Some Issues on Modeling Unobservable Variables, III

不可观测变量建模的一些问题，III

• 批准号: 9409540
• 负责人: Cheng Hsiao
• 金额: $ 12.97万
• 依托单位: University of Southern California
• 依托单位国家: 美国
• 项目类别: Standard Grant
• 财政年份: 1994
• 资助国家: 美国
• 起止时间: 1994-09-15 至 1997-06-30
• 项目状态: 已结题
• 来源: https://www.nsf.gov/awardsearch/showAward?AWD_ID=9409540&HistoricalAwards=false
• 关键词: Some; Issues; Modeling; Unobservable; Variables

Statistics and econometrics are mainly concerned with the quantitative measurement and analysis of real world phenomena. Typically, it is assumed that the data for the variables of interest are available and are measured accurately. In reality, errors of measurement are common because of reporting errors or because behavioral responses are in part determined by unobserved characteristics of agents; thus the variables of interest are latent or unobservable. This project is concerned with developing an analytical framework to extract information from data that is subject to either random or systematic measurement errors. The objective is to refine the methodology to establish empirically valid inferences from imperfectly measured data. In particular, we shall study issues of (i) random measurement errors in time series data that contain (stochastic) trends; (ii) systematic errors due to biased or exaggerated responses of sample respondents or grouping of individual observations into categories for the sake of confidentiality or convenience of collecting data; (iii) extracting information on the implicit market for characteristics (or attributes) from the "hedonic" approach developed by economists to simplify the analysis of differentiated products and quality changes over time; and (iv) modeling unobserved individual characteristics in panel data analysis which directs its focus on individual outcomes.

统计学和计量经济学主要涉及对真实的世界现象的定量测量和分析。 通常，假设感兴趣的变量的数据是可用的，并且被准确地测量。 在现实中，测量错误是常见的，因为报告错误或因为行为反应部分是由未观察到的代理特征决定的;因此，感兴趣的变量是潜在的或不可观察的。 该项目涉及制定一个分析框架，以便从存在随机或系统测量误差的数据中提取信息。 我们的目标是完善的方法，以建立有效的经验推断，从不完善的测量数据。 特别是，我们将研究（i）包含以下内容的时间序列数据中的随机测量误差问题（随机）趋势;（二）抽样答复者的有偏见或夸大的答复或为保密或方便收集数据而将个别意见分类所造成的系统性误差;（iii）提取有关特性隐含市场的资料（或属性）来自经济学家为简化对差异化产品和质量随时间变化的分析而开发的“享乐”方法;以及（iv）在面板数据分析中对未观察到的个体特征进行建模，从而将其重点放在个体结果上。