tmax in case of ties: R vs. R vs. SAS [Nonparametrics]

posted by Jack Homepage – Lancaster, United Kingdom, 2010-09-20 16:03 (5257 d 11:06 ago) – Posting: # 5924
Views: 21,377

Dear d_labes,

here is my attempt to shed some light into the difference in the point estimates.

The coin packages uses the midpoint of the confidence interval as the point estimate for the location difference (ie (0.5-1.5)/2=-0.5 while the exactRankTests packages uses (sort of) the median of the observed differences as the point estimate. Note that the exact procedure of exactRankTests is quite situational which is why the difference in point estimates only occurs sometimes.

The real issue from my point of view is that it is hard (impossible?) to decide which is the "correct" estimate. Intuition tells me that the exactRankTest version is the one that is closer to the spirit of the Wilcoxon test while practicality certainly speaks for the midpoint of the CI to be used.

As for the differences in confidence intervals I presume that it has to do with either the approximation which is uses if n>50 and/or the continuity correction.

Complete thread:

UA Flag
Activity
 Admin contact
23,379 posts in 4,913 threads, 1,661 registered users;
265 visitors (0 registered, 265 guests [including 23 identified bots]).
Forum time: 02:10 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