Stereotype Threat and Math Performance

Stereotype Threat and Math Performance

• When women tackle math problems, they face the risk of being judged by the stereotype that women are not as good at math
• Defined as stereotype threat, the apprehension caused by this stereotype may disrupt women's math performance.
• Study 1 illustrates women underperform on difficult math tests compared to men due to stereotype threat
• Study 2 shows the difference in performance could be eliminated by reducing stereotype threat, noting no gender differences exist
• When told tests produced gender differences, women performed worse than men.
• Study 3 replicated these findings with a less selected sample, exploring stereotype threat mediation.
• Stereotype threat may underlie gender differences in advanced math performance.

Social Psychology and Stereotypes

• Stereotyping and prejudice have been extensively studied in social psychology and is a major concern
• Includes the content of stereotypes, their effects on social perception, how they form and the motivational bases and origins of prejudice
• Previously, less focus was given to those targeted by prejudice and stereotypes.
• Research now focuses on the effects of and reactions to societal devaluation and stigmatization

Stereotype Threat

• Refers to facing judgment based on societal stereotypes.
• Most negative group stereotypes are commonly known throughout society and devalue individuals.
• Where stereotypes apply, any action or characteristic fitting that stereotype increases the likelihood of being evaluated based on it
• Stereotype threat manifests as a situational predicament where individuals risk being judged or fulfilling negative stereotypes.
• It is not confined to specific groups but can affect any group facing adverse stereotypes (e.g., generation "X," the elderly, white males).
• It's situationally specific, differing from the general devaluation of "marking" or stigma
• Stereotype threat impacts performance when stereotypes suggest a group lacks an important ability
• Performance suffers under the added pressure of fulfilling the stereotype
• This research examines women's math performance as a test for the theory and addresses women's underperformance and participation in math.
• Perceived math abilities in women could lead to feeling they don't belong in math, which causes disidentification to avoid possible evaluation
• Such a process originates from stereotype threat, that may influence women's participation in math.

Gender Differences in Math Performance

• Gender differences impact curricula, professions, and performance
• Differences are not always common, women mirror men except when the material is advanced
• Talented junior high school math students, boys outperformed girls on the SAT quantitative
• Hyde, Fennema, and Lamon find males did not outperform females in computational ability but in problem-solving in advanced levels
• Kimball found no gender differences in math unless at college calculus and analytical geometry, males do better
• Gender differences become more likely as students take more difficult coursework, as shown by national surveys

Explanations for Gender Differences

• Differences reflect genetically rooted sex differences in math ability
• Differences reflect gender role socialization, especially as males are encouraged to participate in math/sciences over females
• Socialization and stereotype threat may influence difficult tasks for the performer
• A test doesn't need to be very difficult for stereotype threat to occur, sitting down is the trigger
• Difficult test performance is most likely to interfere with self-evaluation
• Overcoming success on an easier test will counteract any threat a stereotype could have caused.
• Higher pressure from stereotype threat makes solving difficult tasks even harder

Study 1 Design

• Pattern of women underperforming men on difficult tests and performing equally on easy tests
• Highly qualified men and women with strong maths background test sample
• Advanced GRE exam in mathematics containing advanced calculus was given, some questions required real variable theory and abstract algebra
• Quantitative section of the GRE general exam containing advanced algebra, trigonometry, and geometry, excluding calculus was given
• Tests conducted on computer to measure how much time participants took
• 28 men and 28 women from University of Michigan's psychology pool
• Participants had completed at least one semester of calculus with a B grade or higher and scored above the 85th percentile on SAT/ACT maths
• Participants strongly agreed on an 11 point scale that they (1) are good at maths (2) its important to be good at maths.
• 2 (male and female) x 2 (easier and difficult maths test) design measured performance on the maths test and time spent working on the test
• Participants could answer, leave blank or skip questions.
• A microcomputer recorded participants responses and how long each question took

Study 1 Procedure

• Participants told they are developing new tests evaluated across a group of Michigan students, where today they will take a maths test
• The first screen had instructions on scored and computer use with 30 minute test time
• Instructions taken from GRE which defined terms, and symbols, and provided a sample item
• The experimenter assigned a code word to determine condition assignments and stayed blind to assignments, they were also male
• Participants are debriefed after the test

Study 1 Results

• The pattern of women underperforming on difficult tests and performing well on easier tests was replicated
• Two-way ANOVA revealed a significant main effect for sex (F(1, 52) = 3.99, p = .05) and test difficulty (F(1, 52) = 137.27, p < .001)
• The interaction between gender and test difficulty was significant, F(1, 52) = 5.34, p < .05.
• Student-Newman-Keuls posthoc comparisons showed women performed worse on difficult math tests than the other groups (p < .05).
• Women worked 1497s on the difficult test (men 1539s), and 1738s on easy test (men 1599s)
• Two-way ANOVA revealed only a marginal main effect for test difficulty (F(1, 52) = 3.151, p < .10), with more time on the easy test
• Differences observed in literature can be replicated with selected group.
• Women underperformed compared to men on the difficult test, but did just as well on the easy test

