Comparison among three criteria to test the equality of correlation coeficients

Authors

  • José Ruy Porto Carvalho
  • Roger Mead

DOI:

https://doi.org/10.1590/S1678-3921.pab1992.v27.3712

Keywords:

maximum likelihood estimator, likelihood ratio test, split-plot analysis, correlation coefficients, observed power

Abstract

In this paper, the Likelihood Ratio Test is presented. It is shown to be an efficient procedure to test the hypothesis of equality of residual correlation coefficients. In comparing the Likelihood Ratio Test with the largely used U-Fisher statistic test, or the improved Rao version of it for testing che hypothesis of equality of residual correlation coefficients, very interesting result are found. The simulated and experimental results show that the Likelihood Ratio Test rejects the null hypothesis in more experimental situations than the U-Fisher or H-Rao test statistics, with grater observed power.

Published

1992-06-01

How to Cite

Carvalho, J. R. P., & Mead, R. (1992). Comparison among three criteria to test the equality of correlation coeficients. Pesquisa Agropecuaria Brasileira, 27(6), 785–795. https://doi.org/10.1590/S1678-3921.pab1992.v27.3712

Issue

Section

STATISTICS