jomaovhet

The multivariate analysis of variance (MANOVA) is the statistical procedure used to comparing the mean vectors of several multivariate normal populations. If the covariance matrices are assumed to be equal, there are several tests available to test the equality of the mean vectors. The Hotelling T2 test is a powerful invariant test used when there are only two vector population means to compare. According to Krishnamoorthy and Lu (2010), this test may be become seriously biased when the assumption of equality of covariance matrices is not satisfied. Moreover, it is very unlikely that the assumption of covariance equality can be satisfied in practice.

Sevaral test have been proposed to the Behrens-Fisher problem. Among them, there is the Johansen's (1980) procedure proposed as a multivariate test when the covariance matrices are unequal. It generalizes the Welch's (1938; 1951) univariate approximate degrees of freedom solution.

Here, a m-file analytical procedure using the Johansen's test is developed. It works with two or more multivariate samples under heteroscedasticity.

JOMAOVHET treats NaN values as missing values, and removes them.

Syntax: function jomaovhet(X,alpha)

Inputs:
X - data matrix (Size of matrix must be n-by-(1+p); sample=column 1, variables=column 2:p)
alpha - significance level (default = 0.05)

Output:
- Whether or not the equality of mean vectors was met