Is agreeing with a gender stereotype correlated with the performance of female students in introductory physics?
Is agreeing with a gender stereotype correlated with the performanceof female students in introductory physics?
Alexandru Maries, Nafis I. Karim, and Chandralekha Singh Department of Physics, University of Cincinnati, Cincinnati, Ohio 45221, USA Department of Physics and Astronomy, University of Pittsburgh, Pittsburgh, Pennsylvania 15260, USA (Received 23 June 2018; published 15 November 2018)Several prior studies in introductory physics have found a gender gap, i.e., a difference between male andfemale students ’ performance on conceptual assessments such as the Force Concept Inventory (FCI) andthe Conceptual Survey of Electricity and Magnetism (CSEM) with male students performing better thanfemale students. Moreover, prior studies in the context of mathematics have also found that activation of anegative stereotype about a group or stereotype threat, e.g., asking test takers to indicate their ethnicitybefore taking a test, can lead to deteriorated performance of the stereotyped group. Here, we describe twostudies in which we investigated the impact of interventions on the gender gap on the FCI and CSEM inlarge (more than 100 students) introductory physics courses at a large research university. In the first study,we investigated whether asking introductory physics students to indicate their gender immediately beforetaking the CSEM increased the gender gap compared to students who were not asked for this information.We found no difference in performance between male and female students in the two conditions. In thesecond study, which was conducted with several thousand introductory physics students, we investigatedthe prevalence of the belief that men generally perform better in physics than women and the extent towhich this belief is correlated with the performance of both female and male students on the FCI and theCSEM in algebra-based and calculus-based physics courses. We found that at the end of the year-longcalculus-based introductory physics sequence, in which female students are significantly underrepresented,agreeing with a gender stereotype was correlated negatively with the performance of female students on theconceptual physics surveys. The fact that female students who agreed with the gender stereotype performedworse than female students who disagreed with it at the end of the year-long calculus-based physics coursemay partly be due to an increased stereotype threat that female students who agree with the stereotype mayexperience in this course in which they are severely underrepresented. DOI: 10.1103/PhysRevPhysEducRes.14.020119
I. INTRODUCTION
Prior research has found that in introductory physicscourses male students often outperform female studentson conceptual assessments such as the Force ConceptInventory (FCI) [1] and the Conceptual Survey ofElectricity and Magnetism (CSEM) [2], a phenomenonsometimes referred to as the “ gender gap ” [3 – – – –
30] that keepaccumulating from an early age. For example, researchsuggests that even six-year-old boys and girls have genderedviews about smartness in favor of boys [30]. Such stereo-types can impact female students ’ self-efficacy [31 – Published by the American Physical Society under the terms ofthe Creative Commons Attribution 4.0 International license.Further distribution of this work must maintain attribution tothe author(s) and the published article ’ s title, journal citation,and DOI. PHYSICAL REVIEW PHYSICS EDUCATION RESEARCH = = = “ Is Math a Gift?Beliefs that Put Females at Risk ” [32], she argues that a fixedmindset is more detrimental to female students than malestudents. She describes a study in which two groups ofadolescents were taught the same math lesson (whichincluded historical information about the mathematicianswho originated the ideas discussed in the lesson) in twodifferent ways. For one group, the mathematicians wereportrayed as geniuses and their “ innate ability ” and “ naturaltalent ” were highlighted, whereas for the other group, themathematicians ’ commitment and hard work were high-lighted. After the lesson, students were given a difficultmath test and were told that the test would measure theirmathematical ability. Female students who received thelesson which portrayed the mathematicians as geniusesperformed worse than their male counterparts. On theother hand, for students who received the lesson whichhighlighted the mathematicians ’ hard work, there were nogender differences in performance. Dweck argues that whenfemale students receive messages that mathematical ability isa gift, some of them may interpret that this gift is somethingthey do not possess [32,33].Furthermore, prior research has also found that activationof a stereotype about a particular group in a test-takingsituation, i.e., stereotype threat (ST), can alter the perfor-mance of that group in a way consistent with the stereotype[29,34 – et al. conducted a study[44] in which a group of students was told immediatelybefore taking a mathematics test that in prior administra-tions of the test, a gender gap has been found (with femalestudents performing worse than male students), whileanother group was not provided with this information.Female students who were informed about the stereotyperight before the test performed significantly worse thanthose who were not exposed to this stereotype, but theperformance of male students was unaffected. Theresearchers concluded that informing female students aboutthe stereotype acts as a stereotype threat and leads todeteriorated performance [44]. Spence et al. [44] alsodescribe another study in which, when students were toldthat the mathematics test they are about to take is genderneutral, no gender gap was observed, but in the controlcondition, when students were not given any such infor-mation about the gender neutrality of the mathematics test,a gender gap was observed. The researchers hypothesizedthat a stereotype threat may be present for female studentsin a mathematics test-taking situation unless they areexplicitly told that the mathematics test has previouslybeen found to be gender neutral [44].Other researchers have found more subtle stimuli thatcan activate stereotype threat and result in deterioratedperformance (for a comprehensive review of research on effects of stereotype activation on behavior see Ref. [34],and another more recent one on the impact of stereotypethreat on learning see Ref. [29]). For example, priorresearch suggests that asking African American studentsto indicate their ethnicity before taking a difficult test onverbal ability resulted in decreased performance comparedto students of the same race who were not asked for thisinformation [45]. Other studies [47] have found that askingAsian female students to indicate their gender beforecompleting a mathematics test had a negative impact ontheir performance, but asking them to indicate theirethnicity had a positive effect. Researchers suggested thatin the first case, female students may have been experi-encing a stereotype threat (due to negative stereotypesabout the performance of female students), while in thesecond, they may have been experiencing a stereotypeboost (due to positive stereotypes about the performance ofAsian students). However, other studies have found thatasking for gender or ethnicity before taking a standardizedtest did not impact students ’ performance on the test[48,49]. Strickler and Ward [48] conducted a study inwhich they looked at whether asking students to indicateethnicity or gender before taking the Advanced PlacementCalculus AB exam and the Computerized Placement Tests( CPT ). They found that “ the experimental manipulation ofinquiring about ethnicity and gender did not have anydifferential effect on the various ethnic groups or boys andgirls that were both statistically and practically significant. ” We note, however, that previous research has often foundthat inquiring about gender or ethnicity can lead to stereo-type threat and result in deteriorated performance of astereotyped group (see the reviews in Refs. [29,34]).While the impact of stereotype threat has been studiedextensively [29,34 – – et al. [51] suggest that stereotype threat may be partly respon-sible for the gender gap and that the values affirmation mayhave reduced the gender gap by reducing the effects ofstereotype threat. Apart from the Miyake et al. study [51],there is only one study that we are aware of whichinvestigates the connection between stereotype threat andMARIES, KARIM, and SINGH PHYS. REV. PHYS. EDUC. RES.14,
