Data Science for Social Good: Exploring Geographic Factors of Seattle’s Minimum Wage Law

The minimum wage team

By: Emily Keller-O’Donnell

When the Seattle minimum wage law took effect in 2015, it dovetailed with an increase in housing prices as an influx of high-wage technology workers moved to the region. For an interdisciplinary team in the University of Washington Data Science for Social Good (DSSG) program, this raises the question of whether some of the law’s targeted beneficiaries may have simultaneously relocated out of the area due to gentrification and residential displacement.

Their project, “Geography, equity, and the Seattle $15 minimum wage ordinance,” is one of projects hosted this year by the eScience Institute. The DSSG program brings together students, stakeholders, data scientists and domain researchers to work on project teams for a 10-week period.

The project team consists of four student fellows from universities around the country, two data scientists at the eScience Institute, and the project lead, Jennie Romich, a Professor of Social Welfare at the UW School of Social Work and Faculty Director of the West Coast Poverty Center. The DSSG project is part of Romich’s larger collaborative research on wage mandates, which began with the UW Minimum Wage Study.

Conducting Background Research

The team members began their work by reviewing literature and engaging with stakeholders to incorporate grounded perspectives and the regional context into their project. They met with a senior research manager at the Washington State Department of Social & Health Services (DSHS) who managed the collection of agency data used in the project, and a demographer at the City of Seattle’s Office of Planning and Community Development. They also plan to meet with faculty from the Minimum Wage Study and participants from the West Coast Poverty Center’s Roundtable on Housing and Poverty. In these meetings, the team is exploring topics such as the nuances of using administrative data for research, the development of residential displacement measures, and the intersecting issues of wages, gentrification, housing costs and commuting trends.

Christopher Salazar, an incoming Ph.D. student in Industrial and Systems Engineering at UW, said, “Coming from an engineering background, my initial reflex is to begin crunching numbers and getting results as soon as possible. The start of the DSSG program has flipped this upside down by emphasizing engagement with my team, stakeholders, and the qualitative aspects of our work, which I believe will enhance the outcomes of this project.”

Mahader Tamene, a Ph.D. student in the Division of Epidemiology at University of California, Berkeley’s School of Public Health, said, “Stakeholder meetings have been instrumental in building an understanding of the context of this work, particularly around the nuances of our data, the considerations of privacy and confidentiality, and the importance of honoring and uplifting community as expert. All of our conversations have informed how we move forward and really fostered deeper introspection and self-reflection in the decisions we make about our project.”

Exploring Comprehensive and Geocoded Data

From there, the team began diving into the data and establishing the scope of their work. The project utilizes the Washington Merged Longitudinal Administrative Data (WMLAD), a comprehensive state-level geocoded administrative dataset for examining employment and earnings outcomes, which Romich helped create over the past five years with the help of DSHS and six other state agencies. The database, which includes ten million observations in total, uniquely combines unemployment insurance records (containing wages and hours) with driver’s license and state ID data, birth records, voter registration, state criminal records and DSHS client records. The breadth of data on employment and home addresses contained in the database provides the opportunity to explore moving and commuting patterns in new ways.

Given the numerous possible directions available with such a large dataset, the team is focusing on four sub-topics, covering the years 2010-2017. Their research is examining the effects of the minimum wage ordinance on the demographic distribution of workers, residential displacement, the housing price burden (calculated as spending beyond 30% of income on housing per area), and the work commute burden (the change in distance traveled from work to home over the years studied).

The team members’ multi-disciplinary backgrounds have been a significant asset to their work. “The combination of disciplines and perspectives on my team is really intellectually energizing. Our discussions range from the technical challenges of working in the super-secure data environment to the ethical considerations of using big data from public agencies,” Romich said. The DSSG project specifically examines whether trends in the region’s housing market may have reduced the supply of people looking for low-paid jobs as the ordinance took effect.

Lamar Foster, a Ph.D. candidate in the Education Policy, Organizations, and Leadership program at the UW College of Education, said his analytical and critical lens has helped shape his approach to conducting analyses and thinking about the data as a whole. “I am acutely aware of the sensitivity of the data and that each data point represents a person with certain traits and identities,” Foster said, “I am mindful not to contribute to erasure by not accounting for certain people groups or neighborhoods.” One way that the team has thought about inclusion is in creating a variable to designate low-wage workers from other workers that goes beyond the often used poverty line, which misses many low-wage workers who do not qualify for benefits, and considers hourly wages and other factors as well.

Delaney Glass, a Ph.D. student in biological anthropology at UW, provided one takeaway so far. “One might think: more data is always better. However, even with big data, such as the WMLAD that covers over seven million residents, our research questions are still necessarily constrained by that data. Navigating what we can and cannot answer has been very challenging and I anticipate utilizing this experience in the future despite primarily working with smaller, observational data,” she said.

Securing the Data

Working with sensitive data responsibly presents unique challenges. The WMLAD data is housed by the UW Data Collaborative at the Center for Studies in Demography & Ecology (CSDE), governed by the Washington State Institutional Review Board (IRB), and accessible only through a highly-secure remote enclave with no Internet access. The set-up of the project, which represents a partnership between CSDE, the School of Social Work, the West Coast Poverty Center, and the eScience Institute, has required the team to develop a specialized programming workflow. They have also met with a staff member of the state IRB to discuss privacy considerations for conducting ethical research beyond their approval.

Valentina Staneva, a senior data scientist at the eScience Institute, noted the team’s dedication to preserving data privacy. “To provide security, WMLAD is stored on a server without internet access. Working with this dataset has reminded us how reliant we have become on having an internet connection in our data science workflows, and at the same time has asked us to be creative in finding workarounds in the process,” she said.

Jose M. Hernandez, a senior data scientist at the eScience Institute, said, “Working with WMLAD has greatly informed our practice around working with sensitive data responsibly while keeping the greater goals of the project in mind. As a team we understand that having IRB approval to work with such sensitive data is a privilege and that individually it is our responsibility to uphold privacy and confidentiality standards. If we do this right, this will open the door for other researchers to be able to work with this data.”

The team members are hoping their project will not only yield data about the impacts of the minimum wage ordinance, but also enable them to create documentation informing future researchers about considerations and strategies for working with this type of sensitive data. Romich hopes the project will serve as a use case that exhibits the value of conducting research with the administrative data in WMLAD.

The final DSSG presentations will take place on Wednesday, August 18th via Zoom from 1:00 to 2:30 p.m. Pacific. The event is open to the public.