Lack of stereotype threat at a liberal arts college
Mark G. Rivardo
Stereotype threat has been demonstrated to reduce the performance of stereotyped individuals in the threatened domain (Steele & Aronson, 1995). This study attempted to replicate the finding that stereotype threat instruction can erase the performance deficit women experience in math performance (Johns, Schmader, & Martens, 2005) and to further evaluate the arousal hypothesis of stereotype threat (e.g. Ben-Zeev, Fein & Inzlicht, 2005). The study provided no evidence of stereotype threat F(2, 91) = 1.60, p = .208, partial [h.sup.2] = .034. Stereotype threat may be less likely to affect performance at a small, liberal arts institution where the learning environment is both nurturing and personal.
Stereotype threat occurs when members of a negatively stereotyped group are put in a situation where their performance on a given task could confirm the stereotype (Steele & Aronson, 1995). The pressure caused by this knowledge can hinder performance on the task and make confirmation of the stereotype more likely. For example, African-American participants performed worse when a task was described as being diagnostic of intellectual ability, than when it was simply described as an instrument for studying problem solving (Steele & Aronson, 1995). Researchers have also examined stereotype threat in other stereotyped groups and tasks, such as women and math (e.g. BenZeev, Fein, & Inzlicht, 2005; O’Brien & Crandall, 2003; Spencer, Steele & Quinn, 1999), and Black and White athletic performance (Stone, Lynch, Sjomeling, & Darley, 1999).
Research clearly indicates that the effects of stereotype threat can vary based upon individual differences in coping. Some participants engage in self-handicapping such as withdrawing effort (Steele & Aronson, 1995) or engaging in less practice under stereotype threat conditions (Stone, 2002) to provide an alternative explanation for poorer performance. Others may react to the situation with denial, and thus prevent the threat from affecting their performance (von Hippel, et al., 2005). Coping sense of humor can also guard against the negative effects of stereotype threat (Ford, Ferguson, Brooks, & Hagadone, 2004).
Studies conducted on the effects of stereotypes threat on women’s math performance have identified additional parameters for the phenomenon to occur: GRE and SAT items are often used to measure performance because the male advantage in math performance is typically only found in problem solving tasks and does not appear until after the high school years (Hyde, Fennema, & Lamon, 1990) and stereotype threat decreases performance only if the task is sufficiently difficult to challenge the individual (O’Brien & Crandall, 2003; Spencer, Steele, & Quinn, 1999). For women completing math tests, the presence of a female role model who demonstrates strong mathematical ability can improve women’s performance on a difficult mathematical test (Marx & Roman, 2002). Similarly, when female participants read about successful female role models, they did better on quantitative GRE sample items (McIntyre, Paulson, & Lord, 2003). In addition, when the experimenter told the experimental group of female participants that women made better participants in psychology experiments, they did better than females in the control group. Inzlicht and Ben-Zeev (2000) found women performed better when they were tested in same-sex groups of three than when they were tested with two men (experiment 1) and that when women were tested in same-sex groups of three, their performance did not differ from that of males (experiment 2). Aronson, Lustina, and Good (1999) suggest the stereotype threat effect is greatest for people who value the domain being tested. Subsequently, some researchers have limited their samples to students who met an SAT Math cutoff (e.g. Inzlicht, Aronson, Good, & McKay, 2006; Spencer et al., 1999), indicated they valued math ability on a questionnaire (e.g. Ben-Zeev, Fein, & Inzlicht, 2005; Inzlicht & BenZeev, 2003), or were likely to value math based upon their course enrollment (e.g. statistics students, Johns et al., 2005). Although research has demonstrated stereotype threat’s effect on performance to be robust, there certainly are limits to its applicability.