30] that keepaccumulating from an early age. For example, researchsuggests that even six-year-old boys and girls have genderedviews about smartness in favor of boys [30]. Such stereo-types can impact female students ’ self-efficacy [31 – Published by the American Physical Society under the terms ofthe Creative Commons Attribution 4.0 International license.Further distribution of this work must maintain attribution tothe author(s) and the published article ’ s title, journal citation,and DOI. PHYSICAL REVIEW PHYSICS EDUCATION RESEARCH = = = “ Is Math a Gift?Beliefs that Put Females at Risk ” [32], she argues that a fixedmindset is more detrimental to female students than malestudents. She describes a study in which two groups ofadolescents were taught the same math lesson (whichincluded historical information about the mathematicianswho originated the ideas discussed in the lesson) in twodifferent ways. For one group, the mathematicians wereportrayed as geniuses and their “ innate ability ” and “ naturaltalent ” were highlighted, whereas for the other group, themathematicians ’ commitment and hard work were high-lighted. After the lesson, students were given a difficultmath test and were told that the test would measure theirmathematical ability. Female students who received thelesson which portrayed the mathematicians as geniusesperformed worse than their male counterparts. On theother hand, for students who received the lesson whichhighlighted the mathematicians ’ hard work, there were nogender differences in performance. Dweck argues that whenfemale students receive messages that mathematical ability isa gift, some of them may interpret that this gift is somethingthey do not possess [32,33].Furthermore, prior research has also found that activationof a stereotype about a particular group in a test-takingsituation, i.e., stereotype threat (ST), can alter the perfor-mance of that group in a way consistent with the stereotype[29,34 – et al. conducted a study[44] in which a group of students was told immediatelybefore taking a mathematics test that in prior administra-tions of the test, a gender gap has been found (with femalestudents performing worse than male students), whileanother group was not provided with this information.Female students who were informed about the stereotyperight before the test performed significantly worse thanthose who were not exposed to this stereotype, but theperformance of male students was unaffected. Theresearchers concluded that informing female students aboutthe stereotype acts as a stereotype threat and leads todeteriorated performance [44]. Spence et al. [44] alsodescribe another study in which, when students were toldthat the mathematics test they are about to take is genderneutral, no gender gap was observed, but in the controlcondition, when students were not given any such infor-mation about the gender neutrality of the mathematics test,a gender gap was observed. The researchers hypothesizedthat a stereotype threat may be present for female studentsin a mathematics test-taking situation unless they areexplicitly told that the mathematics test has previouslybeen found to be gender neutral [44].Other researchers have found more subtle stimuli thatcan activate stereotype threat and result in deterioratedperformance (for a comprehensive review of research on effects of stereotype activation on behavior see Ref. [34],and another more recent one on the impact of stereotypethreat on learning see Ref. [29]). For example, priorresearch suggests that asking African American studentsto indicate their ethnicity before taking a difficult test onverbal ability resulted in decreased performance comparedto students of the same race who were not asked for thisinformation [45]. Other studies [47] have found that askingAsian female students to indicate their gender beforecompleting a mathematics test had a negative impact ontheir performance, but asking them to indicate theirethnicity had a positive effect. Researchers suggested thatin the first case, female students may have been experi-encing a stereotype threat (due to negative stereotypesabout the performance of female students), while in thesecond, they may have been experiencing a stereotypeboost (due to positive stereotypes about the performance ofAsian students). However, other studies have found thatasking for gender or ethnicity before taking a standardizedtest did not impact students ’ performance on the test[48,49]. Strickler and Ward [48] conducted a study inwhich they looked at whether asking students to indicateethnicity or gender before taking the Advanced PlacementCalculus AB exam and the Computerized Placement Tests( CPT ). They found that “ the experimental manipulation ofinquiring about ethnicity and gender did not have anydifferential effect on the various ethnic groups or boys andgirls that were both statistically and practically significant. ” We note, however, that previous research has often foundthat inquiring about gender or ethnicity can lead to stereo-type threat and result in deteriorated performance of astereotyped group (see the reviews in Refs. [29,34]).While the impact of stereotype threat has been studiedextensively [29,34 – – et al. [51] suggest that stereotype threat may be partly respon-sible for the gender gap and that the values affirmation mayhave reduced the gender gap by reducing the effects ofstereotype threat. Apart from the Miyake et al. study [51],there is only one study that we are aware of whichinvestigates the connection between stereotype threat andMARIES, KARIM, and SINGH PHYS. REV. PHYS. EDUC. RES.14, II. GOALS OF THE INVESTIGATIONS
Since stereotype threat has the potential to exacerbate thegender gap typically found in conceptual physics assess-ments, and inquiring about gender or ethnicity has beenfound to potentially activate stereotype threat, in study 1described here, our goal was to investigate whether askingintroductory physics students to indicate their genderbefore taking the CSEM impacted their performance, bothwhen it was administered as a pretest (before instruction inrelevant concepts) and as a post-test (after traditionallecture-based instruction in relevant concepts). In study 2described here, our goal was to investigate the prevalenceof the belief that men generally perform better in physicsthan women (a gender stereotype) among introductoryphysics students and the extent to which agreeing with thisgender stereotype is correlated with the performance offemale and male students in algebra-based and calculus-based introductory physics I and II on the commonly usedconceptual standardized physics assessments, the FCI andthe CSEM. We also investigated whether there was adifference between the conditions in which the genderstereotype question was asked immediately before orimmediately after students took the FCI or the CSEMfor both male and female students. Based on prior research,we hypothesized that reminding students about the stereo-type before taking the FCI or CSEM (by asking them abouttheir opinion of it) may act as a stereotype threat (orexacerbate the stereotype threat already present) and resultin lower performance of the stereotyped group (femalestudents) compared to when students were not reminded ofthe stereotype.As noted, the research by Marchand and Taasoobshirazi[28] suggests that many female students automaticallyexperience a certain level of stereotype threat while takinga physics test due to the societal stereotypes about physicsbeing a discipline for intelligent men. We hypothesized that while a certain level of stereotype threat may be implicitlypresent for many female students in the introductoryphysics courses, the stereotype threat may be worse, onaverage, for female students who agree with the genderstereotype that men generally perform better in physicsthan women. Moreover, without explicit intervention toimprove women ’ s sense of belonging, self-efficacy, andgrowth mindset, being in a physics course in which they areseverely underrepresented can have worse negative impacton the performance of the female students who believe inthe gender stereotype than those who do not believe in thestereotype. In particular, it is possible that for those femalestudents who agree with the gender stereotype, the eco-system of the physics classrooms in which they areunderrepresented may act as an additional level of stereo-type threat (over and above what female students mayexperience automatically in physics test-taking situationsdue to common societal biases), and they may performworse than female students who do not agree with thestereotype. One of our goals was to investigate this issue. III. METHODOLOGY
