Modelling structured correlation matrices Academic Article uri icon


  • 2017 Biometrika Trust. Ensuring positive definiteness of an estimated structured correlation matrix is challenging. We show that reparameterizing Cholesky factors of correlation matrices using hyperspherical coordinates or angles provides a flexible and effective solution. Once a structured correlation matrix is identified, the corresponding angles and hence the constrained correlations may be estimated by maximum likelihood. Consistency and asymptotic normality of the maximum likelihood estimators of the angles are established. Examples demonstrate the flexibility of the method.

published proceedings


author list (cited authors)

  • Tsay, R. S., & Pourahmadi, M.

citation count

  • 15

complete list of authors

  • Tsay, Ruey S||Pourahmadi, Mohsen

publication date

  • March 2017