Determining multi-level drivers of perceiving undesirable taste and odor in tap water : a joint modeling approach

Spearing, Lauryn Altena
Journal Title
Journal ISSN
Volume Title

Aesthetic considerations of tap water do not necessarily indicate public health threats; however, when noticed by consumers they can affect people’s perceptions about the service received and the water provider. Water aesthetic considerations are caused by nuisance chemicals. Although these chemicals can be measured, the detection of aesthetic issues is relative among people and may vary based on a person’s sociodemographics. This study aims to identify sociodemographic and geographic parameters that influence users’ recognition of select tap water aesthetic issues. A bivariate binary probit model is used to jointly assess whether an individual perceives undesirable taste or odor in tap water. Enabling this study is data from a household survey conducted in August of 2016 in the Austin, Texas area and the 2016 Census. Results indicate that the drivers of noticing undesirable aesthetics, specifically odor and taste, occur at three levels—the individual, household, and regional levels. These distinct levels provide water providers opportunities for different interventions— i.e. leverage points— to address such aesthetic issues. By identifying factors influencing perceptions of these undesirable aesthetic considerations, water providers may develop outreach campaigns and make project decisions that address the groups revealed in this analysis (e.g., users residing in zip-codes with high poverty rates, users over the age of 50). For instance, one driver revealed was that individuals who pay the water bill are associated with higher propensities to notice an undesirable odor. Water providers may use such findings to tailor and disseminate information about water aesthetics through water bills, such as potential aesthetic issues associated with lake turnover at specific times of the year. Additionally, this study demonstrates how joint modeling can be used to more accurately capture relationships in the water sector, providing a method for other researchers to follow when performing statistical analyses in this domain