailing list
> SciPy-User at scipy.org
> http://mail.scipy.org/mailman/listinfo/scipy-user
>
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mail.scipy.org/pipermail/scipy-user/attachments/20100909/41da34ea/attachment.html>