家庭收入跨地区比较的潜变量分析

针对居民家庭消费与社会经济状况调查中户主自报的家庭固定收入可信程度差以及不同地区的不可比问题,利用陕西省2008年第四次国家卫生服务调查数据,将自报的家庭收入作为不能准确测量的潜变量,运用DIHOPIT模型确定反映家庭收入高低的指示变量,通过不同地区共同的指示变量,用相同或近似的指示变量作为截断点评价不同地区家庭收入的差别。用陕西省城乡6个县(区)的调查数据拟合模型,发现在众多的指示变量中,"电话类型"为5个县(区)共同指示变量,同时也发现部分指示变量在部分调查县(区)缺乏效度,需要在今后的调查问卷条目设计时做适当调整和补充。

Analysis on Latent Variable of Family Permanent Income and the Indicator Variables

Focused on the low credibility of the self-reported family permanent income oy noubenl~t~Lcx in household consumption and socio-economic survey along with the incomparable issues, such us the difference between the urban and rural income, this paper made use of the data of the Forth National Health Services Survey in Shaanxi Province in 2008 and regarded the family permanent income as an unobserved latent variable, used the DIHOPIT model to determine the indicator variables exactly reflected the level of family income. By the common indicator variables in different regions, made use of the same or similar indicator variable as the cut-point to evaluate the wealth ranks of family income between regions. The model fitting of 6 counties （districts） indicates ＂the type of phone＂ is the common indicator among numerous variables. On the other hand, some of the indicator variables which are lack of validity in part of the investigated counties （districts） need to be appropriately adjusted or modified.