chapter  6
22 Pages


The methods discussed thus far depend on knowing the measurement error variance, or estimating it, for example, with replicate measurements or validation data. However, it is not always possible to obtain replicates, and thus direct estimation of the measurement error variance is sometimes impossible. In the absence of information about the measurement error variance, estimation of the regression model parameters is still possible, provided the data contain an instrumental variable (IV), T, in addition to the unbiased measurement,W = X+U.