Indoor air pollution and pulmonary performance: investigating errors in exposure assessment.

We use pulmonary function measurements on pre-adolescent children and indoor air pollution measurements in the homes of these children to illustrate estimation techniques for linear regression models containing independent variables measured with error. In our data set, replicate measures of indoor air pollutant concentrations provide one method of estimating measurement error variances. Surrogate information in the form of cigarettes smoked is also available for the pollutant of interest. Several estimation procedures are presented, and we combine two estimators, one based on surrogate information and one based on replication information, using generalized least squares.