TOBIT模型在医疗费用研究中的应用  

Tobit Model and Its Application to Health Expenditure

作　　者：王彤[1] 薛小平[1,2] 郭婷婷[3] 

机构地区：[1]山西医科大学卫生统计教研室,山西太原030001 [2]太原市疾病预防控制中心,山西太原030001 [3]山西医科大学劳动卫生教研室,山西太原030001

出　　处：《数理统计与管理》2009年第6期1047-1051,共5页Journal of Applied Statistics and Management

基　　金：山西省高校青年学术带头人基金[晋教科(2004)13];山西医科大学创新基金

摘　　要：对一般人群进行医疗费用调查研究时,会存在有相当一部分人的医疗费用支出为零,以此作为因变量使用经典的线性回归进行统计建模时将会导致估计值的偏差。本文尝试使用Tobit模型及其半参数估计方法解决因变量存在大量零删失数据的问题,通过对太原市城镇居民医疗费用相关影响因素调查数据的实证分析,用以为医疗费用控制和医疗保险制度改革提供更有效的统计学方法支持和政策依据。In survey study on the health expenditure,there are always some subjects investigated whose medical expenses come up as zeros.Modeling this sort of response variable with classic linear regression techniques will result in estimation bias.The authors of this article aimed to deal with this zero-censored problem using Tobit model and its semi-parametric estimation to analyze an empirical research data from a survey study on health expenditure and its influence factors of urban inhabitants in taiyuan city, meanwhile to provide statistical method support and policy evidence for research on health expenditure control and medicare insurance reform.

关 键 词：TOBIT模型 半参数估计 医疗费用 

分 类 号：O212[理学—概率论与数理统计]