期刊名称：Brazilian Journal of Probability and Statistics
印刷版ISSN：0103-0752
出版年度：2005
卷号：19
期号：1
页码：13-31
出版社：Brazilian Statistical Association
摘要：We propose a class of regression models where the responseis beta distributed and the two parameters that index the beta distributionare related to covariates and regression parameters. The proposed class ofmodels is useful for modeling data that are restricted to the (0, 1) interval.We discuss maximum likelihood estimation of the parameters that define theregression structure of the model, and derive closed-form expressions for thesecond order biases of these estimators. The derived expressions are then usedto define bias-corrected maximum likelihood estimators. Simulation resultsshow that the bias correction scheme yields nearly unbiased estimators
关键词：Beta distribution; bias correction; maximum likelihood es-;timation; regression