The participants in this study were students in algebra-based and calculus-based introductory physics courses.Also, the introductory physics courses (algebra-basedphysics I and II or calculus-based physics I and II) includedin this study were large introductory physics courses (morethan 100 students) at a typical large research university[University of Pittsburgh (Pitt)] except in one study, asdescribed, in which calculus-based introductory physicsstudents from another large research university [Universityof Cincinnati (UC)] participated. The data were collectedover a period of two years and include over 3300 studentsfrom Pitt and over 900 students from UC. We note that bothuniversities are large, public, state, research universitieslocated in neighboring (and similar) states, and the studentstaking introductory physics are similar. The two-semestercalculus-based course sequence at Pitt and Cincinnati ismainly taken by college freshman who are engineering,chemistry, mathematics, or physics majors. Approximately30% of the students in these calculus-based courses arefemales (somewhat higher percentage at Pitt than atCincinnati). The first semester course covers mainlymechanics and waves and the second semester coursecovers mainly electricity and magnetism and some waveoptics. The algebra-based introductory physics coursesequence at Pitt is taken mainly by the biological scienceand neuroscience majors and those who are premedicalstudents. It is taken primarily in the junior or senior year.Introductory mechanics and waves are covered in the firstsemester algebra-based course and electricity and magnet-ism are covered in the second semester algebra-basedcourse, although other topics are also included in thecourse in order to cover the topics in the medical entranceIS AGREEING WITH A GENDER STEREOTYPE … PHYS. REV. PHYS. EDUC. RES.14,
The participants in this study were students in algebra-based and calculus-based introductory physics courses.Also, the introductory physics courses (algebra-basedphysics I and II or calculus-based physics I and II) includedin this study were large introductory physics courses (morethan 100 students) at a typical large research university[University of Pittsburgh (Pitt)] except in one study, asdescribed, in which calculus-based introductory physicsstudents from another large research university [Universityof Cincinnati (UC)] participated. The data were collectedover a period of two years and include over 3300 studentsfrom Pitt and over 900 students from UC. We note that bothuniversities are large, public, state, research universitieslocated in neighboring (and similar) states, and the studentstaking introductory physics are similar. The two-semestercalculus-based course sequence at Pitt and Cincinnati ismainly taken by college freshman who are engineering,chemistry, mathematics, or physics majors. Approximately30% of the students in these calculus-based courses arefemales (somewhat higher percentage at Pitt than atCincinnati). The first semester course covers mainlymechanics and waves and the second semester coursecovers mainly electricity and magnetism and some waveoptics. The algebra-based introductory physics coursesequence at Pitt is taken mainly by the biological scienceand neuroscience majors and those who are premedicalstudents. It is taken primarily in the junior or senior year.Introductory mechanics and waves are covered in the firstsemester algebra-based course and electricity and magnet-ism are covered in the second semester algebra-basedcourse, although other topics are also included in thecourse in order to cover the topics in the medical entranceIS AGREEING WITH A GENDER STEREOTYPE … PHYS. REV. PHYS. EDUC. RES.14, –
20 minutes ofthe recitation class.In order to compare the performances of students underdifferent conditions, we performed t tests [53] on FCI orCSEM pretest and post-test data for male and femalestudents, which are commonly used to investigate whetherthe means of two populations are different from oneanother, which, as discussed in Ref. [53] (see Secs. 12. 9and 12. 10) can be performed even if the samples beingcompared are not close to normal distributions as long asthe numbers of observations (i.e., student scores) are morethan 15. In the cases in which our sample sizes were lessthan 15, we also performed a nonparametric test of meanranks (Mann-Whitney U), which makes no assumptionsabout the distribution of the populations being compared[53]. We also calculated the effect size in the form ofCohen ’ s d defined as j μ − μ j = σ pooled , where μ and μ arethe averages of the two groups being compared and σ pooled ¼ ffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi ð σ þ σ Þ = p (here σ and σ are the standarddeviations of the two groups being compared). We con-sidered d < . as small effect size, . ≤ d < . asmedium effect size and d ≥ . as large effect size, asdescribed in Ref. [53]. A. Study 1
In this study, 170 students in an introductory algebra-based physics II course (as noted, mostly biological andneuroscience majors and premedical students) took theCSEM as a pretest (in the recitation class in the first weekof classes before instruction in relevant concepts) and as apost-test (in the recitation class during the last week ofclasses after instruction in relevant concepts). Studentswere randomly assigned to two conditions, one whichasked them to indicate their gender (checkbox format withoptions male, female, and prefer not to specify) before theytook the CSEM and one in which they were asked for suchinformation after taking the CSEM. We then compared theperformance of students under the two conditions.
B. Study 2
This study involved over 1800 calculus-based students(mainly engineering, mathematics, and physical sciencemajors) and over 1600 algebra-based students (mainly pre-medical and biological and neuroscience majors) enrolled in first and second semester introductory physics courses.To investigate the prevalence of the belief in the genderstereotype, students were asked to indicate the extent towhich they agree with the following statement: “ Accordingto my own personal beliefs, I expect men to generallyperform better in physics than women ” on a five-pointLikert scale (strongly disagree, disagree, neutral, agree, andstrongly agree). To investigate the extent to which belief inthe stereotype is correlated with female and male students ’ performance on the FCI and CSEM, we grouped studentsaccording to their beliefs (agree or strongly agree, neutralwhich was explained to students as neither agree nordisagree, and disagree or strongly disagree) and we investi-gated performance differences (e.g., we compared theperformance of female students who agree with the stereo-type with that of female students in the same class whodisagree with the stereotype) on both the pretest (beforeinstruction) and the post-test (after instruction in relevantconcepts).We note that if students are asked to indicate the extent towhich they agree with the gender stereotype (according tomy own personal beliefs, I expect men to generally performbetter in physics than women) before taking the FCI orCSEM, this may act as an additional stereotype threat (overand above what may be present already due to studentsbeing in a test-taking situation), especially for the femalestudents who agree with this gender stereotype. Thus, toavoid any additional stereotype threat to female students(and potential consequences on performance on the stand-ardized test), all students at one large state-related univer-sity (Pitt) were given the gender stereotype question rightafter they had completed answering the FCI or CSEMquestions. However, we wanted to test whether asking thegender stereotype question before students take the con-ceptual survey qualitatively impacts female and malestudents ’ performance. Since it was agreed that at Pittthe gender stereotype question would be asked at the end(after students had taken the standardized test in therecitation) so that female students do not experience addi-tional stereotype threat, another group of calculus-basedintroductory physics students at the University ofCincinnati was asked the gender stereotype question rightbefore taking the FCI, and the qualitative trends amongstmale and female students who agreed or disagreed with thestereotype were compared with the corresponding calculus-based cohort at Pitt for whom the stereotype question wasasked right after taking the FCI. IV. RESULTS