Johns, Schmader, and Martens (2005) found the advantage male participants had over female participants on GRE word problems disappeared when the women were told about stereotype threat and that the anxiety they might experience during the test could be due to stereotype threat, rather than to any real difficulty with the items. Johns et al. claimed the performance of members in the threatened-instructed group was increased because they externalized the arousal. Their hypothesis is based upon the arousal theory of stereotype threat supported by others (e.g. O’Brien & Crandall, 2003; Blascovich, Spencer, Quinn, & Steele, 2001; Ben-Zeev, Fein, and Inzlicht, 2005), namely that stereotype threat activation leads to arousal that is attributed to difficulty with the test items. The decrease in performance caused by stereotype threat may be due to a reduction in working memory capacity as was found with women and Latinos under stereotype threat conditions (Schmader & Johns, 2003). This reduction in working memory capacity mediates the reduction in performance on standardized quantitative tests. Taken together, these ideas form a reasonable theory of stereotype threat. Stereotype threat activation leads to increased arousal that is attributed to difficulty with the test. The arousal leads to a reduction in working memory capacity, which makes the already difficult test even more difficult for the stereotyped individual, and therefore performance is hindered.
We are unaware of any stereotype threat research that has incorporated a physiological measure of arousal. Individual circumstances producing stress activate the hypothalamic-pituitary-adrenal (HPA) axis (Rhodes & Rubin, 1999). The HPA axis is a dynamic three-gland component of the endocrine system, acting as the central modulator of the body’s response to stress. HPA axis activity is reflected peripherally by plasma concentrations of the pituitary hormone, adrenocorticoptric hormone (ACTH), and the adrenal cortex hormone, cortisol (CORT) (Rhodes et al., 2001). The accumulating effects of stress and associated increases in HPA activity are believed to contribute to increased hippocampal neuronal loss and memory impairment in the brain (Bowman et al., 2003; Luine, 2002; Beck & Luine, 2002; Luine, 1994; Luine et al., 1994; Issa et al., 1990).
Exposure to both chronic and acute stress has substantial effects on learning and memory (for reviews: Cahill & McGaugh, 1998; Kim & Diamond, 2002). In rats, chronic stress generally produces memory deficits in males, while enhancing memory processes in females (Luine et al., 1994; Conrad et al., 1996; Conrad et al., 2004); the effects of acute stress on learning and memory are less consistent, and sex differences vary depending on the task (Conrad et al., 2004). Schmader and Johns’ (2003) findings suggest stereotype threat inhibits working memory capacity, but the lack of an objective, physiological measure of arousal prohibits direct support of the arousal model of stereotype threat. Studies addressing the connection among stress hormones, brain function, and behavior, therefore, are pertinent to the understanding of the arousal model and stereotype threat behaviors.
The present study had two purposes. First, we attempted to replicate the findings of Johns et al. (2005) that teaching participants about stereotype threat can alleviate sex differences in GRE math performance. Second, we determined whether arousal, as measured by salivary CORT, varies between stereotype threatened-instructed, stereotype threatened, and control groups. We hypothesized that if the arousal model of stereotype threat was valid, then CORT in the control group would be lower than CORT in the other two groups. A finding that arousal in the threatened-instructed group was lower than in the threatened group would suggest that instruction limits arousal, which would require minor revisions to the arousal model.
One hundred twelve students at Saint Vincent College, a Catholic, liberal arts college in southwest PA, participated in the experiment. Participants were recruited from psychology, sociology, Spanish, French, elementary education, and biology classes and most received extra credit for their participation. Eighty-five women and 26 men initially participated. One participant was a high school student taking a college class, 22 participants were freshman, 36 were sophomores, 32 were juniors, 17 were seniors, and 3 were continuing education students. Sixty-three participants had majors in the social sciences, 21 in the natural sciences, 16 in the humanities and fine arts, and two students were undeclared. One male participant was unable to produce a sufficient saliva sample within the time allotted, so his data were not included in the analyses.
Materials & Procedures
Participants reported to a small auditorium (280 seats) at 6:30 p.m. and were tested in groups. The participants in each testing session were randomly assigned to one of three conditions similar to those of Johns et al. (2005): problem solving, stereotype threat, or stereotype threat with instruction. After granting their informed consent, participants received a testing packet and materials for saliva collection.
