simul ants with no ties [Nonparametrics]

posted by d_labes  – Berlin, Germany, 2010-09-16 13:19 (5261 d 15:08 ago) – Posting: # 5917
Views: 21,656

Dear Helmut!

❝ Yep. For n1=n2=10 I would expect 0.9108...


Yeah, but ... without ties.
10 000 sims each with round(,5) in the code and only wilcox.exact():
run 1  0.9096
run 2  0.9076
run 3  0.9125
run 4  0.9104
run 5  0.9138
-------------
mean   0.91078

Each run (10 000 sims) only approx. 100 s!

The mean seems very close to your expectation :cool:.
---[nitpicking]---------------------------------
> 1-2*pwilcox(max(1,qwilcox(0.1/2,10,10))-1,10,10)
[1] 0.9107904
---[/nitpicking]--------------------------------


What I'm very curious of is that scatter between the runs. Seems that one has to simulate at least 100 000 or more datasets to get a reliable coverage.
To much beer in case of coin:wilcox_exac() :-D.

BTW: with round to five decimals I have observed anyhow up to 10 sets with ties among the 10 000 sims.

Regards,

Detlew

Complete thread:

UA Flag
Activity
 Admin contact
23,379 posts in 4,913 threads, 1,661 registered users;
289 visitors (0 registered, 289 guests [including 14 identified bots]).
Forum time: 03:27 CET (Europe/Vienna)

Statistics is the art of never having to say you’re wrong.
Variance is what any two statisticians are at.    C.J. Bradfield

The Bioequivalence and Bioavailability Forum is hosted by
BEBAC Ing. Helmut Schütz
HTML5