Weighted least squares estimation for exchangeable binary data
- 124 Downloads
Parametric models of discrete data with exchangeable dependence structure present substantial computational challenges for maximum likelihood estimation. Coordinate descent algorithms such as the Newton’s method are usually unstable, becoming a hit or miss adventure on initialization with a good starting value. We propose a method for computing maximum likelihood estimates of parametric models for finitely exchangeable binary data, formalized as an iterative weighted least squares algorithm.
KeywordsExchangeability Completely monotonic function Weighted least squares Excess risk Benchmark dose
- Bowman D, George EO (2015) Likelihood estimation for exchangeable multinomial data. Commun Stat Theory Methods. doi: 10.1080/03610926.2015.1053934
- Carter RE, Zhang X, Woolson RF, Apfel CC (2009) Statistical analysis of correlated relative risks. J Data Sci 7:397–407Google Scholar
- George EO, Bowman D, An Q (2013) Regression models for analyzing clustered binary and continuous outcomes under the assumption of exchangeability. In: deLeon A, Chough K (eds) Analysis of mixed data: methods and applications. CRC Press, New York, pp 93–107Google Scholar
- Rosen M (1999) A weighted least squares approach to estimating the per unit agreement rate in clustered binary data. In: Proceedings of the biometrics section of the American Statistical Association, pp 65–70Google Scholar