Star ratings cost 9 cents for non-white workers – a thumbs-up system fixes it overnight

A new study shows that the five-star to thumb/fall rating system actually eliminates racism in customer assessments, potentially changing the payments for millions of gig workers.

The researchers found that this simple change eliminated a large wage gap, where non-white workers earn only 91 cents per dollar, each of which was paid to white workers for the same service job, a finding that had profound implications for platforms like Uber, Taskrabbit and countless other digital markets.

How rating differences cause significant income inequality

The study, published in Nature on February 19, 2025, examines data from the Home Services Platform, which connects customers to contractors with repairs and maintenance efforts. When using a traditional five-star rating system, non-white workers always score slightly lower than their white peers despite performing the same quality of work.

While the differences on paper seem small – 83.4% of the time without white workers and 86.9% of white workers, these small differences have huge financial consequences. The platform uses scores to determine how much income each worker receives, resulting in approximately 91 cents per dollar a non-white worker earns for every white worker.

“Although on average, the objective differences between white and non-white workers’ ratings are small, this is important due to its impact on income, which highlights the importance of structure and organizational design for racial equality at work,” said Katherine DeCelles, a professor of organizational behavior at the University of Toronto Rotman, Rotman School in the four-member research team.

How dichotomy eliminates racial bias

When the platform switched to a two-point rating system, the researchers discovered something compelling, which simply asked the client if they would use the contractor again (up or thumbs up). The racial gap in ratings actually disappeared.

This simple change has profound implications:

• Racial gaps that received the highest ratings were eliminated
• There is no racial difference in income for new workers after switching
• Non-white workers who previously earned white workers see income rise to the same level
• Improve immediately after evaluation system changes

What makes this finding particularly important is that it does not require a change in customer attitudes or awareness – changing the assessment structure eliminates discrimination.

Modern racism and subtle discrimination

The research team, including Demetrius Humes, PhD student at Rotman, Tristan Botelho of Yale and Sora Jun of Rice University, conducted other experiments to understand why the two-point system works so efficiently in reducing bias.

Unlike open racism, where someone may deny workers from certain races, modern racism often shows what the evaluator may even admit to himself in a subtle way. The researchers found that the multi-point scale creates perfect conditions for this subtle discrimination.

Their experiments show that people with modern racist beliefs are more likely to slightly lower their assessment of racial minorities when using the five-point scale. For example, 4 instead of 5 stars are offered to well-performing non-white workers.

Why does this happen? The researchers found that the multi-point scale allows evaluators to incorporate their personal opinions and biases without challenging their own self-perceptions. Even if workers are cleverly punished, the four-star rating can still be rationalized as positive.

The focus of performance on two-point scales

When there are only two options, the evaluator must focus only on whether the work performed is good or bad. This structural change fundamentally changes how people evaluate.

“People can assess more clearly whether someone’s job is good than ‘how good is it?’ It’s relatively more subjective and ambiguous – and that’s where we expect a bigger problem of racial bias in the assessment.”

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

Impact on digital platforms and future

As the gig economy continues to grow, millions of workers rely on platform-mediated assessments, these findings suggest direct solutions to persistent inequality. Researchers recommend the platform:

1. Switch to a simpler rating system that focuses assessors on the basic question of whether the service is satisfactory

2. Regularly review whether its system changes in the assessment to indicate bias

3. Provide other ways to provide customers with detailed feedback without affecting workers’ compensation

The application of this study may go beyond the gig economy. The findings show that dichotomy can reduce similar biases in other assessment settings such as hiring decision-making, performance review, and academic evaluation.

What is particularly promising about this approach is its practicality. Unlike many anti-discrimination interventions that require extensive training or building awareness, the solution is simple, direct, and does not require changing people’s attitudes – just the structure that expresses those attitudes.

As rating systems increasingly impact who gets opportunities and how much they earn in the digital economy, the study provides a powerful tool to help ensure that workers are equitable regardless of their race.