[R] wilcox.test and ties
Timo.Makelainen at helsinki.fi
Fri Jul 19 17:59:51 CEST 2002
On 19 Jul 2002, at 15:16, Swanepoel, Christiaan Edward wrote:
> when performing the wilcox.test on a sample (n=100) with the following
> frequency distribution: Value | Frequency
> ---------------------------------------- 0 | 10 0.1 | 88 1 | 1 2.6 | 1
> I get the following error message:
> Error in uniroot(wdiff, c(mumin, mumax), tol = 1e-04, zq =
> qnorm(alpha/2, :
> f() values at end points not of opposite sign
> The error only occurs when I specify that the confidence interval
> should be computed.
> Could anybody suggest how I can compute the confidence interval for
> this sample? Is the problem connected with the large number of ties
> and if so, how can I handle these ties?
>From the help:
"By default (if exact is not specified), an exact p-value is computed
if the samples contain less than 50 finite values and there
are no ties. Otherwise, a normal approximation is used."
The error message tells that a normal approximation is being
attempted. On the other hand, I would rather not rely on a normal
approximation here. With so few observations differing from 0.1, a
hand computation might be easy for a high confidence level.
In general, for the correct handling of ties entering data like yours
as round( x, 1) can be tried.
> Thanks in advance.
> -.-.-.-.- r-help mailing list -- Read
> http://www.ci.tuwien.ac.at/~hornik/R/R-FAQ.html Send "info", "help",
> or "[un]subscribe" (in the "body", not the subject !) To:
> r-help-request at stat.math.ethz.ch
Timo Makelainen, Department of Mathematics
University of Helsinki, P.O. Box 4
E-mail: Timo.Makelainen at Helsinki.Fi
r-help mailing list -- Read http://www.ci.tuwien.ac.at/~hornik/R/R-FAQ.html
Send "info", "help", or "[un]subscribe"
(in the "body", not the subject !) To: r-help-request at stat.math.ethz.ch
More information about the R-help