Measurement error in environmental exposures: Statistical implications for spatial air pollution models and gene environment interaction tests

Measurement error is an important issue in studies of environmental epidemiology. We considered the effects of measurement error in environmental covariates in several important settings affecting current public health research. Throughout this dissertation, we investigate the impacts of measurement error and consider statistical methodology to fix that error.;In Chapter 1, we investigate the effects of measurement error in a linear health effects model with a gene-environment interaction term. We examine these effects under gene-environment dependence. We derive closed-form solutions for the bias in naive parameter estimates, and we find that the resulting bias may be toward or away from the null. We also identify specific cases when the bias will be attenuated and when tests will preserve the Type I error rate.;In Chapters 2 and 3, we consider the problem of measurement error in studies of air pollution health effects, considering the case when air pollution exposure is predicted by kriging or land use regression. Chapter 2 approaches this problem from a more theoretical standpoint, and develops the spatial SIMEX methodology to correct for spatially-correlated classical measurement error. Chapter 3 complements the theoretical work in Chapter 2 in a practical assessment of the effects of measurement error on actual air pollution surfaces. This question is addressed by a simulation study using high-resolution satellite data.