Equality2Cov: Testing the Equality of several Covariance Matrices

In andresfpc/AMUN: Analisis Multivariado Universidad Nacional de Colombia (AMUN)

Description

To test the null hypothesis of equality of covariance of a multinormal population, a slight change over the likelihood-ratio statistic is made, and it is recommended if p and k do not exceed four or five and each n_i is twenty or more (Morrison, 2005).

Usage

Equality2Cov(S, S1, S2, n1, n2)

Arguments

S	a positive define matrix which contains the whole variances and covariances
S1	a positive define matrix related which contains the variances and covariances of the first group
S2	a positive define matrix related which contains the variances and covariances of the second group
n1	the number of observation in the first group
n2	the number of observations in the second group

Value

ChiSquareStatistic	the value of the Chi Square Statistic
DegreeOfFreedom	total normal of degrees of freedom of the Chi Square Statistic
pValue	the p value of the Chi square statistic

Author(s)

Jesus Gonzalez <[e-mail]>, Andres Palacios <[e-mail]>, Campo Elias Pardo <[e-mail]>

References

Morrison, D. F. (2005), Multivariate statistical methods, Series in Probabilty and Statistics, 4 edn, McGraw-Hill, New York

Examples

# Example 5.3 (Morrison, 2005)
S_M <- matrix(c(4.32, 1.88,
                1.88, 9.18), nrow = 2, byrow = TRUE)

S_F <- matrix(c(2.52, 1.90,
                1.90, 10.06), nrow = 2, byrow = TRUE)

S <- matrix(c(3.42, 1.89,
              1.89, 9.62), nrow = 2, byrow = TRUE)

Equality2Cov(S = S, S1 = S_M, S2 = S_F, n1 = 32, n2 = 32)

andresfpc/AMUN documentation built on May 12, 2019, 3:36 a.m.