MBoxtstwod

Multivariate statistical testing for the homogeneity of covariance matrices without data by the Box.

Description

From the covariance matrices (without data), groups size vector and a significance level (default = 0.05), the homogeneity covariance matrices are testing by the Box's M, it gives a Chi-square or F approximation in function of the groups sample-size.

Outputs are the Box's M statistic, approximation statistic test, number of groups, number of variables, degrees of freedom of the approximation statistic test and P-value.

• MATLAB release: MATLAB 5.3 (R11)

[Kadir]

Antonio,

There's a contradiction on the variables P1 & P2 in the following part of the script:

fprintf('%10.4f%11.4f%8.i%10.i%13.i%14.i%13.4f\n',MB,F2,g,p,v1,v22,P2);
        fprintf('--------------------------------------------------------------------------------\n');
        fprintf('With a given significance level of:% 3.2f\n', alpha );
        if P1 >= alpha
            disp('Covariance matrices are not significantly different.');
        else
            disp('Covariance matrices are significantly different.');
        end;

Agree?