rejection.level: rejection level for the test statistic

In multivariance: Measuring Multivariate Dependence Using Distance Multivariance

Description

Under independence the probability for the normalized and Nscaled (squared) multivariance to be above this level is less than alpha. The same holds for the normalized, Nscaled and Escaled (squared) total multivariance and m-multivariance.

Usage

rejection.level(alpha)

Arguments

alpha

level of significance

Details

This is based on a distribution-free approach. The value might be very conservative. This is the counterpart to multivariance.pvalue. For a less conservative approach see resample.rejection.level.

The estimate is only valid for alpha smaller than 0.215.

Examples

rejection.level(0.05) #the rejection level, for comparison with the following values
total.multivariance(matrix(rnorm(100*3),ncol = 3)) #independent sample
total.multivariance(coins(100)) #dependent sample which is 2-independent

# and the p values are (to compare with alpha)
multivariance.pvalue(total.multivariance(matrix(rnorm(100*3),ncol = 3))) #independent sample
multivariance.pvalue(total.multivariance(coins(100))) #dependent sample which is 2-independent

## Not run: 
# visualization of the rejection level
curve(rejection.level(x),xlim = c(0.001,0.215),xlab = "alpha")

## End(Not run)

multivariance documentation built on Oct. 6, 2021, 5:08 p.m.