Salivary cortisol is a reliable indicator of the free cortisol in plasma, which is considered to be the biologically active hormone (Vining et al., 1983). Salivary cortisol increases within minutes in response to acute stressors and has a half-life of approximately 1 hour (HeUhammer et al., 1987). Eight to 10 minutes prior to all sample collections, participants thoroughly rinsed their mouths with water. Saliva was collected via the passive drool method, whereby saliva was allowed to flow down a 2-inch section of an ordinary household plastic straw and collected into a small microcollection tube, and stored on ice. Following each experiment, saliva samples were centrifuged, quickly frozen at -80 [degrees]C, and stored until hormone analysis.
After completing demographic items, each participant read one of three sets of instructions. In the problem solving condition, instructions indicated the upcoming test was a test of general problem solving ability. Instructions informed those in the stereotype threat group that they were about to take a math test that typically shows gender differences. Participants in the stereotype threat with instruction group were told that they were about to take a math test that typically shows gender differences, but they were also told about stereotype threat. Instructions also indicated that women in the latter group should follow these additional instructions, which were identical to those given by Johns, Schmader, and Martens (2005):
It’s important to keep in mind that
if you are feeling anxious while taking
this test, this anxiety could be
the result of these negative stereotypes
that are widely known in
society and have nothing to do with
your actual ability to do well on the
test (p. 176).
Participants had 20 minutes to complete 30 quantitative items taken from different sections of a GRE practice test (ETS, 2001). At the end of the 20-minute testing period, participants were asked to record their score on the quantitative portion of the Scholastic Aptitude Test (SAT) and to indicate their level of agreement on five Likert scale items regarding belief in a sex difference in mathematical ability, perceived belief of the experimenter about a sex difference in mathematical ability, and level of anxiousness experienced during the GRE test. Then a second saliva sample was collected via the passive drool method and participants completed the Ways of Coping Questionnaire (Folkman & Lazarus, 1988).
Saliva samples were analyzed in duplicate for cortisol by a highly specific enzyme immunoassay (EIA) kit for salivary cortisol (Salimetrics, State College, PA). Inter- and intra-assay coefficients of variation were less than 5%, and the minimum detectable cortisol concentration was 0.25 ng/ml.
Results & Discussion
Math performance was measured by dividing the total number of problems correct by the number of problems attempted, a measure referred to as performance accuracy (Steele & Aronson, 1995). This measure is consistently used in the stereotype threat literature (e.g. Inzlicht & Ben-Zeev, 2004; Johns et. al. 2005).
The primary analysis in this study concerns how performance accuracy varies by condition and sex of the participant. A 2(sex) X 3(condition) analysis of covariance (ANCOVA) was conducted with self-reported SAT as the covariate and performance accuracy as the dependent measure. Self-reported SAT was a significant covariate, F(1, 91) = 19.57,p <.001, partial [h.sup.2] = .177. Performance accuracy did not vary by condition, F(2, 91) = 1.60, p = .208, partial [h.sup.2] = .034. Participants in the problem solving condition had a mean performance accuracy of 36.27% (SD = 17.19), those in the stereotype threat condition had a mean of 39.15 % (SD = 17.90), and those in the stereotype threat with instruction condition had a mean of 43.13% (SD = 14.41). The interaction of sex and condition also failed to reach significance, F(2, 91) = 0.54, p = .583, partial [h.sup.2] = .012, (see Table 1 for descriptive statistics).
