In Gutsell and Inzlicht’s study showing physical evidence that white people have difficulty empathizing with non-white people, the researchers studied only white people and made a generalization about “people”:
Our research suggests that people do not mentally simulate the actions of outgroups. That is, those neural networks underlying the simulation of actions and intentions—most likely part of the ‘‘mirror-neuron-system”—are less responsive to outgroup members than to ingroup members.
The Clark Doll Experiment showed that black children prefer white dolls to black dolls during the time of de jure racial segregation. If the researchers instead tested only white children as representative of “children” and found that white children preferred white dolls to black dolls, they might have concluded that all children during Jim Crow prefer dolls of their own race, which would have been completely wrong.
In studies on implicit race bias, white people unconsciously prefer white people to black people, even when they do not consider themselves racist. If the implicit race bias researchers tested only white participants, they might conclude that the preference is due to “people’s” ingroup bias. However, they would be completely wrong, since the same implicit race bias studies on blacks show that blacks prefer whites and blacks equally.