DOI: 10.1080/01621459.1990.10474978
关键词: Linear regression 、 Margin of error 、 Multivariate normal distribution 、 Multivariate statistics 、 Mathematics 、 Statistics 、 Errors-in-variables models 、 Regression analysis 、 Observational error 、 Additive model
摘要: Abstract Increasing attention is being given to measurement error models in which the dimension of proxy or surrogate values different from that missing true values. Even if they are same dimension, model may not be simple additive one observed = + error, where has mean 0. The use broader relating and requires external internal data containing some calibrate/validate model. This article considers a multivariate normal framework allowed any subset variables with broad class regression classical special case. Multiple random regressors (the structural case) treated within this framework. Correcting for made possible through double sampling obtained randomly chosen main ...