TESTING THE EQUALITY OF THE CORRELATION MATRICES WITH THE PERMUTATION TEST
Öz
In this study, the emprical rejection levels of the Shipley's test based on permutations are calculated by using the Monte Carlo procedure in order to test the hypothesis of the equality of correlation matrices of different populations, involving p variables. These values are obtained for normal and mixtures of normal distributed three population and are investigated when the expected rejection level is 0.05, 0.10 and 0.20 and the sample size differs from 3 to 10. In respect of the simulation results, the emprical rejection levels of the normal data seems to be very close to the expected rejection levels.
Separately, since all the results are smaller than a in case of mixtures of normal data, the test statistic appears to be robust for non-normal (mixtures of normal) data too.
Separately, since all the results are smaller than a in case of mixtures of normal data, the test statistic appears to be robust for non-normal (mixtures of normal) data too.
Anahtar Kelimeler
Kaynakça
- Krzanowski, W.J. (1993). Permutational tests for correlation matrices. Statistics and Computing 3, 37- 44.
- Manly, B.F., Rayner, J.C.W. (1987). The comparison of sample covariance matrices using likelihood ratio tests. Biometrika 74, 841-847.
- Riska, B. (1985). Group size factors and geographic variation of morphometric correlation. Ecology 39, 792-803.
- Sakaori, F. (2002). Permutation test for equality of correlation coefficients in two populations. Commun. Statist.- Simula. 31(4), 641-651.
- Shipley, B. (2000). A Permutation procedure for testing the equality of pattern hypotheses across groups involving correlation or covariance matrices. Statistics and Compu- ting 10, 253-257.
- Smith, H., Gnanadesikan, R., Hughes, J.B. (1962). Multivariate analysis of variance (MANOVA). Biometrics. 18(1), 22-41.