[R-sig-eco] data error using "npmc" package
yangwenjing
yangwenjing at ibcas.ac.cn
Fri Feb 4 10:47:04 CET 2011
Dear list members:
I am using the function”npmc” (package “npmc”) to do non-parametric multiple comparison test. But, there are always some error happened to my data. I will illustrate my problems using simulated data.
I create a data frame with four groups, and then do non-parametric pairwise comparison test among the groups.
> mydata <- data.frame( class = factor(c(rep("I", 5), rep("II", 5),rep("III",5), rep("IV", 5))),
+ var = c(1, 1, 1, 1, 1,
+ 2, 2, 2, 2, 2,
+ 5, 5, 5, 5, 5,
+ 4, 4, 4, 4, 4))
> npmc(mydata)
Error in probval.GenzBretz(algorithm, n, df, lower, upper, infin, corr, :
NAs in foreign function call (arg 2)
In addition: Warning message:
In probval.GenzBretz(algorithm, n, df, lower, upper, infin, corr, :
NAs introduced by coercion
But if I change one number in the third group, the error does not happen any more.
> mydata <- data.frame( class = factor(c(rep("I", 5), rep("II", 5),rep("III",5), rep("IV", 5))),
+ var = c(1, 1, 1, 1, 1,
+ 2, 2, 2, 2, 2,
+ 5, 5, 5, 5, 3,
+ 4, 4, 4, 4, 4))
> npmc(mydata)
$BF
cmp gn effect lower.cl upper.cl variance std statistic
1 1-2 10 1.0 0.9998619 1.000138 1e-08 3.162278e-05 15811.39
2 1-3 10 1.0 0.9998619 1.000138 1e-08 3.162278e-05 15811.39
3 1-4 10 1.0 0.9998619 1.000138 1e-08 3.162278e-05 15811.39
4 2-3 10 1.0 0.9998619 1.000138 1e-08 3.162278e-05 15811.39
5 2-4 10 1.0 0.9998619 1.000138 1e-08 3.162278e-05 15811.39
6 3-4 10 0.2 -0.6732195 1.073220 4e-01 2.000000e-01 -1.50
p.value.1s p.value.2s zero
1 0.0000000 0.0000000 TRUE
2 0.0000000 0.0000000 TRUE
3 0.0000000 0.0000000 TRUE
4 0.0000000 0.0000000 TRUE
5 0.0000000 0.0000000 TRUE
6 0.9999252 0.6256789 FALSE
It seems hard to understand, because I only change one number in the third group. Is there anyone who can tell me the reasons?
Thanks a lot.
Regards,
Wenjing
More information about the R-sig-ecology
mailing list