Before discussing the results, we note that whether weuse matched pretest and post-test data or consider allstudents who took the pretest or post-test (unmatched),the qualitative trends are unchanged so we report data fromall students who took the pretest or post-test.MARIES, KARIM, and SINGH PHYS. REV. PHYS. EDUC. RES.14,
Before discussing the results, we note that whether weuse matched pretest and post-test data or consider allstudents who took the pretest or post-test (unmatched),the qualitative trends are unchanged so we report data fromall students who took the pretest or post-test.MARIES, KARIM, and SINGH PHYS. REV. PHYS. EDUC. RES.14, . Study 1
Figure 1 shows the pretest and post-test performance ofintroductory algebra-based female ( N ¼ ) and male( N ¼ ) students on the CSEM in the two conditions:students were or were not asked to provide genderinformation before taking the CSEM (gender salient ornot salient condition, respectively). Figure 1 shows thatthere were no statistically significant differences betweenthe performance of male or female students in the twoconditions (e.g., female students who wrote their genderbefore taking the CSEM did not perform worse than femalestudents who wrote their gender after taking the CSEM) inthe pretest or the post-test. B. Study 2
Table I shows the percentage of male and femaleintroductory students in algebra-based and calculus-based physics I and II courses who agreed or were neutral ordisagreed with the stereotype (according to my ownpersonal beliefs, I expect men to generally perform betterin physics than women). Only 7% –
13% of algebra-basedand calculus-based students (regardless of their gender)agreed with this gender stereotype. Thus, it appears that thisstereotype was not very common amongst college intro-ductory physics students. (We note that the percentage offemale students in algebra-based and calculus-basedclasses was around two-thirds and one-third, respectively,and it was fairly constant over the two years of datacollection.)Before presenting the data from study 2, we note that inall classes regardless of whether they were algebra based orcalculus based, or whether they were introductory physics Ior II, large gender differences were found in our inves-tigation between the performance of male and femalestudents both on the pretest and on the post-test. Thegender gap on the FCI was typically 10% –
20% dependingupon whether it was the pretest or post-test and whether itwas the algebra-based or calculus-based introductorycourse. The gender gap on the CSEM was typically smaller,especially on the pretest where the overall scores of eachgroup (particularly for the algebra-based course) were notsignificantly better than random guessing (20% average).Figures 2 – p values and effectsizes (Cohen ’ s d ) for the comparison of the performance offemale or male students who agreed with the stereotypewith that of female or male students who disagreed with thestereotype.Figures 2 and 3 show that for the algebra-based intro-ductory physics students, neither on the FCI nor on theCSEM were there major differences between the female ormale students who agreed and female or male students whodisagreed with the gender stereotype, on the pretest or onthe post-test.Figure 4 shows that for the calculus-based introductoryphysics students, on the FCI, there were no statistically FIG. 1. Female and male students ’ pretest and post-test per-formance on the CSEM depending on the testing condition. The p values are obtained using a t test and d refers to the effect size(Cohen ’ s d [53]). The comparisons for the post-test have alsobeen done via a nonparametric test due to small sample sizes, theMann-Whitney U test [53], and the p values obtained were 0.320(female students) and 0.975 (male students). The error barsrepresent standard error.TABLE I. Percentage of female ( F ) and male ( M ) students who agreed or were neutral or disagreed with the stereotype that mengenerally perform better in physics than women in algebra-based (Alg.) and calculus-based (Calc.) introductory physics. The totalnumber of female or male students is indicated at the bottom ( N ).Alg. Physics I (FCI) Alg. Physics II (CSEM) Calc. Physics I (FCI) Calc. Physics II (CSEM)Pretest Post-test Pretest Post-test Pretest Post-test Pretest Post-test F M F M F M F M F M F M F M F M
Disagree 77 73 74 73 80 78 76 79 83 72 83 74 83 74 77 73Neutral 14 21 13 21 9 15 12 13 10 21 7 18 9 19 10 20Agree 9 7 13 7 11 7 12 7 8 7 10 8 9 7 13 7 N
668 365 450 251 553 330 348 219 253 453 217 354 231 527 181 396
IS AGREEING WITH A GENDER STEREOTYPE … PHYS. REV. PHYS. EDUC. RES.14,
IS AGREEING WITH A GENDER STEREOTYPE … PHYS. REV. PHYS. EDUC. RES.14, N these results maybecome statistically significant). Figure 5 shows that for thestudents in the calculus-based course, on the CSEM pretest, the trends were similar to the trends on the FCI and thedifferences between female and male students who agreedor disagreed with the gender stereotype were not sta-tistically significant. However, on the CSEM post-test,there was a statistically significant difference (a differenceof 8%) between the calculus-based female students whoagreed and the female students in the same course whodisagreed with the stereotype. Thus, while there was nostatistically significant difference between women who FIG. 2. FCI performance of algebra-based female or malestudents who agreed or disagreed with the stereotype. The p values ( p ) and effect sizes ( d ) were obtained when comparing theaverage performance of female or male students who agreed withthat of female or male students who disagreed with the stereotype.These students answered the stereotype question after taking theFCI. The numbers on the bars represent the number of students ineach group (e.g., 60 female students agreed with the stereotype inthe pretest) and the error bars represent standard errors.FIG. 3. CSEM performance of algebra-based female or malestudents who agreed or disagreed with the stereotype. The p values ( p ) and effect sizes ( d ) were obtained when comparing theaverage performance of female or male students who agreed withthat of female or male students who disagreed with the stereotype.These students answered the stereotype question after taking theFCI. The numbers on the bars represent the number of students ineach group (e.g., 61 female students agreed with the stereotype inthe pretest) and the error bars represent standard errors. FIG. 4. FCI performance of calculus-based female or malestudents who agreed or disagreed with the stereotype. The p values ( p ) and effect sizes ( d ) were obtained when comparing theaverage performance of female or male students who agreed withthat of female or male students who disagreed with the stereotype.These students answered the stereotype question after taking theFCI. The numbers on the bars represent the number of students ineach group (e.g., 19 female students agreed with the stereotype inthe pretest) and the error bars represent standard errors.FIG. 5. CSEM performance of calculus-based female or malestudents who agreed or disagreed with the stereotype. The p values ( p ) and effect sizes ( d ) were obtained when comparing theaverage performance of female or male students who agreed withthat of female or male students who disagreed with the stereotype.These students answered the stereotype question after taking theFCI. The numbers on the bars represent the number of students ineach group (e.g., 20 female students agreed with the stereotype inthe pretest) and the error bars represent standard errors. MARIES, KARIM, and SINGH PHYS. REV. PHYS. EDUC. RES.14,
