Exposure measurement error in air pollution studies: the impact of shared, multiplicative measurement error on epidemiological health risk estimates