Air pollution: do modern statistical model selection techniques make the silent killer speak too loud?

Estimates of the public cost of air pollution are based partly on the estimated mortality effects of air pollution obtained from time series investigations. However, many published studies of this kind have been called into question as they fail to take into account the effects of model selection processes used in the analyses. We will develop methodology for incorporating information from the model selection process into the estimates obtained from air pollution mortality investigations. This research will enable authorities to produce more accurate and defensible estimates of the burden of air pollution and to implement better air pollution standards.