IS AGREEING WITH A GENDER STEREOTYPE … PHYS. REV. PHYS. EDUC. RES.14, N these results maybecome statistically significant). Figure 5 shows that for thestudents in the calculus-based course, on the CSEM pretest, the trends were similar to the trends on the FCI and thedifferences between female and male students who agreedor disagreed with the gender stereotype were not sta-tistically significant. However, on the CSEM post-test,there was a statistically significant difference (a differenceof 8%) between the calculus-based female students whoagreed and the female students in the same course whodisagreed with the stereotype. Thus, while there was nostatistically significant difference between women who FIG. 2. FCI performance of algebra-based female or malestudents who agreed or disagreed with the stereotype. The p values ( p ) and effect sizes ( d ) were obtained when comparing theaverage performance of female or male students who agreed withthat of female or male students who disagreed with the stereotype.These students answered the stereotype question after taking theFCI. The numbers on the bars represent the number of students ineach group (e.g., 60 female students agreed with the stereotype inthe pretest) and the error bars represent standard errors.FIG. 3. CSEM performance of algebra-based female or malestudents who agreed or disagreed with the stereotype. The p values ( p ) and effect sizes ( d ) were obtained when comparing theaverage performance of female or male students who agreed withthat of female or male students who disagreed with the stereotype.These students answered the stereotype question after taking theFCI. The numbers on the bars represent the number of students ineach group (e.g., 61 female students agreed with the stereotype inthe pretest) and the error bars represent standard errors. FIG. 4. FCI performance of calculus-based female or malestudents who agreed or disagreed with the stereotype. The p values ( p ) and effect sizes ( d ) were obtained when comparing theaverage performance of female or male students who agreed withthat of female or male students who disagreed with the stereotype.These students answered the stereotype question after taking theFCI. The numbers on the bars represent the number of students ineach group (e.g., 19 female students agreed with the stereotype inthe pretest) and the error bars represent standard errors.FIG. 5. CSEM performance of calculus-based female or malestudents who agreed or disagreed with the stereotype. The p values ( p ) and effect sizes ( d ) were obtained when comparing theaverage performance of female or male students who agreed withthat of female or male students who disagreed with the stereotype.These students answered the stereotype question after taking theFCI. The numbers on the bars represent the number of students ineach group (e.g., 20 female students agreed with the stereotype inthe pretest) and the error bars represent standard errors. MARIES, KARIM, and SINGH PHYS. REV. PHYS. EDUC. RES.14,
V. DISCUSSION AND SUMMARY
Our investigation in study 1 suggests that asking algebra-based introductory physics students to indicate their genderbefore taking the CSEM did not impact their performance,consistent with a previous study conducted with the APcalculus exam and the Computerized Placement test [48].One possible explanation for this finding supported byprevious research [28] is that stereotype threat for femalestudents occurs implicitly regardless of whether or notstudents are asked to indicate their gender before taking the CSEM test because the stereotype is automatically acti-vated for female students in the test-taking situation inphysics and math. In other words, one possible explanationis that the threat may be present for this group regardless ofbeing explicitly asked about such personal informationexplicitly. Other high-stakes tests (e.g., MCAT, SAT)commonly require students to indicate their gender beforetaking the tests. If the results of study 1 were to hold forthese tests as well, then the common practice of asking forpersonal information such as gender may not impact theperformance of the stereotypically underperforming group.In study 2, we investigated the prevalence of the beliefthat men generally perform better in physics than womenamong introductory physics students and found that thistype of belief was not very common (around 7% –
13% ofalgebra-based and calculus-based students agreed with thisstereotype). We also investigated the extent to whichagreeing with the stereotype was correlated with students ’ performance on the FCI and CSEM. The analysis of datafrom study 2 suggests that in an algebra-based coursefemale students who agreed with the gender stereotype(men generally perform better in physics than women) andfemale students who disagreed with the stereotype hadsimilar performance (within 2%) on both the pretest and thepost-test. In other words, for students in algebra-basedcourses, there were no statistically significant performancedifferences on FCI or CSEM between female students whoagreed with the stereotype and female students who dis-agreed with it. For students in calculus-based courses, therewere no differences on the FCI (although on the post-test,there was a discernable trend of female students who agreedwith the gender stereotype performing worse than femalestudents who disagreed with it and larger number ofstudents may make it statistically significant), but for theCSEM post-test, female students who agreed with thegender stereotype performed worse than female studentswho disagreed with it (even though there was no sta-tistically significant difference between them on the pretestat the beginning of the course). In other words, at the end ofthe full year of a calculus-based introductory physicssequence, a statistically significant difference on theCSEM post-test for the calculus-based students emergedin that the female students who agreed with the stereotypeperformed significantly worse than female students whodisagreed with the stereotype (this result is not onlystatistically significant but also has practical implicationssince there was an 8% difference in female studentperformance between those who agreed and disagreed withthe stereotype).We note that in algebra-based courses, approximatelytwo-thirds of the students were female (compared toapproximately one-third in the calculus-based courses).This suggests that, in a calculus-based course, femalestudents who agree with the stereotype may be impactedmore by the associated stereotype threat since they see FIG. 6. FCI performance of calculus-based female or malestudents who agreed or disagreed with the stereotype. The p values ( p ) and effect sizes ( d ) were obtained when comparing theaverage performance of female or male students who agreed withthat of female or male students who disagreed with the stereotype.These students answered the stereotype question before takingthe FCI. The numbers on the bars represent the number ofstudents in each group (e.g., 35 female students agreed with thestereotype in the pretest) and the error bars represent standarderrors. There was a total of 917 students included in these data. IS AGREEING WITH A GENDER STEREOTYPE … PHYS. REV. PHYS. EDUC. RES.14,