That we failed to replicate the simple stereotype threat effect, much less Johns et al.’s (2005) finding that instructions could help women overcome the effects of stereotype threat, was surprising. Subsequent analyses were conducted to determine potential explanations for this failure. Self-reported SAT scores were analyzed with a 2(sex) X 2(condition: problem solving, stereotype threat, stereotype threat with instruction) analysis of variance (ANOVA) for independent groups. Two participants indicated they did not take the SAT and 11 others did not report their score. SAT did not vary by condition, F(2, 92) = 1.28, p = .289, partial [h.sup.2] = .027, and the sex by condition interaction failed to reach significance, F(2, 92) = 1.80, p = 171, partial [h.sup.2] = .038 so random assignment appeared to control for differences in mathematical ability between conditions (see descriptive statistics in Table 2). However women (M = 555.27 SD = 74.90) reported higher SAT scores than men (M = 514.92 SD = 89.92), F(1, 92) = 6.28, p = .014, partial [h.sup.2] = .064. The sex difference in this sample is the opposite of what was reported by the College Board (Sathy, Barbuti & Mattern, 2006). Aronson, Lustina, and Good (1999) suggested the effect is greatest for students who are interested and skilled in the domain being tested, but even when analyses were limited to participants reporting 500 or better on the SAT Math no performance accuracy results were significant (all p’s > .50). Participants’ responses to questions of whether they thought men or women have lower math abilities, what they thought the experimenter believed about sex differences in math, and how anxious they felt while taking the test were analyzed with a 2(sex) X 3(condition) multivariate analysis of variance (MANOVA). Only the sex difference for the item “I believe males have lower math ability than females” reached even marginal significance, F(1,105) = 3.01, p = .086, partial [h.sup.2] = .028. With negative values indicating disagreement, males (M = -0.54 SD = 0.81) indicated less disagreement than females (M = -0.81 SD = 0.68). All means for the sex difference questions were in the negative direction, indicating participants did not think there was a sex difference in math ability, nor did they think the experimenters believed in a sex difference. The anxiety question received neutral responses (M = 0.22, SD = 1.07).
CORT results mirrored those of the Likert item assessing anxiety. A 2(pre test, post test) x 2(sex) x 3(condition) repeated measures ANOVA indicated CORT was lower at post test (M = 1.18 nmol/L, SD 0.71) than at pre test (M = 1.59, SD 0.96), F(1,104) = 18.87, p < .001, partial = [h.sup.2] 0.15. Participants were less aroused while completing the test items than they were prior to the session. However, change in CORT between pre and post test did not interact with sex, F(1,104) = 0.13, p < .721, partial [h.sup.2] = 0.001, condition, F(2,104) = 0.15 p <.86 partial [h.sup.2] = 0.003, nor with sex and condition combined, F(2,104) = 0.73 p < .49 partial [h.sup.2] = 0.014, (see Table 3 for descriptive statistics).
Characteristics of the testing environment may have reduced the likelihood of finding the stereotype threat effect. More than three fourths of the participants in the present study were female. Consequently, in all sessions females were in the majority, even when counting the three experimenters. As Inzlicht and Ben-Zeev (2000, 2003) found, women perform better when they are not in the minority, and when in same-sex groups they perform as well as men. Experimenters in this study may have been viewed as more sympathetic and supportive than those in other studies. Previous research has been conducted at universities where participants may have been unlikely to know the experimenters. At a liberal arts college of approximately 1600 students, participants were likely to have had one of the experimenters in class, especially because so many of the participants were majoring in the researchers’ fields. With class sizes typically not exceeding 30, students receive more individual attention at the institution, so even if a participant did not get such attention from the experimenters prior to the study, they may still have seen the experimenters as playing the role of supportive instructors who did not believe in a sex difference on this test. The small auditorium where data were collected was also familiar to all students. Orientation meetings, guest lectures, and other events are held in the facility, so participants were likely to feel more comfortable there than if they were at a university looking for a research lab in a building they had rarely, if ever, entered.
Stereotype threat did not affect the performance of participants in the present study. The effect might not have been found because participants did not carefully read the instructions meant to induce the different conditions, because of a non-representative sample, or perhaps because the effect is only found under certain conditions. True, the effect has been found in numerous studies of college students, in highly controlled, well-designed experiments, and research has been conducted to try to understand how stereotype threat can be overcome. Perhaps special efforts are not needed to overcome the effects of stereotype threat on women’s math performance at smaller educational institutions where personal attention of instructors is expected, and women are no longer in the minority (U.S. Census Bureau, 2005).
Aronson, J, Lustina, M. J., & Good, C. (1999). When White men can’t do math: Necessary and sufficient factors in stereotype threat. Journal of Experimental Social Psychology, 35, 29-46.
Beck, K. D., & Luine, V. N. (2002). Sex differences in behavioral and neurochemical profiles after chronic stress: Role of housing conditions. Physiological Behavior; 75, 661-673.
