A Straightforward Change That Could Transform Fairness in the Gig Economy
A recent investigation released in the esteemed journal Nature in February 2025 has revealed an unexpectedly simple approach for addressing racial bias in customer assessments—one that could greatly influence income equality for countless gig workers.
By substituting conventional five-star rating systems with a binary “thumbs-up/thumbs-down” format, researchers found that racial inequities in performance ratings—and thus, compensation—were largely eradicated. The research, conducted by a group of social scientists from the University of Toronto, Yale, and Rice University, has renewed attention on how platform design can inadvertently perpetuate systemic discrimination—and how straightforward design modifications can eliminate it.
The Results: How Ratings Correlate with Income Disparity
The study examined data from a home services platform that connects customers with workers for tasks such as plumbing, electrical repairs, and home maintenance. Under the five-star rating system, a subtle yet persistent racial bias surfaced: non-white workers were somewhat less likely to earn the highest rating compared to their white counterparts, despite providing equally high-quality service.
Particularly, 83.4% of non-white workers obtained five-star ratings in comparison to 86.9% of white workers. Though minor, these differences had significant repercussions because of the platform’s payment algorithm, which directly associated average star ratings with workers’ compensation. Consequently, non-white workers, on average, earned only 91 cents for every dollar earned by white workers.
The Importance of the Rating System
As stated by Katherine DeCelles, a co-author of the study and a professor at the Rotman School of Management at the University of Toronto:
“Even though the objective difference, on average, between the ratings for white and non-white workers is quite small, it is crucial because of its effect on income, underscoring the significance of structure and organizational design for promoting racial equality in the workplace.”
Five-star systems permit variations in judgment that create opportunities for unconscious bias, regardless of whether the evaluator views themselves as prejudiced. A marginal difference between a 4-star and 5-star review grows significant when repeated across thousands of interactions—ultimately affecting job opportunities, platform visibility, customer confidence, and earnings.
The Impact of Dichotomization: Thumbs Up, Just Pay Fairly
After adopting a simpler two-point system—allowing clients to only indicate whether they would hire the service provider again—the racial disparity in ratings and compensation vanished:
– Both white and non-white workers received top ratings at nearly equivalent rates.
– New workers, irrespective of racial background, displayed no pay inequality.
– Previously undercompensated non-white workers experienced a rise in their incomes to align with those of their white colleagues.
– These changes ensued immediately following the transition to the new rating system.
Perhaps most striking is that this change did not necessitate any shift in user attitudes or awareness. The sense of fairness arose purely from altering the structure through which customers evaluated services.
Understanding the Bias: The Mechanisms Behind Subtle Discrimination
To grasp why simple binary rating systems effectively removed racial inequities, the researchers conducted further tests centered on “modern racism”—a more subtle type of bias that operates below overt aggression or bigotry.
Participants displaying modern racist inclinations tended to give slightly lower scores to non-white service providers on a five-point scale (for instance, 4 stars instead of 5), even when the service was exemplary. Researchers suggest this behavior arises from the flexibility provided by multi-point scales: they enable evaluators to convey implicit bias while retaining a self-image of fairness and objectivity.
Conversely, the binary system—where clients must simply respond “yes” or “no” to repeat service—concentrates attention on performance and restricts the opportunity for bias to disguise itself as nuance.
“People can more clearly assess whether someone’s work was acceptable or not, rather than ‘how acceptable was it?’ which tends to be relatively more subjective and ambiguous—that’s where we’d anticipate a greater issue with racial bias in evaluations,” Prof. DeCelles elucidated.
Consequences for Gig Platforms and Beyond
The simplicity and efficacy of this adjustment indicate profound implications for digital platforms that depend on user-generated reviews to determine prospects and compensation for workers. As gig work continues to expand worldwide—through ride-hailing services, freelance job platforms, food delivery, and home maintenance services—rating systems have become quiet gatekeepers of income and opportunity.
The researchers suggest the following steps for platforms aiming to eradicate systemic bias and foster pay equity:
– Implement simplified, binary satisfaction rating systems.
– Conduct routine evaluations for racial disparities in ratings and compensation.
– Decouple earnings from rating scores or establish alternative channels for comprehensive feedback that do not influence pay or visibility.
– Enforce default anonymity in evaluations to decrease identity-based bias.
The research team believes the principle of dichotomization can also enhance fairness in various sectors, including:
– Hiring and promotion assessments
– Academic grading
– Customer satisfaction surveys in healthcare or education
– Annual performance reviews in the workplace
Since this solution does not depend on training.