A New Generalized Ordinal Logit Model for Multicategory Response Data

被引:0
|
作者
Jamroenpinyo, Somsri [1 ]
O'Brien, Timothy E. [2 ]
Bumrungsup, Chinnaphong [1 ]
机构
[1] Thammasat Univ, Fac Sci & Technol, Dept Math & Stat, Pathum Thani 12121, Thailand
[2] Loyola Univ, Dept Math Sci, Chicago, IL 60626 USA
来源
THAILAND STATISTICIAN | 2012年 / 10卷 / 01期
关键词
adjacent-categories; baseline-category logits; continuation-ratios; multinomial distribution; nominal responses; ordinal responses; proportional odds;
D O I
暂无
中图分类号
O21 [概率论与数理统计]; C8 [统计学];
学科分类号
020208 ; 070103 ; 0714 ;
摘要
This paper introduces and illustrates a new generalized ordinal logit (GOL) model which connects the four commonly-used multicategory logit models by using two hyper-parameters. The commonly used models in multicategory models are the adjacent-categories logit model (AC), the proportional odds (PO) model, and two variants of the continuation-ratio logit (CR) models. The GOL model generalizes these four models in the sense that each is a special case of the larger GOL model, and this GOL model is used for multicategory response data. In this article, we discuss (maximum likelihood) estimation and testing related to the GOL model, providing SAS/IML computer programs for the same, and illustrating the use of the proposed model with two real datasets.
引用
收藏
页码:87 / 105
页数:19
相关论文
共 50 条