Stereotype Threat (Classic)
Article Critique of:
Steele, Claude M. & Aronson, Joshua (1995). Stereotype treat and the intelligence test
performance of African Americans. Journal of Personality and Social Psychology, 5,
797-811.
By Alla Andelman
This study comprised of four experiments to discover if stereotype threat had a negative impact on the performance of African Americans on a verbal intelligence test. The authors theorized that a stereotype threat would arise from the anticipation of having to take a difficult test, and would furthermore hinder performance on that test. Test 1 measured 114 Black and White (male and female) participants, randomly assigning them to the diagnostic condition (where participants thought ability was being tested), the non-diagnostic condition (where testing for ability was denied), and the non-diagnostic condition-plus-challenge (where participants were asked to take their challenge seriously. Results showed that Blacks in the diagnostic condition did slightly worse, and Whites in the non-diagnostic condition-plus challenge did slightly better then all the others, which were very similar, though none of these differences were statistically significant. The authors concluded that while they felt their results supported their hypothesis, it needed to be tested for reliability. Test 2 was meant to do that. It tested 20 Black and White female participants in only the diagnostic and non-diagnostic conditions. Results showed Black participants in the diagnostic condition performed worse than all other groups. The authors concluded that this established the reliability of their hypothesis posed in Test 1. In Test 3, 35 Black (9 male), and 33 White (20 male) were assigned to either a diagnostic, non-diagnostic, or control condition (where there was no mention of participants being tested) were given measures to test for stereotype activation, self-doubt activation, stereotype avoidance, indication of race, and self-handicapping were tested. Results showed significantly higher scores of stereotype activation for all conditions for Blacks, and higher self-doubt activation for Blacks in the diagnostic condition. The authors concluded that stereotype threat is ignited in African Americans by the anticipation of taking a test where ability will be tested, and would therefore not want to report their race for various reasons. Test 4 was designed to test if stereotype threat would arise and cause performance difficulties even when participants were not being explicitly tested for ability. Twenty-two Black (6 male) and 22 White (11 male) participants were tested on a verbal intelligence test after either being asked to report their race, or not. Black participants who first reported their race did statistically worse on the test then all other groups (though not significantly less then Whites who did not report race). The conclusion for this test and the final conclusion was that the performance of African Americans would be negatively affected in a situation where stereotype threat arises.
This study is a strong one for a number of reasons. To begin with, whenever a study serves to explore stereotypes and their affect on populations, studying a truly relevant topic, then those researchers must be applauded. Furthermore, the design of the study, using separate studies to build their conclusions, as well as the measures the authors chose to utilize are very innovative. Finally, this study created a strong background of previous research to support their efforts, and well discussed limitations of those works.
While this study is quite well prepared, there are some limitations to it. Primarily, we must look at the original theory that the authors propose. They claim that stereotype threat exists to undermine performance when it arises. However, we must consider the possibility that not all people of a race will react the same way to a perception of being judged by their race, or by essentially being challenged to represent their race. After all, is not that the idea we fight against, that stereotypes are merely descriptions of what people perceive about a race, but that do not actually describe groups of individuals in that race? We can suppose that stereotype threat can be compared to being challenged, and previous research has shown that while some people crumble under pressure, others thrive. The demographics of people who succeed in academics verses those who do not, coming from the same socioeconomic and cultural background, attests to that. Furthermore, the authors base their entire study on the assumption that all Blacks will first feel the stereotype threat, based only on the fact that they are Black, and then, that that will affect their performance. We must wonder, if the participants are not informed of the fact that they are being tested against different races, why would they automatically assume that they are being judged based on their race? In this case, perhaps a measure testing if participants feel they are being measured for their ability based on their race would be helpful. As the studies were designed, we could not really conclude that it was stereotype threat that caused the results in Tests 1 and 2. The instructions for the diagnostic and non-diagnostic groups were very subtle, and self-reports in Test 1 showed that the majority of people believed, from all condition groups, that the test given was a measure of ability. Since this variable was meant to prompt stereotype threat, this compromises the conclusions drawn from the results. Test 3, since it did not have a test to measure actual performance, did establish that some stereotype threat arose, but did not really show that this would then affect performance. Test 4, while seemingly the most stable assessment of results, also had a difficulty in basing its activation of stereotype threat on one measure from Test 3 (reporting race), which was not necessarily enough to cause this threat, even if the threat did have the affect on performance that was hypothesized.
This brings us to another important issue in the study. The authors made conclusions on results that did not show statistical significance in several cases. Particularly important, the first test did not show statistical significance until the authors removed one of the controls, and even then, this showed only marginal significance. The authors used Test 2, which consisted of a much smaller sample, and difference condition groups to prove reliability of a result that they did not actually achieve in the first test. Using the conclusions (the performance on verbal intelligence tests is negatively affected by stereotype threat) in Tests 1 and 2, the authors concluded in Test 3 that stereotype threat does arise when ability is tested for Black verses White participants. This conclusion, of course, is based on the assumption that the diagnostic condition was really different than the non-diagnostic condition, which was questioned in test 1, meaning there could be a different reason altogether for the results. Finally, Test 4 was based on all of these previous conclusions, which are tenuous, at best. Furthermore, the measures used to test for stereotype threat, such as reporting of race, could merely be a test of defensiveness to racial discrimination (an aspect of stereotype threat), without the implication that this would cause people to feel judged, compared based on race, required to defend their race, or all the other attributes associated with stereotype threat.
Finally, we have to consider a few demographic differences in the different groups tested. First, we could see that there were consistently more females in all the groups tested. The authors reported that this was not a factor that affected the results, but it would be helpful to see how they came to that conclusion. A possibly more important distinction, however, was that the White groups had consistently higher SAT scores on which performance was based. The authors claimed to adjust for SAT scores when making assessments of performance in each condition group on the verbal intelligence tests, but what type of adjustments they made, we do not know. We also do not know if the fact that the White participants had higher primary scores made them characteristically different from the Black participants in ways that were not considered, such as more achievement motivation qualities, or better test-taking skills, that would account for the results, even with the adjustments made for actual score differences. Also, SAT score level was not shown for each condition group, only for Black verses White participants, and that may have been a factor that would be important to the results, since it was used as a baseline score. Another issue would be that the samples for all the Tests except Test 1 were so small that it is difficult to trust the results their garnered to be applicable to a larger population. On that note, since the population consisted of all very achievement oriented people in a popular college with relatively high scores on the SATs to begin with, we may assume that they do not represent the general population of both Black and White people.
Clearly, this study had many good features to it. The design of the study was complex, using multiple tests and measures to account for their conclusions. However, some of the conclusions drawn were based on results that were not reliable, causing a chain affect throughout the different designs used in the study. However, this is a unique look at the possible causes of the phenomena of different performance levels of people from different races in academics, which is quite a relevant issue in our society.
Andelman, A. (2008). Stereotype Threat(2). http://psychologyalert.com/2008/11/stereotype-threat-classic.html