Multinomial logit bias reduction via the Poisson log-linear model
- 13 July 2011
- journal article
- research article
- Published by Oxford University Press (OUP) in Biometrika
- Vol. 98 (3) , 755-759
Abstract
For the parameters of a multinomial logistic regression, it is shown how to obtain the bias-reducing penalized maximum likelihood estimator by using the equivalent Poisson log-linear model. The calculation needed is not simply an application of the Jeffreys prior penalty to the Poisson model. The development allows a simple and computationally efficient implementation of the reduced-bias estimator, using standard software for generalized linear models.Keywords
This publication has 8 references indexed in Scilit:
- Bias reduction in exponential family nonlinear modelsBiometrika, 2009
- Confidence intervals for multinomial logistic regression in sparse dataStatistics in Medicine, 2006
- A solution to the problem of separation in logistic regressionStatistics in Medicine, 2002
- A modified score function estimator for multinomial logistic regression in small samplesComputational Statistics & Data Analysis, 2002
- Bias reduction of maximum likelihood estimatesBiometrika, 1993
- The Fisher information matrix for log linear models arguing conditionally on observed explanatory variableBiometrika, 1981
- Maximum Likelihood in Three-Way Contingency TablesJournal of the Royal Statistical Society Series B: Statistical Methodology, 1963
- An invariant form for the prior probability in estimation problemsProceedings of the Royal Society of London. A. Mathematical and Physical Sciences, 1946