文章基本信息

标题：A COMPARISON OF POLYTOMOUS MODEL WITH PROPORTIONAL ODDS AND NON-PROPORTIONAL ODDS MODEL ON BIRTH SIZE CASE IN INDONESIA
本地全文：下载
作者：Yenni Kurniawati ; Anang Kurnia ; Kusman Sadik 等
期刊名称：MEDIA STATISTIKA
印刷版ISSN：1979-3693
电子版ISSN：2477-0647
出版年度：2021
卷号：14
期号：1
页码：79-88
DOI：10.14710/medstat.14.1.79-88
语种：English
出版社：MEDIA STATISTIKA
摘要：The proportional odds model (POM) and the non-proportional odds model (NPOM) are very useful in ordinal modeling. However, the proportional odds assumption is often violated in practice. In this paper, the non-proportional odds model is chosen as an alternative model when the proportional odds assumption is not violated. This paper aims to compare Proportional Odds Model (POM) and Non-Proportional Odds Model (NPOM) in cases of birth size in Indonesia based on the 2017 Indonesian Demographic and Health Survey (IDHS) data. The results showed that in the POM there was a violation of the proportional odds assumption, so the alternative NPOM model was used. NPOM had better use than POM. The goodness of fit shows that the deviance test failed to reject H0, and the value of Mac Fadden R2 is higher than POM. The risk factors that have a significant influence on all categories of birth size are the residence and gender of the child.