Abstract View
Framework for Selecting Data Analytic and Modeling Methods for Environmental Justice Analysis
RIVKAH GARDNER-FROLICK, Amanda Giang, University of British Columbia
Abstract Number: 497
Working Group: Environmental Justice: Technology, Frameworks, and Outcomes
Abstract
Environmental justice (EJ) and health analyses present a unique set of requirements for using data analytic and modeling approaches beyond selecting methods that are suitable for the available data and pollutant of interest. Techniques that may be appropriate for other applications might not be useful for the EJ questions being considered by researchers, communities, and other stakeholders, or worse, they could result in misrecognition of EJ communities. As a result, EJ communities may not receive the aid they need to improve their environmental conditions. Additionally, improper techniques could falsely label a place as harmful or degraded, or unfairly blame community residents for environmental hazards. With this in mind, we perform a scoping review of available literature to create a framework for selecting between common and emerging data analytic and modeling methods for use in EJ analysis, given the purpose of analysis, users, and resources available.
Upon critically synthesizing the literature, we identify three main dimensions on which to evaluate a data analytic or modeling technique when considering its use in EJ analyses. The key components that form the basis of this framework are the accuracy and interpretability of the results, the spatiotemporal features of the method, and the usability of the method. Techniques should be evaluated on accuracy and interpretability by their ability to describe real life concentrations, how reliably they identify EJ communities, and the ease of determining which factors are affecting concentrations and identification of communities. Spatiotemporal features demand appraisal of the available spatiotemporal resolution and domain of a method. Finally, the usability of a method is assessed through an examination of the accessibility, computational efficiency, and cost. We illustrate the framework with case studies from the literature. By locating different methods along these dimensions, we provide a framework for matching methods to analytical needs, resources, and users.