From Diversifying Economic Quality: A Wiki for Instructors and Departments
Stereotype threat is defined as “a situational predicament felt in situations where one can be judged by, treated in terms of, or self-fulfill negative stereotypes about one’s group”. Because of stereotype threat, an individual may not perform according to his or her innate ability; instead, this ability is impacted by generally held beliefs regarding this individual's grouping, whether it is by sex, age, gender, race, etc. The idea of stereotype threat was first developed by Claude Steel and it is discussed in detail in his book, "Whistling Vivaldi: and Other Clues to How Stereotypes Affect Us (Issues of Our Time)".
Examples of Stereotype Threat
When trying to understand certain performance gaps between groups, Steele and his colleagues did not focus on internal psychological factors. Instead, they tried to understand the possible causal role of identity contingencies, the things you have to deal with in a situation because you have a given social identity. Over the years they carried out a series of creative experiments* in which there was a control condition in which a task was given under normal conditions life. In the experimental condition, the identity contingency was either cleverly removed or it was deliberately induced. Here are three examples of experiments to clarify how they worked.
Experiment 1: Steele and Aronson (1995)
In this experiment the researchers had African American and white college students take a very challenging standardized test. In the control condition, the test was presented as these tests are always presented - as a measure of intellectual ability. This condition contained the stereotype that African Americans would be less intelligent. In the experimental condition the test was presented in a non-evaluative way. The test takers were told that the researchers were not interested in measuring their ability with the test but that they just wanted to use the test to examine the psychology of verbal problem solving. In the control condition, the African American test takers, on average, scored much lower than the white test takers. For the white test takers there was no difference in their scores between the control condition and the experimental condition. For the African American test takers there was a big difference between the control condition and the experimental condition. They solved about twice as many problems on the test in the experimental condition. Moreover, there was no difference between the performance of the black test takers and the white test takers.
Experiment 2: Aronson, Lustina, Good, Keough, Steele & Brown (1999)
In this experiment, the researchers asked highly competent white males to take a difficult math test. In the control condition the test was taken normally. In the experimental condition, the researchers told the test takers that one of their reasons for doing the research was to understand why Asians seemed to perform better on these tests. Thus, they artificially created a stereotype threat. In the experimental condition, the test takers solved significantly fewer of the problems on the test and felt less confident about their performance.
Experiment 3: Shih, Pittinsky & Ambady (1999)
In this experiment, a difficult math test was given to Asian women under three conditions. In condition one, they were subtly reminded of their Asian identity, in condition 2 they were subtly reminded of their female identity. In the control condition they were not reminded of their identity. The women reminded of their Asianness performed better than the control group, whereas those reminded of their female identity performed worse than the control group.
Experiments have shown that subtly removing or preventing stereotype threats can completely or largely eliminate performance gaps between stereotyped groups and non-stereotyped groups. This can be achieved by doing the following.
- Make it clear in the way you give critical feedback that you use high standards and let the person know that you expect him or her to be able to eventually succeed.
- Improve the number of people from the social category in the setting so that a critical mass is reached.
- Make it clear that you value diversity.
- Foster inter-group conversations and frame these as a learning experience.
- Allow the stereotyped individuals to use self-affirmations.
- Help the stereotyped individuals to develop a narrative about the setting that explains their frustrations while projecting positive engagement and success in the setting.
Athene Donald's Blog 
Reviews of "Whistling Vivaldi: And Other Clues to How Stereotypes Affect Us (Issues of Our Time)" 
"Linking Stereotype Threat and Anxiety" Osborne, Jason W., Educational Psychology, v27 n1 p135-154 Feb 2007.