Ben-Zeev, T., Fein, S., & Inzlicht, M. (2005). Arousal and stereotype threat. Journal of Experimental Social Psychology, 41, 174-181.
Blascovich, J., Spencer, S. J., Quinn, D., & Steele, C. (2001). African Americans and high blood pressure: The role of stereotype threat. Psychological Science, 12, 225-229.
Bowman, R. E., Beck, K. D., & Luine, V. N. (2003). Chronic stress effects on memory: Sex differences in performance and monoaminergic activity. Hormones and Behavior, 43, 48-59.
Brown, R. P., & Josephs, R. A. (1999). A burden of proof: Stereotype relevance and gender differences in math performance. Journal of Personality, and Social Psychology, 76, 246-257.
Cahill L., & McGaugh J.L. (1998). Mechanisms of emotional arousal and lasting declarative memory. TINS, 21,273-313.
Conrad C. D., Galea, L. A. M., Kuroda, Y., & McEwen, B.S. (1996). Chronic stress impairs rat spatial memory on the Y-maze, and this effect is blocked by tianeptine pretreatment. Behavioral Neuroscience, 110, 1321-1334.
Conrad, C.D., Jackson, J.L., Wieczorek, L., Baran, S.E., Harman, J.S., Wright, R.L., & Korol, D.L. (2004). Acute stress impairs spatial memory in male but not female rats: influence of estrous cycle. Pharm. Biochem. Behav, 78, 569-579.
Folkman, S., & Lazarus, R. S. (1988). Ways of Coping Questionnaire. Consulting Psychologists Press, Inc.
Ford, T. E., Ferguson, M. A., Brooks, J. L., & Hagadone, K. M. (2004). Coping sense of humor reduces effects of stereotype threat on women’s math performance. Personality and Social Psychology Bulletin, 30, 643-653.
GRE: Practice to take the general test (9th ed.). (2001). Princeton, NJ: Educational Testing Service.
HeUhammer DH, Kirschbaum C, Belkien L: Measurement of salivary cortisol under psychological stimulation. In Hingtgen J, Hell-hammer D, Huppmann G (eds), Advanced Methods in Psychobiology. Toronto, Canada, C. J. Hogrefe, Inc., 1987
Hyde, J. S., Fennema, E., & Lamon, S. J. (1990). Gender differences in mathematics performance: A meta-analysis. Psychological Bulletin, 107, 139-155.
Inzlicht, M., Aronson, J., Good, C., & McKay, L. (2006). A particular resiliency to threatening environments. Journal of Experimental Social Psychology, 42, 323-336.
Inzlicht, M., & Ben-Zeev, T. (2000). A threatening intellectual environment: Why women are susceptible to experience problem-solving deficits in the presence of men. Psychological Science, 11,365-371.
Inzlicht, M., & Ben-Zeev, T. (2003). Do high-achieving female students underperform in private? The implications of threatening environments on intellectual processing. Journal of Educational Psychology, 95, 796-805.
Issa, A. M., Rowe, W., Gauthier, S., & Meaney, M. J. (1990). Hypothalamic-pituitary-adrenal activity in aged, cognitively impaired and cognitively unimpaired rats. Journal of Neuroscience, 10, 3247-3254.
Johns, M., Schmander, T., & Martens, A. (2005). Teaching stereotype threat as a means of improving women’s math performance. Psychological Science, 16, 175-178.
Kim, J. J. & Diamond, D. M. (2002). The stressed hippocampus, synaptic plasticity and lost memories. Nature Reviews Neuroscience, 3, 453-462.
Luine, V. (2002). Sex differences in chronic stress effects on memory in rats. Stress, 5, 205-216.
Luine, V. N. (1994). Steroid hormone influences on spatial memory. Annals of the New York Academy of Sciences, 743, 201-211.
Luine, V., Villegas, M., Martinez, C., & McEwen, B. S. (1994). Repeated stress causes reversible impairments of spatial memory performance. Brain Research, 639, 167-170.