13% ofalgebra-based and calculus-based students agreed with thisstereotype). We also investigated the extent to whichagreeing with the stereotype was correlated with students ’ performance on the FCI and CSEM. The analysis of datafrom study 2 suggests that in an algebra-based coursefemale students who agreed with the gender stereotype(men generally perform better in physics than women) andfemale students who disagreed with the stereotype hadsimilar performance (within 2%) on both the pretest and thepost-test. In other words, for students in algebra-basedcourses, there were no statistically significant performancedifferences on FCI or CSEM between female students whoagreed with the stereotype and female students who dis-agreed with it. For students in calculus-based courses, therewere no differences on the FCI (although on the post-test,there was a discernable trend of female students who agreedwith the gender stereotype performing worse than femalestudents who disagreed with it and larger number ofstudents may make it statistically significant), but for theCSEM post-test, female students who agreed with thegender stereotype performed worse than female studentswho disagreed with it (even though there was no sta-tistically significant difference between them on the pretestat the beginning of the course). In other words, at the end ofthe full year of a calculus-based introductory physicssequence, a statistically significant difference on theCSEM post-test for the calculus-based students emergedin that the female students who agreed with the stereotypeperformed significantly worse than female students whodisagreed with the stereotype (this result is not onlystatistically significant but also has practical implicationssince there was an 8% difference in female studentperformance between those who agreed and disagreed withthe stereotype).We note that in algebra-based courses, approximatelytwo-thirds of the students were female (compared toapproximately one-third in the calculus-based courses).This suggests that, in a calculus-based course, femalestudents who agree with the stereotype may be impactedmore by the associated stereotype threat since they see FIG. 6. FCI performance of calculus-based female or malestudents who agreed or disagreed with the stereotype. The p values ( p ) and effect sizes ( d ) were obtained when comparing theaverage performance of female or male students who agreed withthat of female or male students who disagreed with the stereotype.These students answered the stereotype question before takingthe FCI. The numbers on the bars represent the number ofstudents in each group (e.g., 35 female students agreed with thestereotype in the pretest) and the error bars represent standarderrors. There was a total of 917 students included in these data. IS AGREEING WITH A GENDER STEREOTYPE … PHYS. REV. PHYS. EDUC. RES.14, ’ s sense of belonging,self-efficacy and growth mindset, being in a calculus-basedphysics course in which they are severely underrepresentedmay have worse negative impact on the performance of thefemale students who believe the stereotype than those whodo not believe this stereotype. Thus, one reason for theemergence of the statistically significantly different perfor-mance between the female students in the calculus-basedcourse who disagree and those who agree with the genderstereotype on the CSEM post-test may be the cumulativeimpact of increased stereotype threat. In particular, forwomen who agree with the gender stereotype, there may beadditional stereotype threat over and above what femalestudents experience in a physics test taking situationimplicitly. Such an additional threat can create added levelof anxiety that can impact female students ’ performancefrom several angles. For example, due to added level ofanxiety, female students who agree with the stereotype may,on average, be less excited about learning physics and thisdecreased level of excitement can potentially lead to taskavoidance, i.e., less time learning physics. This is consistentwith the review by Appel and Kronenberg which suggeststhat stereotype threat inhibits learning for the stereotypedstudents. Taking into consideration research in cognitivescience related to learning and working memory [54], aswell as the research by Beilock and others on stereotypethreat, anxiety, and cognition [55 – – –14,
13% ofalgebra-based and calculus-based students agreed with thisstereotype). We also investigated the extent to whichagreeing with the stereotype was correlated with students ’ performance on the FCI and CSEM. The analysis of datafrom study 2 suggests that in an algebra-based coursefemale students who agreed with the gender stereotype(men generally perform better in physics than women) andfemale students who disagreed with the stereotype hadsimilar performance (within 2%) on both the pretest and thepost-test. In other words, for students in algebra-basedcourses, there were no statistically significant performancedifferences on FCI or CSEM between female students whoagreed with the stereotype and female students who dis-agreed with it. For students in calculus-based courses, therewere no differences on the FCI (although on the post-test,there was a discernable trend of female students who agreedwith the gender stereotype performing worse than femalestudents who disagreed with it and larger number ofstudents may make it statistically significant), but for theCSEM post-test, female students who agreed with thegender stereotype performed worse than female studentswho disagreed with it (even though there was no sta-tistically significant difference between them on the pretestat the beginning of the course). In other words, at the end ofthe full year of a calculus-based introductory physicssequence, a statistically significant difference on theCSEM post-test for the calculus-based students emergedin that the female students who agreed with the stereotypeperformed significantly worse than female students whodisagreed with the stereotype (this result is not onlystatistically significant but also has practical implicationssince there was an 8% difference in female studentperformance between those who agreed and disagreed withthe stereotype).We note that in algebra-based courses, approximatelytwo-thirds of the students were female (compared toapproximately one-third in the calculus-based courses).This suggests that, in a calculus-based course, femalestudents who agree with the stereotype may be impactedmore by the associated stereotype threat since they see FIG. 6. FCI performance of calculus-based female or malestudents who agreed or disagreed with the stereotype. The p values ( p ) and effect sizes ( d ) were obtained when comparing theaverage performance of female or male students who agreed withthat of female or male students who disagreed with the stereotype.These students answered the stereotype question before takingthe FCI. The numbers on the bars represent the number ofstudents in each group (e.g., 35 female students agreed with thestereotype in the pretest) and the error bars represent standarderrors. There was a total of 917 students included in these data. IS AGREEING WITH A GENDER STEREOTYPE … PHYS. REV. PHYS. EDUC. RES.14, ’ s sense of belonging,self-efficacy and growth mindset, being in a calculus-basedphysics course in which they are severely underrepresentedmay have worse negative impact on the performance of thefemale students who believe the stereotype than those whodo not believe this stereotype. Thus, one reason for theemergence of the statistically significantly different perfor-mance between the female students in the calculus-basedcourse who disagree and those who agree with the genderstereotype on the CSEM post-test may be the cumulativeimpact of increased stereotype threat. In particular, forwomen who agree with the gender stereotype, there may beadditional stereotype threat over and above what femalestudents experience in a physics test taking situationimplicitly. Such an additional threat can create added levelof anxiety that can impact female students ’ performancefrom several angles. For example, due to added level ofanxiety, female students who agree with the stereotype may,on average, be less excited about learning physics and thisdecreased level of excitement can potentially lead to taskavoidance, i.e., less time learning physics. This is consistentwith the review by Appel and Kronenberg which suggeststhat stereotype threat inhibits learning for the stereotypedstudents. Taking into consideration research in cognitivescience related to learning and working memory [54], aswell as the research by Beilock and others on stereotypethreat, anxiety, and cognition [55 – – –14, CKNOWLEDGMENTS