Follow Up Study 2

• Mirrors results of Study 1 but poses concerns about what causes the differences
• Women experience stereotype threat when taking math tests, which undermines difficult tests
• Alternative interpretations remain, such as women equaled men because advanced material reveals true ability differences
• Effects of stereotype threat tested by giving all participants difficult maths exam from study 1
• If the gender stereotype was relevant was varied to test stereotype relevance against performance
• Participants told the test had/had not shown gender differences in the past
• No attack of the validity, just that women do well on it
• Irrelevant stereotype should eliminate underperformance, test is insensitive to gender differences and doesn't cause difficulty
• Ability differences would mean a women should still underperform even if the stereotype is relevant
• Study tests it is a stereotype-guided explanation and difficult performance tests

Study 2 Design & Procedure

• 30 women and 24 men meeting same criteria of Study 1 from the Michigan introductory psychology participator pool
• 2 x 2 mixed model design for sex and test characterisation assessed performance on maths test and time on test
• Tests administered on computers identically to study 1
• Difficult test was divided into 2 halves with 15 minutes per half consisting of 6 questions then 5 questions from the study 1 test as first and second test
• Half participants were told the first test showed gender differences but the second didn't and vice versa, these assignments were randomised
• Participants were told of maths and gender difference controversies despite background
• First test would show gender differences but the second would not depending on the order condition

Study 2 Results

• Two halves of test did not have equal difficulty as mean for both men and women on second test was zero, demonstrating the floor effect
• Only halves with no floor effect were the dependent variable is this experiment
• Test characterisation as people told of gender differences is one level, those told it didn't is the other
• Two-way ANOVA (Sex x Test Characterization) confirmed the experiment's predictions, F(1, 50) = 5.66, p < .05 and significant sex-by-test interaction F(1, 50) = 4.18, p < .05
• Women underperformed when told of gender differences and performed equally when test did not yield a gender difference
• Women in gender difference spent an average of 609 s while men spent 817 s. In no-gender characterisation it was 659 s for women, 697 s for men
• Only half with no floor effects results showed men tended to spend slightly more time on these test items than women with p=0.1

Study 2 Conclusions

• Characterizing a test as insensitive to gender differences was enough to eliminate women’s underperformance
• The same test with gender difference characteristation replicated significant underperformance
• When told differences do not occur, stereotype of women’s math inability was irrelevant
• Evidence that women’s underperformance on difficult math tests is because of stereotype threat and not sex-linked ability
• Women did better when told differences didn't exist demonstrating the test-taking situation accounts for the difference and not abilities
• Characterising test as producing differences benefited men or not producing interfered with their performance
• No significance for women to spend less time, but condition undermined motivation to return

Follow Up Study 3

• Study 2 showed stereotype threat can depress performance on difficult tests, improving them without the threat
• Floor effect raises the effect possibility of certain no. questions, raising generalizability
• Third, explicitly stated gender differences, raising when these allegations occur
• Replicated study 2 but sample came from another university with a test with wider problems and a control with no explicit difference, more confidence
• Aims to explore mediating stereotype threat on women's math performance
• Caused by predicament and is further risk of being judged, literatures show this interferes

Lower Performance and Expectations

• Stemmed from stereotype and from they already had, women have lower expectations than men from previous tests
• Overcoming of the "no gender differences" reduced by raising women's expectation
• Condition differences can be mediated
• Preliminary test where study measured state, self after manipulated
• Variables mediating should vary instructions with performance on the test

Study 3 Method Details

• 36 women and 31 men from State University of New York at Buffalo with an easier test
• Selected based on scores between 400 and 650 on SAT and those who completed 1 year of calc
• Test would be difficult whilst being within ability of participants
• 2 x 2 factorial with sex and test characterisation
• Dependent is performance and collected eval apprehension, state and efficacy
• Experiment given on paper in 20 minutes, except was made easier similar to 1 and 2
• Based math portion of graduate Management Test

General Points of Study 3

• Designed to examine possible mediators collected after the text manipulation
• No gender conditions described from study 2: Participants told there was no differences and performed equal
• They completed 4 questions on eval 5 on efficacy and state trait anxiety
• Instructions similar, ran in mixed-sex by experimenter so she saw
• Problem not completed on a sample problem for difficulty so didn't affect mediation

Study 3 Results

• The results from trials are replicated and a lower selected part, still performs equally with no-gender
• ANOVA is main effects
• Those with lower performance are affected by stereotype
• This shows that woman performs at a higher standard is stereotype decrease

Analyses of Study 3

• There 3 possible mediators in the test, evaluation, apprehension, anxiety and self efficacy
• Factor analysis was used on questionnaires were tested in the study
• 4 factors emerged close to meditational

Investigating the impact of stereotype threat

• Undertook 3 additional analyses to see mediation and effects
• First each potential mediator was tested effect
• Then they were checked to account for effects of the manipulation on test
• Third, individually to see effects of the manipulation test
• Analyses done in regression format
• Series is conducted allowing predicted
• With no evidence or relationship
• It had no discernible impact

Mediator Checks

• Any predicted test and if they made a stereotype perform
• Hierarchical regression: Variables - anxiety and eval
• With what they were
• Efficacy and not only were
• In equation was non existence
• Showed relationship
• Also examine as the test of each alone is mediator
• There is an issue of the 69% as a result can have a suppressor effect so a new one was created

Residual Score and Anxiety

• Made you to what the measure
• After the 3 the potential test is to see
• As shown in in the analysis it was effective where these were the same there was an indication where
• It didn't appear that this had the ability to change

Summary of Study 3 and Discussion Points

• Showed the relationship and more to see and understand
• There needs to small change but it showed and did have
• Also though relate affect has
• Overall tests don't give clear show effects and info
• They weren't
• There for these do not have a compelling for these affect