Using Mass Balance, Factor Analysis, and Multiple Imputation to Assess Health Effects of Water Quality

This dissertation explores the use of three analytical methods to improve the utility of microbial water quality data, collected on the Chicago River from 2007 to 2009, in predicting health risk among water users. The Multiple Imputation(MI) method was applied to fill in microbial missing values and the ability of the method to reduce bias was evaluated, chemical mass balance model and exploratory factor analysis were then utilized to identify sources of fecal contamination in the river system. Sources Identified as contributing to fecal contamination were used in predicting health risk. The results showed that by introducing a 2% bias to the parameter estimates, the MI method was able to recover 24% of missing data. However, in order to fill in 36% of missing values, 33% of bias was introduced. Chemical mass balance model and exploratory factor analysis both identified the water reclamation plant, combine sewer overflows (CSOs), and the precipitation as sources of fecal contamination in the river system. However, no association between pollutant sources and health risk were observed.