We would like to thank the National Science Foundationfor Grants No. DUE-152457 and No. PHY-1505460 andthe members of the physics education research group at the University of Pittsburgh as well as R. P. Devaty for usefuldiscussions and feedback on the manuscript. We also thankL. Pingel and C. Schunn for helpful discussions related tostatistical analysis. [1] D. Hestenes, G. Wells, and M. Swackhammer, ForceConcept Inventory, Phys. Teach. , 141 (1992).[2] D. Maloney, T. O ’ Kuma, C. Hieggelke, and A. VanHeuvelen, Surveying students ’ conceptual knowledge ofelectricity and magnetism, Am. J. Phys. , s12 (2001).[3] A. Madsen, S. McKagan, and E. Sayre, Gender gap onconcept inventories in physics: What is consistent, what isinconsistent, and what factors influence the gap?, Phys.Rev. ST Phys. Educ. Res. , 020121 (2013).[4] J. Docktor and K. Heller, Gender Differences in Both ForceConcept Inventory and Introductory Physics Performance,AIP Conf. Proc. , 15 (2008).[5] A. Traxler, X. Cid, J. Blue, and R. Barthelemy, Enrichinggender in physics education research: A binary past and acomplex future, Phys. Rev. Phys. Educ. Res. , 020114(2016).[6] V. Coletta, J. Phillips, and J. Steinert, FCI normalized gain,scientific reasoning ability, thinking in physics, and gendereffects, AIP Conf. Proc. , 23 (2012).[7] M. Lorenzo, C. Crouch, and E. Mazur, Reducing thegender gap in the physics classroom, Am. J. Phys. , 118(2006).[8] M. Cahill, K. Hynes, R. Trousil, L. Brooks, M. McDaniel,M. Repice, J. Zhao, and R. Frey, Multiyear, multi-instructor evaluation of a large-class interactive-engagement curriculum, Phys. Rev. ST Phys. Educ. Res. , 020101 (2014).[9] J. Day, J. Stang, N. Holmes, D. Kumar, and D. Bonn,Gender gaps and gendered action in a first-year physicslaboratory, Phys. Rev. Phys. Educ. Res. , 020104 (2016).[10] Z. Y. Kalender, E. Marshman, T. Nokes-Malach, C.Schunn, and C. Singh, Motivational characteristics ofunderrepresented ethnic and racial minority students inintroductory physics courses, Proceedings of the PhysicsEducation Research Conference, Cincinnati, OH, 2017 (AIP, New York, 2018), p. 204, https://doi.org/10.1119/perc.2017.pr.046.[11] T. Nokes-Malach, E. Marshman, Z. Y. Kalender, C.Schunn, and C. Singh, Investigation of male and femalestudents ’ motivational characteristics throughout an intro-ductory physics course sequence, Proceedings of thePhysics Education Research Conference, Cincinnati,OH, 2017 (AIP, New York, 2018), p. 276, https://doi.org/10.1119/perc.2017.pr.064.[12] A. Traxler and E. Brewe, Equity investigation of attitudinalshifts in introductory physics, Phys. Rev. ST Phys. Educ.Res. , 020132 (2015).[13] Z. Hazari, G. Potvin, R. Lock, F. Lung, G. Sonnert, and P.Sadler, Factors that affect the physical science career interest of female students: Testing five common hypoth-eses, Phys. Rev. ST Phys. Educ. Res. , 020115 (2013).[14] Z. Hazari, G. Sonnert, P. Sadler, and M. Shanahan,Connecting high school physics experiences, outcomeexpectations, physics identity, and physics career choice:A gender study, J. Res. Sci. Teach. , 978 (2010).[15] Z. Hazari, P. Sadler, and G. Sonnert, The science identity ofcollege students: Exploring the intersection of gender, race,and ethnicity, J. Coll. Sci. Teach. , 82 (2013).[16] M. Besterfield-Sacre, M. Moreno, L. Shuman, and C.Atman, Gender and ethnicity differences in freshmenengineering student attitudes: A cross-institutional study,J. Eng. Educ. , 477 (2001).[17] R. Felder, G. Felder, M. Mauney, C. Hamrin, and E. Dietz,A longitudinal study of engineering student performanceand retention. III. Gender differences in student perfor-mance and attitudes, J. Eng. Educ. , 151 (1995).[18] C. Moss-Racusin, J. Dovidio, V. Brescoll, M. Graham, andJ. Handelsman, Science faculty ’ s subtle gender biases favormale students, Proc. Natl. Acad. Sci. U.S.A. , 16474(2012).[19] E. Marshman, Z. Y. Kalender, C. Schunn, T. Nokes-Malach, and C. Singh, A longitudinal analysis of students ’ motivational characteristics in introductory physicscourses: Gender differences, Can. J. Phys. , 391 (2018).[20] N. I. Karim, A. Maries, and C. Singh, Do evidence-basedactive-engagement courses reduce the gender gap inintroductory physics?, Eur. J. Phys. , 025701 (2018).[21] N. Abramzon, P. Benson, E. Bertschinger, S. Blessing, G.Cochran, A. Cox, B. Cunningham, J. Galbraith-Frew, J.Johnson, L. Kerby, E. Lalanne, C. O ’ Donnell, S. Petty, S.Sampath, S. Seestrom, C. Singh, C. Spencer, K. SparksWoodle, and S. Yennello, Women in physics in the UnitedStates: Recruitment and retention, AIP Conf. Proc. ,060045 (2015).[22] C. Seron, S. Silbey, S. Cech, and B. Rubineau, Persistenceis cultural: Professional socialization and the reproductionof sex segregation, Work and Occupations , 178 (2016).[23] E. Seymour and N. M. Hewitt, Talking about Leaving:Why Undergraduates Leave the Sciences , (Westview Press,Boulder, CO, 1997).[24] S. Bates, R. Donnelly, C. MacPhee, D. Sands, M. Birch,and N. Walet, Gender differences in conceptual under-standing of Newtonian mechanics: A UK cross-institutioncomparison, Eur. J. Phys. , 421 (2013).[25] R. Henderson, G. Stewart, J. Stewart, L. Michaluk, and A.Traxler, Exploring the gender gap in the ConceptualSurvey of Electricity and Magnetism, Phys. Rev. Phys.Educ. Res. , 020114 (2017). IS AGREEING WITH A GENDER STEREOTYPE … PHYS. REV. PHYS. EDUC. RES.14,
We would like to thank the National Science Foundationfor Grants No. DUE-152457 and No. PHY-1505460 andthe members of the physics education research group at the University of Pittsburgh as well as R. P. Devaty for usefuldiscussions and feedback on the manuscript. We also thankL. Pingel and C. Schunn for helpful discussions related tostatistical analysis. [1] D. Hestenes, G. Wells, and M. Swackhammer, ForceConcept Inventory, Phys. Teach. , 141 (1992).[2] D. Maloney, T. O ’ Kuma, C. Hieggelke, and A. VanHeuvelen, Surveying students ’ conceptual knowledge ofelectricity and magnetism, Am. J. Phys. , s12 (2001).[3] A. Madsen, S. McKagan, and E. Sayre, Gender gap onconcept inventories in physics: What is consistent, what isinconsistent, and what factors influence the gap?, Phys.Rev. ST Phys. Educ. Res. , 020121 (2013).[4] J. Docktor and K. Heller, Gender Differences in Both ForceConcept Inventory and Introductory Physics Performance,AIP Conf. Proc. , 15 (2008).[5] A. Traxler, X. Cid, J. Blue, and R. Barthelemy, Enrichinggender in physics education research: A binary past and acomplex future, Phys. Rev. Phys. Educ. Res. , 020114(2016).[6] V. Coletta, J. Phillips, and J. Steinert, FCI normalized gain,scientific reasoning ability, thinking in physics, and gendereffects, AIP Conf. Proc. , 23 (2012).[7] M. Lorenzo, C. Crouch, and E. Mazur, Reducing thegender gap in the physics classroom, Am. J. Phys. , 118(2006).[8] M. Cahill, K. Hynes, R. Trousil, L. Brooks, M. McDaniel,M. Repice, J. Zhao, and R. Frey, Multiyear, multi-instructor evaluation of a large-class interactive-engagement curriculum, Phys. Rev. ST Phys. Educ. Res. , 020101 (2014).[9] J. Day, J. Stang, N. Holmes, D. Kumar, and D. Bonn,Gender gaps and gendered action in a first-year physicslaboratory, Phys. Rev. Phys. Educ. Res. , 020104 (2016).[10] Z. Y. Kalender, E. Marshman, T. Nokes-Malach, C.Schunn, and C. Singh, Motivational characteristics ofunderrepresented ethnic and racial minority students inintroductory physics courses, Proceedings of the PhysicsEducation Research Conference, Cincinnati, OH, 2017 (AIP, New York, 2018), p. 204, https://doi.org/10.1119/perc.2017.pr.046.[11] T. Nokes-Malach, E. Marshman, Z. Y. Kalender, C.Schunn, and C. Singh, Investigation of male and femalestudents ’ motivational characteristics throughout an intro-ductory physics course sequence, Proceedings of thePhysics Education Research Conference, Cincinnati,OH, 2017 (AIP, New York, 2018), p. 276, https://doi.org/10.1119/perc.2017.pr.064.[12] A. Traxler and E. Brewe, Equity investigation of attitudinalshifts in introductory physics, Phys. Rev. ST Phys. Educ.Res. , 020132 (2015).