Despite advances in parity in recent decades, white men in America on average still get paid more than women and minorities for doing the same work, by some accounts about 20 to 25 percent more.

A new study finds one reason: In satisfaction surveys, store customers and medical patients say they prefer white men, and managers frequently hire and set pay based on customer preferences.

In the study, researchers at four business schools showed study subjects a video featuring either a black male, a white female, or a white male actor playing the role of an employee helping a customer. Those viewing the white male were 19 percent more satisfied with the employee's performance, and they were also more satisfied with the store's cleanliness and appearance.

But the actors demonstrated the same scripted behaviors, and the store background, camera angles and lighting were identical.

"Customers, from students buying textbooks to patients in an examining room, are consistently biased in favor of white men," said David Hekman, assistant business professor at the University of Wisconsin–Milwaukee. "Because customer satisfaction is critical for organizational survival, business owners and managers will hire white men when possible and will pay lower salaries to the women and minorities they do hire."

In a separate aspect of the study, Hekman and his colleagues examined more than 10,000 medical patients' ratings of their doctors. Patients who received e-mail from their doctor were more satisfied with their doctor's competence and approachability, but only if the doctor was a white man.

Changes are needed in how customer satisfaction surveys are conducted, Hekman argues.

Surveys should target specific employee behaviors. For example, don't ask customers if they would recommend a physician, he suggests. Ask them how many times the doctor asked a patient if she had additional questions or understood key medical terms.

Anonymous customer feedback surveys should not be done, Hekman said. Customers should be identifiable and therefore at least somewhat accountable for their ratings. "People may do all sorts of bad things when they are anonymous," he said. "Just check out the reader postings on any blog."

The study, announced today, will be published in the Academy of Management Journal.