varjmcm {varjmcm}

varjmcm: Estimations for the Covariance of Estimated Parameters in Joint Mean-Covariance Models

Description

The package provides estimations of the covariance of estimated parameters in joint mean-covariance models, which is fitted in 'jmcm' package. Two methods are available. bootcovjmcm calculates the covariance estimation via a bootstrap based method. covjmcm uses explicit formula, i.e. the inverse of the estimated Fisher's information, to calculate the covariance estimation. The bootstrap method may need large number of replications and thus may be time consuming. The explicit formula in the second method is asymptotically correct, and thus is valid only when the sample size is large.

References

[1] Pan J, Pan Y (2017). "jmcm: An R Package for Joint Mean-Covariance Modeling of Longitudinal Data." Journal of Statistical Software, 82(9), 1–29.

[2] Pourahmadi, M., "Maximum likelihood estimation of generalised linear models for multivariate normal covariance matrix," Biometrika 87(2), 425–435 (2000).

[3] M. Maadooliat, M. Pourahmadi and J. Z. Huang, "Robust estimation of the correlation matrix of longitudinal data", Statistics and Computing 23, 17-28, (2013).

[4] W. Zhang, C. Leng, and C. Y. Tang(2015), "A joint modelling approach for longitudinal studies," Journal of the Royal Statistical Society. Series B. 77, 219-238.

See Also

covjmcm and bootcovjmcm for more details and examples.