In their 1995 study, Steele and Aronson gave Black and white Stanford students a difficult verbal test. When the test was described as a measure of intellectual ability, Black students scored lower than white students with similar backgrounds. When it was described as a problem-solving exercise unrelated to ability, the gap largely disappeared. Later studies found similar effects for women in math and for other groups.
The theory suggests that some measured achievement gaps reflect testing conditions, not ability. Steele's 2010 book Whistling Vivaldi brought the idea to a wide audience. Schools have tested interventions like having students write about their values, which in some studies narrowed gaps.
The research has faced scrutiny during psychology's replication crisis. Some large replications found smaller effects than early studies, and the size of the effect in real-world testing is debated. Researchers broadly agree the phenomenon exists in some settings. It remains useful for understanding how stereotypes affect people even when no one acts on them directly.