Star Ratings Cost Non-White Workers 9¢ on the Dollar—A Thumbs-Up System Fixed It Overnight

A new study reveals that switching from five-star to thumbs-up/down rating systems virtually eliminates racial discrimination in customer evaluations, potentially transforming how millions of gig workers are paid.

Researchers found that this simple change erased a significant wage gap where non-white workers earned just 91 cents for every dollar paid to white workers for identical service jobs—a discovery with far-reaching implications for platforms like Uber, TaskRabbit, and countless other digital marketplaces.

How Small Rating Differences Create Major Income Inequalities

The research, published in Nature on February 19, 2025, examined data from a home services platform that connects customers with contractors for repairs and maintenance work. When using the traditional five-star rating system, non-white workers consistently received slightly lower ratings than their white counterparts, despite performing the same quality of work.

While the difference might seem small on paper—non-white workers received five stars 83.4% of the time versus 86.9% for white workers—these small discrepancies had dramatic financial consequences. The platform used ratings to determine how much of each job’s revenue workers received, creating a compounding effect that resulted in non-white workers earning approximately 91 cents for every dollar earned by white workers.

“While the objective difference, on average, between white and non-white worker ratings is very small, it matters because of the impact it has on income, highlighting the importance of structure and organizational design for racial equality at work,” said Katherine DeCelles, a professor of organizational behaviour at the University of Toronto’s Rotman School of Management who was among the four-member research team.

How Dichotomization Eliminates Racial Bias

The researchers discovered something remarkable when the platform switched to a two-point rating system that simply asked customers if they would use the contractor again (thumbs up or thumbs down). The racial gap in ratings virtually disappeared.

This simple change had profound effects:

• The racial gap in receiving top ratings was eliminated
• New workers joining after the switch showed no racial differences in earnings
• Non-white workers who previously earned less than white workers saw their income rise to equal levels
• The improvement occurred immediately after the rating system changed

What makes this finding particularly significant is that it required no change in customer attitudes or awareness—simply altering the evaluation structure eliminated the discrimination.

Modern Racism and Subtle Discrimination

The research team, which included Demetrius Humes, a PhD student at Rotman, Tristan Botelho of Yale University, and Sora Jun of Rice University, conducted additional experiments to understand why the two-point system worked so effectively at reducing bias.

Unlike overt racism where someone might refuse service from workers of certain races, modern racial discrimination often manifests in subtle ways that evaluators may not even recognize in themselves. The researchers found that multi-point scales create the perfect conditions for this subtle discrimination to emerge.

Their experiments revealed that people holding modern racist beliefs were significantly more likely to slightly downgrade their evaluations of racial minorities when using a five-point scale. For instance, giving a 4 instead of 5 stars to a non-white worker who performed well.

Why does this happen? The researchers discovered that multi-point scales allow evaluators to incorporate their personal opinions and biases without challenging their self-perception as non-prejudiced people. A 4-star rating can still be rationalized as positive, even while it subtly penalizes the worker.

Two-Point Scales Force Focus on Performance

When given just two options, evaluators must focus solely on whether the work performed was good or bad. This structural change fundamentally alters how people approach the evaluation.

“People can more clearly evaluate whether someone’s work was good versus not, instead of ‘how good was it?’ which is relatively more subjective and ambiguous – that’s where we’d expect a larger problem with racial bias in evaluations,” said Prof. DeCelles.

Participants in the experiments confirmed this, reporting that two-point scales made them less likely to incorporate their personal opinions and biases into ratings and more likely to focus only on performance quality.

Implications for Digital Platforms and Beyond

As the gig economy continues to grow, with millions of workers dependent on platform-mediated evaluations, these findings suggest a straightforward solution to a persistent problem of inequality. The researchers recommend that platforms:

1. Switch to simpler rating systems that focus evaluators on the basic question of whether service was satisfactory

2. Regularly audit their systems for systematic variations in evaluations that may indicate bias

3. Provide alternative ways for customers to give detailed feedback without affecting worker compensation

This research has potentially broader applications beyond the gig economy. The findings suggest that dichotomization could reduce similar bias in other evaluation contexts, such as hiring decisions, performance reviews, and academic assessments.

What’s particularly promising about this approach is its practicality. Unlike many anti-discrimination interventions that require extensive training or awareness-building, this solution is simple, immediate, and doesn’t require changing people’s attitudes—just the structure through which those attitudes are expressed.

As rating systems increasingly influence who gets opportunities and how much they earn in the digital economy, this research offers a powerful tool to help ensure that workers are evaluated fairly, regardless of their race.