Marx, D. M., & Roman, J. S. (2002). Female role models: Protecting women’s math test performance. Personality and Social Psychology Bulletin, 28, 1183-1193.
McIntyre, R. B., Paulson, R. M., & Lord, C. G. (2003). Alleviating women’s mathematics stereotype threat through salience of group achievement. Journal of Experimental Social Psychology, 39, 83-90.
O’Brien, L. T, & Crandall, C. S. (2003). Stereotype threat and arousal: Effects on women’s math performance. Personality and Social Psychology Bulletin, 29, 782-789.
Rhodes, M. E., & Rubin, R. T. (1999). Functional sex differences (“sexual diergism”) of CNS cholinergic systems, vasopressin, and hypothalamic-pituitary-adrenal axis activity in mammals: A selective review. Brain Research Reviews, 30, 135-152.
Rhodes, M. E., O’Toole, S. M., Wright, S. L., Czambel, R. K., & Rubin, R. T. (2001). Sexual diergism in rat hypothalamic-pituitary-adrenal axis responses to cholinergic stimulation and antagonism. Brain Research Bulletin, 54, 101-113.
Sathy, V., Barbuti, S., & Mattern, K. (2006). The new SAT and trends in test performance. New York: College Board.
Schmander, T., & Johns, M. (2003). Converging evidence that stereotype threat reduces working memory capacity. Journal of Personality and Social Psychology, 85, 440-452.
Spencer, S. J., Steele, C M., & Quinn, D. M. (1999). Stereotype threat and women’s math performance. Journal of Experimental Social Psychology, 35, 4-28.
Steele, C. M., & Aronson, J. (1995). Stereotype threat and intellectual test performance of African Americans. Journal of Personality and Social Psychology, 69, 797-811.
Stone, J. (2002). Battling doubt by avoiding practice: The effects of stereotype threat on self-handicapping in white athletes. Personality and Social Psychology Bulletin, 28, 1667-1678.
Stone, J., Lynch, C.I., Sjomeling, M., & Darley, J. M. (1999). Stereotype threat effects on Black and White athletic performance. Journal of Personality and Social Psychology, 77, 1213-1227.
United States Census Bureau. (2005). School Enrollment–Social and Economic Characteristics of Students: October 2005. Retrieved January 5, 2007, from http://www.census.gov/ population/www/socdemo/school/cps2005.html
Vining, R. F., McGinley, R. A., Symons, R.G. (1983) Hormones in saliva: Mode of entry and consequent implications for clinical interpretation. Clin Chem, 29, 1752-1756.
von Hippel, W., von Hippel, C., Conway, L., Preacher, K. J., Schooler, J. W., & Radvansky, G. A. (2005). Coping with stereotype threat: Denial as an impression management strategy. Journal of Personality and Social Psychology, 89, 22-35.
MARK G. RIVARDO, MICHAEL E. RHODES AND BRANDI KLEIN
Saint Vincent College
Performance Accuracy by Stereotype Threat Group and Sex
Group Sex M SD
Problem Solving Male 26.76 13.65
Female 38.83 17.36
Stereotype Threat Male 39.31 10.80
Female 39.10 19.68
Stereotype threat Male 42.26 16.34
plus Instruction Female 43.48 13.88
Self-reported SAT MathPerformance by Stereotype Threat Group and Sex
Group Sex M SD
Problem Solving Male 476.86 82.23
Female 560.96 70.38
Stereotype Threat Male 496.67 59.89
Female 551.90 83.03
Stereotype threat Male 549.09 101.24
plus Instruction Female 555.27 74.90
CORT by Stereotype Threat Group, Sex, and Test Time
Sex Pretest Post test
Group M SD M SD
Male Problem Solving 1.48 0.78 1.26 .77
Stereotype Threat 1.28 0.88 1.00 .47
Stereotype Threat 1.83 1.12 1.35 .68
Female Problem Solving 1.56 1.09 1.19 .57
Stereotype Threat 1.82 1.71 1.31 .93
Stereotype Threat 1.31 .79 1.03 .65
COPYRIGHT 2008 Project Innovation (Alabama)
COPYRIGHT 2008 Gale, Cengage Learning