ABSTRACT
The effect of an environmental hazard can be estimated using an observational study, and the risk of exposure to the hazard is often expressed using the relative risk. Finding causes is fundamental to the development of a process model. Depending on what comes through the knowledge channel, a proposed process model can have two classes of covariates, some of which cause change in the process and some not. The aggregation of samples can lead to an issue about causality called Simpson's paradox. Often, health data are only available as aggregated daily counts, meaning that an ecological regression model is required. Concentration response functions are estimated primarily through epidemiological studies, by relating changes in ambient concentrations of pollution to a specified health outcome, such as mortality. The identification of appropriate micro-environments and the associated estimation of levels of the environmental hazard are an essential part of linking external measurements with actual individual exposures.
