Interesting, kind of self-fulfilling prophecy:

Our understanding of group differences in intellectual ability
is furthered by the very large literature on psychological
reactions to negative stereotypes. Steele and Aronson
(1995) argued that when test takers are aware of widespread
stereotypes that impugn a group’s intelligence (e.g.,
“Black people are stupid,” “Girls can’t do math”), they
frequently experience the threat of devaluation—by the
self, by others, or by both. The resulting arousal and
anxiety can impair executive functioning on complex tasks
such as standardized aptitude tests. Steele and Aronson
called this response stereotype threat and demonstrated in
a series of experiments that Black test takers scored considerably
better—sometimes far better—on intellectual
tests when the test was presented in a manner that downplayed
ability evaluation or downplayed the relevance of
race. Since the publication of Steele and Aronson’s 1995
article, some 200 replications of the effect have been published,
extending the findings to women and mathematics
abilities, Latinos and verbal abilities, elderly individuals
and short-term memory abilities, low-income students and
verbal abilities, and a number of nonacademic domains as
well. See Steele, Spencer, and Aronson (2002) and Aronson
and McGlone (2009) for reviews of the literature.
Two recent meta-analyses reported by Walton and
Spencer (2009) that included the data from nearly 19,000
students indicate that stereotype threat can cause tests to
underestimate the true abilities of students likely to experience
stereotype threat (Walton & Spencer, 2009). Walton
and Spencer’s analysis suggests a conservative estimate
that women’s math performance and Black students’ verbal
performance are suppressed by about 0.2 SD. In a number
of the individual studies, however, the suppression was
closer to a full standard deviation.
The stereotype threat formulation has led to a variety
of simple educational interventions conducted in schools
and colleges that have substantially raised the achievement
of Black students (e.g., Aronson, Fried, & Good, 2002;
G. L. Cohen, Garcia, Apfel, & Master, 2006) and the
achievement of girls in mathematics (Blackwell, Trzesniewski,
& Dweck, 2007; Good, Aronson, & Inzlicht, 2003).
The studies suggest that stereotype threat suppresses realworld
intellectual achievements. Some of the interventions
seem remarkably minor on the surface yet produce substantial
gains in academic achievement. For example, simple
efforts at persuading minority students that their intelligence
is under their control to a substantial extent have
nontrivial effects on academic performance (Aronson et al.,
2002; Blackwell et al., 2007).