SOCIAL INNOVATIONS LAB · UNIVERSITY OF MICHIGAN · CSCW 2021
What job platforms don't see.
Job platforms match résumés to openings. But for marginalized job seekers, the real obstacles — housing, transportation, criminal records, stigma — sit outside the matching engine. We interviewed 21 employment-center staff to learn what decades of human practice can teach the design of employment support tools. Published at CSCW 2021 with Tawanna Dillahunt.
The challenge
Technological change is expected to limit decent work for millions, and the brunt falls on socially and economically marginalized job seekers — people already pushed into lower-wage, precarious work. HCI and CSCW had studied these job seekers and built tools for them. But one perspective was missing from the literature entirely: the employment centers that successfully serve up to 15 million job seekers a year.
Before designing tools for marginalized job seekers, we wanted to learn from the people who already help them find work.
We partnered with a non-profit in the American Job Center network — three local centers serving more than 2,500 employers annually, where roughly 70% of job seekers served were experiencing marginalization — and interviewed the staff who do this work every day.
The theory
We framed the study with the Psychology of Working Theory — the only career theory that takes seriously the situation of people with little or no choice in their work. PWT models how marginalization and economic constraints predict whether someone secures decent work, and names the moderators that can buffer that effect.
Most CSCW employment tools target personal factors within the job seeker's control. We asked instead how employment centers work the contextual moderators — and whether those practices could be designed into tools.
What we found
Career advisors and business services coordinators perform significant work far beyond matching — the function most employment tools optimize. The two most common barriers were housing insecurity and transportation, surfaced through intake checklists and informational interviewing. From there, the center works both sides of the market at once:
- Connect to external resourcesHousing programs, paying for rides until the first paycheck, car repairs — even alliances with courts for record expungement.
- Provide encouragement that networks can'tFilling gaps in social support to keep people applying through rejection.
- Turn crisis into strategyFrom "the house is on fire" to long-term goals: O*NET exploration, informational interviews, stepping-stone jobs.
- Reframe stigma"The offender, when given the opportunity, can be the best employee."
- Act as outsourced HRSmall employers lack HR departments — BSCs co-write job calls that attract the right candidates.
- Argue for fair wages with dataQuarterly wage and labor-market reports; a 50¢/hour difference moves workers between companies.
- Engineer the conditionsShuttle and carpool programs, flexible schedules, on-the-job training — fixing turnover at its causes.
- Train employers toward inclusionOpening positions to returning citizens and workers with disabilities.
CAs connected job seekers to external resources specific to their needs — housing, transport, expungement. BSCs worked with employers on more inclusive job calls, wages, and working conditions.
ML that detects discriminatory language in job ads before posting · worker-centered rating systems recognizing fair employers · interfaces connecting seekers to local support organizations · alternative gig platforms with free carpool matching.
The contribution
Two findings mattered most to theory and design. First, we found no evidence of critical consciousness work — the PWT moderator about recognizing and challenging structural barriers. Centers are too busy with urgent needs; the paper argues this is precisely where design can create space ("are certain 'soft skills' truly an issue, or is the requirement rooted in social injustice?").
Second, we proposed a moderator the theory didn't have: mitigating feelings of stigma. Career advisors spent real interpersonal effort reframing setbacks — ex-offender status, employment gaps — so seekers could keep going and recover from rejection. Employment tools don't do any of this today, and there's no technical reason they couldn't.
The outcome
The framework gives employment-tool designers a map: what centers already do for economic conditions, social support, proactivity, and stigma — and where tools could complement rather than replace that human infrastructure. Published amid COVID-19's employment shock, when the economic brunt fell on exactly these job seekers. Supported by NSF award IIS-1717186.
Most of what we "found," the career advisors already knew. The research's job was getting that knowledge into the literature where tool designers will actually see it.