[13] Z. Hazari, G. Potvin, R. Lock, F. Lung, G. Sonnert, and P.Sadler, Factors that affect the physical science career interest of female students: Testing five common hypoth-eses, Phys. Rev. ST Phys. Educ. Res. , 020115 (2013).[14] Z. Hazari, G. Sonnert, P. Sadler, and M. Shanahan,Connecting high school physics experiences, outcomeexpectations, physics identity, and physics career choice:A gender study, J. Res. Sci. Teach. , 978 (2010).[15] Z. Hazari, P. Sadler, and G. Sonnert, The science identity ofcollege students: Exploring the intersection of gender, race,and ethnicity, J. Coll. Sci. Teach. , 82 (2013).[16] M. Besterfield-Sacre, M. Moreno, L. Shuman, and C.Atman, Gender and ethnicity differences in freshmenengineering student attitudes: A cross-institutional study,J. Eng. Educ. , 477 (2001).[17] R. Felder, G. Felder, M. Mauney, C. Hamrin, and E. Dietz,A longitudinal study of engineering student performanceand retention. III. Gender differences in student perfor-mance and attitudes, J. Eng. Educ. , 151 (1995).[18] C. Moss-Racusin, J. Dovidio, V. Brescoll, M. Graham, andJ. Handelsman, Science faculty ’ s subtle gender biases favormale students, Proc. Natl. Acad. Sci. U.S.A. , 16474(2012).[19] E. Marshman, Z. Y. Kalender, C. Schunn, T. Nokes-Malach, and C. Singh, A longitudinal analysis of students ’ motivational characteristics in introductory physicscourses: Gender differences, Can. J. Phys. , 391 (2018).[20] N. I. Karim, A. Maries, and C. Singh, Do evidence-basedactive-engagement courses reduce the gender gap inintroductory physics?, Eur. J. Phys. , 025701 (2018).[21] N. Abramzon, P. Benson, E. Bertschinger, S. Blessing, G.Cochran, A. Cox, B. Cunningham, J. Galbraith-Frew, J.Johnson, L. Kerby, E. Lalanne, C. O ’ Donnell, S. Petty, S.Sampath, S. Seestrom, C. Singh, C. Spencer, K. SparksWoodle, and S. Yennello, Women in physics in the UnitedStates: Recruitment and retention, AIP Conf. Proc. ,060045 (2015).[22] C. Seron, S. Silbey, S. Cech, and B. Rubineau, Persistenceis cultural: Professional socialization and the reproductionof sex segregation, Work and Occupations , 178 (2016).[23] E. Seymour and N. M. Hewitt, Talking about Leaving:Why Undergraduates Leave the Sciences , (Westview Press,Boulder, CO, 1997).[24] S. Bates, R. Donnelly, C. MacPhee, D. Sands, M. Birch,and N. Walet, Gender differences in conceptual under-standing of Newtonian mechanics: A UK cross-institutioncomparison, Eur. J. Phys. , 421 (2013).[25] R. Henderson, G. Stewart, J. Stewart, L. Michaluk, and A.Traxler, Exploring the gender gap in the ConceptualSurvey of Electricity and Magnetism, Phys. Rev. Phys.Educ. Res. , 020114 (2017). IS AGREEING WITH A GENDER STEREOTYPE … PHYS. REV. PHYS. EDUC. RES.14,
26] A. Traxler, R. Henderson, J. Stewart, G. Stewart, A. Papak,and R. Lindell, Gender fairness within the Force ConceptInventory, Phys. Rev. Phys. Educ. Res. , 010103 (2018).[27] T. Majors and P. Engelhardt, Gender & LEAP pedagogy:What does the Gender Force Concept Inventory have tosay?, Proceedings of the Physics Education ResearchConference, Minneapolis, MN, 2014 (AIP, New York,2014), p. 167.[28] G. Marchand and G. Taasoobshirazi, Stereotype threat andwomen ’ s performance in physics, Int. J. Sci. Educ. ,3050 (2013).[29] M. Appel and N. Kronberger, Stereotypes and the achieve-ment gap: Stereotype threat prior to test taking, Educ.Psychol. Rev. , 609 (2012).[30] L. Bian, S. Leslie, and A. Cimpian, Gender stereotypesabout intellectual ability emerge early and influence child-ren ’ s interests, Science , 389 (2017).[31] A. Bandura, Self-efficacy: Toward a unifying theory ofbehavioral change, Psychol. Rev. , 191 (1977).[32] C. Dweck, Is Math a Gift? Beliefs that Put Females at Risk,in Why aren ’ t more women in science? Top researchersdebate the evidence, Washington, DC, 2006 , edited by S. J.Ceci and W. Williams (American Psychological Associa-tion, Washington, DC, 2006).[33] C. Dweck, Mindset: The New Psychology of Success (Random House Inc., New York, 2006).[34] S. Wheeler and R. Petty, The effects of stereotypeactivation on behavior: A review of possible mechanisms,Psychol. Bull. , 797 (2001).[35] M. Inzlicht and T. Schmader,
Stereotype Threat: Theory,Process, and Application (Oxford University Press,Oxford, 2012).[36] I. Dar-Nimrod and S. J. Heine, Exposure to scientifictheories affects women ’ s math performance, Science , 435 (2006).[37] J. W. Osborne, Gender, stereotype threat, and anxiety:Psychophysiological, and cognitive evidence, J. Res. Educ.Psychol. , 109 (2006); P. G. Davies, S. J. Spencer, D. M.Quinn, and R. Gerhardstein, Consuming images: Howtelevision commercials that elicit stereotype threat canrestrain women academically and professionally, Pers. Soc.Psychol. Bull. , 1615 (2002).[38] R. P. Brown and E. A. Day, The difference isn ’ t black andwhite: Stereotype threat and the race gap on Raven ’ sadvanced progressive matrices, J. Appl. Psychol. ,979 (2006).[39] K. E. Ryan and A. M. Ryan, Psychological processesunderlying stereotype threat and standardized math testperformance. Educ. Psychol. , 53 (2005).[40] P. Huguet and I. R´egner, Stereotype threat among school-girls in quasi-ordinary classroom circumstances, J. Educ.Psychol. , 545 (2007).[41] M. Cadinu, A. Maass, A. Rosabianca, and J. Kiesner, Whydo women underperform under stereotype threat?, Psychol.Sci. , 572 (2005).[42] R. B. McIntyre, C. G. Lord, D. M. Gresky, L. L. Ten Eyck,G. D. J. Frye, and C. F. Bond Jr., A social impact trend in the effects of role models on alleviating women ’ s math-ematics stereotype threat. Curr. Res. Social Psychol. ,116 (2005).[43] T. Schmader, Stereotype threat deconstructed, Curr. Dir.Psychol. Sci. , 14 (2010).[44] S. Spence, C. M. Steele, and D. M. Quinn, Stereotypethreat and women ’ s math performance, J. Exp. Soc.Psychol. , 4 (1999).[45] C. Steele and J. Aronson, Stereotype threat and theintellectual test performance of African Americans, J. Pers.Soc. Psychol. , 797 (1995).[46] C. Steele, A threat in the air: How stereotypes shape theintellectual identities and performance of women andAfrican-Americans, Am. Psychol. , 613 (1997).[47] M. Shih, T. L. Pittinsky, and N. Ambady, Stereotypesusceptibility: Identity salience and shifts in quantitativeperformance, Psychol. Sci. , 80 (1999).[48] L. Strickler and C. Ward, Stereotype threat, inquiringabout test takers ’ ethnicity and gender, and standard-ized test performance, J. Appl. Soc. Psychol. , 665(2004).[49] A. Maries and C. Singh, Stereotype threat? Effects ofinquiring about test takers ’ gender on conceptual testperformance in physics, AIP Conf. Proc. , 120008(2015).[50] C. Good, J. Aronson, and M. Inzlicht, Improving adoles-cents ’ standardized test performance: An intervention toreduce the effects of stereotype threat, J. Appl. Dev.Psychol. , 645 (2003).[51] A. Miyake, L. E. Kost-Smith, N. D. Finkelstein, S. J.Pollock, G. L. Cohen, and T. A. Ito, Reducing the genderachievement gap in college science: a classroom study ofvalues affirmation, Science , 1234 (2010).[52] D. S. Yeager and G. M. Walton, Social-psychologicalinterventions in education: They ’ re not magic, Rev. Educ.Res. , 267 (2011).[53] G. V. Glass and K. D. Hopkins, Statistical Methods inEducation and Psychology , 3rd ed. (Allyn & Bacon,Boston, MA, 1996).[54] H. Simon,
Models of Thought (Yale University Press,New Haven, CT, 1979), Vols. 1 and 2.[55] E. Maloney, J. Sattizahn, and S. L. Beilock, Anxiety andCognition, WIREs Cognitive Science , 403 (2014).[56] S. Beilock, W. A. Jellison, R. J. Rydell, A. R. McConnell,and T. H. Carr, On the causal mechanisms of stereotypethreat: Can skills that don ’ t rely heavily on workingmemory still be threatened?, Pers. Soc. Psychol. Bull. , 1059 (2006).[57] S. Beilock, R. J. Rydell, and A. R. McConnell, Stereotypethreat and working memory: Mechanisms, alleviations, andspillover, J. Exp. Psychol. Gen. , 256 (2007).[58] T. Schmader and M. Johns, Converging evidence thatstereotype threat reduces working memory capacity,J. Pers. Soc. Psychol. , 440 (2003).[59] C. McKown and R. Weinstein, The development andconsequences of stereotype consciousness in middlechildhood, Child Development , 498 (2003). MARIES, KARIM, and SINGH PHYS. REV. PHYS. EDUC. RES.14,