You’re only looking at their preliminary evidence. They went on to run experiments on others. Others outside of your objections.
With this preliminary evidence in hand, the researchers set out to further test their ideas.
“My hope is that this work will help include well-intentioned people who see themselves as allies but who may be unwittingly contributing to group divides.”
They designed a series of experiments in which white participants were asked to respond to a hypothetical or presumed-real interaction partner. For half of these participants, their partner was given a stereotypically white name (such as “Emily”); for the other half, their partner was given a stereotypically black name (such as “Lakisha”). Participants were asked to select from a list of words for an email to their partner. For some studies, this email was for a work-related task; for others, this email was simply to introduce themselves. Each word had been previously scored on how warm or competent it appears. The word “sad,” for example, scored low for both warmth and competence. “Melancholy,” on the other hand, scored high for competence and low on warmth.
Participant also completed a variety of measures that assessed how liberal they were.
The researchers found that liberal individuals were less likely to use words that would make them appear highly competent when the person they were addressing was presumed to be black rather than white. No significant differences were seen in the word selection of conservatives based on the presumed race of their partner. “It was kind of an unpleasant surprise to see this subtle but persistent effect,” Dupree says. “Even if it’s ultimately well-intentioned, it could be seen as patronizing.”